TIMP-2: An Endogenous Angiogenesis Inhibitor with Distinct Antitumoral Properties

Introduction

Angiogenesis plays an essential role during development and participates in homeostatic events in adulthood such as tissue repair and reproduction.¹ Angiogenesis is deregulated in certain pathologies described as ‘angiogenesis-dependent diseases’, including heart disease, macular degeneration, rheumatoid arthritis, hemangiomas and cancer.² Tumor growth depends on the formation and recruitment of new blood vessels caused by the expression of proangiogenic proteins within the tumor microenvironment including vascular endothelial growth factor-A (VEGF-A), basic fibroblast growth factor (bFGF), and interleukin-8 (IL-8).³ During the ‘angiogenic switch’ a number of negative endogenous regulators of angiogenesis, including thrombospondin-1 (TSP-1), endostatin, interferon alpha and tissue inhibitor of metalloproteinases (TIMPs) are downregulated. The discovery of proangiogenic molecules led to the development of synthetic angiogenesis inhibitors that target the tumor endothelium and inhibit tumor angiogenesis. In 2004, the Federal Drug Administration (FDA) approved bevacisumab (Avastin), the first antiangiogenic drug for metastatic colon cancer that targets and neutralizes VEGF-A, thereby inhibiting proliferation, migration and survival of vascular endothelial cells within the tumor microenvironment. The FDA has also approved several drugs with antiangiogenic activity including sorafenib (Nexavar), sunitinib (Sutent), pazopanib (Votrient), and everolimus (Afinitor), while others such as angiostatin, thrombospondin and endostatin remain in various stages of development. Studies have shown that the direct angiogenesis inhibitors are less toxic than other chemotherapeutic drugs, but not completely without side effects. Since they target normal endothelium and not tumor cells, they also have a lower risk of introducing drug resistance, although tumors may use genetic instability to switch or escape dependence on VEGF-A as a primary angiogenic stimulus.⁴ There is, however, a group of angiogenesis inhibitors that have been shown to prevent angiogenesis indirectly via the disruption of oncogenic proteins in tumor cells, eg. receptor tyrosine kinases, that drive angiogenesis as part of their signaling pathways. However, oncogenic mutations are easily activated leading to drug resistance and tumor recurrence.⁴

Endogenous Inhibitors of Angiogenesis

It is therefore apparent that therapeutic strategies have to be constantly re-evaluated in order to guarantee success in the clinic. Over the past 30 years a number of laboratories have identified and reported the existence of approximately 30 endogenous angiogenesis inhibitors. These inhibitors are naturally produced by various cell types and exist in the blood circulation (plasma) or in the extracellular matrix making them excellent therapeutic candidates with no or little side effects.⁵ However, none have been sufficiently tested to obtain FDA approval as anti-cancer agents, and only a few are still in various phases of preclinical development and/or clinical trial.

Dr. Judah Folkman once asked the question: ‘Do endogenous angiogenesis inhibitors suppress angiogenesis-dependent disease?’ Indeed, several studies have demonstrated that endogenous angiogenesis inhibitors have the ability to suppress pathologic angiogenesis in the context of disease (eg. rheumatoid arthritis, hemangiomas, cancer).^6,7 The first endogenous angiogenesis inhibitor to be discovered, interferon alpha, is used to treat several cancers including malignant melanoma, and hemangiomas in newborns and infants.⁸ Several potent angiogenesis inhibitors derive from the proteolytic cleavage of larger proteins found within the extracellular matrix or circulation.² Tumstatin is a protein fragment cleaved from the alpha chain of collagen IV.⁹ Genetic studies have shown that tumors grow faster in tumstatin-deficient mice (that have a deletion of the alpha chain of type IV collagen) than in wild-type mice. Also when tumstatin was administered back into these animals, tumor growth slowed down dramatically. Tumor growth was also reduced when the tumstatin receptor, integrin β₃, is deleted.¹⁰ Endostatin is a protein fragment derived from type XVIII collagen and has been shown to completely suppress tumor-induced angiogenesis and tumor growth in vivo including Lewis lung carcinoma and B16F10 melanoma.¹¹ When treatment ceased, tumors began to grow. Thrombospondin-1 null mice showed significantly faster tumor growth than wild type mice.¹² Moreover, transgenic expression of thrombospondin-1 or −2 lead to decreased tumor angiogenesis and reduction of tumor growth in mice with squamous cell carcinoma xenografts.⁷ In many cancers, the downregulation or loss of antiangiogenic protein expression occurs prior to upregulation of proangiogenic stimuli attributed to the ‘angiogenic switch’.¹³ The involvement of exogenous administration of endostatin, tumstatin and thrombospondin-1 peptides in regulating the so called “angiogenic switch”, initiation of tumor growth, tumor progression and survival was recently investigated in a pancreatic islet cell tumorigenesis mouse model (Rip1/Tag2 model) with VEGF-A deletion (RT2/VEGF knock out mice).¹⁴ While VEGF-A activated the ‘angiogenic switch’ in wild type mice, thrombospondin-1 and endostatin, but not tumstatin peptides, prevented VEGFR-2 activation, followed by inhibition of tumor angiogenesis in the early stages of tumor growth. However, all three were effective in inhibiting tumor progression at the intervention stage. In loss of function studies, it was shown that all three were important in controlling tumor cancer progression and survival. These experiments suggest the essential role of endogenous angiogenesis inhibitors in controlling the “angiogenic switch” early on in tumorigenesis, counterbalancing the proangiogenic stimuli and inhibiting tumor growth and tumor progression.

The majority of endogenous antiangiogenic molecules constitute fragments of larger proteins and macromolecules that may have normal functions totally unrelated to angiogenesis.² For instance, endostatin and tumstatin fragments derive from collagen, angiostatin from plasmin/plasminogen and endorepellin from the C terminus of perlecan. Recent studies have also shown that peptides from Tissue Inhibitor of Metalloproteinases −2 (TIMP-2) have antiangiogenic properties.^15,16 TIMP-2 primarily functions to inhibit Matrix Metalloproteases (MMPs), proteins that on their own promote angiogenesis.^17,18 TIMP-2 could potentially be translated into a promising antiangiogenic and anti-cancer therapeutic agent.

Angiogenesis and TIMPs

There are four members in the TIMPs family that share significant homology and structure.^19,20 TIMPs 1–4 are relatively small proteins that by definition inhibit the endoproteolytic activity of MMPs and some members of the adamalysin (ADAMs) family members that lead to degradation of most, if not all extracellular matrix components.²¹ Within the context of physiological and pathological conditions, TIMP functions are primarily explained by their ability to inhibit MMPs, however, several novel biological functions have emerged within the last decade. The details of many of these mechanisms are not dependent on MMP inhibition, but rather through newly identified TIMP-dependent pathways that have yet to be fully characterized.^22,23 Indeed, studies have revealed that TIMPs regulate several biological activities including cell growth, migration, invasion, angiogenesis and apoptosis, and that these cellular effects are mediated independent of their MMP inhibitory activity (reviewed in²⁰).

In cancer, MMPs have been associated with the disruption of the sub-endothelial matrix resulting in enhanced tumor angiogenesis.^24–26 Although there is no evidence that TIMP-4 inhibits endothelial cell proliferation and in vivo angiogenesis, TIMPs 1–3 have been shown to be potent endogenous angiogenesis inhibitors in vitro and in vivo.²⁷ In a TIMP-1 transgenic mouse model, TIMP-1 suppressed Ehrlich tumor cell growth and neovascularization.^28,29 When TIMP-1 was introduced into Burkitt’s lymphoma xenografts, an initial increase and subsequent suppression of tumor growth was observed.³⁰ Tumor regression was associated with a failure of these tumors to develop tumor angiogenesis. TIMP-3 has been shown to act as a VEGF-A antagonist by interacting with vascular endothelial growth factor receptor-2, the primary VEGF-A receptor expressed on endothelial cells.³¹ This interaction prevents VEGF-A binding to its receptor and therefore inhibits endothelial cell proliferation and survival. A possible interaction between TIMP-3 and angiotensin II type 2 receptor, as identified with yeast two-hybrid screening, may also additively inhibit angiogenesis.³² In addition, specific TIMP-3 mutations are associated with Sorsby Fundus Dystrophy, an inherited genetic disease characterized by submacular neovascularization and loss of vision.^33,34

TIMP-2

TIMP-2 is a unique member of the TIMP family in that, in addition to inhibiting MMPs, it interacts with MMP-14 (or MT1-MMP) on the cell surface to facilitate the activation of pro-MMP-2.^35,36 TIMP-2 was first isolated as a complex with pro-MMP-2 from the conditioned media of human A2058 melanoma.³⁷ TIMP-2 isolated from bovine cartilage was shown to have anti-angiogenic activity in 1990.³⁸ It is also the only member of the TIMP family that has been shown to directly inhibit endothelial cell growth. Early studies demonstrated that endothelial cells treated with exogenous TIMP-2 prior to bFGF stimulation had significantly less proliferative potential.³⁹ However, TIMP-1 protein or the potent synthetic MMP inhibitor BB-94, were not shown to have the same effect, suggesting that the anti-angiogenic properties of TIMP-2 were partially MMP-independent. More recently, TIMP-2 mediated inhibition of endothelial cell proliferation was uncoupled from its MMP inhibitory function.⁴⁰ Our laboratory showed that pre-treating endothelial cells with either TIMP-2 or Ala+TIMP-2, a TIMP-2 mutant with an alanine appended at the N-terminus that also lacks MMP inhibitory activity, suppressed VEGF-A or bFGF stimulation of endothelial cell growth. The mechanism involved reduction of the cognate tyrosine kinase growth factor receptor phosphorylation and activation, and translocation of the protein tyrosine phosphatase, Shp-1. More specifically, TIMP-2 was shown to bind to the integrin α₃β₁ receptor on the endothelial surface, while competition experiments with blocking antibodies to each integrin subunit led to loss of TIMP-2 growth suppressive activity. This was the first study to demonstrate that TIMP-2 binds to the endothelial cell surface via a receptor, and that the growth inhibitory activity is dependent on this interaction. Subsequent experiments with exogenous treatment of TIMP-2 or Ala+TIMP-2 demonstrated induction of de novo expression of cyclin dependent kinase inhibitor p27 in microvascular endothelial cells and hypophosphorylation of pRb leading to G1 cell cycle arrest.⁴¹ TIMP-2 also regulates endothelial cell migration via upregulation of the reversion-enhancing, cysteine-rich protein with Kazal motifs (RECK), a membrane bound protein that inhibits MMPs.^42,43 Enhanced RECK expression and inhibition of cell migration occur as a result of Src tyrosine kinase inactivation and modification of paxillin phosphorylation pattern. In turn, this results in small guanosine triphosphatase (GTPase) Rac1 inactivation and subsequent enhanced Rap1 activation leading to RECK upregulation and decreased motility.

Recent studies identified two separate regions of TIMP-2 involved in its antiangiogenic activity, and confirm that the two functions of TIMP-2, the metalloproteinase inhibition and the antiangiogenic activities, are distinct. In the first study, the antiangiogenic function is localized at a 24-mer region at the N-terminus of TIMP-2, within the B-C loop.¹⁶ Synthetic peptides from that region contain the α₃β₁ binding domain that interacts with the endothelial cell surface to suppress VEGF-A-stimulated endothelial mitogenesis. Finally, using a murine Kaposi’s sarcoma model, a tumor derived from the endothelium, the α₃β₁ binding peptides were shown to have significant anti-tumor activity in vivo.

In the second study, the mechanism involved in the antiangiogenic activity implicates a 24-mer at the C-terminal domain of TIMP-2 known as loop 6.¹⁵ Since the MMP inhibitory activity is located at the N-terminus of TIMP-2, inhibition of angiogenesis due to TIMP-2 mediated MMP inhibition is not questioned.^44,45 Further, TIMP-2 loop 6 was reported to directly bind to the insulin-like growth factor receptor I (IGF-IR) on endothelial cells, a tyrosine kinase receptor known to regulate tumor growth and angiogenesis.^46,47 Upon this interaction, both AKT and MAPK pathways were analyzed for Akt and/or Erk phosphorylation in vitro. Both pathways were inhibited when cells were pretreated with loop 6 suggesting that the downstream IGF-IR signaling was disrupted. Moreover, in vivo administration of exogenous loop 6 in SCID mice transplanted with PC-3 tumor cells resulted in inhibition of tumor volume and reduced tumor angiogenesis. This antiangiogenic mechanism occurs independently of TIMP-2 interaction to its receptor α₃β₁ or to MT1-MMP. These results suggest that loop 6 inhibits endothelial cell proliferation and angiogenesis by interacting with IGF-IR and disrupting its downstream signaling through Akt and Erk.

The expression of angiogenesis inhibitors such as TIMP-2 is frequently downregulated in many types of cancer. In a recent study, we overexpressed TIMP-2 and the mutant Ala+TIMP-2 in A549 lung cancer cells.⁴⁸ Our goal was to determine if the apparent MMP-independent antiangiogenic activity of TIMP-2 (Ala+TIMP-2) was sufficient to suppress in vivo tumor growth. In vitro, TIMP-2 or Ala+TIMP-2 did not alter A549 basal cell growth (in the presence of serum, not growth factor stimulated). However, in a chemoattractant migration assay both TIMP-2 and Ala+TIMP-2 inhibited the migration of A549 tumor cells. The inhibitory effects were further demonstrated in a chemoinvasive assay where both cell lines showed significantly lower invasive properties. These experiments suggest that TIMP-2 is able to directly regulate A549 tumor cell motility and invasion, independent of MMP-2 inhibition, without altering the growth in vitro. We then subcutaneously implanted the cells into either nude or NOD-SCID mice in order to investigate the impact of TIMP-2 on the tumor microenvironment and tumor growth (Figure 1). In either mouse model, both TIMP-2 and Ala+TIMP-2 inhibited tumor growth suggesting that the TIMP-2 anti-tumoral effects result from the TIMP-2 interaction with the tumor microenvironment. In addition, the fact that Ala+TIMP-2 was also able to inhibit tumor growth suggests that MMP inhibition and MMP-independent suppression of tumor growth by TIMP-2 are regulated by distinct mechanisms.

Figure 1.

In vivo cell growth of A549 cells, retrovirally transfected to stably overexpress TIMP-2 or Ala+TIMP-2 (TIMP-2 mutant with no MMP inhibitory activity). A549 EV control, TIMP-2 and Ala+TIMP-2 cells were cultured to 80% confluency, counted, and washed twice with PBS. Equal numbers of cells (5×10⁶) were resuspended in 100 µl PBS and injected subcutaneously into the hind flank of 4–6 week old female NOD-SCID mice (n=15 mice per group). Tumor volumes (mm³) were determined using caliper measurements twice a week starting at day 7 up to day 21 post-inoculation. Graph shows effects of TIMP-2 and Ala+TIMP-2 overexpression on A549 xenograft tumor growth. At the end of the experiment, tumor volumes of A549 TIMP-2 and Ala+TIMP-2 xenografts are significantly smaller than that of the EV control (Two-way ANOVA, ***p<0.001, ****p<0.0001) whereas no significant difference is shown between A549 TIMP-2 and Ala+TIMP-2 xenografts.

A detailed immunohistochemical analysis of tumor sections showed decreased tumor angiogenesis and increased tumor cell apoptosis in both TIMP-2 and Ala+TIMP-2 A549 xenografts. These data indicate that TIMP-2 controls tumor growth through its interaction with the microenvironment, i.e. inhibits tumor angiogenesis and as a result induces tumor cell apoptosis (although a direct regulation of apoptosis by TIMP-2 can not be excluded). In order to determine what pathways might be impacted by TIMP-2 during tumor progression, we analyzed the levels and activities of Focal Adhesion Kinase (FAK) and AKT proteins in tumor sections by immunohistochemistry. Since TIMP-2 was able to inhibit A549 migration and invasion we looked at the FAK levels and activation. We showed that FAK levels were reduced and its activation was impaired in the tumor cells. Similarly, total and activated AKT levels were also reduced. Given that the AKT signaling pathway occurs downstream of FAK this might explain the reduced AKT activity. However, as it was previously shown, AKT phosphorylation was also decreased in endothelial cells upon TIMP-2 loop 6 interaction with IGF-IR.¹⁵ Recent studies have demonstrated that in human pancreatic adenocarcinoma cells, FAK and IGF-IR physically interact and colocalize on the focal adhesions.⁴⁹ Decreased phosphorylation status for both molecules resulted in reduced cell viability and increased apoptosis.⁵⁰ It is therefore possible that TIMP-2 loop 6 acts as an antagonist and disrupts this complex, however, further experiments are needed to answer this question.

The TIMP-2 regions and specific peptides/fragments recently identified as responsible for its antiangiogenic and antitumorigenic effects, further supports the view of developing TIMP-2 as a therapeutic. The potential use of endogenous angiogenesis inhibitors as prophylactic or adjuvant regimens for cancer therapy should be further investigated. Cocktails of certain antiangiogenic fragments derived from different angiogenesis inhibitors could synergistically and effectively target not only different proangiogenic or protumorigenic molecules but also prevent the initiation of tumor cell growth.