Carboxymethyl benzylamide dextran inhibits angiogenesis and growth of VEGF-overexpressing human epidermoid carcinoma xenograft in nude mice

Vascular endothelial growth factor (VEGF) expression is elevated in a wide variety of solid tumours. Inhibition of VEGF activities is able to reduce angiogenesis and tumour growth. We have recently shown in vitro that carboxymethyl dextran benzylamide (CMDB7) prevents the binding of VEGF165 to its cell surface receptors and thus inhibits VEGF activities on endothelial cells. In the present study, we explored the effects of CMDB7 on highly aggressive human epidermoid carcinoma A431 cells known to overexpress epidermal growth factor receptors (EGFRs) and produce a high amount of VEGF and a minor quantity of bFGF. In vitro, CMDB7 blocked the mitogenic activity of A431-conditioned medium on endothelial cells. Concerning A431 cells, CMDB7 inhibited their proliferation and the VEGF165 binding to them. In vivo, administration of CMDB7 (10 mg kg−1) three times per week for 2 weeks inhibited the growth of A431 xenografts in nude mice by 73% as compared to the control group. Immunostaining of endothelial cells with mouse-specific GSL-1 lectin in tumour sections revealed that CMDB7 also inhibited the density of intratumour endothelial cells by 66%. These findings demonstrate that CMDB7 has an efficient antiangiogenic and antitumour action in vivo even when tumour cells produce a high level of VEGF and EGFRs.

Neovascularisation is critical for supporting the rapid growth of solid tumours (Folkman, 1990). Tumour angiogenesis appears to be achieved by the overexpression of angiogenic agents within solid tumours that stimulate host vascular endothelial cell mitogenesis and possibly chemotaxis. It is well established now that the induction of vascular endothelial growth factor (VEGF) expression, including via tumour hypoxia (Hlatky et al, 1994), plays a major role in tumour angiogenesis (Dvorak et al, 1991;Millauer et al, 1994;Goldman et al, 1998). In recent years, it has been widely shown that VEGF activity is a key feature during tumour growth and angiogenesis, and that blocking of this signal transduction pathway may inhibit tumour progression (Cheng et al, 1996;Relf et al, 1997). In vivo, VEGFs act as potent mitogenic factors for endothelial cells and as blood vessel permeabilising agents (Senger et al, 1983;Plouët et al, 1989;Klagsbrun and Soker, 1993;Yuan et al, 1996). The VEGF gene family currently includes six members: VEGF-A (prototype VEGF), placenta growth factor (PlGF), VEGF-B, VEGF-C, VEGF-D and VEGF-E, (produced by Orf virus), (reviewed by Veikkola et al, 2000). VEGF is a homodimeric glycoprotein that exists in six isoforms containing 121,145,162,165,189 and 206 amino-acid residues as a result of alternative splicing from a single gene (Ferrara and Davis-Smith, 1997;Lange et al, 2003). The predominant and the best characterised VEGF species is the heparin-binding 165-amino acid-long form VEGF 165 (reviewed by Neufeld et al, 1999).
Carboxymethyl dextran benzylamide (CMDB7) is a noncytotoxic substituted dextran. We have recently shown in vitro that it prevents the binding of VEGF 165 to human umbilical vein endothelial cell surface and thus inhibits VEGF 165 -induced phosphorylation of VEGFR-2 and consequently endothelial cell proliferation (Hamma-Kourbali et al, 2001). In the present study, we explored in vitro and in vivo the effects of CMDB7 on human epidermoid carcinoma A431 cells known to produce a high amount of VEGF and a minor quantity of bFGF (Myoken et al, 1991). The other peculiarity of A431 cells is the production of a newly identified splice form of VEGF, VEGF-162, which binds more efficiently than VEGF-165 to a natural basement membrane of endothelial cells (Lange et al, 2003). Moreover, A431 cells express a high level of epidermal growth factor receptors (EGFRs) activated by EGF, a nonheparin-binding growth factor, which does not interact with CMDB7 (Bagheri-Yarmand et al, 1998b). Interestingly, the resistance of A431 tumours to treatment with EGF receptor-blocking antibodies is associated with an elevated expression of VEGF . Such cells, xenografted in nude mice, provide a model of highly VEGFdependent (Melnyk et al, 1996) and aggressive tumour growth in an in vivo system. Since A431 cells have been recently described to express the VEGF 165 -binding sites (Li et al, 2001), we explored also the possible effect of CMDB7 on radiolabelled VEGF binding. We demonstrate that CMDB7 acts on both tumour and endothelial cells, decreasing in a potent manner the tumour growth and angiogenesis in vivo.

Cell lines and cell culture
Human epidermoid carcinoma cell line (A431) and human umbilical vein endothelial cell line (HUV-EC-C) were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). A431 cells were routinely grown in DMEM (Life Technologies, Inc., Gaithersburg, MD, USA) and HUV-EC-Cs in M199 (Life Technologies, Inc.) and were cultured at 371C in a 5% CO 2humidified atmosphere. Both culture media were supplemented with 10% foetal calf serum (FCS), 2 mM L-glutamine, 1 mM sodium pyruvate, 50 U ml À1 penicillin and 50 mg ml À1 streptomycin (all obtained from Life Technologies, Inc.). The cells were free of mycoplasma, bacteria and viruses.

Preparation of conditioned media (CMs)
To assess the production of VEGF-A, the A431 cells at three different density were seeded into a 24-well culture plate (Falcon, Strasbourg, France) in DMEM supplemented with 10% FCS for 24 h. To obtain CM containing only the growth factors secreted by A431, the cells were washed twice with PBS, and incubated in 1 ml of serum-free DMEM containing 0.1% BSA (Sigma, St Louis, MO, USA). At the indicated time, the media were collected, cleared by centrifugation, and stored at À801C before use. For other experiments, the cells were grown in 150 mm-diameter Petri dishes (Falcon) to 80% confluence in DMEM/10% FCS, washed and incubated in 10 ml dish À1 of serum-free medium.

Determination of VEGF-A concentration in the A431-CMs by radioimmunoassay
The surface of flat-bottomed polystyrene wells (Disposable Immulon 1 Remowawell, Dynatech, Cambridge, MA, USA) were coated overnight at 41C with 200 ml of PBS buffer containing 50 ng polyclonal neutralising anti-VEGF IgG (R&D Systems, Abingdon, UK). The nonspecific interactions were saturated with PBS containing 0.1% BSA and 0.01% Tween-20 (PBT buffer) for an additional overnight at 41C. After blocking, the wells were washed three times with 300 ml of PBT buffer. Then, A431-CM or VEGF 165 (R&D Systems) as standard at increasing concentrations (0 -250 ng ml À1 ) and 50 pM 125 I-VEGF 165 (Amersham Pharmacia Biotech, Orsay, France) were added to a final volume of 200 ml in PBT buffer. After an overnight incubation at 41C, wells were washed three times with 300 ml of PBT buffer and the radioactivity remaining in each well was measured in a g-counter (LKB 1261 Multigamma).

A431-CM effects on HUV-EC-C proliferation
HUV-EC-Cs were seeded at a density of 2 Â 10 4 well À1 into 24-well tissue culture plates (Falcon) in M199-10% FCS. After 24 h, the cells were growth arrested by serum starvation for another 24 h. Then, the cells were incubated for 48 h with A431-CM diluted in a serum-free medium to a final VEGF concentration of 10 ng ml À1 (concentration at which VEGF 165 has the maximal mitogenic effect on HUV-EC-Cs (Hamma-Kourbali et al, 2001) in the presence or absence of 5 mM CMDB7 (optimal concentration at which CMDB7 completely prevents the VEGF 165 mitogenic effect on HUV-EC-Cs (Hamma-Kourbali et al, 2001) or 1 mg ml À1 anti-human VEGF neutralising antibody (Sigma) characterised by neutralisation dose 50 ¼ 0.01 -0.1 mg ml À1 . Cells were washed with PBS, dissociated with 0.025% trypsin-EDTA (Life Technologies) and counted using a Coulter counter (Coultronics, Margency, France). All experiments were performed in triplicate and data illustrate the mean cell numbers7s.e. provided from one representative of three independent experiments.
A431 proliferation assay A431 cells were seeded at a density of 10 4 cells well À1 into 24-well tissue culture plates (Falcon) in DMEM -10% FCS and allowed to adhere to the plastic for 24 h. After washing with DMEM, the cells were incubated with CMDB7 at the indicated concentrations (day zero) in DMEM -1% FCS. At different times, cells were washed with PBS, dissociated with 0.025% trypsin-EDTA (Life Technologies, Inc.) and counted using a Coulter counter (Coultronics). In each case, samples were performed in triplicate, and data illustrate mean cell numbers7s.e. of one representative of three independently performed experiments.

VEGF 165 binding to A431 cells
For displacement binding assays, A431 cells were grown until confluence on 24-well tissue culture plates (Falcon). After an overnight incubation in serum-free medium and two washings with ice-cold binding buffer (PBS/0.1% BSA), the cells were incubated with 7 pM 125 I-VEGF 165 (Amersham, Pharmacia Biotech) and CMDB7 at various concentrations from 8 Â 10 À9 to 4 Â 10 5 M at 41C for 2 h. Incubation was terminated by gently aspirating the medium and washing the cell monolayer three times with ice-cold binding buffer. After cell solubilisation in 0.3 ml of 0.5 N NaOH, the bound radioactivity was measured in a g-counter (LKB 1261 Multigamma). Nonspecific binding was determined in the presence of an excess (5000 pM) of unlabelled VEGF 165 (R&D Systems).

Tumour cell inoculation in nude mice
All in vivo experiments were carried out with ethical committee approval and met the standards required by the UKCCCR guidelines (Workman et al, 1998). A431 cells (5 Â 10 6 ) were inoculated s.c. in the right flank of 4-week-old athymic nude mice (nu/nu) (Charles River Laboratory, Aubin-les-Elbeuf, France). The animals (n ¼ 20) were kept in a temperature-controlled room on a 12 h : 12 h light -dark schedule with food and water ad libitum. All mice developed single s.c. palpable tumours of approximately 50 mm 3 6 days after cell inoculation. Then, mice were placed in control (n ¼ 10) and CMDB7-treated groups (n ¼ 10). Mice were treated by subcutaneous (s.c.) injection of 0.1 ml PBS alone (control) or containing 10 mg kg À1 CMDB7 close to the tumour, three times a week for 2 weeks. Tumours were measured along two major axes with a calliper. Tumour volume was calculated as follows: where R 1 is radius 1, R 2 is radius 2, and R 1 oR 2 .

Tissue preparation and immunohistochemical staining
Immediately after surgical resection, the tumour specimens were fixed with 4% paraformaldehyde and processed to paraffin inclusion. The intratumour mice endothelial cells were specifically stained with GSL-1 lectin (Vector Laboratories, Burlingame, CA, USA) in 5-mm sections as previously described (Bagheri-Yarmand et al, 1999). The GSL-1 lectin binds specifically to galactosyl residues present on vascular endothelium in mice (Alroy et al, 1987;Mattsson et al, 2002). The proliferative index of tumour xenograft was determined by human Ki-67 staining with monoclonal mouse antibody (MIB-1; 1 : 50; Dako, Trappes, France). The epitope retrieval was performed in 10 mM citrate buffer pH ¼ 6.0 for 40 min at 981C. Specific reactions were visualised with 3,3 0diaminobenzidine (DAB) as chromogen.

Image analysis
For each GSL-1-or MIB-1-labelled section of control or CMDB7treated tumour, five fields containing exclusively viable tumour cells, as indicated by the haematoxylin staining, were selected randomly for analysis. Image analysis was performed using the NIH programme (developed at NIH and available on the Internet at http://rsb.info.nih.gov/nih-image/). The endothelial cell density in each field was expressed as the ratio of endothelial cell area and the total viewed area Â 100 (%). To determine the proliferative index, we estimated the percentage of tumour cell nuclei positive for Ki-67 marker. These values were then averaged for untreated (control) and treated-CMDB7 tumours.

Statistical analysis
Multiple statistical comparisons were performed using ANOVA in a multivariate linear model. Statistical comparisons were conducted using the Mann -Whitney t-test. Po0.05 was considered statistically significant.

CMDB7 inhibits, like neutralising anti-VEGF 165 antibody, mitogenic effect of A431-CM on HUV-EC-Cs
According to previous studies (Melnyk et al, 1996), we found that A431 cells secrete in the culture medium large amounts of VEGF-A. Moreover, we showed here that VEGF production is cell number-and time-dependent (Table 1). As expected, A431-CM stimulated the in vitro proliferation of HUV-EC-Cs by 2.5-fold after 48 h of incubation (Figure 1). This mitogenic effect is, at least in part, VEGF-specific since the neutralising antibodies against recombinant VEGF inhibited the A431-CM-induced proliferation of HUV-EC-Cs by 45% after 48 h treatment. A431-CM, used in this experiment, contained 10 ng ml À1 of VEGF 165 as revealed by specific radioimmunoassay. At the same concentration, recombinant VEGF 165 has a similar mitogenic effect on HUV-EC-Cs (Hamma-Kourbali et al, 2001), as described above the addition of 5 mM CMDB7 prevented the stimulatory effect of A431-CM on HUV-EC proliferation (Figure 1). When HUV-EC-Cs were cultivated in serum-free medium, CMDB7 or neutralising anti-VEGF 165 antibodies had no effect.

CMDB7 inhibits A431 cell proliferation in vitro
Next, we tested CMDB7 for its ability to affect the in vitro growth of A431 tumour cells. We demonstrated that treatment with CMDB7 at increasing concentrations, ranging from 0.1 to 20 mM, resulted in a concentration-and time-dependent inhibition of A431 cell number (Figure 2).
In contrast, 1 mg ml À1 anti-VEGF antibody had no effect on A431 proliferation in vitro (data not shown) as reported by others (Melnyk et al, 1996).

CMDB7 inhibits VEGF 165 binding to A431 tumour cells
Since A431 cells produce VEGF-A and binds VEGF 165 on the surface (Li et al, 2001), we explored if CMDB7 is able to compete for VEGF 165 -specific binding (Figure 3). CMDB7 decreased the 125 I-VEGF 165 -specific binding to A431 cells at concentrations ranging from 0.1 to 50 mM with a half-maximum inhibitory effect (IC 50 ) at concentration 2 mM. A431 cells were seeded at the indicated density into 24-well plate in DMEM -10% FCS medium for 24 h. After washing, they were incubated in serum-free medium. At the indicated time, the media were collected. VEGF-A concentration was asessed by radio immunoassay as described in Materials and Methods. Values are in ng ml À1 . The limit of sensitivity of the assay was 0.4 ng ml À1 .

CMDB7 inhibits the growth of A431 cells xenografted in nude mice
The tumours appeared in 100% of mice 6 days after A431 cell inoculation. CMDB7 inhibited the growth of A431 tumours by 73% (Po0.001) after 2 weeks of treatment ( Figure 4).
No apparent toxicity was noticed during treatment with CMDB7. No signs of toxicity such as diarrhoea, infection, weakness or lethargy were observed. The body weight of the inoculated mice was not affected by CMDB7 after 2 weeks of treatment. All treated mice were alive at the end of treatment.

CMDB7 decreases the proliferative index of A431 xenografts
The specific Ki-67 staining was less intense in CMDB7-treated tumours as compared to control (nontreated) ones. The proliferative index for treated and control xenografts were significantly (P ¼ 0.05) diffferent, 2678 and 34710%, respectively (mean7 s.e.m). These data suggest that CMDB7 inhibited directly in vivo the proliferation of tumour cells. In all xenografts, treated as well as nontreated, the areas of necrosis/apoptosis were large, but localised in the centre of tumour. There did not appear to be obvious differences in the degree of necrosis observed in both cases. We had no difficulties in obtaining five fields of viable cells in all tumours.

CMDB7 inhibits the intratumour endothelial cell density
Selective GSL-1 staining showed that CMDB7 treatment reduced the endothelial cell quantity in tumour tissue ( Figure 5B) as compared to control ( Figure 5A). The mean percentage of endothelial cell area (endothelial cell density) in viable fields of CMDB7-treated tumours (2.9 7 0.6; 50 fields in 10 tumours) was inhibited by 66% (Po0.001) as compared to control tumour value (8.670.7; 50 fields in 10 tumours) ( Figure 5C).

DISCUSSION
Antiangiogenesis is a promising therapeutic approach for the treatment of cancer (Folkman, 1995;Schweigerer, 1995). VEGF plays a crucial role in tumour angiogenesis and the inhibition of VEGF action decreases tumour growth in vivo (Kim et al, 1993;Goldman et al, 1998;Lin et al, 1998). Since the human A431 carcinoma cells secrete high amounts of VEGF (Myoken et al, 1991) and develop in nude mice tumours whose growth is highly VEGF-dependent (Melnyk et al, 1996), they provide a good model to test the availability of molecules that inhibit VEGF bioactivity. When tumour volume reached 100 mm 3 (6 day), CMDB7 (10 mg kg À1 ) was administrated s.c. three times a week for 2 weeks. Tumours were measured and the results are presented as the mean tumour volume 7s.e.
In this study, we assessed the anti-VEGF activity of CMDB7, described recently in vitro (Hamma-Kourbali et al, 2001), on A431 xenografted in nude mice, an extremely aggressive tumour model.
CMDB7 is, to our knowledge, the only one of heparin analogues reported to be efficient in A431 xenograft model. This study demonstrated that the s.c. injection of viable A431 cells yielded a 100% tumour uptake rate. The 2 week treatment with CMDB7 resulted in a 73% tumour growth inhibition associated with a 66% decrease in endothelial cell density. Compared to the human breast MDA- MB-435 (Bagheri-Yarmand et al, 1999) and MCF-7ras (Bagheri-Yarmand et al, 1998b) tumours, the magnitude of response of the A431 tumours to CMDB7 treatment was greater. Here, we observed that rapid A431 tumour growth was associated with high intratumour endothelial cell density, suggesting a direct relation between vascularisation within the primary tumour and the tumour growth rate. In 3-weekold A431 control (untreated) tumours, the endothelial cell density was 8.6%, while in 12-week-old MDA-MB-435 and MCF-7ras xenografts this value was 4.9% (Bagheri-Yarmand et al, 1999) and 6.1% (Bagheri-Yarmand et al, 1998b), respectively. Our observations are in agreement with results of Kim (1993), which demonstrated that the inhibitory effect of anti-VEGF antibody on tumour growth was more pronounced in the case of human A673 rhabdomyosarcoma secreting the highest quantity of VEGF and giving the most rapidly growing tumours as compared to G55 glioblastoma and SK-LMS-myosarcoma. In the CMDB7-treated tumours, a reduction of 66% in the density of endothelial cells indicates that this treatment attenuated the rate of neovascularisation, but did not completely reverse the initial activation of angiogenesis. The augmentation of CMDB7 dose did not result in increased efficiency of the drug in vivo (data not shown). Our results demonstrate that CMDB7 inhibited A431 tumour growth by, at least in part, decreasing intratumour endothelial cell density. The mechanism of CMDB7 action on endothelial cells is probably not direct and involves, as we recently described in vitro (Hammakourbali et al, 2001), a direct interaction of the drug with VEGF 165 that becomes unavailable for specific receptors. In agreement, we demonstrate here that CMDB7 inhibits the A431-CM stimulation of endothelial cell proliferation.
The other mechanism by which CMDB7 reduced the A431 tumour growth is direct inhibition of A431 cell proliferation as evidenced by a decrease of proliferative index in treated xenografts compared to nontreated ones. In this study, we demonstrated that CMDB7 inhibited, like VEGF, the binding of 125 I-VEGF 165 to A431 cells with IC 50 similar to the concentration at which CMDB7 inhibits efficiently the A431 proliferation in vitro. These findings could argue for the possible autocrine mitogenic action of VEGF on A431 cells. However, the depletion of VEGF amount in A431conditioned medium by anti-VEGF antibody did not affect the A431 proliferation, although it did inhibit endothelial cell growth. It suggests that VEGF binding sites on the A431 cell surface are not involved in classical, KDR-dependent transmission of mitogenic signal. The A431 growth decrease by CMDB7 in vitro could involve the inhibition of other mitogenic growth factors. This interpretation can be strengthened by our previous studies demonstrating that CMDB7 inhibited the activity of heparin-binding PDGF and TGFb by altering their conformation, but did not change the activity of EGF and IGF1, which are not heparin-binding growth factors (Bagheri-Yarmand et al, 1997, 1998a. Independently, the possible VEGF autocrine pathway in A431 could mediate tumour cell survival by protecting them from apoptosis as it was recently reported for breast cancer MDA-MB-231 cells (Bachelder et al, 2001). Further studies are necessary to understand the mechanisms of direct CMDB7 inhibitory action on A431 proliferation in vitro.
Altogether, our findings demonstrate that CMDB7 has a strong antiangiogenic and antitumour action in vivo, also when tumour cells produce a high level VEGF and EGFRs. CMDB7 acts directly on both tumour and endothelial cells, decreasing in a potent manner the tumour growth by increasing the proliferation of tumour cells and especially angiogenesis in vivo. The development of resistance to antiangiogenic drugs is becoming apparent (Kerbel  et al, 2001). It is very important, now, to enlarge the diversity of molecular targets for antiangiogenic drugs and to use a combination of antiangiogenic therapies. One of the possible mechanisms of this resistance may be due to redundancy of different proangiogenic growth factors made by tumour cells. When one angiogenic factor is targeted, the cancer cells increase production of other angiogenic factors. In this context, we believe that the ability of CMDB7 to interact with several angiogenic factors, including VEGF (Hamma-Kourbali et al, 2001), bFGF (Bagheri-Yarmand et al, 1997, 1998a, TGF-b and PDGF (Bagheri-Yarmand et al, 1998b), will permit to oppose or at least put off the development of resistance. Recently, it was reported that the resistance of tumours to treatment with EGF receptor-blocking antibodies can be associated with an elevated expression of VEGF . Since we show in this study that CMDB-7 efficiently blocks in vivo the effects of VEGF produced at high level, we can speculate that this drug could be useful in the case of failure to anti-EGFR treatment. It is believed now that because angiogenesis is a complex and multistage process, treatment with more than one antiangiogenic agent may be beneficial (Cherrington et al, 2000). Also, the neutralisation of angiogenic growth factors, especially VEGF, in tumour with CMDB7 may increase the effects of a variety of antiangiogenic inhibitors . For example, the reduced ability of Taxotere to induce apoptosis of endothelial cells in the presence of VEGF (Sweeney et al, 2001) could be restored by combined treatment with CMDB7. CMDB7 can be used not only as monotherapy but also especially in combination with other antiangiogenic and anticancer drugs to cause acute tumour regression by delaying development of resistance and by enhancing the effects of other drugs.