Muscle Inflammatory Infiltrates in the Low Dox rNLS8 (TDP-43ΔNLS) Mouse Model of ALS

Research Study Summary: Using multiplex immunofluorescence and Biospective’s PERMITS™ image analysis of gastrocnemius muscle from 14 Low Dox rNLS8 (TDP-43ΔNLS) mice and 14 tTA control mice, we found that CD45⁺ immune cell density was roughly twice that of controls, that the infiltrate appears to be predominantly CD11b⁺ myeloid cells, and that across individual Low Dox mice a higher CD45⁺ burden was strongly associated with more fully denervated neuromuscular junctions (r = 0.81) and lower CMAP amplitude (r = −0.71). Muscle inflammation therefore provides a quantitative, mechanistically informative endpoint that complements NMJ innervation and electrophysiology in preclinical ALS efficacy studies.

Study at a glance: muscle inflammation and NMJ integrity in Low Dox rNLS8 mice (Biospective in-house study).

Skeletal muscle inflammation matters in ALS because immune cells accumulate in patient and rodent muscle around neuromuscular junctions (NMJs), and interventional studies indicate that these cells can accelerate the loss of motor nerve terminals rather than merely clearing up after it (Wang, 2017; Nógrádi, 2025).

Amyotrophic lateral sclerosis (ALS) is defined clinically by the degeneration of upper and lower motor neurons, which produces progressive weakness, paralysis and, ultimately, respiratory failure (Hardiman, 2017). For decades the disease was framed almost entirely around the motor neuron cell body. Spatiotemporal pathology studies changed that view: in SOD1 mutant mice, endplate denervation is already substantial before ventral root axons or spinal motor neurons are lost, and an autopsy from a patient with sporadic ALS showed muscle denervation alongside apparently intact motor neurons (Fischer, 2004). NMJ denervation is also a consistent feature of TDP-43 mouse models, and dying-back mechanisms have been proposed for TDP-43-driven disease, placing the NMJ and its surrounding muscle environment at the centre of disease pathogenesis (Lépine, 2022; Alhindi, 2023).

Muscle is an active participant, not a passive target.

Muscle fibres, fibro-adipogenic progenitors, terminal Schwann cells and resident immune cells all respond to the loss of innervation, and several of these responses feed back on NMJ stability (Shefner, 2023). In denervated muscle, fibro-adipogenic progenitors activate STAT3–IL-6 signalling that promotes atrophy and fibrosis, and blocking this pathway preserved muscle in SOD1 mice (Madaro, 2018). Inflammasome components such as NLRP3 are also altered in ALS mouse muscle and have been explored as a muscle-derived biomarker (Moreno-García, 2021).

Immune infiltrates gather where synapses are lost.

In SOD1^G93A rats, myeloid markers including CD11b and CD68 rise in limb muscle as disease advances, accompanied by elevated IL-1β and TNF-α and reactive glial changes adjacent to NMJs (Van Dyke, 2016). In SOD1 mice, complement C5a signalling through C5aR1 on macrophages drives their recruitment into muscle, and deleting C5aR1 reduced both macrophage numbers and endplate denervation while improving grip strength (Wang, 2017). Mast cells and neutrophils also infiltrate paralytic rat muscle and the quadriceps of ALS patients at autopsy, where they associate with motor endplates (Trias, 2017; Trias, 2018).

Most of this evidence came from SOD1 models. The most direct TDP-43-relevant data now come from a study that found leukocyte and macrophage infiltration in muscle biopsies from people with ALS, reproduced the finding in several TDP-43 mouse models from presymptomatic stages onward, and showed that infiltrates concentrate in NMJ-rich muscle regions (Nógrádi, 2025). Proteomic analysis and the enrichment of CCL2-expressing cells around NMJs in hTDP-43 mice and patient muscle implicated the CCL2–CCR2 axis, and local antibody treatment in hTDP-43 mice (with either CCL2-neutralizing antibodies or non-specific IgG) reduced leukocyte infiltration and lessened NMJ denervation (Nógrádi, 2025).

Immune cells are not uniformly harmful, however. Slow-progressing SOD1 mice mount an early muscle response in which infiltrating macrophages adopt a pro-regenerative phenotype that supports myogenesis and preserves CMAP (Margotta, 2023), and in the rNLS8 model microglia are required for full motor recovery once TDP-43 expression is switched off (Spiller, 2018). Measuring both the magnitude and the identity of the immune infiltrate is therefore essential for interpreting what a therapy is doing in muscle.

The Low Dox rNLS8 model reproduces TDP-43-driven neuromuscular pathology by expressing human TDP-43 with a defective nuclear localization signal in neurons, which drives cytoplasmic TDP-43 accumulation, NMJ denervation and motor decline (Walker, 2015). Biospective's Low Dox induction paradigm produces this phenotype at a slower, more drug-testable pace than the conventional Off Dox paradigm (Biospective in-house characterization).

The rNLS8 mouse is bigenic: a neurofilament heavy chain (NEFH) promoter-driven tTA line is crossed with a tetO-hTDP-43ΔNLS responder line, so transgene expression in neurons is suppressed by doxycycline and induced when doxycycline is removed (Walker, 2015). In the original description, induction produced insoluble, phosphorylated cytoplasmic TDP-43, loss of endogenous nuclear TDP-43, muscle denervation, motor neuron loss and progressive motor impairment, and returning doxycycline after disease onset allowed muscle reinnervation and functional recovery (Walker, 2015). That reversibility makes the model unusually useful for asking whether a pathway is disease-modifying. Because the transgene is driven by a neuronal promoter, the muscle phenotype described on this page most likely arises downstream of neuronal TDP-43 pathology rather than from transgene expression in muscle itself. For a broader comparison of TDP-43 and other ALS mouse models, see our guide to ALS models for drug discovery and our resource on TDP-43ΔNLS (rNLS8) mice for ALS drug development.

Biospective developed and characterized a reduced-doxycycline (“Low Dox”) induction paradigm that slows progression from weeks to months, widening the window in which a therapy can act. Details of the model, our in-house breeding colony and the full set of validated endpoints are on our rNLS8 TDP-43ΔNLS mouse model page. Previous Biospective data from this model, presented in our NMJ denervation Interactive Presentation, showed loss of SV2A–α-bungarotoxin co-localization, simplified presynaptic architecture and fewer axonal projections, together with reduced grip strength, reduced CMAP amplitude and progressive hindlimb muscle atrophy on longitudinal CT imaging. NMJ denervation has also been documented independently in a Thy1-driven hTDP-43 overexpression model, where it was confined to nerve terminals and tracked motor neuron loss (Alhindi, 2023).

In Low Dox rNLS8 gastrocnemius, CD45⁺ immune cells are distributed throughout the muscle at visibly higher density than in tTA controls, and the Interactive Presentation below lets you inspect this whole-section multiplex immunofluorescence data at full resolution, alongside automated NMJ innervation segmentations and the quantitative results.

The presentation is organized so that each microscopy view sits next to the measurement it supports: NMJ structure first, then the CD45⁺ infiltrate in Low Dox and control muscle, the correlation with denervation and CMAP, and finally the CD11b⁺ myeloid component. The sections that follow summarize those results and set them in the context of human disease and drug development.

Muscle inflammation and NMJ integrity were quantified on fixed-frozen gastrocnemius sections using a five-marker multiplex immunofluorescence panel and Biospective's automated PERMITS™ image analysis, with CD11b staining on consecutive sections and CMAP recorded in the same Low Dox animals.

Animals.

The study compared 14 Low Dox rNLS8 mice with 14 tTA single-transgenic control mice, which carry the NEFH-tTA driver but not the human TDP-43ΔNLS responder transgene (Biospective in-house data).

Multiplex immunofluorescence panel.

• Presynaptic nerve terminals: SV2A

• Motor axons and terminal branches: βIII-tubulin

• Postsynaptic acetylcholine receptors: α-bungarotoxin (α-BTX)

• Leukocytes: CD45 (pan-immune-cell marker)

• Nuclei: DAPI

• Myeloid cells (consecutive section): CD11b

Staining was performed using our automated multiplex immunofluorescence platform and whole sections were digitized on a slide scanner, following the workflow described on our NMJ staining and analysis service page.

Quantitative readouts.

PERMITS™ segmented each channel and derived CD45⁺ cell density (cells/mm² of muscle), the percentage of SV2A signal overlapping α-BTX, presynaptic branching complexity and total axonal projections. Each NMJ was then classified by its SV2A/α-BTX overlap as fully denervated (0–20%), partially innervated (20–80%) or fully innervated (80–100%), and class densities were expressed as endplates/mm² (Biospective in-house classification).

Electrophysiology and statistics.

Maximum compound muscle action potential (CMAP) amplitude was recorded from the gastrocnemius following sciatic nerve stimulation. Group differences were tested with unpaired t-tests and are shown as mean ± SEM; relationships between CD45⁺ density and NMJ or CMAP measures were assessed with Pearson correlation within the Low Dox group, one point per animal (Biospective in-house analysis).

Yes. CD45⁺ immune cell density in the gastrocnemius of Low Dox rNLS8 mice was roughly twice that of tTA controls (p < 0.001; Figure 1), indicating a sustained leukocyte infiltrate rather than baseline immune surveillance (Biospective in-house data).

The spatial pattern matters as much as the count. In tTA control muscle, the few CD45⁺ cells present sit mainly in the interstitial spaces between muscle fibres, as expected for resident immune surveillance of healthy skeletal muscle. In Low Dox muscle, CD45⁺ cells are spread throughout the section, which you can confirm by toggling the CD45 channel on the rNLS8 and tTA sections in the Interactive Presentation. Because CD45 labels all leukocytes, this measure captures total inflammatory burden; the identity of the cells is addressed by the CD11b data below (Biospective in-house data). Where the location of immune cells relative to individual endplates matters, our spatial biology analyses can map that neighbourhood directly.

Alongside the inflammatory infiltrate, Low Dox rNLS8 mice showed a pronounced shift from fully innervated to fully denervated NMJs, with reduced presynaptic complexity and fewer axonal projections than tTA controls (all p < 0.0001; Figures 2 and 3; Biospective in-house data).

Where a program needs tau pathology and frank neurodegeneration alongside amyloid, we run an amyloid-β and tau co-pathology model built on the same ARTE10 background, and the wider Alzheimer's disease model portfolio covers adjacent questions. Our Microglia, Astrocytes & Neurodegenerative Diseases Initiative collects the methodology behind the glial endpoints.

These NMJ changes reproduce the denervation phenotype described in our earlier NMJ denervation study in this model, and they provide the structural readout against which the inflammatory infiltrate can be compared animal by animal (Biospective in-house data).

Yes. Across the 14 Low Dox rNLS8 mice, CD45⁺ cell density correlated positively with the density of fully denervated NMJs (Pearson r = 0.81, p = 0.0004) and negatively with CMAP maximum amplitude (r = −0.71, p = 0.0047; Figure 4; Biospective in-house data).

These are within-group correlations: they compare Low Dox animals with one another, not Low Dox with controls. They therefore show that inflammatory burden tracks the severity of neuromuscular disease in individual animals, rather than simply marking genotype. Squaring the coefficients (r² ≈ 0.66 and 0.50), CD45⁺ density is statistically associated with roughly two-thirds of the between-animal variance in fully denervated NMJs and about half of the variance in CMAP amplitude (Biospective in-house analysis). Linking a histological immune measure to an electrophysiological readout of functionally innervated muscle fibres is what gives the infiltrate its functional meaning.

Correlation does not establish direction. Immune cells may be recruited by denervation to clear debris, or they may contribute to it, or both. Interventional work in other models supports at least a contributory role: local antibody treatment that reduced leukocyte infiltration also lessened NMJ denervation in hTDP-43 mice (Nógrádi, 2025), deleting C5aR1 in SOD1 mice reduced macrophage recruitment and denervation (Wang, 2017), and masitinib reduced muscle mast cell and macrophage infiltration and lessened denervation in SOD1^G93A rats (Trias, 2017). The Low Dox model, with its slower progression, is well suited to testing the same causal question for TDP-43-driven disease.

CD11b⁺ myeloid cells, most likely macrophages and related myeloid-lineage cells, appear to make up the bulk of the infiltrate, because CD11b staining on a section consecutive to the CD45-stained section closely mirrored the CD45⁺ distribution in Low Dox muscle and was sparse in tTA controls (Figure 5; Biospective in-house data).

A predominantly myeloid infiltrate is consistent with the macrophage-rich muscle pathology reported in ALS patient biopsies and TDP-43 mice (Nógrádi, 2025) and with macrophage recruitment to denervated muscle in SOD1 models (Van Dyke, 2016; Wang, 2017). Three limitations shape how far this observation can be taken:

• CD11b is also expressed by monocytes, neutrophils and some natural killer cells, so it does not by itself identify macrophages.

• CD11b is largely absent from conventional T and B lymphocytes, which could contribute to the CD45⁺ population and remain to be quantified.

• The CD11b comparison shown here is qualitative; per-animal CD11b density and its relationship to denervation have not yet been quantified.

Adding macrophage markers (F4/80 or CD68), polarization markers that separate pro-inflammatory from pro-regenerative states, Ly6G for neutrophils, CD3 for T cells and CCL2 itself would resolve these questions (the same morphology and phenotyping toolkit we apply to microglia in brain tissue), and matters for interpretation because pro-regenerative macrophages have been associated with slower progression (Margotta, 2023) while mast cells and neutrophils have been implicated in motor axon degeneration (Trias, 2018).

The muscle phenotype in Low Dox rNLS8 mice mirrors findings in people with ALS, whose muscle shows leukocyte, macrophage, mast cell and neutrophil infiltrates near motor endplates together with endplate denervation and progressive loss of electrophysiological measures of motor units (Bruneteau, 2015; Neuwirth, 2015; Trias, 2018; Nógrádi, 2025).

Correspondence between readouts in the Low Dox rNLS8 mouse and observations or measures in human ALS.

A TDP-43 context for muscle immunology.

Cytoplasmic TDP-43 aggregates are present in nearly all people with ALS, whereas SOD1 mutations account for only a small fraction of cases (Hardiman, 2017; Lépine, 2022). Much of the mechanistic literature on muscle inflammation, however, comes from SOD1 rodents (Van Dyke, 2016; Trias, 2017; Wang, 2017). The Low Dox rNLS8 model provides a TDP-43-driven setting in which the same immune readouts can be measured, complementing recent work in other TDP-43 lines (Nógrádi, 2025).

Clinically aligned functional endpoints.

CMAP amplitude is recorded in the clinic with the same principle used in mice, and derived measures such as the Motor Unit Number Index decline measurably over months in multicentre ALS cohorts, even in muscles that are not yet clinically weak (Neuwirth, 2015; Neuwirth, 2017). Pairing a tissue-level immune measure with CMAP therefore gives sponsors a preclinical readout that maps onto an electrophysiological measure already used in ALS clinical research.

Immunomodulation has reached ALS trials.

Peripheral immune targeting is not hypothetical. Masitinib, a tyrosine kinase inhibitor with activity on mast cells and macrophages, reduced muscle immune infiltration and denervation in SOD1 rodents (Trias, 2017; Harrison & Rafuse, 2020) and was reported to slow functional decline in a prespecified subgroup of a randomized ALS trial (Mora, 2020). Modifying peripheral macrophages has also extended survival in ALS mice (Chiot, 2020). A TDP-43 model with a measurable muscle infiltrate offers a way to test such mechanisms in the pathology that most patients share.

Muscle inflammation can be used as a mechanism-of-action and target-engagement endpoint for immunomodulatory ALS therapeutics, and as a disease-severity measure for any therapeutic class, when it is quantified in the same animals as NMJ innervation and CMAP.

Mechanisms this endpoint can read out.

• Chemokine-receptor blockade, for example of the CCL2–CCR2 axis (Nógrádi, 2025)

• Complement inhibition at C5a–C5aR1 (Wang, 2017)

• Tyrosine kinase inhibitors acting on mast cells and macrophages (Trias, 2017; Harrison & Rafuse, 2020)

• Macrophage-targeted delivery, such as dendrimer conjugates that localize to activated muscle macrophages (Tallon, 2022)

• Stromal and cytokine pathways in denervated muscle, such as fibro-adipogenic progenitor STAT3–IL-6 signalling (Madaro, 2018)

• Strategies that reprogramme peripheral macrophages rather than deplete them (Chiot, 2020)

An integrated, same-animal endpoint package.

Because CD45⁺ density, NMJ innervation classes, presynaptic complexity and CMAP come from the same mice, a drug-induced reduction in the infiltrate can be checked directly for downstream benefit to synapses and muscle function, and a drug that rescues NMJs without touching inflammation can be distinguished from one that acts through it. These readouts can be combined with the grip strength, CT muscle volume, plasma and CSF neurofilament light chain (NfL), MRI brain atrophy and brain histopathology endpoints already validated in this model, and can be complemented by cytokine profiling, as described on our TDP-43ΔNLS model page.

Inflammatory burden as a covariate.

Because CD45⁺ density accounted for a large share of between-animal variation in denervation and CMAP in this cohort, it can also be analysed as a covariate or stratification variable in studies of non-immune therapies, helping to separate genuine treatment effects from differences in disease severity among animals (Biospective in-house analysis).

Design considerations we discuss with sponsors.

• Dosing window and route: prevention designs starting at induction versus intervention designs starting once denervation is established, using any of our routes of administration.

• Positive control: reinstating doxycycline halts transgene expression and allows reinnervation and functional recovery, providing a benchmark for disease modification (Walker, 2015).

• Phenotype, not just number: broad myeloid suppression can be counterproductive, since microglia support recovery in this model and pro-regenerative macrophages are associated with slower progression (Spiller, 2018; Margotta, 2023), so polarization markers are recommended for immunomodulatory programmes.

• Sample size: in our earlier NMJ dataset from this model, fewer than 15 mice per group were estimated to detect a 25% change in SV2A–α-BTX overlap; equivalent estimates for the CD45⁺ endpoint can be derived from the present cohort (Biospective in-house data).

Sponsors receive whole-section image data, per-animal quantitative metrics, NMJ class maps, with interactive presentations of study sections available on request.

FAQs

Related Content