TTC17 deficiency is identified as a metastasis driver in BC via high-throughput omics screening

To obtain reliable candidates connected with BC metastasis, we combined biological assays and data mining to screen metastasis-associated genes. The screening flow is outlined in Fig. 1a. We used the GeCKO v2.0 library, which contains sgRNAs targeting genome-wide genes and miRNAs, to construct gene-edited MCF7 cell lines that were then subjected to puromycin selection and Transwell invasion assays (Fig. 1b). Next-generation sequencing of the DNA extracted from the invaded cells revealed enriched sgRNAs targeting 1141 genes (P < 0.05), including TTC17 (Fig. 1c), and functional annotation of the top 286 genes (P < 0.01) showed that genes related to cell motility and the Ras signaling pathway were significantly enriched (Fig. 1d, e).

Fig. 1

TTC17 deficiency is identified as a metastasis driver in breast cancer via high-throughput omics screening. a Flowchart of all screening processes involved in the study. b Schematic representation of loss-of-function screening for metastasis-related genes using the human genome-scale CRISPR–Cas9 knockout library linked to Transwell invasion assays. c Identification of top genes as candidates concerning their invasion-promoting ability in Transwell screening with MCF7 breast cancer cells using MAGeCK analysis. The red line indicates a P value of 0.05. d, e GO (d) and KEGG (e) enrichment for candidate genes identified by GeCKOv.2 library screening in Fig. 1b and c. f Overlap of differentially mutated genes between metastatic and nonmetastatic breast cancer from a genomic data comparison in the TCGA database and whole-exome sequencing of tissue samples from our patient cohort. g Kaplan‒Meier survival plots of relapse-free survival in breast cancer patients with distinct TTC17 transcriptional expression. h Heatmap of representative genes screened by random forest analysis using the transcriptomic data of metastatic and nonmetastatic breast cancer cases from TCGA with low expression promoting metastasis. i, j Schematic illustration (i) and enrichment of candidate genes (j) for a 64-gene CRISPR screening library using in vivo lung metastasis models of MDA-MB-231 breast cancer cells. CRISPR, clustered regularly interspaced short palindromic repeats; WES, whole-exome sequencing; TCGA, the Cancer Genome Atlas; DMGs, differentially mutated genes; RFS, relapse-free survival; KM, Kaplan–Meier; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; GeCKO, genome-scale CRISPR–Cas9 knockout

Then, WES of cancer samples from patients eligible for inclusion with metastatic (n = 13) and nonmetastatic (n = 13) BC revealed 450 differentially mutated genes. Further analysis of the genomic data of tumors from 43 metastatic and 25 nonmetastatic patients in the TCGA database revealed 1530 differentially mutated genes. As shown in Fig. 1f, 69 candidate genes, including typical tumor suppressor genes such as TP53 [28,29,30], CFTR [31,32,33], LRP5 [34], WWC3 [35,36,37], and ZNF423 [38, 39], were common to both cohorts. Among them, 58 genes had the same mutation status in the WES and TCGA data, while the remaining genes had different mutation statuses in the two subsets (Fig. 1f, Additional file 1: Table S2). Subsequently, by combining with the results from the genome-wide CRISPR library screen, the high expression profiles of 33 genes were markedly associated with favorable relapse-free survival (RFS; Fig. 1g, Additional file 2: Fig. S1a-h). TTC17 caught our attention because high TTC17 expression had the strongest correlation with improved RFS (hazard ratio 0.56, 95% confidence interval (CI) 0.48–0.65, P < 0.001; Fig. 1g, Additional file 2: Fig. S1a-h).

For further confirmation, we also identified the DEGs of primary tumors between the metastatic and nonmetastatic BC cases in the TCGA dataset through random forest analysis combined with survival analysis using the Kaplan–Meier Plotter database, among which the decreased expression of 31 genes was linked to worse RFS (Fig. 1h, Additional file 2: Fig. S1i-p). We next constructed a CRISPR–Cas9 library targeting the 33 genes identified from exome sequencing and the 31 DEGs. Three distinct sgRNAs were designed for each gene, and MDA-MB-231 cells infected with the library were used to establish a lung metastasis model in mice by tail vein injection (Fig. 1i). Eleven weeks later, the genes markedly knocked down in the metastatic lung nodules were identified by sequencing the sgRNA library (Fig. 1j). Of note, we also examined the metastasis of cancer cells to other important sites, such as bone, liver, spleen, kidney, brain, and heart, and found no evidence of metastasis to other organs except the lungs. By summarizing all the above screening results, TTC17 was identified as a metastasis suppressor in BC through genome-scale CRISPR screening, WES and RNA-seq of BC tissues and a mouse model of lung metastasis, and its deletion in BC enhanced metastasis potential. Based on the preliminary experimental results that TTC17 had a more obvious and stable effect on the malignant phenotype of breast cancer than other candidate molecules and our literature review assessing the current status of few studies on TTC17, we finally chose TTC17 for further study.

TTC17 is downregulated in BC tissue, and low TTC17 expression predicts worse clinical outcomes

To evaluate the clinical significance of TTC17, we analyzed its transcriptional and translational data using the TCGA, GTEx, UALCAN, and CCLE datasets. Pancancer analysis suggested that TTC17 was mostly downregulated in neoplasms compared to their normal counterparts (Fig. 2a). With regard to prognosis, reduced TTC17 expression was also correlated with shorter RFS in ovarian cancer and testicular germ cell tumors, together with dismal OS in bladder carcinoma, pancreatic ductal adenocarcinoma, rectum adenocarcinoma, and stomach adenocarcinoma (Fig. 2b, c, Additional file 2: Fig. S2a-d). Nevertheless, low TTC17 expression exhibited opposite effects on RFS in kidney renal papillary cell carcinoma and OS in cervical squamous cell carcinoma, kidney renal clear cell carcinoma, kidney renal papillary cell carcinoma, and pheochromocytoma and paraganglioma (Fig. 2d, Additional file 2: Fig. S2e-h).

Fig. 2

TTC17 is downregulated in breast cancer tissue, and low TTC17 expression predicts worse clinical outcomes. a Pancancer analysis of TTC17 expression using data from TCGA combined with the GTEx database. b–d Kaplan‒Meier curves of relapse-free survival in patients with ovarian cancer (b), testicular germ cell tumor (c), or kidney renal papillary cell carcinoma (d), stratified by TTC17 expression. Data were obtained from the KM plotter program. e–i Association of TTC17 mRNA expression with primary lesion (T stage, e), lymph node involvement (N stage, f), distant metastasis (M stage, g), clinical stage (h), or molecular subtype (i) in breast cancer patients based on bulk RNA-seq data archived in TCGA. j TTC17 expression level in breast cancer with distinct molecular subtypes using single-cell RNA-seq data from the PanglaoDB project. k TTC17 expression in luminal or basal/TN breast cancer cell lines based on the CCLE dataset. *P < 0.05, **P < 0.01, ***P < 0.001. TCGA, The Cancer Genome Atlas; GTEx, Genotype-Tissue Expression; RNA-seq, RNA-sequencing; CCLE, Cancer Cell Line Encyclopedia

In BC specifically, TTC17 mRNA and protein expression levels were decreased in the tumor tissues relative to juxta-tumoral tissues (Fig. 2a, Additional file 2: Fig. S3). Moreover, compared to their respective counterparts, breast neoplasms with larger primary lesions, lymph node involvement, distant metastasis, or advanced TNM stage had lower TTC17 expression (Fig. 2e–h). Additionally, the determination of TTC17 based on molecular subtypes indicated that human epidermal growth factor receptor 2 (HER2)-positive and triple-negative BC, representing the intractable subpopulations, had even lower levels of TTC17 expression than luminal subsets with lower aggressiveness (Fig. 2i). scRNA-seq analysis of BC also showed this trend (Fig. 2j). Likewise, TTC17 expression was lower in basal/TNBC cell lines than in luminal BC cell lines (Fig. 2k, Additional file 1: Table S3) [40,41,42]. Given that higher TTC17 levels were associated with a longer RFS, we surmised that decreased TTC17 expression in breast tissues predisposed individuals to disease progression and metastasis.

Loss of TTC17 function promotes the metastatic properties of BC cells in vitro and in vivo

To determine the biological role of TTC17 in BC, we deleted or silenced TTC17 in MDA-MB-231 and MCF7 cells using the CRISPR–Cas9 system and an shRNA technique. Loss of TTC17 enhanced not only the in vitro migration of MDA-MB-231 cells in wound healing assays but also their invasiveness through Matrigel in Transwell assays (Fig. 3a–d). Forced expression of TTC17 induced the opposite effects (Fig. 3e, f, Additional file 2: Fig. S4). Furthermore, MCF7 cells overexpressing TTC17 formed much smaller and fewer colonies than the controls (Additional file 2: Fig. S5). Thus, TTC17 deficiency promoted the metastatic progression of BC cells.

Fig. 3

Loss of TTC17 function promotes the metastatic properties of breast cancer cells in vitro and in vivo. a–d Representative images and quantitative analysis of wound healing assays (a, TTC17 knockout; b, TTC17 knockdown) and Transwell invasion assays (c, TTC17 knockout; d, TTC17 knockdown) using MDA-MB-231 cells with TTC17-deficient and control cells. Scale bars, a, b 500 μm; c-d 200 μm. e, f Representative images and quantitative analysis of wound healing assays (e) and Transwell invasion assays (f) using MCF7 cells with TTC17 overexpression and control cells. Scale bars, e 500 μm; f 200 μm. g Schematic for evaluating the effect of TTC17 loss on the formation of pulmonary metastasis in the orthotopic mammary carcinoma model. h Representative histological images (H&E staining) and quantitative analysis of metastatic nodules in the lungs of BALB/c mice that received 4T1 cells with or without TTC17 depletion. Arrowheads indicate lung metastasis nodules (n = 6 for each group). Scale bar, 5 mm. i Schematic for assessing the effect of TTC17 silencing on the initiation of lung colonization through the tail vein injection model. j–l Representative gross images and quantitative analysis of metastatic nodules in the lungs (j), lung weight (k), and bodyweight (l) of NOD/SCID mice injected with TTC17-silenced MDA-MB-231 cells and their counterparts (n = 5 for each group). Scale bar, 5 mm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. H&E, hematoxylin and eosin

We next established an orthotopic mouse mammary carcinoma model in female BALB/c mice using TTC17-knockout 4T1 cells to determine the influence of TTC17 on metastatic lung colonization (Fig. 3g). TTC17 depletion substantially accelerated the colonization of BC cells in the lungs, as evaluated by histopathological changes and the number of metastatic nodules (Fig. 3h). Furthermore, we confirmed this property by constructing a lung metastasis model by tail vein injection in NOD/SCID mice using TTC17-silenced MDA-MB-231 cells (Fig. 3i). After 13 weeks, compared to the control group, pronounced increases in metastatic lung nodules and lung weight, in addition to a striking reduction in body weight, were observed in the mice injected intravenously with TTC17-knockdown cancer cells (Fig. 3j-l). Taken together, loss of TTC17 function could boost the metastatic capacity of BC in mouse and human cell-derived models.

Activation of RAP1/CDC42 signaling is required for TTC17 deficiency to facilitate metastasis

Subsequently, we investigated the mechanism by which TTC17 modulated malignant behaviors. Transcriptome sequencing of the control and TTC17-knockout MDA-MB-231 cells revealed DEGs that were notably enriched in biological processes related to metastasis, such as regulation of angiogenesis, cell movement, extracellular matrix, cell adhesion, and cell migration (Fig. 4a, b). In the KEGG pathway enrichment analysis, we observed that the RAP1 signaling pathway was prominently upregulated in TTC17-knockout cells, suggesting its possible involvement in TTC17-mediated BC cell motility and invasiveness (Fig. 4c, d). Multiple RAP1 signaling pathway factors, such as RAP1A, RAP1B, CDC42, MAPK2K6, RHOA, ITGB1, ITGB2, ACTB, and ACTG1, were inversely correlated with TTC17 in the coexpression assessment using the METABRIC dataset, which underscored the role of this pathway in the TTC17-mediated metastatic capacity of BC (Fig. 4e).

Fig. 4

Activation of RAP1/CDC42 signaling is required for TTC17 deficiency to facilitate metastasis. a Profile of DEGs (logFC > 1.5, FDR < 0.05) between TTC17-knockout MDA-MB-231 cells and control cells. b Enrichment of GO biological processes for up- and downregulated genes. c Enrichment of KEGG pathways for up- and downregulated genes. d Heatmap of key molecules in the RAP1 signaling pathway enriched by DEGs described in Fig. 4a. e mRNA expressional correlation of TTC17 with members of the RAP1/CDC42 cascade in breast cancer, including RAP1A, RAP1B, CDC42, MAP2K6, RHOA, ITGB1, ITGB2, ACTB, and ACTG1. f, g Western blot analysis of RAP1/CDC42 signaling activation in MDA-MB-231 cells with or without TTC17 knockdown by increased RAP1A, CDC42 (f), and RAP1A-GTP (g). h, i Representative images and quantitative analysis of wound healing assays (h) and Transwell invasion assays (i) using TTC17-silenced MDA-MB-231 cells treated with the CDC42 inhibitor ML141 and their counterparts. Scale bars, h 500 μm; i 200 μm. j Representative immunofluorescence images of actin cytoskeleton and Golgi morphology in MDA-MB-231 cells with TTC17 silencing or scramble control. Scale bar, 10 μm. *P < 0.05, **P < 0.01, ***P < 0.001. DEGs, differentially expressed genes; FDR, false discovery rate; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; BRCA, breast invasive carcinoma

Among the RAP1 pathway network members, CDC42 is a GTPase markedly regulated by TTC17. CDC42 was expressed at a higher level in BC than in related normal tissues per the TCGA and GTEx databases (Additional file 2: Fig. S6). TTC17 knockdown in BC cells significantly increased the expression of RAP1A, RAP1-GTP, and CDC42 levels (Fig. 4f, g, Additional file 2: Fig. S7a), while overexpressing TTC17 produced the opposite phenomena (Fig. S7b). These findings suggest that TTC17 silencing activates the RAP1 signaling pathway by releasing RAP1A and CDC42. To validate the contribution of this pathway to cancerous behaviors, we treated TTC17-knockdown BC cell lines with the CDC42 inhibitor ML141. Pharmacological blockade of CDC42 abrogated the compelling difference in migration and invasion ability between MDA-MB-231 cells with TTC17-knockdown shRNA or scramble control, thereby confirming that the TTC17-mediated RAP1/CDC42 pathway was responsible for tumor progression (Fig. 4h, i). Biologically, TTC17 is localized to the Golgi membrane and is required to process cargo proteins and maintain Golgi homeostasis, and its loss leads to Golgi swelling, dilation, and polarization disorders [13]. Consistent with these findings, we also observed dilated and enlarged Golgi in the cells with silenced TTC17, as well as altered actin cytoskeletal organization, as reported previously (Fig. 4j). In conclusion, the RAP1/CDC42 cascade exerted an indispensable effect on the metastatic capability of TTC17 impairment.

TTC17/RAP1/CDC42 pathway activation correlated with the clinical metastasis and aggressive characteristics of BC

To address the contribution of TTC17/RAP1/CDC42 activation to metastasis in patients with BC, we detected the protein expression of TTC17 and CDC42 in a large cohort of nonmetastatic and metastatic primary breast tumors together with their metastatic lymph node tissues. Overall, the primary tumor lesions exhibited higher H-scores (evaluation of IHC staining) for TTC17 than the metastatic lymph nodes, whereas CDC42 expression demonstrated the opposite trend: nonmetastatic primary lesions displayed substantially lower CDC42 expression than metastatic tumors and corresponding metastatic lymph nodes (Fig. 5a, b). Meanwhile, data from another of our BC cohorts suggested that the TTC17Low arm had significantly reduced expression of ATP5A, a molecule that has been reported to have an inverse correlation with metastasis (Fig. 5c) [43].

Fig. 5

TTC17/RAP1/CDC42 pathway activation correlated with the clinical metastasis and aggressive characteristics of breast cancer. a, b Representative images and quantitative analysis of IHC staining of TTC17 (a) and the RAP1 signaling pathway effector CDC42 (b) in the primary tumor and metastatic lymph node tissues of breast cancer patients. Scale bars, a 50 μm, b 50 μm. c Representative IHC images and quantitative analysis of ATP5A expression in patients with breast cancer. Scale bar, 100 μm. d, e Correlation between the expression of TTC17 and tumor histopathologic grade (d) together with the Ki-67 index (e) in our breast cancer cohort. f Univariate and multivariate logistic regression analysis of the linkage between TTC17 expression and clinicopathological features in breast cancer patients using data archived in TCGA. Bold text represents statistically significant difference. IHC, immunohistochemistry; non-mPT, non-metastatic primary tumor; mPT, metastatic primary tumor; mLN, metastatic lymph node; TCGA, The Cancer Genome Atlas CI, confidence interval; ILC, infiltrating lobular carcinoma; IDC, infiltrating ductal carcinoma; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; TNBC, triple-negative breast cancer; lum A, luminal A; lum B, luminal B

Moreover, we observed that low expression of TTC17 resulted in higher susceptibility to a higher histopathologic grade of foci as well as a Ki-67 index ≥ 40 than high TTC17 expression (grade 3: 53.6% for TTC17Low, 19.2% for TTC17High, P = 0.009; Ki-67 ≥ 40: 54.3% for TTC17Low, 36.7% for TTC17High, P = 0.025; Fig. 5d, e). Furthermore, we divided the TCGA BC cohort into two groups based on their high or low expression of TTC17. Univariate logistic regression analysis demonstrated that low TTC17 expression was more common in Asians or African Americans versus Caucasians (odds ratio (OR) 0.456, 95% CI 0.156–0.756, P < 0.001), postmenopausal versus premenopausal women (OR 1.421, 95% CI 1.121–1.721, P = 0.022), patients with T stage 3 or 4 versus 1 or 2 (OR 1.385, 95% CI 1.060–1.709, P = 0.049), patients with N stage 2 or 3 versus stage 0 or 1 (OR 1.396, 95% CI 1.083–1.708, P = 0.036), patients with negative versus positive ER status (OR 0.478, 95% CI 0.178–0.777, P < 0.001), patients with negative versus positive PR status (OR 0.583, 95% CI 0.321–0.845, P < 0.001), and patients with TNBC or HER2amp subtype versus luminal subtype (OR 0.395, 95% CI 0.059–0.731, P < 0.001). Subsequent multivariate logistic regression analysis revealed that postmenopausal versus premenopausal status (OR 1.507, 95% CI 1.114–1.900, P = 0.041), N stage 2 or 3 versus stage 0 or 1 (OR 1.633, 95% CI 1.151–2.115, P = 0.046), pathologic stage II, III or IV versus stage I (OR 1.714, 95% CI 1.238–2.189, P = 0.026), and TNBC or HER2amp subtype versus luminal subtype (OR 0.432, 95% CI − 0.39–1.263, P = 0.048) were independent factors associated with the low expression of TTC17 (Fig. 5f, the threshold for inclusion in the multivariate analysis after the univariate analysis was controlled for a P value < 0.1). The association between the expression of TTC17 and clinicopathological characteristics regarding aggressive malignancy and a poor prognosis further verified its role in facilitating BC progression.

Loss of TTC17 sensitizes BC to rapamycin and paclitaxel

To determine whether TTC17 affects the response of BC cells to antineoplastic drugs, we screened an anticancer drug library consisting of 313 commercially available drugs in control and TTC17-knockdown MDA-MB-231 cells (Fig. 6a, b). Compared to controls, cells losing TTC17 showed increased sensitivity to 25.8% of the tested drugs (lower IC50) and decreased sensitivity to 11.2% of the drugs (higher IC50), which acted on different signaling pathway targets (Fig. 6c, d). Specifically, TTC17 deficiency enhanced the intrinsic sensitivity of cancer cells to rapamycin, paclitaxel, pirfenidone, palbociclib, cobimetinib, sorafenib tosylate, fluorouracil, NVP-LDE225, and oxaliplatin and diminished the sensitivity of cells to floxuridine, dasatinib monohydrate, and estramustine phosphate sodium (Fig. 6e).

Fig. 6

Loss of TTC17 sensitizes breast cancer to rapamycin and paclitaxel. a Schematic showing the screening of drugs with a discrepancy in the efficacy on breast cancer cells by TTC17 silencing in the listed antineoplastics through CCK8 assays. b Rose plot of the composition of the signaling pathways targeted by the compounds being tested. c–d Proportions (c) and profiles of signaling networks (d) of drugs with inhibitory disparity for TTC17-knockdown and control cells. e–f Representative agents, whose inhibitory effects on breast cancer cells were enhanced (e) or attenuated (f) by TTC17 silencing, including rapamycin and paclitaxel. g-j Disease control rates of the breast cancer patients treated with rapamycin in the overall cohort (g), in subgroups with patients < 60 years of age (h), with pathological grade 1 or 2 (i), or with a combination therapy of nonsteroidal aromatase inhibitor or exemestane (j). k–m Gross image (k) and statistical analysis of volume (l) and weight (m) of the transplanted tumors extracted from BALB/c mice injected with 4T1 cells. n–o Association between TTC17 expression and pathologic complete response in breast cancer patients subjected to paclitaxel-based chemotherapy as a neoadjuvant therapy (n) or relapse-free survival at five years in breast cancer patients receiving paclitaxel-based chemotherapy as an adjuvant therapy (o)

Notably, rapamycin is an inhibitor of CDC42 and has been applied to treat patients with advanced hormone receptor-positive, HER2-negative BC. A higher disease control rate (DCR) was achieved in the TTC17Low arm of BC patients receiving rapamycin (P = 0.029), and stratified analysis demonstrated that in patients aged < 60 years (P = 0.006), with pathological grade 1 or 2 (P = 0.020), and a combination regimen of nonsteroidal aromatase inhibitor or exemestane (P = 0.001), low expression of TTC17 exhibited a more pronounced correlation with higher DCR (Fig. 6g–j). Furthermore, considering the regulatory role of TTC17 on the actin cytoskeleton, we wondered whether TTC17 expression would affect the efficacy of the anti-microtubule chemotherapeutic agent paclitaxel, which is a common chemotherapeutic drug in routine BC treatment and was also indicated by drug library screening in BC. Paclitaxel treatment significantly reversed the accelerated BC cell growth caused by TTC17 knockdown in mice, showing better efficacy in TTC17 low-expressing BC (Fig. 6k–m). The data concerning therapy responses suggested that reduced TTC17 conferred better pathological complete response (pCR) and five-year RFS rates in BC patients subjected to paclitaxel-based chemotherapy in the neoadjuvant and adjuvant settings, respectively (Fig. 6n, o). Taken together, decreased TTC17 levels in BC promoted metastatic behaviors but sensitized the cancers to a panel of routine anticancer drugs, such as rapamycin and paclitaxel, as verified, providing a treatment opportunity for patients with TTC17-related molecular subtypes (Fig. S8). Drugs losing sensitivity to BC cells with low TTC17 expression are suggested to be avoided in future use in TTC17-deficient breast cancers.