Appropriate nutrition is often an issue in the care of critically ill patients especially during the first days in ICU that often lead to missed caloric goals. Adequate feeding and outcome are closely connected with each other. Both underfeeding and overfeeding is often observed in critically ill patients, which can be harmful to critically ill patients suggesting that caloric goals need to be correctly determined and reached (83). Underfeeding and overfeed are related to uncertainties about the energy expenditure (resting metabolic rate, additional requirements due to the illness that is also dynamic with the disease course), the underlying disease itself, unknown body composition, difficulties to administer food, and the route of enteral versus parenteral nutrition. Patients biological age and sex differences may also have an effect on energy expenditure. Uncertainties about the right composition of the administered calories complicate the situation. As we describe later in this review hyperglycemia and insulin resistance are risk factors for ICUAW. It is therefore reasonable to assume that treating these risk factors may improve clinical outcome. This hypothesis led to the clinical concept of permissive underfeeding, which has been shown to reduce blood glucose and insulin levels indicative for an improved insulin sensitivity (266). However, Arabi et al., investigated if permissive underfeeding has an effect on clinical outcome in critically ill patients. They showed that administration of a hypocaloric diet with 70% of calculated caloric requirements (permissive underfeeding) did not affect mortality, ICU stay and other secondary outcomes when compared to standard nutrition (266). The study design required that both groups received the same amount of proteins implicating that not the number of calories administered but that the amount of proteins could have an effect on outcome of critically ill patients. The problematic situation on feeding of critically ill patients has been reviewed elsewhere (267–270). Although experimental models suggest an association between the feeding status and UPR, its involvement in this complex situation is far from being understood.

Importantly, fasting is a strong stimulus for skeletal muscle atrophy (29), which shares common pathways with other causes of muscle wasting (84). The UPS and the ALP are the best described pathways involved in starvation-induced muscle atrophy. Recent work indicates that also the UPR is involved in this pathology (85). Specifically, 24 hours of starvation caused an elevation of mRNA and protein expression of UPR markers such as Atf4, Chop, Grp94, and Gadd34 and an increased splicing of XBP1 in the gastrocnemius muscle of mice. Also, 48 hours of starvation in cows leads to an increased IRE1α phosphorylation and an increased XBP1 splicing in hepatocytes, indicating a connection between ER stress and starvation (86).

Additionally, TNF Receptor Associated Factor 6 (TRAF6) an E3 ubiquitin ligase that belongs to a family of conserved intracellular adaptor proteins has been shown to be involved in UPR. In response to cytokines and microbial products, TRAF6 regulates multiple signaling pathways, such as NF-κB, MAPK, and PI3K/Akt. Skeletal muscle specific deletion of Traf6 attenuates skeletal muscle atrophy caused by starvation, denervation, and cancer cachexia (15, 85). Of note, TRAF6 is essential for the inducible expression of ER stress response-related genes in response to fasting. Specifically, the starvation-induced increase in mRNA and protein expression of the UPR markers ATF4, CHOP, GRP94, and GADD34 in the gastrocnemius muscle was attenuated in muscle specific Traf6 knockout mice (85). However, the mechanism by which TRAF6 and the UPR are connected in skeletal muscle during starvation remains to be uncovered.

The activating transcription factor 4 (ATF4) that is downstream of the PERK arm, was also shown to mediate starvation-induced muscle atrophy. Of note, 24 hours of fasting increased the Atf4 mRNA expression in tibialis anterior of male mice (87). Inhibition of ATF4 expression by electroporation of a small interfering RNA targeting ATF4 or overexpression of a phosphorylation-resistant form of eIF2α (eIF2α-S51A) attenuated myofiber atrophy in tibialis anterior muscle during fasting (87). Overexpression of ATF4 caused a reduction in myofiber size even in the absence of fasting. Importantly, electroporation of a transcriptionally inactive ATF4 construct (ATF4ΔbZIP) did not reduce myofiber size, suggesting that ATF4-dependent gene expression is required for ATF4-mediated atrophic effects (87). These data show that ATF4 mediates starvation-induced muscle atrophy but it can also cause muscle atrophy in the absence of starvation.

Starvation has also been shown to activate the nutrient-sensitive transcription factors transcription factor EB (TFEB) and transcription factor E3 (TFE3). Both TFEB and TFE3 belong to the microphthalmia/transcription factor E (MiT/TFE) family of basic bZIP transcription factors that recognize a unique E-box motif within the proximal promoters of lysosomal and autophagy genes (88, 89), and regulate cellular catabolism and nutrient-dependent lysosomal response (90, 91). Both TFEB (92) and TFE3 (89) are predominantly localized in the cytoplasm but also to the nucleus (93, 94). In the presence of nutrients, mTORC1 (mammalian target of rapamycin complex 1) phosphorylates TFEB and TFE3, which facilitate their binding to 14-3-3 chaperone proteins and mediates their retention in the cytoplasm. Conversely, during starvation reduced mTORC1 activity increases the shuttling of TFEB (89) and TFE3 (89) into the nucleus increasing the expression of TFEB- and TFE3-dependent genes. Importantly, ER stress also induces the nuclear translocation of TFEB and TFE3 and this effect was shown to be dependent on PERK and calcineurin but not mTORC1 (95). We recently described that TFEB (93) and TFE3 (94) are also localized to the nucleus where TFEB is bound to conserved E-box elements in the MuRF1/Trim63 promoter to regulate its expression (93) implicating a role in muscle atrophy. Both TFEB and TFE3 are inhibited by class IIa histone deacetylases (i.e., HDAC4, 5, and 7) and this inhibition is reversed by the Protein kinase D family (i.e., PKD1, 2 and 3). However, if this pathway is important in ER stress induced muscle atrophy has not been investigated.

Insulin resistance accompanied by hyperglycemia is common in critically ill patients and strongly associated with increased morbidity and mortality (96). The underlying mechanism is still elusive but includes a decreased activity at the level of the insulin receptor and the post-receptor level (97). The insulin receptor is a receptor tyrosine kinase and its activation by insulin causes the cytosolic domains to autophosphorylate. The activated insulin receptor phosphorylates tyrosine residues in proteins such as insulin receptor substrate 1 (IRS-1) and IRS-2 that mediate downstream signaling. IRS-1 activates PI3K and Akt, which regulate muscle protein homeostasis (98). In addition, activation of the insulin receptor via the Akt substrate of 160 kDa leads to translocation of the glucose transporter (GLUT)-4 into the plasma membrane, which increases cellular glucose uptake (99). Insulin resistance causes decreased glucose uptake in skeletal muscle. Under physiological conditions myocytes internalize circulating glucose via their main glucose transporter glucose transporter-4 (GLUT4). GLUT4 is contained in storage vesicles that are localized near the sarcolemma, the transverse tubular system and in the trans-Golgi region (100). Normally, these GLUT4 storage vesicles are translocated to the plasma membrane and glucose uptake is increased in skeletal muscle in response to insulin (101). In insulin resistance insulin receptor signaling is perturbed, GLUT4 translocation is attenuated and circulating glucose cannot be taken up. Since glucose is the main source of energy in muscle, less energy-rich phosphates (e.g., ATP) are formed, which reduces muscular energy supply (102). This impairs muscle function, especially strength and endurance. At the same time, insulin has an anabolic effect on muscle by increasing protein synthesis and inhibiting protein breakdown (98, 103), so that insulin resistance leads to disturbed protein homeostasis with increased protein breakdown and reduced protein synthesis, resulting in muscle atrophy and weakness (27, 37). Importantly, insulin resistance has been reported to be a predictor for ICUAW (47). Indeed, previously we reported that GLUT4 was trapped in the perinuclear space of myofibers of critically ill patients whereas it was localized to the sarcolemma of myofibers of control subjects (102). This observation was most pronounced in patients with critical illness myopathy, a severe form of ICUAW further supporting that insulin resistance is involved in ICUAW.

Although the contribution of the UPR to insulin resistance in skeletal muscles of the critically ill is uncertain, published evidence supports this assumption. For example, ER stress has been shown to be associated with insulin resistance, which was accompanied by a suppression of insulin receptor signaling and serine phosphorylation of insulin receptor substrate-1 (IRS-1), in vivo and in vitro (104). Insulin resistance was also described in Xbp1 knockout mice (104). Inhibition of ER stress by the chemical chaperones 4-phenyl butyric acid (4-PBA) or taurine-conjugated ursodeoxycholic acid (TUDCA) restored systemic insulin sensitivity, enhanced the action of insulin on muscle, and normalized hyperglycemia in obese and diabetic mice (105). These data indicate that inhibition of ER stress by the chemical chaperones 4-PBA and TUDCA counteracts insulin resistance.

The adenosine 5’-monophosphate (AMP)-activated protein kinase (AMPK) that senses and regulates cellular energy states increases the expression of GLUT4 (106). Deletion of Ampk2a (encoding AMPK2α) in mice causes a reduction in glucose utilization in skeletal muscle (107). Activation of ER stress by tunicamycin or thapsigargin (an inhibitor of ER-specific calcium ATPase) reduces AMPK activity, insulin signaling and glucose uptake in rat-derived L6 myotubes, mainly due to an increased extracellular signal-regulated kinase (ERK) phosphorylation. Inhibition of ERK restored AMPK phosphorylation and glucose uptake that were deteriorated in response to ER stress (108). These data implicate a cross talk between ER stress and the major energy sensor AMPK in myocytes.

Insulin resistance caused by the saturated fatty acid palmitic acid also caused ER stress as indicated by an increased mRNA and protein expression of the ER stress-markers ATF6, ATF3, GRP78, CHOP, and Tribbles 3 (TRIB3) in C2C12 myotubes. Palmitic acid-induced ER stress, insulin resistance and, inflammation were attenuated by low concentrations of the proteasome inhibitor Bortezomib and these effects were mediated by AMPK. Importantly, knockdown of AMPK abolished the protective effects of Bortezomib (109). These studies indicate that ER stress induced insulin resistance is at least partially mediated by inhibition of AMPK activity.

The pseudokinase Tribbles 3 (TRIB3) that is expressed in various tissues such as liver, adipose tissue, heart, and skeletal muscle (110) has also been shown to impede insulin signaling (111). TRIB3 is increased in skeletal muscle of patients with type 2 diabetes mellitus known to be insulin resistant (110). TRIB3 binds to Akt and inhibits Akt phosphorylation eventually preventing insulin-stimulated glucose uptake in C2C12 myocytes (112). Activation of ER stress by tunicamycin and thapsigargin causes an increase in TRIB3 in C2C12 cells and skeletal muscle, which possibly mediates ER stress-induced insulin resistance. Accordingly, when TRIB3 was overexpressed in tibialis anterior of mice insulin signaling was greatly impaired (113). In addition, deletion of Trib3 in mice prevented high-fat diet induced insulin resistance in skeletal muscle (113). If Tribbles 3 plays a role in insulin resistance in skeletal muscle of the critically ill is not known.

The Protein tyrosine phosphatase 1B (PTP1B) that is a key negative regulator of insulin signaling by dephosphorylation of tyrosine residues of the insulin receptor which inhibits insulin receptor signaling (114). PTP1B that is localized on the ER membrane has also been shown to mediate insulin resistance in response to ER stress (115). Specifically, high-fat diet (for 20-weeks) induced ER stress (e.g., GRP78) caused an increase in PTP1B expression in the gastrocnemius of mice. This response was attenuated by administration of TUDCA for 5-weeks. Likewise, tunicamycin-induced ER stress caused an increase in PTP1B levels in C2C12 myotubes. In addition, tunicamycin-induced phosphorylation of eIF2α and JNK2 were attenuated in Ptpn1 knockout mice, encoding PTP1B. Downregulation of PTP1B or inhibition of ER stress by TUDCA treatment improved glucose uptake in tunicamycin treated C2C12 myotubes (116). These data indicate that ER stress activates PTP1B and that PTP1B is required for activation of ER stress pathways to mediate insulin resistance in skeletal muscle. In addition, it has been reported that mice harboring a germline deletion of Ptpn1 were protected against high-fat diet induced insulin resistance, weight gain, and hepatic steatosis. These mice also exhibited improved glucose uptake and Akt phosphorylation in the gastrocnemius. In addition, high-fat diet-induced increase in GRP78/BiP and phosphorylation of eIF2α as well as JNK2 were significantly blunted in Ptpn1 knockout mice. These data support the hypothesis that ER stress induces the expression of PTP1B which mediates insulin resistance in the skeletal muscle in response to high-fat diet (117). It has also been shown that PTP1B potentiates IRE1-mediated ER stress signaling pathways in mouse embryonic fibroblasts (118).

Hyperglycemia leads to elevated glucosamine levels that are also able to induce insulin resistance and ER stress as indicated by an increased expression of the ER stress markers GRP78/BiP, XBP1 and ATF6 in both human and L6 myotubes (119). Pretreatment of both rat and human myotubes with 4-PBA or TUDCA, attenuated glucosamine-induced ER stress and insulin resistance. Glucosamine reduced the expression of GLUT4, myocyte enhancer factor 2A (MEF2A) and peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC-1α) expression, which was prevented by 4-PBA and TUDCA in both rat and human myotubes. Importantly, glucosamine-mediated downregulation of GLUT4, MEF2A and PGC-1α was attenuated by ATF6 silencing. These data strongly indicate that the ATF6 arm of the UPR mediates glucosamine-induced insulin resistance (119).

In summary, these studies indicate that ER stress and UPR pathways are involved in systemic and muscular insulin resistance and muscular glucose uptake. However, if this pathomechanism is involved in ICUAW and if its targeting has therapeutic potential requires further investigation.

Immobilization of critically ill patients will inevitably lead to disuse and unloading of skeletal muscle which are major risk factors for ICUAW (21). Although immobilization allows complex treatments in ICU, the accompanying unloading of the skeletal muscle will lead to skeletal muscle atrophy and a reduction in muscle force. The loss of muscle mass and function delays recovery and attempts to prevent muscle atrophy through physical therapy or other interventions have been insufficient. Despite the fact that skeletal muscle atrophy due to immobilization is a prevalent and severe clinical problem in many different medical situations, the underlying mechanisms are not well understood. Of note, bed rest in humans and immobilization of animals have been shown to cause skeletal muscle atrophy mainly by an increased activity of UPS-mediated protein degradation with enhanced expression of MuRF1 and atrogin1 (27, 29–32, 120). Since skeletal muscle atrophy due to disuse leads to an upregulation of genes encoding sarcoplasmic reticulum (SR) calcium-handling proteins and considering that many of the proteins that are induced with ER stress are also calcium-handling proteins a connection between disuse atrophy and ER stress is likely. This hypothesis was tested in soleus muscle of rats after seven days of unloading. In this model, the expression of the UPR markers GRP78/BiP, calreticulin, CHOP and eIF2α remained unchanged in response to inactivity (121). Similar observations have been made by Baehr et al., were hindlimb unloading did not cause ER stress in soleus of adult (9 month) and old (29 month) male rats (122). However, ER stress was different between soleus and tibialis anterior of the same animals and revealed age-specific effects. The authors also investigated if reloading of the muscle show age-specific effects on ER stress. Indeed, in response to 14 days of reloading, both GRP78/BiP and CHOP were increased in soleus of adult but not old mice. Also, CHOP increased upon reloading in tibialis anterior of old but not adult rats, whereas GRP78/BiP remained unchanged (122). Contrary to the mentioned data, an upregulation of ER stress and UPR pathway genes in response to immobilization was reported by other groups. For example, 10-days of bed rest resulted in a persistent upregulation of the UPR in skeletal muscle of young men, as shown by microarray analysis (123). In addition, the expression of ATF4, a downstream transcription factor of PERK branch, increases in response to 3 days of unilateral hindlimb immobilization in tibialis anterior in mice (124). Furthermore, conditional Atf4-knockout mice lacking ATF4 in muscle are resistant to immobilization-induced muscle atrophy, fasting-induced muscle atrophy, and age-related muscle atrophy (87, 124–126). Conversely, forced expression of ATF4 is sufficient to induce muscle atrophy even in the absence of inactivity (87, 126). Further studies that address the individual UPR pathways are needed to elucidate their specific role of ER stress in disuse-induced skeletal muscle atrophy.

Denervation is another model to investigate mechanisms of immobility-induced muscle atrophy. To achieve this phenotype the sciatic nerve is cut and dissected in rodents (29). It has been shown that denervation leads to an increased phosphorylation of eIF2α and splicing of XBP1 mRNA in the gastrocnemius 3 days and 7 days post denervation indicative for an activation of the PERK and IRE1 arm of the UPR. In denervation, also the expression of Grp78/BiP and Chop was also increased (127). These data show that denervation activates the PERK and IRE1 arm of the UPR in skeletal muscle. It also showed that the proapoptotic transcription factor CHOP, also a target gene of the PERK arm of the UPR, is involved in denervation induced muscle atrophy. Specifically, CHOP deficient mice exhibited an increased skeletal muscle wasting in response to denervation (127). Parveen et al., demonstrated an increased mRNA expression of Grp78, Atf4, and Atf6 and an increased protein content of PERK, ATF4, CHOP, IRE1α, and XBP1s in the gastrocnemius of mice in response to 14 days of denervation (128). They also found that a myofiber-specific ablation of Xbp1 attenuated denervation-induced skeletal muscle atrophy in mice (128). A further study in mice revealed that the PERK target gene Sestrin2 (Sesn2) that is a stress responsive protein and protects cells from cell death by maintaining mitochondrial homeostasis (129) is upregulated in the gastrocnemius two weeks after denervation (130). Knockdown of Sesn2 aggravated denervation-induced skeletal muscle atrophy. These data indicate that the PERK arm of the UPR via upregulation of Sestrin2 attenuates denervation-induced skeletal muscle atrophy (130). However, because the IRE1/XBP1 arm of the UPR has also been shown to increase Sesn2 expression, this effect is not unique for the PERK arm (131). In summary, these data reveal a complex involvement of the different UPR arms in various atrophy models depending on age and muscle types. Further studies are needed to decipher the involvement of the individual UPR branches in specific atrophy conditions. Generally, it seems that dependent on the muscle atrophy trigger, different branches of the UPR are activated. One further example for this hypothesis is the differential expression of the UPR in sarcopenic patients suffering from either chronic obstructive pulmonary disease (COPD) or lung cancer cachexia. While lung cancer cachexia-patients show an increase in many ER stress markers and all three UPR branches, only few ER stress markers and the IRE1α branch of the UPR are increased in COPD patients (132).

Almost all major ER stress related genes have been found to be upregulated in sepsis, as summarized by Khan and colleagues (133). Moreover, the UPR is able to regulate cytokine production at various steps, from stimulation of pattern recognition receptors (PRR), to modulation of inflammatory signaling pathways, and the regulation of cytokine transcription factors (134). These reviews (133, 134) based on a significant number of original articles demonstrating that there is a connection between sepsis, inflammation and the UPR pathway. However, if these UPR pathways are also relevant for ICUAW has not been shown.

In sepsis, expression of the pattern recognition receptors Toll-like receptor (TLR) 2 and TLR4 as well as their downstream pathways are induced in skeletal muscle (31). Specifically, we recently showed that the acute phase protein serum amyloid A1 (SAA1) via TLR2 and TLR4 activates NF-κB p65, that translocates to the nucleus, binds to NF-κB response elements in the promoter region of its target genes and increases their expression (31). Inhibition of NF-κB inhibited sepsis-induced muscle wasting in mice (31). In addition, TLR2 and TLR4 have been shown to engage IRE1α to promote cytosolic splicing and activation of XBP1 to increase the expression of proinflammatory cytokines, such as IL-6, to regulate innate immune responses in macrophages (135). Our data also revealed that SAA1 via TLR2 and TLR4 and the NF-κB pathway increased the expression of IL-6 in myocytes (28, 31). However, if SAA1 binding to TLR2 and TLR4 also affects the IRE1α/XBP1/IL-6 axis in myocytes has not been tested. Moreover, the expression of several TLR as well as their downstream signaling adaptor, myeloid differentiation primary response 88 (MyD88), are increased in response to denervation in skeletal muscle of mice. Importantly, deletion of Myd88 inhibited the activation of XBP1 and attenuated muscle atrophy in response to denervation (128). These data indicate that TLR and TLR signaling are connected to UPR during inflammation.

The skeletal muscle has been shown to be affected by and to react to immunologic conditions such as proinflammatory stimuli (30–32, 136). The involvement of the UPR in the pathogenesis of different myopathies such as Sporadic inclusion body myositis, Myasthenia gravis, Duchenne muscular dystrophy has been reviewed recently (137). In immune cells, ER stress and the UPR are implicated in the regulation of the inflammatory state (138). A couple of studies demonstrated that also in myocytes, proinflammatory stimuli regulate UPR signaling. For example, acute muscle injury by cardiotoxin injections resulted in an inflammatory response and an increased expression of ER stress and UPR genes, including chaperones and genes from all three branches of the UPR. A link between increased cytokines and ER stress in skeletal muscle has also been suggested by the observation that the UPR is disrupted in inflammatory muscle diseases (139).

Interestingly, Interferon gamma (IFN-γ) treatment of myocytes causes a muscular immune response and activates the IRE1α and PERK but not the ATF6 arm of the UPR. Specifically, IFN-γ caused ER stress in mouse myogenic progenitor cell-derived myotubes in vitro. In addition, ER stress was also caused by cardiotoxin injections in tibialis anterior in vivo. At base line myoblasts and differentiated myotubes do not express class I major histocompatibility complex (MHC) I and MHCII (140). However, when myotubes were treated with IFN-γ the expression of MHCI, TLRs and proinflammatory cytokines (i.e., IL-1β, IL-6) was considerably increased. This response was further triggered by inhibition of ER stress through 4-PBA. The ER stress inducing agents tunicamycin and thapsigargin prevented the IFN-γ-induced expression of MHCII and MHCI molecules. Inhibition of IRE1α attenuated the UPR stressor-mediated reduction of these immunobiological molecules (141). These data indicate that the IRE1α arm mediates immunobiological suppression in myocytes during inflammation. Notably, not only does inflammation affect the UPR but this also works the other way around. IRE1α induces JNK signaling, which regulates many inflammatory genes (68). IRE1α signaling also activates the NF-κB signaling pathway that induces multiple inflammatory cytokines, such as IL-6. In addition it has been demonstrated that IRE1α and IKK are linked by TRAF2 to activate NF-κB (142) and this pathway was also activated in inflammatory skeletal muscle atrophy (30, 31).

There is strong evidence that ER stress induces inflammation by activating inflammasomes (143). Inflammasomes are multiprotein complexes that act as scaffolds for caspase-1-mediated maturation and secretion of inactive pro-IL-1β and pro-IL-18 to active IL-1β and IL-18 (144). Both IL-1β and IL-18 are involved in sepsis (145), sepsis-induced muscle failure (30) and sepsis-induced cardiomyopathy (35). Cytoplasmic receptors of the nucleotide-binding domain (NOD)-like receptor (NLR) family are key inflammasome components (144). The NOD-, leucine-rich repeat (LRR)- and pyrin domain-containing protein-3 (NLRP3) is one of the best-characterized NLR proteins able to form an inflammasome complex and all components of the NLRP3 inflammasome are contained in the skeletal muscle (146). Importantly, ER stress has been reported to activate the NLRP3 inflammasome with a subsequent increase in IL-1β secretion. Because this effect was independent of the UPR a contribution of a novel ER stress response signaling pathway was proposed (143). These findings taken together strongly indicate that there is a direct link between ER stress and inflammation in skeletal muscle and given the role of proinflammatory cytokines in muscle atrophy also a connection between ER stress and ICUAW is very likely. This however needs to be proven.

The role of circulating and tissue residing immune cells for ICUAW in general and ER stress and UPR in particular is not well understood. However, novel technical advances may provide new insights into the role of the various different cell types contained in muscle for the pathogenesis of muscle wasting in critically ill patients. For example, using a combined approach of single-cell RNA sequencing and single-cells mass cytometry, ten different mononuclear cell types (i.e., muscle satellite cells, fibro-adipogenic progenitors, macrophages, neutrophils, endothelial cells, B cells, T cells, glial cells, smooth muscle and mesenchymal cells and scleraxis-positiv tenocytes) were mapped in adult mouse muscle (271). Another study used single-cell RNA-sequencing to classify single cells isolated from tibialis anterior muscle into myocytes, endothelial cells, fibroblasts, mesenchymal stem cells, macrophages, neutrophils, T-cells, B-cells, and dendritic cells. These data indicate that several classes of immune cells are contained in muscle. Importantly, early during sepsis, a decrease in the proportion of most non-immune cell populations and an increase in immune cells (i.e., neutrophils and macrophages) was observed in muscle of male and female mice subjected to the cecal ligation and puncture (CLP) and faecal-induced peritonitis (FIP) models of sepsis. These findings persisted for one month after sepsis (272). In summary, these data implicate that immune cells may be involved in sepsis-induced muscle wasting. However, if UPR in any of these muscle residing cells contributes to ICUAW has not been investigated and is far from being understood. Likewise, if ER stress and UPR affect the levels of immune cells in muscle tissue during acute or chronic injury is also unknown. Nevertheless, these novel techniques will allow us to investigate the function of UPR in any of these cells and its importance for ICUAW.

Both hypoxemia and hypercapnia are associated with an increased mortality in critically ill patients, particularly in those patients with acute respiratory distress syndrome (ARDS).

Many ICU survivors experience a long-lasting impairment of muscle function that may even be apparent five years after their acute illness (44, 45), which is indicative for a perturbed muscle regeneration. This is surprising given the fact that skeletal muscle has a tremendous capacity for regeneration (147). Regeneration of skeletal muscle is realized by muscle residing stem cells, called satellite cells, that are located between the sarcolemma and the basal membrane of myofibers in a quiescent state (147). In response to myofiber damage, satellite cells get activated, proliferate and differentiate into myoblasts that in turn fuse with injured myofibers (147). Although the majority of satellite cells differentiate into the myogenic lineage, few of them undergo self-renewal, return to a quiescent state and replenish the satellite cell pool. Myogenic differentiation requires a well-controlled expression of myogenic transcription factors (148), myoblast-fusion proteins (149, 150) and contractile proteins (151). This involves the sequential expression of the transcription factors paired box gene 3 (Pax3) and Pax7 followed by the expression of myogenic regulatory factors (MRFs) such as Myf5, MyoD, Myogenin, and MRF4. The role of myogenesis in maintaining skeletal muscle size is well described (152); however, its importance in critical illness in general and ICUAW in particular is unexplored. Previously is has been reported that regenerative muscle capacity is impaired in ICU patients (153) and septic mice (154); although, genes involved in skeletal muscle regeneration are upregulated in muscle of survivors of critical illness (155). The process of myogenesis and the underlying mechanisms are investigated in vitro by using primary satellite cells and myoblasts, mouse-derived C2C12 myoblasts or rat-derived L6 myoblasts, which form multi-nucleated myotubes in low serum conditions. To investigate myogenesis in vivo, intramuscular injections of barium chloride (156) or cardiotoxin, a snake venom toxin that causes myofiber necrosis without effects on satellite cells viability (157), and subsequent time course analyses are used.

Previously it has been reported that the UPR mediates selective apoptosis of differentiation-incompetent myoblasts during skeletal muscle formation (158). In these cells, ER stress activated caspase-12, which in turn activated caspase-9 and caspase-3 to mediate apoptosis. Previously it was shown that exclusively the ATF6 arm of the UPR was increased in those myoblasts that undergo apoptosis in response to induction of myogenic differentiation (159). When ATF6 or caspase-12 were inhibited, myoblast apoptosis was reduced and the formation of multi-nucleated C2C12 myotubes was attenuated. An increase in caspase-12 was also observed in skeletal muscle during embryonic development. These data suggest that the ATF6 arm of the UPR is involved in selective apoptosis of those myoblasts that are susceptible to stress (159). However, caspase-12 deficient mice develop normally (160), indicating that caspase-12–dependent apoptosis is not essential for muscle development. Interestingly, activation of ER stress has been shown to promote myofiber formation (161). When ER stress was activated by tunicamycin or thapsigargin differentiation-associated apoptosis of myoblasts was increased. Nevertheless, the surviving myoblasts were resistant to apoptosis and differentiated well into contracting myotubes. These data suggest that during myogenesis differentiation incompetent myoblasts are specifically targeted by ER stress (161).

The other arms of the UPR are also implicated during myogenesis. A transient activation of the PERK/eIF2α arm of the UPR and its downstream target, CHOP, has been observed in a subset of myoblasts during differentiation (162). Knockdown of CHOP expression lead to an earlier and more robust myogenic differentiation, whereas its overexpression delayed differentiation. Cells that expressed CHOP did not express the myogenic regulatory factors MyoD and Myogenin indicating that CHOP inhibits myogenic differentiation by inhibition of Myod gene expression.

The IRE1 arm of the UPR leads to the generation of XBP1s, a potent transcription factor (163). XBP1 was shown to be a target gene of the transcription factors MyoD and Myogenin (164) that are essential for myogenic differentiation suggesting a connection between XBP1 and myogenesis. XBP1 regulates the expression of genes involved in maintenance of ER function, growth, and DNA damage and repair pathways in myoblasts as well as in myotubes (165). Overexpression of XBP1 inhibited the expression of myogenic differentiation markers, such as Myh7, Myh4 and Mef2c, and led to formation of shorter and less mature C2C12 myotubes during differentiation. This effect was caused by XBP1-mediated induction of Muscle, Intestine and Stomach Expression 1 (Mist1)/Basic Helix-Loop-Helix Family Member A15 (BHLHA15) expression. Specifically, XBP1 binds to the Mist1 promoter in myoblasts, XBP1s and ER stress increase the expression of Mist1 and downregulation of XBP1 reduced the ER stress-induced Mist1 expression (166). Mist1 is a transcription factor that inhibits the activity of MyoD (166). In summary, these data indicate that XBP1 plays a role in myogenic differentiation by induction of Mist1-mediated inhibition of MyoD (165).

The role of satellite cells in general and the UPR in particular in ICUAW is not well defined. Nevertheless, previously it has been shown that the downstream target of the PERK arm of the UPR eIF2α is constitutively phosphorylated in quiescent satellite cells and that this phosphorylation is decreased in activated satellite cells (167). When a phosphorylation resistant mutant of eIF2α was expressed (i.e., eIF2αS51A) or PERK was deleted satellite cell quiescence was lost as indicated by an increased Myf5 and MyoD expression. Satellite cells that express eIF2αS51A are able to undergo terminal differentiation and subsequent fusion with myofibers. However, these satellite cells are incapable to undergo self-renewal. Interestingly, this study also showed that inhibition of eIF2α dephosphorylation using the salubrinal-derivative Sal003 promoted satellite cell self-renewal ex vivo (168). Sal003 inhibits the serine/threonine phosphatase GADD34/PP1, which dephosphorylates phosphorylated eIF2α, and therefore keeps eIF2α in a phosphorylated state (169). In addition, inhibition of eIF2α dephosphorylation by Sal003 retained the myogenic capacity of satellite cells when transplanted in skeletal muscle of a mouse model of Duchenne muscular dystrophy in vivo (168). However, these data are in contrast to recently published work were eIF2αS51A overexpression was shown to cause cell death and diminish the differentiation of cultured myoblasts (162). Differences in the experimental approaches and techniques as well as the cell types used for analyses might account for these discrepancies (162, 168).

In summary, these data show that ER stress and UPR play a key role in satellite cell function and fate, skeletal muscle regeneration and myogenesis. However, it is currently unknown to which extend satellite cell function and regeneration are involved in ICUAW and even more so if ER stress and UPR are involved in this process.

To relive ER stress higher amounts of ATP are needed especially for a proper function of newly synthesized ER chaperone proteins. This suggests a close connection between the ER and mitochondria, which are the main site of ATP production in myocytes. Because muscle is a highly energy consuming tissue the interaction between the ER and mitochondria during ER stress possibly affects its function and adaptation processes. This is supported by the observation that tunicamycin-induced ER stress causes a translocation of mitochondria to the ER. These translocated mitochondria were reported to increase ATP production, which increases the intracellular ATP pool (282). The transport of ATP and signaling molecules between the ER and mitochondria occurs at interfaces between both organelles, which are called mitochondria-associated ER membranes (MAM) (283). This interface is realized by specific proteins localized to the outer mitochondrial membrane and the ER (284). MAM are involved in ER stress, calcium signaling, mitochondrial dynamics, lipid trafficking, initiation of apoptosis, and autophagosome formation [for review (284–286]. Interactions between ER and mitochondria in skeletal muscle and its effects on calcium homeostasis and reactive oxygen species has been reviewed recently (287). MAM have also been shown to influence insulin signaling through different pathways, including those associated with ER stress responses, mitochondrial function, and inflammation. A recent review addressed possible mechanisms underlying MAM-associated insulin resistance (Cheng et al., 2020). In addition, MAM have also been shown to be affected by insulin resistance, obesity, and type 2 diabetes. However, a decrease (Tubbs et al., 2018) as well as an increase (Arruda et al., 2014;Thoudam et al., 2019) of MAM in murine and human skeletal muscle as well as cultivated myocytes has been shown in response to metabolic and ER stress. These conflicting results are possibly due to differences in the techniques used to investigate MAM structure and function in skeletal muscle and myocytes. Likewise, the data on changes of the amount of proteins that mediate tethering of mitochondria to the ER in response to metabolic stress are not uniform. It is therefore difficult to assess the functional impact of MAM modifications for muscle. Also, no data are available for the role of MAM in muscle of critically ill patients, therefore, further analyses are needed to conclude if MAM are involved in the pathogenesis of ICUAW.

To prevent sequelae of ICU treatment, critically ill patients receive physiotherapy and if applicable early mobilization. The goal is to prevent muscle loss and to maintain and support muscle function, as reviewed elsewhere (41, 170). Protocol-based early mobilization has been reported to improve muscle function and reduce insulin resistance in ICU patients, and is recommended in guidelines (171–174). In general ICU patients, early mobilization has been shown to be clinically beneficial (174–176). If additional physiotherapeutic measures, such as neuromuscular electrical stimulation (NMES), have an effect on muscular structure or function is debated. Some reports described that such interventions prevent muscle atrophy and improve glucose metabolism and physical function, whereas others did not find any effect (102, 177–182). Whole-body vibration (WBV) has been shown to be effective for maintenance of muscle mass and strength in elderly individuals. However, its impact on long-term outcome of ICU patients has not been investigated, although it is feasible in critically ill patients (40, 183, 184). We recently investigated if additional muscle activating measures (e.g., NMES or WBV) have an effect on muscle structure and function when applied in combination with daily protocol-based physiotherapy in critically ill patients (SOFA ≥ 9). We observed that muscle integrity was maintained and found an increased myocyte cross sectional area in the intervention group as compared to those patients that received daily protocol-based physiotherapy only. However, early administration of additional muscle activating measures did not improve muscle strength or function in the critically ill (41). Although early mobilization, physiotherapy, muscle activating measures and physical training are quite distinct in terms of the intended goals and the applied work load, it is noticeable that acute and chronic exercise have been associated with ER stress. For example, a single strong bout of exercise and chronic exercise caused an activation of different branches of the UPR. However, whether or not mobilization therapy increases or decreases ER stress and UPR, if this interferes with the effects of other ICUAW risk factors on ER stress and if this has an effect on muscle structure and function is not known. It is possible, but not proven, that muscle activation measures aggravate UPR that is already induced by immobilization, inflammation, and treatment and that this is time, context, dose and frequency dependent.

If applicable critically ill patients receive a structured physiotherapy during ICU treatment (185). The effects of physiotherapy and mobilization on ER stress-induced UPR with regards to muscle structure and function are uncertain. Although not directly comparable to physiotherapy endurance and resistance exercise have been shown to regulate UPR in muscle. The effects of exercise on UPR were often investigated in rodent models that combined exercise with metabolic stress induced by, among others, type 2 diabetes mellitus and or high-fat diet. In these models, type 2 diabetes mellitus induced by streptozotocin injection and high-fat diet caused glucose intolerance, induced ER stress and endoplasmic reticulum-associated degradation (ERAD) in skeletal muscle of male mice. Swimming (1h per day, 5 days per week for 6 weeks) of these mice improved glucose intolerance and alleviated ER stress in skeletal muscle (186). Specifically, type 2 diabetes mellitus significantly increased the phosphorylation of IRE1α and ATF6 expression in skeletal muscle and swimming alleviated this response. In contrast, the PERK branch of the UPR remained unaffected by type 2 diabetes mellitus and exercise. Similar findings have been reported in muscle biopsy specimens from vastus lateralis from young and old untrained men in response to resistance exercise. Acute resistance exercise led to an increase in both XBP1 and XBP1s, indicating that exercise may stimulate the IRE1α and ATF6 arms of the UPR. In contrast, acute resistance exercise did not affect the PERK pathway as no increase in Atf4, Chop, Gadd34, or eif2a mRNA expression and eIF2α phosphorylation were found 24 hours and 48 hours after exercise (187). In contrast to swimming and resistance exercise, it was shown that endurance exercise, when applied in addition to high-fat diet in mice, further increased the expression of several UPR markers when compared to high-fat diet alone. This response was different between the muscle (i.e., soleus vs. tibialis anterior) and organs (i.e., skeletal muscle, liver, pancreas) investigated (188). Importantly, it has been reported that UPR is absent in the heart in response to short-term exercise in adult male Sprague Dawley rats (189) and male and female mice (190). It is also interesting to note that a single bout of running exercise activates the UPR in some (e.g., quadriceps, gastrocnemius) but no other muscles (e.g., erector spinae) (190) indicating a muscle specific regulation and possibly function of the UPR. Muscle specific (soleus vs. tibialis anterior) and age-related regulations of ER stress markers have also been observed in adult (9 month) and old (29 month) male rats following 14 days of hindlimb unloading by tail-suspension and 13 days of reloading (122). In summary, these data imply that the UPR is differentially regulated in a muscle- and tissue-specific manner that depends on the type of exercise applied (e.g., swimming, resistance- or endurance exercise) and its duration (e.g., short- or long-term, single bout) as well as the preexisting metabolic situation and age. An additional level of complexity is added when the different branches of the UPR are considered. Whether or not physiotherapy and early mobilization are able to increase or decrease UPR, if this adds on to preexisting UPR caused by inflammation, immobilization, insulin resistance, and if this has beneficial or detrimental effects on skeletal muscle in critically ill patients needs further investigation. Given the kinetics of UPR it will be interesting to know if there is a time window where critical illness-induced and therapy related ER stress and UPR interfere with each other and aggravate muscle damage, which most certainly will have clinical implications.

These data show that exercise as well as immobilization are able to activate ER stress and UPR and that UPR is involved in the control of muscle mass and function. However, to which degree immobilization- vs. mobilization-mediated UPR activity affects the occurrence of ICUAW and if this knowledge can be used therapeutically is not known.

Given its importance in multiple cellular processes and human diseases such as cancer, neurodegenerative disease (e.g., Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis, Huntington’s disease) and metabolic disorders (e.g., type 2 diabetes mellitus) attempts have been undertaken to pharmacologically target the UPR. These efforts have been summarized recently (191). However, UPR modulators have rarely been tested in critically ill patients or ICUAW, which is not surprising as several critical functions of the UPR in muscle are not fully understood and sometimes controversial. As already mentioned, preclinical animal models revealed that targeting specific UPR branches or mediators may have beneficial or detrimental consequences for disease severity or progression which depends on the disease context. Among others, disease-, tissue- and fiber type-specific activation of distinct UPR pathways that often follow a time course complicate decision making for the therapeutic usage of UPR modulators. Many preclinical disease models are based on otherwise healthy young rodents that do often not have contributing risk factors that may affect the disease course, which is in contrast to the critically ill especially on medical ICUs. These studies are mainly performed in male, some in female rodents but usually not in animals of both sexes. In order to evaluate therapeutic effects especially in muscle results from multiple muscles (e.g., slow- and fast-twitch muscles) need to be obtained, which is often not the case. Finally, although there are many valid reasons to choose a preemptive treatment in situations where the onset of a disease and even its cause is very well defined, this is in great contrast to the clinical reality. These limitations need to be considered when decisions are made to translate from preclinical to clinical applications of UPR modulators.

Novel UPR modulators were found to be contained in natural products (210), which could serve as a source for novel lead compounds. Such natural UPR modulators provide an innovative and novel strategy to prevent or treat ICUAW. We have here listed a few promising natural compounds, their source and the model systems they were investigated in ( Table 2 ). However, for most of these compounds their effects on ER stress have been investigated in heart-failure models in vivo or in cardiomyocytes in vitro. Although favorable effects of these compounds have been shown in human patients and animal models, further analyses are needed to elucidate if they also inhibit ER stress and UPR pathways in skeletal muscle as part of their mode of action and if this has favorable effects for the critically ill.