Angiostatin generating capacity and anti-tumour effects of D-penicillamine and plasminogen activators

Angiogenesis is an essential step in both tumour growth and the formation of distant metastases. Therefore, inhibition of tumour angiogenesis is currently being explored as a novel anti-tumour treatment [1‐5].

One class of anti-angiogenic compounds consists of proteolytic fragments of larger parent molecules. Examples of this type of angiogenesis inhibitors are angiostatin and endostatin, that are fragments of plasminogen and collagen XVIII, respectively [6, 7]

Angiostatin has been shown to possess anti-angiogenic effects both in vitro and in vivo. In vitro, angiostatin inhibited endothelial cell migration, proliferation and tube formation [8‐13]. In vivo, administration of angiostatin – either alone, or in combination with radiotherapy or chemotherapy – resulted in potent inhibition of tumour growth and metastasis formation in several animal tumour models [14, 14‐19].

Application of naturally occurring anti-angiogenic molecules in a clinical setting may be especially attractive, as these compounds have been shown to combine limited toxicity with the absence of development of drug resistance in several animal tumour models [20, 16]. The step from animal model to clinical application, however, has in the case of angiostatin been hampered by problems associated with producing sufficient amounts of biologically active recombinant protein. In addition, several different angiostatin moieties have been described, possibly with different clinical efficacies [9, 21, 22]. As angiostatin is also present in the circulation of healthy individuals [11]; Westphal, manuscript in preparation], boosting the angiostatin generating capacities of the host might provide an alternative for exogenous angiostatin administration.

In vitro, angiostatin is generated via a two-step process [23]. First, plasminogen activators (PAs) convert plasminogen into the active protease plasmin. Next, plasmin excises angiostatin from other plasmin molecules or from plasminogen, in a process requiring a donor of a free sulfhydryl (-SH) group (FSD). In several recent studies, administration of tissue type plasminogen activator (tPA) and/or a FSD has actually been shown to result in anti-angiogenic activity in vitro and additional anti-tumour activity in murine models. Merchan et al. reported that human plasma treated with tPA and captopril displayed anti-angiogenic activity in an in vitro tube formation assay. In vivo, tPA/captopril treatment inhibited angiogenesis in Matrigel plugs. In addition, anti-angiogenic activity was present in serum from human cancer patients treated with tPA and captopril. Remarkably, the anti-angiogenic activity present in the plasma samples was not completely attributable to angiostatin, as plasma, from which angiostatin was depleted, retained its anti-angiogenic properties [24]. Agarwal et al [25] showed that administration of the FSD N-acetyl-cysteine to MDA-MB-435 breast carcinoma xenograft bearing mice resulted in angiostatin accumulation in the tumour, the induction of apoptosis in endothelial cells, a reduction of vascular density in the core of the tumour and, as a consequence, a reduced tumour growth rate. Studies performed in our laboratory have demonstrated that administration of tPA in combination with captopril to mice resulted in an increased level of circulating angiostatin. In addition, this treatment resulted in anti-tumor activity in a human melanoma xenograft model [26]. Finally, Soff reported potent anti-tumour activity in human cancer patients after treatment with uPA and captopril [27].

However, several questions remain to be addressed. Most FSDs have other characteristics beside their capability to donate a SH group that may interfere with their application in the clinical setting. Moreover, these properties may influence, either positively or negatively, their angiostatin generating capacity. Captopril, for instance, is a potent ACE inhibitor, and administration of captopril lowers the blood pressure. Impaired blood flow towards and in the tumour may have contributed to the observed reduction in tumour growth. In addition, captopril inhibits matrix metalloproteinases [28, 29], molecules that are prominently involved in angiogenesis [30].

N-acetyl-cysteine (NAC), a glutathione precursor and FSD used in the study by Agarwal et al [25], not only affects angiostatin formation, but also possesses antioxidant activity, and inhibits MMPs [31]. Clearly, there is a need for preclinical studies to determine the net effects of the FSDs, and to select the FSD that is best suited for clinical studies.

In this paper we have compared the in vitro angiostatin generating capacities of a panel of FSDs. In our hands, D-penicillamine, a structural analog of cysteine, displayed a higher angiostatin generating capacity on a molar basis than either captopril, NAC, or other FSDs. Next, we monitored circulating angiostatin levels in mice treated with D-penicillamine and tPA, and, finally, we assessed the anti-tumour activity of the FSD/plasminogen activator treatment in an human melanoma xenograft model. Surprisingly, the superior in vitro angiostatin generation capacity of D-penicillamine was not reflected in either efficient in vivo angiostatin generation or superior anti-tumour activity. In both assays, D-penicillamine was less effective than captopril, indicating the need for further research into the most optimal free sulfhydryl group donor for clinical applications.

In a series of recent publications, the positive effect of a combination of a plasminogen activator (PA) and a free sulfhydryl group donor (FSD) on in vitro and in vivo angiostatin generation was described. In addition, treatment with PA plus FSD, or FSD alone, displayed anti-angiogenic and anti-tumour effects in murine xenograft tumour models [24‐26] and in a limited number of human patients [27]. This novel approach to anti-angiogenic therapy may utilize approved drugs such as captopril as FSD.

In this paper we set out to identify the most potent FSD with regard to angiostatin generation. In addition, testing different FSDs with different additional properties next to donating an SH group, may shed more light on the actual working mechanism of the anti-angiogenic effects induced by these drugs.

Our in vitro data demonstrated that, in the presence of plasminogen activators, all tested FSDs were able to induce angiostatin generation in vitro, in accordance with data reported by Gately et al [Gately et al., 1997]. In line with previous experiments with tPA and captopril? [26], we observed formation of angiostatin moieties with molecular weights varying between 30 and 50 kDa under reducing conditions. The 36 and 48 kDa moieties may represent the 3 and 4 kringle forms of angiostatin, respectively. A dose response curve was observed for the angiostatin formation induced by tPA and D-penicillamine. Remarkably, increasing uPA concentrations in the presence of FSDs did not result in increased angiostatin levels, but rather in an inhibition of the formation of this molecule. The reason for this phenomenon was unclear. Hypothesizing, it could be that (too) high levels of uPA, cause the activation of negative feedback mechanisms, resulting in the inactivation of plasmin, and as a consequence in a reduced generation of angiostatin. For this reason, we decided to perform our in vivo studies with tPA.

Although the tPA-induced conversion of plasminogen to plasmin in vivo was also increased by co-treatment with D-penicillamine, no increased angiostatin formation compared to treatment with tPA alone was observed. In contrast to the multiple molecular weight forms of angiostatin that we observed in our in vitro experiments, in vivo only the formation of a single angiostatin species of 36 kDa, comparable to angiostatin generated by tPA and captopril in previous studies, could be observed. Different angiostatin moieties might express different anti-angiogenic effects in vivo, although this is at present unclear. Thus far, only in vitro studies have compared the anti-angiogenic properties of plasminogen fragments containing kringles 1–3, 1–4, 1–5 and 5 [Cao et al., 2002;Lucas et al., 1998;Cao et al., 1996].

The fact that plasminogen, plasmin and angiostatin levels had returned to normal within two hours after the tPA injection, suggested a relatively short biological T_1/2 of both tPA and angiostatin. Whether angiostatin is subject to quick renal excretion, is degraded to shorter fragments, or disappears from the circulation due to other mechanisms, cannot be concluded from our experiments.

When treating mice bearing human melanoma xenografts, administration with either tPA or D-penicillamine alone resulted in an inhibition of tumour growth (44% and 23%, respectively). Treatment with both drugs resulted in a cumulative inhibition (58%). It should be noted, however, that the observed inhibition of tumour growth was not statistically significant, and therefore was far from convincing. In accordance with this finding, mean vascular densities in treated tumours where somewhat lower than in untreated tumours, but again not statistically significant. Matsubara et al (1989) [33] found that D-penicillamine only reduced endothelial cell proliferation in the presence of copper sulfate. Treatment of D-penicillamine alone did not alter the proliferation. The lack of administration of copper sulfate during our experiments might be the cause of the somewhat disappointing in vivo effects. However, Matsubara et al (1989) [33] also treated rabbits with D-penicilamine, and these experiments did result into inhibition of vessels growth in the cornea. But their administration route was intravenously, and not intraperitoneal as our route was. This could made the differences in the anti-angiogenic effects of D-penicillamine observed in both experiments.

In contrast to our D-penicillamine data, co-administration of captopril to mice treated with tPA in similar experiments, resulted in a clearly enhanced formation of angiostatin, and in a significant inhibition of tumour growth [26]. We therefore concluded that the superior capacity of D-penicillamine to induce angiostatin formation in vitro compared to captopril and other FSDs, was antagonized by environmental factors present in the living organism, resulting in a failure to significantly inhibit tumour growth.

Several explanations for this finding can be offered. Both captopril and D-penicillamine possess other characteristics that may influence angiogenesis and tumour growth.

First, as discussed previously, the drop in blood pressure induced by captopril-mediated ACE inhibition may account for some of its anti-tumour activity. D-penicillamine does not alter blood pressure.

Second, captopril may inhibit MMP function, thus preventing angiogenesis. D-penicillamine, on the other hand, does not affect blood pressure, but might share with captopril the ability to inhibit MMPs, by chelating metal ions. MMPs require zinc to express their activity. Therefore, chelation of these zinc molecules by D-penicillamine prevents the activation of MMPs. Moreover, it is known that D-penicillamine suppresses the inhibitory effects of hypochlorous acid (HOCl) on TIMP-1 (tissue inhibitor of metalloproteinase-1). As a consequence, TIMP-1 is not inhibited anymore, and could express its activity: limiting MMP activity [34].

Third, the fact that D-penicillamine binds with high affinity to albumin [35, 36] migt be of importance. By binding to albumin, the bioavailability, and therefore efficacy, of free circulating D-penicillamine is reduced. D-penicillamine sequestered to albumin may be less capable, or even unable, to donate its SH groups.

The fact that we could not detect angiostatin in tumour-bearing mice treated with tPA or tPA plus D-penicillamine suggested that circulating angiostatin is taken up by the tumour, for instance by binding of angiostatin to tumour blood vessels. As, however, immunohistochemical staining for angiostatin requires a specific antibody that does not stain the precursor (plasminogen) and this is not available, this could not be confirmed.

Another explanation for the absence of angiostatin could be that other anti-angiogenic factors might be present in our model. In our studies with tPA and captopril [26], and D-penicillamine (this paper), as well as in the study published by Agarwal with N-acetyl-cysteine (NAC), it is unclear whether angiostatin was in fact the effector molecule responsible for the observed anti-angiogenic and anti-tumour effects. Although Agarwal described accumulation of angiostatin in tumors derived from mice treated with NAC, no depletion studies were performed.

Agarwal et al. only administered the FSD N-acetyl-cysteine to achieve anti-angiogenic and anti-tumour activity, without the need for additional plasminogen activator administration. In vitro it was shown that tumour-derived VEGF induced expression of uPA and uPAR on HUVEC. In addition, uPA and uPAR expression was also demonstrated on tumour endothelium of NAC-treated mice. Apparently, this endothelium-derived uPA was sufficient to produce angiostatin. This notion is supported by our observations that both captopril and D-penicillamine are able to induce some (although not significant) inhibition of tumour growth. Additionally, captopril alone has been shown to possess anti-tumour activity in a human Karposi sarcoma patient [37], and in murine tumour models [38, 28, 29].

Furthermore, plasminogen activators expressed by the tumours used in these experiments, may have contributed to the observed anti-tumour effects after treatment with captopril or D-penicillamine alone. The human melanoma cell line BLM we used in our xenograft growth inhibition experiments expresses high levels of tPA. In both cases, however, administration of exogenous tPA improved anti-tumour activity, suggesting that the efficacy of the N-acetyl-cysteine treatment as reported by Agarwal may benefit from administration of exogenous plasminogen activators as well.

In conclusion, recent studies have provided evidence that upregulation of endogenous anti-angiogenic molecules as a novel anti-tumour therapy is feasible. Further studies are required to identify the most efficient plasminogen activators and FSDs, their optimal concentrations, and the appropriate application strategies. Regarding the dramatically different angiostatin generating capacities of D-penicillamine in vitro versus in vivo, our results indicate that selecting the most appropriate free sulfhydryl donor for anti-angiogenic therapy in a (pre)clinical setting should be performed by in vivo rather than by in vitro studies. Finally, we conclude that D-penicillamine is not suitable for this type of therapy.