The Chaperone System in Salivary Glands: Hsp90 Prospects for Differential Diagnosis and Treatment of Malignant Tumors

Salivary gland tumors represent a serious medical problem and new tools for differential diagnosis and patient monitoring are needed. Here, we present data and discuss the potential of molecular chaperones as biomarkers and therapeutic targets, focusing on Hsp10 and Hsp90. The salivary glands are key physiological elements but, unfortunately, the information and the means available for the management of their pathologies, including cancer, are scarce. Progress in the study of carcinogenesis has occurred on various fronts lately, one of which has been the identification of the chaperone system (CS) as a physiological system with presence in all cells and tissues (including the salivary glands) that plays a role in tumor-cell biology. The chief components of the CS are the molecular chaperones, some of which belong to families of evolutionarily related molecules named heat shock protein (Hsp). We are quantifying and mapping these molecular chaperones in salivary glands to determine their possible role in the carcinogenetic mechanisms in these glands and to assess their potential as diagnostic biomarkers and therapeutic targets. Here, we report recent findings on Hsp10 and Hsp90 and show that the quantitative and topographic patterns of tissue Hsp90 are distinctive of malignant tumors and differentiate benign from malignant lesions. The Hsp90 results show a correlation between quantity of chaperone and tumor progression, which in turn calls for negative chaperonotherapy, namely, elimination/inhibition of the chaperone to stop the tumor. We found that in vitro, the Hsp90 inhibitor Ganetespib is cytotoxic for the salivary gland UM-HACC-2A cell line. The drug, by interfering with the pro-survival NF-κB pathway, hampers cellular proliferation and migration, and favors apoptosis, and can, therefore, be considered a suitable candidate for future experimentation to develop a treatment for salivary gland tumors.


Introduction
Salivary gland tumors are phenotypically heterogeneous and diverse in nature encompassing over 21 different subtypes [1]. While their incidence has been on the rise during the past decades [2,3], they still constitute a rarity among tumors, accounting for 8.1% of all head and neck cancers and 0.5 to 1.2% of all carcinomas [4][5][6]. Consequently, their identification and classification present a major challenge for histopathologists, especially for the untrained in routine microscopy. As a result, misdiagnoses of salivary gland tumors with erroneous identification and/or classification can be expected. Thus, developing novel diagnostic criteria and tools that can be applied without the need of a prerequisite experience in salivary gland histopathology is a necessity. adverse effects (AE) in patients [43][44][45]. Its anti-cancer potential was never tested in salivary gland malignancies.
In this study, we investigated the pattern of expression of two molecular chaperones, Hsp10 and Hsp90, in tumors of the major salivary glands and we examined the effect of Ganetespib on the PI3K/Akt/NF-κB axis in adenoid cystic carcinoma (ACC), aiming to begin identification of efficacious drugs for its treatment.

Immunohistochemical Assessment of Hsp10 and Hsp90
All results are summarized in Table 1, in which the differences between the benign and malignant pathologies relevant to Hsp90 and presented in detail below, can be visualized. The anti-Hsp90 antibody used for IHC is described by the supplier as able to recognize both cytosolic Hsp90 isoforms, alpha and beta. 2 Ducts and acini were no discernable in tumor specimens.

Hsp10 Levels and Topography in the Tissue of Normal and Pathological Salivary Glands
The immunohistochemical reaction of Hsp10 showed a granular and diffuse cytoplasmic positivity in all specimens ( Figure 1). The Hsp10 positivity in the ducts and acini was of high intensity ( Figure 1B). A moderate to high intensity was found in the epithelium of the ducts of sialadenitis while the acini displayed a weaker intensity ( Figure 1E). In the epithelial component of WT and PA, the immunopositivity was predominantly granular cytoplasmic with a moderate intensity ( Figure 1H,K). A moderate to strong granular cytoplasmic Hsp10 positivity was found in the epithelium of the neoplastic cells of EX-PA ( Figure 1N), and moderate to intense granular cytoplasmic positivity was found in the cells of MUC ( Figure 1Q). In the ACC cancerous cells, the positivity was also granular but more intense than in MUC ( Figure 1T). In the histograms to the right, it can be seen that the average percentage of Hsp10 immunopositivity was highest in normal SMG and PG (Figure 1a  H&E of mucoepidermoid carcinoma (MUC) (P). Hsp10 (Q,i) and Hsp90 (R,j) are present in the cytoplasm of the epithelium of the neoplastic cells (arrowheads). H&E of adenoid cystic carcinoma (ACC) (S). Hsp10 (T,k) and Hsp90 (U,l) occur in the cytoplasm of the neoplastic cells (arrowheads). * Significantly different (p ≤ 0.005), ** (p ≤ 0.05), *** (p ≤ 0.001), **** (p ≤ 0.0005). Bar 20 µm.

Hsp90 Tissue Levels Decrease in Sialadenitis
The Hsp90 reaction showed a diffuse cytoplasmic positivity in all specimens ( Figure 1). The Hsp90 positivity observed by IHC and IF in the ducts and acini of normal SMG and PG was of moderate intensity ( Figures 1C and 2A). A weak Hsp90 positivity was found in the epithelium of the ducts and acini of sialadenitis ( Figure 1F). In this chronic inflammatory condition, there was a decrease in Hsp90 positivity by comparison with normal SMG and PG (p = 0.0012) (Figure 1b). Right. Histogram. The staining intensity for Hsp90 (bars) in Normal (G), WT (G), PA (H), EX-PA (I), MUC (J), ACC (K) was expressed as the mean pixel intensity (PI) normalized to the CSA (crosssectional area), using the software Leica Application Suite Advanced Fluorescences software. Data are presented as the mean ± SD. * Significantly different (p ≤ 0.005), ** (p ≤ 0.05). Abbreviations: Normal, normal submandibular and parotid gland; WT, Warthin's tumor; PA, pleomorphic adenoma; EX-PA, carcinoma ex-pleomorphic adenoma; MUC, mucoepidermoid carcinoma; ACC, adenoid cystic carcinoma.

Hsp90 Tissue Levels Decrease in Benign Tumors
In the epithelial component of WT ( Figures 1I and 2B) and PA ( Figures 1L and 2C), the Hsp90 immunopositivity was predominantly diffuse cytoplasmic with a weak intensity but slightly higher than in sialadenitis. A marked decrease in the Hsp90-immunopositive tumor cells occurred in WT and PA by comparison with normal SMG and PG, as shown by IHC (Figure 1d

Ganetespib Reduces UM-HACC-2A Cell Viability by Inducing Apoptosis
Ganetespib displayed a cytotoxic effect on UM-HACC-2A cells after 24 h by reducing their viability in a dose-dependent manner from 0 to 60 nM, after which it plateaued at around 50% viability even at 150nM concentration ( Figure 3A). Twenty-four hours treatment with 60 nM Ganetespib was selected for the rest of the experiments. Ganetespib induced apoptosis in cells treated with 60 nM concentration for 24 h. Typical morphological features of nuclear apoptosis were assessed by DAPI staining; nuclear fragmentation, chromatin condensation, and apoptotic bodies were apparent in Ganetespib-treated cells (p < 0.0001) ( Figure 3B,C). These results were confirmed by TUNEL assay that shows apoptosis through DNA strand-breaks labeling. Consistently, Ganetespib-treated cells were more intensely and numerously stained when compared to non-treated cells (p = 0.0061) ( Figure 3D,E).

Ganetespib Reduces Cell Proliferation and Migration
Ganetespib significantly (p = 0.0016) reduced cellular proliferation by decreasing UM-HACC-2A cell number ( Figure 4A,B). Moreover, Ganetespib significantly reduced cellular migration, as UM-HACC-2A-treated cells were unable to fill the wound in comparison with the non-treated cells (p = 0.0005) ( Figure 4C,D).

Expression of Hsp90 in Ganetespib-Treated UM-HACC-2A Cells
Ganetespib is a potent second generation Hsp90 inhibitor and its anti-cancer properties has prompted its entry into clinical trials [43][44][45]. However, its effect on the chaperones in adenoid cystic carcinoma of the salivary glands has not yet been reported. After treatment with Ganetespib, Hsp90 (p = 0.0350) ( Figure 5A

Expression of Caspase 3 in Ganetespib-Treated UM-HACC-2A Cells
The mitochondrial cytochrome c-caspase 3 intrinsic apoptotic pathway is an important cell-death pathway [52]. The effect of Ganetespib on caspase 3 activation has been reported in carcinogenesis but not for the salivary gland carcinoma [40,41,[53][54][55][56]. We examined the expression of the precursor of caspase 3 (pro-caspase 3) and its activated form (cleaved caspase 3) in Ganetespib-treated UM-HACC-2A cells. The results indicate that both procaspase 3 and the activated caspase 3 protein levels are unchanged in Ganetespib-treated cells compared with non-treated cells (p = 0.8302) ( Figure 5A,F).

Expression of VEGF in Ganetespib-Treated UM-HACC-2A Cells
Vascular endothelial growth factor (VEGF) is a potent cytokine involved in the induction of neovascularization that participates in cellular migration, invasion, and ultimately metastasization [57,58]. A few studies report the effect of Ganetespib on VEGF in cancer but none of them in salivary gland carcinoma [58][59][60]. We examined the expression of VEGF protein monomer and dimer in Ganetespib-treated UM-HACC-2A cells. VEGF levels showed a tendency to increase in Ganaetespib-treated cells by comparison with non-treated cells but this was not statistically significant (p = 0.3012) ( Figure 5A,G).

Discussion
Current methods and information are insufficient for proper and rapid diagnosis of tumors of salivary glands and, consequently, patient management is difficult. Our work is a contribution toward remedying this situation for the benefit of health professionals and patients in as much as it adds new criteria for the differential diagnosis of tumors of SMG and PG. We standardized the methods to assess the levels of Hsp10 and Hsp90 in normal and tumoral tissues, and present examples of the type of results the methods provide. This study demonstrates for the first time that assessing Hsp90 can help to discern between malignant and benign tumors of major salivary glands, using two independent immunomorphological techniques, i.e., IF and IHC, with separate analytical methodologies. Our data show that Hsp90 is a promising diagnostic biomarker for salivary gland pathology because it can help differentiate not only between normal and tumorous glands but it can also help discern between normal and inflamed salivary glands, i.e., sialadenitis. Our data show an up-regulation in Hsp90 tissue levels in three types of salivary gland carcinomas, i.e., ACC, MUC, and EX-PA, when compared to normal salivary glands. In contrast, we found a down-regulation in Hsp90 tissue levels in benign salivary gland tumors, WT and PA, and in inflamed salivary glands when compared to their healthy counterpart. Among the four molecular chaperones we have thus far assessed in tumors of salivary glands [12,16] and this work, Hsp90 was the only chaperone that displayed a specific quantitative pattern of expression at the tissue level depending on salivary gland disease, which might indicate that this chaperone plays a role in pathogenesis, including tumorigenesis and inflammation. Based on those results, screening for Hsp90 tissue levels by IF and IHC in patients with inflamed or tumorous salivary glands can be considered useful and should be added to conventional diagnostic tools in pathology units to reduce false positives and increase accuracy of diagnosis.
Hsp90 is considered to be an important molecular chaperone in cancer because of its support of tumor cells and ability to stabilize mutated proto-oncogenes. Consistent with our results, Hsp90 levels did not vary in lobular neoplasia of the breast [61], while Hsp90 positive cells were significantly increased in invasive breast carcinoma [62], again indicating a role of Hsp90 in promoting malignancy. Hsp90 overexpression has been documented as a potential diagnostic marker in hepatocellular carcinoma [63]; cervical [64], colorectal [65], and breast [62] cancers; melanomas [66]; leiomyosarcomas [66]; gastrointestinal stromal tumors (GIST) [66]; and malignant peripheral nerve sheath tumors [66]. Tissue Hsp90 up-regulation was correlated with higher risk GIST and extragastric locations, and served as an independent predictor of recurrence in patients with GIST after complete surgical resection [66]. Hsp90 high IHC scores in oral squamous cell carcinoma (OSCC) [67][68][69], nasopharyngeal carcinoma [70] were associated with lymph node metastasis and a dismal survival rate. Worse OSCC clinical parameters were seen in Hsp90 high-tissue level specimens, which emphasizes the possible role of Hsp90 as a potential independent prognostic biomarker that can effectively predict poor prognosis in OSCC [71]. High Hsp90 expression was associated with worse overall survival in breast cancer, although no correlation between Hsp90 tissue levels and node metastasis was found [72]. Hsp90 increased tissue expression is prominent in melanoma malignancies and metastases when compared with melanocytic nevi [73]. Interestingly, assessing Hsp90 tissue levels was useful in differentiating between low-grade and high-grade ocular surface squamous lesions [74]. Conversely, Hsp90 immunostaining decreased in infiltrative lobular carcinoma confirming once again the specificity of the pattern of expression of Hsps based on tissue, cell, and cancer type [75].
Anti-cancer targeted therapy via Hsp90 inhibition gained significant interest in the last two decades. Geldanamycin (GA) and Radicicol (RC) were the first discovered natural Hsp90 inhibitors that showed promising results in cancer therapy [76,77]. First generation Hsp90 inhibitors GA-derivatives were then synthesized with improved stability, solubility, and binding affinity to Hsp90. These early Hsp90 inhibitors had success in prompting an anti-cancer response and several entered clinical trials. Unfortunately, their severe adverse effects ranging from hepatotoxicity to ocular toxicity resulted in discontinuation of the trials [44]. Ganetespib is a second generation Hsp90 inhibitor with improved affinity and efficiency. In addition, Ganetespib is highly potent at low doses whilst showing minimal adverse effects such as liver toxicity and retinal apoptosis [43,44].
There is a need of salivary gland carcinoma cell lines to make progress in the development of anti-cancer therapies. One available cell line derived from the minor salivary glands adenoid cystic carcinoma is the UM-HACC-2A cell line, established in 2018 [78]. Our study is the first that investigates the effect of an anti-cancer drug such as Ganetespib in this cell line. Ganetespib treatment resulted in an antitumorigenic effect on UM-HACC-2A cells by diminishing cellular proliferation and migration and, furthermore, induced cell death via apoptosis. At 24 h after treatment with Ganetespib, the cells' viability was approximately 50%, although there was no further reduction at higher doses, i.e., 120 and 150 nM. This could be due to UM-HACC-2A cells possessing the mutant BRAF gene that was reported to confer resistance to conventional chemotherapeutic drugs [79][80][81][82].
A body of evidence suggests that carcinoma development is largely linked to accumulation of somatic missense mutations and activation of mutated proto-oncogenes [81]. A prominent example of genetic alterations in cell growth regulatory genes that has been associated with specific colon cancers and melanomas is a single point mutation in the BRAF gene [79][80][81][82]. RAF is part of the signaling cascade of the MAPK pathway which regulates cellular growth. RAS binds to RAF (or MAP kinase kinase kinase) and activates it. RAF will in turn phosphorylate MEK (or MAP kinase kinase) that ends up activating ERK (or MAP kinase), allowing its translocation into the nucleus to phosphorylate and activate transcription factors [83]. Mutated BRAF has 10-fold greater kinase activity than wild-type BRAF [80]. Following Hsp90 inhibition, both ERK and Akt activation is dampened but due to mutated BRAF's increased kinase activity, a rebound activation of Akt and ERK takes place and confers resistance against Hsp90 inhibition by activating mitogenic and survival pathways [82]. This leads us to hypothesize that applying a dual inhibition, or even a triple inhibition, for Akt and/or ERK and Hsp90 simultaneously, may prove to be effective to sensitize carcinomas possessing BRAF mutations to Hsp90 inhibitors.
The Ganetespib mechanism of action consists in blocking the Hsp90 ATP binding region on the N-terminus, thus competing with ATP and inhibiting Hsp90, which prompts Hsp90 clients proteins to adopt aberrant conformations, thereby triggering their dephosphorylation and/or UPS-mediated degradation [67,84]. Hsp90 inhibition is characterized by an up-regulation of Hsp70 (a phenomenon called heat shock response), which confirms the success of Hsp90 inhibition. Our data corroborate this since both Hsp90 and Hsp70 proteins were increased as a result of Hsp90 inhibition by Ganetespib in UM-HACC-2A cells. The heat shock response is also reported in OSCC [67], pancreatic cancer [85], hepatocellular carcinoma [86,87], thyroid carcinoma [88] and colon adenocarcinoma [89] cell lines. Hsp90 and Hsp70 form a complex in the cytoplasm and are bound to HSF1, the master stress-inducible transcription factor regulator, keeping its transcriptional activity repressed. Once Hsp90 inhibition occurs, HSF1 is released from the complex and its activity is no longer suppressed, thus enabling its translocation to the nucleus to induce transcription of several Hsp genes, including Hsp90 [67]. This could explain the increase in Hsp90 protein expression post-inhibition. The loss of functional active Hsp90 multichaperone complexes in cancer cells elicit an HSF1-dependent anti-stress response, which confers cell resistance and offsets the cytocidal effects of Hsp90 inhibitors [67,85].
Hsp90 tightly regulates and is indispensable for the stabilization and activation of both Akt and NF-κB pathways [90]. We studied the PI3K-Akt-NF-κB pathway in salivary gland adenoid cystic carcinoma, using the UM-HACC-2A cell line, and our data indicate that activated Akt (p-akt) is upregulated in the Ganetespib-treated cells. Knockdown of Hsp90 was shown to inhibit autophagy by activation of the PI3K/Akt/mTOR pathway in osteosarcoma [91]. The persistence of Akt activation after Hsp90 inhibition may be linked to the UM-HACC-2A cells having the mutant BRAF gene. Hsp90 inhibition by AUY922 was shown to inhibit Akt phosphorylation in wild-type BRAF colon cancer cells but not in mutant BRAF colon cancer cells [82]. While mutant BRAF was shown to be among Hsp90 several clients, Hsp90 inhibition failed to abolish mutant BRAF in colon cancer, suggesting that it may not be solely, nor heavily dependent on Hsp90 for its stabilization [92,93]. CDC37 is a co-chaperone of Hsp90 and has the ability to stabilize mutant BRAF independently of Hsp90 [82]. CDC37 specializes in the co-chaperoning of kinases as it mediates the bridging between Hsp90 and the client kinases before entering the chaperone cycle [94,95]. Stabilization of Akt in mutant BRAF colon cancer requires both Hsp90 and CDC37. The combined inhibition of Hsp90 and CDC37 was efficient in abrogating the phosphorylation of Akt in mutant BRAF colon cancer cells [82]. Conversely, other studies highlight the suppression of cellular proliferation, migration, and progression through deactivation of the PI3K/Akt pathway by Hsp90 knockdown or inhibition in colorectal [96], gastric [97], osteosarcoma [98], ovarian [99] and cervical [64] cancers. The discrepancy in the results may lie in the type of tissue, type of cancer and the cell lines that were used. Therefore, a strategy involving CDC37 inhibition in combination with Hsp90 inhibitors may be considered to potentiate the effect of Hsp90 inhibitors and combat BRAF mutation-acquired resistance.
Our results show that Hsp90 inhibition by Ganetespib in UM-HACC-2A cells downregulated the level of total NF-κB. This was expected, as Hsp90 is a key regulator of the IKK complex. Similar results, in which phosphorylated NF-κB is also downregulated as a consequence of Hsp90 inhibition, were reported for prostate carcinoma [100]. Hsp90 inhibition decreased constitutive and induced NF-κB activity in human myeloid leukemia [49,[101][102][103], chronic lymphocytic leukemia [104,105], primary effusion lymphoma [106], and melanoma [107], which led to cytotoxicity; cell-cycle arrest; reduction in proliferation, migration, and invasion; and induced apoptosis. Moreover, blocking NF-κB and Akt survival pathways using the Hsp90 inhibitor potentiates chemotherapeutic drug cytotoxicity in lung cancer [108]. Abolishing activation of the pro-survival NF-κB pathway in UM-HACC-2A cell by Ganetespib may have hindered cellular proliferation and migration and ultimately, led to apoptosis.

Ethics Statement
Human salivary gland tissues embedded in paraffin were selected from the archives of the Department of Pathology within the University of Palermo (Italy), Civico Hospital of Palermo (Italy), and Cervello Hospital of Palermo (Italy). All the specimens dated at least 10 years from the moment of the biopsy and therefore, no informed consent from the patients was needed. The study protocol conformed to the ethical guidelines for traceability, collection, transport, conservation, and archiving of cells and tissues for diagnostic investigations of the Pathology Unit issued on May 2015 by the Italian Ministry of Health.

Histopathology
Sections of salivary glands with a thickness of 5 µm were obtained from paraffin blocks and stained with hematoxylin and eosin (H&E) for histological examination. In brief, sections were de-waxed in xylene for 10 min and rehydrated by sequential immersion in decreasing ethanol concentrations. Then, the sections were stained with H&E [109] and analyzed using a Leica DM5000 upright microscope (Leica Microsystems, Heidelberg, Germany). All sections were examined by two independent observers (F.C. and F.R.) in a blind manner, using coded slides and not knowing their source.

Immunohistochemistry
Immunohistochemistry (IHC) reactions for Hsp10 and Hsp90 were carried out on 5-7 µm thick paraffin-embedded tissue sections. The IHC reactions were performed using the automated IHC system (IntelliPath Flx, Biocare Medical, distributed by Bio-Optica, Milan, Italy) of the Biotechnology Laboratory of the Euro-Mediterranean Institute of Sciences and Technologies (IEMEST). The primary antibodies used were anti Hsp10 (mouse monoclonal antibody, Santa Cruz Biotechnology, Dallas, TX, USA, D-8, sc-376313, dilution 1:100) and anti Hsp90 (mouse monoclonal antibody, Santa Cruz Biotechnology, F-8, sc-13119, dilution 1:200). The latter anti-Hsp90 antibody is described by the supplier as able to recognize the two cytosolic Hsp90 isoforms, alpha and beta. Nuclear counterstaining was carried out using hematoxylin (Hematoxylin aqueous formula, DAKO, Golstrup, Denmark. Cat. no. S2020). The slides were then re-dehydrated using ascending concentrations of alcohol followed by complete immersion in xylene. Finally, the slides were prepared for routine microscopy observation with coverslips and an aqueous mounting solution. The observation of the sections was performed with an optical microscope (Leica DM 5000 B, Heidelberg, Germany) connected to a digital camera (Leica DC 300 F). Each tissue section was examined on two separate occasions by two independent observers (FR and FC) to determine the percentage of cells positive for Hsp10 and Hsp90. The evaluation of immunopositivity percentage was calculated in a high-power field (HPF) at 400× of magnification and repeated for 10 HPFs. The average of the percentages of all immunoquantifications performed in each case for each described group was used for the statistical evaluation. This method of quantification/evaluation was chosen instead of using pertinent software, for example, ImageJ, because software is unable to differentiate between actual positivity and background signal/artifacts, thus reducing the credibility of the results and leading to false positives. Moreover, the software is unable to differentiate between tumor cells and other type of cells (e.g., infiltrating cells). The quantification was conducted by two trained pathologists with extensive experience in chaperonopathies, as described in previous publications from our laboratories [24,[110][111][112].
Positive Standards. The positivity of the sections for each of the chaperones was assessed following two main criteria: The percentage of positive cells (≤5% being considered as negative) and the intensity of staining, ranging from + (weak signal intensity) to +++ for the most intense signal, following established procedures [24].

Immunofluorescence and Confocal Microscopy
For immunofluorescence (IF), deparaffinized 5-10 µm salivary gland sections were incubated with antigen-retrieving solution (10 mM tri-sodium citrate, 0.05% Tween-20) for 8 min at 85 • C, and treated with a blocking solution (3% BSA in PBS) for 30 min. Next, the primary antibody anti-Hsp90 mouse monoclonal, Santa Cruz Biotechnology, F-8, sc-13119, dilution 1:50, was applied, and the sections were incubated in a humidified chamber at 4 • C overnight. Then, the sections were incubated for 1 h at 25 • C with a conjugated secondary antibody (anti-mouse IgG-FITC antibody produced in goat, F5897, Sigma-Aldrich, St. Louis, MO, USA). Nuclei were stained with Hoechst Stain Solution (dilution 1:1000, Hoechst 33258, Sigma-Aldrich). The slides were treated with PermaFluor Mountant (Thermo Fisher Scientific, Waltham, MA, USA) and covered with a coverslip. The images were captured using a Leica Confocal Microscope TCS SP8 (Leica Microsystems). Staining intensity for Hsp90 was expressed as the mean pixel intensity (PI) normalized to the CSA (cross-sectional area expressed in pixel) using the software Leica application suite advanced fluorescence software, as previously described [110,111,113]

Wound Healing Assay
UM-HACC-2A cell motility assays were carried out as previously described (Jiali Zhang & Peng, 2009) [114]. Two × 10 5 cells were seeded in a T25 flask for 24 h and left to reach confluency. The monolayer of cells was scraped using a 1 mL stripette, creating a clear surface area (wound). Then, OSGM was replaced by low serum medium deprived of growth factors (5% FBS and 1% PenStrep) in the control flask and 60 nM of Ganetespib diluted in the same low-serum medium was added in the treated flask and incubated for 24 h. To evaluate the migration of the cells, images of the wound were taken under a light microscope (Leica DMI 3000 B) at 0 time and at intervals of 24 h. The distance of the gap between the two edges of the wound were assessed using ImageJ software.

Cell Proliferation
Two × 10 5 cells were seeded in a T25 flask for 24 h and left to reach confluency. OSGM was then replaced by 5% FBS OSGM deprived of insulin for the control flask, while OSGM was replaced by 60 nM of Ganetespib diluted in 5% FBS OSGM deprived of insulin in the treated flask. To assess cellular proliferation, the cells were trypsinized and were solubilized in trypan blue and then loaded onto a hemocytometer. The cells were counted under the microscope.

DAPI Staining
One × 10 4 cells was seeded in a cell culture chamber slide for 48 h and left to reach confluency. OSGM was then replaced by 5% FBS OSGM deprived of insulin in the control well, while OSGM was replaced by 60 nM of Ganetespib diluted in 5% FBS OSGM deprived of insulin in the treated well and incubated for 24 h. After 24-h treatment, cells were washed three times with PBS and then fixed with cold methanol on ice for 30 min followed by two washes with PBS. Then, the cells were stained by DAPI for 15 min in the dark. The stained cells were washed twice with PBS and examined by fluorescence microscopy. Apoptotic cells were identified by chromatin condensation, nuclear fragmentation, and apoptotic bodies. The percentage of apoptotic cells was assessed by calculating the apoptotic index (A.I.) using the following formula: A.I. = number of apoptotic cells total number of cells × 100.

TUNEL Assay
One × 10 4 cells were seeded in a cell culture chamber slide for 24 h and left to reach confluency. OSGM was then replaced by 5% FBS OSGM deprived of insulin in the nontreated well, while OSGM was replaced by 60 nM of Ganetespib diluted in 5% FBS OSGM deprived of insulin in the treated well and incubated for 24 h. After 24-h treatment, cells were washed one time with PBS and then fixed in 4% paraformaldehyde for 1 h at 25 • C followed by one wash with PBS. Then, the cells were permeabilized for 10 min at 4 • C with permeabilizing solution (0.1% Triton X-100 in 0.1% sodium citrate) and then washed twice with PBS. Positive control was treated with 200 U/mL DNase I for 10 min. Then TUNEL staining was performed using the In Situ Cell Death Detection Kit, Fluorescein (Cat# 11684795910, Roche, Mannheim, Germany). In brief, the cells were incubated with 50 µL of TUNEL reaction mix (terminal deoxynucleotidyl transferase + nucleotide mixture) for 1 h at 37 • C. Negative control cells were incubated with 50 µL of the nucleotide mixture devoid of the terminal deoxynucleotidyl transferase enzyme. Thereafter, cells were washed 3 times with PBS, then mounted with a cover slip and examined by fluorescence microscopy. Apoptotic cells were identified by the stain intensity. The percentage of apoptotic cells was assessed by calculating the apoptotic index (A.I.) using the following formula: A.I. = number of intensely stained cells total number of cells × 100.

Statistical Evaluation
All data are presented as mean ± standard error. One-way Anova and t-test were used for statistical analysis (GraphPad Prism). p values equal or less than 0.05 were considered statistically significant.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to our preference in personal interaction with those interested in our work and data and we are open to dialogs with colleagues who identify themselves and show genuine honest interest. We have no problems sharing data in this way.

Acknowledgments:
A.J.L.M. and E.C.d.M. were partially supported by IMET and IEMEST. A special thanks to Kristy Warner who has been of great help with the optimization of UM-HACC-2A cells culture.

Conflicts of Interest:
The authors declare no conflict of interest.