5,6-Dimethylxanthenone-4-acetic acid, synthesised in this laboratory, reduces tumour blood flow, both in mice and in patients on Phase I trial. We used TUNEL (TdT-mediated dUTP nick end labelling) assays to investigate whether apoptosis induction was involved in its antivascular effect. 5,6-Dimethylxanthenone-4-acetic acid induced dose-dependent apoptosis in vitro in HECPP murine endothelial cells in the absence of up-regulation of mRNA for tumour necrosis factor. Selective apoptosis of endothelial cells was detected in vivo in sections of Colon 38 tumours in mice within 30 min of administration of 5,6-Dimethylxanthenone-4-acetic acid (25 mg kg −1 ). TUNEL staining intensified with time and after 3 h, necrosis of adjacent tumour tissue was observed. Apoptosis of central vessels in splenic white pulp was also detected in tumour-bearing mice but not in mice without tumours. Apoptosis was not observed in liver tissue. No apoptosis was observed with the inactive analogue 8-methylxanthenone-4-acetic acid. Positive TUNEL staining of tumour vascular endothelium was evident in one patient in a Phase I clinical trial, from a breast tumour biopsy taken 3 and 24 h after infusion of 5,6-Dimethylxanthenone-4-acetic acid (3.1 mg m −2 ). Tumour necrosis and the production of tumour tumour necrosis factor were not observed. No apoptotic staining was seen in tumour biopsies taken from two other patients (doses of 3.7 and 4.9 mg m −2 ). We conclude that 5,6-Dimethylxanthenone-4-acetic acid can induce vascular endothelial cell apoptosis in some murine and human tumours. The action is rapid and appears to be independent of tumour necrosis factor induction. British Journal of Cancer (2002) 86 , 1937–1942. doi: 10.1038/sj.bjc.6600368 www.bjcancer.com © 2002 Cancer Research UK

5,6-Dimethylxanthenone-4-acetic acid (DMXAA) is currently undergoing clinical evaluation as an antivascular agent for the treatment of cancer. We have previously demonstrated that DMXAA induces apoptosis of vascular endothelial cells in murine tumour sections and in a breast carcinoma biopsy from one patient in a Phase I trial. We wished to determine the tissue selectivity of this effect and its relationship to induced blood flow changes. Mice with Colon 38 tumours were treated with DMXAA and tissues were examined for apoptosis by TdT-mediated dUTP nick-end labelling (TUNEL). Hoechst 33342 was used to stain functional vessels, with the loss of stained vessels used as a measure of tumour vascular collapse. Treatment with DMXAA at 25 mg kg À1 , its maximum tolerated dose (MTD), showed, after 3 h, a 12-fold increase in TUNEL staining of tumour vascular endothelial cells. In contrast, tissue from the heart, brain, liver and spleen showed no increase. Induction of apoptosis in tumour tissue was both dose-dependent, observable at doses as low as 5 mg kg À1 , and time-dependent. Apoptosis was significantly lower in Colon 38 tumours of mice, with a targeted disruption in the TNF gene (TNF À/À ), or in the TNF receptor 1 gene (TNFR À/À ), as compared with that in wild-type mice. Increasing the DMXAA dose to 50 mg kg À1 in these knockout mice raised tumour apoptosis to a level comparable to that induced in wild-type mice given DMXAA at the MTD. For all the data, a significant correlation (r ¼ 0.94; Po0.001) was found between logarithmic percentage apoptosis induction and the logarithmic density of Hoechst-stained vessels. These results suggest that blood flow inhibition caused by DMXAA is tumour tissue-specific and is a consequence of induction of apoptosis in tumour vascular endothelial cells.

A series of nitrogen mustard analogues of the DNA minor groove binding fluorophore pibenzimol (Hoechst 33258) have been synthesized and evaluated for antitumor activity. Conventional construction of the bisbenzimidazole ring system from the piperazinyl terminus, via two consecutive Pinner-type reactions, gave low yields of products contaminated with the 2-methyl analogue which proved difficult to separate. An alternative synthesis was developed, involving construction of the bisbenzimidazole from the mustard terminus, via Cu(2+)-promoted oxidative coupling of the mustard aldehydes with 3,4-diaminobenzonitrile to form the monobenzimidazoles, followed by a Pinner-type reaction and condensation with 4-(1-methyl-4-piperazinyl)-o-phenylenediamine. This process gives higher yields and pure products. The mustard analogues showed high hypersensitivity factors (IC50AA8/IC50 UV4), typical of DNA alkylating agents. There was a large increase in cytotoxicity (85-fold) across the homologous series which cannot be explained entirely by changes in mustard reactivity and may be related to altering orientation of the mustard with respect to the DNA resulting in different patterns of alkylation. Pibenzimol itself (which has been evaluated clinically as an anticancer drug) was inactive against P388 in vivo using a single-dose protocol, but the short-chain mustard homologues were highly effective, eliciting a proportion of long-term survivors.