key: cord-0979596-v80epetm
authors: Lagoutte, Priscillia; Bettler, Emmanuel; Vadon-Le Goff, Sandrine; Moali, Catherine
title: Procollagen C-proteinase enhancer-1 (PCPE-1), a potential biomarker and therapeutic target for fibrosis
date: 2021-04-20
journal: Matrix Biol Plus
DOI: 10.1016/j.mbplus.2021.100062
sha: 7a3eb2e3b28d239f3cc34d500ac52db8f6c2ec22
doc_id: 979596
cord_uid: v80epetm

The correct balance between collagen synthesis and degradation is essential for almost every aspect of life, from development to healthy aging, reproduction and wound healing. When this balance is compromised by external or internal stress signals, it very often leads to disease as is the case in fibrotic conditions. Fibrosis occurs in the context of defective tissue repair and is characterized by the excessive, aberrant and debilitating deposition of fibril-forming collagens. Therefore, the numerous proteins involved in the biosynthesis of fibrillar collagens represent a potential and still underexploited source of therapeutic targets to prevent fibrosis. One such target is procollagen C-proteinase enhancer-1 (PCPE-1) which has the unique ability to accelerate procollagen maturation by BMP-1/tolloid-like proteinases (BTPs) and contributes to trigger collagen fibrillogenesis, without interfering with other BTP functions or the activities of other extracellular metalloproteinases. This role is achieved through a fine-tuned mechanism of action that is close to being elucidated and offers promising perspectives for drug design. Finally, the in vivo data accumulated in recent years also confirm that PCPE-1 overexpression is a general feature and early marker of fibrosis. In this review, we describe the results which presently support the driving role of PCPE-1 in fibrosis and discuss the questions that remain to be solved to validate its use as a biomarker or therapeutic target.

Fibrillar collagens are the most abundant proteins in the human body and their biosynthesis is a critical pathway controlling development, growth, homeostasis and aging. Fibrillar collagens and the numerous regulators of their synthesis and degradation also represent potential targets in several diseases, especially those derived from altered collagen turnover. The latter include but are not limited to defective tissue repair [1, 2] , cancer [3] and osteoarthritis [4] . Here, we focus on fibrosis, a general term referring to the common pathogenic outcome of a great diversity of uncon-trolled tissue repair responses leading to the aberrant accumulation of extracellular matrix (ECM) [5] . Collagens are the most abundant components of fibrotic lesions and they are known to be responsible for tissue stiffening and impaired function, potentially leading to death when a vital organ is affected. Despite the intense efforts of the scientific community to identify relevant and druggable protein targets to alleviate the deleterious effects of fibrosis, very few drugs have demonstrated an efficient and direct anti-fibrotic activity [5] . As a key regulator of collagen fibrillogenesis, the protein named procollagen C-proteinase enhancer-1 (PCPE-1) attracts more and more attention. PCPE-1 is a potent and specific enhancer of fibrillar collagen proteolytic maturation, the rate-limiting step in collagen fibrillogenesis. In this review, we show that in addition to the prominent role of PCPE-1 in collagen fibrillogenesis, its specific mechanism of action, its expression pattern, which closely matches that of collagen I, and the fact that it is relatively more dispensable to tissue homeostasis than most other players in collagen synthesis, also make it a very promising target for the treatment or prevention of fibrotic disorders.

Fibrillar collagens are usually classified into "major" (collagens I, II and III) and "minor" (collagens V and XI) collagen types based on their relative abundance. The common view is that minor fibrillar collagens co-assemble with major fibrillar collagens to regulate fibril growth [6, 7] . This classification does not take into account the more recently discovered members of the fibril-forming collagens (collagens XXIV and XXVII), which exhibit a more restricted expression pattern and produce fibrils with distinct characteristics [8, 9] . Collagens I, V and XI form heterotrimers composed of two to three different polypeptide (a) chains while collagens II and III form homotrimers composed of three identical chains [10, 11] . With the notable exception of normal cartilage ECM which mainly consists of fibrils made of collagens II and XI, the ECM of connective tissues is usually composed of heterotypic fibrils where collagens I, III and V are present in various proportions [10] . Collagen I is the major fibrillar component while collagen III is most abundant in blood vessels and collagen V in the cornea [7, 12] . All three collagen types are interdependent and the lack of one collagen has broad consequences on fibrillogenesis [13, 14] .

The various steps leading from individual collagen chains to fibrils are well described for major and minor collagens [15] . This pathway starts with the synthesis of collagen precursor molecules, also known as procollagens, which then undergo several post-translational modifications (e.g. proline and lysine hydroxylations) and assemble into trimers. At this stage, trimeric procollagen molecules are kept in soluble form through their charged Nand C-terminal domains called propeptides. During or after secretion into the extracellular environment, the propeptides are cleaved by specific proteolytic enzymes and fibrillogenesis can occur (Fig. 1A) . Fibril formation is supposed to be a spontaneous process, as observed in vitro [16] , but several additional protein partners are also involved in vivo to obtain the specific fibril characteristics that best fit to collagen tissue function. Among these partners, FACITs (fibril-associated collagens with interrupted triple helices) and SLRPs (small leucine-rich proteoglycans) play major roles by regulating fibril diameter, spacing and interactions [7, 17] . The final step of fibrillogenesis is the stabilization of the collagen network by the formation of intermolecular cross-links, a process initiated by lysyl oxidases LOX and LOXL [18] .

The proteolytic enzymes that cleave procollagen propeptides are known as procollagen N-and Cproteinases (PNPs and PCPs). Ten different proteinases, belonging to the astacin, ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) and SPC (subtilisin-like proprotein convertase) subgroups, can cleave at least one propeptide in at least one fibrillar collagen [19] [20] [21] [22] . Of specific interest here are the bone morphogenetic protein-1 (BMP-1)/tolloid-like proteinases or BTPs [22] . They form a group a four astacin-like proteinases comprising BMP-1 and mTLD (mammalian tolloid), two splicing variants of the same BMP1 gene, and mTLL-1 and -2 (mammalian tolloid-like 1 and 2) which are encoded by two additional genes (TLL1 and TLL2). These are the main enzymes in charge of the C-terminal maturation of fibrillar collagens [23] , the most important proteolytic event to trigger fibrillogenesis [16] ( Fig. 1A) . They can also cleave the N-termini of minor fibrillar procollagens V and XI, releasing the first thrombospondin-like N-terminal (TSPN) domain of their a1 chain but leaving most of their N-propeptide attached to the main triple helix [21, [24] [25] [26] . In addition, they process the SLRPs decorin, biglycan and mimecan/osteoglycin by removing a short sequence at their N-terminus [27] [28] [29] . Both partial collagen V and XI Npropeptide retention and SLRP binding to fibrils are recognized as key mechanisms to control fibril lateral growth [7] . Finally, BTPs process the lysyl oxidase members LOX and LOXL [30] [31] [32] , thereby promoting their activation and the formation of covalent cross-links between collagen monomers (Fig. 1A) .

PCPE-1 was first described by Adar and colleagues in 1986 as a contaminant in PCP preparations [33] where they observed that it was capable of significantly increasing procollagen I Cterminal processing. For this reason, the name procollagen C-proteinase enhancer (PCPE) is most often used while the other commonly found abbreviation PCOLCE is kept for the gene name. PCOLCE is localized to the same region of chromosome 7 that contains the COL1A2 gene [34] and leads to a 50 kDa glycoprotein composed of two Nterminal CUB (complement, Uegf, BMP-1) domains and one netrin-like (NTR) domain, separated by a long linker (Fig. 1B) . Homologues of PCPE-1 are rather ubiquitous in vertebrates and their amino acid sequences are well conserved with the exception of the linker sequence [34] [35] [36] . In human and mouse proteins, this linker is prone to proteolysis and can be cleaved in vitro by a wide range of proteases including trypsin [37] , plasmin [38] , MMP2 [39] and meprins [40] . Thus, proteolytic fragments corresponding to both CUB1CUB2 and NTR domains have been detected in vivo and in cell culture [33, 38, 41, 42] . In addition, proteolysis of human PCPE-1 within CUB1 was reported under the action of soluble ADAM17 [43] but this cleavage seems less predominant.

Interestingly, PCPE-1 is very specific to procollagen maturation and was found to stimulate the C-terminal processing of procollagens I, II and III by BTPs [33, 44, 45] . There are some hints that it could also promote the C-terminal maturation of the a2 chain of procollagen V in the supernatant of mouse embryonic fibroblasts [46] but this has never been confirmed in vitro and is theoretically not possible if the a2 chain is included in a heterotrimer, as will be discussed below. In contrast, the N-terminal removal of the TSPN domain of the procollagen V a1 chain by BTPs was shown to be enhanced by PCPE-1 [47] , although some less conclusive data had first been obtained, probably because of the presence of contaminating PCPE-1 in the procollagen preparation [44] . The extent of stimulation seems somewhat lower than for C-terminal processing events, although this has not been precisely quantified. Also, the mechanism of Nterminal enhancement is not well understood for the moment and it will not be discussed further here. Specificity to procollagen maturation is however a very striking feature of PCPE-1 and, despite the ever-growing list of reported BTP substrates [22, [48] [49] [50] , no other cleavage event seems to be affected by PCPE-1. The lack of effect of PCPE-1 was thus demonstrated for other BTP substrates involved in collagen fibrillogenesis such as osteoglycin and LOX but also for other ECM components processed by BTPs such as procollagen VII, laminin 332, dentin matrix protein-1 and dentin sialophosphoprotein or for growth factor antagonists which are inactivated by BTPs such as chordin [44, 51, 52] .

PCPE-1 expression is high in tissues rich in collagen I such as bones, tendons, skin or cornea and its abundance is lower when the collagen content is less dominant [34, 37, 53, 54] . In early studies, this protein was not detected in brain and liver and it was barely detected in heart and kidneys [37] but this was tempered in later studies which showed that PCOLCE RNA and its derived protein were relatively high in the heart [45, 55] . Interestingly and in contrast to what is described for Bmp1-and Tll1-null mice which die around birth or even before [25, 56, 57] , Pcolce-null mice are viable, fertile and do not show any obvious defects [46] . However, a closer inspection of bones and tendons revealed that their collagen fibrils were abnormal with irregular and fluted shapes [46] . In the left femora that were analysed by transmission electron microscopy, some broadening and branching of fibrils was also observed. This results in macroscopic alterations of cortical and trabecular bone geometries which are explained as a compensatory mechanism for inferior material properties [46] . The importance of the Pcolce gene for efficient tissue repair was also demonstrated in several models of corneal injury, all of which were characterized by a more or less pronounced delay in wound closure in KO mice [54] . These results will be discussed in more detail below but, overall, they seem to relate to a relatively mild phenotype compared to the effect of mutations in genes encoding collagen chains, BMP-1 or other enzymes involved in collagen synthesis [58] . Notably, no human genetic disease seems to be associated with mutations in PCOLCE.

Another human PCPE protein has been described in 2000 which shares 43% identity with PCPE-1 and has the same domain composition (Fig. 1B) . It was called PCPE-2 and the corresponding gene PCOLCE2 [45, 59] . From the few studies that have analysed PCPE-2 activity, it seems to be endowed with the ability to increase the C-terminal processing of procollagen II by BMP-1 [45] but less clearly of procollagen I [60] . PCPE-2 has also been linked to high density lipoprotein (HDL)-cholesteryl ester uptake and atherosclerosis [61, 62] . Finally, it was found that Pcolce2-null mice are protected against cardiac fibrosis in a model of transverse aortic constriction [63] . However, since the data about its PCP or PNP enhancing activity and its possible involvement in fibrosis are scarce, we have decided to focus here on PCPE-1.

Structure and mechanism of action of PCPE-1 PCPE-1 is a secreted protein containing multiple post-translational modifications [34, 64] . It is decorated with sialylated O-and N-glycosylations that may differ in healthy and diseased states, resulting in a complex migration pattern in isoelectric focusing (IEF) experiments [45, 65, 66] . Two Nglycosylation sites are predicted but only the most N-terminal one seems functional, its absence inducing lower protein stability [67] but having no impact on PCPE-1 activity [68] . The mature protein also contains 14 cysteines and correct disulfide bonding is necessary for activity [69] . Other modifications such as phosphorylations or myristoylations have been detected or predicted [59, 70] and may contribute to the complex pattern observed in IEF.

PCPE-1 is a monomer in solution and the first low-resolution structural data showed an elongated molecule of about 150 A in length [71] . This finding was somewhat unexpected since CUB and NTR domains have very different isoelec- After secretion, procollagens remain soluble until they are processed by N-and C-proteinases which remove their N-and C-propeptides (NPs, CPs). The BTPs play a major role in these maturation steps as they are in charge of cleaving the C-propeptides of collagens I, II and III. They can also process the C-terminus of the a1 and a2 chains of procollagens V and XI, although SPCs might act before on the a1 chains. Finally, BTPs mature the N-terminus of the a1 chain of procollagens V and XI but this proteolytic event preserves a large part of their N-propeptide which, after incorporation into fibrils, protrudes from the fibril surface and inhibits further fibril lateral growth. PCPE-1 enhances the C-terminal maturation of collagens I, II and III and, to a lesser extent, the N-terminal maturation of the a1 chain of collagen V. BTPs control other aspects of collagen fibrillogenesis through the proteolytic maturation of decorin, biglycan and osteoglycin (SLRPs) which interact with collagen fibrils and regulate their diameter and spacing. They also activate the lysyl oxidases LOX and LOXL, thereby triggering the formation of cross-links between collagen monomers to increase fibril stability. (B) Domain composition of the BTPs and PCPEs. BMP-1 and mTLD are splice variants differing by the presence of a short C-terminal sequence in BMP-1 and of one additional EGF and two additional CUB domains in mTLD. mTLL-1 and mTLL-2 domain structures are similar to that of mTLD. All BTPs are synthesized as proenzymes and are activated by SPCs (site indicated by arrowhead) in the secretory pathway. The catalytic activity is conferred by the astacin-like domain and other domains contribute to regulate substrate specificity. PCPE-1 and PCPE-2 have the same domain composition with two N-terminal CUB domains and a C-terminal NTR domain connected by a long linker. The CUB domains of PCPE-1 can be proteolytically released (site indicated by arrowhead) and retain full enhancing activity in vitro. P = propeptide.

3 tric points (around 5 for individual CUB domains and 9.5 for NTR) whereas the full-length protein is rather neutral [65, 69] , suggesting possible interactions between the two ends of the molecule.

The original purification protocol of PCPE-1 from mouse fibroblast supernatants [33, 64] involved the use of a Sepharose column coupled to the Cpropeptide of procollagen I (CPI). PCPE-1 was coeluted from this column with the so-called "PCP", already suggesting a strong affinity for CPI and/or the protease. Interestingly, although this procedure also enabled the co-purification of a shorter 36 kDa fragment (corresponding to the CUB domains, Fig. 1B ), only the latter and not the full-length PCPE-1 was separated from the protease by size exclusion chromatography [33, 64, 72] . This was the first indication that the NTR domain of PCPE-1 was probably necessary for the interaction of the enhancer with the protease, a finding which was later confirmed by surface plasmon resonance experiments [73] even though the interaction remains weak (K D > 100 nM).

PCPE-1 is devoid of catalytic activity on its own but, as already mentioned above, is able to specifically stimulate CP removal from fibrillar procollagens I, II or III [44, 45, 64] by all four BTPs [25, 45, 74, 75] . PCPE-1 has nanomolar affinity for procollagen molecules [76] [77] [78] and kinetic studies have shown that it can increase the catalytic efficiency of BMP-1 by a factor of 15 on procollagen I and by a factor of 12 on a mini-procollagen III [68, 79] . Of note, this increase in catalytic efficiency derives from both the increased affinity of the enzyme for the procollagen substrate and the increased turnover number of the hydrolysis reaction.

Several studies demonstrated that the enhancing activity of PCPE-1 is provided by its CUB domains [33, 64, 75, 76] and is due to a direct and strong calcium-dependent interaction with the procollagen CPs (K D < 30 nM) [75] [76] [77] [78] . Through mutagenesis studies with deletion and domain-swapping mutants, it was then observed that contiguous CUB1 and CUB2 domains are both necessary and sufficient for PCPE-1 activity [76] . This led to the isolation of the complex formed by CUB1CUB2 and CPIII which was first analysed by small angle X-ray scattering (SAXS) [68] . The derived lowresolution structure of the complex revealed that CUB1CUB2 binds to the N-terminal/"stalk" region of CPIII [80] , close to the BTP cleavage site, and confirmed the previously evidenced and intriguing 1:1 (PCPE-1 monomer:procollagen trimer) stoichiometry [35, 69, 77] . The recent crystal structure of the same complex [81] gave additional information and unveiled the key determinants of the mechanism of action of PCPE-1 ( Fig. 2A) . It shows that a single PCPE-1 molecule winds around the CPIII trimer with each CUB domain interacting mainly with a specific procollagen chain (e.g. in Fig. 2 , CUB1 with chain A and CUB2 with chain B). The complex is stabilized around the two calcium binding sites of CUB1 and CUB2 through interactions with two conserved and distinct CP lysines (one in each procollagen chain; upper left inset in Fig. 2A ) that are essential for PCPE-1 binding, and therefore, activity [35, 68, 81] . Importantly, this mode of action mobilizes two procollagen chains simultaneously, precluding the binding of a second PCPE-1 molecule. Also, it is interesting to note that the conserved lysines are found in all chains of major fibrillar procollagens and in the a2 chain of procollagen V but not in other chains of minor fibrillar procollagens [80] , making it unlikely that PCPE-1 modulates their Cterminal maturation by BTPs. However, it cannot be excluded that an alternate mechanism is at play for these collagens.

The structure also shows that upon CUB1CUB2 binding, the stalk region of the CPIII trimer is distorted, with the end of chain A being "pulled" into the groove between the CUB domains. The BMP-1 active site being too small to accommodate more than one procollagen chain [82] , this loss of symmetry offers two possibilities for the protease to bind to the procollagen/PCPE-1 complex. Separated from chain A, chain B or C could become more accessible to enter the enzyme active site ( Fig. 2B and C) or the BMP-1 catalytic domain could bind directly to the locally unraveled chain A ( Fig. 2A) . Molecular modelling favors this last mechanism since it would create a new interaction surface between PCPE-1 and the protease [81] and contribute to the previously observed decrease in K M [68, 79] . Once the first chain is cleaved, the sequential cleavage of the remaining chains is probably facilitated, independently of the presence of PCPE-1.

It should be noted that the current model does not include BMP-1 or tolloid non-catalytic domains. Their exact role remains to be determined but BMP-1 CUB and EGF domains have been shown to be necessary for full enhancement by PCPE-1 ([51,52,74] and our unpublished results). This implies that a direct interaction between PCPE-1 and the protease is involved at one stage of the catalytic process. If the interaction between BMP-1 and PCPE-1 is actually weak for the full-length enhancer and null for CUB1CUB2 [68, 73, 74] , it is probably strengthened in the presence of the procollagen molecule even if the exact details of this three-partner interaction are presently unknown. The involvement of the protease auxiliary domains is also supported by the fact that PCPE-1 does not enhance the cleavage of procollagen III by another family of PCPs, the meprins [20, 40] , that have a similar astacin-like catalytic domain but no CUB or EGF domains [40] . Instead, PCPE-1 inhibits meprin-mediated release of CPIII, most probably because its binding to procollagen III hinders protease access to the cleavage site.

The role of the NTR domain of PCPE-1 is less directly connected to the enhancing mechanism but is still very important. This domain is homologous to tissue inhibitors of metalloproteinases (TIMPs) [83] and it can be released proteolytically. However, an early report suggesting that it was a weak inhibitor of matrix metalloproteinases [38] was not confirmed by later studies [36, 73] . In contrast, the NTR domain can strongly interact with heparin and heparan sulfate proteoglycans (HSPGs), i.e. syndecans-1, -2 and -4, and mediates PCPE-1 attachment to the cell surface [69, 73, 84, 85] . In vitro, the presence of heparin induces a "super-stimulation" of procollagen processing, through the increase of PCPE-1 enhancing activity (around two-fold), but only when the NTR domain is attached to the CUB domains [73] . Since procollagens and BTPs are also able to interact with HSPGs, PCPE-1 may contribute to forming large protein complexes at the cell surface or in the Fig. 2 . Current models of the procollagen C-terminal maturation complex. Starting from the crystal structure of CPIII in complex with the CUB1CUB2 region of PCPE-1 (PDB code: 6FZW; [81] ), modelling was used to extend chains A (in pink), B (in yellow) and C (in grey) of procollagen III towards the C-telopeptide region and the collagen triple helix (see [81] for details). Full-length PCPE-1 (in green) was rebuilt using YASARA (http://www.yasara.org) for homology modeling, based on Uniprot/Swiss-Prot sequence Q15113, PDB code 6FZW for CUB1CUB2 and PDB code 1UAP for the NTR domain. YASARA was also used to successively model the BMP-1 catalytic domain (BMP-1 cat; PDB code: 6BSL, in blue) in interaction with chain A (A), chain B (B) or chain C (C) of procollagen III. Cyan spheres represent calcium ions and yellow spheres are zinc ions. Rendering was obtained using UCSF ChimeraX [178] . In (A), the inset shows a close-up view of the interface between PCPE-1 and procollagen III. The two procollagen lysines (K1 and K2) involved in the interaction are highlighted in grey. As discussed in the text, the model shown in (A) is the most likely. Note that the NTR domain of PCPE-1 is shown to give a full view of the protein and an idea of the volume occupied by this domain but it is not known how it could interact with other partners in the complex. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) matrix, where anchoring through its NTR domain would allow optimal orientation of its CUB domains and maximal activity. In line with this, PCPE-1 was also found to interact with fibronectin [84] , fibulin-4 [86] and thrombospondin-1 [87] , all of which are involved in collagen supramolecular maturation.

Enhancement of PCP activity could also be amplified in the presence of secreted frizzled related protein 2 (sFRP2). sFRP2 has been described as another enhancer of procollagen I cleavage by BTPs [88] and, even if the activity of this protein alone remains highly controversial [52, 89, 90] , it was recently reported that sFRP2 is able to interact with mouse PCPE-1 and that both proteins could act in synergy to increase procollagen processing in vitro [91] . Indeed, while Sfrp2 knock-out mice are viable and do not show any major defects [92] , the simultaneous knockdown of Sfrp2 and Pcolce is embryonic lethal and cultures of mouse embryonic fibroblasts derived from these mice show a significant decrease in collagen maturation. Similarly, double knockdown of Sfrp2 and Pcolce in zebrafish led to a strongly dorsalized phenotype with decreased procollagen processing and survival limited to 72 h. The mechanism is unclear but seems to involve the CUB domains of PCPE-1 and the frizzled domain of sFRP2. Other partners have been reported for PCPE-1, including b2microglobulin [93] , laminin-111, endostatin (a fragment of collagen XVIII), collagen IV and collagen VI [87] but the functional implications of these interactions still require further investigation.

Finally, it was recently shown, in fibroblasts, that the production, secretion and cell surface localization of PCPE-1 are boosted by ascorbic acid (vitamin C) [94] , an essential co-factor for correct hydroxylation, folding and secretion of procollagens. In the absence of evidence for the presence of hydroxyprolines in PCPE-1, the authors suggest that the effect of ascorbic acid on PCPE-1 is indirect and could derive from its enhanced co-secretion with procollagens. Such an early association would facilitate efficient subsequent extracellular procollagen maturation. Whether PCPE-1 could also act as a procollagen chaperone is more debatable, as an early report showing that PCPE-1 binds to the triple helix [45] could not be confirmed [78] and it is not known if its sole binding to the CP of procollagens [78] could have a significant stabilizing effect. Also, the presence of PCPE-1 or CUB1CUB2 during fibrillogenesis does not seem to affect the kinetics of fibril formation nor the type of fibrils that are formed [46, 75] , indicating that the role of the enhancer probably ends after procollagen processing.

Involvement of PCPE-1 in tissue repair and fibrosis

As described above, fibrosis is mainly defined by the aberrant tissue accumulation of extracellular matrix components [5, 95, 96] , among which fibrillar collagens I and III are the most abundant. It can be triggered by a great diversity of acute or chronic injuries, ranging from apparently benign skin wounds, leading to hypertrophic scars after a few weeks, to chronic hypertension causing myocardial fibrosis and heart failure after several years of silent remodelling. If fibrosis can be a beneficial reparative component of wound healing and tissue repair in the initial phase, chronic and uncontrolled fibrosis becomes highly pathogenic and impairs normal organ function [97] . Main players in fibrosis are myofibroblasts which derive from different cell types depending on the organ but share similar features of high contractility and ability to secrete large amounts of ECM proteins [98, 99] . These cells are required to allow tissue repair but due to persistent activating mechanical and biochemical factors, they fail to leave the injured area, survive beyond the normal phase of tissue healing and continue to synthesize ECM.

Fibrosis principally refers to excessive ECM accumulation in internal organs such as heart, lung, kidney or liver. However, in tissues where the ECM content is already high in normal conditions such as skin and cornea, fibrosis can also reflect the abnormal organization of the ECM network with no or a limited increase in total ECM content. As a specific and potent regulator of collagen fibrillogenesis, PCPE-1 has early been considered as a potential contributor to fibrotic disorders [100] [101] [102] . However, it is only in recent years that the accumulation of data from various types of patient samples and animal models has made it evident that there is a strong link between fibrosis and PCPE-1 overexpression. These results are described below and show that up-regulation of PCPE-1 can be observed in all major organs known to develop fibrosis.

Cardiac fibrosis is associated with most cardiovascular diseases (CVDs) and is one of the main factors that predispose to heart failure [103, 104] . In the heart, myocardial stiffness leads to mechanical, electrical and vasomotor dysfunctions [105] which are typical features of heart failure. Myocardial infarction (MI), hypertension, coronary artery disease and valvular lesions are examples of CVD that trigger a tissue response to injury and often lead to aberrant tissue remodelling with excessive accumulation of ECM. While in some CVDs such as MI, damaged cells are replaced by ECM and a scar is formed through a mechanism often referred to as replacement (reparative) fibrosis, in other conditions affecting the heart such as dilated cardiomyopathy or hypertension, it is more a reactive perivascular and interstitial fibrosis which develops, i.e. the diffuse accumulation of collagen between myocytes [103, 106] . Although the pathophysiological mechanisms leading to fibrosis (replacement or reactive) may differ, common effectors are involved. These include fibroblasts and myofibroblasts, components of the reninangiotensin-aldosterone system and the transforming growth factor-b1 (TGF-b1) [96, 104, 107] .

In this context, it is interesting to note that treatment of adult rat heart fibroblasts with 2.5 ng/ ml TGF-b1 was shown to increase the amount of PCPE-1 in culture medium by a factor around 2 [102] . In the same study, Pcolce mRNA and secretion of PCPE-1 protein were found to be modestly but significantly increased (around 1.5-fold) in the presence of 10 nM aldosterone. Also, spironolactone, an aldosterone receptor antagonist, restored the basal level of Pcolce mRNA which was upregulated 3-fold in a rat model of myocardial infarction after coronary artery ligation [100] . Normalization in the presence of spironolactone was also observed at the protein level, confirming that aldosterone directly stimulates PCPE-1 expression. Following this seminal study establishing that PCPE-1 expression is induced after MI, this finding was confirmed in other reports [108, 109] . Atorvastatin, a cholesterol-lowering agent among the most prescribed drugs in the world, also proved its efficacy in counteracting the increased expression of PCPE-1 in a rat model of MI [109] . Similarly, rosuvastatin reduced myocardial fibrosis and Pcolce mRNA in a mouse model of metabolic syndrome induced by deficiency of leptin and LDL receptor [110] .

Pressure overload secondary to chronic hypertension is another well-described cause of cardiac fibrosis. It can be induced in animal models by a continuous subcutaneous infusion of angiotensin II, a potent vasoconstrictor and upstream regulator of aldosterone production. After the first reports showing that PCPE-1 is more abundant in mouse heart after angiotensin II treatment [111, 112] , the relationship between angiotensin II, cardiac fibrosis and PCPE-1 induction was recently confirmed in a more extensive study combining the effect of angiotensin II with a NO synthase inhibitor in rats [113] . In this model, PCPE-1 protein was increased 6-fold at the heart tissue level and to a lesser extent in the circulation. It was also co-localized with fibrotic areas in the myocardium. Notably, both in this study and in the model of MI described above, there was an excellent correlation between the PCPE-1 protein quantified in the heart by immunoblotting and the amount of collagen I measured through immunoblotting or Sirius Red staining (r = 0.82 and 0.92) [100, 113] .

Another leading cause of reactive fibrosis with chronic hypertension is aortic valve stenosis (AS), the most common valvular disease in Western countries. Of great interest to establish the link between this disease and PCPE-1 was the study by Beaumont and colleagues [114] who analysed myocardial samples from the interventricular septum in a cohort of 28 AS patients versus 10 control samples by western-blotting. Although the data are not shown in the manuscript, these authors report a marked increase in the PCPE-1 levels of AS patients (11-fold) . When the same patients were stratified between severe and less severe AS, PCPE-1 was also more abundant in the most severe cases. Finally, PCPE-1 appeared to be wellcorrelated with the collagen volume fraction and to a lesser extent with TGF-b expression in the same samples. In agreement with the induction of PCPE-1 in the context of left ventricular pressure overload, Pcolce mRNA was also shown to be elevated in murine models of transverse aortic constriction [115] or ascending aortic banding leading to dilated cardiomyopathy [116] .

More generally, PCPE-1 has also been used to assess myocardial ECM remodelling, most of the time together with collagens I and III and with the cross-linking enzyme lysyl oxidase, and its upregulation was for example demonstrated when MI in sedentary rats was compared to MI in exercised rats [108] , after subtotal nephrectomy in rats [117] or after ischemia-reperfusion injury in mice [118] and pigs [119] . All these results demonstrate that PCPE-1 overexpression broadly overlaps collagen overexpression and suggest a central role of PCPE-1 in both replacement and interstitial fibrosis of the heart.

The second most well-studied fibrotic condition in relation with PCPE-1 is hepatic fibrosis. In the liver, the main causes of fibrosis are chronic viral infections, drug or alcohol consumption and metabolic disorders such as NASH (nonalcoholic steatohepatitis) which can elicit chronic inflammation and lead to the activation of effector cells, mainly hepatic stellate cells (HSCs) [120, 121] . In mouse NASH liver for example, the fibrogenic pathway was shown to be enriched with a more than 3-fold increase in COL1A1, COL1A2, COL3A1 and PCOLCE transcripts [122, 123] . Liver fibrosis often develops into cirrhosis with several life-threatening complications such as portal hypertension. In 1997, Ogata et al. carried out the first experiments suggesting a link between liver fibrosis and elevated PCPE-1 expression. They showed that HSCs derived from cirrhotic rat livers, obtained after 7 weeks of CCl 4 intraperitoneal injections, had higher Pcolce RNA content than normal liver HSCs [101] . A similar observation was made in endothelial cells freshly isolated from cirrhotic rats [124] . These findings were confirmed directly on cirrhotic livers where RNA and protein were detected [101] whereas PCPE-1 remained undetectable in normal liver [34, 37, 101] .

More recently, the expression of PCPE-1 was monitored over time in mice both during the onset of fibrosis, over the 6 weeks of CCl 4 treatment, and during liver recovery after the treatment was stopped [125] . This analysis revealed that the amounts of circulating and tissue PCPE-1 closely reflected fibrosis severity with a 9-fold increase in tissue level quantified by western-blotting at the end of CCl 4 injections and a peak in PCPE-1 plasma concentration, about 60% higher than normal, at the same time point. Furthermore, PCPE-1 expression fell to basal levels at the end of the recovery period, showing the reversibility of the up-regulation upon termination of the pro-fibrotic stimulus. In another study in rats, comparing oral administration of CCl 4 or two other pro-fibrotic chemicals for a short period of 5 days with compounds known to cause non-fibrotic liver injury, Pcolce was found to be the most differentially expressed gene among a set of 67 selected fibrogenic genes [126] . This result suggested that its expression pattern was specific to the fibrotic phenotype and that its up-regulation could be detected early after the induction of the fibrogenic response. As in cardiac fibrosis, Pcolce is beginning to be used to evaluate the fibrotic status of the liver and the efficacy of anti-fibrotic compounds in various preclinical models [127, 128] .

Despite the rather consistent results obtained in animal models, conclusions from the few available studies dealing with patient samples are presently rather confusing. A first study comparing PCPE-1 plasma concentration in five healthy controls and five patients with liver fibrosis caused by hepatitis B or C infection showed that the mean level of circulating PCPE-1 was higher in the liver fibrosis group, although this remained below statistical significance due to the small size of the cohort [129] . In contrast, compared to healthy individuals, the plasma concentration of PCPE-1 was lower in a cohort of 126 patients diagnosed with chronic hepatitis B for at least 6 months, thus indicating a negative correlation with fibrosis stage [130] . In addition, there was a huge difference in measured PCPE-1 concentrations between the two studies (basal levels in control subjects around 300 ng/ml in [129] , in agreement with most other measurements of PCPE-1 plasma concentration [66, 131] , versus 4.5 ng/ml in [130] ) and this questions the reliability of some of the commercial ELISA kits used to detect PCPE-1.

Fibrosis is also a characteristic of many respiratory diseases induced by environmental pollutants, infections, including tuberculosis [132] and COVID-19 [133] , immune diseases or certain medications. Another common type of lung fibrosis is idiopathic pulmonary fibrosis (IPF), a chronic and progressive interstitial lung disease of unknown origin which is often fatal with a survival time of less than 5 years after diagnosis [134] . It is characterized by alterations of cellular composition and ECM accumulation with patchy scarring that cause irreversible damage to lung structure and function leading to respiratory failure [135] .

In recent years, the matrisome of normal and diseased lung has been extensively described [136, 137] and PCPE-1 was found to be a component of normal lung. The most complete set of data concerning its expression in lung fibrosis derives from the proteomic study performed by Schiller and colleagues in a classical mouse model of lung fibrosis induced by bleomycin [136] . The first interesting finding was that the solubility profile of PCPE-1 was significantly altered upon bleomycin treatment, with a global increase in protein solubility, suggesting a loss of its potential binding partners [87] in the fibrotic lung ECM. Despite this increased solubility, PCPE-1 was not detected in the bronchoalveolar lavage fluid (BALF) proteome of these mice. In terms of kinetics, the abundance of PCPE-1 was shown to peak between days 3 and 14 after bleomycin administration, corresponding to the fibrogenic phase, and then to diminish during the recovery phase. The maximum foldchange (2.7) was somewhat lower than in the models of cardiac and liver fibrosis described above, a characteristic that was also observed in patients when freshly extracted IPF fibroblasts were compared to normal lung fibroblasts [138] or when IPF lung biopsies were compared to control lung biopsies at the RNA [139] or protein level [140] . This suggests that PCPE-1 might be less directly necessary to the progression of pulmonary fibrosis than in cardiac or liver diseases.

Chronic kidney disease (CKD) is mainly a consequence of diabetes, hypertension or glomerulonephritis and is highly prevalent in the adult population. A common pathway in CKD is fibrosis which can ultimately lead to end-stage kidney failure with ECM accumulating in the glomeruli (glomerulosclerosis) and/or in the tubulo-interstitial space (renal interstitial fibrosis) [141] . PCPE-1 expression in the kidney is low in normal conditions [34, 37] but seems to increase with the progression of CKD.

First, the PCOLCE gene was found to be located in close proximity to a single nucleotide polymorphism associated with the development of diabetic kidney disease and was included in a panel of 226 transcripts for which correlation with kidney function (quantified through the estimated glomerular filtration rate or eGFR) was analysed [142] . A moderate but significant inverse correlation was observed between PCOLCE expression and eGFR in the glomeruli, indicating increased expression of PCPE-1 when kidney function was decreased. In contrast, there was no correlation between PCOLCE transcript and kidney function in the tubules. Furthermore, the mean serum level of PCPE-1 measured by ELISA was 76% higher in a separate cohort of 131 CKD patients compared to 34 healthy controls [131] . Of note, the highest recorded serum concentration was 325 ng/ml in the control group and 840 ng/ml in the patient group albeit that large variations were observed in both groups. Finally, circulating PCPE-1 was also inversely correlated with glomerular filtration rate and positively correlated with body mass index. What remains to be established to consolidate present data is a direct link between PCPE-1 levels and histological evidence of kidney fibrosis.

Like all the organs described above, ligaments and muscles can be affected by injury and activate a wound healing response. In some cases, healing will result in scar formation and fibrosis. Thus, the semi-quantitative mRNA analysis of some targets related to collagen fibrillogenesis in a rabbit model of ligament injury and scarring showed that the PCOLCE transcript had the highest level of up-regulation after 3 weeks when compared to BMP1, LOX and TGFB1 transcripts [143] . It then went down but was still somewhat higher than normal levels after 14 weeks, a behaviour that differed substantially from the progression observed during normal ligament development and could explain why ligaments often fail to regenerate.

In the field of inherited muscle diseases, Duchenne muscular dystrophy (DMD) is known to lead to skeletal muscle degeneration and necrosis, inflammation and fibrotic remodelling [144] . Interestingly, PCPE-1 expression has been well characterized in a widely-used model of DMD, the mdx mouse which has a spontaneous mutation in the dystrophin gene and develops a mild form of the disease. In the fibrotic diaphragms of the mdx mice collected at 4 and 8.5 months of age, there was an overall increase in PCPE-1, detected by immunoblotting and immunofluorescence, which was not quantified but seemed to follow the marked increase in collagen I [125] . Notably and similarly to the observations made in models of cardiac and liver fibrosis [100, 113, 125] , PCPE-1 and collagen I were also co-localized in diaphragm tissue. Finally, this study showed that the elevation of PCPE-1 levels in mdx mouse plasma was comparable to that observed for the N-propeptide of collagen III (PIIINP), an established marker of fibrosis [145] .

Another group analysed the link between PCPE-1 and muscle fibrosis in OPMD (oculopharyngeal muscular dystrophy), a genetic disorder first affecting muscles of the eyelid, throat and proximal limbs due to a polyalanine expansion at the N-terminus of the PABPN1 (poly(A)-binding protein nuclear 1) protein. The results point to a completely novel mechanism by which mutant PABPN1 entraps PCPE-1 in nuclear aggregates leading to its exhaustion from the ECM [146] . ECM depletion and nuclear co-localization with PABPN1 aggregates can be observed in symptomatic OMPD patient muscle biopsies and in a mouse model of the disease. These observations are accompanied by a significant reduction in PCOLCE/Pcolce mRNA in symptomatic but not pre-symptomatic OPMD patients/mice when compared to age-matched controls. Importantly, the mechanism of PCPE-1 sequestration in the nucleus seems to be shared with another inherited muscle disease, myotonic dystrophy type 1, but not with DMD and facioscapulohumeral muscular dystrophy [146] . This study reveals that disease-specific mechanisms can also play a role and challenge the general trend of PCPE-1 overexpression in fibrosis-related conditions.

The cornea can also be exposed to different types of injuries such as refractive surgery, physical traumas, burns or infections. Due to the very specific organization of the collagen network required to maintain corneal transparency [7] , defects in corneal wound healing lead to persistent non-transparent scars associated with a loss of visual acuity [147] . Soon after its discovery, PCPE-1 was shown to be highly expressed in the cornea [37] and, more recently, Malecaze et al. demonstrated that it was further increased at the RNA level between days 14 and 28 in a murine model of full-thickness excision leading to scarring, both in the corneal stroma and epithelium [53] . The protein detected by immunostaining also seemed substantially more abundant in the stroma, a trend that was qualitatively confirmed in four human corneal scars of various etiologies [53] . A subsequent study extended these results to other murine models of corneal injuries [54] . Five days after an alkali burn, PCPE-1 seemed rather decreased in the wound area but was highly concentrated at the epithelial-stromal junction juxtaposed to the burn zone. Another model of epithelial abrasion led to a stronger immunofluorescence signal for PCPE-1 in the stroma at day 5 but, intriguingly, this was preceded by the transient accumulation of the protein just beneath the basement membrane between days 1 and 3. In these various models, the lack of Pcolce did not change the grade of the opacity but delayed epithelial closure and neutrophil recruitment and promoted angiogenesis [54] . The mechanisms underlying these observations are presently not understood.

Hypertrophic scars (HTSs) and keloids are the most common manifestations of skin fibrosis. They occur after a wound healing response has been triggered, when activating signals fail to disappear. Whereas normal scars usually appear as fine and flat lines and regain up to 80% of original skin tensile strength, HTSs are raised, erythematous, sometimes pruritic and painful, and can cause contractures affecting mobility. However, they remain restricted to injury area in contrast to keloid scars which can grow outside injury boundaries and evolve into benign dermal tumors that usually recur after excision [148, 149] . Like in other fibrotic disorders, TGF-b signaling is largely demonstrated to be a critical regulator of skin scarring [150, 151] . In this context, it is interesting to note that the addition of 10 ng/ml TGF-b1 to normal human dermal fibroblasts had no effect on PCPE-1 synthesis ( [152] and our unpublished results), in contrast to the positive effect described for heart fibroblasts [102] , suggesting that the effect of the growth factor on fibroblasts depends on tissue origin. To the best of our knowledge, the effect of TGF-b1 on the synthesis of PCPE-1 by HTS and keloid fibroblasts has never been directly tested but there are some data suggesting that PCPE-1 could still be an important driver of collagen aberrant accumulation in skin scars. First, secreted PCPE-1 protein seems to be significantly more abundant (more than ten-fold) in cultures of HTS fibroblasts, as assessed by immunoblotting and compared to normal fibroblasts [153] . Second, PCOLCE mRNA was up-regulated in skin biopsies from HTSs and keloids [154, 155] . Finally, treatment of fibroblasts extracted from the same type of skin scars by rapamycin, an mTOR inhibitor known to hinder fibroblast proliferation and ECM/collagen synthesis, decreased PCOLCE transcription between 58 and 69% in normal, HTS and keloid fibroblasts [155] .

The studies described above suggest that PCPE-1 could be an excellent tissue marker to reveal fibrotic lesions. Tissue biopsies remain widely used for diagnosis and disease monitoring in the field of fibrosis-related diseases despite their invasiveness and possible sampling errors. However, there are already established collagenspecific histological methods to detect fibrosis in tissue sections and it is unlikely that PCPE-1 immunostaining would bring any added-value. More promising therefore is the development of methods allowing its detection through noninvasive approaches such as quantitation in easily-accessible body fluids. The search for such new biomarkers in the field of fibrosis is extremely active, either because reliable and sensitive markers are presently missing or because they are not specific or insufficient. In this context, PCPE-1 has a strong potential even if several steps will have to be validated before its clinical use can be considered.

What is presently well established is that PCPE-1 is found in human serum and plasma and can be easily detected and quantified using several techniques such as immunoblotting, ELISA and mass spectrometry [66, 113, 129, 131, 156] . According to present literature and public databases, PCPE-1 is also found in cerebrospinal fluid [157] [158] [159] [160] but almost undetectable in saliva, urine or broncho-alveolar fluid. Even though the most com-prehensive studies establishing the relationship between the severity of tissue fibrosis and circulating levels of PCPE-1 were performed in murine models [113, 125, 126] , measuring PCPE-1 in human plasma has also revealed elevated concentrations in diseased individuals with a strong suspicion of liver or kidney fibrosis [129, 131] . Moreover, there are some good hints that PCPE-1 monitoring allows the early detection of fibrogenic/pre-fibrotic stages and is closely correlated with fibrosis severity [125, 126, 136] . The most urgent need is hence to strengthen patient studies by the orthogonal evaluation of tissue fibrosis and to extend them to larger cohorts or new fibrotic disorders.

While an ideal biomarker should be specific to the targeted disease, it is likely that several concomitant conditions occurring in parallel to fibrosis will affect PCPE-1 blood levels. This is the case for closelyrelated collagen metabolites (collagen I and III propeptides or telopeptides) that show variations associated with age, sex, bone remodelling or pregnancy [161] [162] [163] . Whether PCPE-1 circulating levels display the same variations remains to be analysed but it is already known that bone disorders can affect its concentration and glycosylation pattern in human serum [66] . However, it should be noted that truly specific biomarkers of fibrosis are very rare and combinations of biomarkers are often considered to be more reliable [164] [165] [166] [167] . In summary, PCPE-1 represents a promising biomarker of fibrosis, alone or in combination with other biomarkers, but a more extensive evaluation in fibrotic patient samples is required before its detection can be implemented in clinical practice.

As broadly exemplified above, PCPE-1 is found in larger amounts in fibrotic tissues than in normal tissues, both at the RNA and protein levels, with very few exceptions. This does not show whether PCPE-1 overexpression is a causative factor of fibrosis or just an indirect consequence of the fibrotic environment. However, current knowledge about its direct contribution to the last step of procollagen maturation just preceding fibril formation, its specificity for collagen-related processes, its co-localization with collagen in several types of fibrotic scars and its mechanism of action requiring a 1:1 ratio with procollagen for maximal enhancement, all point at a direct and prominent contribution to aberrant collagen deposition in fibrosis. Curiously, the effect of the absence of PCPE-1 has only been studied in models of corneal scarring and no clear positive effect on collagen abundance or corneal transparency could be evidenced [54] . Due to the high content of collagen in cornea, this might not be the best model to establish that PCPE-1 depletion or inhibition can significantly affect further collagen integration in ECM and other models dealing with less collagen-rich organs might be more informative. Thus, what is critically missing at this point is the combination of a model of cardiac, kidney or liver fibrosis with PCPE-1 overexpression, deficiency or inhibition to demonstrate that the modulation of PCPE-1 activity can significantly impact the collagen volume fraction and the function of these organs.

Nevertheless, we can already speculate that a great advantage of targeting PCPE-1 in reparative fibrosis processes would be that BTPs will remain active at their basal level and can ensure that a minimum amount of properly matured collagen fibrils continue to be synthesized to repair the injured tissue. Indeed, the complete blockade of collagen deposition is known to have detrimental effects on normal and injured adult tissues, as demonstrated by the analysis of the conditional double Bmp1/Tll1 knock-out mice [168] [169] [170] . Very illustrative is the fact that BTP ablation in young adult mice leads to a progressive reduction in the thickness of skin dermis and to never-healing wounds [168] .

Another obvious strong point of PCPE-1 as a target, especially in contrast with PCPs and PNPs, is its specificity for procollagen maturation. As described above, it does not seem to affect the processing of any of the BTP substrates beside fibrillar procollagens [44, 51, 52] and the potential TIMP-like activity of its NTR domain was not confirmed [73] . The procollagen specificity also implies that other pro-fibrogenic BTP substrates such as pro-lysyl oxidases or those involved in the regulation of TGF-b signalling [22, 48] are not under the direct regulation of PCPE-1. However, the role of matrix stiffening in the amplification of fibrogenesis was clearly demonstrated in recent years [171, 172] and we can speculate that PCPE-1 also contributes to fibrosis by increasing collagen fibril density and therefore tissue mechanical strains.

At this point, it is important to mention that PCPE-1 specificity has been challenged by some reports suggesting possible roles of the protein in angiogenesis [54, 87, 173] , cell proliferation [174, 175] or RNA stabilization [176] . Among these, the role in angiogenesis is the most-well supported and should be considered seriously. First, PCPE-1 was found to belong to a group of five ECM proteins synthesized by fibroblasts (together with collagen I) required to promote endothelial cell lumen formation [173] . Second, PCPE-1 can strongly bind to several anti-angiogenic factors such as endostatin and thrombospondin-1 and displays antiangiogenic properties itself, both in in vitro assays [54, 87] and in an in vivo model of corneal wound healing induced by alkali treatment [54] . There is presently no clear mechanism to explain these observations and it remains possible that the control of angiogenesis by PCPE-1 is an indirect consequence of its role in collagen maturation.

In terms of drug development, we know that PCPE-1 forms a tight complex with procollagens I and III [33, 44, 77] . Even if protein-protein interactions are known to be difficult targets, it should be possible to develop molecules capable of binding either the enhancer or the procollagen motifs involved in the interaction with the latter. Both can be considered and should lead to efficient slowdown of procollagen maturation but with slightly different consequences. A pharmacological molecule targeting PCPE-1 would block its interaction with the CPs of all three major fibrillar collagens (I, II and III) while targeting the CP of one procollagen would, in principle, diminish the maturation of this specific molecule without affecting the maturation of other procollagens. Given the previously described interdependence between fibrillar collagens [13, 14] , the latter approach would require determination of how the collagen network can cope with the delayed incorporation of one collagen type. However, considering the high level of homology between the CPs of procollagens I, II and III (46% identity), it might actually be easier to develop a pan-CP inhibitory molecule and, in this case, both strategies should have similar outcomes.

Other related questions remain regarding the fate of unprocessed procollagens in the context of a fibrosis-related rapid procollagen synthesis combined with a decrease in maturation efficiency. In the preferred scenario, they would be rapidly degraded by MMPs or other ECMdegrading proteases released by fibroblasts and inflammatory cells. Overall collagen degradation, which is thought to be insufficient in the context of fibrosis, could here benefit from collagen intermediates being more easily accessible for proteolytic degradation than fibril-embedded and highly cross-linked collagen monomers. In another scenario, unprocessed procollagens would persist in the tissue and could potentially trigger signalling in surrounding cells, including in those involved in the resolution of fibrosis. Along this line, addition of CPI to cultures of human lung fibroblasts was early found to induce a dose-dependent decrease in collagen synthesis [177] and, although this result has never been confirmed or explained, it suggests that released C-propeptides could exert a negative transcriptional feedback on collagen synthesis. Whether the same type of signalling is also induced when CPs are retained in procollagens or whether the lack of CP sensing by cells would rather increase collagen synthesis remains to be analysed. In addition to the previous outcomes, it remains possible that unprocessed or partiallyprocessed procollagen molecules are aberrantly incorporated into the few newly formed fibrils. However, if C-propeptide retention in collagen fibrils was previously observed in the Bmp1 À/À /Tll1 À/À mouse embryonic cell layers, this was not the case in the Pcolce À/À cell layers [46] , giving a good indication that collagen processing intermediates do not become abnormally integrated in fibrils when PCPE-1 activity is missing.

As summarized in Table 1 , PCPE-1 has many clear advantages as an anti-fibrotic target but several questions still need to be addressed to confirm its potential. Especially, future work in the field should aim at clarifying the overall effect of modulating PCPE-1 expression or activity on collagen abundance and organization.

The authors declare that they have no competing financial interests.

This work was supported by the CNRS, the University of Lyon and the Agence Nationale de la Recherche (ANR grant 17-CE14-0033). We thank Jeannette Meunier and Lucie Essayan (LBTI, Univ Lyon, France) for their contribution to the preliminary analysis of PCPE-1 expression data in fibrosis and David Hulmes (LBTI, Univ Lyon, France) for his critical reading of the manuscript. Abbreviations: ADAMTS, a disintegrin and metalloproteinase with thrombospondin motifs; AS, aortic valve stenosis; BMP, bone morphogenetic protein; CVD, cardiovascular disease; CKD, chronic kidney disease; CP, C-propeptide; CUB, complement, Uegf, BMP-1; DMD, Duchenne muscular dystrophy; ECM, extracellular matrix; EGF, epidermal growth factor; eGFR, estimated glomerular filtration rate; ELISA, enzyme-linked immunosorbent assay; HDL, high-density lipoprotein; HSC, hepatic stellate cell; HTS, hypertrophic scar; IPF, idiopathic pulmonary fibrosis; LDL, low-density lipoprotein; MI, myocardial infarction; MMP, matrix metalloproteinase; mTLD, mammalian tolloid; mTLL, mammalian tolloid-like; NASH, nonalcoholic steatohepatitis; NTR, netrin; PABPN1, poly(A)-binding protein nuclear 1; OPMD, oculopharyngeal muscular dystrophy; PCP, procollagen C-proteinase; PCPE, procollagen C-proteinase enhancer; PNP, procollagen N-proteinase; SPC, subtilisin proprotein convertase; TIMP, tissue inhibitor of metalloproteinases;

TGF-b, transforming growth-factor b; TSPN, thrombospondin-like N-terminal Table 1 Pros and cons of targeting PCPE-1 activity to prevent collagen accumulation in fibrosis.

Delaying procollagen processing could not be sufficient to observe a significant reduction of tissue fibrosis Basal collagen assembly remains possible to allow tissue repair Absence of PCPE-1 does not lead to C-propeptide retention in collagen fibrils

Consequences of the possible accumulation of unprocessed procollagen molecules in fibrosis models are unknown PCPE-1 is highly specific to collagen biosynthesis and increased PCPE-1 expression is found at sites of high collagen deposition PCPE-1 unexplained role in angiogenesis remains to be explored Structural data are available to facilitate drug development Protein-protein interactions are difficult to target with small-molecule inhibitors

Anti-PCPE strategies can be applied to multiple potential indications in fibrosis Deleterious effects of long-term PCPE-1 inhibition on tissue homeostasis or in case of co-morbidity (e.g. wounds, bone fractures) cannot be excluded

Molecular and tissue alterations of collagens in fibrosis

Skin tissue repair: Matrix microenvironmental influences

Collagens and cancer associated fibroblasts in the reactive stroma and its relation to cancer biology

Type II collagen degradation and its regulation in articular cartilage in osteoarthritis

Fibrosis: from mechanisms to medicines

Collagen fibrillogenesis in vitro: interaction of types I and V collagen regulates fibril diameter

Composition, structure and function of the corneal stroma

Collagen XXIV (Col24a1) gene expression is a specific marker of osteoblast differentiation and bone formation

Collagen XXVII is developmentally regulated and forms thin fibrillar structures distinct from those of classical vertebrate fibrillar collagens

Fibrillar collagens

The collagen family

Type III collagen (COL3A1): Gene and protein structure, tissue distribution, and associated diseases

Type III collagen is crucial for collagen I fibrillogenesis and for normal cardiovascular development

Collagen V is a dominant regulator of collagen fibrillogenesis: dysfunctional regulation of structure and function in a corneal-stroma-specific Col5a1-null mouse model

Collagens, modifying enzymes and their mutations in humans, flies and worms

Assembly of collagen fibrils de novo by cleavage of the type I pCcollagen with procollagen C-proteinase. Assay of critical concentration demonstrates that collagen self-assembly is a classical example of an entropy-driven process

The "other" 15-40%: the role of non-collagenous extracellular matrix proteins and minor collagens in tendon

Lysyl oxidases: from enzyme activity to extracellular matrix cross-links

The procollagen Nproteinases ADAMTS2, 3 and 14 in pathophysiology

Metalloproteases meprin alpha and meprin beta are C-and N-procollagen proteinases important for collagen assembly and tensile strength

Biosynthetic processing of the pro-alpha 1(V)2pro-alpha 2(V) collagen heterotrimer by bone morphogenetic protein-1 and furin-like proprotein convertases

BMP-1/tolloid-like proteinases synchronize matrix assembly with growth factor activation to promote morphogenesis and tissue remodeling

Roles and regulation of BMP-1/tolloid-like proteinases: collagen/matrix assembly, growth factor activation and beyond

Biosynthetic processing of the Pro-alpha1(V) Pro-alpha2(V)Pro-alpha3(V) procollagen heterotrimer

Use of Bmp1/Tll1 doubly homozygous null mice and proteomics to identify and validate in vivo substrates of bone morphogenetic protein 1/tolloid-like metalloproteinases

Bone morphogenetic protein-1 processes the NH2-terminal propeptide, and a furin-like proprotein convertase processes the COOH-terminal propeptide of pro-alpha1(V) collagen

Bone morphogenetic protein-1/tolloid-related metalloproteinases process osteoglycin and enhance its ability to regulate collagen fibrillogenesis

Bone morphogenetic protein-1 processes probiglycan

Decorin is processed by three isoforms of bone morphogenetic protein-1 (BMP1)

Lysyl oxidase-like protein from bovine aorta. Isolation and maturation to an active form by bone morphogenetic protein-1

The proteolytic processing site of the precursor of lysyl oxidase

Metalloproteinase activity secreted by fibrogenic cells in the processing of prolysyl oxidase. Potential role of procollagen C-proteinase

Evidence for a protein that enhances the activity of type I procollagen C-proteinase

Type I procollagen COOH-terminal proteinase enhancer protein: identification, primary structure, and chromosomal localization of the cognate human gene (PCOLCE)

Insights into how CUB domains can exert specific functions while sharing a common fold: Conserved and specific features of the CUB1 domain contribute to the molecular basis of procollagen C-proteinase enhancer-1 activity

NMR structure of the netrin-like domain of human type I procollagen Cproteinase enhancer defines structural consensus of NTR domains and assesses potential proteinase inhibitory activity and ligand binding

Procollagen type I C-proteinase enhancer is a naturally occurring connective tissue glycoprotein

Posttranslational proteolytic processing of procollagen Cterminal proteinase enhancer releases a metalloproteinase inhibitor

Proteomics discovery of metalloproteinase substrates in the cellular context by iTRAQ labeling reveals a diverse MMP-2 substrate degradome

Processing of procollagen III by meprins: new players in extracellular matrix assembly

Identification of a potential cardiac antifibrotic mechanism of torasemide in patients with chronic heart failure

Collagen matrix deposition is dramatically enhanced in vitro when crowded with charged macromolecules: the biological relevance of the excluded volume effect

Degradome of soluble ADAM10 and ADAM17 metalloproteases

Substrate-specific modulation of a multisubstrate proteinase. C-terminal processing of fibrillar procollagens is the only BMP-1-dependent activity to be enhanced by PCPE-1

PCOLCE2 encodes a functional procollagen Cproteinase enhancer (PCPE2) that is a collagen-binding protein differing in distribution of expression and posttranslational modification from the previously described PCPE1

Procollagen C proteinase enhancer 1 genes are important determinants of the mechanical properties and geometry of bone and the ultrastructure of connective tissues

Identification of binding partners interacting with the alpha1-N-propeptide of type V collagen

BMP-1 disrupts cell adhesion and enhances TGF-beta activation through cleavage of the matricellular protein thrombospondin-1

Proteolysis of the low density lipoprotein receptor by bone morphogenetic protein-1 regulates cellular cholesterol uptake

C-terminal proteolysis of the collagen VI alpha3 chain by BMP-1 and proprotein convertase(s) releases endotrophin in fragments of different sizes

Identification of the minimal domain structure of bone morphogenetic protein-1 (BMP-1) for chordinase activity: chordinase activity is not enhanced by procollagen Cproteinase enhancer-1 (PCPE-1)

Dentin sialophosphoprotein (DSPP) is cleaved into its two natural dentin matrix products by three isoforms of bone morphogenetic protein-1 (BMP1)

Upregulation of bone morphogenetic protein-1/mammalian tolloid and procollagen C-proteinase enhancer-1 in corneal scarring

Procollagen C-proteinase enhancer 1 (PCPE-1) functions as an anti-angiogenic factor and enhances epithelial recovery in injured cornea

Region and celltype resolved quantitative proteomic map of the human heart

The mammalian Tolloid-like 1 gene, Tll1, is necessary for normal septation and positioning of the heart

Failure of ventral body wall closure in mouse embryos lacking a procollagen Cproteinase encoded by Bmp1, a mammalian gene related to Drosophila tolloid

Osteogenesis imperfecta

Identification and expression of a novel type I procollagen C-proteinase enhancer protein gene from the glaucoma candidate region on 3q21-q24

Regulation of apoAI processing by procollagen C-proteinase enhancer-2 and bone morphogenetic protein-1

Disruption of the murine procollagen C-proteinase enhancer 2 gene causes accumulation of pro-apoA-I and increased HDL levels

Procollagen Cendopeptidase enhancer protein 2 (PCPE2) reduces atherosclerosis in mice by enhancing scavenger receptor class B1 (SR-BI)-mediated high-density lipoprotein (HDL)-cholesteryl ester uptake

Effects of the absence of procollagen C-endopeptidase enhancer-2 on myocardial collagen accumulation in chronic pressure overload

Type I procollagen Cproteinase from mouse fibroblasts. Purification and demonstration of a 55-kDa enhancer glycoprotein

Different patterns of human serum procollagen C-proteinase enhancer1 (PCPE1)

Quantification of human serum procollagen C-proteinase enhancer (hsPCPE) glycopattern

Up-regulation of PCOLCE by TWIST1 promotes metastasis in Osteosarcoma

Procollagen C-proteinase enhancer grasps the stalk of the C-propeptide trimer to boost collagen precursor maturation

Folding and activity of recombinant human procollagen C-proteinase enhancer

Quantitative phosphoproteomics of vasopressin-sensitive renal cells: regulation of aquaporin-2 phosphorylation at two sites

Partial purification and characterization of a procollagen C-proteinase from the culture medium of mouse fibroblasts

Role of the netrin-like domain of procollagen Cproteinase enhancer-1 in the control of metalloproteinase activity 1

Mammalian tolloid-like 1 binds procollagen Cproteinase enhancer protein 1 and differs from bone morphogenetic protein 1 in the functional roles of homologous protein domains

The CUB domains of procollagen C-proteinase enhancer control collagen assembly solely by their effect on procollagen Cproteinase/bone morphogenetic protein-1

Strong cooperativity and loose geometry between CUB domains are the basis for procollagen Cproteinase enhancer activity

Interaction properties of the procollagen C-proteinase enhancer protein shed light on the mechanism of stimulation of BMP-1

Procollagen C-proteinase enhancer stimulates procollagen processing by binding to the C-propeptide region only

Data comparing the kinetics of procollagen type I processing by bone morphogenetic protein 1 (BMP-1) with and without procollagen C-proteinase enhancer 1 (PCPE-1). Data Brief

Structural basis of fibrillar collagen trimerization and related genetic disorders

Structural basis for the acceleration of procollagen processing by procollagen C-proteinase enhancer-1

Structural basis for the substrate specificity of bone morphogenetic protein 1/tolloid-like metalloproteases

The NTR module: domains of netrins, secreted frizzled related proteins, and type I procollagen C-proteinase enhancer protein are homologous with tissue inhibitors of metalloproteases

The NTR domain of procollagen C-proteinase enhancer-1 (PCPE-1) mediates PCPE-1 binding to syndecans-1, -2 and -4 as well as fibronectin

Binding of procollagen C-proteinase enhancer-1 (PCPE-1) to heparin/heparan sulfate: properties and role in PCPE-1 interaction with cells

Loss of fibulin-4 disrupts collagen synthesis and maturation: implications for pathology resulting from EFEMP2 mutations

Extended interaction network of procollagen C-proteinase enhancer-1 in the extracellular matrix

Secreted Frizzled-related protein 2 is a procollagen C proteinase enhancer with a role in fibrosis associated with myocardial infarction

Sizzled is unique among secreted frizzled-related proteins for its ability to specifically inhibit bone morphogenetic protein-1 (BMP-1)/tolloid-like proteinases

Exogenously administered secreted frizzled related protein 2 (Sfrp2) reduces fibrosis and improves cardiac function in a rat model of myocardial infarction

Synergistic effect of PCPE1 and sFRP2 on the processing of procollagens via BMP1

Sfrp1 and Sfrp2 regulate anteroposterior axis elongation and somite segmentation during mouse embryogenesis

Procollagen C-proteinase enhancer-1 (PCPE-1) interacts with beta2-microglobulin (beta2-m) and may help initiate beta2-m amyloid fibril formation in connective tissues

Ascorbic acid promotes procollagen C-proteinase enhancer 1 expression, secretion, and cell membrane localization

Therapeutic approaches to control tissue repair and fibrosis: Extracellular matrix as a game changer

Cardiac interstitium in health and disease: the fibrillar collagen network

Fibrosis-a common pathway to organ injury and failure

Fibroblasts: The arbiters of extracellular matrix remodeling

The myofibroblast, multiple origins for major roles in normal and pathological tissue repair

Expression of procollagen Cproteinase enhancer-1 in the remodeling rat heart is stimulated by aldosterone

Up-regulation of type I procollagen C-proteinase enhancer protein messenger RNA in rats with CCl4-induced liver fibrosis

Expression of procollagen C-proteinase enhancer in cultured rat heart fibroblasts: evidence for co-regulation with type I collagen

Towards better definition, quantification and treatment of fibrosis in heart failure. A scientific roadmap by the Committee of Translational Research of the Heart Failure Association (HFA) of the European Society of Cardiology

Cardiac fibrosis: Cell biological mechanisms, molecular pathways and therapeutic opportunities

Myocardial fibrosis: biomedical research from bench to bedside

Mechanisms of cardiac collagen deposition in experimental models and human disease

Advances in understanding the renin-angiotensin-aldosterone system (RAAS) in blood pressure control and recent pivotal trials of RAAS blockade in heart failure and diabetic nephropathy

Exercise training alters the molecular response to myocardial infarction

Atorvastatin improves ventricular remodeling after myocardial infarction by interfering with collagen metabolism

HMGCoA reductase inhibition reverses myocardial fibrosis and diastolic dysfunction through AMP-activated protein kinase activation in a mouse model of metabolic syndrome

Age-and hypertensionassociated protein aggregates in mouse heart have similar proteomic profiles

Differential role of TIMP2 and TIMP3 in cardiac hypertrophy, fibrosis, and diastolic dysfunction

Procollagen C-Proteinase Enhancer 1 (PCPE-1) is a marker of myocardial fibrosis and impaired cardiac function in a murine model of pressure overload

microRNA-122 down-regulation may play a role in severe myocardial fibrosis in human aortic stenosis through TGF-beta1 upregulation

The scaffold protein muscle A-kinase anchoring protein beta orchestrates cardiac myocyte hypertrophic signaling required for the development of heart failure

Differential patterns of replacement and reactive fibrosis in pressure and volume overload are related to the propensity for ischaemia and involve resistin

Gene expression profiling on global cDNA arrays gives hints concerning potential signal transduction pathways involved in cardiac fibrosis of renal failure

OPLAH ablation leads to accumulation of 5-oxoproline, oxidative stress, fibrosis, and elevated fillings pressures: a murine model for heart failure with a preserved ejection fraction

Proteomics analysis of cardiac extracellular matrix remodeling in a porcine model of ischemia/reperfusion injury

Mechanisms of hepatic fibrogenesis

Mechanisms of hepatic stellate cell activation

Landscape of intercellular crosstalk in healthy and NASH liver revealed by single-cell secretome gene analysis

Bioinformatics analysis of key differentially expressed genes in nonalcoholic fatty liver disease mice models

Microarray analysis of endothelial differentially expressed genes in liver of cirrhotic rats

Procollagen C-Proteinase Enhancer 1 (PCPE-1) as a plasma marker of muscle and liver fibrosis in mice

Gene expression patterns associated with histopathology in toxic liver fibrosis

Evaluating the antifibrotic potency of galunisertib in a human ex vivo model of liver fibrosis

ASK1 inhibition reduces cell death and hepatic fibrosis in an Nlrp3 mutant liver injury model

Data comparing the plasma levels of procollagen C-proteinase enhancer 1 (PCPE-1) in healthy individuals and liver fibrosis patients

Evaluation of serum procollagen C-proteinase enhancer 1 level as a fibrosis marker in patients with chronic hepatitis B

Elevated serum levels of procollagen C-proteinase enhancer-1 in patients with chronic kidney disease is associated with a declining glomerular filtration rate

Heterogeneity in tuberculosis pathology, microenvironments and therapeutic responses

A systematic review of pathological findings in COVID-19: a pathophysiological timeline and possible mechanisms of disease progression

Idiopathic pulmonary fibrosis: epidemiology, clinical features, prognosis, and management

Idiopathic pulmonary fibrosis: evolving concepts

Time-and compartment-resolved proteome profiling of the extracellular niche in lung injury and repair

Extracellular matrix in lung development, homeostasis and disease

Genomic phenotype of non-cultured pulmonary fibroblasts in idiopathic pulmonary fibrosis

Expression of ciliumassociated genes defines novel molecular subtypes of idiopathic pulmonary fibrosis

Conditional deletion of Nedd4-2 in lung epithelial cells causes progressive pulmonary fibrosis in adult mice

Common pathophysiological mechanisms of chronic kidney disease: therapeutic perspectives

Functional genomic annotation of genetic risk loci highlights inflammation and epithelial biology networks in CKD

Molecular events surrounding collagen fibril assembly in the early healing rabbit medial collateral ligament-failure to recapitulate normal ligament development

The role of fibrosis in Duchenne muscular dystrophy

Dynamic changes in myocardial matrix and relevance to disease: translational perspectives

Nuclear entrapment and extracellular depletion of PCOLCE is associated with muscle degeneration in oculopharyngeal muscular dystrophy

How the cornea heals: cornea-specific repair mechanisms affecting surgical outcomes

Extracellular matrix molecules implicated in hypertrophic and keloid scarring

Hypertrophic scars and keloids-a review of their pathophysiology, risk factors, and therapeutic management

Transforming growth factor beta (TGF-beta) isoforms in wound healing and fibrosis

The role of TGFbeta in wound healing pathologies

Transforming growth factorbeta regulation of bone morphogenetic protein-1/ procollagen C-proteinase and related proteins in fibrogenic cells and keratinocytes

Comparative proteomic analysis of extracellular matrix proteins secreted by hypertrophic scar with normal skin fibroblasts

Gene expression in human keloids is altered from dermal to chondrocytic and osteogenic lineage

Transcriptional profiling of rapamycin-treated fibroblasts from hypertrophic and keloid scars

Detection of bone and cartilage-related proteins in plasma of patients with a bone fracture using liquid chromatography-mass spectrometry

Towards two-dimensional electrophoresis mapping of the cerebrospinal fluid proteome from a single individual

Protein identification with Teflon as matrix-assisted laser desorption/ionization sample support

In-depth characterization of the cerebrospinal fluid (CSF) proteome displayed through the CSF proteome resource (CSF-PR)

Proteomic analysis of cerebrospinal fluid from patients with idiopathic temporal lobe epilepsy

Agerelated collagen turnover of the interstitial matrix and basement membrane: Implications of age-and sexdependent remodeling of the extracellular matrix

Changes in urinary excretions of C-telopeptide and cross-linked Ntelopeptide of type I collagen during pregnancy and puerperium

Bone turnover: Biology and assessment tools

Circulating lung biomarkers in idiopathic lung fibrosis and interstitial lung diseases associated with connective tissue diseases: Where do we stand?

Biochemical markers of myocardial remodelling in hypertensive heart disease

Non-invasive diagnosis of liver fibrosis and cirrhosis

Biomarkers to predict reverse remodeling and myocardial recovery in heart failure

BMP1-like proteinases are essential to the structure and wound healing of skin

Induced ablation of Bmp1 and Tll1 produces osteogenesis imperfecta in mice

BMP1 and TLL1 are required for maintaining periodontal homeostasis

The extracellular matrix and transforming growth factor-beta1: Tale of a strained relationship

Mechanotherapeutics: targeting mechanical signaling in fibrosis and tumor stroma

The requirement for fibroblasts in angiogenesis: fibroblast-derived matrix proteins are essential for endothelial cell lumen formation

The regulatory expression of procollagen COOH-terminal proteinase enhancer in the proliferation of vascular smooth muscle cells

Effects of procollagen C-proteinase enhancer protein on the growth of cultured rat fibroblasts revealed by an excisable retroviral vector

Stabilization of RNA strands in protein synthesis by type I procollagen C-proteinase enhancer protein, a potential RNA-binding protein, in hepatic stellate cells

Evidence for pretranslational regulation of collagen synthesis by procollagen propeptides

UCSF ChimeraX: Structure visualization for researchers, educators, and developers