Background
Frontotemporal lobar degeneration (FTLD) is a clinically and pathologically complex neurodegenerative disorder defined as progressive behavioral abnormality, frontal executive dysfunction, and selective language impairments associated with frontal and anterior temporal lobe degeneration [
1,
2]. Frontotemporal dementia (FTD), the most common clinical manifestation of FTLD, has been recognized as a prominent cause of dementia, especially in patients under 65 [
1]. Since the first description of the link between a pathogenic variant (PV) of the progranulin gene (
GRN) and FTD in 2006 [
3], more than 70 different pathogenic
GRN variants in FTD have been reported [
3,
4].
GRN encodes progranulin (PGRN), a highly conserved, cysteine-rich, secreted glycoprotein [
5,
6]. PGRN is involved in many cellular processes, including inflammation, wound healing, tumorigenesis, and neuroprotection [
6,
7]. PGRN haploinsufficiency is caused by heterozygous loss-of-function (LOF) mutations of
GRN, leading to autosomal dominant FTD with TAR DNA-binding protein 43 (TDP-43) positive inclusions in neuron and glial cells [
3,
8,
9].
Various studies have demonstrated that PGRN associated with microglia can serve as a critical regulator of inflammation [
7,
10‐
13]. It is well-known that PGRN plays a role in the anti-inflammatory process by reducing pro-inflammatory cytokines and suppressing disease-associated microglial activation, which can lead to neuronal loss [
6,
14]. Activated inflammatory response, accumulation of myelin debris in microglial lysosomes, and excessive synaptic pruning via complement activation have been identified in a
Grn knockout mice model [
10‐
12]. In addition, both global
Grn knockout mutant mice and microglia-specific
Grn knockout mutant mice demonstrate extended pro-inflammatory microglial activation and neuronal loss [
14]. Likewise, most previous studies have evaluated microglia function in mice models with complete PGRN deficiency. Mice models with heterozygous loss of
Grn failed to develop gliosis and inflammation. They only exhibited minimal behavior and neuropathologic changes [
15‐
17]. Therefore, microglial function should be identified in human cell models of PGRN haploinsufficiency to investigate the pathology of FTD–
GRN. In diseases such as Nasu-Hakola disease and hereditary diffuse leukoencephalopathy with spheroids where microglial dysfunction is considered the primary pathomechanism, the term “microgliopathy” has been introduced, emphasizing the pivotal role of microglia [
18,
19]. This concept underscores the significance of elucidating pathological mechanisms that give rise to abnormal microglial activation. In the context of FTD–
GRN, unraveling the pathological phenomena responsible for proinflammatory microglia activation would be crucial for comprehending the disease precisely.
Neuronal and glial cytoplasmic TDP-43 aggregation with a ubiquitinated state is a pathological hallmark of FTD–
GRN. In FTD–
GRN cases, it is still unclear how
GRN dysfunction causes TDP-43 pathology and neurodegeneration. Recent evidence suggests that TDP-43 is involved in neuroinflammatory and immune-mediated mechanisms in FTD pathogenesis [
20]. In addition, TDP-43 has relationships with immune and inflammatory pathways, including NF-κB/p65, cGAS/STING, and NLRP3 inflammasome that centers around microglia [
20,
21]. Nevertheless, research investigating the presence and mechanisms of TDP-43 pathology in microglia has been scarce.
To investigate whether microglial pathology and dysfunction are present in PGRN haploinsufficiency, we generated monocyte-derived microglial-like cell (iMGs) from two patients diagnosed with FTD–GRN (p.M1? and p.W147*). Herein, transcriptional and functional analyses of FTD–GRN patient-derived iMGs demonstrated that PGRN deficiency could lead to cytoplasmic TDP-43 deposition with persistent pro-inflammatory environment by microglia activation, dysregulation of lysosomal function, and altered lipid metabolism. This study also suggests new evidence for the relationship between TDP-43 aggregation and microglia-mediated excessive inflammatory reactions, elucidating the underlying mechanism of TDP-43 proteinopathy in FTD–GRN. These pathological and functional abnormalities found in human microglia harboring PGRN haploinsufficiency could provide crucial insight into the development of therapeutic strategies for FTD–GRN.
Discussion
Microglia are innate immune cells of the central nervous system. As major components of neuroinflammation, they have recently emerged as targets for neurodegenerative diseases [
47]. PGRN is highly expressed in microglia. It is activated in reactive states due to injuries, aging, or disease pathology [
9,
48]. FTD–
GRN patients present increased disease-associated reactive microglia, pro-inflammatory cytokines, and microglial dystrophy [
49]. However, whether human microglial functional defects caused by FTD-linked
GRN–LOF variant directly contribute to FTD pathogenesis remains unclear. The current study investigated functional and pathological properties of microglia associated with PGRN haploinsufficiency by utilizing monocyte-derived iMGs from two FTD–
GRN patients, one with a recurrent pathogenic variant and the other with a novel, likely pathogenic variant. Here, we demonstrated that LOF variants of
GRN in the human monocyte-derived microglia-like cell model caused microglial dysfunction with abnormal TDP-43 aggregation induced by inflammatory milieu as well as impaired lysosomal function, thus representing an exacerbated disease phenotype. In contrast to the findings in patients with
GRN variants, control subjects did not demonstrate PGRN-associated biomarkers or cell abnormalities. This contrast further substantiates the genetic integrity of our control group, affirming their absence of
GRN-related abnormalities and underscoring the significance of the genetic differences observed in the study. Age-matched control subjects have been confirmed not to carry pathogenic variants of the
GRN gene. Furthermore, these individuals did not exhibit any diseases associated with
GRN abnormalities, nor was there any family history of such conditions reported. Microglial TDP-43 alterations were presumably a compositive phenotype reflecting exaggerated immune responses by activated complement and lysosomal abnormalities in PGRN-deficient microglia. Functional impairments in microglia due to
GRN variants associated with FTD appear to be essential pathophysiological mechanisms underlying FTD–
GRN.
iMGs from patients with FTD–
GRN exhibited a reduction in phagocytic function compared to control iMGs, indicating a pro-inflammatory state. This observation aligns with broader findings that show alterations in morphology, cytokine production, secretion, and phagocytic ability in microglia when they are aberrantly activated or in a pro-inflammatory or diseased state [
50]. Chronic pro-inflammatory conditions or pathological aberrant activation, particularly prevalent in neurodegenerative diseases, lead to a shift in microglial behavior characterized by impaired phagocytosis [
51]. Consequently, we can assume that PGRN deficiency likely hinders their ability to maintain homeostatic molecular signatures and impairs their phagocytic capacity, further exacerbating neuroinflammation.
Accumulation of TDP-43 aggregates in the central nervous system is a common feature of many neurodegenerative diseases, such as amyotrophic lateral sclerosis (ALS), FTD, Alzheimer’s disease, and limbic-predominant age-related TDP-43 encephalopathy with dementia [
52,
53]. Hereditary FTD–
GRN shows ubiquitin-positive inclusions composed of TDP-43 in neuron and glial cells [
3,
8]. Although aggregated TDP-43 has not been defined in microglia, this study demonstrates cytoplasmic TDP-43 positive inclusions in FTD–
GRN patient-derived iMGs cultures under basal conditions without additional stressors. Microglial TDP-43 aggregates showed phosphorated TDP-43 (S409/410) and colocalization with ubiquitin protein. In addition, we identified elevated levels of TDP-43 and pathological TDP-43 (S409/410) in the insoluble fraction of FTD–
GRN patient-derived iMGs by western blots (Fig.
3). To support this result, whether identical FTD–
GRN patient-derived fibroblasts showed TDP-43 pathology comparable to those in FTD–
GRN patient-derived iMGs was investigated.
GRN patient-derived fibroblasts showed an increase in pTDP-43 (Ser409/410) immunoreactivity within the cytoplasm (Additional file
1: Fig. S3A). Furthermore, the expression of pTDP-43 (Ser409/410) in insoluble fractions of identical patient-derived fibroblasts carrying
GRN variants and increased levels of TDP-43, ubiquitinated proteins, and p62 may support microglial TDP-43 proteinopathy (Additional file
1: Fig. S3B, C).
In a study using TDP-43-depleted BV2 microglial cells subjected to
GRN knockdown, there was a notable accumulation of TDP-43 in the cytoplasm (Fig.
3F). This finding indicates a disposition for TDP-43, usually located in the nucleus of microglial cells, to aggregate in the cytoplasm when PGRN is deficient. In addition, recent research has shown that in the postmortem brains of patients with motor neuron disease with TDP-43 pathology, phosphorylated TDP-43 aggregates were present in the Iba1-positive microglial cells [
54]. While TDP-43 is known to spread in a prion-like manner, moving from cell to cell in a seed-dependent and self-templating process [
55], it appears feasible that cytoplasmic TDP-43 accumulation might be initiated by PGRN haploinsufficiency in a microglial environment where TDP-43 is typically confined to the nucleus under normal conditions.
Several studies have shown that inflammatory stimuli can promote TDP-43 aggregation and cytoplasmic mislocalization in microglial cells [
20,
56‐
58]. In this study, we found several remarkable microglial phenotypes caused by PGRN haploinsufficiency in human microglia, which could not maintain homeostasis. They transformed into an inflammatory state mainly characterized by pro-inflammatory cytokine and complement activation with impaired phagocytosis, finally inducing exaggerated immune responses. The complement system is a rapid and efficient immune surveillance system. Its imbalance can contribute to various immune, inflammatory, and neurodegenerative diseases [
59]. Upregulation of C1q and C3b is not only present in
GRN mutation carriers, but also in genetically unexplained FTLD–TDP subtype A patients [
60,
61]. Therefore, this study focused on complement activation in microglia from
GRN variants to define the linkage between the complement system and TDP-43 proteinopathy. We found that direct complement C1q treatment in BV2 microglial cells triggered abnormal cytoplasmic aggregation of microglial TDP-43. In addition, complement C1q treatment with
GRN loss condition markedly increased cytoplasmic aggregation of microglial TDP-43. Apart from the possibility that innate immune activation of microglial cells may exacerbate neuronal TDP-43 proteinopathy through the release of inflammatory cytokines, it was noteworthy that treatment with complement C1q in microglial cells self-triggered abnormal cytoplasmic aggregation of microglial TDP-43. TDP-43 is shuttled from the nucleus to the cytoplasm. It transiently forms cytoplasmic condensates through phase separation. This process can also lead to irreversible formation of permanent aggregates and fibrils in neurodegenerative diseases [
62‐
64]. Recently, cytoplasmic TDP-43 mislocalization in monocyte-derived microglia-like cells of patients with ALS [
65] and in lymphoblasts of patients with FTD–
GRN [
66] has been reported. In addition,
GRN-deficient microglia exhibit extranuclear TDP-43 condensates with lipid droplets in a zebrafish model of traumatic brain injury [
58]. Granulins have been shown to exacerbate TDP-43 toxicity in vivo in
Caenorhabditis elegans and mice [
67] and to alter the solubility of TPD-43, thereby modulating its phase separation and aggregation properties [
68,
69]. The cell model of iMGs, similar to lymphocytes, may reflect the inflammatory state of FTD–
GRN, including complement activation, inflammation, and other aging factors, which could result in cytoplasmic TDP-43 accumulation. Therefore, formation of cytoplasmic TDP-43 aggregates by complement activation suggests that
GRN–LOF microglia are sufficient to trigger the pathological process of FTD–
GRN.
Due to inflammation induced by PGRN haploinsufficiency, upregulated cell death might obscure TDP-43 accumulation in human brain microglia with FTD–
GRN. Investigating this, we utilized cleaved capsace-3 as a marker of apoptosis to access cytotoxicity. As a result, we found that iMGs from patients with FTD–
GRN showed significantly increased cleaved caspase-3 positive immunoreactivity compared to control iMGs (Additional file
1: Fig. S4). Furthermore, we examined endolysosomal membrane permeabilization of iMGs from patients with FTD–
GRN and control iMGs, hypothesizing that impairment of lysosomal membrane integrity would cause lysosomal-dependent cell death [
70]. Utilizing immunofluorescence staining techniques, we examined the cellular distribution of galectin-3 (Gal-3). Gal-3 is a cytosolic protein known to localize to damaged lysosomes, functioning as a sensitive marker for lysosomal leakage [
71]. Our observations revealed that in iMGs obtained from patients with FTD–
GRN, a substantial proportion of Gal-3 was present as punctate formations. These formations indicate intracellular vesicle rupture, commonly triggered by amyloid proteins such as α-synuclein, tau, and mutant Huntingtin [
72]. In contrast, such punctate formations of Gal-3 were scarcely observed in control microglia, highlighting a distinct pattern in FTD–
GRN patient-derived cells (Additional file
1: Fig. S5). The observations indicate potential challenges in detecting TDP-43 accumulation in human brain microglia, which might be attributed to activated cell death, potentially influenced by diminished lysosomal membrane integrity.
PGRN is an intracellular and extracellular precursor protein that undergoes proteolytic cleavage, forming individual granulin peptides [
73]. In the context of PGRN cleavage, it is noteworthy that this process yields paragranulin, with an approximate molecular weight of 3.5 kDa, and granulins A–G, each approximately 7 kDa in size [
73,
74]. Recent reports have suggested that granulin peptides may be critical in generating TDP-43 toxicity in FTD–
GRN [
67]. Notably, despite the reduction of its precursor PGRN due to haploinsufficiency, granulin F levels have been found to be increased in regions of the human FTD–
GRN brain [
74]. Despite the controversy surrounding the diverse functions and variable expression of individual granulin peptides in pathological states, emerging evidence that links granulin peptides to prion-like TDP-43 cytoplasmic inclusions supports the hypothesis of their potential pathognomonic role in FTD–
GRN [
68,
75,
76].
PGRN is a critical lysosomal chaperone required for lysosomal function and the ability of microglia to counteract misfolded proteins [
7]. It has been reported that PGRN deficiency is linked to lysosomal dysfunction, which can influence lysosomal acidification and enzymatic activity, defective autophagy, and lipofuscinosis [
77,
78]. Our findings are consistent with recent studies showing that PGRN protein is expressed in lysosomes of human microglial cells. FTD–
GRN patient-derived iMGs reveal lysosomal abnormalities, including enlarged lysosomes, alteration of lysosomal genes, abnormal lipid droplet accumulation, and TFEB activation. Activation of TFE3/TFEB has been shown to drive expression of inflammation genes [
79]. These data suggest that lysosome abnormalities in microglia can result in a feedback loop through activation of the TFEB pathway, which could drive the expression of inflammatory genes and the activation of target genes by lysosomal damage. PGRN levels have been linked to expression of several genes, including
SORT1 and
TMEM106B in lysosomes [
42,
80]. Induced lysosome dysfunction caused by increased expression of TMEM106B can inhibit the processing of PGRN into granulins [
81]. Consistent with previous studies, we found that iMGs from FTD–
GRN patients resulted in lysosomal enlargement and dysregulated markers of lysosomes along with increased lysosomal protein levels such as SORT1 and TMEM106B. Overexpression of TMEM106B can cause translocation of TFEB to the nucleus and induces upregulation of coordinated lysosomal expression and regulation network [
82]. The present study demonstrated a significant elevation in the relative mRNA expression of
LGALS3, encoding Gal-3, in iMGs from patients with FTD–
GRN. Furthermore, the occurrence of Gal-3 and its puncta formations, which are absent in control iMGs, were evident in the iMGs derived from patients with FTD–
GRN (Additional file
1: Fig. S5). Our results align with existing data indicating an upregulation of Gal-3 in both patients with FTD–
GRN and
Grn LoF mice [
83,
84]. These observations, coupled with our previous findings of lysosomal membrane permeabilization triggered by PGRN deficiency in human iPSC-derived
GRN−/
− microglia, strongly suggest the occurrence of lysosomal damage in iMGs associated with FTD–
GRN.
Furthermore, we examined the co-localization of lipid droplets (BODIPY) with lysosomes immunostained with LAMP1 to investigate the lipophagic delivery of lipid droplets to lysosomes. Lipid droplets in control-derived iMGs clearly showed co-localization with lysosomes, whereas partial co-localization of lipid droplets with lysosomes was present in iMGs from patients with FTD–
GRN (Additional file
1: Fig. S6). The current results demonstrate that PGRN haploinsufficiency-induced abnormal lipid droplets in microglia may interfere with lipid degradation in microglial lysosomes.
In addition, TDP-43 pathology may disrupt lysosomal function, driving further pathology. Loss of nuclear TDP-43 is a key aspect of TDP-43 pathology that may disrupt the autophagy–lysosomal and endolysosomal systems [
85‐
89]. Therefore, defective microglial lysosome by PGRN loss might lead to impaired phagocytic and autophagic clearance of cellular waste and debris as well as toxic protein aggregates. Conversely, TDP-43 aggregation might be further exacerbated by lysosomal abnormalities in PGRN-deficient microglia.
It has been reported that inflammatory and metabolic changes in immune cells involved in upregulated fatty acid production can cause formation of lipid droplets [
46]. Accumulation of lipid droplets in microglia is known to represent a dysfunctional and pro-inflammatory state in the aging brain [
90,
91].
GRN knockout by gene editing can promote lipid droplet accumulation in microglia, resulting in phagocytic dysfunction and activation of pro-inflammatory responses [
90]. This study revealed that FTD–
GRN patient-derived iMGs could induce lipid-droplet formation accompanied by activation of inflammatory cytokines, including complement. Furthermore, direct complement C1q treatment induced lipid droplet formation in BV2 microglial cells. Additional
GRN loss increased lipid droplet sizes. These results suggest that an impaired lysis mechanism caused by lysosomal abnormalities can lead to the excessive accumulation of lipid droplets with activated inflammatory conditions in FTD–
GRN patient-derived iMGs.
Recent studies have aimed at a therapeutic approach for FTD–
GRN to restore CNS PGRN levels [
92] using adenovirus-associated virus-based gene therapy, SORT1-binding antibodies, and small molecules modulators (such as suberoylanilide hydroxamic acid, methyltransferase inhibitors, nor-binaltorphimine dihydrochloride, and dibutyryl-cAMP, sodium salt [
93‐
96]. Despite encouraging success in preclinical studies, a barrier remains due to the lack of suitable human and mouse models for therapeutic development of FTD–
GRN. Mice with heterozygous
Grn deletions do not exhibit behavioral or neuropathological changes typical of
GRN heterozygosity in humans [
97,
98]. Recent approaches using induced pluripotent stem cell-derived microglia are now available. However, relative complexity, high variability, and extended timeframe are required to generate cell models. Moreover, iPSC-derived microglia may not accurately contain the heterogeneity of clinical features observed in the disease process by pathogenic variants due to loss of epigenetic factors during reprogramming [
99]. In this study, we used an iMGs model derived from human monocytes, a rapid and minimally invasive system that allows for multiple sampling at various stages of the disease. This cell model can recapitulate changes in microglia during disease progression. Such changes can be correlated with clinical data (brain imaging and clinical disease progression), which may bridge the gap between clinical studies by providing a better clinical outcome [
25,
30,
65]. Therefore, iMGs could be used as an in vitro platform method or a preclinical study tool to analyze their functional defects through genetic mutations and to evaluate therapeutic drugs.
However, this study has some limitations. First, it was not possible to enroll various types of patients diagnosed with FTD–
GRN. In contrast to the Caucasian population, the Asian population revealed a significantly lower frequency of FTD–
GRN [
100‐
105]. Furthermore, differences in clinical severity of the disease, patients’ states, other genetic modifiers, and sex-based microglial effects might have affected results since this study only investigated two types of patient-derived iMGs models. Additional patients with different FTD–
GRN pathogenic variants might provide more valuable experimental results. Second, microglial TDP-43 aggregates in FTLD–
GRN human brain tissues have not been reported yet. However, biochemical TDP-43 phenotypes closely resemble those observed in neuron. Given that the possibility of TDP-43 aggregation in microglial cells has recently been reported, further studies are needed to confirm the formation of TDP-43 aggregates according to different phenotypic markers of microglial cells in patient tissues. Third, we could not analyze the effects of individual granulin peptides produced through the proteolytic cleavage of PGRN. Specific granulin peptides have been implicated in liquid–liquid phase separation associated with TDP-43 accumulation [
68]. However, our research did not extend to investigating the impact of individual granulin peptides on TDP-43. Given the conflicting results emerging from various studies regarding individual granulin peptides, there is a clear need to develop antibodies that can specifically detect these peptides and further research into their interactions [
74‐
76]. Lastly, although this study focused only on the role of PGRN in human microglia, its effect on interactions with various types of neuronal cells in the brain environment was not determined. Therefore, further studies are necessary to elucidate the impact of PGRN, such as utilization of 3D modeling that incorporates a brain microenvironment with different neuronal cell types, embodying the complexity of a brain’s homeostatic and diseased states.