Macroautophagy (hereafter referred to as autophagy) is a catabolic pathway that contributes to cellular homeostasis by mediating routine protein and organelle turnover through lysosomal degradation [1, 2]. An early step in autophagy induction is lipidation of the microtubule-associated protein 1 light chain 3 (LC3, a mammalian orthologue of yeast ATG8); LC3-II (the lipidated form of LC3) associates with the newly generated autophagosome membrane and is thus commonly used as a marker of autophagosome formation [3]. LC3-II also binds multifunctional adapter protein SQSTM1/p62, which through this interaction targets ubiquitinated protein aggregates for lysosomal degradation [4]. Because SQSTM1 is selectively degraded through autophagy, its accumulation can be used as another measure of disrupted autophagic flux [3].

Autophagic disruption is a hallmark of several inherited skeletal myopathies including X-linked myopathy with excessive autophagy and infantile autophagic vacuolar myopathy [5, 6]; these disorders are characterized by defects in lysosomal degradation that lead to secondary accumulation of autophagic vacuoles [5]. Autophagic vacuolar myopathies (AVMs) can also develop as a toxic side effect of treatment with autophagy inhibiting drugs such as chloroquine (CQ; used for treatment of malaria), hydroxychloroquine (HCQ; used for treatment of rheumatologic disorders), and colchicine (used for treatment of gout) [7–9]. CQ and its derivative HCQ are thought to inhibit autophagy through lysosome alkalinization, which interferes with function of the pH-sensitive lysosomal enzymes [10]; in contrast, colchicine disrupts microtubule cytoskeleton, thus inhibiting autophagosome-lysosome fusion (which requires translocation of autophagosomes along microtubules) [9]. Interestingly, all AVMs clinically present with muscular weakness but only a minority of AVM muscle biopsies show muscle fiber degeneration (a classic feature of both inherited and toxic myopathies); thus, the causal mechanism linking autophagy impairment with muscle weakness remains poorly understood.

Skeletal muscle from patients with drug-induced/toxic AVMs exhibits accumulation of both LC3-II and SQSTM1, either in the fiber center or in association with rimmed vacuoles (sarcoplasmic vacuoles that on hematoxylin & eosin or trichrome stains show a basophilic rim) [11]. Intriguingly, rimmed vacuoles are also a hallmark feature of inclusion body myositis (IBM), a common and currently untreatable myopathy of the elderly that is characterized by both inflammatory and degenerative features as well as evidence of abnormal protein aggregation [12–14]. As with muscle from patients with toxic AVMs, skeletal muscle from IBM patients shows abundant LC3-II and SQSTM1 sarcoplasmic puncta that can be used to morphologically distinguish IBM from polymyositis (PM), a histologically similar inflammatory myopathy that generally occurs in younger patients and responds well to immunosuppressive therapy [15]. While IBM pathogenesis is not well understood, the presence of autophagy defects in this disorder suggests a possible mechanistic link with inherited and toxic AVMs.

The Keap1/Nrf2 stress response pathway controls redox homeostasis in eukaryotic cells. Nrf2 is a Cap’n’Collar basic leucine zipper transcription factor that regulates expression of a battery of antioxidant enzymes, chaperones, and other stress response proteins [16, 17]. At baseline, the Nrf2 pathway is negatively regulated by Keap1, a cytoplasmic regulatory protein that targets Nrf2 for proteasomal degradation through interaction with the Cullin-E3 ubiquitin ligase [17, 18]. Canonical activation of Nrf2 signaling–for example, during oxidative stress or in response to various phytochemicals–is mediated by a conformational change in Keap1 that leads to disruption of the Keap1/Nrf2 protein complex, allowing Nrf2 to evade degradation, translocate to the nucleus, and initiate transcription of genes with Antioxidant Response Element (ARE) in their promoter [17]. Recent work has demonstrated that Nrf2 signaling can also be activated through a non-canonical, SQSTM1-mediated pathway: because SQSTM1 competes with Nrf2 for Keap1 binding, SQSTM1 accumulation disrupts Nrf2 degradation, ultimately leading to increase in cytoplasmic and nuclear Nrf2 levels and elevated transcription of the Nrf2-regulated genes [19–22]. Canonical Nrf2 activation is generally cytoprotective [17], but persistent Nrf2 activation caused by autophagy disruption has been linked to liver toxicity and development of hepatocellular carcinoma [19, 21–23]. In the current study, we investigated whether autophagy inhibition in AVMs leads to abnormal sequestration of Keap1 protein and subsequent activation of the Keap1/Nrf2 stress response pathway.

Non-canonical activation of Nrf2 signaling is mediated through direct interaction of the Nrf2 regulator Keap1 and the adapter protein SQSTM1, which competes with Nrf2 for Keap1 binding [19–21]. Given that SQSTM1 accumulation is a hallmark of human AVMs [11, 15, 30, 31], we performed a retrospective case-control study to evaluate whether SQSTM1-Keap1 interaction leads to activation of the Nrf2 pathway in this category of disorders. Using two previously described cohorts of human AVM subjects and their controls (toxic AVM cohort [11] and IBM cohort [15]), we found that Keap1 is diffusely distributed in the sarcoplasm of normal muscle fibers (Fig. 1 and Additional file 2: Figure S1) but is sequestered into the SQSTM1-positive sarcoplasmic protein aggregates in the AVM muscle (Figs. 1, 3 and 5); in fact, Keap1-positive sarcoplasmic puncta/inclusions can be used as a diagnostic marker for both toxic AVMs (Fig. 2) and IBM (Fig. 4). In addition, we found that Keap1 sequestration was associated with an increase in mRNA and protein expression of the Nrf2-regulated genes (Fig. 6), indicating that the Nrf2 signaling pathway is activated in human AVM muscle; this phenomenon could be replicated in vitro by pharmacologic inhibition of autophagy in cultured murine myotubes (Fig. 7). Together, these findings indicate that autophagy disruption in skeletal muscle results in activation of the Keap1/Nrf2 stress response pathway both in vitro and in vivo.

Historically, pathologic diagnosis of toxic AVMs and IBM required electron microscopy to demonstrate disease-linked ultrastructural abnormalities (autophagosome accumulation in toxic AVMs [11] and presence of 15–18 nm tubulofilamentous inclusions in IBM [14, 32]). More recently, immunohistochemical stains for the autophagosome marker LC3-II and/or protein aggregate marker SQSTM1 have emerged as a more practical alternative: in the appropriate diagnostic setting, these immunohistochemical tests show high sensitivity and specificity for both toxic AVMs and IBM while lowering the cost and improving the turn-around-time compared to ultrastructural analysis [11, 15, 24]. In the current study, we found that immunohistochemical detection of Keap1-positive sarcoplasmic aggregates can be used as an alternate diagnostic test for both toxic AVMs (Figs. 1 and 2) and IBM (Figs. 3 and 4), with sensitivity and specificity that is comparable to LC3-II and SQSTM1 immunohistochemistries [11, 15]. Given that SQSTM1- and Keap1-immunopositive sarcoplasmic aggregates are co-localized in both AVMs (Fig. 5), the close correspondence in diagnostic utility of SQSTM1 and Keap1 immunohistochemistries is not surprising. On the other hand, we found little overlap between LC3-II- and Keap1-immunopositive structures in either AVM disorder (Fig. 5); however, both types of inclusions were generally found in the same muscle fibers, suggesting the explanation for a close correspondence in the diagnostic utility of LC3-II and Keap1 immunohistochemistries. Interestingly, the published research lacks consensus on whether LC3-II co-localizes with SQSTM1/Keap1 protein aggregates. Using cell-free biochemical assays and mammalian cell lines, one study found that Keap1 acts as a competitive antagonist of LC3-SQSTM1 interaction, with increase in Keap1 expression leading to attenuated SQSTM1 degradation via autophagy [33]. In apparent contrast, another study used stably transfected cell lines to show that Keap1 interacts with both LC3-II and SQSTM1 in a stress-inducible manner, facilitating SQSTM1-mediated autophagic clearance of ubiquitin aggregates [34]. Our finding that LC3 does not co-localize with Keap1-immunopositive structures in human AVM muscle (Fig. 5) is more consistent with the former observation; however, additional work will be required to fully elucidate the nature of Keap1/SQSTM1/LC3-II interaction.

What is the effect of chronically elevated Nrf2 signaling on skeletal muscle function and integrity? The canonical activation of the Nrf2 pathway (which is mediated through a transient oxidation-induced change in the Keap1 conformation [35]) is generally beneficial, leading to life span extension in both C. elegans and D. melanogaster [36, 37]. In agreement with the cytoprotective role of canonical Nrf2 signaling, pharmacologic activation of the Nrf2 pathway is myoprotective in a mouse model of dystrophinopathy [38], while Nrf2 deletion enhances muscle pathology in a mouse model of dysferlinopathy [39], possibly because intact Nrf2 signaling is required for muscle regeneration [40]. In contrast, sustained activation of the Nrf2 pathway generally has negative consequences on organ function and organismal life span; for example, elevated Nrf2 activity promotes tumorigenesis [21, 41] and is associated with a loss of glucose-stimulated insulin secretion in a β cell line [42]. The aberrant effect of sustained Nrf2 activation is particularly evident in the context of impaired autophagy, perhaps because SQSTM1 (an Nrf2-regulated gene) generates a positive feedback amplification loop in the Keap1/Nrf2 pathway that can only be interrupted via its autophagic degradation [33, 43, 44]; in agreement with this notion, chronic activation of Nrf2 signaling in autophagy-deficient livers leads to hepatotoxicity that can be abrogated by deletion of either SQSTM1 or Nrf2 [19, 23]. In the heart, sustained activation of the Nrf2 pathway via Keap1 sequestration into mutant αB-crystallin aggregates leads to reductive stress, a shift of the cellular redox potential to a more reduced state that is both necessary and sufficient for development of fatal cardiac disease [27, 28]. The consequences of chronic Nrf2 activation in skeletal muscle are less well understood; however, mutations in nuclear lamina proteins were recently reported to cause SQSTM1 accumulation, activation of the Nrf2 pathway, and reductive stress in both human and fly muscle [45]. Together, these findings raise the possibility that the aberrant activation of the Nrf2 pathway we observed in the AVM muscle is maladaptive and contributes to AVM pathogenesis; however, further work using cell culture and mouse models will be required to test this hypothesis.

In summary, we found that SQSTM1-mediated sequestration of Keap1 is associated with chronic activation of the Nrf2 stress response pathway in the AVM muscle, raising the possibility that toxic AVMs and IBM should be added to a growing list of disorders that exhibit dysregulation of cellular redox homeostasis.