Regulation of cellular senescence by extracellular matrix during chronic fibrotic diseases

The extracellular matrix (ECM) is a complex network of macromolecules surrounding cells, traditionally recognised for providing structural support and tissue stability [1]. Our understanding of the diverse roles of the ECM has greatly advanced in the last three decades, particularly in the provision of essential biochemical and biomechanical cues required to direct tissue morphogenesis during development, homoeostasis and injury. Mesenchymal cell types, most notably the resident fibroblasts, have the role of maintaining the ECM within tissues throughout an organism’s lifespan. Therefore, mesenchymal cells need to respond to localised cues to deposit, maintain and remodel the appropriate ECM to meet the functional needs of the tissue. To facilitate the ECM in its role as a biologically active structure, cells interact with individual matrix proteins through surface receptors such as cell surface proteoglycans (PGs), discoidin domain receptors and specific integrins resulting in outside-in signalling which dictates cell function, fate and phenotype [2,3]. Cells are able to communicate through the ECM by generating mechanical forces or by altering mechanical properties of the ECM such as the stiffness [4].

The ECM is crucial for maintaining normal tissue homoeostasis and many pathological conditions arise from dysregulated ECM remodelling as a result of ageing or as an attempt to preserve or restore organ function. Remodelling of the ECM during fibrotic disease, which changes the physical state of the tissue, may contribute to altered cellular responses to mechanical forces such as pressure, stretch and shear force [5,6]. Ageing is characterised by functional decline of tissue over time, leading to progressive deterioration that eventually leads to tissue dysfunction. Many diseases have age as a risk factor such as cardiovascular disease, diabetes, cancer, dementia, glaucoma, chronic obstructive pulmonary disease (COPD) and idiopathic pulmonary fibrosis (IPF) [7–14]. The changes that occur during ageing and/or disease influence the composition, topography and biomechanics of the ECM, thereby contributing to abnormal cellular activation and dysregulated behaviour. For example, aberrant ECM deposition and increased stiffness are observed in fibrotic diseases and cancer, while excessive ECM degradation is linked to osteoarthritis (OA) and COPD [15–17]. The ECM changes in fibrotic lung disease are recognised as a major driver of aberrant cellular responses [18]. In addition, a population of resident fibroblasts in aged tissue show growth arrest and resistance to apoptotic cues, which are indicative of cellular senescence, a hallmark of ageing [19,20]. Cellular senescence is characterised as irreversible cell-cycle arrest in response to various sources of stress such as DNA damage or reactive oxygen species (ROS). During physiological homoeostasis, cellular senescence benefits embryogenesis, tissue repair and defence against tumorigenesis. In contrast, it was recently postulated that senescence contributes to the pathogenesis of chronic fibrotic diseases in heart, kidney, liver and lung [21–25]. Here, senescent cells negatively impact tissue regeneration while creating a pro-inflammatory environment, as part of a senescence-associated secretory phenotype (SASP) that favours disease progression. Whether the ECM changes in ageing and fibrotic disease are also drivers of the senescent phenotype associated with these conditions is as yet unexplored.

During remodelling and injury, breakdown products of the ECM are released which often act as damage-associated molecular patterns (DAMPs) that activate pattern recognition receptors (PRRs) on cells of the innate immune system. Interestingly, several of these ECM DAMPs have been found to be significantly increased in multiple chronic fibrotic diseases [26–28]. Activation of PRRs by DAMPs induces nuclear factor κB (NF-κB)-mediated pro-inflammatory cytokine release, with a similar cytokine release profile to that associated with the SASP. Moreover, NF-κB is known as a master regulator of the SASP in fibroblasts, which suggests a role of DAMPs in regulating cellular senescence.

In this review, we explore and summarise the current knowledge around how aberrant ECM potentially contributes to the senescent phenotype in chronic fibrotic diseases. A broad overview will be provided about the roles of ECM and senescence during physiological homoeostasis and in chronic fibrotic diseases. An overview of ECM changes in senescence during ageing has previously been described in detail elsewhere [29], and hence will not be extensively covered herein. In this review, the possible mechanisms through which the ECM drives pathological senescence will be outlined, including interactions between cells and the ECM, disruption to mechano-transduction upon ECM remodelling, alterations in integrin–adhesion complexes and the release of ECM DAMPs. Lastly, we will explore the possibility for interventions in the ECM-senescence regulatory pathways for therapeutic potential in chronic fibrotic diseases. While this review will portray several diseases, IPF will be focused upon as an exemplar for chronic fibrotic disease.

Cellular senescence is a hallmark of ageing and involves marked changes in cell morphology, phenotype, metabolism and fate [30]. While irreversible cell cycle arrest is the defining feature of senescent cells, they are also resistant to apoptosis and are highly secretory; driving a range of different functions through the SASP [31]. Resistance to apoptosis is mediated by up-regulation of B-cell lymphoma 2 (Bcl-2) and epigenetic suppression of Bcl-2-associated X protein (BAX) [32]. The paracrine activities of senescent cells are crucial for many facets of normal development and tissue function as well as pathologies including fibrosis and tumorigenesis. Senescence is primarily an outcome of genomic damage comprising DNA double-stranded breaks (DSBs). As part of a subsequent DNA damage response (DDR), ataxia-telangiectasia mutated kinase (ATM) phosphorylates p53, leading to the induction of p53-p21 and/or p16-pRB pathways that mediate cell cycle arrest through cyclin-dependent kinase (CDK) inhibition [33]. The transcription factor NF-κB is a pivotal mediator of the SASP and activated by ATM-dependent phosphorylation of NF-κB essential modifier (NEMO) [34].

Replicative senescence (RS) is a consequence of mitotic ageing and arises from the shortening of telomeres; the repetitive DNA sequences at the end of chromosomes that protect DNA from damage [35]. In a process in somatic cells that was originally described by Hayflick and Moorhead in diploid fibroblasts in vitro, telomeres shorten with each successive cell division until DNA damage and senescence eventuate [36,37]. Stress-induced premature senescence (SIPS) is another form of senescence in which insults such as oxidative stress, radiation, mutagens and cytokines elicit directly or indirectly genomic damage and the subsequent DDR [38–41]. While many of these stressors are of external sources, ROS derived from dysfunctional mitochondria can induce and/or reinforce the DDR in SIPS in a cell autonomous manner [42,43]. ROS and SASP mediators released by senescent cells can also evoke ‘secondary’ senescence in surrounding naïve (non-senescent) cells; a paracrine process termed as ‘senescence-induced senescence’ [44,45]. Oncogenic-induced senescence (OIS), a form of SIPS comprises a complex senescence program involving the induction of oncogenes such as the Ras/Raf family [46]. OIS restricts the excessive mitotic signalling of DNA-damaged cells that may otherwise undergo malignant transformation to provide a tumour-suppressing function [47]. Figure 1 summarises the characteristics of cellular senescence. RS is also an anti-tumour mechanism that prevents cell immortalisation. The anti-cancer properties of senescence can be viewed as a ‘trade off’ with its role in age-related stem cell loss/tissue degeneration [47].

Cellular senescence is an essential, yet transitory feature of normal tissue homoeostasis. In this role, the senescent cell has a relatively short half-life, with its removal by immune cells occurring once its task is completed. This limits overproduction of ECM and contributes to active remodelling to maintain homoeostasis. In embryogenesis, senescence in the apical ectodermal ridge provides an instructive role for tissue development and patterning by the prevention of cell proliferation. This effect is mediated via the release of secreted factors that regulate processes such as neovascularisation and cell migration [48]. Under normal conditions, SASP functionality in tissue homoeostasis changes with time, culminating in the provision of cues that direct senescent cell removal. For example, in skin wounds, senescent cells accumulate and promote wound healing and resolution firstly through secretion of platelet-derived growth factor (PDGF) AA (PDGF-AA) and then subsequently drive their own immune-mediated clearance; suggesting a temporal switch from orchestration of wound repair to inflammatory recruitment of immune cells [49]. Cytokines and other secreted factors from senescent cells are involved in the recruitment of immune cells such as neutrophils, macrophages, natural killer cells and CD4⁺ T cells that clear senescent cells. The expression of immunogenic stimulatory ligands by senescent cells that control their removal is regulated by autocrine/paracrine activities of the SASP.

The numbers of senescent cells in tissue increases with age, contributing to age-related tissue degeneration and disease. Deficiencies in immune surveillance and clearance facilitated by an immune system of deteriorating function and capacity (i.e., a phenomenon termed as ‘immuno-senescence’) contribute to this accumulation [50]. Senescent cells from aged tissues also express molecules such as human leukocyte antigen E (HLA-E) that inhibit the function of infiltrating immune cells; providing another means by which senescent cells can evade immune-driven apoptosis [51]. Ageing tissues exhibit chronic senescence mediated by SASP that can be increasingly harmful with time, leading to a loss of regenerative potential and subsequent disease [52]. Evidence for a causal role of senescence in ageing and disease comes from murine studies that show that the ablation of senescent cells in vivo prolongs lifespan, has a restorative effect on aged tissues and induces protection against insults associated with disease [53–55]. It was shown that glycosaminoglycans (GAGs) content significantly increases with clearance of senescent cells in the knee joint of mice [55]. Accelerated senescence in specific cell types is a feature of certain respiratory and vascular diseases, as well as in OA and in cancer. For example, senescent epithelial cells and fibroblasts accumulate in the lungs of IPF and COPD patients [56,57]. Senescent epithelial cells are proposed to be the crux of the aberrant wound healing response observed in these lung diseases; possibly by hindering re-epithelialisation and/or by the activation of resident fibroblasts [58–62]. A failure to eliminate senescent fibroblasts by apoptosis or immune cell clearance may also impede wound resolution and contribute to disease progression in fibrotic diseases [58,63,64]. In OA, senescent chondrocytes with a pro-inflammatory SASP accumulate and contribute to the damage of articular cartilage tissue and joints [55]. Senescent endothelial cells at atherosclerotic lesions are a prominent feature of vascular disease, potentially contributing to aberrations in barrier function, platelet aggregation and fibrinolysis. While senescence limits cancer initiation in a cell autonomous manner, it can also promote cancer in ageing tissues non-autonomously via the SASP by altering the microenvironment to allow for cancer progression [7,65]. The remodelling of ECM plays a critical role in facilitating tumour cell migration and spread via metastasis to other tissues. A causal role of senescent cells in cancer is supported by studies that show senescent fibroblasts stimulate human tumour cell proliferation when co-injected into immunocompromised mice [65,66]. The deleterious contribution of senescence in age-related pathology is an example of antagonistic pleiotropy; a process that has evolved to be beneficial in youth and increases reproductive fitness (i.e., by the prevention of malignancies); but is detrimental in older, post-reproductive age [52].

The composition of ECM comprises well over 300 proteins including fibrous proteins (collagens and elastins), PGs, GAGs and glycoproteins such as fibronectin [67]. Based on the location and function, ECM can be separated into two main types: interstitial connective tissue matrix and basement membrane. Interstitial ECM is mainly composed of collagen I, elastin and fibronectin to give support for surrounding cells, and to provide structural scaffolding for tissue. Interstitial ECM is deposited, maintained and remodelled by mesenchymal cells. The basement membrane is a specialised form of compact ECM that separates the epithelium or endothelium from the surrounding stroma. The basement membrane consists mainly of collagen IV, laminins, heparin sulphate PGs (HSPs) and proteins such as entactin and nidogen that are synthesised by epithelial cells and endothelial cells [1].

In addition to giving support to cells, the ECM provides important mechanical and biomechanical cues that direct tissue morphogenesis during development, homoeostasis and injury. More evidence is accumulating that ECM plays an active role in pathological conditions such as cancer, OA and fibrosis across multiple organs which emphasises the importance of understanding the developmental origins of ECM during tissue morphogenesis, homoeostasis and remodelling [17,68–70]. During development, many organs such as lung, kidney, nervous system and several internal glands undergo several changes such as lengthening, narrowing, bending, folding and branching which is guided by the ECM by a process called branching morphogenesis [71]. Formation of epithelial clefts and buds involves invasion of surrounding embryonic ECM, which is subjected to continuous change and distribution over time. Many different factors have been identified that are involved in branching, such as bone morphogenetic factors (BMPs), sonic hedgehog (Shh), fibroblast growth factors (FGFs), wingless and int-1 (Wnt) and retinoic acid [72]. Not only are these factors involved in development, they also play an important role during injury, remodelling and disease in adult tissue.

Mechanical forces play an important role in tissue morphogenesis and an essential role in generating cytoskeletal prestress contributing to cell, tissue and organ shape stability [73]. Cells are able to interact and respond to changes in the local microenvironment by contracting and pulling on the ECM or neighbouring cells. The contractile forces generate signals that propagate rapidly within the cell showing an advantage in activating signalling pathways compared with chemical stimuli [74]. Contractile forces within a cell are generated by the interaction of activated non-muscle myosin II and actin creating the actomyosin machinery regulating cellular shape and thus cellular function. Cellular contractility is controlled by the Rho-associated coiled-coil containing kinase (ROCK). ROCK consists of two subtypes, ROCK 1 and 2, which both have a critical function in regulating the actin cytoskeleton assembly [75]. ROCK may directly phosphorylate myosin light chain 2 (MLC2) or indirectly regulates MLC by phosphorylating the myosin-binding subunit of myosin phosphate (MYPT1) which is necessary for its interaction with actin [76]. In the embryonic lung of mice, inhibition of ROCK reduced branching and resulted in loss of basement membrane thickness and cell proliferation. Interestingly, ROCK inhibition in ureteric bud formation reported promotion of branching [77]. Another study reported that the lack of both ROCK1 and 2 protein using conditional knockout mice lead to a flattened and bigger morphology and loss of proliferation suggesting the potential influence on senescence [76]. It was shown that ROCK inhibition in vitro induced cellular proliferation and alleviation of senescence-associated cell-cycle arrest [78]. These studies suggest the cell-type specific role of ROCK in the cellular response to the environment, especially during pathological conditions in fibrosis.

ECM homoeostasis is a complicated balance of synthesis, modification and degradation. The importance of this intricate balance has been demonstrated in various diseases that originate from alterations in ECM [79]. Specific proteins which constitute the ECM are well conserved across species but show dramatic changes between individual tissue and organs in the same species. This suggests that the function of cells, tissues and organs is directly related to the mechanical and biomechanical cues of the surrounding ECM [80]. Several studies evaluating cardiac, renal and lung fibrotic disease have demonstrated that ageing, compositional and structural changes to the ECM can greatly influence cell phenotype, and function contributing to disease progression [25,68,81]. The intricate balance is maintained by cells residing in the ECM that produce various key enzymes such as metalloproteinases (MMPs) and its inhibitor endogenous tissue inhibitor of metalloproteinases (TIMPs), a disintegrin and metalloproteinase (ADAM), and ADAMs with thrombospondin motifs (ADAMTS) [67,82]. While the basement membrane is maintained by epithelial or endothelial cells, fibroblasts can contribute to basal lamina by producing nidogen while the interstitial matrix in the lung, kidney or heart is maintained by resident fibroblasts and hepatic stellate cells in the liver [83,84]. Activity of proteolytic enzymes is generally low during homoeostasis but increases during specific developmental events, injury and disease. The expression and activity of proteolytic enzymes is tightly regulated and is cell type- and tissue-specific [82]. During development, when ECM remodelling is necessary for tissue growth and alignment, there is a greater proteolytic profile compared with adult tissue which shows a higher anti-proteolytic profile. In diseases such as cancer, fibrosis and cardiovascular disease the intricate balance of homoeostasis is lost leading to an altered ECM and studies have shown that this is associated somewhat paradoxically with increased MMP activity. In cardiac remodelling, high levels of MMP1 have been linked to decreased collagen expression which contributes to loss of contractility and ultimately cardiomyopathy [85,86]. In IPF, increased MMP expression and activity is an important feature of the disease which has been identified as part of the SASP in fibrosis [25].

Fibrotic diseases are characterised by deposition of excessive ECM after injury resulting in organ dysfunction. While little is known about the exact mechanisms that result in excessive ECM deposition and accumulation in fibrotic disease, there is increasing evidence that cellular senescence is implicated in fibrosis [22,23,69]. One major under-recognised element in fibrosis is the contribution of the ECM itself. It is known that the altered composition and increased cross-linking leads to an altered topography and stiffness. These changes alter cellular behaviour and might potentially drive cells to become senescent contributing to an environment favouring disease progression. Increased ECM stiffness is a feature of fibrosis and is thought to result from the quantity of ECM deposition and the degree of its cross-linking. The mechanical properties of areas of wounded tissue significantly change in rat liver fibrosis. It was suggested that this was the result of increased cross-linking of the ECM fibres, rather than increased ECM deposition after the initial insult [87]. These results suggest that cross-linking of ECM has a greater impact on stiffness and thus altered mechanical properties than the quantity of deposited ECM alone. In normal homoeostasis, cross-linking of collagen I and III provides stability which leads to the generation of strong fibrils providing tensile strength and elastic properties. Cross-linking of ECM can happen spontaneously by glycation which is a slow process or by further processing via post-translational modifications by enzymes such as lysyl oxidases (LOs) and transglutaminases (TGs) which will be described in more detail in the next section and its potential regulation of cellular senescence.

The LO family of enzymes consist of lysyl oxidase (LOX) and four additional LOX-like proteins (LOXL1–4) which are copper-dependent amine oxidases responsible for covalent cross-linking of collagen and elastin. The process of ECM cross-linking is tightly regulated in homoeostasis with each oxidase being differently expressed across organs. LOX, LOXL-1 and LOXL-2 are expressed in nearly all tissues while LOXL-3 is not expressed in cardiac and skin tissue. LOXL-4 is only expressed in lung, kidney, pancreas and adipose tissue [88]. LOX family proteins are involved in several biological processes including tumour suppression, cell motility, cell transduction and cell adhesion. Dysregulation of LOX activity has been linked to pathological conditions such as cancer and fibrosis in lungs and liver [89–93]. TG is an enzyme responsible for cross-linking proteins such as collagen and fibronectin between glutamine and lysine residues or a primary amino group in a calcium-dependent manner. There are eight isoforms of TG, factor XIIIa and TG isoform 1–7 which are expressed throughout the whole body. The post-translational modification mediated by TG is essential for several processes including signal transduction, cell adhesion, skin barrier formation and the ECM assembly. It has been demonstrated that TGs are overexpressed in lungs of IPF patients and inhibition of TGs results in reduced inflammation and pulmonary fibrosis in bleomycin-treated mice [94]. In a recent study it was shown that TG isoform 1 was increased in both epithelium and interstitial space while TG isoform 2 was only increased in the extracellular space in the fibrotic kidney [95]. Inhibition of TGs reduced and preserved function in chronic kidney disease demonstrating the importance of cross-linking in pathological conditions [96].

In fibrotic disease, the increased cross-linking has not only been associated with increased stiffness but also in higher resistance to proteolytic degradation as the conformational changes lead to less accessible epitopes for MMPs [97,98]. The stiffness of the ECM directly influences the behaviour and function of cells including increased fibroblast proliferation, migration and contraction [17]. Furthermore, a stiffer ECM leads to an increase in latent transforming growth factor-β (TGF-β) activation which reinforces fibrosis and it has been linked to cellular senescence [99–101]. TGF-β consists of three different isoforms (TGF-β1–3) which are produced in a precursor form bound to the latency associated peptide which are attached to the latent TGF-β binding protein (LTBP). This keeps TGF-β in an inactive form until it is activated/released from the matrix by proteases such as MMP2, MMP9 and thrombospondin. Non-protease release of the active form is facilitated by fibroblasts pulling on the matrix in response to increased ECM stiffness. Upon activation, TGF-β binds the TGF-β receptor 2 (TGFR2) and subsequently recruits TGFR1 to active downstream signalling via activated mother against decapentaplegic proteins (Smads). A role for TGF-β in senescence has already been suggested by Frippiat et al. in 2001 while Acosta et al. demonstrated the importance of TGF-β in paracrine induced-senescence of fibroblasts [44,102]. More recently it was described that TGF-β contributes to the induction of senescence or the acceleration of transformation into senescence in various cell types [103,104]. The senescence promoting effect of TGF-β could be explained by down-regulation of human telomerase reverse transcriptase (hTERT) and proliferation factor c-Myc. Additionally, there is up-regulation of CDKs p15, p21, p27 and p53 and ROS production which are characterised as part of the senescent phenotype [101]. In a recent study, it was shown that signal transducer and activation of transcription 3 (STAT3) was a downstream target of TGF-β in a SMAD-independent pathway suggesting an important role of STAT3 in liver fibrosis [105]. Indeed, not only has STAT3 been linked to fibrosis in the liver but it is also an important regulator in a model of oxidative stress-induced senescence in primary human lung fibroblasts [41]. Recently, it was suggested that STAT3 potentially is involved in epithelial cell senescence [106]. In addition to regulation by TGF-β, it was concluded that matrix stiffness has a substantial impact on the intrinsic and extrinsic activation of the STAT3 pathway in hepatocellular carcinoma cells [107]. Moreover, activation of TGF-β by a stiffer matrix leads to up-regulation of LOX gene expression and enzyme activity [108]. Several studies have linked the increase in LOX expression by TGF-β to decreased expression of the microRNA (miR) 29 family [109,110]. The miR-29 family has been described to have strong antifibrotic effects in organs such as heart, kidney and lung. It is thought that miR-29 can regulate and modulate apoptosis, cell proliferation, differentiation, migration and the immune response [109,111,112]. Decreased levels of miR-29 have been shown to play a critical, common role in myocardial infarction, systemic sclerosis and patients with advanced fibrosis [112]. Furthermore, miR-29 can increase growth factor and inflammatory cytokine expression including PDGF, tumour necrosis factor α (TNF-α), interleukin (IL) 4 and master regulator of the SASP transcription factor NF-κB. The interplay between these factors and miR-29 maximises up-regulation of collagen and ECM related genes and induction of senescence.

Pathological increase in ECM deposition and remodelling contributes to an increase in ECM stiffness causing a mechanical gradient between fibrotic and normal tissue. It is known that these changes in stiffness favour proliferation and migration in different cell types. However, currently little is known how the pathological change in composition specifically contributes to the senescent phenotype. Several core and matricellular ECM proteins have been reported to be differentially expressed and these will now be highlighted and discussed as to how they potentially could contribute to cellular senescence.

Collagens are the most common ECM protein and are composed of three procollagen chains configured in a triple helix. In connective tissue collagen I and III are the most abundant types of collagens providing structure and tensile strength. Increased collagen III deposition is associated with wound healing, in renal fibrosis both type I and type III are increased while in IPF collagen type I is found in end-stage fibrosis [113,114]. Several studies have reported that collagen expression is differentially expressed in senescent cells. Schafer et al. reported that human IPF lungs express more collagen type I α-1 (COL1A1) while collagen expression in senescent HSCs was reduced [25,115]. Other studies have shown an up-regulation of collagen gene expression in fibroblasts in RS and Werner Syndrome fibroblasts implying a possible role for COL1A1 in cell-cycle arrest [116]. These studies suggest bi-directional cross-talk between senescent cells and collagen type I. Murine vascular smooth muscle cells expressing a collagenase-resistant collagen type I were more sensitive to stress-induced senescence compared with wildtype (WT) cells suggesting a role for increased collagen deposition in premature senescence. In a study investigating glycated collagen, it was shown that early-passage human umbilical vein endothelial cells (HUVECs) grown on glycated collagen-coated dishes expressed hallmarks of premature cell senescence including p14, p53 and increased senescence-associated β-galactosidase (SA-β-Gal) staining compared with non-coated dishes [117]. In cardiac fibroblasts, glycated collagen is associated with integrin α11 expression which modulates activation of TGF-β2 signalling pathway through Smad2/3 which is important for myofibroblast activation and fibrosis [118]. There is accumulating evidence that collagen in different cell types might contribute to premature senescence, however, the exact mechanisms remain unclear. Taken together, it appears that increased deposition of collagen and a stiffer ECM might contribute to the senescent phenotype in fibrosis.

Elastin is the main component of elastic fibres which provides compliance and elasticity to a range of tissues. In addition to providing mechanical integrity it has critical regulatory functions of cell behaviour. It is produced by a wide variety of cells including fibroblasts, epithelial cells and smooth muscle cells. Elastin is secreted as tropoelastin monomers and subsequently aligns with collagen fibres before cross-linking by LOX to become part of the fibril assembly [16,119]. As part of the elastic fibre, elastin has several functions; providing elasticity and resilience to connective tissue, regulating the activity of TGF-β as well as having roles in cell migration, survival and differentiation [120,121]. During ageing the expression of elastin decreases. For example in older tendons, there is less elastin while the arterial wall stiffens due to degradation and fragmentation of elastic fibres [122,123]. Pathological increase in deposition and degradation of elastin fragments have been associated with fibrotic disease and poor survival in IPF [124–126]. A more recent study in mice demonstrated that elastin was increasingly expressed in active areas of lung fibrosis induced by bleomycin. In addition, the increased expression was associated with fibroblast activation and differentiation into myofibroblasts that potentially creates a feedback loop to reinforce fibrosis [127]. Elastin does not directly interact with cell surface receptors such as integrins, instead it uses a receptor complex named elastin-binding protein (EBP). During hydrolysis, elastin is cleaved into small peptides such as Val-Gly-Val-Ala-Pro-Gly (VGVAPG) and Val-Val-Gly-Pro-Gly-Ala (VVGPGA). It has been shown that these peptides induce several biological effects such as cell proliferation, migration and differentiation [128]. Furthermore, these peptides lead to chemotaxis of neutrophils and monocytes contributing to a self-propagating cycle of inflammation and ECM proteolysis [108,129]. A recent study reported that VVGPGA can induce expression of IL-α, IL-1β and IL-6, which have been identified as part of the SASP, in ligamentum flavum cells [130–132]. In addition, VGVAPG increased the caspase-1 activity in astrocytes while decreasing IL-1β release in vitro [132]. Caspase-1 has been identified as part of the SASP and is strongly associated with fibrosis [133,134]. The most promising receptor activated by VGVPAPG in relation to fibrosis and cellular senescence is peroxisome proliferation-activated receptor-γ (PPARγ) [135]. PPARγ activation leads to the induction of senescence, characterised by increased levels of G₁ cell-cycle arrest, p16 protein expression and an increased SA-β-Gal expression all of which are characteristics of senescent cells [136]. During fibrosis there is increased production and degradation of elastin which might pathologically increase released peptides and contribute to the senescent phenotype.

Fibronectin is a glycoprotein that is expressed by multiple cell types and plays a significant role in cell adhesion and migration. Fibronectin as well as collagens are needed for normal embryonic development. During wound healing the first active secreted matrix is fibronectin which replaces the fibrin clot formed after the initial insult. To restore homoeostasis, fibroblasts produce fibronectin and collagen as a provisional matrix which is eventually replaced by collagen abundant ECM. Fibronectin is necessary for collagen fibril formation and in vitro added collagen was organised into matrix fibres that co-localised with pre-existing fibronectin fibrils indicating the important function of both ECM proteins [137]. In addition, fibronectin can bind to itself or other ECM proteins such as fibrin, heparin and tenascin. Fibronectin can act as a reservoir for growth factors by binding vascular endothelial growth factor (VEGF), hepatocyte growth factor (HGF), FGF-2, PDGF and BMP [138]. Dysregulated fibronectin has been implicated in diseases such as cancer and fibrosis [139,140]. A recent study showed that fibronectin is able to influence cell-cycle progression and could inhibit cellular senescence and apoptosis in glioma cells via the PI3K/Akt signalling pathway [141]. Fibronectin acts through integrin αvβ3 and it was recently demonstrated that the β3 subunit is involved in cellular senescence via its role in TGF-β activation. During OIS there is a dynamic increase in subunit β3 while down-regulation of β3 reduced OIS and therapy-induced senescence. An αvβ3 antagonist was shown to attenuate the SASP without affecting proliferation in glioma cells. While the exact mechanism remains unclear, there is strong evidence showing that fibronectin is involved in senescence and its increased abundance in fibrosis could potentially contribute to the senescent phenotype.

Laminins are proteins that are secreted by epithelial cells into the basement membrane. They contain an α-chain, a β-chain and a γ-chain and there are several splice variants found. Laminins are an integral part of the ECM structure and they mediate cell adhesion, migration, differentiation and proliferation [142]. During wound repair they are critical players in re-epithelialisation and angiogenesis, reviewed in detail by Lorio et al. [143]. The expression of laminins is altered in patients with IPF, and in a small number of patients the distribution of laminin α3 subunit was lost [144]. Interestingly, increased markers of senescence were found in fibroblastic foci and localised clusters of basal cells expressing high levels of laminin 5 γ2 [145], suggesting a potential association between cellular senescence and laminin deposition. Moreover, fibroblastic foci show co-localised expression between laminin and markers of senescence. Increased expression of laminin was also reported in other diseases such as cancer and Sjörgren’s syndrome [146,147]. Laminin is able to activate various signal pathways; for example in murine melanoma cells, laminin-1 results in activation of phospholipase D and subsequent induction of phosphatidic acid leading to MMP-2 production favouring tumour invasion [148]. Laminin signalling is mediated via integrins and the 67-kDa laminin receptor [149]. Laminin subunit α1 (Lama1), part of the laminin 111 complex was originally thought to be only expressed during morphogenesis. However, Lama1 was recently shown to be a genetic modifier of TGF-β activity in the context of pulmonary fibrosis [150]. In kidney development, the absence of Lama1 increased TGF-β induced Smad2 phosphorylation in knockout mesangial cells in addition to increased proliferation [151]. These studies demonstrate a strong association between Lama1 and TGF-β. In a study using Schwann cells, Armstrong et al. reported that NF-κB is induced by laminin which had a positive impact on neurite outgrowth [152]. When looking at ageing in endothelial cells of murine origin, a shift was found in expression of laminin β1 and laminin β2. Culture of endothelial cells on laminin produced a difference in response to β1 and β2-containing laminin showing reduced endothelial cell proliferation and migration [153]. While the crucial function of laminin has been described extensively in morphogenesis and wound repair, its involvement in senescence remains poorly understood. There is some evidence suggesting that laminin is involved in senescence and this might be in a TGF-β-dependent way or via activation of NF-κB, but this needs to be elucidated.

Matricellular proteins are dynamically expressed and secreted into the extracellular environment but they do not contribute directly to the major fibrillar structure of the ECM. These proteins modulate cell function by binding to a large variety of cell-surface receptors, cytokines and ECM proteins. They are well-recognised proteins such as CCN family member 1 (CCN1), fibulin, periostin, thrombospondin, tenascin-C and decorin.

The best characterised matricellular protein in relation to senescence is CCN1. The CCN family consists of six different types of which CCN1 and CCN2 are the most studied in relation to senescence, fibrosis and cancer [154,155]. During each stage of wound repair CCN1 plays a role in inducing senescence in fibroblasts and endothelial cells. During the proliferation phase fibroblasts undergo senescence to promote formation of granulation tissue. During the maturation phase myofibroblasts undergo senescence to initiate matrix remodelling and limiting fibrosis through up-regulation of MMPs and down-regulation of ECM proteins [156]. Downstream targets ras-related C3 botulinum toxin substrate 1 (RAC1) and NADPH oxidase 1 (NOX1) of CCN1 are activated through the interaction with α6β1 integrin resulting in ROS accumulation leading to senescence via the p53 and p16 pathways [157,158]. Recently, Leask et al. reported that loss of CCN1 expression in fibroblasts caused resistance to bleomycin-induced fibrosis but did not impair myofibroblast induction [159]. Interestingly, endogenous CCN1 leads to matrix organisation and stability while CCN2 induces myofibroblast differentiation [160]. In contrast, Grazioli et al. reported that CCN1 was found to be increased in COPD, IPF and animal models of lung fibrosis exacerbating lung injury while inhibition was found to be protective of fibrosis in the kidney [161,162]. In a study using IPF fibroblasts the essential role of CCN1 in mediating profibrotic gene expression via the TGF-β/SMAD3 pathway was demonstrated [154]. CCN2 has been reported to be a pro-fibrotic mediator in several organs and acts downstream of TGF-β [163]. The role of CCN1 and CCN2 has been demonstrated to be tissue- and cell type-specific and how they contribute to fibrotic disease remains unclear. It is plausible that activation of CCN1 and CCN2 contributes to fibrosis by inducing senescence followed by failure of immune clearance during repair leading to pathological senescence.

Periostin is a matricellular protein which has been associated with various disease and pathological conditions including cancer and congestive heart failure and is highly expressed in fibrotic disease [164–168]. In IPF, periostin is highly expressed in fibroblastic foci and in vitro upon stimulation with IL-4 or IL-13, increased secretion in the supernatant of fibroblasts but not epithelial cells was found [169]. In a murine model, neutralising antibodies prevented bleomycin-induced fibrosis indicating the importance in early onset fibrosis. Periostin binds to ECM including collagen type 1 and fibronectin and transfers signals from the ECM to the cell via specific integrins (i.e., αvβ1, αvβ3, αvβ5 and α6β4) to regulate cell adhesion, proliferation and migration [164]. In cancer, high expression of periostin is associated with increased proliferation, cell invasion and metastasis through activation of the integrin-PI3K/Akt pathway [170,171]. Periostin up-regulates TGF-β and in turn TGF-β may up-regulate periostin providing a complex positive feedback [172]. In a recent study using human fibroblasts, Nanri et al. found that the cross-talk between periostin and TGF-β via SMAD3 resulted in up-regulation of CTGF, IGFBP-3, IL-11 and SERPINE1 in pulmonary fibrosis [173]. Furthermore, it was shown by Li et al. using rat cells that age-related periostin expression from cardiac fibroblasts promotes senescence in cardiomyocytes via the angiotensin II-TGF-β/mitogen-activated protein kinase (MAPK)/ERK pathway [174]. More recently Kim et al. demonstrated that angiotensin II mediated CCN1 induced senescence via the angiotensin II type 1 receptor (AT1R)-p53 dependent pathway in human coronary artery smooth muscle cells. Angiotensin II induces senescence via activation of Erk1/2 and p38 MAPK leading to p16 and p53 expression while CCN1 mediates between Erk1/2 and p53 [175]. While there is clear evidence about the crucial role of periostin in fibrosis and its regulatory effect on TGF-β, there is little knowledge about the influence on cellular senescence.

Fibulin is a secreted glycoprotein important for the stabilisation and binding of other ECM proteins such as fibronectin, PGs and laminins during collagen deposition [172]. The human fibulin family consist of seven members (fibulin-1–7) of which fibulin-1 expression is associated with fibrosis [176,177]. Fibulin-1 is predominantly expressed in connective tissue that highly expresses elastic fibres such as vessels, skin and lung [178]. The soluble form of fibulin-1 can be found in plasma and interacts with fibrinogen. Soluble fibulin-1 is associated with disease severity in IPF and has been identified as a potential marker of kidney disease and diabetes [176,179]. Four different isoforms of fibulin-1 (FBLN1A, B, C and D) are known, each with a different function [178]. In a recent study it was shown that the activation of fibroblasts to myofibroblasts by TGF-β was substantially reduced in the absence of FBLN1C [177]. Furthermore, FBLN1C can increase MMP activity leading to increased production of ECM fragments which could act as DAMPs [177]. Histological staining in healthy and control lung has shown that fibulin-1c is highly expressed in fibrotic areas of the lung while non-fibrotic areas show little expression. Given that fibulins are responsible for the stabilisation of ECM proteins, it is possible that fibulin-1c creates or contributes to an environment that favours induction of senescence in fibroblastic foci. Whether this is via TGF-β activation or by direct binding of specific integrins is unknown. While the evidence suggests a strong role of fibulin-1 in fibrosis, other members within the family are of interest as regulators of cell function. Fibulin-5 can block angiogenesis by direct binding to the fibronectin receptors α5β1 and α4β1, and in endothelial cells it can increase the integrin-induced production of ROS [180]. In addition, fibulin-5 can interact with LOXL-1, -2 and -4 to tether LOXL on to elastic fibres [181]. A role for fibulin-3 is suggested in the pathogenesis of OA where it has been shown that reduced levels are associated with age and high levels with bone marrow mesenchymal stromal cells (BM-MSCs) [182]. Moreover, Chen et al. reported that fibulin-3 functions as a suppressor of cancer invasion via inhibitory effects on ERK and Wnt/β-catenin and that both fibulin-3 and 5 are down-regulated in invasive lung cancer and metastasis [183]. Over the years several lines of evidence have suggested that Wnt/β-catenin signalling plays crucial roles in ageing and cellular senescence [184–187]. Modulation of Wnt/β-catenin signalling in systemic lupus erythematosus BM-MSCs leads to a decrease in SA-β-Gal expression and the p53/p21 pathway, and increased proliferation [188]. More recently, chronic Wnt/β-catenin signalling was linked to senescence induction in lung epithelial cells [189]. Evidence is accumulating that the fibulin family might have a potential role in regulation of cellular senescence by altering expression of the Wnt/β-catenin signalling and increasing ROS production both of which are linked to senescence and age-related pathologies.

Aggrecan is a major glycoprotein predominantly found in articular cartilage of the joint providing a hydrated gel structure that allows the joint to withstand tremendous pressure. Aggrecan consists of three disulphide bonded globular regions with keratin-sulphate and chondroitin-sulphate linked to its core. Aggrecan does not exist in free form but is found in PG aggregates, which is composed of hyaluronic acid (HA) with multiple aggrecan molecules attached to it. While aggrecan is best understood for its function in cartilage and pathogenesis of OA, there is evidence suggesting a role in arterial remodelling [190]. In a recent study by Yasmin et al. it was suggested that loss of aggrecan functionality could impact viscoelastic properties, ECM disorganisation and collagen fibril redistribution to modulate vessel stiffness [191]. Additionally, it was unexpectedly shown that aggrecan was expressed in normal skin and to be significantly increased during ageing [192]. While it has been reported that PG deposition is altered in disease, the exact role of aggrecan in fibrosis is poorly understood [193,194]. It is possible that aggrecan in serum could play a role in fibrosis. CD44 is able to bind aggrecan and is necessary for an invasive fibroblast phenotype in fibrosis [195]. Furthermore, Tsuneki and colleagues described that CD44 influences fibroblast behaviour via modulation of the Survivin and Hippo pathways [196]. The contribution of aggrecan remains poorly understood with little to no evidence present that could suggest a role in fibrosis let alone cellular senescence.

Tenascins are a family of giant proteins found in the ECM of embryonic tissues, wounds, tumours and in fibrosis. There are four different proteins named tenascin-C, -R, -W and -X, which somewhat overlap in adult tissue [197]. Tenascins bind to different ECM proteins including fibronectin, PGs and cell-surface receptors. Their function is cell type specific and tenascin can promote or inhibit cell adhesion. Tenascin-R is predominantly found in the central nervous system during development while the others are more widespread and can be found in connective and soft tissue [198]. Tenascin-X and -Y are mainly found in skeletal muscle connective tissue during development while tenascin-C and -W are expressed in a variety of developing tissues [199]. The focus of tenascin research has predominantly been in cell migration and tissue remodelling, tumour development or fibrosis. This review will focus on tenascin-C because of its association in pathological conditions such as vascular hypertension, myocardial infarction and fibrosis [200–204]. There is little or no tenascin-C expression in healthy adult tissue but there is de novo expression after tissue injury. Tenascin-C can interact with several cell-surface receptors such as integrins α9β1, αvβ3 and αvβ6, and toll-like receptor 4 (TLR4) to control cell adhesion, motility and proliferation. Cell adhesion can be modulated by tenascin-C by direct interaction with fibronectin and integrin α5β1 to diminish focal adhesions. Cell proliferation can be modulated through direct binding and activation of EGF-receptor and tenascin-C can act as a DAMP by activating TLR-4 [205,206]. It has been reported that TLR4 is able to up-regulate TGF-β expression in fibrosis across multiple organs [207,208]. In a recent study persistent tenascin-C deposition was shown to lead to activation of TLR4-dependent profibrotic responses in skin and lung fibrosis [208]. This included enhanced expression of collagen, myofibroblast activation, fibroblast migration and contraction, and secreted factors such as IL-6 and TGF-β. Activation of TLR4 using lipopolysaccharide (LPS) and TGF-β has shown a different profibrotic response between control and IPF lung fibroblasts, perhaps as a consequence of reduced TGF-β receptor type 1 expression on IPF fibroblasts [209]. In pancreatic cancer Shi and colleagues described that tenascin-C can induce resistance to apoptosis through activation of the Erk1/2/NF-κβ/p65 pathway [210]. Moreover, it was reported that tenascin-C is able to bind several soluble factors including PDGF, VEGF, FGF and TGF-β [211]. The affinity for growth factors could be due to the large role it plays during tissue repair and disease. Furthermore, growth factors bound by tenascin-C have distinct roles in fibrosis, have been described to be part of the SASP and more importantly they are targeted by the antifibrotic drug nintedanib [212]. Taken together it may suggested that tenascin-C plays an important role during fibrosis by increasing collagen production, fibroblast activation and creating a profibrotic environment potentially contributing to cellular senescence.

Decorin is a small leucine-rich PG that is widely expressed in different tissues by fibroblasts and its function is required for normal fibrillogenesis of collagen. Decorin is involved in many different biological and physiological processes including proliferation, differentiation, muscular development, regulation of inflammation and autophagy, and fibrosis of kidney and liver [213,214]. Decorin interacts with other matrix proteins and a variety of cell surface receptors such as EGF-receptor, insulin-growth like receptor, HGF-receptor, VEGF-receptor and PDGF-receptor. Reszegi and colleagues recently reported that decorin has a protective role in liver cancer and could theoretically be used as a physical tyrosine kinase receptor inhibitor [215]. The antifibrotic properties of decorin have been well characterised in the last two decades. Decorin binds and forms a complex with TGF-β, which prevents activation of the TGF-β receptor type 1 and subsequent signalling via SMAD2/3-Erk1/2 protein [216]. The antifibrotic properties further extend by down regulating production of fibronectin, thrompospondin-1, inhibition of collagen maturation and increased production of MMP-1 [217]. Using a colon carcinoma cell-line it was shown that decorin can induce growth suppression by up-regulating p21 and CDK in a p53 and TGF-β independent pathway [218]. Overexpression of decorin in the adenocarcinoma cell-line A549 resulted in an increase in p21 and p53 expression and decreased phosphorylation of EGF-receptor leading to inhibition of metastasis [219,220]. In a study comparing COPD to control fibroblasts it was suggested that cellular senescence is associated with deregulation of ECM expression. A negative correlation was shown between decorin and p16 expression [221]. The reported antifibrotic effect of decorin has been documented in the literature but the effect on cell-cycle progression has been given much less attention. Interestingly, decorin also has the ability to induce premature apoptosis in dermal fibroblasts contributing to hypertrophic scarring [222]. It is possible that decorin induces cell-cycle inhibition in pre-senescent cells but fails to induce apoptosis in this specific phenotype. In addition it was reported that cleaved fragments from decorin in serum correlate with disease severity in both cancer and fibrosis [223]. In the next section of the review the role of decorin as a DAMP will be discussed in more detail.

It has become increasingly evident that the ECM has important immunomodulatory properties. The ECM can act as a reservoir for cytokines and matrikines which can be liberated from the ECM. Matrikines are small peptides proteolytically cleaved from the ECM macromolecules in order to regulate cellular functions, including cell migration, proliferation, and death [224]. Another important immunomodulatory function of the ECM is to signal damage and danger, and subsequently alert and activate the immune system. The signalling of damage and danger is done by the release of danger signals or DAMPs from the ECM [225]. DAMPs were first proposed by Polly Matzinger in 1994 when she described the danger theory [226]. This theory states that not only molecular products derived from ‘strangers’, or pathogen-associated molecular patterns (PAMPs) can activate the immune system, but similar molecular patterns derived from damaged or death host-cells can also activate the immune system using the same receptor and signalling pathways. The list of currently known DAMPs is extensive and still growing [225,227,228]. However, all DAMPs have in common that they are actively or passively released upon cellular or structural damage and that they elicit pro-inflammatory functions by activation of PRRs upon release which are distinct from their physiological functions [225]. Currently, there are five main classes of PRRs: (1) TLRs, transmembrane proteins located at the cell surface or intracellularly on endosomes; (2) NOD-like receptors (NLRs), located in the cytoplasm; (3) C-type lectin receptors (CLRs), which are transmembrane receptors located at the cell surface; (4) RIG-I-like receptors (RLRs), which are also located intracellularly and (5) the Receptor for Advanced Glycation End products (RAGE) a transmembrane protein located at the cell surface [229]. Upon activation of PRRs by DAMPs, downstream signalling pathways are activated, including NF-κB, MAPK and type I interferon pathways, leading to the release of pro-inflammatory cytokines and chemokines (e.g. IL-6, chemokine ligand (CXCL) 8 and TNF-α), ultimately resulting in activation of the immune system and attraction of immune cells to the site of damage [230]. Next to DAMPs derived from intracellular compartments upon release during cell death, it has been shown that the proteolytic cleavage of ECM macromolecules leads to the release of pro-inflammatory peptides, coined ECM DAMPs [231,232]. Examples of ECM DAMPs are peptides from (1) GAGs such as, low-molecular weight hyaluronan and heparan sulphate, (2) PGs such as, decorin, biglycan, aggrecan and versican and (3) glycoproteins such as, fibronectin, fibrinogen and tenascin-C. All these peptides can be cleaved by proteases, such as MMPs, ADAMs or more specific proteases, including hyaluronidase and heparanase and that they can induce a pro-inflammatory response by activation of PRRs such as TLRs, RAGE and NLRP3. Which proteases cleave the different ECM DAMPs and the exact receptors and downstream signalling pathways are reviewed in detail elsewhere [231,232].

Little is known about the role of extracellular DAMPs in the pathophysiology of IPF. Given the fact that IPF is characterised by the exaggerated deposition of ECM within lung tissue it is enticing to speculate about the role of ECM DAMPs in IPF. Although many studies have been performed investigating individual ECM DAMPs in the context of IPF, no studies have been performed investigating the role of ECM DAMPs in a more structured and controlled fashion. Most studies have investigated the levels of specific ECM DAMPs by either measuring the intracellular expression levels by western blot or mRNA expression or the protein concentration within the ECM by immunohistochemistry. While an increased intracellular expression or increased ECM deposition of ECM DAMPs may be indicative for increased cleavage of pro-inflammatory ECM DAMP fragments it is not directly correlated. In order to study ECM DAMPs they should be assessed in the context of fragments released into the extracellular space, for instance by measuring ECM DAMPs in serum, plasma, sputum, bronchoalveolar lavage fluid (BALF) or epithelial lining fluid.

Several studies have been performed investigating the extracellular fluid levels of ECM proteins such as fibronectin. Fibronectin has several isoforms, which all originate from the same gene. One of these isoforms is referred to as fibronectin-EDA [233]. Fibronectin-EDA is enriched at sites where tissue remodelling due to inflammation is present and has the ability to activate the immune system by binding TLR4, thereby acting as a DAMP [234,235]. Although the levels of fibronectin-EDA have not been measured in extracellular fluids of IPF patients to date, using a targeted mass-spectrometry approach it was shown that the plasma levels of fibronectin were significantly lower in plasma of IPF patients compared with matched healthy controls [236]. In contrast, the levels of fibronectin in BALF were shown to be two- to three-fold increased in IPF patients compared with healthy controls [237]. This was supported by a second study showing a two-fold increase in fibronectin in BALF of IPF patients, however the study lacked the statistical power to reach significance [238]. Together, these studies indicate that fibronectin is locally released in lung tissue in IPF patients, but the leakage of extracellular fibronectin into the circulation is limited. This may lead to a local inflammatory response attracting innate immune cells to the site of release.

Another ECM glycoprotein with the ability to activate TLR4 upon release from the ECM is fibrinogen. Bargagli and colleagues described that the levels of fibrinogen were more than three times as high in patients with IPF compared with healthy controls. Interestingly, the levels were not different between IPF patients in stable disease and IPF patients which experienced an acute exacerbation [239]. Tenascin-C is another glycoprotein with DAMP properties, that can activate TLR4 when released from the ECM [231]. One study investigating the levels of tenascin-C in serum and BALF in patients with idiopathic interstitial pneumonias found no differences in both serum and BALF levels of tenascin-C between healthy controls and IPF patients [240]. However, the serum tenascin-C levels were significantly increased in patients with cryptogenic organising pneumonia (COP) and the BALF tenascin-C levels were increased in both COP and non-specific interstitial pneumonia patients [240].

Several studies have been performed investigating decorin in the context of IPF. Decorin has been shown to be immunostimulatory when it is proteolytically cleaved by MMP3, MMP13 or cathepsin-S and released from the ECM into the extracellular space [241,242]. Cleaved decorin has a small protein core which is up to 42 kDa and contains a leucine-rich-repeat that is also present on TLRs and NLRs, enabling it to ligate and activate these PRRs [243]. In soluble form, decorin can activate TLR2 and TLR4, increasing activity of the p38 MAPK and ERK pathways and resulting in the synthesis of pro-inflammatory cytokines like IL-1β and TNF-α [244] (Figure 2). Interestingly, Kehlet and colleagues reported in two independent cohorts that decorin fragments specifically degraded by cathepsin-S are significantly increased in serum of IPF patients compared with healthy controls [223]. This was shown by an elegantly designed ELISA assay that specifically detects decorin fragments that are cleaved by cathepsin-S [223]. Another study found that serum decorin levels significantly correlated with oxygenation, and that IPF patients with low serum decorin levels showed a significantly higher survival rate compared with IPF patients with high serum decorin levels [223]. These data indicate that in IPF there is increased breakdown of decorin which produces increased circulating levels of pro-inflammatory decorin fragments, which may contribute to the development of IPF.

Although the data listed above indicate that IPF patients display increased ECM breakdown, that increases the concentration of pro-inflammatory ECM DAMPs both systemically as well as locally, much is unknown about the role of ECM DAMPs in the pathophysiology of IPF. Furthermore, although the link between increased senescence and increased ECM DAMP production seems logical, as both are associated with release of the same pro-inflammatory mediators, no studies have been performed to date to investigate the connection between ECM DAMP release and senescence in IPF. A systematic study investigating the levels of multiple ECM DAMPs both locally and systemically in the same cohort is needed to assess the importance of ECM DAMPs in IPF.

Current therapeutic approaches for modulating senescence aim to specifically kill aberrant cells that have entered the senescent state and are having a detrimental effect on the tissue microenvironment. Such agents are being newly developed for this purpose, but also include several drugs (for example quercetin and dasatinib) that were originally developed for targeting tumours. However, as some of the markers that are present in malignant cells are also specifically expressed in senescent cells these agents have now been repurposed as senolytic agents [245]. These drugs have been shown to effectively reduce the number of senescent cells, and in some cases decrease fibrosis, at least in preclinical models, potentially through modulation of the pro-inflammatory SASP released by the senescent cells [61,246]. In a first human, open label pilot study, the potential application of dasatinib and quercetin was tested in 14 patients with IPF [247]. One patient had a serious adverse event during the three-week trial and mild-moderate adverse events included respiratory symptoms, skin irritation/bruising and gastrointestinal complaints. There was a significant improvement in the physical condition of the patients, but the pulmonary function of the patients did not change within this short trial period. Circulating SASP factors were measured, and while no significant change was reported there was a suggestion of reduced levels of selected proteins important for fibrotic remodelling, including IL-6, MMP and TIMP2 in at least 8 of the patients. Intriguingly, changes in pulmonary function and physical condition of the patients correlated with changes in the circulating levels of matrix remodelling proteins. While this is a very early study that needs to be validated in a much larger population in a randomised controlled trial, these observations further suggest that the role of the ECM should not be overlooked when striving to understand the regulation of senescence in fibrosis.

The second approach for modulating senescence in vivo strives to interfere with the impact of the released elements of the SASP using drugs named senomorphics. This approach aims to block the components of the SASP from signalling to surrounding cells and in this way reduce the damaging effects of the senescent cells. As ECM remodelling components are recognised elements of the SASP, an innovative approach would be to begin exploring how blocking the activity of these proteins and enzymes impacts the fibrotic disease phenotype. In line with this, there is a need to investigate the therapeutic options of inhibiting either the release or activity of ECM DAMPs in fibrosis. DAMPs, have been well characterised in terms of their role in inflammaging, particularly RAGE ligands, which have a well characterised role in innate immune driven pathologies such as COPD, OA and asthma [248–251]. However, the contribution of DAMPs and the effect of targeting this therapeutically in fibrosis is less well explored. Majewski and colleagues examined the levels of IL-25, IL-33 and thymic stromal lymphopoietin (TSLP) in patients with IPF and followed them after one year of treatment with the antifibrotic drugs nintedanib or pirfenidone [252]. The baseline levels of TSLP were elevated in IPF patients compared with controls, while IL-25 and IL-33 levels did not vary. Strikingly, patients showed reduced levels of both TSLP and IL-25 in exhaled breath condensate after twelve months of treatment with pirfenidone, but not nintedanib. The investigators did not measure the levels of ECM DAMPs in this study leaving an open question as to the likely impact of these antifibrotic therapeutics on ECM DAMPs.

An unexplored realm for modulating senescence in fibrotic diseases, is through modulating the ECM changes directly. It is recognised that the SASP released by senescent cells can induce ECM structural changes and stiffening of the local tissue microenvironment [253] (reviewed in [254]). These changes are well characterised in fibrotic tissues and are known to drive aberrant cell functions including, but not limited to, enhanced proliferation, migration, cytokine/growth factor production and ECM deposition and crosslinking [17,255,256]. Given the role of senescence in regulating tissue repair, it would seem logical that the altered ECM would also give rise to a greater number of senescent cells as part of the natural defence response. Cell interactions with the ECM, via integrin α6β1 and the matricellular protein CCN1 [157,257,258], or integrin αvβ3 [259] exemplify interactions that induce senescence. Therefore, exploring therapeutic options for targeting these cell-matrix interactions orchestrated via these, and possibly other, integrins may hold promise for regulating senescence.

The influence of the currently approved antifibrotic therapeutics nintedanib and pirfenidone on the induction of senescence has not been directly examined. Both of these drugs are known to modulate the production of major ECM proteins including collagen I and fibronectin in in vitro or ex vivo models [260,261]. In vivo, administration of both antifibrotic therapies did not modulate the senescence markers p16 or p21 nor the deposition of dense collagen deposits in lung tissues collected at the time of lung transplantation. The lack of effectiveness of these therapies in this instance may reflect the end stage disease tissues examined in this study [262]. Encouragingly, both nintedanib and pirfenidone inhibited collagen fibril formation in vitro, an important step in the early stages of the deposition of collagen fibres driving the development of fibrosis, and returned the microscopic appearance of the fibres deposited by fibroblasts derived from donors with IPF to an appearance more like those deposited by fibroblasts from a healthy donor [261]. The structural organisation and crosslinking of the ECM, as directed by LOX family enzymes, deposited by IPF fibroblasts dictated the subsequent response, in terms of attachments and proliferation, of other fibroblasts that were then exposed to this environment [263]. While this study did not examine the senescence characteristics of cells exposed to the IPF-derived ECM it would be an important follow up for understanding the potential for modulating the ECM microenvironment as a tool for controlling senescence in the fibrotic environment.

This review has highlighted the potential feedback loop resulting in enhanced senescence driven by cellular responses to their altered ECM microenvironment. As our understanding of the mechanisms underlying the effectiveness of the approved antifibrotic therapies for lung fibrosis, nintedanib and pirfenidone, and further exploration of novel antifibrotic therapies advances, new possibilities for modulating levels of senescence with increasing finesse should emerge.

In this review, we discussed the current knowledge about the ECM as a modulator of cell function and how this could potentially contribute to the senescent phenotype observed in chronic fibrotic diseases. The potential of ECM as a regulator of cellular senescence is described in several ECM components across a variety of tissues while some ECM components may prevent cellular senescence (Figure 3). Moreover, the ability to regulate cell function was not always directly attributed to the ECM component itself i.e. binding to cell-surface receptors, but rather by the ability to bind SASP factors and serve as a reservoir for growth factors/cytokines. Furthermore, some ECM components are described as having a strong regulatory effect once cleaved and released as danger molecules in the environment (Figure 3).

While our knowledge has tremendously increased over the last three decades, much remains unclear as to how the specific ECM components play a role in regulating or contributing to pathological disease states such as fibrosis. Much of what we know about the function of ECM has been discovered in relation to cancer research, which shares similar mechanisms with fibrosis. It is interesting that the majority of matricellular proteins described in the literature show a stronger regulatory effect towards cellular senescence than the core proteins collagen, elastin and laminin. Although these proteins have been extensively investigated, their association with senescence remains poorly understood considering the important role they play for providing structure and stability. The matricellular proteins CCN1, decorin, periostin and tenascin-C all have in common that they might contribute to senescence by regulating cell-cycle, apoptosis and production of secreted factors (Figure 4). In contrast fibronectin and fibulin can act both as inhibitor and promotor of cellular senescence. Lastly most DAMPs released from matricellular proteins lead to a pro-inflammatory response via TLR receptors contributing to up-regulation of secreted factors which are shared with the SASP from senescent cells.

Cellular senescence and the regulation or induction by aberrant ECM contributes to a negative feedback loop in which both reinforce each other. It is difficult to know if senescence is a result of dysregulated ECM deposition, or dysregulated ECM is a consequence of senescent cells. There remain many mysteries as to how the different ECM components can regulate cellular senescence in chronic fibrotic disease. New therapies that combine antifibrotics and senolytics might be able to interrupt the negative feedback loop and interfere with both the aberrant production of ECM and aid in clearance of senescent cells to ultimately halt or reverse disease progression.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the Rosalind Franklin Fellowship from the University of Groningen and the European Union.

Open access for this article was enabled by the participation of University of Groningen in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society.

K.E.C.B. prepared the figures. K.E.C.B., S.D.P., M.S. and J.K.B. drafted the manuscript. K.E.C.B., S.D.P., M.S., D.A.K. and J.K.B. edited and revised the manuscript. K.E.C.B., S.D.P., M.S., D.A.K. and J.K.B. approved the final version of manuscript.

ADAM

a disintegrin and metalloproteinase

ATM

ataxia-telangiectasia mutated kinase

BALF

bronchoalveolar lavage fluid

Bcl-2

B-cell lymphoma 2

BM-MSC

bone marrow mesenchymal stromal cell

BMP

bone morphogenetic factor

CCN1

CCN family member 1

CDK

cyclin-dependent kinase

COL1A1

collagen type I α-1

COP

cryptogenic organising pneumonia

COPD

chronic obstructive pulmonary disease

DAMP

damage-associated molecular pattern

DDR

DNA damage response

ECM

extracellular matrix

FGF

fibroblast growth factor

GAG

glycosaminoglycan

HGF

hepatocyte growth factor

IL

interleukin

IPF

idiopathic pulmonary fibrosis

LO/LOX

lysyl oxidase

LOXL

LOX-like protein

MAPK

mitogen-activated protein kinase

miR

microRNA

MMP

metalloproteinase

NF-κB

nuclear factor κB

NLR

NOD-like receptor

OA

osteoarthritis

OIS

oncogenic-induced senescence

PDGF

platelet-derived growth factor

PG

proteoglycan

PPARγ

peroxisome proliferation-activated receptor-γ

PRR

pattern recognition receptor

RAGE

receptor for advanced glycation end product

ROCK

Rho-associated coiled-coil containing kinase

ROS

reactive oxygen species

RS

replicative senescence

SASP

senescence-associated secretory phenotype

SA-β-Gal

senescence-associated β-galactosidase

SIPS

stress-induced premature senescence

STAT3

signal transducer and activation of transcription 3

TG

transglutaminase

TGF-β

transforming growth factor-β

TIMP

thrombospondin motif

TLR4

toll-like receptor 4

TNF-α

tumour necrosis factor α

TSLP

thymic stromal lymphopoietin

VEGF

vascular endothelial growth factor

Wnt

wingless and int-1