MicroPathID: Development of an In-Field Microfluidic System for Detection of Pathogens

UNIVERSIDADE DE LISBOA INSTITUTO SUPERIOR TÉCNICO MicroPathID: Development of an In-Field Microfluidic System for Detection of Pathogens Catarina Raquel Fernandes Caneira Supervisor: Doctor João Pedro Estrela Rodrigues Conde Co-supervisor: Doctor Elisabete Ramos Fernandes Thesis approved in public session to obtain the PhD Degree in Biomedical Engineering Jury final classification: Pass with Distinction 2023 i ii UNIVERSIDADE DE LISBOA INSTITUTO SUPERIOR TÉCNICO MicroPathID: Development of an In-Field Microfluidic System for Detection of Pathogens Catarina Raquel Fernandes Caneira Supervisor: Doctor João Pedro Estrela Rodrigues Conde Co-supervisor: Doctor Elisabete Ramos Fernandes Thesis approved in public session to obtain the PhD Degree in Biomedical Engineering Jury final classification: Pass with Distinction Jury Chairperson: Doctor Duarte Miguel de França Teixeira dos Prazeres, Instituto Superior Técnico da Universidade de Lisboa Members of the Committee: Doctor Pedro Miguel Ribeiro Viana Baptista, Faculdade de Ciências e Tecnologia Universidade Nova de Lisboa; Doctor João Pedro Estrela Rodrigues Conde, Instituto Superior Técnico Universidade de Lisboa; Doctor Duarte Miguel de França Teixeira dos Prazeres, Instituto Superior Técnico Universidade de Lisboa; Doctor Domenico Caputo, Dipartimento di Ingegneria dell'Informazione, Elettronica e Telecomunicazioni, Sapienza Università di Roma, Italy; Doctor Frederico Castelo Alves Ferreira, Instituto Superior técnico Universidade de Lisboa. Funding Institution Fundação para a Ciência e Tecnologia (FCT) – PhD Grant PD/BD/135274/2017 2023 iii iv Abstract The detection of pathogens in primary care settings is increasingly critical, particularly in the fight against antimicrobial resistance, and has applications in areas such as food and feed safety, environmental monitoring and control, biodefense, to animal and human health. Currently, the methods used for detection are time-consuming or expensive and require specialized personnel, an elevated amount of hands-on time, and laboratory infrastructures. The increased burden of antimicrobial resistance and the predictions of future impact on human health motivate an increasing demand for a fast, simple, portable, cost-effective, sensitive, and specific device for the early detection of antibiotic-resistant bacteria, with application in the early detection in hospital environments. This thesis reports the development of several miniaturized modules that included (1) cell lysis and (2) nucleic acid extraction, (3) amplification, and (4) optical target detection using miniaturized photosensors for optical transduction and signal acquisition to perform bacterial detection based on molecular assays. The use of nanoporous microbeads on the microfluidic device allowed for device simplification and an increase in sensitivity. Fluidic control and sensor integration can also be employed in the device to achieve a sample-to-answer system. We were able to achieve a lysis module capable of lysing bacterial and virus-like particles, and successfully extracting DNA/RNA in under 20 min. The development of an isothermal amplification module allowed the detection of sub-femtomolar concentrations of DNA on a chip in less than 2 hours targeting a specific gene that allows the identification of the presence of Staphylococcus aureus. These developments contribute significantly to the state-of-the-art and pave the way toward the development of a fully integrated molecular-based device, to be used at the point-of-care to allow rapid detection and screening of patients that present antibiotic-resistant bacteria. Keywords Nucleic acid detection, antibiotic resistance, microfluidics, nanoporous microbeads, integration v vi Resumo A deteção de patógenos em ambientes de cuidados primários de saúde é cada vez mais crítica, particularmente no combate à resistência antimicrobiana mas tem aplicações em áreas tão vastas como segurança de alimentos e rações, monitoramento e controlo ambiental, biodefesa e saúde animal e humana. Atualmente, os métodos utilizados para a deteção são demorados ou caros, requerem utilizadores especializados, uma quantidade elevada de tempo de trabalho e infraestruturas laboratoriais. O aumento da carga de resistência antimicrobiana e as previsões de impacto na saúde humana num futuro próximo, motivaram uma procura crescente por um dispositivo rápido, simples, portátil, econômico, sensível e específico para a deteção precoce de bactérias resistentes a antibióticos, em particular, com importante aplicação na deteção antecipada e rápida destas bactérias em ambiente hospitalar. Esta tese descreve o desenvolvimento de vários módulos miniaturizados que incluíram (1) lise celular e (2) extração de ácidos nucleicos, (3) amplificação de material genético e (4) deteção ótica da molécula de interesse usando fotodetectores miniaturizados para transdução ótica e aquisição de sinal. O uso de microesferas nanoporosas no dispositivo microfluídico permitiu a simplificação do dispositivo e aumento da sensibilidade. Sistemas de controlo fluídico e sensores também podem ser integrados ao dispositivo para obter um sistema verdadeiramente da amostra para resposta. Conseguimos obter um módulo de lise capaz de lisar bactérias e partículas semelhantes a vírus e efetuar extração e DNA/RNA em menos de 20 min com sucesso. O desenvolvimento de um módulo de amplificação isotérmica permitiu a deteção no chip levando à confirmação da presença de Staphylococcus aureus através de um gene específico. Estes desenvolvimentos contribuíram significativamente para o estado da arte e abrem caminho para o desenvolvimento de um dispositivo, totalmente integrado, baseado em testes moleculares, para ser usado no local de atendimento o que permitirá uma rápida deteção e triagem de doentes com a presença de bactérias resistentes a antibióticos. Palavras-Chave Deteção de ácidos nucleicos, resistência a antibióticos, microfluídica, microesferas nanoporosas, integração vii viii Acknowledgements I express my most sincere thanks to all who made this journey possible. To my supervisor, Professor João Pedro Conde, I am deeply grateful for the opportunities given to me to explore my academic career, from the master's thesis to the moment of this doctoral thesis. Thank you for the scientific guidance, for believing in my abilities along the way, and for the patience and understanding I always felt I received in good and bad times. It will forever be a reference for me of what it means to be an excellent Teacher, Advisor, and Human Being. I would also like to thank Dr. Virgínia Chu for her availability to always discuss my project, for her precious help and suggestions for the dissemination of the scientific work that was being carried out, and for having also been an example and inspiration for me how it is possible to manage a research group. I want to thank both of them for welcoming me into what I consider to be more than the BioMEMS research group at INESC MN, but my scientific family. A family that helped me grow as a researcher and as a person. To my co-supervisor at INL, Doctor Elisabete Fernandes, I thank her for her sympathy, availability to talk, and good heart. Of all our encounters, she was always available to help in any way he could. Thank you so much for this. In addition to my formal supervisors, I could not fail to mention Dr. Silvia Monteiro and Dr. Ricardo from LAIST for having welcomed me and sharing countless discussions about my project, science, life, and so many other topics with me. Thank you for your scientific guidance, company, and friendship. To my former colleagues and friends at INESC MN. In particular, my deepest thanks to Dr. Ruben Soares for having been a mentor, a scientist I greatly admire and who inspires me, and a good friend. Professionalism, focus and determination, scientific acumen, rectitude, and good heart are characteristics that define him and that I know have led him to fly high. And, of course, I would like to thank my friend and companion during the afternoons of experience at Biolabo, Dr. Inês Pinto. It was a pleasure and a privilege to meet Inês and work alongside her. A scientist and person that I will always look up to and that I know will have a bright future. To Doctor Eduardo Brás, thank you for the company, friendship, and philosophical and nerd conversations. I would also like to thank Dr. Narayanan Madaboosi Srinivasan for being a reference to what it meant to be an excellent researcher when I started working at INESC MN. To all the other people who have been in the group over the years and who have contributed to making it a family, Dr. Denis Santos, Dr. Rui Pinto, Dr. Ricardo Fradique, Cristiana xv 2.5. Fundamentals of microfluidics .............................................................................. 40 2.6. Assay requirements to develop a fit for purpose biosensor ................................... 44 3. General fabrication and experimental methods ................................................ 49 3.1. Overview of a microfluidic device fabrication process ........................................... 49 3.2. Fabrication of a-Si:H p-i-n photodiodes ................................................................. 56 3.3. Microfluidic structure general handling ................................................................. 57 3.4. Optical signal transduction .................................................................................... 58 4. Development of a rapid bead-based microfluidic platform for DNA hybridization using singleand multi-mode interactions for probe immobilization ................................. 59 4.1. Introduction .......................................................................................................... 59 4.2. Materials and experimental methods ................................................................... 61 4.3. Results and Discussion .......................................................................................... 68 4.4. Chapter conclusions .............................................................................................. 80 5. Regenerable Bead-based Microfluidic Device with Integrated Thin-film Photodiodes for Real-time Monitoring of DNA Detection ............................................................. 83 5.1. Introduction .......................................................................................................... 84 5.2. Materials and experimental methods ................................................................... 86 5.3. Results and Discussion .......................................................................................... 92 5.4. Chapter Conclusions ............................................................................................. 99 6. Silica bead-based microfluidic device with integrated photodiodes for the rapid capture and detection of rolling circle amplification products in the femtomolar range 103 6.1. Introduction ........................................................................................................ 104 6.2. Materials and experimental methods ................................................................. 105 6.3. Results and Discussion ........................................................................................ 111 6.4. Chapter Conclusions ........................................................................................... 116 xvi 7. Chemical lysis and nucleic acid capture of Gram-positive, Gram-negative, and viruslike particles using a bead-based microfluidic chip ................................................. 119 7.1. Introduction ........................................................................................................ 120 7.2. Materials and experimental methods ................................................................. 122 7.3. Results and discussion ........................................................................................ 129 7.4. Chapter Conclusions ........................................................................................... 142 8. Systematic implementation of padlock probing-based rolling circle amplification in an integrated microfluidic device for quantitative biomolecular analyses ................... 145 8.1. Introduction ........................................................................................................ 146 8.2. Materials and experimental methods ................................................................. 148 8.3. Results and discussion ........................................................................................ 160 8.4. Chapter Conclusions ........................................................................................... 176 9. Conclusion and Future developments .............................................................. 177 9.1. Novelty and main improvements to the state-of-the-art ..................................... 177 9.2. Future developments toward a fully integrated microfluidic system for the detection of pathogens ...................................................................................................................... 178 Bibliography .......................................................................................................... 181 xvii xviii List of Figures Figure 1.1 - Research project timeline. ........................................................................................................ 5 Figure 1.2 - Doctorate structure. .................................................................................................................. 6 Figure 2.1 - Schematics and comparison of pathogen detection using conventional laboratory detection methods - A - and PoC methods - B. .................................................................................................. 13 Figure 2.2 – Schematics of an ideal µTAS device for nucleic acid detection. ............................................. 14 Figure 2.3 – Sample types and possible interferents of each sample type that may interfere with downstream processes which make sample preparation I crucial for the success of assay development. ........... 17 Figure 2.4 – Types of cell lysis that includes mechanical methods ............................................................ 18 Figure 2.5 – Optical e non-optical detection methods used in µTAS device. ............................................. 28 Figure 2.6 - Representative illustration of the cell wall components of gram-negative, gram-positive bacteria and non-enveloped and enveloped viruses. ...................................................................................... 40 Figure 2.7 - Biosensor characterization.. .................................................................................................... 45 Figure 3.1 – Schematics of the microfabrication process to obtain the PDMS microfluidic devices. ........ 51 Figure 3.2 - Schematics of the thin-film a-Si:H photodiodes. ..................................................................... 57 Figure 4.1 - General overview of the microfluidic setup and the DNA detection strategies using nanoporous microbeads. ....................................................................................................................................... 62 Figure 4.2 – Schematics of the assays for blocking optimization using Q Sepharose beads. ..................... 65 Figure 4.3 - Summary of the hybridization assays performed for the three different optical detection labels used, fluorescence using ATTO 430LS coupled with Q SepharoseTM and CaptoTM Adhere beads, QdotsTM 605 coupled with CaptoTM Adhere beads and chemiluminescence using horseradish peroxidase (HRP) coupled with Q SepharoseTM beads. .................................................................................................. 66 Figure 4.4 - Schematics of the mass balance method used to calculate the mass of probe DNA immobilized on the beads and captured target DNA. ................................................................................................. 68 Figure 4.5 - Quantification of the target DNA non-specifically bound to the PDMS walls for (A) cDNA and (B) ncDNA.. .............................................................................................................................................. 68 Figure 4.6 - BSA-FITC adsorption isotherms and effect of conductivity on molecular adsorption. ............ 70 Figure 4.7 - Optimization of probe DNA-FITC immobilization on a positively charged surface, silanized with (3Aminopropyl)triethoxysilane (APTES). ............................................................................................... 71 Figure 4.8 - Optimization of the blocking time. .......................................................................................... 72 Figure 4.9 - Effect of different blocking agents on blocking efficacy and hybridization selectivity on QS beads. ........................................................................................................................................................... 73 Figure 4.10 - Comparison of fluorescence intensity of probe DNA immobilized on QS beads and CA beads before and after the blocking step. ................................................................................................... 74 Figure 4.11 - (A,B) Calibration curves for known concentrations of DNA, used to determine !"#$%&' . 75 xix Figure 4.12 - Biosensing of DNA in QS or CA beads using three different optical detection labels: (A) fluorescence using Atto 430LS in QS or QA bead-based assays, (B) Qdots labels and CA beads or (C) chemiluminescence using HRP and QA beads. .................................................................................. 77 Figure 4.13 - Comparison of background fluorescence for QS beads and CA beads before and after the blocking step. ................................................................................................................................................... 78 Figure 4.14 - Comparison of the chemiluminescence signal profile when using CA beads or QS beads. .. 79 Figure 5.1 - Bead-based microfluidic assays. .............................................................................................. 88 Figure 5.2 - Integrated optical transduction after target capture in regeneration assays. ........................ 91 Figure 5.3 - Optimization of microcolumn regeneration conditions. ......................................................... 94 Figure 5.4 - Measurements of the hybridization assay cycles, including detection of specific complementary target DNA, column regeneration, non-complementary target DNA control, and column regeneration. ........................................................................................................................................................... 97 Figure 5.5 - Fluorescence A and B and chemiluminescence C and D monitoring using photosensors. ..... 98 Figure 6.1 - Schematics of the off-chip target detection and amplification, followed by on-chip RCA product (RCP) capture and fluorescence signal transduction. ...................................................................... 110 Figure 6.2 A - Bright field (top) and fluorescence (bottom) microscopy imaging of the silica bead microcolumn and captured RCPs and B - flow-rate optimization. ......................................................................... 112 Figure 6.3 - Monitoring of RCP capture on the microfluidic device using the integrated photodiode .... 114 Figure 6.4 - Calibration curve for increasing target RNA concentrations and correlation with a commercial RCP counter (Q-linea Aquila 400).. ......................................................................................................... 115 Figure 6.5 - Detection of Ebola vRNA using RCA and the miniaturized read-out device. ......................... 116 Figure 7.1 - Schematics of the microfluidic device for pathogens lysis and genetic material detection. . 124 Figure 7.2 - Schematics of the protocol to obtain the total copies of genetic material captured by the beads packed in the microfluidic chamber. ............................................................................................... 128 Figure 7.3 - Confirmation of results using TSA plates. (A) and (B) BPER lysis of Gram-positive and negative bacteria. (C) and (D) Combination of Genolyse and BPER solutions to perform chemical lysis. ..... 131 Figure 7.4 - SEM micrographs. .................................................................................................................. 132 Figure 7.5 - Fluorescence results of the controls for the EvaGreen® detection. ...................................... 133 Figure 7.6 - Calibration curve of the lysis device for Gram-positive and Gram-negative bacteria using a chemical lysis strategy (Mixture of GenoLyse® (GL) and B-PPERTM (BPER) solutions). ................................... 134 Figure 7.7 - Summary of the lysis results obtained using a mass balance approach for quantification. .. 139 Figure 8.1 - (A) Schematics of the microfluidic device used for rolling circle amplification (RCA) on-chip.149 Figure 8.2 - Half-normal plot for the various effects and interactions. .................................................... 161 Figure 8.3 - Calibration curve for the detection of biotinylated S. aureus target DNA using a solid-phase capture and rolling circle amplification on-chip. .......................................................................................... 163 Figure 8.4 - Flow rate optimization for the capture of ssDNA. ................................................................. 164 xx Figure 8.5 - Target single-strand DNA capture studies. ............................................................................ 165 Figure 8.6 - Capture capacity for the first column of Q Sepharose beads and a second column. ............ 166 Figure 8.7 - Detection of rolling circle amplification products for strategy A and strategy B. ................. 171 Figure 8.8 - Comparison of PLP-RCA performed on a well plate and on-chip for both strategies A and B.174 Figure 8.9 - Fluorescence signal obtained by microscopy and image analysis for the different strategies used to detect S. aureus gDNA. .................................................................................................................... 175 xxi List of Tables Table 1.1Structure of this report. ............................................................................................................... 6 Table 2.1– PoC devices in the market for pathogen detection using nucleic acid amplification. .............. 33 Table 2.2 – The leading pathogens of that cause hospital acquired infections and relevant characteristics. 38 Table 3.1 - Materials and equipment required for device fabrication and operation. .............................. 53 Table 3.2 - General timeline of the sequence of steps required to perform the fabrication of the microfluidic device. ............................................................................................................................................... 55 Table 4.1 - Description of the oligonucleotide sequences used in this work with respective modifications.62 Table 4.2 - Description of on-chip steps performed in the microfluidic device with the respective flow rates and assay times for fluorescence and chemiluminescence assays. Steps that were not applied in a given assay are identified as not applicable (n.a.). Off-chip steps are descried in the text. ................................. 63 Table 5.1 - Description of the oligonucleotide sequences used in this work and respective modifications.87 Table 5.2 - Critical comparison of similar methodologies in the literature. ............................................. 100 Table 7.1 - Real-time RT-qPCR primers and fluorogenic probes sequences ............................................ 126 Table 7.2 - Summary of the quantitive results obtained for the different lysis treatments. ................... 137 Table 8.1 - DNA oligonucleotide sequences used throughout this work to perform RCA. ...................... 151 Table 8.2 - Description of on-chip steps performed in the microfluidic device with the respective flow rates and assay times to detect ssDNA with model strategy. ......................................................................... 153 Table 8.3 - Description of on-chip steps performed in the microfluidic device with the respective flow rates and assay times for the different strategies tested to detect ssDNA. .................................................... 154 Table 8.4 - Concentration of reagents used in each step of the PLP-RCA assay, and flow conditions. .... 156 Table 8.5 - Description of on-chip steps performed in the microfluidic device with the respective flow rates and assay times for the different strategies tested to detect gDNA. ..................................................... 156 Table 8.6 - Description of on-well RCA for gDNA, solutions prepared for each step, and incubation times on well. ................................................................................................................................................. 158 Table 8.7 - Fractional factorial design. ..................................................................................................... 160 xxii List of acronyms and abbreviations µ TAS Micro total analysis systems a-Si:H Hydrogenated amorphous silicon a-SiC:H Hydrogenated amorphous silicon carbide A. baumanii Acinetobacter baumanii AC Alternating current ACK Ammonium-Chloride-Potassium ACV Alternating current voltammetry Al Aluminum AMF Alternate current magnetic field AMR Antimicrobial resistance ARB Antibiotic-resistant bacteria ASMD Amplified single-molecule detection ASSURED Affordable, Sensitive, Specific, User-friendly, Rapid and robust, Equipment-free, and Deliverable ATP Adenosine triphosphate AU Arbitrary unitts B. cereus Bacillus cereus BHQ Black Hole Quencher BP Band pass BPER Bacterial protein extraction reagent BSA Bovine serum albumin C2CA Circle to circle amplification CA Capto Adhere CCD Charged-coupled device cDNA Complementary deoxyribonucleic acid xxiii CFU Colony forming units COVID-19 Coronavirus disease CPA Cross-priming amplification CRE Carbapenem-resistant Enterobacteriaceae Ct Cycle threshold CV Coefficient of variation DC Direct current DI Deionized DNA Deoxyribonucleic acid dNTPs Deoxynucleotide triphosphates DO Detection oligo DoE Design of experiment dsDNA Double-stranded deoxyribonucleic acid DWL Direct write lithography E. coli Escherichia coli EDTA Ethylenediamine tetraacetic acid EG EvaGreen dye ESKAPE-E Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp and Escherichia coli EXPAR Exponential amplification reaction FAM 6-Carboxyfluorescein FITC Fluorescein isothiocyanate GAS Group A streptococcus GC Genome copies gDNA Genomic deoxyribonucleic acid GL GenoLyse xxiv GPIB General purpose interface bus GUI Graphical user interface HIV Human immunodeficiency virus HRP Horseradish peroxidase IAV/IBV Influenza virus A/B IPA Isopropyl Alcohol ITO Indium tin oxide ITP Isotachophoresis JOE 4-5-Dichloro carboxy fluorescein LAMP Loop-mediated isothermal amplification LED Light-emitting diode LFA Lateral flow assay LLE Liquid-liquid extraction LNA Locked nucleic acid LoB Limit of blank LoC Lab-on-a-chip LoD Limit of detection LoQ Limit of quantification LP Long pass LS Lysis solution MDRTB Multi-drug resistant tuberculosis MERS-CoV Middle east respiratory syndrome coronavirus MRSA methicillin-resistant Staphylococcus aureus MTB Mycobacterium tuberculosis NA Nucleic acids NAAT Nucleic acid amplification test 4 1.3. Development of an amplification module. 1.4. Development of optical detection module. 1.5. To perform pre-validation against the gold standard techniques in central laboratories for each module. 1.3.1.2. Milestones M1.1. Sensitivity curve for fluorescence using integrated photodiodes. M1.2. Demonstration of the capture of the DNA in microbeads trapped in the microfluidic channel. M1.3. Demonstration of bacterial lysis in a microfluidic system. M1.4. Demonstration of rolling circle amplification (RCA) in a microfluidic system. 1.3.2. Task 2 1.3.2.1. Objectives 2. To integrate the individual modules successfully towards a complete analysis platform still using model solutions and conditions. To test and evaluate the device requirements. 2.1. Development of an integrated module for DNA amplification and detection. Photodiodes should be included to perform optical signal transduction. 2.2. Coupling of DNA extraction to the device obtained in 2.1. 2.3. Addition of the module of cell lysis and pre-concentration to the previous devices. 2.4. To test the integrated system for sensitivity, specificity, and time of analysis, using a model solution with a different titter of the bacteria of interest and an other bacterium. 2.5. To perform the pre-validation of the integrated system against the gold standard molecular biology methods. 1.3.2.2. Milestones M2.1. Demonstration of integrated detection of DNA amplification. M2.2. Demonstration of amplification from on-chip captured DNA. M2.3. Demonstration of DNA detection from bacterial cell culture. 5 1.3.3. Task 3 1.3.3.1. Objectives 3. Perform a test of the device in a clinical environment. 3.1. To optimize the integrated system, namely for sensitivity and time of analysis. 3.2. Detection of positive spiked samples (swab sample) with a different titter. 3.3. Test in clinical environment done by trained hospital laboratory technical staff. 1.3.3.2. Milestones M.3.1 Demonstration of DNA detection from positive clinical samples and assay verification with negative control samples. The milestones are identified in the timeline presented in the following subsection. The milestone's original delivery time is first represented in solid blue color. When the expected deliverable time was not accomplished, the effective delivery time is represented in front in a lighter blue (Figure 1.1). 1.4. Research Project Timeline This section presents a Gantt chart with the project timeline. Figure 1.1 - Research project timeline. 6 1.5. Research Project Structure This section schematizes the project development and how each of the chapters fit in in the modules required for the device. The research doctorate structure is illustrated below in Figure 1.2. Figure 1.2 - Doctorate structure. 1.6. Structure of this thesis In this section, the structure of the thesis is presented in Table 1.1. Table 1.1Structure of this report. Chapter Description 1 Ph.D. project contextualization Introduces the challenges of detecting pathogens at the point of care and presents the aims, objectives, and structure of the project to develop a novel integrated microfluidic device. 2 Introduction Presents a more detailed review on the problem, state of the art and theoretical concepts required to develop the project. 3 General fabrication and experimental methods Presents the general fabrication, such as standard fabrication techniques, and experimental method used throughout the project. 4 Bead-based microfluidic platform for DNA hybridization Presents the rationale for the use of the specific methodologies for the development of a bead-based module to detect DNA via hybridization, and the results obtained. [Article published] 7 5 Regenerable beadbased microfluidic platform for DNA detection Presents the development of a microfluidic device able to detect DNA and perform regeneration. Integration of photosensors was also achieved. [Article published] 6 Bead-based microfluidic device for the capture and detection of DNA amplification products Presents the results of a bead-based device developed to capture product of an isothermal amplification technique. Integration of photosensors was also achieved. [Article published] 7 Bead-based microfluidic chip for chemical lysis and nucleic acid detection of bacteria and virus-like particles Presents the results of the development of the lysis module. Extraction of nucleic acids was also achieved in this chapter. [Article submitted to publication] 8 Rolling circle amplification in an integrated microfluidic device for quantitative biomolecular analyses Presents the results from the development of the amplification module using an isothermal technique on-chip (RCA). [Article submitted to publication] 9 Conclusion and Future developments Discusses the research impact of the work, concludes with remarks around the essential findings and contributions of this work and future developments. 8 9 Chapter 2 Introduction 2. Introduction This introductory chapter describes and discusses micro total analysis systems (µTAS) and the several modules required to identify pathogens via nucleic acid detection. We present a comprehensive analysis of the several modules required and enhanced relevant works in the literature demonstrating these modules. We also review commercially available µTAS highlighting the most critical features and gaps that can be filled with new detection devices. Furthermore, we highlighted the pathogens more prominent in nosocomial infections and pathogens characteristics that can influence the detection performance of µTAS. Finally, we lay the theoretical background that allows us to study microfluidics and characterize biosensors. 2.1. The technology, the problem, and the proposed solution – A brief description Micro total analysis systems (µTAS), also called lab-on-a-chip (LoC), are defined as systems that include all the steps required to perform biochemical analysis of a sample. This includes all the steps usually performed in a standard laboratory (e.g., sample preparation, purification, preconcentration, chemical reaction or biological event, and detection) in an automated, miniaturized, single device [4]. The first µTAS was reported in 1979 by Terry et al. and consisted of a miniaturized gas chromatographic analyzer fabricated on silicon [5]. However, with the novel work of Manz et al., published in 1990, the scientific community got interested in the technology. Manz described a miniaturized high-pressure liquid chromatography system fabricated on a silicon wafer and simultaneously proposed the concept of µTAS [6], [7]. By that time, the main argument for pursuing miniaturization of the chemical analytical systems was enhancing the analytical performance. Nonetheless, it was also acknowledged that miniaturization would lower reagent consumption and possibly monitor several components using a single device. Such devices soon saw their field of applications broadening from analytical chemistry to biomedical [8]–[10], food and environmental control [11]–[14], or biosafety [15]. Over the years, LoC devices have demonstrated the potential to be used in loco, 10 responding to a pressing need to decentralize the sample analysis in the fields of applications mentioned. Furthermore, improving the analytical turnaround time was a significant advance for these devices compared to standard laboratory techniques. This led to the concepts of point-of-care and point-of-need. A point-of-care (PoC) test is usually defined as a diagnostic test performed near or at the actual patient site, producing a rapid and reliable result (turnaround time of up to 1 hour) used in urgent patient care decisions. For instance, short turnaround times are crucial for patient triage in healthcare settings. Similarly, point-of-need (PoN) testing is commonly used when the application and sample is not a patient but an environmental, food, or biosafety sample and on-site (e.g., municipal sewage and water supply, farms, crop fields, etc.). In this case, rapid analysis response is vital to contain pathogen dissemination in animals, crops, or even through the air. LoC devices used for PoC testing and PoN testing should follow the ASSURED criteria implemented by the WHO in 2003 [16], [17]. The requirements establish a general guideline for benchmarking a diagnostic test for a given application. This can be useful when selecting a PoC/PoN testing device already available in the market and when developing a new device by trying to fulfill the ASSURED requirements. The ASSURED criteria stand for Affordable, Sensitive, Specific, User-friendly, Rapid and robust, Equipment-free, and Deliverable to endusers. It should be affordable to those that require the test, sensitive enough to display few or no false negatives, and specific to avoid false positives. It should deliver a fast result to enable an immediate medical decision on triage or treatment to perform and be robust enough to withstand the changes in the environment, such as temperature altering the assay result. Moreover, it should be easy enough to perform, requiring minimal training, equipmentfree, or minimal equipment that can be portable, and demanding no plug into an electric outlet. The acronym REASSURED was proposed to accompany the technology's evolution [18]. Real-time connectivity should be embedded in the device, and the device should be connected to a display showing real-time test results, or a cell phone can be used to receive and analyze the data. Finally, the Ease of specimen collection should be considered, and if possible, the collection should be non-invasive or minimally invasive. Another concern coupled with the requirement of Equipment free or simple is that ideally, the tests should be Environment friendly and fabricated from recycled material. After completion, they should be easy to dispose of. 11 In this spirit, microfluidics technology enables LoC devices for PoC/PoN testing. Microfluidic technology is fluid manipulation in channel design with at least one channel's dimensions within tens of micrometers [19]. And while at the macroscale, inertial forces such as gravity dominate the fluid mechanics, at the microscale, it is the viscosity that most influences the fluid regime. Microfluidics physics is governed by different forces than those that govern fluids at the macroscale mainly because as the systems go down in size, we see an increased dominance of surface effect over volumetric effects. This is a crucial feature of microfluidics devices, a high surface-tovolume ratio typically in 106 m. Thus, the fluid mechanics at this scale is controlled by viscous forces, surface tension, and electrostatic/electrodynamics forces [20]. Usually, when working with human clinical samples, analyte concentration varies between 1014 to 1021 copies/mL. For immunoassays is between 108 to 1018 copies/mL, and detection of DNA for genomic assays, bacteria, or virus detection ranges between 102 to 107 copies/mL [21]. Depending on the analyte concentration, we can decrease the sample volume from femtoliter (with higher analyte concentration) to milliliters range (for lower analyte concentration). Therefore, when working with nucleic acid detection in microfluidics, steps of preconcentration and amplification are required. Nevertheless, microfluidic devices offer several advantages for detecting bacterial pathogens. These systems allow the reduction of sample volume and reagents consumption (10–6 to 10–18 liters), increased speed of reactions, increased sensitivity, a decrease in cost, and simplification of the detection process by developing autonomous systems that allow for pathogen isolation, lysis and nucleic acid extraction, nucleic acid amplification and amplicon detection in one single chip [22], [23]. Pathogens can be defined as a microorganism or an agent of infection that causes infectious disease, also referred to as communicable disease. The most common pathogens include parasites, fungi, bacteria, and viruses. Overall, the spread of these pathogens increases due to global traveling and the growing population, increasing the risk of an infectious outbreak. Many specialists predicted that a major pandemic was sure to happen before COVID-19, caused by SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2), and now we are experiencing the effects of not having prepared for such a scenario. In the last report from WHO, lower respiratory infections remained the world's most deadly communicable disease (the 4th leading cause of death globally). And according to the report, infectious diseases are more likely to lead to death in low-income countries, primarily due to poor hygiene and health conditions and barriers to health care services [24], [25]. However, there is also 12 a threat of these pathogens in high-income countries, and it is particularly important, for instance, in nosocomial infections (hospital-acquired infections). Antibiotic-resistant organisms, such as MRSA and multi-drug Resistant Tuberculosis (MDRTB), are significant threats to global health due to years of misuse of antibiotics. Then there is also the concern of emerging pandemic viral infections, recently with the ongoing pandemic SARS-CoV-2, but previously with SARS-CoV-1, MERS-CoV (Middle East respiratory syndrome coronavirus), Ebola, Zika, Influenza, and Chikungunya infections. And if, in the case of bacteria, the misuse of antibiotics is a problem for viral infections, not having antiviral drugs to treat these emerging viral infections is a major concern. There is also a significant fear revolving around foodborne and waterborne pathogens. Some of the most common foodborne pathogens include Salmonella, Shiga bacillus, Escherichia coli (E. coli) O157:H7, Bacillus cereus (B. cereus), Staphylococcus aureus (S. aureus), and Listeria monocytogenes. Hence, detecting pathogens is essential to help decrease antimicrobial resistance, the spread of pandemics, and foodborne contaminations and spreads. Some approaches can be implemented to fight this emergent threat for infectious disease control. The first one is vaccines, antibiotics, curative treatments, which already exist, and the continued search for new ones that can help eradicate when possible. Another measure of the highest importance is a rapid and differential diagnosis. Current detection methods rely on cell culture methods, enzyme immunoassay, and PCR; the test result will often take half a day to 3 days or more to be obtained. These tests are usually centralized in laboratories traditionally located in large cities (Figure 2.1 - A). Because of the long time to result, the spread of the infection can occur, creating a chain of infection in the community. With the PoN/PoC devices, we can remove the necessity of performing sample transportation to the laboratory, and the assay simplicity helps improve the turnaround time to a window frame between 15 min to 3 hours (Figure 2.1 - B). As a consequence, the possible infection chain in the community is contained. There is clearly an interest in developing point-of-need devices to perform the detection. Adopting the rapid and easy-to-use test for malaria detection shows the success of adopting this strategy. The development of quick point-of-care tests is one of our significant assets in this war against infectious diseases [26]. Micro total analysis systems (µTAS) can be the answer to achieving such tests. In the following subchapters, we review the several modules and strategies in microfluidic device 13 technology developed in the last years that allow pathogen identification in a sampleto-answer type of assay. Figure 2.1 - Schematics and comparison of pathogen detection using conventional laboratory detection methods - A - and PoC methods - B. Transportation into a central laboratory is required for conventional methods, and the laboratory assay steps are labor and time intensive. In contrast, for point-of-care devices, the analysis can be performed next to or near the patient. The point-of-care devices allow assay automation, decreasing user intervention and turnaround time. 2.2. Microfluidic devices for the detection of pathogens The detection of a pathogen can be done by searching for different biomarkers. Two of the most common biomarkers used to detect pathogens are antibodies (immunological tests) and specific nucleic acid sequences (molecular tests) [27], [28]. It is also worth mentioning that besides antibodies, aptamers have also been used to detect these pathogens [29], [30]. The binding properties of aptamers can be easily tunable since they are synthesized in the lab without resorting to animals for production, and they are more chemically stable than antibodies. There is also a new trend of using bacteriophage-based techniques since phage replication only occurs in living cells, thus identifying viable pathogens [31]–[33]. Overall, in all the microfluidic devices developed, we can find some closer to µTAS devices or simple modules of a 20 components are released. Lo et al. reported a rectangular microchannel with a planar electrode built on its bottom wall and actuated by alternating current (AC) voltages between neighboring electrodes [55]. Human whole blood was pumped through the device; the cells were lysed entirely (lysis efficiency 100%) within 7 s after the application of a 20 V. While electrical lysis methods usually present the fastest times of the lysis methods, with reasonable efficiencies, they require electrode integration and power consumption, increasing the complexity of the devices. Thermal lysis usually requires the integration of a heater and a temperature sensor on chip, increasing the complexity of the device. Since the heaters require a power source, the tendency is to use low-power heating, such as ohmic heating, to make system integration easier. Nevertheless, this lysis strategy cannot be applied if the microfluidic device material is permeable to gases (like polydimethylsiloxane (PDMS)), leading to sample evaporation and creating air bubbles inside the device. On the other hand, thermal lysis also eliminates the need for chemical buffers that can compromise downstream processing. Geissler et al. developed a lab-on-chip device that integrates thermal lysis and PCR amplification [56]. The chip is made of a thermoplastic material, allowing the sample solution to heat to 95 ºC for 5 min. The disk is rotated at high speed during this step (800 rpm) to handle condensation and air bubbles. The lysis efficiency of E. coli cells for all the concentrations tested was above 99%. Burklund et al. reported a system with a magnetic polymer substrate (Mag-Polymer microchip) that allows for highly controlled, on-chip heating of biological targets when an alternate current magnetic field (AMF) is applied [57]. After being concentrated using magnetic beads coated with a selective antibody from 1 mL to 5 µL, the bacteria were lysed by applying an AMF for 60 sec. The temperature onchip ranged from 100 ºC - 105 ºC. This strategy proved to have the efficiency to lyse S. aureus from 87% to 100% for a bacteria concentration between 105 CFU/mL and 103 CFU/mL, respectively. Liu et al. developed a single-cell lysis protocol for both gram-positive and gram-negative bacteria, combining thermal, chemical, and enzymatic lysis [58]. The device was made of PDMS, but since a combination of lysis methods was used, the temperature required was 65 ºC. In fact, the strategy used was a cycle of - 20 ºC / 65 ºC heat shock treatment to alter the fluidity of the cell membrane, creating pores due to thermal shock. Interestingly, an ice / 65 ºC heat shock slightly increased the single-cell amplification rate compared to no heat-shock treatment. A more aggressive heat shock was tested (ice/90 ºC), but no DNA was amplified. The authors postulate that over-denaturation of DNA may have occurred rather than DNA degradation. DNA degradation is associated with temperatures 21 above 100 ºC. Using -20 ºC for 2 min and 65 ºC for 2 min led to a 90% rate of singlecell amplification with 6.81 ng of the DNA detection limit. Due to the simplicity of integration in microfluidic channels, there are many reports of chemical cell lysis integrated on-chip. The reagents are commercially available and well-established, and implementing this strategy requires minimal modification of the chip design. The device has a separate inlet for the sample and the chemical lysis reagents for chemical lysis implementation. The channel length is also an important feature to allow enough residence time for lysis. Mixers should also be considered to improve the mixing of the samples and, consequently, the efficacy of the lysis. Recently, Fradique et al. developed a microfluidic device to lyse E. coli using a commercial bacterial protein extraction reagent (BPER) solution and an enzymatic solution of lysozyme [59]. The retail solution BPER showed efficiencies in the range of 100%. Kaba et al. overcame microfluidic devices' low mixing capacity and enhanced chemical lysis by applying cavitation-microstreaming [60]. The device used commercially available lysis buffer and proteinase K (PK) to lyse CHO K1 mammalian cells and perform DNA extraction in under 25 min, with a 77% efficiency. Such a method could also be considered to be applied to bacteria and viruses. Shamloo et al. described the process of chemical lysing of cells within a droplet-based microfluidic chip by modeling the assay in a computational fluid dynamics program. The lysis simulation was defined to happen inside a droplet. They conclude that increasing the volume fraction of the lysis reagent (the ratio of the volume initially filled with the lysis reagent to the total volume of the droplet) to 97% achieved a complete cell solution lysis within 0.25 s in a 2 mm long microchannel [61]. Ma et al. developed an integrated self-driven device capable of virus isolation, virus lysis, and isothermal nucleic acid amplification. The lysis was achieved by adding the lysis buffer and incubating it with Influenza A virus for 5 min. The detection was performed on-chip by reverse transcription LAMP, achieving an LoD of 87 copies of the H1N1 virus per reaction [62]. In this case, a 2-fold dilution with phosphate-buffered saline (PBS) was performed after lysis to decrease amplification inhibition. Berger et al. developed a portable pathogen diagnostic cartridge that detects E. coli from whole blood by lysing and amplifying, via LAMP, the extracted DNA. The effect of saponin, AmmoniumChloride-Potassium (ACK), SDS, and Triton X-100 lysis buffer was studied, and the saponin formulas demonstrated a faster amplification threshold time of around 25 min. Triton X and SDS had threshold times ∼30 min, while all samples were amplified by 45 min. The saponin lysis buffer showed an LoD of 10 CFU/μL after 50 min of 22 amplification in buffer, and when tested directly from whole blood, the sensitivity dropped to 50 CFU/μL [63]. To conclude, several key lysis aspects must be considered when integrating a lysis strategy on the chip. If we are developing a device to detect low concentrations of pathogens, a method with high efficiencies, such as chemical or electrical, should be preferred. If the goal is to have a portable device, the ease of integration should be increased, which is the case with chemical lysis. Electrical methods are a good choice considering the time for the result and if the application requires a fast assay. The lysis method selected should also be compatible with downstream processing, such as nucleic acid purification and amplification. 2.2.2. Sample preparation II - Nucleic acid isolation and purification After lysis, nucleic acid extraction and purification are crucial steps that can enhance the sensitivity and specificity of the detection. Standard methods usually follow liquidliquid extraction (LLE) or solid-phase extraction (SPE) techniques. Liquid-liquid extraction uses two liquids, where one of them will be preferred by the target analyte. Zhang et al. developed a microfluidic platform for nucleic acid purification of gram-positive and gram-negative bacteria (Pseudomonas aeruginosa (P. aeruginosa) and S. aureus) using a phase partitioning technique [64]. The aqueous-phase bacterial lysate was isolated in an array of microwells, and the immiscible organic phase was then introduced into a headspace channel connecting the microwell array. At the optimum flow rate of the organic phase, DNA or (ribonucleic acid) RNA from as few as five bacteria could be selectively recovered in the aqueous phase. Although LLE presented better results than the conventional Qiagen solidphase nucleic acid purification technique, the use of hazardous organic solvents, which require safe handling and disposal, can be a disadvantage when integrating this extraction method on-chip, particularly when a point-of-care and resource-limited application is required. On the other hand, SPE is faster and less complex, holding higher nucleic acid yield and purity. On common material used in SPE to extract nucleic acids is silica, either membranes or beads. In commercial kits, such as QIAmp nucleic acid purification kits from Qiagen, a spin column of silica particles is used to purify nucleic acids before nucleic acid amplification. In microfluidics, SPE methods using membranes [65] and bead-based particles (magnetic [41] and non-magnetic) are commonly employed. 23 Silica membranes can be integrated into microfluidic devices to perform nucleic acid extraction. Loo et al. reported a microfluidic disk to complete sample-to-answer detection of Mycobacterium tuberculosis (MTB) and Acinetobacter baumanii (A. baumanii). A silica membrane, made by adding silica gel on a polycarbonate track etch membrane, allowing for 90% nucleic acid extraction of the pathogens and an LoD of 102 CFU/mL A. baumanii in blood within two hours after sample loading [66]. Lee et al. reported a silica membrane-based DNA extraction-based device [67]. A 3 mm-diameter membrane was selected to process 1 mL plasma with a recovery DNA rate of 78%. Magnetic silica particles are also a frequent SPE support for microfluidic nuclei acid extraction. Li et al. reported a fully integrated cassette that uses magnetic silica beads to extract DNA from E. coli, Proteus mirabilis, S. typhimurium, and S. aureus. This device achieved LoDs in the order of 1 to 10 CFU/μL in urine samples. The complete process of DNA extraction, including the washing step, lasted 18 min. There was no difference between nucleic acid extraction performed on the microfluidic device and benchtop extraction for the four bacteria tested [68]. Similarly, Wang et al. used a continuous flow extraction device to extract DNA from large samples (10 mL) [69]. The channel comprises a magnetic DNA extractor with two concentric half-ring magnets to generate a homogeneous magnetic field. It leads to the formation of magnetic silica beads chains in a serpentine channel of the PDMS flexible chip. This device extracted more than 90% of the DNA of S. typhimurium in less than an hour with an LoD of as low as 102 CFU/mL. Paper nucleic acid extraction in microfluidic devices is also reported and commonly used for simple and low-resource applications [70]. Recently, Sullivan et al. reported a paper-based device for purifying nucleic acids from whole blood using isotachophoresis (ITP) [71]. The paper-based ITP purifies and concentrates target nucleic acids that are fractionated and filtered via an integrated plasma separation membrane, achieving 88% plasma extraction efficiency. After the extraction, nucleic acids are added directly to recombinase polymerase amplification (RPA) reactions. The device detected as low as 3 × 103 copies of nucleic acid per mL input blood, with extraction and purification taking only 30 min. Although paper extraction is more straightforward and cheaper, for low-concentration detection of nucleic acid, it is essential to select a method with a high yield of extraction efficiency. Silica has proven stable with high extraction yields, regardless of application types, such as membrane or microbead. After nucleic acid extraction and purification of possible downstream inhibitors, the sample is ready to go under an amplification technique. 24 2.2.3. Nucleic acid amplification When the analyte concentration in the sample is very low, amplification is indispensable for nucleic acid detection. Implementing nucleic acid amplification test (NAAT) on-chip is of great interest since it increases the test's sensitivity. PCR is the most common DNA amplification technique in centralized labs and has also become popular for implementation in microfluidic devices [72]–[74]. However, the several heating-cooling cycles that allow DNA denaturation, primer annealing, and strand extension require a precise temperature cycler and controller integrated within the microfluidic device. This integration leads to the increased complexity of the device. Isothermal amplification strategies have also been integrated on-chip since they do not require a complex temperature profile to amplify, thus simplifying chip design [75]– [77]. 2.2.3.1. PCR In conventional PCR, the target sequence is amplified using two oligonucleotide primers (reverse and forward primers) that hybridize in opposite strands and flank the region of interest in the target DNA and DNA polymerase (e.g., Taq polymerase). The PCR product (amplicon) is visualized by gel electrophoresis using DNA intercalating fluorescence dyes (e.g., ethidium bromide). In the case of real-time PCR (qPCR), the reaction solution contains a fluorescently labeled probe and a passive reference dye (e.g., carboxy-X-rhodamine (ROX)). The quantification of the amplicon is performed at each cycle, which allows the monitoring of the exponential amplification phase to obtain the quantity of initial template DNA. On the other hand, in conventional PCR, the detection and quantification are only performed at the end of the amplification process and require extra analysis steps. A typical qPCR consists of an initial denaturation step and amplification cycle that includes three major phases (1) denaturation, (2) annealing, and (3) extension, and can be repeated up to 40 times. The initial denaturation is performed at a high temperature, usually 95 ºC, and the solution is incubated for a few minutes to ensure that the double-stranded DNA molecules are separated into single-stranded DNA to allow amplification. In the second step of the process occurs the annealing of the primers, facilitated by the change in the temperature to approximately the melting temperature of the primer. Finally, the last step of the cycle occurs extension at 70-72 ºC, where the activity of DNA polymerase is optimal [78]. Several groups integrated PCR on-chip in microfluidics as a standalone device or in conjunction with other modules [79]. We can divide the microfluidic devices used for 25 PCR into stationary/reactor type and flow type devices [80], [81]. The amplification occurs in the static/reactor type by thermocycling to the reagent inside the chamber [64], [82]. For the flow-type device, the sample goes through several channel zones with the distinctive temperatures required in the PCR cycle, or the solution can be pumped between reactors at different temperatures [83]. Recently, Huang et al. developed a microfluidic chip-based PCR-array system to detect 21 pathogens within 1.5 h with an LoD of 103 copies/mL [84]. Hung et al. adopted a strategy of performing PCR in a solid phase coupled with a supercritical angle fluorescence (SAF) microlens array embedded in a microchip for signal detection [85]. The found limit of detection of the LoC system was 0.8 fluorophores/µm2 with a sensitivity of the solid phase (SP)- PCR as low as 1.6 copies/µL for the Salmonella Enteritidis DNA template. In recent years droplet, digital PCR has also been applied to microfluidics [86]. This amplification method results in higher sensitivity. However, for all PCR-based methods, the requirement of several temperatures that include high temperatures (95 ºC) makes the integration more challenging than for isothermal techniques. 2.2.3.2. Isothermal To overcome some of the limitations displayed by PCR, isothermal nucleic acid amplification has gained much interest, particularly for point-of-care applications. Rolling circle amplification (RCA) amplifies the nucleic acids at a temperature of 37 ºC, and it is based on the rolling circle replication mechanism present in bacteria and viruses. RCA requires a circular DNA template termed padlock (PLP), a short DNA or RNA primer/target, and a DNA polymerase to start the reaction. The amplification generates a long single-stranded molecule containing thousands of repeated copies of the circular template tethered to the original circular DNA. PLPs are linear oligonucleotides composed of two end-sequences and a linker sequence. The two end-sequences complement a target DNA, such as the DNA of a pathogen. Upon hybridization with the target, the PLP becomes circularized with a nick between the two linker sequences that a DNA ligase can seal, usually T4 DNA ligase of Thermus thermophilus (Tth) DNA ligase. DNA ligases demand a perfect hybridization to efficiently join the 5’-phosphorylated end of the DNA strand to the 3’-end of another strand. Phi29 is a replicative polymerase from Bacillus subtilis phage phi29 with a 3' - 5' proofreading exonuclease activity. An advantage of padlock probing-based rolling circle amplification (PLP-RCA) is that it can also detect RNA directly, simplifying the process by eliminating the need for reverse transcription when detecting a virus. Although this linear amplification process already provides high sensitivity, it can be 26 further enhanced by performing another variation of RCA that allows for exponential amplification, such as hyperbranched RCA, multiprimed RCA, or circle-to-circle (C2C) RCA [87]. Sato et al. reported the integration of PLP and RCA in microfluidic, used beads as solid support to capture the target DNA of Salmonella enterica (S. enterica), and performed the amplification [77]. The amplification on-chip took 3.5 h, and 88 ng of Salmonella genomic DNA (30 amol) could be successfully detected. RCA has also been combined with aptamers and CRISPR-Cas12a technology to detect pathogens [88], [89]. One of the most widely studied isothermal techniques is LAMP, performed at a constant temperature of 60 ºC to 65 ºC. LAMP requires 4 to 6 primers, complementary to six specific regions on the target gene, and a Bst polymerase enzyme with high displacement activity but lacks a 3’ to 5’ exonuclease proofreading capability. There are at least two primers present in the reaction, the inner and outer primers [90]. LAMP can also amplify RNA simply by using a reverse transcriptase without requiring additional time for the reaction. However, the method also carries several disadvantages: false positives are still an issue, the complicated design of the four specific primers, and the amplified products cannot be used for downstream experiments. In microfluidics, LAMP has been frequently reported to perform multiplex detection [91]–[93]. Meng et al. developed a LAMP-based microfluidic system that can identify S. aureus, Staphylococcus epidermidis (S. epidermidis), Staphylococcus haemolyticus (S. haemolyticus), and Staphylococcus hominis (S. hominis) within 70 min, including the hands-on time. LoDs of 20 CFU/reaction for S. aureus, S. epidermidis, S. hominis, and MRSA or 200 CFU/reaction for S. haemolyticus were achieved [94]. Compared with other well-known techniques, on-chip LAMP assay provides low sample and reagent consumption, ease-of-use, accelerated analysis, multiple bacteria on-site detection, and high reproducibility. However, the temperature required is superior to the one needed by RCA, which can still be a hurdle to implement on-chip. Recombinase polymerase amplification (RPA) [95]–[103] and isothermal crosspriming amplification (CPA) [104] are also found in microfluidic devices. RPA is widely implemented since the amplification temperature is 25 ºC - 40 ºC with a comparable LoD to PCR and can amplify the target in less than 30 min. But the main disadvantage is that it can lead to false-positive results. CPA is a novel isothermal DNA amplification technology developed by Ustar Biotechnologies Co., Ltd. It can be carried out by a strand displacement DNA polymerase at the assay temperature of 63 °C and does not require a nicking enzyme or an initial denaturation step [105]–[108]. 27 After the amplification is performed, transducing the amplicon signal is an essential step and miniaturizing the components required to do so defines the portability of the devices. In the next section, the detection of the amplification product is discussed. 2.2.4. Detection 2.2.4.1. Optical detection Fluorescence is the most common optical method for molecular sensing in microfluidic systems due to the well-established, highly sensitive, and highly selective fluorescent labeling techniques from the conventional genomic and proteomic analysis [109], [110]. Boissinot et al. described a bead-based DNA microfluidic device to capture and detect amplicons via DNA hybridization and fluorescence optical detection. The device had an LoD of 5.6 to 5.8 x 10-5 M [109]. Fluorescence is also the most common detection method commercially available for µTAS devices for pathogen detection. Colorimetry is another attractive option for optical detection in which analyte binding causes photochemical emission, either directly or with the help of an enzyme label [111]–[114]. One of the advantages of using enzymes as a biological recognition element in biosensor technology is that they are highly selective to a specific substrate or a class of substrates. The second advantage is that enzymes can produce ions, protons, heat, photons, and/or electrons during catalytic turnover, which are all measurable parameters in colorimetric detection. Surface plasmon resonance (SPR) is a common optical detection technique applied in microfluidic devices. Yuan et al. reported a precise and replicable DNA sensing platform for specific target DNA oligo detection with a detection limit down to 3.21 fM [115]. 2.2.4.2. Others An electrochemical signal read-out has been proposed for highly sensitive detection of nucleic acids by portable and miniaturized POCT devices. Integrating electrochemical sensors on a microfluidic chip provides continuous measurement and collection of electrical signals from nucleic acids. Ichzan et al. developed a DNA detection method using a combination of solid-phase RPA and electrochemical detection [99]. To obtain a high electrochemical signal-to-background ratio, electrochemical-enzymatic redox cycling employing 1,4-naphthoquinone was used. The detection limit for synthetic template DNA was measured using microfabricated indium tin oxide (ITO) electrodes was approximately 0.1 fM. Mass-based sensing devices can also be used to detect pathogens on-chip [116]. For instance, Hong et al. 28 developed a Quartz Cristal Mass (QCM) sensor with a LoD detection limit of 1.6 x 109 M [117]. Magnetic detection is also a growing detection mode on-chip [118]. Figure 2.5 – Optical e non-optical detection methods used in µTAS device. Overall, all the detection methods featured and presented in Figure 2.5 have the potential for integration into µTAS devices. While some will require more instrumentation of the device or even a stand-alone system to perform the detection, such as the case of fluorescence, others could be implemented easily without the need to resource complex instrumentation, e.g., colorimetric detection. However, the sensitivity can also be affected by choice of method. Usually, electrochemical detection methods can achieve higher sensitivities. When selecting the appropriate detection method, a balance between these characteristics should be considered. When all the assay modules are developed (sample extraction, lysis, nucleic acid extraction, amplification, and detection), a µTAS can be achieved. The following section presents a review of commercially available µTAS for pathogen detection. 2.3. State of the art µ TAS device The main goal when developing a device for pathogen detection is to integrate all the steps involved in a molecular diagnostic test in a single sample-in/answer-out device. Ideally, the user would only be required to collect a sample, insert it in the device, and push a bottom. Recent advances in microfluidic molecular diagnostics tests and electric components integration (smaller and cheaper instruments), how these instruments connect to laboratory information systems, providers, and patients, combined with the SARS-CoV-2 pandemic and the pressing need to decentralize the detection of pathogens, had led to an increase in the development, approval by public health agencies, and commercialization of µTAS devices. Most of these devices have as their primary application the detection of SARS-CoV-2, but not only. The adaptability of these devices (changing the target gene to be detected) allows 29 translating the technology, with few modifications, to other pathogens and infectious diseases and antibiotic resistance testing, for instance. In this subsection, recent devices approved in the market are presented, discussed, and compared. Products available and characteristics such as turnaround time, LoD, cost, techniques employed, and sample throughput can be found in Table 2.1. A standard key feature of these devices is that they are composed of one (or more) cartridges with all the reagents already sealed inside and adapted for the pathogen to be detected and usually a separated analyzer instrument. Each cartridge is single-use, and after the assay, it is discarded, and no µ TAS device presented here can be fully reutilized. Apart from Lucira's Covid test, PortNAT and EasyNAT from Ustar, and Visby Medical from Click Diagnostics Inc., all the devices required a separate analyzer instrument. This analyzer allows for mechanical actuation in the cartridge, controlled thermocycling, and signal readout, usually in the form of a fluorescence signal. In that sense, we cannot affirm that they are truly portable. Still, they satisfy the requisites to be considered point-of-need devices and allow for decentralization in the diagnostics workflow. As highlighted in Table 2.1, some devices require steps off-chip regarding automation. The EasyNAT, RTisochipTM-A by Capitalbio, and the Novodiag developed by Mobidiag are examples that require ex-situ sample preparation. The EasyNAT requires nucleic acid extraction using magnetic beads outside the device. Similarly, the RTisochipTM-A doesn't have a lysis on-chip, and lysis is achieved using commercial kits off-chip, requiring several steps with pipetting and handling. Novodiag device requires sample inactivation with a step of vortexing, chemical lysis, and nucleic acid extraction, all performed off-chip. Most products still rely on skilled operators and central laboratories due to their multiple operation steps with bulky analyzers. Abbot ID NOW tried to simplify the device's complexity by bypassing a sample preparation step and allowing the assay to occur directly in the swab sample. Also, Simplexa by DiaSorin, Visby Medical, and the all-in-one test kit® for COVID-19 diagnosis delivered by Lucira Health requires the collection of the sample, followed by lysis, where the reagents can be lyophilized in the sample vial or the cartridge. The user only needs to insert the sample into the device to obtain a result. Although this simplification can be an advantage to simplify the apparatus, for samples with complex matrixes, the sensitivity, specificity the ability to detect the pathogens can be compromised. Regarding fluidic flow actuation, RTisochipTM-A uses centrifugal forces, enabling efficient miniaturization, parallelization, and integration of assays. Pressure valves (Visby Medical) and rotating valves (GeneXpert®) are used in other devices. 36 System Company Targets/ Diseases Time (min) LoD Type of system Sample Preparation Lysis Amplificat ion technique Signal Readout Effective sample volume tested (µL) Throughput Cost Ref. Simplexa DiaSorin SARS-CoV-2/ Influenza/ respiratory syncytial virus/ C. difficile 60 167 copies/mL Sample-toanswer n.a. Chemical lysis “reaction mixture” Real-Time PCR Fluorescence signal 50 1 to 8 $9.56/test; $35000 for instrument [146] Visby Medical Click Diagnostics Inc SARS-CoV-2 30 500 copies/mL Sample-toanswer n.d. Heat and chemical lysis Continuousflow PCR Colorimetric (LFA) *initial sample volume 500-1000 1 $50/test [147]–[150] 37 Recent advancements in sample preparation have improved the recovery of pure NAs while reducing the time required for sample preparation. Lysis methods have also evolved to simplify the chip requirements. A move towards automation has also occurred by integrating sample preparation steps in microfluidic devices supported by the advancement and miniaturization of electronic components. COVID-19 also propelled the approval and commercialization of more µ TAS devices. But, although a certain number of integrated µ TAS products exist on the market, there is still plenty of room for improvement. Firstly, the detection cost on these products is still relatively high, taking tens to hundreds of dollars, due to the high cost of reagents and consumables. Secondly, most commercial products suffer from an insufficient capacity for quantitative signal readout, especially those relying on isothermal amplification technologies. Thirdly, the high throughput detection on such products is based on introducing multiple parallel consumables with preloaded reagents, which will increase additional consumables' cost and operation complexity. Fourthly, most products still rely on skilled operators and central laboratories due to their multiple operation steps with bulky analyzers, with some exceptions. Since sample preparations are usually performed off-chip, a fully integrated device is required with sample preparation on-chip and can achieve detection limits in the order of 1 - 10 CFU/mL. And finally, expanding the panel of pathogens these devices can detect is also urgent to fight viral infections and antibiotic-resistant bacteria. The development of a microfluidic-based system that, starting from the raw biological sample (namely a buccal swab), will identify the presence of specific pathogens, such as viruses or bacteria, by performing steps of sample preparation exploiting SPE methods (beadbased), universal chemical cell lysis for bacteria and viruses, isothermal amplification (RCA) and integrated optical detection using photosensors will help to satisfy a present need in the market that was still not fulfilled. 2.4. Pathogens of interest Bacterial pathogens of interest to human health and with clinical relevance are presented in this subsection. As mentioned, in 2017, the WHO created a priority list of antibiotic-resistant bacteria that pose the most significant risk to human health to guide research and development of new detection systems and antibiotics [2], [151]. The WHO has also included in the priority list of antibiotic-resistant bacteria to guide research, discovery, and development of new antibiotics for the ESKAPE pathogens. The ESKAPE pathogens can' escape’ the biocidal action of antibiotics and collectively represent new paradigms in pathogenesis, transmission, and resistance [152]. And recently, the ESKAPE-E was adopted to define the leading pathogens that cause 38 hospital-acquired infections [153]. The ESKAPE-E includes Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp and Escherichia coli and main characteristics of these pathogens are presented in Table 2.2. Table 2.2 – The leading pathogens causing hospital-acquired infections and relevant characteristics. Bacterial pathogen Class and Family Relevant diseases Gram stain test Nucleic acids WHO Priority Enterococcus faecium (vancomycinresistant) Bacilli, Enterococcaceae nosocomial bacteremia, infective endocarditis, and intraabdominal and urinary tract infections Grampositive Circular Chromossome 2 (HIGH) Staphylococcus aureus (methicillinresistant, vancomycinintermediate and resistant) Bacilli, Staphylococcaceae pneumonia, nosocomial bacteremia, and infections of the bone and wounds. Toxic shock syndrome Grampositive Circular Chromossome 2 (HIGH) Klebsiella pneumoniae (carbapenemresistant, ESBLproducing) Gammaproteobacteria, Enterobacteriaceae pneumonia, bloodstream infections, wound or surgical site infections, and meningitis Gramnegative Circular Chromossome 1 (CRITICAL) Acinetobacter baumannii, carbapenemresistant Gammaproteobacteria, Moraxellaceae pneumonia, bloodstream infections, brain infections, urinary tract and wound infections Gramnegative Circular Chromossome and circular plasmid 1 (CRITICAL) Pseudomonas aeruginosa (carbapenemresistant) Gammaproteobacteria, Pseudomonadaceae urinary tract infections, respiratory system infections, dermatitis, soft tissue infections, bacteremia, bone and joint infections, gastrointestinal infections Gramnegative Circular Chromossome 1 (CRITICAL) Enterobacter spp, (carbapenemresistant, ESBLproducing) Gammaproteobacteria, Enterobacteriaceae cerebral abscess, pneumonia, meningitis, septicemia, and wound, urinary tract and abdominal cavity/intestinal infections Gramnegative Circular Chromossome 1 (CRITICAL) Escherichia coli (carbapenemresistant, ESBLproducing) Gammaproteobacteria, Enterobacteriaceae traveler’s diarrhea and dysentery, pneumonia, bacteremia, and abdominal infections such as spontaneous bacterial peritonitis Gramnegative Circular Chromossome 1 (CRITICAL) In the context of developing a new device for early detection, we can direct the device to be as universal as possible, enabling the detection of the pathogens of interest 39 independently of the differences between them. In this sense, two essential characteristics should be considered when developing a biosensor to detect pathogens, especially when using a nucleic acid amplification test-based biosensor. The first aspect should consider the easiness of sample preparation and nucleic acid extraction, which is affected by the cell wall. The second aspect is related to the efficient capture of the nucleic acid of interest, which is usually more complex when the pathogen's genome is double-stranded. Regarding the first aspect, the gram stain provides an important classification that correlates with the bacteria's cell wall structure. The ESKAPE-E list includes both pathogens classified as gram-positive and pathogens classified as gram-negative. The gram-positive bacteria have a ticker cell wall (20-80 nm) [154]. Briefly, they are composed of two main components the inner membrane and peptidoglycan. They can also present an optional S-layer external to the peptidoglycan layer (Figure 2.6). The inner membrane surrounds and encloses the cell cytoplasm and acts as a semipermeable barrier. Then the peptidoglycan layer is a single, highly interconnected macromolecule outside the inner membrane that gives rigidity to the cell. Peptidoglycan is composed of repeating units of N-acetylglucosamine-Nacetylmuramic acid disaccharides, also called glycan strands. The degree of chains crosslinked varies from the organism and can go up to almost 90% of available peptide stems in s. aureus to 16% in Bacillus anthracis. It also presents amino acids attached to the glycan strand, and two or more side chains can be crosslinked to one another by a covalent bond. The peptidoglycan layer also contains glycopolymers such as teichoic acids. Teichoic acid is a negative molecule that can work as a protective cloak against antimicrobial compounds. For instance, for S. aureus, teichoic acids protect the bacteria against the antibacterial effects of fatty acids [155]. In contrast, Gram-negative bacteria have a relatively thin (<10 nm) layer of a cell wall made of peptidoglycan. Still, they harbor an additional lipid-protein bilayer encompassing the outer membrane with several pores and appendices. The outer membrane is constituted of different phospholipids in the inner face and lipopolysaccharide in the outer face. Proteins are a significant component of the outer membrane and contribute to the integrity and stability of the gram-negative cell wall. Gram-negative bacteria can also present an S-layer after the outer membrane (Figure 2.6). An array of paracrystalline proteins constitutes the S-layer and usually contributes to making the bacterial surface more hydrophobic [155]. Depending on the lysis method, gram-positive bacteria may be more challenging to lyse than gramnegative or the opposite. For instance, ethylenediamine tetraacetic acid (EDTA) or 40 CaCl2 are outer membrane disrupting agents that promotes electrostatic repulsion, which destabilizes the cell wall structure. Figure 2.6 - Representative illustration of the cell wall components of gram-negative, gram-positive bacteria, and non-enveloped and enveloped viruses. Besides bacterial detection, detecting viruses responsible for infectious diseases such as Dengue, Zika, Chikungunya, Ebola, human immunodeficiency virus (HIV), or Influenza through a PoC device would also have a positive impact on the control of emerging infectious diseases worldwide [156]. In this case, detecting enveloped and non-enveloped viruses may affect the lysis efficiency. In the case of non-enveloped viruses, only present a capsid that protects the virus's genetic material (Figure 2.6). The capsid is formed by proteins called capsomere. However, in enveloped viruses, besides the capsid, an outer lipid membrane composes the viral envelope (Figure 2.6). Regarding nucleic acid extraction and capture, we can have pathogens that present double-stranded DNA genomes (both for bacteria and some viruses). In this case, there is a need to extract and, previous to amplification, denature the double-strand DNA to allow the star of the amplification. For RNA-based genomes, a reverse transcription step is usually required to synthetise a cDNA that will work as a template for amplification. This increases the complexity of implementing the detection in a PoC device. 2.5. Fundamentals of microfluidics This subsection summarizes the fundamentals of microfluidics technology, the main physical characteristics of these systems, and the advantages and disadvantages of using these systems for analytical assays. Microfluidic is the technology of fluid manipulation in channels, designed with at least one of the channel’s dimensions within tens of micrometers [19]. While at the macroscale, inertial forces such as gravity dominates the fluid mechanics, at the microscale it is the viscosity that most influence 41 the fluid regime. Microfluidics physics is governed by different forces than the ones that govern fluids at the macroscale mainly because as the systems go down in size, we see an increased dominance of surface effect over volumetric effects. This is a key feature of microfluidics devices with a high surface-to-volume ratio typically in the order of 106 m. Thus, the fluid mechanics at this scale is controlled by viscous forces, surface tension, and electrostatic/electrodynamics forces [20]. The characterization of the fluid flow regime in microsystems uses the Reynolds number ( " ) (Equation 2.1). This adimensional parameter is a ratio between kinetic energy ( #$! ) and viscous energy ( %$ & ⁄). It uses the hydraulic diameter of the channel ( ( ), the average velocity of the liquid ( $ ), the mass density of the liquid ( # ), and the viscosity of the fluid ( % ), giving: () =+!"# $ Equation 2.1 For non-circular systems, such as rectangular cross-section microchannels, the hydraulic diameter is given by Equation 2.2: , =+%& ' Equation 2.2 Where ) = +, , is the area of a microchannel cross-section with a width + and height ,, and . is the total perimeter of the walls that are in contact with the liquid [157]. The Reynolds number can be used to evaluate the type of flow regime inside a microfluidic channel. For "/ >4000 , it is expected a turbulent flow, 4000 >"/ > 2300 , we are in the range of transitional flow, and for "/ <2300 , we are in the laminar flow regime. Usually, in microchannels, "/ ≈ 1 , meaning that we are inside the laminar flow regime with predictable fluid dynamics. Due to this, there is no convective mixing inside the microchannels, and the transport of molecules is done solely by diffusion. It is essential to understand if the species under analysis can diffuse across the channel to get to a sensor area, for instance, or if it will be drawn downstream by the fluid flow. The Péclet number ( ./ ), which is the ratio between the advective and diffusive transport, in Equation 2.3 asses the molecular diffusion: 42 -) =+./ " Equation 2.3 Where 8 and 9 are the characteristics length and characteristic velocity, respectively, and & is the diffusion coefficient of the species under analysis. When the ./ >1000 the diffusion of the species can be neglected, and the advection is the main process. If the ./ <10 , then the diffusion process dominates, and the advection can be neglected [158]. Microfluidic devices tend to present high Peclet and low Reynolds numbers with simple and steady boundary conditions. As such, the laminar flow does not shorten the diffusion length scales [159]. The laminar flow regime and the short diffusion lengths make mixing solutions challenging, particularly along the microfluidic channel's width. This can be verified for each channel by calculating the characteristic diffusion time along the width (τDW). Starting with the Stokes-Einstein, Equation 2.4, we can estimate the & or the diffusion coefficient: " =+ 0(' 1234 Equation 2.4 Where :" is Boltzmann’s constant, T is the absolute temperature, % is the dynamic viscosity, and r is the particle radius. For a bacterial cell, such as E. coli, the diffusion can be modeled by a 2 µm spherical particle. The diffusion time (τD) is defined as the time that it takes one molecule to travel the distance 8 (length of the mixing path) by diffusion as expressed in Equation 2.5: #)=+.* " Equation 2.5 The movement of a single molecule in a fluid is characterized by the EinsteinSmoluchowski relation of Equation 2.6: 5 =+ √ 7"8 Equation 2.6 Where ; is the root-mean-square distance transverse by a particle during the time interval < for a given diffusion coefficient = . For instance, in the case of adjacent 43 streams, one can assume that along the width, the diffusion distance corresponds to half of the channel width. With Equation 2.6 it is possible to estimate the amount of time required for each particle of interest to achieve complete diffusion in the microchannel [160]. When the mixing of the solutions is dependent on molecular diffusion, the time for diffusion is proportional to the square of the diffusion distance that limits the mixing. The time for diffusion can be incompatible high with the assay application. In this case, mixing enhancement of the solutions is required, and strategies for mixing should be considered. In some devices, there is an advantage in designing pillars, using hydrogel mesh or even microporous beads inside the channel to shorten the molecules' diffusion length, leading to a faster assay reaction [161]. Moreover, another important parameter is the thermal Péclet number ( ./$% ) being, in this case, a ratio between heat transport due to advective transport (heat transfer by bulk fluid flow) and diffusive transport (heat conduction), given by Equation 2.7: -)+, =+,/9:- ; Equation 2.7 Using the diameter of the channel ( ( ), the average velocity of the liquid ( $ ), mass density of the liquid ( # ), the specific heat capacity ( >&) and the thermal conductivity (A) . In the case of chemical reactions that require heat, and we apply heat onto the channel walls to increase the temperature inside the device, this ratio allows us to understand how hastily the heat is displaced by the moving fluid relative to how fast it is transferred from the channel wall to the center of the fluid flow where the reaction can be occurring [23]. When ./$% ≪ 1, the heat transfer occurs due to conduction (equivalent to diffusion for mass transfer), while for ./$% ≫ 1 , heat transfer occurs due to convection. Usually, microfluidic devices display a low ./$% and the thermal convection is neglected. Since we have a high surface-to-volume ratio in microfluidic channels, the heat/cooling time is usually considered faster than at the macroscale. Nevertheless, the heat transfer could be enhanced by creating a turbulent flow inside the microfluidic channel which can be required when performing assays that require fast temperature cycles, such as PCR for DNA amplification. On the other hand, there is also a question of the maximum sensitivity a sensor can have. This is limited by the amount of analyte that is present in a sample volume ( D ) and can be expressed as Equation 2.8: 44 < =+ = 3.#&$/ Equation 2.8 Using as a parameter the sensor efficiency ( 0 < %'< 1 ) Avogadro’s number ( E( ) and the concentration of analyte ( )) ). Usually, when working with human clinical samples, analyte concentration varies between 1014 to 1021 copies/mL. For immunoassays is between 108 to 1018 copies/mL, and for genomic assays to detect bacteria or viruses the range is between 102 to 107 copies/mL [21]. Depending on the analyte concentration, we can decrease the sample volume from femtoliter (with higher analyte concentration) to milliliters range (for lower analyte concentration). Therefore, when working with nucleic acid detection in microfluidics, steps of preconcentration and amplification are required. With all things considered microfluidic devices offer several advantages for the detection of bacterial pathogens. These systems allow the reduction of sample volume and reagents consumption (10–6 to 10–18 liters), increase speed of reactions, increase sensitivity, a decrease of cost, and simplification of the detection process by developing autonomous systems that allow for pathogen isolation, lysis and nucleic acid extraction, nucleic acid amplification and amplicon detection in one single chip [22][23]. 2.6. Assay requirements to develop a fit for purpose biosensor This section briefly summarizes the fundamentals of standard procedures in analytical chemistry to evaluate if an assay is fit-for-purpose for a specific DNA detection threshold. This is particularly important to discuss the trade-offs between performance and other considerations, such as simplicity of use and cost when developing a pointof-need analytical device. The microfluidic system should be studied for performance. The relevant parameters to consider when looking at biosensor performance are sensitivity, the limit of detection (LoD), specificity, reproducibility, and dynamic range [162]. To understand the adequacy of the biosensor and to optimize the detection capability of the biosensor for a specific application, cut-off values should be known. The sample matrix could also be an important parameter affecting the biosensor detection capacity. Figure 2.7 summarizes in a schematic fashion the parameters for biosensor characterization. 45 Figure 2.7 - Biosensor characterization. A - Schematics of the main flow on the Biosensor analysis and signal output. B - Representation of the signal output for biomolecular recognition. C - Definition of the limit of blank (LoB), the limit of detection (LoD), and the limit of quantification (LoQ). Specificity is distinguishing a particular substance from others(selective) and capturing a particular target (specific). Therefore, the biosensing system should be configured to detect a particular target by binding with the specific biorecognition element through strong and selective affinity. To evaluate the specificity performance, the measurement of the non-complementary samples should be compared to the specific target. A functional biosensor should be able to distinguish these different samples producing significantly different response signals. Ideally, the non-specific target’s resulting signal level is similar to the zero-concentration measurement of the biosensor. One solution to improve the assay's specificity is developing a blocking strategy for the non-specific signal. Sensitivity can be defined as the slope of the analytical calibration function in the linear region, which can be seen as the response per unit concentration. We can assume that the higher the signal reaction for a given target concentration, the higher the sensitivity [163]. There is also the dynamic range which helps characterize the interval between the minimum and maximum concentration that is detected by the system. In other words, it can be seen as the interval between the detection limit and the saturation level of the signal [162], [164]. To be noted that in the lower concentrations and higher concentrations, the signal is not linearly proportional to the target concentration. However, one can define the dynamic linear range that only considers the linear region of a fitting curve. The dynamic linear range and the sensitivity influence each other. When sensitivity is high, the saturation level will be 52 with the aluminum surface facing down, to prevent a loss in resolution due to scattering effects. The stack is exposed to a 400 W UV light with an energy per unit area of 178 mJ/cm2, baked for 5 min at 95 °C, and cooled down to room temperature (rt) for 2 min. The development of the non-exposed photoresist is achieved by immersion of the SU-8 in a propylene glycol monomethyl ether acetate (PGMEA) solution for 2 min with manual orbital agitation. After the development, the substrate is rinsed with IPA and dried with compressed air. The second layer with 100 μm height is defined by spin-coating a SU-8 50 film on top of the previous layer at 10 s at 500 rpm with an acceleration of 100 rpm/s, followed by 30 s at 1000 rpm with an acceleration of 300 rpm/s. A pre-exposure bake process is then performed, which involves baking at 65 °C for 10 min, followed by a gradual ramping-up of the temperature to 95 °C, where it is baked for 30 min and then allowed to cool for 1 min. Then, the second hard mask for the 100 μm features is manually aligned to the previous layer using a stereomicroscope and placed on the SU-8, again with the aluminum surface facing down and exposed again to the UV light with an energy per unit area of 416 mJ/cm2. The second photoresist layer is developed after a postexposure bake at 65 °C for 1 min, followed by 10 min at 95 °C and 2 min of cooling down. The mold is developed in PGMEA for 10 min with manual orbital agitation (see Table 3.2), rinsed with IPA, and dried. Finally, the mold is hard baked for 15 min at 150 °C and left to slowly cool down on top of the hot plate until the temperature drops below 50 °C. 53 Table 3.1 - Materials and equipment required for device fabrication and operation. Reagents/Materials Equipment /Facilities Substrate Cleaning - Glass / Silicon substrate - Alconox solution, Alconox Inc. (White Plains, NY/USA) - Acetone (99.6%), LabChem Inc. (Zelienople, PA/USA) - Isopropyl Alcohol (IPA, 99.9%), LabChem Inc. (Zelienople, PA/USA) - DI water - Petri dish - Tweezers - Crystallizing dish - Kerry Ultrasonic Cleaning Bath, Guyson (Skipton, North Yorkshire, UK) - Automatic Dicing Saw DAD-321, Disco Corporation, (Tokyo, JP) Hard Mask Fabrication Aluminum Deposition - Clean glass substrate, Corning Inc. (Corning, NY/USA) - Class 10/ 100 clean room - Nordiko 7000 magnetron sputtering system, Nordiko Technical Services Ltd (Havant, Hampshire, UK) Lithography - Photoresist PFR 7790G, JSR (Sunnyvale, CA/USA) - Photoresist developer TMA238WA, JSR (Sunnyvale, CA/USA) - Silicon wafer (150 mm diameter), University Wafer (South Boston, MA/USA) - Adhesive tape - TechniEtch Al80 Aluminum etchant, Microchemicals (Ulm, DE) - Crystallizing dish - Class 10/100 clean room - SVG Resist coater and developer track, Silicon Valley Group Inc. (San Jose, CA/USA) - Heidelberg DWLii direct write laser lithograph, Heidelberg Instruments (Heidelberg, DE) Mold Fabrication - Clean silicon substrate, University Wafer (South Boston, MA/USA) - SU-8 50 photoresist, Microchem Corp. (Newton, MA/USA) - SU-8 2015 photoresist, Microchem Corp. (Newton, MA/USA) - Propylene glycol monomethyl ether acetate (PGMEA, 99.5%), Sigma-Aldrich (St. Louis, MO/USA) - Isopropyl Alcohol (IPA, 99.9%), LabChem Inc. (Zelienople, PA/USA) - Styrofoam box - 2 crystallizing dishes - Vertical laminar airflow cabinet, FASTERBSC-EN (Cornaredo, IT) UVO Cleaner 1444AX-220, Jelight Company, Inc. (Irvine, CA/USA) - Spin coater, Laurell Technologies Corp. (North Wales, PA/USA) - Digital hotplate, Stuart (Staffordshire, UK) - UV Light (254 nm, 400 W), UV Light Technology Limited (Birmingham, UK) 54 3.1.4. PDMS mold-replication and sealing The SU-8 mold is first placed at the bottom of a Petri dish, with the patterned surface facing up, and taped on each corner of its silicon substrate. To prepare the PDMS elastomer, a 10:1 weight ratio of PDMS to curing agent is mixed, degassed for 30 min, and poured into the Petri dish containing the mold up to 0.5-1 cm height. The Petri dish is then left to cure at 70 °C for 90 min. The cured PDMS is then cut using a scalpel and peeled off from the mold using tweezers. Access holes are punched with blunt 20 and 18 Gauge needles for the outlets and inlets, respectively (see Table 3.2). A PDMS slab (500 μm thick) is prepared by spin coating the PDMS mixture on top of a silicon wafer at 250 rpm for 25 s with an acceleration of 100 rpm/s. This membrane is baked as described above and then cut into pieces at least the size of, or a little larger than, the PDMS piece with patterned structures. The PDMS structures are sealed against the PDMS slabs by first oxidizing both sides using an oxygen plasma - Tweezers - Aluminum foil - Stereo microscope, AmScope (Irvine, CA/USA) PDMS Casting and Sealing - Plastic cup - Sylgard 184 poly(dimethyl)siloxane, Dow Corning (Midland, MI/USA) - Spatula - Petri dish - Scalpel - Blunt syringes tips (18 and 20 Gauge), Instech Laboratories, Inc. (Plymouth Meeting, PA/USA) - Analytical Balance, Scientech (Bradford, MA/USA) - Vacuum desiccator, Bel-Art Products (South Wayne, NJ/USA) - Oven loading model 100-800 (70°C), Memmert (Schwabach, DE) - Expanded oxygen plasma cleaner PDC-002CE (200 W), Harrick Plasma (Ithaca, NY/USA) Microfluidic Handling - Insulin syringe 1 mL U-100 Luer-Lock, Codan (Lensahn, DE) - Luer stub adapter (20 Ga), Instech Laboratories, Inc. (Plymouth Meeting, PA/USA) - Polyethylene tubing (BTPE-90), Instech Laboratories, Inc. (Plymouth Meeting, PA/USA) - Tubing couplers (SC20/15), Instech Laboratories, Inc. (Plymouth Meeting, PA/USA) - Syringe filter (0.2 µm), Whatman GE healthcare Life Sciences (Piscataway, NJ/USA) - Syringe pump NE1002X, New Era Pump Systems, Inc. (Farmingdale, NY/USA) - Inverted Fluorescence Microscope CKX41, Olympus (Shinjuku, Tokyo, JP) - CCD color camera XC30, Olympus (Shinjuku, Tokyo, JP) - Leica DMLM Microscope, Leica Microsystems (Wetzlar, DE) - Digital color camera DFC300FX, Leica Microsystems (Wetzlar, DE) 55 cleaner at the medium power setting (11 W applied to the radiofrequency coil) for 60 s. The membrane is placed in contact with the PDMS structure within a few min of the plasma treatment. After the sealing step, the PDMS becomes relatively hydrophilic for a few hours due to the plasma treatment. To allow hydrophobic recovery and stabilization by diffusion of the unreacted siloxane oligomers to the surface, the PDMS structures are stored for at least 24 h before being used. Table 3.2 - Timeline of the sequence of steps required to fabricate the microfluidic device. Step Time Frame Designing the mask using AutoCAD software As needed Cleaning of glass and silicon substrates 30 min Hard mask fabrication (two hard masks) 3 h 10 min (Total time) Aluminum deposition 15 min Photoresist spin-coating 5 min DWL exposure 2 x 60 min Development of exposed photoresist 5 min Aluminum etching 10 min Photoresist removal 5 min Mold fabrication Approximately 2 h 15 min (Total time) Photoresist spin-coating 1 44 s Pre-exposure bake 4 min Mask 1 alignment As needed UV exposure (5,94 mW/cm2) 30 s Post-exposure bake 7 min Photoresist development 2 min Hard bake 15 min Photoresist spin-coating 2 40 s Pre-exposure bake 40 min Mask 2 alignment (manual) As needed UV exposure (5,94 mW/cm2) 1 min 10 s Postexposure bake 11 min Photoresist development 10 min Final hard bake 15 min PDMS casting and sealing 3 h 30 min (Total time) Preparation of PDMS 10 min 56 Degassing PDMS mixture 40 min Baking PDMS 90 min Peeling off PDMS and punching holes 60 min PDMS sealing to glass/PDMS 10 min Device stabilization 24 hours (recommended) 3.2. Fabrication of a-Si:H p-i-n photodiodes Unless stated otherwise, the a-Si:H p-i-n photodiodes were fabricated following the procedure next described. The p-i-n a-Si:H photodiodes with (Figure 3. 2 – A – I) and without (Figure 3. 2 – A – II) an amorphous silicon carbide (a-SiC:H) absorption filter were used for fluorescence and chemiluminescence assays, respectively. In summary, the fabrication of the photodiodes started with the deposition of a bottom Al contact layer (200nm) by DC magnetron sputtering and patterned using DWL. Then a sequence of thin films was deposited, first an n-type a-Si:H (100 nm), then an intrinsic a-Si:H (500 nm), and finally a p-type a-Si:H (100 nm). This deposition was done on top of a bottom Al contact by radio frequency plasma enhanced chemical vapor deposition (rf-PECVD) by decomposition of silane for the intrinsic layer and a gas mixture of silane and phosphine or diborane for nor p-type doping, respectively. Using direct-write optical lithography (Heidelberg Instruments DWLii), mesa junctions (200 × 200 μm) were patterned and defined by reactive ion etching (RIE). After this, a passivation layer of SiNx (100 nm) was deposited by rf-PECVD and patterned through a lift-off process to open a via through which electric contact could be made between the p-type a-Si:H layer and the indium tin oxide (ITO) transparent conductive top contact. The 50 nm ITO layer top contact was deposited by DC magnetron sputtering and circumscribed by lift-off. After depositing and patterning Al lines to address the top contact, the device was again passivated with 200 nm SiNx. A 1.8 μm-thick a-SiC:H filter was deposited by rf-PECVD and patterned by lift-off for the fluorescence assays. The a-SiC:H layer acts as an absorption filter to block the excitation light (λex = 405 nm) while being almost transparent to the emission light of the Atto 430 LS fluorophore (λem = 540 nm). The photodiode chips were then diced and wire-bonded to printed circuit boards (PCBs). Data regarding the performance and spectral properties of the filter is depicted in detail and can be found elsewhere [22]. 57 Figure 3. 2 - Schematics of the thin-film a-Si:H photodiodes. A – I thin-film a-Si:H photodiode with integrated a-SiC:H filter used for fluorescence assays. A – II a-SiC:H filter photodiode used for chemiluminescence and colorimetric assays. This device were fabricated at INESC MN by former Ph.D. student Denis Santos. 3.3. Microfluidic structure general handling Commercially available beads are typically provided as a slurry in a storage buffer (ethanol 20%), so the first step is to homogenize the bead stock using a pipette, ensuring thorough mixing. Then, a specific volume of stock solution is added to a polyethylene glycol (PEG) 8000 30% (w/w) solution to obtain a final solution with 12% bead volume. A viscous solution allows the beads to remain suspended and homogeneously dispersed without significant settling, thus avoiding clogging problems when flowing the beads inside the micro-columns (see Table 3.2). The flow is driven by applying negative pressure at the outlet via a syringe pre-filled with water up to about half the total capacity, adapted to a syringe pump, and connected to the microfluidic structure via capillary tubing and a metal coupler. It is essential that the syringe and capillary tubing always remain free of air gaps that may hinder the rapid decrease in pressure; therefore, the syringe should be appropriately purged until the liquid reaches the tip of the metal coupler before inserting the coupler about 2/3 of the way (see Table 3.2) into the 20 Gauge access holes. The pipette tip (2-200 µL) containing the bead suspension should be inserted roughly halfway (see Table 3.2) into the inlet access holes punched using the 18 Gauge blunt syringe, and the syringe pump subsequently turned on with the appropriate flow rate. Within approximately 40 s, the beads accumulate at the interface region of the microchannels and fill the entire field of view of the microscope. After the first liquid flowing step, it is critical that the metal adapter connected to the syringe pump be removed before removing the pipette tip. Otherwise, any accumulated negative pressure can quickly trap air bubbles at the interface between the liquid column in the inlet hole and the subsequent solution. Subsequently, the PEG solution is washed from the micro-columns using an appropriate buffer solution (e.g., PBS). 58 3.4. Optical signal transduction 3.4.1. Microscopy detection Fluorescence or chemiluminescence measurements are performed using an inverted fluorescence microscope (Olympus CKX41) coupled to a charged couple device (CCD) color camera (Olympus XC30) and 50 W short arc mercury lamp, or a fluorescence microscope (Leica DMLM) coupled to a digital camera (DFC300FX) and a 100 W short arc mercury lamp. For each microscope, three different filters were available. Olympus microscope presented a U filter (lex Band Pass (BP): 360 – 370 nm; lem Long Pass (LP): 420 nm), B filter (lex BP: 460 – 490 nm; lem LP: 520 nm), and G filter (lex BP: 480 – 550 nm; lem LP: 590 nm). Leica microscope presented a D filter (lex BP: 355 – 425 nm; lem LP: 470 nm), I3 filter (lex BP: 450 – 490 nm; lem LP: 515 nm), and TX2 filter (lex BP: 560/40 nm; lem BP: 645/75 nm). Details of the acquisition of images are also presented throughout the results chapters, as they are specific to each assay. Images are analyzed using ImageJ software, and, in each case, the average grayscale quantification is obtained by considering the entire endsection of the micro-columns. 3.4.2. Photodiodes detection To measure the optical signal using the photodiodes, an optical table was used to set the setup. The photodiodes were bonded to an oriented circuit board (PCB), and the photodiode pads were wire bonded to the PCB pads. The photocurrent (measured at 0V) was recorded using a picoammeter (Model 237; Kiethley Instruments, Inc.) connected to the photodiode PCB via coaxial and triaxial connections was used. This connection allowed noise mitigation. The measurements from the picoammeter are acquired using a general interface bus (GPIB) by a computer graphical user interface (GUI) programmed in Python and PyQT4, which is used to read, process, store and plot the raw data. Former Ph.D. student Denis Santos developed the interface. The microfluidic devices, particularly the areas with the packed beads, were manually aligned directly on top of the photodiode. A laser with the appropriate excitation wavelength was shone on top of the microfluidic device with the packed beads for fluorescence measurements. Details of the acquisition of images are also presented throughout the results chapters, as they are specific to each assay. 59 Chapter 4 Bead-based microfluidic platform for DNA hybridization 4. Development of a rapid bead-based microfluidic platform for DNA hybridization using singleand multi-mode interactions for probe immobilization This chapter reports the development of rapid and simple bead-based microfluidic platform to detect a specific short DNA strand (22-mer DNA sequence) via hybridization using singleand multi-mode interactions for probe immobilization. The microfluidic device uses commercial nanoporous chromatography beads as solid support for probe DNA immobilization, using either a single-mode electrostatic interaction or multi-mode interactions (electrostatic and hydrophobic). Using a mass balance approach, a probe density of 2.4 × 1013 ± 18 (± RSD%) molecules/cm2 was quantified in optimized conditions, which was found to provide hybridization efficiencies above 95% and a hybridization dissociation constant below 1 nM. Comparing targets with different optical labels, namely Atto 430LS, quantum dots, and horseradish peroxidase, the lowest limit of detection of 9.5 ±1.1 pM was achieved with an assay time of 10 min using a quantum dot label coupled with a multi-mode immobilization. These results highlight the system's potential to be applied as a simple and highly sensitive DNA hybridization platform, achieving low pM sensitivities without needing a DNA amplification procedure. The contents of this chapter are summarized and reproduced from one original research article [171], based on experimental results obtained during my MSc project [173] and the beginning of the Ph.D. project. I fabricated the microfluidic devices. Co-conceived, designed, performed, and analyzed the experimental results. I was the main writer of the publication. 4.1. Introduction Integrated DNA hybridization-based miniaturized biosensors are currently of great interest to allow decentralized DNA testing that provides high specificity and sensitivity while being faster, simpler, and cheaper than conventional lab-based methods typically comprising PCR, performed in real-time or followed by gel- 60 electrophoresis. Furthermore, these biosensors have been used in a wide range of applications not only in health monitoring and diagnostics [174]–[176] but as well in food analysis [177], [178], environmental monitoring [179] and bioterrorism [180]. Despite the potential of those biosensors, only a few have reached the market [181], [182]. One of the reasons for this is that the direct detection of DNA without preamplification is very challenging since target concentrations in a real sample are typically in the fM to pM range [183]. Thus, the development of portable DNA biosensors providing low LoD in a simple and fast manner without needing a preamplification step is in great demand for integration in LoC devices. Several recent reports in the literature have described the coupling of microbeads with microfluidic systems to improve further the sensitivity of the detection assays [184]–[189]. Much of this research has focused on using magnetic beads since they are biocompatible, can be easily functionalized, have a large surface area, and can be controlled by a magnetic field. For example, a microfluidic device for the separation of target DNA from a complex mixture was developed by Wang et al. using streptavidin-coated magnetic beads to immobilize probe DNA, used to capture the target DNA within 15 min [184]. In another work, a microfluidic system with integrated loop-mediated isothermal amplification was developed, incorporating specific probe-conjugated magnetic beads to recognize target DNA from MRSA in a lysate solution [185]. The system presented a LoD ~6 × 10-18 M after a total analysis time of 60 min. More recently, a microfluidic system to detect RNA from the Ebola virus was developed using 4-formyl benzamide functionalized magnetic beads with covalently immobilized probe DNA [186]. This system achieved a detection limit below the value required for pre-symptomatic detection of the infection (0.04 plaque-forming units (PFU)/mL) for clinical samples and a detection time of roughly 60 min. Other groups have focused on the use of non-magnetic beads using physical entrapment to confine the beads to specific regions, such as filters [190], chambers [191], membranes [192] or two-level microchannels [193]. Many reported bead-based microfluidic systems for DNA detection use biotin-streptavidin interactions for probe immobilization. For instance, Kim et al. successfully used SuperAvidin-coated beads to immobilize probe DNA (biotin-streptavidin bond), allowing an enhanced detection of target DNA within 14 min with an LoD of 10-10 M [187]. A different study by Sochol et al. presented a microfluidic system that used streptavidin-coated polystyrene beads to immobilize a molecular beacon probe DNA [188]. This system could parallelly detect three different target DNA (3 × 10-5 M). In another work, polystyrene and Sepharose beads were used for primer immobilization and rolling circle amplification [189]. This microfluidic system detected viral pathogens within 61 15 min with an LoD of 10-13 M. Other groups have focused on using nanoporous beads, offering increased surface-to-volume ratios and potentially further improving the sensitivity of the system [194]. These have been employed in microfluidics to study protein breakthrough curve analysis [195], [196], toxin screening [193], and detection of protein cancer biomarkers [197]. However, few have focused on using nanoporous beads for DNA detection, and in all these cases, an amplification step was included in the assay [189], [198]. Furthermore, electrostatic, or multi-mode commercial chromatography beads explored for probe DNA immobilization in microfluidics systems potentially allow a rapid, cost-effective, and single-step analysis using a highly stable solid support, which can be highly advantageous compared to other strategies requiring expensive biomolecules (such as thiolmodified probes) and lengthy covalent functionalization procedures, long incubations times and several washing steps [186], [197]. This chapter reports the development of a bead-based microfluidic optical biosensor for detecting DNA using commercial electrostatic or multi-modal chromatography beads, tested without pre-amplification, to specifically detect a model 22-mer DNA strand. Two different types of microbeads were used as an immobilization platform providing (i) single-mode (electrostatic) or (ii) multi-mode (electrostatic, hydrogen bonding, and hydrophobic) interactions with probe DNA. To improve the non-specific signal, a blocking step using sodium polyacrylate (PA) molecules was employed and optimized. Moreover, different optical detection modes were tested and compared using (i) an organic fluorophore, (ii) a quantum-dots, and (iii) an enzyme label. Finally, the probe DNA density, hybridization efficiency, and equilibrium dissociation constant (Kd) were quantified to characterize this system. 4.2. Materials and experimental methods 4.2.1. DNA oligonucleotides and beads Synthetic oligonucleotide sequences (Table 4.1) were purchased from StabVida Genomics Lab. The stock solution was prepared by resuspension of the lyophilized DNA in 1x TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0.) (Sigma-Aldrich) to a final concentration of 100 μM and stored at -20ºC. The streptavidin (Sigma-Aldrich) stock solution was diluted to a 1 mg/mL working solution in PBS and held at -20°C. Q Sepharose Fast Flow (QS) beads and Capto Adhere (CA) beads were purchased from GE Healthcare as a slurry in 20% ethanol (Figure 4.1). 68 Figure 4.4 - Schematics of the mass balance method used to calculate the mass of probe DNA immobilized on the beads and captured target DNA. (A) Measurement of calibration curves. A fixed volume (Vi) of a known concentration (Ci) of the probe or target DNA flowed through a channel with packed beads at a flow rate (Qqt) of 3 µL/min for probe DNA and 1 µL/min for target DNA. This calibration curve is then used to measure the :(<=>)!"# by correlating fluorescence intensity with concentration. (B) Measurement of DNA non-specifically bound to the PDMS walls ( :(<=>)$%&' ). In this case, a fixed volume (Vi) of known concentrations (Ci) of the probe or target DNA flowed through a bare channel at an experimental flow rate (Qex) of 5 µL/min for probe DNA and 7 µL/min for target DNA. (C) Determination of the total mass of DNA ( :(<=>)#()*+ ) immobilized on the beads and adsorbed to the PDMS walls. A fixed volume (Vi) of a known concentration (Ci) of the probe or target DNA flowed through a channel with packed beads at an experimental flow rate (Qex) of 5 µL /min. The solution was collected at the outlet and quantified according to the schematics in Figure 4.2 - A. Figure 4.5 - Quantification of the target DNA non-specifically bound to the PDMS walls for (A) cDNA and (B) ncDNA. The error bars in all plots correspond to the standard deviation of three individual measurements. For both cDNA and ncDNA, the amount of DNA non-specifically bound to the PDMS walls was considered non-significant (ns) (p-value >0.05). Therefore, ( :(<=>)$%&' ) was not included in the mass balance equations. 4.3. Results and Discussion 4.3.1. Electrostatic immobilization of probe DNA and optimization of the blocking step Immobilization of probe DNA on the beads was achieved via (1) electrostatic interaction only using QS beads or (2) a combination of electrostatic and hydrophobic 69 interactions, using CA beads. QS beads (Figure 4.1 - B I) are functionalized with quaternary ammonium (Q), which are strong anion exchange groups, resulting in a positively charged surface that allows an electrostatic immobilization of the complex streptavidin-probe DNA. CA beads (Figure 4.1 – B II), which have a strong anion exchanger with a primary amine group and a phenyl group, allow multi-modal electrostatic and hydrophobic interactions during the immobilization of the complex streptavidin-probe DNA. Typically, the multi-modal interactions confer a higher salt tolerance, which can expand the range of stringency conditions tolerated by the immobilized probe. In our tests, however, the electrostatic and multi-mode ligands' binding capacities were comparable when using BSA labeled with FITC as a model molecule since it is negatively charged at pH 7 (Figure 4.6). From Figure 4.6 – A, it is possible to see that the amount of BSA-FITC immobilized on the two types of beads is comparable in the tested range of concentrations and that saturation is achieved when using 200 μg/mL of BSA-FITC. This concentration was fixed and used for the subsequent experiments. To evaluate the effect of buffer conductivity, the fluorescence of BSA-FITC was measured after flowing solutions with 0 M, 0.137 M, 0.660 M, and 3.300 M of NaCl. The concentration of 0.137 M corresponds to PBS buffer, 0.660 M corresponds to the NaCl in 4x SSC buffer, commonly used as hybridization buffer, 3.300 M of sodium ion concentration corresponds to the SSC stock solution. Finally, 0 M of sodium ion concentration corresponds to MilliQ water, considered the most stringent condition. Figure 4.6 – B shows that higher concentrations of sodium ions in solution will significantly decrease BSA-FITC adsorbed onto the QS beads, resulting in complete desorption at 3.300 M. On the other hand, BSA-FITC is effectively captured on the CA beads independently of the sodium ion concentration. At the same time, a peak of adsorption was obtained using a NaCl concentration of 0.660 M. Overall, both QS and CA beads may be used when higher stringency levels (lower buffer conductivity) are required. However, lower stringency conditions are only achievable using CA beads. 70 Figure 4.6 - BSA-FITC adsorption isotherms and effect of conductivity on molecular adsorption. A - Optimization of BSA-FITC concentration adsorbed on QS and CA beads. The error bars represent the standard deviation of two independents assay. B - Measurement of the fluorescence intensity of BSA-FITC adsorbed on QS and CA beads after flowing solutions of increasing sodium ion concentrations. The error bars represent the standard deviation of two independent measurements. For the two solutions with higher concentrations of sodium ion, the adsorbed BSA onto the QS or CA beads was statistically different with a confidence level of 95% (p-value < 0.05 and highlighted with *). This work used a complex streptavidin-biotinylated probe DNA (hereafter mentioned as complex streptavidin-probe DNA) in a molar ratio of 1 : 4 for DNA immobilization. This molar ratio proved optimal in probe surface density and 3-fold superior for immobilizing only probe DNA (Figure 4.7). Therefore, streptavidin contributes to a favorable conformation of the probe DNA to facilitate hybridization by promoting a more upright orientation instead of a horizontal immobilization on the bead surface via the phosphate groups. 71 Figure 4.7 - Optimization of probe DNA-FITC immobilization on a positively charged surface, silanized with (3-Aminopropyl)triethoxysilane (APTES). (1) Electrostatic immobilization of probe DNA to a PDMS surface functionalized with APTES. (2) APTES functionalization of the PDMS channel followed by streptavidin immobilization and sequential flow of probe DNA. (3.1-4) APTES functionalization of the PDMS channel followed by a flow of a pre-mixed solution of the complex streptavidin : probe DNA in a molar ratio of 1:1, 1:2, 1:3, and 1:4 (illustrated in (3.1), (3.2), (3.3), (3.4), respectively). The error bars correspond to the standard deviation of two individual measurements, and the experiments were performed using an exposure time of 1 s and a 1x gain. These results highlight a statistical difference between the immobilization strategy (1) and (3.3) with a confidence level of 95% (p-value < 0.05) and between immobilization strategy (1) and (3.1), (3.2) and (3.4) with a confidence level of 99% (p-value < 0.01). Furthermore, we found a statistical difference between immobilization strategy (3.1) and (3.2), (3.1) and (3.4), and (3.2) and (3.4) with a confidence level of 95% (p-value < 0.05). Although no statistical difference was found between the immobilization strategy (3.3) and (3.4), the last condition was chosen as the immobilization strategy since it presented a smaller variability and was used throughout this work. Using a streptavidin : probe DNA complex in a molar ratio of 1:4 to perform the immobilization, the fluorescence intensity is increased by ~3-fold when compared to electrostatic immobilization of free probe DNA. A critical step in biosensor development is guaranteeing that the blocking agent effectively minimizes non-specific interactions with the free electrostatic or multi-mode ligands on the surface. The rationale behind the choice of blocking agents was to use molecules resembling the DNA probes in length and charge. Therefore, several polyacrylate (PA) molecules with different molecular weights were tested, as well as standard procedures, including saDNA and BSA (4%). It was observed that all blocking agents induced a time-dependent release of the previously immobilized probe. Therefore, a total blocking time of 2 min was selected to minimize probe loss (Figure 4.8). 72 Figure 4.8 - Optimization of the blocking time. This experiment was performed to assess whether the blocking agents removed the DNA probe from the surface of the beads over time. Each blocking agent was flowed at 5 µL/min for 10 min after immobilizing probe DNA using the procedure described in section 2.4. This allowed the selection of an optimal time for the blocking step. The images were acquired with an exposure time of 50 ms and 2.5 dB gain. The two-minute time point was chosen since, for the majority of the blocking agents, at least 50% of the initial DNA probes remained bound to the beads. The efficacy of the blocking was then tested by first blocking the beads with each blocking agent and then flowing probe DNA (Figure 4.9 – A and B). From Figure 4.9 – A it was possible to conclude that the most effective blocking agents were PA 15000 (5%), PA 15000 (1%), PA 8000 (5%), and PA 8000 (1%) with blocking efficiencies of 85.6%, 91.7%, 77.1% and, 74.6%, respectively, calculated as the percentage of the signal after washing relative to the control signal obtained using PBS only (Figure 4.9 – B). Three combinations were chosen from the previous blocking agent candidates and concentrations to be evaluated when performing a hybridization assay (Figure 4.9 – C and D). According to Figure 4.9 – C, the blocking agent that allowed a higher fluorescence ratio (5.9-fold difference) between cDNA and ncDNA was PA 8000 (5%), which was selected as the blocking agent for further experiments. 4.3.1. Quantitative analysis of DNA immobilization and hybridization The quantitative analysis of DNA immobilization and hybridization was performed for the more versatile CA beads since these beads displayed higher binding capacity towards the streptavidin–probe DNA complex after the blocking step (Figure 4.10). Before the blocking steps, we can see that the binding capacity for streptavidin-probe DNA is similar for CA and QS beads by the results obtained with BSA-FITC (Figure 4.6 – A). The accurate quantification of immobilized DNA probes is critical in biosensor development since it is known that there is an optimal probe density for hybridization kinetics and capture efficiency that peaks before decreasing for higher probe density values due to stereochemical effects [199]. 73 Figure 4.9 - Effect of different blocking agents on blocking efficacy and hybridization selectivity on QS beads. PBS was used as a control. A - Blocking efficiency – fluorescence from immobilized DNA for different blocking agents during immobilization and washing; B - Absolute fluorescence values at the end of each immobilization and washing step. C - Hybridization selectivity – fluorescence intensity ratio of cDNA hybridization to ncDNA measured during hybridization and washing for different blocking agents. D - Comparison of the fluorescence ratio at the end of the hybridization and washing steps for the different blocking agents. 74 Figure 4.10 - Comparison of fluorescence intensity of probe DNA immobilized on QS and CA beads before and after the blocking step. Probe DNA labeled with Atto 430LS was used to monitor the probe DNA immobilized onto the beads. The images were acquired using a fluorescence microscope and an exposure time of 50 ms,1x gain, and 160x total magnification. Before the blocking step, the probe density immobilized onto the beads was similar for both QS (A) and CA beads (C). However, for QS beads before (A) and after (B) the blocking steps, there was a loss of approximately 50% of probe density immobilized onto the beads, while for CA beads before (C) and after (D) the blocking step there was a loss of approximately 25%. This difference in probe density loss leads to a higher probe density in the CA beads compared to the QS beads in the step before target DNA detection. To quantify the mass of probe and target DNA, a mass balance approach was used to correlate the fluorescence signal intensity with the mass of molecules immobilized on the beads. First, calibration curves, one for the probe DNA immobilization onto the beads (Figure 4.11 - A) and two for target DNA (complementary DNA (cDNA) and non-complementary (ncDNA)) were obtained (Figure 4.11 – B). For the probe DNA calibration curve, concentrations ranging from 0 to 9.4 μM were tested, and the amount of probe DNA immobilized onto the beads was calculated as (70.4 ± 2.0) × 10@A! mol for a probe DNA concentration of 9.4 μM. The microbeads used in this work were 4% crosslinked agarose beads with an estimated total surface area of 5 m2/mL, allowing to estimate a probe surface density of ( 40.4 ± 7.5) × 10@A! mol/cm2. Comparing the probe density found in the microbeads with other 3D immobilization platforms, such as paramagnetic beads, the value found here is 23-fold higher. One possible explanation for this difference is the extra surface in porous beads where the probes can also be immobilized. In fact, from the total estimated surface area of the 75 agarose beads (5 m2/mL), only 2% corresponds to the outer surface of a solid sphere with an average diameter of ~90 μm. Figure 4.11 - (A,B) Calibration curves for known concentrations of DNA, used to determine : ( <=> ) !"# . (A) Standard curve obtained for streptavidin - probe DNA complex as a function of probe concentration. The vertical dotted line represents the probe concentration used in subsequent assays. (B) Calibration curves for cDNA and ncDNA as a function of increasing concentrations of the target concentration. The fluorescence intensity is derived from propidium iodide as a specific intercalator of double-stranded DNA. Higher concentrations of cDNA were statically different from the blank with a confidence level of 95% (pvalue < 0.05 and highlighted with *) or a confidence level of 99% (p-value < 0.01 and highlighted with **). Concentrations of ncDNA were not statically different from the blank assay. (C) Fluorescence intensity as a function of hybridized target DNA labeled with Atto 430LS. The vertical dotted line corresponds to probe DNA immobilized on the beads, and the orange band represents the standard error. (D) Target DNA molecules bound (left Y-axis) as a function of target concentration in solution introduced in the microfluidic channel, which was used to calculate the hybridization efficiency (right Y-axis). Error bars represent the standard deviation of two (A) or three independent experiments (B, C, and D). For the target DNA calibration curves, concentrations ranging from 0.3 to 100 nM were used. From Figure 4.11 – B, it is possible to conclude that, for the range of concentrations tested, the signal from the ncDNA was not significantly above the blank assay, thus in conditions where the mass measured at the outlet was not significantly different from the inlet, the hybridized mass was considered zero. For the cDNA target, it was possible to calculate the number of molecules hybridized to the probe DNA for the entire range of concentrations tested, from 3.4 × 10@AB ± 5.0 × 10@AC mol to K1.3 × 10@A! ± 1.8 × 10@AB mol (Figure 4.11 – C). For target concentrations above 3 nM, the average hybridization efficiencies varied between ( 95.3 ± 4.0 ) % to ( 97.3 ± 4.1 ) % (Figure 4.11 – D). This result demonstrates that the 76 probe density is in a regime that allows all probes to be available to bind the target DNA in solution without significant steric hindrance. In contrast, for 1 nM of target concentration, the hybridization efficiency is only (75.6K ± 11.4)K% , since we are approaching the equilibrium dissociation constant (KD) of the assay. Therefore, the expected KD for this system is lower than 1.0 × 10@D M (Experimental temperature: 20 ºC; GC content: 52%). This estimated value of KD is below a previously reported value of ([2.9 ±K0.9]K×K10@D M) using a 22-mer target with a 40.9% GC content. This difference could be due to the difference in the GC content, since a higher GC content for short strands results in a higher KD value. Moreover, in another study focusing on detecting a 16-mer target (50% GC content) with electrostatically adsorbed probe DNA at a density of 2.5 ×K10AE cm-2, the KD value obtained 0.2 K×K10@F M was above of the KD estimated in this work. Another reported system reports an average KD of (434.5 K±K18.8)K×K10@A! M for a 20-mer target with a 50% GC content. In this case, the probe DNA was covalently immobilized to the biosensor, and the authors estimated a probe density of (1.1 ×K10AA ) cm-2. One possible explanation for this low KD could be the probe conformation and structural difference of the hybridization strand due to different immobilization strategies used, covalent versus electrostatic. Due to the covalent immobilization strategy, the probes can be more accessible to the target. In addition, it is possible that the energy necessary to form the double DNA strand differs when the probe is immobilized electrostatically to a positively charged surface or to a covalently bound probe. 4.3.2. DNA detection in model buffer solutions After optimizing the assay architecture, calibration curves for increasing concentrations of target DNA were measured using three different labels for optical detection: Atto 430LS, Qdots, and HRP. The results are shown in Figure 4.12 – A, B, and C, respectively. Figure 4.12 – A shows the absolute fluorescence intensity for the hybridization of target DNA (cDNA and ncDNA) in both QS and CA beads. It can be observed that the CA beads provide lower minimum detectable limits for cDNA, above the 3σ threshold (three times the standard deviation ( s ) of a blank assay), despite the relatively higher nonspecific signal than QS beads. Furthermore, it can be noted that the CA bead assay provides a significantly higher background signal associated with the higher intrinsic fluorescence of the CA beads relative to the QS beads and due to the PA molecules used in the blocking step, which causes an increase in the baseline fluorescence of the assay (Figure 4.13). 77 Figure 4.12 - Biosensing of DNA in QS or CA beads using three different optical detection labels: (A) fluorescence using Atto 430LS in QS or QA bead-based assays, (B) Qdots labels and CA beads or (C) chemiluminescence using HRP and QA beads. The shaded regions represent the 3 s thresholds. The 3 s thresholds are calculated as the average of three blank assays, and the error bars represent the standard deviation of three independent measurements. (D) The signal intensity ratio between target cDNA and target ncDNA for each optical label. The horizontal dashed lines represent the 3 s thresholds. (E-G) Microscopy images for each detection method for target cDNA (top images) and target ncDNA (bottom images). (E) Fluorescence image of the Atto 430LS label, (F) fluorescence image of the Qdots label, an, (G) optical microscopy image of the HRP label. All images were contrast-enhanced for visualization purposes. 84 5.1. Introduction Detecting specific nucleic acid strands, such as DNA or RNA, is critical for developing nucleic acid biosensors. These sensors can answer the urgent need, especially in a global pandemic, for devices capable of detecting a viral and bacterial presence in a fast, sensitive, and specific fashion [202], [203]. Techniques for sensitive and rapid detection of bacteria and viruses in complex matrices, such as bodily fluids, environmental samples, or even from surfaces, are vital for treating and controlling infectious diseases [204], [205]. Current pathogen detection methods include culture-based methods, which are time-consuming, or molecular-based methods which include PCR-based detection assays performed in centralized labs, requiring specialized technical personnel and equipment. These detection methods can delay identification and treatment, leading to inappropriate therapy or ineffective pathogen suppression. The molecular assays are the most rapid, sensitive, and specific from the standard detection methods. PCR is the gold-standard method used in central labs and has also been increasingly implemented in point-of-care microsystems [74], [206], [207]. Microfluidic devices offer several advantages for detecting bacterial pathogens. These devices allow the reduction of sample volume and reagent consumption, increase the speed of analysis, and simplify the detection process through autonomous systems that will enable bacterial cell isolation, cell lysis, nucleic acid extraction, and amplification and amplicon detection through hybridization of a complementary detection oligo on a single chip [12], [208]. For this reason, several recent efforts have been to design DNA hybridization-based microfluidic biosensors. Many use electrochemical detection techniques that are prone to variability due to changes in the surrounding environment and have complex and time-consuming protocols [209], [210], or magnetic-bead-based detections that usually require a bulky apparatus for the actuation of the beads and the detection of a magnetic signal, increasing the cost of the device [211]. In addition, few studies have focused on the regeneration of the microfluidic device [209], [212]. In Good et al. biosensor regeneration review, there is no description of regeneration studies for DNA optical biosensors. In contrast, other analytes, such as antibodies and proteins, are well described [213]. Regeneration allows more assays on a single chip and enables the measurement of a single control sample under identical experimental conditions, thus providing more precise control of the assay and avoiding false-positive or false-negative results. This aspect is crucial to bringing DNA biosensors to the market. One of the main barriers preventing this type of biosensor from translating from a laboratory prototype to a commercial 85 product involves safety concerns with the risk of misdiagnosis [214]. The ability to regenerate with effectiveness the device also enables the reuse of the device, leading to a decrease in costs. DNA biosensors are usually designed for single use. Thus, there are few studies of regeneration conditions and optimal immobilization strategies to produce a high-performance DNA biosensor for the assay and the control. Furthermore, for the majority of regeneration protocols, the capture DNA strand is removed from the sensor surface, and the sensors would then require a new functionalization every time they are reused, increasing the complexity of the regeneration protocol [213] and eliminating the advantage of having a single, consistent control for multiple assays. Several techniques can reconstitute biosensors, such as chemical, thermal, or electrochemical. This work presents a simple chemical regeneration mediated by a basic medium (NaOH). This type of regeneration process is simple to implement on-chip and compatible with our assay since it will not interfere with the sensor output signal and the device's baseline, as would be the case with electrochemical systems or risk denaturation and damage of the capture molecule. The main goal of this work was to increase the reliability of the assay by having internal quality control. In these systems, having a control in the same microfluidic channel is relevant because inside the microchannel, there is a packed microbead column that can contribute to variability factors such as the degree of column packing and the probe density on the bead surface. The possibility of reusing the device was suggested due to the effectiveness of the regeneration obtained. Increasing the surface-to-volume ratio of the biosensor through the use of microbeads packed inside the microfluidic channel allows more DNA probes to be immobilized on the matrix, promoting higher sensitivities. We have previously demonstrated that using nanoporous microbeads inside a microfluidic network increases biosensors' sensitivity due to the microbeads' high surface-to-volume ratio [165], [171]. Moreover, it is possible to perform regeneration of the microbeads after antibody capture to perform chromatography cycles [215]. However, this methodology has yet to be applied to DNA biosensors. The regeneration of the microfluidic biosensing system results in a further decrease in costs. At the same time, integrating on-chip signal transduction allows for increased sensitivity and flexibility in system design. Recently, Boissinot et al. described a bead-based DNA microfluidic device to capture and detect amplicons via DNA hybridization and optical detection [109]. No regeneration of the device was directly attempted. These authors showed a competition between a target strand already immobilized to a probe DNA and a complementary strand flowing 86 through the device, which results in a release of the target strand from the probe DNA attached to the beads. However, the signal does not return to a baseline value, meaning that not all probes become free to capture a new target DNA. Ariffin et al. developed a DNA biosensor using hollow silica spheres as a surface to immobilize the ssDNA probe covalently. This device was regenerated four times using 100 mM of NaOH, with a 4-6% signal decrease after rehybridization with target DNA, and the detection was made via differential pulse voltammetry [212]. To the best of our knowledge, this is the first report of an optical read-out bead-based DNA biosensor with the ability to perform the regeneration of the device without removing or changing the conditions of the capture DNA layer (density, capture efficiency, and sensor baseline). This study can be transposed for short-strand DNA detection via hybridization, and different sequences can be employed. This paper presents a regenerable bead-based microfluidic device with integrated optical detection via thin-film photodiodes for the real-time monitoring of cDNA and ncDNA molecular recognition to a DNA probe. The ncDNA control assay is performed in the same microfluidic channel as the cDNA assay and thus represents a robust internal control of the assay. The optimal probe DNA immobilization to the beads was achieved by covalent bonding. Covalent bonding immobilization allowed the system's regeneration without significant loss of probe DNA while maintaining the same molecular recognition efficiency. This was confirmed by performing several highsensitivity assay cycles. This work demonstrates the potential for the system's reusability, portability, and reproducibility. 5.2. Materials and experimental methods 5.2.1. DNA oligonucleotides and beads used for DNA immobilization Synthetic 23 base-pair single-strand oligonucleotide sequences (Stabvida Genomics Lab, Portugal) were used to perform the immobilization and molecular recognition of 100% cDNA and 100% ncDNA strands (Table 5.1) on the surface of the nanoporous agarose beads that were packed inside the microfluidic device (Figure 5.1 - A). Three different DNA immobilization strategies were tested (Figure 5.1 - B). For QS Fast Flow beads and CA beads (GE Healthcare Life Sciences) (~90 μm average size), the beads were suspended in a solution of biotinylated probe DNA (9.4 μM) : streptavidin (2.35 μM) (Sigma-Aldrich) in a 1:4 M ratio (previously incubated for 10 min), to allow adsorption of the probe DNA : streptavidin complex to the beads through electrostatic interaction (QS) or electrostatic and hydrophobic interaction (CA), as described in 87 Chapter 4. Aminolink C6-modified probe DNA was covalently immobilized to Nhydroxysuccinimide-Activated Sepharose (NHS-Activated Sepharose 4 Fast Flow beads, GE Healthcare Life Sciences). The coupling of the aminolink C6-modified DNA to the NHS-Activated Sepharose 4 Fast Flow beads was performed as instructed by the supplier and using a 1 : 1 volume ratio of bead suspension and DNA solution [50 μM]. The coupling had an incubation step of 3 hours at rt with an agitation of 750 rpm/min. All three immobilization strategies were used to verify the optimization of the microcolumn regeneration for DNA detection. Table 5.1 - Description of the oligonucleotide sequences used in this work and respective modifications. Oligonucleotide Sequence 5’ Mod 3’ Mod Atto 430LS - probe DNA - biotin 5’-CAGGTCAAAAGGGTCCTTAGGGA-3’ Atto 430LS Biotin Atto 430LS - probe DNA - aminolink 5’-CAGGTCAAAAGGGTCCTTAGGGA-3’ Atto 430LS Aminolink-C6 probe DNA - aminolink 5’-CAGGTCAAAAGGGTCCTTAGGGA-3’ None Aminolink-C6 probe DNA - biotin 5’-CAGGTCAAAAGGGTCCTTAGGGA-3’ None Biotin cDNA – Atto 430LS 5’-TCCCTAAGGACCCTTTTGACCTG -3' None Atto 430LS ncDNA – Atto 430LS 5’-CGTGTCGTTCACATCTGTCCGT -3' None Atto 430LS cDNA – biotin 5’-TCCCTAAGGACCCTTTTGACCTG -3' None Biotin ncDNA – biotin 5’-CGTGTCGTTCACATCTGTCCGT -3' None Biotin 5.2.1. Packing of beads and optimization of microcolumn regeneration for DNA detection For the different types of beads used in this work, the packing was achieved by diluting the bead stock in PBS to obtain a solution with a final concentration of 0.12% bead volume. We used 20 μL of the PBS diluted bead solution to pack a bead column inside the microfluidic channel by applying negative pressure at the outlet using a syringe pump (NE-1002X, New Era Pump System Inc.) at a flow rate of 7 μL/min. 88 Figure 5.1 - Bead-based microfluidic assays. A - Schematics of the microfluidic setup with integrated thinfilm p-i-n photodiodes used for optical transduction. The photodiodes were aligned with the channel and positioned underneath the packed beads. B - The three types of nanoporous agarose microbeads evaluated for probe DNA immobilization: Q SepharoseTM (QS) Fast Flow (I-1) and CaptoTM Adhere (CA) beads (I-2) provided reversible immobilization of streptavidin – biotinylated probe DNA, while NHS Activated SepharoseTM beads (II-1) were used to immobilize amino-modified probe DNA covalently. To optimize microcolumn regeneration, we tested two conditions: 3 M and 0.3 M of sodium hydroxide (NaOH), and two types of assays were performed. These concentrations were selected based on similar values reported in the literature [216]. In the first assay, an Atto 430 LS labeled probe DNA was used to determine the effect of the two regeneration conditions on probe density. The following protocol was used when the probe DNA was immobilized via electrostatic or electrostatic and hydrophobic interactions: first, a solution with streptavidin and biotinylated probe DNA Atto 430 LS labeled, was incubated in an [1:4] M ratio for 10 min, as stated before. CA beads or QS beads were added to the previous solution to obtain a final solution with 0.12% bead volume and were incubated for an additional 10 min with orbital agitation at 3000rpm. After this, sodium polyacrylate (PA) 45% (w/w) (8000 average molecular weight) was added to the previous solution, obtaining a final solution with PA 8000 5% volume, which was incubated for 2 min with QS beads or 5 min with CA beads. Using 20 μL of this solution, the beads were packed inside the microchannel. In the case of the beads with covalently immobilized probe DNA - Atto 430 LS, no incubation steps were required, and the bead solution was packed into the microchannel. After a washing step with PBS at 7 μL/min for 1 min, 0.3 M of NaOH 89 regeneration solution was pumped through the channel at 7 μL/min for 3 min, followed by a washing step with PBS at 7 μL/min for 1 min. Another regeneration condition was tested in the same microcolumn by flowing 3 M of NaOH regeneration solution at 7 μL/min for 3 min, followed by a washing step with PBS at 7 μL/min for 1 min. The probe DNA fluorescence was monitored using a Leica DMLM microscope equipped with a DFC300FX CCD camera and a 100 W mercury arc lamp coupled to a blue light excitation filter with a band-pass excitation of 450–490 nm and a long-pass emission filter of 515 nm. Images were captured in 30 seconds intervals with an exposure time of 50 ms, 1× gain, and 160× magnification. The fluorescence signal was analyzed via measurement of the Green channel in the acquired RGB images using the software ImageJ (National Institutes of Health, Bethesda, MA, USA). In the second assay, we tested the effect of regeneration on the denaturation of the complementary and non-complementary DNA strands labeled with Atto 430 LS. In this case, the assay flow process was similar to that previously described for the probe DNA immobilization. After packing the beads inside the microchannel, 1 μM of complementary Atto 430 LS labeled DNA was flowed at 7 μL/min for 5 min, followed by a flow of a regeneration solution of 0.3 or 3 M of NaOH at 7 μL/min for 3 min concluding with a PBS washing step for 1 min. After that, the non-complementary Atto 430 LS labeled DNA flowed into the same bead-packed microchannel at 7 μL/min for 5 min, followed by another regeneration step of 0.3 or 3 M of NaOH and PBS washing step. The assay was monitored continuously and analyzed using the same strategy described before. 5.2.2. Calibration curves of one hybridization cycle using a fluorescence microscope The one hybridization cycle assay consisted of flowing a solution with the strand of complementary Atto 430 LS labeled DNA through a microchannel packed with the beads with a probe DNA covalently immobilized to its surface. Then, a step of regeneration with a 0.3M solution of NaOH followed by a flow of non-complementary Atto 430 LS labeled DNA, followed by another step of regeneration with 0.3M of NaOH, with PBS washing steps in between. The flow rates and times used for each step are the same as in section 2.3, and the target DNA concentrations tested varied from 3 to 1000 nM. Images were acquired using an exposure time of 1 second, 1× gain, and 320× magnification with a frame rate of 60 seconds. 90 5.2.3. Chemiluminescence calibration curves of one hybridization cycle Chemiluminescence detection was also performed to improve the sensitivity of the assay further. Moreover, chemiluminescence detection is simpler to implement in a point-of-care context when compared to fluorescence assays. The simplicity is associated with the setup required (no need for excitation light) and the device itself (the photosensor does not require a filter layer). We used streptavidin horseradish peroxidase conjugate (HRP - streptavidin) as a labeling molecule for the chemiluminescence detection assays, purchased from Invitrogen and diluted from the stock solution to 1 mg/mL in PBS. Luminol SuperSignal® West Femto Chemiluminescent Substrate kit (Thermo Scientific) was used as substrate. All the steps were similar to the ones performed in the assay of section 2.4, but, in this case, biotinylated complementary and non-complementary DNA strands were utilized. Then, HRP - streptavidin was diluted to 500 nM in PBS and flowed through the beads at 5 μL/min for 3 min, followed by a PBS washing step. After that, Luminol was introduced into the channel by applying positive pressure at a flow rate of 20 μL/min for 2 min. The emitted chemiluminescence light was acquired using thin-film p-i-n aSi:H photodiodes under dark background conditions. The regeneration of the microcolumn was performed using 0.3 M of NaOH after each Luminol step. The target DNA concentrations tested ranged from 10 000 – 30 pM. 5.2.4. Signal transduction using a-Si:H photodiodes and for photocurrent acquisition Here, p-i-n a-Si:H photodiodes with (Figure 5.2 - A I) and without (Figure 5.2 - A II) an amorphous silicon carbide (a-SiC:H) absorption filter were used for fluorescence and chemiluminescence assays, respectively. 91 Figure 5.2 - Integrated optical transduction after target capture in regeneration assays. A - Schematics of the p-i-n a-Si:H photodiode aligned with the bead-packed microchannels. An integrated a-SiC:H excitation light filter is deposited on top of the sensor (I) for fluorescence measurements. An identical sensor was used without the filter (II) for chemiluminescence measurements. B - Schematics of the hybridization assay cycle with regeneration for fluorescence (I) and chemiluminescence (II) detection. For chemiluminescence assays after the biotinylated target DNA, it was necessary to flow the streptavidin-HRP followed by luminol substrate to obtain the optical signal. To record the photocurrent (measured at 0V), a picoammeter (Model 237; Kiethley Instruments, Inc.) connected to the photodiode PCB via coaxial and triaxial connections was used. The devices, particularly the areas with the packed beads, were manually aligned directly on top of the photodiode. For the fluorescence assays, a 405 nm laser excitation light (photon flux (ϕ) = 1.04 × 1017 cm-2×s-1) was used, and the laser beam was run through a 1.5 neutral density filter before being aligned and focused on top of the area with the packed beads. The solutions flowed inside the 92 microchannel, and the current was continuously measured throughout the assay duration. Moreover, the dark current and the current upon flowing the luminol solution through the channel were also acquired for the chemiluminescence measurements. 5.3. Results and Discussion 5.3.1. Strategies for probe DNA immobilization onto the beads and optimization of microcolumn regeneration conditions Three probe DNA immobilization strategies for the beads were tested to optimize regeneration: electrostatic interaction, multimodal interaction, and covalent immobilization. The optimized conditions for electrostatic and multimodal immobilization were presented in Chapter 4. In the case of covalent immobilization, the highest DNA probe concentration was obtained using a 1 : 1 volume ratio of bead suspension and DNA solution [50 μM] incubated for 3 hours. This condition was obtained in an optimization assay where 1 : 1, 1 : 0.5, and 1 : 0.25 volume ratios of bead suspension and DNA solution [50 μM] were tested, as well as different coupling incubation times from 2 – 4 hours. Using an Atto 430 LS labelled probe DNA, we tested three probe DNA immobilization strategies aforementioned, under two regeneration conditions by flowing 0.3 M of NaOH (condition 1) followed by a flow of 3.0 M NaOH (condition 2) (Figure 5.3 - A). The goal was to choose a regeneration condition to minimize the removal of probe DNA immobilized on the microbeads. After flowing the regeneration solution, a washing step with PBS is performed. This step is required since the fluorophore's fluorescence can change under different salt conditions. Thus, measuring after the washing step ensures that the fluorescence is not affected by the salt condition of the various regeneration solutions, allowing them to be compared. For condition 1, the covalently immobilized and electrostatically immobilized probe DNA maintained the same initial fluorescence, 97.6% ± K3.9% and 99.8% ± 5.0% , respectively. These results were obtained using the fluorescence values at 4 minutes (Figure 5.3 - A). However, for the multimodal immobilization (electrostatic and hydrophobic) using the CA beads, the fluorescence decreased by 10% from the initial fluorescence. Immediately after condition 1 and the washing step, the same channel with the same beads was used to test condition 2. After the PBS washing step, we can observe that the loss of probe DNA is higher than in condition 1. In the case of the electrostatic interaction, there was a loss of 65.6% ±10.3% , while for the multimodal interaction, there was a loss of 49.3% ± 1.1% . As expected, due to the covalent bond between the probe DNA and the NHS-Activated Sepharose 4 Fast Flow beads, the probe DNA loss was minimal (approximately 0% ). These results 93 were obtained using the fluorescence values at 8 minutes (Figure 5.3 - A). We can also conclude that using the regeneration condition with 0.3 M NaOH is the best option to avoid probe DNA removal. However, in this case, we want to ensure that we can de-hybridize and remove the cDNA and the ncDNA from the channel walls and the surface of the beads. To do that, we performed regeneration cycle experiments, in which we evaluated the different probe DNA immobilization strategies through a cycle of hybridization with Atto 430 LS labeled cDNA, followed by 0.3 M NaOH for column regeneration, Atto 430 LS labeled ncDNA and another regeneration step (Figure 5.3 - B). This experiment was repeated but used the regeneration condition with 3 M NaOH (Figure 5.3 - C). We can conclude that we cannot fully regenerate the bead column using electrostatic and multimodal immobilization for both conditions. Moreover, since the regeneration is not complete after flowing the cDNA and the interaction between the DNA and the beads is due to electrostatic and hydrophobic interactions, the ncDNA can be attached to the beads, giving a final value higher than for cDNA. However, when using the covalent immobilization of probe DNA, even with the lowest concentration of NaOH tested, we can regenerate the bead column after both the cDNA and ncDNA hybridization steps, achieving fluorescence values of the order of the background level after regeneration. Since there was no difference between using 0.3 M or 3 M of NaOH for the covalent immobilization, we selected the lowest concentration as the regeneration condition for the subsequent experiments. This type of regeneration requires only flowing a regeneration solution through the packed beads inside the microfluidic channel, making it compatible with PoC devices. Furthermore, we can also observe that the system is more stable with covalent immobilization than with the other immobilization strategies used, based on the minor variations between assay repetitions. Selecting the covalent immobilization strategy and the regeneration with 0.3M NaOH allowed to include a ncDNA control assay in the device in the same microfluidic channel as the cDNA assay, leading to robust internal control. This does not always happen when using two different microfluidic channels in parallel to perform in one detection of a target and in another, a control. The robustness of the assay in other channels could be guaranteed only if the assay conditions were exactly the same (probe density, packing of the beads, sample, photosensor output, channel geometry), and fluctuations of these parameters are expected in real assay conditions. 100 Table 5.2 - Critical comparison of similar methodologies in the literature. Authors Ref Biosensor system Immobilization surface Probe DNA immobilization Analyte Reg conditions Repeats Signal loss Assay time LoD This work Fluorescence tagged ssDNA target optical detection Agarose pourous microbeads (90 µm diameter) Covalent bond ssDNA 300 mM of NaOH; probe is maintained 4 nonsignificant 10 min 7.8 x 10-11 M Boissinot et al. [109] Fluorescence tagged DNA target optical detection Streptavidin coated polystyrene beads (22.7 µm diameter) StreptavidinBiotin ssDNA amplicons complementary DNA strand, probe is maintained 1 n.a 5-15 min 5.6-5.8 x 10-5 M* Ferguson et al. [209] Label-free ssDNA electrochemical sensor Planar gold sensor surface Thiol-gold ssDNA amplicons 8 M guanidine hydrochloride 1 2% 20 min < 2 x 10-18 M Bronder et al. [69], [210] Label-free electrical detection of ssDNA Planar SiO2 sensor surface modified with poly(allylamine hydrochloride) layer Electrostatic interaction ssDNA adsorption of a new poly(allylamine hydrochloride) layer 5 72% to 88% signal loss after 5 repeats 55 min 5 x 10-6 M* Wang et al. [211] ssDNA preparation out of dsDNA amplicons for Streptavidincoated magnetic beads (1.05 µm diameter) Streptavidinbiotin ssDNA amplicons 100 mM NaOH; probe is maintained 1 n.a 45 min 2.5 ng 101 Authors Ref Biosensor system Immobilization surface Probe DNA immobilization Analyte Reg conditions Repeats Signal loss Assay time LoD magnetic beadbased microarray analysis with optical detection Ariffin et al. [212] Label-free lectrochemical sensor for ssDNA detection Hollow silica spheres (HSiSs) (50-200 µm diameter) Covalent bond ssDNA 100 mM NaOH; probe is maintained 4 4-6% 60 min 1.74 x 10-18 M Hong et al. [117] Quartz crystal microbalance (QCM) sensor Avidin modified gold-coated quartz surface Streptavidinbiotin ssDNA amplicons 0.8 M Trisglycine, pH 2.3; probe is maintained 32 30 % signal decrease at cycle 32 50 min 1.6 x 10-9 M Pursey et al. [222] Multiplex electrochemical sensor for ssDNA detection Planar gold sensor surface Thiol-gold ssDNA 50mM NaOH; probe is maintained 4 nonsignificant 20 min 2.50 x 10-13 M 102 103 Chapter 6 Bead-based microfluidic device for the capture and detection of DNA amplification products 6. Silica bead-based microfluidic device with integrated photodiodes for the rapid capture and detection of rolling circle amplification products in the femtomolar range This chapter demonstrated the combination of an isothermal amplification technique, RCA-based nucleic acid amplification off-chip, with on-chip size-selective trapping of amplicons on silica beads (~ 8 nL capture chamber) coupled with a thin-film photodiode (200 × 200 μm area) fluorescence readout. This module can be used as a standalone detection module for labeld products of amplification, acting as a sample concentrator. Parameters such as the flow rate of the amplicon solution and trapping time were optimized, as well as the photodiode measurement settings, providing minimum detection limits below 0.5 fM of targeted nucleic acids and requiring only 5μL of the pre-amplified sample. Finally, the analytical performance of this approach is evaluated by benchmarking it against a commercial instrument for RCA product (RCP) quantification and further investigated the effect of the number of RCA cycles and elongation times (ranging from 10 to 120 min). Moreover, we demonstrate the application for diagnostic purposes by detecting RNA from influenza and Ebola viruses, thus highlighting its suitability for integrated PoC systems. The contents of this chapter are summarized and reproduced from one original research article entitled “Silica bead-based microfluidic device with integrated photodiodes for the rapid capture and detection of rolling circle amplification products in the femtomolar range” [223] co-authored (equal contribution) with Doctor Ruben R. G. Soares (at the time Ph.D. student at INESC-MN), Doctor Felix Neumann and Doctor Narayanan Madaboosi. I was responsible for the development of the experimental microfluidic assays and troubleshooting. 104 6.1. Introduction Simple diagnostic devices are currently in high demand to address clinical needs at the PoC, particularly concerning the screening of single-stranded RNA viruses such as Influenza and Ebola, having high mutation rates [224], [225] and the potential to cause devastating pandemics [226]–[228]. Fluorescence-based optical read-outs are the most common transduction method used for bioassays since they are characterized by high sensitivity and specificity [229]. However, fluorescence detection equipment is typically expensive, non-portable, and requires trained personnel, thus being challenging to interface with miniaturized bioassays for field or PoC diagnostic applications [193], [230], [231]. Semiconductor technology allows for true lab-on-a-chip (LoC) applications since optical systems can be miniaturized and integrated with the assay on-chip practically and economically [165]. In these systems, micro-fabricated photodiodes are ideal since they are portable, affordable, reusable, and highly multiplexable for PoC devices [232], [233]. Thin-film hydrogenated amorphous silicon (a-Si:H) photodiodes, as integrated optical signal transducers, have demonstrated promising properties, such as low dark current, fabrication at low temperatures compatible with glass and plastic substrates, high quantum efficiency in the visible spectrum, as well as high sensitivity and a wide dynamic range [234], [235]. Nucleic acid biosensors have revolutionized the field of diagnostics since they simplify traditional testing methods, both in terms of assay time and complexity, offering high specificity and sensitivity. In particular, those based on PCR have shown outstanding performance in the diagnosis of genetic and infectious diseases in healthcare facilities [204]. However, PCR requires specialized instruments for power-intensive thermocycling, thus limiting their use at the PoC [236]. In this context, isothermal amplification methods have been developed to overcome the need for thermocycling [237]. In particular, RCA has proven to be a simple isothermal amplification technique that does not require extensive assay optimization and is typically robust against interferents [238], [239]. Upon combining RCA with PLPs, a powerful molecular detection tool is created, finding numerous applications in the diagnostic field [240]– [242]. RCA products (RCPs) comprise discrete molecules of ~1 μm in diameter, which by hybridizing fluorophore-labeled complementary oligonucleotides, display a high local fluorescence intensity that can be imaged using standard low-magnification epifluorescence microscopy, thus making linear RCA-based amplification intrinsically digital, since each molecular recognition event generates one long ssDNA (~60 kb at ~1 kb/ min[243]). Different read-outs and biosensing strategies have been reported to 105 detect RCPs, such as electrical [240], [244], [245], colorimetric [240], or optomagnetic [241]. However, a simple and integrated RCP fluorescent read-out strategy has not been described. It could be highly advantageous to simplify on-chip fluorescence signal transduction using linear RCA-based methods. In this context, the use of silica microbeads to selectively isolate and concentrate nucleic acids is a practical, cost-effective, and well-known procedure [246] previously integrated into commercial DNA extraction kits [247] and miniaturized devices [248]. The interaction of DNA with silica at neutral pH occurs mainly by hydrophobic and ionic phosphate-silanol interactions [249], largely overcoming the negative charge repulsion between both molecules. These interactions are also known to be significantly promoted in the presence of high concentrations of chaotropic salts such as guanidinium chloride (GdnHCl) [250]. Moreover, the DNA capture efficiency is known to significantly decrease for short fragments, in the range of tens to a few hundred base pairs, even at high salt concentrations, as is the case of the non-specific free detection oligonucleotides [251]. This combination of properties makes silicabased extraction ideal for the selective capture and enrichment of RCPs. In this work, we develop and optimize a silica bead-based microfluidic device integrated with a miniaturized thin-film photodiode for the enrichment and fluorescent detection of RCPs. For this, we characterize the microfluidic trapping of RCPs on silica beads and evaluate the sensitivity of the integrated a-Si:H photodiodes. We further demonstrate the diagnostic capabilities of our biosensor using RNA extracted from infected cell culture isolates of Influenza and Ebola viruses. Thus, this work points towards developing simple bioassay schemes for microfluidic devices with integrated opticalto-electrical signal transduction to address the need for compact and portable biosensing platforms for biomedical applications. 6.2. Materials and experimental methods 6.2.1. Rolling circle amplification (RCA), circle-to-circle amplification (C2CA), and quantification of RCA products (RCPs) Influenza isolates A/Stockholm/8/2015 grown in MDCK cells was kindly provided by the Karolinska Institute. The RNA was extracted with a MagNA Pure 96 instrument and MagNA Pure 96 DNA and Viral RNA Large Volume Kit (Roche, Sweden). RNA was stored in an elution buffer at -80 °C until further use. The influenza RNA titer was determined by limiting dilution analysis in the Simplexa™ assay and Poisson distribution using the Simplexa™ Flu A/B & RSV Direct Kit (Focus Diagnostics Inc., 106 Cypress, USA) according to instructions from the manufacturer. Ebola samples were obtained from the Public Health Agency of Sweden, under BSL-4 biocontainment facilities, as inactivated cell culture isolates. The RNA was extracted using the Directzol™ RNA isolation kit (Zymo Research, California, USA) according to instructions from the manufacturer. To perform the reverse transcription, two different optimized protocols were used for either influenza or Ebola samples. For influenza, cDNA synthesis, 100 nM of 5´ biotin-modified primers (Integrated DNA Technologies, Coralville, USA) and 500 μM dNTPs (Thermo Fisher Scientific, Waltham, USA) were added to the viral RNA (vRNA). The mixture was preheated at 65 °C for 5 min and snap-cooled on ice. After that, 100 mU/μL RiboLock, an RNase inhibitor, 250 mU/μL Transcript ME reverse transcriptase, and reverse transcription buffer (Blirt S. A., Gdansk, Poland) were added to a total volume of 20 μL. Reverse transcription was performed at 55 °C for 20 min, followed by 5 min heat inactivation at 85 °C. Ebola cDNA synthesis was performed using the SuperScript™ III Reverse Transcriptase kit according to the instructions from the manufacturer (Invitrogen, Thermo Fisher Scientific, Waltham, USA). The reverse transcribed influenza and Ebola samples were subjected to C2CA and RCA, respectively, using sets of 8 PLPs each. The oligonucleotide sequences used are summarized in Table S1. The ligation of PLPs was performed in Φ29 DNA polymerase buffer (33 mM Tris-acetate, pH 7.9, 10 mM Mg-acetate, 66 mM K-acetate, 1 ‰ (v/v) TWEEN® 20 and 1 mM DTT) (Mgacetate, K-acetate, TWEEN® 20 and DTT were obtained from Sigma-Aldrich, Darmstadt, Germany, while Tris-HCl, PBS, and NaCl were acquired from Karolinska Institute Substrat, Stockholm, Sweden) containing 0.2 μg/μL BSA (Sigma-Aldrich, Darmstadt, Germany), 0.68 mM ATP (Thermo Fisher Scientific, Waltham, USA), 500 mU/μL T4 DNA ligase (Blirt S. A., Gdansk, Poland) and 1 nM PLPs at 37°C for 20 min and inactivated at 65 °C for 2 min. Amplification was performed in Φ29 DNA polymerase buffer containing 125 μM dNTPs, 0.2 μg/μL BSA, and 400 mU/μL Φ29 DNA polymerase (Monserate Biotechnology Group, San Diego, USA) at 37 °C for 10 to 120 min and inactivated at 65 °C for 2 min. 107 Table 6.1 – Oligonucleotide sequences used. Eight PLPs were designed to target each of the H3N2 influenza subtype genome segments and 7 Ebola genes. Synthetic target DNA and a separate PLP were designed for the time point experiments. Furthermore, 4 detection oligonucleotides were designed for RCP detection. The PLP backbones include the binding site for detecting oligonucleotide and the Alu I restriction oligonucleotide (needed for C2CA). Type of Oligo Target Sequence (5' to 3' direction) RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT AGCGTAGACGCTTTGTC RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT CACGATGGAAAAGCATG RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT TCCTCATCGGAGGACTT RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT GACCAAAGTCTCCCACC RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT TTTGCGAAAAGCTTGAA RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT GTAACATGGTGGAATAGGAA RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT TAGGGGTTACTTCAAAATACGA RT Primer H3N2 Biotin-CTCTCTCTCTCTCTCTCTCT TCCTTTATGACAAAGAAGAAATAAG PLP H3N2 PO4GGGTGCATCTGATCTCATTAGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCAGACTTGCATTTGCCAAT PLP H3N2 PO4TGTTGGCTTGGCGCCAGATTGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCATGTAGCATCCTCACCAT PLP H3N2 PO4CATTCCCATTGAGGGCATTTGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCTCCATGTTATTTGGGTCTC PLP H3N2 PO4GCATTTTTATCATCCCCCGTGTGTATGCAGCTCCTCAGTAATAGTGTCTT ACATAGCACCGGAATAAGGCCCTGTAAATGAAGCTAGCAGTT PLP H3N2 PO4TTCTCAAGGCAGGAGAAGTTGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCCATCCACATAGGCTCTAAAA PLP H3N2 PO4AGAGACTCGAACTGTGTTATGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCGCGAATCTCTGTAGATTTTT PLP H3N2 PO4TTTCATTACCCCCAACCGGAGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCATTTGCCAGTTTGGCCTTCT 108 PLP H3N2 PO4TAATGGACCGTACTTGTCACGTGTATGCAGCTCCTCAGTAATAGTGTCT TACATAGCACCGGAATAAGGCCCAAGTTTTGTACACTTTTGGG Synthetic Target Ebola BiotinCTCTCTCTCTCTCTCTTTCCTGATACTTTGTGCACATACCGGCACCGGC TCTCTTCTGGCGCTGCTGGTAGACACTCACTCCCGTCA PLP Ebola PO4CCAGCAGCGCCAGAGTGTATGCAGCTCCTCAGTAATAGTGTCTTACATA TCGAGGATCGTCCGCATACGGGAGTGAGTGTCTA PLP Ebola PO4CAATTCATTTTCTCGGTAGTGTATGCAGCTCCTCAGTAATAGTGTCTTAC AGTAGCCGTGACTATCGACTAGGATTATTGTCATAAAT PLP Ebola PO4TTTGTGATTCGTCCTTTTGTGTATGCAGCTCCTCAGTAATAGTGTCTTAC AGTAGCCGTGACTATCGACTAAATAATCATTGACTAGT PLP Ebola PO4CCAGTGTAACAAATCTATGTGTATGCAGCTCCTCAGTAATAGTGTCTTAC AGTAGCCGTGACTATCGACTAATAAAATTGTTGACCAT PLP Ebola PO4TCTGATCAATTTTGTCTGGTGTATGCAGCTCCTCAGTAATAGTGTCTTAC AGTAGCCGTGACTATCGACTCAACAAAATCATGAATAA PLP Ebola PO4CGACACCTAGAGGAAGCCAAGTGTATGCAGCTCCTCAGTAATAGTGTCT TACCCTCAATGCTGCTGCTGTACTACGCTGTAGGTCTTTTGATCAG PLP Ebola PO4TCTTGGAAGATTGGAACTCTGTGTATGCAGCTCCTCAGTAATAGTGTCT TACCCTCAATGCTGCTGCTGTACTACGGATGACAGGTGGAGCAGCA DO RCA product AF430or Cy3-TTTTTCCTCAGTAATAGTGTCTTAC DO C2CA product AF430or Cy3-TTTTTGTAAGACACTATTACTGAGG After PLP ligation, RCA was performed in Φ29 DNA polymerase buffer containing 125 μM dNTPs, 0.2 μg/μL BSA, and 400 mU/μL Φ29 DNA polymerase (Monserate Biotechnology Group, San Diego, USA) at 37 °C for 10–120 min followed by inactivation at 65 °C for 2 min. In the case of C2CA, ligation of PLPs was performed using a mix of 10 nM of each PLP, 0.2 μg/μL of BSA, 500 mU/μL Ampligase, and 1 U/μL RNase H in Ampligase buffer (Epicentre, Nordic Biolabs AB, Täby, Sweden). 10 μL of cDNA were added and incubated for 10 min at 37 °C followed by 5 min incubation at 55 °C. Then, the biotinylated cDNA with ligated PLPs was first captured on streptavidin-coated DynabeadsTM MyOneTM T1 (Life Technologies, Oslo, 109 Norway). The beads were resuspended in 20 μL of amplification mixture (0.2 μg/μL BSA, 125 μM dNTPs, and 200 mU/μL Φ29 DNA polymerase in Ampligase buffer) and incubated for 20 min at 37 °C, followed by 2 min enzyme inactivation at 65 °C. Afterward, the liquid was discarded, and bead-bound RCPs were digested with 120 mU/μL Alu I (New England Biolabs, Bionordika, Stockholm, Sweden), 120 nM restriction oligonucleotide and 0.2 μg/μL BSA in 20 μL Ampligase buffer at 37 °C for 5 min followed by 2 min enzyme inactivation at 65 °C. The supernatant, containing monomerized RCPs, was transferred to new vials for the second amplification while discarding the magnetic beads. The second amplification mixture contained 0.2 μg/μL BSA, 0.68 mM ATP, 14 mU/μL T4 DNA ligase, 125 μM dNTPs, and 200 mU/μL Φ29 DNA polymerase in 10 μL Ampligase buffer. Amplification was performed for 60 min at 37 °C, and heat-inactivated at 65 °C for 2 min. Experimental assay steps of C2CA are illustrated in Figure 6.1 - A. For quantification, RCPs were labeled by adding 30μL hybridization buffer (1.4 M NaCl, 0.01% TWEEN® 20, 20 mM of Tris-HCl, pH 8, and EDTA) containing 5 nM of the respective detection oligonucleotide (Table S1) for 2 min at 75 °C and 20 min at 55 °C. RCPs from RCA and C2CA were quantified by amplified single-molecule detection (ASMD) using the dedicated instrument Aquila 400 from Q-linea (Uppsala, Sweden). The measurement average of technical duplicates was obtained, and the standard deviation was calculated. 6.2.1. Packing of silica beads and capture of RCPs in the microfluidic device and fluorescence measurements using fluorescence microscopy A suspension of mesoporous spherical silica beads (dried flash silica with 40–75 μm diameter and ~7 nm pores, purchased from Sigma-Aldrich) was first prepared in a solution of 20% (w/w) polyethylene glycol 8000 in water. After first pre-filling the device with water, the bead suspension was flowed into the 100 μm tall chamber at 10 μL/min by exerting a negative pressure at one of the outlets. The inlet of the 100 μm tall chamber was then sealed with a 20 ga metallic plug, and the PEG solution was washed with water through the 20 μm tall channel. The RCP solution obtained in subsection 6.2.1 was then diluted 5fold in an aqueous solution of 6 M guanidinium hydrochloride (≥99%, Sigma-Aldrich) and flowed through the 20 μm tall channel at flow rates ranging from 1.25 to 50 μL/min. 116 signal. For this, we first compared the photodiode response when performing RCA and C2CA on Ebola vRNA extracts. Figure 6.5 - A shows that both conditions result in a detectable fluorescence signal after only a few minutes of microfluidic trapping, thus demonstrating the suitability of both assays with the developed read-out. However, in this particular assay setup, RCA comes with the advantage of fewer assay steps (3 vs. 7 steps) when compared to C2CA, which facilitates eventual integration with the reported miniaturized device towards PoC applications. Moreover, to optimize the total assay time, we further investigated the sensor response for increasing times of RCA. Figure 6.5B shows a linear correlation (R2 = 0.99) between elongation time and the measured slope. From the obtained linear regression, the minimum amplification time necessary to provide a significant signal (3.3σ) above the background was calculated as 41.5 min. The linearity of generated signal was maintained throughout all elongation times proving the suitability of our trapping method for a wide range of RCP sizes, estimated to range between 10,000 and 120,000 bp [243], without a simultaneous loss of capture efficiency. Figure 6.5 - Detection of Ebola vRNA using RCA and the miniaturized read-out device. A - Detection of resulting RCPs after one (RCA) or two (C2CA) cycles of amplification for positive (P) (14.3 ng.µL-1 of viral RNA) or negative (N) (type I water only) samples. The inset plot shows the slope for technical replicates. B - Detection of RCPs generated using increasing RCA times. The inset plot shows the slope measured in duplicate for the same sample. 6.4. Chapter Conclusions In conclusion, we have presented a novel and integrated read-out strategy for RCAbased assays in a microfluidic format by combining size-selective trapping on silica beads with miniaturized p-i-n a-Si:H photodiodes. Our device provided a practical, portable, and affordable alternative over commercially available instruments for RCP detection with a wide dynamic range, resorting to sampling volumes below 5 μL and achieving total assay sensitivities in the low femtomolar range (LoD < 0.5 fM). These figures of merit demonstrate comparable, or superior sensitivities to recently reported 117 paper-based diagnostic devices using reverse transcription loop-mediated isothermal amplification (RT-LAMP) for viral detection, combined with a dramatic decrease in sample requirements (5–10-fold) and automated signal transduction [253], [254]. On the other hand, while superior sensitivities (~10-fold) were previously reported using fully integrated lab-on-disk devices coupled with RT-PCR amplification [255], significantly higher sample volumes (200 μL) were required. Immediate future perspectives include the full integration of the RCA assay with the reported read-out, bringing our approach closer to the PoC by avoiding lengthy and laborious off-chip amplification procedures. Furthermore, since the use of PLPs offers high multiplex capability, the assay can be extended to detect several pathogens [252] and implemented in a multiplexed format for a broader selection of targets, including antimicrobial resistance markers. 118 119 Chapter 7 Bead-based microfluidic chip for chemical lysis of bacteria and viruslike particles with integrated nucleic acid capture and detection 7. Chemical lysis and nucleic acid capture of Gram-positive, Gram-negative, and virus-like particles using a bead-based microfluidic chip Point-of-care tests can provide an appropriate medical diagnosis for pathogens, and they should be robust and universal enough to detect Gram-positive and Gram-negative bacteria and even nonenveloped and enveloped viruses. In this sense, cell lysis and nucleic acid extraction are critical steps with a profound impact on the success of the diagnosis. In this chapter, we describe the development of a simple, continuous, rapid, and universal chemical lysis module featuring an integrated microfluidic bead-based capture chamber for nucleic acid extraction. Escherichia coli was used as a model for Gram-negative bacteria, Staphylococcus aureus was used as a model for Gram-positive bacteria, and MS2 was used as a model for nonenveloped virus like-particle. Cell lysis and on-chip genomic DNA and RNA detection were performed in under 30 minutes. Using a mass balance approach and quantitative polymerase chain reaction, we could quantify lysis and capture efficiencies on-chip for the different lytic solutions tested. The strategy that provided the best results for Gram-negative and Grampositive bacteria and nonenveloped virus like-particle was with GenoLyse® + BPER + Proteinase K (50 % ± 21 %, 58 % ± 20 %, and 47 % ± 27 % of capture efficiency, respectively). This work was presented at the 23rd International Conference on Miniaturized Systems for Chemistry and Life Sciences (µTAS 2019) as a Poster. The contents of this chapter are summarized and reproduced from one original research paper draft submitted for publication. I fabricated the microfluidic devices. Conceived, designed, performed, and analyzed the experimental results. I was the main writer of the publication and creator of the Poster. 120 7.1. Introduction Two global health threats of increasing urgency for public health are viral infections and AMR. Viral infections can spread swiftly across the globe and cause a pandemic, as was observed with the COVID-19 pandemic, with significant consequences to human health, the global economy, and social interactions. On the other hand, infections caused by antibiotic-resistant bacteria (ARB) that were disregarded during the pandemic crisis will become more evident again as the world returns to a new normal, becoming a significant threat to global health [256]. AMR is considered a One Health problem that caused more than one million deaths worldwide in 2019, with predictions to reach 10 million deaths annually by 2050 [1]. One proven way to fight these threats is by controlling the source of infection, reducing the spread, and containing the transmission of these pathogens via early detection [257]. In this sense, accurate diagnostic systems are mandatory for successful global health management [258], [259]. Current detection methods rely on culture methods, molecular methods (such as PCR), and enzyme immunoassays, and the test result can often take 2 to 4 days to be obtained [204]. PoC tests can provide an appropriate medical diagnosis for bacterial and viral pathogens. However, PoC tests must be robust and universal enough to detect both Gram-positive bacteria (e.g., MRSA) and Gram-negative bacteria (e.g., CRE) or even enveloped (such as the SARS-CoV-2) and nonenveloped (such as the Norovirus) viruses. PoC devices based on molecular methods would confer the sensitivity required for detection and the specificity to distinguish infections presenting similar symptoms [257], [260]. Although there are already a few commercially available PoC devices to detect infectious diseases, significant improvement is still needed. For instance, sample preparation (including a lysis step) in these devices is usually performed off-chip. A fully integrated device is required, with on-chip sample preparation, that can achieve detection limits of 1-10 CFU or PFU/mL [39], [261]. Expanding the panel of pathogens these devices can detect is also urgent to fight viral infections and antibiotic-resistant bacteria. There is still a need to have a device that will detect specific pathogens, such as viruses or bacteria, starting from the raw biological sample (such as a swab). The first steps in a POC test, which uses a molecular method approach to detect pathogens, are sample preparation and cell lysis. These steps will affect all the downstream processes and impact the success of the diagnosis in terms of sensitivity and specificity. The efficiency of the lysis of bacteria and virus cells may differ due to structural differences, which compromises subsequent analysis steps. Thus, developing a universal lysis and extraction module 121 that can lyse and capture nucleic acids from Gram-positive and Gram-negative bacteria and virus-like particles is a challenge that holds the key to successful sampleto-answer PoC tests. This work aims at simplifying both steps in a lysis module that can be used in a fully integrated PoC device. Here we exploit universal chemical cell lysis for bacteria and viruses coupled with SPE (bead-based) of the nucleic acids present in the solution. Non-mechanical lysis methods can be divided into physical, chemical, and enzymatic processes. In the first case, physical disruption of the membrane is accomplished by non-contact methods that utilize external forces such as heat, pressure, and sound energy. In the case of high temperatures, the proteins in the membrane are denatured, leading to the release of the intracellular components [261]. Thermal lysis was used and considered the gold standard for lysis in this work. To disrupt the membranes, chemical methods use lysis buffers, such as alkali buffers. The main constituents of these buffers are sodium hydroxide and SDS. The OHions break the fatty acid glycerol ester bonds, and the SDS solubilizes the membrane leading to cell disruption. One downside of this method is that the process is usually long (6 - 12 h). Detergents such as BPER can also be used to disrupt lipid-lipid, lipid-protein, and protein-protein interactions to affect the membrane. Finally, enzymes such as lysozyme and protease can be used to lyse specific cells. Lysozyme is usually used to lyse Gram-positive bacteria by attacking the peptidoglycan layer and breaking down the glycosidic bond [262]. Commercial lysis kits are also available and can be chosen according to the type of cell to be lysed. Usually, commercial kits use an admixture of standard lysis methods. For instance, GenoLyse® (GL), used to extract DNA from bacteria, uses a combination of alkaline lysis buffer and thermal heating. The current extraction methods typically use incubation steps with enzymes such as lysozyme and proteinase K (PK) to digest cell wall components and interfering proteins, respectively. They are often user-dependent, complex, time-consuming, or suffer from a low yield of nucleic acid extraction [263]. Recent works of on-chip lysis methods have shown that chemical lysis combined with mechanical lysis has a good DNA recovery [264]–[267]. For instance, in the paper by Sciuto et al., the lysis and extraction required 15 minutes, with an LoD of 8 copies/reaction for the Hepatitis B virus [268]. We used S. aureus and E. coli as our Gram-positive and Gram-negative bacteria models, respectively, and bacteriophages M13 and MS2 as nonenveloped virions 122 models to test and compare different chemical lysis methods. The goal was to optimize lysis efficiency for the various organisms tested. We demonstrate a mass balance approach using a standard molecular method, qPCR, which allows us to calculate the number of nucleic acids extracted and therefore calculate the efficiency of the lysis methods. This method can be generalized to other pathogens and lysis methods to characterize such devices. Moreover, this allowed the quantification of the nucleic acid copies captured on-chip via an SPE method. To the best of our knowledge, this is the first time a continuous on-chip chemical lysis strategy has been applied to virus and bacteria cells. This mass-balance quantification approach can be helpful when selecting a lysis strategy to implement on-chip, especially when high sensitivities are required. Finally, all the tests were performed without amplification on-chip. The next step would be integrating an amplification module with this lysis strategy. The lysis method implemented on-chip will also influence subsequent modules such as amplification. By performing the qPCR outside the chip after the onchip lysis, we can already indicate if the on-chip amplification would be inhibited due to the on-chip lysis. 7.2. Materials and experimental methods 7.2.1. Bacteria and Bacteriophages S. aureus (ATCC® 19433) and E. coli (ATCC® 13706) were used as Gram-positive and Gram-negative bacteria models, respectively. S. aureus and E. coli were grown in tryptic soy agar (TSA, Biokar) and incubated at 36 (±2) ºC for 18 (±4) hrs. Following incubation, a cell suspension was prepared in sterile PBS 1×. A solution containing approximately 1.6×108 CFU/mL was obtained for both S. aureus and E. coli by measuring the optical density (OD) of the solutions in an 1102 UV/vis spectrophotometer (Techcomp, CN) at a wavelength of 600 nm (OD600). Bacteria calculations were performed using the following correlation [269]: 1+OPCDD += +8+ ×+10E+cell+per+mL+culture Equation 7.1 Adjustments to the solutions were made to reach a final absorbance of 0.200, which corresponds to a concentration of 1.6×108 CFU/mL, according to Equation 7.1. Bacteriophage MS2 (ATCC® 15597-B1TM) was quantified using reverse transcription digital PCR (RT-dPCR) carried out on a QuantSudio 3D Digital PCR System (ThermoScientific, US). The stock of MS2 bacteriophage was found to have a concentration of 1011 genome copies (GC)/mL or 109 PFU/mL. The M13 123 bacteriophage strain used in this study was based on the helper phage M13KO7 (New England Biolabs), engineered in pIII and propagated in E. coli 2737 (New England Biolabs). Cultures were grown overnight at 37°C with orbital shaking at 220 rpm in SOB Media supplemented with KCl (2.5 mM), MgCl2 (10 mM), ampicillin (100 µg/mL), tetracycline (10 µg/mL), and kanamycin (50 µg/mL). For phage recovery, PEG/NaCl was used [270]. M13 phages were obtained from the culture at 1011 PFU/mL in 1 × PBS. 7.2.2. Lysis assays on-chip The lysis in this work was performed at rt. The on-chip lysis was performed using the chip illustrated in Figure 7.1. Inlet 1 contained 60 µL of the microorganisms to be lysed. The concentration inserted was 108 CFU or PFU/mL unless stated otherwise. Inlet 2 was used to insert the mixture of lysis solutions. For the chemical lysis, different combinations were tested. A mixture of 1:1 (v/v) of GL (Hain Lifescience, Nehren, Germany) lysis solution with BPER solution (Thermo Fisher Scientific Inc., Rockford, IL). A mixture of 1:1 (v/v) of GL lysis solution with BPER solution and 0.1 of Proteinase K (Thermo Fisher Scientific Inc., Rockford, IL) and a solution of RNA rapid extraction solution (InvitrogenTM, MA, USA). For all cases, the total volume of solution inserted in inlet 2 was 60 µL. Inlet 3 was used to add 60 µL of a neutralization buffer after lysis into the solution. For the mixtures with GL present, the neutralization buffer used was supplied by the GL Kit. PBS 1× was used as a neutralization buffer for the RNA extraction solution. The solutions flowed simultaneously by applying a positive pressure with the conditions given in the bead packing and liquid handling subsection. Thermal lysis was also performed. The solution with the microorganisms was lysed in a microcentrifuge tube and placed at 95 ⁰C (Heating Block QBT1, Grant Instruments, UK) for 10 min. After this, the lysed microorganism was introduced into inlet 1, while inlets 2 and 3 were used to flow PBS 1 ×. After flowing the solutions through the beads, EvaGreen® (EG) dye was flowed through the microfluidic chip to intercalate in the nucleic acids captured by the beads allowing for subsequent on-chip fluorescence detection. A final washing step with PBS was performed prior to the fluorescence measurement. The fluorescence signal was monitored with a Leica DMLM microscope equipped with a DFC300FX CCD camera and a 100 W mercury arc lamp coupled to a blue light excitation filter with a band-pass excitation of 450– 490 nm and a long-pass emission filter of 515 nm. The setup used a 10× magnification, and 1× gain to acquire the images. The fluorescence signal was analyzed via measurement of the green channel in the acquired RGB images using the software ImageJ (National Institutes of Health). 124 Figure 7.1 - Schematics of the microfluidic device for pathogens lysis and genetic material detection. A - Photograph of the actual microfluidic device. B - Detailed image of parallelogram barriers mixer and dimensions of the mixer repeating unit. C - Detailed image of the microfluidic chamber used to pack the beads and respective sizes. The positively charged beads captured the free genomic DNA in the solution after lysis. 7.2.3. Lysis assays of-chip Several non-mechanical methods were applied: a physical method that uses high temperatures to lyse the organisms (thermal lysis) and three variations of chemical lysis, GL + BPER, GL + BPER + PK, and an RNA extraction kit, performed at rt. The lysis assays off-chip were performed in 1.5 mL microcentrifuge tubes (VWR). All the volumes were maintained the same as those used for on-chip lysis. For the quantification analysis, the outflow was collected in a microcentrifuge tube. Prior to qPCR or reverse transcription qPCR (RT-qPCR), the solution was centrifuged at 9700 g for 10 min. 7.2.4. Nucleic acid purification The lytic solutions where Proteinase K (PK) was added to the mixture required a purification step before quantification via qPCR. The purification was performed using the QIAamp DNA Blood Mini Kit (Qiagen, MD). The kit allows for lysis and purification of nucleic acids. Still, for the purpose of this work, we skipped the lysis steps in the 125 protocol supplied by the manufactured and applied only the steps to purify the solutions using the kit columns, following the manufacturers’ instructions. 7.2.5. Real-time quantitative PCR for microfluidic lysis characterization qPCR, RT-qPCR, and data analysis were performed in a 7300 Real-Time PCR detection system (Applied Biosystems). For the DNA-based pathogens (E. coli and S. aureus), qPCR was used to quantify the starting DNA quantity in the solutions after the lysis protocol. Amplifications were performed in a 25 μL reaction mixture containing 12.5 μL of Maxima Probe/ROX qPCR Master Mix (2×) (Thermo Scientific), 0.2 μL of each primer (800 nM), 0.05 μL of probe (200 nM), 5 μL of DNA template. The volume was adjusted to 25 μL with nuclease-free water. The primers and probes used in this study are presented in Table 7.1. Cycling conditions were as follows: 50 °C for 2 min; 95 °C for 10 min; and 40 cycles of 95 °C for 15 s and 60 °C for 1 min. To amplify the RNA-based microorganism (MS2), RT-qPCR was performed. In this case, the DNA concentration measured by the RT-qPCR was assumed to correspond directly to the total extracted RNA concentration of the bacteriophage. The quantitative analysis was carried out in 20 μL of the reaction mixture, where 5 μL was RNA extract. The complete mix contained 10 μL of specific one-step RT-qPCR (Luna® Universal One-Step RT-qPCR kit, New England Biolabs Inc.) buffer and 1 μL of RT-enzyme, 0.16 μL of forward primer and reverse primer each (800 nM), 0.04 μL of probe (200 nM), and 3.64 μL of nuclease-free water (ThermoFisher). Amplification was performed using the following protocol: 10 min at 55 °C to allow the reverse transcription to occur, followed by denaturation at 95 °C for 10 min, 45 cycles of denaturation at 95 °C for 15 s, and annealing at 60 °C for 45 s. For all the PCR runs, negative control without target DNA (water was used as a non-template control) was added for each microorganism tested. Positive controls were also added in each reaction and consisted of the genetic material extracted from each pathogen via thermal lysis at 95 °C for 10 min. The solutions were centrifuged at 12,000 rpm for 10 min, and dilutions were prepared and analyzed in the same run. Dilutions were performed to avoid false-negative results due to PCR inhibition. Results, such as amplification plots, cycle threshold values (Ct), and initial DNA concentration, were obtained using 7300 Fast System SDS software. All the primers and probes used for amplification are described in Table 7.1.