Abstract
A better understanding of the interplay between the tumor environment, the immune system, and hemostatic apparatus is essential to effectively improve cancer treatment. Histidine-rich glycoprotein (HRG) is an abundant plasma protein with a wide array of functions. HRG has the ability to bind multiple ligands thereby modulating immunity, cell adhesion, angiogenesis, and thrombosis. Many of these functions are involved in tumor progression and antitumor response. We outline current data on HRG as an important potential player and as a potential future target for cancer therapy.
The interactions of the tumor environment, the hemostatic apparatus, and anticancer immune control are of paramount importance in cancer development and outcome (1-4). Histidine-rich glycoprotein (HRG) is a relatively abundant plasma protein with two cystatin-like domains and a wide spectrum of targets and functions (5, 6). First described in 1972, the role of HRG was originally identified as regulating clotting and fibrinolysis due to the association of venous thrombosis with elevated levels of HRG (7). As an adaptor protein, HRG is able to modulate multiple interactions and many other roles have since been described.
One of the more recent discoveries has been the role of HRG in pathogen control, which has provided additional insight into the function of HRG in disease states. HRG has anti-inflammatory effects as demonstrated in HRG-deficient mice that had higher bacterial burden and higher levels of the pro-inflammatory cytokine interleukin-6 than wild-type mice after Streptococcus pyogenes challenge (8). The anti-fungal properties of HRG, particularly at low pH, were demonstrated in a mouse model of candidiasis in which HRG-deficient mice were susceptible to infection and wild-type mice were resistant (9).
While the full spectrum of physiological roles of HRG is not yet fully described, its importance in hemostasis and angiogenesis, influence on tumor progression and metastasis formation, and the antitumor immune response are gradually coming to light.
Structure and Isolation of HRG
Structure of HRG. The human HRG gene consists of seven exons and six introns totaling approximately 12 kb with the C-terminal half of HRG encoded in exon 7 (5). Located on human chromosome 3q27, the HRG gene is located adjacent to the other cystatin type 3 proteins: fetuin-A, fetuin-B, and kininogen (5).
The protein has two cystatin-like domains, a histidine-rich region (HRR) that is surrounded by two proline rich regions, and a C-terminal domain (Figure 1). Multiple tandem repeats of the motif GHHPH in the HRR contribute to the high histidine content (10). In humans, two polymorphisms of HRG have been identified; both 77-kDa and 75-kDa proteins have been isolated from human blood, which differ by a serine or proline at position 186, respectively (5). The serine allows for a N-glycosylation site, increasing the molecular weight by the addition of carbohydrates (11). There are five disulfide bonds which protect the protein from proteolytic cleavage (Figure 1) (12).
Purification of HRG from plasma. HRG was originally identified in serum as a protein that had a high affinity for carboxymethylcellulose (13, 14). The most common method for purifying HRG has been with phosphocellulose and nitrilotriacetic acid chromatography, and these preparations have been used to determine the biological activities of HRG in vitro (15, 16). However, isolation of HRG using these methods can result in retention of HRG-binding partners, including immunoglobulin, plasminogen, and fibrinogen (16). For example, HRG purified from plasma may contain plasminogen contaminants, which may stimulate angiogenesis (16-18). Additionally, these purification methods often require further downstream steps that reduce protein yields and create a mixture of proteolytically-cleaved HRG species. These cleavage species may have biologically different effects those of the parent protein.
Patel et al. have recently addressed this issue by purifying HRG using a two-step process combining metal ion affinity and ion exchange chromatography that enables further study of HRG without contaminating proteins or cleavage products (16). The metal ion affinity step using cobalt sepharose beads takes advantage of the histidine-rich sequences within HRG that efficiently bind immobilized metal ions. Combined with ion exchange chromatography, Patel et al. were able to generate intact HRG without detectable cleavage products (16). Subsequent studies of the purified protein suggest that some previously-reported functions of HRG, such as the binding of heparin and phosphatidic acid are consistent with previous reports (16). However, other previously reported functions, such as mediating phagocytosis of necrotic and apoptotic cells, may be due to contaminating plasma proteins (6, 10, 13, 16, 19-22). In vitro studies to fully-elucidate the interactions of HRG will require for contaminant-free purified protein.
HRG Deficiency
Hrg-deficient mice. In the steady state, Hrg-deficient mice have no gross abnormalities and are viable and able to reproduce (23). HRG deficiency is gene-dosage dependent with HRG+/− mice having half the serum concentration of Hrg as Hrg+/+ mice (23, 24). Targeted deletion of Hrg in mice is associated with increased anti-thrombin activity, increased fibrinogen, and dysregulation of platelet function, leading to shorter bleeding time that is more remarkable in females, indicating that HRG functions both as an anti-coagulant and anti-fibrinolytic in vivo (23). Mice deficient in HRG have fewer platelets in the peripheral blood, which is also indicative of increased platelet activation (25).
HRG deficiency in humans. Congenital deficiencies in HRG have been reported in several families (26, 27). The Tokushima allele is a single missense mutation in exon 3 which generates a glycine amino acid instead of glutamic acid in the first cystatin-like domain, leading to a reduction in circulating HRG of 20% that of normal levels and is associated with thrombophilia (26, 28).
Disease states also alter HRG levels. HRG levels have been measured in patients with sepsis, heart disease, liver cirrhosis, and in the acute-phase response (22, 29, 30). During pregnancy, and as a result of low dose oral contraception HRG levels are also lowered (31). In women with pre-eclampsia, reduced levels of HRG and increased levels of fibrinogen have been detected in the placenta suggesting that both factors are involved in the angiogenic alterations observed during pregnancy (32).
Binding Partners of HRG. The HRR within HRG is able to bind several hemostatic factors that include heparan sulfate, heparin, tropomyosin, and heme reviewed in Poon et al. (6). HRG also binds divalent metal ions including Cd2+, Ni2+, and Cu2+, and many of its known interactions require or are enhanced by the presence of Zn2+, including binding to heparan sulfate (21, 33). In vitro experiments showed that in the presence of Zn2+, HRG prevents binding of a murine T-cell line to lamin, collagen, or fibronectin-coated tissue culture dishes (34). Furthermore, cell surface binding of HRG is mediated through glycosaminoglycans (GAG) as demonstrated using GAG-sufficient and -insufficient Chinese hamster ovary cell lines (21). These results indicate that HRG modulates adhesion of cells both to each other and with components of the extracellular matrix.
HRG may interfere with heparan sulfate degradation by heparanase through the occlusion of cleavage sites. Importantly for tumor control, HRG binding may also prevent heparanase-mediated fibroblast growth factor release from the extracellular matrix (12).
Heparin binding to HRG is zinc-dependent and is enhanced at low pH levels, which is of particular importance in wound healing, infection, and the low pH environment of the extracellular space in tumors (35). There is a high degree of sequence homology between HRG and antithrombin at the putative N-terminal heparin binding site, which allows HRG to effectively limit the anti-coagulant activity of antithrombin by binding heparin (22).
HRG is able to bind tropomyosin under very specific conditions, such as platelet activation during coagulation. Zn2+ is released by α-granules and the local concentration of zinc increases to sufficient levels to support HRG binding to tropomyosin (12). This also has implications for tumor control as HRG binding of tropomyosin may be anti-angiogenic (12).
Functionally, HRG was initially thought primarily to inhibit fibrinolysis and have pro-thrombotic potential (22, 29, 30, 36). Fibrinogen-plasminogen binding is reduced in the presence of HRG, leading to a reduction in available plasminogen that can be converted to plasmin and degrade fibrin clots (29, 37). This enables HRG to act as an anti-fibrinolytic (29). HRG has also been thought of as an anti-thrombotic, as it can replace fibrin-bound thrombin in fibrin clots, in a Zn2+, dependent manner (18, 38). In the presence of Zn2+, HRG has a high affinity for factor XIIa and its addition increases clotting times, but not FIXa or FXIa suggesting that the interaction is specific to the intrinsic coagulation pathway (39).
Effects of HRG on Angiogenesis: Pro- and Anti-angiogenesis
Angiogenesis is essential for tumor growth and metastasis. During angiogenesis, blood vessels are destabilized when pericytes detach from endothelial cells forming the vessel walls. New vessels are formed by endothelial cell sprouting and pericytes then re-attach to form mature vessels. Proper contact between endothelial cells and pericytes ensures vessel integrity and limits metastasis (40). The angiogenic switch occurs when pro-angiogenic factors, such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF), out-weigh anti-angiogenesis and tumor progression occurs. HRG is reported to have both pro- and antiangiogenic properties, owing to its multi-domain structure and the activities of its proteolytically-released fragments, notably the histidine/proline rich region (5, 15, 17, 41).
Potential pro-angiogenic activity. HRG localizes in the stroma of some tumors upon release from platelets and may block interaction of anti-angiogenic proteins such as thrombospondins (42). In a study utilizing the thrombospondin (TSP-1) receptor CD36 and Hrg-deficient animals, it was hypothesized that HRG interacts with TSP-1 and promotes angiogenesis by blocking binding to CD36 via thrombospondin type-I repeat domains (TSR) (43). CD36 is expressed on a variety of cell types, including macrophages, monocytes, platelets, and some endothelial cells, and upon binding of TSP-1, counteracts angiogenic signals mediated through fibroblast growth factor receptors. Hrg−/− mice did not control TSR-transfected B16F1 tumors as well as did wild-type control mice (43).
Klenotic et al. showed that HRG can also have pro-angiogenic properties in a brain cancer model by acting as a decoy receptor for CD36 (44). HRG was expressed primarily in the basement membrane in 83% of grade IV human glioblastomas (44). In the absence of HRG, the angiostatic protein vasculostatin, produced by glial cells, binds CD36 on endothelial cells resulting in apoptosis. In the setting of vascular inflammation, HRG from plasma or platelets binds vasculostatin preventing caspase- mediated apoptosis and encouraging angiogenesis through endothelial cell proliferation. HRG was found to co-localize with TSP-1 in the tumor matrix of human breast cancer and masked the anti-angiogenic epitope of TSP-1, inhibiting its anti-angiogenic properties (42).
Anti-angiogenic activity. The quality of tumor vasculature is impaired in Hrg−/− mice. In contrast to the reported pro-angiogenic activity of HRG, Hrg−/− mice bearing T241 fibrosarcomas had tumor vessels with increased area as well as decreased coverage and vessel perfusion (45). T241 fibrosarcomas were transduced with HRG to force stromal deposition of HRG and evaluate overexpression of HRG on tumor growth. HRG-transduced tumors showed improved tumor vasculature including increased vessel perfusion and percentage of pericyte covered vessels (46). A lack of pericyte coverage on tumor vessels has been associated with increased hypoxia, epithelial to mesenchymal transition, and reduced overall survival (47). Increased hypoxic regions and hemorrhage were associated with lack of HRG (46). About 60% of HRG-transduced Panc02 tumors had normal vasculature in comparison to nearly all control tumors, which exhibited multiple layers of disconnected endothelial cells and luminal protrusions. Accordingly, the vessel area was increased and vessels had less coverage and perfusion (45). These reports are also consistent with the orthotopic RipTag2, model where HRG deficiency led to decreased pericyte coverage and vessel perfusion (48).
Hrg−/− mice also had increased production of VEGF and placental growth factor (PIGF) (45). HRG inhibited both VEGF and FGF-mediated angiogenesis, and inhibited chemotaxis of endothelial cells in vitro (41). The minimal active domain of the HRG (amino acids 330-364) was identified from the histidine/proline rich region and has potent anti-angiogenic activity through inhibition of endothelial cell chemotaxis (15).
Sub-cutaneous administration of HRG to mice with established T241 fibrosarcomas resulted in reduced vascularization and improved tumor control (41). Additionally, the presence of angiogenic islets is increased in insulinomas in Hrg−/− mice with increased tumor proliferation and a corresponding two-fold increase in tumor volume (48). The development of gliomas was delayed and reduced in severity in HRG-treated mice, which coincided with improved tumor vasculature (49). These data suggest that HRG partially mediates tumor control through normalization of blood vessels.
The different results between the studies that deemed HRG to be pro-angiogenic and those that found it to have antiangiogenic properties may be explained by the absence of CD36 on larger vessels, or in that the HRR domain of HRG may have qualities that are independent of the parent protein. Whether or not HRG is pro-angiogenic or anti-angiogenic may be highly contextual depending on the components of the tumor environment, or dependent on proteolytic cleavage in vivo (Table II) (41, 50).
Effects Due to Interactions with Platelets and Other Coagulation Factors
The expression of tissue factor by tumor cells promotes platelet activation and coagulation. Tumor-derived tissue factor also supports the metastatic potential of tumors (2, 51). Platelet activation and coagulation play an important role in tumor metastasis, possibly by shielding circulating tumor cells from immune surveillance and reducing endothelial barrier function (52). Increased platelet counts are associated with tumor metastasis and poor prognosis. Platelet-stored HRG is released upon activation (24, 36) increasing local concentrations at selected sites, such as sites of infection or tumors.
The histidine/proline-rich fragment of HRG was detected in human kidney cancer sections and co-localized with platelets, identified by CD41 and CD42b (24). This was associated with anti-angiogenic properties, indicating that platelets could be a major source of HRG for tumors (24). In the absence of HRG, platelet activation becomes enhanced and the angiogenic switch was accelerated in an orthotopic pancreatic cancer mouse model described by Ringvall et al. (25).
Cedervall and colleagues recently investigated the role of platelet-released HRG in epithelial-to-mesenchymal transition in a mouse model of pancreatic cancer (48). They demonstrated that HRG is a potent tumor suppressor by regulating platelet signaling. In Hrg−/− mice, platelet depletion, early during tumor development suppressed the enhanced angiogenic switch associated with Hrg-deficiency (48). When platelets were depleted in more advanced tumors there was also improved tumor control in Hrg−/− mice (48). Taken together, these data suggest that HRG counteracts proangiogenic factors associated with platelet activity (Table II). Platelet depletion also restored levels of tissue factor regulated E-cadherin (48). Loss of the cell-cell adhesion molecule E-cadherin is associated with the promotion of metastasis, and its restoration offers an explanation for the decreased numbers of metastases in platelet depleted Hrg−/− mice (51).
Impact on Other Hemostatic Factors
Fibrinolysis, coagulation, and angiogenesis are modified by the interaction of HRG with multiple targets including thrombospondins, heparin, and complement. Either directly or indirectly, the regulation of these processes can all have an effect on tumor control (5, 6, 17, 45, 46, 53).
HRG is able to limit the anti-coagulant activity of anti-thrombin by binding to heparin at the same sites as anti-thrombin (54). In circulation, HRG is bound to plasminogen, reducing plasminogen available to bind fibrin (53). HRG further functions as an anti-coagulant by incorporating into fibrin clots, mediated by the HRR, which is consistent with the shorter bleeding times observed in Hrg-deficient mice (18, 23).
HRG and the Modulation of Antitumor Immune Responses
Modulation of tumor associated macrophages (TAMs). Myeloid- derived TAMs are a key component of tumor stroma and depending on the tumor microenvironment may possess anti-tumor (M1) or pro-tumor (M2) characteristics (55-57). TAMs are involved in regulating the angiogenic switch (58) and their presence in tumors are associated with poor outcomes (59-62). The classically-activated M1 TAMs are stimulated by lipopolysaccharide and interferon gamma (IFN-γ) to produce chemokines such as chemokine (C-X-C motif) ligand 9, chemokine (C-X-C motif) ligand 10, and high levels of interleukins IL-12 and IL-23. These cytokines then promote Th1 T-cell mediated anti- tumor responses via STAT4 signaling in naïve and memory CD4+ T-cells (63). In contrast, M2 TAMs, which can express programmed death ligand 1 (PD-L1) and produce chemokine (C-C motif) ligand 17, chemokine (C-C motif) ligand 22, and high levels of IL-10 and IL-1RA, are implicated in tumor angiogenesis, metastasis and immune suppression through promotion of Th2 T-cell responses and inhibition of Th1 CD4+ T-cells (55, 56). As macrophage polarization is plastic and can shift based on environmental cues in different tissues (64, 65), there has been considerable interest in therapeutic interventions that promote skewing towards an M1 phenotype (57, 63, 64, 66-68). A recent study in a mouse model of established gliomas indicated that inhibition of colony- stimulating factor 1 receptor resulted in re-polarizing TAMs from an M1 to a M2 phenotype, and in improved tumor control and increased survival (69).
In Hrg-transduced tumors, TAMs were skewed towards an M1 phenotype, including lower levels of IL-10 production, which permitted increased accumulation of DC, natural killer (NK), and CD8+ T-cells (46). Highlighting the importance of HRG on TAM skewing, clodronate depletion of macrophages resulted in equivalent tumor control in HRG-transduced tumors and control tumors (46).
The ability of HRG to modulate TAMs may be due to the regulation of the VEGF family member PlGF produced by TAMs. HRG specifically down regulates PIGF in TAMs, leading to reduced production of IL-10 and the observed skewing towards an M1 phenotype (46). In vivo studies with PIGF-deficient animals indicated that the improved tumor control observed in Hrg-transduced tumors was a direct cause of PIGF down-regulation, including tumor vessel normalization, as discussed previously (46). Using pimonidazole to detect hypoxic regions, Rolny et al. were able to detect a significant reduction in hypoxia in Hrg-transduced tumors (46). As macrophages are more likely to skew towards an M2 phenotype under hypoxic conditions, HRG may exert its effects partially through affecting changes in oxygenation of the tumor environment. Expression of genes associated with hypoxia, including Arg1, Mmp8, and Ccl12 were increased in TAMs from Hrg−/− mice (45).
These results suggest an important role for HRG in programming TAMs (Table III). The signature of HRG-influenced TAMs appears to be specific for inhibition of tumor vessel growth and metastasis, with reductions in pro-angiogenic cytokines such as IL-1β and of TNF-α (46). Targeting of TAM polarization to improve tumor control through modulation of the tumor microenvironment is of great interest (63, 68, 70). Tumors that are particularly affected by TAMs, such as pancreatic cancer, may be especially sensitive to TAM-targeted therapies. HRG may prove to be a valuable tool for modulating TAMs towards an antitumor phenotype (67, 71).
Tumor infiltrating lymphocytes affected by HRG. Analysis of tumor immune infiltrates indicated that HRG deficiency resulted in less infiltration of cells able to mount an anti-tumor response, and reduced levels of angiostatic and T-cell attractant factors, explaining the reduced number of tumor infiltrating CD8+ T cells (45). Expression of genes associated with DC maturation and T-cell priming are specifically up-regulated in TAMs upon HRG stimulation (45). Correspondingly, HRG-transduced tumors showed increased infiltration of antitumor immune cells (46). Increased accumulation of activated DC were observed in HRG-transduced tumors, which corresponded with increased accumulation of CD8+ T-cells (46). NK cells were also increased in HRG-transduced tumors (46). Figure 2 illustrates the multi-faceted effects of HRG.
Does HRG Have Prognostic Value for Patients with Cancer?
HRG is easily detectable in the steady-state at high concentrations in plasma and lowered levels are detected in certain disease conditions, including some types of cancers in comparison to healthy tissue (6, 46). Proteomics approaches have been employed to determine if HRG could be useful as a prognostic tool in several types of cancers. In a study to identify potential glycoprotein biomarkers in ovarian cancer, lower levels of HRG compared to healthy donor controls were associated with stage I/II ovarian cancer (72). Lectin-specific glycosylated proteins in the serum of patients with ovarian cancer and normal donor controls were isolated using a lectin array. The proteins were further identified by liquid chromatrography and tandem mass spectrometry (LC-MS/MS) and confirmed by enzyme-linked immunosorbent assay (ELISA). Fucosylated HRG levels were significantly higher in patients with stage III ovarian cancer compared to normal and benign donors, but was not significantly higher in patients with stage I/II disease (72). Overall, levels of HRG were lower in patients with ovarian cancer. The combination of HRG and corticosteroid-binding globulin, also identified by lectin array, differentiated between normal donors and patients with early-stage ovarian cancer. These glycosylated proteins could potentially be used in conjunction with CA125 screening to detect ovarian cancer.
A Phase III trial (NCT01664169) that compared patients with pancreatic cancer receiving gemcitabine with or without bevacizumab, anti-VEGF monoclonal antibody, identified HRG as a weakly-positive prognostic biomarker (73). LC-MS/MS on patient serum samples initially identified HRG as a predictive protein along with complement factor H. Further screening by ELISA on these samples suggested that HRG was weakly-associated with improved overall survival, but not affected by bevacizumab treatment (73). This effect may be due to the anti-angiogenic properties of HRG.
A proteomics study using isobaric tagging and LC-MS/MS analysis of 20 patients with endometrial disease found that those with atypical hyperplasia had significantly lower levels of HRG in comparison to normal donor controls, potentially indicating increased likelihood of progression (74). In a matrix-assisted laser desorption/ionization time-of-flight spectrometry analysis of preoperative biopsies of breast cancer, HRG was elevated in samples from obese patients, potentially identifying an inflammatory response (75). Given the difficulty of early identification of both ovarian and pancreatic cancer, future studies of HRG as a biomarker are warranted.
HRG as Adjuvant Therapy
In pre-clinical settings, administration of HRG synthetic peptides have demonstrated anti-angiogenic activity (41). This suggests that HRG supplementation is a potential therapeutic and has the added benefit of positively modulating the immune system in the tumor microenvironment (15, 25, 41, 45, 76).
Tumors transfected with HRG had improved responses to sub-therapeutic doses of chemotherapy, likely through the normalization of tumor blood vessels. Adjuvant HRG could improve responses to chemotherapy or make it possible to titrate chemotherapy doses in order to limit toxicities (46). The normalization of tumor blood vessels also makes HRG supplementation attractive as part of an adoptive cell transfer strategies and antibody therapies. Normalized tumor vasculature makes it possible for immune cells to infiltrate the tumor and exert and an antitumor effect, as well as allowing efficient delivery of therapeutic agents to the tumor site (77). In HRG-transduced tumors, numbers of activated DC, NK, and CTL were increased, suggesting that HRG therapy may help facilitate trafficking of both endogenous and adoptively transferred cells.
As HRG modulates both angiogenesis and antitumor immunity (Figure 3), further study targeting HRG, or downstream targets, such as PIGF, is warranted. Indeed, Pigf−/− mice are not affected by HRG supplementation, indicating an important link between HRG and PIGF (46).
Concluding Remarks
Modulation of the tumor microenvironment as a part of cancer treatment has great potential and there are a multitude of targets that can affect tumor immunity, angiogenesis, and coagulation. Studies have demonstrated roles for HRG in hemostasis, angiogenesis, and immunity, which in turn can greatly affect tumor control and metastasis (Table IV). In vivo, HRG has significant effects on tumor growth and metastasis through modulation of TAMs towards an M1 phenotype, activation of platelets, and normalization of angiogenesis. Few studies have tested for HRG as a prognostic factor, and the results have not been definitive. However, tumor biopsies have shown that HRG deposits are present in the tumor stroma, indicating that the effects are likely dependent on their concentration in the tumor and type of tumor (24).
The importance of HRG in pre-clinical tumor models, through the use of genetically deficient mice, has been clearly demonstrated. However, the significance of HRG has yet to be validated in clinical settings. The role of HRG may be highly dependent on both the tumor type and factors, such as pH, within the tumor. This may explain why some groups have observed that HRG-deficiency promotes tumor growth (25, 45, 46) and others have seen the opposite effect (42, 43). Proteolytic cleavage of HRG may alter activity dependent on the tumor microenvironment (41-43, 50). Based on the elegant studies by Rolny et al. and Tugues et al., a key element of HRG in tumor control is its ability to modulate the tumor microenvironment through polarization of TAMs (45, 46). Optimal strategies to utilize a protein that can both create an immunostimulatory environment through polarization of macrophages towards an antitumor phenotype as well as modulate blood vessel normalization, should be explored.
- Received December 6, 2013.
- Revision received December 27, 2013.
- Accepted December 31, 2013.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved