Comparative Analysis of Gene Expression in RNA-Seq Studies of Salmonella enterica: Focus on the Type VI Secretion System

Máster Universitario en Análisis de Datos Ómicos y Biología de Sistemas Universidad de Sevilla Universidad Internacional de Andalucía 1 Comparative Analysis of Gene Expression in RNA-Seq Studies of Salmonella enterica: Focus on the Type VI Secretion System Clara Muñoz Capilla, Francine Amaral Piubeli 1 *, Joaquín Bernal Bayard 1 * 1 1 Bacterial Genetics Research Group – BIO-116, Department of Genetics, University of Seville, 2 41012 Seville, Spain 3 * Correspondence: 4 Francine Amaral Piubeli 5 [email protected] 6 Joaquín Bernal Bayard 7 [email protected] 8 Keywords: Type VI Secretion System, Salmonella Typhimurium 14028S, RNA-seq, SPI-6 regulation, 9 Differential expression analysis, Galaxy platform. 10 Abstract 11 The Type VI Secretion System (T6SS) is a complex multiprotein apparatus present in many Gram12 negative bacteria, involved in injecting effectors into both microbial competitors and host cells. 13 Encoded by Salmonella Pathogenicity Island-6 (SPI-6) in Salmonella enterica subsp. enterica serovar 14 Typhimurium strain 14028S, its transcriptional regulation remains poorly defined under diverse 15 conditions. To fill this gap, publicly available RNA-seq data from 23 distinct experimental settings 16 were retrieved from the NCBI Sequence Read Archive (SRA) and reanalysed using a robust, 17 reproducible pipeline on the Galaxy platform. The results were organized in a relational database to 18 facilitate subsequent queries. Condition-specific expression profiles revealed a mosaic of T6SS gene 19 activation, often confined to subsets of the cluster rather than the entire system. For instance, co20 culturing with Acanthamoeba castellanii, L-arabinose supplementation, and HeLa cell infection 21 triggered selective upregulation of structural genes, demonstrating a nuanced transcriptional 22 responsiveness to microbial interactions and environmental signals. Downstream functional 23 enrichment analysis further emphasised the role of secretion and virulence-associated pathways under 24 these stimuli. This study highlights the value of leveraging public transcriptomic datasets to uncover 25 condition-dependent regulatory dynamics in bacterial systems. It also illustrates how integrate large26 scale data with curated analysis pipelines can inform future experimental hypotheses, while pointing 27 to the importance of consistent data annotation to fully exploit the potential of open-access biological 28 resources. 29 1 Introduction 30 The Type VI Secretion System (T6SS) is a protein translocation nanomachine, widely distributed 31 among Gram-negative bacteria, and structurally analogous to an inverted bacteriophage tail structure 32 (Cascales & Cambillau, 2012). This dynamic structure is composed of at least 13 conserved core com33 ponents (Silverman et al., 2012), which together form a contractile tube anchored to the cell envelope, 34 Transcriptomic Analysis of Salmonella T6SS 2 capable of puncturing neighbouring cells and delivering toxic effectors into bacterial or eukaryotic 35 targets (Jurėnas & Journet, 2021). The structure is organized into three main components: a membrane 36 complex, a baseplate, and a tail tube assembly, which is mostly composed of Hcp proteins and sur37 rounded by a contractile sheath formed by TssB and TssC. At the distal end, VgrG and PAAR proteins 38 form the tip and also function as effector carriers (Silverman et al., 2012). In Salmonella enterica subsp. 39 enterica serovar Typhimurium strain 14028S, the T6SS is encoded in Salmonella Pathogenicity Island 40 6 (SPI-6), and organised into two major operons that contain structural, regulatory, and accessory 41 genes, including effector and immunity pairs (Brunet et al., 2015). Numerous physiological functions 42 have been linked to this system, particularly in survival and replication within macrophages as well as 43 systemic dissemination in murine infection models (Mulder et al., 2012). Although not as extensively 44 studied as the Type III Secretion Systems (T3SS) encoded in SPI-1 and SPI-2, there is growing evi45 dence that the SPI-6 T6SS has a condition-dependent role in microbial competitiveness and virulence 46 (Wang et al., 2019). Furthermore, T6SS loci in S. enterica are not evenly distributed throughout all 47 serovars, as demonstrated by comparative genomic studies, and frequently represent evolutionary ac48 quisition through horizontal gene transfer. Some strains harbour additional T6SSs beyond SPI-6, such 49 as those in SPI-19 or SPI-20, each with potentially distinct functional specialisations (Blondel et al., 50 2009). The diversity of T6SS-associated effectors and immunity proteins, many of which display lytic 51 enzymatic activity targeting bacterial cell walls, supports the idea that these systems serve key roles in 52 interbacterial competence, although some T6SS also take part in eukaryotic interactions and nutrient 53 acquisition (Hespanhol et al., 2023). 54 Regulation of T6SS is highly complex and multilayered, including transcriptional and post-transcrip55 tional mechanisms that react to environmental factors such interbacterial contact, cell density, stress, 56 and ion availability (Hespanhol et al., 2023). Despite its potential influence in pathogenesis and bacte57 rial adaptation, the expression of SPI-6 T6SS under host-like or laboratory settings remains elusive 58 (Brunet et al., 2015). Studies have reported that many Salmonella strains exhibit low or undetectable 59 expression of T6SS components under typical in vitro growth circumstances, and regulatory factors 60 that might promote or repress activation are still not fully characterized (Mulder et al., 2012). A par61 ticularly relevant study by Kröger et al., (2013) provided a transcriptome resource for S. Typhimurium 62 across 17 infection-relevant settings, such as intracellular, environmental, and stress-related scenarios. 63 Although this dataset provides a useful tool for understanding transcriptional landscape of Salmonella, 64 it also revealed an unexpected result: the SPI-6 T6SS cluster was not significantly activated under any 65 of the examined conditions. These findings highlight the possibility that T6SS induction depends on 66 host-microbe interactions or very specific stimuli that have not yet been replicated in vitro (Mulder et 67 al., 2012). 68 In this regard, the research presented in this study was motivated by ongoing efforts in our group to 69 determine the environmental or host-like conditions that cause T6SS activation in S. Typhimurium 70 14028S. Building on publicly available transcriptomic resources and the need for systematic 71 experimental validation, we examined the expression profiles of T6SS-associated genes under a panel 72 of carefully selected in vitro and infection-mimicking conditions using RNA-seq. In order to replicate 73 a wide range of ecological and physiological scenarios pertinent to Salmonella's survival, adaptation, 74 and virulence, these conditions included intracellular infection, plant-associated environments, 75 exposure to antibiotics and oxidative stress, nutrient limitation, varying temperatures and pH levels, 76 and the presence of specific carbon sources. 77 The results obtained in this study reveal that T6SS expression in S. Typhimurium 14028S is 78 heterogeneous across the gene cluster and highly condition-dependent. Specific subsets of genes within 79 the SPI-6 locus are differentially activated under distinct experimental conditions, suggesting modular 80 Transcriptomic Analysis of Salmonella T6SS 3 regulation and functional compartmentalisation of T6SS. These findings contribute to a more detailed 81 understanding of how Salmonella modulate secretion systems in response to environmental variation 82 and may contribute to future studies aimed at understanding the regulatory network governing T6SS 83 activity. 84 2 Materials and Methods 85 2.1 Data acquisition and metadata analysis 86 To examine the range of publicly accessible transcriptomic data related to Salmonella enterica, and to 87 subsequently build a robust dataset for this research, a structured examination of the NCBI Sequence 88 Read Archive (SRA) repository was conducted. The resulting metadata, including fields such as 89 BioProject accession numbers, sample identifiers, library strategies (paired-end or single-end), 90 sequencing platforms (e.g., Illumina HiSeq), and other experimental parameters, were downloaded. 91 The metadata were standardized and formatted using R (R: A Language and Environment for Statistical 92 Computing, 2024) and Microsoft Excel to unify various data types and annotations. The curated 93 metadata was then loaded into Power BI environment (Microsoft Power BI, 2025) to perform an 94 exploratory data analysis. Interactive dashboards were created to illustrate the distribution of different 95 serovars and strains, with particular attention on the prevalence of S. enterica subsp. enterica serovar 96 Typhimurium. Strain 14028S was chosen for further analysis due to its clinical importance, high 97 frequency in the database, and its relevance to the research interests of our group, specifically in 98 relation to the T6SS. From this dataset, a curated selection of 14 representative studies covering 23 99 experimental conditions was chosen for downstream analysis, as detailed in Table 1, where the 100 corresponding BioProject accession codes and SRA sample identifiers are also provided. The 101 corresponding raw sequencing data (FASTQ files) for the selected datasets were downloaded using the 102 fasterq-dump tool (v3.1.1) (Leinonen et al., 2011) within the Galaxy server (v25.0.2.dev0) (Afgan et 103 al., 2016). 104 2.2 RNA-Seq data processing and quantification 105 A reproducible bioinformatics pipeline was built on Galaxy server to analyse all the raw sequencing 106 reads, with the workflow was differentially designed to handle both single-end and paired-end library 107 formats. Initial quality assessment of the raw FASTQ files was verified using FastQC (Afgan et al., 108 2016; Andrews, n.d.) to evaluate key parameters such as per-base quality scores, adapter content, GC 109 content distribution, and sequence duplication levels. Read trimming and filtering were carried out 110 using Trimmomatic (v0.39) (Bolger et al., 2014) with default parameters to remove adapters and low111 quality bases, ensuring high-quality reads for subsequent analysis. These processed reads were then 112 aligned to the reference S. enterica subsp. enterica serovar Typhimurium strain 14028S genome (NCBI 113 accession: GCA_000022165) using the STAR aligner (v2.7.11b) (Dobin et al., 2013) under default 114 parameters, generating coordinate-sorted BAM files. Summary statistics from the STAR log output 115 files are presented in Table 2 to provide an overview of mapping quality across all datasets. The cor116 responding genome annotation file in GTF format was also downloaded from the NCBI database as a 117 support for read assignment. Finally, gene-level quantification was performed using featureCounts 118 (v2.1.1) (Liao et al., 2014) to produce a table of read counts per gene for each sample, where reads 119 were assigned to features annotated as “gene” and grouped by gene identifier. 120 2.3 Differential gene expression analysis 121 The resulting gene-level count files were downloaded from Galaxy server and processed using a 122 custom R script to generate a count matrix for each experimental condition. Differential gene 123 Transcriptomic Analysis of Salmonella T6SS 4 expression (DGE) was subsequently conducted by comparing each condition against a common 124 reference consisting of samples cultured under Luria-Bertani (LB) medium (Hoang & Stoebel, 2024), 125 which served as the baseline control in all comparisons. The statistical approach for DGE analysis was 126 selected based on the number of available biological replicates: for experiments with two or more 127 biological replicates (n ≥ 2), differential expression was assessed using the DESeq2 package (v1.46.0) 128 (Love et al., 2014), with genes considered significantly differentially expressed if they exhibited an 129 adjusted p-value below 0.05 and an absolute log₂ fold-change greater than 2; in experiments with only 130 a single biological replicate (n = 1), the NOISeq package (Sonia Tarazona, 2017) was employed, where 131 genes were classified as differentially expressed when the probability of differential expression 132 exceeded 0.90 and the absolute log₂ fold-change exceeded 1. 133 Although the statistical frameworks of the two methods, differ, with DESeq2 applying a parametric 134 model based on the negative binomial distribution and NOISeq relying on a non-parametric, data135 driven approach, both generate log₂ fold-change values that reflect the magnitude and direction of gene 136 expression changes. These values provide an interpretable metric for comparative analysis and 137 visualisation across all conditions, even though the statistical measures of significance they produce 138 are not directly equivalent. The use of both approaches allowed the inclusion of a greater number of 139 experimental conditions in the study, enabling the analysis of datasets that appeared biologically 140 relevant but lacked sufficient replication to support parametric methods. 141 A summary of the differential expression results, including the log₂ fold-change values for each gene 142 and condition, the adjusted p-value obtained from DESeq2 analyses, and the differential expression 143 probabilities from NOISeq analyses, is provided in Supplementary Table S3. From the resulting lists 144 of upregulated and downregulated genes, particular attention was given to those located inside the 145 T6SS gene cluster. To explore overall transcriptional dynamics, a heatmap was generated using the 146 pheatmap package (v1.0.13) (Kolde, 2010) in R to visualise the expression profiles of both upregulated 147 and repressed T6SS genes, based on their log₂ fold-change values. A more targeted visualization was 148 created using the gggenes package (Wilkins, 2017) in R to construct a gene map highlighting the 149 specific genes within the cluster that were transcriptionally activated, along with the experimental 150 conditions in which this activation occurred. 151 2.4 Functional enrichment analysis 152 To identify the biological processes associated with transcriptional activation, a functional enrichment 153 analysis was performed on the lists of up-regulated genes from every experimental condition. This was 154 through the implementation of the plugin ClueGO (v2.5.10) (Bindea et al., 2009) under the Cytoscape 155 application environment (v3.10.3) (Shannon et al., 2003). The annotation source used was the Gene 156 Ontology (GO) terms of Biological Process. In order to capture specific yet interpretable functional 157 terms, the analysis was restricted to GO terms within levels 8 to 13 of the ontology hierarchy. A 158 significance threshold of p < 0.05 (after Benjamini-Hochberg correction) was applied to identify 159 enriched terms. Related GO categories were grouped into functionally organized networks, enabling 160 the visualization and interpretation of biological processes that are specifically associated with 161 transcriptional activation across conditions. 162 2.5 Database construction and data management 163 Finally, to ensure structured data integration and facilitate access to all results generated in this study, 164 a relational database was developed using Microsoft SQL Server (Microsoft SQL Server, 2024). The 165 database stores comprehensive information including sample metadata, study identifiers, experimental 166 conditions, differential gene expression results and gene feature annotations. A relational schema was 167 Transcriptomic Analysis of Salmonella T6SS 5 designed to link these datasets efficiently, enabling cross-referencing between samples, comparisons, 168 and gene-level outcomes. Additionally, the database supports targeted queries, such as retrieving the 169 expression profiles and annotations of specific gene clusters. Users can perform focused searches for 170 genes within the T6SS cluster, allowing streamlined access to their expression data and enrichment 171 context across all analysed conditions. 172 3 Results and Discussions 173 3.1 Exploratory analysis of publicly available Salmonella enterica transcriptomic datasets 174 In order to analyse T6SS transcriptional patterns of the T6SS under varied environmental and host175 associated conditions, a general analytical procedure was proposed. This workflow, described in Figure 176 1, integrates data set selection, quality control, genome alignment, differential gene expression and 177 functional enrichment analysis. This ensures homogeneous and reproducible data handling on all data 178 and facilitates valid comparison among accessible RNA-seq datasets in multiple biological contexts. 179 As a first step, an exploratory analysis of RNA-seq metadata in NCBI SRA was performed to assess 180 the distribution of Salmonella enterica serovars and strains in publicly available datasets. As Figure 181 2A illustrates, S. Typhimurium is the best-represented serovar, exceeding 2,000 samples, well ahead 182 of other serovars such as Enteritidis, Newport, or Typhi. Focusing on the Typhimurium subset, strain183 level analysis revealed that SL1344 and 14028S are among the most sequenced strains, followed by 184 LT2 and other variants. Figure 2B shows the distribution of these strains, where 14028S represents 185 approximately 16% of all Typhimurium submissions. This strain was selected as the reference for this 186 study due to its high representation in public databases, its widespread use in infection models, and its 187 relevance to the ongoing research conducted in our group. In addition, Tables S1 and S2 188 (Supplementary Material) summarise the main experimental conditions associated with strains 14028S 189 and 14028, respectively. The distribution highlights that 14028S is linked with a broader range and 190 greater number of conditions, supporting its suitability for comparative transcriptome analysis. 191 Note, however, that a notable percentage of entries lacked specific annotation, appearing as 192 "undefined" in metadata. These non-curated entries reflect inconsistent or missing sample annotation 193 processes in submission to SRA. This lack of data standardization reduces usability in a significant 194 percentage of accessible data and highlights the importance of thorough curation in large-scale studies. 195 3.2 Transcriptional landscape of the T6SS across diverse experimental contexts 196 To find out at what levels T6SS responds transcriptionally to selected environmental and host-related 197 conditions, we conducted a comparative analysis based on the expression patterns of the T6SS-related 198 genes encoded in the SPI-6 in S. Typhimurium 14028S. The differential expression values of the T6SS199 related genes were visualized as a heatmap (Figure 3A), while a principal component analysis (PCA) 200 (Figure 3B) tracked similarity between the expression patterns of entire conditions. Both 201 representations were constructed using log2 fold-change values as a common metric across conditions 202 to highlight transcriptional variation. 203 The heatmap demonstrates a broad spectrum of regulatory behaviours, ranging from coordinated 204 upregulation to general repression, depending on the experimental conditions. A similar expression 205 behaviour has been observed among subsets of conditions that share analogous characteristics. In 206 particular, pH 3, MOPS EZ Rich Defined Medium (MOPS), and Mg at pH 5.8 and 7.7 form a tightly 207 group where genes within the first operon tend to be downregulated. These genes correspond to 208 baseplate components, as shown in Figure 4A, which was adapted from Li et al. (2022). In contrast, 209 Transcriptomic Analysis of Salmonella T6SS 6 the second operon demonstrates a substantial increase in expression under these conditions. This 210 opposite regulation provides a model in which the processed of assembly, disassembly and secretion 211 component priming occur even in the absence of fully formed membrane complex. These data support 212 previous observations, where stress response modulation of T6SS components under to environmental 213 challenge was reported (Kurasz et al., 2018). Moreover, the presence of divergent regulation between 214 operons leads further support the hypothesis that, under specific conditions such as pH and nutrient 215 availability, the system may be selectively fine-tuned. This finding is also consistent with the 216 established repressive effect of H-NS on virulence gene expression in neutral conditions, where acid 217 pH provides a depressive cue (Brunet et al., 2015). Furthermore, analogous examples can be found in 218 the context of SPI-2-encoded T3SS virulence determinants, where upregulation occurs under mild 219 acidic conditions (Colgan et al., 2016). Exposure to lower pH (e.g. pH 3) has been observed during the 220 phases of infection and intracellular adaptation (Ren et al., 2015). 221 A second group of conditions, including exposure to lettuce exudates, autoclaved diluvial sand (DS) 222 soil, oxidative stress, and starvation at high Colony-Forming Units (CFU), exhibits a more 223 heterogeneous transcriptional profile. The first T6SS operon exhibits an irregular expression under 224 these conditions, indicating a strong possibility of context-dependent or competing signal regulations. 225 Conversely, specific regions within the second operon show more pronounced patterns of activation. 226 In particular, those regions spanning tssB to STM14_0323, coding for core sheath and tube subunits 227 (Figure 4A) (Silverman et al., 2012), along with the cluster downstream from PAAR to STM14_0343, 228 become upregulated in several of these conditions. These regions contain T6SS-specific structural 229 deployment and effector delivery functions. This transcriptional pattern may indicate a regulatory 230 maintenance state in which partial activation of this secretion apparatus is sustained during periods of 231 environmental persistence or stress adaptation. Such a strategy has been proposed for bacteria facing 232 sub-lethal stress, enabling a “primed” state of secretion readiness without full activation (Hespanhol et 233 al., 2023). This finding is also consistent with the general inference that T6SS expression is likely to 234 vary in environmental contacts, in biofilm formation, or in plant surface colonisation, such as 235 demonstrated in other non-host contexts (Silverman et al., 2012). This activation of structural modules 236 in these conditions may thus be indicative of this poised state leading to an enhancement in 237 competitiveness or survival in reactive or nutrient-poor environments (Hespanhol et al., 2023). 238 The remaining conditions, including exposure to HeLa cells, novobiocin, MgCl₂, cold or heat shock, 239 have more dispersed transcriptional profiles, without a unified regulatory signature. Despite the 240 manifestation of mild activation or repression in certain genes, a uniform operon-wide pattern remains 241 absent. Among all tested contexts, co-culture with Acanthamoeba castellanii and L-arabinose exposure 242 were identified clearly as the strongest T6SS gene expression inducers, leading to widespread 243 upregulation in multiple functional modules. Both conditions will be described in detail in later 244 sections. 245 Notably, the gene tssA, which codes for a conserved cap protein that initiates and regulates sheath 246 polymerisation (Hespanhol et al., 2023), is predominantly repressed across various conditions. 247 However, in the group previously mentioned (pH 3, MOPS, Mg conditions), it is observed to be 248 overexpressed. Conversely, tssB, which is responsible for assembling sheath in coordination with tssC 249 (Cascales & Cambillau, 2012), is among the strongest, consistently upregulated genes, specifically in 250 those conditions where tssA is repressed. This apparent decoupling in expression may be indicative of 251 independent regulatory controls, where tssA and tssBC play differential roles in initiating versus 252 elongating secretion sheath (Silverman et al., 2012). As shown in Figure 4A, these proteins possess 253 specific structural positions within the T6SS apparatus. In this configuration, tssA constructs the distal 254 sheath tip to establish the cap, whereas tssB and tssC become integral to the contractile sheath itself. 255 Transcriptomic Analysis of Salmonella T6SS 7 As illustrated in Figure 4B, clpV, which codes for the ATPase responsible for sheath disassembly 256 (Cascales & Cambillau, 2012) and is likely a promoter region for the second T6SS operon, exhibits 257 only weak upregulation in most conditions. In consideration of the significant central role it plays in 258 T6SS apparatus recycling and system reset to accommodate multiple firing events (Wang et al., 2019), 259 a substantially greater induction would have been anticipated. Nevertheless, due to statistical 260 limitations, in particular in those datasets having few, if any, biological replicates, differential 261 expression detection could be compromised in those contexts. 262 3.3 T6SS module activation and associated genome-wide functional responses 263 Environmental signals not only regulate the transcriptional profile of the T6SS locus (Wang et al., 264 2019) but also impact broader cellular processes, as shown in patterns of enrichment in conditions 265 where subsets of specific genes become turned on (Liu et al., 2013). In the group including low pH, 266 MOPS, and magnesium-supplemented conditions, induction of one subset of T6SS structural genes 267 correlates with a characteristic upregulation in RNA biosynthesis, nucleotide metabolism, and arginine 268 biosynthesis pathways (see Supplementary Figure S1, panels E, J, K and L). This finding indicates an 269 apparent coordinated transcriptional response, encompassing both secretion-related and metabolic 270 components, which are presumably involved in supporting apparatus assembly or control. However, a 271 surprising finding was that regulatory subunits such as tssA or vgrG became strongly uninduced, 272 suggesting an incomplete, potentially primed, but in large part, immobilised activation state. Such 273 decoupled regulation has been previously reported under nutrient limitation (Hespanhol et al., 2023) 274 or in low-pH stress (Ren et al., 2015), where post-transcriptional controls or hierarchical gene 275 activation may delay or restrict full system deployment (Colgan et al., 2016). 276 In the second group of conditions related to environmental persistence (lettuce exudates, autoclaved 277 DS soil, oxidative stress, starvation at elevated CFU) the transcriptional signature demonstrates a shift 278 towards metabolic reprogramming, specifically involving amino acid synthesis and catabolism 279 processes. In the most consistently enriched biological processes, alpha-amino acid metabolic process, 280 glutamine and histidine metabolism, branchedand sulfur amino acid metabolism, and carboxylic acid 281 synthesis rank among those strongly identified with nitrogen and carbon homeostasis (see 282 Supplementary Figure S1, panels A, B, D, I and N). This finding indicates that under conditions of 283 environmental pressure, such as nutrient limitation or oxidative lesions, Salmonella may modify its 284 metabolic processes to support cellular functions and enhance its competitive capacity (Schierstaedt et 285 al., 2020). Although T6SS core components such as tssB, tssC, and hcp1 show transcriptional 286 activation in these contexts (Sana et al., 2016), regulatory or structural genes like tssA remain not 287 induced, indicative of an incomplete or condition-dependent activation module. The implication of 288 oxidative or nutrient-linked cues is consistent with reports suggesting T6SS might be differently 289 controlled in soil-like or stress-prone conditions (Jechalke et al., 2019), potentially favouring 290 persistence or competitive fitness over direct antagonism. These broader metabolic corrections may 291 thus represent preparatory or supportive processes that enable T6SS activity only under tightly 292 controlled parameters or in response to additional stimuli. 293 3.4 Environmental competition triggers broad T6SS activation in co-culture with 294 Acanthamoeba castellanii 295 Across all circumstances examined, the transcriptome response of Salmonella Typhimurium to co296 cultivation with Acanthamoeba castellanii showed the strongest activation of the SPI-6-encoded T6SS 297 gene cluster. In addition to a block of downstream genes within the second operon, the expression of 298 important structural elements such as hcp1, vgrG, and PAAR was found to be significantly upregulated, 299 Transcriptomic Analysis of Salmonella T6SS 8 as represented in the heatmap (Figure 3A) and gene map (Figure 4B). In response to stress associated 300 with protists, this pattern suggests a coordinated transcriptional programme that may be directed 301 towards the assembly and deployment of the T6SS. Contact with predatory protozoa such as A. 302 castellanii has been shown to mimic natural niches where bacteria must protect themselves from 303 engulfment, and this activation most likely reflects an adaptation to a competitive biotic environment 304 (Balkin et al., 2021). Salmonella employs the T6SS to mitigate amoebal predation and enhance survival 305 in phagocytic environments (Bao et al., 2020), as indicated by the original study associated to this 306 dataset. These interactions bear a resemblance to those observed in real soil and aquatic environments, 307 where bacteria encounter protozoa as a key selective force (Balkin et al., 2021). 308 This condition also exhibits increased expression of genes associated with the T3SS in addition to the 309 upregulation of T6SS-associated genes (Balkin et al., 2021). This suggests the possibility of a shared 310 transcriptional activation mechanism in both systems, potentially as part of a broader anti-predator or 311 virulence-related response (Colgan et al., 2016). This finding is consistent with research showing that 312 when environmental or host-derived dangers are recognised, T3SS and T6SS, despite their functional 313 differences, may act in parallel or be controlled under overlapping stimuli (Hespanhol et al., 2023). 314 The extensive and coordinated activation of structural genes supports the hypothesis that S. 315 Typhimurium primes its T6SS machinery under competing biotic stress (Silverman et al., 2012), even 316 though transcriptome data alone cannot validate secretion or effector delivery. According to Basler et 317 al. (2013), this finding aligns with the T6SS's ecologically well-defined role in interspecies competition 318 and defence against eukaryotic predation. Collectively, these results highlight the relevance of protist 319 interactions in shaping bacterial adaptative responses, thereby reinforcing the T6SS's role as a crucial 320 effector system in dynamic microbial environments (Kurasz et al., 2018). 321 3.5 Metabolic input through L-arabinose selectively activates central T6SS components 322 As demonstrated in Figure 4B and illustrated in the heatmap (Figure 3A), the transcriptional profile 323 under L-arabinose supplementation indicates a preferential activation of the first operon of the T6SS 324 gene cluster, which includes the tssE, tssF, tssG, and tag genes. These genes encode critical elements 325 of the structural scaffold, including the cap (tssA), wedge components (tssE-F-G), and associated 326 accessory proteins, which are essential for the proper initiation and assembly of the secretion system 327 (Silverman et al., 2012). Notably, genes from the second operon, including tssB, tssC and to a lesser 328 extent clpV, which are involved in sheath elongation and disassembly, also display transcriptional 329 induction under this condition. These results indicate that L-arabinose triggers a selective 330 transcriptional response within the T6SS gene cluster, with induction of most genes from the first 331 operon (except tssA) and a subset of genes from the second, suggesting an incomplete or condition332 specific activation rather that a global upregulation of the entire system. Notably, regulatory 333 information for the first operon is limited in the existing literature, in contrast to the more extensively 334 characterised control of downstream genes, which are frequently modulated by specific stressors or 335 quorum-sensing cues (Journet & Cascales, 2016). 336 As indicated by this dissociated activation pattern, early structural components may be selectively 337 primed in response to L-arabinose, potentially reflecting a preparatory state rather than full system 338 activation. L-arabinose is a pentose sugar derived from plant cell wall polysaccharides, which becomes 339 available in the gastrointestinal tract during dietary fibre degradation. In Escherichia coli, this sugar 340 has been demonstrated to function as a favoured carbon source in vivo, thereby facilitating efficient 341 colonisation and persistence within the murine intestine (Fabich et al., 2008). It is possible that S. 342 enterica may also utilise L-arabinose in analogous host-associated or plant-related environments. In 343 this context, the presence of L-arabinose could act as a metabolic signal that exerts influence over the 344 Transcriptomic Analysis of Salmonella T6SS 9 transcriptional landscape, thereby potentially contributing to the partial induction of T6SS operons. 345 This hypothesis is supported by the enrichment analysis (Figure 5B), which reveals upregulation of 346 pathways related to carbon metabolism and amino acid biosynthesis, consistent with a general 347 metabolic shift towards adaptation to nutrient-rich or host-like conditions. 348 3.6 Host-like conditions induce partial T6SS activation during HeLa cell infection 349 S. Typhimurium 14028S showed minimal transcriptional activation of the T6SS under conditions that 350 approximated gut epithelium environment. As illustrated by the heatmap (Figure 3A), during HeLa 351 cell infection, the expression levels of the majority of T6SS genes remained close to baseline, with no 352 extensive induction across the operons. While significant structural and regulatory elements such as 353 tssA, hcp, and vgrG were essentially insensitive, only a limited number of loci, including tssB and tae4, 354 exhibited weak or irregular upregulation. This is in stark contrast to the prevailing predictions 355 concerning the function of T6SS, which is frequently associated with bacterial adaptation and 356 persistence in intracellular or competitive niches (Bao et al., 2020). The muted response under these 357 conditions may also point to an active repression mechanism, preventing unnecessary expenditure of 358 cellular resources in environments where the T6SS does not confer a selective advantage. 359 When evaluated in the context of the broader transcriptional landscape, this absence of T6SS 360 expression is remarkable. A demonstrated in Figure 5C, the application of functional enrichment 361 analysis yielded a substantial augmentation in the prevalence of secretion-related pathways, 362 particularly those associated with the T3SS. In the initial transcriptome investigation, S. Typhimurium 363 inside HeLa cells exhibited significant activation of SPI-1 and SPI-2 regulons, but not of the SPI-6364 encoded T6SS cluster (Cohen et al., 2022). The discrepancy between the expected and observed T6SS 365 activation may reflect a tightly regulated, condition-specific deployment of this system. In contrast to 366 T3SS, which is necessary for vacuolar survival and host invasion (Bao et al., 2020), T6SS may require 367 additional environmental or interbacterial cues, absent in the HeLa intracellular niche, to initiate its 368 production (Mulder et al., 2012). Overall, these results suggest that while host-like conditions do 369 promote the activation of general secretion pathways, they do not suffice to trigger T6SS expression 370 in S. Typhimurium, highlighting the system’s specialised and context-dependent regulation. 371 4 Conclusions 372 This study conducted a systematic comparative transcriptomic analysis of the T6SS in Salmonella 373 enterica subsp. enterica serovar Typhimurium strain 14028S by leveraging publicly available RNA374 seq datasets. Through the implementation of a reproducible and customisable bioinformatics pipeline, 375 coupled with a relational database architecture for results integration and exploration, transcriptional 376 activity across the SPI-6-encoded T6SS gene cluster was profiled under 23 distinct experimental 377 conditions reflecting a spectrum of ecological, stress-related, and host-associated environments. The 378 analysis revealed that T6SS gene expression is tightly modulated in a condition-specific manner, with 379 differential activation patterns affecting discrete genes and subregions of the cluster. Notably, the co380 culture with Acanthamoeba castellanii and exposure to L-arabinose emerged as strong inducers of 381 transcriptional activity within the system, particularly affecting structural and assembly-associated 382 components. Furthermore, transcriptional dissociation between the two operons and between core 383 structural modules, such as the baseplate and the contractile sheath, suggests a modular architecture in 384 regulatory control, potentially linked to distinct environmental triggers or metabolic cues. These 385 findings expand our understanding of the regulatory dynamics governing T6SS expression in S. 386 Typhimurium and underscore its probable role in environmental sensing, niche colonisation, and 387 interspecies interactions. This regulatory complexity highlights the T6SS not only as a static virulence 388 Transcriptomic Analysis of Salmonella T6SS 16 including various amino acid biosynthetic processes, inorganic anion transport, and pathways related 619 to protein secretion, notably the Type III Secretion System. Statistical significance is indicated by 620 p < 0.05 (*) and p < 0.01 (**). 621 Table Headers. 622 Table 1. Summary of experimental conditions and publicly available RNA-seq datasets used in this 623 study. The table lists BioProject identifiers, associated study conditions, and corresponding SRA 624 accession numbers for Salmonella enterica subsp. enterica serovar Typhimurium strain 14028S. 625 Table 2. Summary of STAR mapping statistics across all analysed BioProjects. For each project, the 626 table shows the number of RNA-seq samples, the average number of input reads, and the average 627 percentage of reads that mapped uniquely, to multiple loci, or remained unmapped. These values 628 provide an overview of mapping quality and read distribution and were extracted from the STAR log 629 output files. 630