Shear stress activates ATOH8 via autocrine VEGF promoting glycolysis dependent-survival of colorectal cancer cells in the circulation

Peripheral blood samples were collected from 156 CRC patients (clinical cohort 1) with detailed information of blood pressure before any anti-tumour therapy in Nanfang Hospital (Guangzhou, China) from August 2016 to July 2017. CTC isolation and classification were performed as described previously [23]. Details are available in the supplementary materials (Additional file 1: Supplementary methods and materials). Meanwhile, 12 pairs of CRC and adjacent non-tumour tissues (clinical cohort 2) were collected from patients who underwent surgery at Nanfang Hospital between May 2018 and September 2018 to verify the expression of ATOH8. All samples were taken under the approval of the Ethics Committee of Nanfang Hospital along with obtainment of written informed consent from patients.

All animal experiments were conducted in accordance with the Public Health Service Policy in Humane Care and Use of Laboratory Animals and were approved by the Ethical Committee of Southern Medical University. BALB/c female nude mice 4–5 weeks of age were purchased from the Experimental Animal Centre, Southern Medical University (Guangzhou, China) and maintained in specific pathogen-free conditions. Subcutaneous tumour and metastatic tumour mouse models were generated as described previously [24]. Details are available in the supplementary materials (Additional file 1: Supplementary methods and materials).

The CRC cell lines (LoVo, SW480, SW620, DLD1, HT29, and HCT116) and immortalised intestinal epithelial cell line NCM460 were purchased from Foleibao Biotechnology Development Company (Shanghai, China). Cells were cultured with RPMI 1640 medium with 10% foetal bovine serum (Hyclone, USA) at 37 °C under 5% CO₂.

The microfluidic system fabricated by 7 tandem μ-slides I 0.4 (Ibidi, GmbH, Martinsried, Germany) and infusion bump was used to load different levels of shear stress on colorectal cancer cells. Details are available in the supplementary materials (Additional file 1: Supplementary methods and materials).

CTCs are vital to tumour metastasis, while the number of CTCs is sparse. To solve this research dilemma, previous researchers have used alternative strategies, such as adapted suspension tumour cells or tumour cells suspended and exposed to LSS [25, 26]. Therewith, we simulated the mechanical fluid microenvironment of CTCs using a device that could induce continuous cyclic shear stress on suspended tumour cells and we verified the stability of flow velocity in this flow system, using ANSYS software (Additional file 3: Figure S1a-b). According to previous reports, we set parameters to control LSS within a physiological range of 0–20 dyn/cm² [7]. Most CTCs maintained their original morphology, while some other cells’ edges became indiscernible (Additional file 3: Figure S1c). Importantly, we have identified these suspended colorectal cancer cells with molecular features like CTCs, which are CK8+/CD45−/DAPI+ (Additional file 3: Figure S1d). In conclusion, we defined the above suspension cells exposed to physiological LSS as mimic circulating tumour cells (m-CTCs) and use them as an alternative to CTCs in related experiments in this study.

Firstly, LoVo and SW480 suspended cells were loaded into the shear stress device, and the expression of ATOH8, an LSS response molecule was detected. After size-gradient and time-gradient shear stress stimulation, the results of immunofluorescence analysis, quantitative polymerase chain reaction (qPCR) and western blotting (WB) were concurrent, implying that the expression levels and the nuclear localisation of ATOH8 were increased in CRC m-CTCs (Fig. 1a-f). The mRNA level of ATOH8 in CRC m-CTCs increased obviously after 15 min of LSS and reached a maximum at around 4 h (Fig. 1e, Additional file 3: Figure S1e).

To further investigate changes of ATOH8 in response to LSS in vivo, we harvested peripheral blood samples from clinical cohort 1 and carried out CTC assessment. A previous study has reported that patients with hypertension often have high blood LSS levels [27]. Therefore, after excluding patients receiving antihypertensive therapy (n = 15), we divided patients into hypertension (HP, n = 27) and non-hypertension (NHP, n = 114) groups according to their previous history of hypertension. Other baseline clinical characteristics of the two groups were compared and no significant differences were detected (Additional file 4: Table S2). In both groups, CTCs with three different phenotypes (epithelial phenotype, mixed epithelial/mesenchymal phenotype, and mesenchymal phenotype) and different ATOH8 expression levels were observed and enumerated (Fig. 1g). As predicted, the proportion of CRC patients with a total number of CTCs ≥5 cells/5 mL was higher in the HP group (Additional file 3: Figure S1f-g). Moreover, the total number of ATOH8 (+) CTCs increased in the HP group (Fig. 1h). Together, these data indicate that ATOH8 expression in CRC CTC is sensitive to LSS.

Additionally, ATOH8 expression was further assessed in samples from clinical cohort 2 via WB, revealing that ATOH8 was markedly upregulated in tumour tissues relative to adjacent normal tissues (ANTs) (Additional file 3: Figure S2a). ATOH8 expression levels were also higher in CRC cell lines than in NCM460 (Additional file 3: Figure S2b). Next, Kaplan-Meier analysis was conducted, revealing a significant correlation between ATOH8 upregulation and poor overall survival (OS) (P = 0.0335, TCGA) (Additional file 3: Figure S2c). Together, these results indicate that ATOH8 is upregulated in CRC tissues and may predict a poor prognosis.

Furthermore, we quantified ATOH8 expression levels in another cohort including 333 primary and 167 metastatic colorectal tumours, and ATOH8 was upregulated in metastatic CRC tissues (P < 0.0001, GSE131418) (Additional file 3: Figure S2d). We then established mouse subcutaneous and metastatic tumour models (Additional file 3: Figure S2e) and used serial sections, and immunohistochemical (IHC) staining to confirm ATOH8 upregulation in metastatic tumours in comparison with that in primary tumours (Fig. 1i, Additional file 3: Figure S2f). The results suggested that ATOH8 upregulation in CRC cells may happen during the “CTC stage” and be associated with tumour metastasis. Additionally, Kaplan-Meier analysis of progression-free survival (PFS) in 153 surgically treated patients with stage II-III colorectal cancer from GSE103479 was conducted according to the expression of the ATOH8, revealing a significant correlation between ATOH8 upregulation and poor PFS (P = 0.0169, GSE103479) (Additional file 3: Figure S2 g). And unexpectedly, in the clinical cohort 1, we found that the proportion of ATOH8 (+) CTCs was higher in the subgroup with high metastatic mesenchymal CTCs or a total CTC number ≥ 5 cells/5 mL (HP group) (Additional file 3: Figure S2 h), suggesting that ATOH8 (+) CTCs are potentially associated with a high risk of metastasis.

In summary, LSS can trigger the ATOH8 upregulation in m-CTCs, which may affect hematogenous metastasis and prognosis of colorectal cancer.

Using a mouse model of lung metastasis (Additional file 3: Figure S3a), we found that ATOH8 overexpression markedly increased tumour volume, tumour weight and metastatic foci in the lungs of nude mice, as expected (Fig. 2a-b, Additional file 3: Figure S3b). Furthermore, haematoxylin and eosin staining revealed a sharp increase in the rate of pulmonary metastasis in mice with ATOH8 overexpression (Fig. 2c-d, Additional file 3: Figure S3c). IHC staining of tumours indicated that ATOH8 and Ki-67 were upregulated upon ATOH8 overexpression, while cleaved caspase-3, an apoptotic marker, was downregulated (Fig. 2e-f). Interestingly, we found a significant increase in the number of CTCs in peripheral blood of mice in the ATOH8 overexpression group (Fig. 2g), which means that the increased lung metastasis after overexpression of ATOH8 may be related to the elevated CTC number.

Previous studies have reported that the number of CTCs is an independent predictor of PFS and OS in patients with metastatic colorectal cancer [28]. However, metastatic colonization is a highly inefficient process wherein most CTCs die, and the survived CTCs are rare [4]; therefore, it is important to identify the reason for upregulated CTC number in the ATOH8-overexpressing group of mice. Our further experiments found that ATOH8 efficiently promoted the migratory and invasive abilities (Additional file 3: Figure S4a-b), respectively. Furthermore, the effect of suppression of ATOH8 on apoptosis was detected by MTT and flow cytometry in suspended LoVo and SW480 cells, while cell cycle didn’t change significantly in ATOH8-overexpression group (Additional file 3: Figure S4c-e). Additionally, the qPCR results indicated that the anoikis markers N-cadherin, Vimentin and Laminin5 were increased after ATOH8 overexpressing in LoVo and SW480 cells, while E-cadherin is decreased (Additional file 3: Figure S4f). The data above suggested that ATOH8 overexpressing may increase the number of CTCs by inhibiting death rather than promoting proliferation. To monitor cell death in real time and reduce additional LSS interference, we added live/dead cell dye into the culture of m-CTCs, and the experiment determined that overexpression of ATOH8 in CRC m-CTCs leads to a decreased cell death rate (Fig. 2h). Furthermore, we aimed to examine the survival advantage of ATOH8-overexpressing CTCs in vivo. Thus, vector or ATOH8-overexpressing SW480 cells with luciferase were intravenously injected into nude mice (Additional file 3: Figure S3a). Whole-body imaging indicated that ATOH8 overexpression decelerated the reduction in CTCs (Fig. 2i). Moreover, tracking of GFP-labelled SW480 cells in mice revealed that CTCs were extremely rare in blood, with the ratio fluctuating between 0.1 and 1.42% (Fig. 2j-k, Additional file 3: Figure S3a). As shown in Fig. 2l, the percentage of CTCs undergoing cell death (PI-positive CTCs) gradually decreased in the ATOH8-overexpressing group, with a reduction of approximately 10–20%.

Concisely, these in vitro and in vivo experimental data suggest that ATOH8-overexpressing m-CTCs have a distinctive capacity to resist death and exert their vital effects in CRC metastasis.

To evaluate the potential mechanisms underlying the pro-survival effects of ATOH8 in m-CTCs, single sample Gene Set Enrichment analysis (ssGSEA) were performed in the metastatic colorectal cancer cohort from GSE131418 (Fig. 3a, Additional file 3: Figure S5a, Additional file 5: Table S3). The results indicated that the gene set of positive regulation of anoikis could be enriched in ATOH8_low group (Fig. 3a), supporting our hypothesis that ATOH8_high CTCs is prone to survival in the circulation. On the other hand, our previous studies have found that metabolic reprogramming is a key factor mediating the anoikis resistance of tumour cells [24]. Further, to explore the association between metabolism and the ATOH8 mediated CTC survival, ssGSEA analyses were performed, and the data revealed that only glycolysis, a vital metabolic pathway in tumour cells, was significantly different between ATOH8_high and ATOH8_low, rather than fatty acid metabolism, oxidative phosphorylation and amino acid metabolism (Fig. 3a), etc. A previous study has reported that activated glycolysis was closely associated with anoikis tolerance and cell survival in prostate cancer [29]. So, we hypothesised that activated glycolysis may be related to the pro-survival potential of ATOH8.

In fact, the present results confirmed that ATOH8 accelerates glucose uptake, via a 2-NBDG assay (Additional file 3: Figure S5b). Moreover, HK2 enzyme activity and both ATP and lactate production were induced via ATOH8 overexpression, while the opposite result was observed upon silencing ATOH8 (Fig. 3b-c, Additional file 3: Figure S5c). In short, ATOH8 did activate glycolysis in suspended CRC cells. Additionally, a significant positive correlation was observed between ATOH8 and the key glycolytic enzymes HK2, GLUT1 (Additional file 3: Figure S5d, Additional file 6: Table S4). To elucidate the molecular mechanism underlying ATOH8-induced glycolysis, we screened the expression of HK2, GLUT1, and LDHA on both transcriptional and translational level in ATOH8-overexpressing or -silenced suspended CRC cells. Our results showed that ATOH8 significantly upregulated glycolytic factors HK2 and GLUT1 at the mRNA level, rather than LDHA and MCT1 (Additional file 3: Figure S5e). Further, among the candidate factors, only HK2 showed an enhanced expression at the protein level (Fig. 3d), concurrent with the tissue IHC analysis mentioned above (Fig. 1i, Fig. 2e). More importantly, HK2 was also increased in tumour cells under LSS independent of ATOH8 overexpression (Fig. 3e). These results indicate that ATOH8 probably maintains CTC survival by promoting HK2.

Reduced ROS production and mitochondrial-associated HK2 inhibit glycolysis-mediated cell survival [30]. Our results showed that ATOH8 overexpression decreased ROS accumulation and potentially promoted mitochondrial localisation of HK2 binding to mitochondrial VDAC, contributing to cell survival (Additional file 3: Figure S6a-b). Furthermore, functional experiments revealed that promotion of glucose uptake, HK2 enzyme activity, and ATP and lactate production by ATOH8 overexpression was partially recovered after using HK2 inhibitors (2-DG and 3-BrPA) in suspended CRC cells (Additional file 3: Figure S6c-f). And as expected, in the live/dead cell vitality assay, HK2 inhibitors almost completely reversed the ATOH8-induced m-CTC survival (Fig. 3f). These results further corroborate the pro-survival function of ATOH8 in CRC m-CTCs by upregulating HK2.

ATOH8 is a bHLH domain transcription factor that binds to E-box sequences and activates transcription [19]. To determine whether there is a direct regulation relationship between ATOH8 and HK2, we designed three different sets of primers around the TSS (− 1000 to + 1 bp). In ATOH8-overexpressing CRC cells, ChIP-qPCR data revealed that binding of ATOH8 to DNA fragment 2 (HK2-p2, nt − 702 and nt − 524) was increased, with no significant enrichment at DNA fragments 1 (HK2-p1, nt − 866 and nt − 794) and 3 (HK2-p3, nt − 222 and nt − 145) (Fig. 3g-h). The HK2 promoter region contains two E-Box sequences (both in DNA fragment 2), as predicted using Genomatix (https://​www.​genomatix.​de/​, Fig. 3i). To further analyse the exact binding sites of ATOH8 on the HK2 promoter, we designed five plasmids for the HK2 promoter region, which are pGL4.10, pGL4.10-HK2-wt, pGL4.10-HK2-mut1, pGL4.10-HK2-mut2, pGL4.10-HK2-mut1 + 2 (Fig. 3i). Next, we conducted a dual luciferase reporter assay and the results showed that the luciferase activities of pGL4.10-HK2-wt and pGL4.10-HK2-mut1 were significantly increased in ATOH8-overexpressing 293 T cells, but not pGL4.10-HK2-mut2 and pGL4.10-HK2-mut1 + 2, suggesting that the E-box site (nt − 563 and nt − 558, CATATG) is essential for ATOH8-induced HK2 promoter activation (Fig. 3j). Together, our data show that ATOH8 promotes CTC survival by binding to HK2 and directly increasing its transcriptional activity.

Cellular plasticity is important for understanding tumorigenesis and tumour progression [31]. Accordingly, we observed that the molecular plasticity of ATOH8 driven by LSS facilitated m-CTC survival. Emerging evidence has revealed that cytokine secretion contributes to LSS-related mechanotransduction [32]. Furthermore, bioinformatics analysis revealed that VEGF is secreted by LSS-stimulated endothelial cells (GSE13712 and GSE52211) (Fig. 4a, Additional file 3: Figure S7a, Additional file 7: Table S5). We investigated cytokine and cytokine receptor levels in CRC cells and found that LSS notably upregulated VEGF (Additional file 3: Figure S7b). Consistent with ATOH8, VEGF upregulation and increased secretion were observed in both CRC m-CTCs through gradual increases in the intensity and duration of LSS (Fig. 4b-e, Additional file 3: Figure S7c). It is worth noting that previous literature has shown that when human exfoliated deciduous teeth were exposed to 4 dyn/cm² for about 4 h, VEGF secretion gradually reached platform phase [33], which is similar to the trend of ATOH8 mRNA in CRC m-CTCs (Additional file 3: Figure S1e).

VEGF, a key cytokine mainly secreted by endothelial cells and tumour cells, reportedly promotes angiogenesis, activates glycolysis, and induces anoikis tolerance [34]. Indeed, in VEGF-treated suspended CRC cells, we detected less ROS accumulation and anoikis (Additional file 3: Figure S7d-e). Meanwhile, VEGF improved the cell viability of CRC m-CTCs CTCs, either evaluated by live or dead cell assay (Fig. 4f). These results suggest that autocrine VEGF secretion from CRC m-CTCs may contributes to LSS-induced ATOH8 upregulation and changes in cell survival.

To examine this assumption, we measured ATOH8, HK2, BCL2, and BAX expression levels in LoVo and SW480 cell suspensions cultured in media supplemented with VEGF for 24 h (Fig. 4g, Additional file 3: Figure S7f). As predicted, VEGF upregulated ATOH8, HK2, and BCL2/BAX ratio in CRC cells. Especially, increased nuclear translocation of ATOH8 and HK2 enzyme activity was observed in VEGF-treated suspended CRC cells (Fig. 4hg, Additional file 3: Figure S7 g). These results indicated that VEGF could promote ATOH8 expression and activate downstream glycolysis. In addition, Bevacizumab, a humanised mouse anti-human VEGF antibody, could inhibit the CRC m-CTCs survival mediated by VEGF (Fig. 4i) and block the LSS-induced ATOH8 and HK2 upregulation (Fig. 4j), implying the ATOH8 upregulation induced by LSS is related to the secretion of VEGF.

Furthermore, as the rescue experiments shown, siATOH8 partially reversed VEGF-induced upregulation of HK2 activity and ATOH8, HK2, and BCL2/BAX ratio (Fig. 4k, Additional file 3: Figure S7 h) and restored VEGF-induced reduction of ROS production and anoikis (Additional file 3: Figure S7i-j) in suspended CRC cells. And ATOH8 suppression by siRNA partially reversed the pro-survival phenotype of CRC m-CTCs owing to VEGF stimulation (Fig. 4l). Together, these findings suggest that the promotion of CRC m-CTC survival mediated by ATOH8 partially depends on LSS-induced autocrine VEGF signalling.

VEGF exerts its effects by binding to VEGF receptor 2 (VEGFR2) in CRC cells [35]; similarly, the present results show that VEGFR2 was upregulated in CRC m-CTCs exposed to LSS (Fig. 5a). Moreover, VEGFR2 inhibitor ZM323881 and Apatinib markedly downregulated ATOH8 in CRC cell suspensions (Fig. 5b). Then, we investigated whether VEGFR2 regulates VEGF-ATOH8 signalling in CRC m-CTCs. Indeed, blocking of VEGFR2 signals partially reversed HK2 activity upregulation and ATOH8, HK2, and BCL2/BAX expression induced by VEGF in suspended CRC cells (Fig. 5c, Additional file 3: Figure S8a). Furthermore, the VEGF-induced reduction in cellular ROS levels and CTC death of CRC m-CTCs was partly restored upon treatment with VEGFR2 inhibitor (Fig. 5d, Additional file 3: Figure S8b). As predicted, ATOH8 overexpression partially reversed the downregulation of ATOH8, HK2, and BCL2/BAX ratio via inhibition of VEGFR2 signals (Fig. 5e) and CTC death induced by ZM323881 (Fig. 5f). Taken together, these data suggest that VEGFR2 is relatively responsible for VEGF-mediated ATOH8 upregulation in CRC cells.

Moreover, the AKT and ERK signalling pathways, both of which are downstream of VEGFR2, were reportedly associated with cell survival [36, 37]. However, it remains unclear whether AKT or ERK signalling is responsible for CTC survival induced by the VEGF/VEGFR2/ATOH8 axis. We found that the treatment of CRC cell suspensions with AKT inhibitors (AZD5363 and MK-2206), rather than ERK inhibitor (SCH772984), downregulated ATOH8 (Additional file 3: Figure S8c-e). Consistent with this, the results of ssGSEA also indicated that VEGF-mediated ATOH8 upregulation may primarily depend on the AKT signalling pathway (Fig. 5g, Additional file 5: Table S3). As illustrated, blockade of AKT signalling partially reversed VEGF-induced upregulation of HK2 activity, and the expression of ATOH8, HK2, and BCL2/BAX, but restored the reduced ROS production caused by VEGF in CRC cells (Fig. 5h, Additional file 3: Figure S8f-g). And AKT inhibition could also partially reversed the protective effects of VEGF on cell survival in CRC m-CTCs (Fig. 5i). What’s more, ATOH8 overexpression attenuated AKT inhibitor (AZD5363) induced downregulation of ATOH8, HK2, and BCL2/BAX ratio, and partly reversed cell death in CRC m-CTCs (Fig. 5j-k). These findings reveal that VEGF upregulates ATOH8 by selectively activating VEGFR2-AKT signalling, with major implications for understanding and targeting the mechanism underlying m-CTC survival (Fig. 6).