Inhibition of LIPG phospholipase activity suppresses tumor formation of human basal-like triple-negative breast cancer

The endothelial lipase LIPG possesses serine phospholipase activity and is involved in lipoprotein metabolism. Our previous studies have revealed that LIPG overexpression is required for tumor formation and metastasis of human basal-like triple-negative breast cancer (TNBC). We also demonstrated that LIPG differentially regulates TNBC malignancy through its enzymatic and non-enzymatic functions. The present studies were aimed at determining how XEN445, a specific inhibitor targeting LIPG phospholipase activity, impacts on TNBC tumor formation and malignant features. We established a cell-based LIPG enzymatic assay system to measure the inhibitory effect of XEN445 on LIPG phospholipase activity and determine its IC50. We found that XEN445 preferentially inhibited the proliferation of LIPG-expressing TNBC cells but not LIPG-negative luminal breast cancer cells. XEN445 inhibited the self-renewal of cancer stem cells (CSCs) in vitro and TNBC tumor formation in vivo. However, XEN445 had no inhibitory effect on the invasiveness and CSC stemness of TNBC cells. Our studies suggest that targeting both LIPG enzymatic and non-enzymatic functions is an important strategy for the treatment of TNBC.

. Analysis of the inhibitory effect of XEN445 on the phospholipase activity of LIPG in a cell-based assay. (A) A diagram shows the functional role of LIPG in breast cancer. LIPG possesses both enzymatic and non-enzymatic functions to specifically regulate different aspects of breast cancer malignancies. XEN445 is a pharmacological inhibitor that specifically targets the enzymatic function of LIPG. (B) Kinetically studying LIPG phospholipase activity of LIPG-overexpressing MDA-MB-468 cells with or without XEN445 treatment. Parental and LIPG-overexpressing MDA-MB-468 cells were analyzed in LIPG enzymatic assays. Either vehicle (DMSO) or XEN445 (200 μM) was included in assays. Kinetic studies were performed for 30 minutes with 1 cycle/min. Duplicate experiments were performed and their kinetic data were plotted separately for showing experimental accuracy and reproducibility. Two biological replicates for each assessed sample include 468-CTRL + DMSO (indicated by dark blue and sky blue), 468-CTRL + XEN445 (red and pink), 468-LIPG OE + DMSO (dark green and light green) and 468-LIPG OE + XEN445 (dark brown and orange). (C) Dosedependent analysis of the inhibitory effect of XEN445 on LIPG phospholipase activity. The phospholipase activity of LIPG was measured under a series of 2-fold diluted XEN445 doses as indicated in the figure. The end-point enzymatic activity of LIPG was plotted against the XEN445 dose. The relative LIPG enzymatic activity expressed as a percentage was calculated from the measured enzymatic fluorescence data as described in "Materials and Methods". Triplicate experiments were performed. Error bars shown in the plot are standard deviation (SD). (D) Linear regression analysis of log2-transformed LIPG enzymatic activity against log2transformed XEN445 doses. Data shown in (C) were log2-transformed and their linear regression relationship was analyzed. The fitted linear regression line was plotted based on the linear modeled equation log2(LIPG activity) = −0.514 X log2(XEN445 dose) + 6.203 (multiple R-squared: 0.9825, adjusted R-squared: 0.979). The predicted IC50 is 2.172 μM.

Results
Analysis of the inhibitory effect of XEN445 on the phospholipase activity of LIPG. XEN445 has shown to inhibit the phospholipase activity of LIPG (Fig. 1A) 8 . To validate the inhibitory effect of XEN445 on the enzymatic activity of LIPG at a cellular level, we performed LIPG enzymatic assays using parental and LIPG-overexpressing MDA-MB-468 cell lines 16 . Figure 1B shows that LIPG-overexpressing MDA-MB-468 cells (468-LIPG OE) manifested significantly higher phospholipase A1 (PLA1) activity than parental cells (468-CTRL) (approximately 5-fold). Treatment with XEN445 (200 μM) led to the approximate 80% reduction of PLA1 activity in 468-LIPG OE cells compared to the estimated 25% reduction of PLA1 activity in 468-CTRL cells (Fig. 1B). Given that 200 μM XEN445 nearly inhibited the total overexpressed LIPG activity and one-fourth cellular endogenous PLA1 activity, these results suggest that XEN445 is a specific LIPG inhibitor and endogenous LIPG in MDA-MB-468 cells contributes to approximate 25% of total PLA1's activity. These findings indicate that the 468-LIPG OE cell line is a suitable cell model for the analysis of the XEN445 effect on LIPG enzyme activity.
We performed LIPG enzymatic assays to determine the IC50 of XEN445. Figure 1C shows the relationship between LIPG activity and XEN445 doses was not linear. Data shown in Fig. 1C were log2-transformed, and a linear regression relationship was analyzed. We found that log2 values for both variables (LIPG activity and XEN445 dose) showed a linear relationship with intercept (6.203), slope (−0.514) and adjusted R-squared (0.979) values (Fig. 1D). Based on the linear regression equation, the predicted IC50 of XEN445 is 2.172 μM, which is approximately 9-fold higher than the reported IC50 (0.237 μM) of XEN445 8 . This difference may be due to a different approach used in the assay system.

XEN445 preferentially inhibits the proliferation of LIPG-expressing TNBC cells. Our previous
studies have shown that the enzymatic function of LIPG is required for promoting and sustaining the proliferation of breast cancer cells 16 . Therefore, we expected that XEN445 treatment may inhibit the proliferation of LIPG-expressing breast cancer cells. To test this hypothesis, we performed cell viability assays on MCF10DCIS cells, a LIPG-expressing TNBC cell line 16 , treated with either vehicle (DMSO) or XEN445 (1 to 200 μM). Figure 2A shows that XEN445 (>=100 μM) significantly inhibited the cell viability of MCF10DCIS cells.
To reveal whether XEN445 inhibits cell viability in a LIPG-dependent manner, we treated two LIPG-expressing TNBC cell lines (MCF10DCIS and MDA-MB-468) and two LIPG-deficient luminal breast cancer (LuBC) cell lines (MCF7 and T47D) 16 with XEN445 at two different doses (200 and 250 μM). The data demonstrate that XEN445 treatment selectively decreased the cell viability of LIPG-expressing TNBC lines rather than LIPG-deficient LuBC lines (Fig. 2B). This result suggests that XEN445-elicited inhibition of cell viability is LIPG-dependent.
To further confirm the result of Fig. 2B, we performed the same XEN445 treatment on LIPG-overexpressing MDA-MB-468 cells, LIPG-knockdown MDA-MB-468 cells and LIPG-overexpressing MCF7 cells. LIPG overexpression and knockdown were confirmed previously 16 . Figure 2C shows that LIPG overexpression sensitized MDA-MB-468 cells to the inhibitory effect of XEN445 when compared to parental cells (Fig. 2B). However, LIPG knockdown rendered MDA-MB-468 resistant to the XEN445 impact (Fig. 2C). Consistently, LIPG overexpression made MCF7 cells sensitive to the inhibitory effect of XEN445 (Fig. 2C), whereas parental MCF7 cells were highly resistant to XEN445 (Fig. 2B). These data, taken together, demonstrate that the antitumor activity of XEN445 is highly LIPG-dependent.
To understand how XEN445 inhibits the viability of TNBC cells, we performed cell cycle analysis of TNBC cells treated with or without XEN445. Figure 2D shows that XEN445 treatment led to an increase in G1 (increased from 47.48% of the control to 64.24%) and decreases in S-G2/M populations (S: decreased from 26.75% of the control to 17.38%; G2/M: decreased from 18.15% of the control to 11.71%) in MCF10DCIS. There was no significant difference between sub-G1 percentages (indicating apoptotic cells) of vehicle-treated (6.27%) and XEN445-treated (4.81%) MCF10DCIS cells, indicating that XEN445 had no significant effect on cell apoptosis. Consistent with MCF10DCIS data, XEN445 treatment of MDA-MB-468 cells induced G1 arrest (G1: increased from 44.14% of the control to 55.74%) and reduced S-G2/M populations (S: decreased from 29.22% of the control to 19.34%; G2/M: decreased from 19.07% of the control to 17.45%), but failed to induce apoptosis (4.16% of the control vs. 3.58%), in MDA-MB-468 cells (Fig. 2E). We further examined whether XEN445 impacts cell cycle with a LIPG-dependent manner. Our FACS's data demonstrate that ectopic LIPG overexpression sensitized MDA-MB-468 cells to XEN445 treatment, leading to increased G1-phase (increased from 47.55% of the control to 62.80%) and decreased G2/M-phase (decreased from 18.86% of the control to 12.87%) cell populations to a greater extent (Fig. 2F) than those observed in parental cells (Fig. 2E). Similar to parental cells, XEN445 treatment caused a reduction in the S phase by approximately 10% (decreased from 25.52% of the control to 15.71%) (Fig. 2F). In contrast, XEN445-induced differences in G1 (56.38% of control vs. 59.11% of XEN445), S (19.62% of control vs. 17.68% of XEN445) and G2/M (17.69% of control vs. 16.52% of XEN445) phases of MDA-MB-468 cells with LIPG knockdown (Fig. 2G) were minor when compared to parental (Fig. 2E) and LIPG-overexpressing cells (Fig. 2F), indicating that LIPG knockdown rendered MDA-MB-468 cells resistant to the effect of XEN445. These findings, taken together, revealed that XEN445 reduced breast cancer cell viability through G1 cell cycle arrest.

XEN445 augments the invasiveness of TNBC cells in a LIPG-dependent manner.
Our previous studies have shown that the invasiveness of TNBC cells is independent of the lipase function of LIPG 16 . To further confirm our findings, we examined the migratory activity of control and LIPG-knockdown MDA-MB-468 cells treated with either vehicle or XEN445 (250 μM) using a Boyden assay. Untreated cells were also included in the study to evaluate the effect of the vehicle (DMSO). Unexpectedly, XEN445 treatment significantly enhanced the migration of MDA-MB-468 cells (~2.84-fold, p < 0.001) when compared to vehicle-treated cells (Fig. 3A). In contrast, LIPG knockdown almost completely abolished this enhancing effect of XEN445 on the migration of www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ MDA-MB-468 (Fig. 3A). The migratory activity of vehicle-treated cells was similar to untreated cells, indicating that DMSO has no significant effect on cell migration. To rule out the possibility that these observed outcomes were cell-line-specific, we performed the same XEN445 treatment experiment on MCF10DCIS cells with or without LIPG knockdown. In line with the result from studying MDA-MB-468 (Fig. 3A), XEN445 treatment also significantly promoted migration of MCF10DCIS cells (~3-fold, p < 0.001) and LIPG knockdown blocked this effect in a same manner (Fig. 3B).
To examine whether XEN445 can also promote invasion of TNBC cells, we performed invasion assays on control and LIPG-knockdown MDA-MB-468 cells treated with either vehicle or XEN445 (250 μM). Consistent with the migration data ( Fig. 3A), the invasive activity of MDA-MB-468 cells was also enhanced by XEN445 treatment (~1.9-fold, p < 0.001), which again was abolished by LIPG knockdown (Fig. 3C). The same as the migration data, the vehicle (DMSO) has no significant effect on cell invasion.
To examine whether the XEN445-enhanced TNBC invasiveness resulted from any changes in LIPG protein levels after XEN445 treatment, we performed western blot analysis of LIPG protein expression in MDA-MB-468 and MCF10DCIS cells treated with either vehicle or XEN445 for 4 days. We included control lysates from cells with LIPG knockdown to confirm the specificity of the LIPG antibody. The western blot data demonstrated that XEN445 treatment slightly decreased LIPG protein levels in MDA-MB-468 cells and had no significant impact on LIPG protein levels in MCF10DCIS cells (Fig. 3D,E) (Original western blot images can be found in Supplementary Materials). These results suggest that the XEN445-enhanced TNBC invasiveness does not result from altering LIPG protein levels. In conclusion, we unexpectedly observed that XEN445 enhances the invasiveness of TNBC cells in a LIPG-dependent manner.

XEN445 has no inhibitory impact on stemness and basal-like features in LIPG-expressing TNBC cells.
In our previous studies, we demonstrated that the lipase-independent function of LIPG regulates the stemness of TNBC cells 16 . To address the impact of XEN445 on stemness of TNBC cells, we performed flow cytometry analysis of the CD44+ CD49f+ cell population in MCF10DCIS cells, representing a CSC-enriched cell fraction 17 . We first performed the flow cytometry analysis of MCF10DCIS cells with or without LIPG knockdown. LIPG knockdown led to a reduction in the CD44+ CD49f+ cell population from 37.95 ± 1.05% to 33.88 ± 1.00% (p < 0.01) (Fig. 4A). This result indicated that loss of LIPG led to a reduction in the stemness of MCF10DCIS. In contrast, XEN445 treatment (250 μM) significantly increased the percentage of CD44+ CD49f+ cells from 37.61 ± 0.86% to 45.88 ± 4.07% (p < 0.05) (Fig. 4B). To examine the impact of XEN445 on the basal-like characteristics of MCF10DCIS cells (a hallmark of the mesenchymal phenotype in TNBC), we performed flow cytometry analysis of EpCAM (a well-known epithelial/luminal protein marker inversely related to the cell basal-like feature) on MCF10DCIS cells treated with either vehicle or XEN445. As shown in Fig. 4C, XEN445 treatment reduced the percentage of EpCAM high cells in MCF10DCIS from 31.51 ± 0.40% to 26.09 ± 2.73% (p < 0.05), indicating that XEN445 promoted the basal-like phenotype of MCF10DCIS cells. We were surprised for this data because our previous LIPG knockdown experiments showed that LIPG knockdown significantly altered the basal-like characteristic of MCF10DCIS cells to the luminal phenotype 16 .
To confirm whether the findings described above can be recapitulated in MDA-MB-468 cells, we first performed flow cytometry analysis of the CD44+ CD24-cell population, representing a CSC-enriched cell fraction in MDA-MB-468. We found that XEN445 treatment had no impact on the CD44+ CD24-cell population in MDA-MB-468 (96.00 ± 0.41% of control vs. 95.27 ± 1.51% of XEN445), whereas LIPG knockdown significantly decreased the percentage of CD44+ CD24-cells (from 96.00 ± 0.41% of the control to 72.54 ± 3.24%, p < 0.001) (Fig. 4D). Moreover, flow cytometry analysis of EpCAM-positive (EpCAM+) cells showed that XEN445 was unable to change the basal-like feature of MDA-MB-468 cells to the luminal phenotype because we did not observe an increase in the EpCAM+ cell percentage (Fig. 4E). Actually, EpCAM analysis showed a reverse result that XEN445 treatment gave rise to a reduction in the EpCAM+ cell population (decreased from 3.01 ± 0.45% of the control to 1.51 ± 0.16%, p < 0.01), similar to the result from the MCF10DCIS study (Fig. 4C). These findings indicate that XEN445 has no inhibitory effect on stemness and basal-like traits in LIPG-expressing TNBC cells.
XEN445 suppresses CSC self-renewal of LIPG-expressing TNBC. Self-renewal of CSCs is a process of renewing CSC, which involves both activated stemness and cell division. The sphere formation assay is a frequently performed method to measure the self-renewal capacity of CSCs. To examine the impact of XEN445 on self-renewal of CSCs in LIPG-expressing TNBC, we performed sphere formation assays on MCF10DCIS cells with either XEN445 treatment or LIPG knockdown compared to control cells. We found that the total number of formed CSC spheres from MCF10DCIS cells was significantly decreased by XEN445 treatment (~ 69% reduction, p < 0.001) (Fig. 5A). Consistently, the total number of sphere cells was decreased by XEN445 treatment (~59% reduction, p < 0.001) (Fig. 5B). Furthermore, LIPG knockdown in MCF10DCIS cells dramatically reduced the total numbers of CSC spheres (~ 96% reduction, p < 0.001) (Fig. 5A) and sphere cell viability (~90% reduction, p < 0.001) (Fig. 5B). These findings demonstrate that XEN445 attenuates the self-renewal of CSCs in MCF10DCIS.
To determine the effect of XEN445 on CSC self-renewal of MDA-MB-468 cells, we performed sphere formation assays on XEN445-treated cells compared to control and LIPG-knockdown cells. Consistent with the result from the MCF10DCIS study (Fig. 5A), the treatment of MDA-MB-468 with XEN445 significantly decreased the total number of formed CSC spheres (~83% reduction, p < 0.001) (Fig. 5C). In line with the sphere data, the total number of viable MDA-MB-468 sphere cells was significantly decreased by XEN445 treatment (~68% reduction, p < 0.001) (Fig. 5D). Similarly, LIPG knockdown in MDA-MB-468 cells led to dramatic decreases in the total numbers of formed CSC spheres (~97% reduction, p < 0.001) and viable sphere cells (~85% reduction, p < 0.001) (Fig. 5C,D). Given that the flow cytometry data showed that XEN445 had no inhibitory impact on  www.nature.com/scientificreports www.nature.com/scientificreports/ the subpopulation of TNBC CSCs (Fig. 4), the data from sphere formation assays suggest that XEN445 inhibited self-renewal of CSCs in LIPG-expressing TNBC through suppressing the cell division of CSCs, but not CSC stemness properties.

XEN445 impedes in vivo tumor growth of basal-like TNBC cells.
To address the impact of pharmacological inhibition of LIPG enzymatic function by XEN445 on TNBC tumor growth in vivo, we performed XEN445 therapy (50 mg/kg) on nude mice with MDA-MB-468 xenograft tumors. Consistent with the result from in vitro cell studies shown in Fig. 2B, XEN445 treatment significantly inhibited tumor growth in nude mice (p < 0.001) (Fig. 6A). To determine whether tumor cell proliferation is inhibited in vivo by XEN445 treatment, we performed immunohistochemistry (IHC) analysis of Ki67, a cell proliferation marker, on vehicleand XEN445-treated tumors. In line with the tumor growth data, XEN445 therapy significantly decreased the number of Ki67-positive cells in xenograft tumors (150 ± 18/1000 tumor cells, n = 3, p < 0.001) when compared to vehicle-treated tumors (423 ± 27/1000 tumor cells, n = 3) (Fig. 6B). To examine the EMT status of XEN445-treated xenograft tumors, we performed IHC analysis of vimentin on isolated tumors treated with either vehicle or XEN445. As shown in Fig. 6C, there was no significant difference in vimentin staining between vehicleand XEN445-treated tumors. This finding suggests that XEN445 therapy has no inhibitory impact on the EMT/ mesenchymal phenotype of MDA-MB-468 xenograft tumors, consistent with the result from the qRT-PCR study (Fig. 5E). These in vitro and in vivo findings from XEN445 therapy studies contrast with our previous findings from LIPG knockdown studies showing that LIPG loss-of-function led to downregulation of vimentin expression in MDA-MB-468 cells 16 .

Discussion
Several studies have revealed that histone H3 K36 demethylase KDM4A, caspase-8, and lysyl oxidase have enzymatic and non-enzymatic functions [18][19][20] . Mechanistically, these enzymes execute their non-enzymatic functions through protein-protein interactions. Our previous studies have shown that LIPG also possesses both enzymatic and non-enzymatic functions in breast cancer cells 16 . The phospholipase function of LIPG is responsible for supporting cell growth and promoting cell proliferation rate. In contrast, the phospholipase-independent function of LIPG is involved in oncogenic DTX3L-ISG15 signaling and promotes invasiveness, stemness and basal/EMT features of breast cancer cells 16 . Although the mechanism by which LIPG executes its non-enzymatic function is unknown, it is likely through protein-protein interactions.
The only currently approved targeted therapy for TNBC is the immunotherapy with atezolizumab for patients whose tumors express PD-L1, which was found to increase progression-free survival. Since our prior studies have shown that LIPG is essential for the malignancy and metastasis of TNBC 16 , it is clinically imperative to investigate the therapeutic effects of currently available chemical inhibitors targeting LIPG. In this study, we for the first time explored the therapeutic impacts of XEN445, a chemical inhibitor specific to the phospholipase activity of LIPG 8 , on TNBC malignancy. We first examined the effect and IC50 of XEN445 on LIPG phospholipase activity. Consistent with the previous finding 8 , XEN445 specifically inhibited the enzymatic activity of LIPG in our cell-based LIPG enzymatic activity assay system, and IC50 is approximate 2 μM, which is approximately 9-fold higher than the previously reported IC50 8 . Furthermore, we found that the high dose of XEN445 (200 μM) was needed to observe its significantly inhibitory effect in cell function studies. This dose is approximately 100-fold higher than IC50 observed in the LIPG enzymatic study. The serum in the cell culture medium is a possible cause After a week, the numbers of formed tumorspheres were counted as described above. After sphere counting, tumorspheres were dissociated into single cells as described above. The obtained tumorsphere and single cell number data were used to plot the bar graphs of sphere formation (shown in C) and cell viability (shown in D). (E) Expression profiling analysis of stemness-related and EMT programming genes in vehicle-treated and XEN445-treated MDA-MB-468 cells. (F) Expression profiling analysis of stemness-related and EMT programming genes in vehicle-treated and XEN445-treated MCF10DCIS cells. Errors are SD; n = 3; **p < 0.01; ***p < 0.001.

Scientific RepoRtS |
(2020) 10:8911 | https://doi.org/10.1038/s41598-020-65400-7 www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ leading to the need of the high dose of XEN445 in cell function studies as serum proteins are reported to bind XEN445 8 and therefore it is expected that serum affects the concentration of the free form of XEN445 in the cell culture medium.
With regard to the impact of XEN445 on malignancy of breast cancer cells, we found that in concordance with findings from studies of the catalytically inactive LIPG mutant 16 , XEN445 suppressed TNBC cell proliferation and CSC self-renewal, but not other TNBC properties such as invasiveness, and EMT/basal-like features. In line with these in vitro findings, XEN445 therapy inhibited in vivo tumor growth of TNBC xenografts, but not their basal-like characteristics. Therefore, our studies suggest that pharmacologically inhibiting LIPG phospholipase activity is not sufficient to hinder every aspect of TNBC malignancy.
In addition to our finding that XEN445 was unable to impede the non-enzymatic function of LIPG, our studies have revealed another important finding that XEN445 actually enhanced the non-enzymatic function of LIPG, evidenced by XEN445-induced invasiveness, stemness and EMT/basal-like features. It remains unknown how XEN445 can enhance tumor cell invasiveness. However, our previous studies have shown that ectopic overexpression of the catalytically inactive LIPG mutant also enhanced migration and stemness of luminal breast cancer cells to a greater degree than overexpression of wild-type LIPG 16 . These prior findings indicate that inactivating LIPG phospholipase activity promotes the non-enzymatic function of LIPG. Based on our studies, we hypothesize that LIPG phospholipase activity has an intrinsically inhibitory effect on the non-enzymatic function of LIPG in a direct and/or indirect manner. If this is true, inactivating both enzymatic and non-enzymatic functions of LIPG will be essential for inhibiting the whole oncogenic function of LIPG in breast cancer.
Accumulating evidence indicates that disseminated tumor cells (DTCs) can acquire stem-cell-like properties as well as the mesenchymal/EMT feature and their quiescent stemness (also known as stem-cell dormancy) is activated during their metastasis and after DTCs settle down at distant tissues 21 . These stem-cell-like DTCs can stay in the dormant state at distant tissues for many years before they are activated to proliferate and differentiate in the formation of metastatic tumors 21 . The essential role of LIPG in TNBC metastasis 16 suggests that both enzymatic and non-enzymatic functions of LIPG participate in modulating the metastatic process, the dormancy of DTCs and metastatic outgrowth in a highly plastically regulatory manner. Our study suggests that inhibition of LIPG phospholipase activity only impedes primary tumor growth as well as metastatic outgrowth and has no impact on dormant DTCs and quiescent primary tumor cells. Furthermore, blocking LIPG enzymatic activity may potentially promote CSC properties as well as EMT/basal-like features in tumor cells and induce the dormancy of DTCs. Therefore, the development of new therapeutic agents that can target both enzymatic and non-enzymatic functions of LIPG is necessary in the future for TNBC therapy.

Materials and Methods
Cell culture and XEN445. We  LIPG enzymatic assay. Both parental and LIPG-overexpressing MDA-MB-468 cell lines were analyzed by the LIPG enzymatic assay. To perform this assay, single cells were prepared and washed with the assay buffer (HBSS with 25 mM HEPES), and cells were eventually suspended in an appropriate volume of the assay buffer. Cells were diluted with the assay buffer to make the 2X cell prep (6 × 10 5 /mL). To make the 2X working substrate solution, 1 mM PED-A1 (Thermo Fisher Scientific), a specific substrate for phospholipase A1,was diluted with HBSS-HEPES (1:125 dilution) to 8 μM. For analysis of the inhibitory effect of XEN445, a series of 2 times XEN445 doses in the 2X working substrate solution were prepared. 50 μl of the 2X working substrate solution with or without XEN445 was loaded into each well of the 96-well plate. 50 μl of the cell suspension (3 × 10 4 ) from the 2X cell prep was loaded into each well and mixed with the working substrate solution. Enzymatic assays were performed at 37 °C in a kinetic mode (488 nm for excitation, 520 nm for emission) using the setting of 30 cycles (1 cycle/min). For the calculation of LIPG enzymatic activity, the measured fluorescence data from 468-LIPG OE cell samples were subtracted by the average basal fluorescence data from DMSO-treated 468-CTRL cell samples to obtain LIPG OE-contributed fluorescence measurements, and then the average fluorescence value (after subtraction) from DMSO-treated 468-LIPG OE cell samples was set as default 100% to calculate the relative LIPG enzymatic activities of other treated 468-LIPG OE cell samples. siRnA transfection. The siRNA transfection was performed with 40 nM of siRNA using Lipofectamine RNAiMAX (Thermo Fisher Scientific) according to manufacturer's instructions. siRNAs were obtained from Sigma-Aldrich. The siRNA sequences used in the study are: siControl, 5′-UAACUCGCUCGAAGGAAUC-3′; siLIPG, 5′-GCCGCAAGAACCGUUGUAA-3′ 16 .

Establishment of stable LIPG-overexpression and LIPG-knockdowncell lines. Stable
LIPG-overexpression and LIPG-knockdown MDA-MB-468 cell lines were established as described previously 16 . cell viability assay. 2 × 10 4 cells were seeded into the well of a 6-well cell culture plate with culture media and cultured overnight. Cells were then treated with either vehicle (DMSO) or XEN445 for four days. After harvested from mice treated with either vehicle or XEN445. Representative staining pictures are shown. Scale bars indicate 50 μm. The quantitative bar graph for the vimentin staining data was generated as described in (B). ns: not significant. (2020) 10:8911 | https://doi.org/10.1038/s41598-020-65400-7 www.nature.com/scientificreports www.nature.com/scientificreports/ four-day treatment, cells were trypsinized and spun down. The cells were resuspended in the 0.5 ml medium and 10 μl of cells was taken for cell counting by staining with trypan blue. The viability of drug-treated cells relative to vehicle-treated cells was determined by the ratio of drug-and vehicle-treated cell counts. cell cycle analysis. Cell cycle analysis was performed as previously described 22 . In brief, trypsinized cells at a density of 10 6 cells/ml in phosphate buffered saline (PBS) were fixed by 70% ethanol. Ethanol-fixed cells were stained in the staining buffer made of propidium iodide (0.4 mg/ml), RNase A (0.1 mg/ml) and 0.1% Triton X-100 in PBS. The cell-cycle profile and sub-G1 fraction were analyzed by flow cytometry using the Becton Dickinson FACSAria II system (Franklin Lakes, NJ).
Migration and invasion assays. Transwell-based migration and invasion assays were implemented as previously described 23 .
Western blot analysis. Western blot analysis was performed as previously described 16 . Protein expression of LIPG was examined by western blotting using a mouse antibody against LIPG (Abcam, ab56493). Protein expression was detected by chemiluminescence (ECL, Pierce). Expression of α-tubulin (Thermo Fisher Scientific) was used as a protein loading control. Western blot data were quantified by densitometric analysis of autoradiograms, using a computerized densitometer (Typhoon System; Molecular Dynamics, Inc., Sunnyvale, CA). The quantitative protein level data were normalized by the α-tubulin protein levels. Sphere formation assay. We performed sphere formation assays as previously described 23 . After one-week sphere culture, formed primary spheres were counted under a microscope according to the sphere size criterion (≥100 μm). For sphere cell viability assays, spheres were collected and dissociated with accutase (BioLegend) for obtaining single sphere cells. Collected sphere cells were spun down and resuspended in the 0.2 ml sphere culture medium. 10 μl of dissociated, single sphere cells was taken for cell counting by staining with trypan blue. The viability of treated sphere cells relative to control sphere cells was determined by the ratio of treated and control sphere cell counts.
RnA isolation and quantitative Rt-pcR (qRt-pcR) analysis. Total RNA of cultured cells was isolated using the Ambion TRIzol reagent (Thermo Fisher Scientific, Halethorpe, MD) according to manufactures' instructions. qRT-PCR analysis of mRNA expression was performed as described previously with normalization to GAPDH 23 . The sequence information of gene primers used in qRT-PCR experiments has been included in our publications 16,17 . In vivo tumorigenicity assay. The fourth mammary glands of 8-week-old female nude mice were transplanted with MDA-MB-468 cells (1 × 10 6 cells per injection) for xenograft mammary tumor formation. After cell transplantation, mice were subsequently split into two groups that were treated with vehicle and XEN445 (50 mg/ kg), respectively, via intraperitoneal (IP) injection on the same day and later three times per week. Athymic nude mice (NU/NU mice, Stock No: 002019) were obtained from the Jackson Laboratory (Bar Harbor, ME). The length and width of tumors were measured every four days with a caliper to calculate tumor volume using the formula: V = 1/2 (Length × Width 2 ) 24 . At the endpoint, xenograft tumors were isolated and processed for immunohistochemistry analysis. Xenograft tumor experiments were performed according to the animal protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Maryland School of Medicine, which is in accordance with the guidelines established by the USPHS. immunohistochemistry assay. The immunohistochemistry (IHC) analysis was performed using the avidin biotin peroxidase complex (ABC) method as previously described 17 . Anti-Ki67 (27309-1-AP) and anti-vimentin (10366-1-AP) rabbit polyclonal antibodies used in IHC experiments were obtained from Proteintech (Rosemont, IL). Statistical analysis. Statistical analysis was performed as previously described 16 . Data are presented as mean ± S.D. Statistical analysis of general experimental datasets was performed by Student's t test. Statistical analysis of tumor growth curves was performed by Two-way ANOVA. The p values of <0.05 were considered significant. Data were analyzed using the GraphPad Prism software (version 6.0; GraphPad Software, Inc, La Jolla, CA). DVA/US Government disclaimer. The contents do not represent the views of the U.S. Department of Veterans Affairs or the United States Government.