Characterization of a Novel MAP4K4-SASH1 Kinase Cascade Regulating Breast Cancer Tumorigenesis and Metastasis


 Background: The SAM and SH3 domain containing protein 1(SASH1) was previously described as a candidate tumor-suppressor gene in breast cancer and colon cancer to mediate tumor metastasis and tumor growth. However，the underlying mechanisms by which SASH1 implements breast cancer tumorigenesis and the question why SASH1 is downregulated in most solid cancers remain unexplored. Methods: The expression and clinical relevance of SASH1 and mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4) were analyzed by immunohistochemistry(IHC). The bindings of SASH1 and MAP4K4 were investigated by pull down assay, nano-flow LC-MS/MS , immunoprecipitation- western blot(IP-WB) and immunofluorescence(IF). The knockdown of SASH1- and/or MAP4K4-induced cell proliferation, apoptosis, migration and invasion were investigated by flow cytometry and transwell assays. The cell proliferation and apoptosis treated by GNE-495 were assessed by flow cytometry. The in vitro regulation relationship between SASH1 and MAP4K4 were identified by in vitro kinase assay. The functional effects of the silencing of SASH1 and/or MAP4K4 on tumorigenesis and metastasis of T47D cells-xenograft tumors were assessed by HE staining and IHC .Results: SASH1 and MAP4K4 was significantly downregulated in hormone- dependent subtypes of breast cancers and SASH1 was significantly correlated with MAP4K4 in hormone-dependent subtypes. SASH1 is identified to be a novel serine/threonine protein kinase that binds to MAP4K4, and is phosphorylated by MAP4K4. SASH1 and MAP4K4 synergistically regulate cell proliferation，migration and invasion of hormone-dependent breast cancer cells and mediate liver and lung metastasis of T47D cells-xenograft tumors.Ser801 in MAP4K4 might be the serine site for MAP4K4 to phosphorylate SASH1. GNE-495， a MAP4K4-specific inhibitors by upregulating the expression of MAP4K4 and SASH1 might be a potential reagent in treating hormone-dependent breast cancers. Conclusion: Our findings characterize a novel MAP4K4-SASH1 kinase cascade to mediate the tumorigenesis and metastasis of breast cancer, which can be of targeted intervention by GNE-495.

Background SASH1 was originally identi ed as a candidate tumor-suppressor gene in breast cancer and colon cancer, regulating tumorigenesis of breast and other solid cancers and the adhesive and migratory behavior of cancer cells in tumor formation [1,2]. A growing number of reports have con rmed the tumor suppressor roles of SASH1 in breast and other solid cancers [3][4][5][6][7][8]. Treatment of breast cancer cells treated with chloropyramine results in higher SASH1 protein levels [9]. SASH1 was recently suggested to be potential therapeutic target for cancer [10]. SASH1 was con rmed to bind to 14-3-3 proteins and phospho-Ser90 of SASH1 provides the 14-3-3-binding sites to promote SASH1 phosphorylation by phosphatidylinositol 3kinase and MAPK/p90RSK signaling [11]. And the most recent study has indicated that SASH1 was phosphorylated by LATS1 [12].
MAP4K4, is also known as a hepatocyte progenitor kinase-like/germinal center kinase-like kinase (HGK).HGK is a member of the human STE20/ mitogen-activated protein kinase kinase kinase kinase family of serine/threonine kinases and is the ortholog of mouse NIK (Nck-interacting kinase). Strikingly increasing rates of invasion and morphogenesis are induced by enhanced HGK kinase activity through overexpression [13]. MAP4K4 overexpresses in many types of human cancer [14][15][16]. MAP4K4 is commonly overexpressed in TNBC(triple negative breast cancer)-breast tumor-initiating cells(BTICs) and suggested to be a common regulator in TNBC [17]. The present study indicates that SASH1 is identi ed to be a novel protein kinase interacts with MAP4K4. MAP4K4 and SASH1 coordinately mediate tumorigenesis and metastasis of breast cancer in hormone-dependent breast cancer cell lines and xenograft tumors. SASH1 is phosphorylated and regulated by MAP4K4, which forms a novel MAP4K4-SASH1 phosphokinase cascade and this cascade can be inhibited by GNE-495 .

Materials And Methods
Collection of breast cancer tissues and human breast carcinoma tissue arrays.
Fresh primary breast cancer tissues from breast carcinoma patients undergoing resection and normal tissues were collected between July 2015 and May 2020 at the Chongqing Cancer Hospital. A set of human breast cancer tissue arrays and a set of matched adjacent tissue arrays of human breast cancer were provided by Shanghai Biochip(Shanghai, China).
Antibodies, cell lines, recombinant DNA or shRNA and siRNAs, and primers for gene cloning and sitedirected mutagenesis and siRNA construction.
The detailed information on chemical reagents, antibodies, cell lines, BALB/c-nude mice, recombinant DNA or shRNA and siRNAs, and primers for gene cloning, shRNA silencing of lentivirus, site-directed mutagenesis , siRNA construction used in this study is indicated in Table S1.
Gene cloning SASH1-Pegfp-C3 vectors were constructed as described previously [18]. MAP4K4 cDNA was cloned into Pbabe-Flag-puro and Pegfp-C3 using KpnI and XhoI restriction sites. SASH1 mutant plasmids were constructed as described previously [18]. Full-length cDNA of SASH1 was cloned into the pSec Tag vector using XhoI and HindIII restriction sites to construct His-SASH1-pSec Tag vector.
Human breast cancer cell lines and HEK-293T cells were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai China). After passages, cells were transfected recombined vectors. Cells were transfected with two unique and e cient SASH1-shRNA or MAP4K4-shRNA constructs in a GFP vector (Origene) using PEI(prepared by us) according to our previous reports [18,19]. The sequences used for silencing of SASH1-shRNA vector, silencing of MAP4K4-shRNA vector, silencing of MAP4K4-shRNA lentivirus and silencing of SASH1-shRNA lentivirus were shown in Table S1.
Pull down assay, Nano-ow LC-MS/MS and Database search The Pull down assay, nano-ow LC-MS/MS and the database search were constructed as previously described. The binding partners of SASH1 and the phosphorylation sites on SASH1 in SK-BR-3 cells were identi ed with nano-LC MS/MS analysis on an HPLC system as previously described [20].
Most of western blot were mainly performed as our previous reports [18,19].The primary antibodies used for western blotting were as listed in Table S1.
Cell cycle and apoptosis assays Transfected cells were harvested , xed , stained and analyzed by ow cytometry (Navious, Beckman Coulter) for cell cycle pro le determination [21]. Apoptosis was measured by staining with 7AAD/APC.
The results were analyzed using Flowjo software (CT, USA).

Transwell migration and Invasion assay
Transwell migration assay and invasion assays were performed as previously described [22].

Immuno uorescence
Immuno uorescence (IF) was performed as previously described by us [20].

In vitro Kinase assay
For the SASH1 kinase assay, HEK-293T cells were transfected with GFP-MAP4K4 and treated with 150 nM GNE-495 for 24hr and MAP4K4 was immunoprecipitated. His-SASH1 was also introduced into HEK-293T cells and at 36hr after transfection, His-SASH1 was immunoprecipitated with His Ab.The puri ed His-SASH1 was used as the substrate for the immunoprecipitated MAP4K4. The protocol of this study were mainly referred to the previous report [23] and the detailed protocol of In vitro kinase assay was indicated in the Supplementary Materials and Methods .

Immunohistochemical staining and quanti cation
Tissue sections (5μm) were dehydrated and subjected to peroxidase blocking. of human breast cancer tissues, normal breast tissues, corresponding adjacent tissues and T47D cells-xenograft tumors were immunohistochemically stained with anti-SASH1, anti-MAP4K4, anti-phospho-MAP4K4, anti-Ki67, anti-MMP2 anti-caspase9 and anti-MMP9. Detailed information on these antibodies can be found in Table   S2. The IHC analyses results of p53 and the breast cancer patients' information including age, tumor size, tumor location, tumor stage, node status and clinical stage were provided by the manufacturer of the breast cancer tissue arrays. IHC were mainly performed as previously described [18]. The staining intensity of SASH1, MAP4K4, phospho-MAP4K4, Ki67, MMP2 and MMP9 of the positive cells, and the positive areas' percentage of the positive cells , and total scores of each visual eld were calculated as previously described [18].

Xenograft
Animals were housed and maintained under speci c pathogen-free experimental animal center. T47D stable cells containing SASH1-shRNA or/and were suspended in Matrigel (BD)-DMEM (5 ×10 6 ) and injected subcutaneously into the right side fossa axillaris of each 5-to 6-week-old female BALB/c nude mice in each group. Tumor volume was calculated and tumor diameter was measured. All mice were sacri ced 3 weeks after observation, and tumor weights were measured.

Statistical analyses
Data are presented as mean ± SD. Pearson's chi-squared Tests was used to analyze the relationship between IHC scores of the relevant proteins and between clinicopathological parameters and expression levels of SASH1 and p53. The statistical signi cance of differences was assessed using SPSS16.0 software. The densitometry values of protein bands on western blot were analyzed using a homogeneity of variance test and one-way ANOVA was used for multiple comparisons using the least signi cant difference in SPSS16.0 to generate the required P-values. P-values of less than 0.05 were considered statistically signi cant. The cartograms were made and plotted using GraphPad Prism 5 (GraphPad, Inc, CA, USA). Additional methods can be found in the Supplementary Materials and Methods.

Results
SASH1 is down-regulated in four breast cancer subtypes and its expression is associated with the pathological diagnosis index and receptor expression of breast cancer.
To examine the expression of SASH1 in different breast cancer subtypes, 284 carcinoma specimens of four breast cancer subtypes including 92 luminal A, 105 luminal B, 45 HER2 positive and 42 TNBC specimens, and 79 normal breast tissues including tissues of breast broadenoma and hyperplasia of mammary glands were collected to compare SASH1 expression in normal breast tissues and breast cancer tissues. Positive SASH1 staining was found in both ductal and lobular epithelial cells. SASH1 expression was observed predominantly in the cytoplasm (Fig. 1A). SASH1 was signi cantly downregulated in all four subtypes as compared with normal breast tissues( Fig. 1A and Fig. 1B). SASH1 staining of human breast carcinoma tissue arrays indicated that SASH1 was downregulated in four subtypes as compared with the matched breast tissues ( Fig. 1C and Fig. 1D). Immunoblot analyses indicated that in 18 benign breast tissues and 18 cancer tissues, SASH was downregulated in different subtypes of breast cancer (Fig. 1G). Statistical analyses based on IHC staining scores of SASH1 and Ki67 in human breast carcinoma tissue arrays suggested that SASH1 expression was signi cantly related to that of Ki67 in subtypes of luminal and Her2 positive, however, not associated with that of Ki67 in TNBC subtype (Fig. 1E). The relationship between SASH1 expression and p53 expression obtained by the human breast cancer tissue array was assessed by Pearson's chi-squared test. SASH1 expression was signi cantly related to p53 in subtypes of luminal and Her2 positive, however, not related to Ki67 in TNBC subtype (Fig. 1F). SASH1 downregulation was signi cantly associated with early tumor stages,early lymph node metastasis, and early clinical stage; however, it was not correlated with tumor size (Table S2).
We also analyzed the relationship of SASH1 expression with ER, PR, EGFR, p53 and CK5/6 in human breast carcinoma tissue arrays. Pearson's χ2 test based their IHC scores and their positive rate indicated that SASH1 expression was inversely correlated with the ER of nucleus and cytoplasm( Fig.S1A and S1B). SASH1 expression was negatively correlated with the nucleic and cytoplasmic PR ( Fig.S1C and S1D).However, SASH1 expression was positively correlated with the nucleic HER2 (Fig.S1E). SASH1 expression was positively correlated with the nucleic and cytoplasmic p53 ae well as Ki67 (Fig.S1F, Fig.S1G and Fig.S1H).However, SASH1 expression was not correlated with the EGFR of cytoplasm and cytoplasm membrane( Fig.S1I and S1J) and CK5/6( Fig.S1K).

SASH1 is associated with MAP4K4.
To investigate the molecular network in which SASH1 involved, pull down assays and LC-MS/MS spectrometry analysis were performed. Our results indicated that SASH1 may interact with MAP4K4 in stable SK-BR-3 cells. MAP4K4 was shown to have a high possibility to bind to SASH1 ( Fig. 2A), and the interacting peptide sequences of SASH1 and MAP4K4 were identi ed (Fig. 2B). IP-WB analysis further con rmed that endogenous SASH1 was associated with endogenous MAP4K4 (Fig. 2C). We also found exogenous SASH1 binds to endogenous MAP4K4 using IP with anti GFP antibody. As demonstrated in Fig. 2D, exogenous SASH1 was shown to immunoprecipitate with endogenous MAP4K4 in SK-BR-3 cells. We further identi ed the binding domains of SASH1 to MAP4K4. Deleted SASH1 constructs were created, transfected into HEK-293T cells and immunoprecipitated with anti-GFP antibody. The associated endogenous MAP4K4 were identi ed by IP-WB. Compared with the binding of full-length SASH1 to endogenous MAP4K4, the N-terminal domain of SASH1 (SASH1-ΔC terminal) was shown to bind to endogenous MAP4K4 ( Fig. 2F and Fig. 2G). In addition, we observed co-localization between endogenous SASH1 and endogenous MAP4K4 in T47D and MCF-7 cells ( Fig. 2H and Fig. 2I).
SASH1 expression is positively correlated with MAP4K4 expression and downregulation of SASH1 is positively associated with increased phospho-MAP4K4 levels in breast cancer.
Since SASH1 was shown to bind to MAP4K4, the expression of MAP4K4 and the relationship between SASH1 and MAP4K4 levels in breast cancers were investigated. A total of 284 carcinoma specimens of breast cancer and 79 normal breast tissues including tissues of breast broadenoma and hyperplasia of mammary glands were collected to analyze the relationship between SASH1 and MAP4K4 in normal breast tissues and breast cancer tissues. MAP4K4 expression was detected predominantly in the cytoplasm and on the plasma membrane in some cases. In four breast cancer subtypes especially the luminal and HER2 positive subtypes as well as normal breast tissues,SASH1 was signi cantly associated with MAP4K4. Statistical analysis of IHC scores of SASH1 and MAP4K4 revealed that SASH1 was obviously correlated with MAP4K4 in normal breast tissues (Fig. 3A). In the subtypes of luminal A, luminal B and HER2 positive (Fig. 3B, Fig. 3C and Fig. 3D) SASH1 was obviously correlated with MAP4K4. However,SASH1 was also correlated with MAP4K4 in TNBC subtype (Fig. 3E). Although MAP4K4 mRNA levels in TNBC cells were higher than in non-TNBC cells MAP4K4 protein levels were not higher than in non-TNBC cells [17]. Statistical analysis of IHC scores indicated that MAP4K4 expression of 284 breast carcinoma specimens was lower than that of 79 normal breast tissues( Fig. 3G and Fig. 3H). Western blot analysis indicated that in 10 pairs of breast cancer tissues and normal breast tissues, in most breast cancer specimens downregulation of MAP4K4 was accompanied by decreased SASH1 levels ( Fig. 3I). In addition, statistical analysis of IHC scores indicated that SASH1 expression was correlated with phospho-MAP4K4 levels in 75 breast cancer tissues (Fig. 3J). These IHC results revealed that in breast cancer tissues SASH1 expression is correlated with MAP4K4 expression and increased phospho-MAP4K4 levels are correlated with SASH1 expression.
MAP4K4 and SASH1 co-regulate proliferation, migration and invasion of luminal subtype and HER2 positive subtype cell lines.
Since SASH1 expression is correlated with MAP4K4 expression in the luminal subtype and the HER2 positive subtype, we further assessed their roles in breast cancer tumorigenesis. We rst performed lossof-function studies using lentiviruses expressing two distinct short hairpin RNAs (shRNAs) against SASH1 to knockdown SASH1 in two luminal subtype cell lines, T47D and MCF-7. The distribution of G1, G2/M, and S-phases of T47D cells were determined by ow cytometry and SASH1 silencing increased the proportion of S-phase cells and decreased the proportion of G1-phase cells (Fig. 4A). Moreover, ow cytometry analyses indicated that SASH1 knockdown induced less proportion of early apoptotic cells (Fig. 4B). We further analyzed the effects of loss-of-function of SASH1 on the migration and invasion of T47D cells. Transwell assays revealed that migration and invasion of T47D cells were induced by SASH1 deletion (Fig. 4C and Fig. 4D). Flow cytometry analyses suggested that MAP4K4 knockdown increased the proportion of G2/M-phase cells and induced more proportion of early apoptotic cells ( Fig. 4F and Fig. 4G). Transwell assays indicated that MAP4K4 knockdown inhibited the cell numbers of migrating and invasive cells when T47D cells were silenced by MAP4K4-shRNA lentivirus ( Fig. 4H and Fig. 4I). We further co-silenced SASH1 and MAP4K4 and evaluated the effects of co-silencing of SASH1 and MAP4K4 on cell growth migration and invasion of T47D. The cell cycle pro les indicated that although MAP4K4 deletion did not signi cantly change the inducement of SASH1 silencing to the increased proportion of S-phase cells MAP4K4 knockdown could signi cantly increase the proportion of early apoptotic cells (Fig. 4K and Fig. 4L). Meanwhile, MAP4K4 deletion signi cantly abolished the inducement of SASH1 silencing to enhance the migration and invasion ability of T47D cells (Fig. 4N and  Fig. 4O). In the other luminal subtype cell line, MCF-7, SASH1 silencing also increased the proportion of Sphase cells and decreased the proportion of G1-phase cells (Fig.S2A) and SASH1 knockdown also induced decreased proportion of apoptotic cells of MCF-7 cells (Fig.S2B). The cell cycle pro les indicated that MAP4K4 deletion decreased the proportion of S-phase cells (Fig.S2C) and induced increased proportion of apoptotic cells of MCF-7 cells (Fig.S2D). Similarly, in the co-silenced MCF-7 cells, although MAP4K4 deletion did not signi cantly change the inducement of SASH1 silencing to the increased proportion of S-phase cells, MAP4K4 silencing signi cantly increased the proportion of early apoptotic cells ( Fig.S2E and Fig.S2F).
In the HER2 + subtype cell line,SK-BR-3 cells, SASH1 was silenced by two pairs of speci c SASH1-shRNA, and the distribution of G1,G2/M and S-phases was determined by ow cytometry. SASH1 silencing increased the proportion of S-phase cells and decreased the proportion of G2/M-phase cells ( Fig. S3A and Fig.S3C). Moreover ow cytometry analysis revealed after SASH1 silencing a signi cantly smaller proportion of early apoptotic cells was observed ( Fig. S3B and Fig.S3C). Speci c MAP4K4-shRNAs were also introduced into SK-BR-3 cells; cell cycle analyses showed that the interference with MAP4K4 expression increased proportion of G2/M-phase cells and decreased the proportion of S-phase cells which suggested MAP4K4 silencing induced cell cycle arrest in the G2/M-phase ( Fig. S3D and S3F). Flow cytometry analyses also demonstrated that a signi cantly higher proportion of early apoptotic cells were observed after MAP4K4 silencing ( Fig. S3E and S3F). Co-silencing of SASH1 and MAP4K4 induced cell cycle arrest in the G2-phase and a decrease in the proportion of S-phase cells, which suggested that the increase in S-phase cells upon SASH1 silencing in SK-BR-3 cells was prevented by MAP4K4 silencing and MAP4K4 silencing resulted in G2/M-phase cell cycle arrest (Fig. S3G and S3I). In addition, increased proportion of apoptotic cells was induced by co-silencing of SASH1 and MAP4K4 ( Fig. S3H and  S3I).
Loss-of-function of MAP4K4 impairs the ability of SASH1 silencing to promote cell proliferation and metastasis of the xenograft of ER + breast cancer cells.
We further identi ed the effects of loss-of-function of SASH1 and/or MAP4K4 on T47D cells xenograft. T47D cells with stable SASH1 knockdown, MAP4K4 knockdown and co-silencing of SASH1 and MAP4K4 were cultured and injected subcutaneously into the left side fossa axillaris of BALB/c female nude mice. MAP4K4 deletion inhibited the tumor growth including tumor size and tumor weight in the xenograft mouse model with MAP4K4 silencing and co-silencing of SASH1 and MAP4K4 (Fig. 5A, 5B, 5C and 5D). We also evaluated the in uence of silencing of SASH1 and/or MAP4K4 on the metastasis of T47D cells xenograft. SASH1 silencing induced more liver metastatic lesions in the xenograft mouse models with SASH1 silencing compared with those in the xenograft mouse models with negative control(NC), MAP4K4 silencing and co-silencing of SASH1 and MAP4K4 (Fig. 5E). We further identi ed the micro-metastases of liver using H&E staining by counting the number of metastatic lesions in the liver tissues of each mouse. H&E staining revealed that knockdown of SASH1 substantially enhanced the invasiveness of the T47D derived xenograft tumors (Fig. 5F and Fig. 5G). MAP4K4 deletion did not signi cantly reduce the number of liver metastatic lesions, however, MAP4K4 deletion could signi cantly reduce the increased numbers of liver metastatic lesions induced by SASH1 silencing (Fig. 5F and Fig. 5G). Furthermore, pulmonary micrometastases were examined by H&E staining and the total number of pulmonary metastatic lesions in each mouse were counted. The results showed that more and larger pulmonary micro-metastases were detected in the mice model with SASH1-depleted T47D cells than in the control mice. Although MAP4K4 deletion did not signi cantly reduce the number of pulmonary metastatic lesions, MAP4K4 deletion could signi cantly reduce the increased numbers of liver metastatic lesions induced by SASH1 silencing (Fig. 5H and Fig. 5I). We further evaluated the in uence of SASH1 and/or MAP4K4 silencing on the molecular index of proliferation and metastasis. MAP4K4 knockdown in vivo reversed the SASH1 downregulation induced by SASH1 silencing in the T47D cells-xenograft tumors co-silenced with SASH1 and MAP4K4-shRNA lentivirus, and MAP4K4 silencing upregulated SASH1 expression in the xenograft tumors silenced with MAP4K4-shRNA lentivirus (Fig. 6A). SASH1 silencing induced upregulated expression of Ki67 and increased numbers of Ki67-positive cells in the xenograft tumors silenced with SASH1-shRNA lentivirus. MAP4K4 knockdown abolished the inducement of Ki67 expression and increased Ki67-positive cells by SASH1 silencing in the xenograft tumors co-silenced with the lentivirus of SASH1-shRNA and MAP4K4-shRNA (Fig. 6C). However ,SASH1 silencing did not change the expression of caspase 9 MMP2 and MMP9 in the T47D cells-xenograft tumors silenced with SASH1-shRNA lentivirus(data not shown).

SASH1 acts as a protein kinase and is phosphorylated by MAP4K4.
To identify the reason that SASH1 is downregulated in most of tumors and tumor cells, The phosphorylation modi cation sites on SASH1 were identi ed by LC-MS/MS analysis. Ser355, Ser359, Ser914 and Ser918 are potential phosphorylation modi cation sites ( Fig. S4A and S4B). Actually, previous report has suggested that Ser90 of SASH1 is the phosphorylation modi cation site of SASH1 [11]. We further investigated whether SASH1 is phosphorylated by MAP4K4 and which serine sites of SASH1 were the key phosphorylation sites. SASH1 was identi ed to be a very unstable protein and exogenous SASH1 and endogenous MAP4K4 began to degrade after cycloheximide (CHX) treatment ( Fig. 7A). Decreased exogenous SASH1 expression and a backward shift of the SASH1 protein bands in GFP-SASH1-transfected SK-BR-3 cells were observed after treatment with the serine/threonine phosphatase inhibitor calyculin A (Fig. 7B), which indicates that SASH1 is a novel serine protein kinase. Calyculin A-induced SASH1 degradation and calyculin A-induced MAP4K4 degradation were blocked upon the proteasome inhibitor MG132 (Fig. 7C), which indicated that MAP4K4 and SASH1 are both phosphorylated. GNE-495 is a potent and selective MAP4K4 inhibitor and inhibit MAP4K4 function in vivo [24]. To identify the inhibitory effects of GNE-495 to MAP4K4 in breast cancer cells and investigate whether GNE-495 can block the phosphorylation of MAP4K4. SK-BR-3 cells were treated with different concentrations of GNE-495 and western blot showed that 100 nM GNE-495 could e ciently inhibit phospho-MAP4K4 expression and upregulated the protein levels of endogenous MAP4K4 and SASH1 (Fig. 7D). Treatment with GNE-495 not only increased expression of SASH1 and MAP4K4 but also enhanced bindings of SASH1 to MAP4K4 (Fig. 7E). The calyculin A-induced decreases in phosphorylation of exogenous SASH1 and exogenous MAP4K4 were abolished by GNE-495 treatment (Fig. 7F). The SASH1 phosphorylation sites Ser355, Ser359, Ser914, and Ser918 are highly conserved, as analyzed by Clustal X2 software (Fig. 7G). Immunoblot analysis indicated that Ser359 and Ser914 are the key phosphorylation sites of SASH1 because the mutations of S359A and S914A caused little changes in SASH1 expression after GNE-495 treatment as compared with wild type SASH1 and other SASH1 mutants. Especially S914A caused a forward shift of the SASH1 protein band (Fig. 7H, 7I and 7J). IP-WB revealed that these two mutations increased binding of SASH1 to MAP4K4 and GNE-495 not only promoted the binding of wild type SASH1 but also of mutated SASH1 to MAP4K4 (Fig. 7K).
GNE-495 may reverse MAP4K4-mediated phosphorylation of SASH1 and upregulates both protein levels of MAP4K4 and SASH1.
MAP4K4 over-expression downregulated endogenous SASH1 expression in hormone-dependent breast cancer cell lines,SK-BR-3 and MCF-7 cells (Fig. 8A). Exogenous SASH1 was upregulated upon MAP4K4 knockdown (Fig. 8B). Bioinformatics analysis by Clustal X2 software indicated that Ser629, Ser631 and Ser801 of MAP4K4 are highly conserved phosphorylation sites (Fig. 8C). Western blot analysis suggested that the S801A mutation of MAP4K4 induced a forward shift of the MAP4K4 protein bands and unlike exogenous SASH1 was upregulated after the phosphorylation inhibition of GNE-495 treatment to MAP4K4 S801A mutation of MAP4K4 could not upregulate or attenuated the expression of exogenous SASH1 in SK-BR-3 and HEK-293T cells (Fig. 8D and Fig. 8E). In addition, the S801A mutation of MAP4K4 not only induced a forward shift of MAP4K4 main bands but also caused the appearance of multiple SASH1protein (Fig. 8D and Fig. 8E). Similarly, the S801A mutation in MAP4K4 attenuated the GNE-495induced-endogenous SASH1 upregulation ( Fig. 8F and Fig. 8G). Further IP-WB analyses revealed that S801A mutation in MAP4K4 attenuated the bindings of MAP4K4 to SASH1 compared with wild type MAP4K4 (Fig. 6H) in vivo. We performed in vitro kinase assay to identify the phosphorylation of MAP4K4 to SASH1. GNE-495 not only upregulates the expression of wild type MAP4K4 but also that of S801A-MAP4K4. Being similar to the in vivo results indicated above, GNE-495 promoted the bindings of wild type MAP4K4 to his-SASH1, however, attenuated the association S801A-MAP4K4 tohis-SASH1 in vitro (Fig. 6I). In addition GNE-495 treatment caused a signi cant apoptosis in hormone-dependent breast cancer cells and MAP4K4 deletion can impair the induced apoptosis by GNE-495(supplementary Fig.S5).

Discussion
Although growing evidence suggests that SASH1 is a tumor suppressor in breast cancer [25] [2,26,27], the mechanism by which SASH1 regulates tumor behaviors including tumorigenesis,adhesion and migration on cancer cells in cancer development and progression remain completely unclear [25]. Recent evidence reveal that SASH1 might be phosphorylated by LATS1 and LATS2 [12,28]. In this study, we rstly reveal that SASH1 acting as a novel protein kinase is phosphorylated by MAP4K4, to constitute a novel MAP4K4-SASH1 phosphokinase cascade to mediate tumorigenesis and metastasis of hormonedependent breast cancer. Taken the previous ndings into consideration, a MAP4K4-LATS1/LATS2-SASH1 signal cascade may mediate the tumorigenesis and metastasis of breast cancer. Functionally MAP4K4 existence is necessary to SASH1. MAP4K4 deletion will abolish the SASH1-induced tumorigenesis and metastases in vitro and in vivo. Importantly, GNE-495 speci cally inhibits MAP4K4 phosphorylation and not only upregulate MAP4K4 protein level but also that of SASH1. GNE-495 by preventing the phosphorylation of the MAP4K4-SASH1 phosphorylation cascade might be a novel effective inhibitor, inhibits tumorigenesis of hormone-dependent breast cancer. Therefore, our ndings rstly reveal a novel kinase cascade that can be targeted by small molecular compounds to treat hormone-dependent breast cancers. Previous reports have revealed that Ser90 in SASH1 serves as the 14-3-3-binding sites to induce SASH1 phosphorylation by phosphatidylinositol 3-kinase and MAPK/p90RSK signaling [17] and Ser407 in SASH1 is the key serine site for LATS1-mediated phosphorylation of SASH1 [12]. Ser 914 in SASH1 in the present study is identi ed to be another serine site for MAP4K4mediated phosphorylation of SASH1.
The serine/threonine kinase MAP4K4 is a member of the Ste20p (sterile 20 protein) family and increasing evidence suggests that MAP4K4 also plays an important role in cancer [29]. Current evidence suggests that MAP4K4 potentially serves as a negative prognostic indicator in patients with colorectal cancer [30], hepatocellular carcinoma [31], pancreatic ductal adenocarcinoma [32], lung adenocarcinoma [33] and prostate cancer [34]. MAP4K2, MAP4K4 and MAPK1 have been identi ed as kinases that mediate paracrine growth signaling in ER-negative breast cancers [35]. In this study, our ndings indicate that in breast cancer, especially the luminal A and TNBC subtypes, MAP4K4 is downregulated compared to normal breast tissues. Generally, cell apoptosis [31,[36][37][38][39], cell cycle arrest [31,38,39], and migration and invasion [40] [38,39,41,42] are induced by MAP4K4 downregulation in cancer cells. In this study, MAP4K4 silencing in induces cell cycle arrest, apoptosis and block migration and invasion in hormonedependent breast cancer cell lines and T47D cells-xenograft tumors. MAP4K4, in a kinase activitydependent manner, positively regulate cell transformation and invasion and negatively regulates cell spreading and adhesion [13]. MAP4K4 physically interacts with both LATS1 and LATS2 and phosphorylates LATS1 and LATS2 [28] and SASH1 was phosphorylated by LAST1 [12]. Our ndings indicated that MAP4K4 physically interacts with SASH1 and MAP4K4 may indirectly phosphorylate SASH1. However, MAP4K4 deletion will impair the inducement of loss-of-function of SASH1 to tumorigenesis,adhesion and migration of hormone-dependent breast cancer in breast cancer cell lines and xenograft tumors. So functionally, the existence of MAP4K4 is necessary to SASH1 and there may be a balance of expression levels between MAP4K4 and SASH1. Ser801 in MAP4K4 are identi ed might to be the key sites for MAP4K4-mediated phosphorylation of SASH1.
Given the diverse roles of MAP4K4 in many cell processes, the therapeutic potential of MAP4K4 inhibition will be appealing and it's worth identifying the MAP4K4 inhibitor that would function in vivo [24]. Some of the inhibitors show promise in treating pathological angiogenesis in mice [24,43]. MAP4K4 may represent a novel actionable cancer therapeutic target. Whether these inhibitors possess potent antitumor properties remains to be determined [29]. In this study, GNE-495 not only inhibits phospho-MAP4K4 protein level and upregulates MAP4K4 but also increases SASH1 protein level, which eventually induce the apoptosis of hormone-dependent breast cancer cells.

Conclusion
Therefore, our ndings rstly reveal a novel SASH1-MAP4K4 phosphoprotein kinase cascade that may serve as a therapeutic intervention to treat hormone-dependent breast cancers. showed, respectively. bar:10μm and Magni cation: 10x. B The IHC scores of SASH1 of four cancer subtypes and normal breast cancer were analyzed statistically and statistical analyses indicated that SASH1 expression was downregulated in four subtype breast cancers compared with that of normal breast cancer. C IHC staining of an array with 139 human breast carcinoma tissues (bottom panels) and matched adjacent tissues (upper panels). A representative image of SASH1 IHC staining in breast cancer subtypes and adjacent tissues. bar: 10μm and Magni cation: 10x. D Statistical analysis of SASH1 IHC scores in tissue arrays showed that SASH1 expression was decreased in different of breast carcinoma subtypes as compared with adjacent breast tissues*P<0.05, ** P <0.01, *** P <0.001. E Statistical analyses of IHC scores of SASH1 and Ki67 in tissue arrays showed that SASH1 expression was negatively associated with that of Ki67 in subtypes of luminal and Her2 positive. However, SASH1 expression was not associated with that of Ki67 in TNBC subtype. F Statistical analyses of IHC scores of SASH1 and 53 tissue arrays showed that SASH1 expression was positively associated with that of p53 in subtypes of luminal and Her2 positive. However, SASH1 expression was not associated with that of p53 in TNBC subtype. G Immunoblotting analyses identi ed that SASH1expression was decreased in most breast carcinoma subtypes compared with normal breast tissues. N = benign tissues. C = cancer tissues. and MCF-7 cells were stained with SASH1 antibody (green) and MAP4K4 antibody (red) with immuno uorescence and photographed by confocal microscope.    Values are mean ± SD. ***p<0.001. E Liver tissues were photographed and the black arrows indicate the liver metastatic lesions. F More live metastatic lesions were observed in the nude mice bearing SASH1 knockdown-xenograft tumors. However, MAP4K4 knockdown impaired the inducement of more live metastatic lesions by SASH1 silencing in the nude mice with co-silencing SASH1 and MAP4K4xenograft tumors. Liver tissues were photographed, xed, and stained with hematoxylin and eosin (H&E).
The numbers of metastatic lesions of each mouse were counted and compared statistical analyses. ***p<0.001. G Increased pulmonary metastatic lesions were observed in the nude mice bearing SASH1 deletion-xenograft tumors. However, MAP4K4 silencing abolished the inducement of more pulmonary metastatic lesions by SASH1 silencing in the nude mice with co-silencing SASH1 and MAP4K4-xenograft tumors. Lung tissues were photographed, xed and subjected to HE staining. The total numbers of pulmonary metastatic lesions in one lung tissues of each mouse were counted and analyzed statistically. *p<0.05, **p<0.01. bar:10cm. The parameter index of cell proliferation induced by SASH1 deletion were impaired by MAP4K4 silencing in vivo. A MAP4K4 deletion in vivo reversed the SASH1 downregulation induced by SASH1 silencing in the T47D cells-xenograft tumors bearing SASH1 and MAP4K4-co-silencing , and MAP4K4 silencing upregulated SASH1 expression in the xenograft tumors stable expressing MAP4K4-shRNA lentivirus. The SASH1 expression of the xenograft tumors containing SASH1 and/or MAP4K4 -shRNA lentivirus as well as the negative control lentivirus were analyzed by IHC. The SASH1 expression were scored according to the positive areas and positive intensity of SASH1.The SASH1 scores were analyzed using one-way ANOVA and the cartograms were plotted. The mice numbers in each group were indicated in the cartograms and the SASH1 scores between the xenograft tumors expressing negative control-shRNA lentivirus(NC-shRNA LV) and SASH1-shRNA LV, those expressing SASH1-shRNA LV and MAP4K4-shRNA LV, those expressing SASH1-shRNA LV and SASH1-shRNA+ MAP4K4-shRNA LV and those NC-shRNA LV and MAP4K4-shRNA LV were compared. **p<0.01,*p<0.05,ns no signi cance. B SASH1 silencing may promote MAP4K4 expression in the xenograft tumors containing SASH1-shRNA lentivirus. **p<0.01,