Mutual reinforcement of lymphotoxin-driven myositis and impaired autophagy in murine muscle

Abstract

Inclusion body myositis (IBM) is a progressive muscle disorder characterized by inflammation and degeneration with altered proteostasis. To better understand the interrelationship between these two features, we aimed to establish a novel preclinical mouse model. First, we used quantitative PCR, in situ hybridization and immunohistochemistry to determine the expression of pro-inflammatory chemokines and cytokines including lymphotoxin (LT)-signalling pathway components in human skeletal muscle tissue diagnosed with myositis. Based on these results, we generated a mouse model that we analysed at the histological, ultrastructural, transcriptional, biochemical and behavioural level. Lastly, we subjected this model to anti-inflammatory treatments. After confirming and extending previous data on activation of LT-signalling in human myositis, we generated distinct transgenic mouse lines co-expressing LTα and -β in skeletal muscle fibres. Transgenic mice displayed chronic myositis accompanied by dysregulated proteostasis, including an altered autophagolysosomal pathway that initially showed signs of activation and later exhaustion and decreased flux. To enhance the latter, we genetically impaired autophagy in skeletal muscle cells. Autophagy impairment alone induced a pro-inflammatory transcriptional state, but no obvious cellular inflammation. However, the combination of LT-driven myositis with autophagy impairment induced the full spectrum of characteristic molecular and pathological features of IBM in skeletal muscle, including protein inclusions with typical ultrastructural morphology and mild mitochondrial pathology. Our attempts to treat the pathology by subjecting these mice to corticosteroids or anti-Thy1.2 antibodies mirrored recent treatment failures in humans, i.e. none of these treatments resulted in significant clinical improvement of motor performance or the transcriptional profile of muscle pathology. In summary, these data provide evidence that inflammation and autophagy disruption play a synergistic role in the development of IBM-like muscular pathology. Furthermore, once established, IBM-like pathology in these mice, as in human IBM patients, cannot be reverted or prevented from progression by conventional means of immunosuppression. We expect that this novel mouse model will help to identify future treatment modalities for IBM.

Keywords: NF-κB signalling; autophagy; inclusion body myositis; lymphotoxin; lymphotoxin signalling; myositis.

Figure 1

Histological features and chemokine as well as cytokine expression in human inflammatory myopathies. (A) Trichrome staining showing endomysial inflammatory infiltrates (white arrowheads) as well as muscle fibre atrophy and endomysial fibrosis (black arrowheads), which is most severe in this case of inclusion body myositis (IBM). Yellow arrowheads: rimmed vacuoles. Immunohistochemistry (brown signal): CD3⁺ T cells occasionally invading intact muscle fibres in polymyositis (PM) and IBM (blue arrowheads); CD20⁺ B cells are most frequent in dermatomyositis (DM), but are occasionally also seen in PM and IBM; CD68⁺ macrophages are frequent in all of these idiopathic inflammatory myopathies (IIMs); phospho-TDP43-, p62- and ubiquitin-positive inclusions in IBM (black arrowheads). (B) Heat map of chemo- and cytokine expression in PM, DM and IBM compared with controls. P-values were determined using the two-tailed Mann–Whitney test and are shown if the expression was significantly different from control after Bonferroni correction (P < 0.17). Upregulation of lymphotoxin (LT)α and LT-signalling target genes including CCL19 and CXCL10. (C) Schematic drawing of LTα/LTβ heterotrimer and LTα homotrimer activation of both tumor necrosis factor (TNF) and LTβ receptor, leading to the translocation of nuclear factor-kappa B (NF-κB) into the nucleus. (D) In situ hybridization of LTα and LTβ receptor (LTβR). Controls (i and iii) show no LTα signal and little LTβR signal in muscle fibres (arrows in enlarged images shown below in iii). In IBM, little LTα is detected in inflammatory infiltrates (black arrowheads) as well as in myonuclei (black arrows in enlarged images shown below in ii). In IBM, LTβR is also detected in both, inflammatory infiltrates (black arrowheads) as well as in myonuclei (black arrows, in enlarged images shown below in iv as well as in another area in v). (E) Immunohistochemistry (brown signal) of LTβ in muscle cells of IBM cases (ii–iv) compared with controls (i). Bottom row shows enlarged areas of i–iv. LTβ is detected in inflammatory infiltrates (black arrowheads) in IBM. In muscle fibres, the overall LTβ signal is slightly increased in IBM with a particular enrichment in rimmed vacuoles (black arrows) and where lymphocytes infiltrate intact muscle fibres (white arrowheads). (F) Normalized read counts of LTα, LTβ as well as LT target genes CXCL10, CXCL13, CCL5 and CCL19 compared with the housekeeping gene HPRT1 in controls and IBM samples derived from two previously published RNA-Sequencing datasets, showing a strong and significant increase of LT-related gene expression.

Figure 2

Generation, histology and chemokine and cytokine expression of HSA-LTα/β transgenic mice. (A) Schematic drawing of the transgenic constructs of the coding sequences (CDS) of lymphotoxin (LT)α and LTβ both being cloned downstream of the human skeletal muscle actin (HSA) promoter (−2000 to +239) followed by an SV40 poly A site. (B) Four founder mice carrying both transgenes, HSA-LTα and HSA-LTβ, and transmitting them through the germline were obtained (Lines #8, #19, #22, #44). (C and D) LTα and LTβ transgene expression was detected in quadriceps muscle fibres of HSA-LTα/β #19 mice by RNA in situ hybridization of paraffin sections (brown signal) in the muscle fibres at 3, 6 and 10 months of age and also in inflammatory infiltrates in skeletal muscle, shown at 10 months of age (C). LTα ELISA with skeletal muscle tissue homogenate from indicated time points and genotypes. Data are shown as nanogram LTα3/LTα1β2/LTα2β1 protein per miligram total protein. Statistical significance was tested using the two-tailed Student's t-test. (E) Histologically, HSA-LTα/β transgenic mice show endomysial inflammatory infiltrates and occasional muscle fibre necrosis. Especially at the age of 9 months, there is fibrosis and partial replacement by fatty tissue within the endomysium. Fibre size variation is also increased in transgenic mice [haematoxylin and eosin (H&E) and trichrome]. In addition to the normal MHCI localization on endomysial capillaries seen in wild-type mice, transgenic mice display focal sarcolemmal major histocompatibility complex (MHC)I upregulation (red signal). The inflammatory infiltrates are mostly composed of B220⁺ B cells and CD68⁺ macrophages and—to a lesser extent—of CD4⁺ and CD8⁺ T cells (brown signals). (F) Heat map of chemo- and cytokine expression in HSA-LTα/β transgenic mice determined by quantitative PCR of quadriceps femoris muscle tissue at 3, 6 and 10 months of age. P-values were determined using two-tailed Mann–Whitney tests. In the case of significant differences between wild-type and transgenic mice, the P-values are displayed between the group of wild-type and transgenic line of the respective age group. (G) Quantitative PCR determining LTα and LTβ expression in different organs of HSA-LTα/β transgenic mice. P-values were determined using the ANOVA test with Šídák's multiple comparison test, significant differences compared with wild-type are displayed. (H) H&E-stained sections of three different muscles from 9-month-old HSA-LTα/β mice.

Figure 3

Muscle fibre size distribution, muscle hypotrophy and motor impairment in HSA-LTα/β transgenic mice. (A) Muscle fibre diameters were examined morphometrically using cross sections of the quadriceps femoris muscle [at 3 months, 596 fibres of n = 4 wild-type (wt) and 840 fibres of n = 4 human skeletal muscle actin (HSA)-lymphotoxin (LT)α/β transgenic mice (LT); at 6 months: 981 fibres of n = 4 wt mice and 1075 fibres of n = 5 LT mice; at 10 months: 1266 fibres of n = 5 wt mice and 1732 fibres of n = 5 LT mice]. For statistical analysis, we used the chi-squared test and grouped fibres into atrophic (<50 µm), normal (50–109 µm) and hypertrophic (>109 µm). Fibre size distribution was broader in HSA-LTα/β transgenic mice at 3 and 6 months, with both, more atrophic and hypertrophic fibres; there were more atrophic fibres at 10 months of age. (B) Body weight was determined every 4 weeks and is displayed separately for female and male mice. (C) Weight of quadriceps femoris, gastrocnemius and triceps brachii muscles. Two-sided unpaired Student's t-test was used for statistical analysis. Differences were significant for all analysed time points for all muscles; ****P < 0.0001. Motor performance of wild-type and HSA-LTα/β transgenic mice was determined by the grip strength test (D) and by the hanging wire test (E). Groups were compared using the two-sided unpaired Student's t-test. P-values are shown in case of a significant difference.

Figure 4

Organelle stress and signs of altered autophagolysosomal pathways in chronic myositis. (A–C) Gene expression in human skeletal muscle actin (HSA)-lymphotoxin (LT)α/β transgenic mice determined by quantitative PCR of quadriceps femoris muscle tissue at 3, 6 and 10 months of age. Expression of genes related to endoplasmic reticulum (ER) stress, autophagolysosomal pathway, heat shock response (A) as well as oxidative stress-related genes (B) are displayed in heat maps. P-values were determined using two-tailed Mann–Whitney tests. In the case of significant differences between wild-type and transgenic mice, the P-values are displayed between the group of wild-type and transgenic line of the respective age group. To visualize the gene expression changes of the marked upregulated autophagy-related gene Atg5 at 6 months of age, we also display it together with another autophagy-related gene Becn1 and lysosomal protease Ctsb in transgenic compared with wild-type mice in C. Both autophagy-related genes are upregulated at 3 months (Atg5) and 6 months (Atg5, Becn1, see A), but downregulated (Atg5) along with lysosomal Ctsb at 10 months. (D) Downregulation of ATG5, BECN1 and HSPB1 as well as upregulation of CTSB in human inclusion body myositis (IBM) muscle tissue compared with controls determined by analysing published RNA-Seq data. (E and F) We determined amounts of LC3-I and LC3-II proteins in the muscle of 6- and 10-month-old HSA-LTα/β transgenic compared with wild-type mice by western blot (E) and observed a significant reduction of total LC3 and LC3-II/ α-tubulin ratio in transgenic mice at 6 months and a significant increase in total LC3 and a decrease in the LC3II/ LC3-I ratio at 10 months. Quantification of western blot bands (F); values for total LC3 and LC3II are relative to wild-type values; n = 5 biological replicates at 10 months; one to two technical replicates from three to four biological replicates at 6 months. P-values were determined using two-tailed Student's t-tests. Dotted line shows where membranes were cut; solid line shows cropping for representation in this figure; full size membranes are shown in the Supplementary Figs 7 and 8. (G) In line with disturbed proteostasis, we occasionally observed focal accumulation of ubiquitin in muscle fibres of HSA-LTα/β transgenic mice at 6 and 10 months of age (immunohistochemistry). Distribution of LAMP2⁺ lysosomes was even in wild-type mice, but uneven with less, but larger lysosomes in HSA-LTα/β transgenic mice (immunohistochemistry). Scale bar = 50 µm. (H) Ultrastructural examination at 9 months of age did not show any IBM-like inclusions in HSA-LTα/β transgenic mice, but mitochondria showed slight swelling. The chi-squared test was used for statistical testing.

Figure 5

Behavioural and histological consequences of autophagy depletion in lymphotoxin (LT)-induced chronic myositis. (A) Breeding scheme to obtain human skeletal muscle actin (HSA)-LTα/β⁺ Ckmm-Cre⁺ Atg5^fl/fl (HSA-LT;CreAtg5) mice. (B and C) Muscle volume was determined by MRI at 3 and 6 months of age of male and female HSA-LT;CreAtg5 compared with Ckmm-Cre⁺ Atg5^fl/fl (CreAtg5) mice. Representative fast low angle shot MRI (FLASH) images and 3D reconstructions of calf muscles are shown along with volumetric quantification. Muscle atrophy was observed in HSA-LT;CreAtg5 compared with CreAtg5 mice. Two-sided unpaired Student's t-test was used to determine P-values (****P < 0.0001). (D–F) Histological and immunohistochemical analysis for inflammatory markers revealed chronic myositis in HSA-LT;CreAtg5, but not in CreAtg5 mice, characterized by endomysial infiltrates composed of CD4⁺ and CD8⁺ T cells, B220⁺ B cells and CD68⁺ macrophages (brown signals) as previously observed in HSA-LTα/β transgenic mice. Sarcolemmal upregulation of major histocompatibility complex I (MHCI) was detected in HSA-LT;CreAtg5, but MHCI was either not (one of four) or expressed weakly/only on single fibres (three of four) in CreAtg5 mice (red signal). Quantification of most affected fields of view is shown in F. P-values were determined using the ANOVA test with Šídák's multiple comparison test. (G and H) Motor performance of CreAtg5 and HSA-LT;CreAtg5 mice was determined using the grip strength test (G) and the hanging wire test (H). Groups were compared using the two-sided unpaired Student's t-test. P-values are shown.

Figure 6

Consequences of autophagy depletion in lymphotoxin (LT)-induced chronic myositis on proteostasis, ultrastructure and mitochondria. (A) Immunohistochemistry for ubiquitin, LAMP2, p62 and phospho-TDP43 (brown signals). Human skeletal muscle actin (HSA)-LT;CreAtg5 mice show numerous ubiquitin- and p62-positive inclusions in all muscle fibres. Ubiquitin-positive inclusions were not observed in wild-type mice. One mouse showed little focal p62-positivity (n = 6 of 6 HSA-LT;CreAtg5, but only focally in n = 1 of 6 wild-type; P = 0.0152, Fisher's exact test). The distribution of LAMP2⁺ lysosomes in HSA-LT;CreAtg5 compared with wild-type with less, but often larger lysosomes in most fibres and increased lysosomal density in other fibres. Occasionally, phospho-TDP43-positive inclusions are observed in individual fibres of HSA-LT;CreAtg5 mice (n = 6 of 6 HSA-LT;CreAtg5, Fisher's exact test when compared with wild-type: P = 0.0152). (B) Electron microscopy at 6 months of age revealed disintegration of sarcomeric structure/myofibrils (i), accumulation of granular and fibrillar material (between black arrows in ii and iii, and black arrowheads in vii), probably corresponding to myofibrillar fragments, often containing inclusions resembling tubulofilamentous inclusions characteristic for human inclusion body myositis (IBM; between black arrows in iv and vii), diameter measured in iv: 106 tubulofilaments showed a mean diameter of 16.9 ± 2.3 nm in n = 3 HSA-LT;CreAtg5 mice. Subsarcolemmal accumulation of mitochondria (v) next to accumulation of granular material (black arrowheads in v) with foci of mitochondria in different stages of abnormal mitophagy (vi) with some mitochondria showing almost normal structure of cristae (black arrows in vi) and those with cristae remnants in double membranes characteristic for autophagosomes (abnormal mitophagy, white arrowheads in vi). Focal deposits of abnormal myelin-like phospholipid are frequently seen (white arrowheads in vii). Pyknotic, abnormally invaginated myonucleus (viii). Scale bars = 500 nm. (C) Long-read nanopore sequencing of PCR-amplified mitochondrial DNA, showing no obvious deletions in any of the genotypes tested. Coverage is displayed in grey. Single nucleotide substitutions are shown in green (adenine), orange (guanine), blue (cytosine) and red (thymine). (D) Combined cytochrome c oxidase (COX, brown signal) and succinate dehydrogenase (SDH, blue signal) enzyme histochemistry. Even distribution of COX-positive mitochondria in wild-type (i) with higher density in type 1 fibres (brown signal). Focal irregularities in the distribution of COX-positive mitochondria in HSA-LTα/β (ii), more pronounced in HSA-LT;CreAtg5 mice that also display single fibres with mosaic COX-positive (brown) and COX-negative mitochondria (blue, due to preserved SDH activity, black arrowheads, iii and iv, with enlarged images of these fibres to the right of iii and iv). Several fibres also focally lack both, COX and SDH activity (black arrows).

Figure 7

Transcriptional consequences of depleting autophagy in lymphotoxin (LT)-induced chronic myositis mice. (A–D) Heat maps of gene expression in wild-type, CreAtg5, human skeletal muscle actin(HSA)-LTα/β and HSA-LT;CreAtg5 mice determined by quantitative PCR of quadriceps femoris muscle tissue at 6 months of age. P-values were determined using Mann–Whitney U-test and are displayed above the group of mice in case of a significant difference after Bonferroni correction (P-values <0.016). Cytokine and chemokine expression is shown in A, endoplasmic reticulum (ER) stress and autophagolysosomal gene expression is displayed in B. (C) Genes related to muscle de- and regeneration and Alzheimer disease/neurodegeneration including App and Cryab. (D) Genes related to oxidative stress (gene functions overlap).

Figure 8

Inclusion body myositis (IBM)-like inflammatory myopathy in mice is resistant to anti-inflammatory treatments. (A) Timeline of treatment. Density of CD3⁺ T cells in the liver of untreated mice and after treatment; for anti-Thy1.2 (B) and prednisolone (C) treatment. P-values were determined using two-tailed Students t-tests. Grip strength of mice at Day 150. The results show genotype-dependent differences, but no treatment effect after prednisolone (D) and anti-Thy1.2 (E) treatment. (F) The heat map shows expression examined by quantitative PCR (qPCR) relative to the respective untreated controls (treated compared with untreated wild-type, human skeletal muscle actin (HSA)-lymphotoxin (LT)α/β and HSA-LT;CreAtg5, respectively)—here in the case of prednisolone treatment. Except for a significant downregulation of Cxcl10, Ncf2 and Gpx2 following prednisolone treatment of HSA-LTα/β mice compared with untreated HSA-LTα/β and non-significant trends for some other genes, there was a counterintuitive upregulation of Ccl5 in treated compared with untreated HSA-LT;CreAtg5 mice, but otherwise stable gene expression. n.d. = not determined. Mann–Whitney U-test was performed as a statistical test since some values were not normally distributed. P-values are only displayed in the case of statistical significance between untreated and treated mice. Behavioural phenotypes and gene expression patterns following these treatments were stable (Supplementary material).