The Matrisome during Aging and Longevity: A Systems-Level Approach toward Defining Matreotypes Promoting Healthy Aging

Cells and tissues are embedded within extracellular matrices (ECMs), which are important for tissue geometry, integrity, and function [1]. Besides providing structural support, the ECM controls intercellular commu-nication by either storing or transporting signaling -molecules, such as growth factors, hormones, and neuropeptides [1-3]. Receptors on the cell surface, such as integrins, directly link ECM to cell signaling to regulate various biological functions (Fig. 1). Cells synthesize and secret ECM proteins, such as collagens, extracellular glycoproteins, and proteoglycans that are integrated into matrices by extracellular enzymes [2]. Cells also continuously monitor and integrate signals from the ECM via mechanotransduction and adjust their surrounding environment by synthesizing and remodeling their ECM [2]. Thus, the ECM at least during development and normal tissue homeostasis is a dynamic system.

During aging, the connective tissue or ECM integrity declines through accumulation of damage from collagen fragmentation, oxidation, glycation, crosslinking, and accumulation of protein aggregates (Fig. 1), leading to a decline in ECM dynamics and loss of organs support and function [3, 4]. This decline in ECM integrity has been implicated in many age-dependent diseases, such as atherosclerosis, diabetes, and cancer [2]. Accumulation of damage to the ECM drives cellular aging and disease progression. Below, I discuss how collagen fragmentation, collagen glycation, collagen crosslinking, and aggregation of proteins in the extracellular space affect ECM integrity to accelerate aging. Moreover, longevity-assurance pathways might slow aging by activating collagen and ECM remodeling to protect from this progressive decline in ECM integrity.

Protein homeostasis is defined as the cellular regulation of synthesis, folding, trafficking and degradation of proteins. In contrast to unfolded intracellular proteins that can be refolded through chaperones, damaged, or unfolded ECM proteins need to be excised out of the matrix, discarded, and replaced by de novo synthesized ECM proteins (Fig. 1a). Since this turnover requires cells to be adjacent to the ECM, ECM structures that are acellular lack this dynamic turnover. For example, certain collagens in the eye lenses or cartilages of humans are only synthesized once, integrated into the ECM, and remain there lifelong with spectacular half-lives of over 117 years [4, 5]. By contrast, collagens in other tissues can be turned over extremely fast, for example, within 72 h in the Achilles tendons after exercise [6]. During aging, either through collagen fragmentation or loss of adherence proteins [7], cells detach from the ECM potentially leading to cell dysfunction and loss of ECM synthesis and turnover. In fact, the loss of ECM-to-cell connection might start a vicious downward spiral. For instance, during aging, there is an increase in activity of ECM-degrading enzymes, such as matrix metalloproteases (MMP) [7]. Increased MMP activity leads to collagen fragmentation, causing cell detachment, which leads to altered -integrin signaling and an increase in mitochondrial -reactive oxygen species, which in turn promotes the -expression of more MMPs, leading to further ECM fragmentation [7].

Since cells synthesize the ECM (Fig. 1a), this self-amplifying downward spiral might underlie the observed decline in collagen production during aging across species [3, 8, 9]. For instance, collagen mass continuously declines at a rate of 1% per year in the human skin [10], as best illustrated by wrinkles and sagging skin. In general, during aging there is a progressive decline in ECM biosynthesis, accompanied by an increase in ECM degradation. There are two striking pieces of evidence suggesting that this loss of ECM biosynthesis and increase in ECM degradation might drive or accelerate aging. First, treating young human skin samples with MMPs quickly alters the skin morphology to look like old skin [7]. Second, deficits in an ECM-remodeling enzyme MMP14 in the skin lead to premature aging, short lifespan, and cellular senescence in mice [11]. MMPs cleave collagens. We might expect that mice genetically engineered to express modified collagen COL1A1^r/r that cannot be cleaved by MMPs, would not show age-depended ECM degradation, and would live longer. However, collagen COL1A1^r/r mice showed accelerated aging, short lifespan, and cellular senescence [12], suggesting that blocking of ECM turnover is also detrimental. This points toward a balanced ECM turnover being important to maintain health during aging. Moreover, mouse models with premature aging disease (Hutchinson-Gilford Progeria Syndrome) fail to produce a functional ECM [13]. Interestingly, their progeric cellular phenotypes and short lifespan are rescued by a functional ECM [14]. This suggests an integrated system of ECM-to-cell-to-nucleus linkage, translating outside mechanical forces to changes in cellular gene transcription to balance ECM and tissue homeostasis. Taken together, the fragmentation or gradual weakening of the ECM causes alterations in mechanotransduction signaling to the nucleus, thereby initiating a self-amplifying process, disrupting cell-ECM interaction and accelerating aging.

The Maillard reaction is the major nonenzymatic glycation pathway for advanced glycation end-products (AGE) formation [4]. Simplistically described, glucose interacts with lysine and arginine side chains of a protein (e.g., collagen) to form reversible Schiff base products within hours, which are stabilized to keto amine (Amadori products) within days. Within weeks to months, in a series of chemical reactions these products are transformed into AGE (Fig. 1b), such as carboxymethyl-lysine, 3-deoxyglucosone-lysine dimer, glyoxal, glucosepane, pentosidine, and so on [4]. These AGEs are able to crosslink and thus stiffen collagen structures. Glycation and crosslinking of collagens impair wound healing and is a major problem for diabetes patients with high blood sugar levels [4]. Cells sense this increase in ECM stiffness and respond by expressing ECM-remodeling enzymes to adjust to their preferred matrix stiffness [15]. However, AGEs or crosslinks hinder enzymatic cleavage, potentially leaving these stiffer matrix parts, but cleave the surrounding intact matrix resulting in collateral damage. Thus, certain parts of the ECM become stiffer through crosslinking and other parts of the ECM become mechanically weaker through collagen fragmentation during aging. Ultimately, aged ECMs are categorically stiffer, but also mechanically weaker.

Aggregation of proteins is a hallmark of many neurodegenerative diseases and is thought to be a crucial factor in their pathogenesis. Postmortem studies of brains from Parkinson’s patients show an intracellular aggregation of α-synuclein. Recent evidence suggests that α-synuclein aggregates might be actively secreted or released after cell death and taken up by other neurons to spread α-synuclein aggregation in a “prion-like fashion” throughout the brain. Injecting mice with antibodies against α-synuclein helps to clear extracellular α-synuclein [16]. Importantly, this kind of “prion-like” spreading has been observed in models for Tau pathologies or models for SOD1 mutant amyotrophic lateral sclerosis. Intriguingly, targeting and clearing extracellular SOD1 by intraventricular infusion of a SOD1-specific antibody delayed the disease-onset and increased the lifespan of these SOD1 (G93A) transgenic mice [17]. The accumulation of extracellular amyloid beta has been suggested to be involved in the etiology of neurodegeneration in Alzheimer’s disease [18]. How these extracellular aggregates are cleared and how these mechanisms are lost during aging is still unclear. Microglia (“brain macrophages”) and autophagy have been implicated in the clearance of extracellular aggregates. However, components of the extracellular matrix itself, such as collagen COL25A1, have emerged from whole-genome sequencing of healthy elderly people to be protective against amyloid beta pathology [19]. This suggests that the ECM might not only be a by-stander that accumulates these aggregate-prone proteins but might also have a more functional and active role in either safe storing or clearing these aggregates (Fig. 1c). For instance, these aggregate-prone proteins that stick to the ECM might be turned over in young developing organisms (Fig. 1c). Assuming that this physiological ECM turn-over-process progressively declines during aging, this could explain the accumulation of these aggregation-prone proteins in the ECM during aging. Thus, understanding how this ECM turn-over-process works and how to reactivate it during aging might be a strategy to protect against these neurodegenerative diseases.

Aged ECMs are characterized by a loss of collagen mass, formation of AGEs, accumulation of protein aggregates, becoming stiffer and fragmented, and by losing their protective and mechanical functions that are important for cellular integrity. Many interventions that increase lifespan of Caenorhabditis elegans prolong the -expression and synthesis of collagens during aging [9]. Accompanying this prolonged-collagen-synthesis are MMPs and other enzymes, suggesting a proper integration and remodeling of collagens into the ECM [9]. This prolonged expression of collagens is required to extend lifespan of longevity interventions, such as reduced insulin/insulin-like growth factor (IGF)-1 signaling, reduced TOR signaling, germ-stem cell ablation, and dietary restriction [9]. Furthermore, overexpression of certain key collagens is sufficient to increase the lifespan of C. elegans [9]. Supplementing the diet with collagen peptides increases collagen expression and lifespan of C. elegans [20] and rats [21, 22]. In human osteoblasts, supplementation with collagen peptides increases collagen expression [23]. It is unknown how collagen supplementation or collagen overexpression induces de novo collagen biosynthesis and how this promotes longevity. I speculate that certain collagen levels or their integrity are monitored as indicators to maintain collagen and ECM homeostasis.

A fascinating finding is that placing senescent cells [24] or aged stem cells [25] in a “younger ECM” has been shown to rejuvenate these old cells. The ECM provides instructive signals that change cellular function and identity. For instance, placing tumor cells into an embryonic ECM reprograms them to lose their tumorigenicity [26]. Furthermore, the lost regenerative potential of old muscles is rejuvenated by grafting them into young, but not old hosts [27, 28]. Heterochronic parabiosis, that is, stitching together a young with an old circulatory system, or simply injecting young blood plasma, has been shown to rejuvenate several organs of old mice. Moreover, administrating human umbilical cord plasma, which is enriched in TIMP2, rejuvenates the brain of old mice [29]. Tissue inhibitor of metalloproteinase TIMP2 inhibits MMPs, which suggest that alteration of the ECM might underlie this rejuvenation. Although the underlying rejuvenating mechanisms are unknown, these findings point toward the importance of maintaining ECM integrity for healthy aging. This raises the question of what constitutes a “young” ECM? Which are the key ECM components important to maintain cellular homeostasis and promote health during aging? In order to answer these questions, we first need to identify the key molecular components, modifications, and proteins that make up a “young or healthy ECM.” Recent advances in the field of proteomics and data analysis enable such a quantification of ECM protein compositions. I build on these advances to define a novel concept that will help grasping changes in ECM composition during disease, health, aging, and longevity.

The entire set of proteins that can be expressed by the genome is known as the proteome. The proteome of ECMs has been recently defined for humans, mice, and C. elegans and is called the “matrisome” (http://matrisomeproject.mit.edu; [30, 31]). Thus, the matrisome is a subset of the entire proteome dedicated to ECMs [30]. The matrisome is divided into 2 main categories: core- and associated-matrisome [30]. The core-matrisome is the set of proteins that are synthesized and secreted by cells to form ECMs. The human genome encodes 44 collagen genes, 195 ECM glycoproteins (including, fibronectins, laminins, etc.), and 35 proteoglycans, forming the core-matrisome [30]. The “associated-matrisome” comprises of proteins that are secreted and either localize to the ECM or remodel the ECM. Thus, the “associated-matrisome” is further divided into secreted factors localizing to ECMs (344 proteins, including transforming growth factor β, bone morphogenetic proteins, Wnts, cytokines), ECM-regulators (238 proteins, including the proteases MMPs and cathepsins, the ECM-crosslinkers lysyl oxidases and transglutaminases), and ECM-affiliated (171 proteins, including annexins, semaphorins, syndecans, glypicans, C-type lectins) [30]. The matrisome of humans comprises of 1,027 genes, of mice 1, of 110 genes, and of C. elegans 719 genes, which for each of these organisms is about 4% of their protein-encoding genome [30, 31]. The mouse and human matrisome are quite similar, whereas the evolutionary more distant nematode C. elegans shares about 45% of conserved matrisome proteins with humans [31]. These include proteins usually found in metazoans, such as laminins; collagen types IV, XVIII, and XXV; perlecans; and syndecans [31]. For instance, human COL25 associated with healthy aging [19] is col-99 in C. elegans [31]. This in silico definition and comprehensive compendium of ECM proteins is key for the analysis of transcriptomic and proteomic datasets in order to identify matrisome gene products associated with young ECM and/or with healthy aging.

The matrisome encompasses all the ECM and ECM-associated proteins that are or can be expressed by a genome. However, different cell types express different ECM proteins [3]. Organismal phenotypes and physiological stages are characterized by distinct sets of expressed ECM proteins and by the occurrence of different posttranslational protein modifications [3]. Thus, the compositions and modifications of different matrisome proteins reflect cellular identity, physiological status, and phenotype. Based on this, I propose here a new term: the “matreotype,” which is the acute state of an ECM composition and/or modification associated with or causal for a given physiological condition or phenotype. Similar to the proteotype reflecting a specific state of the proteome [32], the matreotype reflects a specific state of the matrisome linked to a phenotype.

Below, I discuss technical challenges and theoretical considerations of the matreotype. As a proof-of-concept, I reanalyzed gene expression and proteomics data to identify a preliminary matreotype during aging and longevity, demonstrating that by combining transcriptomic, proteomic, and genetic approaches, the matreotype of healthy aging can be revealed across species in future approaches. Furthermore, I highlight that identifying and defining matreotypes will be of translational value.

Aging is characterized by a progressive physiological decline over time. This physiological decline is reflected in the matreotype as for instance in changes in composition of ECM proteins, accumulation of AGEs on collagens, and collagen crosslinking. During development and growth, cells constantly remodel the ECM by degrading parts of their ECM and through de novo synthesis of matrisome components in order to maintain homeostasis. This dynamic and energy-intensive process might decline after reproduction. Either because natural selection, as defined as reproductive fitness, is ineffective after reproduction, or because of a shift in resource allocation from somatic to germline tissue during the onset of reproduction. Irrespective of the etiology, this predicts that after reproduction the homeostasis of matrisome components, that is, de novo synthesis and ECM remodeling, would decline and should be reflected in the temporal change of matreotypes during aging. Thus, the matreotype of a young ECM is different compared to an old ECM.

The gradual decline of ECM over time should be reflected by distinct matreotypes at any given time point during aging. This raises the question whether in long-lived animals this gradual decline in matreotype is simply slowed during aging (temporal scaling) or longevity-assurance pathways produce an alternate matreotype to maintain youthfulness? The latter model of an alternate matreotype could be conceptually similar to the Waddington model during development; that is, just an alternative or parallel path or fate during aging. Either model, temporal scaling or alternate path/fate, would have implications on how to design strategies to improve health during aging. If its temporal scaling, interventions might be started right after reproduction to slow down this progressive decline in physiology. If it is an alternate fate, it could be activated during old age, assuming that there has not been excessive damage accumulation in order to promote some cellular function and reprogramming during aging.

There are several technical challenges for quantifying matreotypes that in principle could be overcome. First, quantifying the matreotype through standard proteomics needs to be adopted to enrich for ECM proteins. Since ECM proteins are crosslinked and form high molecular networks, they become insoluble and this insoluble fraction is often discarded. However, there are currently several proteomic protocols that have been successfully used to enrich for ECM proteins [33]. Furthermore, a targeted proteomics approach is needed for the quantification and reproducibility, such as SWATH-MS [34] or other comparable approaches. Second, although mRNA levels might be indicative for de novo synthesis of ECM proteins, mRNA expression levels largely do not reflect protein levels, which is very likely to be true for ECM proteins. For instance, even if a collagen is synthesized, correctly modified, forms a triple helix, and is secreted, this collagen needs to be further processed and incorporated into the matrix. Otherwise, these soluble collagens accumulate amorphously in the extracellular space as seen in fibrosis. Since collagens can act as a glue (“kolla” means glue, “gen” means producing), this property of filling the extracellular space is important for wound closure. Moreover, inflammation proceeds fibrosis [35] and during aging there is a steady increase of inflammation. This inflammation-induced matreotype, which is characteristic for fibrosis and amorphous collagen deposition during aging, can be assessed through tissue staining and biochemical techniques [36]. A third aspect for consideration is that there is also heterogeneity within cell types and tissues due to randomness in cellular gene expression including the expression and biosynthesis of ECM proteins. While single-cell RNA-sequencing is feasible, single-cell proteomics is more challenging. Alternatively, tagging ECM proteins with fluorescent proteins could reveal the matreotypes during aging in vivo and noninvasively [37]. Thus, these few examples already indicate that it might be challenging but feasible to quantify and interpret matreotypic changes during aging. A possible strategy would be to focus on functional or “causal” matreotypic changes during aging, which need to be experimentally identified and validated.

A systematic and longitudinal quantification of matreotype during aging or longevity is missing. However, here I present a preliminary matreotype by reanalyzing publicly available expression profiles and a proteomics dataset. Budovskaya et al. [38] measured the mRNA expression levels during C. elegans aging (day 0–11 of adulthood). They found that of the approximately 20,000 C. elegans genes, about 900 genes decline and 300 are increased in expression during aging [38]. To reanalyze this dataset, we used the C. elegans Matrisome Annotator (http://ce-matrisome-annotator.permalink.cc/; [31]). About 150 out of these 1,254 age-regulated genes [38] are matrisome genes, comprising of 92 collagen genes (Fig. 2a, b). Most matrisome genes are expressed during development and growth, but their expression rapidly declines during the reproductive phase (days 1–4 of adulthood; Fig. 2c). There are only 3 matrisome-associated genes that increase in expression during aging (cpr-2, chil-14, lec-2; Fig. 2c). CPR-2 is the cathepsin B orthologue and might be involved in ECM degradation. Consistent with observations in mammals, tissue inhibitor of metalloproteases, TIMP-1, expression is also progressively lost during aging (Fig. 2c). Taken together, we found a general decline of matrisome gene expression during aging for the model organism C. elegans (Fig. 2c) that is similar to the progressive decline in collagen biosynthesis observed in human skin [10] and across species [8, 9]. A comprehensive assessment of not only the transcriptional matreotype but also of the protein levels in mammalian tissues during aging might be feasible by using an in situ decellularization protocol coupled with ECM proteomics [39]. This could reveal the longitudinal signature of the matreotype during aging.

Comparing the gene expression or protein levels at a given chronological age between wild type and long-lived animals revealed several molecular mechanisms at work, which were then experimentally assessed for their functional importance for longevity. For instance, reduced insulin/IGF-1 signaling upregulates genes involved in antimicrobial, oxidative stress and xenobiotic responses, protein homeostasis, and metabolism, all of which are required to promote healthy aging [40]. Reanalyzing the expression profile of these long-lived mutants with reduced insulin/IGF-1 signaling [9] revealed that almost one-fifth, that is, 79 of the 426 upregulate genes are matrisome genes [31]. Out of the 79 upregulated matrisome genes, 48 are collagens and 15 are ECM proteases (cathepsins, astacin-like metalloendopeptidases, MMPs) [31], suggesting an activation of collagen remodeling [9]. Such a mobilization of matrisome genes also occurs through other longevity-assurance pathways. Reanalyzing the proteomics dataset comparing long-lived germ stem cell mutants (glp-1) with wild-type C. elegans [41] revealed an increase of 177 proteins including 25 matrisome proteins in long-lived C. elegans [31]. The 25 matrisome proteins include 2 basement membrane-forming laminins, 10 collagens, 1 prolyl 4-hydroxylase (DPY-18), which is important for collagen stability, and 3 ECM-remodeling enzymes [31], suggesting an increase in ECM turnover and homeostasis. Surprisingly, I could not find any proteomics attempts comparing long-lived mice with wild type during aging. Whether longevity-assurance pathways in mammals alter the matreotype toward reactivating ECM homeostasis needs to be investigated. Taken together, based on the data from C. elegans, it appears that longevity-assurance pathways invest in collagen or ECM turnover to maintain a youthful matreotype.

The physiological state of a cell or tissue is reflected in a unique ECM composition or matreotype [2]. For instance, fibroblasts that become senescent or are dedifferentiated into myofibroblasts or cancer-associated fibroblast express a distinct set of ECM proteins. Based on the ECM signature or matreotype, it is even possible to identify tumor types [42]. Thus, distinct matreotypes could be developed into biomarkers or prognostic indicators for disease and health status with implications for personalized medicine. Targeting ECM-cell surface receptors might provide an entry point to remodel ECM and matreotype, since cell-surface receptors read-out ECM stiffness and ECM properties to reprogram cells. Biologics or antibodies that act only on the cellular surfaces could target and alter these ECM-cell surface receptors to change intracellular signaling and to induce the desired gene expression program. Currently, there are about 11 different targets being investigated in 27 clinical trials with primary end points specific to ECM stiffness [43] that might reprogram the matreotype.

Major efforts have revealed how proteins are maintained within cells and cellular compartments and how longevity interventions improve protein homeostasis during aging (Fig. 1). Here, I propose that collagen homeostasis or ECM turnover is a process that works efficiently when the organisms are young to maintain their somatic tissue. Since collagen biosynthesis is costly and energy-intensive, upon reproduction resources might be allocated to produce high-quality offsprings. Moreover, ECM turnover might be the repair mechanism to cleave-out, digest, and degrade damaged ECM proteins from the matrix. This requires cellular contact, proper mechanotransduction, and de novo synthesis of ECM components from cells. Damage to the ECM or to cells will start a vicious downward spiral of ECM fragmentation and cell detachment leading to a progressive decline in cellular and ECM homeostasis during aging. Longevity interventions might also maintain protein homeostasis of extracellular proteins reflected in changes of the matreotype (Fig. 1). Thus far, we have taken the first steps to define the matrisome and the matreotypes during aging and longevity (Fig. 2). Since the matreotype reflects the cellular, tissue, and disease status, quantifying and defining the matreotype could become a valuable biomarker for health assessment. Rejuvenating the matreotype might systemically rejuvenate cellular and tissue functions. Identifying druggable targets and understanding how to trigger rejuvenation of the aged matreotype have broad implications for clinical applications.

I thank Cyril Statzer for making the figures, Katrien De Bock, Gabriele Ewald, Nancy Hynes, and members of the Ewald lab for discussion and comments on the manuscript. My inspiration for the term matreotype came from reading about the matrisome by Alexandra Naba and Richard O. Hynes and from Ruedi Aebersold’s definition of proteotype. Furthermore, this work was also inspired by Fritz Verzár (1886–1979), the founder of this journal and a pioneer in investigating collagens in aging research (https://www.unibas.ch/en/Research/Uni-Nova/Uni-Nova-128/Uni-Nova-128-An-almost-forgotten-pioneer.html). Standing on the shoulders of giants; I apologize for omitting or not citing individual original work and simply referring to reviews due to reference limitation. This work was supported by the Swiss National Science Foundation (163898).