Our present findings demonstrate that both the α and the β isoforms of HSP90 are secreted by cultured MDAMB453 human breast cancer cells and interact with matrix metalloproteinases, MMP2 and MMP9. Additionally, we show that a monoclonal antibody against HSP90, mAb 4C5, prevents maturation of the two metalloproteinases in vitro and inhibits metastatic deposition of breast cancer cells in vivo.
We have previously reported the presence of HSP90α and HSP90β on the cell surface of MDAMB453 breast cancer cells . Here we further examined the extracellular localization of HSP90. More specifically, we demonstrated the secretion of both HSP90α and HSP90β in the culture medium of MDAMB453 cells using antibodies specific for the two isoforms. At this point it is important to note that specificity of the anti-HSP90β antibody was further confirmed since, Western blot analysis of human recombinant HSP90α using this antibody gave negative results (see additional files 1 and 2). Our findings, for the first time to our knowledge, reveal secretion of HSP90β by breast cancer cells and are only partly in agreement with existing knowledge. Indeed, previous studies have reported secretion only of HSP90α from various cell types including HT 1080 fibrosarcoma cells , human dermal fibroblasts under hypoxia conditions and transforming growth factor alpha stimulated human keratinocytes 
Controlled degradation of the extracellular matrix is essential for the invasion and metastasis of malignant tumors. In this context matrix metalloproteinases and in particular MMP2 and MMP9 are of crucial significance for tumor development and progression . Indeed increasing experimental evidence indicates involvement of both enzymes at multiple steps of the metastatic process It is important to note that these metalloproteinases are secreted as inactive pro- enzymes and acquire their active form extracellularly [3, 22]. Taking these into account together with accumulating data demonstrating extracellular activity of HSP90 , we next examined the possible interaction of this molecule with the two metalloproteinases using co-immunoprecipitation experiments. Our results showed that both the α and the β isoforms of HSP90 interact with the secreted, inactive forms of MMP2 and MMP9. Interestingly, reverse co-immunoprecipitation experiments revealed that HSP90 also interacts, with the extracellularly activated forms of the two MMPs, The above observations indicate that HSP90 participates in the activation of MMP2 and MMP9. Moreover, they suggest a second level of chaperoning for this molecule associated with the mature forms of the metalloproteinases examined. Our findings are in accordance with previously reported data showing association of the secreted pool of HSP90α with MMP2 [15, 23]. Moreover they provide new evidence for a further increase in the number of clients of this chaperone molecule by adding MMP9 to the list of its extracellular substrates.
We have previously demonstrated that mAb 4C5 inhibits the invasive capacity of MDAMB453 breast cancer cells by disrupting the interaction of cell surface HSP90 with the extracellular domain of HER-2, a member of the ErbB family of epidermal growth factor receptors . Taking this into account together with the above mentioned findings we next investigated whether mAb 4C5 independently of its effect on HER2 has also an effect on the secretion and activation of MMP2 and MMP9. Our results showed that when the cell- impermeable mAb 4C5  is added to the culture medium of MDAMB453 cells, secretion of pro-MMP2 and pro-MMP9 is not affected; however and as shown by zymography and western blot analysis, the activation of these metalloproteinases is dramatically inhibited as compared to controls where mAb 4C5 is not included in the culture medium. Moreover, co-immunoprecipitation experiments performed as described in Methods, revealed that mAb 4C5 effectively disrupts the interaction of HSP90 with MMP2 and MMP9 respectively. Our present data indicate that mAb 4C5 while not affecting secretion of the inactive forms of MMP2 and MMP9, prevents their maturation most probably by disrupting their interaction with the extracellular pool of HSP90. Similar to our results have been previously reported by Eustace et al  who have shown that inhibition of extracellular HSP90α decreases activation of MMP2. At this point it should be noted that preliminary experiments performed using the MDAMB231 breast cancer cells which lack the HER2 receptor, indicate not only that these cells too secrete both isoforms of HSP90 but also that MMP2 and MMP9 activation by extracellular HSP90 may be independent of the HER2/HSP90 interaction previously reported , since in this system as well mAb 4C5 prevented activation of these metalloproteinases (unpublished data). Nevertheless our overall observations do not exclude the possibility that in combination with HSP90 additional molecules, such as other HSPs, HSP90 co-chaperones etc may be necessary for the activation of the MMPs studied in this work. This however, needs further investigation.
We have previously demonstrated that mAb 4C5 significantly inhibits melanoma metastasis in C57BL/6 mice inoculated with B16 F10 melanoma cells, . supporting accumulating data showing that antibodies or other small molecules that inhibit HSP90 can be used as anti-cancer agents  This prompted us to explore the anti-metastatic activity of mAb 4C5 with respect to MDAMB453 breast cancer cells. In line with the recently reported in vitro results , in this work we showed that mAb 4C5 strongly inhibits the metastatic depositions of MDAMB453 cells into the lungs of SCID mice. More specifically an 86.67% inhibition of metastatic depositions of MDAMB453 cells was observed in the mAb 4C5 treated mice as compared with control animals. It is interesting to note that in the experimental group of mice that received mAb 4C5, MDAMB453 cells were often observed stagnating on the inner surface of large pulmonary vessels, whereas such a phenomenon was very rarely detected in the control mice. Quantification of this occurrence confirmed a statistically significant difference (p < 0,001) between the two experimental groups. This remarkable observation together with the above mentioned data, tempts us to reason that mAb 4C5 exerts its activity by disrupting the interaction of extracellular HSP90 with MMP2 and MMP9 and thus preventing activation of these matrix metalloproteinases which, as is well documented [7, 8], is necessary for the degradation of type IV collagen, a major constituent of the basement membrane associated with the pulmonary blood vessel. As a consequence, tumor cells remain limited to the inner surface of the vessel and their infiltration into the lung tissue is impaired. Regulation of MMP activation has been previously correlated with cancer cell extravasation. In particular Cruz-Munoz et al have reported that TIMP-3 decreases MMP2 activation which in turn limits tumor cell extravasation and subsequent colonization of the lung.