The Role of Galectin-3 in the Regulation of the Immune-Brain Response in Alzheimer's Disease and Frontotemporal Dementia

THE ROLE OF GALECTIN-3 IN THE REGULATION OF THE IMMUNE-BRAIN RESPONSE IN ALZHEIMER'S DISEASE AND FRONTOTEMPORAL DEMENTIA Jesús Soldán Hidalgo Departamento de Bioquímica y Biología Molecular Facultad de Farmacia. Universidad de Sevilla Instituto de Biomedicina de Sevilla Tesis doctoral, 2025 THE ROLE OF GALECTIN-3 IN THE REGULATION OF THE IMMUNEBRAIN RESPONSE IN ALZHEIMER'S DISEASE AND FRONTOTEMPORAL DEMENTIA Directores José Luis Venero Recio Francisco Javier Vitorica Ferrández Memoria presentada por Jesús Soldán Hidalgo para optar al grado de Doctor por la Universidad de Sevilla Dissertation submitted by Jesús Soldán Hidalgo in fulfilment of the requirements for the degree of Doctor at the University of Seville Tesis doctoral financiada por el VI PPIT-US “Como no sabían que era imposible, lo lograron” - María de Villota Agradecimientos Ni yo mismo imaginaba, al comenzar la escritura de este texto, que el apartado más difícil de redactar fuesen los agradecimientos. No es que no disponga de familiares, compañeros, amigos ni allegados que me hayan acompañado en este largo camino de aventuras (y alguna que otra desventura), pues he tenido la enorme suerte conocer y aprender de personas fantásticas a diario. Tampoco se trata de una falta de modales por los cuales me resista a agradecer todo lo compartido y disfrutado en esta etapa tan significativa de mi vida. Créeme cuando digo que lo más difícil de esta tesis reside en saber encontrar las palabras adecuadas para cada una de aquellas personas que han hecho posible que hoy esté redactando estas líneas, pues todo lo que aquí exprese será una mera nimiedad en comparación con lo que ellos me han aportado. A vosotros, papá y mamá, que habéis celebrado cada pequeño éxito de esta tesis y que, en los momentos de tormenta, habéis recogido los restos de mi naufragio para levantar un nuevo barco que llegue más lejos con vuestras enseñanzas, consejos, algún que otro tirón de orejas y el refugio de vuestro abrazo. La expresión “no sería posible sin vosotros” cumple de manera literal vuestra aportación a este trabajo, pues habéis sido mi modelo a seguir desde el primer segundo de mi vida. Es por ello por lo que esta tesis es también vuestra. A mis abuelos, a quienes el capricho de vivir en una época con menos facilidades les ha otorgado una serie de virtudes y cualidades que no se enseñan en la escuela ni se descubren en los laboratorios. El inmenso corazón que atesoran es difícil de encontrar en ningún otro sitio. Gracias por vuestro apoyo incondicional, por vuestras visitas, vuestros ánimos y, cómo no, vuestros muchos almuerzos y dulces que tanto han acompañado al que escribe estas líneas. Pero más importante aún, gracias por darme el privilegio de tener en mi vida a quienes considero mis “segundos padres”. Al mayor descubrimiento, a mi compañera de vida, a Rosa. Gracias por tu cariño incondicional, por escucharme cuando más lo he necesitado, por tus consejos, por ser la calma y alegría en mis días más negativos, por tu capacidad para convertir lo gris en memorable, por tu “originalidad”. Gracias por estar a mi lado en los momentos difíciles, por apoyarme y sujetarme cuando no he podido más. Gracias por ser hogar, por hacer que sobren las palabras, por tu seguridad en mi incertidumbre, por ser salvavidas en el mar de mis miedos, por sacarme las carcajadas cuando más escondidas estaban, por iluminar mis sombras y por hacer que los planes fueran lo de menos y la compañía lo que más. A mis titos y titas Mari, Juan, Almudena y Nacho, y a mis primos, Kike, María, Nachito y Candela. Por ser la mayor fuente de eventos extravagantes, improvisaciones y merendolas que nadie se pueda imaginar. Gracias por hacer que las reuniones familiares siempre estuvieran cargadas de anécdotas. Gracias también por hacer que ni el tiempo ni la FTD: Frontotemporal Dementia FTLD: Frontotemporal Lobal Degeneration FUS: Fused in sarcoma GS: Gene Set HAM: Human AD Microglia HIF-1: Hypoxia-Inducible Factor 1 Gal1: Galectin-1 Gal3: Galectin-3 GAM: Glioma-Associated Microglia GCase: β-glucocerebrosidase GPNMB: Glycoprotein Non-metastatic Melanoma protein B GRN: Granulin GS: Gene set GSEA: Gene Set Enrichment Analysis HGF: Hepatocyte Growth Factor IGF1: Insulin growth factor-1 IL-1β: Interleukin-1β IL-2: Interleukin-2 IL-4: Interleukin-4 IL-5: Interleukin-5 IL-6: Interleukin-6 IL-10: Interleukin-10 IL-12p70: Interleukin-12 IL-18: Interleukin-18 IRM: Interferon-Responsive Microglia KC/GRO: Keratinocyte Chemoattractant/Human Growth Regulated Oncogene KEGG: Kyoto Encyclopedia of Genes and Genomes LAMP1: Lysosome-Associated Membrane Protein 1 LDAM: Lipid-Droplet-Accumulating Microglia LYZ: Lysozime 2 MAPT: Microtubule-associated protein tau MSigDB: Molecular Signature Database MFG-E8: Milk Fat Globule–Epidermal Growth Factor 8 MGnD: Microglia Neurodegenerative MIM: Microglia Inflamed in Multiple Sclerosis MRI: Magnetic Resonance Imaging Ndufa1: NADH:Ubiquinone Oxidoreductase Subunit A1 Ndufa2: NADH:Ubiquinone Oxidoreductase Subunit A2 NGF: Nerve Growth Factor PAMP: Pathogen-Associated Molecular Pattern PCA: Principal Component Analysis PCR: Polymerase Chain Reaction PD: Parkinson's Disease PDGF: Platalet-Derived Growth Factor PET: Positron Emission Tomography PRR: Pattern Recognition Receptor PSEN1: Presenilin 1 PSEN2: Presenilin 2 PSP: Progressive Supranuclear Palsy ROS: Reactive Oxygen Species RNS: Reactive Nitrogen Species snRNP: Small Nuclear Ribonucleoproteins TCR: T-Cell Receptor TDP-43: TAR DNA-Binding Portein 43 ThS: Thioflavin-S TLR: Toll-Like Receptor TNFα: Tumor Necrosis Factor-Alpha WAM: White matter-Associated Microglia Table of Content Abstract ................................................................................................................................ 1 Introduction ......................................................................................................................... 5 1. Alzheimer’s Disease – “Sitting on the bed, eyes filled with anguish” ..................... 7 a. Historical Perspective ........................................................................................... 7 b. Histopatology ........................................................................................................ 7 c. Symptoms ........................................................................................................... 12 d. Diagnostic ............................................................................................................ 12 e. Epidemiology ...................................................................................................... 14 f. Risk factors .......................................................................................................... 14 g. Treatment ........................................................................................................... 14 2. Frontotemporal Dementia – “So the beginning began” ....................................... 15 a. Historical Perspective ......................................................................................... 15 b. Histopathology .................................................................................................... 16 c. Symtomps ........................................................................................................... 19 d. Diagnosis ............................................................................................................. 20 e. Epidemiology ...................................................................................................... 20 f. Treatment ........................................................................................................... 21 g. Genetic Causes and Risk Factors ........................................................................ 21 3. Microglia – Beyond the Neuron ............................................................................ 22 a. Historical perspective ......................................................................................... 22 b. Origin ................................................................................................................... 22 c. Homeostatic functions ........................................................................................ 24 d. Neuroinflammation ............................................................................................ 26 e. Microglial Heterogeneity .................................................................................... 27 f. Microglia in Alzheimer's Disease ........................................................................ 29 g. Microglia in Frontotemporal Dementia.............................................................. 30 4. Galectin-3 – When Science and Mythology Unite ................................................ 31 a. The Galectin Family ............................................................................................. 31 b. Functions of Galectin-3 ....................................................................................... 32 c. Galectin-3 and Microglia ..................................................................................... 33 d. Galectin-3 and Neurodegenerative Diseases ..................................................... 36 Objectives .......................................................................................................................... 39 Materials and methods ...................................................................................................... 43 1. Animal models ....................................................................................................... 45 2. Genotyping ............................................................................................................ 46 3. Immunofluorescence ............................................................................................. 48 4. Immunohistochemistry ......................................................................................... 49 5. Cell Separation ....................................................................................................... 50 6. RNASeq .................................................................................................................. 50 7. RNA Extraction ....................................................................................................... 51 8. Expression Array .................................................................................................... 51 9. Protein Analysis...................................................................................................... 52 10. Quantification of Inflammatory Cytokines ............................................................ 52 11. Proteomic Analysis ................................................................................................. 53 12. Behavioural Tests ................................................................................................... 55 13. Image Analysis ....................................................................................................... 55 14. Statistical Software ................................................................................................ 56 Results ................................................................................................................................ 57 CHAPTER 1: Role of Galectin-3 in Alzheimer's Disease ................................................. 59 1. Galectin-3 is involved in the morphology and size of Aβ plaques in APP model 59 2. The Presence of Galectin-3 Determines the Microglial Activation Profile ........ 60 3. Galectin-3 Modifies the Microglial Expression Profile ....................................... 65 4. The Toxicity of Aβ Plaques Decreases in the Absence of Galectin-3 ................. 68 CHAPTER 2: The Role of Galectin-3 in Frontotemporal Dementia ................................ 69 1. The Absence of Progranulin Leads to Increased Galectin-3 Levels in Various Brain Regions .............................................................................................................. 69 2. The Absence of Progranulin Induces Microgliosis in Myelinated Brain Regions 72 3. PGRNKO Animals Exhibit Increased Levels of Various MGnD Microglia Markers 74 4. The Exacerbated Immune Response in PGRNKO Animals Is Modulated by the Absence of Galectin-3 ................................................................................................ 78 5. The Absence of Progranulin Induces Numerous Transcriptomic Changes in the Thalamus .................................................................................................................... 79 6. PGRNKO Animals Exhibit Lysosomal Dysfunction .............................................. 94 7. Progranulin is Essential for Lysosomal Integrity ................................................. 97 8. Proteomic Profile Alterations in White Matter Due to the Absence of Progranulin and Galectin-3 ........................................................................................ 97 9. The absence of Gal3 induces phagocytosis in microglia .................................. 103 10. PGRNKO Animals Exhibit Exacerbated Neuronal Death ............................... 105 11. The Absence of Gal3 Reduces Lipofuscinosis in PGRNKO Animals .............. 106 CHAPTER 3: Modification of Tau Pathology by the Absence of Progranulin .............. 108 1. The Absence of Progranulin Modifies Tau Pathology in the P301S Model ..... 108 2. PGRNKO/P301S Animals Exhibit Reduced Microgliosis in the Hippocampus .. 112 3. Effect of Progranulin Deficiency on the Behaviour of the P301S Model ......... 114 Discussion ........................................................................................................................ 117 1. Galectin-3 Modifies the Morphological Characteristics of Aβ Plaques .............. 119 2. Changes in the Microglial Expression Profile in the Absence of Galectin-3 ....... 120 3. Galectin-3 Determines the Microglial Activation Profile in the APP Model ....... 120 4. The Absence of Gal3 Contributes to the Reduction of Dystrophic Neurites ...... 122 5. Microglial Phenotype in the PGRNKO Model ...................................................... 123 6. The absence of progranulin reduces tauopathy in P301S animals. .................... 125 7. Alterations of the brain immune response due to the absence of Galectin-3 ... 127 8. Dysregulation of energy metabolism and its relationship with progranulin and Galectin-3 ..................................................................................................................... 128 9. Alterations of the Cell Cycle in Neurodegeneration Conditions ......................... 128 10. Changes in Lipid Metabolism and Its Relationship with Galectin-3 .................... 129 11. The Absence of Galectin-3 Reduces Lipofuscinosis in the PGRNKO Model ....... 129 Conclusion ........................................................................................................................ 133 Bibliography ..................................................................................................................... 137 Abstract ABSTRACT 3 Abstract Abstract Neurodegenerative pathologies such as Alzheimer's disease and frontotemporal dementia are two neurodegenerative processes that affect millions of people worldwide, generally in old age. Despite being discovered over a century ago, these pathologies lack not only an effective treatment capable of reversing the disease or significantly improving the quality of life of patients and, by extension, their families, but also a clear understanding of their origin, which remains an unresolved challenge for the scientific community. Both Alzheimer's disease and frontotemporal dementia are characterized by a marked presence of microglial cells in specific brain regions. This microglia exhibits a characteristic activation phenotype, in which galectin-3 (Gal3) plays a prominent role. Previous studies from our group have described how Gal3 acts as a ligand for TREM2, which is necessary to induce a shift in the microglial phenotype from homeostatic to reactive microglia. Furthermore, Gal3 has been proposed as a potential biomarker in some neurodegenerative diseases. In the present study, we investigated how the microglial response can modify some of the most relevant histopathological features in an animal model of Alzheimer's disease characterized by APP overexpression. Specifically, we evaluated how the absence of Gal3 can induce changes in the morphology and toxicity of β-amyloid peptide plaques, one of the key pathological hallmarks of the disease. We observed that the absence of Gal3 leads to a modification in the microglial activation profile, with a reduction in the expression of certain characteristic markers. In this context, we highlight that the absence of Gal3 significantly attenuated the type I interferon signaling pathway, which is closely linked to disease progression and plays a key role in amyloid pathology. Additionally, our study demonstrates that the absence of Gal3 induced an increase of TREM2 in plaqueassociated microglia, one of the most relevant microglial proteins in the context of Alzheimer's disease. Regarding frontotemporal dementia, this study employed a knockout animal model for the progranulin gene (PGRNKO). Our findings point to the thalamus and the thalamocortical tract as two of the most affected regions in this animal model. We describe how the absence of progranulin leads to the overexpression of genes associated with reactive microglia, with Gal3 being one of the most prominent alongside GPNMB. It has been reported that PGRNKO animals exhibit marked lysosomal dysfunction, positioning progranulin as an essential element for proper lysosomal homeostasis. Additionally, PGRNKO animals present severe neuronal ceroid lipofuscinosis, with the thalamus being one of the most affected areas. The presence of lipofuscin in these animals is not limited to neurons; rather, microglial cells, through the phagocytosis of lipofuscin-positive neurons, accumulate this lipid-rich substance in their lysosomes. Our transcriptomic analysis of the thalamic region reveals a marked enrichment of pathways 10 INTRODUCTION effect of mutations in presenilin 1 gene (PSEN1) is qualitative rather than quantitative, as proposed by the amyloidogenic hypothesis (De Strooper & Karran, 2016). Another key player in AD is tau protein. The cytoskeleton of cells is composed of microtubules, which maintain cell structure and serve as intracellular transport pathways. The proper assembly of microtubules is mediated by tau protein, which binds to their surface via its C-terminal domain. Phosphorylation of tau, mediated by kinases, leads to its detachment from microtubules. However, hyperphosphorylation can cause tau selfaggregation, forming neurofibrillary tangles inside neurons. This phenomenon results in reduced tau binding to microtubules, weakening their structure and leading to loss of function, which may trigger cell death. At a larger scale, neuronal loss can contribute to brain atrophy, a hallmark of AD (Ashrafian et al., 2021). Illustration I-3: Neurofibrillary tangle formation. First, one phosphate group is transferred to microtubule attached tau by tyrosin. Secondly, phospho-tau leave the microtubule. Thirdly, phospho-tau aggregates forming neurofibrillary tangles. Finally, the lack of tau leads to microtubule breakage. Extracted from Ashrafian (Ashrafian et al., 2021). The distribution of tau deposits and Aβ plaques in AD patients was described by Braak and Braak in 1991 (Braak & Braak, 1991). They concluded that the emergence of both histopathological hallmarks of the disease occurs before the onset of clinical symptoms. Regarding amyloid deposits, their appearance represents one of the earliest changes in the brains of AD patients. These deposits vary in shape, size, and distribution patterns. Despite this variability, Braak and Braak were able to define three stages (A, B, and C) of Aβ plaque formation. In Stage A, Aβ deposits are restricted to the basal region of the isocortex. In Stage B, they expand to all association areas of the isocortex, with a mild involvement of the hippocampus. In Stage C, Aβ plaques can be found throughout the isocortex, including sensory and motor areas, as well as in the striatum, thalamus, and hypothalamus. However, Braak and Braak highlighted that some non-demented 11 INTRODUCTION individuals also exhibited Aβ plaques in the cerebral cortex, suggesting that Aβ deposition does not necessarily precede the development of neurofibrillary changes. Illustration I-4: Distribution of amyloid deposits. Stage A. Amyloid deposits can be found in the basal portions of the isocortex. Stage B. All isocortical association areas present amyloid deposits. Hippocampal areas are midly involved. Stage C. Final stage characterised by the presence of amyloid deposits in all isocortical areas. Shading indicates the presence of amyloid deposits in each region. Extracted from Braak (Braak & Braak, 1991). In contrast to Aβ deposition, neurofibrillary changes, which include neurofibrillary tangles and neuropil threads, exhibit a well-defined distribution with little variation among individuals, allowing for the clear differentiation of six stages. Neuritic plaques, considered neurofibrillary events, have a more irregular distribution and are therefore less relevant in the staging of the pathology. In the early stages (I and II), neurofibrillary changes are restricted to the transentorhinal region and, to a lesser extent, the CA1 region of the hippocampus. In stages III and IV, the previously mentioned regions, along with the entorhinal area, become more severely affected, while the isocortex is only mildly involved. Finally, in stages V and VI, the isocortex becomes profoundly affected. The formation and accumulation of neurofibrillary tangles and neuropil threads ultimately result in neuronal death and the appearance of "ghost tangles" in cortical regions and subcortical nuclei. The loss of neurons in gray matter areas is believed to underlie the functional deficits and clinical symptoms characteristic of AD (Braak & Braak, 1991). 12 INTRODUCTION Illustration I-5: Distribution pattern of neurofibrillary changes (neurofibrillary tangles and neuropil threads) divided in six stages. Stage I – II: Transentorhinal region is the only affected with mild involvement of hippocampus. Stage III – IV: Entorhinal and transentorhinal layer Preα are severe involved. Stage V – VI: Isocortial areas are highly affected. Shading indicates increasing severity of neurofibrillary changes. Extracted from Braak (Braak & Braak, 1991). c. Symptoms AD is the leading cause of dementia worldwide (Scheltens et al., 2021) and is characterized by a gradual and continuous decline in memory and other cognitive abilities. This pathology typically begins with a selective memory deficit (amnesia), often accompanied by difficulty in language use, word recognition, or even mutism (aphasia), reduced performance in daily activities such as cooking, dressing, drawing or driving (apraxia), inability to recognize objects and people (agnosia), and deficits in executive functions including reasoning, judgment, planning, execution, and monitoring of certain activities. Occasionally, patients may develop delusions. Additionally, unlike age-related memory decline in non-pathological aging, AD patients typically lack awareness of their condition (anosognosia) and often fail to take proactive measures to mitigate the risks associated with the disease, such as financial management or personal safety. In advanced stages, patients become bedridden, suffering from severe confusion, incontinence, and an inability to feed themselves independently. The disease generally lasts eight to ten years, with the cause of death being directly related to dementia or indirect causes such as pneumonia or accidents (Reiman & Caselli, 1999). d. Diagnostic The diagnosis of AD has evolved significantly over the past century, from Alois Alzheimer’s exclusively pathological approach focused solely on the dementia stage to the molecular perspective that began in the late 20th century and has extended to the present day 13 INTRODUCTION (Scheltens et al., 2021). Currently, clinical and biochemical aspects converge to establish the diagnosis of AD. According to the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2022), for a patient to be diagnosed with AD, beyond the typical characteristics of mild and major neurocognitive disorders, they must present a progressive and gradual cognitive and behavioural decline. Typically, patients with this disease are characterized by amnesia. However, a significant proportion of patients initially exhibit behavioural symptoms that precede cognitive symptoms. In advanced stages, visuoconstructive/perceptual-motor abilities and language become compromised. Social cognition is usually preserved until the final stages of the disease. The diagnosis of AD specifies the degree of diagnostic certainty through the nomenclature of “probable” and “possible” AD. The diagnosis of probable AD is established in cases where there is evidence of a causative gene in autosomal dominant family history, confirmed either through autopsy or genetic testing of an affected family member. Additionally, recent advances in biomarkers allow for improved and complementary diagnostic certainty of the disease through techniques such as positron emission tomography (PET) and the detection of tau and/or amyloid deposits via imaging techniques or cerebrospinal fluid analysis. Histologically, Alzheimer’s disease can be diagnosed post-mortem by the presence of characteristic features such as neuritic amyloid plaques, tau neurofibrillary tangles, neuronal loss, or cortical atrophy in regions such as the hippocampus, parietal lobe, and frontal cortex. Additionally, other relevant diagnostic biomarkers that do not require postmortem examination include mutations in APP gene, PSEN1, and presenilin 2 (PSEN2), which are characteristic of familial or early-onset Alzheimer’s disease. A recent study has also revealed that the allele ε4 of the apolipoprotein E gene (APOE4) is an important genotype, as its homozygosity represents a genetic form of Alzheimer’s disease (Fortea et al., 2024). Furthermore, the deposition of β-amyloid peptide and tau and phosphorylated tau, fundamental characteristics of Alzheimer’s disease, can be assessed using PET scans and cerebrospinal fluid analysis, showing diagnostic potential. Neuronal damage can also be evaluated through magnetic resonance imaging (MRI) and PET. Currently, biomarker research with diagnostic, prognostic, and theranostic potential is focused on blood-based measurements. The development of biomarkers has led to their classification for disease diagnosis. There are A-type biomarkers (amyloid), T-type biomarkers (phosphorylated tau), and N-type biomarkers (neurodegeneration) (Scheltens et al., 2021). However, the diagnosis of AD is often complicated by high comorbidity. Elderly patients commonly present with other conditions, including cerebrovascular diseases, which impact both diagnosis and disease progression. 14 INTRODUCTION e. Epidemiology Currently, approximately 55 million people worldwide suffer from some form of dementia, with AD accounting for around 60-70% of cases. Since 1990, the prevalence of the disease has increased by 160.84%, and it is estimated that by 2050, 139 million people worldwide will be affected. However, a slight decline in the incidence of this pathology is expected in developed countries by then (Li et al., 2022). It is important to note that Alzheimer’s disease is more common in women than in men, as they represent 63.77% of cases. Additionally, both the prevalence and incidence of the disease are higher in countries with a high sociodemographic index (Li et al., 2022). In countries with lower cultural and socioeconomic levels, the incidence is lower because memory loss in the elderly is often considered normal in these regions, daily cognitive demands are lower, and objective cognitive assessment of patients is challenging due to lower levels of education. f. Risk factors The risk of developing AD can be significantly influenced by both genetic and environmental factors. The most relevant risk factor is age, as well as sex, since the incidence of this pathology is higher in older individuals and women. However, AD presents many other risk factors, such as midlife hypertension, obesity, hearing loss, low educational level, smoking in old age, depression, physical inactivity, social isolation, and diabetes. Additionally, traumatic brain injuries could increase the risk of developing AD, particularly in men. Similarly, the impact of genetic predisposition on the risk of developing AD should be highlighted. Rare mutations in chromosomes 1, 14, and 21, as well as alterations such as trisomy 21, can lead to the development of AD if individuals reach middle age. Moreover, the most common risk factors today are polygenic, with more than 45 genes identified that cause a slight increase in the risk of developing AD. Specifically, the APOE gene has three possible alleles: ε2, ε3, and ε4, with the presence of ε4 being a determining factor in the pathology, while ε2 is considered a protective factor that decreases the risk (Scheltens et al., 2021). It should also be noted that many of these genetic alterations do not have the same impact across all ethnic groups and races. Mutations such as Gly206Ala in the PSEN1 gene lead to early-onset AD in the Puerto Rican population, whereas mutations in the ABCA7 gene, which encodes a transporter protein, are associated with the African American population (American Psychiatric Association, 2022). g. Treatment Treatments for AD have an intrinsic limitation due to the very nature of the pathology, as its clinical manifestation is preceded by a preclinical stage that lasts for years or even decades. In other words, by the time the disease presents its first symptoms, significant 15 INTRODUCTION neurodegeneration has already occurred. This explains why there is still no effective cure for the disease or a treatment that significantly slows its progression in the long term. Currently, AD is treated with three types of drugs: acetylcholinesterase inhibitors, NMDA modulators, and monoclonal antibodies targeting Aβ. Among them, both acetylcholinesterase inhibitors and NMDA modulators such as memantine provide only a palliative effect on the patient’s symptoms, lacking the ability to attenuate or reverse the pathology. Additionally, due to the absence of early diagnosis, treatments are often administered at an inadequate time frame, potentially making them ineffective while still maintaining their side effects. Regarding monoclonal antibodies targeting Aβ, the recent approval of aducanumab (June 2021) and lecanemab (July 2023) represents a step forward in the search for an effective treatment to combat the pathology itself and not just its symptoms. Specifically, lecanemab administration delays cognitive decline, extending the quality of life of patients with mild to moderate AD for approximately six months. However, while the relationship between familial AD and Aβ plaque deposition is well accepted, these treatments may not be equally effective in an ethnically diverse population. Moreover, their long-term effects remain unknown, making further and longer-duration studies necessary (Kim et al., 2024). 2. Frontotemporal Dementia – “So the beginning began” a. Historical Perspective Frontotemporal dementia (FTD) is a term encompassing a broad spectrum of clinical syndromes characterized by progressive changes in behaviour, executive function, or language, resulting from a neurodegenerative process affecting the frontal and/or temporal lobes, as well as one or more proteinopathies. Three distinct syndromes are identified, classifying FTD into three variants: behavioural, non-fluent, and semantic (Boeve et al., 2022; Sociedad Española de Neurología, 2018). The first documented case of FTD was studied by Austrian neuropsychiatrist Arnold Pick in 1892, in an article published under the title "On the relationship between senile cerebral atrophy and apraxia." In this work, Pick evaluated a 71-year-old patient who had been experiencing a gradual decline in mental condition for two years. The patient exhibited alternating aggressive and threatening behaviour with childlike attitude. His wife reported that he had initially suffered from fainting episodes lasting several minutes. Later, he developed a flu-like illness that persisted for three months. His cognitive deterioration began with delusions accompanied by fever; he confused people and showed initial mild speech difficulties that later became more pronounced. Clinical examination revealed severe memory deficits, significant facial weakness on the right side, and marked aphasia. Although the patient was talkative, his speech was notably impaired. At times, he used words in an incorrect order, making comprehension difficult. Additionally, he rearranged syllables and digits in numbers greater than ten, complicating 16 INTRODUCTION communication beyond simple sentences. He also struggled to repeat spoken words, managing to do so only with great effort and at a slow pace. Similarly, his writing and ability to copy written text were also affected. The patient passed away just two weeks after being evaluated (Pick et al., 1994; Spatt, 2003). The post-mortem study posed a challenge for Arnold Pick. The language impairments and absence of motor symptoms were attributed to the previously mentioned strokes. However, he noted that the most affected region was the left temporal lobe. Anatomically, the brain was surrounded by a firm dura mater, with venous sinuses containing coagulated blood. The inner surface of the pia mater, over the convexity of the cerebral hemispheres, was lined with newly formed hard connective tissue. The inner meninges were oedematous and thickened. The walls of the basal vessels exhibited uneven thickness. Additionally, the cerebral gyri were narrow, with pronounced atrophy in the left hemisphere, particularly in the temporal lobe. This hemisphere also weighed 30 grams less than the right. Other notable characteristics of the patient’s brain included enlarged ventricles, the presence of granulation tissue, and the absence of focal pathology or granular cells in the temporal lobe (Pick et al., 1994). In his article, Arnold Pick did not name the pathology he described; it was later designated as Pick’s disease, marking the first documented case of FTD worldwide (Berrios & Girling, 1994). b. Histopathology While the term frontotemporal dementia is reserved for clinical diagnosis, the pathological diagnosis of the disease employs the term frontotemporal lobar degeneration (FTLD). FTLD is characterized by histopathological features such as neuronal loss, gliosis, and microvacuolar changes in the frontal and anterior temporal lobes, anterior cingulate cortex, and insular cortex. As a concept encompassing various disorders, FTD can be classified into different subtypes depending on the abnormal accumulation of proteins in the patient’s brain. Primarily, these include aggregates of the microtubule-associated protein tau (MAPT), the TAR DNA-binding protein with a molecular weight of 43 kDa (TDP-43), and the fused in sarcoma (FUS) protein. The accumulation of these proteins occurs in most cases of frontotemporal lobar degeneration, while cases involving exclusive accumulations of ubiquitin or p62 protein, as well as cases without detectable protein deposits, are uncommon (Bang et al., 2015). FTLD – Tau The tau protein plays a fundamental role in tubulin assembly for microtubule formation, which is essential for maintaining cell morphology, enabling axonal transport, and linking microtubules with other proteins. FTLD with tau protein deposits accounts for approximately 36% to 50% of FTLD cases (Mandelkow & Mandelkow, 2012; Sato-Harada 17 INTRODUCTION et al., 1996). Different subtypes can be distinguished, including Pick’s disease, corticobasal degeneration, and progressive supranuclear palsy. Pick’s disease represents 5% of all dementia cases and 30% of FTLD cases associated with tau, characterized by frontal, temporal, and cingulate gyrus atrophy. Its primary histopathological hallmark is the presence of so-called "Pick bodies" (intracellular inclusions composed of tau filaments) and "Pick cells" (ballooned or swollen cells). This pathology involves significant pyramidal neuron loss in cortical layers II, III, and IV, as well as synaptic loss in the superficial layers of the frontal cortex (Bang et al., 2015; Mann et al., 1993). Corticobasal degeneration accounts for 35% of FTLD cases associated with tau and involves structures such as the dorsal prefrontal cortex, supplementary motor area, the region surrounding the Rolandic fissure, and the subcortical nucleus. It is characterized by early-stage neurofibrillary tangles of hyperphosphorylated tau, particularly in white matter regions, neuritic threads, ballooned neurons (though in smaller quantities than in Pick’s disease), astrocytic plaques, and cytoplasmic inclusions in oligodendrocytes called oligodendrocyte coiled bodies (Bang et al., 2015; Dickson, 1999). Lastly, progressive supranuclear palsy accounts for 31% of FTLD cases with tau accumulation. It is marked by astrocytic aggregates containing small tau inclusions near the cell body, coiled bodies in oligodendrocytes, and neurofibrillary tangles in neurons. The most affected regions include the brainstem, basal ganglia, dentate nucleus, and superior cerebellar peduncle (Bang et al., 2015; Dickson, 1999). FTLD – TDP-43 The TDP-43 protein is typically localized in the nucleus, where it plays essential roles in splicing regulation, nucleic acid stability, and trafficking (Vanden Broeck et al., 2014). However, in FTD, TDP-43 accumulates in the cytoplasm, undergoes hyperphosphorylation, ubiquitination, and cleavage, leading to the formation of cytoplasmic granules. FTLD with TDP-43 accumulation accounts for 50% of cases and is classified into four subtypes (A, B, C and D) based on the distribution of cytoplasmic or intranuclear inclusions (Bang et al., 2015; Mackenzie et al., 2011). Subtype A accounts for 50% of non-fluent FTLD cases, 25% of suspected corticobasal degeneration cases, and a small portion of the behavioral variant. It is the only subtype associated with mutations in the progranulin-coding gene. The relationship between progranulin and frontotemporal dementia remains an area of active research, but it is known that haploinsufficiency of this protein, caused by GRN gene mutations, leads to FTLD with TDP-43 inclusions (subtype A), while homozygous mutations cause neuronal ceroid lipofuscinosis (Baker et al., 2006; Gass et al., 2006). Histopathologically, this subtype is characterized by small intracellular aggregates of TDP-43 in neurons, thin neuropil threads, and lentiform nuclear inclusions in layer II or III cortical neurons. 18 INTRODUCTION Additionally, it is associated with volume reduction in the dorsal frontal and anterior temporal regions (Bang et al., 2015; Perry et al., 2017; Roher et al., 2009; Whitwell et al., 2011). Subtype B accounts for two-thirds of FTD cases linked to motor neuron diseases and 25% of behavioral variant cases. It is characterized by diffuse intracellular granules in neurons, with few neuropil threads in both superficial and deep cortical layers, moderate cortical atrophy, and pronounced atrophy in the basal ganglia (Bang et al., 2015; Perry et al., 2017). Subtype C represents 90% of semantic variant FTLD and behavioral variant cases (Bang et al., 2015). It is distinguished by long, tortuous dystrophic neurites in the superficial cortical layers, along with atrophy in the anterior temporal region (Younes & Miller, 2020). Some authors describe a fourth subtype, D, which is associated with lentiform intranuclear inclusions in neurons, short neuritic dystrophies, and intracellular inclusions across all cortical layers (Younes & Miller, 2020). FTLD – FUS The accumulation of FUS protein occurs intracellularly and accounts for only 10% of FTLD cases. It is characterized by the presence of abundant FUS protein inclusions, particularly in the dentate gyrus, as well as severe atrophy in the striatal nuclei (Bang et al., 2015). Illustration I-6: Neuropathology in FTLD-tau and FTLD-TDP. A. Pick bodies in Pick’s disease. B. Astrocytes in progressive supranuclear palsy. C. Astrocites in FTLD-TDP. D. Vermiform neuronal inclusion in the dentate gyrus of a granule cell. E. Small compact neuronal cytoplasmic inclusions in FTLD-TDP. F. Diffuse or granular cytoplasmic inclusions in neurons in FTLD-TDP. G. Long distrophic neurites in TFLD-TDP. H. Small juxtanuclear ubiquitin-positive, TDP-negative inclusions. Extracted from Bang (Bang et al., 2015). 19 INTRODUCTION c. Symtomps FTD is the third most common cause of dementia worldwide, following AD and Lewy body dementia. It encompasses a group of syndromes characterized by neurodegeneration resulting from localized atrophy of the frontal and temporal lobes. Therefore, depending on the predominant symptoms, different variants can be distinguished. i. Behavioural Variant Frontotemporal Dementia Behavioural variant frontotemporal dementia (bvFTD) is the most common form of FTD, accounting for 60–70% of FTLD cases. Phenotypically, bvFTD is primarily characterized by behavioral and personality changes, with affected individuals displaying varying degrees of apathy or disinhibition. Patients often exhibit impulsivity, stereotyped behaviors, lack of empathy, impaired social conduct, hyperorality, and changes in eating habits. Executive functions are impaired in the early stages, while visuoperceptual abilities and memory remain relatively preserved initially. Language disturbances also occur, progressing to mutism in advanced stages (Sociedad Española de Neurología, 2018). ii. Primary Progressive Aphasia Primary progressive aphasia is characterized by a predominant language impairment in the early stages of the disease, while other cognitive functions remain preserved. It is classified into three subtypes: the agrammatic or non-fluent variant, the semantic variant, and the logopenic variant, each with distinct symptoms and neuropathology. The agrammatic/non-fluent variant is marked by reduced speech fluency, which becomes effortful, with agrammatism, difficulty in understanding complex grammatical structures, apraxia, and phonetic paraphasias. The semantic variant is characterized by a progressive loss of semantic knowledge of words. Patients maintain or even enhance speech fluency and preserve grammatical structure but exhibit impaired word comprehension. The logopenic variant is distinguished by speech disruptions, pauses, and reduced fluency. Affected individuals tend to repeat phrases, while articulation, prosody, and grammar remain intact (American Psychiatric Association, 2022; Sociedad Española de Neurología, 2018). iii. Corticobasal Degeneration Corticobasal degeneration (clinically referred to as corticobasal syndrome) is characterized by a marked asymmetry of motor symptoms, with prominent rigidity and cortical dysfunction. However, cognitive and behavioral symptoms, which may also exhibit asymmetry, often predominate over motor impairments. Key symptoms of corticobasal degeneration include rigidity or akinesia affecting one or more limbs, dystonia, and apraxia, which can manifest either symmetrically or asymmetrically. Additionally, patients often exhibit executive dysfunction, behavioral changes, and visuospatial deficits. It is important to note that many symptoms of corticobasal degeneration (especially asymmetric extrapyramidal and cognitive impairments) can also appear in other 26 INTRODUCTION TREM2. TREM2 signaling induces DAP12 phosphorylation, leading to cytoskeletal reorganization and enabling phagocytosis (Takahashi et al., 2005). Thus, microglia are essential for clearing apoptotic cells and debris, ensuring the proper functioning of the CNS. iii. Synaptogenesis Beyond regulating neuronal survival and death, microglia play a crucial role in synaptic homeostasis (Paolicelli et al., 2011; Wake et al., 2009). These cells actively participate in synaptic pruning, responding to synaptic activity and plasticity. Proper synaptic function depends on various trophic factors and synaptogenic signals, many of which are released by microglia. A key molecule in synaptic function and plasticity is DAP12, whose expression in the mouse brain is restricted to microglia. The absence of DAP12 leads to synaptic dysfunction and deficits in synaptic plasticity (Roumier et al., 2004). Microglia regulate synaptic density, the presence of glutamatergic receptors, and the number of dendritic spines (Ji et al., 2013). Additionally, microglia are involved in experience-dependent synaptic plasticity and actively remodel synaptic architecture (Tremblay et al., 2010). Programmed axonal pruning, the elimination of defective synapses, and the removal of redundant neuronal processes are fundamental events in the establishment of functional neural circuits during embryonic development. These processes are mediated, in part, by CX3CR1 and the complement system (Paolicelli et al., 2011; Stevens et al., 2007). d. Neuroinflammation Beyond its homeostatic functions, microglia play a crucial role in the central nervous system by orchestrating inflammatory responses. Inflammation is a biological process aimed at resolving tissue damage or eliminating pathogens through various molecular cascades. As a result, inflammation is often associated with both tissue damage and repair. In the brain, neuroinflammation is primarily mediated by microglia, which undergo activation in response to immunogenic molecules. Upon activation, microglia adopt an amoeboid morphology with fewer ramifications, alongside changes in their expression profile and membrane receptors (Woodburn et al., 2021). Pattern recognition receptors (PRRs), including Toll-like receptors (TLRs), are responsible for detecting damage-associated molecular patterns (DAMPs) and pathogen-associated molecular patterns (PAMPs), determining the microglial response and phenotype (Chen & Nuñez, 2010). This activation leads to a significant increase in the production and release of pro-inflammatory cytokines, such as tumor necrosis factor-alpha (TNFα), multiple interleukins and chemokines, interferon-gamma, and reactive oxygen and nitrogen species (ROS and RNS) (Cherry et al., 2014; Woodburn et al., 2021). 27 INTRODUCTION Although inflammation plays a critical role in tissue defense and repair, it also carries a cytotoxic component, leading to the death of endogenous cells, with neurons being particularly vulnerable. Therefore, anti-inflammatory mechanisms, also orchestrated by microglia, are essential in mitigating tissue damage. The release of molecules such as interleukin-4 (IL-4) and interleukin-10 (IL-10) is crucial in this process (Gadani et al., 2012). An insufficient anti-inflammatory response or chronic inflammation can result in significant tissue damage (Cherry et al., 2014). A key component of neuroinflammation is the inflammasome, a protein complex that regulates caspase-1 protease activity. Caspase-1 is responsible for the proteolytic maturation of interleukin-1β (IL-1β) and interleukin-18 (IL-18) and induces pyroptosis, a rapid form of inflammatory cell death (Rathinam & Fitzgerald, 2016). Inflammasome assembly is dependent on PRR recognition of PAMPs and DAMPs, leading to caspase-1 recruitment and activation. This process involves caspase-1 auto-cleavage and conformational changes, facilitated by the adaptor protein ASC (Van Opdenbosch & Lamkanfi, 2019). Once activated, caspase-1 processes pro-IL-1β and pro-IL-18 into their biologically active forms. Additionally, transcription factors such as NF-κβ regulate the expression of inflammasome components like NLRP3 (Hayden & Ghosh, 2011). The inflammasome's activity can also be modulated by post-translational modifications and interactions with signaling pathways such as the TLR pathway (Heneka et al., 2018; Wen et al., 2013; Xia et al., 2023). Due to their functional similarities with macrophages, microglia have historically been classified into M1 (pro-inflammatory) and M2 (anti-inflammatory) phenotypes (Orihuela et al., 2016). However, transcriptomic studies over the past decade have demonstrated that microglial activation is not confined to this binary model. Instead, microglia exhibit a broad spectrum of activation states, influenced by various factors, including the activating stimulus, CNS region, organismal age, and microenvironmental conditions (Stratoulias et al., 2019). e. Microglial Heterogeneity Since its discovery, the identification and characterization of microglia have been subjects of ongoing debate. It is now known that these cells possess self-renewal capabilities through the CSF1R receptor, whose ligands are IL-34 and CSF1 (Chitu et al., 2016). As a result, CSF1R has been recognized as a key marker for microglia identification. Moreover, mammalian microglia express a distinct set of molecules referred to as "microglial markers," including IBA1, PU.1, P2RY12, TMEM119, and CSF1R (Grabert et al., 2020). However, these markers may also be present in other cell types, such as border-associated macrophages (BAMs), CNS-associated macrophages (CAMs), and other glial cells (Chappell-Maor et al., 2020; Kaiser & Feng, 2019; Masuda et al., 2020; McKinsey et al., 2020; Parkhurst et al., 2013; Yona et al., 2013). Additionally, microglial marker expression may be downregulated under pathological conditions (Kim et al., 2021). 28 INTRODUCTION In a healthy CNS, microglia are not in a dormant state but instead maintain a homeostatic condition characterized by high plasticity and continuous surveillance of their surroundings, cellular environment, and structural components (Davalos et al., 2005; Nimmerjahn et al., 2005). The homeostatic microglial population exhibits significant heterogeneity, with their state influenced by factors such as age, sex, and both CNS and peripheral signals, including microbiota composition (Erny et al., 2015; Thion et al., 2018). Microglia are highly adaptable and respond to diverse stimuli, making them critical players in neurodegenerative diseases and various neurological and developmental processes. Depending on genetic background, environmental factors, and disease stage, microglia undergo morphological and transcriptional changes, altering the expression of both classical microglial markers like P2RY12 and other functional proteins (Chen & Colonna, 2021; Paolicelli et al., 2022). These variations result in the emergence of multiple microglial states, including Disease-Associated Microglia (DAMs), Microglial Neurodegenerative phenotype (MGnD), Activated Response Microglia (ARMs), Interferon-Responsive Microglia (IRMs), Human AD Microglia (HAMs), Microglia Inflamed in Multiple Sclerosis (MIMS), Lipid-Droplet-Accumulating Microglia (LDAMs), GliomaAssociated Microglia (GAMs), ALS-associated microglia, Parkinson Disease microglial signature, White Matter-Associated Microglia (WAMs), Axon Tract-Associated Microglia (ATMs), and Proliferative-region-Associated Microglia (PAMs), among others (Paolicelli et al., 2022). Among the microglial states associated with pathological conditions, DAM and MGnD microglia stand out. These phenotypes share a common molecular signature, including upregulation of Trem2, Apoe, Itgax, Spp1, and Clec7a, along with downregulation of homeostatic genes such as P2ry12, Hexb, Cx3cr1, and Tmem119. The transition from homeostatic to DAM/MGnD microglia is mediated by the TREM2 receptor, although its precise role remains unclear (Keren-Shaul et al., 2017; Krasemann et al., 2017; McQuade et al., 2020; Zhou et al., 2020). Potential functions of TREM2 include β-amyloid sensing and regulation of microglial energy metabolism and anaerobic glycolysis (Ulland et al., 2017; Xiang et al., 2021). The DAM/MGnD transition has been observed in multiple neurodegenerative diseases, including Alzheimer’s disease, amyotrophic lateral sclerosis, and multiple sclerosis, as well as in aging brains and developmental myelination regions (Chiu et al., 2013; Keren-Shaul et al., 2017; Safaiyan et al., 2021; Sobue et al., 2021). These findings raise important questions regarding the role of DAM/MGnD microglia, whether they exert neuroprotective or neurotoxic effects, whether they are restricted to specific pathologies or represent a universal response to certain stressors, and whether their functions differ between the developing and aging brain or in disease contexts (Paolicelli et al., 2022). 29 INTRODUCTION f. Microglia in Alzheimer's Disease Microglia's ability to sense environmental changes and respond accordingly positions them as key players in neurodegenerative diseases such as Alzheimer’s disease (AD). However, the exact role of microglia in AD pathology remains unclear. Microglia recognize Aβ peptide through various receptors, including CD36, TLR2, TLR4, TLR6, and NLRP3, triggering an inflammatory response (El Khoury et al., 2003; Heneka et al., 2013, 2015). In fact, microglia cover approximately 80% of Aβ plaques in Alzheimer’s patients (Wisniewski et al., 1989). Additionally, microglia detect and respond to DAMPs such as ATP and DNA (Colonna & Butovsky, 2017). These cells form a physical barrier around Aβ plaques, phagocytose oligomeric Aβ, and limit Aβ polymerization through secreted enzymes such as neprilysin (Bolmont et al., 2008; Condello et al., 2015; Frautschy et al., 1992; Mandrekar et al., 2009; Miners et al., 2011; Simard et al., 2006; Yang et al., 2011). Recent studies propose that microglia may be the source of Aβ plaques. Within microglial lysosomes, the accumulation of APOE lipoprotein promotes Aβ aggregation, initiating plaque nucleation and subsequent growth through phagocytosed Aβ. These primitive plaques are then released into the extracellular space via exocytosis or following microglial cell death (Kaji et al., 2024). Microglia also play a critical role in maintaining neuronal networks, but this function may be compromised under neuroinflammatory conditions due to the cytotoxic effects of cytokines and reactive oxygen species released by microglia (Kettenmann et al., 2013; McGeer & McGeer, 2010; Nayak et al., 2014). Aβ presence further amplifies inflammatory responses by activating NF-κβ signaling, and Alzheimer's patients exhibit elevated levels of IL-1β, IL-6, and TNFα (Bauer et al., 1991; Combs et al., 2001). Furthermore, Aβ induces the secretion of chemokines such as CCL2 and CCL3, promoting microglial and astrocytic recruitment (Arfaei et al., 2024; Ridolfi et al., 2013). Interestingly, studies in Alzheimer’s mouse models suggest that the absence of NLRP3, TLRs, and IL-1β reduces Aβ deposition and prevents cognitive decline (Heneka et al., 2015). However, other research indicates that inflammation may have a neuroprotective role. The complete depletion of microglia in animal models leads to significant brain parenchymal alterations, including a shift from compact to diffuse Aβ plaques, cerebral amyloid angiopathy, calcification, hemorrhages, transcriptional changes, and premature lethality (Kiani Shabestari et al., 2022). Moreover, some findings suggest that IL-1β and IFN-γ expression may trigger a beneficial inflammatory response that facilitates Aβ clearance (Chakrabarty et al., 2010; Shaftel et al., 2007; Weitz & Town, 2012). These contradictory findings highlight the complex and dual role of microglia in Alzheimer’s pathology, where their activity can be both beneficial and detrimental. Consequently, modulating microglial function has been proposed as a potential therapeutic target for Alzheimer's disease (Condello et al., 2015). 30 INTRODUCTION g. Microglia in Frontotemporal Dementia Like AD, microglia appear to play a crucial role in FTD. Using PET, Cagnin and colleagues (Cagnin et al., 2004) detected microglial activation in frontotemporal lobar degeneration. Moreover, activated microglia express progranulin in response to specific stimuli, such as in experimental traumatic spinal cord injury models (Byrnes et al., 2011; Naphade et al., 2010; Y. Tanaka et al., 2013). Progranulin deficiency induces an exacerbated inflammatory response in microglia (Tanaka et al., 2013). Several studies have attempted to explain how loss-of-function mutations in the GRN gene lead to FTLD. This deficiency results in increased production of proinflammatory cytokines, including CCL2, CXCL1, IL-6, IL-12p40, and TNF, while reducing antiinflammatory cytokines like IL-10 in macrophages. Additionally, GRN-deficient mice exhibit impaired resistance to bacterial infections compared to controls, as well as heightened microglial activation with age and an exaggerated inflammatory response, leading to increased microglial cytotoxicity. Furthermore, GRN knock-out neurons are more vulnerable to damage and stress conditions. Aged GRN knock-out mice also show increased TDP-43 phosphorylation in the hippocampus and thalamus. Behaviourally, these animals display deficits in social recognition tests, which emerge as early as one month of age (Yin, Banerjee, et al., 2010). Upon exposure to damaging stimuli such as MPTP, LPS, and IFN-γ, microglia exhibit dysregulated activation, contributing to neuronal death. It has been observed that GRN functions as a neuroinflammation modulator and that its deficiency may contribute to neurodegenerative processes (Martens et al., 2012). Some studies hypothesize that frontotemporal lobar degeneration results from the combination of both increased microglial activation and reduced neuronal resilience (Yin, Banerjee, et al., 2010). One of the most relevant histopathological features of GRN-deficient mice and FTD patients is the accumulation of lipofuscin or lipofuscinosis, a neuropathological marker associated with lysosomal dysfunction (Root et al., 2023). Lipofuscin mainly consists of lipoproteins localized within secondary lysosomes and exhibits autofluorescent properties (Miller & Zachary, 2017). In addition to neurons, lipofuscin can be found inside microglia due to phagocytic processes (Stillman et al., 2023). Moreover, progranulin deficiency leads to significant lysosomal hypertrophy and dysfunction, with elevated lipofuscin levels (Evers et al., 2017; Stillman et al., 2023). A recent study indicates that GRN loss triggers molecular cascades and cellular events in microglia, shifting them from a homeostatic to a specific neurotoxic pathological phenotype. This transition is associated with cytoplasmic accumulation of TDP-43, nuclear pore formation, and excitatory neuronal death during aging. Notably, this neurotoxic microglial state can be mitigated by deleting genes encoding C1q and C3 or by blocking complement system activation (Zhang et al., 2020). 31 INTRODUCTION Progranulin may also play a role in microglial activation in other neurodegenerative conditions, such as AD, where it colocalizes with Aβ plaques (Gliebus et al., 2009; Pereson et al., 2009). Progranulin promotes the secretion of lectin and neuroprotective cytokines such as IL-4, IL-5, and IL-10 while reducing the secretion of TRAIL, an apoptosis regulator. Additionally, progranulin enhances chemotaxis and endocytosis of extracellular peptides by microglia, suggesting that its regulation could be a potential therapeutic target for neurodegenerative diseases (Pickford et al., 2011; Uberti et al., 2004). In tau-associated FTD, Bellucci and colleagues (Bellucci et al., 2011) investigated the role of mutated tau protein (P301S) in patients with FTDP-17T, a spectrum of disorders ranging from FTLD to parkinsonism. In the cortex and hippocampus of these patients, CD68positive activated microglia and infiltrating macrophages were observed. These cells were found surrounding neurons containing intracellular aggregates of hyperphosphorylated tau, contributing to reactive microgliosis. This microglial activation, accompanied by a proinflammatory environment, is thought to drive neuronal death and disease progression in tau-related neurodegenerative disorders. 4. Galectin-3 – When Science and Mythology Unite a. The Galectin Family Galectins are proteins belonging to the lectin family, characterized by their ability to bind β-galactosidase through a carbohydrate recognition domain (CRD) present in their peptide chain. Galectins perform numerous functions via protein-carbohydrate and protein-protein interactions. Currently, 15 different galectins have been identified in mammals, 11 of which are present in humans. Based on their quaternary structure, they can be classified into three types. The prototypic galectins (Gal-1, Gal-2, Gal-5, Gal-7, Gal10, Gal-11, Gal-13, Gal-14 and Gal-15) have a single CRD and can form monomers or dimers, although they recognize only one ligand. Tandem-repeat galectins (Gal-4, Gal-6, Gal-8, Gal-9, and Gal-12) possess two CRDs, enabling them to bind two distinct ligands simultaneously. A special mention goes to Galectin-3, the only member of the chimeratype galectins. It is distinguished by having three domains: an N-terminal domain similar to collagen, involved in oligomerization and regulatory functions; a proline-rich domain; and a single CRD at the C-terminal end. Galectins play essential roles in inflammatory responses and neuroprotection mechanisms. However, the precise regulatory mechanisms governing their expression remain unknown (Arthur et al., 2015; GarcíaRevilla et al., 2022; Rabinovich & Toscano, 2009; Suarez & Meyerrose, 2014). 32 INTRODUCTION Illustration I-8: A. Classification of galectins in three families. B. Binding partners of Gal3. C. Gal3 interactions. Extracted from Soares (Soares et al., 2021). b. Functions of Galectin-3 Galectin-3 (Gal3) exhibits a wide range of functions depending on both its cellular localization and the specific cell type it acts upon. In the nucleus, Gal3 is involved in mRNA splicing by binding to U1 snRNP (Small Nuclear Ribonucleoproteins) and pre-mRNA, forming the spliceosome (Haudek et al., 2016). It also plays a role in the activation of transcription factors such as CREB (cAMP Response Element Binding Protein) (Lin et al., 2002). Additionally, Gal3 participates in the Wnt signaling pathway through its interaction with β-catenin (Dumic et al., 2006; Shimura et al., 2004). In the cytoplasm, Gal3 is involved in the induction of proliferation, differentiation, and survival via K-RAS-mediated signaling (Elad-Sfadia et al., 2004). It also regulates apoptosis by either preventing it through interaction with Bcl-2 or inducing it by binding to CD95, leading to caspase-8 activation (Fukumori et al., 2004; Yang et al., 1996). In a tumor environment, Gal3 promotes proliferation and adhesion (Rivera-Ramos et al., 2024; Vladoiu et al., 2014). Furthermore, Gal3 interacts with endosomes, phagosomes, and damaged lysosomes and is involved in autophagy (Thurston et al., 2012; Weng et al., 2018). In the extracellular space, Gal3 can be secreted through unconventional pathways such as exosomes (Bänfer et al., 2018; Théry et al., 2001). It influences microglial cell 33 INTRODUCTION morphology through interactions with glycosylated growth factors like IGF-1 (Rahimian et al., 2018). In some tumor cells, Gal3 inhibits immune system function by immobilizing Tcell receptors (TCRs), preventing their interaction with co-receptors like CD8, and inducing apoptosis (Demotte et al., 2008; Fukumori et al., 2003). It is also involved in neutrophil activation and recruitment and plays a significant role as an opsonin (Bhaumik et al., 2013; Fermino et al., 2011; Karlsson et al., 2009). On the cell surface, Gal3 interacts with numerous receptors, modifying their dynamics by increasing their retention time on the cell surface, producing different effects depending on the cell type (Dennis et al., 2009; Partridge et al., 2004; Rabinovich & Toscano, 2009). Among its attributed functions, Gal3 has been suggested to have an anti-inflammatory role by interacting with TCRs, limiting migration (Demetriou et al., 2001). However, other studies propose that Gal3 is essential for cytokine signaling and leukocyte migration (Partridge et al., 2004). In microglia, Gal3 has been reported to interact with TLR4 and TREM2 (Boza-Serrano et al., 2019; Burguillos et al., 2015). Focusing on innate immune system cells, Gal3 exerts its effects on neutrophils, monocytes, macrophages, and microglia. In neutrophils, Gal3 is involved in oxidative burst and cell activation, promoting phagocytosis and extending cell lifespan (Farnworth et al., 2008). Other studies suggest a role in inflammatory factor release via MAPK activation, as well as a reduction in neutrophil lifespan due to apoptotic mechanisms and lower ROS production (Fernández et al., 2005; Wu et al., 2017). In monocytes, Gal3 has been linked to increased phagocytosis, acting as an opsonin. This process can be inhibited by lactose, which blocks Gal3 activity (Karlsson et al., 2009). Additionally, Gal3 is involved in monocyte-monocyte interactions and is upregulated when monocytes differentiate into macrophages, while its levels decrease upon further differentiation into dendritic cells. In macrophages, Gal3 plays a crucial role in alternative activation through interactions with IL-4 and CD98 (MacKinnon et al., 2008). Its inhibition enhances the inflammatory response to stimuli like LPS (Li et al., 2008). c. Galectin-3 and Microglia Although homeostatic microglia do not significantly express Gal3, there is a strong association between activated microglia and this protein (Krasemann et al., 2017; Mathys et al., 2017). Recent single-cell transcriptomics studies have revealed that activated microglia exhibit a much more complex activation phenotype than the simplified M1/M2 classification (Ransohoff, 2016; Stratoulias et al., 2019). In conditions of neuroinflammation, including AD, amyotrophic lateral sclerosis, and aging models, specific microglial clusters have been identified where Gal3 emerges as a relevant marker (Holtman et al., 2015). Holtman and colleagues found a distinct expression profile in these microglial cells compared to the classical pro-inflammatory phenotype, characterized by 34 INTRODUCTION the induction of genes such as Itgax, Axl, Clec7a, and Lgals3 (Holtman et al., 2015). Additionally, they highlighted Csf1, Axl, Igf1, and Lgals3 as key regulators of essential microglial functions, including phagocytosis, proliferation, and activation, playing a role in the polarization state of these cells across different pathologies. Gal3 is one of the most upregulated genes in pathological conditions in the brain (Boza-Serrano et al., 2019; Yip et al., 2017). In fact, both in AD patients and in experimental models, Gal3-expressing microglia are frequently found surrounding amyloid plaques (Boza-Serrano et al., 2019). This not only underscores the importance of Gal3 in microglial function and neurodegenerative diseases but also supports the idea that microglial activation states are highly diverse, extending beyond the simplistic M1/M2 classification (Rangaraju et al., 2018). Recent studies on FTD animal models deficient in the progranulin encoding gene (Grn) have shown that the absence of this protein leads to an accumulation of TDP-43, with the most enriched proteins in the brains of these animals being GPNMB (Glycoprotein nonmetastatic melanoma protein B) and Gal3. Both proteins are predominantly expressed by microglia and are particularly abundant in the cortex, hippocampus, and thalamus (Huang et al., 2020; Zhang et al., 2020). While no direct interaction between Gal3 and lipid metabolism or myelin has been described, Gal3 does interact with certain membrane lipids, suggesting a potential role in lipid clearance (Lukyanov et al., 2005). Additionally, Gal3 is closely linked to TREM2, a receptor primarily expressed in DAM or MgnD states (Keren-Shaul et al., 2017; Krasemann et al., 2017), which interacts with anionic and zwitterionic lipids (Y. Wang et al., 2015). Furthermore, MGnD/DAM microglia are characterized by the overexpression of lipid metabolism-related genes, including APOE (Keren-Shaul et al., 2017; Krasemann et al., 2017). 35 INTRODUCTION Illustration I-9: A. Microglial can be isolated by antibodies agains ITGAX (CD11c), CLEC8A and MHCII. Microglia associated with Aβ plaques adquire an especific phenotipe defines as DAM or MGnD. B. TREM2 induces microglial shift from homeostatic to DAM. TREM2 senses phosphatidylserine on damage or apoptotic cells, leading to ApoE signaling, thus inducing DAM phenotype, which is associated with inflammatory molecules like CLEC7A, GONMB, ITGAX, MSR1 and LGALS3. Image extracted from Butovsky (Butovsky & Weiner, 2018). A recently identified subtype of microglia named WAM (White-matter Associated Microglia), shares part of its molecular signature with DAM microglia but is TREM2dependent and APOE-independent. WAM microglia are enriched in genes related to HIF1 signaling, lysosomal pathways, and cholesterol metabolism, exhibiting overexpression of Clec7a, Axl, Itgax, and Lgals3 (Poliani et al., 2015; Safaiyan et al., 2021). This type of microglia has been proposed to actively participate in the clearance of myelin debris generated during aging and in neurodegenerative diseases. Given that WAM microglia show a marked expression of Gal3 in animal models of aging and AD, it has been suggested that this protein plays a relevant role in myelin clearance. Gal3, through its interaction with various proteins such as TLRs, TREM2, IGF1R, or MERTK receptors, can modulate the microglial activation phenotype (García-Revilla et al., 2022). In various neurodegenerative diseases, sterile inflammation develops, characterized by the absence of pathogens and promoted by the accumulation of protein aggregates such as Aβ or α-synuclein. Many of these aggregates interact with TLR receptors, suggesting a key role for these receptors in the microglial inflammatory response (Agalave et al., 2014; Burguillos et al., 2015; Hughes et al., 2020; Jana et al., 2008; Jin et al., 2008; Kim et al., 2013; Maezawa et al., 2011; Yip et al., 2017). Materials and methods MATERIALS AND METHODS 45 MATERIALS Y METHODS Materials and methods 1. Animal models All experiments were conducted in compliance with European (2010/63/EU) and Spanish (BOE 34/11370-421, 2013) regulations, which establish the fundamental guidelines governing the use of animals in experimentation. Additionally, these experiments were approved by the Scientific Ethics Committee of the University of Seville. All animals used in this thesis were bred, maintained, and crossed at the Óscar Pintado Animal Production Center and, when necessary, housed at the animal facility of the Faculty of Pharmacy at the University of Seville. The animals were kept under controlled conditions: a constant temperature of 22°C, 60% humidity, a 12-hour light/dark cycle, and ad libitum access to food and water. Although various mouse lines were used in this doctoral thesis, all of them share the same genetic background: C57BL/6. The specific details of each strain will be described below. a. PGRN Knockout Model The PGRN knock-out (-/-) (B6(Cg)-Grntm1.1Aidi/J, strain 013175, Jackson Laboratory), referred to in this text as PGRNKO, carries a deletion of exons 1-4 of the progranulin (Grn) locus. Homozygous mice exhibit behavioural and neuropathological features similar to FTD. These include microglial and astroglial activation, as well as cytoplasmic accumulation and ubiquitination of phosphorylated TDP-43 in hippocampal and thalamic neurons. At 18 months of age, these mice exhibit spatial learning and memory deficits. Macrophages from these animals show reduced interleukin-10 (IL-10) secretion but increased inflammatory cytokine production in response to microbial agents. Nevertheless, these mice struggle to clear infections due to a dysregulated inflammatory response. b. APPsl Model The hAPP751sl (+/-) murine model, referred to in this text as APP, is a transgenic mouse model overexpressing a mutated form of human amyloid precursor protein (APP, 751 isoform). Specifically, this protein carries both the Swedish (K670N/M647L) and London (V717I) mutations, under the control of the murine Thy1 promoter. This model exhibits an age-dependent increase in Aβ40 and Aβ42 levels. Soluble Aβ40 and Aβ42 appear in the brain around six months of age, alongside an increase in Aβ oligomers. Plaque formation occurs early, between three and six months, in the frontal cortex, later extending to the hippocampus, thalamus, and olfactory region. 46 MATERIALS Y METHODS c. Tau P301S Model The Tau (+/-) model (B6;C3-Tg(Prnp-MAPT*P301S)PS19Vle/J) expresses the P301S mutant form of human microtubule-associated protein tau (MAPT) under the control of the Prnp promoter. In these animals, hyperphosphorylated and insoluble MAPT protein accumulates in the brain with age. By three months of age, these mice display limb retraction when lifted by the tail. By ten months, they develop a hunched posture and paralysis, leading to feeding difficulties. The average lifespan of this model is nine months, with 80% mortality by 12 months. Histological analyses show neuronal degeneration and ventricular dilation by eight months. Neuronal loss extends to the amygdala, isocortex, and entorhinal cortex by 12 months. Neurofibrillary tangles, similar to those found in AD and tauopathies, emerge in these regions, as well as the brainstem and spinal cord, from five months of age. Additionally, this model exhibits microgliosis, astrogliosis, and synaptic degeneration. d. Galectin-3 Knockout Model The Galectin-3 knockout (-/-) (Gal3KO) mice were obtained from Dr. K. Sävman at the University of Gothenburg (Sweden). Through various breeding strategies involving the previously described models, the following experimental lines were generated: • PGRNKO/Gal3KO • PGRNKO/P301S • Gal3KO/APP 2. Genotyping The use of multiple animal lines, along with the need to work with heterozygous individuals in some cases, requires a genotyping process to determine the genotype of each animal. To achieve this, a biopsy sample is collected, preferably from the phalanges or ears, allowing for both identification and numbering of the animal. DNA is then extracted from the tissue using 25 µL of the "QuickExtract™ DNA Extraction Solution" kit (Lucigen, QE09050). Following the manufacturer’s instructions, the sample undergoes incubation at 65°C for 15 minutes in a Biometra T-personal Thermal Cycler, followed by a second incubation at 98°C for 5 minutes. Once DNA extraction is complete, a Polymerase Chain Reaction (PCR) is performed using the MyTaq™ Red DNA Polymerase kit (Bioline, 21109) and the corresponding primers (Table 1) at a concentration of 10 µM. The reaction mix is prepared as follows: 47 MATERIALS Y METHODS • 10 µL of 5x MyTaq Red Reaction Buffer • 0.5 µL of MyTaq Red DNA Polymerase • 1 µL of DNA sample • 2 µL of each primer (10 µM concentration) • Nuclease-free water to a final volume of 50 µL Table 1: Sequence of primers used for genotyping the different animal models. Primer Sequence (5’ – 3’) APP (FW) GGCTGAGGAACCCTACGAAGA APP (RV) CAAAGTACCAGCGGGAGATCA Gal3 (comon) CTACTCCTTGGCCCTCTAGGTC Gal3 (WT) TGAAATACTTACCGAAAAGCTGTCTGC Gal3 (KO) GCTTTTCTGGATTCATCGACTGTGG PGRN (comon) AGAGGGTGAGCTGCAATGTT PGRN (WT) TCTCCCAGGTAGCCCCTACT PGRN (KO) AAGGGCATTAGCCAAGTGTG MAPT (FW) GGCATCTCAGCAATGTCTCC MAPT (RV) GGTATTAGCCTATGGGGGACA The PCR reaction is carried out in an Eppendorf Mastercycler epgradient S Realplex 2, following the corresponding thermal cycling program for each genotype (Table 2). Following amplification, PCR products are loaded onto a 1.5% agarose gel prepared with TAE buffer (Tris base/acetic acid/EDTA) and stained with RedSafe™ Nucleic Acid Staining Solution 20000X (iNtRON Biotechnology, 21141). After solidification, electrophoresis is performed, and the results are visualized under ultraviolet light using a Chemi-Doc MP Imaging System (BIORAD). 48 MATERIALS Y METHODS Table 2: PCR program used for genotyping the different animal models. Gene Temperature Time Cycles Lgals3 94°C 5 min 1 cycle 94°C 45 s 40 cycles 55°C 30 s 72°C 1.3 s 72°C 10 min 1 cycle 25°C 30 min 1 cycle App 94°C 15 min 1 cycle 94°C 45 s 20 cycles 60°C 45 s 72°C 45 s 72°C 10 min 1 cycle Mapt 95°C 1 min 1 cycle 95°C 15 s 35 cycles 58°C 15 s 72°C 10 s Grn 94°C 1 min 1 cycle 94°C 15 s 10 cycles 65°C 15 s 68°C 10 s 94°C 15 s 28 cycles 60°C 15 s 72°C 10 s 3. Immunofluorescence Mice were anesthetized using isoflurane and transcardially perfused with saline buffer (0.9% NaCl in water). The brains were extracted and placed in 4% PFA for one week at 4°C, then stored in a 30% sucrose solution with 0.1% sodium azide for at least 48 hours. Next, the brains were frozen in dry ice for 15 minutes and sectioned at 30 µm using a Leica CM1850 cryostat. These sections were preserved at -40°C in antifreeze solution (30% sucrose, 30% ethylene glycol in 0.1M phosphate buffer, pH 7.4). After selecting the desired section, the tissues were placed in baskets positioned over 24well plates (floating immunofluorescence). Three five-minute washes were performed in phosphate-buffered saline (PBS, NZYtech, MB18201), followed by one hour of incubation in PBS with 1% Triton X-100 (Sigma-Aldrich, X100) for tissue permeabilization. Blocking was then carried out by incubating the samples for two hours in PBS-T 1% with 5% bovine serum albumin (BSA, Sigma-Aldrich, A7906). After blocking, tissues were incubated with primary antibodies (Table 3) in PBS-T 0.1% with 1% BSA for 30 minutes at room 49 MATERIALS Y METHODS temperature and 18 hours at 4°C. Once this time elapsed, the tissues were brought to room temperature for 30 minutes and washed six times in PBS-T 0.1%, with each wash lasting five minutes. Subsequently, the samples were incubated for one hour in PBS-T 0.1% with 1% BSA containing secondary antibodies conjugated to Alexa Fluor fluorophores (Thermo Fisher Scientific). For thioflavin-S (ThS) staining (0.01 mg/mL, Sigma-Aldrich, T1892-25G), the solution was added after incubation with secondary antibodies at a 1:1000 dilution in PBST 0.1% for five minutes. The tissues were then washed three times in PBS-T 0.1% and three times in PBS, each for five minutes. Finally, the sections were mounted on glass slides using a 50% glycerol solution (PanReac AppliChem, A1123,1000) in PBS and sealed with nail polish. Table 3: Antibodies utilised for immunofluorescence techniques and usage conditions. Antibody Host Dilution Citrate (Recommended) Reference 6E10 Mouse 1:1000 No SID-39320 BAM10 Mouse 1:1000 No A3981 (Sigma-Aldrich) CD68 Rat 1:250 No 14-0681-82 (Invitrogen) IBA1 Rabbit 1:1000 No 019-19741 (Wako) LAMP1 Rat 1:250 Yes 1D4B-S (DSHB) TREM2 Sheep 1:250 Yes AF1729 (R&D) GPNMB Rabbit 1:250 Yes 56898S (Cell Signaling) CLEC7A Rat 1:250 Yes mabg-mdect-2 (Invivogen) GAL3 Goat 1:500 No AF1197 (R&D) 4. Immunohistochemistry Immunohistochemistry is a technique that shares great procedural similarity with the previously described immunofluorescence, as both are based on the detection of antigens present in animal tissue through the use of antibodies. The technique begins with 30 µm-thick mouse brain sections, which are mounted on gelatin-coated slides and left to dry. They are then outlined with an Immedge Pen (Vector Laboratories, H-4000) and washed three times with PBS 1X for 5 minutes per wash. Next, the tissues are incubated for 15 minutes in a solution of hydrogen peroxide (Scharlau, HI01351000) diluted to 0.33% in methanol (PanReac AppliChem, 131091.1214), followed by three additional washes in PBS 1X for 5 minutes each. Subsequently, tissue blocking and permeabilization are carried out by incubating the samples for 2 hours in PBS-T 1% BSA 5%, after which they are incubated overnight at 4°C with the corresponding primary antibody (Table 4) in PBS-T 1% BSA 1%. Once the primary antibody incubation is complete, the tissues are washed three times with PBS 1X as previously described and then 50 MATERIALS Y METHODS incubated with the secondary antibody, diluted in PBS-Triton X100 0.1% BSA 1%, for 2 hours at room temperature. Following this, three 5-minute washes with PBS 1X are performed, and the reagent from the VECTASTAIN® Elite® ABC kit (Vector Laboratories, PK-6100) (98 µL of PBS + 1 µL of reagent A + 1 µL of reagent B) is applied for 1 hour. After three additional washes with PBS 1X for 5 minutes each, the tissue is incubated with the reagent prepared according to the DAB Substrate Kit (Vector Laboratories, SK-4100) protocol (5 mL of Milli-Q water + 2 drops of reagent 1 + 4 drops of reagent 2 + 2 drops of reagent 3) for a specific time to allow proper visualization of each antigen. The reaction is stopped with three 5-minute washes in PBS 1X, and the samples are mounted with DPX Mounting Medium for Histology (Scharlau, DP00500100) and left to dry for several days. Table 4: Antibodies utilised for immunohistochemistry techniques and usage conditions. Antibody Host Dilution Reference AT8 Mouse 1:500 MN1020 (Invitrogen) IBA1 Rabbit 1:1000 019-19741 (Wako) GAL3 Goat 1:500 AF1197 (R&D) 5. Cell Separation For microglia isolation, the process begins with animal perfusion and the extraction of the cerebral cortex. The tissue is then mechanically dissociated using a scalpel and passes through 20G needles in HBSS 1X with 10% FBS. Next, the suspension is filtered through a 70 µm strainer pre-moistened with dissociation buffer and centrifuged at 160 x g for 10 minutes at 4°C with brake. The pellet is then resuspended in a 30% isotonic Percoll solution (90% Percoll + 10% HBSS 1X), creating a gradient with HBSS 1X. After centrifugation at 800 x g for 10 minutes at 4°C with brake, the supernatant is discarded, and the pellet is washed with PBS 1X. A subsequent centrifugation at 800 x g for 10 minutes at 4°C with brake is performed, and the pellet is resuspended in 500 µL of PBS 1X. Next, 400 µL of the sample is mixed with 1 µL of anti-CD11b-APC antibody (MO11BA, Immunostep S.L.) and 1 µL of anti-CD45-PE antibody (M45PE, Immunostep S.L.). This mixture is incubated for 20 minutes at room temperature in darkness. Finally, the samples are processed using the Facs Aria Fusion cell sorter (Becton Dickinson). 6. RNASeq Microglia isolated through the previously described cell separation process was analyzed using RNA Sequencing (RNASeq). RNA was extracted from the isolated cells using a cell sorter and assessed with the Bioanalyzer 2100 (Agilent Technologies). Library preparation was then performed using the SMART-Seq Stranded kit from Takara Bio USA, starting with 10 ng of total RNA. To ensure proper size distribution, the libraries were evaluated using the Bioanalyzer DNA High Sensitivity chip, with prior quantification by fluorometry (Qubit). The pooled sequencing libraries were denatured and diluted to the required 51 MATERIALS Y METHODS concentration of 1.6 pM. A control library (PhiX) was always included to monitor sequencing quality, detect potential issues, or troubleshoot unexpected problems with the equipment. Cluster generation (monoclonal amplification) was performed on the Flow Cell using the NextSeq500 MIDOutput platform with a read length of 2x75bp. Sequencing quality results were analyzed using Illumina's BaseSpace Hub software. The FASTQ Toolkit v1.0.0 application was used for filtering and trimming (removal of adapters, poly-A/T sequences, etc.), generating the final FASTQ files. 7. RNA Extraction RNA extraction was performed using the gTPzol reagent (GTPZOL01, gTPbio, C-Viral, Spain). Once the brain structures were extracted from the animal, they were mechanically dissociated using a micropipette and 20G and 27G needles in 800 µL of the aforementioned reagent at 4°C. After tissue homogenization, the sample was centrifuged at 12000 x g for 10 minutes at 4°C, and the supernatant was transferred to a new tube. Following a 5-minute incubation at room temperature, 160 µL of chloroform (Alfa Aesar, J67241) was added, and the tubes were shaken for 15 seconds. The mixture was then incubated for 3 minutes at room temperature and centrifuged at 12000 x g for 15 minutes at 4°C. This process resulted in the formation of two phases, with the upper, colorless phase containing the RNA. This phase was transferred to a new tube, and 400 µL of cold isopropanol (Fisher Bioreagents, BP2618-500) was added. The tubes were incubated for 10 minutes at room temperature, followed by centrifugation at 12000 x g for 10 minutes at 4°C, leading to RNA precipitation. The supernatant was then discarded, and two washes were performed with 75% absolute ethanol (PanReac AppliChem, 131086.1211) in nuclease-free water (Thermo Scientific, R0581). Each wash involved resuspending the pellet in 75% ethanol, centrifuging at 7500 x g for 5 minutes at 4°C, and discarding the supernatant. After the second wash, the pellet was left to dry and resuspended in 30 µL of water. Finally, RNA quantity and purity were assessed using a Nanodrop 2000 spectrophotometer (Thermo Scientific). 8. Expression Array To study gene expression in tissues, the Clariom S Assay hybridization array (Thermo Fisher Scientific) was used, providing a rapid and highly reliable global overview of annotated gene expression in a given tissue. This procedure began with RNA extracted as described in section 6. RNA Extraction, and its concentration was normalized to 50 ng/µL. RNA sample quality was assessed using the TapeStation 4200 (Agilent). RNA amplification and labeling were performed using the GeneChip® WT PLUS Reagent Kit (Thermo Fisher Scientific). Amplification was carried out with 100 ng of total RNA, following the user manual's instructions for the WT PLUS Reagent Kit. The amplified cDNA was quantified, fragmented, and labeled for hybridization with the GeneChip® Clariom S Mouse Array (Thermo Fisher Scientific), using 5.5 μg of single-stranded cDNA and following the protocols outlined in the user manual. 59 RESULTS Results CHAPTER 1: Role of Galectin-3 in Alzheimer's Disease 1. Galectin-3 is involved in the morphology and size of Aβ plaques in APP model Recent studies conducted by our group on neurodegenerative diseases, such as AD and PD, have revealed the potential involvement of Gal3 in the formation of protein aggregates, including Aβ peptide plaques and Lewy bodies, in both animal models and human tissues (Boza-Serrano et al., 2019; García-Revilla et al., 2023). Given the early onset of pathological features in the 5xFAD model, this study used an animal model with a progressive pathological course (APP). This approach allowed us to study animals that reached an advanced age (18 months), thereby avoiding a saturation scenario and providing a condition more closely resembling that of human AD patients. To assess the involvement of Gal3 in the morphological characteristics of Aβ plaques, immunofluorescence techniques were used with different markers of the Aβ peptide. To analyse the structural and compaction characteristics of the plaques, thioflavin-S (ThS), a fluorophore with affinity for β-sheet structures widely used in tissues containing Aβ plaques, was employed (Boza-Serrano et al., 2019; Shin et al., 2021). As shown in Figure 1A, Aβ plaques present a compact β-sheet folded core, positive for ThS, as well as microglial cells surrounding the aggregate. To study plaque size, antibodies against Aβ were used (BAM10 and 6E10), both of which exhibited a smaller area in animals lacking Gal3 (Gal3KO/APP) (Fig. 1B and 1C). Likewise, the ThS core of plaques in Gal3KO/APP animals was smaller than that in APP animals (Fig. 1D). However, the ThS/Aβ ratio increased under conditions of Gal3 absence (Fig. 1E). Notably, the absence of Gal3 reduced the circularity of the plaque core (Fig. 1F). However, the amount of microglia surrounding the plaque did not show significant differences between APP and Gal3KO/APP animals (Fig. 1G). These results suggest that Gal3 influences the morphology of Aβ plaques, as these aggregates are smaller in Gal3-deficient animals. At this point, it remains unknown whether the effect of this protein on plaques is direct, by mediating Aβ peptide aggregation as occurs in α-synuclein aggregation events, or indirect, by modifying the action of microglia on the plaque due to its relevance in the MGnD phenotype (GarcíaRevilla et al., 2023; Keren-Shaul et al., 2017). 60 RESULTS Figure 1: Lack of Gal3 modify the morphology of Aβ plaques and diminish their size. A. Immunofluorescence on 30 µm thick sections of 18-month-old APP and Gal3KO/APP mice. Scale bar: 25 µm. B. Area of Aβ of each plaque measured on µm2 using BAM10 antibody. C. Area of Aβ of each plaque measured on µm2 using 6E10 antibody. D. Area of ThS of each plaque measured on µm2. E. Percentage of the plaque occupied by the nuclei expressed as ThS area / BAM10 area. F. Circularity of the plaque core of each plaque. G. Ratio IBA1 area / BAM10 area expressed as percentage. N = 5. Mann-Whitney. ** p < 0.01; *** p < 0.001; **** p < 0.0001. Region: Cortex. 2. The Presence of Galectin-3 Determines the Microglial Activation Profile After evaluating the morphological changes in Aβ plaques and the presence of microglia surrounding them, it was important to understand how the microglial profile might change in the presence or absence of Gal3. To this end, immunofluorescence techniques were used to assess three classic markers of microglial reactivity: CD68, TREM2, and CLEC7A. First, an immunofluorescence assay was conducted to determine whether Aβ plaques, identified by their ThS-positive core, exhibited phagocytic activity by microglia. CD68, a marker typically associated with lysosomes and indicative of phagocytic events, was used for this purpose (Hopperton et al., 2018). Images such as the one shown in Figure 2 were obtained. In the APP model, CD68-positive staining was observed around the plaque, generally presenting a punctate appearance. Interestingly, Gal3 was also found in the tissue of these animals. Notably, Gal3 staining occasionally colocalized with CD68, particularly in regions where Gal3 exhibited higher intensity and a cellular-like APP Gal3KO/APP 0 500 1000 1500 2000 8000 10000 12000 A (6E10) Area per plaque (m2) ✱✱✱ APP Gal3KO/APP 0 50 100 Nuclei ratio ThS Area / BAM10 Area per plaque (%) ✱✱✱✱ APP Gal3KO/APP 0 500 4000 6000 8000 ThS Area per plaque (m2) ✱✱ APP Gal3KO/APP 0 1000 2000 3000 4000 5000 10000 15000 20000 Aβ (BAM10) Area per plaque (m2) ✱✱✱✱ APP Gal3KO/APP 0.00 0.25 0.75 1.00 ThS Circularity ✱✱✱✱ APP Gal3KO/APP 0 50 100 150 200 500 1500 IBA1 ratio IBA1 Area / BAM10 Area per plaque (%) APP GAL3KO/APPThS IBA1 BAM10 MERGED C B A D E F G 61 RESULTS appearance. These regions are indicated by yellow arrows in Figure 2. The colocalization of both markers, along with the observed cellular morphology, suggests the presence of a relevant cellular subpopulation. Conversely, Gal3KO/APP animals exhibited a complete absence of Gal3 in the tissue and a generally reduced presence of CD68 around Aβ plaques. These findings confirm that Aβ plaques are positive for CD68, regardless of the presence or absence of Gal3. However, the occasional colocalization of CD68 with Gal3, both of which are typical microglial markers, along with the observed staining morphology, suggests a potential relationship between them. Figure 2: Lack of Gal3 reduces the amount of CD68 around Aβ plaques. Immunofluorescence on 30 µm thick sections of 18-month-old APP and Gal3KO/APP mice. Scale bar: 25 µm. Region: Cortex. Following this, it was necessary to verify that the presence of both CD68 and Gal3 corresponded to microglial cells. To this end, an immunofluorescence assay was performed on APP and Gal3KO/APP tissues, as illustrated in Figure 3. Yellow arrows indicate microglial cells positive for both CD68 and Gal3, suggesting that these markers correspond to these cells. The presence of CD68+ microglia in the Gal3KO/APP tissue shows that CD68 may be independent of Gal3. Thus, two microglial subpopulations were identified: cells positive for both CD68 and Gal3, and cells positive only for CD68. Additionally, a higher amount of CD68 was observed in APP tissues compared to Gal3KO/APP animals. Taken together, these findings suggest that microglia surrounding Aβ plaques may play an active role in modulating plaque morphology. The prominent CD68 staining indicates that microglia engage in phagocytic activity, a process in which Gal3 appears to play a significant role. Furthermore, the colocalization of CD68 and Gal3 in microglia, along with the apparent reduction of CD68 in Gal3KO/APP tissues, suggests that the absence of Gal3 may induce changes in the microglial phenotype. 62 RESULTS Figure 3: CD68 labelling colocalizes with microglia, occasionally positive for Gal3. Immunofluorescence on 30 µm thick sections of 18-month-old APP and Gal3KO/APP mice. Scale bar: 10 µm. Region: Cortex. Once the presence of CD68 in the vicinity of Aβ plaques and its relationship with microglia and Gal3 was confirmed, a quantification of CD68 and TREM2 around the plaques was performed to test the previously mentioned hypothesis. It has been demonstrated that TREM2 plays a crucial role in the transition from homeostatic microglia to the MGnD/DAM phenotype. Additionally, loss-of-function mutations in Trem2 are Alzheimer's disease risk factors, as they lead to impaired microglial migration toward plaques, reduced phagocytosis, and increased plaque size (Hou et al., 2022; Jay et al., 2017; McQuade et al., 2020). Conversely, increased TREM2 expression has been associated with the reduction in Aβ plaque size and fewer dystrophic neurites surrounding them (Lee et al., 2018). Immunofluorescence was used to quantify both markers (Fig. 4), which are characteristic of MGnD microglia, using ThS staining as an indicator of Aβ plaques (KerenShaul et al., 2017). In APP animals, CD68 and TREM2 staining was observed around the plaques, indicating the presence of DAM microglia in these areas. The absence of Gal3 led to a reduced presence of CD68 around the plaques, consistent with the findings presented in Figures 2 and 3, as well as previous studies from our group (Boza-Serrano et al., 2019). In contrast, TREM2 levels were significantly elevated in Gal3KO/APP animals. These results suggest that Gal3 plays a key role in shaping the microglial profile around Aβ plaques. Previous studies from our group identified Gal3 as an endogenous ligand of TREM2, which is considered a cornerstone of neuroinflammation and the development of the MGnD/DAM microglial phenotype (Boza-Serrano et al., 2019; Jin et al., 2008; Keren-Shaul et al., 2017; Krasemann et al., 2017; Leyns et al., 2017). The observed increase in TREM2 is highly relevant, as elevated TREM2 levels in 5xFAD mice have been reported to attenuate amyloid burden, whereas its deletion results in increased amyloid deposition (Lee et al., 2018; Parhizkar et al., 2019). On the other hand, the decrease in CD68 levels, a hallmark of phagocytic microglia, may be a consequence of either the smaller plaque size and/or increased plaque compaction in Gal3KO/APP animals, leading to a scenario with reduced phagocytosis. 63 RESULTS Figure 4: Lack of Gal3 reduces the amount of CD68 but increases the TREM2 levels around Aβ plaques in thalamus. Immunofluorescence on 30 µm thick sections of 18-month-old APP and Gal3KO/APP mice. A. CD68 area expressed as percentage of the plaque nuclei. CD68 area divided by the ThS area. B. TREM2 area expressed as percentage of the plaque nuclei. TREM2 area divided by the ThS area. C. Immunofluorescence of ThS, CD68 and TREM2. CD68 area diminish in absence of Gal3 around plaques with similar nuclei size. TREM2 area around the plaque increases when there is no Gal3 in the animal. Scale bar: 25 µm. N = 4 – 5. Mann-Whitney. * p < 0.05; **** p < 0.0001. Region: Thalamus. Finally, to complement the previously evaluated MGnD microglial markers, a third classical marker of this microglial phenotype, CLEC7A, was analysed. This marker is widely recognized for its overexpression under neuroinflammatory conditions, occurring in parallel with Gal3, and its expression levels decrease in the absence of this galectin (BozaSerrano et al., 2019; García-Revilla et al., 2022; Keren-Shaul et al., 2017). Given the APP Gal3KO/APP 0 50 100 150 200 CD68 Area per plaque (%) ✱ APP Gal3KO/APP 0 10 20 30 40 50 150 200 TREM2 Area per plaque (%) ✱✱✱✱ ThS CD68 TREM2 MERGED APP GAL3KO/APP C B A APP Gal3KO/APP 0 50 100 150 200 CD68 Area (%) ✱ APP Gal3KO/APP 0 50 150 200 TREM2 Area (%) ✱✱✱✱ 64 RESULTS observed alterations in the MGnD microglial profile, it was essential to assess the presence of CLEC7A in the vicinity of Aβ plaques through immunofluorescence. Consistent with the findings for CD68, CLEC7A expression was significantly reduced in Gal3KO/APP animals in both the cortex and the dentate gyrus (Fig. 5). Thus, we can conclude that the APP model not only exhibits a marked presence of MGnD microglial markers such as CD68, TREM2, and CLEC7A but also that this MGnD phenotype can be modulated by the presence or absence of Gal3. Given the relationship between Gal3 and TREM2, the absence of the former may lead to dysregulation of TREM2 function, which is responsible for the transition from homeostatic microglia to the MGnD phenotype. Consequently, markers such as CD68 and CLEC7A exhibit reduced tissue levels. APP ThS CLEC7A BAM10 MERGED GAL3KO/APP APP ThS CLEC7A BAM10 MERGED GAL3KO/APP B A 65 RESULTS Figure 5: Lack of Gal3 reduces the amount of Clec7a around Aβ plaques. Immunofluorescence on 30 µm thick sections of 18-month-old APP and Gal3KO/APP mice. A. Clec7a and Aβ immunohistochemistry with ThS staining on 30 µm thick brain sections. Cortex (blue), hippocampus (yellow) and thalamus (red) outlined by dotted lines. B. Aβ plaques observed under X63 objective on confocal microscope. Scale bar: 10 µm. C. Quantification of Clec7a area around plaques measured in µm2 and percentage over the ThS nuclei, both in cortex and dentate gyrus. N = 4 – 5. Mann-Whitney. **** p < 0.0001. 3. Galectin-3 Modifies the Microglial Expression Profile After confirming how Gal3 influences the MGnD microglial profile and considering the pleiotropic nature of this protein, it was inferred that the effects of its absence in Gal3KO/APP animals would not be limited solely to the phenotypic determination of microglial cells. Given the multiple functions of microglia and their relevance in the progression of neurodegenerative diseases, it was of interest to assess the transcriptomic profile of these cells. To this end, the cortex of 18-month-old APP and Gal3KO/APP animals was extracted and processed for microglial isolation. Subsequently, RNA was isolated from these cells for analysis via RNA sequencing (RNASeq). The RNASeq results were analyzed using Gene Set Enrichment Analysis (GSEA), examining a collection of gene sets (GSs) compiled in the Molecular Signature Database (MSigDB), as well as a set of selected genes derived from the transcriptomic study by Krasemann and colleagues (Krasemann et al., 2017). First, the GS corresponding to DAM microglial signature genes was evaluated. In this case, the Enrichment Score (ES) and Normalized Enrichment Score (NES) were 0.59 and 1.52, respectively. The p-value for the GS was 0.018. As shown in Figure 6, Gal3KO/APP animals exhibit a general decrease in the expression of this GS. This finding aligns with the immunofluorescence analysis of DAM markers, which demonstrated a reduction of this phenotype in Gal3KO/APP animals. APP Gal3KO/APP 0 50 100 150 200 250 300 600 900 Dentate gyrus CLEC7A Area per plaque (m2) ✱✱✱✱ APP Gal3KO/APP 0 50 100 200 400 Dentate gyrus CLEC7A Area / ThS Area per plaque (%) ✱✱✱✱ APP Gal3KO/APP 0 10 20 30 40 50 100 200 300 Cortex CLEC7A Area / ThS Area per plaque (%) ✱✱✱✱ APP Gal3KO/APP 0 50 100 150 200 600 900 Cortex CLEC7A Area per plaque (μm2) ✱✱✱✱ C APP Gal3KO/APP 0 50 100 150 200 250 300 600 900 Dentate gyrus CLEC7A Area per plaque (m2) ✱✱✱✱ APP Gal3KO/APP 0 50 100 200 400 Dentate gyrus CLEC7A Area / ThS Area per plaque (%) ✱✱✱✱ APP Gal3KO/APP 0 10 20 30 40 50 100 200 300 Cortex CLEC7A Area / ThS Area per plaque (%) ✱✱✱✱ APP Gal3KO/APP 0 50 100 150 200 600 900 Cortex CLEC7A Area per plaque (μm2) ✱✱✱✱ 66 RESULTS Figure 6: Lack of Gal3 reduces DAM genes. All core enrichment genes displayed in the heatmap. 3 APP animals and 3 Gal3KO/APP animals analysed. Subsequently, the Hallmark Gene Sets collection was evaluated. Two sets of GSs were identified, as presented in Tables 5 and 6. Table 5 lists the GSs enriched in APP animals compared to Gal3KO/APP, among which the genes expressed in response to interferonalpha stand out. On the other hand, Table 6 highlights one of the main GSs repressed in APP animals compared to Gal3KO/APP. Conversely, the NF-κβ-regulated genes in response to TNF and the genes overexpressed in response to TGFβ were among the most downregulated GSs in APP animals. Table 5: List of GSs differentially upregulated on APP over Gal3KO/APP. SIZE: Number of genes of the GS. ES: Enrichment score. NES: Normalized enrichment score. NAME SIZE ES NES NOM p-val FDR q-val FWER p-val RANK AT MAX LEADING EDGE INTERFERON ALPHA RESPONSE 90 0.51 1.52 0.005 0.019 0.094 6260 tags=56%, list=25%, signal=74% As shown in Figure 7, APP animals lacking Gal3 exhibit a decrease in genes involved in the interferon-alpha response (Fig. 7A). Some studies have suggested an interaction between Gal3 and interferon-gamma, although the relationship between Gal3 and interferonalpha remains poorly defined (Gordon-Alonso et al., 2017; Ruvolo, 2019). Therefore, it appears that Gal3 is involved in modulating the microglial response under pathological conditions, being associated with the production and release of inflammatory cytokines. LGALS3 LYZ2 SIGLEC1 SOX7 ALCAM CST7 COLEC12 ITGAX LGALS3BP LOX FABP5 GPX3 VEGFA FABP3 ZBP1 EGLN3 AXL CD68 LPL TYROBP TREM2 APOE CLEC7A LYZ1 C3 S100A1 LAMP1 APP GAL3KO/APP APP vs GAL3KO/APP 67 RESULTS Figure 7: Lack of Gal3 modify genes upregulated in response to alpha interferon proteins. Enrichment plot and heatmap. Top 30 ranked genes displayed. 3 APP animals and 3 Gal3KO/APP animals analysed. Table 6: List of GSs differentially downregulated on APP over Gal3KO/APP. SIZE: Number of genes of the GS. ES: Enrichment score. NES: Normalized enrichment score. NAME SIZE ES NES NOM p-val FDR q-val FWER p-val RANK AT MAX LEADING EDGE TNFA SIGNALING VIA NFKB 195 -0.55 -1.92 0 0 0 4400 tags=49%, list=18%, signal=59% G2M CHECKPOINT 192 -0.5 -1.75 0 0.011 0.012 4938 tags=45%, list=20%, signal=56% TGF BETA SIGNALING 54 -0.57 -1.63 0.005 0.021 0.033 4596 tags=46%, list=18%, signal=57% NOTCH SIGNALING 31 -0.61 -1.59 0.02 0.022 0.046 4186 tags=42%, list=17%, signal=50% E2F TARGETS 191 -0.45 -1.56 0 0.022 0.059 4646 tags=44%, list=19%, signal=54% MITOTIC SPINDLE 197 -0.43 -1.52 0 0.027 0.084 5522 tags=45%, list=22%, signal=58% PROTEIN SECRETION 93 -0.4 -1.28 0.066 0.213 0.568 5215 tags=38%, list=21%, signal=47% On the other hand, the absence of Gal3 leads to an increase in TGF-β signaling, which is associated with cellular survival processes. This suggests that Gal3KO/APP animals exhibit greater overexpression of genes related to cell viability and survival. IFI44 IRF7 IFIT2 SAMD9L IFI35 MX1 NCOA7 HELZ2 IFIT3 BST2 STAT2 IFITM3 TRIM14 EIF2AK2 B2M DHX58 PARP9 USP18 RSAD2 ADAR RTP4 MVB12A PARP14 HERC6 MOV10 EPSTI1 LGALS3BP GBP2 UBA7 CMPK2 APP GAL3KO/APP APP vs GAL3KO/APP 74 RESULTS Figure 12: Grn deficiency increases the amount of microglia in different brain regions. A. IBA1 immunohistochemistry on 30 µm thick brain sections from 18-month-old mice. B. Different brain regions illustrated in rows. Groups pictured in columns. 18-month-old mice. Scale bar: 20 µm. C. Number of microglia per mm2 in the different regions of study. * p < 0.05; ** p < 0.01; *** p < 0.001. 3. PGRNKO Animals Exhibit Increased Levels of Various MGnD Microglia Markers Given the prominent role of Gal3 as a MGnD microglia marker and its unavailability for evaluating this microglial phenotype in Gal3KO and PGRNKO/Gal3KO animals, alternative markers typically upregulated in this microglial subtype were considered. Therefore, immunofluorescence staining with antibodies against CLEC7A and TREM2 was employed (Fig. 13). As expected, CLEC7A levels were significantly elevated in the thalamus of PGRNKO and PGRNKO/Gal3KO animals compared to WT and Gal3KO controls. Similarly, TREM2 staining was also increased in animals lacking progranulin, indicating that the PGRNKO model harbours a substantial DAM microglial population in the thalamus. Surprisingly, despite the absence of Gal3 in PGRNKO/Gal3KO animals, CLEC7A and TREM2 levels did not differ from those observed in the thalamus of PGRNKO animals (Fig. 13). This suggests that these two MGnD markers are not altered by the presence or absence of Gal3 in the context of frontotemporal dementia. WT PGRN 0 200 400 600 Corpus callosum Nº microglia/mm2 WT Gal3KO ✱✱ ✱✱✱ WT PGRN 0 200 400 600 Thalamocortical tract Nº microglia/mm2 WT Gal3KO ✱ ✱✱ WT PGRN 0 200 400 600 Thalamus Nº microglia/mm2 WT Gal3KO ✱✱ ✱✱✱ WT PGRNKO 0 1 2 3 4 Lipofuscin Area (%) WT GAL3KO ✱✱✱✱✱✱ ✱✱✱✱ 75 RESULTS Figure 13: Gal3 deficiency could produce an increase in TREM2 levels. Immunofluorescence on 30 µm thick brain sections from 18-month-old mice. A. High magnification images of thalamus (DORsm). Scale bar: 25 µm. B. Percentage of area occupied by CLEC7A. C. Percentage of area occupied by TREM2. Two-way ANOVA. **** p < 0.0001. Region: DORsm. At this point, it becomes evident that progranulin deficiency not only triggers microgliosis in specific brain regions but also induces a shift in microglial phenotype. More specifically, focusing on the thalamus, the expression of MGnD markers such as Gal3, CLEC7A, and TREM2 is markedly elevated in the absence of progranulin, a phenomenon also observed in other pathologies such as AD and PD (Boza-Serrano et al., 2019; García-Revilla et al., 2023; Keren-Shaul et al., 2017). However, the absence of Gal3 in the frontotemporal dementia model does not affect CLEC7A levels but may increase TREM2 presence. Beyond the evaluation of the aforementioned MGnD markers, we expanded the study. Specifically, we aimed to assess the presence of GPNMB in the thalamus, an endogenous glycoprotein involved in inflammatory processes and induced by TREM2 and APOE (Krasemann et al., 2017; Saade et al., 2021). Previous studies have reported GPNMB overexpression in progranulin-deficient animals, with this increase being accompanied by elevated Gal3 levels (Huang et al., 2020). Therefore, it was of particular interest to determine whether the absence of Gal3 would alter GPNMB tissue levels through immunofluorescence analysis using anti-GPNMB and anti-Gal3 antibodies. C B A WT PGRNKO 0 1 2 3 CLEC7A Area (%) WT GAL3KO ✱✱✱✱ ✱✱✱✱ WT PGRNKO 0.0 0.2 0.4 0.6 0.8 1.0 TREM2 Area (%) WT GAL3KO 0.05020.0012 <0.0001 WT PGRN 0 200 400 600 Thalamocortical tract Nº microglia/mm2 WT Gal3KO ✱ ✱✱ 76 RESULTS Figure 14: Progranulin deficiency produces an increase in GPNMB levels, which are exacerbated in absence of Gal3. Immunofluorescence on 30 µm thick brain sections from 18-month-old mice. A. Thalamus of 18-month-old animals observed at 20 magnifications. Scale bar: 250 µm. B. Magnified images from A of the PGRNKO and PGNKO/Gal3KO animals. Scale bar: 25 µm. C. Percentage of the area of DORsm covered by GPNMB. N = 4 – 6. Two-way ANOVA. * p < 0.05. The absence of progranulin has been found to induce GPNMB upregulation in the sensorimotor region of the thalamus (DORsm), similar to what is observed with Gal3 (Fig. 14A). Thus, we identified cells with a typically microglial morphology that were positive for both Gal3 and GPNMB and colocalized with lipofuscin, an autofluorescent aggregate primarily composed of lipoproteins, present in the PGRNKO model. However, Gal3 and PGRNKO GAL3 LIPOFUSCIN GPNMB MERGED PGRNKO/GAL3KO B C A WT PGRNKO 0 5 10 15 GPNMB Area (%) WT GAL3KO ✱✱✱ 77 RESULTS GPNMB labelling did not always colocalize, as there were cellular populations that were positive for only one of these markers (Fig. 14B). Notably, in the absence of both Gal3 and progranulin, GPNMB levels were significantly elevated in the DORsm region. At this point, it was crucial to determine whether GPNMB expression was microglial, a key criterion for its consideration as a specific DAM marker. To this end, we performed immunofluorescence for this marker alongside IBA1 (Fig. 15). We observed that GPNMB labelling colocalized with IBA1, confirming that GPNMB is exclusively expressed by microglia. Likewise, GPNMB-positive cells contained lipofuscin accumulations. Thus, GPNMB emerges as a DAM microglial marker in both the presence and absence of Gal3 in the PGRNKO model. Interestingly, tissue levels of GPNMB appeared to be modulated by the presence or absence of Gal3. Recent studies have reported that both Gal3 and GPNMB are overexpressed in similar brain regions of aged Grn-/- animals (Huang et al., 2020). These findings suggest that both proteins are lysosome-associated and predominantly localized to myelin-rich regions, such as the corpus callosum and the thalamus examined in this study, indicating a close relationship between them. Here, we observed a notable increase in GPNMB levels in PGRNKO animals lacking Gal3 (Figs. 14C and 15B), particularly in the thalamus. This observation is of great interest, as recent studies highlight GPNMB as a key player in neuroinflammation due to its anti-inflammatory properties (Saade et al., 2021). Figure 15: Gal3 deficiency produces an increase of GPNMB+ microglial population. Immunofluorescence on 30 µm thick brain sections from 18-month-old mice. A. High magnification images of thalamus. Scale bar: 25 µm. B. GPNMB and IBA1 colocalized area expressed as percentage over IBA1 area. N = 4 – 6. Two-way ANOVA. ** p < 0.01. A B WT PGRNKO 0 20 40 60 80 IBA1 positive for GPNMB Area (%) WT GAL3KO ✱✱✱✱ 78 RESULTS 4. The Exacerbated Immune Response in PGRNKO Animals Is Modulated by the Absence of Galectin-3 After characterizing the MGnD/DAM microglial phenotype and given the direct relationship between microglia and the production of inflammatory cytokines in neurodegenerative diseases, we proceeded to examine the presence of several cytokines in the cortex of the animals. This is a region that is affected in the FTD and where Gal3 was also detected. For this purpose, the cortices of WT, Gal3KO, PGRNKO, and PGRNKO/Gal3KO animals were dissociated. A Meso Scale inflammatory cytokine panel was developed for the sensitive detection of IFN-γ, IL-1β, IL-2, IL-5, IL-6, IL-10, IL-12p70, KC/GRO, and TNFα (Fig. 16). Figure 16: PGRN deficiency produces an exacerbated inflammatory response in the cortex of 18month-old mice. The double lack of PGRN and Gal3 diminish the amount of some inflammatory citokines (expressed in pg/mL). N = 4 – 9. Two-way ANOVA. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. PGRNKO animals exhibit a significant increase in the expression of cytokines such as IL1β, IL-5, IL-6, KC/GRO, and TNFα. Notably, the absence of Gal3 in the PGRNKO model resulted in a reduction of IFN-γ, IL-2, IL-5, IL-6, IL-10, IL-12p70, and KC/GRO levels WT PGRNKO 0.00 0.02 0.04 0.06 IFN-γ (pg/mL) WT GAL3KO ✱✱ WT PGRNKO 0.0 0.1 0.2 0.3 0.4 0.5 IL-10 (pg/mL) WT GAL3KO ✱ ✱✱ WT PGRNKO 0 5 10 15 IL-12p70 (pg/mL) WT GAL3KO ✱✱✱ WT PGRNKO 0.0 0.2 0.4 0.6 0.8 IL-1β (pg/mL) WT GAL3KO ✱✱ ✱✱✱ WT PGRNKO 0.0 0.1 0.2 0.3 0.4 IL-2 (pg/mL) WT GAL3KO ✱✱✱ WT PGRNKO 0.0 0.2 0.4 0.6 0.8 IL-5 (pg/mL) WT GAL3KO ✱✱ ✱✱✱✱ ✱✱✱✱ ✱✱ WT PGRNKO 0 2 4 6 IL-6 (pg/mL) WT GAL3KO ✱✱ ✱✱ ✱✱✱ WT PGRNKO 0 2 4 6 8 KC/GRO (pg/mL) WT GAL3KO ✱✱✱✱✱ WT PGRNKO 0.0 0.1 0.2 0.3 0.4 0.5 TNFα (pg/mL) WT GAL3KO ✱✱ ✱ WT PGRN 0 200 400 600 Thalamocortical tract Nº microglia/mm2 WT Gal3KO ✱ ✱✱ 79 RESULTS compared to PGRNKO. Among these cytokines, IL-6, IL-12p70, and KC/GRO were the most highly expressed, standing out for their pro-inflammatory role, and their levels were elevated in the frontotemporal dementia model (Korbecki et al., 2022; Schwarz & Carson, 2022; West et al., 2019). This indicates that progranulin deficiency increases cortical levels of these cytokines. However, the Gal3-deficient PGRNKO model exhibited a reduction in these three pro-inflammatory cytokines, restoring their levels to those observed in the WT condition. This highlights the critical role of Gal3 in shaping the inflammatory profile of microglia, favouring an anti-inflammatory phenotype in its absence, which aligns with the findings from GPNMB immunofluorescence. 5. The Absence of Progranulin Induces Numerous Transcriptomic Changes in the Thalamus The various immunohistochemistry and immunofluorescence analyses conducted in PGRNKO animals identified the thalamus as one of the most relevant brain structures. Therefore, understanding the transcriptomic changes occurring in this tissue from a functional perspective is crucial. To this end, RNA was extracted from the thalamus of 18month-old male WT, Gal3KO, PGRNKO, and PGRNKO/Gal3KO animals, followed by an RNA array study (Clariom S Assay). A preliminary analysis was performed using the Transcriptome Analysis Console software (Thermo Scientific), followed by a Gene Set Enrichment Analysis to examine a collection of gene sets curated in the Molecular Signature Database (MSigDB). Specifically, the Hallmark Gene Sets collection was analysed, as it includes gene sets associated with well-defined biological processes and conditions. First, a principal component analysis (PCA) revealed how the different animals clustered based on three distinct components (PCA1, PCA2, and PCA3), which together accounted for 42.4% of the explained variance (Fig. 17). Subsequently, a volcano plot was generated for several group comparisons. Figure 18 presents the volcano plot for the PGRNKO vs WT comparison, highlighting genes of interest with a p-value below 0.05 and a fold change greater than 2. Notably, genes associated with DAM microglia, such as Gpnmb, Lyz2, Cd68, and Clec7a, were overexpressed in PGRNKO animals. In addition to these genes, others involved in developmental processes, such as Prl (prolactin) and Cga (Glycoprotein Hormones, Alpha Polypeptide), were also elevated in the absence of progranulin. Importantly, there was a marked decrease in the expression of the Grn gene, which encodes progranulin, an expected result given the knock-out nature of PGRNKO animals in this comparison. 80 RESULTS Figure 17: PCA of the samples studied by RNA microarray. WT N = 4; Gal3KO N = 4; PGRNKO N = 6; PGRN/Gal3KO N = 6. Only male animals. Figure 18: Volcano plot of the samples studied by RNA microarray. PGRNKO and WT animals analysed. Blue indicates downregulation in PGRNKO group. Red indicates upregulation in PGRNKO group. Fold change > 2; p-value < 0.05. WT N = 4; PGRNKO N = 6. Only male animals. Regarding the volcano plot for the PGRNKO vs PGRNKO/Gal3KO comparison (Fig. 19), the overexpression of several genes encoding proteins associated with developmental processes is particularly noteworthy. Specifically, genes encoding growth hormone (Gh), as well as Prl and Cga, both also overexpressed in PGRNKO compared to WT animals, were GRN GPNMB LYZ2 PRL CD68 GFAP CGA APOBEC1 CLEC7A PGRNKO vs WT 81 RESULTS significantly upregulated. Additionally, Acadl, which encodes long-chain acyl-CoA dehydrogenase and is involved in mitochondrial fatty acid oxidation, showed increased expression. Similarly, Sod3, encoding superoxide dismutase 3 and linked to the regulation of reactive oxygen species, was also overexpressed. Notably, Lgals3, the gene encoding Gal3, exhibited a marked increase in expression. Figure 19: Volcano plot of the samples studied by RNA microarray. PGRNKO and PGRNKO/Gal3KO animals analysed. Blue indicates downregulation in PGRNKO group. Red indicates upregulation in PGRNKO group. Fold change > 2; p-value < 0.05. PGRNKO N = 6; PGRNKO/Gal3KO N = 6. Only male animals. Next, an initial functional analysis was conducted using GSEA, focusing on two contrasts: PGRNKO vs WT and PGRNKO vs PGRNKO/Gal3KO. The top 50 upregulated and downregulated genes for both comparisons are summarized in Figure 20. The study then focused on the broader characteristics of PGRNKO animals compared to WT. As shown in Table 7, some of the most upregulated gene sets in PGRNKO animals are those related to apoptosis and cell proliferation, as well as immune response pathways, including complement system genes, genes upregulated in response to interferon alpha and gamma, and those regulated by NF-κβ in response to TNF. Additionally, gene sets associated with fatty acid metabolism and adipogenesis were significantly enriched. CGA PRL GH ACADL LGALS3 SOD3 PGRNKO vs PGRNKO/GAL3KO 82 RESULTS Figure 20: Heatmaps of the top 50 upregulated and top 50 downregulated genes in the comparisons PGRNKO versus WT (left) and PGRNKO versus PGRNKO/Gal3KO (right). PGRNKO vs WT PGRNKO vs PGRNKO/GAL3KO LGALS3 TIMP1 GADD45B CDKN1A BCL2L11 TNFRSF12A HMGB2 CASP8 IFNB1 KRT18 CASP3 CASP1 LMNA SAT1 RNASEL CCND2 DNAJC3 CASP9 CD14 BIRC3 PMAIP1 HMOX1 CASP7 CD44 ISG20 DPYD IL18 CTH BAX TSPO PGRNKO WT PTPN6 VAMP8 CD69 CD74 IFIT1 ISG20 ISG15 GZMA IFIT3 CASP1 IL7 SSPN IFI27 UPP1 OAS2 IFI35 SAMD9L PARP14 TNFAIP3 IFITM3 MYD88 NLRC5 EIF2AK2 MVP IRF7 MX1 GBP4 EIF4E3 PSME2 CASP3 PGRNKO PGRNKO/GAL3KO 83 RESULTS Table 7: List of GSs differentially upregulated on PGRNKO over WT. SIZE: Number of genes of the GS. ES: Enrichment score. NES: Normalized enrichment score. GENE SET SIZE ES NES NOM p-val FDR q-val FWER p-val RANK AT MAX LEADING EDGE KRAS SIGNALING UP 189 0.406 1.821 0.000 0.012 0.008 4065 tags=32%, list=19%, signal=39% P53 PATHWAY 188 0.373 1.692 0.000 0.032 0.044 4477 tags=33%, list=20%, signal=41% ALLOGRAFT REJECTION 178 0.363 1.615 0.000 0.045 0.09 3158 tags=27%, list=14%, signal=31% APOPTOSIS 157 0.368 1.609 0.000 0.037 0.097 4615 tags=34%, list=21%, signal=43% MTORC1 SIGNALING 186 0.344 1.561 0.003 0.043 0.14 5621 tags=36%, list=26%, signal=48% FATTY ACID METABOLISM 143 0.354 1.546 0.005 0.041 0.159 5649 tags=39%, list=26%, signal=52% COMPLEMENT 184 0.337 1.530 0.002 0.041 0.182 5211 tags=36%, list=24%, signal=47% ANDROGEN RESPONSE 92 0.379 1.520 0.004 0.039 0.194 3492 tags=24%, list=16%, signal=28% INTERFERON ALPHA RESPONSE 89 0.359 1.463 0.015 0.062 0.324 5729 tags=45%, list=26%, signal=61% OXIDATIVE PHOSPHORYLATION 178 0.323 1.448 0.002 0.064 0.369 8393 tags=48%, list=38%, signal=78% ADIPOGENESIS 193 0.315 1.442 0.010 0.063 0.388 6089 tags=33%, list=28%, signal=45% INTERFERON GAMMA RESPONSE 182 0.312 1.415 0.005 0.074 0.465 5208 tags=34%, list=24%, signal=44% IL2 STAT5 SIGNALING 189 0.312 1.402 0.012 0.079 0.524 4346 tags=30%, list=20%, signal=37% MITOTIC SPINDLE 196 0.306 1.388 0.010 0.083 0.573 3896 tags=22%, list=18%, signal=26% TNFA SIGNALING VIA NFKB 195 0.303 1.367 0.018 0.094 0.642 4245 tags=28%, list=19%, signal=35% BILE ACID METABOLISM 110 0.326 1.363 0.030 0.091 0.653 5559 tags=37%, list=25%, signal=50% PI3K AKT MTOR SIGNALING 103 0.330 1.352 0.042 0.094 0.687 5983 tags=32%, list=27%, signal=44% 90 RESULTS Table 8: List of GSs differentially upregulated on PGRNKO over PGRNKO/Gal3KO. SIZE: Number of genes of the GS. ES: Enrichment score. NES: Normalized enrichment score. NAME SIZE ES NES NOM p-val FDR qval FWER pval RANK AT MAX LEADING EDGE OXIDATIVE PHOSPHORYLATION 178 0.383 1.717 0.000 0.022 0.023 7463 tags=47%, list=34%, signal=70% FATTY ACID METABOLISM 143 0.350 1.532 0.002 0.093 0.176 5707 tags=37%, list=26%, signal=50% INTERFERON ALPHA RESPONSE 89 0.376 1.528 0.009 0.064 0.183 6225 tags=45%, list=28%, signal=63% INTERFERON GAMMA RESPONSE 182 0.326 1.451 0.007 0.111 0.371 6282 tags=41%, list=29%, signal=57% SPERMATOGENESIS 130 0.336 1.439 0.012 0.099 0.397 1890 tags=18%, list=9%, signal=19% APOPTOSIS 157 0.302 1.334 0.041 0.228 0.757 6107 tags=33%, list=28%, signal=46% COAGULATION 132 0.311 1.321 0.049 0.217 0.81 6149 tags=36%, list=28%, signal=50% KRAS SIGNALING UP 189 0.289 1.297 0.036 0.230 0.874 5359 tags=31%, list=24%, signal=41% a. Apoptosis and Cell Survival Regarding GSs involved in cell survival, such as apoptosis and genes overexpressed upon KRAS activation, a reduction of these pathways was observed in PGRNKO/Gal3KO animals (Fig. 27). Notably, genes encoding various caspases, including Casp1, Casp3, Casp6, and Casp7, were suppressed. Other significant genes, such as Bax, Bcl2l11, Dap, and Tnfs10, also exhibited reduced expression. These genes were overexpressed in PGRNKO tissues but displayed lower levels in both WT and PGRNKO/Gal3KO animals. The role of Gal3 in apoptotic processes and its interaction with certain pathway members, such as CASP3, has been previously described by our group (Alonso Bellido et al., 2023). Additionally, the effect of Gal3 on the KRAS pathway, which is essential for cell survival, is noteworthy. The interaction of Gal3 with this pathway has been reported in other studies (Levy et al., 2011; Ruvolo, 2019). However, the present findings highlight Gal3 as a key regulator of cell viability by modulating apoptotic and survival pathways. 91 RESULTS Figure 27: Lack of Gal3 in PGRNKO model upregulates gene sets associated with cell survivance. A. Genes mediating programmed cell death (apoptosis) by activation of caspases. B. Genes upregulated by KRAS activation. Enrichment plot and heatmap of each gene set. Top 30 ranked genes displayed. 6 PGRNKO animals and 6 PGRNKO/Gal3KO animals analysed. b. Lipid Metabolism A noteworthy finding is the overexpression of the gene set related to fatty acid metabolism in PGRNKO animals compared to PGRNKO/Gal3KO (Fig. 28), one of the pathways also overexpressed in PGRNKO animals relative to controls. Key genes include Acadl and Acadvl, involved in the oxidation of longand very-long-chain fatty acids, as well as Fabp2, which facilitates intracellular fatty acid transport. Additionally, Lgals1, encoding galectin-1, stands out due to its role in lysosomal integrity, lipid accumulation, and triglyceride metabolism (Aits et al., 2015; Fryk et al., 2024). The relationship between Gal3 and fatty acid metabolism has not been extensively described. However, it is known that Gal3 can interact with membrane lipids (Lukyanov et al., 2005). Likewise, TREM2 has been implicated in myelin debris clearance (Poliani et al., 2015). These findings suggest not only alterations in lipid processing but also potential changes in myelin within the thalamus, a moderately myelin-rich brain region. Figure 28: Lack of Gal3 in PGRNKO model upregulates genes encoding proteins involved in metabolism of fatty acids. Enrichment plot and heatmap of the gene set. Top 30 ranked genes displayed. 6 PGRNKO animals and 6 PGRNKO/Gal3KO animals analysed. LGALS3 CCNA1 GUCY2D IL1A CD69 CTH DCN GADD45B LUM ISG20 CCND2 CASP6 DAP KRT18 CASP1 HSPB1 LMNA CCND1 IFITM3 HMGB2 PMAIP1 CASP3 TNFSF10 SAT1 PTK2 ATF3 CASP7 IL18 ANXA1 CD38 PGRNKO PGRNKO/GAL3KO KCNN4 PRDM1 CBR4 TMEM100 GNG11 GALNT3 LY96 LAT2 CCND2 PPBP USH1C ANGPTL4 VWA5A PRRX1 SPRY2 SOX9 CIDEA SEMA3B TNFAIP3 AKT2 HOXD11 GYPC HSD11B1 HBEGF ETV4 PSMB8 ATG10 IRF8 TNNT2 MTMR10 PGRNKO PGRNKO/GAL3KO ACADL FABP2 HPGD HSDL2 ALDH1A1 NTHL1 LGALS1 PRDX6 INMT CEL DECR1 SUCLG2 S100A10 ECI1 CIDEA HADH ODC1 HADHB ACSS1 GPD1 ECI2 GCDH ECH1 UBE2L6 MGLL IDI1 BMPR1B ACADVL PPARA FMO1 PGRNKO PGRNKO/GAL3KO A B PGRNKO vs PGRNKO/GAL3KO PGRNKO vs PGRNKO/GAL3KO PGRNKO vs PGRNKO/GAL3KO 92 RESULTS c. Oxidative Phosphorylation Given the critical role of oxidative phosphorylation in cellular energy metabolism and its dysregulation in the absence of progranulin, it was particularly relevant to examine how this pathway was affected by the absence of Gal3 (Fig. 29). A notable finding was the reduced expression of genes such as Acadvl in PGRNKO/Gal3KO mice, previously mentioned in the context of fatty acid metabolism. Additionally, genes encoding components of the electron transport chain, including Uqcr11, Ndufa1, Ndufa2, Ndufb2, Ndufb6, Cox4i1, Cox7a2, and Cox17, exhibited decreased expression. These findings suggest that Gal3 plays a significant role in mitochondrial function. This aligns with previous studies demonstrating the interaction between Gal3 and mitochondrial membrane proteins and its involvement in mitochondrial morphology (Coppin et al., 2020). Figure 29: Lack of Gal3 in PGRNKO model upregulates genes encoding proteins involved in oxidative phosphorylation. Enrichment plot and heatmap of the gene set. Top 30 ranked genes displayed. 6 PGRNKO animals and 6 PGRNKO/Gal3KO animals analysed. d. Interferon Pathways Given the overexpression of interferon alpha and gamma pathways in the PGRNKO model and the known interaction between Gal3 and interferon gamma, it was relevant to explore how these pathways might be affected by the absence of Gal3 in an inflammatory context (Fig. 30) (Gordon-Alonso et al., 2017; Ruvolo, 2019). Our analysis revealed that the deletion of Gal3 in the PGRNKO model led to a reduced expression of interferonstimulated genes such as Herc6 and Ifi27, as well as cytokine-encoding genes like Il7 and Il15, and the transcription factor Stat4. These findings are particularly significant as they highlight the role of Gal3 as a key modulator of the inflammatory response in the frontotemporal dementia model. PGRNKO PGRNKO/GAL3KO UQCR11 NQO2 BDH2 DECR1 MRPL34 BCKDHA NDUFA1 ECI1 TCIRG1 NDUFA2 HADHB MRPL35 NDUFB2 TIMM8B ETFA ECH1 ACADVL HADHA CASP7 COX7A2 PHYH NDUFB7 MAOB POLR2F COX17 MTX2 MRPS22 COX4I1 NDUFB6 SURF1 PGRNKO vs PGRNKO/GAL3KO 93 RESULTS Figure 30: Lack of Gal3 in PGRNKO model upregulates gene sets associated with immune response. A. Genes upregulated in response to alpha interferon proteins. B. Genes upregulated in response to IFNγ. Enrichment plot and heatmap of the gene set. Top 30 ranked genes displayed. 6 PGRNKO animals and 6 PGRNKO/Gal3KO animals analysed. e. DAM and WAM Microglial Phenotypes Finally, given the prominent presence of DAM and WAM microglial phenotypes in PGRNKO animals and the role of Gal3 as a representative marker of these phenotypes in various neurodegenerative diseases, it was relevant to investigate whether these profiles exhibited significant transcriptomic changes in the absence of Gal3. Our analysis revealed a reduction in the expression levels of genes associated with DAM and WAM microglial gene sets (Fig. 31). Specifically, we observed the suppression of DAM-related genes such as Clec7a, Lyz2, Trem2, and Axl (p-value = 0.062; FDR q-value = 0.039). However, WAMrelated genes did not show a statistically significant decrease (p-value = 0.412; FDR qvalue = 0.249). These findings suggest that certain DAM-associated genes exhibit reduced expression in the PGRNKO/Gal3KO model compared to PGRNKO, indicating a potential role of Gal3 in sustaining the DAM microglial phenotype. CD74 PROCR ISG20 ISG15 IFIT3 LPAR6 CCRL2 CASP1 IL7 IFI27 IFI35 SAMD9L PARP14 IFITM3 EIF2AK2 IRF7 MX1 TRIM5 GBP4 PSME2 UBE2L6 PSMB8 PARP12 LAP3 SELL RIPK2 BST2 IFITM2 CASP8 MVB12A PGRNKO PGRNKO/GAL3KO PTPN6 VAMP8 CD69 CD74 IFIT1 ISG20 ISG15 GZMA IFIT3 CASP1 IL7 SSPN IFI27 UPP1 OAS2 IFI35 SAMD9L PARP14 TNFAIP3 IFITM3 MYD88 NLRC5 EIF2AK2 MVP IRF7 MX1 GBP4 EIF4E3 PSME2 CASP3 PGRNKO PGRNKO/GAL3KO A B PGRNKO vs PGRNKO/GAL3KO PGRNKO vs PGRNKO/GAL3KO 94 RESULTS Figure 31: Lack of Gal3 in PGRNKO model upregulates gene sets associated with reactive microglia. A. Genes encoding proteins upregulated in DAM microglia. B. Genes encoding proteins upregulated in WAM microglia. Enrichment plot and heatmap of each gene set. Most significant ranked genes displayed. 6 PGRNKO animals and 6 PGRNKO/Gal3KO animals analysed. 6. PGRNKO Animals Exhibit Lysosomal Dysfunction The lysosomal localization of granulin and the association of Gal3 with lysosomal function, along with the transcriptomic changes observed in the previous section, prompted us to assess lysosomal status in the frontotemporal dementia animal models (Aits et al., 2015; Root et al., 2023). To this end, we utilized anti-LAMP1 and anti-IBA1 antibodies. Our findings revealed that lysosomes in progranulin-deficient animals exhibit pronounced hypertrophy. Furthermore, this hypertrophy is predominantly observed in microglia (Fig. 32). B PGRNKO vs PGRNKO/GAL3KO DDO PDCD1 ITGAX APOC1 AXL CLEC7A AHNAK2 OLR1 PGRNKO PGRNKO/GAL3KO LGALS3 S100A1 S100B ITGAX AXL CLEC7A TYROBP LYZ2 TREM2 PGRNKO PGRNKO/GAL3KO 95 RESULTS B A 96 RESULTS Figure 32: Progranulin deficiency leads to lysosomal dysfunction in mouse brain at 18 months of age. Immunofluorescence on 30 µm thick brain sections from 18-month-old mice. A. DORpm. Scale bar: 25 µm. B. DORsm. Scale bar: 25 µm. C. Magnified images from B and after IMARIS software processing. D. Percentage of the area of DORpm or DORsm covered by LAMP1. N = 4 – 5. Two-way ANOVA. As shown in figures 32A and 32B, LAMP1 appears as small puncta in WT and Gal3KO animals. Since it is a lysosomal protein, this pattern indicates that lysosomes in these animals remain relatively small compared to the cell body, which is typical in a healthy organism. In contrast, in progranulin-deficient animals, this labelling appears as large granules within microglial cells located in the thalamic regions DORpm and DORsm. These hypertrophic lysosomes may serve as a potential indicator of lysosomal dysfunction (Cao et al., 2021). Additionally, colocalization between lysosomal labeling and lipofuscin autofluorescence suggests that these enlarged lysosomes contain lipofuscin. Figure 32C further illustrates the quantification of the lysosomal area within the analyzed tissue. These findings indicate that progranulin is essential for lysosomal homeostasis, as its absence leads to significant morphological changes that particularly impact microglia. Regarding the role of Gal3 in lysosomal morphology, it is noteworthy that lysosomes in Gal3KO animals resemble those in WT controls. However, PGRNKO/Gal3KO animals may exhibit a reduced degree of lysosomal hypertrophy compared to PGRNKO animals, suggesting that Gal3 could be involved in the process of lysosomal hypertrophy and subsequent dysfunction. In a previous study by our group (García-Revilla et al., 2023), we reported that Gal3 can be found within the lumen of damaged lysosomes, further supporting its potential role in conditions where lysosomal homeostasis is compromised. C D WT PGRNKO 0 1 2 3 4 5 LAMP1 - DORpm Area (%) WT Gal3KO 0.0002 0.0364 0.1415 WT PGRNKO 0 5 10 15 LAMP1 - DORsm Area (%) WT GAL3KO <0.0001 0.0003 0.0951 97 RESULTS 7. Progranulin is Essential for Lysosomal Integrity The pronounced lysosomal hypertrophy observed in progranulin-deficient animals prompted us to assess whether lysosomal integrity was preserved under these conditions. To this end, we performed immunofluorescence labelling for galectin-1 (Gal1), a lectinfamily protein that binds to the glycocalyx of ruptured lysosomes (Aits et al., 2015). Its presence in microglial cells would therefore indicate clear lysosomal dysfunction. As shown in Figure 33, Gal1 labelling is detected in microglia as well as in other cell types. However, it is particularly relevant to examine the relationship between lysosomal hypertrophy and Gal1 expression. Based on LAMP1 labelling, we can distinguish two patterns of lysosomal hypertrophy. On one hand, some microglial lysosomes appear as large, granular structures occupying the entire cytoplasm as a single entity. On the other hand, a more diffuse LAMP1 signal is observed, preserving the overall shape of the cell. In the latter case, Gal1 labelling is more pronounced. These findings suggest that the lysosomal hypertrophy resulting from progranulin deficiency is severe enough to compromise lysosomal stability, ultimately leading to rupture and subsequent dysfunction. Figure 33: Progranulin deficiency leads to failure in the maintenance of lysosomal integrity. Immunofluorescence on 30 µm thick brain sections from 18-month-old WT and PGRNKO mouse in the thalamic area. Scale bar: 15 µm. 8. Proteomic Profile Alterations in White Matter Due to the Absence of Progranulin and Galectin-3 Given the lysosomal alterations observed both through immunofluorescence and transcriptomic changes in the thalamus, as well as their relationship with lipids, we proceeded with the extraction of white matter from the thalamocortical tract and corpus callosum of 18-month-old animals. Following this, protein extraction was performed, and proteomic analysis was conducted using mass spectrometry. 98 RESULTS The sPLS-DA (sparse Partial Least Squares Discriminant Analysis) method allowed us to evaluate how the analysed animals are distributed across two variables, which account for 15% of the explained variance (component 1) and 5% of the explained variance (component 2). The results of this analysis are presented in Figure 34, which illustrates how animals of each genotype cluster distinctly within these two components. Figure 34: sPLS-DA of 18-month-old mice. White matter from thalamocortical tract and corpus callosum. Components represented = 2. Variables for component 1 = 10. Variables for component 2 = 20. WT N = 10; Gal3KO N = 9; PGRNKO N = 24; PGRNKO/Gal3KO N = 8. Next, differentially expressed proteins were identified in PGRNKO animals compared to WT controls (Fig. 35). A total of 16 upregulated proteins and 240 downregulated proteins were found in the frontotemporal dementia model. Among the upregulated proteins, notable examples include lysozyme 2 (LYZ2) and complement protein C4B, both implicated in inflammatory processes, as well as S100A4, CD44, and CRADD, which are associated with cell cycle regulation and apoptosis. All these processes had already been observed as upregulated in the thalamic transcriptomic studies presented earlier in this manuscript. 99 RESULTS Figure 35: Volcano plot of the samples studied by proteomics. PGRNKO versus WT comparison. Differentially upregulated proteins in PGRNKO represented in pink. Differentially downregulated proteins in PGRNKO represented in blue. Fold change > 2; p-value < 0.05. WT N = 10; PGRNKO N = 24. The functional analysis of differentially expressed proteins between PGRNKO animals and WT controls revealed the involvement of various biological processes, with particular emphasis on those related to the regulation of hydrolase activity (Fig. 36 and 37). Hydrolases are a group of enzymes present in lysosomes. More specifically, one of the most relevant processes identified was the regulation of lipid catabolism, as well as the regulation of proteolysis and apoptosis. This indicates that the absence of progranulin leads to dysregulation of lysosomal processing, which could be a consequence of the lysosomal hypertrophy and rupture observed through immunofluorescence techniques, as well as certain gene sets analysed at the transcriptomic level. From a more detailed perspective, the downregulation of AKT1 and AKT2 proteins stands out. These proteins are part of the Akt complex, which not only acts as a mediator of cell survival but also plays a crucial role in autophagy regulation and is associated with lysosomes. Hirata and colleagues (Hirata et al., 2018) observed that both AKT1 and AKT2 interact with VRK2, a member of the VRK protein kinase family. Through this interaction, the VRK2-Akt complex regulates lysosome size, acidification, and degradation of phagocytosed material, among other processes. Consequently, the reduction of Akt or VRK2 results in the induction of autophagy, as well as the dysregulation of other processes in which they are involved, such as lysosomal acidification, which is essential for proper hydrolase activity. PGRNKO vs WT 106 RESULTS Figure 42: Progranulin deficiency increases phagocytic pouches in DORsm. A. Immunofluorescence on 30 µm thick brain sections from 18-month-old mice. Yellow arrow indicates a neuron positive for lipofuscin being phagocyted by microglia. Scale bar: 20 µm. B. Number of phagocytic pouches present in DORsm. C. Percentage of microglial cells that present phagocytic pouches in DORsm. N = 4 – 5. Two-way ANOVA. * p < 0.05; ** p < 0.01. 11. The Absence of Gal3 Reduces Lipofuscinosis in PGRNKO Animals The findings presented earlier in this manuscript, such as the pronounced microglial reactivity observed in PGRNKO animals, as well as the evident lysosomal hypertrophy and dysfunction, prompted us to evaluate one of the main histopathological characteristics of progranulin-deficient animals: lipofuscinosis. It has been described that the absence of progranulin compromises protein and lipid homeostasis, leading to the accumulation of lipofuscin (Root et al., 2023). Due to its autofluorescent properties, lipofuscin can be visualized through confocal microscopy (Miller & Zachary, 2017; Stillman et al., 2023). Its analysis in 18-month-old mice (Fig. 43) revealed a clear accumulation of lipofuscin in the brains of PGRNKO animals compared to WT animals, with the thalamus being one of the structures where lipofuscinosis is most pronounced. Furthermore, within the subthalamic regions, this phenomenon is particularly prominent in the DORsm region compared to DORpm in PGRNKO animals (Fig. 43E). That is, in the area where pronounced microglial reactivity and lysosomal dysfunction were observed, there is also a marked accumulation of lipofuscin. Additionally, lipofuscin is absent in regions predominantly associated with white matter, such as the thalamocortical tract, fimbria, and corpus callosum. This suggests that lipofuscin originates in neurons, accumulating in neuronal somas before being relocated to the interior of microglial cells through phagocytic processes. IBA1UCH-L1 LIPOFUSCIN MERGEDWT GAL3KO PGRNKO PGRNKO/GAL3KO A B C WT PGRNKO 0 10 20 30 40 50 Phagocytic pouches (%) WT GAL3KO ✱ ✱ WT PGRN 0 200 400 600 Thalamocortical tract Nº microglia/mm2 WT Gal3KO ✱ ✱✱ 107 RESULTS In this study, we observed for the first time that the pronounced lipofuscinosis present in PGRNKO animals is attenuated in PGRNKO/Gal3KO subjects, with their levels approximating those of WT and Gal3KO conditions (Fig. 43B). In other words, the mere absence of Gal3 is sufficient to significantly reduce one of the main histopathological hallmarks of the PGRNKO model. Likewise, lipofuscin aggregates exhibit differences in morphology, with PGRNKO mice displaying larger accumulations, whereas the other conditions exhibit a more diffuse and punctate staining pattern. A recent study from our group identified Gal3 in lipofuscin-associated lysosomes in the brains of Parkinson’s disease patients as well as in controls (García-Revilla et al., 2023). This, combined with the association of Gal3 with lysosomal damage and Gal3-positive microgliosis in regions where lipofuscin is abundant, establishes a close relationship between Gal3 and lipofuscinosis. However, the mechanisms by which Gal3 influences lipofuscinosis remain unknown. A B C D WT PGRNKO WT PGRNKOGAL3KO PGRNKO/GAL3KO Thalamus DORsm DORpm WT PGRNKOGAL3KO PGRNKO/GAL3KO Thalamus WT PGRNKO 0 1 2 3 4 Lipofuscin Area (%) WT GAL3KO ✱✱✱✱✱✱ ✱✱✱✱ 108 RESULTS Figure 43: Progranulin deficiency causes the accumulation of lipofuscin in mouse brain at 18 months of age. Lipofuscin burden is reduced in the absence of Gal3. A. Lipofuscin emission in brain tissue in wild type and PGRN knock-out animal. Thalamus is delineated by orange dotted line. B. Area of the thalamus covered by lipofuscin. N = 3 – 8. PGRNKO mice accumulates more lipofuscin than WT, GAL3KO and PGRN/GAL3KO mice. C. Magnified images from A, focused on thalamus. Scale bar: 250 µm. D. Lipofuscin presence on thalamus observed at 63 magnifications. Scale bar: 20 µm. E. Area of subthalamic regions covered by lipofuscin. N = 4 – 5. Two-way ANOVA. ** p < 0.01; **** p < 0.0001. CHAPTER 3: Modification of Tau Pathology by the Absence of Progranulin 1. The Absence of Progranulin Modifies Tau Pathology in the P301S Model The P301S animal model is characterized by the presence of intracellular aggregates of hyperphosphorylated tau protein, a process typical of neurodegenerative diseases such as AD and certain types of FTD. Various studies indicate that this tau accumulation results in neurodegeneration and gliosis in different regions, with the hippocampus being one of the most affected (Mohamed et al., 2024; Takahashi et al., 2024). In the context of tauopathies, microglia have been identified as the main drivers of neurodegeneration and disease progression, rather than neurotoxicity induced directly by tau accumulation. In fact, studies have shown that depletion of microglial cells blocks neurodegeneration (Shi et al., 2019). Furthermore, Shi and colleagues (Shi et al., 2019) suggest that neurodegeneration in tauopathies is dependent on microglial activation status. Previous studies have demonstrated that tau phosphorylation occurs through the activation of tau kinases via IL-1 signaling, a process mediated by microglial activation (Kitazawa et al., 2005; Li et al., 2003). In this regard, granulin has been shown to play a key role in limiting microglial reactivity, with its deficiency leading to prolonged microglial activation and impaired neurogenesis (Zambusi et al., 2022). Given the importance of progranulin in modulating the microglial response and considering that microglia are the primary mediators of protein aggregate clearance in various pathologies, we hypothesized that the absence of progranulin might enhance the clearance of tau protein in the hippocampus of P301S mice (Boza-Serrano et al., 2019; Janda et al., 2018; Martens et al., 2012; Reifschneider et al., 2022). To investigate this, 12month-old animals were perfused, and their brains were fixed and sectioned using a WT PGRNKO 0 2 4 6 Lipofuscin Area (%) WT Gal3KO ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ Thalamus DORpm WT PGRNKO 0 1 2 3 4 Lipofuscin Area (%) WT Gal3KO ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ ✱ Thalamus DORsm E WT GAL3KO PGRNKO PGRNKO/GAL3KO 0 2 4 6 Lipofuscin Area (%) DORpm DORsm ✱✱✱✱ ns 0.0551 WT PGRNKO 0 1 2 3 4 Lipofuscin Area (%) WT GAL3KO ✱✱✱✱✱✱ ✱✱✱✱ WT PGRNKO 0 1 2 3 4 Lipofuscin Area (%) WT GAL3KO ✱✱✱✱✱✱ ✱✱✱✱ 109 RESULTS cryostat. The presence of hyperphosphorylated tau aggregates was assessed through immunohistochemistry using the AT8 antibody, which binds to tau phosphorylated at Ser202 and Thr205. Figure 44 shows that at 12 months of age, WT and PGRNKO mice lack hyperphosphorylated tau, with AT8 staining being completely absent. In contrast, P301S and PGRNKO/P301S animals exhibit high levels of hyperphosphorylated tau in different hippocampal regions. More specifically, the staining is most prominent in regions where most neuronal somas are located (Fig. 44A and 44B), such as the dentate gyrus, CA1, CA2, and CA3, indicating the presence of aggregates in both granular and pyramidal cells. Additionally, AT8 staining is diffusely distributed throughout the hippocampus, with taupositive fibres or bundles being distinguishable (Fig. 44A). However, both P301S and PGRNKO/P301S models exhibit marked phenotypic variability, with some animals showing no tau pathology and others presenting significant accumulation. This variability is illustrated in figure 44C, where animals from the same group are distributed into different subgroups based on their levels of hyperphosphorylated tau. This figure suggests not only the existence of distinct subgroups but also differences in the area occupied by phosphorylated tau in the dentate gyrus between P301S and PGRNKO/P301S animals. Consequently, a double-blind classification of the animals was performed based on immunohistochemistry images. Based on tau pathology in the dentate gyrus, piriform cortex, and surrounding areas, animals were classified into four pathological subgroups (Types 1–4; Fig. 45A), a method previously used in other studies (Shi et al., 2017). "Type 1" animals lack pathology in both the hippocampus and cortex. "Type 2" animals exhibit tau aggregates in the piriform cortex with very faint or absent staining in the dentate gyrus. "Type 3" animals show moderate staining in both the cortex and dentate gyrus, whereas "Type 4" animals present a high burden of hyperphosphorylated tau in both regions. The "Type 1" group primarily consists of WT and PGRNKO animals, with occasional cases of P301S and PGRNKO/P301S. The remaining subgroups include only P301S and PGRNKO/P301S animals. 110 RESULTS Figure 44: Progranulin deficiency reduces phospho-tau burden in P301S mice at 12 months of age. A. AT8 immunohistochemistry on 30 µm thick brain sections. Hippocampus is outlined by a blue dotted line. Dentate gyrus is delineated by an orange dotted line. Dark blue boxes indicate images magnified on B. Scale bar 250 µm. B. AT8 immunohistochemistry of dentate gyrus. Orange arrows show AT8 positive labelling. Scale bar: 20 µm. C. Reduced p-Tau load on P301S animals compared to PGRNKO/P301S animals. N = 13 – 18. t-test. *p < 0.05. Region: Dentate gyrus. P301S P301S/PGRNKO 0 5 10 15 AT8 p-Tau area (%) ✱ A B C 111 RESULTS Figure 45: Classification of tau pathology based on phospho-tau burden in dentate gyrus and cortex piriform (CPi). A. AT8 immunohistochemistry on 30 µm thick brain sections. Scale bars: 250 µm and 50 µm. B. Distribution of animals according to tau pathology. C. Distribution of P301S and PGRNKO/P301S in the different groups of pathology considering sex differences. N = 20 P301S (12 male, 8 female), 12 PGRNKO/P301S (8 male, 4 female), 5 PGRNKO (3 male, 2 female) and 4 WT (3 male, 1 female). Chi-Square p-value = 0.3136 (B) and 0.0021 (C). Despite the noticeable trend observed in Figure 45B, where subgroups 2 and 4 appear to be more populated by P301S animals and subgroups 1 and 3 by PGRNKO/P301S animals, no statistically significant differences were found. However, a sex-related effect was identified, with males being more prone to exhibiting higher levels of hyperphosphorylated tau than females. Interestingly, both P301S and PGRNKO/P301S females rarely present Type 4 pathology. Altogether, these findings suggest that the absence of progranulin leads to a reduced accumulation of hyperphosphorylated tau in the P301S model. Other studies have indicated that progranulin, expressed in both neurons and microglia, plays a crucial role in lysosomal function and microglial response (Tanaka et al., 2013). Therefore, the reduction in tau levels may be attributed to changes in the microglial population within the hippocampus or a differential processing of tau aggregates in this region. B A C Type 1 50.00% Male 50.00% Female Total=14 Type 2 41.67% Male 58.33% Female Total=12 Type 3 100.00% Male Total=5 Type 4 90.00% Male 10.00% Female Total=10 Type 4 80.00% P301S 20.00% PGRNKO/P301S Total=10 Type 3 40.00% P301S 60.00% PGRNKO/P301S Total=5 Type 2 66.67% P301S 33.33% PGRNKO/P301S Total=12 Type 1 14.29% P301S 21.43% PGRNKO/P301S 35.71% PGRNKO 28.57% WT Total=14 112 RESULTS 2. PGRNKO/P301S Animals Exhibit Reduced Microgliosis in the Hippocampus After assessing the presence of hyperphosphorylated tau protein, the next step was to determine whether these conditions led to a microgliosis scenario that could explain the differential accumulation of tau. To this end, IBA1 immunohistochemistry was performed (Fig. 46). Figures 46A and 46B show that microglia preferentially localize in the hippocampus of P301S and PGRNKO/P301S animals, possibly in response to tau accumulation in this region. However, due to the high heterogeneity of the studied models, no significant differences were generally found in the amount of microglia per square millimetre or in the area occupied by microglia (Fig. 46C and 46D). Nevertheless, when exclusively analysing P301S and PGRNKO/P301S animals belonging to subgroups 3 and 4, along with the controls, it becomes evident that microgliosis is more pronounced in P301S animals (Fig. 46E). This indicates that the absence of progranulin significantly reduces the amount of hyperphosphorylated tau in the brains of P301S animals. This reduction may decrease the microgliosis associated with its accumulation. Despite this, PGRNKO animals exhibit notable microgliosis in the thalamus at 12 months of age. This phenomenon does not appear to be caused by an accumulation of phosphorylated tau in the thalamus. This microglial response remains present in the PGRNKO/P301S model, even though hippocampal microglia levels are reduced. In other words, the PGRNKO/P301S model shows less hippocampal pathology while retaining the pathology associated with lipofuscinosis and thalamic neuronal death (Fig. 47). Interestingly, microgliosis, a hallmark of progranulin-deficient animals, is not present in the hippocampus, a finding supported by other studies (Takahashi et al., 2024). However, it is observed in the P301S model, suggesting that the high presence of microglia in the region is primarily due to phosphorylated tau accumulation. Therefore, it can be inferred that the lower tau burden in the dentate gyrus may result in reduced associated microgliosis. However, it remains to be determined whether the microgliosis observed in the hippocampus of PGRNKO/P301S animals is a cause or a consequence of the reduced tau levels. 113 RESULTS Figure 46: Progranulin deficiency may cause the reduction of hippocampal microglia in mouse brain at 12 months. A. IBA1 immunohistochemistry on 30 µm thick brain sections. P301S and PGRNKO/P301S tissues belongs to Type 4 subgroup. B. Magnified images from A focusing on hippocampus, delineated by the blue dotted line. Scale bar: 250 µm. C. Magnified images from B. Scale bar: 25 µm. D. IBA1 area expressed in percentage occupied on the hippocampus. N = 4 – 20. E. Number of microglia per mm2 in the hippocampus. N = 4 – 20. F. Number of microglial cells per mm2 considering just P301S animals with burden of phospho-tau. N= 4 – 11. Two-way ANOVA. * p < 0.05; *** p < 0.001. Region: Hippocampus. WT PGRNKO PGRNKO/P301SP301S PGRNKO/P301S WT P301SPGRNKO PGRNKO/P301SWT P301SPGRNKO WT PGRNKO 0 10 20 30 40 IBA1 Area (%) WT P301S ✱ WT PGRNKO 0 200 400 600 Number of microglía per mm2 WT P301S WT PGRNKO 0 200 400 600 800 Number of microglía per mm2 WT P301S 0.0044 0.1042 A B C D E F WT PGRN 0 200 400 600 Thalamocortical tract Nº microglia/mm2 WT Gal3KO ✱ ✱✱ 114 RESULTS Figure 47: Progranulin deficiency preserves microgliosis in thalamus despite reduction of hippocampal microglia in mouse brain at 12 months of age. AT8 and IBA1 immunohistochemistry on 30 µm thick brain sections. Scale bar: 50 µm. Regions: Hippocampus and thalamus. 3. Effect of Progranulin Deficiency on the Behaviour of the P301S Model After identifying some histological features of P301S animals and how these are altered in the absence of progranulin, the animals' performance in various behavioural tests was assessed (Fig. 48). Both the P301S and PGRNKO models represent different approaches to frontotemporal dementia, either due to tau protein accumulation or progranulin deficiency. Therefore, understanding how different histopathological traits impact mouse behaviour is of great interest. First, depressive-like behaviour was evaluated using the tail suspension test. In this test, the animal is suspended by its tail, and the time spent immobile is measured, with increased immobility indicating depressive-like behaviour. It was observed that both Hippocampus WT PGRNKO P301S PGRNKO/P301S AT8 IBA1 AT8 IBA1Thalamus 115 RESULTS P301S and PGRNKO animals at 12 months of age displayed longer immobility times compared to WT animals, indicating that both models exhibit depressive-like behaviour. Next, the open field test was conducted. This test assesses not only the animal's motor capacity but also its anxiety and stress levels. Unhealthy animals typically show reduced activity in this test, which could indicate either motor deficits or increased anxiety/stress. As expected, WT animals displayed the highest overall activity. Although no significant differences were found in some comparisons, a trend toward reduced activity was observed in P301S and PGRNKO animals. Moreover, statistically significant differences were found between PGRNKO/P301S and WT animals. This suggests that tau pathology in the absence of progranulin may lead to anxiety and/or motor impairments, which should be further evaluated using tests such as the elevated plus maze and the rotarod test. To further investigate potential anxiety-related effects, the elevated plus maze test was performed (Fig. 48C). In this test, a healthy animal typically spends more time in the enclosed arms of the maze, avoiding the open arms. It was found that P301S animals, and especially PGRNKO/P301S animals, spent more time in the open arms compared to WT animals. In contrast, no differences were observed between WT and PGRNKO animals. This indicates that anxiety levels are elevated in the P301S model and that the absence of progranulin further exacerbates this behaviour. Thus, these findings, together with previous test results, suggest that P301S animals exhibit anxiety-related behaviour, which is more pronounced in double-mutant PGRNKO/P301S animals. Social behaviour was assessed using a social interaction test. This test measured the time the subject spent in different environments when exposed to familiar and unfamiliar mice, as well as the number of interactions with these animals. A healthy subject would typically spend more time interacting with the unfamiliar mouse. However, in this study, no differences were found between groups in terms of time spent interacting with either the familiar or unfamiliar mouse, nor in the number of approaches to both (data not shown). This indicates that tauopathy and/or progranulin deficiency do not impair the animals' social behaviour. 122 DISCUSSION Additionally, the administration of soluble TREM2 in the hippocampus of 5xFAD mice has a similar effect, reducing Aβ plaque burden and decreasing the number of dystrophic neurites in the injected tissue compared to controls (Zhong et al., 2019). On the other hand, a decrease in TREM2 is a risk factor in the development of the pathology, leading to a deficit in microglial migration to Aβ plaques, poor synapse phagocytosis, and increased amyloid burden (Hou et al., 2022; McQuade et al., 2020; Parhizkar et al., 2019). Some studies have demonstrated that the absence of TREM2 in the 5xFAD model reduces the microglial population associated with Aβ plaques, causes structural alterations in the plaques, and increases the presence of dystrophic neurites around these aggregates (Wang et al., 2016). Furthermore, genome-wide association studies (GWAS) have observed that different cohorts of Alzheimer's disease patients present loss-of-function mutations in genes coding for immune response receptors. Among them, the R47H (arginine 47 to histidine) variant of the gene encoding TREM2 stands out (Guerreiro & Hardy, 2014; Karch & Goate, 2015). Therefore, some studies suggest TREM2 agonists as a therapeutic approach for TREM2 signaling in cases of TREM2 deficiency (Schlepckow et al., 2023). One example is the anti-TREM2 antibody 4D9, whose main effect at the histological level is the reduction of the Aβ plaque halo, something also observed in our approach through the absence of Gal3 (Schlepckow et al., 2020). Another example is the anti-TREM2 antibody AL002c, whose systemic administration in animal models leads to a reduction in filamentous plaque burden, as well as in dystrophic neurites, has a positive impact on exploratory and risk-taking behavior in mice, and alleviates the inflammatory response (Wang et al., 2020). 4. The Absence of Gal3 Contributes to the Reduction of Dystrophic Neurites An important histopathological feature of Alzheimer's disease is the formation of dystrophic neurites. These dystrophies originate from Aβ plaques, with these aggregates being the main cause of neuronal/axonal damage in Alzheimer's disease models. In fact, β-amyloid pathology precedes the formation of dystrophic neurites (Trujillo-Estrada et al., 2014). Over the last decade, the hypothesis that the microglia surrounding Aβ plaques functions as a physical barrier, preventing their growth, has gained support. However, regions not covered by microglia become key points for the fibrillation of Aβ peptides and present a marked neurotoxic environment, leading to the formation of dystrophies (Condello et al., 2015). The preservation of this barrier largely depends on the proper functioning of TREM2, as loss-of-function mutations in this gene cause a breakdown in microglial barrier integrity. This is accompanied by reduced plaque compaction, an increase in the presence of axonal dystrophies, and represents a risk factor for disease development (Yuan et al., 2016). Due to the close relationship between TREM2 and Gal3, as well as the previously mentioned increase in TREM2 in our tissues lacking Gal3, we were particularly interested in examining how the presence or absence of Gal3 could result in a change in the neurotoxic tissue environment (Boza-Serrano et al., 2019). In this study, we observed that the presence of dystrophic neurites, evaluated by LAMP1 123 DISCUSSION immunofluorescence, was reduced in conditions where Gal3 was absent. Therefore, our study demonstrates that the absence of Gal3 significantly influences the degree of compaction of Aβ plaques, an effect that could be related to the induction of TREM2 and the subsequent reduction in plaque toxicity in terms of neuritic dystrophies. These findings reinforce the therapeutic potential of Gal3 inhibitors in Alzheimer's disease. 5. Microglial Phenotype in the PGRNKO Model Microglia plays a leading role in neurodegenerative disease scenarios, performing functions such as tissue repair in response to damage and leading processes like neuroinflammation and phagocytosis (Gao et al., 2023). Given the notable presence of these cells surrounding amyloid plaques and regions with tau accumulation, it was of interest to assess whether PGRNKO animals, a model of frontotemporal dementia, exhibit a phenomenon of microgliosis in any brain regions. In this study, a marked gliosis was demonstrated, via immunohistochemistry, in areas associated with white matter, such as the corpus callosum, the thalamocortical tract, and notably, the DORsm region of the thalamus. Previous studies had already identified a significant increase in the microglial population in the thalamus of progranulin-deficient animals (Ghoshal et al., 2012; Lui et al., 2016). This study shows that the prominent presence of microglia in these brain areas is accompanied by a significant overexpression of various markers associated with neurodegeneration, such as CD68, CLEC7A, TREM2, GPNMB, and, notably, Gal3. Thus, the use of progranulin and Gal3 knockout animals (PGRNKO/Gal3KO) becomes of great interest. Although microgliosis is a key feature in certain brain regions in PGRNKO animals, the absence of Gal3 in this model does not lead to changes in the amount of microglia present in these regions. However, due to the close relationship between Gal3 and TREM2, and the relevance of TREM2 in the appearance of the DAM/MGnD phenotype; and considering that Gal3 is one of the main genes overexpressed in this phenotype, it is possible that microglial activity is altered, leading to phenotypic changes that may affect the progression of the disease (Boza-Serrano et al., 2019; Keren-Shaul et al., 2017; Krasemann et al., 2017). In fact, modulation of microglial activity in neurodegenerative contexts is currently being explored as a potential therapeutic approach for various diseases (Gao et al., 2023). In the last decade, several studies have shown that microglia produce elevated levels of Gal3 under conditions of neuroinflammation (Boza-Serrano et al., 2014, 2019; Burguillos et al., 2015; García-Revilla et al., 2023). Generally, these studies have focused on diseases such as Alzheimer's disease or Parkinson's disease. More recently, some subtypes of frontotemporal dementia have also been shown to exhibit elevated levels of Gal3 in both brain tissue and cerebrospinal fluid (Borrego–Écija et al., 2024). In this study, we describe the presence of Gal3 in the brains of PGRNKO animals. Overall, Gal3 is localized in specific brain structures, with the thalamic region and the 124 DISCUSSION thalamocortical tract being the main areas of interest. Additionally, this protein is also found abundantly in the cortex and corpus callosum, with the hippocampus and hypothalamus being regions where Gal3 is present in lower amounts, although still higher than in WT animals. Interestingly, Gal3 deposition in 18-month-old control tissue occurs diffusely in white matter regions, while in the absence of progranulin, the labelling is much more intense and exhibits a star-shaped, typically cellular morphology. This pattern suggests that Gal3 expression is linked to abundant cells in these regions, and supported by previous studies, as well as the previously observed abundance of microglia in these areas, it is reasonable to conclude that microglia are responsible for the abundant presence of Gal3 in PGRNKO animals (Burguillos et al., 2015). Given all the above, we focused on evaluating the microglial phenotype present in the PGRNKO model, assessing whether there are similarities with the main microglial phenotypes observed in neurodegenerative conditions, such as MGnD and DAM, and understanding how the presence or absence of Gal3 could affect phenotypic alterations in microglia in the absence of progranulin. To do this, we approached the study from both a transcriptomic perspective, using an RNA array, and a proteomic approach, using confocal microscopy with markers such as TREM2, CLEC7A, GPNMB, and CD68 (Ferwerda et al., 2008; Fitz et al., 2020; Krasemann et al., 2017; Mecca et al., 2018; Poliani et al., 2015; Root et al., 2023; Saade et al., 2021; Shah et al., 2008). We found that the absence of progranulin results in an increase in markers such as TREM2, CLEC7A, CD68, GPNMB, and Gal3 in the thalamic region and, in some cases, also in the thalamocortical tract. Additionally, the expression of genes encoding for these markers, along with others like Lyz2, Tyrobp, and Spp1, was increased in the thalamus of PGRNKO animals compared to controls. Thus, the presence of the MGnD/DAM microglial phenotype can be confirmed, being localized in certain brain structures, with the DORsm region showing the most prominent expression of these markers. Given the roles of the evaluated markers and the intrinsic characteristics of the MGnD/DAM phenotype, we can deduce that there is significant microglial activation in the thalamus, accompanied by inflammation and phagocytosis processes. Following this, and considering the previously described prominent presence of Gal3 in multiple regions of the PGRNKO mouse brain, it was of interest to evaluate how the presence of different MGnD/DAM markers might be modified by the absence of Gal3. In this study, we observed that many of the previously mentioned MGnD/DAM microglial markers (TREM2, CLEC7A, CD68, GPNMB) were also elevated in the thalamus of the PGRNKO/Gal3KO model compared to WT control animals, further confirming the presence of the MGnD/DAM phenotype in these animals as well. At the proteomic level, we found that the absence of Gal3 leads to an overexpression of certain genes associated with DAM, such as GPNMB and CD68, both of which were already elevated in the PGRNKO model. Notably, GPNMB is the most overexpressed gene in progranulin-deficient animals in our study, similar to findings in other studies focused 125 DISCUSSION on other neurodegenerative diseases (Krasemann et al., 2017). Recently, the important role of GPNMB in the neuroinflammation process has been described. GPNMB can modulate the PI3K/Akt and p38 MAPK pathways, which are fundamental not only in inflammation but also in cell survival, apoptosis, cell differentiation, stress response, and glucose transport (Huang et al., 2021; Ping et al., 2025). Additionally, GPNMB knockdown animals exhibit elevated NF-κβ expression, a central transcriptional regulator of inflammation (Ping et al., 2025; Saade et al., 2021). Thus, the overexpression of this gene and the increased protein presence in the thalamus of PGRNKO/Gal3KO animals suggest an attenuation of the pro-inflammatory response. On the other hand, CD68 expression is also elevated in the thalamus of PGRNKO/Gal3KO animals compared to PGRNKO animals. Due to its relationship with phagocytic processes, this suggests an increase in phagocytosis in Gal3-deficient animals (Hopperton et al., 2018). It is important to emphasize that the presence of CD68 is not limited to the thalamus but is also located in the thalamocortical tract, a region where Gal3 is also highly present. This is not the case with GPNMB, which is primarily found in the DORsm region of the thalamus and is sparsely present in other brain regions. This could suggest the presence of different microglial subpopulations in the PGRNKO model. At the transcriptomic level, we found that the DAM phenotype gene set was reduced in the absence of Gal3 in the PGRNKO model. Thus, it is concluded that the absence of Gal3 in neurodegeneration conditions leads to a change in the microglial phenotype, with a reduction in the expression of some genes such as Itgax, Lyz2, or Axl, and an increase in the thalamic presence of GPNMB and CD68. 6. The absence of progranulin reduces tauopathy in P301S animals. A common histopathological feature of Alzheimer's disease and frontotemporal dementia is the development of marked tauopathy (Reiman & Caselli, 1999; Silva et al., 2019). The formation of intracellular aggregates of hyperphosphorylated tau protein is associated with a series of phenomena such as neuronal damage, gliosis, and neuroinflammation, with the hippocampus being one of the most affected regions (Mohamed et al., 2024; Takahashi et al., 2024). Another common feature between both pathologies lies in the gene encoding progranulin, whose loss-of-function mutations represent a risk factor for Alzheimer's disease as well as a genetic cause of frontotemporal dementia (American Psychiatric Association, 2022; Rhinn et al., 2022; Sociedad Española de Neurología, 2018). The deficiency of progranulin creates a microglial reactivity scenario (Cagnin et al., 2004; Root et al., 2023; Tanaka et al., 2013). Some authors have described that microglia play a relevant role in the development of tau pathology, with the state in which these cells are being a key element (Maphis et al., 2015; Shi et al., 2019). In this regard, it has been described that microglial activation promotes tau phosphorylation via the activation of kinases through the IL-1 pathway. Additionally, it was found that the absence of microglia slows down the progression of tauopathy (Kloske et al., 2024; Shi et al., 2019). Therefore, this study evaluated the hypothesis of whether an activated microglial state due to the 126 DISCUSSION lack of progranulin could result in a change in tau protein accumulation in the hippocampus. In this study, it was observed that P301S animals accumulate significantly hyperphosphorylated tau protein in different brain regions. One of the most affected regions is the hippocampus, where phosphorylated tau aggregates were found in the dentate gyrus, as well as in CA1, CA2, and CA3. Furthermore, this protein can also be found accumulated in different cortical regions, with the entorhinal cortex and piriform cortex being two highlighted areas. However, it was observed that PGRNKO/P301S animals show less tau accumulation in the hippocampus. As described earlier in this text, the absence of progranulin leads to a change in the microglial phenotype, causing microglia to leave its homeostatic state and present an MGnD phenotype, with Gal3 playing a key role. In fact, we studied how Gal3 is present in various brain structures, with the hippocampus being one of them. This, along with the fact that the state of microglia is relevant in the deposition of tau protein, further confirms the importance of these cells in the aggregation of this protein (Maphis et al., 2015; Shi et al., 2019). Papegaey and colleagues (Papegaey et al., 2016) described how FTD patients with GRN mutations have less accumulated tau protein and pointed to post-translational modifications as responsible for this decrease in tauopathy. Similarly, during the development of our experiments, other study showed how tauopathy model animal PS19 lacking progranulin (Grn-/-) presents a smaller area of phosphorylated tau in the hippocampus compared to the PS19 Grn+/+ mouse (Takahashi et al., 2024). However, the absence of progranulin led to a greater number of tau inclusions. Another relevant aspect observed in this study is the effect on microgliosis. Specifically, P301S animals show significant microgliosis in the hippocampus, which is attenuated in PGRNKO/P301S animals. The role of microglia in tauopathies is a subject of study and current debate. Some studies suggest that microgliosis occurs prior to tau deposition (Yoshiyama et al., 2007), while others point out that microglial activation is a consequence of protein accumulation (Holmes et al., 2014; van Olst et al., 2020). Additionally, it is thought that microglia play a relevant role in the propagation of tau, with the phagocytosis of this protein being a key event (Bolós et al., 2016; Luo et al., 2015; Maphis et al., 2015). Therefore, considering that the absence of progranulin triggers phagocytic markers such as CD68, among other microglial activation markers, our results suggest that the microglial activation that could occur due to the absence of progranulin would result in a more efficient clearance of tau protein deposits and, consequently, a decrease in gliosis (Huang et al., 2020; Root et al., 2023; Takahashi et al., 2024). The histological changes observed in P301S and PGRNKO/P301S animals translate into modifications in their behavior. This study observed that both P301S and PGRNKO/P301S models show behavioral alterations, resulting in changes in depressive and anxiety-like behaviors in the animals. P301S animals show a worsening of depressive-like behaviors, which could be slightly improved in the PGRNKO/P301S model. However, both P301S and 127 DISCUSSION PGRNKO/P301S animals show aggravated disinhibited behavior, particularly pronounced in the PGRNKO/P301S model. Similar results have been observed by Takahashi and colleagues (Takahashi et al., 2024). 7. Alterations of the brain immune response due to the absence of Galectin-3 One of the main features of neurodegenerative pathologies is the activation of the immune response, primarily driven by microglia. At the transcriptomic level, in the thalamus of PGRNKO animals, pathways intrinsically associated with the immune response are triggered, highlighting the complement pathway, the response to interferons alpha and gamma (Class I and II), IL-2/STAT5 signaling, and TNF signaling regulated by NF-κβ. On the other hand, in view of these results and to determine the level of pro-inflammatory activation in the brains of PGRNKO animals, we conducted a study on inflammatory cytokines present in the cortex. This region is moderately affected in the PGRNKO model, considering the presence of Gal3 in it. Our study points to an abundance of cytokines such as IL-6, IL-12p70, and KC/GRO, which are markedly pro-inflammatory, in PGRNKO animals (Korbecki et al., 2022; Schwarz & Carson, 2022; West et al., 2019). Through various approaches, the literature has described how FTD patients and/or models of this pathology due to the absence of progranulin exhibit an exacerbated immune response. On one hand, FTD patients are notable for the elevation of complement proteins such as C1Q and C3 in cerebrospinal fluid (van der Ende et al., 2022). Additionally, PGRNKO animals show overexpression of interferon and NF-κβ pathways, which are associated with TDP-43 accumulation (Yu et al., 2020). Furthermore, the IL-2/STAT5 pathway has been described as overrepresented in these animals (Zuppe & Reed, 2024). Therefore, we can conclude that the absence of progranulin leads to an exacerbated immune response both in the thalamus and in the cortex, from a transcriptomic and proteomic perspective, respectively. Among the mentioned pathways, those related to the response to interferons alpha and gamma show a decrease in the expression of the genes involved in the absence of Gal3. Some studies have described the interaction between Gal3 and interferon gamma (Gordon-Alonso et al., 2017; Ruvolo, 2019). Likewise, it has been demonstrated that Gal3 induces the production of various inflammatory cytokines such as TNFα, IL6, IL1B, IL12p40, IP-10, and IFNγ. Moreover, Gal3 activates the phosphorylation of the JAK-STAT pathway through its binding to IFNGR1. This process occurs independently of IFNγ itself (Jeon et al., 2010). Additionally, the expression of the mentioned pro-inflammatory cytokines is reduced in these Gal3-deficient animals. It is also important to note the increase of GPNMB in the thalamus of these animals, as this protein is known for its antiinflammatory properties (Ping et al., 2025; Saade et al., 2021). Thus, the absence of Gal3 would result in lower microglial activation and associated immune response, favouring a less pro-inflammatory environment. 128 DISCUSSION 8. Dysregulation of energy metabolism and its relationship with progranulin and Galectin-3 Oxidative phosphorylation is an essential process for proper energy metabolism and is altered in various neurodegenerative pathologies such as Parkinson’s disease or amyotrophic lateral sclerosis (Kawamata & Manfredi, 2018). In their study, Kawamata and Manfredi also pointed to frontotemporal dementia, establishing a relationship between TDP-43 accumulation and dysfunctional endoplasmic reticulum – mitochondria communication. In this work, it has been observed that the expression of genes involved in energy metabolism and related to the electron transport chain is overexpressed in thalamic tissue in the absence of progranulin. Some examples are Ndufa1 and Ndufa2, which encode subunits of NADH dehydrogenase. This alteration in energy metabolism is reversed or attenuated in conditions of Gal3 deficiency in frontotemporal dementia model animals. Genes encoding NADH dehydrogenase, including the mentioned Ndufa1 and Ndufa2, as well as other proteins involved in the electron transport chain, show reduced expression levels in PGRNKO/Gal3KO animals. This not only reverses a pathological characteristic observed in various proteinopathies but also establishes a relationship between Gal3 and the mitochondria, something suggested by Coppin and colleagues (Coppin et al., 2020). In their study, they postulated that Gal3 interacts not only with various mitochondrial membrane proteins but could also determine morphological aspects of the organelle itself. 9. Alterations of the Cell Cycle in Neurodegeneration Conditions Oxidative phosphorylation and energy metabolism are closely linked to cell survival and proliferation pathways (Raimondi et al., 2020). Following the observations made in the previous section, our study revealed that in the thalamus of PGRNKO mice there is an overregulation of genes involved in pathways associated with cell proliferation and survival, such as the KRAS and mTOR pathways. It has been described that, under neurodegenerative conditions, these pathways become dysregulated, leading to the production of inflammatory cytokines (Keane et al., 2021). Additionally, the involvement of mTOR and its interaction with Akt is crucial in the PGRNKO model. The Akt complex has been shown to play a significant role not only in cell survival but also in the processes of phagocytosis and lysosomal integrity, which are compromised in the frontotemporal dementia model (Hirata et al., 2018; Huang et al., 2021). Conversely, the absence of Gal3 in the PGRNKO model triggers a decrease in the expression of some genes involved in cell survival and proliferation, with Akt2 being among them. Furthermore, some studies have already pointed out the importance of Gal3 in the KRAS pathway (Levy-Lahad et al., 1995; Ruvolo, 2019). Thus, and considering all that has been mentioned previously, it is essential to highlight the role of Gal3 as a modulator of cell survival. However, given the pleiotropic nature of some affected genes, such as those encoding for Akt, Gal3 might be implicated in immune response through 129 DISCUSSION the suppression of cytokine release, as well as in the reduction of the expression of genes associated with phagocytosis and lysosomes. 10. Changes in Lipid Metabolism and Its Relationship with Galectin-3 Transcriptomic analysis of the thalamus revealed changes in the expression of genes related to fatty acid metabolism and adipogenesis. Previous studies have pointed out that frontotemporal dementia in humans and animal models lacking progranulin presents a dysregulation of the lipid profile: accumulation of polyunsaturated triglycerides, decreased diacylglycerols, and phosphatidylserines. Moreover, this phenomenon is associated with lysosomal dysfunction along with an increase in the lysosomal lipidome (Evers et al., 2017). This lipid accumulation must be processed by lysosomal hydrolases, many of which show deregulated activity, as observed through proteomic studies of white matter. Furthermore, progranulin plays a key role in lysosomal acidification and acts as a chaperone for various proteases present in this organelle (Tanaka et al., 2017). An example is cathepsin D, whose maturation under acidic conditions is influenced by progranulin (Butler et al., 2019). Literature has described that the absence of progranulin leads to decreased cathepsin D activity, such that the deficiency of both proteins leads to the accumulation of myelin debris and TDP-43 in white matter regions (Beel et al., 2017; Valdez et al., 2017; Ward et al., 2017; Wu et al., 2021). Additionally, progranulin is involved in the neuronal uptake and lysosomal transport of prosaposin, which regulates various sphingolipid hydrolases, as well as its cleavage into saposin C, responsible for activating β-glucocerebrosidase (GCase) (Valdez et al., 2020; X. Zhou et al., 2017). It is also relevant to highlight that one of the primary processes associated with the regulation of lysosomal hydrolases is lipid catabolism, which is compromised in the absence of progranulin. Thus, from a transcriptional perspective, lipid metabolism alterations and its association with lysosomes are demonstrated in PGRNKO animals. The relationship between Gal3 and lipid metabolism is scarcely described in the current literature. However, it is known that Gal3 has the ability to interact with membrane lipids and is associated with dysfunctional lysosomes (Aits et al., 2015; Lukyanov et al., 2005). On the other hand, TREM2, a receptor with which Gal3 physically interacts, is associated with the clearance of myelin debris, which is abundant in the studied region (BozaSerrano et al., 2019; Poliani et al., 2015). In this work, we found that the absence of Gal3 in a lipid dysregulation scenario leads to a decrease in fatty acid metabolism by reducing the expression of genes such as ACADL, ACADVL, and FABP2, among others. 11. The Absence of Galectin-3 Reduces Lipofuscinosis in the PGRNKO Model One of the main phenomena observed in the PGRNKO model is the lysosomal hypertrophy localized in regions such as the thalamus and the thalamocortical tract, with microglial cells showing the most pronounced features. The increase in lysosomal size 130 DISCUSSION compromises the integrity of its membrane, potentially triggering a process of cell death (Cao et al., 2021; Wang et al., 2018). The lysosomal dysfunction caused by hypertrophy in PGRNKO animals is supported by the presence of Gal1 in microglial cells where hypertrophic lysosomes are found. It has been described that Gal1, like Gal3, can bind to the glycocalyx of lysosomes whose integrity is compromised (Aits et al., 2015). Similarly, Gal3 is highly localized in both the thalamus and white matter regions of PGRNKO animals. This lysosomal dysregulation is also reflected in the proteomic study of white matter. One of the main biological processes that is altered in PGRNKO animals compared to controls is the regulation of hydrolase activity, enzymes present in the lysosome. Therefore, the absence of progranulin in mice results in marked lysosomal hypertrophy, which is closely related to lysosomal dysfunction itself. The compromise of lysosomal homeostasis is also reflected through the high presence of the marker CD68 via immunofluorescence. This protein is associated not only with lysosomes but also with the phagocytosis process carried out by microglia (Hopperton et al., 2018). In line with recent findings by Root and colleagues (Root et al., 2023), the absence of progranulin triggers elevated levels of CD68 in the thalamus and white matter regions. From the above, it can be deduced that the absence of progranulin compromises lysosomal integrity through the increase in lysosomal size and elevated phagocytosis by microglia. The lysosomal dysfunction in PGRNKO animals, as well as the widely described microglial activation through not only CD68 but other multiple MGnD markers such as TREM2, CLEC7A, GPNMB, and Gal3 itself, contribute to creating an environment intrinsically associated with neuronal death. In this study, we describe for the first time the development of cell death processes through the phagocytosis of thalamic neurons via immunofluorescence, as well as the transcriptomic study of a set of genes associated with apoptosis. Although the cause of the death of these neurons is still unknown, the absence of progranulin and its effect on lysosomal integrity suggests that a notable change in the homeostasis of these organelles leads to cell death processes through apoptosis (Wang et al., 2018). Through confocal microscopy of PGRNKO tissues, we observed how phagocytosed thalamic neurons contained notable amounts of lipofuscin inside them. Surprisingly, the absence of Gal3 in the PGRNKO model leads to reduced expression of genes involved in the apoptosis pathway in the thalamus. Among the various genes most affected in this pathway are several effector caspases, such as Casp3, Casp6, and Casp7, as well as Dcn, which induces apoptosis by increasing p53 levels, and Stat1, responsible for inhibiting cell growth and inducing apoptosis (Hu et al., 2021; Yoon et al., 2017). These results are consistent with other phenomena studied in this text. Thus, a lower presence of inflammatory cytokines and the decrease in the expression of pathways associated with 131 DISCUSSION inflammation in PGRNKO/Gal3KO animals could translate into less cell death (Zhang et al., 2023). Similarly, the reduction of some MGnD/DAM markers and the increase of others associated with a reduction in inflammation, such as GPNMB, would point in the same direction (Tanaka et al., 2012). Additionally, the potential reduction in lysosomal hypertrophy, as evidenced by LAMP1 immunofluorescence, would allow for better lysosomal integrity. It has been described that lysosomal hypertrophy can lead to the release of the organelle's contents into the cytosol, resulting in corresponding cell death (Wang et al., 2018). One of the main histopathological features of PGRNKO animals is the notable accumulation of lipofuscin, or lipofuscinosis. Lipofuscin is an accumulation of lipoproteins that resides in the interior of secondary lysosomes of certain cells, such as neurons, and is especially present in aging organisms. Lipofuscin has autofluorescent properties, which allows it to be visualized using microscopy techniques, appearing as granular in nature. Additionally, due to its lipid content, the autofluorescence of lipofuscin can be attenuated or blocked using lipid stains such as Sudan black (Miller & Zachary, 2017). In this study, we observed that PGRNKO animals exhibit a significant amount of lipofuscin in brain tissue, with the thalamus, particularly the DORsm region, standing out, present both in neurons and microglial cells. Also, due to the lysosomal nature of lipofuscin, it is interesting to note its absence in regions such as the thalamocortical tract and the corpus callosum, areas rich in axonal fibers and showing notable microgliosis, but lacking neuronal somas. This suggests that, in PGRNKO animals, lipofuscinosis is primarily found in the soma of neurons, while microglia do not intrinsically form these accumulations. Thus, it is proposed that the lipofuscin observed in microglia in thalamic regions likely originates from neurons and could reach the interior of microglia via endocytosis. It is important to highlight that this study demonstrates for the first time that the lipofuscinosis generated by progranulin deficiency is reduced in the absence of Gal3 in the thalamus. It is known that Gal3, along with other galectins, acts as markers of lysosomal permeability, binding to the glycocalyx covering the lysosomal membrane and participating in the process of eliminating damaged lysosomes through autophagy in a phenomenon known as lysophagy (Aits et al., 2015). Recently, Gal3 has been observed in lysosomes containing lipofuscin in the brain of Parkinson's disease patients and controls (García-Revilla et al., 2023). Given the relationship between Gal3 and lysosomal damage, and considering the intracellular localization of lipofuscin, as well as the presence of Gal3+ microgliosis in regions of the brain rich in lipofuscin, a direct association between Gal3 and lipofuscin is proposed. Lysosomes containing lipofuscin in microglia could present damage or alterations in their permeability, possibly due to the absence of progranulin and the granulins derived from it, resulting in elevated levels of Gal3 in these cells. The observed microglial phenotypic changes, the decrease in inflammatory response, and the increase in phagocytic marker levels could translate into a less inflammatory environment