The C-terminal HSP90 inhibitor NCT-58 kills trastuzumab-resistant breast cancer stem-like cells

N-terminal HSP90 inhibitors in development have had issues arising from heat shock response (HSR) induction and off-target effects. We sought to investigate the capacity of NCT-58, a rationally-synthesized C-terminal HSP90 inhibitor, to kill trastuzumab-resistant HER2-positive breast cancer stem-like cells. NCT-58 does not induce the HSR due to its targeting of the C-terminal region and elicits anti-tumor activity via the simultaneous downregulation of HER family members as well as inhibition of Akt phosphorylation. NCT-58 kills the rapidly proliferating bulk tumor cells as well as the breast cancer stem-like population, coinciding with significant reductions in stem/progenitor markers and pluripotent transcription factors. NCT-58 treatment suppressed growth and angiogenesis in a trastuzumab-resistant xenograft model, concomitant with downregulation of ICD-HER2 and HSF-1/HSP70/HSP90. These findings warrant further investigation of NCT-58 to address trastuzumab resistance in heterogeneous HER2-positive cancers.


INTRODUCTION
Heat shock protein 90 (HSP90) is a chaperone that governs the maturation, stabilization, and activation of client proteins [1,2]. HSP90 also interacts with a variety of pathways that can become oncogenic, and its aberrant activity is implicated in tumorigenesis and cancer progression [2][3][4]. For example, HSP90 has been shown to regulate the tyrosine kinase activity of human epidermal growth factor receptor 2 (HER2), facilitating pro-survival signaling [5][6][7].
Trastuzumab is a humanized anti-HER2 monoclonal antibody used for the treatment of HER2-positive breast cancer, however, most patients eventually develop resistance to the drug within 1-2 years [8][9][10]. Key mechanisms responsible for trastuzumab resistance include HER2/HER3 and HER2/EGFR interactions, hyperactivation of PI3K/Akt signaling, and accumulation of truncated forms of HER2 [10,11]. Truncated p95HER2 is a constitutively active form of the tyrosine kinase that activates downstream signaling through dimerization with other HER family members [11][12][13]. These trastuzumab resistance-related factors are HSP90 client proteins [5][6][7]14], and therefore the inhibition of HSP90 may suppress several potent oncogenic drivers and trastuzumab-refractory factors.
HER2-positive tumors are typically highly heterogeneous consisting of differentiated tumor bulk cells and a smaller subset of breast cancer stem cells (BCSCs) with tumorigenic potential and asymmetric cell division capability [15,16]. A positive association has been reported between CSCs and trastuzumab resistance in HER2-positive cancer [17][18][19]. Expression of the stem/progenitor cell marker ALDH1 is highly elevated in HER2-positive breast cancer and is associated with an aggressive phenotype [20,21]. Subpopulations with CD44 high /CD24 low mesenchymal stem-like phenotypes are often resistant to trastuzumab [18,22,23]. Therefore, new therapeutic strategies that effectively target both cancer stem cells and trastuzumab resistance are needed to improve clinical outcomes. In addition, recent studies have shown that HSP90 confers stability to the pluripotent transcription factors Oct4 and Nanog by preventing degradation via the ubiquitinproteasome pathway, further highlighting the potential advantages of HSP90 inhibition with regard to attenuating pluripotency and self-renewal capacity [24,25].
Although HSP90 is a promising target for cancer treatment, there are no approved inhibitors due to issues including heat shock response (HSR) induction and off-target effects [26][27][28]. Inhibition of HSP90 by binding to its N-terminal domain also triggers heat shock factor-1 (HSF-1)-mediated HSP transcription, leading to upregulation of HSP27, HSP40, HSP70, and HSP90 [2,3,27]. Collectively, this series of reactions is called the HSR and is an intracellular defense mechanism that promotes survival of malignant cells.
We developed the novel C-terminal HSP90 inhibitor NCT-58 to address the shortcomings of N-terminal HSP90 inhibitors. NCT-58 (compound 80) is one of 90 synthesized O-substituted analogues of the B-and C-ring truncated scaffold of deguelin, which is a naturally occurring C-terminal inhibitor (Fig. 1A). NCT-58 specifically binds to the C-terminal domain of HSP90 and was identified through a molecular docking study [29]. We sought to further evaluate its efficacy against cancer stem cells in trastuzumab-resistant HER2-positive breast cancer.

RESULTS
NCT-58-dependent apoptosis is mediated by caspase activation in HER2-positive breast cancer cells We first sought to evaluate the effect of NCT-58 on cell viability and apoptosis in HER2-positive breast cancer cells. NCT  for 72 h) as seen through phase-contrast microscopy. D Cells were treated with NCT-58 (2-10 μM) for 72 h and the sub-G1 population was assessed by flow cytometry (***p < 0.01, n = 3). E Early and late apoptotic cells in the presence or absence of NCT-58 were quantified by annexin V/PI staining (right panel, **p < 0.01; ***p < 0.001, n = 3). F Effect of NCT-58 on expression of cleaved-caspase-3, cleaved-caspase-7, cleaved-PARP and survivin in BT474 and SKBR3 cells. G Quantitative graphs of these protein levels (*p < 0.05; **p < 0.01; ***p < 0.001, n = 3). GAPDH was used as an internal loading control. H The sub-G1 fraction of the normal human mammary epithelial MCF10A cells was analyzed through flow cytometry after exposure to 10 μM of NCT-58 for 72 h (NS, not significant, versus DMSO control, n = 3). The results are presented as mean ± SD of at least three independent experiments and analyzed by one-way ANOVA or Student's t-test followed by Bonferroni's post hoc test.
To investigate whether NCT-58 induces the HSR, the subcellular localization of HSF-1 and the expression levels of HSP70 and HSP90 following NCT-58 treatment were determined by immunocytochemical analysis. SKBR3 cells were treated with geldanamycin (300 nM), a known N-terminal HSP90 inhibitor, and NCT-58 (300 nM and 10 μM) for 24 h and immunostained for HSF-1, HSP70, and HSP90. No increase in HSF-1 was observed following exposure to NCT-58, whereas a marked increase in nuclear accumulation of HSF-1 occurred after geldanamycin treatment (Fig. 2C). NCT-58 did not affect HSP70 or HSP90, while geldanamycin markedly upregulated HSP70 (Fig. 2D) and HSP90 (Fig. 2E). We further confirmed that NCT-58 had no effect on HSP70 and HSP90 expression in HER2-positive breast cancer cells at 24 h, as determined by immunoblotting ( Supplementary Fig.  S2). NCT-58 competitively inhibited binding between HSP90α CTD and its co-chaperone peptidylprolyl isomerase D (PPID) to a greater extent than geldanamycin and novobiocin, a wellcharacterized C-terminal HSP90 inhibitor (Fig. 2F).
The Ras/Raf/mitogen-activated protein kinase pathway (MAPK) signaling cascade is activated by dimerization of HER family members and plays a crucial role in various aspects of breast cancer progression including the cell cycle, proliferation, apoptosis, and angiogenesis [11,30]. Furthermore, Ras, Raf, Mek, and Erk are client proteins of HSP90, and their expression and activation are also controlled by HSP90 [31]. We observed that NCT-58 significantly impaired their expression as well as the phosphorylation of Ras, Raf (Ser338), Mek (Ser217/221) and Erk (Tyr202/204) in JIMT-1 cells ( Fig. 3H and Supplementary Fig. S6).
NCT-58 eradicates HER2-positive BCSCs without triggering the HSR HSF-1 plays a major role in the maintenance of BCSC-like properties [32]. ALDH1-positive or -negative cells were sorted from HER2-positive BT474 and JIMT-1 cell populations, respectively, and the levels of HSR-related factors were examined ( Fig. 4A and Supplementary Fig. S8A). In agreement with previous findings [33], HER2 was preferentially overexpressed in the ALDH1-positive population ( Fig. 4B and Supplementary Fig. S8B). Immunocytochemistry analysis and intensity profiling revealed that HSF-1 is highly elevated and accumulates in the nucleus of ALDH1-positive cells, whereas ALDH1-negative cells exhibit comparatively lower levels of HSF-1 ( Fig. 4C and Supplementary Fig. S8C). Considerable overexpression of HSP70 was found in ALDH1-positive cells, implying that abundant HSF-1 likely enhances the transcription of HSP70 in the BCSC subpopulation ( Fig. 4D and Supplementary  Fig. S8D).
Mammospheres are enriched in mammary stem/progenitor populations that possess self-renewal and differentiation potential, as well as higher ALDH1 activity [21,34]. The mammosphere-forming ability of BT474 and JIMT-1 was diminished in the presence of NCT-58 (Fig. 4E). It is noteworthy that HSF-1 protein content was significantly upregulated in BCSC-enriched mammospheres and this effect was considerably downregulated by NCT-58 treatment. NCT-58 also significantly downregulated HSP70 and HSP90 protein content (Fig. 4F). Levels of the pluripotent transcription factors Nanog, Oct4, and Sox2 as well as ALDH1 were markedly diminished in the presence of NCT-58 (Fig. 4G).

NCT-58 administration suppresses trastuzumab-resistant tumor growth
To confirm the physiological relevance of our in vitro observations, we evaluated the impact of NCT-58 on tumor angiogenesis, expression of BCSC makers and tumor growth in a trastuzumab-resistant xenograft model. JIMT-1 cells (3 × 10 6 ) were orthotopically injected into the right fourth mammary fat pads of BALB/c female nude mice (n = 6, each group). The mice were treated with NCT-58 (30 mg/kg body weight, every other day) or vehicle control (1:9, DMSO:corn oil). NCT-58 administration caused a significant impediment of tumor growth (Fig. 5A) and a marked decrease in tumor weight (Fig. 5B). There were no significant differences in body weight between the groups (Fig. 5C). No histological abnormalities were observed in the lungs, liver, and kidneys after administration of NCT-58 (Fig. 5D). To examine the potential organ toxicity of NCT-58, aspartate aminotransferase (AST), alanine aminotransferase (ALT) and blood urea nitrogen (BUN) assays were performed with serum samples from the animals. No significant changes were found between the control and treatment groups, suggesting that the inhibitor does not overtly affect liver or kidney function (Fig. 5E).
The antitumor effect of NCT-58 was observed concurrently with a marked reduction in Ki-67-positive cells (Fig. 5F) and a significant increase in apoptosis as detected by TUNEL-positivity (Fig. 5G). To further evaluate the effect of NCT-58 on tumor angiogenesis, a microvessel density (MVD) assay was performed using the endothelial-specific marker CD31 [35]. NCT-58 administration significantly reduced the number of CD31-positive vessels in both peritumoral and intratumoral areas (Fig. 5H). Consistent with in vitro observations, NCT-58 administration significantly suppressed the expression of both ICD-HER2 (Fig. 5I) and full-length HER2 (Fig. 5J).
Anti-tumor effect of NCT-58 is accompanied by the suppression of BCSC-like characteristics and downregulation of HSF-1/HSP70/HSP90 The cell surface glycoprotein CD44, as a marker of trastuzumab resistance [11,18], is highly expressed in the plasma membrane of JIMT-1 control tumor cells. Animals receiving NCT-58 exhibited a noticeably lower level of CD44 in vivo (Fig. 6A). Furthermore, ALDH1 expression was also significantly different between the groups (Fig. 6B).
HSF-1 activity was suppressed after NCT-58 treatment, as evidenced by a significant decrease in the signal intensity of nuclear HSF-1 (Fig. 6C) as well as downregulation of its downstream client proteins HSP70 (Fig. 6D) and HSP90 (Fig. 6E).

DISCUSSION
HSP90 orchestrates the activation and stability of more than 200 potential client oncoproteins [1]. However, N-terminal inhibitors of this target have been shown to induce the HSR, representing a considerable obstacle in clinical development [26,27].
The master transcriptional determinant HSF-1 plays a pleiotropic role not only in regulating the HSR, but also during tumorigenesis, tumor cell migration, and metastasis [36][37][38]. Clinical evidence has shown significant associations between HSF-1 and HER2 in HER2-positive breast cancer, while higher levels of nuclear HSF-1 are correlated with histologic grade, larger tumor sizes and reduced survival [39]. HSF-1 in HER2/Neu+ transgenic mice also stimulates tumorigenesis in the mammary glands, and metastasis via the promotion of epithelial to mesenchymal transition (EMT) [38]. Activated HSF-1 translocates to the nucleus to promote the transcription of HSPs during oncogenesis, facilitating rapid tumor cell proliferation [40]. Meanwhile, HSP70 hinders the apoptosis pathway by interfering with the release of cytochrome c, inhibiting apoptosome complex formation and caspase activation [41].
Treatment with NCT-58 did not induce the HSR, as evidenced by the absence of nuclear accumulation of HSF-1 and upregulation of HSP70 expression in HER2-positive breast cancer cells. NCT-58 significantly increased apoptosis via caspase-3/caspase-7 activation, but elicited no such toxicity in non-malignant cells. Our in vivo findings show that NCT-58 administration suppresses tumor growth, concomitant with increased apoptosis and simultaneous downregulation of HSF-1/HSP70/HSP90. Meanwhile, no histological changes were seen in liver and kidney tissue sections.
Evidence suggests that trastuzumab resistance can be attributable to the existence of a subpopulation of BCSCs in HER2positive breast cancers [42,43]. The CD44/hyaluronan complex masks the cognate epitope, interfering with trastuzumab binding to HER2, and leading to PI3K/Akt activation and trastuzumab resistance [11,18]. We have previously found that trastuzumabsensitive cell lines harbor limited numbers of CD44 high /CD24 low cells at less than 1% [44], whereas trastuzumab-resistant JIMT-1 cells harbor considerably higher numbers at more than 50%, suggesting that the CD44 high /CD24 low phenotype is a marker of trastuzumab resistance. Trastuzumab-resistant JIMT-1 cells when treated with NCT-58 exhibited a noticeable reduction in the CD44 high /CD24 low population and ALDH1 activity, as well as suppression of mammosphere formation. HSP90 physically interacts with Nanog and Oct4, protecting them from ubiquitinmediated proteasomal degradation, implying that HSP90 may contribute to BCSC self-renewal [24,45,46]. Elevated levels of Nanog, Oct4, and SOX2 in BCSC-enriched mammospheres were overwhelmingly abolished after NCT-58 challenge.
Emerging evidence suggests that HSF-1 plays a pivotal role in regulating the BCSC phenotype [32,45]. The BCSC-enriched population exhibited relatively higher levels of HSF-1 in both mammospheres and ADLH1-positive cells. Highly abundant HSF-1 is believed to enhance the expression of HSPs such as HSP70 and HSP90 via HSF-1-mediated transcription, which can promote BCSC survival. In agreement with our observations, forced overexpression of HSF-1 increased mammosphere-forming ability as well as CD44, Sox2 and ALDH1 expression, and conferred drug resistance in breast cancer, while HSF-1 knockdown attenuated these phenomena [32,47]. Further defining the role of HSF-1 in regulating the BCSC phenotype will likely provide important clues into developing effective CSC-targeted therapeutic strategies.
The emergence of acquired trastuzumab resistance remains an urgent unmet medical need. Oncogenic p95HER2 retains tyrosine kinase activity and interacts with HER3, enhancing activation of Akt signaling and tumor cell survival [11,12,48]. Akt directly interacts with and phosphorylates HSF-1 at S326, which confers HSF-1 activation, leading to EMT via Slug upregulation [38,49,50]. Importantly, NCT-58 downregulates the levels of truncated p95HER2 and Akt levels and phosphorylation in trastuzumabresistant HER2-positive cells, an effect also seen in p95HER2overexpressing MDA-MB-231 cells. It is conceivable that Akt downregulation by NCT-58 may attenuate the transcriptional ability of HSF-1, resulting in downregulation of HSP70/HSP90.
In conclusion, NCT-58 does not induce the HSR as targeting of the C-terminal region is not accompanied by HSF-1-mediated HSP70 and HSP90 upregulation that occurs when targeting the N-terminus [27]. NCT-58 suppresses HER2-positive breast cancer cells with simultaneous inhibition of important trastuzumab resistance factors including HER2/HER3/Akt and truncated p95HER2 as well as downregulation of Ras/Raf/Mek/Erk. In Fig. 7, we present a hypothetical model illustrating several key actions of NCT-58 on cancer stem-like properties, HER2 signaling, and trastuzumab resistance in HER2-positive breast cancer. Our findings support rationale for further investigation of NCT-58 as a new therapeutic approach for trastuzumab-resistant HER2positive breast cancers.

Cell cycle analysis and Annexin V/PI assay
Cells were harvested and fixed with 95% ethanol containing 0.5% Tween-20 for 24 h, and incubated with propidium iodide (PI, 50 mg/mL) and RNase (50 mg/mL) for 30 min. For the Annexin V/PI assay, a FITCconjugated Annexin V apoptosis detection kit (BD Biosciences, Franklin Lakes, NJ) was used according to the manufacturer's protocol. Stained cells were analyzed by flow cytometry using BD LSRFortessa TM X-20 Cell Analyzer (BD Biosciences).

In vivo xenograft and mammosphere formation assays
All animal procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals, approved by the Korea University Institutional Animal Care and Use Committee (IACUC, KOREA-2018-0135). Five-week-old female BALB/c nude mice were obtained from the Shizuoka Laboratory Animal Center (Shizuoka, Japan) and housed in a specific pathogen-free environment. JIMT-1 cells (3 × 10 6 ) were implanted into the right fourth mammary fat pads of 6-week-old BALB/c nude female mice. When average tumor volumes reached 100 mm 3 , the animals were randomized into 2 groups (n = 6/each group), solvent control (DMSO/Corn oil, 1:9) or NCT-58 (30 mg/kg BW) was administered intraperitoneally every other day for 47 days. Tumor Fig. 5 NCT-58 inhibits tumor growth in trastuzumab-resistant JIMT-1 xenografts. A-C Effect of NCT-58 on tumor growth in vivo. JIMT-1 cells (3 × 10 6 ) were injected into the mammary fat pads of BALB/c nude mice. Mice were administered intraperitoneally with NCT-58 (30 mg/ kg·BW, every other day) or solvent control for 47 days (n = 6/each group). Tumor volumes were measured with a caliper at the intervals indicated. A, B NCT-58 administration resulted in significant decreases in tumor growth (A, ***p < 0.001, n = 6) and tumor weight (B, *p < 0.05, n = 6). C Changes in body weight of the xenografted mice after exposure to NCT-58 or control vehicle (NS; not significant, n = 6). D Representative histological analysis of lung, liver, and kidney sections stained with hematoxylin and eosin (H&E) and analyzed by microscope slide scanner. E Effects of NCT-58 on serum biochemical parameters of liver and kidney function. Blood biochemical analysis indicated there was no significant change in serum ALT, AST, BUN (NS; not significant). ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, blood urea nitrogen. F Influence of NCT-58 on Ki-67 expression. Tumor tissue sections were immunostained for Ki-67 (red) and DAPI (blue). The graph shows the percentage of Ki-67-positive cells (***p < 0.001, n = 6). G NCT-58-induced apoptosis was determined by TUNEL assay. Extent of apoptosis expressed as the percentage of total TUNEL-positive cells (***p < 0.001, n = 6). H Effect of NCT-58 on tumor angiogenesis, as determined by a microvessel density (MVD) assay. Tumor tissues were immunostained with a specific endothelial marker CD31 (red) and DAPI (blue). The number of CD31-positive microvessels in the intratumoral and peritumoral areas were quantified, respectively (**p < 0.01; ***p < 0.001, n = 6). I, J Immunohistochemical analysis for the intracellular domain (ICD)-HER2 (green, I) and full-length HER2 (green, J) in vivo. Quantitative graphs of signal intensities shown in the right panels (***p < 0.001, n = 6). The results are presented as mean ± SD and data were analyzed by unpaired Student's t-test and two-way ANOVA followed by Bonferroni's post hoc test. Fig. 6 NCT-58 suppresses BCSC-like properties and downregulates the expression of HSF-1, HSP70, and HSP90 in vivo. A, B Following exposure to NCT-58, CD44 and ALDH1 expression levels were markedly downregulated in JIMT-1 xenograft tumors. The fluorescence intensities of CD44 (A, ***p < 0.001, n = 6) and ALDH1 (B, ***p < 0.001, n = 6) were analyzed using a histogram tool. C-E Effect of NCT-58 on induction of the HSR in vivo. C NCT-58 administration resulted in a decrease in nuclear expression of HSF-1 (green). Fluorescence intensity of HSF-1 localized in the nuclei is represented in arbitrary units as defined by the software using the intensity profile tool. D, E NCT-58 downregulated both HSP70 and HSP90 expression. Quantitative graphs of fluorescence intensities of HSP70 (D, **p < 0.01, n = 6) and HSP90 (E, ***p < 0.001, n = 6) signal, shown in the bottom panels, respectively. Tumors were harvested within 24 h of the last administration of NCT-58 for immunohistochemistry assessment. Data were analyzed by unpaired Student's t-test. volumes were measured twice weekly after the initial treatment and calculated using the formula V = (Length × Width 2 )/2.
For the in vivo mammosphere-forming assay, tumors were harvested when volumes reached 200-250 mm 3 and dissociated with type III collagenase (2 mg/mL). The digested tissues were filtered through a 40 mm cell strainer, centrifuged at 200 × g for 5 min and washed with medium containing 0.2% bovine serum albumin (BSA). Dissociated single cells (5 × 10 4 ) were plated in ultralow attachment dishes and cultured in the presence or absence of NCT-58 (5 and 10 µM) for 8 days.

Serum biochemistry profiles for biomarkers of liver and renal injury
At sacrifice, blood samples from each animal were collected, and serum enzyme activities of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and blood urea nitrogen (BUN) levels were determined with an assay kit following the manufacturer's protocol (Sigma-Aldrich).

Statistical analysis
All data were analyzed using GraphPad Prism 5.0 statistical software (San Diego, CA). The results are presented as mean ± SD of at least three independent experiments. Data were analyzed by student's t-test, and one-or two-way ANOVA as appropriate. Significance between multiple experimental groups was determined using the Bonferroni post hoc test and defined at p < 0.05. Fig. 7 Hypothetical model illustrating multiple actions of NCT-58 on cancer stem-like properties, HER2 signaling, and trastuzumab resistance in HER2-positive breast cancer. i NCT-58 targets the C-terminal domain of HSP90 independent of the HSR and effectively induces apoptosis via the activation of cleaved caspase-3/7. ii NCT-58 downregulates the expression and phosphorylation of HER2, HER3, and EGFR as well as suppresses truncated p95HER2 accumulation and Akt activation, concomitant with downregulation of the Ras/Raf/Mek/Erk signaling pathway. iii NCT-58 kills not only proliferating tumor cells, but also eliminates BCSC-like cells. The latter responses are accompanied by the reduction of stem/progenitor markers CD44/ALDH1 and expression of the pluripotent transcription factors Nanog/Oct4/Sox2 as well as downregulation of HSF-1/HSF70/HSP90. [Heat shock response, HSR; Heat shock factor-1, HSF-1, Heat shock element, HSE].