Clinical correlation analysis of TSP50 expression with breast cancer

TISIDB (http://cis.hku.hk/TISIDB/index.php) [25], an integrated repository portal for tumor-immune system interactions, was used to analyze the expression of TSP50 in different molecular subtypes of breast cancer. Furthermore, we evaluated the prognostic significance of TSP50 in KM plotter online database (http://kmplot.com/) [26], patients were split by auto select best cutoff. The ROC of therapy-related survival and differences in TSP50 expression between responders and non-responders were investigated using the ROC plotter (www.rocplot.org), a web-based tool that connects gene expression to therapeutic response [27]. In addition, the GEPIA2 database (http://gepia2.cancer-pku.cn/#index) [28] was used to perform expression correlation analysis of TSP50 with breast cancer stem cell (BCSC) markers, including OCT4, ALDH1, NANOG and CD44.

Cell culture, drug treatment and plasmid transfection

The human breast cancer cell lines MDA-MB-231, MCF7, T47D, ER7530 and SKBR3 were obtained from the Chinese Academy of Sciences and cultured in RPMI 1640 medium with 10% FBS, 100 units/mL penicillin and 100 mg/mL streptomycin. All the cells were maintained at 37 °C in an incubator with 5% CO2. For drug treatment, breast cancer cells were incubated with 10 µM SC79, 20 µM LY294002 or 5 µM BYL-719 for subsequent experiments. For transfection, 200 µL of RPMI 1640 serum-free medium was mixed with 6 µL of X-tremeGENE HP and 2 µg of plasmid when the cell confluence reached 80%. The cells were harvested for subsequent experiments after the specified transfection time.

Lentiviral-mediated stable transfection of breast cancer cells

We utilized lentiviral vectors obtained from Miaoling Bio., which encode control, shTSP50 for TSP50 knockdown, and TSP50 overexpression sequences, to establish stable transfection in breast cancer cells. The shRNA sequence designed to target TSP50 was as follows: GGAACTCAAGTACAGCAATTATTCAAGAGATAATTGCTGTACTTGAGTTCCTT. For the transfection process, breast cancer cells were plated in a 24-wells plate at a density of 5×10⁴ cells per well and incubated for 24 h. Subsequently, the cells were infected with the lentiviral particles for an additional 24 h. To enhance the infection efficiency, polybrene was added according to the manufacturer’s protocol. Finally, stably transfected cell clones were established by the addition of puromycin.

Mammosphere formation and analysis

Breast cancer cells were digested into single cells and subsequently seeded in ultralow attachment plates at a density of 500 or 3000 cells per well. The cells were cultured in epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) added DMEM/F12 serum-free culture medium for 10–14 days. The number and diameter of formed mammospheres were then quantified. For the assessment of self-renewal capacity, the mammospheres were collected, dissociated into single cells, and then reseeded under the same culture conditions as described above for secondary sphere formation.

Extreme limiting dilution analysis (ELDA) for mammosphere formation

Breast cancer cells were seeded in 96-well ultralow attachment plates at densities of 2, 4, 8, 16, 32 and 64 cells per well for the limited dilution experiment. The number of mammospheres formed in each well was counted after 10 days. To quantitatively analyze the results of the limited dilution assay, we employed ELDA (http://bioinf.wehi.edu.au/software/elda) to determine the mammosphere synthesis efficiency [29].

qRT-PCR detection and Western blot

Total RNA of cells or tissues was extracted using Trizol reagent and reverse transcribed into cDNA using a reverse transcription kit after the integrity was confirmed by electrophoresis. A SYBR Green kit was then employed for qRT-PCR experiment according to the manufacturer’s instructions. The specific primers were designed and listed in Table 1.

RIPA buffer was used to extract the total protein from the cells or tissues, which was then separated by SDS-PAGE and transferred to a PVDF membrane. Nonspecific antigen sites were inhibited with 5% skim milk and the membranes were incubated with primary antibodies at the suggested dilutions. Finally, ECL luminous fluid was used to track the protein signals. Gray values of the proteins were analyzed using Image J software and normalized to GAPDH.

IP, Co-immunoprecipitation (Co-IP) and IF

The protein A/G magnetic beads were used for IP and Co-IP detection. The IP, Co-IP and IF were performed with the specific primary antibodies (IP and Co-IP experiments: anti-TSP50, anti-p110α and anti-flag. IF experiment: anti-TSP50 1:1000, anti-p110α 1:1000, anti-β-catenin 1:1000, anti-NICD 1:1000) as described previously [12].

PI3K p110α activity assay

The total protein of the transfected breast cancer cells was extracted, and PI3K p110α was purified using the IP method. The activity of PI3K p110α was determined using a human PI3K elisa kit according to the manufacturer’s instructions.

Aldehyde dehydrogenase (ALDH) activity assay

The cells were placed on ice, followed by the detection of ALDH using an ALDH test kit, according to the manufacturer’s instructions. The ALDH activity was evaluated at 340 nm using a microplate reader by measuring NAD+production. Stronger activity was indicated by higher optical density (OD) value.

Flow cytometry

The transfected breast cancer cells were dispersed into individual cells and incubated with anti-CD44-PE and anti-CD24-APC antibodies for 30 min in the dark. The number of CD44⁺/CD24⁻ cells was analyzed by flow cytometry. The flow cytometry was also used for intracellular ADR pumping rate detection, as previously reported [30].

Colony formation

The transfected breast cancer cells (200 cells/well) were seeded in 6-well plates and cultured for 14 days. The medium was refreshed every 3 days. The formed colonies were fixed with 4% paraformaldehyde for 15 min, and subsequently, stained with 0.1% crystal violet for 15 min. After abstersion with PBS, the colonies were photographed. Furthermore, the crystal violet was completely eluted by 33% glacial acetic acid. The eluent OD value was measured at 570 nm using a microplate reader, which represents the clone-forming ability.

Wound healing

A pipette tip was used to generate a scratch in the monolayer of transfected breast cancer cells (approximately 100% confluence) for the wound healing assay. The adherent cells were then gently washed twice with PBS to wipe out the floating cells and cultured in serum-free medium. The migration distance of breast cancer cells at 0 and 48 h was observed using a microscope, and the wound closure rate was then evaluated.

Transwell invasion and migration

To measure cell invasive and migratory capabilities, the transwell chambers were pre-coated with or without Matrigel. The membrane of upper chamber was infiltrated with serum-free medium and plated with the transfected breast cancer cells. Meanwhile, the bottom of the chamber was supplemented with FBS-added culture medium. At 48 h post-incubation, the cells that crossed the membranes into the lower chamber were fixed with 4% paraformaldehyde for 15 min, and subsequently, stained with 0.1% crystal violet for 15 min. An inverted microscope was used to photograph the invasive and migratory cells. Finally, the crystal violet was completely eluted with 33% glacial acetic acid. The eluent OD value was measured at 570 nm using a microplate reader, which represented the number of invaded and migrated cells.

Mouse xenograft and drug treatment assays

For serial dilution assays, lentivirus-infected experimental group and control group mammospheres were digested with trypsin (2×10³–2×10⁶ cells) and injected subcutaneously into both flanks (MDA-MB-231 cells, left flank for shNC group and right flank for shTSP50 group) and fat pads (MCF7 cells, left fat pad for LV-Con group and right fat pad for LV-TSP50 group) of 6-week-old female BALB/C nude mice (n=2 in 2×10⁶ cell groups, n=6 in other groups), respectively. To analyze the effects of drugs on tumorigenesis, lentivirus-infected mammospheres were injected subcutaneously into both flanks (MDA-MB-231 cells, 1×10⁶ cells, left flank for shNC group and right flank for shTSP50 group) and fat pads (MCF7 cells, 1×10⁶ cells, left fat pad for LV-Con group and right fat pad for LV-TSP50 group) of the nude mice (n=6 in each group), respectively. SC79 was administered at 0.04 mg/g/d (i.p.) [31], LY294002 at 0.05 mg/g (i.p.), twice weekly [32], and BYL-719 at 50 mg/kg/d (i.g.) [33], after the injection of tumor cells. The tumors were weighed using an electronic balance and a vernier caliper was used for tumor size measurement. Furthermore, the volumes were estimated using the equation (a²×b)/2 (a, short diameter; b, long diameter). All animal studies were approved by the Animal Research Ethics Committee of Northeast Normal University (NENU/IACUC, AP20230315) of China and performed in accordance with the established guidelines.

Lung metastasis model assay

A total of twenty 4-week-old nude mice were randomly divided into four groups (n=5 in each group), and 2×10⁶ transfected MDA-MB-231 cells were collected and incubated with the Did solution for 15 min at 37 in the dark. The cells were then tail vein injected into the nude mice. Lung metastasis was measured by an in vivo imaging system after 4 weeks of treatment with 0.04 mg/g/d (i.p.) SC79. Lung tissues were collected for Haematoxylin–eosin (HE) staining.

IHC staining

Nude mouse tumor tissues were fixed, then paraffin embedded and sectioned into slices (5 μm). Tissue microarrays (HBreD136Su02) were obtained from Outdo BioTech (Shanghai, China), containing 136 cases of breast cancer specimens with the clinical data of patient survival, age, tumor volume, Ki67 level, axillary lymph node metastasis, pathologic stage, TNM stage, ER level, PR level and HER-2 level. After dewaxing and rehydration, IHC was performed as previously described [20].

Overexpression of TSP50 increases CSC-like phenotypes in breast cancer cells

We created stable TSP50 or negative control (NC) overexpressed MCF7 and ER7530 cells using lentivirus to evaluate the role of TSP50 in the regulation of CSC-like properties. Notably, when compared to NC cells, TSP50 overexpression dramatically increased the mRNA and protein expression levels of the BCSC-related markers OCT4, NANOG, ALDH1 and CD44 (Fig. 2A, Fig. S1A, Fig. S2A and B). Sphere-forming experiments showed that TSP50 overexpression significantly increased the number and diameter of spherical cells, as well as the frequency of sphere-forming cells, as detected by an in vitro limiting dilution assay (Fig. 2B-E, Fig. S2C-F). Self-renewal was then measured according to the secondary mammosphere formation assay and colony formation experiments. Consistent with our hypothesis, TSP50 significantly promoted secondary mammosphere formation, and colony formation experiments revealed a much higher number of colonies in TSP50 overexpressed breast cancer cells (Fig. 2F-J, Fig. S2G-K). Additionally, BCSCs are characterized by important cell-surface and functional markers, such as CD44⁺/CD24^−/low and high ALDH activity [34, 35]. In this study, the proportion of CD44⁺/CD24⁻ cells and ALDH activity in the TSP50 overexpressed cells were markedly increased compared with the control (Fig. 2K and L, Fig. S2L-N). Adriamycin (ADR) is a common chemotherapy drug for breast cancer, however, it is more likely to develop drug resistance during clinical treatment due to the presence of CSCs [36]. Also, ABCG2 and P-gp have been shown to be frequently expressed in human cancer and act as xenobiotics efflux pumps to transport various classical chemotherapeutic agents out of cells [37]. Interestingly, we found that TSP50 overexpression increased the ADR pumping rate and the expression levels of drug resistance proteins ABCG2 and P-gp (Fig. 2M and N, Fig. S2O and P). And the sensitivity of breast cancer cells to ADR was reduced (Fig. 2O and Fig. S2Q).

Open in a separate window

Fig. 2

Overexpression of TSP50 increases CSC-like phenotypes in MCF7 cells. A The protein expression levels of BCSC-related markers OCT4, NANOG, ALDH1 and CD44 in MCF7 cells stably overexpressing NC or TSP50. B Representative mammosphere images derived from the NC or TSP50-overexpressed MCF7 cells. Scale bar, 25 μm. C Number of primary mammospheres. D Size of primary mammospheres. E The linear regression plot generated by ELDA for in vitro limiting dilution assay with NC or TSP50 transfected MCF7 cells. F Representative mammosphere images derived from the NC or TSP50-overexpressed mammospheres. Scale bar, 25 μm. G Number of secondary mammospheres. H Size of secondary mammospheres. I, J Representative colony formation images derived from the stable TSP50 or NC overexpressed MCF7 cells and the number of colonies analysis results. K The CD44⁺/CD24⁻ cell subpopulation proportion analysis results by flow cytometry. L ALDH activity detection results. M–O The ADR pumping rate (M), expression levels of drug resistance proteins ABCG2 and P-gp (N) and IC50 values of ADR (O) detection results in MCF7 cells stably overexpressing NC or TSP50. N=3 biologically independent replicates. ^*p<0.05 and **p<0.01, as compared with control group by Student’s t-test

The limited dilution and tumorigenesis assay in vivo serves as the gold standard for CSCs testing. As shown in Fig. 3A-C, experiments using in vivo xenografts revealed that breast cancer cells overexpressing TSP50 had a considerably greater tumor-forming efficiency and a significantly larger tumor volume and weight than control cells. The qRT-PCR analysis results of tumor tissue lysates showed that the overexpression of TSP50 promoted the mRNA expression of BCSC-related markers NANOG, OCT4 and ALDH1 (Fig. S1B-E). Further IHC staining results proved that overexpressed TSP50 induced Ki67 expression in tumor tissues (Fig. 3D). These findings imply that the CSC-like properties of breast cancer cells are enhanced by TSP50.

Open in a separate window

Fig. 3

Overexpression of TSP50 promotes limited dilution tumorigenesis. A-C Stable overexpression of TSP50 or NC mammospheres were seeded into nude mice. The incidence of tumors was listed according to the number of seeded cells (A). Tumor volume (B) and weight (C) of each group. D Ki67 IHC analysis results in control or TSP50 overexpressed tumor tissues of nude mice. Scale bar, 50 μm. N=3 biologically independent replicates for IHC assay (2×10⁵ cell groups). N=2 biologically independent replicates for IHC assay (2×10⁶ cell groups). ^*p<0.05 and **p<0.01, as compared with the NC group by Student’s t-test

Inhibition of TSP50 attenuates CSC-like phenotypes in breast cancer cells

To further establish TSP50’s role in CSC-like properties maintenance of breast cancer cells, a lentivirus vector expressing shRNA for TSP50 knockdown was developed. Following TSP50 suppression, the expression of stemness genes was markedly downregulated in MDA-MB-231 and T47D cells (Fig. 4A, Fig. S1F, Fig. S3A and B). Smaller and fewer mammospheres were produced as a result of TSP50 suppression during primary and secondary mammosphere formation (Fig. 4B-H, Fig. S3C-I). Additionally, TSP50 knockdown breast cancer cells formed significantly fewer colonies than cells carrying an empty vector, as demonstrated by the colony formation assay (Fig. 4I and J, Fig. S3J and K). Meanwhile, TSP50 shRNA-stably transfected breast cancer cells revealed a lower CD44⁺/CD24⁻ cell ratio and ALDH activity (Fig. 4K and L, Fig. S3L-N). Furthermore, lower expression of TSP50 also decreased the ADR pumping rate, drug resistance protein levels, and the sensitivity of breast cancer cells to ADR was enhanced (Fig. 4M-O, Fig. S3O-Q). The subsequent in vivo results indicated that breast cancer cells with decreased expression of TSP50 had a lower tumor-forming efficiency, besides tumor volume and weight (Fig. 5A-C). Meanwhile, NANOG, OCT4 and ALDH1 mRNA and Ki67 protein levels were dropped in tumor tissues of the TSP50 knockdown group (Fig. 5D, Fig. S1G-J). These results suggest that the inhibition of TSP50 attenuates breast cancer CSC-like properties.

Open in a separate window

Fig. 4

Inhibition of TSP50 attenuates CSC-like phenotypes in MDA-MB-231 cells. A The protein expression levels of BCSC-related markers OCT4, NANOG, ALDH1 and CD44 in MDA-MB-231 cells transfected with TSP50 shRNA (shTSP50) and control (shNC). B Representative mammosphere images of MDA-MB-231 cells transfected with shNC or shTSP50. Scale bar, 25 μm. C Primary mammospheres number. D The size of primary mammospheres. E The linear regression plot generated by ELDA for in vitro limiting dilution assay with MDA-MB-231-shNC or MDA-MB-231-shTSP50 cells. F Representative mammosphere images derived from the mammospheres transfected with shNC or shTSP50. Scale bar, 25 μm. G Secondary mammospheres number. H The size of secondary mammospheres. I, J Representative colony formation images derived from the shNC or shTSP50 transfected MDA-MB-231 cells and the number of colonies analysis results. K The CD44⁺/CD24⁻ cells subpopulation proportion analysis results by flow cytometry. L ALDH activity detection results. M–O The ADR pumping rate (M), expression levels of drug resistance proteins ABCG2 and P-gp (N) and IC50 values of ADR (O) detection results in shNC or shTSP50 transfected MDA-MB-231 cells. N=3 biologically independent replicates. ^*p<0.05 and **p<0.01, as compared with control group by Student’s t-test

Open in a separate window

Fig. 5

TSP50 knockdown inhibits limited dilution tumorigenesis. A-C Stable TSP50-knockdown and shNC mammospheres were seeded into nude mice. The incidence of tumors was listed according to the seeded cell number (A). Tumor volume (B) and tumor weight (C) for each group. D Ki67 IHC analysis results in control or TSP50 knockdown tumor tissues of nude mice. Scale bar, 50 μm. N=3 biologically independent replicates for IHC assay (2×10⁵ cell groups). N=2 biologically independent replicates for IHC assay (2×10⁶ cell groups). ^*p<0.05 and **p<0.01, as compared with shNC group by Student’s t-test

TSP50 promotes breast cancer cell metastasis and EMT

Metastasis is the primary cause of mortality in breast cancer patients. Therefore, we investigated TSP50’s role in breast cancer metastasis. According to the results of investigations on wound healing, breast cancer cells that overexpressed TSP50 repaired wounds more quickly than the control cells (Fig. 6A and B, Fig. S4A and B). However, TSP50 knockdown breast cancer cells showed the opposite behavior (Fig. 6C and D, Fig. S4C and D). Studies on migration and invasion revealed that, in contrast to TSP50 knockdown, TSP50 overexpression significantly boosted both cell migration and invasion (as measured by the Transwell assay and Matrigel experiment, respectively) (Fig. 6E-L, Fig. S4E-L). Moreover, studies have revealed that EMT cells have CSC-like properties, and CSCs exhibit a mesenchymal-like phenotype [38]. We further investigated the role of TSP50 in EMT regulation of breast cancer cells. Western blot analysis was used to examine the expression of EMT-related markers, and the results showed that when TSP50 was overexpressed in breast cancer cells, E-cadherin protein level was dramatically reduced, whereas MMP9, Slug and Snail protein levels were significantly elevated (Fig. 6M, Fig. S4M). In contrast to control cells, TSP50 knockdown breast cancer cells showed significantly increased E-cadherin protein level and significantly decreased MMP9, Slug and Snail protein levels (Fig. 6N, Fig. S4N). However, variations in mRNA levels of the EMT-related markers were inconsistent (Fig. S1K and L, Fig. S4O and P). Furthermore, we evaluated the expression levels of EMT-related markers in tumor tissues formed by limited dilution and tumorigenesis assay using breast cancer cells with TSP50 stably overexpressed or knocked down. As shown in Fig. S1M-T, the alterations in the expression levels of EMT-related markers in tumor tissues formed by TSP50 overexpression or knockdown of breast cancer cells matched our cell-based detection findings. These results demonstrate that TSP50 promoted the metastasis and EMT of breast cancer cells (Fig. 6O).

Open in a separate window

Fig. 6

TSP50 promotes breast cancer cell metastasis and EMT. A, B Wound healing of MCF7 cells stably overexpressing NC or TSP50. C, D Wound healing of MDA-MB-231 cells transfected with shNC or shTSP50. E, F Migration of MCF7 cells stably overexpressing NC or TSP50. G, H Migration of MDA-MB-231 cells transfected with shNC or shTSP50. I, J Invasion of MCF7 cells stably overexpressing NC or TSP50. K, L Invasion of MDA-MB-231 cells transfected with shNC or shTSP50. M The protein expression of EMT-related markers E-cad, MMP9, Slug and Snail in MCF7 cells stably overexpressing NC or TSP50. N The protein expression of EMT-related markers E-cad, MMP9, Slug and Snail in MDA-MB-231 cells transfected with shNC or shTSP50. O Schematic diagram of TSP50, which promotes the migration, invasion and EMT of breast cancer cells. Scale bar, 25 μm. N=3 biologically independent replicates. Student’s t-test statistical analysis was used (**p<0.01)

TSP50 and PI3K p85α regulatory subunit competitively bind with PI3K p110α catalytic subunit to enhance its catalytic activity

The EMT of breast cancer cells is regulated by a variety of signaling pathways, including Wnt/β-catenin [39], Notch [40], Hedgehog [41] and PI3K/AKT [42], which have been confirmed to be involved in CSC-like properties maintenance. Therefore, we examined the effects of TSP50 on the activation of these pathways. The IF assay results showed that overexpression or knockdown of TSP50 did not affect the nuclear translocation of β-catenin (Wnt/β-catenin signal factor) or NCID (Notch signal factor) in breast cancer cells (Fig. S5A-D). Additionally, the expression levels of SHH and PTCH1, two essential components of the Hedgehog pathway, were also not regulated by TSP50 (Fig. S5E and F). These results indicate that TSP50 does not mediate the activation of the Wnt/β-catenin, Notch or Hedgehog pathways. However, we found significantly altered p-AKT levels in response to the overexpression or suppression of TSP50 in breast cancer cells, with constant AKT levels, indicating that PI3K/AKT signaling is stimulated by TSP50 (Fig. 7A and B, Fig. S5G and H).

Open in a separate window

Fig. 7

TSP50 and PI3K p85α regulatory subunit competitively bind with PI3K p110α catalytic subunit to enhance its catalytic activity. A, B Western blot detection results of PI3K/AKT signal-related markers expression in MCF7 cells stably overexpressing NC or TSP50 and MDA-MB-231 cells transfected with shNC or shTSP50. C, D MCF7 and MDA-MB-231 cells were harvested and subjected to Co-IP with anti-TSP50 antibody, followed by Western blot analysis with anti-AKT, anti-p85α or anti-p110α antibody. E, F MCF7 and MDA-MB-231 cells were harvested and subjected to Co-IP with anti- p110α antibody, followed by Western blot analysis with an anti-TSP50 antibody. G MCF7 and MDA-MB-231 cells were fixed and subjected to IF analysis. TSP50 co-localizes with p110α in MCF7 and MDA-MB-231 cells. Scale bar, 50 μm. H The five domains in the p110α subunit. I In vitro Co-IP analysis of different domains of p110α (p85BD, RBD, C2, helical, and catalytic domains) with Flag-TSP50. The p85BD domain of p110α interacts with TSP50. J, K The p110α was purified from MCF7 cells stably overexpressing NC or TSP50 and MDA-MB-231 cells transfected with shNC or shTSP50. The catalytic activity of p110α was measured. L, M MCF7 cells stably overexpressing NC or TSP50 and MDA-MB-231 cells transfected with shNC or shTSP50 were harvested and subjected to Co-IP with anti-p110α antibody, followed by Western blot analysis with anti-p85α and anti-TSP50 antibodies. N A model of TSP50 regulating PI3K/AKT signaling through interaction with p110α. N=3 biologically independent replicates. Student’s t-test statistical analysis was used. **p<0.01

We further investigated how TSP50 activated PI3K/AKT signaling. Studies have shown that PTEN is an important regulatory factor in PI3K/AKT signal [43]. In this study, Western blot analysis results showed that PTEN expression remained stable after TSP50 was overexpressed or knocked down in breast cancer cells (Fig. 7A and B, Fig. S5G and H). PI3Ks are a family of lipid kinases (Class I, II and III), among which class I PI3K is the most widely studied. Class I PI3K can be formed by different catalytic subunits, namely p110α, p110β, p110γ or p110δ, which are involved in the production of phosphatidylinositol 3-phosphate (PI3P) [44]. The most characterized catalytic and regulatory subunits of PI3K, p110α and p85α, are closely associated with cancer progression and tumorigenesis [45]. The p110α kinase, which is involved in the production of PI3P from phosphatidylinositol 4,5-biphosphate (PI2P), activates the AKT signaling cascade [46]. We speculated that TSP50 might bind with the aforementioned PI3K/AKT key factors, such as p85α, p110α and AKT, which were selected for Co-IP analysis to identify potential target molecules regulated by TSP50. The results showed that endogenous TSP50 interacted with p110α rather than with p85α or AKT (Fig. 7C-F, Fig. S5 I and J). As confirmed by IF, TSP50 and p110α were colocalized in the cytoplasm of MCF7 and MDA-MB-231 cells (Fig. 7G). The p110α subunit consists of p85 binding domain (p85BD), Ras-binding domain (RBD), C2, Helical and Catalytic domains (Fig. 7H). The MyC-tagged domain plasmids were synthesized and co-transfected with Flag-TSP50 overexpression plasmid into MDA-MB-231 cells for Co-IP analysis. Our results showed that the p85BD was responsible for the interaction of p110α with TSP50 (Fig. 7I). Next, the p110α protein was purified from breast cancer cells with TSP50 overexpressed or knocked down using an IP assay, and its activity was then determined. We found that compared with the control group, p110α activity was significantly increased when TSP50 was overexpressed, while decreased markedly in TSP50 knockdown breast cancer cells (Fig. 7J and K). It is confirmed that the increased disruption of p110α binding to the inhibitory subunit p85α relieves its catalytic inhibition, and increases p110α membrane lipid binding [47]. Intriguingly, we have demonstrated that TSP50 binds to p110α through p85BD, however, p85α was not detected in the complex obtained using Co-IP with anti-TSP50 (Fig. 7C, D and I). On the basis of the findings, we hypothesized that TSP50 might enhance PI3K activity by reducing complex formation between p85α subunit and p110a. To verify this hypothesis, we analyzed the effect of TSP50 on the p110α-p85α interaction. When TSP50 was overexpressed, the interaction between p110a and p85α was reduced, and the level of TSP50 bound to p110a increased significantly in the complex obtained by Co-IP assay. Meanwhile, an enhanced p110α-p85α interaction was found in TSP50 knockdown cells compared with the shNC group, accompanied with a reduced TSP50 binding level to p110α (Fig. 7L and M, Fig. S5 K and L). Therefore, these data indicate that TSP50 positively modulates PI3K activity by preventing p85α from binding to p110a (Fig. 7N).

The TSP50 D206A mutation enhances the binding ability of p85α with p110α, thereby inhibiting the activity of p110α

Our previous study has shown that the catalytic triad of TSP50 is essential for its function in cell proliferation, and mutants (H153A, D206A or T310A) in the catalytic triad abolish the enzyme activity of TSP50 [48]. Therefore, we investigated the effects of different TSP50 mutants on p110α activity and found that p110α activity was significantly decreased only in TSP50 D206A mutated breast cancer cells compared with the TSP50 wild-type group (Fig. 8A and B). Further analysis results revealed that TSP50-p110α binding was weakened by the D206A mutation in TSP50 (Fig. 8C-F). These results indicate that the TSP50 D206A mutation may affect its binding ability with p110α through a conformational change rather than an activity alteration in TSP50. We also analyzed the effect of TSP50 D206A mutant on the p110α-p85α interaction, and the results showed that p110α and p85α interacted more favorably in breast cancer cells with the TSP50 D206A mutation compared with the TSP50 wild-type breast cancer cells (Fig. 8G and H), further verifying the mechanism by which TSP50 activates PI3K/AKT signal.

Open in a separate window

Fig. 8

TSP50 D206A mutation enhances the binding ability of p85α with p110α. A, B The p110α was purified from MCF7 and MDA-MB-231 cells transfected with wild-type TSP50, H153A, D206A or T310A mutant. The catalytic activity of p110α was measured. C, D MCF7 and MDA-MB-231 cells transfected with Flag-TSP50 D206A mutant were harvested and subjected to Co-IP with anti-Flag antibody, followed by Western blot analysis with anti-p110α antibody. E, F MCF7 and MDA-MB-231 cells transfected with Flag-TSP50 wild-type or Flag-TSP50 D206A mutant were harvested and subjected to Co-IP with an anti-Flag antibody, followed by Western blot analysis with the anti-p110α antibody. G, H MCF7 cells and MDA-MB-231 cells transfected with Flag-TSP50 wild type or Flag-TSP50 D206A mutant were harvested and subjected to Co-IP with anti-p110α antibody, followed by Western blot analysis with anti-p85α and anti-Flag antibodies. N=3 biologically independent replicates. Student’s t-test or one-way ANOVA statistical analysis was used. ^*p<0.05 and **p<0.01

TSP50-mediated PI3K/AKT signaling activation is essential for maintaining CSC-like properties, EMT and metastasis in breast cancer cells

We treated cells with the p110α inhibitor BYL-719, AKT inhibitor LY294002, or AKT agonist SC79 to determine whether TSP50 promotes breast cancer CSC-like traits, EMT and metastasis through the PI3K/AKT signaling pathway. As shown in Fig. 9A, S6A, S7A and B, BYL-719 or LY294002 treatment inhibited the phosphorylation level of AKT and partly reversed the TSP50-induced promotion of AKT phosphorylation and CSC-like marker expression. BYL-719 or LY294002 treated cells exhibited a decrease in sphere formation efficiency (both number and diameter), the number of colonies, the proportion of CD44⁺/CD24⁻ cells, ALDH activity and ADR pumping rate, even in the presence of TSP50 (Fig. 9B-G, Fig. S8A and B, Fig. S6B-G, Fig. S8C and D, Fig. S7C-G, Fig. S8E and F). Treatment with BYL-719 or LY294002 also reversed the decreased expression of E-cadherin and increased expression of MMP9, Slug and Snail induced by TSP50 (Fig. 9H, Fig. S6H and Fig. S7H and I), and impaired TSP50-induced cell migration and invasion (Fig. 9I and J, Fig. S9A and B, Fig. S6I and J, Fig. S9C and D, Fig. S7J and K, Fig. S9E and F). In addition, SC79 enhanced the phosphorylation level of AKT, meanwhile, SC79 treatment restored AKT phosphorylation and CSC-like marker protein levels (Fig. 10A, Fig. S10A), CSC-like properties (Fig. 10B-G, Fig. S11A and B, Fig. S10B-G, Fig. S11C and D), EMT level (Fig. 10H, Fig. S10H), cell migration and invasion activity (Fig. 10I and J, Fig. S12A and B, Fig. S10I and J, Fig. S12C and D) and tumor metastasis in vivo (Fig. 10K) after TSP50 knockdown in breast cancer cells. These results reveal that TSP50 increases breast cancer CSC-like traits, EMT and metastasis partially through the PI3K/AKT signaling pathway.

Open in a separate window

Fig. 9

Inhibition of PI3K/AKT signaling partially reversed TSP50-mediated CSC-like properties, EMT and metastasis maintenance in MCF7 cells. TSP50 or NC stably overexpressed MCF7 cells were treated with BYL-719. A The levels of indicated BCSC-related markers were determined by Western blot. B, C Primary mammosphere number and size were calculated. D The number of colonies analysis results. E The subpopulation proportion analysis results of CD44⁺/CD24⁻ cells. F ALDH activity detection results. G The ADR pumping rate. H The levels of EMT-related markers were determined by Western blot. I Cell migration detection results. J Cell invasion detection results. N=3 biologically independent replicates. Student’s t-test or one-way ANOVA statistical analysis was used. ^*p<0.05, **p<0.01 and ns, no significance

Open in a separate window

Fig. 10

PI3K/AKT signaling activation partially restored the CSC-like properties, EMT and metastasis inhibited by TSP50 knockdown. The shNC or shTSP50 transfected MDA-MB-231 cells were treated with SC79. A The levels of indicated BCSC-related markers were determined by Western blot. B, C Primary mammospheres number and size were calculated. D The number of colonies analysis results. E The subpopulation proportion analysis results of CD44⁺/CD24⁻ cells. F ALDH activity detection results. G The ADR pumping rate. H The levels of EMT-related markers were determined by Western blot. I Migration assay detection results. J Cell invasion detection results. K Bioluminescence images of lung metastasis in mice that were injected with cells through the tail veins as indicated, and the metastasis was quantified by measuring the photo flux. Macroscopically or HE staining was used to examine lung metastatic nodules. The red arrows denoted the metastatic nodules. N=3 biologically independent replicates. Student’s t-test or one-way ANOVA statistical analysis was used. ^*p<0.05, **p<0.01 and ns, no significance

PI3K/AKT signaling activation is crucial for TSP50 tumor-promoting effects in vivo

To explore the effect of TSP50 on pathological progression of breast cancer, an in vivo tumor formation assay was performed (n=6/group). The tumor weight, volume, Ki67 and p-AKT levels were higher in TSP50 stable overexpression group than in the control group (Fig. 11A-F, Fig. S13A-F). As expected, BYL-719 or LY290042 treatment reversed the tumorigenic effects of TSP50 and the TSP50-induced promotion of AKT phosphorylation in tumor tissues (Fig. 11A-F, Fig. S13A-F). In contrast, in MDA-MB-231-derived mammospheres, TSP50 knockdown significantly inhibited tumor formation, accompanied with a dramatic reduction in Ki67 and p-AKT levels (Fig. 11G-L). Xenografted mice treated with SC79 reversed the tumorigenesis inhibition mediated by TSP50-knockdown through enhancing the phosphorylation level of AKT (Fig. 11G-L).

Open in a separate window

Fig. 11

TSP50 promotes tumorigenesis through PI3K/AKT signaling in vivo. A-F The nude mice were injected with NC or TSP50 stably overexpressed mammospheres and then treated with BYL-719. Photography of xenograft tumor (A), tumor weight (B), tumor volume (C), Ki67 level (D, E) and PI3K/AKT signal-related markers levels (F). G-L The nude mice were injected with stable shNC or TSP50-knockdown mammospheres and then treated with SC79. Photography of xenograft tumor (G), tumor weight (H), tumor volume (I), Ki67 level (J, K) and PI3K/AKT signal-related markers levels (L). Scale bar, 50 μm. N=3 biologically independent replicates. Student’s t-test statistical analysis was used. ^*p<0.05, **p<0.01 and ns, no significance

TSP50 expression is positively correlated with p-AKT and ALDH1 levels in human breast tumor tissues

Based on the finding that TSP50 regulates PI3K/AKT signaling and promotes CSC-like phenotypes in breast cancer cells, we investigated whether TSP50 is correlated with the levels of p-AKT and ALDH1 in breast cancer specimens. The expression levels of TSP50, p-AKT and ALDH1 were examined in 136 breast cancer tissues using IHC (Fig. 12A). As expected, p-AKT and ALDH1 protein levels in 136 breast cancer specimens increased significantly with increasing TSP50 expression (p-AKT and TSP50, r=0.42, p<0.01; ALDH1 and TSP50, r=0.46, p<0.01). In particular, a significant proportion of the samples with high TSP50 expression also exhibited elevated levels of p-AKT (40%, 27/68) and ALDH1 (44%, 30/68) (Fig. 12B and C). Accordingly, lower levels of p-AKT and ALDH1 were found in 41% (28/68) and 37% (25/68) of the samples with low TSP50 expression (Fig. 12B and C). Collectively, these findings provide compelling evidence that TSP50 expression positively correlates with p-AKT and ALDH1 protein levels in human breast cancer.

Open in a separate window

Fig. 12

TSP50 expression is positively correlated with p-AKT and ALDH1 levels, and the TSP50/p-AKT/ALDH1 positive relationship is a potential diagnostic index for human breast cancer. A TSP50, p-AKT and ALDH1 expression levels were analyzed using IHC in human breast cancer tissues (n=136). Scale bar, 100 μm. B, C The expression levels of p-AKT and ALDH1 were positively correlated with TSP50 expression levels. Cancer tissues with high TSP50 expression (n=68) and low TSP50 expression (n=68) were analyzed using IHC. D-S The relationship analysis results between TSP50/p-AKT/ALDH1 expression and patient survival time (D), tumor volume (E), Ki67 level (F, G), age (H), axillary lymph node metastasis (I), pathologic stage (J), TNM stage (K-M), ER level (N, O), PR level (P, Q) and HER-2 level (R, S) in human breast cancer tissues. Spearman, Chi-square or one-way ANOVA statistical analysis was used. ^*p<0.05, **p<0.01 and ns, no significance

We sincerely thank the team members for their dedication to this project.