Human urinary bladder cancer is considered an increasingly significant public health issue in the industrialized countries, with a worldwide estimate of about two million patients . Due to the importance of Hsp90 molecular chaperone on client protein maturation and function, along with its voluminous and highly diverse clientele of cancer-related proteins, a variety of Hsp90 inhibitors have emerged as promising anticancer agents [6, 12]. In the present study, we have comparatively examined the effects of 17-AAG-induced Hsp90 inhibition on multiple protein targets implicated in signaling pathways critically regulating cell proliferation, apoptosis and motility, in RT4 (grade I), RT112 (grade I-II) and T24 (grade III) human urothelial bladder cancer cells. The data presented herein clearly demonstrate that, upon 17-AAG treatment, cell type-specific downregulation of multiple signaling molecules is followed by cell cycle arrest, finally resulting in Caspase-mediated cell death.
Depending on the cellular context and malignancy grade, 17-AAG has been shown to facilitate arrest in all checkpoints of the cell cycle, as for example, in human malignant pleural mesothelioma (G1 or G2/M block)  and breast cancer cells (G1 block) overexpressing HER2 . In all human bladder cancer cell lines examined in this study, apoptotic cell death was found to be preceded predominantly by a drug dose-dependent G1/S cell cycle block, with arrest in other phases of the cell cycle appearing in a cell type-specific manner. The unpredictability of cell cycle arrest induced by 17-AAG in bladder cancer cells is in agreement with previous reports and might be related to differences in client protein repertoires and cellular contexts . To elucidate the 17-AAG-induced block of the cell cycle, we undertook analysis of expression and/or activation profiles of several key-modulators of cell cycle progression. This demonstrated that, in response to 17-AAG exposure, the drug-dependent protein downregulation patterns correlate well with the observed G1 arrest of the cell cycle, as well as with the reduction in cell proliferation capacity.
Implementation of apoptosis, due to the effect of 17-AAG, has previously been reported in glioblastoma  and colon cancer . In the bladder cancer cell lines used in this study, cell type-specific and drug dose-dependent activation of a Caspase-induced cell death program proved to be initiated upon 17-AAG administration. These findings are in accordance with the survival rates observed in the cytotoxicity tests, although, in these experiments, 17-AAG-induced cell death percentages in the three bladder cancer cell lines were not found to differ significantly. In contrast, the cell-type specific profile of Caspase repertoire activation, and especially the diminished levels of processed Caspase-3 in RT112 and T24 cells, could possibly implicate other types of executive Caspases not studied here (i.e. Caspase-6 or -7) or even Caspase-independent cell death mechanisms such as autophagy [25, 26].
Hsp90 expression levels seem to be upregulated in cancer, resulting in addiction of tumor cells to multiple oncogenic pathways in which Hsp90 clients play a critical role. In bladder cancer, Hsp90 was found to be expressed in more than 90% of human tumor specimens, with high-grade and muscle invasive tumors expressing significantly higher levels of Hsp90 than low-grade and superficial tumours . Nevertheless, in 10% of the tumor samples Hsp90 expression was found to be downregulated and this was associated with infiltrating recurrences and poor prognosis [28, 29], most likely due to the overall molecular profile of the individual tumors. Besides the importance of Hsp90 expression levels, specific conformations of the chaperone have been implicated in cancer versus normal cell sensitivity to Hsp90 inhibitors: Hsp90 was shown to display higher binding affinity for 17-AAG exclusively in cancer cells , leading to the formation of 17-AAG-sensitive Hsp90-containing "superchaperone" complexes in malignant cells, whereas normal cells bearing a predominantly uncomplexed Hsp90 are significantly less sensitive to these types of inhibitors [13, 30]. This feature is likely exploited by Hsp90 targeting with the use of 17-AAG and subsequent effects on multiple Hsp90 targets.
Hsp90 inhibition and subsequent Hsp70 and Hsp27 upregulation, due to 17-AAG, have been reported in human colon , prostate  and cervical cancer cells . As presented in this study, even though a 17-AAG-induced Hsp90 downregulation was detected in all bladder cancer cell lines over a 24-hours treatment period, a cell type-specific pattern of inhibition was observed. In RT4 and RT112 cells, after exposure to the highest dose of the drug, an additional protein band was generated, whereas no such band could be detected in T24 cells. This novel finding in relation to Hsp90 structural integrity, upon high dose of 17-AAG administration, is presented herein for the first time. We suggest that this fragment may well be a product of Hsp90 proteolytic processing by Granzyme B [14, 15]. Use of the GrabCas algorithm has revealed a putative Granzyme B recognition and cleavage site in the amino acid sequence of both Hsp90α and Hsp90β protein isoforms, indicating that Hsp90 must be a bona fide substrate of Granzyme B. On the contrary, no Caspase cleavage site could be identified, with the help of GrabCas, fitting to the molecular weight of the possible Hsp90 cleavage fragment under discussion. Interestingly, Hsp90 cleavage has been reported previously, as a response to oxidative stress factors , arsenic-based compounds  and exposure to doxorubicin and cisplatin chemotherapeutic agents [14, 15]. Yet, it is not known whether the putative cleavage product is associated, somehow, with malignancy grade or p53 genetic status of the cells, since RT4 and RT112 are grade I and I-II, respectively, harboring a wild-type p53, whereas T24 are grade III, bearing a mutant p53 (Figure 6A). Intriguingly, the RT4- and RT112-specific production of a ~ 65 kDa putative proteolytic fragment could further enhance the functional amputation effect of 17-AAG on Hsp90, likely acting as a putative dominant negative component able to severely impair Hsp90 chaperoning properties. Thus, despite the Hsp90 upregulation observed in response to the highest 17-AAG concentration in grade I (RT4) and I-II (RT112) cell lines, the protein, due to its functional titration by the ~ 65 kDa processed product, seems unable to support its numerous clients thoroughly analyzed here. Therefore, we suggest that the chaperosomes containing these Hsp90 truncated forms are most likely inefficient to exert their cellular tasks.
The three bladder cancer cell lines seemed to follow a distinct and cell type-dependent downregulation profile of the Hsp90 molecular chaperone. However, as shown herein, despite 17-AAG administration, gene expression at the level of transcription remained unaffected for both isoforms of Hsp90 (α and β), clearly indicating that the regulation of Hsp90 is beyond transcriptional control, but occurs more likely at the post-translational level, via ubiquitination and subsequent proteasomal degradation or autophagy.
Hsp90 inhibition was suggested to be tightly associated with a compensatory upregulation of Hsp70  and/or Hsp27 protein levels, likely inducing resistance to 17-AAG . In this work, upon exposure to 17-AAG, total Hsp70 expression levels proved to exhibit a dose-dependent increase and generation of an ~ 65 kDa protein fragment in all three cell lines, reaching peak value at dose 10 μΜ. Using the GrabCas software, we propose that, similarly to Hsp90, the lower molecular weight band could likely represent a product derived from Hsp70 proteolytic processing by 17-AAG-induced Granzyme B activity, but not Caspase protease function.
CHIP was studied in order to illuminate the intriguing pattern of Hsp90 protein level alterations after 17-AAG treatment. CHIP levels were found to be downregulated in a dose-dependent manner in all three bladder cancer cell lines, suggesting a CHIP-regulated effect on proteasomal degradation of associated target proteins, such as Hsp90 and its clients. However, the higher dose-dependent upregulation of Hsp90 and α-tubulin implies a likely redundant, or non-essential, role of CHIP and, therefore, other ubiquitin ligases must be critically implicated in this type of response. An alternative scenario is that affinity threshold phenomena are at play here, with CHIP, although downregulated, still being able to implement its ubiquitin ligase activities regarding Hsp90 clients, but not Hsp90 itself.
The critical role of IGF-IR/Akt signaling pathway deregulation in tumor cell proliferation, survival and migration has been well documented . It has been previously reported that 17-AAG administration causes severe inhibition of the Akt-dependent signaling pathways in osteosarcoma  and gastric cancer . As demonstrated here, in human urinary bladder cancer cells, 17-AAG-induced inhibition of Hsp90 resulted in a cell type-specific downregulation of several proteins involved in Akt-dependent signaling, critically contributing to the negative regulation of proliferation, survival and motility. As a consequence, NF-κB transcription null activation potential was significantly compromised, mainly due to the sequestration of the factor into the cytoplasm, as clearly illustrated in Figure 8A. Reduced NF-κB activity was indirectly assessed by measuring the mRNA expression levels of Survivin and cIAP1, two well known bona fide NF-κB target genes. Thus, we have demonstrated that 17-AAG-dependent inhibition of NF-κB activity is tightly associated with transcriptional repression of Survivin and cIAP1 anti-apoptotic genes, thus decisively contributing to the cytotoxic potency of 17-AAG by decreasing the required "apoptotic threshold" in bladder cancer cells .
Moreover, 17-AAG-mediated Hsp90 inhibition resulted in alterations of the phosphorylation status of members of the Forkhead family of transcription factors (FOXO), immediate downstream substrates of Akt kinase, in bladder cancer cells. As shown in this study, FOXO factors proved to be strongly phosphorylated in the highly malignant T24 cells, whereas extremely low, but detectable, levels were also observed in RT112 cells. Administration of 17-AAG caused a notable downregulation of phosphorylated FOXO1 and FOXO3 family members, likely inducing an enhancement of their apoptotic activity.
Interestingly, the undetectable phosphorylation of the IGF-I-dependent downstream mediators (i.e. IGF-IR, Akt, IKKs and FOXOs) in RT4 cells strongly suggests the deactivated character of the pathway under the particular growth conditions, whereas, on the contrary, in T24 cells the IGF-IR/Akt pathway seems to be constitutively activated. RT112 cells proved to display an intermediate pattern of signaling potency, with the IGF-IR/Akt pathway being activated at very low levels. This novel finding of cell type-specific activation of the IGF-IR/Akt-dependent signaling repertoire, herein demonstrated for the first time, could be tightly associated with the underlying differences in various features of the malignant phenotype observed in the three bladder cancer cell lines examined.
Hsp90 inhibition and ensuing Akt inactivation in bladder cancer cells was accompanied by downregulation of Erk1/2-dependent signaling. Exposure to 17-AAG has been previously reported to cause inhibition of the Raf/MEK/ERK signaling cascade in Hodgkin's lymphoma  and leukemia . Although total Erk1/2 protein levels exhibited a cell type-specific and drug dose-dependent response similar to the one of α-tubulin and Hsp90, phosphorylated p44/42 levels were severely downregulated in all bladder cancer cell lines, implying the differential control between total and phosphorylated protein destabilization processes in response to the high drug dose treatments.
Invasion and metastasis are one of the hallmark traits of cancer involved in the advanced stages of tumor progression. Hsp90 inhibition by ansamycins has been reported to suppress cancer cell motility and invasion through depletion of the HGF/c-Met signaling pathway in both leiomyosarcoma and glioblastoma cell lines . Another novel finding of the present study is the notable expression and constitutive activation of c-Met receptor in T24 bladder cancer cells, whereas in RT4 and RT112 cells total c-Met protein levels were either absent (RT4) or barely detectable (RT112). Finally, in this study, we have clearly demonstrated that T24 cell line expresses high amounts of phosphorylated IGF-IR, Akt, FOXOs, p44/42 and c-Met proteins and exhibits strong migration dynamics, which could well be associated with a more invasive and metastatic potency, exactly as a result of this over-activated signaling network. Nevertheless, we have shown that the inhibitory effect of 17-AAG on T24 cells is reflected on the significant decrease of both total and phosphorylated c-Met protein levels, with subsequent suppression of other oncogenic parameters, such as increased cell proliferation and motility, hence critically contributing to the impairment of aggressive cancer cell phenotype [41–43].