Gelatinase B/MMP-9 in Tumour Pathogenesis and Progression

Since its original identification as a leukocyte gelatinase/type V collagenase and tumour type IV collagenase, gelatinase B/matrix metalloproteinase (MMP)-9 is now recognised as playing a central role in many aspects of tumour progression. In this review, we relate current concepts concerning the many ways in which gelatinase B/MMP-9 influences tumour biology. Following a brief outline of the gelatinase B/MMP-9 gene and protein, we analyse the role(s) of gelatinase B/MMP-9 in different phases of the tumorigenic process, and compare the importance of gelatinase B/MMP-9 source in the carcinogenic process. What becomes apparent is the importance of inflammatory cell-derived gelatinase B/MMP-9 in tumour promotion, early progression and triggering of the “angiogenic switch”, the integral relationship between inflammatory, stromal and tumour components with respect to gelatinase B/MMP-9 production and activation, and the fundamental role for gelatinase B/MMP-9 in the formation and maintenance of tumour stem cell and metastatic niches. It is also apparent that gelatinase B/MMP-9 plays important tumour suppressing functions, producing endogenous angiogenesis inhibitors, promoting inflammatory anti-tumour activity, and inducing apoptosis. The fundamental roles of gelatinase B/MMP-9 in cancer biology underpins the need for specific therapeutic inhibitors of gelatinase B/MMP-9 function, the use of which must take into account and substitute for tumour-suppressing gelatinase B/MMP-9 activity and also limit inhibition of physiological gelatinase B/MMP-9 function.


Gelatinase B/MMP-9 Hemopexin Domain
The gelatinase B/MMP-9 hemopexin domain exhibits a relatively unique covalent structure in which Cys516 and Cys704 form a disulphide bridge, which is involved in domain function but is not required for gelatinase B/MMP-9 secretion [45,46]. This domain facilitates interactions with substrates, gelatinase B/MMP-9 oligomerisation, binds the carboxyl terminal of TIMP-1, binds cell surface proteins such Ku70/80 and LRP, and upon binding appropriate substances, such as heme, also mediates autocatalytic gelatinase B/MMP-9 activation [47]. Divergent disulphide bridging between the 17-cysteine residues within gelatinase B/MMP-9 regulates structure and function. Disulphide bridging within fibronectin repeats are essential for gelatinase B/MMP-9 secretion. Hemopexin domain function depends upon disulphide bridging and disulphide bridging between the O-glycosylation or hemopexin domains facilitates gelatinase B/MMP-9 dimerization or oligomerisation, promoting CD44 binding, which results in activation of the EGF receptor and subsequent ERK/1/2 mediated cancer cell migration [46,48]. Gelatinase B/MMP-9 hemopexin domain hetero-dimerization with proteins such as TIMP-1 and NGAL protects gelatinase B/MMP-9 against proteolytic degradation.

Gelatinase B/MMP-9 O-Glycosylation Domain
The O-glycosylated domain of gelatinase B/MMP-9, also known as the type V collagen-like domain, represents a 64 amino acid linker containing 22 proline residues, six glycine residues and approximately 12-14 O-linked glycans [49]. This domain is active in hemopexin domain orientation, which is important for molecular interactions with exogenous proteins, including gelatinase B/MMP-9 substrates [49]. The removal of this domain reduces gelatinase B/MMP-9 specificity for macromolecular substrates, including gelatin [13].

Truncated Gelatinase B/MMP-9 Isoforms
Several truncated gelatinase B/MMP-9 isoforms have been described that include proteolytically active fragments derived from autocatalysis and exogenous proteolytic degradation. The 65 kDa gelatinase B/MMP-9 catalytically active fragment generated by MMP-3 is deleted of COOH terminal sequence and escapes TIMP-1 inhibition. KLK7 and meprin-α also remove this domain from gelatinase B/MMP-9 [50][51][52]. A novel 82 kDa inactive pro-gelatinase B/MMP-9 form has been described in human leukaemic cells, which also escapes TIMP inhibition [53,54] and a similar sized human pro-gelatinase B/MMP-9 isoform is generated by the action of plasmin [41].

Gelatinase B/MMP-9 Substrates
Gelatinase B/MMP-9 was originally characterised as a gelatinase/V collagenase [1][2][3][4][5], and was later attributed type IV collagenolytic activity [6,11]. Although there is controversy surrounding the susceptibility [55] or resistance [8,40,56] of triple helical domains of collagens to degradation by gelatinase B/MMP-9, the capacity of gelatinase B/MMP-9 to degrade native type IV collagen may be limited, therefore, to non-triple helical, less-disulphide cross-linked or pre-digested molecular forms of type IV collagen [8,57,58]. It remains debatable whether activated gelatinases alone degrade type IV collagen within the context of an insoluble basement membrane [8,[57][58][59]. Gelatinase B/MMP-9 does, however, degrade basement membrane laminin, disrupting basement membrane structure, tissue architecture [60] and inducing apoptosis [61]. In addition to its capacity to degrade extracellular matrix components, recent reports have characterized an ever-increasing array of substrates susceptible to degradation by gelatinase B/MMP-9, dramatically widening the potential physiological and pathological sphere of gelatinase B/MMP-9 influence. Gelatinase B/MMP-9 exhibits substrate specificity for cytokines, chemokines and growth factors within the extracellular compartment and may also degrade nuclear, mitochondrial and cytoplasmic substrates (Table 1). For a broad spectrum of gelatinase B/MMP-9 substrates, both old and new, we direct reader to the following articles [52,. Table 1. Update of Gelatinase B/MMP-9 substrates. Substrates (sub) and MMP-9 origins are provided: human (hu), mouse (mu), bovine (bo) and rabbit (ra).

Gelatinase B/MMP-9 Transcription and Translation
The 2.2 kb human gelatinase B/MMP-9 promoter contains a TATA-like motif at position −29 but no CAAT-like motif. Relative to the transcriptional start site, functional transcription sites include: an SP1 binding GC box located at −563, a retinoblastoma binding element or GT box that also binds SP1 at position −54, and three additional GT boxes. In addition to a TGF-β1 inhibitory element at −474 bp and 4 potential AP-1 binding elements, the functional AP-1 site at position −79 is essential for basal and jun/Fos induced expression in HT-1080 and osteosarcoma cells [119], three functional PEA3/Ets binding sites localise between −599 and −531 are also involved in basal gelatinase B/MMP-9 transcription [119,120]. A functional NF-κB binding site is located at −600 and a second site at −328 bp [121], and potentially functional inhibitory AP-2-like binding sites immediately upstream of the GC-box that interferes with Sp-1 binding [122], an alternating microsatellite CA sequence in close proximity to the AP1 site at position −79 [12] (Figure 1).
The gelatinase B/MMP-9 protein is constitutively expressed by only a limited number of cell types, such as keratinocytes, macrophages, polymorphonuclear leukocytes and some malignant tumour cell lines, including MDA-MB-231 breast cancer, HT-1080 fibrosarcoma and A2058 melanoma cell lines [12,15] and is readily induced in wide range of normal and tumour cell types by pro-inflammatory cytokines, activators of PKC and growth factors with gelatinase B/MMP-9 expression regulated through inhibitory STAT and SMAD pathways and stimulatory PKC, Ras/MAPK, TRAD/TRAF, MEK/JNK, ASK/MKK and IRAK/TRAF pathways [12,13,15].
Gelatinase B/MMP-9 enzymatic activity is inhibited by the universal systemic protease inhibitor α2-macrogloblin [160], members of the tissue inhibitors of metalloproteinases (TIMPs) family [161,162] and is also antagonized by its own isolated hemopexin domain [41,163]. TIMPs 1-4 are 20-30 kDa glycoprotein MMP inhibitors that depend upon disulphide bridges between 6 cysteine pairs for their inhibitory activity [161,162,164]. TIMP-1 exhibits a unique binding interaction with gelatinase B/MMP-9 and, with the exception of human neutrophils, exhibits a high level of coordinated expression with TIMP-1, is frequently secreted as a TIMP-1/gelatinase B/MMP-9 complex and binds gelatinase B/MMP-9 with high affinity, in contrast to TIMP-2 and TIMP-3 [12,15,41,123,157,162]. The interaction between pro-form gelatinase B/MMP-9 and TIMP-1 involves the C-terminal domains of both proteins and in this form TIMP-1 is available to inhibit other MMPs. Upon gelatinase B/MMP-9 activation, TIMP-1 inhibits gelatinase B/MMP-9 catalytic activity through N-terminal interaction with the gelatinase B/MMP-9 catalytic site, with inhibition facilitated by the gelatinase B/MMP-9 C-terminus, since it does not readily occur in gelatinase B/MMP-9 C-terminus deletion mutants. In contrast to TIMP-1, TIMP-2 inhibition of gelatinase B/MMP-9 depends upon the N-terminal domain, but does not involve C-terminal interaction and is less effective that TIMP-1. TIMP-3 is a matrix-associated inhibitor that interacts with and inhibits gelatinase B/MMP-9 to a lesser extent than either TIMP-1 or TIMP-2 [161,162].

The Gelatinase B/MMP-9/TIMP-1 Protease-Antiprotease Equilibrium
Mechanisms that alter the equilibrium between gelatinase B/MMP-9 and its TIMP-1 inhibitor in favour of protease activity, facilitate gelatinase B/MMP-9 involvement in tumour pathology, and include differential expression, evasion from TIMP inhibition, and TIMP-1 inactivation.
Gelatinase B/MMP-9 and TIMP-1 are frequently co-ordinately expressed in a large number of cell types and secreted as a pro-gelatinase B/MMP-9/TIMP-1 complex [12,15]. The tumor environment is however complex, and composed of tumor, stromal and inflammatory elements that also contribute to the modulation of this important equilibrium. Tumor infiltrating neutrophils release gelatinase B/MMP-9 in TIMP-free form, facilitating tumor-associated differential gelatinase B/MMP-9 and TIMP-1 expression [168,169] Furthermore, the differential up-regulation of gelatinase B/MMP-9 but not TIMP-1 expression has been reported in human ovarian cancer [170], skin cancer [171], squamous cell carcinoma of the hypopharynx [172] and colon and rectal tumours in vivo [173], and has also been demonstrated in malignant melanomas induced in metallothionin/RET transgenic mice [174]. In vitro, differential up-regulation of gelatinase B/MMP-9 but not TIMP-1 expression characterises PC-3 prostate tumor cell/stromal cell co-cultures and endothelial cells co-cultured with fibroblasts [175,176], cervical carcinoma cells in response to CD40L activation [177], human head and neck squamous carcinoma cells in response to c-erbB ligands [178], spontaneous epithelial to neuroblast transformation of human neuroblastoma cells [123], retinoic acid treatment of differentiation resistant human neuroblastoma cells [124], peroxiredoxin expression in metastatic human MDA-MB-231 breast cancer cells [179], thioredoxin expression in human MDA-MB-231 breast cancer cells [132] and bFGF treatment of human retinoblastoma cells [180].
Differential gelatinase B/MMP-9 and TIMP-1 regulation may also involve promoter SNPs and/or 3'-UTR micro RNA binding sites. Indeed, gelatinase B/MMP-9 SNPs that augment gelatinase B/MMP-9 expression have been associated with increased risk of different forms of cancer (See Section Gelatinase B/MMP-9 SNPs on page 241), as have altered levels of miRs that bind miR binding sites within the 3'-UTR region of gelatinase B/MMP-9. The miRs -211, 491-5p and 885-5p target and inhibit gelatinase B/MMP-9 expression and are down regulated in human glioblastoma multiforme, in association with increased gelatinase B/MMP-9 expression [17,18], and miR-19a has been reported to regulate gelatinase B/MMP-9 expression in colon cancer cells [181]. Furthermore, a recent report has shown that miR-17 targets the TIMP-1 protein-coding region and its inhibition enhances TIMP-1 expression and decreases gelatinase B/MMP-9 activity [182]. It is likely, therefore, that altered patterns of miR expression may also facilitate the differential expression of gelatinase B/MMP-9 and TIMP-1 in malignant tumours.
Extracellular activation of the thioredoxin redox system, up-regulated in malignant tumours, has been shown to inhibit TIMP but not MMP activity in vitro and in models of human neuroblastoma and UV irradiated dermal fibroblasts [164,183]. Furthermore, the myeloperoxidase/H 2 0 2 /hypochlorous acid (HOCl) system of inflammation induces the oxidative inactivation of TIMPs, whilst promoting the activation of MMPs, at concentrations found during inflammation [184,185], providing mechanisms through which the gelatinase B/MMP-9/TIMP equilibrium within tumours can be altered in favour of proteolytic activity even under conditions of high level TIMP expression [186]. TIMP MMP-inhibitory activity, furthermore, can be destroyed by neutrophil elastase, trypsin and -chymotrypsin, all of which activate gelatinase B/MMP-9 [12,187,188], providing an additional mechanism for irreversible TIMP inhibition combined with gelatinase B/MMP-9 activation within inflammatory tumour environments and also environments such as the pancreas, in which trypsin and trypsin-like enzymes are expressed [189]. Finally, truncated gelatinase B/MMP-9 isoforms generated by enzymatic digestion or present on the cell surface of human leukemic cells have been shown to escape TIMP inhibition (see Section 3.4).
Gelatinase B/MMP-9 has also been reported to promote liver tumour initiation by the proteolytic release and activation of matrix-associated TGFβ and VEGF [206], and in human mammary epithelial cells induces cell surface expression of the HER2/Neu oncoprotein, inhibiting apoptosis and shifting normal mammary cells towards a transformed phenotype, in the presence of oestrogen [207].
In contrast, gelatinase B/MMP-9 optimises non-homologous end joining (NHEJ) DNA repair in human glioma cells. Indeed, down-regulation of gelatinase B/MMP-9 expression, combined with either urokinase or cathepsin B, delays DNA repair by lowering KU70/80 recruitment to damaged DNA. This reduces NHEJ DNA repair function, increases the levels of DNA damage and promotes apoptosis [208].

Gelatinase B/MMP-9 and Tumour Initiating Cell Proliferation and Expansion
Clonal expansion of transformed cells is also an essential step in tumour progression and is facilitated by inflammation and involves a change in equilibrium between proliferation, apoptosis and angiogenesis [209,210].
In the heterozygous APC knockout mouse model (APC-min), neutrophil-derived gelatinase B/MMP-9 stimulates adenoma initiating cell proliferation, promoting adenoma expansion, and implicating gelatinase B/MMP-9 in the expansion of tumour cell populations that lack full APC function. It is likely that this involves gelatinase B/MMP-9-mediated release and activation of non-matrix cytokines, such as TNFα and IL-1β and matrix-associated growth factors, such as VEGF, TGFβ and FGFs and/or the degradation of growth inhibitors [211][212][213][214]. Indeed, gelatinase B/MMP-9 degrades IGF-BPs augmenting the circulating levels of IGF, promoting astrocytoma growth [215], and increasing circulating VEGF and EGF levels [216], which also promote adenoma cell proliferation in APC-min mice. Furthermore, transcriptional silencing of gelatinase B/MMP-9 inhibits human glioma cells proliferation [208] and Wnt signalling induced by hypoxia stimulating gelatinase B/MMP-9 expression and promotes neural stem cell proliferation [217], suggesting that a hypoxia/Wnt/gelatinase B/MMP-9 axis may also promote proliferation of the cancer stem/progenitor cell component of neural-related tumours.

Gelatinase B/MMP-9, Stem Cells and the Cancer Stem Cell Niche
The stem cell niche is a unique, specialised location responsible for maintaining stem cells. Stem cells within the niche are anchored by intracellular and cell matrix adhesive interactions, which regulate stem cell numbers, stem cell self-renewal and potentially asymmetrical stem cell division. Normal stem cells and cancer stem cells exhibit similar behaviour [218,219]. Cancer stem cell niches have been identified in tumours and implicated in tumour heterogeneity, metastatic progression and therapeutic resistance, and are regulated by conditions within the tumour and promoted by tumour associated stress such as hypoxia [220,221]. Gelatinase B/MMP-9 has been implicated in regulating stem cell niche behaviour and within the bone marrow, degrading extracellular matrices within the stem cell niche, resulting in the activation and mobilisation of haemopoetic stem cells. This is facilitated by the conversion of stem cell factor from its membrane bound to free form, promoting c-Kit receptor-mediated stem cell proliferation [90]. Gelatinase B/MMP-9 also releases circulating endothelial precursor stem cells from the bone marrow, contributing to angiogenesis [90]. Interaction between stroma-derived factor (SDF)-1 and the chemokine receptor CXCR4 is essential for stem/progenitor cell function [222] and induces gelatinase B/MMP-9 expression. A similar interaction induces gelatinase B/MMP-9 expression in cancer cells, promoting dissemination and metastasis to bone [222][223][224]. Wnt signalling induces gelatinase B/MMP-9 expression and maintains stem cell niche integrity [225,226]. Wnt signalling is up regulated in cancer, and also stimulates cancer stem cell proliferation, resistance to apoptosis, tumour invasion and metastasis [227][228][229]. Furthermore, Wnt-induced gelatinase B/MMP-9 expression has been implicated in embryonic neural stem cells proliferation in conditions of hypoxia [217], a similar mechanism may, therefore, regulate cancer stem cells proliferation within neural tumours [218].

Gelatinase B/MMP-9 and Epithelial to Mesenchymal Transition (EMT)
Epithelial to mesenchymal transition (EMT) represents the conversion of polarized immotile epithelial cells into motile mesenchymal progenitor cells. This mechanism is important in development (type 1), normal wound healing or pathological fibrosis (type 2) and in the metastatic transformation of cancer cells (type 3) [230]. Type 3 EMT is fundamental for tumour progression to metastasis, and is either re-activated in de-differentiated epithelial cancer cells or activated in epithelial cancer stem cells, inducing a more motile and invasive phenotype [228]. It may also be transient, with metastatic cells reverting back to an epithelial phenotype at destination.
Gelatinase B/MMP-9 is an important EMT-related gene, and is not only a consequence but also a cause of EMT ( Figure 3). Gelatinase B/MMP-9 cooperates with Snail transcription factor to induce EMT in epidermoid carcinoma cells [231], is involved in medulloblastoma cell EMT [232], is induced by Twist transcription factor an essential inducer of EMT [233][234][235] and by krupple like factor (KLF)-8, a critical component of FAK-regulated breast cancer EMT, which induces gelatinase B/MMP-9 expression in human breast cancer cells, promoting migration, invasion, angiogenesis and metastasis [234][235][236]. EMT in gastric cancer involves a Shh/PI3K/Akt/gelatinase B/MMP-9 pathway, which promotes metastatic dissemination to lymph nodes [237]. In human neuroblastoma cells, spontaneous EMT-like phenotypic conversion from a less invasive epithelial to more invasive neuroblast phenotype, associates with the induction of gelatinase B/MMP-9 expression and increased gelatinase B/MMP-9-mediated invasion [123].

Gelatinase B/MMP-9 and Cancer-Related Inflammation
Inflammation is now considered to be a hallmark of tumour progression, and regulates tumour-associated angiogenesis, tumour cell proliferation, invasion and metastasis [209,210]. Gelatinase B/MMP-9 is considered to be a tuner and amplifier of inflammatory and immune functions [106,238] and is up regulated by pro-inflammatory cytokines such as TNFα, IL-1β, IL-6 and TGFβ in a wide variety of human tumour cells, stromal and endothelial cells [12,13,15]. Gelatinase B/MMP-9 activates pro-inflammatory cytokines TNFα and IL-1β, increases the activity of chemokines CXCL1, CXCL4, CXCL7 and CXCL8, releases TGFβ from matrix stores, is released by activated neutrophils in TIMP-1-free form and acts as a nanomolar effector of tumour associated inflammation [12,13]. Neutrophil-derived gelatinase B/MMP-9 also interacts with neutrophil NGAL, which prevent autolytic gelatinase B/MMP-9 processing but does not impair gelatinase B/MMP-9 activity, promoting tumour progression [130,161]. CXCL8 interaction with the chemokine receptor CXCR2 induces gelatinase B/MMP-9 release from neutrophils [239], and activation of the chemokine receptor CXCR4 up-regulates gelatinase B/MMP-9 expression in prostate tumour cells, promoting invasion and metastasis [240]. Furthermore, myeloperoxidase/H 2 O 2 /HOCl system activation in neutrophils activates gelatinase B/MMP-9 and inhibits TIMP activity [184,185]. Gelatinase B/MMP-9, therefore, exhibits an integral relationship with tumour-associated inflammation. Indeed, the inhibition of gelatinase B/MMP-9 expression by inhibitors of pro-inflammatory cyclooxygenase-2 reduces tumour cell proliferation, invasion and metastasis [241,242]. In addition to its relationship with neutrophils, gelatinase B/MMP-9 also promotes macrophage and tumour cell invasion by cleaving the TGF-β-induced protein βig-h3, releasing it from the extracellular matrix, whereas non-degraded βig-h3 inhibits cell migration by promoting cell-cell contact and cell-extracellular matrix interactions [243].
In contrast to its tumour-promoting effects, high-level gelatinase B/MMP-9 expression has also been reported to promote tumour regression in a breast cancer model, augmenting neutrophil infiltration and promoting tumour-associated macrophage anti-tumour activity [244].

Gelatinase B/MMP-9 and Angiogenesis
The formation of new blood vessels is a highly orchestrated process that depends upon mitogenic and non-mitogenic angiogenic factors and involves matrix remodelling, cell migration, and regulated adhesive interactions between vascular cells and with the matrix. Tumour neovascularisation is fundamental for primary tumour expansion, metastatic progression and metastatic growth, and occurs via processes including sprouting angiogenesis, vasculogenesis, co-option inter-susception and/or vascular mimicry. Unlike normal vessels, blood vessels within tumours are abnormal, immature and inflammatory in nature [245].
Gelatinase B/MMP-9 is a critical pro-angiogenic molecule [246] and triggers the -angiogenic switch‖ in the quiescent vasculature [247,248] (Figure 4). Both host inflammatory and vascular gelatinase B/MMP-9 has been shown to be crucial for the development of the tumour angiogenic vasculature in models of pancreatic, ovarian and skin cancer [170,211,249]. Neutrophil gelatinase B/MMP-9 regulates pericyte proliferation, apoptosis and recruitment during angiogenesis [170] and mobilises the recruitment of bone marrow-derived angiogenic precursors to the tumour stroma enhancing the tumour angiogenic and vasculargenic process [90,[250][251][252]. Gelatinase B/MMP-9 also triggers -the angiogenic switch‖ by mobilising and activating angiogenic mitogens from matrix stores at the onset of tumour-associated angiogenesis [169,211,251,253]. This process is facilitated by the release of TIMP-1-free gelatinase B/MMP-9 from neutrophils, which acts as an exceptionally potent nanomolar angiogenic factor, releasing both FGF and VEGF from matrices [169,254].
The gelatinase B/MMP-9/VEGF axis not only supports angiogenesis but also promotes hyperactive haematopoiesis, [255,256], which also promotes tumour progression by expanding myeloid-derived suppressors that suppress T-lymphocyte proliferation and activation, promoting tumour evasion of immune surveillance [257][258][259]. Mouse gelatinase B/MMP-9 has been shown to cleave VEGF to a truncated VEGF 121 form that promotes irregular neovascularisation by altering interactions with heparan sulphate and other matrix components [118]. This, however, does not appear to extend to human gelatinase B/MMP-9 [80].
In addition to promoting pericyte recruitment, gelatinase B/MMP-9 also promotes the recruitment of pro-angiogenic monocytes and CD34 + endothelial cell progenitors, which express VE-cadherin and VEGFR2, to tumours, markedly influencing angiogenesis [252,[260][261][262][263] and induces the release of circulating endothelial precursor stem cells from the bone marrow by degrading c-kit ligand, contributing to both angiogenesis and vasculogenesis [90]. In human neuroblastoma, gelatinase B/MMP-9 recruits bone marrow-derived leukocytes and support cells to tumour vessels, regulating vessel maturation [264] and the VEGF/gelatinase B/MMP-9 axis has been implicated in the robust angiogenic response associated with TrkAIII oncogene promotion of neuroblastoma tumorigenicity [265]. In gelatinase B/MMP-9 knockout mice, impaired vascularisation associates with reduced pericyte-recruitment [266] and vascular pericytes have been shown to express gelatinase B/MMP-9 in human breast tumours [267]. In general, knockout technology has implicated MMP9 in skeletal growth plate vascularisation [268] and in bone marrow derived CD11b + myelomonocytic cell-mediated vasculogenesis in irradiated tumour tissues, with the absence of gelatinase B/MMP-9 associated with small tumours containing mature vessels [269]. Gelatinase B/MMP-9 has also been implicated in capillary branching during ischemia-induced revascularisation [270]. Although, bone marrow-cell derived gelatinase B/MMP-9 appears to be sufficient for tumour vasculogenesis, it is not essential and can be substituted by gelatinase B/MMP-9 from either stromal, smooth muscle or tumour cell components. Indeed, fibroblast gelatinase B/MMP-9 enhances endothelial cell survival and function [271], gelatinase B/MMP-9 from circulating macrophages promotes angiogenesis in a model of pancreatic cancer [272] and increased tumour cell gelatinase B/MMP-9 promotes angiogenesis in a model of neuroblastoma [265].

Gelatinase B/MMP-9 and Lymphangiogenesis
Lymphagiogenesis is also an important component of tumour progression, with lymphatic vessels providing important routes for metastatic dissemination [311,312]. Although gelatinase B/MMP-9 is not required for normal skin lymphangiogenesis [313], tumour induced lymphangiogenesis has been reported to involve a sonic hedgehog/PI3K/Akt/gelatinase B/MMP-9 pathway, leading to lymph node metastases in gastric cancer [237]. Furthermore, neutrophil-derived gelatinase B/MMP-9 has been implicated in inflammation-associated lymphangiogenesis, promoting VEGF-A bioavailability and bioactivity [314] and, together with VEGF-C, has been implicated in lymphangiogenesis and lymph node metastasis in breast cancer [315].

Gelatinase B/MMP-9 and Disruption of Tissue Architecture
The loss of tissue architecture is one of the earliest hallmarks of premalignant epithelial cancer and results in tumour proliferation, local invasion and angiogenesis [316]. In malignant breast cancer, tumour cells loose their capacity to form ordered structures and proliferate as disorganised colonies [317]. Raf/MEK/ERK-mediated induction of gelatinase B/MMP-9 expression results in the destruction of breast tissue architecture, during breast cancer initiation, by degrading basement membrane laminin and destroying basement membrane integrity. This results in de-regulated tissue polarity and the loss of growth control (Figure 4). Gelatinase B/MMP-9 inhibition halts this process by preserving basement membrane integrity, which in turn reverses phenotype, arrests growth and re-establishes a differentiated acinar polarity [60].

Gelatinase B/MMP-9, Tumour Cell Invasion and Motility
Within the tumour context, gelatinase B/MMP-9 was originally identified as a novel type IV collagenolytic protease secreted by metastatic Ras transformed fibroblasts and implicated in basement membrane disruption required for tumour invasion and metastasis [6,7]. Although it remains debatable as to whether gelatinase B/MMP-9 alone can effectively degrade type IV collagen within the context of an insoluble basement membrane matrix, gelatinase B/MMP-9 promotes invasion by disrupting basement membrane structure by degrading basement membrane laminin and type IV collagen, in concert with other enzyme systems [8,57,59,60]. Interaction between tumour cells and stroma at the invasive edge regulates gelatinase B/MMP-9 expression, which combined with gelatinase B/MMP-9 released by tumour-associated neutrophils and macrophages, increases invasiveness [175,[328][329][330] ( Figure 5). Tumor cell invasion is, however, a complex process that depends upon alterations in protein expression, interaction between tumour, inflammatory and stromal cells, altered intercellular and extracellular adhesive interactions, and changes in the tumour microenvironment. It is regulated by pro-inflammatory cytokines, chemokines, growth factors, matrix components, integrin and non-integrin receptors, proteases and inhibitors, and depends upon the cellular motile response. Cellular motility is achieved by different mechanisms and can reversibly switch between mesenchymal and amoeboid migration, which promote invasion as either single cells or collective chains, sheets, columns, tubes or clusters [331].
Protease involvement in migration and invasion is relatively restricted to mesenchymal motility, whereas amoeboid motility does not require proteolytic activity but involves a high level of cellular deformability, low affinity substrate binding and cycles of morphological contraction and expansion [331,332]. Mesenchymal migration, on the other hand, requires high affinity binding to integrin and non-integrin receptors. During mesenchymal migration, integrin or non-integrin receptors concentrate to membrane lamellipodia, filopodia, pseudopodia and invadopodia, promoting adapter protein-mediated intracellular interaction with the actin cytoskeleton. This results in the formation of focal contacts and adhesions with extracellular matrix components, the maturation of which activates intracellular focal adhesion kinases (FAKs) that form transient signalling complexes with Src kinases, promoting movement by inducing the turnover of focal contact providing the propulsive force for movement by continually modifying of cell-matrix interactions. These events depend upon proteolytic activity and involve the fibronectin integrin receptors α5β1 or αVβ6, the laminin integrin receptors α6β1 or α6β4, the fibronectin/vitronectin integrin receptor αVβ3 and the fibrillar collagen receptor α2β1 [331].
Gelatinase B/MMP-9 also interacts with the non-integrin receptor CD44, concentrating gelatinase B/MMP-9 to cell extensions, which control the turnover of adhesive interactions and extracellular matrix degradation required for motility, in a coordinated process that also involves ezrin, actin and Krp1 [336]. Gelatinase B/MMP-9 interaction with CD44 also promotes breast cancer cell migration and invasion in association with EGFR activation [48]. An N-cadherin/FGFR/gelatinase B/MMP-9 axis has been implicated in breast cancer cell invasion and metastasis, bypassing E-cadherin invasion and metastasis suppressing signals [155]. Gelatinase B/MMP-9 degradation of protease nexin-1 has also been implicated in a novel pathway through which gelatinase B/MMP-9 regulates tumour cell invasion, impairing the capacity of nexin to bind and down-regulate the activity of uPA [345].
In contrast to these reports, gelatinase B/MMP-9 has also been shown to degrade the β4 component of α6β4 integrin, de-regulating sheet migration of epithelial cells [346]. Furthermore, gelatinase B/MMP-9 interaction with α4β1 integrin and CD44 on the surface of chronic leukemic cells has been shown to inhibit migration in response to chemotactic gradients [318]. Tumour cells, furthermore, can undergo mesenchymal to amoeboid transition (MAT). Gelatinase B/MMP-9 is not required for amoeboid movement through 3D interstitial matrices [332] and MAT has been shown to increase metastatic capacity in association with reduced gelatinase B/MMP-9 expression [347].

Gelatinase B/MMP-9 and Primary Tumour Cell Escape
In order to escape the constraints of the primary tumour, tumor cells invade, move and alter their adhesive interactions. Chemotactic motile responses may direct tumour cells to lymphatic or blood vessels and tumour interaction, with tumour associated macrophages facilitating directional movement within tumours [348]. HGF activation of tumour cell c-met induces gelatinase B/MMP-9 expression, increasing tumor cell motility and scattering [349].

Gelatinase B/MMP-9 and Immunological Surveillance
The capacity to evade elimination by immunological/inflammatory mechanisms is an essential feature of tumour progression to metastasis. Gelatinase B/MMP-9 is an important regulator of both innate and tumour immune responses [12]. This is illustrated in gelatinase B/MMP-9 knockout mice, which do not resolve contact hypersensitivity reactions, implicating gelatinase B/MMP-9 in the down regulation of the immune response [350], suggesting an analogous role for gelatinase B/MMP-9 in cancer. In support of this, gelatinase B/MMP-9 expression associates with that of anti-pathogen immune-response-related genes in late stage compared to early stage lung tumours, although it remains to be determined whether any of these represent novel gelatinase B/MMP-9 substrates [351]. Gelatinase B/MMP-9 degrades ICAM-1, down-regulating leukocyte homing [80] and promotes evasion of the immune system by chronic myeloid leukemia cells by solubilizing cell membrane ICAM-1 [352]. Gelatinase B/MMP-9 degrades the IL-2 receptorα, repressing activation and proliferation of tumour infiltrating T-lymphocytes in cervical cancer [85,86]. Gelatinase B/MMP-9 degrades Surfactant protein D (SP-D), an important component of innate immune defence, leading to loss of innate immune function, limiting SP-D involvement in tumour immunology and renders oncology patients more susceptible to infection [353]. Gelatinase B/MMP-9 digests C1q complement component at a site required for interaction with the C1qR 02 receptor, repressing C1q/C1qR 02 involvement in tumour immunology [109] and may also degrade complement component C1r [65]. C5a induces the expression of the gelatinase B/MMP-9 stimulator IL-1β in monocytes [15,354] and the complement membrane attack complex induces gelatinase B/MMP-9 expression in cells protected against MAC-mediated lysis by CD59 [355], suggesting that activation of the complement system may promote tumour-associated gelatinase B/MMP-9 expression. Gelatinase B/MMP-9 also degrades the β2 subunit of macrophage CD18 integrin receptor, important for macrophage recruitment [91].

Gelatinase B/MMP-9 Haematogenous and Lymphatic Metastatic Dispersal
Tumour cell intravasation represents an important mechanism for haematogenous metastatic dissemination. The breaching of the vascular wall is considered to represent a rate limiting step for intravasation and consequently for haematogenous metastasis [311].
Gelatinase B/MMP-9 has been implicated in intravasation and subsequent metastasis formation [362] (Figure 6), with a particular role for inflammatory neutrophil-derived gelatinase B/MMP-9 highlighted in the promotion of haematogenous tumour cell dispersion of HT1080 fibrosarcoma and PC3 prostate carcinoma cells in xenograft models [12]. Neutrophil-derived gelatinase B/MMP-9 involvement in tumour-associated intravasation requires neutrophil attraction to the activated endothelial cell surface, neutrophil activation and release of TIMP-1-free gelatinase B/MMP-9. Activation of TIMP-1 free gelatinase B/MMP-9 releases angiogenic factors stored within the extracellular matrix, which promote endothelial sprouting and new vessel formation, and gelatinase B/MMP-9-assisted tumour cell intravasation and dissemination [363]. In support of this, gelatinase B/MMP-9 expression correlates with the intravasation and metastatic dissemination of HT-1080 fibrosarcoma cells, which is inhibited by the gelatinase B/MMP-9 inhibitor marimistat [362]. Furthermore, keratinocyte growth factor induces gelatinase B/MMP-9 expression and venous invasion by pancreatic cancer cells [364]. Gelatinase B/MMP-9 has also been implicated in lymphatic dissemination of colon cancer to lymph nodes. This mechanism involves gelatinase B/MMP-9 and the chemokine receptor CCR7. C-C chemokine interaction with CCR7 promotes gelatinase B/MMP-9 expression and lymphatic dissemination of colon cancer, whereas CCR7 knockdown reduces gelatinase B/MMP-9 expression lymphatic dissemination and lymph node metastases, implicating the CCR7/gelatinase B/MMP-9 axis in lymphatic metastatic dissemination of colon cancer [365]. In gastric cancer, lymphatic dissemination and lymph-node metastasis associate with increased expression of both Twist and gelatinase B/MMP-9 [233].

Gelatinase B/MMP-9 and Extravasation
After tumour cells that arrest in the microvasculature of distant organs they either extravasate or grow within vessels [366,367], adding to the debate as to whether extravasation is indeed a critical step in the metastatic process [368,369]. Due to the positive contribution made by inflammatory cells to the metastatic process, inflammatory cell-derived gelatinase B/MMP-9 may promote extravasation, as may tumour cell derived gelatinase B/MMP-9. Alternatively, endothelial cell clusters within metastatic sites may be primed to produce gelatinase B/MMP-9 by circulating VEGF through VEGF receptors [296], which may facilitate tumour cell extravasation across an already compromised vascular BM.

Gelatinase B/MMP-9 and the Metastatic Niche
Gelatinase B/MMP-9 appears to be one of the genes required for tumor metastasis [6,9,10,12,369]. Organ specific metastatic tropism characterises the metastatic process and is a complex process that involves interaction between infiltrating cancer cells and the local environment [220]. Within the bone marrow, gelatinase B/MMP-9 regulates the recruitment and mobilization of hematopoietic stem and progenitor cells from the quiescent bone marrow niche to the proliferative niche, suggesting that gelatinase B/MMP-9 may play a similar role in cancer stem behaviour within the bone environment. In this process, gelatinase B/MMP-9 activated within the bone marrow, degrades anchorage proteins enabling haematopoietic stem cell (HSC) migration from the osteoblastic to the vascular niche, which promotes their proliferation [90]. This involves gelatinase B/MMP-9 degradation of soluble kit-ligand mobilizing factor from its membrane-associated moiety and also degradation of osteopontin, which together induce stem cell cycling and reduces anchorage to the osteoblastic niche [370]. Furthermore, osteoclasts activated within bone enhance gelatinase B/MMP-9 proteolytic activity, inducing further degradation of the endosteal-niche components osteopontin and membrane bound-stem cell factor [371]. Since cancer stem and normal stem cells share molecular machinery and cancer stem cells hijack physiological stem cell trafficking mechanisms [218], gelatinase B/MMP-9 is likely to play a similar role in stimulating the proliferation cancer stem cells that locate to the bone metastatic niche.
Within non-bone metastatic niches, increased circulating levels of gelatinase B/MMP-9 have been shown to enhance the frequency of colon cancer metastasis to lung in a mouse model. This also associates with reduced size of metastases resulting from reduced tumour vascularisation associated with increased circulating angiostatin levels [372]. Furthermore, distant primary tumours have been shown to induce gelatinase B/MMP-9 in pre-metastatic lung endothelial cell clusters via VEGF receptor-1 signalling, pre-conditioning lungs to metastatic growth, indicating that distant tumours can aggressively determine specific metastatic sites by activating endothelial cells at secondary sites [296,373], in a process involving gelatinase B/MMP-9 expressed by endothelial cells and tumor-associated macrophages that fertilizes the soil necessary for metastatic growth [374]. Many metastatic tumours also release membrane vesicles that gain access to the circulation. Micro-vesicles shed by renal cancer stem cells contain pro-angiogenic factors, including gelatinase B/MMP-9, and promote the formation of a pre-metastatic niche, which is associated with unfavourable outcome [375]. Circulating hematopoietic CD45 and Col1a positive fibrocytes have also been shown to predispose the lung to B16/F10 metastases by recruiting Ly-6C (+) monocytes, in a chemokine and gelatinase B/MMP-9-dependent manner [376]. Furthermore, stromal derived factor (SDF)-1 interaction with the chemokine receptor CXCR4, which is essential for normal stem/progenitor cell function, promotes carcinogenesis, metastasis [377] and trans-endothelial migration of cancer cells by stimulating gelatinase B/MMP-9 secretion, disrupting basement membrane and inducing vascular permeability, promoting tumour cell extravasation. This mechanism promotes cancer stem cell homing to specific metastatic niches and in particular to the bone metastatic niche [222,378].

Gelatinase B/MMP-9, Apoptosis, Survival and the Mitochondria
Gelatinase B/MMP-9 regulates cellular survival and apoptosis [195,379]. Pro apoptotic effects have been described for gelatinase B/MMP-9 in the presence of proneurotrophins [380], in cerebellar neurons and retinal ganglion apoptosis [381,382], in hypertrophic growth plate chondrocytes [268] and in HL60 pro-myelocytic leukaemia cells [383]. Pro-survival effects of gelatinase B/MMP-9 have also been described during angiogenesis, through the release and activation of mitogens from matrix stores [211]. Gelatinase B/MMP-9 localises to mitochondria via Hsp70/Hsp60, and can disrupt mitochondrial structure, function and induce mitochondrial mtDNA damage, leading to diabetic retinal capillary cell apoptosis and gelatinase B/MMP-9 inhibition protects mitochondria from ultra-structural, functional and DNA damage [384][385][386], suggesting that gelatinase B/MMP-9 inhibitors may protect against mitochondrial apoptosis. Within the extracellular environment, gelatinase B/MMP-9 promotes neuronal apoptosis by degrading basement membrane laminin [61]. In contrast, siRNA down regulation of gelatinase B/MMP-9 expression induces apoptosis in human glioblastoma cells in association with Fas death receptor-mediated caspase 3 and caspase 8 cleavage, implicating gelatinase B/MMP-9 in protecting glioblastoma cells against Fas ligand-mediated apoptosis [387]. Methylation of the miR-211 gene up-regulates gelatinase B/MMP-9 expression in glioblastoma stem cells and increases their resistance to radiotherapy and chemotherapy-induced death [18]. The activation of αβ1 and CD44 bound gelatinase B/MMP-9 induces lyn/STAT/MCL-1 signalling and apoptosis in chronic lymphocytic leukemia cells, that depends upon gelatinase B/MMP-9 hemopexin and O-glycosylation domains [318,322] and in human medulloblastoma cells inhibition of gelatinase B/MMP-9 expression promotes apoptosis through β1 integrin and ERK activation [325]. In human mammary epithelial cells gelatinase B/MMP-9 expression reduces apoptosis by up regulating cell surface Her2/Neu expression [207].

Lessons from Gelatinase B/MMP-9 Knockout and Transgenics
Gelatinase B/MMP-9 knockout reduces intestinal adenoma formation and progression within the context of the APC-min mouse model, and has identified an important role for gelatinase B/MMP-9 released by inflammatory neutrophils in the formation, proliferation and progression of intestinal adenomas in cells, exhibiting compromised APC oncosoppressor function [202]. In contrast, gelatinase B/MMP-9 elimination in Myc/BclXl and RIP1-Tag2 models of pancreatic neuroendocrine carcinogenesis impairs tumour angiogenesis but promotes tumor invasion in association with a shift in inflammatory cell content to cathepsin expressing CD11b/Gr1 positive cells at the invasive front. Plasticity in tumour inflammatory infiltrates, therefore, can alter tumour-associated protease expression to compensate for gelatinase B/MMP-9 loss, helping to explain the MMP inhibitor-induced tumour progression described in human late stage tumor clinical trials [388]. Tumours, unable to grow in gelatinase B/MMP-9 knockout mice, grow readily following auto-transplantiation of normal mouse bone marrow by a mechanism independent of endothelial cell progenitors but involving CD11b positive myelomonocytic cells. In this model gelatinase B/MMP-9 is required for tumor-associated vasculogenesis [269]. Human pancreatic cancer cell growth, impaired in gelatinase B/MMP-9 knockout mice, is promoted by gelatinase B/MMP-9 produced by parabiosed normal stromal cells, implicating stromal gelatinase B/MMP-9 in tumour progression [272]. Gelatinase B/MMP-9 knockout mice also exhibit substantial inhibition of spontaneous metastasis due to impaired triggering of the -angiogenic switch‖ [389], and in experimental metastasis models, lung metastasis formation by both melanoma and lung carcinoma cells is reduced [390,391]. The inhibition of skin and ovarian cancer metastasis formation in gelatinase B/MMP-9 knockout mice can be reversed by transplantation of normal bone marrow cells, implicating inflammatory cell gelatinase B/MMP-9 in the metastatic process and adding to the role of gelatinase B/MMP-9 in primary tumour initiation, promotion and expansion [170,390].
In contrast to these reports, transgenic α1 integrin mice exhibit increased gelatinase B/MMP-9 expression and produce high level of circulating angiostatin, which reduces primary and metastatic growth of orthotopic cancers, in association with reduced angiogenesis. This identifies an anti-angiogenic, tumour suppressing function for gelatinase B/MMP-9 [392][393][394].

Gelatinase B/MMP-9 Inhibitors and Future Directions
More than 50 broad-spectrum MMP inhibitors have been subjected to clinical trials. However, despite impressive results in non-randomized clinical trials, phase II and III clinical trials in patients with a range of different cancers have not been positive, due to a combination of factors that include a lack of understanding of the complexities of MMP involvement in tumour pathogenesis and progression, the lack of inhibitor specificity, drug intolerance and problems with drug dosage [399][400][401]. This was somewhat expected considering reports that several MMPs, including gelatinase B/MMP-9, display anti-tumor activity [72,112,113,284,285,402,403], different MMPs may be involved in different stages of tumour progression and the most potent endogenous gelatinase B/MMP-9 inhibitor, TIMP-1 may promote scattered micro-metastases in the liver [404]. Therefore, the detailed characterisation of exact roles played by the different MMPs within tumour pathogenesis and progression is required, as is the development of highly specific MMP inhibitors.
The impressive quantity of data concerning gelatinase B/MMP-9 involvement in the different phases of tumor progression, reviewed in this article, highlights particularly important roles for inflammatory leukocyte-derived gelatinase B/MMP-9 in tumour initiation and early progression and a more complex involvement of gelatinase B/MMP-9 from inflammatory, stromal and tumour sources in the continued progression of tumours to metastasis. Furthermore, reports also suggest that under certain conditions gelatinase B/MMP-9 may also protect against tumour progression by promoting the formation of systemic inhibitors of angiogenesis, may promote apoptosis and also facilitate anti-tumor inflammatory and immunological reactions (see Sections 11 and 16). Furthermore, it is also evident that under conditions of MMP inhibition malignant tumour cells compensate by undergoing mesenchymal to amoeboid transition, facilitating protease-independent progression [405].
However, with the exception of REGA-3G12, there are few specific inhibitors of gelatinase B/MMP-9 catalytic activity, attesting to the close similarity exhibited by MMP catalytic sites. A HFDDDE motif of the gelatinase B/MMP-9 catalytic domain interferes with pro-gelatinase B/MMP-9 binding of β2 integrin, inhibits OCI-AML3 leukemia cells transmigration across a human endothelial cell layers and inhibits gelatinase B/MMP-9-mediated extracellular matrix degradation, suggesting potential use in therapeutic reduction of acute myeloid leukemia cells extra medullary infiltration [408]. The peptides CTTHWGFTLC and GRENYHGCTTHWGHTLC also inhibit gelatinase B/MMP-9 catalytic activity but not gelatinase B/MMP-9 activation and inhibit primary HSC-3 tongue carcinoma growth but not tumor spread in a mouse model [409].
Recent approaches have also focussed on molecules that interfere with gelatinase B/MMP-9 activity mediated by the hemopexin and/or O glycosylation domains. Recombinant or isolated gelatinase B/MMP-9 hemopexin domain inhibits gelatinase B/MMP-9 activity [41,163], hampers colorectal cancer cell adhesion and migration [410], inhibits gelatinase B/MMP-9-induced functions in chronic lymphocytic Leukemia B cells [411], and inhibits angiogenesis in glioblastoma xenografts [291]. A peptide mimic of integrin beta chain that binds the gelatinase B/MMP-9 hemopexin domain inhibits gelatinase B/MMP-9 binding to αVβ5 integrin, prevents progelatinase B/MMP-9 activation, inhibits HT-1080 fibrosarcoma cell invasion in vitro and HSC-3 tongue carcinoma xenograft growth in vivo but does not inhibit gelatinase B/MMP-9 activity [412]. Peptide mimics of the first and fourth blades of the gelatinase B/MMP-9 hemopexin domain block gelatinase B/MMP-9 dimerization and inhibit HT-1080 and MDA-MB-435 tumour cell motility [48]. The compound N- [4-(difluoromethoxy) phenyl]-2-[(4-oxo-6-propyl-1H-pyrimidin-2-yl) sulfanyl]-acetamide also binds to the gelatinase B/MMP-9 hemopexin domain, inhibits gelatinase B/MMP-9 homo-dimerization, blocks gelatinase B/MMP-9 mediated migration and reduces xenograft tumorigenicity and metastasis of MDA-MB-435 human breast cancer cells [46]. Deletion of the O-glycosylation domain inhibits macromolecular substrate specificity of gelatinase B/MMP-9 [413], suggesting that inhibitors of O-glycosylation domain function may also be effective inhibitors of gelatinase B/MMP-9 function. Therefore, molecules that interact, interfere or compete with these two domains hold some promise in the development of specific therapeutic inhibitors of gelatinase B/MMP-9 activity and function. When considering the potential therapeutic use of specific gelatinase B/MMP-9 inhibitors, however, potential anti-tumor activity of gelatinase B/MMP-9 (i.e., production of anti-angiogenic molecules) must be taken into account and attempts made also to limit inhibitor interference with physiological gelatinase B/MMP-9 functions.

Conclusions
There is no doubt that gelatinase B/MMP-9 plays a fundamental role in tumour biology, ranging from initiation/promotion to angiogenesis, dissemination, immunological surveillance and metastatic growth. Gelatinase B/MMP-9, however, also exhibits anti-tumor activity and plays important physiological functions. It is therefore essential that specific inhibitors of gelatinase B/MMP-9 proteolytic and non-proteolytic functions are developed in order to determine the potential therapeutic efficacy of inhibiting gelatinase B/MMP-9 function in cancer therapy. The difficulty will be to inhibit the tumour promoting functions of gelatinase B/MMP-9, whilst substituting for anti-tumor gelatinase B/MMP-9 effects and minimising the inhibition of physiological gelatinase B/MMP-9 function.