Links between autophagy and tissue mechanics

Multicellular organisms have to constantly adapt to external stimuli and modifications in their intracellular milieu. Responses that maintain homeostasis are integrated at the tissue and cellular level. Organs, tissues and cells have to face both chemical (e.g. Ca²⁺, amino acids, cytokines, chemokines and hormones) and physical challenges. Physical challenges can be external (e.g. radiofrequency, X-rays, UV, electric fields, radioactivity and microgravity) or internal. Among the internal physical and mechanical forces, several are tissue-dependent, such as compression in bones; tension in the muscles, lung and kidney; as well as shear stress in the blood and lymphatic circulatory systems (blood flow and lymph flow) and the kidney (urinary flow) (Janmey and Miller, 2011). Some are common to all solid tissues, such as interactions with the extracellular matrix (ECM) and intercellular adhesion (Kechagia et al., 2019; Pinheiro and Bellaϊche, 2018; van Helvert et al., 2018), while others are common to all cells, with or without persistent matrix interactions, such as organelle deformation during migration, intravasation and extravasation (Feng and Kornmann, 2018; Garcia-Arcos et al., 2019; Moreau et al., 2018). Mechanical forces are integrated by all cell types, including stem cells, to control their differentiation program (Argentati et al., 2019; Vining and Mooney, 2017). Finally, several pathological states are dependent on changes in mechanical forces. This paradigm is particularly relevant in tumor development and metastasis (Follain et al., 2020), cardiovascular diseases (Conway and Schwartz, 2013; Garoffolo and Pesce, 2019) and inflammatory diseases (Cambré et al., 2018). Additionally, pressure injury is a major risk in critically ill populations (Alderden et al., 2017).

Similar to chemical stimuli, which are recognized at the cell surface by receptors, channels and transporters, mechanical forces are in many cases detected by cell-surface sensors that can be associated with specialized membrane structures, such as cell–cell adhesion sites, focal sites and the primary cilium (Pinheiro and Bellaϊche, 2018; Ferreira et al., 2019; Spasic and Jacobs, 2017; Sun et al., 2019). Moreover, remodeling of the different cytoskeleton filaments in the cytoplasm and nucleus triggers cytoplasmic signaling and, in certain circumstances, the onset of transcriptional programs (Janota et al., 2020; Kirby and Lammerding, 2018; Mammoto et al., 2012; Uhler and Shivashankar, 2017).

Among the cytoplasmic responses to mechanical forces, recent studies have uncovered the role of macroautophagy (hereafter referred to as autophagy) in the translation of mechanical forces into biological responses (Dupont and Codogno, 2016; Hernández-Cáceres et al., 2021; Kenific et al., 2016b; King, 2012). Autophagy is one of the major degradative pathways in cells, culminating in the lysosomal compartment (see Box 1). At the basal level, autophagy controls the quality and quantity of intracellular biomass (Mizushima and Komatsu, 2011). Autophagy is also a stress-responsive catabolic process that is induced by many different physiological stimuli and pathophysiological situations, including nutrient deprivation, growth factor withdrawal, hypoxia, oxidative stress, protein accumulation and infection (Levine and Kroemer, 2019).

Although the regulation and role of autophagy in diverse situations has been the focus of much attention, less is known about the autophagic response to physical cues (Dupont and Codogno, 2016; King, 2012). However, recent studies have illuminated the role of autophagy in the translation of mechanical forces into a wide range of biological responses, from the control of cell size and metabolism, to anti-atherogenic effects. The aim of this Review is to present and discuss these recent findings that constitute a novel aspect of the role of the autophagic machinery in physiology and pathophysiology.

Many cells in multicellular organisms are embedded in a three-dimensional microenvironment where they are constantly challenged with different forms of stress arising from intrinsic biological processes or from the extrinsic physical environment. For example, mesenchymal stem cells preferentially undergo osteogenesis on stiff ECM, whereas they are able to differentiate into neurons on soft ECM (Engler et al., 2006). At the tissue level, mechanical signals contribute to morphogenesis during development and organ functions (Mammoto et al., 2013). For example, forces generated by blood flow directly impact the physiology of the cardiovascular system, which transports pressurized blood, through the process of mechanotransduction (Chien, 2007; Garoffolo and Pesce, 2019). Physical forces from mechanical loading also regulate the balance between bone formation and bone resorption (Alfieri et al., 2019; Curtis et al., 2020; Duncan and Turner, 1995), whereas mechanical forces from inflation play a central role in the regulation of the structure and function of lung tissue (Spieth et al., 2014; Tschumperlin et al., 2010). Furthermore, the relevance of mechanics is obvious in some pathological processes with the most obvious being tissue fibrosis, where the ECM is stiffened. More importantly, altered tissue mechanics are now also recognized as having an active role in driving human diseases, such as cancer (Mohammadi and Sahai, 2018; Papalazarou et al., 2020; Tschumperlin et al., 2018; van Helvert et al., 2018) and microbial pathogenesis (Charles-Orszag et al., 2016; Dufrêne and Persat, 2020).

Mechanotransduction is the process by which cells sense and translate external mechanical forces into internal biochemical signals (Wolfenson et al., 2019). Cells are subjected to a myriad of mechanical cues, such as forces generated by changes in ECM composition, compression, stretching, tension and shear stress; these are either sensed directly by mechanoreceptors or indirectly relayed via the cytoskeleton (Essig et al., 2001; Wang et al., 1993), and transduced to intracellular signaling pathways that regulate fundamental cellular processes, such as cell proliferation, differentiation (Pinheiro and Bellaϊche, 2018) and autophagy (Dupont and Codogno, 2016).

Cells are able to sense extracellular mechanical inputs in multiple ways. For instance, cellular adhesion complexes (both with the ECM and other cells) containing integrins (Kechagia et al., 2019; Sun et al., 2019) or cadherins (Yap et al., 2018) have been identified as mechanosensors. Other mechanosensors include proteoglycans localized in the glycocalyx (Tarbell and Ebong, 2008; Weinbaum et al., 2007), as well as mechanically activated [such as transient receptor potential (TRP) and PIEZO proteins] and mechanosensitive ion channels (Martinac and Poole, 2018; Murthy et al., 2017; O'Neil and Heller, 2005). Although the molecular basis of organelle-based mechanosensitivity remains mostly unclear, several plasma membrane-associated structures, such as the primary cilium (see Box 2) (Praetorius, 2015), microvilli (Du et al., 2004; Guo et al., 2000), caveolae (Sinha et al., 2011) and clathrin-coated pits (Baschieri et al., 2018), have been described as being involved in mechanosensing. Intracellular organelles, including the nucleus (Aureille et al., 2017; Kirby and Lammerding, 2018; Lomakin et al., 2020; Uhler and Shivashankar, 2017; Venturini et al., 2020), endoplasmic reticulum (ER) (Kim et al., 2015), mitochondria (Helle et al., 2017), the Golgi (Guet et al., 2014; Romani et al., 2019), as well as autophagosomes (see Box 1) are also known, or highly suggested, to sense mechanical forces. For example, shear or stretch directly deform organelles such as cilia or nuclei (Praetorius, 2015; Uhler and Shivashankar, 2017), whereas physical forces are capable of indirectly affecting the dynamics of specific organelles (e.g. mitochondria; Helle et al., 2017).

Once sensed by one of these mechanoreceptors, mechanical forces are able to activate ion influx, as is the case for PIEZO channels (Murthy et al., 2017), or intracellular cytoskeleton-associated signaling molecules, as described for integrins, such as the Rho-family GTPases that connect the cytoskeleton to the plasma membrane lipid bilayer, allowing the transmission of mechanical cues inside the cell (Kechagia et al., 2019; Sun et al., 2019). Downstream intracellular effectors, such as the transcriptional coactivators YAP1 (Yes-associated protein 1) and TAZ (WW domain-containing transcription regulator 1) have also been shown to be involved in the translation of external mechanical forces into internal biochemical signals (Dupont et al., 2011; Panciera et al., 2017; Pocaterra et al., 2020). Readers who are interested in a broader and in-depth discussion of the different types of mechanosensors and intracellular signaling pathways should consult the following relevant reviews (Hoffman et al., 2011; Kechagia et al., 2019; Lim et al., 2018; Panciera et al., 2017; Pinheiro and Bellaϊche, 2018).

Here, we summarize our knowledge of the role of autophagy in the integration of mechanical forces, highlighting the mechanosensors involved and the cellular signaling pathways controlling autophagy, with a discussion of the relevance of this novel aspect of autophagy in physiology and pathophysiology.

In contrast to tension (see below), compression squeezes and shortens cells in the direction of the applied force. The archetypes of tissues that constantly bear compressive forces are the bone, cartilage and teeth. Many physiological (e.g. tissue growth during development) or pathological situations (e.g. inflammation or cancer) are associated with an increase of compression. For example, tumor growth generates compression force that acts on adjacent non-tumorous cells, activating tumorigenic pathways (Fernández-Sánchez et al., 2015).

Compression-induced autophagy has been observed in Dictyostelium (King et al., 2011) and in mammalian cells, including the cervical cancer cell line HeLa (Das et al., 2019; Wang et al., 2019), chondrocytes (Xiang et al., 2019; Xu et al., 2014) and nucleus pulposus cells derived from intervertebral discs (He et al., 2021; Ma et al., 2013). Studies have also confirmed the stimulation of autophagy upon compression in several ex vivo experimental conditions, including bovine and human cartilage explants (Caramés et al., 2012), mouse spinal cords (Tanabe et al., 2011; Zhang et al., 2020), porcine eyes (Porter et al., 2014; Shim et al., 2021) and rat skeletal tibial muscles (Teng et al., 2011) (Fig. 1 and Table 1). Despite this knowledge, the exact role of autophagy induction upon compression remains poorly defined. So far, it has been shown that compression-dependent autophagy inhibits apoptosis in the spinal cord (Lee et al., 2009; Wang et al., 2012; Zhang et al., 2020). The mechanosensors involved and the cellular signaling pathways that control the regulation of compression-dependent autophagy are far from being well understood. The primary cilium has been recently proposed to participate in the stimulation of compression-induced autophagy in vitro in chondrocytes, where ERK and mTOR signaling are suggested to contribute to this stimulation (Xiang et al., 2019) (Fig. 1). However, in breast cancer cell lines subjected to pressure, autophagy induction is independent of the mTOR pathway (King et al., 2011). There is evidence indicating that compression-induced autophagy could also be supported by reactive oxygen species (ROS)-dependent signaling in nucleus pulposus cells (Ma et al., 2013) and lipid-raft-mediated p38 MAPK-dependent processes in HeLa cells (Das et al., 2019). Hypoxia-inducible factor 1α (HIF1α) has also recently been shown to upregulate autophagy in nucleus pulposus-derived stem cells under compression (He et al., 2021).

The brain is another organ exposed to compression, which occurs in traumatic brain injury (TBI) or subarachnoid hemorrhage (SAH). Interestingly, several laboratories have observed an accumulation of autophagic markers in experimental SAH samples (Lee et al., 2009; Wang et al., 2012) and TBI samples from rodents (Lai et al., 2008; Zhang et al., 2008), as well as in human samples (Clark et al., 2008). Autophagy induction appears to be neuroprotective in SAH by inhibiting apoptosis (Fang et al., 2018; Zhang and Liu, 2020). However, the signaling pathways and mechanosensors involved remain to be uncovered.

Accumulating evidence indicates that compression contributes to the development of other human diseases, including cancer. Indeed, uncontrolled proliferation of cancer cells generates compression that affects migration during metastasis (Cheng et al., 2009; Tse et al., 2012). Tumor growth in vivo has also been shown to exert mechanical compression forces on adjacent non-tumorous cells (Fernández-Sánchez et al., 2015). Since the autophagy pathway is dysregulated during tumorigenesis and compression can upregulate autophagy in cancer cell lines, as mentioned above (Das et al., 2019; King et al., 2011; Wang et al., 2019), it will be of interest to study the role of compression-induced autophagy in tumor biology. Finally, inflammation can be caused by many different factors, including heat, freezing, trauma, chemicals or infectious agents, and is generally associated with an increase in interstitial osmotic pressure (Grimble, 2003; Schwartz et al., 2008, 2009). Considering that autophagy acts mainly through mitophagy-mediated suppression of inflammatory cytokine production (Deretic and Levine, 2018; Levine et al., 2011; Matsuzawa-Ishimoto et al., 2018) in inflammatory diseases, studying the role of mechanical force-induced autophagy should increase our understanding of the pathogenesis of these diseases.

Shear stress is a frictional force caused by fluid flow that is parallel to, and acts over, the surface of cells; it is therefore directly linked to fluid viscosity, flow rate and geometry. Thus, by definition, every single cell facing fluid in the human body is constantly dealing with shear stress, in particular endothelial and endocardial cells within cardiovascular tissue, as well as other cells experiencing fluid flow from sources that include urinary, interstitial, lymphatic, nodal, lacunocanalicular and cerebral spinal fluids.

As summarized in Table 1, studies of endothelial cells have concluded that unidirectional or laminar fluid flow promotes autophagy. Importantly, the shear stress applied in these studies (above 10 dyn/cm², 1 Pa) corresponds to forces observed at atheroprotective sites, such as the descending thoracic aorta (Chiu and Chien, 2011; Gimbrone and García-Cardeña, 2013; Hahn and Schwartz, 2009). However, autophagy appears to be inhibited when oscillatory or low-magnitude fluid flow (below 4 dyn/cm², 0.4 Pa) is used (Table 1), conditions prevalent at atherosclerosis-prone sites, such as arterial bifurcations and the inner part of curvatures (Chiu and Chien, 2011; Gimbrone and García-Cardeña, 2013; Hahn and Schwartz, 2009). Analyses of ex vivo perfused animal vessels and in vivo animal models have confirmed the effect of fluid flow-induced mechanical force on the autophagic pathway (Ding et al., 2015; Guo et al., 2014; Li et al., 2015; Santovito et al., 2020; Vion et al., 2017; Yang et al., 2016). Importantly, regarding the effect of autophagy induced by high shear stress on atherosclerotic plaque development, it has been demonstrated that hypercholesterolemic mice bearing an endothelial-specific deletion of Atg5 (Apoe^−/−; Atg5^flox/flox;VE-cadherin-Cre) develop larger atherosclerotic lesions only in normally atheroprotected areas, when compared to littermate controls (Vion et al., 2017). Interestingly, the atheroprotective effect of shear-stress-induced autophagy is mediated by the inhibition of endothelial cell apoptosis and senescence, as well as the inhibition of inflammatory cytokine production (Vion et al., 2017) (Figs 1, 2). However, the exact autophagy-dependent molecular mechanisms driving this protective cell phenotype remain unclear. What we know so far is that under conditions of high shear stress, autophagy downregulates endothelial cell death by facilitating the nuclear shuttling and activation of the anti-apoptotic microRNA miR126-5p (Santovito et al., 2020).

It has been suggested that autophagy inhibits the activity of the transcriptional coactivators YAP1 and TAZ (referred to collectively as YAP/TAZ; Yuan et al., 2020), which are known to regulate inflammation. These data are in line with the fact that YAP/TAZ are active in atheroprone regions that are exposed to disturbed low blood flow and characterized by inefficient autophagy, compared to the atheroprotective regions, which are exposed to undisturbed high laminal blood flow with high levels of autophagy (Nakajima et al., 2017; Wang et al., 2016; Yuan et al., 2020). Shear-stress-dependent autophagy is also important for nitric oxide (NO) production, a crucial molecule that maintains vascular homeostasis in endothelial cells (Bharath et al., 2017, 2014). Indeed, in zones of low shear stress that are prone to develop atherosclerotic plaques, especially in the artery tree, the impairment of autophagic flux induces endothelial NO synthase (eNOS, also known as NOS3) uncoupling, resulting in the production of O₂⁻ instead of NO. Restoration of the autophagic flux favors the production of NO by eNOS (Zhang et al., 2018).

Nevertheless, the mechanosensors upstream of autophagy induction in endothelial cells are not well known. Several structures and molecules, including the apical glycocalyx, primary cilia, proteins such as glycoprotein CD31 (also known as PECAM-1) or ion channels, have been suggested to sense fluid shear stress in endothelial cells (Baeyens et al., 2016). However, neither CD31 nor the primary cilium appear to be implicated in autophagy induced by high shear stress (20 dyn/cm², 2 Pa) in human umbilical vein endothelial cells (HUVECs; Vion et al., 2017). Other key endothelial mechanosensors, including integrins (Tzima et al., 2001; Weinbaum et al., 2007) and specialized plasma membrane-associated organelles (such as caveolae; Yu et al., 2006) have been shown to play a role in the regulation of autophagy, but whether they are involved in fluid-flow-induced autophagy remains to be investigated. Regarding the cellular signaling pathways downstream of mechanosensors, there is evidence indicating that induction of autophagy by fluid flow depends on the redox potential, the NAD-dependent deacetylase SIRT1, the lectin-like oxidized low-density lipoprotein scavenger receptor LOX-1 (also known as OLR1; Ding et al., 2015; Liu et al., 2015; Yuan et al., 2020) and the AMP-activated protein kinase (AMPK)–mTOR pathway (Vion et al., 2017). Other key factors, including phosphoinositide 3-kinase (PI3K) (Wang et al., 2018), focal adhesion kinase (FAK) (Wang et al., 2018; Yan et al., 2019), Rho GTPases (Wang et al., 2018; Yan et al., 2019) and p38 MAPK (Das et al., 2018) have also been suggested to induce autophagy upon shear stress in cancer cell lines, such as the hepatocarcinoma cell line HepG2.

As mentioned above, kidney tubular cells are also subjected to shear stress, which is a crucial factor in renal physiological function (Raghavan and Weisz, 2016; Verschuren et al., 2020). For the past few years, we have been investigating the effect of mechanical forces on kidney epithelial cells and have demonstrated that physiologically relevant shear stress (1 dyn/cm², 0.1 Pa) induces autophagy in a primary cilia-dependent manner in mouse and human cell lines (Boukhalfa et al., 2020; Miceli et al., 2020; Orhon et al., 2016; Zemirli et al., 2019), as well as in mouse models (Orhon et al., 2016). Autophagy induction in this model is important for the regulation of cell size and stimulation of metabolic reprogramming. Here, mechanotransduction is based on the induction of a selective autophagic pathway, known as lipophagy, that is required for lipid droplet degradation (Singh et al., 2009). Lipophagy induction is accompanied by an upregulation of mitochondrial biogenesis driven by the PGC1α (also known as PPARGC1A) signaling pathway and a stimulation of ATP production, which is needed to support energy-consuming processes such as cell size reduction and renal glucose reabsorption (Miceli et al., 2020) (Fig. 2). This is in line with another study showing that cell metabolism is sensitive to the physical cell microenvironment (Romani et al., 2021).

In response to shear stress, autophagy is stimulated in kidney epithelial cells by the mTOR complex 1 (mTORC1)–AMPK pathway and by Ca²⁺ influx dependent on the primary cilium Ca²⁺ channel polycystin 2 (PC2, also known as PKD2) (Orhon et al., 2016). PC2-dependent autophagy is not involved in cell size regulation, in contrast to mTORC1–AMPK-dependent autophagy. The role of PC2-dependent autophagy remains to be determined (for a recent discussion see Peña-Oyarzun et al., 2020). The activation of mTORC1–AMPK depends on the upstream kinase LKB1 (also known as STK11) located in the axoneme (Boehlke et al., 2010; Orhon et al., 2016). More recently, folliculin (FLCN) has been demonstrated to act upstream of LKB1 in controlling AMPK activity, autophagy and cell size regulation (Zemirli et al., 2019).

The most intriguing question that remains to be answered is which ciliary mechanosensitive protein upstream of the FLCN–LKB1–AMPK–mTORC1 axis is involved in inducing autophagy. Along the same lines, studying the role of actin-rich microvilli structures, already known to trigger autophagy upon shear stress in intestinal epithelial cells (Kim et al., 2017), may also help to improve our understanding of mechanotransduction in kidney cells. More generally, it will be of interest to determine to what extent the observations of fluid-flow-induced autophagy in renal tubules help us to better decipher the early steps involved in acute and chronic renal diseases, in which autophagy and urinary fluid flow are known to be deregulated (Choi, 2020; Verschuren et al., 2020). Based on a number of in vivo studies pointing to a reno-protective role of autophagy in autophagy-related (ATG) protein-deficient mice subjected to chronic or acute kidney injury (Choi, 2020), we believe that the deregulation of autophagy induced by mechanical alterations could participate in driving the progression of renal diseases.

Finally, cancer cells can also experience shear stress. Among the first studies to show that shear stress can modulate cell fate by inducing autophagy were in vitro analyses of osteoblast and cancer cell lines subjected to flow by using fluidic pumps (Kim and Yoo, 2013; Lien et al., 2013). For instance, at the primary tumor site, interstitial flow can promote the distribution of tumor-derived cells, whereas circulating tumor cells are also subjected to and exploit the shear stress generated by body fluids (blood, lymph and interstitial fluid) during metastasis (Follain et al., 2020; Swartz and Lund, 2012). Interestingly, shear stress induces autophagy in cancer cell lines (Das et al., 2018; Lien et al., 2013; Wang et al., 2019, 2018; Yan et al., 2019) (see Table 1), which leads by as-yet-unknown means to the stimulation of cell migration and cell invasion at early time points of shear (Wang et al., 2018; Yan et al., 2019) and the induction of cell death at later time points (Lien et al., 2013). Thus, studying how mechanical forces influence autophagy in cancer cells may help to improve our understanding of the progression of tumorigenesis. Overall, even if the exact molecular mechanisms upstream or downstream of autophagic induction upon shear are still not fully understood, accumulating evidence clearly shows that shear stress-induced autophagy plays an important role in controlling a variety of cell responses from the regulation of cell size and metabolism to inflammation and cell death.

Stretching and tension are related to the force needed to extend a tissue or a cell uniaxially or biaxially. Stretching usually refers to a passive deformation due to the application of external forces, whereas tension is an active force that depends on the elastic properties of cells. Tissues such as muscles are constantly subjected to tension. While the heart muscle is the best characterized example, many other tissues, including bone, tendons, and ligaments, are also exposed to stretching.

Two independent laboratories have provided the first evidence that mechanical forces induce autophagy in vivo by demonstrating that physical exercise is a potent inducer of autophagy in skeletal and cardiac muscles when mice are subjected to training on a treadmill (Grumati et al., 2011; He et al., 2012a). Whether the effect of tension directly modulates autophagy in this setting is still to be investigated. Autophagy induction in muscles during physical activity is important for muscle homeostasis and represents an adaptive response to exercise that ensures effective mitochondrial quality control (Lo Verso et al., 2014). The health benefits of exercise, including life span extension and protection against cardiovascular diseases and diabetes, are clear, but whether these benefits are due in part to autophagy activation is an interesting open question. Physical exercise has been shown to induce autophagy in peripheral tissues (liver, pancreas and adipose tissue) and in the brain (He et al., 2012b). By using stretching devices, stretch-induced autophagy has also been observed in vitro in human patellar tendon fibroblasts isolated from tendon pieces (Chen et al., 2015), trabecular meshwork cells derived from corneal rims (Porter et al., 2014; Shim et al., 2021, 2020) and smooth muscle cells (Ulbricht et al., 2013) (Fig. 1 and Table 1).

Cardiac cells are also subjected to tension. Several pathophysiological conditions, including myocardial infarction, systemic hypertension and pulmonary embolism, lead to an increase in cardiac workload and mechanical forces that are usually associated with pathological cardiac hypertrophy (Nakamura and Sadoshima, 2018). In vitro and in vivo studies have analyzed the effects of mechanical forces on the autophagic pathway in cardiac cells. First, mechanical tension-induced autophagy has been observed in vitro in cultured cardiomyocytes of neonatal rats (Lin et al., 2015, 2014; Yu et al., 2019). Second, autophagy has been shown to be upregulated in the heart during myocardial infarction (Kanamori et al., 2011; Matsui et al., 2007; Yan et al., 2005) and pressure overload induced by transverse aortic constriction (TAC) in a mouse model (Nakai et al., 2007; Zhu et al., 2007). The role of autophagy in this setting appears to be quite complex, because its induction can be protective or detrimental depending on the context, level and duration. Indeed, autophagy protects cardiomyocytes during ischemia (Kanamori et al., 2011; Matsui et al., 2007; Yan et al., 2005), whereas inhibition of autophagy has been shown to improve cardiac function after reperfusion in an ischemia–reperfusion mouse model (Matsui et al., 2007). In parallel, activation of autophagy has been shown to prevent TAC-induced ventricular hypertrophy and improve ventricular function (Ha et al., 2005; McMullen et al., 2004). However, excessive autophagy induction in failing hearts could also lead to cardiomyocyte cell death (Gao et al., 2018; Knaapen et al., 2001; Kostin et al., 2003; Takemura et al., 2006).

Currently, the mechanosensors and cellular signaling pathways that control the regulation of stretch- or tension-induced autophagy are far from being well understood. As discussed above, the primary cilium has been recently shown to regulate stretch-induced autophagy in trabecular meshwork cells (Shim et al., 2021) via AKT-dependent and SMAD2 and SMAD3 (SMAD2/3)-dependent signaling (Fig. 1). However, the role of the primary cilium in sensing stretching and/or tension remains unclear. Studying the role of stretch-sensitive or stretch-activated channels, such as TRP channels including TRPC1 (Bai et al., 2008), that are localized in the cilium could help to better understand this mechanotransduction pathway. In addition to the primary cilium, the muscle-stretch sensor LIM protein (MLP; also known as cysteine and glycine-rich protein 3, CSRP3), which was discovered in the sarcomeric Z-disc and shown to be defective in a subset of cardiomyopathies (Knöll et al., 2002), also regulates autophagy (Rashid et al., 2015). Other mechanosensing proteins that are also able to sense stretching are the stretch-activated ion channels (SACs) (Nilius and Honoré, 2012), including the TRP channel polycystin 1 (PC1, also known as PKD1 or TRPP1), which is required for cardiac hypertrophy (Pedrozo et al., 2015). It will be interesting to study the role of SACs in stretch- or tension-induced autophagy. Furthermore, several molecular chaperones have been found to be essential for tension-induced autophagy in muscle cells, including Hsc70 (heat shock cognate 70, also known as HSPA8) and HspB8 (heat shock protein family B member 8), as well as the co-chaperones BAG3 (BAG co-chaperone 3) and synaptopodin-2 (Arndt et al., 2010; Ulbricht et al., 2013). This particular type of autophagy is referred to as chaperone-assisted autophagy (CASA) (Kettern et al., 2010). In addition, exercise-induced autophagy is controlled by a pro-oxidant complex consisting of regulated in development and DNA damage responses 1 (REDD1, also known as DDIT4) and thioredoxin-interacting protein (TXNIP), which regulates the activity of ATG4B (Qiao et al., 2015). Thus, based on these studies, it would be interesting to determine whether there is a common molecular signature of stretch- or tension-induced autophagy.

Although the ECM is not a physical force per se, its meshwork of collagens, glycoproteins and proteoglycans has viscoelastic properties that generate mechanical cues, such as resistance of tension-mediated forces. In tissues, cells do not exist in isolation but are surrounded by other cells and by the ECM produced by the cells themselves. The biochemical composition and biophysical properties of the ECM vary according to the organ type, physiological situation (such as tissue repair) or pathological state (such as fibrosis).

Evidence that ECM stiffness interferes with the autophagic pathway comes from different studies showing that several ECM constituents can modulate autophagy. For example, collagen VI or proteoglycans, including decorin and biglycan, have been identified as autophagic inducers (Castagnaro et al., 2018; Gubbiotti and Iozzo, 2015; Tuloup-Minguez et al., 2011), while the glycoprotein laminin α2 (LAMA2) is known to inhibit the autophagy pathway (Carmignac et al., 2011). ECM-driven autophagy plays a central role in several biological processes, such as angiogenesis, tissue repair and fibrosis (Chen et al., 2020a; Madhu et al., 2020). Conversely, it is worth mentioning that autophagy is able to regulate the turnover of ECM receptors, including integrins and focal adhesion-associated proteins, such as paxillin and vinculin, which are important to support cell migration (Kenific et al., 2016a; Sharifi et al., 2016; Tuloup-Minguez et al., 2011). Based on this, the effect of ECM stiffness itself on autophagy has been investigated, and it has been shown that the autophagic pathway is compromised in non-tumorigenic epithelial cells (MCF10A) cultured in a soft ECM microenvironment (Pavel et al., 2018). Interestingly, this ECM-dependent modulation of autophagy is important to negatively control proliferation and induce cell death. Subsequent studies have confirmed that soft ECM impairs autophagy in both non-malignant and cancer cells (Li et al., 2020; Totaro et al., 2019). However, in contrast to the previously observed role in autophagy initiation (Pavel et al., 2018), the specific effect here is an induction of autophagic flux (Totaro et al., 2019). The reason for this discrepancy remains unknown, but it could be due to differences in the quantitative composition of the ECM used in the in vitro studies. More recently, it has also been reported that changing the substrate stiffness significantly and differentially impacts the autophagy pathway in vascular endothelial and smooth muscle cells (Hu et al., 2021).

Several key factors are known to be involved in the signaling pathways that lead to the induction of ECM-dependent autophagy. First, ECM stiffness interferes with autophagy through modulation of cytoskeleton dynamics and actin-associated regulators, such as the Rho small GTPase and its downstream effector ROCK (Rho-associated protein kinase), which has been previously shown to regulate the endomembrane remodeling process required for the formation of double-membrane autophagic structures (Li et al., 2020). In addition, the transcriptional coactivators YAP/TAZ have been shown to control ECM-dependent autophagy (Pavel et al., 2018; Totaro et al., 2019), and YAP/TAZ directly modulate autophagy by controlling the transcription of actomyosin genes (MLC2, MYH10, MYH14) (Pavel et al., 2018). In addition, YAP/TAZ control the expression of factors that are essential for the autophagic pathway, such as the RAB7 GTPase-activating protein Armus (also known as TBC1D2A; Totaro et al., 2019). However, despite these insights, direct evidence for an involvement of the known mechanosensors that control autophagy in soft or stiff microenvironments is lacking. More generally, other important questions regarding ECM-dependent mechanics and autophagy remain unanswered, such as exactly how autophagy exerts an effect in the differentiation of stem cells that are exposed to different types of ECM. Although ECM-induced autophagy has been suggested to play a role in the acquisition of stemness properties (Li et al., 2020; Totaro et al., 2019), it remains unclear how autophagy impacts cell fate.

The role of autophagy in the integration of mechanical constraints is an emerging field of research, but we are only at the tip of the iceberg, and future studies will be necessary to elucidate this aspect of autophagy in tissue homeostasis and disease. Although it is clear that autophagy is an effector of mechanical forces, it is sometimes difficult to assess whether the accumulation of autophagic markers reflects a true induction of autophagy or a blockade of autophagic degradation. Nevertheless, accumulating in vitro and in vivo evidence indicates that mechanical force-induced autophagy is important in normal physiology, and that its dysregulation is associated with various diseases, despite the fact that this research field is still in its infancy, as the outcomes of mechanical force-dependent autophagy remain challenging to analyze.

Another complication is that, in vivo, most cells are not subjected to a single type of force but to a combination of forces. For example, although fibroblasts in tendons and ligaments are mainly exposed to stretching forces in vivo, endothelial cells lining the vessel surface are subjected to a combination of mechanical forces, including shear, hydrostatic pressure and cycling stretch. Various two-dimensional experimental devices have been used to study changes in the autophagy pathway in response to mechanical forces in vitro (Roccio et al., 2021), but they neither take into account the complexity of mechanical forces, nor reflect what occurs in vivo. Physiologically relevant three-dimensional systems, including organs-on-chips, allow cells to reside in a more in vivo-like environment (Bhatia and Ingber, 2014; Ross et al., 2021). Exploiting this technology might help to better understand the role of mechanical force-induced autophagy in the future, with the aim of promoting tissue repair and regeneration.

We thank Nicolas Kuperwasser for editing and proofreading.