Cystatin M/E (Cystatin 6): A Janus-Faced Cysteine Protease Inhibitor with Both Tumor-Suppressing and Tumor-Promoting Functions

Simple Summary Cystatin M/E is a low molecular mass protein and a potent inhibitor of proteolytic enzymes (cathepsins B, L, V, and legumain). Cystatin M/E participates in epidermal homeostasis and its deregulation is associated with several skin diseases. Cystatin M/E was initially identified as a tumor suppressor in breast cancer. Its gene is epigenetically downregulated and its methylation status holds a prognostic significance. Moreover, it may also serve as a biomarker for clinical diagnosis. This tumor-suppressing role was documented in cutaneous carcinoma, melanoma, lung, cervical, brain, prostate, gastric and renal cancers. Conversely, increased levels of cystatin M/E in triple-negative breast cancer tissues correlate with a higher risk of metastasis and a lower disease-free survival rate. Beside its orthodox tumor-suppressing role, cystatin M/E may operate as a tumor-promoting effector as reported for thyroid, oral and pancreatic cancer and hepatocellular carcinoma. Given its seemingly contradictory results, in-depth analysis of the regulatory mechanisms of the expression and activity of cystatin M/E during tumorigenesis have to be expanded. Likewise, the possible involvement of cystatin M/E in signaling pathways, beside its protease inhibitor function, remains to be scrutinized. Abstract Alongside its contribution in maintaining skin homeostasis and its probable involvement in fetal and placental development, cystatin M/E (also known as cystatin 6) was first described as a tumor suppressor of breast cancer. This review aims to provide an update on cystatin M/E with particular attention paid to its role during tumorigenesis. Cystatin M/E, which is related to type 2 cystatins, displays the unique property of being a dual tight-binding inhibitor of both legumain (also known as asparagine endopeptidase) and cysteine cathepsins L, V and B, while its expression level is epigenetically regulated via the methylation of the CST6 promoter region. The tumor-suppressing role of cystatin M/E was further reported in melanoma, cervical, brain, prostate, gastric and renal cancers, and cystatin M/E was proposed as a biomarker of prognostic significance. Contrariwise, cystatin M/E could have an antagonistic function, acting as a tumor promoter (e.g., oral, pancreatic cancer, thyroid and hepatocellular carcinoma). Taking into account these apparently divergent functions, there is an urgent need to decipher the molecular and cellular regulatory mechanisms of the expression and activity of cystatin M/E associated with the safeguarding homeostasis of the proteolytic balance as well as its imbalance in cancer.


Gene and Protein Structures of Cystatin M/E 2.1. CST6
Almost simultaneously, cystatin M was characterized as the product of a downregulated gene in breast cancer [27], while Abrahamson and colleagues first identified cystatin E as a cDNA corresponding to a rare mRNA species in epithelial cells [28]. The nucleotide sequence encoded human pre-cystatin E and showed the presence of an open reading frame containing a typical consensus sequence for the initiation of translation around the start ATG codon, followed by a poly(A) signal, 78 nucleotides downstream from the stop codon, after which a poly(A) sequence was evident a further 20 nucleotides downstream. Northern blot analysis indicated that the distribution pattern was deeply different from the distribution of cystatin C mRNA and that the cystatin E gene was expressed ubiquitously in the uterus, liver, placenta, pancreas, heart, spleen, small intestine and blood leucocytes, and in a lesser amount in the brain, testis and kidney. Moreover, all cDNA clones originated either from amniotic membrane cells or fetal skin epithelium libraries, which suggests that its gene expression is strongly upregulated in, or specific for, epithelial cells. The authors proposed that cystatin E may play a protective role during fetal development [28]. Alternatively, a partial cDNA of cystatin M was pinpointed to be expressed specifically in the primary tumor cell line but not in metastatic breast tumor cells. The authors identified a full-length cystatin M cDNA containing an open reading frame of 447 nucleotides (24-470), a short 5 -untranslated sequence (1-23) and a 3 -untranslated sequence of 128 nucleotides, with a polyadenylation signal AATAAA (552-557) and a poly(A) tail. Besides its expression by normal mammary cells, consistent levels of cystatin M mRNA were distributed in the placenta, lung, skeletal muscle, kidney and pancreas [27]. These twin papers paved the way for further successful research on cystatin M/E, also known as cystatin 6.
The gene encoding cystatin M/E (CST6) is located on the 11q13 region, while the cystatin C gene (CST3) is located on 20p11.2 [29]. The complete sequence of the human CST6 gene was first elucidated by Zeuween et al. [30]. Briefly, CST6 encompasses three exons within~1.4 kb of genomic sequence from the translation start site to the translation termination site and a specific genomic organization within the cystatin 2 family, since the TATA box is lacking in the 5 -flanking sequence. Additionally, the two introns of CST6 (542 bp and 365 bp) are remarkably short compared to introns of cystatins C, D or S genes (respectively, 1.1 to 2.2 kb). The mouse CST6 ortholog was characterized by the same group and was assigned to the proximal end of chromosome 19. Mouse and human CST6 retain 69% amino acid identity and 82% amino acid conservation [31].

Protein Structure
Cystatin M/E is synthesized with a signal peptide (1-28) and a main peptide chain . Mature cystatin M/E is composed of 121 amino acid residues (theoretical pI = 6.78). Its sequence has some unusual characteristics compared to cystatin C, the archetypal member of type 2 cystatins, specifically a five-residue insertion between amino acids 76 and 77, and a deletion of residue 91 (cystatin C numbering). Likewise, cystatin M/E displays two disulfide bridges at positions 73-83 and 97-117, respectively, and an N-glycosylation site at Asn108 [28]. Cystatin M/E is found both in a~14 kDa nonglycosylated form and a~17 kDa glycosylated form, and carbohydrates correspond to a complex mannose-linked type [32,33].
In a preliminary study, the three-dimensional structure of cystatin M/E was generated by homology modelling and refined by molecular dynamics simulation. The N-terminal region of cystatin M/E contributes equally to the inhibition of papain together with the conserved L1 and L2 loops [34]. Alternatively, a three-dimensional model of human cystatin M/E was generated on the basis of the crystal structure of human cystatin D, supporting that the overall predicted structure of cystatin M/E narrowly matches that of cystatin D [35]. Moreover, bioinformatics data support that legumain and cysteine cathepsins are inhibited by two distinct non-overlapping binding sites. Finally, the three-dimensional crystal structure of cystatin M/E was solved by Brandstetter and collaborators. Cystatin M/E is characterized by a typical 5-stranded anti-parallel β-sheet that is wrapped around an almost central, perpendicular five-turn α-helix. Two disulfide bridges additionally stabilize the structure by clamping strands β4 and β5 and the cystatin M/E-specific appending structure that is inserted between strands β3 and β4 [36].

Legumain (Asparaginyl Endopeptidase, AEP)
Cystatin M/E demonstrates a compelling affinity (Ki = 0.0016 nM) for pig legumain [10]. It appears to be the most effective physiological legumain inhibitor, binding more tightly as compared with cystatin C (Table 1). Among type 2 cystatins, cystatin C is also an effective inhibitor of legumain (Ki = 0.2 nM), while cystatin F displays a weaker potency (Ki = 10 nM). Conversely, cystatin D as well as stefins A and B (type 1 cystatins) and LK (type 3 cystatin) did not inhibit legumain. Legumain is a caspase-related protease (clan CD, family C13). As elegantly mentioned by Dall and Brandstetter [11], "legumain resembles a caspase in cathepsin's clothing". The legumain crystal structure confirmed its caspase-like fold [40]. Legumain harbors a strict substrate specificity, cleaving exclusively after Asn residue (P1 position), despite Asp being also tolerated at low pH conditions [41]. Crucially, legumain holds a unique dual protease-ligase activity, associated with a pH-dependent reversible on-off switching [36], which could lead to biologically relevant functional amendments with respect to its regulation by cystatins, in particular cystatin M/E. Site-directed mutagenesis of cystatin M/E revealed that Asn39 is crucial for the inhibition of legumain but not for CatL and CatV, albeit Trp106 was identified to be a key residue to inactivate CatV but not legumain (cystatin C numbering). Data support that legumain and papain-like cysteine proteases are inhibited by two distinct non-overlapping sites [35] and correlate with a seminal study that demonstrated the formation of a stable ternary complex between legumain, cystatin C and papain [10]. The crystal structure of cystatin M/E bound to its protease target was further deciphered and confirmed kinetics data [36]. While the papain-interacting region is composed of the N-terminal substrate-like region and L1 and L2 loops, cystatin M/E interacts with legumain according to a totally different mechanism [6] (Figure 1). The binding of cystatin M/E to legumain is mediated by a reactive center loop (RCL) according to a canonical, substrate-like binding mode with Asn39 on RCL serving as P1 residue and stabilized via exosite interactions by the so-called legumain exosite loop (LEL), which makes contact in primed substrate binding sites [36]. Cystatin M/E also inhibits autoactivation of secreted prolegumain, supporting a regulatory role in controlling both intraand extracellular legumain activity [42]. Both glycosylated and unglycosylated cystatin M/E forms exhibit a similar legumain inhibitory potency, endorsing that the inhibition of legumain does not rely on its glycosylation status [32].
While it was shown that MDA-MB-435S, A375 and C8161 cells internalized substantial amounts of cystatin C, a low internalization of cystatin M/E was observed. Nevertheless, it may significantly affect intracellular legumain activity and cell migration [43]. On the other hand, secreted cystatin M/E does not reveal consistent effects on Fas-induced apoptosis [17]. Recently, it was shown that a domain-swapped dimer of cystatin M/E retained its capacity to inhibit legumain while binding sites toward papain-like proteases were buried. Alternatively, cystatin M/E amyloid fibrils retain the functional ability to inhibit both legumain and papain under in vitro conditions [44].

Cystatin M/E in Skin Homeostasis and Skin Disorders
Under biological conditions, the expression pattern of human cystatin M/E is distinctive among type 2 cystatins, as elevated protein levels are chiefly restricted to epidermal keratinocytes, hair follicles, sebaceous glands and sweat glands [38,45]. The role of cystatin M/E in skin has been outstandingly summarized elsewhere (see for review: [46]). Briefly, cystatin M/E colocalizes specifically with CatV in the stratum granulosum and in the root sheets of the hair follicle. Moreover, cystatin M/E is found in the extracellular space in the stratum corneum associated with corneodesmosomes, where it is thoroughly associated with CatV [47]. Cystatin M/E may directly control the activity of cysteine proteases, and both CatV-cystatin M/E and CatL-cystatin M/E balances have a critical regulatory role for epidermal differentiation, hair follicles, desquamation process of the stratum corneum and skin barrier formation, as validated by genetic ablation of cystatin M/E [46,48]. Accordingly, human CST6 deficiency prevents the development of a multilayered epidermis in a 3D reconstructed skin model and may be incompatible with normal human foetal development [49]. Cystatin M/E also regulates the crosslinking of structural proteins by governing the CatL-dependent processing of transglutaminase 3 during the cornification process of the epidermis and the hair follicle [38,46]. A recent report showed that interleukin-17A downregulated the gene expression of CST6 and profilaggrin, a major protein of keratohyalin granules in the granular layer [50]. Moreover, both CatL and CatV together with cystatin M/E may participate in cell shedding, which is associated with proteolytic impairment of skin adhesive structures, and notably of desmogleins [51].
Cystatin M/E deficiency in mice causes disturbed epidermal cornification, impaired barrier function and neonatal lethality because of excessive trans-epidermal water loss (for review: [46]). Harlequin ichthyosis is a relentless congenital skin disorder that usually leads to a stillborn fetus or neonatal death. Its clinical features at birth include a thickened fissured epidermis and are characterized by excessive epidermal and follicular hyperkeratosis. Concordantly, a null mutation in the cystatin M/E gene of harlequin ichthyosis (ichq) mice, which recapitulates most of the features of the human disease, was identified to be responsible for juvenile lethality and defects in epidermal cornification and desquamation [31]. Harlequin ichthyosis is associated with an unlimited activity of its target protease legumain in hair follicles and epidermis, leading to a dysregulation of legumain-cystatin M/E balance and an enhanced crosslinking of loricrin monomers by transglutaminase 3 [52]. Interestingly, the rescue of the skin ichq phenotype by transgenic, epidermis-specific, re-expressed cystatin M/E murine model does not completely restore the wild-type phenotype. Indeed, mice survive the neonatal phase, but display severe eye pathology and alopecia [53]. During chronic hand eczema, the desquamation-related kallikrein-related peptidases (KLK5 and KLK7) are downregulated, like filaggrin and cystatin M/E [54]. Inflammatory skin disorders (atopic dermatitis and psoriasis) also cause an extended expression of cystatin M/E, which is constitutively found in the stratum granulosum of normal skin and in the spinous cell layers (stratum spinosum) where it colocalizes with transglutaminase for which it serves as a substrate [55]. Yet, a decrease in both transcription and translation of cystatin M/E and CatV was reported in lesional atopic dermatitis and psoriasis epidermis. Conversely, mRNA levels of CatL and transglutaminase 3 were increased [45]. A homozygous variant, resulting in a stop codon in exon 2 of CST6, was associated with hypotrichosis, eczema and impaired sweating. Functionally, the corresponding recombinant mutant cystatin M/E protein lacks its ability to inhibit CatV, CatL and legumain [56]. Interestingly, analysis of the murine skin proteome showed that the deletion of CatL induced increased levels of acidic CatD, stefin B and cystatin M/E, and accumulation of the extracellular glycoprotein periostin [57].
Alternatively, Lyme disease results from infection with the spirochete Borrelia burgdorferi. The first and common clinical manifestation, called erythema migrans, corresponds to a circular, inflamed skin lesion. Proteomic analysis was conducted on an independent cohort of patients with erythema migrans, leading to the identification of some candidate molecules, including cystatin M/E, as putative acute-phase blood markers for the diagnosis of early Lyme disease [58].

Cystatin M/E and Reproductive System
In an early report, it was proposed that cystatin M/E in the amniotic fluid may participate in fetal development [28]. During the estrous cycle, CST6 was identified as a pig pregnancy-related gene and dynamic changes in its mRNA level were observed in bovine endometrium [59,60]. On the other hand, interferon tau and prostaglandins, which are key players of the regulation of the peri-implantation period of pregnancy in sheep, increase ex-pression of endometrial cystatin M/E, which in turn could play a role in tissue remodeling, in placental and fetal development and drive conceptus elongation in beef heifers [60][61][62]. Cystatin M/E, which is expressed at the maternal-fetal interface together with legumain, may participate in the establishment and maintenance of pregnancy in pigs [63]. Additionally, analysis of the ovarian transcriptome suggested that the differentially expressed CST6 gene could be a candidate gene for prolificacy traits in sheep [64].

Cystatin M/E in Neurodegenerative Progressive Diseases and Atherosclerosis
Type 1 cystatins and cystatin C were proposed to play a role in Alzheimer's disease according to their ability to form amyloid fibrils [65][66][67]. Interestingly, the conformational destabilization of cystatin M/E leads to the formation of a domain-swapped dimer, still acting as an effective legumain inhibitor while cathepsin binding sites are buried. Then, cystatin M/E dimers may further convert to amyloid fibrils [44]. Aside from being an oncogene, DJ-1 was also identified as a syndrome-associated protein of a familial form of Parkinson's disease, working as a coactivator to diverse transcription factors. An increased level of cystatin M/E was detected in the cerebrum, spleen and heart from DJ-1-deficient mice, leading to a decreased activity of legumain, which participates in the DJ-1-regulated activation and cleavage of annexin A2 [68]. Elsewhere, an unsTable 15-to 18-mer minisatellite repeat expansion in the promoter region of CST6 was identified as a disease-causing mutation of progressive myoclonus epilepsy of Unverricht-Lundborg type, a rare autosomal recessive disorder [69]. Yet, a rat homolog of cystatin M/E, which was expressed by embryonal hippocampal cells, was proposed to contribute to neuronal cell differentiation [70]. Together with cystatins C and M/E, legumain expression is increased in both plasma and plaques of patients with carotid stenosis and was proposed as an early biomarker of atherosclerosis [71].

Cystatin M/E: Tumor Suppressor and Biomarker in Breast Cancer
In a seminal in vitro study, cystatin M was identified as a downregulated transcript in a metastatic breast cancer cell line when compared to a matched primary tumor cell line. Consequently, loss of cystatin M/E expression was proposed to be associated with the progression of human breast cancer [27]. Transfection of cystatin M/E in highly tumorigenic and metastatic human breast cancer cells (MDA-MB-435S cell line) impaired proliferation, migration, matrix invasion and tumor-endothelial cell adhesion, providing the first evidence that cystatin M/E plays a protective role in safeguarding against breast cancer [72]. Consistently, SCID (severe combined immunodeficiency) mice that have been orthotopically implanted with breast cancer cells expressing cystatin M/E revealed a delayed primary tumor growth and lower metastatic burden in the lung and liver [73,74]. Conversely, deficiency of cystatin M/E expression was reported in invasive ductal carcinoma cells from stage IV patients compared to normal human breast epithelial cells, sustaining that cystatin M/E could be a candidate tumor suppressor gene [73]. An analysis of differential gene expression of triple-negative breast cancer (TNBC)-derived MDA-MB-231 cells demonstrated that CST6 was also strongly repressed, in addition to type 2 cystatin genes (CST1, CST2 and CST4), which are all encoded by a gene cluster on 20p11 [75].
Bone metastasis is a common cause of morbidity and mortality in breast cancer patients. Jin and colleagues demonstrated that CST6 was downregulated in bone metastatic (MDA-MB-231) derivative cell lines. Ectopic expression of CST6 suppressed proliferation, colony formation, migration and invasion in vitro. This overexpression rescued mice from osteolytic metastasis and death, while CST6 knock-down markedly enhanced cancer cell bone metastasis and shortened animal survival, sustaining that cystatin M/E is a bona fide suppressor of breast cancer osteolytic metastasis [76].
The loss of cystatin M/E expression has been strongly correlated with CST6 promoter hypermethylation in a panel of breast cancer cell lines (MDA-MB-231, ZR75-1, MCF-7, T47D, BT-549, Hs578T, MDA-MB-436 and MDA-MB-453) and in primary breast carcino-mas, indicating that this aberrant characteristic is common in breast malignancies [77][78][79]. Hypermethylation, which arises before the development of invasive breast cancer, occurs within the vicinity of a large and dense CpG-rich promoter island in the 5 region of CST6 [79,80]. An aberrant methylation profile of CST6 was also found in axillary lymph node metastasis in comparison to primary tumor tissue and adjacent normal tissue from the same breast cancer patients [81,82]. Treatment with a DNA demethylating agent (5-aza-2 -deoxycytidine, 5-aza-dC) reactivated CST6 transcription and expression, substantiating that cystatin M/E is an epigenetically-inactivated tumor suppressor gene [77][78][79].
Basal-like breast tumors (~25% of all breast cancers) contribute disproportionately to breast cancer deaths, as they have the propensity to display aggressive tumor characteristics (increased size, rapid tumor growth, increased rate of metastasis, higher incidence of relapse and lower overall patient survival). A cluster of tumors, identified as basal-like breast cancers, expresses a hypermethylator signature with a low expression level of six specific genes including CST6 and is associated with an aberrant overexpression of DNA (cytosine-5)-methyltransferase 3β (DNMT3b) [83]. Secondary to DNMT3b overexpression, the molecular mechanism governing DNMT3b gene-mediated aberrant DNA hypermethylation involves the loss of post-transcriptional regulation of DNMT3b by regulatory microRNAs [84].
The prognostic significance of the methylation status of CST6 as a cancer biomarker in breast tumors was statistically evaluated. Multivariate analysis showed that CST6 promoter methylation status was an independent prognostic factor for disease-free-interval (DFI) and overall survival (OS) [85]. Data support that CTS6 hypermethylation may serve as a biomarker for clinical molecular diagnosis and is associated with unfavorable prognosis in operable breast cancer [85,86]. The CST6 promoter methylation status of circulating tumor cells (CTCs) isolated from peripheral blood of breast cancer patients was also analyzed. CST6 hypermethylation for both operable and metastatic breast cancer significantly differed from that of the control population [87]. Moreover, metastatic breast cancer patients with CTCs unmethylated for CST6 have a significantly longer progression-free survival compared to patients with corresponding enriched methylated CTCs [88]. A panel of 13 genes, including CST6, may enable breast cancer prediction and patient stratification, based on well-defined DNA methylation profiles, and might be a valuable biomarker for early detection of breast cancer [89]. In a pilot study, it was observed that CST6 promoter is hypermethylated in cell-free DNA circulating in the plasma of breast cancer patients, but not in healthy individuals. Data support that CST6 promoter methylation in cell-free DNA could be a putative novel plasma tumor marker for breast cancer [90]. Alternatively, a clear association between the epithelial cell adhesion molecule (EpCAM)-positive CTC-fraction and circulating tumor DNA (ctDNA) for promotor methylation of the transcriptional regulator SOX17, but not for CST6, was found for both patients with early and metastatic breast cancer [91]. At the molecular level, this hypermethylation is associated with aberrant serine/threonine kinase AKT1 activation in epithelial cells, as well as the disabled inositol polyphosphate-4-phosphatase, type II (INNP4B) regulator that results from suppression by cancer-associated fibroblasts. Accordingly, microenvironmental stimuli may trigger this epigenetic cascade and lead to the downregulation of cystatin M/E in breast tumors [92]. Moreover, hypermethylation of the CST6 gene is influenced by the alteration of estrogen receptor (ER) and receptor tyrosine-protein kinase erbB-4 (HER4), and cystatin M/E was identified as a downstream target of ER and HER4 [93].
Cystatin M/E may modulate malignant properties of human breast carcinoma cells (MDA-MB-435S cells) partly through the downregulation of two major secreted signaling molecules, autotaxin/lyso-phospholipase D and interleukin-8, which combine mitogenic, motogenic and angiogenic properties [94]. CST6 was identified as a repressed TBX2 (a Tbox family oncogenic transcription factor) target gene through a mechanism involving Early Growth Response 1 (EGR1). An increase in TBX2 expression is known to drive breast cancer proliferation and metastasis. In contrast, the exogenous expression of cystatin M/E in TBX2-expressing breast cancer cells resulted in upregulated apoptosis. Remarkably, induction of apoptosis depends on the cystatin M/E legumain-inhibitory domain, but not on the cysteine cathepsin-inhibitory domain, emphasizing that legumain is a crucial oncogenic driver. Secretion or glycosylation of cystatin M/E is not a prerequisite to elicit cell killing effects. Thus, cystatin M/E represents an important barrier to breast tumorigenesis which could be by-passed by TBX2 transcriptional repression. Data also sustain that the "TBX2-cystatin M/E-legumain" pathway may represent a novel promising axis for the development of innovative therapies to target TBX2-driven breast cancers [95]. It could be noticed that CST6 was also repressed through a mechanism involving both TBX2 and EGR1 in rhabdomyosarcoma, a pediatric malignant cancer which is thought to arise from myogenic precursors in the skeletal muscle lineage [96].

Lung Cancer
Type I cystatins were proposed as independent prognostic markers for patients with non-small cell lung cancer (NSCLC), since high levels of both stefins A and B, but not cystatin M/E, are associated with a better survival probability. Compared to cystatin C (median concentration: 3.7 ng.mg −1 of protein), the estimated median level of cystatin M/E in NSCLC primary tumors is 14.6 pg.mg −1 of protein [97]. On the other hand, CST6 is frequently transcriptionally downregulated in NSCLC tumors, while its ectopic expression suppresses NSCLC growth in culture [98]. Transcriptomics of lung adenocarcinoma cell lines derived from both smokers (S) vs. never smokers (NS) demonstrated that the methylation of CST6 gene differed significantly based on smoking practice [99]. Like cystatin C, immunoreactive cystatin M/E levels (median concentration into the pleural space: 0.14 nM) are significantly higher in effusions of primary pleural tumors (mesotheliomas) compared to secondary pleural tumors (3.4 µg/L vs. 2.5 µg/L) [22].

Cervical Cancer
The loss of expression of cystatin M/E that is associated with CST6 inactivation by somatic mutations and promoter hypermethylation was also detected in primary cervical tumors. Interestingly, expression of cystatin M/E was recovered after 5-aza-dC and/or trichostatin A treatment, while its ectopic expression in cell lines resulted in growth suppression, supporting that cystatin M/E is a cervical cancer suppressor [100]. An inverse correlation between the expression of cystatin M/E and CatL was detected in cervical carcinoma and primary tumors. On the other hand, a direct relationship between the deficiency of cystatin M/E and the regulation of the nuclear factor-kappa B (NF-κB) signaling pathway and its nuclear expression was reported, suggesting that cystatin M/E participated in the reduction of tumor cell growth through cytoplasmic retention of NF-κB [101].

Brain and Neck Cancer
Malignant gliomas are the most common primary brain tumors in adults and the second most common tumor in children, and are characterized by a high morbidity and mortality. It was found that CST6 promoter was hypomethylated in normal brain samples, albeit most primary brain tumors demonstrated deficient cystatin M/E expression, in agreement with CST6 promoter hypermethylation. Moreover, CST6 expression by glioma tumor initiating cells (TIC) was fully hampered by promoter methylation, while the ectopic expression of cystatin M/E reduced TIC motility and invasion. Qiu et al. [102] proposed that analysis of CST6 epigenetic silencing may therefore represent a novel prognostic marker of gliomas. Interestingly, together with CST6, the apoptosis-inducer Bcl-2-interacting killer (BIK) is also a target of epigenetic silencing in malignant gliomas. BIK, which is a member of the BH3-only family of pro-apoptotic proteins, is known to suppress glioma cell growth in culture [103]. The treatment of SCC-25 neck squamous carcinoma cells with a pleiotropic calcemic vitamin D3 analog (EB1089) lead to a 20-fold increase in cystatin M/E expression compared to a control [104].

Cutaneous Squamous Cell Carcinoma and Melanoma
Conversely to benign hyperplasia as observed during psoriasis vulgaris, a unique cancer-specific gene expression signature was identified for human cutaneous squamous cell carcinoma (cSCC) associated with a gene downregulation of granzyme B, and a differential gene upregulation (fold change > 3) of matrix metalloproteinase MMP-1, kallikrein 7, CatV and cystatin M/E [105]. Contrariwise, proteome profiling of primary and metastatic cSCC lesions by mass spectrometry-based proteomics demonstrated a decrease in cystatin M/E in metastatic cSCC lesions relative to the primary phenotypes (fold change = −3.9), and the authors suggested that cSCC metastasis may be partly attributed to the decreased level of cystatin M/E in metastatic lesion phenotype [106]. Accordingly, further investigation should be conducted to clarify the regulation of CST6 and cystatin M/E protein expression, as well as its pathophysiological (deleterious vs. protective) relevance in cSCC metastasis. On the other hand, CST6 is involved in the suppression of proliferation, migration and metastasis of melanoma [107]. Interestingly the glycosylated~17 kDa form of cystatin M/E, but not the non-glycosylated~14 kDa form, is predominantly detected in melanoma cell lines secreting cystatin M/E, while cystatin C is lacking or expressed at a very low level. Invasion is suppressed in cystatin M/E overexpressing melanoma cell lines and intracellular legumain activity is drastically impaired. Conversely, CatB activity is not affected, which suggests that cystatin M/E specifically regulates legumain activity and therefore the invasive potential of human melanoma cells [108]. Additionally, in the tumor acidic microenvironment, secreted cystatin M/E could prevent auto-activation of secreted prolegumain. Alternatively, in its absence, extracellular prolegumain could auto-activate and further cleave target substrates (e.g., fibronectin), thus promoting tumor invasion and metastasis [42].

Prostate and Gastric Cancer
Expression of cystatin M/E is extensively decreased in metastatic prostate cell lines and prostate tumor tissues compared to normal human prostate epithelium [109]. Treatment of metastatic prostate cell lines with a histone deacetylase inhibitor transcriptionally upregulates CST6 and increases cystatin M/E protein expression. Its overexpression significantly reduces prostate cell proliferation and inhibits tumor growth and the incidence of lung metastasis. The authors proposed that CST6 downregulation was associated with promoter histone modifications and may have participated in prostate cancer progression during the invasive and metastatic stages. Analysis of the methylation status of CST6 in gastric carcinomas and their paired adjacent non-tumor tissues revealed a loss of expression of cystatin M/E in 70% of gastric carcinomas due to promoter hypermethylation. Moreover, patients with hypermethylated CST6 had a significantly shorter survival time than those with hypomethylated promoter [110]. In a close way, a CpG island methylator phenotyperelated prognostic gene signature, including CST6, may allow the classification of patients with gastric cancer into high-and low-risk groups with a significant OS difference [111].

Renal Carcinoma
Epigenetic study by high-density gene expression microarrays of renal cell carcinoma (RCC) cell lines identified eight genes, including CST6, exhibiting tumor-specific promoter region hypermethylation associated with transcriptional silencing [112]. Hypermethylation of CST6 is associated with both a shortened progression-free and OS of patients with metastasized RCC undergoing anti-vascular endothelial growth factor (VEGF)-based therapy [113]. Following transfection and subsequent expression of cystatin M/E, the growth of RCC cell lines was compromised, supporting the status of cystatin M/E as a candidate kidney tumor suppressor [112] (summarized in Table 2).  [116] Lung cancer CST6 ectopic expression suppresses NSCLC cell lines growth [98] Melanoma CST6 involved in the suppression of proliferation, migration and metastasis of melanoma cells [42,107,108] Oral cancer CST6 is overexpressed in oropharyngeal metastatic cells.
Its silencing increases cell proliferation and invasion [117,118] Pancreatic cancer Cystatin M/E expression is associated with PDAC cells growth and proliferation [119] Papillary thyroid carcinoma (PTC) Increase of CST6 in PTC is associated with lymph node metastasis [120] Prostate cancer Overexpression of cystatin M/E inhibits cell growth and lung metastasis incidence [109] Renal cell carcinoma (RCC) Promoter hypermethylation of CST6 is associated with a shortened progression-free and overall survival of patients with metastasized RCC [112,113]

Tumor-Promoting Role of Cystatin M/E 7.1. Triple-Negative Breast Cancer
Besides its tumor-suppressive function, cystatin M/E transcripts were detected in both primary and metastatic breast cancer cells. Increased expression of cystatin C and cystatin M/E, which correlated significantly with a larger tumor size, was observed in breast cancer cells isolated by laser capture microdissection [114]. Analysis of patients suffering from TNBC (10-20% of human ductal adenocarcinomas), which is the most aggressive breast cancer type, with the poorest prognosis and highest mortality rates, demonstrated that cystatin M/E was unexpectedly elevated in TNBC tissues vs. adjacent normal breast tissues. Moreover, increased levels of cystatin M/E correlated with a high risk of lymphnode metastasis and a low disease-free survival rate, supporting that cystatin M/E could be used as an independent predictor of disease-free survival in TNBC. Conversely to the archetypal role of cystatin M/E as a tumor suppressor in breast cancer, the authors suggested that cystatin M/E may be involved in the pathophysiology of TNBC, possibly acting as a tumor-promoting molecule [115]. Nevertheless, these unanticipated results need to be further confirmed.

Oral Cancer
Analysis of cell lines derived from primary and metastatic lesions of oropharyngeal squamous cell carcinomas indicated that cystatin M/E was expressed 40-fold higher in the metastatic vs. the primary tumor cell line. Unexpectedly, upregulation of cystatin M/E may favor metastasis by blocking intrinsic CatB activity and rescuing tumor cells from TNF-α-induced apoptosis [117]. The silencing of CST6 expression in tumor oral cancer cell line increases both invasion and motility, but also renders metastatic MDA-686Ln cells hyperproliferative [118].

Pancreatic Cancer
No or limited expression of cystatin M/E was observed in normal pancreas, while immunohistochemical analysis confirmed cystatin M/E overexpression in pancreatic ductal adenocarcinoma (PDAC). Its constitutive expression in CST6-null cells promotes their growth in vitro and in vivo, and the addition of exogenous active cystatin M/E, but not recombinant cystatin M/E lacking its proteinase-inhibitor domain, promotes cell proliferation. Conversely, knock-down of cystatin M/E attenuates PDAC cell growth. The authors endorse that cystatin M/E may be involved in the proliferation and survival of pancreatic cancer, probably through the inhibition of intracellular pro-apoptotic CatB [119].

Hepatocellular Carcinoma
Survival analysis indicated that CST6 correlated with the OS and recurrence-free survival (RFS) of patients with hepatitis B virus-related hepatocellular carcinoma (HBVrelated HCC) [116]. CST6 may function as a prognostic biomarker for HCC, with a high CST6 expression being associated with poor survival of patients, in correlation with a previous report claiming that cystatin M/E could be a pancreatic tumor promotor [119]. Taken together, data suggest that cystatin M/E may participate in diverse molecular mechanisms and exert distinct effects in different types of cancer [116].

Conclusions
Cystatin M/E (type 2 cystatin) is a specific and dual tight-binding inhibitor of both cathepsins L, V and B (family C1) and legumain (or asparaginyl endopeptidase, family C13), via two distinct mechanisms of inhibition, which involve non-overlapping binding sites. Cystatin M/E plays a critical role in epidermal homeostasis, and its deregulation is associated with various skin diseases such as ichthyosis or psoriasis. Cystatin M/E is also probably involved in fetal and placental development and could participate in the pathogenesis of neurodegenerative progressive diseases, according to its ability to form amyloid fibrils. CST6 was initially identified as a tumor suppressor and an epigenetically downregulated gene in breast cancer. The methylation status of CST6 holds a prognostic significance and may also serve as a biomarker for clinical diagnosis. Afterward, this tumorsuppressing role was corroborated with other cancers (cutaneous carcinoma, melanoma, lung, cervical, brain, neck, prostate, gastric and renal cancers) and the use of CST6 as a valuable biomarker was confirmed. Conversely, increased levels of cystatin M/E in TNBC tissues correlate with a higher risk of metastasis and a lower disease-free survival rate. Thus, beside its conventional tumor-suppressing role, cystatin M/E may act as a tumor-promoting effector. This outward antagonistic function was further testified for thyroid, oral and pancreatic cancer and hepatocellular carcinoma. Accordingly, molecular and cellular analysis of the regulatory mechanisms of the expression and activity of cystatin M/E associated with the maintenance of the proteolytic balance, as well its imbalance under pathophysiological conditions, has to be urgently deepened. Moreover, the putative involvement of cystatin M/E, during tumorigenesis, in signaling pathways independent of its protease inhibitor function, still remains to be deciphered.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the writing of the manuscript.