Human TDP-43 expression worsens FTD-related phenotypes in progranulin-insufficient mice Authors: Anna K Cook, Benjamin Lin, Yumo Song, Kelsey M Greathouse, Azariah K Kaplelach, Mackenzie L Love, Skylar E Davis, Anna C Stoll, Justin A Hall, Ahmad R Hakim, Jakub F Hel, Giacynta A Vollmer, Alexandria C Howard, Noelle H Cooper, Phaedra N Manuel, Juliana M Eberhardt, C Ryan Miller, Ashley S Harms, Jeremy H Herskowitz, Lindsay F Rizzardi, Andrew E Arrant Published: 2026 ABSTRACT AbstractLoss-of-function progranulin (GRN) mutations cause frontotemporal dementia with TDP-43 pathology (FTD-TDP). Nearly all pathogenic GRN mutations cause progranulin haploinsufficiency, but it is unclear how progranulin insufficiency causes FTD-TDP. To address this question, we crossed progranulin-insufficient mice with a human TDP-43 transgenic mouse line (RRID:IMSR_JAX:012836) in which homozygous mice (hTDP++) develop TDP-43 aggregates at an early age, but hemizygous mice (hTDP+) do not develop TDP-43 aggregates. We therefore analyzed the effects of progranulin insufficiency on both hTDP+ and hTDP++ mice. Progranulin insufficiency did not induce TDP-43 aggregation in hTDP+ mice, but interacted with hTDP expression to worsen FTD-related phenotypes. Grn+/−:hTDP+ mice exhibited more dramatic impairment of social dominance than either Grn+/− or hTDP+ mice, which was associated with combined effects of progranulin insufficiency and hTDP expression on dendritic spines of neurons in the medial prefrontal cortex (mPFC). Despite a lack of TDP-43 aggregation, progranulin insufficiency altered the RNA splicing events induced by hTDP overexpression in frontal cortex of hTDP+ mice. Progranulin insufficiency also did not alter TDP-43 aggregation in hTDP++ mice, but Grn−/−:hTDP++ mice exhibited an abnormal neuroinflammatory response characterized by increased markers of disease-associated microglia and signs of an impaired adaptive immune response. These results highlight dysfunction of mPFC neurons as a potential mechanism of behavioral changes in FTD-GRN and implicate dysregulated inflammation as a potential driver of disease progression in FTD-GRN.Keywords: TDP-43, Progranulin, Frontotemporal dementia Introduction Heterozygous loss-of-function mutations in the progranulin gene (GRN), which typically result in haploinsufficiency of progranulin, cause frontotemporal dementia (FTD) (Baker et al., 2006; Cruts et al., 2006). The underlying pathology of FTD due to GRN mutations (FTD-GRN) is frontotemporal lobar degeneration with TDP-43 pathology (FTLD-TDP) type A (Mackenzie et al., 2011), a form of TDP-43 proteinopathy which is characterized by mislocalization of TDP-43 from the nucleus to the cytoplasm and the formation of TDP-43 aggregates (Arai et al., 2006; Neumann et al., 2006). TDP-43 proteinopathy may directly lead to neurodegeneration, as pathologic TDP-43 may be toxic to neurons and depletion of TDP-43 from the nucleus has a loss-of-function effect that disrupts RNA expression and splicing (Arai et al., 2006; Barmada et al., 2010; Igaz et al., 2011; Ling et al., 2015; Neumann et al., 2006). Understanding how GRN mutations cause TDP-43 proteinopathy is therefore important to understanding pathogenesis of FTD-GRN. GRN mutations are thought to cause FTLD-TDP through progranulin haploinsufficiency, as carriers of nearly all known pathogenic GRN mutations have roughly 50% loss of progranulin in blood and cerebrospinal fluid (Finch et al., 2009; Meeter et al., 2016b; Swift et al., 2024). In support of this possibility, some progranulin-insufficient iPSC-derived models exhibit phenotypes consistent with TDP-43 proteinopathy (de Majo et al., 2023; Marsan et al., 2023; Sung et al., 2024). Progranulininsufficient mice also recapitulate aspects of FTD-GRN. Grn+/− mice or mice with knock-in of an FTD-causing mutation (Grn+/R493X) develop FTD-related behavioral deficits, but no pathology (Filiano et al., 2013; Nguyen et al., 2018; Smith et al., 2024). Grn−/− and GrnR493X/R493X mice model the neuroinflammation of FTD-GRN, but exhibit modest neuronal loss and TDP-43 proteinopathy in the thalamus and pons at advanced ages (Nguyen et al., 2018; Tanaka et al., 2014; Wu et al., 2021; Zhang et al., 2020). It is not clear how progranulin insufficiency causes FTLD-TDP, but experimental models indicate several possibilities. Progranulin is necessary for maintaining lysosomal function, as homozygous mutations in GRN, resulting in complete loss of progranulin protein, cause the lysosomal storage disorder neuronal ceroid lipofuscinosis (NCL) (Smith et al., 2012). Patients with FTD-GRN exhibit signs of lysosomal impairment (Arrant et al., 2019; Boland et al., 2022; Götzl et al., 2014; Ward et al., 2017), which might contribute to buildup of pathologic TDP-43 in progranulin-insufficient neurons (Chang et al., 2017). Patients with FTD-GRN also exhibit neuroinflammation in affected brain regions, which is replicated in Grn−/− mice, and may contribute to development of TDP-43 pathology (Lant et al., 2014; Lui et al., 2016; Marsan et al., 2023; Sakae et al., 2019; Yin et al., 2010a; Zhang et al., 2020). However, the limited neurodegeneration and TDP-43 pathology of progranulin-insufficient mice has limited their use for investigating mechanisms of FTLD-TDP caused by GRN mutations. To investigate mechanisms of TDP-43 dysregulation due to progranulin insufficiency, we crossed progranulin-insufficient mice (Martens et al., 2012) with a human TDP-43 (hTDP) transgenic mouse line, which expresses wild-type human TDP-43 under the Thy1 promoter (Wils et al., 2010). Hemizygous (hTDP+) mice of this line, which are reported to express hTDP-43 at a level roughly equal to endogenous Tardbp, have a mild phenotype with no TDP-43 aggregation (Wils et al., 2010). Homozygous (hTDP++) mice, which express hTDP-43 at roughly twice the level of endogenous Tardbp, develop TDP-43 aggregates and severe neuroinflammation in motor and somatosensory cortex (Wils et al., 2010). Thus, analysis of both hTDP+ and hTDP++ mice allowed us to investigate whether progranulin insufficiency could either induce or exacerbate the TDP-43 aggregation and other phenotypes. Here, we describe characterization of FTD-related behavior and pathology in Grn:hTDP mice. Methods Animals Grn+/− mice (RRID:MMRRC_036771-JAX) (Martens et al., 2012) on a congenic C57BL/6J background were crossed with hemizygous human TDP-43 transgenic mice (hTDP+) (RRID:IMSR_JAX:012836) (Wils et al., 2010) on a C57BL/6J/SJL background. The resulting pups were crossed with Grn+/− mice on a C57BL/6J background for 2–6 generations before using for studies with hemizygous mice, and 7–9 generations before using for studies with homozygous mice. Grn+/−:hTDP+ mice from these crosses were bred with either Grn+/+ or Grn+/− mice to generate nontransgenic or hemizygous TDP-43 transgenic (hTDP+) mice, or bred with each other to generate nontransgenic or homozygous TDP-43 (hTDP++) transgenic mice with either Grn+/+, Grn+/−, or Grn−/− genotypes. Approximately equal distributions of male and female littermates were used for all experiments in this study. Animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care accredited facility with a 12-h light/dark cycle. Mice had free access to food (Envigo #7917) and water. All experiments were approved by the University of Alabama at Birmingham’s Institutional Animal Care and Use Committee (APN# 21572 & 22553). Behavioral assays Behavioral studies of hemizygous hTDP+ mice were conducted on three cohorts of mice (Fig. S1), with the first two cohorts having all six possible genotypes. The third cohort excluded Grn−/− mice and was intended to define age-of-onset for the low-dominance tube test phenotype of Grn+/−:hTDP+ mice. All behavior was performed during the light phase of the light/dark cycle after mice had habituated to the experimental room for at least 30 min. All materials used in the behavioral assays were cleaned with 2% chlorhexidine at the beginning and end of each day and 70% ethanol between each mouse. Rotarod Motor performance was assessed using a rotarod (Med Associates) where the speed accelerated from 4 rpm to 40 rpm over 5 min. Mice were tested once daily for three consecutive days, and the time to fall off the rotarod was averaged across the three days. Pole test Motor performance was assessed via the pole test (Li et al., 2014b), using a 60.96 cm vertical wooden pole with a 0.9525 cm diameter. Lab tape was wrapped around the pole every 3.81 cm to assist with grip. Mice were placed at the top of the pole, facing up, and the time it took for the mouse to turn around and descend the pole was recorded. Trials exceeding 60 s were considered a failure and assigned a time of 60 s. Mice were tested over 3 non-consecutive rounds conducted on the same day, and time to descend the pole was averaged over the 3 trials. Wire Hang Mice were placed on the bottom of a wire cage top, which was then gently inverted over a cage filled with standard bedding. A blinded experimenter recorded the time until mice released the cage top and fell to the bedding. Mice were tested over 3 non-consecutive rounds conducted on the same day, and time to fall was averaged over the 3 trials. Grooming Mice were placed in a recording chamber (Med Associates) and habituated for 10 min, then recorded for 10 min. An experimenter blinded to genotypes later calculated grooming time from the video recording. Open field Mice were placed into an open field apparatus (Med Associates) for 15 min. Infrared beam breaks recorded total ambulatory counts and time spent near the edges or in the center of the apparatus. Tube Test As previously described (Arrant et al., 2016), mice were paired against sex-matched non-cagemates of a different genotype. Dominance was assessed over 3 non-consecutive trials in a round robin design so that each mouse was tested against 3 different opponents. Mice were placed in either end of a small tube and allowed to meet in the middle. The first mouse that placed at least two paws outside of the tube was considered the loser, and then both mice were returned to their home cages. The tube was transparent and 30.5 cm long, but of varying diameters to account for the size of the mice. The side each mouse was initially placed on was alternated each trial, and tests that lasted longer than 2 min were aborted and run again at the end of the session. For examination of c-Fos in response to the tube test, mice were solo housed for one week prior to testing. A set of control mice was also solo housed, habituated, and kept in the room during testing, but did not participate in the tube test. Brains were collected 90 min after completing the first match. Tissue preparation Mice were anesthetized with approximately 200 mg/kg pentobarbital (Euthasol, MWI Animal Health, Catalog #O11344), then transcardially perfused with 0.9% saline using a peristaltic pump. The brains were removed and the right hemisphere was flash frozen in 2-methylbutane on dry ice, then placed into a cryotube for storage at −80 °C. The left hemisphere was placed into 4% paraformaldehyde for 48 h before being transferred to PBS. The left hemibrain was later cryoprotected with 30% sucrose and sliced into 30 μm sections using a sliding microtome (Leica) then stored in cryoprotectant (30% glycerol, 30% ethylene glycol in PBS) at −20 °C. qPCR RNA was isolated from brain or spleen tissue using Trizol reagent (ThermoFisher) and treated with DNase to remove any genomic DNA (Invitrogen TURBO DNA-free kit, ThermoFisher). cDNA was generated using iScript reverse transcriptase (Bio-Rad) then analyzed by qPCR with PowerTrack SYBR green master mix (ThermoFisher) using a QuantStudio3 system (ThermoFisher). Grn (Mm.PT.58.16608371.g), mouse Tardbp (Mm.PT.58.31054883), and human TARDBP (Hs. PT.58.26912658) were detected with PrimeTime probes (Integrated DNA Technologies). Autoregulatory splicing of mouse Tardbp was analyzed using primers previously described to detect exclusion of Tardbp intron 7 (White et al., 2018) (F: TTCATCTCATTTCAAATGTTTATGGAAG, R: ATTAACTGCTATGAATTCTTTGCATTCAG), or developed to detect multiple unique autoregulatory splicing events in Tardbp (D’Alton et al., 2015) (mTDP-1 F:TGATAAGGTTGCCCAGTCTCTTTGTG, R:TTGAAATGAGATGAACACCAAATCTTCC, mTDP-2 F:TGATAAGGTTGCCCAGTCTCTTTGTG, R:ATGAGATGAACTGGATTACCACCAAATCT, mTDP-5 F:AGATGATAAGGTTGCCCAGTCTCTTTG, R:GGATTACCACCAAATCTTCCACTTCTTT, mTDP-8 F:CAGATGATAAGGTTGCCCAGTCTCTTT, R:TTGTACTTTCATGCACGCTGATTCC). All transcripts were normalized to levels of Gapdh (F: GGGAAGCCCATCACCATCTT, R: GCCTTCTCCATGGTGGTGAA). Progranulin Enzyme Linked Immunosorbent Assay (ELISA) Progranulin levels were determined by ELISA (Adipogen). Brain tissue was homogenized in a mild lysis buffer (10 mM Tris, 10 mM NaCl, 3 mM MgCl2-6H2O, 1 mM EDTA, 0.05% NP-40) with protease inhibitors (Halt protease inhibitor cocktail, Thermo Scientific). Sample concentration was determined by BCA assay (ThermoFisher). Samples were diluted 1:1 with ELISA buffer and 50 μg of protein was loaded per well. Samples were quantitated based on a standard curve of recombinant progranulin run on the same plate. Progranulin levels were normalized to the amount of total protein loaded per well and corrected for background signal using samples from Grn−/− mice. Immunoblots Microdissected tissue was homogenized in either Triton X-100 lysis buffer (50 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate) or Ripa buffer (50 mM Tris, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 0.5% sodium deoxycholate, 5 mM EDTA, pH 7.5) then centrifuged at 5000 xg for 10 min. The supernatant was isolated as the soluble fraction and the pellet was dissolved in 8 M urea and used as the insoluble fraction. Sample concentrations were determined by BCA assay (ThermoFisher). Samples were diluted to a uniform concentration with lysis buffer and 2× Laemmli buffer (120 mM Tris, 4% SDS, 20% glycerol, 0.02% bromophenol blue, pH 6.8) before being loaded into 10% TGX polyacrylamide gels (Bio-Rad). Electrophoresis was performed with Tris-Glycine-SDS running buffer. Proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Immobilon F, MilliporeSigma) then blocked with protein free blocking buffer (ThermoFisher). Blocked membranes were incubated with primary antibodies at 4 °C overnight then incubated with IR-dye-conjugated secondary antibodies (Li-COR Biosciences) and scanned with an Odyssey scanner (Li-COR Biosciences). Antibodies used for these immunoblots were: TDP-43 (Proteintech, #10782–2-AP, RRID:AB_615042), pTDP-43 (Proteintech, #22309–1-AP, RRID:AB_11182943) and α-tubulin (MilliporeSigma, #T5168, RRID:AB_477579). Immunostaining 30 μm brain sections were blocked (10% normal goat serum, 1% powdered milk, 0.2% gelatin) then incubated with primary antibodies overnight at 4 °C. AlexaFlour-conjugated secondary antibodies (ThermoFisher) were used for immunofluorescence and the sections were then quenched with 1% Sudan Black B (Acros Organics). For chromogenic labeling with diaminobenzidine (DAB) (MP Biomedicals), biotinylated secondary antibodies (Vector Laboratories) and VectaStain Elite ABC reagent (Vector Laboratories) were used. The primary antibodies used in these experiments were: TDP-43 (Proteintech, #10782–2-AP, RRID:AB_615042), pTDP-43 (S409/410) (Proteintech, #22309–1-AP, RRID:AB_11182943), Ubiquitin (MilliporeSigma, #U5379–1VL, RRID:AB_477667), GFAP (Agilent, #Z0334, RRID:AB_10013382), CD68 (Bio-Rad, #MCA1957, RRID:AB_322219), MHC II (I-A/I-E), clone M5/114.15.2 (eBioscience, #14–5321, RRID:AB_467562), NeuN (MilliporeSigma, #ABN91, RRID:AB_11205760), Olig2 (MilliporeSigma, #AB9610, RRID:AB_570666), CD3 (Invitrogen, #16–0032-85, RRID: AB_468852), c-Fos (Cell Signaling Technology, #2250, RRID: AB_2247211), and Subunit C of Mitochondrial ATP Synthase (SCMAS) (Abcam, #ab181243, RRID:AB_2935765). Imaging and analysis Stained brain sections were imaged at a 20× magnification using an EVOS M5000 imaging system (ThermoFisher). 2–3 images were taken of each brain region for each mouse, then averaged to give final values. Images of cortical regions were centered on layer V, which is the primary site of TDP-43 aggregates in hTDP++ mice. The proportion of the image that was positively stained was calculated with FIJI (Schindelin et al., 2012) using a constant threshold. For analysis of TDP-43 subcellular localization, 2–3 20× z-stack images per mouse were taken using a Nikon Ti2 C2 confocal microscope and Nikon Elements 4.20.02 image capture software. The images were analyzed using CellProfiler (Carpenter et al., 2006; Stirling et al., 2021) in which DAPI staining was used to define the nucleus and NeuN staining to define the soma, because NeuN strongly labels neuronal nuclei and also labels the soma (Mullen et al., 1992). The resulting regions of interest were then overlaid onto the TDP-43 channel and the integrated density (pixel intensity × area) of the TDP-43 staining within nucleus and soma was measured. For experiments counting c-Fos + cells, NeuN+ cells, or pTDP-43 aggregates, images were taken of 2–3 sections per region then counted with the particle analyzer plugin on FIJI (Schindelin et al., 2012). Tissue area was consistent between images, and counts were averaged across the images. Tissue was sliced to give a distribution of sections around 300 μm apart to ensure that the same cell was not counted twice. Lysosomal enzyme activity assays Microdissected motor cortex tissue was homogenized in lysis buffer without protease inhibitors (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate) and centrifuged at 5000 xg for 10 min. Sample concentration was determined by BCA (ThermoFisher). Samples were run in duplicate with 5 μg of protein loaded per well for all assays. Cathepsin D (CatD) activity was determined as previously described (Arrant et al., 2018b), with a fluorogenic CatD/E substrate (7-methoxycoumarin-4-yl)acetyl-GKPILF FRLK(2,4-initrophenyl)-D-R-NH2 (Millipore/Sigma-Aldrich). Samples and substrate were incubated at 37 °C for 60 min and read on a Biotek Synergy LX plate reader. Fluorescence was measured with and without the CatD inhibitor pepstatin A (Fisher Scientific) and normalized to control samples run on the same plate. β-hexosaminidase A activity was determined as previously described (Arrant et al., 2019), using 4-methylumbelliferyl-2-acetamido-2-deoxy-6-sulfo-β-D-glucopyranoside fluorogenic substrate (Research Products International) that was conjugated to the fluorophore 4-methylumbelliferone (4-MU). Samples and substrate were incubated at 37 °C for 60 min before stopping the reaction with sodium carbonate buffer. Fluorescence was read on a Biotek Synergy LX plate reader and quantitated with a 4-MU standard curve. Dendritic spine morphology Analysis of dendritic spine morphology was performed as previously described (Cook et al., 2024; Greathouse et al., 2018; Weber et al., 2021). 10-month–old mice were anesthetized with 200 mg/kg pentobarbital (Euthasol, MWI Animal Health, Catalog #O11344) then transcardially perfused for 1 min with cold 1% paraformaldehyde (PFA, Sigma Aldrich), then for 10 min with cold 4% PFA with 0.125% glutaraldehyde (Fisher Scientific). Brains were immediately removed and placed into 4% PFA with 0.125% glutaraldehyde at 4 °C for 8–12 h. A Leica vibratome (VT1000S, speed 70, frequency 7) was used to slice brains into 250 μm sections which were then stored with 0.1% sodium azide (Fisher Scientific) at 4 °C. Iontophoretic microinjections were performed using a Nikon Eclipse FN1 upright microscope on an air table. Prior to injection, sliced brains were incubated in 1× PBS and Hoechst 33347 (Life Technologies) for 5 min at room temperature. LuciferYellow dye (ThermoFisher) was injected into layer II/III neurons in the mPFC using 2 nA of negative current. If the neuron was not completely filled, it was not used in the experiment. Multiple neurons were injected per mouse as quality of the tissue allowed. A total of 2–10 neurons were injected per mouse, averaging 3.4 neurons per mouse. Injected slices were stored in the dark at 4 °C until imaging. Injected neurons were imaged on a Nikon Ti2 C2 confocal microscope, collecting 60× z-stacks from dendrites that were parallel to the surface of the tissue section, within 80 μm working distance of microscope, between 40 and 120 μm from the soma, and did not overlap with other dendrites. Settings for collection of z-stacks were as follows: step size = 0.503 μm, image size 1024 × 1024 pixels, line averaging = 2. Z-stacks were deconvolved using Huygens Deconvolution System (16.05, Scientific Volume Imaging) then traced with Neurolucida 360 (2.70.1, MBF Biosciences). A voxel-clustering algorithm was used to reconstruct the dendritic spines then examined by experimenters to verify that all reconstructed spines were real and that no spines were excluded. Dendrite reconstruction images were exported to Neurolucida Explorer (2.70.1, MBF Biosciences) then analyzed for spine density, length, head diameter, and volume. Apical and basal arbors were analyzed separately. Only one apical and one basal dendrite were imaged per neuron. Experimenters remained blind to mouse genotype until the analysis was completed. RNA-sequencing RNA was isolated from mouse frontal cortex using RNeasy kits (Qiagen) and further purified using RNA Clean & Concentrator kits (Zymo Research). Samples were sent to Novogene, Inc. for sequencing. RNA libraries were prepared using poly-A selection and sequenced as 150 bp paired-end reads on an Illumina NovaSeq 6000 at ~80 million reads per sample. Reads were aligned using STAR to a custom genome derived from Gencode vM32 genome with vH43 TARDBP with the following settings: (–-twopassMode Basic –outSAMstrandField intronMotif –outSAMunmapped Within –outSAMtype BAM Unsorted SortedBy-Coordinate –quantMode TranscriptomeSAM) (Dobin et al., 2013). Bam files were indexed with samtools (Li et al., 2009). STAR aligned transcripts were quantified using Salmon (Patro et al., 2017). Salmon quantification files were imported into R using tximport (Soneson et al., 2015). Non-protein coding genes and genes with low reads were filtered out. After initial quality control checks, one Grn+/+:hTDP+ mouse was excluded as an extreme outlier from differential expression analysis. Differential expression analysis were performed using the R package DESeq2 (Love et al., 2014). Significance was defined as a log2 fold change greater than ±0.5 and an FDR-corrected (Benjamini-Hochberg method) p value of less than 0.05. Splicing analysis Splicing analysis was performed with rMats-Turbo (v4.3.0) (Shen et al., 2014; Wang et al., 2024) using a publicly hosted Docker Image (Xing Lab). STAR aligned bam files were organized into groups by genotype. Splicing analysis with rMats-Turbo between each group was performed using the following settings: (–-gtf gencode.vM32.hTARDBP. annotation.gtf -t paired –readLength 150 –variable-read-length –allow-clipping). Significance was defined as a difference in percent splice index (ΔPSI) greater than ±0.1 and an FDR-corrected (Benjamini-Hochberg method) p value of less than 0.05. To determine if differential splicing events identified by rMATS might be related to TDP-43 overexpression, differential splice sites were overlapped with TDP-43 binding regions obtained from CLIPdb (Yang et al., 2015) and separately from TDP-43 eCLIP-seq peak datasets generated from young (1.5-month-old), middle-aged (6-month-old), and old (24-month-old) mouse brains (Rhine et al., 2025) using bedtools intersect (v2.31.1) (Quinlan and Hall, 2010) with the parameters -wa -wb -s. Splice site coordinates were extended by ±500 bp flanking regions before overlap analysis. The TDP-43–associated motif “TGAATG”, which is highly bound by TDP-43 in mouse brain (Li et al., 2014a), was identified within significant splice sites extended by ±500 bp using the vcountPattern function from the Biostrings package (v2.74.1) in R (v4.4.3). Coordinates from datasets not aligned to the mm39 genome assembly were converted from mm10 to mm39 using UCSC LiftOver (Kent et al., 2002). Sashimi plots were generated using ggsashimi (v1.1.5) with the parameters -M 10 —C 2. To gain insight into associated cell types, differentially spliced genes located within 500 bp of either a TDP-43–associated motif or of a documented TDP-43 binding site were screened against a database of mouse brain cell-type markers (Yao et al., 2021) using Enrichr (Chen et al., 2013; Kuleshov et al., 2016). NanoString mouse neuroinflammation panel RNA was isolated from the motor and somatosensory cortex with a QIAGEN RNeasy Mini kit, treated with DNase, and analyzed using the NanoString Mouse Neuroinflammation Panel on an nCounter analysis system (NanoString). Gene expression was normalized to the house-keeping genes Aars, Ccdc127, Cnot10, Csnk2a2, Fam104a, Lars, Mto1, Supt7l, Tada2b, Tbp, and Xpnpep1 and analyzed using nSolver 4.0 software. Differential gene expression was determined using the ROTS package in R (Suomi et al., 2017), with significance defined as a log2 fold change greater than ±0.5 and an FDR-corrected (Benjamini-Hochberg method) p value of less than 0.05. Pathway significance scores and global significance scores for NanoString-annotated pathways were calculated using nSolver 4.0 software and graphed using the ggplot2 package in R (Wickham, 2016). Heatmaps were generated using the pheatmap package in R. Statistics and reproducibility All experiments were designed to compare littermates, and most experiments were repeated in at least two cohorts of mice (except for analysis of dendritic spines and RNA sequencing). Investigators were blinded by assigning mice unique numbers that were used to identify mice and samples for all experiments. Samples were decoded after data collection and analysis were complete. For most experiments individual mice comprised the experimental unit for sample sizes, which are provided in the figure legends. For analysis of dendritic spines, individual neurons comprised the experimental unit due to wide variability of neurons within mice. Data were generally only excluded from analysis for technical reasons such as poor sample quality, with the exception of one Grn+/+:hTDP+ mouse that was excluded as an outlier from RNA sequencing analysis as described above. Approximately equal numbers of male and female mice were analyzed for all outcome measures. We did not detect sex differences, so data from male and female mice are combined in all figures. Most data are shown with a bar graph representing group mean ± SEM, and individual data points shown in an overlaying scatter plot. Data presented in other formats are further described in the figure legends. Except where noted, data were analyzed using GraphPad Prism 10.0.2 (GraphPad Software). Prior to analysis, data were tested for normality (D’Agostino-Pearson test) and equal variance (Brown-Forsythe test). Data not meeting these assumptions were log transformed before further analysis. Data that did not meet these assumptions and contained many zero values (such as tube test winning percentage) were converted to ranks before analysis. This study generated several large datasets with repeated measures of age or brain region. Thus, most data were initially analyzed with a global analysis testing for effects and interactions of Grn and hTDP genotypes. Significant findings in this analysis were followed up with lower-order analyses and/or post-hoc testing to investigate pre-specified comparisons of greatest interest. Except for the tube test, behavior tests of hemizygous hTDP+ mice were analyzed by linear mixed effects model with factors of age, Grn, and hTDP using the lme4 (Bates et al., 2015) and lmertest (Kuznetsova et al., 2017) packages in R. Datasets with significant main effects or interactions were further analyzed for pre-planned pair-wise comparisons at each age using Sidak’s post-hoc test with the emmeans package in R. These comparisons investigated the effect of hTDP within each Grn genotype and the effect of Grn genotype within hTDP+ mice. All comparisons that reached statistical significance (p < 0.05) are noted in the figures. Win percentage per mouse in the tube test was analyzed by converting win percentages to ranks and analyzing by mixed effects analysis in GraphPad Prism. Due to the dependent nature of matches between two genotypes, each genotype pairing was analyzed separately. Data with significant main effects or interactions were further analyzed by Sidak’s post-hoc test for comparison of genotype differences at each age. Most pathology data were analyzed using 3-way repeated measures ANOVA in IBM SPSS Statistics, version 29.0.0 (IBM) with factors of Grn genotype and hTDP genotype, and either age or brain region as a within-subjects factor. Significant main effects or interactions were followed by lower-order 2-way ANOVA to further investigate the factors driving statistical significance. If an experiment was not a part of a repeated-measures design, results were analyzed with 2-way ANOVA with factors of Grn genotype and hTDP genotype. CellProfiler analysis of TDP-43 localization was analyzed using one-way ANOVA on the median integrated density value for each mouse. The differences in the number of c-Fos + cells were analyzed by region with t-test due to previously established regional specificity. For all analyses, significance was defined as a two-tailed p value of less than 0.05. Data availability RNA sequencing data have been deposited on GEO with accession number GSE305500. NanoString data have been deposited on Zenodo at https://doi.org/10.5281/zenodo.17214781. Results Generation and characterization of Grn+/−:hTDP+ mice We crossed Grn+/− mice with hemizygous human TDP-43 transgenic mice (hTDP+) to generate Grn+/−:hTDP+ mice. We then crossed these Grn+/−:hTDP+ mice with Grn+/− mice to generate nontransgenic (hTDP−) and hTDP+ mice on Grn+/+, Grn+/−, or Grn−/− backgrounds. Analysis of progranulin and TDP-43 transcript and protein levels in tissue from frontal cortex revealed the expected changes with each genotype, with no interaction between the two (Fig. 1). Since progranulin and TDP-43 were expressed as expected, we next examined behavioral phenotypes and pathology in these mice. Validation of progranulin and hTDP expression in Grn × hTDP+ mice. A) In Grn mice crossed with hemizygous hTDP+ mice, progranulin transcript decreased as expected in Grn+/− and Grn−/− mice and was not affected by hTDP expression (ANOVA effect of Grn, F(2,62) = 943.7, p < 0.0001, effect of hTDP, F(1,56) = 1.518, p = 0.2231, n = 5–17 mice per genotype). B) Mouse Tardbp transcript was not affected by either Grn (ANOVA effect of Grn, F(2,61) = 1.254, p = 0.2926) or hTDP genotype (ANOVA effect of hTDP, F(1,61) = 0.02592, p = 0.8726, n = 5–17 mice per genotype). C) Human TARDBP was only detected in hTDP+ groups, and levels were not significantly affected by Grn genotype (ANOVA effect of hTDP, F(1,60) = 812.6, p < 0.0001, effect of Grn, F(2,60) = 1.93, p = 0.154, Grn × hTDP, F(2,60) = 1.071, p = 0.3493, n = 5–17 mice per genotype). D) Progranulin protein levels decreased with loss of the progranulin allele and were not affected by hTDP expression (ANOVA, effect of Grn, F(2,28) = 69.08, p < 0.0001, effect of hTDP, F(1,28) = 0.2556, p = 0.6171, n = 5–7 mice per genotype). E) Ripa-soluble TDP-43 levels (detected with a non-species–specific antibody) were elevated in frontal cortex of hTDP+ mice, but were not changed with progranulin insufficiency (ANOVA, effect of hTDP, F(1,69) = 272, p < 0.0001, effect of Grn, F(2,69) = 1.173, p = 0.3156, Grn × hTDP, F(2,69) = 1.497, p = 0.2311, n = 5–19 mice per genotype). F) Representative immunoblot of the Ripa-soluble fraction, immunoblotted for TDP-43 and α-tubulin. Data obtained from frontal cortex of 10-month-old mice. In initial studies we analyzed two cohorts of mice in several behavior tests at 6 and 9 months of age (Fig. S1A). These tests assessed FTD-related behavior that is disrupted in progranulin-insufficient mice and motor behavior that is mildly impaired in hTDP+ mice. Some mice were euthanized for brain collection at around 10 months of age, with the rest continuing to further behavior testing at 12 and 15 months of age. These mice were then euthanized for brain collection at 18 months of age. Several phenotypes were observed at the 6 month time point in these first two cohorts, so to define age-of-onset we analyzed a third cohort of mice beginning at 3 months of age, with mice euthanized for brain collection at around 10 months of age (Fig. S1B). Progranulin insufficiency did not worsen motor deficits in hTDP+ mice The primary behavioral phenotype reported for hTDP+ mice is mild motor impairment (Wils et al., 2010). We therefore investigated whether progranulin insufficiency might exacerbate this phenotype by analyzing behavior with the pole test, rotarod, and open field. We observed mild impairment on the pole test in hTDP+ mice at all ages from 6 to 15 months (Fig. S1C). This impairment was not altered by progranulin insufficiency. Similarly, we observed mild impairment of rotarod performance in hTDP+ mice from ages 6–15 months (Fig. S1D), which was not altered by progranulin insufficiency. We did not observe clear effects of either hTDP expression or Grn genotype on total ambulations in the open field test (Fig. S1E). We further analyzed open field behavior by measuring the percentage of ambulations detected in the center of the open field (Fig. S1F). Movement in the center of the field was modestly reduced by hTDP expression, regardless of Grn genotype. We then analyzed a third cohort with testing beginning at 3 months of age (Fig. S1B). In this cohort, we replaced the open field test with the wire hang test, due to lack of a robust phenotype in the open field in the first two cohorts. In this new cohort, hTDP+ mice also exhibited impairment of pole test performance (Fig. S1H), which was not affected by progranulin insufficiency. Pole test deficits were detected as early as 3 months of age, and worsened as mice aged to 6 and 9 months. In contrast to our earlier data, hTDP+ mice in this cohort did not exhibit impairment of rotarod performance (Fig. S1I). We also observed impaired performance in the wire hang test in hTDP+ mice (Fig. S1J), which was not affected by Grn genotype. In summary, we observed mild motor impairment in hTDP+ mice from 3 to 15 months of age. This impairment was not affected by progranulin insufficiency, as both Grn+/− and Grn−/−:hTDP+ mice exhibited similar performance as Grn+/+:hTDP+ mice. Having observed no effect of progranulin-insufficiency on motor phenotypes of hTDP+ mice, we next assessed the interaction of progranulin insufficiency and hTDP expression on FTD-related behavior. hTDP overexpression worsened social dominance deficits in progranulin-haploinsufficient mice Using the cohorts described above, we investigated FTD-relevant phenotypes that have been previously reported in progranulin-insufficient mice (Arrant et al., 2016; Filiano et al., 2013; Frew and Nygaard, 2021; Ghoshal et al., 2012; Lui et al., 2016; Smith et al., 2024; Yin et al., 2010b). We assessed grooming behavior in all genotypes of cohorts 1 and 2, but none of the groups had significant changes in grooming time (Fig. S1G). We next performed the tube test for social dominance using only Grn+/+ and Grn+/− nontransgenic and hTDP+ mice, as we have not previously observed social dominance deficits in Grn−/− mice (Arrant et al., 2016). Data from all three cohorts were combined for analyzing tube test performance, as all genotypes were represented in all three cohorts. We observed the expected age-dependent low-dominance phenotype of Grn+/−:hTDP− mice (Arrant et al., 2016), which gained statistical significance at 12 months of age (Fig. 2A). Grn+/+:hTDP+ mice did not have a significant change in social dominance at any single age, but did exhibit an age-dependent trend for reduced social dominance with a similar effect size as Grn+/−:hTDP− mice (Fig. 2B). In contrast, Grn+/−:hTDP+ mice exhibited a robust low-dominance phenotype at all ages tested from 3 to 15 months (Fig. 2C). To further determine if hTDP expression reduced dominance in Grn+/− mice, we tested Grn+/−:hTDP− mice against Grn+/−:hTDP+ littermates at each age, and found that Grn+/−:hTDP+ mice also exhibited low dominance against Grn+/−: hTDP− mice at several ages (Fig. 2D). hTDP expression exacerbates social dominance deficits in Grn+/− mice. A–C) We assessed the effects of progranulin insufficiency, hTDP expression, and progranulin insufficiency + hTDP expression on social dominance by testing Grn+/−:hTDP–, Grn+/+:hTDP+, and Grn+/−:hTDP+ mice against wild-type littermates. A) Analysis of mouse winning percentage revealed development of an age-dependent low-dominance phenotype in Grn+/− mice, with significantly low dominance at age 12 months (Mixed effects analysis, effect of genotype, F(1,50) = 6.863, p = 0.0116, age × genotype, F(4,85) = 3.467, p = 0.0113, * = p = 0.0106 by Sidak’s post-hoc test, n = 9–22 mice per genotype per age). B) hTDP+ mice also exhibited age-dependent changes in dominance behavior (Mixed effects analysis, age × genotype, F(4,78) = 6.462, p = 0.0002, n = 9–20 mice per genotype per age), though they did not exhibit a statistically significant phenotype at any single age upon post-hoc testing. C) In contrast, Grn+/−:hTDP+ mice exhibited a robust low-dominance phenotype throughout the testing period (Mixed effects analysis, effect of genotype, F(1,43) = 29.51, p < 0.0001, * = p < 0.05, *** = p < 0.001, **** = p < 0.0001 by Sidak’s post-hoc test, n = 8–19 mice per genotype per age). D) To further assess the shift in dominance induced by hTDP expression in Grn+/− mice, we also tested Grn+/−:hTDP+ mice against Grn+/−:hTDP− littermates. Analysis of winning percentages for each mouse also revealed overall lower dominance in Grn+/−:hTDP+ mice, with individual significance at ages 9 and 15 months (Mixed effects analysis, effect of genotype, F(1,50) = 10.5, p = 0.0021, age × genotype, p = 0.0002, ** = p = 0.0074 and **** = p < 0.0001 by Sidak’s post-hoc test, n = 11–20 mice per genotype per age, which were tested in three independent cohorts (Fig. S1)). Tube test illustrations were created at biorender.com Arrant, A. (2026) https://BioRender.com/hzhv9y3. In summary, testing three cohorts of mice at ages ranging from 3 to 15 months of age revealed dramatic impairment of social dominance in Grn+/−:hTDP+ mice. Though less robust, we observed the expected low social dominance phenotype of nontransgenic Grn+/− mice, and trends for impaired social dominance in Grn+/+:hTDP+ mice. Progranulin insufficiency and hTDP expression may thus exert additive effects on social dominance in Grn+/−:hTDP+ mice. Progranulin insufficiency did not promote TDP-43 aggregation in cortex of hTDP+ mice To determine if the impaired social dominance behavior of Grn+/−:hTDP+ mice might be related to development of FTD-related pathology, we investigated whether progranulin insufficiency promoted TDP-43 aggregation in hTDP+ mice. Our initial analysis focused on motor cortex, which is one of the regions with highest TDP-43 aggregation in hTDP++ mice (Wils et al., 2010). We thus reasoned that it might be most vulnerable to formation of TDP-43 aggregates in this line. We analyzed levels of Ripa-insoluble TDP-43 in hTDP+ mice at 10 and 18 months of age. All hTDP+ mice exhibited elevated levels of insoluble full-length TDP-43 (Fig. 3A–C), but these levels were not increased by progranulin insufficiency. Instead, Grn−/−:hTDP+ mice had lower levels of insoluble TDP-43 than Grn+/+:hTDP+ mice at 18 months of age (Fig. 3B). We also attempted to analyze levels of insoluble pTDP-43 (Ser409/410), which accumulates in hTDP++ mice, but observed only a faint signal in hTDP+ mice that did not differ from nontransgenic mice (Fig. 3C). We also conducted immunostaining for pTDP-43, which reveals TDP-43 aggregates in hTDP++ mice (Fig. 3D–F). We observed no clear pTDP-43 signal in 10-month–old mice (Fig. 3D). While we observed sparse pTDP-43–immunoreactive puncta in 18-month–old mice (Fig. 3E, F), these did not differ by hTDP genotype and were not affected by progranulin insufficiency. Progranulin insufficiency did not induce TDP-43 aggregation in hTDP+ mice. A) Motor cortex of 10-month–old hTDP+ mice had higher levels of Ripa-insoluble TDP-43 than hTDP− mice, but progranulin insufficiency did not impact levels of insoluble TDP-43 (ANOVA, effect of hTDP, F(1,30) = 187.1, p < 0.0001, effect of Grn, F(2,30) = 0.3519, p = 0.7062, Grn × hTDP, F(2,30) = 0.412, p = 0.666, n = 5–7 mice per genotype). B) In motor cortex of 18-month–old hTDP+ mice, progranulin deficiency reduced levels of insoluble TDP-43 (ANOVA, effect of hTDP, F(1,50) = 157.6, p < 0.0001, effect of Grn, F(1,50) = 1.568, p = 0.2186, Grn × hTDP, F(2,50) = 2.965, p = 0.0607, n = 5–12 mice per genotype, * = p = 0.0377 by Sidak’s post-hoc test). C) Representative immunoblot of Ripa-insoluble TDP-43 and pTDP-43 with hTDP++ mice as a positive control showing that motor cortex of 18-month–old hTDP+ mice failed to accumulate pTDP above background levels, regardless of Grn genotype. D) Similarly, pTDP-43 immunostaining revealed no TDP-43 aggregation in motor cortex of 10-month–old hTDP+ mice, a region in which hTDP++ mice develop numerous TDP-43 aggregates. E, F) We observed sparse pTDP-43 immunoreactive puncta in motor cortex of 18-month–old hTDP+ mice, but there was no effect of hTDP (ANOVA, F(1,46) = 1.379, p = 0.2463) or Grn (ANOVA, effect of Grn, F(2,46) = 1.51, p = 0.2317, Grn × hTDP, F(2,46) = 1.461, p = 0.2426) on the number of these aggregates (n = 5–12 mice per genotype). G, H) We observed no changes in ubiquitin immunoreactivity in motor cortex of 18-month–old hTDP+ mice (ANOVA effect of hTDP, F(1,53) = 0.008754, p = 0.9258, effect of Grn, F(1,53) = 0.715, p = 0.4939, n = 5–13 mice per group), a region in which hTDP++ mice exhibit robust increases in ubiquitin. Due to the highly significant effects of hTDP in A and B, post-hoc testing was limited to assessing the effect of Grn within hTDP+ mice using Sidak’s test. All significant results are noted on the graphs. Representative images are shown in D, F, and H with 50 μm scale bars. Boxes denote insets shown in higher magnification on the bottom row with 25 μm scale bars. Arrowheads in D and F denote pTDP-43–immunoreactive puncta. As a final measure of protein aggregation, we immunostained 18-month–old mice for ubiquitin and quantified immunoreactivity in motor cortex, a region in which hTDP++ mice exhibit elevated ubiquitin immunoreactivity (Fig. 3G, H). Similar to results with pTDP-43, we observed no elevation in ubiquitin immunoreactivity in any hTDP+ group, regardless of Grn genotype. Taken together, these data provide no evidence that progranulin insufficiency induced TDP-43 aggregation in hTDP+ mice up to 18 months of age. Social dominance behavior is heavily dependent on medial prefrontal cortex (mPFC) (Wang et al., 2011), so we also analyzed levels of Ripa-insoluble TDP-43 in tissue from frontal cortex of hTDP+ mice, which included mPFC and surrounding cortical regions. Similar to motor cortex, all hTDP+ mice exhibited increases in Ripa-insoluble TDP-43, but these were not affected by Grn genotype (Fig. S2A, B). Thus, we observed no signs of TDP-43 aggregation that might drive the exacerbated social dominance deficits of Grn+/−:hTDP+ mice. Progranulin insufficiency modestly increases insoluble TDP-43 in spinal cord of hTDP+ mice In addition to the motor cortex, hTDP++ mice develop TDP-43 aggregation in the spinal cord (Wils et al., 2010). To determine if progranulin insufficiency might thus induce TDP-43 aggregation in hTDP+ mice, we analyzed levels of Ripa-insoluble TDP-43 in spinal cord of 10-month–old hTDP+ mice. Similar to motor cortex, we observed increases in insoluble TDP-43 in hTDP+ mice of all Grn genotypes (Fig. S2C, D). Among hTDP+ mice, Grn+/−:hTDP+ mice showed no signs of increased insoluble TDP-43 versus Grn+/+:hTDP+ mice, but Grn−/−:hTDP+ mice had modestly increased levels of insoluble TDP-43. As in motor cortex, Ripa-insoluble pTDP-43 levels were very low in hTDP+ mice, indicative of a lack of robust TDP-43 aggregation (Fig. S2D). Thus, complete progranulin deficiency modestly increases insoluble TDP-43 in the spinal cord, but this is not associated with worsening of motor function (Fig. S1). hTDP expression did not worsen lysosomal phenotypes in progranulin-insufficient mice We next investigated whether the social dominance deficits of Grn+/−:hTDP+ mice could be driven by worsening of the lysosomal phenotypes of progranulin-insufficient mice. Consistent with the development of NCL in people homozygous for GRN mutations, Grn−/− mice accumulate lysosomal storage material beginning around 6 months of age, which is immunoreactive for the lipofuscin marker SCMAS (Ahmed et al., 2010; Arrant et al., 2018b; Filiano et al., 2013; Ghoshal et al., 2012; Yin et al., 2010a). Grn−/− mice also exhibit increases in lysosomal proteins and enzyme levels (Arrant et al., 2018b; Arrant et al., 2019; Gotzl et al., 2018; Götzl et al., 2014). Grn+/− mice do not accumulate SCMAS, but do exhibit subtle increases in lysosomal proteins and enzyme activity (Arrant et al., 2017; Arrant et al., 2018a). We therefore conducted SCMAS immunostaining in both 10- and 18-month–old hTDP+ mice and measured activity of the lysosomal enzymes β-Hexosaminidase A (HexA) and Cathepsin D (CatD) in cortical tissue. We observed the expected increase in SCMAS throughout the brain of Grn−/− mice at both ages, though this was not affected by hTDP expression (Fig. S3A—C). We also observed no increase in SCMAS with hTDP expression in Grn+/− mice. Similarly, HexA and CatD activity were both increased in Grn−/− mice but were not affected by hTDP genotype (Fig. S3D, E). Thus, we observed no effect of hTDP expression on lysosomal phenotypes in progranulin-insufficient mice. Hemizygous hTDP expression did not exacerbate neuroinflammation in progranulin-insufficient mice We next investigated whether progranulin insufficiency and hemizygous hTDP expression might interact to worsen neuroinflammation. Progranulin suppresses inflammation, and people with GRN mutations and Grn−/− mice develop elevated markers of gliosis (Ahmed et al., 2007; Ahmed et al., 2010; Filiano et al., 2013; Ghoshal et al., 2012; Heller et al., 2020; Lant et al., 2014; Lui et al., 2016; Sakae et al., 2019; Woollacott et al., 2020; Yin et al., 2010a). However, Grn+/− mice do not develop signs of increased inflammation, even at advanced ages (Ahmed et al., 2010; Filiano et al., 2013). To determine if hTDP expression could either induce reactive gliosis in Grn+/− mice or exacerbate gliosis in Grn−/− mice, we immunostained for CD68 as a marker of reactive microglia and GFAP as a marker of reactive astrocytes in 10- and 18-month−old mice. We observed the expected increases in CD68 and GFAP in Grn−/− mice at both ages, but hTDP expression did not significantly worsen either of these measures (Fig. S3F—I). We also observed no signs of induction of reactive gliosis by hTDP expression in Grn+/− mice. Grn+/−:hTDP+ mice had decreased c-Fos + cells in the mPFC after the tube test In summary, neither TDP-43 aggregation, severe lysosomal impairment, nor reactive gliosis appear to have caused the low social dominance of Grn+/−:hTDP+ mice. These negative results are consistent with development low social dominance in Grn+/− mice without any of these phenotypes (Filiano et al., 2013). We therefore investigated whether Grn+/−:hTDP+ mice exhibited changes in neuronal activity that might explain their behavioral deficits. The neural circuits driving tube test behavior have been well-characterized, and include mediodorsal thalamic (MDT) input to the medial prefrontal cortex (mPFC), with additional influence of the basolateral amygdala (BLA), nucleus accumbens (NAc), and other regions (Fetcho et al., 2023; Scheggia et al., 2022; Shan et al., 2022; Wang et al., 2011; Xin et al., 2025; Zhang et al., 2022; Zhou et al., 2017). We therefore used c-Fos immunoreactivity to probe neuronal activation in these brain regions to determine whether abnormal neuronal activity might drive the decreased social dominance of Grn+/−:hTDP+ mice. After a week of solo-housing to minimize background c-Fos expression, 10-month− old wild-type and Grn+/−:hTDP+ littermates were tested in the tube test assay, then euthanized 90 min after completing their first match. These Grn+/−:hTDP+ mice replicated the low social dominance phenotype that was identified in previous cohorts (Fig. 4B). We did not observe differences between Grn+/−:hTDP+ mice and their wild-type opponents in the number of c-Fos + cells in the mediodorsal thalamus (MDT), the basolateral amygdala (BLA), or the nucleus accumbens (NAc) (Fig. 4E–G). However, there was a significant decrease in the number of c-Fos + cells in the medial prefrontal cortex (mPFC) of Grn+/−:hTDP+ mice (Fig. 4C,D). Thus, among several brain regions selected for relevance to the tube test, Grn+/−:hTDP+ mice exhibited hypoactivation only in the mPFC. Hypoactivation of medial prefrontal cortex in Grn+/−:hTDP+ mice after performing the tube test. A) 10-month–old Grn+/−:hTDP+ mice were subjected to three unique matches against wild-type littermate opponents, then euthanized 90 min later for assessment of c-Fos immunoreactivity. B) As with prior cohorts, these Grn+/−:hTDP+ mice had a low social dominance phenotype (Binomial test, * = p = 0.047). C) Representative c-Fos staining of the medial prefrontal cortex (mPFC). D – G) Grn+/−:hTDP+ mice had fewer c-Fos + cells than their wild-type opponents in the mPFC (* = p = 0.0376, t-test), but had a similar number of c-Fos + cells in the mediodorsal thalamus (t-test, p = 0.9089), basolateral amygdala (t-test, p = 0.8287), and ventral striatum (t-test, p = 0.7301). Data in D–G were obtained from 11 wild-type mice and 13 Grn+/−:hTDP+ mice. Panel A was created at biorender.com Arrant, A. (2026) https://BioRender.com/mz2aslh. Independent effects of Grn and hTDP on dendritic spines in mPFC In previous work, we observed abnormal dendritic arborization and spine morphology in mPFC neurons of Grn+/− mice (Arrant et al., 2016; Cook et al., 2024). We therefore investigated whether dendritic spine morphology and density were altered in mPFC of Grn+/−:hTDP+ mice. We analyzed dendritic spines from layer II/III mPFC neurons of an additional cohort of Grn:hTDP mice using LuciferYellow dye-filling and confocal microscopy. Apical and basal dendrites of layer II//III neurons in the mPFC receive projections from different brain regions, so we analyzed these separately (Anastasiades and Carter, 2021; Collins et al., 2018). The apical dendritic spines, which primarily receive projections from other cortical regions (Anastasiades and Carter, 2021; Collins et al., 2018; Larkum et al., 2007), had no significant changes in spine density, length, head diameter, or volume (Fig. 5C–E). However, we observed changes in basal dendrites (Fig. 5F–H), which receive input from the mediodorsal thalamus (MDT) and interneurons (Anastasiades and Carter, 2021; Collins et al., 2018; Larkum et al., 2007). hTDP expression significantly reduced basal spine density (Fig. 5F), regardless of Grn genotype. In contrast, Grn genotype affected both both spine length (Fig. 5G) and head diameter (Fig. 5H), with Grn+/− mice having longer spines with smaller diameter than Grn+/+ mice, regardless of hTDP genotype. These independent effects of hTDP and Grn genotype may combine to contribute to hypoactivation of mPFC in Grn+/−:hTDP+ mice during performance of the tube test. Distinct effects of progranulin insufficiency and hTDP expression on dendritic spine density and morphology in medial prefrontal cortex. A) We analyzed dendritic spine density and morphology by filling layer II-III pyramidal neurons in medial prefrontal cortex of 10-month–old mice with Lucifer Yellow dye. A representative neuron is shown with a 30 μm scale bar. B) Dendrites were analyzed by 3D reconstruction. The scale bar represents 50 μm. Mushroom spines are shown in blue, all other spines are shown in gray. C) On apical dendrites, we detected no significant change to spine density (ANOVA effect of Grn, F(1,114) = 0.7172, p = 0.3988, effect of hTDP, F(1,114) = 0.5093, p = 0.4769, Grn × hTDP, F(1,114) = 0.04647, p = 0.8297). D) Apical spine length was also similar among all genotypes (ANOVA effect of Grn, F(1,114) = 0.9587, p = 0.3296, effect of hTDP, F(1,114) = 0.6074, p = 0.4374, Grn × hTDP, F(1,114) = 0.3913, p = 0.5329). E) We also detected no significant change to apical spine head diameter (ANOVA effect of Grn, F(1,114) = 2.163, p = 0.1441, effect of hTDP, F(1,114) = 1.706, p = 0.1941, Grn × hTDP, F(1,114) = 0.4481, p = 0.5046). However, on basal dendritic spines we found independent effects of both hTDP and Grn genotype, resulting in additive abnormalities in spines of Grn+/−:hTDP+ mice. F) hTDP expression significantly reduced spine density (ANOVA, effect of hTDP, F(1,103) = 4.245, p = 0.0419). G, H) In contrast, progranulin insufficiency altered spine morphology, with a significant increase in spine length (ANOVA, effect of Grn, F(1,103) = 8.074, p = 0.0054) and a significant decrease in spine head diameter (ANOVA, effect of Grn, F(1,103) = 3.995, p = 0.0483). n = 21–37 neurons from 5–6 mice per genotype, with 2− 10 dendrites analyzed per mouse. Progranulin haploinsufficiency altered splicing of TDP-43 RNA targets To further investigate mechanisms underlying the phenotypes observed in Grn+/−:hTDP+ mice, we microdissected frontal cortical tissue and performed bulk RNA sequencing at a depth of ~80 million paired end reads per sample, with the goal of analyzing differential gene expression and splicing. We detected very few differentially expressed genes after correcting for multiple comparisons (FDR < 0.05) and a Log2 fold change threshold (Log2FC > ± 0.5) (Table S1). A major function of TDP-43 is regulating RNA splicing (Polymenidou et al., 2011; Tollervey et al., 2011), so we also analyzed alternative splicing in our sequencing dataset using Replicate Multivariate Analysis of Transcript Splicing Turbo (rMATS-Turbo) (Shen et al., 2014). We identified many differentially-spliced junctions in Grn+/+:hTDP+ mice relative to wild type mice (Fig. 6A, Table S1), defined as a difference in percent splice index (ΔPSI) > 0.1 and an FDR-corrected p value <0.05. To determine if these splicing events might be attributed to TDP-43, we screened for events occurring within 500 bp of either a TDP-43 binding site identified by CLIP-seq (Yang et al., 2015) or a TDP-43 associated binding motif. Approximately 40% of the differentially-spliced junctions in Grn+/+:hTDP+ mice occurred near a TDP-43 binding site or a TDP-43 motif. We also detected differential splicing in known TDP-43 targets such as autoregulation of endogenous Tardbp and a recently reported target of TDP-43 overexpression, Psmd14 (Carmen-Orozco et al., 2024) (Fig. 6C). The majority of these splicing events were skipped exons, which is consistent with prior reports of TDP-43 overexpression and gain-of-function mutations (Fig. S4A, Table S1) (Carmen-Orozco et al., 2024; Fratta et al., 2018). Thus, our initial analysis revealed splicing changes in Grn+/+:hTDP+ mice that are consistent with TDP-43 overexpression. Progranulin haploinsufficiency altered splicing of TDP-43 targets. A, B) rMATS-turbo analysis of alternative splicing detected many differentially-spliced junctions in both Grn+/+ and Grn+/−:hTDP+ mice versus Grn+/+:hTDP− littermates. In both cases, approximately 40% of differential splicing events occurred within 500 bp of TDP-43 binding sites listed in CLIP-seq databases (Rhine et al., 2025; Yang et al., 2015) or of a common TDP-43 binding motif in mouse brain (Li et al., 2014a). C) Differential splicing events also included exon skipping events known to occur with TDP-43 overexpression in mice, such as in Psmd14 (Carmen-Orozco et al., 2024). D) Despite the relatively similar number of differential splicing events, the differentially-spliced genes detected in Grn+/+ and Grn+/−:hTDP+ mice exhibited little overlap. E) Subsequent direct comparison of Grn+/+ and Grn+/−:hTDP+ mice with rMATS-turbo also revealed many differential splicing events, with nearly 40% occurring within 500 bp of either a TDP-43 binding site or motif. For volcano plots in A, B, and E, significance was defined as a Delta PSI > ± 0.1 and an FDR-corrected p value <0.05. PSI = percent spliced in. n = 5–6 mice per genotype. In all panels Ctrl refers to Grn+/+:hTDP− littermates. We observed a roughly similar number of differentially-spliced junctions in Grn+/−:hTDP+ mice relative to controls, with a similar percentage occurring within 500 bp of known or potential TDP-43 binding sites (Fig. 6B). As in Grn+/+:hTDP+ mice, the majority of these events were skipped exons, including in the known TDP-43 target Psmd14 (Fig. 6C) (Carmen-Orozco et al., 2024). Despite these similarities, there was little overlap between the differential splicing events in Grn+/+:hTDP+ mice and Grn+/−:hTDP+ mice versus control mice (Fig. 6D). To better understand this discrepancy, we directly compared Grn+/−:hTDP+ mice to Grn+/+:hTDP+ mice and identified 332 differentially spliced junctions, 129 of which occurred near known or potential TDP-43 binding sites (Fig. 6E). The genes containing these 129 differentially spliced junctions were enriched for neuronal markers, consistent with the neuronal expression of TDP-43 in this transgenic mouse line. However, gene ontology analysis did not reveal enrichment for specific pathways. The differential splicing in Grn+/−:hTDP+ mice versus Grn+/+:hTDP+ mice did not appear to indicate a general change in TDP-43 activity. For example, there were 96 differential exon skipping events near TDP-43 binding sites or motifs, which may be some of the most likely events to be caused by TDP-43 overexpression (Carmen-Orozco et al., 2024; Fratta et al., 2018). Of these events, Grn+/+:hTDP+ mice had higher exon skipping in 47 events and Grn+/−:hTDP+ mice had higher exon skipping in 49 events. Thus, these data are more consistent with discordant changes in RNA splicing between Grn+/+:hTDP+ mice and Grn+/−:hTDP+ mice rather than a general change in TDP-43 function. To determine if the discordant splicing of Grn+/−:hTDP+ mice versus Grn+/+:hTDP+ mice reflected general effects of progranulin haploinsufficiency, we analyzed splicing in Grn+/−:hTDP− mice versus controls. We observed fewer differentially spliced junctions in Grn+/−: hTDP− mice than were detected in either hTDP+ group (Fig. S4A, B). However, a similar proportion (36%) occurred near TDP-43 binding sites or TDP-43 motifs. Only a few of these differential splicing events were also observed in Grn+/−:hTDP+ mice (Fig. S4C), so progranulin haploinsufficiency alone likely does not explain the discordant splicing of Grn+/−:hTDP+ mice. Finally, we analyzed for signs of TDP-43 dysregulation that might explain the discordant splicing of Grn+/−:hTDP+ mice. We analyzed TDP-43 cytoplasmic and nuclear localization in frontal cortex of Grn+/+:hTDP+ mice and Grn+/−:hTDP+ mice. Both hTDP+ groups had increased nuclear/cytoplasmic ratios relative to nontransgenic mice. However, progranulin haploinsufficiency did not significantly alter TDP-43 subcellular localization (Fig. S5A—D). We also considered whether aberrant TDP-43 autoregulation might affect TDP-43 function in Grn+/−:hTDP+ mice. While the human TDP-43 transgene consists only of TARDBP cDNA and is not subject to autoregulatory splicing (Wils et al., 2010), we did detect signs of autoregulatory splicing of endogenous Tardbp transcripts in cortex of both Grn+/−:hTDP+ and Grn +/+:hTDP+ mice. We thus used qPCR to measure changes in multiple autoregulatory splice sites in Tardbp (D’Alton et al., 2015; White et al., 2018). We detected autoregulatory Tardbp splicing at most of these sites in Grn+/+:hTDP+ mice, and Grn+/−:hTDP+ mice exhibited generally similar changes (Fig. S5E—I). In summary, we observed signs of differential TDP-43− related RNA splicing in Grn+/−:hTDP+ mice compared to Grn+/+:hTDP+ mice in the absence of clear changes to TDP-43 subcellular localization or aggregation, and with generally similar signs of autoregulation of endogenous Tardbp. These data might indicate a subtle influence of progranulin insufficiency on TDP-43 function. Progranulin insufficiency did not increase TDP-43 aggregation in hTDP++ mice Given our finding that progranulin insufficiency does not induce TDP-43 pathology in hTDP+ mice, but may exert subtle influences on TDP-43-dependent splicing activity, we next investigated whether progranulin insufficiency would exacerbate the TDP-43 pathology of hTDP++ mice. Similar to hTDP+ mice, Grn × hTDP++ mice exhibited the expected changes in Grn and hTDP expression, with no interaction between the two genotypes (Fig. S6). Since hTDP++ mice develop the highest levels of TDP-43 aggregates in motor and somatosensory cortex (Wils et al., 2010), our analysis of hTDP++ mice focused primarily on these regions. To assess the burden of TDP-43 aggregates in hTDP++ mice, we first immunoblotted the Ripa-insoluble fraction of motor cortex for TDP-43. As expected, all hTDP++ groups had higher levels of insoluble TDP-43 than nontransgenic mice (Fig. 7A). However, levels of insoluble TDP-43 were not affected by progranulin insufficiency (Fig. 7A). hTDP++ mice also had elevated levels of lower molecular weight TDP-43–immunoreactive bands, but these did not differ by Grn genotype (Fig. 7B). hTDP++ mice also exhibited a significant increase in pTDP-43 (S409/410), but this was not affected by Grn genotype (Fig. 7C). We also immunostained brain slices with a pTDP-43 (S409/410) antibody and observed the expected increase in pTDP-43–immunoreactive puncta in the cortex of hTDP++ mice, which was particularly heavy in layer V (Wils et al., 2010). However, the burden of these pTDP-43 aggregates was not affected by Grn genotype (Fig. 7E, F). Finally, we observed ubiquitin accumulation in cortical layer V of hTDP++ mice, but this was also not affected by Grn genotype (Fig. 7G, H). Progranulin insufficiency did not worsen TDP-43 aggregation in hTDP++ mice. A) Motor cortex from 21-day–old hTDP++ mice contained higher levels of Ripa-insoluble TDP-43 than nontransgenic mice (ANOVA effect of hTDP, F(1,65) = 363.8, p < 0.0001), but this was not significantly affected by Grn genotype (ANOVA effect of Grn, F(2,65) = 0.5357, p = 0.5878, Grn × hTDP, F(2,65) = 1.552, p = 0.2196, n = 9–16 mice per group). B) These TDP-43 immunoblots also revealed low-molecular weight (LMW) TDP-43 immunoreactivity that was increased in hTDP++ mice but not influenced by Grn genotype (ANOVA effect of hTDP, F(1,65) = 199.2, p < 0.0001, effect of Grn, F(2,65) = 0.07928, p = 0.9239, Grn × hTDP, F(2,65) = 1.055, p = 0.3541). C) Motor cortex of hTDP++ mice also exhibited a significant increase in Ripa-insoluble pTDP-43 (Ser409/410), but this was not affected by Grn genotype (ANOVA effect of hTDP, F(1,64) = 66.76, p < 0.0001, effect of Grn, F(2,64) = 0.5068, p = 0.6048, Grn × hTDP, F(2,64) = 1.538, p = 0.2226, n = 8–16 mice per group). Representative immunoblots are shown in D. E, F) hTDP++ mice exhibited an increase in pTDP-43− immunoreactive puncta in layer V of several cortical regions (3-way RM ANOVA effect of hTDP, F(1,40) = 79.984, p < 0.001, n = 9–18 mice per genotype), but this was not affected by Grn genotype (3-way RM ANOVA effect of Grn, F(2,40) = 0.048, p = 0.954, 2-way RM ANOVA of only hTDP++ mice effect of Grn, F(2,26) = 0.2769, p = 0.7603). G, H) hTDP++ mice also had increased ubiquitin immunoreactivity in layer V of several cortical regions (3-way RM ANOVA effect of hTDP, F(1,84) = 65.513, p < 0.001), which was not affected by progranulin insufficiency (3-way RM ANOVA, effect of Grn, F(2,84) = 1.646, p = 0.199, 2-way RM ANOVA of only hTDP++ mice effect of Grn, F(2,47) = 0.01618, p = 0.984, n = 10–19 mice per genotype). Due to the highly significant effect of hTDP in all analyses, we limited post-hoc testing in A–C, E, and G to comparison of Grn+/−:hTDP++ and Grn−/−:hTDP++ mice against Grn+/+:hTDP++ mice to conserve statistical power for the comparisons of greatest interest. Sidak’s test was used in A–C and Dunnett’s test was used to compare these genotypes within each region in E and G. However, none of these tests revealed significant differences between Grn genotypes of hTDP++ mice. Representative images are shown in F and H with 50 μm scale bars for the top two rows of images. Boxes in F denote insets shown in higher magnification on the bottom row with 25 μm scale bars. Arrowheads in F denote pTDP-43–immunoreactive puncta. We next measured accumulation of cytoplasmic TDP-43, another hallmark of TDP-43 pathology (Neumann et al., 2006), using immunostaining and confocal imaging. To validate this method, we initially assessed TDP-43 subcellular localization in hTDP+ and hTDP++ mice on a Grn+/+ background (Fig. S7A—D), and found that most TDP-43 remained in the nucleus even in transgenic mice, though hTDP++ mice exhibited an increase in cytoplasmic TDP-43. We therefore investigated whether progranulin insufficiency might promote accumulation of cytoplasmic TDP-43 by comparing TDP-43 localization in hTDP++ mice on Grn+/+, Grn+/−, and Grn−/− backgrounds. However, we observed no effect of progranulin insufficiency on levels of cytoplasmic TDP-43, nuclear TDP-43, or the nuclear/cytoplasmic TDP-43 ratio (Fig. S7E—H). In summary, progranulin insufficiency did not impact the degree of cortical TDP-43 aggregation or mislocalization in hTDP++ mice. Progranulin haploinsufficiency increased markers of reactive glia in TDP-43 transgenic mice Due to their severe phenotype, hTDP++ mice do not reach ages at which progranulin insufficiency induces neuroinflammation (Wils et al., 2010). However, hTDP++ mice independently develop striking neuroinflammation, particularly in layer V of the motor and somatosensory cortices, in which they have a high burden of TDP-43 aggregates. We therefore measured markers of reactive glia and neuroinflammation in the Grn:hTDP++ mouse cross. Consistent with the initial report (Wils et al., 2010), we observed significant increases in CD68 and GFAP immunoreactivity throughout cortical layer V of all 21-day–old hTDP++ mice versus nontransgenic littermates (Fig. 8A–D). In motor and somatosensory cortex, Grn+/−:hTDP++ mice had more CD68 (Fig. 8A, B) and GFAP (Fig. 8C, D) immunolabeling than Grn+/+:hTDP++ mice, showing that progranulin haploinsufficiency such as occurs in FTD-GRN can exacerbate reactive gliosis in the context of TDP-43 aggregation. Dysregulated neuroinflammation in Grn−/−:hTDP++ mice. A, B) 21-day–old hTDP++ mice exhibited increased CD68 immunoreactivity in several cortical regions (3-way RM ANOVA effect of hTDP, F(1,78) = 82.537, p < 0.001, region × hTDP interaction, F(1.718,134.014) = 7.557, p = 0.001). Due to the young age of the mice, we observed no independent effect of Grn genotype (3-way RM ANOVA effect of Grn, F(2,78) = 2.287 p = 0.108). A follow-up analysis of all hTDP++ mice revealed regional differences in CD68 immunoreactivity (2-way RM ANOVA effect of region, F(1.581,72.72) = 6.339, p = 0.0055, n = 11–19 mice per genotype), with significantly increased CD68 immunoreactivity in Grn+/−:hTDP++ mice versus Grn+/+:hTDP++ mice in motor and somatosensory cortex (* = p < 0.05 and ** = p < 0.01 by Dunnett’s post-hoc test). C, D) Similarly, 21-day–old hTDP++ mice exhibited increased GFAP immunoreactivity in several cortical regions (3-way RM ANOVA effect of hTDP, F(1,69) = 134.543, p < 0.001, region × hTDP interaction, F(1.843,127.179) = 21.848, p < 0.001). There was no main effect of Grn genotype (3-way RM ANOVA effect of Grn, F(2,69) = 1.63, p = 0.203) on GFAP immunoreactivity, but we did observe a region × Grn interaction (3-way RM ANOVA, F(3.686,127.179) = 2.529, p = 0.048). Follow-up analysis of only hTDP++ mice again revealed an effect of Grn genotype on GFAP immunoreactivity (2-way RM ANOVA effect of Grn, F(2,40) = 4.733, p = 0.0143, n = 8–18 mice per genotype), with significantly increased GFAP immunoreactivity in Grn+/−:hTDP++ mice versus Grn+/+:hTDP++ mice in motor and somatosensory cortex (* = p < 0.05 by Dunnett’s post-hoc test). Due to the highly significant main effect of hTDP genotype on CD68 and GFAP immunoreactivity in A and C, lower order ANOVA and post-hoc testing focused on analyzing the effect of Grn within hTDP++ mice. Representative images are shown in B and D with 50 μm scale bars for the top two rows of images. Boxes denote insets shown in higher magnification on the bottom row with 25 μm scale bars. E) NanoString analysis of tissue from motor and somatosensory cortex showed that all three hTDP++ groups clustered separately from hTDP− mice on principal component analysis. F) A heatmap of the z-score of each gene in the NanoString panel again showed a distinct pattern of transcriptional changes in Grn−/−:hTDP++ mice versus the other hTDP++ groups. G) All three hTDP++ groups shared 136 differentially expressed genes (46 upregulated, 90 downregulated) versus control littermates (log2 > ± 0.5, FDR < 0.05). Another 50 genes were upregulated and 38 downregulated only in Grn−/−:hTDP++ mice relative to control mice. H) NanoString-assigned pathway analysis showed divergent changes in pathway scores of Grn−/−:hTDP++ mice than in other hTDP++ groups versus controls. E) Volcano plot of differentially expressed genes in Grn−/−:hTDP++ mice versus Grn+/+:hTDP++ mice. Colors indicate a log2 fold change greater than ±0.5 and an FDR-corrected p value of less than 0.05. n = 4–8 mice per genotype. Grn−/−:hTDP++ mice showed similar trends for increased CD68 relative to Grn+/+:hTDP++ mice (Fig. 8A). This trend did not reach statistical significance, likely due to high variability in Grn−/−:hTDP++ mice. In contrast, Grn−/−:hTDP++ mice showed no trend for increased GFAP (Fig. 8B). An important factor in interpreting these data is that at age 21 days, these mice are too young to exhibit the typical age-dependent increases in reactive glial markers of Grn−/− mice (Ahmed et al., 2007; Filiano et al., 2013; Ghoshal et al., 2012; Yin et al., 2010a). Grn−/−:hTDP++ mice had a dysregulated inflammatory response to cortical TDP-43 pathology We next conducted transcriptomic analysis of tissue from motor and somatosensory cortex of wild-type (Grn+/+: hTDP−) and Grn+/+, Grn+/−, and Grn−/−:hTDP++ littermates using a NanoString mouse neuroinflammation panel. All hTDP++ mice exhibited widespread changes in gene expression relative to wild type mice, with 136 genes that were significantly upregulated or downregulated (log2FC > ± 0.5, FDR < 0.05) in all three hTDP++ groups (Fig. 8G). However, Grn−/−:hTDP++ mice exhibited a distinct pattern of gene expression relative to the other hTDP++ groups (Fig. 8E–G). Analysis of functional pathways annotated by NanoString revealed multiple pathways in which Grn−/−:hTDP++ mice exhibited divergent changes than the other hTDP++ groups (Fig. 8H). Notably, Grn−/−:hTDP++ mice exhibited an upregulation of genes related to NF-κB signaling, DNA damage, and astrocyte function, but downregulation of the oligodendrocyte and microglial function pathways compared to Grn+/+:hTDP++ and Grn+/−:hTDP++ mice (Fig. 8H). Full pathway scores, including gene lists, have been deposited on Zenodo.com at https://doi.org/10.5281/zenodo.17214781. We then performed direct pair-wise comparison to determine differentially-expressed genes between Grn−/−:hTDP++ mice and Grn+/+:hTDP++ mice (Fig. 8I). Grn was the most downregulated gene, and most of the other downregulated genes were associated with oligodendrocytes or myelination (Fig. 8I). We therefore performed immunostaining for the oligodendrocyte marker Olig2 to investigate potential loss of oligodendrocytes, but observed no loss of Olig2+ cells in any group (Fig. S8A, B). Among the upregulated genes, we considered Ms4a4a to be of particular interest (Fig. 8I). Ms4a4a regulates levels of Trem2, and thus can influence microglial activation states (Deming et al., 2019; Rosner et al., 2024; You et al., 2023). Trem2 is a key regulator of microglial transition to disease-associate microglia (DAM) phenotypes (Deczkowska et al., 2018; Keren-Shaul et al., 2017), and mediates age-dependent increases in DAM markers in older Grn−/− mice (Reifschneider et al., 2022). We therefore individually assessed levels of several DAM markers on the NanoString panel (Fig. S8C). This analysis revealed that Grn−/−:hTDP++ mice had higher expression of multiple DAM markers than Grn+/+:hTDP++ mice. Thus, even at this young age (21 days) progranulin insufficiency may drive microglia toward DAM phenotypes. Taken together, these data indicate distinct effects of progranulin haploinsufficiency and complete progranulin deficiency on neuroinflammation in hTDP++ mice. Grn+/−:hTDP++ mice had higher immunoreactivity for reactive glial markers, but exhibited similar transcriptomic changes as Grn+/+:hTDP++ mice. In contrast, Grn−/−:hTDP++ mice did not have higher immunoreactivity for reactive glial markers, but had distinct transcriptomic changes that may indicate oligodendrocyte impairment and microglial transition toward DAM phenotypes. Altered neuroinflammatory markers in hippocampus of Grn−/−:hTDP++ mice Having found evidence that progranulin dysregulated neuroinflammation in the cortex of hTDP++ mice, we analyzed neuroinflammation in additional regions. While the cortex of hTDP++ mice develops many TDP-43 aggregates, most of the neurons remain alive by 21 days of age. In contrast, the CA3 region of the hippocampus is severely degenerated by 21 days, with almost no detectable neurons remaining and thus very few TDP-43 aggregates (Fig. 9A, B). Thus, the cortex (high TDP-43 aggregation, low neuronal loss) and CA3 (low detectable TDP-43 aggregation, high neuronal loss) provided opportunities for analyzing neuroinflammation under distinct conditions. Neuroinflammatory markers in CA3 of hTDP++ mice. A) Loss of NeuN immunoreactivity showed that the CA3 region of the hippocampus was severely degenerated in hTDP++ mice. B) pTDP-43 immunostaining showed that the CA3 region lacked pTDP-43 puncta, likely due to the dramatic loss of neurons. C, D) hTDP++ mice had more CD68 immunoreactivity in the CA3 region than nontransgenic mice, but Grn−/−:hTDP++ mice had a trend for CD68 staining than Grn+/+:hTDP++ mice (ANOVA, effect of hTDP, F(1,73) = 256.5, p < 0.0001, effect of Grn, F(2,56) = 3.919, p = 0.0242, listed p value indicates result of Sidak’s post-hoc test, n = 8–18 mice per genotype). E, F) hTDP++ mice had more GFAP staining in CA3, but this was not affected by Grn genotype (ANOVA, effect of hTDP, F(1,65) = 13.44, p = 0.0005, effect of Grn, F(2,65) = 0.8622, p = 0.427, Grn × hTDP, F(2,65) = 0.6887 p = 0.5059, n = 6–18 mice per genotype). G, H) hTDP++ mice had IgG deposition in CA3, but Grn−/−:hTDP++ mice lacked IgG deposition, suggesting an impaired adaptive immune response (2-way ANOVA, main effect of hTDP, F(1,23) = 49.07, p < 0.0001, hTDP × Grn interaction, F(2,23) = 3.474, p = 0.048, * = p = 0.0269 by Sidak’s post-hoc test, n = 3–7 mice per genotype). I, J) Further immunostaining of hippocampal tissue revealed no signs of antigen presenting cells (I, MHC-II) or infiltrating T cells (J, CD3), which did not indicate a local adaptive immune response. K) However, qPCR of spleen from nontransgenic and hTDP++ mice demonstrated expression of human TARDBP, which is consistent with expression of Thy1-hTDP in peripheral immune cells such as T cells (* = t-test, p = 0.0054, n = 3 mice per genotype). This supports the potential for a peripheral immune response to explain IgG deposition in hTDP++ mice. Representative images in D, F, H, I, and J show images of CA3 with 100 μm scale bars, with magnified images from the hTDP++ mice in the lower panels of D, F, and H shown with 50 μm scale bars. Immunostaining of CA3 for reactive glial markers revealed a trend for increased CD68, but not GFAP immunoreactivity in Grn−/−:hTDP++ mice (Fig. 9C–F). We also identified IgG deposition, a marker of an adaptive immune response, in the CA3 of Grn+/+ and Grn+/−:hTDP++ mice (Fig. 9G, H) (Hemmer et al., 2002; Lim et al., 2007; Zhou et al., 2023). This IgG deposition was almost entirely absent from Grn−/−:hTDP++ mice (Fig. 9G, H), suggesting an impaired adaptive immune response. We did not detect MHC II or CD3+ cells that would indicate a local immune response (Fig. 9I, J). However, analysis of splenocytes revealed robust hTDP expression in hTDP++ mice, consistent with Thy1 expression patterns (Fig. 9K). The IgG deposition in hTDP++ mice might therefore be driven by a peripheral immune response, and the lack of IgG deposition in Grn−/−:hTDP++ mice may be related to progranulin’s known role in regulating B- and T-cell function, as well as antigen presentation (Chen et al., 2020; Hu et al., 2014; Schmitz et al., 2020; Wei et al., 2014; Zhang et al., 2024). Discussion Despite the fact that GRN mutations cause FTD with TDP-43 pathology, the relationship between progranulin and TDP-43 remains unclear. While progranulin-insufficient mice model many aspects of FTD-GRN, they develop only sparse TDP-43 pathology at advanced ages, and only with complete loss of progranulin (Tanaka et al., 2014; Wu et al., 2021; Zhang et al., 2020). In an effort to investigate progranulin’s regulation of TDP-43 in vivo, we crossed progranulin-insufficient mice with human TDP-43 transgenic mice and characterized behavioral phenotypes, TDP-43 mislocalization and aggregation, and neuroinflammation. Despite a lack of change in TDP-43 localization or aggregation, neuroinflammation, or lysosomal phenotypes, we observed worse social dominance deficits in Grn+/−:hTDP+ mice versus nontransgenic Grn+/− littermates (Fig. 2). This low social dominance was associated with hypoactivation of the mPFC during social dominance testing, and with changes to dendritic spine density and morphology in mPFC neurons (Figs. 4, 5). Transcriptomic analysis of Grn+/−:hTDP+ frontal cortex revealed altered splicing of TDP-43 RNA targets (Fig. 6), perhaps indicating a subtle effect of progranulin haploinsufficiency on TDP-43 function. We also observed an increase in reactive glial markers in the cortex of Grn+/−:hTDP++ mice (Fig. 8), and signs of dysregulated immune function in Grn−/−:hTDP++ mice (Figs. 8, 9). Taken together, these data provide insight into potential mechanisms by which progranulin insufficiency drives FTD progression. A major goal of this study was to determine if progranulin insufficiency could either induce or worsen TDP-43 aggregation in TDP-43 transgenic mice. It remains unclear why nearly all people with GRN mutations specifically develop FTLD-TDP type A pathology (though often with a variety of other co-pathologies) (Mackenzie, 2007). Though Grn−/− mice accumulate only sparse TDP-43 aggregates at advanced ages (Tanaka et al., 2014; Wu et al., 2021; Zhang et al., 2020), other animal models demonstrate regulation of TDP-43 aggregation or toxicity by progranulin. Overexpression of progranulin reduces axonopathy and neurodegeneration in animal models of TDP-43 overexpression or knockdown (Beel et al., 2018; Chitramuthu et al., 2017; Laird et al., 2010). Progranulin overexpression also reduces insoluble TDP-43 in transgenic mice expressing mutant TDP-43 (Beel et al., 2018). In Caenorhabditis elegans, progranulin haploinsufficiency worsens TDP-43–induced phenotypes (Laird et al., 2010; Salazar et al., 2015). While these models do not develop TDP-43 aggregation, they do support the hypothesis that progranulin regulates some aspect of TDP-43 function. However, we did not observe any significant increases in TDP-43 insolubility, cytoplasmic localization, or aggregation in the brain of progranulin-insufficient hTDP+ or hTDP++ mice (Figs. 3, 7). These data are consistent with a recent study that examined the effects of progranulin insufficiency on TDP-43 transgenic mice and mutant TDP-43 knock-in mice (Yang et al., 2025). The lack of TDP-43 aggregation might be due to shortcomings of each mouse cross. For example, the modest TDP-43 expression of hTDP+ mice may not be sufficient to produce aggregation even in Grn+/− or Grn−/− mice, and hTDP++ mice do not survive long enough to experience age-dependent deficits due to progranulin insufficiency that might exacerbate their pathology. While the transgenic mouse line used in this study expressed human TDP-43 (Wils et al., 2010), it is likely that we would have obtained similar results with expression of mouse TDP-43, as transgenic mice expressing mouse TDP-43 exhibit similar phenotypes as lines expressing human TDP-43 (Arnold et al., 2013; Cannon et al., 2012; Carmen-Orozco et al., 2024; Gordon et al., 2019; Shan et al., 2010; Stallings et al., 2010; Swarup et al., 2011; Tsai et al., 2010; Xu et al., 2010; Xu et al., 2011; Yang et al., 2022). Additionally, we do not expect that human and mouse TDP-43 might differentially interact with progranulin, as progranulin is not thought to regulate TDP-43 through direct protein-protein interaction. Our negative findings might be due to species differences in regulation of TDP-43 localization or aggregation, though it is notable that human models of progranulin insufficiency also report variable levels of TDP-43 aggregation (Almeida et al., 2012; de Majo et al., 2023; Raitano et al., 2015; Sung et al., 2024; Valdez et al., 2017). The major phenotype that emerged in hTDP+ mice was the low social dominance of Grn+/−:hTDP+ mice (Fig. 2). Analysis of c-Fos expression after performance of the tube test suggested the mPFC as a key region underlying these deficits (Fig. 4). This is consistent with studies showing that social dominance in the tube test is primarily driven by circuits driving excitability of mPFC pyramidal neurons (Fetcho et al., 2023; Wang et al., 2011; Zhou et al., 2017). Further analysis of mPFC neurons revealed changes to both the morphology and density of basal dendritic spines in Grn+/−:hTDP+ mice (Fig. 5), with distinct effects of progranulin haploinsufficiency and TDP-43 overexpression on mPFC dendritic spines. The changes to spine morphology appeared to be driven by progranulin haploinsufficiency, similar to our prior work with Grn+/− mice (Cook et al., 2024). The reduction of spine density was primarily driven by TDP-43 overexpression, which is similar to studies showing reduction of dendritic arborization and spine density with TDP-43 overexpression (Dyer et al., 2021; Handley et al., 2017; Herzog et al., 2020). Thus, our findings suggest an association of mPFC hypoactivation after the tube test with reduced spine density and abnormal morphology. While we did not assess neuronal activity, others have reported deficits in MDt-mPFC conductivity in Grn+/− mice (Nagy et al., 2019), and patients with FTD-GRN exhibit thalamocortical hyperconnectivity prior to disease onset that converts to hypoconnectivity with symptomatic FTD (Flagan et al., 2025; Lee et al., 2019). Thus, the tube test deficits of Grn+/− mice and Grn+/−:hTDP+ mice may be driven by impaired cortical activation that models that of symptomatic patients with FTD-GRN. Further investigation will be necessary to test this hypothesis. TDP-43 regulates splicing of thousands of transcripts, and TDP-43 overexpression increases exon skipping in both mouse and human cells (Carmen-Orozco et al., 2024). We analyzed splicing of TDP-43 RNA targets in hTDP+ mice, which based on results from other mouse models may provide a more sensitive indicator of TDP-43 function than development of TDP-43 pathology (Carmen-Orozco et al., 2024; Fratta et al., 2018; White et al., 2018). Consistent with prior studies, we observed increases in autoregulatory splicing of endogenous Tardbp and exon skipping of many TDP-43 RNA targets in hTDP+ mice (Fig. 6A–C). However, there was little overlap in the differential splicing of Grn+/−:hTDP+ and Grn+/+:hTDP+ mice versus controls, and Grn+/−:hTDP+ exhibited many differential splicing events versus Grn+/+:hTDP+ mice (Fig. 6D, E). These findings may indicate subtle changes in TDP-43 function in Grn+/−:hTDP+ mice, but are subject to a notable caveat. TDP-43 overexpression has opposite effects on RNA splicing as TDP-43 proteinopathy, which causes loss of TDP-43 function and inclusion of cryptic exons (Mehta et al., 2023). These cryptic splicing events are poorly conserved between mice and humans (Ling et al., 2015), so progranulin’s effects on TDP-43–mediated splicing should be further investigated using human cells and models of TDP-43 loss-of-function. Many prior studies have reported that Grn−/− mice have elevated neuroinflammatory responses to infection, brain injury or pathology (Martens et al., 2012; Menzel et al., 2017; Takahashi et al., 2017; Tanaka et al., 2013; Yin et al., 2010a). We observed increased markers of reactive glia in Grn+/−:hTDP++ mice (Fig. 8), with similar trends in Grn−/−:hTDP++ mice. These data indicate that even partial loss of progranulin can exacerbate neuroinflammation under some conditions. Grn−/−:hTDP++ mice exhibited high variability in cortical CD68 immunoreactivity that hinted at a potentially more complex phenotype. Transcriptomic analysis revealed that Grn−/−:hTDP++ mice exhibited complex changes in markers of neuroinflammation consistent with a dysregulated glial response to TDP-43 aggregation. These changes included an increase in DAM markers, which is similar to changes observed with aging in Grn−/− mice (Gotzl et al., 2019; Reifschneider et al., 2022; Tejwani et al., 2024). This suggests that Grn−/− microglia are more prone to adopting DAM phenotypes in response to both aging and neurodegenerative pathology. In CA3, Grn−/−:hTDP++ mice failed to develop the IgG deposition observed in Grn+/+:hTDP++ and Grn+/−:hTDP++ mice. This may indicate impairment of an adaptive immune response, which could be caused abnormal antigen processing or presentation (Houser et al., 2022; Schmitz et al., 2020) or by dysregulated T cell responses. Progranulin regulates cytokine secretion and is involved in T cell maturation and proliferation (Chen et al., 2020; Hu et al., 2014; Wei et al., 2014), so Grn−/−:hTDP++ mice may not produce appropriate T cell populations to induce B cell class switching or restrain the immune response. Progranulin insufficiency may also impact adaptive immune function in humans, as patients with FTD-GRN have a high rate of comorbid autoimmune diseases and increased expression of genes associated with adaptive immunity in peripheral immune cells (Miller et al., 2013; Sawyer et al., 2022). The neuroinflammatory phenotypes of Grn+/−:hTDP++ and Grn−/−:hTDP++ mice, despite a similar burden of TDP-43 aggregation as Grn+/+:hTDP++ mice, may provide a model of the excessive inflammation and aggressive disease course of FTD-GRN. Patients with FTD-GRN have faster progression of brain atrophy and decline in social cognition, increased inflammatory markers, and higher serum neurofilament light chain than patients with other FTD subtypes (Bocchetta et al., 2023; Bossu et al., 2011; Fenoglio et al., 2024; Heller et al., 2020; Meeter et al., 2016a; Poos et al., 2022; Sakae et al., 2019; Saracino et al., 2021; Staffaroni et al., 2022; van der Ende et al., 2019; Woollacott et al., 2020). The degree to which aberrant glial and immune responses to TDP-43 proteinopathy contribute to this aggressive disease course will be an important topic for future study. In conclusion, we have shown that crossing progranulin-insufficient mice and human TDP-43 transgenic mice results in worse phenotypes than occur in either parent line. Data from Grn+/−:hTDP+ mice highlight the importance of the mPFC in mediating FTD-related social dominance deficits of progranulin-insufficient mice. Data from Grn−/−:hTDP++ mice showed that progranulin insufficiency dysregulates neuroinflammatory responses to TDP-43 aggregation, which may have implications for progression of FTD-GRN. These data raise questions for future studies on progranulin’s regulation of immune function and TDP-43–dependent RNA splicing. Additionally, the more robust phenotypes of Grn+/−:hTDP+ mice may provide a useful model for preclinical testing of therapies that require the presence of an intact Grn allele. Supplementary Material Acknowledgements This work was supported by the National Institute of Neurological Disorders and Stroke grants R01NS128031, T32NS095775, and R01NS134798, National Institute on Aging grants F30AG085889, R01AG061800, R01AG054719, and P30AG086401, National Cancer Institute grants R01CA258248 and P30CA013148, and National Institute of General Medical Sciences grant T32GM008361. NanoString RNA analysis was performed in collaboration with UAB’s NanoString laboratory. We thank Dr. Sami Barmada for advice on RNA sequencing for alternative splicing. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.nbd.2026.107475. Footnotes CRediT authorship contribution statement Anna K. Cook: Conceptualization, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Benjamin Lin: Conceptualization, Formal analysis, Writing – original draft, Writing – review & editing. Yumo Song: Formal analysis, Writing – review & editing. Kelsey M. Greathouse: Formal analysis, Investigation, Writing – review & editing. Azariah K. Kaplelach: Investigation, Writing – review & editing. Mackenzie L. Love: Investigation, Writing – review & editing. Skylar E. Davis: Investigation, Writing – review & editing. Anna C. Stoll: Investigation, Writing – review & editing. Justin A. Hall: Investigation, Writing – review & editing. Ahmad R. Hakim: Investigation, Writing – review & editing. Jakub F. Hel: Investigation, Writing – review & editing. Giacynta A. Vollmer: Investigation, Writing – review & editing. Alexandria C. Howard: Investigation, Writing – review & editing. Noelle H. Cooper: Investigation, Writing – review & editing. Phaedra N. Manuel: Investigation, Writing – review & editing. Juliana M. Eberhardt: Investigation, Writing – review & editing. C. Ryan Miller: Conceptualization, Formal analysis, Writing – review & editing. Ashley S. Harms: Conceptualization, Formal analysis, Writing – review & editing. Jeremy H. Herskowitz: Conceptualization, Formal analysis, Writing – review & editing. Lindsay F. Rizzardi: Formal analysis, Writing – review & editing. Andrew E. Arrant: Conceptualization, Formal analysis, Writing – review & editing. Conflict of interest The authors declare no competing financial interests. Data availability RNA sequencing data have been deposited on GEO with accession number GSE305500. NanoString data have been deposited on Zenodo at https://doi.org/10.5281/zenodo.17214781. References Associated Data Supplementary Materials Data Availability Statement RNA sequencing data have been deposited on GEO with accession number GSE305500. NanoString data have been deposited on Zenodo at https://doi.org/10.5281/zenodo.17214781. RNA sequencing data have been deposited on GEO with accession number GSE305500. NanoString data have been deposited on Zenodo at https://doi.org/10.5281/zenodo.17214781.