PG545, a dual heparanase and angiogenesis inhibitor, induces potent anti-tumour and anti-metastatic efficacy in preclinical models

Background: PG545 is a heparan sulfate (HS) mimetic that inhibits tumour angiogenesis by sequestering angiogenic growth factors in the extracellular matrix (ECM), thus limiting subsequent binding to receptors. Importantly, PG545 also inhibits heparanase, the only endoglycosidase which cleaves HS chains in the ECM. The aim of the study was to assess PG545 in various solid tumour and metastasis models. Methods: The anti-angiogenic, anti-tumour and anti-metastatic properties of PG545 were assessed using in vivo angiogenesis, solid tumour and metastasis models. Pharmacokinetic (PK) data were also generated in tumour-bearing mice to gain an understanding of optimal dosing schedules and regimens. Results: PG545 was shown to inhibit angiogenesis in vivo and induce anti-tumour or anti-metastatic effects in murine models of breast, prostate, liver, lung, colon, head and neck cancers and melanoma. Enhanced anti-tumour activity was also noted when used in combination with sorafenib in a liver cancer model. PK data revealed that the half-life of PG545 was relatively long, with pharmacologically relevant concentrations of radiolabeled PG545 observed in liver tumours. Conclusion: PG545 is a new anti-angiogenic clinical candidate for cancer therapy. The anti-metastatic property of PG545, likely due to the inhibition of heparanase, may prove to be a critical attribute as the compound enters phase I clinical trials.

Tumour angiogenesis has become a prominent drug target for cancer therapy (Carmeliet, 2005) and agents designed to target the vascular endothelial growth factor (VEGF) family and the fibroblast growth factor (FGF) family, along with their corresponding receptor tyrosine kinases, have received considerable attention (Sparano et al, 2004;Folkman, 2007). Anti-angiogenic therapies directed against these targets have resulted in promising and wellvalidated therapeutics, with both biologicals and small-molecule drugs that target VEGF/VEGFR already approved for a number of cancers. Nonetheless, most cancer deaths are due to the development of metastases (Eccles and Welch, 2007) and angiogenesis inhibitors do not necessarily affect this process. Indeed, recent pre-clinical studies have raised some concerns about the impact of such agents and metastatic disease (Ebos et al, 2009;Paez-Ribes et al, 2009;Steeg et al, 2009).
Cleavage of HS by the endo-b-glucuronidase heparanase (HPSE) is strongly implicated in cell dissemination associated with tumour metastasis (Barash et al, 2010). Endo-b-glucuronidase heparanase is also involved in angiogenesis, where it releases sequestered HSbinding angiogenic growth factors (GFs) from the extracellular matrix (ECM) (Nakajima et al, 1988;Vlodavsky et al, 1996;Zetter, 1998;Goldshmidt et al, 2002). It also induces VEGF expression (Zetser et al, 2006). The upregulation of HPSE in many tumours has been linked to poor prognosis (Chen et al, 2008;Mikami et al, 2008;Rivera et al, 2008;Zheng et al, 2009). The use of heparin, heparin derivatives, sulfated oligosaccharides and suramin and its analogues to inhibit HPSE activity have shown promise as antitumour or anti-metastatic agents (Vlodavsky et al, 1994;Marchetti et al, 2003;Basche et al, 2006;Norrby, 2006). The clinical utility of a mixture of sulfated oligosaccharides (PI-88) with dual heparanase and angiogenic inhibitory activity has been previously demonstrated in a phase II clinical trial for hepatocellular carcinoma (HCC) . The limited utility of current angiogenesis inhibitors and the critical need to address the metastatic component of cancer together provide a strong rationale to develop new HPSE inhibitors typified by aglyconfunctionalised sulfated oligosaccharides (Karoli et al, 2005;. We report here the pharmacological profile of PG545, a synthetic, fully sulfated HS mimetic. We have previously demonstrated that PG545 is a potent heparanase inhibitor (K i ¼ 6 nM) and a potent angiogenesis inhibitor (p1 mM) using endothelial cell assays in vitro and the ex vivo the rat aortic ring assay. Moreover, PG545 was found to have mild anti-coagulant properties using the activated partial thromboplastin time test and the Heptest compared with PI-88 or earlier PG500 series compounds . In contrast to other HS mimetics which are mixtures, PG545 is a single molecular entity. PG545 induces potent antiangiogenic activity in vivo together with broad acting pre-clinical anti-tumour and anti-metastatic efficacy with a pharmacokinetic (PK) profile to support less frequent dosing compared with other HS mimetics. An open-label, single centre phase I study of the safety and tolerability of PG545 in patients with advanced solid tumours commenced in late 2010 (http://www.clinicaltrials.gov).

Compound synthesis and preparation
PG545 is a fully sulfated tetrasaccharide functionalised with a cholestanyl aglycon ( Figure 1) designed at Progen Pharmaceuticals Ltd (Brisbane, QLD, Australia). PG545 and radiolabeled PG545, the latter being synthesised from [1a,2a(n)-3 H]-cholesterol (GE Healthcare Australia Pty. Ltd., Rydalmere, NSW, Australia), were manufactured at Progen Pharmaceuticals Ltd and PharmaSynth (Brisbane, Australia). The tyrosine kinase inhibitor (TKI) sorafenib is an inhibitor of Raf-1 and other kinases involved in neovascularisation and tumour progression (Wilhelm et al, 2004) indicated for renal cell carcinoma and HCC. It was approved by US FDA in 2007 for patients with unresectable HCC and is the first approved systemic drug therapy for liver cancer. Sorafenib was sourced from Bayer Health Care (Leverkusen, Germany). PG545 was dissolved using cell culture medium for in vitro or ex vivo experiments and phosphate buffered saline (PBS), pH 7.2 for in vivo studies.

Animals
Athymic (nude), SCID, FvB, BALB/c and C57Bl6 mice were obtained from the Animal Resources Centre (Perth, Australia). All animal studies were approved by the University of Queensland or the University of Adelaide Animal Ethics Committees. During the studies, the care and use of animals was conducted in accordance with the principles outlined in the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, 7th Edition, 2004 (National Health and Medical Research Council). For the HT-29 study, the protocol was approved by the Institutional Animal Care and Use Committee and the study was performed at the AAALAC certified animal facility in Pharma Legacy (Shanghai, China).

In vivo angiogenesis
Angiogenesis in vivo was evaluated using the Angiosponge model in which subcutaneously implanted, agarose-captured FGF2 gelfoam sterile sponges (Pfizer, West Ryde, Australia) induces vascularisation over a 10-day period responsive to angiogenesis inhibition as determined by the number of vessels with identifiable lumen that stained positive for CD31/PECAM-1, an antigen present specifically on endothelial cells (McCarty et al, 2002). PG545 was administered s.c. on the flank distal to the AngioSponge implant. This method was also used to examine five medium-sized primary tumours from each group of Lewis Lung Carcinoma (LL/2)-bearing mice.

Tumour models and drug administration
In the orthotopic Hep3b2.1-7 human liver cancer model, treatment of mice (n ¼ 10) began 14 days after inoculation of 2.5 Â 10 6 cells. Mice received PG545 (or [ 3 H]-PG545 for the efficacy/PK study) using either a once-or twice-weekly dosing schedule. To initiate tumour xenografts, MDA-MB-231, PC3, Cal27 and HepG2 cells were injected s.c. into nude mice (n ¼ 8) at 4, 5, 2 and 2 Â 10 7 cells ml À1 , respectively. PG545 was injected s.c. in a volume of 100-200 ml PBS once the average tumour volume was B100 -200 mm 3 . In the B16F0 lung metastasis model, 2 Â 10 5 cells were injected via the tail vein of C57Bl6 mice (n ¼ 8). In the syngeneic LL/2 model, single dose or weekly treatments with PG545 at 20 or 40 mg kg À1 and sorafenib daily at 60 mg kg À1 commenced on the day of inoculation with 2 Â 10 6 cells. Surface lung macrometastases were counted for each group (n ¼ 9) upon study termination. In the intrasplenic HT-29 xenograft model, 1 Â 10 6 cells in 100 ml were injected into the spleen. The next day, PG545 (20 mg kg À1 ) was administered to mice (n ¼ 12) each week (qw) over 8 weeks and surface liver and colon metastatic nodules were counted for each group (n ¼ 12) upon study termination. Solid tumours were measured in two dimensions (length and width) and the tumour volume (cm 3 ) was calculated using the equation: V ¼ length Â width 2 Â p/6. The anti-tumour activity was expressed as percent tumour growth inhibition (%TGI). Further details are available in Supplementary Methods.    Figure 1 The structure of PG545.

Statistics
For multiple comparisons, a one-way analysis of variance (ANOVA) followed by a post hoc Holm-Sidak test or a Dunnett's test was used, provided the data passed a normality test or the equal variance test. For comparison between a control and a treatment group, an unpaired t-test was used. A P-value o0.05 was considered significant.

PG545 inhibits angiogenesis in vivo
In the AngioSponge model, PG545 was administered either daily or twice-weekly. Significant inhibition of CD31 staining was observed ( Figure 2A). As the duration of the AngioSponge model is 10 days in total, once-weekly treatment was not examined in this model. However, the effect of once-weekly dosing with PG545 on tumour angiogenesis was determined in tumour-bearing mice (LL/2 model). Assessment of CD31-positively stained vessels with identifiable lumen in the primary (subcutaneous) tumours of mice confirmed that once-weekly treatment with PG545 significantly inhibited angiogenesis ( Figure 2B).
PK profile of radiolabeled PG545 indicates a long half-life in blood with evidence that efficacious doses penetrate solid tumours PG545 significantly reduced solid tumour growth in the orthotopic liver cancer model Hep3b2.1-7 under two different regimens: a twice-weekly s.c. administration schedule at 20 mg kg À1 ( Figure 3A) or using 20 mg kg À1 as a loading dose followed by a twice-weekly 10 mg kg À1 maintenance dose ( Figure 3B). For compounds with long half-lives, a loading dose is often used to quickly increase the plasma concentration of a drug to the desired steady state levels (Paw, 2006). These dosing schedules resulted in body weight loss of 9% in the 20 mg kg À1 twice-weekly group and 4% in the 20 mg kg À1 followed by twice-weekly 10 mg kg À1 group (data not shown). The data generated in separate experiments using 20 mg kg À1 twice-weekly or as a loading dose (followed by 10 mg kg À1 as a maintenance dose) exhibited a similar efficacy profile (Tumour Growth inhibition (TGI) values of 52% versus 55%). In combination, twice-weekly dosing with PG545 and sorafenib (30 mg kg À1 , qd) led to a statistically significant reduction in tumour weight compared with control (TGI ¼ 85%).
Once-weekly dosing with PG545 at 20 mg kg À1 followed by 10 mg kg À1 increased the TGI to 65% compared to the twiceweekly regimen but the enhanced anti-tumour effects observed in combination with sorafenib were not improved with once-weekly dosing (data not shown).
The PK profile of PG545 in tumour-bearing animals was determined using 20 mg kg À1 [ 3 H]-PG545 administered s.c. every 96 h for a total of three doses. The concentration of PG545 was assessed in both blood and tumour samples. The PK profile in blood indicated that the mean elimination half-life (t1 2 ) of [ 3 H]-PG545 was 450 h for mice administered up to three s.c. bolus doses at 96 h intervals (Table 1). Visual inspection of the mean blood [ 3 H]-PG545 concentrations vs time curve data showed that the mean blood C max values were in the range 25.9 -29.0 mg ml À1 and the corresponding mean T max values were in the range 2 -6 h ( Figure 3C).
Visual inspection of the mean tumour [ 3 H]-PG545 concentration vs time curve data showed that the tumour C max values were in the range 36.6 -45.1 mg g À1 and that the corresponding mean T max values in the tumour tissue were 12 h (one dose), 48 h (two doses) and X96 h (three doses) ( Figure 3D). Using a one-compartment PK model, the estimated mean rate constant for uptake of [ 3 H]-PG545 from the bloodstream into the tumour (t1 2 b ), derived from the mean blood and tumour concentration vs time data, was in the range 12.8 -16.5 h. The corresponding estimated mean t1 2 for return of [ 3 H]-PG545 from the tumour into the bloodstream (t1 2 t ), derived from the mean blood and tumour concentration vs time data, was in the range 15.7 -16.3 h (Supplementary data, Supplementary Table S1). The C max values in liver and kidney reached 106 and 39 mg g À1 , respectively, following a single dose (data not shown). This equates to a tumour:liver ratio of 0.4 and a tumour:kidney ratio of 1.3, 12 h after a single dose, suggesting good distribution into tumour tissue compared with other organs. As a comparison, gefitinib (Iressa, Astra Zeneca, London, UK) has a tumour:liver ratio of 0.15 or 0.18 and a tumour:kidney ratio of 0.35 or 0.56, 2 h or 8 h, respectively, after a single dose in LoVo tumour xenografts (McKillop et al, 2005).

PG545 inhibits tumour progression in a variety of xenograft models
Further investigation of once-or twice-weekly s.c. dosing schedules was performed in a variety of xenograft or syngeneic cancer models (Figure 4). Tumour growth inhibition in the MDA-MB-231 model ( Figure 4A) was maximal at 20 mg kg À1 onceweekly (77%) or 20 mg kg À1 twice-weekly (71%) with 30 mg kg À1 groups inhibiting tumour growth by 69% (30 mg kg À1 , onceweekly) or 59% (30 mg kg À1 , twice-weekly). Body weight loss at the end of the study was 4% for 20 mg kg À1 once-weekly, 6% for 20 mg kg À1 twice-weekly, 11% for 30 mg kg À1 once-weekly and 10% for 30 mg kg À1 twice-weekly (data not shown). Employing the 20 mg kg À1 dose only in subsequent models, the differences noted between once-and twice-weekly schedules were not significant in the PC3 or HepG2 models. For example, in the PC3 model, onceweekly dosing led to a TGI of 74% while twice-weekly dosing resulted in a TGI of 83% ( Figure 4B). Although body weight loss at 20 mg kg À1 once-weekly was 6%, the twice-weekly regimen was 16% (data not shown). However, in the HepG2 model ( Figure 4C), the once-weekly (TGI ¼ 55%), but not twice-weekly (TGI ¼ 45%) schedule, significantly inhibited tumour growth to a similar extent as 20 mg kg À1 once-weekly treatment in the Cal27 model which produced a TGI of 57% ( Figure 4D). In both models, 20 mg kg À1 once-weekly (and twice-weekly in the HepG2 model) led to body weight loss between 2% and 4% (data not shown). PG545 also significantly reduced primary tumour growth in the HT-29 model (Supplementary Figure S1a). The concentration of PG545 that is lethal to 50% of cells (LC 50 ) using HUVEC (78 mM), B16 (4200 mM) or HepG2 cells (4100 mM), indicated that PG545 is not cytotoxic in vitro (data not shown).

PG545 induces potent anti-metastatic effects in experimental and spontaneous models of metastasis
Twice-weekly injections of PG545 significantly inhibited the formation of metastases in the B16F0 experimental lung metastasis model, when administered either before or on the day of tumour cell inoculation ( Figure 5A). When administered as a once-weekly schedule, PG545 also significantly inhibited the occurrence of spontaneously arising liver metastases ( Figure 5B; Supplementary Figure S1b) and colon metastases ( Figure 5C) in the HT-29 colon model. In terms of body weight loss, the nadir was 8% in the B16 model and 12% in the HT-29 model (data not shown).
The Lewis Lung Carcinoma model (LL/2) provided another model to investigate the effect of PG545 on spontaneous metastasis. PG545 was administered at weekly intervals or in some cases, as a single injection from the day of tumour inoculation. The body weight nadir during the study constituted a loss of 5% (data not shown). In addition to significant inhibition of primary tumour growth ( Figure 6A), PG545 significantly reduced the number of lung metastases, with complete abrogation  in the single dose 40 mg kg À1 group and 96% inhibition in the 20 mg kg À1 once-weekly group ( Figure 6B). However, it is not clear why 40 mg kg À1 once-weekly failed to reach significance but interestingly sorafenib demonstrated no effect on metastasis in this model. Plasma concentration vs time curves indicated that efficacious exposure levels should be between 20 and 70 mg ml À1 ( Figure 6C) and PK analysis revealed the apparent t1 2 to range between 28 and 39 h with a C max , T max and to a lesser extent, an AUC, comparable to that observed in the liver cancer model using the radiolabeled compound ( Table 2). The difference between the half-life estimation of [ 3 H]-PG545 and PG545 by LC/MS/MS may be explained by experimental variations/matrices and the possibility that the radiolabel was no longer attached to the parent molecule, thus estimating the t1 2 for removal of a metabolite to which the label is attached.

DISCUSSION
Heparanase-mediated degradation of HS is involved in tumour angiogenesis and metastasis, making it a major target for cancer research (Hammond et al, 2006;McKenzie, 2007;Barash et al, 2010). Clinical studies have typically noted a poorer prognosis in patients with high expression of heparanase, presumably due to its association with an aggressive, metastatic phenotype (Chen et al, 2008;Mikami et al, 2008;Rivera et al, 2008;Zheng et al, 2009). The pivotal role of VEGF and its receptors in promoting angiogenesis is also well established, and both antibody and small-molecule inhibitors of VEGF signaling have shown clinical utility. Despite no specific FGF inhibitors being clinically available, their relevance as a cancer therapy has become increasingly clear due to the increased expression of FGF family members in late-stage tumours following phenotypic resistance to VEGFR2 blockade (Casanovas et al, 2005). We previously identified a series of HS mimetics, of which PG545 was one, as potent inhibitors of heparanase, GFs including VEGF and FGFs and ex vivo angiogenesis . Herein, we extend those findings to the pre-clinical arena to demonstrate the in vivo potency of PG545 in angiogenesis, solid tumour and metastasis models and provide initial data on the pharmacokinetics associated with the compound.
The AngioSponge model highlighted that daily and twice-weekly dosing with PG545 induced a potent anti-angiogenic effect in vivo, as measured by CD31 expression. In the LL/2 model, decreases in CD31 expression were also statistically significant using onceweekly dosing. Unfortunately, as only five representative mediumsized tumours from each group were available, any dosedependent correlation between reduction in CD31 staining and inhibition of solid tumour growth is not apparent. Overall, the data confirmed that a once-or twice-weekly dosing schedule is anti-angiogenic in vivo and this schedule is less frequent compared with those of other heparin-derived products or HS mimetics (Gandhi and Mancera, 2010). The body weight loss in mice may be species-specific because PG545 is not cytotoxic in vitro or in vivo (vacuolation was most common pathology finding) and did not lead to significant body weight loss in other species during toxicity studies (data not shown). The absence of bruising in PG545-treated nude mice illustrates the mild anti-coagulant properties of PG545 . Although high doses of PG545 increased activated partial thromboplastin time (which was reversible) in toxicology studies, it is not clear whether this is due to a depletion of clotting factors or a mild anti-coagulant effect (data not shown).
PK data were generated using radiolabeled PG545 administered to satellite groups of tumour-bearing mice (Hep3b2.1-7). A relatively long half-life (t1 2 ) in blood (450 h) was observed. The concentrations of [ 3 H]-PG545 measured within the liver tumours of mice confirmed entry into tumour tissue. A one-compartment PK model found each of the estimated parameter values was internally consistent following administration of one, two or three doses (Supplementary data; Supplementary Table S1). However, visual inspection of the graph for tumour-bearing mice administered three doses of [ 3 H]-PG545, with each dose separated by a 96-h interval, showed that the tumour concentration vs time curve for the third dose is different from that for the preceding two doses. The reason for this is unclear though it is possible that PG545 could have induced biological change within the tumour Vehicle control PG545 20 mg kg -1 1× wk, TGI = 71% PG545 20 mg kg -1 2× wk, TGI = 77% PG545 30 mg kg -1 1× wk, TGI = 69% PG545 30 mg kg -1 2× wk, TGI = 59% PG545 20 mg kg -1 2× wk, TGI = 83% PG545 20 mg kg -1 1wk, TGI = 57% PG545 20 mg kg -1 1× wk, TGI = 74%

Vehicle control
Vehicle control PG545 20 mg kg -1 1× wk, TGI = 55% PG545 20 mg kg -1 2× wk, TGI = 45% vasculature -such an assumption would not necessarily be applicable to the one-compartment PK model. Tumour growth in the Hep3b2.1-7 model was significantly reduced following the once-and twice-weekly dosing schedule for PG545 or using a loading and maintenance dose approach, to an extent similar to that seen following treatment with sorafenib, a TKI approved for HCC. The combination of PG545 and sorafenib produced a TGI up to 85%, thus illustrating some potential utility for combination modalities using PG545. Taken together with the PK profile, the efficacy data indicated that once-weekly dosing may provide an optimal dosing frequency for PG545. This was further confirmed using a variety of subcutaneously implanted xenograft   or syngeneic models of prostate, breast, liver and head and neck cancer in the mouse. We next tested the anti-metastatic activity of PG545 in three murine models of metastasis. In the B16 melanoma model, shortterm therapy with PG545 was initiated either before or immediately after i.v. tumour inoculation of B16 cells and confirmed that both schedules significantly inhibited the development of lung metastases. In the spontaneously metastasising HT-29 colon model, complete abrogation of liver metastases was achieved using a weekly regimen of 20 mg kg À1 PG545 over 8 weeks (a dose in week 6 was omitted due to body weight concerns). In the spontaneously metastasising Lewis Lung Carcinoma model, PG545 significantly reduced the number of metastases and PK data estimated the efficacious plasma concentration range starts at B20 mg ml À1 . This potent anti-metastatic effect of PG545 may differentiate it from other angiogenesis inhibitors, which have been associated with an acceleration of metastasis under certain circumstances (Ebos et al, 2009;Paez-Ribes et al, 2009). It is hypothesised that a metastatic 'conditioning' effect may occur via multiple mechanisms following treatment with angiogenesis inhibitors (Ebos et al, 2009;Paez-Ribes et al, 2009) which posed the question -are there any alternative anti-angiogenic agents that might not evoke this malignant behaviour? (Loges et al, 2009). Thus, further elucidation of the contrasting mechanisms of action and possible differences in efficacy between HS mimetics and TKI/ biologics is warranted.
In conclusion, we have clearly demonstrated the anti-tumour and anti-metastatic properties of PG545, which to our knowledge is the first HS mimetic that is a single molecular entity to emerge as a clinical candidate. In addition to its anti-angiogenic properties, it is the anti-heparanase/anti-metastatic activity that may differentiate PG545 from other anti-angiogenic agents. Particularly in light of recent preclinical findings relating primarily to anti-VEGF therapies, this property may become a critical attribute as PG545 progresses towards the clinic. Table 2 Model independent pharmacokinetic analysis of PG545 in the plasma of tumour-bearing mice (Lewis Lung Carcinoma) Week 1 Week 2 Parameter 20 mg kg À1 40 mg kg À1 60 mg kg À1 20 mg kg À1 40 mg kg À1 60 mg kg À1 C max (mg ml À1 ) 2 3 5 5 7 9 2 7 7 1 9 4 T max (h)