Structure and dendrogram of SYNE3
SYNE3 located at 14q32.13, containing 18 exons and 17 introns (Fig. 1a). SYNE3 encoded 2 isoforms, nesprin-3α and nesprin-3β. Nesprin-3α was the dominant isoform and the main undertaker of nesprin-3 functions. The protein structure of nesprin-3α consisted of two parts of spectrin repeats, a KASH domain and a coiled-coil region (Fig. 1b). To study the conservation of SYNE3 among distinct species, we compared protein sequences encoded by SYNE3 among six different species (Fig. 1c), showing that Homo sapiens SYNE3 shared 77, 30, 78, 77 and 79% identity to Mus musculus, Danio rerio (zebrafish), Oryctolagus cuniculus, Rattus norvegicus and Ovis aries, respectively. It presented that SYNE3 was highly conserved in mammals, but varied significantly between human and zebrafish, a common animal model to study SYNE3 functions in neurons and stem cells. Moreover, a phylogenetic tree was constructed to analyze the conservative relationship among SYNE family members (Fig. 1d). In this tree, the SYNE family was divided into two main clusters; one included two SYNE2 isoforms, and the other contained the other SYNEs. More specifically, SYNE3, SYNE1, and the left SYNE2 isoforms were in the same subset, while SYNE4 was not.
Expression of SYNE3 in BALB/c mice tissues
As BALB/c mice was a widely-used animal model in studies on tumorigenesis and tumor migration, we first detected the SYNE3 expression of various normal tissues in BALB/c mice (Fig. 2a). SYNE3 staining was positive in the gut, lung, trachea, esophagus, stomach, and heart, while not obvious in the liver, thyroid, brain, spleen, kidney, and pancreas. In the gut, weak positive staining of SYNE3 was found in the cytoplasm of the intestinal villi and gland. The lung presented moderate staining in the nucleus of epithelial cells. SYNE3 was strongly detected in the nucleus in all layers of the trachea, and moderate in the cytoplasm in the mucosal layer. For the esophagus, nucleus staining was strong in all layers, and cytoplasmic staining was moderate in all layers except submucosal cells. In the stomach, weak staining was observed in the cytoplasm of chief cells and parietal cells. In the heart, only the nucleus of cardiomyocytes exhibited weak staining.
Expression of SYNE3 in normal tissues and tumor tissues of the human
We carried out IHC on samples of eight types of normal human tissues and their corresponding tumor tissues (Fig. 2b). In liver, only the cancer cells presented weak SYNE3 staining in the cytoplasm. No SYNE3 was detected in both the normal and tumor cervix tissues. Weak positive staining was found in the cytoplasm of normal gut epithelial cells, compared with moderate positive staining in cytoplasm of colon adenocarcinoma cells. Both small intestine and esophagus showed no staining in normal tissue and weak staining in their cancer cells. In terms of the kidney, moderate to strong positive staining was present in both the nucleus and cytoplasm of normal renal tubular epithelial cells, while only weak staining was found in whole cells of renal clear cell carcinoma. Breast and lung presented the most obvious expression differences between the normal and the tumor tissues. In normal tissues, strong positive staining was observed in the nucleus and cytoplasm of breast luminal epithelial cells and the nucleus of lung alveolar epithelial cells. While in the corresponding tumor samples, SYNE3 staining was only weakly present in cancer cells.
We also referred to UCSC database (Additional file 1: Fig. S1a). Compared with our experimental data, SYNE3 mRNA was expressed mostly in the breast and lung as well. It was contradictory that the SYNE3 mRNA level in the kidney was low, though our IHC result showed moderate to strong positive staining.
Meanwhile, according to GEPIA database, SYNE3 was significantly expressed less in 9 cancer types, including invasive breast carcinoma (BRCA), lung adenocarcinoma (LUAD), and lung squamous cell carcinoma (LUSC)(P < 0.0001), which were consistent with our IHC results in these tumors. However, only in acute myeloid leukemia (LAML), SYNE3 expression level is higher in tumor tissue compared with normal one (P < 0.0001). The expression change in LAML is the most noticeable, with its fold change reaching 9.26 (Additional file 1: Fig. S1b).
Prognostic value of SYNE3
To discover prognostic cancer value of SYNE3, we investigated the connection between SYNE3 expression level and DFS of patients with cancer, finding that higher SYNE3 expression brought longer lifespan to patients with renal clear cell carcinoma (KIRC) (P = 0.033, HR = 0.67) or with cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC) (P = 0.046, HR = 0.56) (Additional file 2: Fig. S2a). In terms of analysis of OS, five out of 33 kinds of the tumor showed significant results (Additional file 2: Fig. S2b). Specifically, we found that in KIRC (P < 0.001, HR = 0.58, Kaplan–Meier), LUAD (P = 0.011, HR = 0.67), CESC (P = 0.02, HR = 0.57) and squamous cell carcinoma of head and neck (HNSC) (P = 0.046, HR = 0.76), patients with high SYNE3 expression survive longer than others with low SYNE3 expression level. However, the case was in reverse in brain lower-grade glioma (LGG) (P = 0.034, HR = 1.5), in which patients with higher SYNE3 expression live even shorter. These results suggested SYNE3 prognostic value in various cancer types.
Construction of ceRNA network of SYNE3 in LUSC
We analyzed the upstream regulation of SYNE3, screening miRNAs or lncRNAs, which targeted SYNE3. We found 37 miRNAs possibly targeting SYNE3 from DIANAmT database, 2289 from miRWalk database, 317 from TargetScan database, and 45 from mirDIP database (Fig. 3a). As a result, we screened hsa-miR-330-3p and hsa-miR-149-5p as the most vital miRNA regulators by overlapping predictions of four databases. We also presented the complementary sequences between SYNE3 and miR-330-3p and miR-149-5p (Fig. 3b). Then, we predicted eleven lncRNAs that can bind with hsa-miR-330-3p and 19 lncRNAs targeting miR-149-5p. In this way, we constructed a lncRNA-miRNA-mRNA network (Fig. 3c), in which lncRNA competitively bound with miRNA and weakened the suppression from miRNA to SYNE3.
The regulatory mechanisms we predicted were only limited to the genetic level for using TCGA data, as there should be many mechanisms after transcription that can also influence the gene expression. Therefore, to better verify our predicted mechanisms, we tend to choose a cancer tissue whose mRNA difference is consistent with protein difference to minimize influence of modifications after transcription. Based on our results in IHC and bioinformatic analysis, BRCA, LUSC and LUAD presented most significant differences in both mRNA and protein level of SYNE3 between normal and tumor tissues. As previous studied have revealed SYNE3 connection with lung cancer development [14, 16] compared with few reports on BRCA, so we chose to focus more on SYNE3 role in lung cancer in this article. Considering that SYNE3 expression difference can lead to OS difference in LUAD, we used LUAD as an example to illustrate our ceRNA network here.
SYNE3 expression (Fig. 3d), miRNAs expression (Fig. 3e) and expression of lncRNAs were compared between tumor tissues of LUAD and normal ones. SYNE3 expressed significantly lower in tumor tissues of LUAD, and both two miRNAs expressed higher in tumor tissues instead. In terms of lncRNAs, RP11-2B6.2 and RP11-67L2.2 presented a significant decrease in their expression (Fig. 3f). Combining with these analyses and interactions predicted by the database above (Fig. 3g), we might suppose that RP11-2B6.2 and RP11-67L2.2 bind with hsa-miR-149-5p and hsa-miR-330-3p respectively to weaken miRNA suppression toward SYNE3 in LUAD (Fig. 3h).
Transcriptome analysis of SYNE3 in LUSC
To better understand the upstream mechanisms of SYNE3, we continued to do some transcriptome analysis on SYNE3, still using the example of LUAD. First, we constructed a transcriptional regulatory network involving 100 transcriptional factors predicted by GCBI database (Fig. 4a). Then, we selected 21 TFs with significant expression differences in normal tissues and tumor tissues of LUAD (Fig. 4b). Among them, SATB1 was correlated with SYNE3 expression (Fig. 4c, r = 0.418, P < 0.0001). Combined with the analysis of ceRNA network above, we wondered if the predicted SYNE3-related miRNAs also participated in TF regulation. By using the data of Tarbase, mirDIP, miRWalk and our predicted results, we identified miR-149-5p as an upstream regulator of SATB1 (Fig. 4d). Moreover, in LUAD, SATB1 and SYNE3 were both downregulated, while miR-149-5p was expressed more. Based on these, a SATB1-miR-149-5p-SYNE3 transcriptional network in LUAD was constructed (Fig. 4e).
KEGG and GO pathway enrichment analysis of SYNE3
To investigate the downstream regulation of SYNE3, a PPI network (Fig. 5a) was built with 40 interacting genes of SYNE3 predicted by the STRING database. We performed KEGG analysis of SYNE3 and its 40 interacting genes on KOBAS. Finally, We acquired 6 KEGG pathways these 41 genes were enriched in, specifically ribosome, apoptosis, arrhythmogenic right ventricular cardiomyopathy (ARVC), hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy and p53 signaling pathway, with KEGG map of ribosome especially presented (Fig. 5b).
For GO enrichment analysis, SYNE3 and its interacting genes were significantly and credibly enriched in 12 Biological processes (BPs), 11 Cellular components (CCs), and 3 Molecular functions (MFs) (Fig. 5c). These genes mainly encoded protein from LINC, cytoskeleton, nucleoskeleton and ribosome, likely participating in nucleus adjustment and transcription.
Immune infiltration level analysis in cancer associated with SYNE3
Higher SYNE3 expression was linked to better clinical outcomes in our analysis, suggesting its tumor-suppressing functions. Here, we explored this function in the aspect of immunity. First, we screened cancer types whose immune infiltration was correlated with SYNE3 expression. Accordingly, we found that SYNE3 expression was significantly correlated with dendritic cell, neutrophil, CD4 + T cell, macrophage, CD8 + T cell and B cell in 30, 30, 27, 26, 25, 25 and 22 types of cancer respectively.
We then analyzed each cancer type and found ten types of cancer with significant results in all immune cell types and purity. Mostly, in HNSC, KIRC and LUAD, both OS and immune infiltration were positively correlative with SYNE3 expression (Fig. 6a). Among these three cancer types, the survival of LUAD presented a closer relationship with the level of immune cell infiltration (Fig. 6b), which was positively correlated with the infiltration of B cell and dendritic cell.
Hence, we tended to figure out the role of SYNE3 expression in the immune infiltration of LUAD. In LUAD, higher SYNE3 expression corresponded with better clinical outcome (Fig. 6c), and the expression of SYNE3 was significantly downregulated (Fig. 6d) overall. We performed GSEA analysis on 535 TCGA samples divided into a high group (267 samples) and a low group (268 samples) by expression level. The result revealed that pathways indicative of DC cell and B cell activation significantly correlated with SYNE3 expression (Fig. 6e). To further confirm this point, we also analyzed the relationship between SYNE3 and biomarkers of DC cell [17, 18] and B cell [17] (Fig. 6f), with all correlations significant.
