Desflurane induced EMT, migration, and invasion in CRC
To evaluate the potential effect of desflurane on CRC metastasis, the epithelial-like CRC cell line DLD-1 and HT29 were exposed to gas mixture constituting 21% oxygen, 5% carbon dioxide and 10.3% desflurane (an equal 1.7 minimum alveolar concentrations (MAC) in human of desflurane anesthetic concentrations) for 2 hours, followed by normal cell culture condition. After 48 hours, the morphology of both cell lines had changed to spindle-like, which is a typical characterization of EMT (Figure 1A). To further confirm and figure out whether this effect is time-dependent or independent, DLD-1 cells were exposed to desflurane for 0.5h, 2h, and 3h, respectively, and the EMT marker genes were quantified by qPCR. Mesenchymal markers SNAIL, ZEB1, and VIM were upregulated, while the epithelial marker CDH1 (coding E-cadherin) was decreased (Figure 1B). Interestingly, these changes were not found when cells exposed to desflurane for 0.5 hours. In addition, no significant difference in the EMT marker between the 2-hour-treated group and the 3-hour-treated group. Besides the mRNA level, the protein level of SNAIL, ZEB1 and VIM were increased, accompanied by a decrease of E-cadherin. Therefore, exposure to desflurane induced EMT in CRC cells in a time-independent manner (Figure 1C).
Since EMT initiates the spreading of tumor cells from the primary site to other organs via the bloodstream or lymphatic system, we asked whether desflurane affected the migration and invasion in CRC cells. Indeed, treatment with desflurane accelerated the closures in the scratch assay (Figure 1D). Moreover, transwell assay containing with or without matrigel showed that desflurane increased the ability of migration and invasion in DLD-1 cells (Figure 1E-F). To further confirm the EMT induction by desflurane, DLD-1 cells with or without treatment of desflurane were injected into mice through the tail vein. After 8 weeks, the mouse was sacrificed and the lung metastasis was analyzed by HE staining. Compared with the control group, the mouse injected with desflurane-treated formed more tumors in the lungs (Figure 1G-H). Therefore, desflurane-induced EMT was confirmed in vivo.
Downregulation of miR-34a was required for desflurane-induced EMT
Despite the EMT transcriptional factors such as SNAIL and ZEB1 which directly drive cells from epithelial to mesenchymal, miRNAs have been shown a critical role in the regulation of EMT or MET (mesenchymal-epithelial transition, a reverse biological process of EMT ) 19,20. To address whether miRNAs were involved in desflurane-induced EMT, we examined the expression changes of miRNAs after treatment with desflurane for 2 hours. According to literature, a panel of miRNAs with pro-metastasis or anti-metastasis ability were analyzed 21,22. Among those miRNAs, miR-34a was the only one that dramatically decreased after treatment with desflurane, while other miRNAs showed no significant changes, suggesting that miR-34a was a potential mediator in desflurane-induced EMT (Figure 2A). As expected, restoration of miR-34a by a miR-34a mimic prevented the migration and invasion induced by desflurane (Figure 2B). Furthermore, the changes in EMT-markers induced by desflurane were also neutralized by the ectopic expression of miR-34a (Figure 2C-D). Antagomir mediated downregulation of miR-34a in a mesenchymal-like CRC cell line SW620, which has a higher level of endogenous miR-34a, reversed the changes in EMT markers altered by desflurane (Figure 2E). Moreover, antagomir miR-34a prevented the induction in cell invasion induced by desflurane (Figure 2F). Taken together, desflurane induced EMT through downregulation of miR-34a in CRC cells.
LOXL3 is a direct target of miR-34a
miR-34a is a tumor suppressor microRNA and exhibits anti-tumor function by repressing its downstream targets that contribute to cancer progression. To figure out which targets of miR-34a were involved in desflurane-induced EMT in CRC, we first analyzed the correlation between all the putative targets of miR-34a and EMT markers based on data from The Cancer Genome Atlas (TCGA). According to the correlation, several genes were picked up and further analyzed. Before experimental validation of the putative targets of miR-34a, we checked the effects of those genes in desflurane-induced cell migration and invasion by using siRNAs. siRNA-mediated downregulation of LOXL3 dramatically prevented the increased migration and invasion capacity of DLD-1 cells induced by desflurane (data not shown). Therefore, we focused on LOXL3.
The seed-matching sequence in 3’-UTR of LOXL3 for miR-34a is conserved in human, mouse and other species, implying the evolutional importance of miR-34a/ LOXL3 axis (Figure 3A). Analysis of the TCGA database, which contains 461 primary CRC samples, showed mRNA expression of LOXL3 was inversely correlated with miR-34a (Figure 3B). In addition, ectopic expression of miR-34a mimics resulted in the downregulation of LOXL3 at both mRNA and protein levels (Figure 3C-E). Furthermore, reporter constructs containing the 3’-UTR of LOXL3 including the seed-matching sequence were repressed by co-transfection of miR-34a mimic, but not when the seed-matching sequence was mutated, demonstrating that LOXL3 was a direct target of miR-34a (Figure 3F).
LOXOL3 induced EMT, migration, and invasion in CRC cells
Gene Sets Enrichment Analysis (GSEA) showed LOXL3 expression from TCGA was highly enriched in the EMT gene signature, which was further validated by another two cohorts (Figure 4A). In further, LOXL3 was positively associated with EMT-TFs, while negatively associated with epithelial related genes, such as CDH1 and TJP1 (also known as ZO-1; Figure 4B). Ectopic expression of LOXL3 in DLD-1 cells induced SNAIL, ZEB1, VIM, and repressed E-cadherin (Figure 4C). Moreover, overexpression of LOXL3 increased cell capacity of migration and invasion (Figure 4D). siRNA-mediated knockdown of LOX3 in a mesenchymal-like cell line SW480 (with high expression of SNAIL/ZEB1/VIM, and high migration and invasion capacity) downregulated EMT marker genes and inhibited cell migration and invasion (Figure 4E). Therefore, LOXL3 itself has the ability to induce EMT, migration, and invasion in CRC.
Deregulation of miR-34/LOXL3 axis contribute to desflurane-induced EMT
Next, we addressed whether miRNA-34a/LOXL3 was involved in desflurane-induced EMT. As expected, LOXL3 was upregulated after desflurane treatment, while it was repressed by ectopic expression of miR-34a (Figure 5A). Furthermore, the siRNA-mediated knockdown of LOXL3 protected cells from desflurane-induced EMT, migration, and invasion (Figure 5B-C). Overexpression of LOXL3 lacking miR-34a binding site rescued the decrease of EMT, migration, and invasion induced by ectopic miR-34a expression after exposure to desflurane (Figure 5D-E). Taken together, the disorder of the miR-34a/LOXL3 axis contributed to the EMT, migration, and invasion induced by exposure of desflurane.
The clinical relevance of LOXO3
Since EMT is required for cancer metastasis which is the leading cause of cancer-related death, we wondered whether miR-34a/LOXL3 axis was clinical relevance. As miR-34a is a powerful tumor suppressor in CRC and had been well studied previously 23-25, here we investigated the role of LOXL3 in CRC clinical outcomes and features. Recently, the molecular subtypes of CRC base on gene expression have been proposed, which correlate gene expression with tumor behavior. Consensus molecular subtypes (CMS) of CRC represents one of the robust subtypes classifications and have been wildly accepted. Therefore, we first performed an analysis of the correlation between LOXL3 and molecular subtypes in CRC. Analysis of primary tumors from TCGA and another two datasets showed higher expression of LOXL3 were enriched in the CMS4 subtype (Figure 6A), which was characterized by EMT gene signature and poor survival. Since the CMS classification is base-on gene expression of primary tumor samples, the samples might be contaminated by stromal cells. Besides the patients with higher expression of LOXL3 in the whole tumors have poor survival, we also wanted to figure out that the features of tumor cells with a higher level of LOXL3. Therefore, we used another molecular subtype classification which bases on patient’s derived xenografts (PDX), and the contaminated mRNA from mouse stromal cells had been filtered, thus the classification was based on cancer cell-intrinsic gene expression. Among the five CRC intrinsic subtypes (CRISA-B), LOX3L was enriched in the CRIS-B subtype, which was also characterized by EMT and poor prognosis (Figure 6B).
As both CMS and CRIS classification showed LOXL3 was strongly associated with EMT and poor survival, we next analyzed the association of LOXL3 with CRC patients' survival. Analysis of the TCGA database showed patients with a higher mRNA level of LOXL3 had poor survival (Figure 6C). Furthermore, increased expression of LOXL3 was associated with CRC patients' lymph node metastasis (Figure 6D), and the expression level of LOXL3 was positively associated with the degree of lymph node metastasis (Figure 6E).