Tweak and Fn14 are dysregulated in two SMA mouse models

We firstly investigated the expression of Tweak and Fn14 in skeletal muscle of the severe Taiwanese Smn−/−;SMN2 mouse model [22], using muscles with reported differential vulnerability to neuromuscular junction (NMJ) denervation (vulnerability: triceps brachii > gastrocnemius > TA > quadriceps femoris) [35]. Muscles were harvested from Smn−/−;SMN2 and WT mice at several time points during disease progression: birth (post-natal day (P) 0, pre-symptomatic (P2), early symptomatic (P5), late symptomatic (P7), and end stage (P10)). Muscle pathology in this SMA mouse model during disease progression has been well documented [36, 37].

We assessed the expression of parvalbumin, a high affinity Ca2+-binding protein, which is downregulated in denervated muscle [38, 39] and a marker of muscle atrophy in skeletal muscle of SMA patients and Smn−/−;SMN2 mice [40]. We observed a significant decreased expression of parvalbumin mRNA during disease progression (Fig. 1a) in SMA mice compared to WT animals, further confirming parvalbumin as a bona fide marker of muscle atrophy in SMA [40]. Furthermore, we noted that parvalbumin expression was downregulated at earlier time points in the two most vulnerable muscles (triceps and gastrocnemius) [35] of SMA mice compared to WT animals (Fig. 1a).

We next evaluated the expression of Tweak and Fn14 and observed significant decreased levels of Tweak mRNA in muscles of Smn−/−;SMN2 mice during disease progression, except in the quadriceps (Fig. 1b). Similarly, we found significantly lower levels of Fn14 mRNA in all muscles of Smn−/−;SMN2 mice during disease progression (Fig. 1c) compared to WT animals. Interestingly, the decreased expression of Fn14 in denervated and atrophied muscles of neonatal animals is different to previous reports in adults where denervation-induced atrophy stimulates its expression [15, 16].

As mentioned above, the TWEAK/Fn14 pathway has been reported to negatively influence the expression of metabolic effectors Klf15, Pgc-1α, Mef2d, Glut-4 and HKII [18]. Given that we have previously published a concordant increased expression of Klf15 in skeletal muscle of SMA mice during disease progression [41], we next evaluated if the additional metabolic targets proposed to be modulated by Tweak and Fn14 were similarly dysregulated in the predicted directions. We indeed observed that the mRNA expression of Pgc-1α, Mef2d, Glut-4 and HKII was significantly upregulated in muscles of Smn−/−;SMN2 mice at symptomatic time points (P5–P10) compared to WT animals (Fig. 1d–g), showing an expected opposite pattern to both Tweak and Fn14 (Fig. 1b–c) [18]. Notably, we also found that in most muscles, mRNA levels of Pgc-1α, Mef2d, Glut4 and HKII were significantly decreased in pre-symptomatic Smn−/−;SMN2 mice (P0–P5) compared to WT animals (Fig. 1d–g), independently of Tweak and Fn14 (Fig. 1b–c).

TWEAK and Fn14 have also been reported to impact the canonical and non-canonical NF-κB pathways in skeletal muscle [42, 43]. In pre-symptomatic (P2) TA muscle, we observed no significant difference in the expression of NF-κB1 (p50), a component of the canonical NF-κB pathway, between Smn−/−;SMN2 mice and WT animals (Fig. 1h), consistent with normal Tweak and Fn14 levels (Fig. 1b–c). Conversely, there was a significant decreased expression of NF-κB1 (p50) in TA muscle of symptomatic Smn−/−;SMN2 mice compared to WT animals at P7 (Fig. 1i), in line with reduced levels of Tweak and Fn14 (Fig. 1b). These findings are validated in P7 quadriceps, where NF-κB1 (p50) levels are also significantly decreased in Smn−/−;SMN2 mice compared to WT animals (Fig. 1j). We found no significant difference for the p105 NF-κB1 component. Of note, for all NF-κB1 p50/105 westerns, the p105 component was always more difficult to detect and sometimes even undetectable such as was the case for P7 TAs. We also investigated the expression of NF-κB-inducing kinase (NIK), involved in the non-canonical NF-κB activation pathway [44]. We observed that mRNA levels of NIK were significantly increased in TA muscle of P7 Smn−/−;SMN2 mice compared to WT animals (Fig. 1k), suggesting that dysregulated activity of Tweak and Fn14 in skeletal muscle of SMA mice may influence both the canonical and non-canonical NF-κB pathways, which play key regulatory roles in muscle health and metabolism [11, 12].

Finally, we evaluated the expression of Tweak and Fn14 in skeletal muscle of the less severe Smn2B/− mouse model of SMA [23]. TA muscles were harvested from Smn2B/− mice and age-matched WT animals at P0 (birth), P2 (early pre-symptomatic), P4 (late pre-symptomatic), P11 (early symptomatic), and P19 (end stage). Similar to the Smn−/−;SMN2 mice, muscle pathology in this SMA mouse model during disease progression has been well documented [36, 37]. We found a significant decreased expression of parvalbumin (Fig. 2a), Tweak (Fig. 2b) and Fn14 (Fig. 2c) in muscle from Smn2B/− mice during disease progression compared to WT animals, similar to that observed in the more severe Smn−/−;SMN2 SMA mouse model (Fig. 1a–c). We have previously reported the aberrant increased expression of Klf15 in the TA muscle of Smn2B/− mice during disease progression [41]. However, Pgc-1α expression was increased at P11 only (Fig. 2d), Mef2d at P2 only (Fig. 2e), Glut-4 at P11 only (Fig. 2f), while HKII was significantly decreased at P0 and P19 and significantly increased at P4 (Fig. 2g), suggesting that the proposed negative impact of Tweak and Fn14 activity on these metabolic effectors may be dependent on disease severity, age, and/or genetic strain. Tweak downregulation in triceps of P18 Smn2B/− mice was confirmed by western (Fig. 2h). Furthermore, contrary to what was observed in the Smn−/−;SMN2 mice, there was no significant difference in the NF-κB1 p50 component but a significant decreased expression of the NF-κB1 p105 component in skeletal muscle of Smn2B/− mice compared to WT animals (Fig. 2i). For the NF-κB2 pathway, we found no significant difference for either the p52 or the p100 components (Fig. 2j). Thus, our results point to distinct profiles of the NF-κB1 and 2 pathways in skeletal muscle of the two SMA mouse models, which could be due to differential expression and/or processing of the components and to non-Tweak/Fn14 pathways.

Fig. 2

Aberrant expression of Tweak and Fn14 in skeletal muscle of Smn2B/−SMA mice. a–g qPCR analysis of parvalbumin (a), Tweak (b), Fn14 (c), Pgc-1α (d), Mef2d (e), Glut-4 (f), and HKII (g) in TA muscles from P0 (birth), P2 (pre-symptomatic), P4 (pre-symptomatic), P11 (early symptomatic), and P19 (end stage) Smn2B/− and WT mice. Normalized relative expressions are compared to WT P0. Data are mean ± SEM, n = 3–4 animals per experimental group, two-way ANOVA, Sidak’s multiple comparison test between genotypes, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. h Quantification of Tweak protein levels normalized to total protein in the triceps of late-symptomatic (P18) Smn2B/− mice and age-matched WT animals. Images are representative immunoblots. Data are mean ± SEM, n = 6–7 animals per experimental group, unpaired t-test, p = 0.014. i Quantification of NF-κB1 p50 and p105 protein levels normalized to total protein in the triceps of late-symptomatic (P18) Smn2B/− mice and age-matched WT animals. Images are representative immunoblots. Data are mean ± SEM, n = 6–7 animals per experimental group, unpaired t-test, ns, not significant (p50), p = 0.0354 (p105). j Quantification of NF-κB2 p52 and p100 protein levels normalized to total protein in the triceps of late-symptomatic (P18) Smn2B/− mice and age-matched WT animals. Images are representative immunoblots. Data are mean ± SEM, n = 3–4 animals per experimental group, unpaired t test, ns, not significant (p52), p = 0.0532 (p100)

To determine if the dysregulated expression of Tweak, Fn14, and the previously reported metabolic effectors in SMA muscle is independent of disease status, we investigated the mRNA expression of Tweak, Fn14, Pgc-1α, Mef2d, Glut-4, HKII and Klf15 in triceps of P7 WT, Smn2B/2B and Smn+/− mice (Supplementary Fig. 1), a time point at which significant changes were already observed in the Smn−/−;SMN2 mice. Smn2B/2B and Smn+/− mice express ~70% and 50% of full-length functional Smn protein compared to WT animals, respectively, and do not display a canonical SMA phenotype [23, 45]. While we found some instances of differential expression (Glut-4: Smn2B/2B vs Smn+/−, HKII: Smn2B/2B vs Smn+/− and Klf15: WT vs Smn+/−), there is no clear correlation between non-pathological Smn levels (WT vs Smn2B/2B vs Smn+/−) and expression of molecular components associated with the Tweak/Fn14 pathway (Supplementary Fig. 1).

We have thus demonstrated that Tweak, Fn14, and associated metabolic effectors are dysregulated during progressive muscle atrophy in two SMA mouse models, and that this is most likely due to pathological levels of Smn depletion.

Denervation does not affect Tweak and Fn14 during the early stages of muscle development

As SMA muscle pathology is defined by both intrinsic defects and denervation-induced events, we set out to determine which of these may influence the dysregulation of Tweak and Fn14 in SMA muscle. We firstly addressed the denervation component by performing nerve crush experiments in which the sciatic nerves of P7 WT mice were crushed and the muscle harvested at P14 [46]. Of note, the sciatic nerve was crushed in only one hind limb, leaving the other control hindlimb intact. Quantification of myofiber area in TA muscles showed a significant decrease in myofiber size in the nerve crush muscle compared to the control hind limb (Fig. 3a–c).

Fig. 3

Tweak and Fn14 are not dysregulated in denervated (nerve crush) muscles of pre-weaned mice. A sciatic nerve crush was performed on postnatal day (P) 7 WT FVB/N mice, and both ipsilateral (nerve crush) and contralateral (control) TA muscles were harvested at P14. a Representative images of hematoxylin and eosin-stained cross sections of control and nerve crush TA muscles. Scale bars, 100 μm. b Myofiber area in control and nerve crush TA muscles. Data are mean ± SEM, n = 3–6 animals per experimental group, unpaired t-test, p = 0.0020. c Myofiber size distribution in control and nerve crush TA muscles. d qPCR analysis of parvalbumin, Tweak, Fn14, Pgc-1α, Mef2d, Glut-4, HKII, Klf15, and Smn in control and nerve crush TA muscles. Normalized relative expressions for each gene are compared to control muscle. Data are mean ± SEM, n = 4–6 animals per experimental group, two-way ANOVA, uncorrected Fisher’s LSD, ns, not significant

Expression analyses further revealed that there were no significant changes in mRNA levels of parvalbumin, Tweak, Fn14, PGC-1α, Mefd2, Glut-4 and HKII in the denervated muscle compared to the control TA muscle (Fig. 3d). Interestingly, while denervation in adult muscle has previously been reported to induce a dramatic surge in Fn14 expression [15, 16], this did not occur in the denervated muscles of our pre-weaned mice, suggesting an age and/or development regulatory element to this response. We also investigated the expression of Klf15 and Smn and similarly observed no significant differences between the nerve crush and control muscles (Fig. 3d). To ensure that our results were not influenced by the potential reinnervation of muscles following a nerve crush, we repeated the experiments by performing a nerve cut instead. We observed that this complete denervation of TAs in pre-weaned mice does not significantly impact the mRNA expression of Tweak, Fn14, PGC-1α, Mefd2, Glut-4, HKII and Klf15 compared to uninjured control hind limbs (Supplementary Fig. 2).

Overall, these results suggest that the dysregulation of parvalbumin, Tweak, Fn14, and the proposed metabolic effectors in SMA muscle during disease progression is most likely not denervation dependent.

Intrinsic muscle injury affects Tweak and Fn14 during the early stages of muscle development

We next investigated what impact impairing intrinsic muscle integrity would have on Tweak and Fn14. To do so, we used cardiotoxin to induce myofiber necrosis. Cardiotoxin was injected in P10 WT mice into the left TA, while the right TA was injected with equal volumes of 0.9% saline and used as a control. TAs were harvested after 6 days, a time point where muscles are still in an immature and regenerating mode [47]. Indeed, analysis of centrally located nuclei showed a significantly increased percentage of regenerating myofibers in cardiotoxin-treated muscles compared to saline-treated TAs (Fig. 4a–b).

Fig. 4

Tweak and Fn14 are dysregulated in cardiotoxin-induced muscle necrosis in pre-weaned mice. Cardiotoxin was injected in the left TA muscle of postnatal day (P) 10. The right TA muscle was injected with equal volumes of 0.9% saline. TA muscles were harvested 6 days later. a Representative images of hematoxylin and eosin-stained cross sections of saline- and cardiotoxin-injected TA muscles. Scale bars, 100 μm. b Percentage of muscle fibers with centrally located nuclei in saline- and cardiotoxin-injected TA muscles. Data are mean ± SEM, n = 3 animals per experimental group, unpaired t--test, p = 0.0020. c qPCR analysis of parvalbumin, Tweak, Fn14, Pgc-1α, Mef2d, Glut-4, HKII, Klf15, and Smn in saline- and cardiotoxin-injected TA muscles. Normalized relative expressions for each gene are compared to saline-treated muscle. Data are mean ± SEM, n = 3 animals per experimental group, two-way ANOVA, uncorrected Fisher’s LSD, ns, not significant, *p < 0.05, ***p < 0.001, ****p < 0.0001

We then proceeded with molecular analyses and observed that the atrophy marker parvalbumin was significantly downregulated in cardiotoxin-treated TA muscles compared to saline-treated TA muscles (Fig. 4c). Fn14 mRNA expression was significantly increased after cardiotoxin injury (Fig. 4c), in accordance with previous research showing that muscle damage conditions activate Fn14 [15]. Conversely, Pgc-1α, Glut-4, HKII and Klf15 mRNA levels were significantly downregulated (Fig. 4c), supporting their previously reported negative response to active Tweak and Fn14 [18]. Interestingly, Tweak mRNA expression remained unchanged (Fig. 4c), contrary to previous reports of upregulation following cardiotoxin injury in adult muscle [48], suggesting a differential response in early developmental stages of skeletal muscle. Notably, Smn expression was significantly increased in the regenerating muscles compared to saline-treated TA muscles (Fig. 4c), perhaps due to SMN’s reported role during muscle fiber regeneration [49].

Together, these results suggest that intrinsic muscle injury in pre-weaned mice induces a dysregulation of Tweak, Fn14 and previously reported proposed metabolic effectors. However, the changes were in the opposite direction than that observed in SMA muscles (Fig. 1b), perhaps due to the necrosis and regeneration events that occur following cardiotoxin injury [50], which are not typically found in muscles of SMA mice.

Genetic interactions between Smn, Tweak, and Fn14 in muscle

We next wanted to further understand the potential relationship between dysregulated expression of Tweak, Fn14, and Smn in skeletal muscle of SMA mice. To do so, we evaluated the impact of Tweak and Fn14 depletion in the early stages of muscle development by performing molecular analyses on P7 triceps from Fn14−/−, Tweak−/− and WT mice. In Tweak−/− mice, we observed a significant increased expression of Fn14 with a concomitant significantly decreased expression of Klf15 compared to WT animals (Fig. 5a). Notably, we found a significant decreased expression of Smn in Tweak−/− triceps compared to WT mice (Fig. 5a), suggesting a direct or indirect positive interaction between Tweak and Smn levels. For their part, Fn14−/−mice displayed a significant downregulation of parvalbumin and a significant upregulation of Pgc-1α (Fig. 5b). These analyses further support the previously reported negative influence of Fn14 on Pgc-1α and Klf15 expression as well as the absence of overt pathological muscle phenotypes in young Tweak−/− and Fn14−/− mice [15, 51].

Fig. 5

Smn, Tweak, and Fn14 depletion impact each other’s expression. a–b qPCR analysis of parvalbumin, Tweak, Fn14, Pgc-1α, Mef2d, Glut-4, HKII, Klf15, and Smn in triceps muscle from postnatal day (P) 7 Tweak−/− (a) and Fn14−/− (b) mice. Normalized relative expressions for each gene are compared to WT. Data are mean ± SEM, n = 4 animals per experimental group, two-way ANOVA, uncorrected Fisher’s LSD, ns, not significant, *p < 0.05, ***p < 0.001, ****p < 0.0001. c–j qPCR analysis of Smn (c), Tweak (d), Fn14 (e), Pgc-1α (f), Mef2d (g), Glut-4 (h), HKII (i), and Klf15 (j) in siRNA-mediated Tweak-, Fn14-, and Smn-depleted and control proliferating (day 0) and differentiated (day 7) C2C12 cells. Normalized relative expressions for day 1 experimental groups are compared to day 1 untreated group, and normalized relative expressions for day 7 experimental groups are compared to day 7 untreated group. Data are mean ± SEM, n = 3 per experimental group, two-way ANOVA, Dunnett’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. k Proposed model of the relationship between Smn and the Tweak/Fn14 signaling pathway. Red lines represent inhibition, and blue lines represent activation

To further dissect the relationship between Smn, Tweak, and Fn14 during myogenic differentiation, we performed siRNA-mediated knockdown of Smn, Tweak and Fn14 in C2C12 myoblasts and evaluated the effect on the expression of Tweak, Fn14, and the previously reported proposed metabolic effectors in undifferentiated (day 0) and differentiated (day 7) cells. Reduced levels of Smn, Tweak and Fn14 were significantly maintained in both proliferating and differentiated cells following transfection with siSmn, siTweak, and siFn14, respectively (Fig. 5c–e). We observed an interaction between Smn, Tweak and Fn14 specifically in differentiated C2C12s, whereby Smn expression was significantly upregulated in Fn14-depleted D7 cells (Fig. 5c), Tweak expression was significantly reduced in Smn-depleted D7 cells (Fig. 5d) and Fn14 levels were significantly decreased in Tweak- and Smn-depleted D7 cells (Fig. 5e). Similarly, the effects of siRNA-mediated knockdown of Smn, Tweak and Fn14 on the metabolic effectors were only apparent in differentiated C2C12s (Fig. 5f–j). Indeed, both knockdown of Tweak and Fn14 resulted in a significant upregulation of Pgc-1α (Fig. 5f) and Mef2d (Fig. 5g). While Glut-4 expression was neither affected by depletion of Smn, Tweak, or Fn14 (Fig. 5h), HKII mRNA levels were significantly decreased following knockdown of all three (Fig. 5i). Finally, Klf15 expression was significantly increased in siRNA-mediated knockdown of Fn14 only (Fig. 5j). The upregulation of Pgc-1α, Mef2d and Klf15 in Tweak- and/or Fn14-depleted differentiated C2C12 cells is in accordance with the previously reported downregulation of these genes when Tweak and Fn14 are active, while the unchanged Glut-4 and downregulated HKII levels were not [52].

Thus, using both in vivo and in vitro models, we have provided evidence for a potential interaction between Smn, Tweak and Fn14 and subsequent impact on the previously proposed downstream metabolic effectors (Fig. 5k). Our results suggest that the aberrant expression of Tweak and Fn14 in SMA muscle during disease progression may be due to a dynamic interplay between muscle-specific conditions and the molecular impact, individual and combined, of reduced expression of Smn, Tweak and Fn14 in the early developmental stages of skeletal muscle.

Overlap of dysregulated myopathy and myogenesis genes and glucose metabolism genes in SMA, Fn14−/− and Tweak−/− mice

To further decipher the potential contribution(s) of Smn, Tweak, and Fn14 depletion to SMA muscle pathology, we used commercially available mouse myopathy and myogenesis qPCR arrays (SABiosciences), which measure expression levels of a subset of 84 genes known to display and/or regulate myopathy and myogenesis. We used triceps (vulnerable) and quadriceps (resistant) from P7 Smn−/−;SMN2, Tweak−/−and Fn14−/− mice. WT FVB/N mice were compared to SMA animals and WT C57BL/6 mice were compared to Tweak−/− and Fn14−/− mice to account for differences due to genetic strains. Unsurprisingly, we observed a larger number of significantly dysregulated myopathy and myogenesis genes in triceps of Smn−/−;SMN2 mice than in the more resistant quadriceps, some of which overlapped with the subset of genes aberrantly expressed in Fn14−/− mice and Tweak−/− mice (Fig. 6a, Table 1, Supplementary file 1). We also used the publicly available database STRING [34] to perform network and enrichment analysis of the shared differentially expressed genes in both triceps and quadriceps (Table 1), which revealed that there were no known protein-protein interactions between any of the dysregulated genes and Smn, Fn14, or Tweak (Fig. 6b). Interestingly, the central connectors Myod1 and Myf6 were upregulated in Tweak−/− and Fn14−/− mice and Pax7 was downregulated in the triceps of all three experimental groups (Table 1). Myod1 and Myf6 are key myogenic regulatory factors (MRFs) and are normally upregulated after skeletal muscle injury [53]. Pax7 is a canonical marker for satellite cells, the resident skeletal muscle stem cells [53], and reduced activity of Pax7 leads to cell-cycle arrest of satellite cells and dysregulation of MRFs in skeletal muscle [54]. Furthermore, Titin (Ttn) was downregulated in the quadriceps muscles of all three mouse models and plays major roles in muscle contraction and force production, highlighted by titin mutations leading to a range of skeletal muscle diseases and phenotypes [55].

Fig. 6

Overlap between dysregulated genes involved in myopathy, myogenesis, and glucose metabolism in skeletal muscle of Smn−/−;SMN2, Fn14−/−, and Tweak−/− mice. a Venn diagram showing overlap of genes involved in myopathy and myogenesis that are significantly dysregulated in the same direction (either up or downregulated, p < 0.05) in triceps and quadriceps muscle from postnatal day (P) 7 compared to Smn−/−;SMN2, Fn14−/−, and Tweak−/− mice to age- and genetic strain-matched wild-type (WT) mice. b Network and enrichment analysis of the overlap of significantly dysregulated myopathy and myogenesis genes in triceps and/or quadriceps of P7 Smn−/−;SMN2, Fn14−/−, and Tweak−/− mice using STRING software. Smn (Smn1), TWEAK (Tnfsf12), and Fn14 (Tnfrsf12a) are included in the analysis. Colored nodes represent query proteins and first shell of interactors. Filled nodes indicate that some 3D structure is known or predicted. Connection colored lines between nodes represent either known interactions (turquoise: from curated databases, magenta: experimentally determined), predicted interactions (green: gene neighborhood, red: gene fusions, dark blue: gene co-occurrence) or other interactions (yellow: textmining, black: co-expression, light blue: protein homology). c Venn diagram showing overlap of genes involved in glucose metabolism that is significantly dysregulated in the same direction (either up or downregulated, p < 0.05) in triceps and quadriceps muscle from P7 compared to Smn