PNSA, a Novel C-Terminal Inhibitor of HSP90, Reverses Epithelial–Mesenchymal Transition and Suppresses Metastasis of Breast Cancer Cells In Vitro

Metastasis accounts for the vast majority of deaths in breast cancer, and novel and effective treatments to inhibit cancer metastasis remain urgently developed. The expression level of heat shock protein 90 (HSP90) in invasive breast cancer tissue is higher than in adjacent non-cancerous tissue. In the present study, we investigated the inhibitory effect of penisuloxazin A (PNSA), a novel C- terminal inhibitor of HSP90, on metastasis of breast cancer cells and related mechanism in vitro. We found that PNSA obviously affected adhesion, migration, and invasion of triple-negative breast cancer (TNBC) MDA-MB-231 cells and Trastuzumab-resistant JIMT-1 cells. Furthermore, PNSA was capable of reversing epithelial–mesenchymal transformation (EMT) of MDA-MB-231 cells with change of cell morphology. PNSA increases E-cadherin expression followed by decreasing amounts of N-cadherin, vimentin, and matrix metalloproteinases9 (MMP9) and proteolytic activity of matrix metalloproteinases2 (MMP2) and MMP9. Comparatively, the N-terminal inhibitor of HSP90 17-allyl-17-demethoxygeldanamycin (17-AAG) had no effect on EMT of MDA-MB-231 cells. PNSA was uncovered to reduce the stability of epidermal growth factor receptor (EGFR) and fibroblast growth factor receptor (FGFR) proteins and thereby inhibiting their downstream signaling transductions by inhibition of HSP90. In addition, PNSA reduced the expression of programmed cell death-ligand 1 (PD-L1) to promote natural killer (NK) cells to kill breast cancer cells with a dose far less than that of cytotoxicity to NK cell itself, implying the potential of PNSA to enhance immune surveillance against metastasis in vivo. All these results indicate that PNSA is a promising anti-metastasis agent worthy of being studied in the future.


Introduction
Breast cancer is the most malignant and major cause of cancer-related deaths among women worldwide [1], which is classified into hormone-receptor-positive, human epidermal growth factor receptor-2 overexpressing (HER2+) and triple-negative breast cancer (TNBC) based on histological features. Therapy strategies vary according to the classification, including endocrine modulators, agents that target estrogen receptor (ER), and human epidermal growth factor receptor 2 (HER2) such as tamoxifen and trastuzumab. Much progress has been made in early detection and better treatment of breast cancer, leading to improved survival. However, a considerable number of patients will relapse as a result of organ metastasis, specifically lung, liver, bone, and brain [2]. It is reported that 20-30% of breast cancer patients may develop metastases after diagnosis and primary tumor treatment, and approximately 90% of cancer-related deaths are attributed to metastasis [3,4]. TNBC accounts for about 16% of all breast cancers, and is more aggressive than mor treatment, and approximately 90% of cancer-related deaths are attributed to metas tasis [3,4]. TNBC accounts for about 16% of all breast cancers, and is more aggressive than other breast cancer, and has a higher three-year recurrence rate and five-year mortalit rate after treatment [5,6]. Due to the lack of estrogen receptor (ER), progesterone recepto (PR), and human epidermal growth factor receptor 2 (HER2), the conventional treatmen used for other breast cancers is not effective in TNBC. In addition, the high heterogeneity of TNBC on the molecular level, pathology, and clinical characteristics also lead it to b hard to deal with [7]. HER2-positive tumors reach about 20-30% of breast cancers [8]. Th application of HER2-directed antibodies, such as trastuzumab, has been suggested a standard therapy for HER2 positive advanced breast cancer. However, the resistance to anti-HER2 antibody has resulted in antibody drug-refractory metastatic and disease pro gression [9,10]. Therefore, novel and effective treatments to inhibit the metastasis of breas cancer cells remain urgently developed.
Epithelial-mesenchymal transition (EMT) is a process in which epithelial cells ac quire mesenchymal features. In cancer, EMT is associated with tumor initiation, invasion metastasis, and resistance to therapy. During EMT, epithelial cells undergo multiple mor phologic, biochemical, and genetic rearranges that gradually enable them to acquire mesenchymal phenotype [11,12]. The hallmark of EMT is the upregulation of N-cadheri followed by the downregulation of E-cadherin, and this process is regulated by a comple network of signaling pathways and transcription factors [13,14]. Reversal of EMT has been suggested as a valuable strategy for therapeutic intervention of metastatic cancer [15].
Heat shock protein 90 (HSP90) is a ubiquitous molecular chaperone protein and play an important role in multiple biological functions and processes such as cell survival, pro liferation, tumor progression, and metastasis by modulating the stability, maturation, and conformational changes of various proteins [16][17][18]. The expression level of HSP90 in in vasive breast cancer tissue is higher than in adjacent non-cancerous tissue [19]. The ele vated expression of HSP90 protein had a significantly positive connection with metastasis advanced stage, and poor survival in patients [20]. In previous studies, we have reported that compound penisuloxazin A (PNSA), a new epipolythiodiketopiperazines (ETPs from the mangrove endophytic fungus Penicillium janthinellum, is a novel C-termina HSP90 inhibitor with the binding site at cysteine residues C572/C598 of HSP90 via disul fide bonds [21,22] (Figure 1), which is different from currently reported inhibitors tha bind directly to HSP90 at ATP-binding pocket. The purity of PNSA was determined to b 98% ( Figure S1) and its structure was verified by 1H NMR spectroscopy ( Figure S2) [22] Herein, we probe into the potential inhibitory effects of PNSA on EMTs and metastasis o triple-negative breast cancer cells and trastuzumab-refractory cells in vitro. Our result indicate that the novel inhibitor of HSP90 has the ability to reverse EMT and block cel migration and invasion.

PNSA Impairs the Migration and Invasion of Breast Cancer Cells In Vitro
To evaluate the antimetastatic activity of PNSA, we first assessed the inhibitory effects of PNSA on the migration and invasion of MDA-MB-231 cells, triple-negative breast cancer cells, by the wound healing assay and the transwell assay. The results showed that PNSA obviously inhibited the two processes of cancer metastasis in a dose-dependent manner by PNSA. The Wound healing assay revealed that PNSA significantly decreased the migrations of MDA-MB-231 cells with the migration rates of 13.51% and 9.49% at 1, and 2 µM of PNSA, respectively (Figure 2A). Similarly, transwell assay results showed PNSA meaningfully inhibited invasive capabilities of MDA-MB-231 cells with inhibitory rates of 50.62% and 24.88% at indicated concentrations of PNSA ( Figure 2C). 17-allyl-17demethoxygeldanamycin (17-AAG) is a well-known N-terminal inhibitor of HSP90, we found that 17-AAG suppressed the migration and invasion of MDA-MB-231 cells dosedependently. The migration rates were significantly decreased to 8.21% and 4.56%, and the invasion rates were 30.8% and 16.40% at 4, 8 µM of PNSA ( Figure 2B,D). Within the range of dose in the experiments, PNSA and 17-AAG did not exhibit obviously cytotoxic activity as detected by sulforhodamine B (SRB) assay, which ruled out the influence of cell viability on cellular motility. (Figure 2E,F). Collectively, these results demonstrate that PNSA and 17-AAG all impair the migration and invasion of breast cancer cells.

PNSA Impairs the Migration and Invasion of Breast Cancer Cells In Vitro
To evaluate the antimetastatic activity of PNSA, we first assessed the inhibitory effects of PNSA on the migration and invasion of MDA-MB-231 cells, triple-negative breast cancer cells, by the wound healing assay and the transwell assay. The results showed that PNSA obviously inhibited the two processes of cancer metastasis in a dose-dependent manner by PNSA. The Wound healing assay revealed that PNSA significantly decreased the migrations of MDA-MB-231 cells with the migration rates of 13.51% and 9.49% at 1, and 2 μM of PNSA, respectively (Figure 2A). Similarly, transwell assay results showed PNSA meaningfully inhibited invasive capabilities of MDA-MB-231 cells with inhibitory rates of 50.62% and 24.88% at indicated concentrations of PNSA ( Figure 2C). 17-allyl-17demethoxygeldanamycin (17-AAG) is a well-known N-terminal inhibitor of HSP90, we found that 17-AAG suppressed the migration and invasion of MDA-MB-231 cells dosedependently. The migration rates were significantly decreased to 8.21% and 4.56%, and the invasion rates were 30.8% and 16.40% at 4, 8 μM of PNSA ( Figure 2B,D). Within the range of dose in the experiments, PNSA and 17-AAG did not exhibit obviously cytotoxic activity as detected by sulforhodamine B (SRB) assay, which ruled out the influence of cell viability on cellular motility. (Figure 2E,F). Collectively, these results demonstrate that PNSA and 17-AAG all impair the migration and invasion of breast cancer cells. The bar graph represents the average ± SD of at least three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001; ns, not significant (relative to DMSO-treated cells).

PNSA Reverses EMT of MDA-MB-231 Cells
Previous studies have demonstrated that during the EMT process of cancer, intercellular interaction and the adherence of cancer cells to ECM components play important roles in the process of tumor metastasis [23]. As shown in Figure 3A, cellular polymers became larger with increasing concentrations of PNSA, the percentages of aggregated cells were significant compared to DMSO control at 28.80% and 45.93% with 1 and 2 μM of PNSA, indicating that PNSA promotes spontaneous cell aggregation ( Figure 3A). In contrast, we found that PNSA inhibited attachment of MDA-MB-231 cells to coated Matrigels including many ECM components ( Figure 3B). In addition, after PNSA treatment, the cell morphology gradually changed from spindle-shaped or polygonal mesenchymal morphology to flat polygonal epithelial-like cell morphology ( Figure 3C). These results suggest that the PNSA may reverse the EMT process of MDA-MB-231 cells. N-cadherin is upregulated while E-cadherin is downregulated during an EMT in cancers, and this "cadherin switch" is associated with enhanced migratory and invasive traits [13]. Besides, EMT-related proteins vimentin and MMP9 were up-regulated, and the proteolytic activity of MMP was enhanced in tumor metastasis [24]. Meanwhile, C-Myc, a transcription factor, is also increased, strongly contributing to invasion and migration [25]. As shown in Figure  3D, PNSA induced a significant increase in E-cadherin expression and decreases in Ncadherin, vimentin, C-Myc, and MMP-9 expressions compared to DMSO group. Furthermore, the proteolytic activities of MMP-2 and MMP-9 in decomposing extracellular matrix were investigated after treatment of PNSA by gelatin zymography assay. As shown in Figure 3E, PNSA efficiently inhibited enzyme activities of MMP-2 and MMP-9 in MDA-MB-231 cells in a dose-dependent manner. For comparison, 17-AAG was revealed to have no influence on MDA-MB-231 morphology and E-cadherin protein level, although inhibiting N-cadherin expression ( Figure 3C,D). These results indicate that PNSA with the new binding site on HSP90 has the ability to reverse EMT of MDA-MB-231 cells. and cell viability was measured using sulforhodamine B (SRB) assay. The bar graph represents the average ± SD of at least three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001; ns, not significant (relative to DMSO-treated cells).

PNSA Reverses EMT of MDA-MB-231 Cells
Previous studies have demonstrated that during the EMT process of cancer, intercellular interaction and the adherence of cancer cells to ECM components play important roles in the process of tumor metastasis [23]. As shown in Figure 3A, cellular polymers became larger with increasing concentrations of PNSA, the percentages of aggregated cells were significant compared to DMSO control at 28.80% and 45.93% with 1 and 2 µM of PNSA, indicating that PNSA promotes spontaneous cell aggregation ( Figure 3A). In contrast, we found that PNSA inhibited attachment of MDA-MB-231 cells to coated Matrigels including many ECM components ( Figure 3B). In addition, after PNSA treatment, the cell morphology gradually changed from spindle-shaped or polygonal mesenchymal morphology to flat polygonal epithelial-like cell morphology ( Figure 3C). These results suggest that the PNSA may reverse the EMT process of MDA-MB-231 cells. N-cadherin is upregulated while E-cadherin is downregulated during an EMT in cancers, and this "cadherin switch" is associated with enhanced migratory and invasive traits [13]. Besides, EMT-related proteins vimentin and MMP9 were up-regulated, and the proteolytic activity of MMP was enhanced in tumor metastasis [24]. Meanwhile, C-Myc, a transcription factor, is also increased, strongly contributing to invasion and migration [25]. As shown in Figure 3D, PNSA induced a significant increase in E-cadherin expression and decreases in N-cadherin, vimentin, C-Myc, and MMP-9 expressions compared to DMSO group. Furthermore, the proteolytic activities of MMP-2 and MMP-9 in decomposing extracellular matrix were investigated after treatment of PNSA by gelatin zymography assay. As shown in Figure 3E, PNSA efficiently inhibited enzyme activities of MMP-2 and MMP-9 in MDA-MB-231 cells in a dose-dependent manner. For comparison, 17-AAG was revealed to have no influence on MDA-MB-231 morphology and E-cadherin protein level, although inhibiting N-cadherin expression ( Figure 3C,D). These results indicate that PNSA with the new binding site on HSP90 has the ability to reverse EMT of MDA-MB-231 cells.  The bar graph represents the average ± SD of at least three independent experiments. * p < 0.05; ** p < 0.01; ns, not significant (relative to DMSO-treated cells).

PNSA Inhibits Expressions and Activations of Signaling Molecules Related to EMT
It has been reported that multiple signaling pathways play a crucial role in the process of tumor metastasis such as PI3K/Akt, MAPK/ERK, JAK/STAT, and Wnt/β-catenin [26][27][28][29]. Our previous study showed that PNSA inhibited the metastasis of breast cancer at 12 h. In order to further explore the mechanism of PNSA underlying metastasis of MDA-MB-231 cells, we tested the effects of PNSA on several signaling pathways through Western blotting experiments at different concentrations and times. We found that PNSA significantly reduced active phosphorylated forms of ERK, Stat3, and Akt without changes in the total levels of these proteins. PNSA also downregulated the amounts of β- and MMP-2 were measured by gelatin zymography assay. The bar graph represents the average ± SD of at least three independent experiments. * p < 0.05; ** p < 0.01; ns, not significant (relative to DMSO-treated cells).

PNSA Inhibits Expressions and Activations of Signaling Molecules Related to EMT
It has been reported that multiple signaling pathways play a crucial role in the process of tumor metastasis such as PI3K/Akt, MAPK/ERK, JAK/STAT, and Wnt/β-catenin [26][27][28][29].
Our previous study showed that PNSA inhibited the metastasis of breast cancer at 12 h. In order to further explore the mechanism of PNSA underlying metastasis of MDA-MB-231 cells, we tested the effects of PNSA on several signaling pathways through Western blotting experiments at different concentrations and times. We found that PNSA signifi- cantly reduced active phosphorylated forms of ERK, Stat3, and Akt without changes in the total levels of these proteins. PNSA also downregulated the amounts of β-catenin and upstreaming membrane receptors in these signaling pathway epidermal growth factor receptor (EGFR), fibroblast growth factor receptor (FGFR)1 in MDA-MB-231 after 12 h treatment ( Figure 4A). At 24 h, PNSA had more obvious effects on the expression levels or activations of these proteins ( Figure 4B).

Mar. Drugs 2021, 19, x FOR PEER REVIEW
6 of 17 catenin and upstreaming membrane receptors in these signaling pathway epidermal growth factor receptor (EGFR), fibroblast growth factor receptor (FGFR)1 in MDA-MB-231 after 12 h treatment ( Figure 4A). At 24 h, PNSA had more obvious effects on the expression levels or activations of these proteins ( Figure 4B).

PNSA Inhibits the Protein Expressions and Activations in the Signaling Pathway by Targeting HSP90
In order to explore whether the declines in the protein expressions and activations as mentioned earlier were related to HSP90 inhibition, we further examined the influences of HSP90 on these proteins of MDA-MB-231 cells. MDA-MB-231 cells were treated with different fragments for siRNA-mediated depletion of HSP90 and related protein levels were determined by Western blotting. As shown in Figure 5A, Si-HSP90 was found to decrease protein levels of FGFR1, EGFR, C-Myc, and phosphorylation levels of ERK and AKT, but the amounts of N-Cadherin, ERK, AKT, and Stat3 were not changed in MDA-MB-231 cells. HSP90 inhibition induces degradation of its client proteins through the proteasome pathway [30]. We further investigated whether the down-regulation of these proteins were the result of proteasome degradation after PNSA treatment. EGFR and C-Myc were selected to be observed after MDA-MB-231 cells treated with the proteasome inhibitor MG132 before the addition of PNSA. We found that PNSA-induced reductions of EGFR and C-Myc were rescued by MG132 ( Figure 5B). Furthermore, we found that PNSA treatment strongly protected HSP90 protein from destabilization at the indicated temperatures in MDA-MB-231 cells compared to DMSO control detected by cellular thermal shift assay (CETSA) method ( Figure 5C). CETSA method directly detects the interaction between the drug molecule and target protein in cells, as the binding of the molecule can increase the thermal stability of protein [31]. Altogether, these results indicate that PNSAinduced down-regulations of these signaling molecules related to EMT are associated

PNSA Inhibits the Protein Expressions and Activations in the Signaling Pathway by Targeting HSP90
In order to explore whether the declines in the protein expressions and activations as mentioned earlier were related to HSP90 inhibition, we further examined the influences of HSP90 on these proteins of MDA-MB-231 cells. MDA-MB-231 cells were treated with different fragments for siRNA-mediated depletion of HSP90 and related protein levels were determined by Western blotting. As shown in Figure 5A, Si-HSP90 was found to decrease protein levels of FGFR1, EGFR, C-Myc, and phosphorylation levels of ERK and AKT, but the amounts of N-Cadherin, ERK, AKT, and Stat3 were not changed in MDA-MB-231 cells. HSP90 inhibition induces degradation of its client proteins through the proteasome pathway [30]. We further investigated whether the down-regulation of these proteins were the result of proteasome degradation after PNSA treatment. EGFR and C-Myc were selected to be observed after MDA-MB-231 cells treated with the proteasome inhibitor MG132 before the addition of PNSA. We found that PNSA-induced reductions of EGFR and C-Myc were rescued by MG132 ( Figure 5B). Furthermore, we found that PNSA treatment strongly protected HSP90 protein from destabilization at the indicated temperatures in MDA-MB-231 cells compared to DMSO control detected by cellular thermal shift assay (CETSA) method ( Figure 5C). CETSA method directly detects the interaction between the drug molecule and target protein in cells, as the binding of the molecule can increase the thermal stability of protein [31]. Altogether, these results indicate that PNSA-induced down-regulations of these signaling molecules related to EMT are associated with protein degradations due to inhibition of HSP90. By the way, we noticed that HSP70 expression levels were not obviously affected by the treatment with PNSA. However, 17-AAG could induce an unfavorable increase of cytoprotective HSP70 expression ( Figure 5D). Up-regulation of HSP70 was a striking disadvantage induced by HSP90 N-terminal inhibitors [32,33].
Mar. Drugs 2021, 19, x FOR PEER REVIEW 7 of 17 with protein degradations due to inhibition of HSP90. By the way, we noticed that HSP70 expression levels were not obviously affected by the treatment with PNSA. However, 17-AAG could induce an unfavorable increase of cytoprotective HSP70 expression ( Figure  5D). Up-regulation of HSP70 was a striking disadvantage induced by HSP90 N-terminal inhibitors [32,33].

PNSA Inhibits the Migration, Invasion, and EMT-related Signaling Pathway Proteins in JIMT-1 Cells
Previous studies have shown that PNSA inhibits the metastasis of MDA-MB-231 cells which lack drug target receptors such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (Her2) [5]. To further explore the effect of compound PNSA on metastasis ability of other breast cancer cells, JIMT-1 cells, a trastuzumab-resistant form of breast cancer, were chosen to study further. As shown in Figure 6A,B, we found that PNSA suppressed the migration and invasion of JIMT-1 cells dose-dependently. PNSA at 0.5, 1, and 2 µM significantly decreased the migrations of JIMT-1 cells and the migration rates were 15.33%, 8.94%, and 6.60%, respectively ( Figure 6A). Similarly, the transwell assay showed that PNSA meaningfully inhibited invasive capabilities of JIMT-1 cells with inhibitory rates at 56.75%, 40.08%, and 38.10% with indicated concentrations of PNSA ( Figure 6B). No cell viability was affected by PNSA ( Figure 6C). Next, we detected the influences of PNSA on the expressions and activations of several signaling proteins by Western blotting and the results showed that PNSA decreased active phosphorylated forms of ERK, Stat3, and Akt without changes in the total protein levels of these molecules, and downregulated the amounts of EGFR and FGFR1 at 12 h ( Figure 6D). These results suggest that PNSA also inhibits metastasis of trastuzumab-resistant breast cancer cells in vitro.

PNSA Inhibits the Migration, Invasion, and EMT-related Signaling Pathway Proteins in JIMT-1 Cells
Previous studies have shown that PNSA inhibits the metastasis of MDA-MB-231 cells which lack drug target receptors such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (Her2) [5]. To further explore the effect of compound PNSA on metastasis ability of other breast cancer cells, JIMT-1 cells, a trastuzumab-resistant form of breast cancer, were chosen to study further. As shown in Figure 6A,B, we found that PNSA suppressed the migration and invasion of JIMT-1 cells dose-dependently. PNSA at 0.5, 1, and 2 μM significantly decreased the migrations of JIMT-1 cells and the migration rates were 15.33%, 8.94%, and 6.60%, respectively ( Figure  6A). Similarly, the transwell assay showed that PNSA meaningfully inhibited invasive capabilities of JIMT-1 cells with inhibitory rates at 56.75%, 40.08%, and 38.10% with indicated concentrations of PNSA ( Figure 6B). No cell viability was affected by PNSA ( Figure  6C). Next, we detected the influences of PNSA on the expressions and activations of several signaling proteins by Western blotting and the results showed that PNSA decreased active phosphorylated forms of ERK, Stat3, and Akt without changes in the total protein levels of these molecules, and downregulated the amounts of EGFR and FGFR1 at 12 h ( Figure 6D). These results suggest that PNSA also inhibits metastasis of trastuzumab-resistant breast cancer cells in vitro.

PNSA Inhibits PD-L1 Protein Expression in MDA-MB-231 and Enhances NK Cells Cytotoxic Activity
Recent studies suggest that EMT processes induce upregulation of PD-L1 expression [34]. Interestingly, other reports also have confirmed that PD-L1 signaling plays an important role in maintaining the EMT state of various cancers including breast cancer [35,36]. Therefore, we further tested the effect of PNSA on PD-L1 protein level by Western blotting. As shown in Figure 7A, PNSA obviously downregulated the expression of PD-L1 protein in a dose-dependent manner. However, with immunoprecipitation analysis, PD-L1 was not found to combine with HSP90, which suggests that decrease of PD-L1 is not involving in HSP90 inhibition by PNSA and may be related to the reversal of EMT after PNSA treatment. (Figure 7B).
Natural killer (NK) cells play a critical role in host immune responses against tumor growth and metastasis [37]. PD-L1/PD-1 is a pair of immune checkpoints, the reduction of PD-L1 protein expression can enhance the ability of NK cells to attack tumor cells by releasing more perforin and granzyme B 37 . We first digested the MDA-MB-231 cells, replanted them into the plates, and then added PNSA and NK92 cells for 5h to observe whether PNSA increased the killing ability of NK cells. Our results showed that NK92 cells exhibited enhanced-cytotoxicity activity against cancer cells after PNSA treatment compared with DMSO control at effector-to-target ratio T:E of 5:1 ( Figure 7C). Furthermore, the releases of granzyme B and perforin were increased in PNSA-treated MDA-MB-231 cells ( Figure 7D,E). During the time, no change of NK92 cell viability upon PNSA treatment was observed ( Figure 7F). We further examined the effect of PNSA on NK92 cells viability for 72h and found that PNSA was much less cytotoxic to NK92 than to tumor cells with IC 50 value at 12.96 µM, and the IC 50 value of PNSA for MDA-MB-231 cells was 0.86 µM ( Figure 7G). For comparison, the IC 50 value of 17-AAG on NK92 cells for 72h was 9.08µM, meanwhile, the IC 50 value for MDA-MB-231 cells was 3.46 µM ( Figure 7H). These results suggest that PNSA may be better to trigger cancer immunology in vivo to deal with cancer metastasis.

PNSA Inhibits PD-L1 Protein Expression in MDA-MB-231 and Enhances NK Cells Cytotoxic Activity
Recent studies suggest that EMT processes induce upregulation of PD-L1 expression [34]. Interestingly, other reports also have confirmed that PD-L1 signaling plays an important role in maintaining the EMT state of various cancers including breast cancer [35,36]. Therefore, we further tested the effect of PNSA on PD-L1 protein level by Western blotting. As shown in Figure 7A, PNSA obviously downregulated the expression of PD-L1 protein in a dose-dependent manner. However, with immunoprecipitation analysis, PD-L1 was not found to combine with HSP90, which suggests that decrease of PD-L1 is not involving in HSP90 inhibition by PNSA and may be related to the reversal of EMT after PNSA treatment. (Figure 7B).
Natural killer (NK) cells play a critical role in host immune responses against tumor growth and metastasis [37]. PD-L1/PD-1 is a pair of immune checkpoints, the reduction of PD-L1 protein expression can enhance the ability of NK cells to attack tumor cells by releasing more perforin and granzyme B 37 . We first digested the MDA-MB-231 cells, replanted them into the plates, and then added PNSA and NK92 cells for 5h to observe whether PNSA increased the killing ability of NK cells. Our results showed that NK92 cells exhibited enhanced-cytotoxicity activity against cancer cells after PNSA treatment compared with DMSO control at effector-to-target ratio T:E of 5:1 ( Figure 7C). Furthermore, the releases of granzyme B and perforin were increased in PNSA-treated MDA-MB-231 cells ( Figure 7D,E). During the time, no change of NK92 cell viability upon PNSA treatment was observed ( Figure 7F). We further examined the effect of PNSA on NK92 cells viability for 72h and found that PNSA was much less cytotoxic to NK92 than to tumor cells with IC50 value at 12.96 μM, and the IC 50 value of PNSA for MDA-MB-231 cells was 0.86 μM ( Figure 7G). For comparison, the IC50 value of 17-AAG on NK92 cells for 72h was 9.08μM, meanwhile, the IC 50 value for MDA-MB-231 cells was 3.46 μM ( Figure 7H). These results suggest that PNSA may be better to trigger cancer immunology in vivo to deal with cancer metastasis. for 72 h and cell viability was measured using SRB assay or CCK-8. The bar graph represents the average ± SD of at least three independent experiments. ** p < 0.01; *** p < 0.001; ns, not significant (relative to DMSO-treated cells).

Discussion
In the presence of the paper, we prove that PNSA, a novel c-terminal inhibitor of HSP90 with the binding site at cysteine residues C572/C598 of HSP90, has the potential to cope with metastasis of breast cancer cells, which is superior to 17-AAG, a well-known Nterminal inhibitor binding to ATP pocket of HSP90.
Breast cancer is a complex and heterogeneous disease. In recent years, various molecular targets are being explored including epidermal growth factor receptor (EGFR), poly(ADP-ribose) polymerase (PARP), and vascular endothelial growth factor (VEGF). Receptors, protein tyrosine kinases, phosphatases, proteases, PI3K/Akt signaling pathway, microRNAs (miRs), and long noncoding RNAs (lncRNAs) are potential therapeutic targets [38]. It has been reported that HSP90 empowers the evolution of resistance to target therapy in human breast cancer models and pharmacological inhibition of Hsp90 shows great promise in breast cancer treatment [39][40][41]. HSP90 consists of three different domains, an N-terminal ATP binding domain, middle domain, and C-terminal dimerization domain [17]. So far, most N-and C-terminal HSP90 inhibitors target the ATP-binding region of HSP90 to blocking ATP binding, thereby inhibiting the formation of HSP90clients complex to promote degradations of these client proteins [42]. Inhibition of HSP90 can disrupt multiple signaling pathways that are important for the growth and metastasis for 72 h and cell viability was measured using SRB assay or CCK-8. The bar graph represents the average ± SD of at least three independent experiments. ** p < 0.01; *** p < 0.001; ns, not significant (relative to DMSO-treated cells).

Discussion
In the presence of the paper, we prove that PNSA, a novel c-terminal inhibitor of HSP90 with the binding site at cysteine residues C572/C598 of HSP90, has the potential to cope with metastasis of breast cancer cells, which is superior to 17-AAG, a well-known N-terminal inhibitor binding to ATP pocket of HSP90.
Breast cancer is a complex and heterogeneous disease. In recent years, various molecular targets are being explored including epidermal growth factor receptor (EGFR), poly(ADP-ribose) polymerase (PARP), and vascular endothelial growth factor (VEGF). Receptors, protein tyrosine kinases, phosphatases, proteases, PI3K/Akt signaling pathway, microRNAs (miRs), and long noncoding RNAs (lncRNAs) are potential therapeutic targets [38]. It has been reported that HSP90 empowers the evolution of resistance to target therapy in human breast cancer models and pharmacological inhibition of Hsp90 shows great promise in breast cancer treatment [39][40][41]. HSP90 consists of three different domains, an N-terminal ATP binding domain, middle domain, and C-terminal dimerization domain [17]. So far, most N-and C-terminal HSP90 inhibitors target the ATP-binding region of HSP90 to blocking ATP binding, thereby inhibiting the formation of HSP90-clients complex to promote degradations of these client proteins [42]. Inhibition of HSP90 can disrupt multiple signaling pathways that are important for the growth and metastasis of cancer cells [43]. However, the effects of different inhibitors of HSP90 on cancer metastasis and related mechanism are not very clear. Our data show that PNSA and 17-AAG all have the potential to inhibit migration and invasion of MDA-MB-231 cells. However, only PNSA is able to reverse EMT of MDA-MB-231 cells. We found that the cell morphology of MDA-MB-231 cells changed to epithelial cell shape obviously after treated with PNSA for 12 h. Furthermore, the molecule marker of epithelial cell E-cadherin was up-regulated while the molecule marker of mesenchymal cells N-cadherin, Vimentin, and MMP9 proteins PNSA were significantly down-regulated. However, 17-AAG did not obviously cause changes in E-cadherin protein expression and cell morphology. Our previous study found that slightly different binding [39] inhibitor-stirred HSP90 possesses selective inhibition potential on client proteins [22], we deduce that PNSA and 17-AAG may differently affect certain key proteins related to EMT development, which is worthy of being studied in the future. Increases in EMT process in primary breast cancer cells lead to enhance plasticity, enabling tumor progression, distant metastatic spread, and resistance to current chemotherapy and immunotherapy [44,45]. PNSA can reverse EMT of breast cancer cells and enhance the killing ability to breast cancer cells of NK cells by downregulation of PD-L1, which is better than 17-AAG, let alone PNSA is less toxic to NK cells than 17-AAG.
HSP90 is responsible for the maturation of more than 200 client proteins, and HSP90 inhibition usually induces degradation of its client proteins through the proteasome pathway [46]. The protein EGFR, FGFR, AKT, ERK, STAT3, vimentin, C-Myc, and MMP in our study are all reported to be client proteins of HSP90 (https://www.picard.ch/downloads/ Hsp90interactors.pdf, accessed on 11 February 2021). Our results showed that in MDA-MB-231 cells, PNSA was able to bind to HSP90, and protein levels of EGFR, FGFR, vimentin, C-Myc, and MMP were reduced upon PNSA treatment. As an example, EGFR and C-Myc were also degraded via the proteasome pathway. However, the protein levels of AKT, ERK, and STAT3 were unchanged. With siRNA-mediated depletion of HSP90, we noticed that the amounts of AKT, ERK, and STAT3 were indeed not altered, which may due to the different cell lines to be used, as HSP90 inhibitor performs the inhibitory effect depending on cell context [22]. Other proteins such as N-cadherin and PD-L1 were not reported to be client proteins of HSP90. Consistently, protein levels of these two proteins were influenced by siRNA of HSP90, especially, PD-L1 was not associated with HSP90 by immunoprecipitation assay.
Multiple signaling pathways play a crucial role in the process of tumor metastasis [47,48]. EGFR frequent overexpression and/or hyperactivation in breast carcinomas play an active role in facilitating brain-specific metastatic spread [49,50]. In addition, FGFR aberrations increase the risk of brain metastases and predict poor prognosis in metastatic breast cancer patients [51,52]. We found that PNSA distinctly suppressed the protein levels of EGFR and FGFR, and subsequently inhibiting down-streaming signaling pathway with reduction of P-ERK, P-STAT3, and P-AKT. Wnt/β-catenin signaling contributes to EMT and metastasis in breast cancer. Deyet al. has found that the patients identified by the Wnt/β-catenin classifier had a greater risk of lung metastasis in TNBC [53]. Our data showed that the amount of β-catenin was decreased, and its target gene protein C-Myc was also reduced.
In summary, PNSA inhibits migration and invasion not only in TNBC cells but also in trastuzumab-resistant JIMT-1 cells. Furthermore, PNSA reverses EMT of breast cancer cells and promotes the cytotoxicity of NK cells. All these results indicate that the novel C-terminal inhibitor of HSP90 PNSA is a promising anti-metastasis agent worthy of being studied in the future.

Cell Proliferation Assay
MDA-MB-231 and JIMT-1 cells were evaluated by SRB assay. NK92 cells were measured by CCK-8 assay. Briefly, Logarithmic growing MDA-MB-231 and JIMT-1 cells were plated in 96-well plates at an initial density of 5 × 10 3 per well. After 24 h, cells were treated with PNSA for 12 h or 72 h and SRB assay was used to evaluate the cell viability. The absorbance at 515 nm was measured using a microplate reader (BioTek, Winooski, VT, USA). NK92 cells were plated in 96-well plates at an initial density of 1 × 10 4 per well. Cells were incubated with PNSA for 5 h or 72 h and CCK-8 assay was used to evaluate the cell viability. The absorbance value was detected by a microplate reader at 450 nm.

Migration and Invasion Assay
MDA-MB-231 and JIMT-1 cells were seeded in 96-well plates (5 × 10 4 cells/well) and cultured for 24 h, obtaining a 90-100% confluent monolayer. Wounds were made with a p10 pipette tip and washed with PBS to eliminate non-adherent cells and cell debris, and a fresh medium with PNSA or DMSO was added. At 0 h and after 12 h, cells were photographed with a citation imaging reader. The empty area in each wound was quantified using Image J software (NIH) and compared to the corresponding initial wound. For invasion assay, the two kinds of breast cancer cells were tested with a Boyden chamber with 8 µm pore-size polycarbonate membranes coated with Matrigel (Corning). MDA-MB-231 and JIMT-1 (3 × 10 5 cells/well) were seeded in a serum-free medium with the indicated concentration of PNSA in the upper chamber; the lower chamber contained 20% fetal bovine serum. After 12 h, the cells on the upper side of the membrane were removed with a cotton swab, and cells on the underside were fixed with methanol for 30 min and then stained with 0.1% crystal violet for 15 min. Photos were taken on an EVOS, and cells were counted with Image J Software. Migration (%) = [1 − (wound healing area at 12 h/wound healing area at 0 h)] × 100%. Invasion rate (%) = (the number of invading cells in PNSA group /the number of invading cells in DMSO group) × 100%

Cell Aggregation Assay
The cell aggregation assay was performed essentially as described previously [54]. Briefly, a total of MDA-MB-231(2.5 × 10 5 ) cells in 1 mL of L-15 (serum-free) with different concentrations of PNSA were placed in polystyrene microtubes and shaken gently every 5 min for 1 h at 37 • C. Finally, glutaraldehyde (at a final concentration of 1% (v/v)) was used to stop the aggregation process. The percentage of aggregated cells was calculated as (1 − Ne/Nc) ×100%, where Ne is the number of single cells after incubation at 37 • C and Nc is the number of single cells before incubation.

Cell Adhesion Assay
MDA-MB-231 cells (1 × 10 4 cells/well) were plated into 96-well plates. After 24 h, cells were treated with PNSA at different concentrations and incubated for 12 h. Then, cells were suspended in serum-free L-15 to form a single-cell suspension and were seeded into 96-well cell culture plates that had been precoated with Matrigel. After 45 min incubation at 37 • C, the wells were washed three times with PBS to remove non-adherent cells. 5mg/mL of MTT was added to each well for 4 h. After 4 h, the DMSO was added to each well and the absorbance value was detected by a microplate reader at 450 nm. Adhesion rate = (absorbance of PNSA group/absorbance of DMSO group) × 100%.

Gelatin Zymography Assay
MDA-MB-231 cells (3 × 10 6 cells/well) were seeded in 6-well plates and the cells were allowed to attach for 24 h. The cells were then washed with PBS and incubated with serum-free medium containing PNSA for 12 h. Conditioned media from cell cultures treated with PNSA or DMSO were collected, centrifuged, and mixed with sample buffer (60 mM Tris-HCl pH 6.8, 25% glycerol, 2% SDS, 0.1% bromophenol) without reducing agents. The corresponding samples were loaded on 10% polyacrylamide gels with gelatin (1 mg/mL) and separated by gel electrophoresis. Gels were then washed with 2.5%Triton X-100 for 40 min and incubated in incubation buffer (50 mM Tris, 0.15 M NaCl, 10 mM CaCl 2 , 1 mM ZnCl 2 ) at 37 • C for 36 h. The cell lysates derived from the same separation gel were transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA). The membranes were blocked in Rapid protein blocking solution for 20 min and then washed three times in TBST (Tris-buffered saline with 0.1% Tween-20). The membranes were incubated overnight at 4 • C with indicated primary antibodies and then probed with an HRP-conjugated secondary antibody for 1 h. Blots detected by chemiluminescence with the enhanced chemiluminescence (ECL) detection reagents.

Western Blotting
MDA-MB-231 cells were incubated with PNSA at indicated concentrations for 12 h, then the medium was removed, and cells were washed with ice-cold phosphate-buffered saline (PBS) twice, disrupted on ice for 35 min in loading buffer, and boiled for 15 min. Equal amounts of protein in cell lysates were separated by SDS-PAGE, followed by transferring to a PVDF membrane (Bio-Rad, Hercules, CA, USA). The membranes were blocked in Rapid protein blocking solution for 20 min and then washed three times in TBST (Trisbuffered saline with 0.1% Tween-20). The membranes were incubated overnight at 4 • C with indicated primary antibodies and then probed with an HRP-conjugated secondary antibody for 1 h. Blots were detected by chemiluminescence with the enhanced chemiluminescence (ECL) detection reagents.

Cell Transfection
RNA interference assay was performed as previously described [55]. MDA-MB-231 cells (3 × 10 5 cells/well) were seeded in 6-well plates and the cells were allowed to attach for 24 h. The

Cellular Thermal Shift Assay (CETSA)
MDA-MB-231 cells (3 × 10 5 cells/well) were seeded in 6-well plates and the cells were allowed to attach for 24 h. After 24 h, cells were treated with PNSA at different concentrations and incubated for 3 h. After incubation, cells were resuspended in PBS and divided into several aliquots. Four aliquots were treated with DMSO and other four aliquots were treated with PNSA. Then, cells were heated at different temperatures by Biometra TOne PCR (Analytikjena, Jena, Germany). The heated cells were freeze-thawed three times with liquid nitrogen for 4 min every time and followed by centrifugation at 20,000× g for 20 min at 4 • C. The supernatants were harvested and loading buffer was added before boiling for 15min. Protein levels were assessed by Western blotting.

Immunoprecipitation
MDA-MB-231 cells were seeded in 6-well plates and the cells were allowed to attach for 24 h. After 24 h, the medium was removed, then washed twice with PBS, and disrupted on ice for 30 min. The lysates were centrifuged at 10,000× g for 15 min, and then the supernatants were harvested. Next, the lysates were incubated with 1 µg HSP90 antibody or control IgG overnight at 4 • C. The mixtures were incubated with a 40 µL slurry of protein A/G-agarose for 3 h at 4 • C. The immunoprecipitates were gathered by centrifugation and then washed six times with wash buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton, pH 7.5). Finally, loading buffer was added and followed by boiling. Protein levels were assessed by Western blotting.

Cytotoxicity Assays of NK92 Cell
The cytotoxicity assay was performed essentially as described previously [56]. MDA-MB-231 cells (3 × 10 4 cells/well) were seeded in 96-well plates and the cells were allowed to attach for 24 h. After 24 h, cells were treated with PNSA at different concentrations and incubated for 24 h. Then, we digested the MDA-MB-231 cells, replanted the plates, and after the cells adhered to the plates, added NK92 cells at an effector-to-target ratio of 5:1 for 5 h at 37 • C. After 5 h, 20 µL CCK-8 was added to each well, and the incubation continued for 1 h. The absorbance value was detected by a microplate reader at 450 nm. Cytotoxicity (%) = [1 − (experimental group OD/effector group OD)/ targeted group OD] × 100%

Analysis of Perforin and Granzyme by ELISA
MDA-MB-231 cells (3 × 10 4 cells/well) were seeded in 96-well plates and the cells were allowed to attach for 24 h. After 24h, cells were treated with PNSA at different concentrations and incubated for 24 h. Then, we digested the MDA-MB-231 cells, replanted the plates, and after the cells adhered to the plates, added NK92 cells at an effector-to-target ratio of 5:1 for 8 h at 37 • C. After 8 h, supernatant was obtained from 96-well plates by centrifugation at 1500× g for 10 min at 4 • C and immediately frozen at −80 • C for further analysis. Perforin and Granzyme B levels in the serum were measured using human Perforin and Granzyme B ELISA kits (Dakewe Biotech Co, Ltd., Beijing, China), according to the manufacturer's instructions.

Statistical Analysis
Data were presented as mean values ± standard deviation. Student's t-test and oneway ANOVA analysis were used to analyze the differences between groups. Differences were considered statistically significant when p < 0.05.