3.1 LncRNAs differentially expressed in esophageal squamous cell carcinoma

Previously, our group had demonstrated 296 differentially expressed lncRNAs in ESCC patients’ blood samples that may have the potential to be utilized as liquid biopsy markers in ESCC [6]. Taking clues from our previous study, herein, we have explored the expression of the SOX2-OT and NEAT1 lncRNAs in ESCC (n = 50). We used qRT-PCR to assess the SOX2-OT and NEAT1 expression levels for this. Following 40 cycles of qRT-PCR, we look at SOX2-OT was expressed in only 36 out of 50 ESCC samples, and NEAT1 was expressed in 34 out of 50 ESCC samples. Further, we examine that SOX2-OT was significantly downregulated (fold-change = 2.02, p-value = 0.0001) [Fig. 1A] in blood samples of ESCC patients compared to healthy matched individuals. Similarly, NEAT1 was also down-regulated (fold change = 1.79, p-value = 0.0052) [Fig. 1B] in ESCC blood samples compared to matched healthy individuals. The downregulation of both SOX2-OT and NEAT1 suggest that both lncRNAs may act as tumor suppressors in ESCC patients.

Fig. 1

Expression profile of lncRNAs in ESCC patient’s blood sample. A Normalized expression of SOX2-OT (∆Ct values). B Normalized expression of NEAT1 (∆Ct values)

3.2 Downregulated SOX2-OT and NEAT1 possess a good diagnostic potential in ESCC

Using GEPIA online tools, we first examined how SOX2-OT and NEAT1 were expressed in relation to one another in esophageal squamous cell cancer. After that, we look at the SOX2-OT and NEAT1 have the potential to be used as clinical diagnostics, with Area under the ROC curve (AUC) values of 0.736 (Fig. 2A) [95% CI = 0.6479 to 0.8737; p-value = 0.0001] and 0.695 (Fig. 2B) [95% CI = 0.5701 to 0.8201; p-value = 0.0057] respectively. Additionally, we explored for the combined efficiency of the lncRNA SOX2-OT and NEAT1 panel in diagnosing ESCC. We reported that the combined Area under the ROC curve (AUC) value came out to be 0.6217 (95% confidence interval [CI] = 0.5245 to 0.7189; p-value = 0.0175]) [Fig. 2C]. The combined value of SOX2-OT and NEAT1 is lower than individual AUC values for SOX2-OT and NEAT1. Therefore, we predict that SOX2-OT and NEAT1 individually may have a better diagnostic potential than being used as a combined diagnostic panel.

Fig. 2

Diagnostic potential of lncRNAs in ESCC patient’s blood sample. A ROC curve to demonstrate the diagnostic potential of SOX2-OT. B ROC curve to demonstrate the diagnostic potential of NEAT1. C ROC curve to demonstrate the diagnostic potential of SOX2-OT and NEAT1 as a panel

3.3 Association of SOX2-OT and NEAT1 expression with the lifestyle and clinicopathological status of ESCC patients

Since it is a well-known fact that some lifestyle factors affect the pathogenesis of ESCCs, our next goal was to assess how altered SOX2-OT and NEAT1 levels related to the clinicopathological traits and lifestyle aspects of ESCC patients, including age, gender, tobacco use, alcohol use, consumption of hot drinks, tumor grade, and TNM stages. We found that ESCC patients' lifestyles and their clinicopathological features impacted SOX2-OT expression more significantly than healthy persons.

Notably, we observed a ~ 1.69-fold reduction in SOX2-OT expression in ESCC patients aged 18 to 50 years (n = 17; p = 0.0028) compared to age-matched healthy individuals. In contrast, no significant difference in SOX2-OT expression was detected in ESCC patients over 50 years of age (n = 19; p = 0.1871) relative to their age-matched healthy counterparts. Additionally, there was no statistically significant difference in SOX2-OT expression between the two age groups of ESCC patients (> 50 years and ≤ 50 years; p = 0.3420) [Fig. 3A].

Fig. 3

Association of SOX2-OT expression with ESCC patients’ lifestyle status and clinicopathological characteristics compared to healthy individuals. A Age of ESCC patient. B Gender of ESCC patients. C Tobacco smoking status of ESCC patients. D Alcoholic status of ESCC patients. E Consumption of hot beverages status. F Histopathological grading of the ESCC patients. G TNM staging of the ESCC patients. Bar graphs represent the normalized expression of SOX2-OT in ESCC patients compared to healthy individuals. The data are expressed as mean ± SEM where *represents p-value < 0.05, **represents p-value < 0.01, ***represents p-value < 0.001, and ****represents p-value < 0.0001 calculated using unpaired (Mann–Whitney test), paired t-tests and Chi-square test

The gender differences between the two groups of ESCC patients were also not discernible (p-value = 0.7961) in SOX2-OT expression. However, compared to healthy male individuals, male ESCC patients demonstrated a ~ 2.30-fold downregulation of SOX2-OT expression (n = 19; p-value = 0.0018) [Fig. 3B], In contrast, there was no discernible difference in the expression of SOX2-OT between female ESCC patients and healthy females (n = 17; p = 0.1086) [Fig. 3B]. In addition, ESCC patients with a history of tobacco use displayed a ~ 1.85-fold downregulation of SOX2-OT (n = 17; p-value = 0.0008), while non-smokers displayed a ~ 2.38-fold downregulation (n = 19; p-value = 0.0001) compared to healthy people (Fig. 3C). Smokers and non-smokers with tumors did not differ significantly from one another (p-value = 0.7961) [Fig. 3C]. Similar to this, ESCC patients with a history of alcohol use showed a ~ 1.85-fold downregulation (n = 22; p-value = 0.0030), whereas patients who were not alcoholics showed a ~ 2.4-fold downregulation (n = 14; p-value = 0.0001) compared to healthy people (Fig. 3D). As opposed to this, no distinction between the two groups of ESCC patients was seen (p-value = 0.2370) [Fig. 3D]. Further, patients with a history of using hot beverages displayed a ~ 2.03-fold downregulation (n = 22; p-value = 0.0055), whereas patients without a history of consuming hot beverages displayed a ~ 1.72-fold downregulation (n = 14; p-value = 0.0093) as compared to healthy individuals. (p-value = 0.7067) [Fig. 3E].There was no discernible difference between the two groups of ESCC patients.

There were no discernible changes between the groups of ESCC patients when we assessed the expression of SOX2-OT with different tumor grades (Fig. 3F). However, The TNM stages (I + II) and (III + IV) of ESCC patients demonstrated a significant difference between the two groups of ESCC patients (p-value = 0.0128) (Fig. 3G).

In keeping with the previous section, we specifically looked at the relationships between the dysregulated levels of NEAT1 and the clinicopathological traits and lifestyle factors, including age, gender, smoking, drinking, consuming hot beverages, tumor grade, and TNM stages of ESCC patients.

According to statistical analysis, we discovered no discernible difference between the expression of NEAT1 in ESCC patients between the ages of 18 and 50 (n = 16, p-value = 0.4077) compared to age-matched healthy individuals. However, compared to the healthy individuals, patients over 50 demonstrated a ~ 2.44-fold downregulation in NEAT1 expression (n = 18, p-value = 0.0106). Moreover, there was a significant downregulation of ~ 3.04-fold in the expression of NEAT1 between the two age groups of ESCC patients (p = 0.0078) [Fig. 4A]. Similarly, we could not observe any significant difference in the expression of NEAT1 in ESCC females (n = 17, p-value = 0.3479), whereas ~ 2.32-fold downregulation was noted in ESCC males (n = 17, p-value = 0.0023) compared to male healthy controls. However, no significant difference in the expression of NEAT1 was also observed between the two gender groups of ESCC patients (p-value = 0.2083) [Fig. 4B]. Additionally, in ESCC patients having a history of tobacco smoking, it was determined that there is no significant difference in NEAT1 expression in tobacco smokers (n = 9, p-value = 0.2584) as compared to their matched healthy individuals. However, ~ 1.89-fold downregulation of NEAT1 was examined in non-tobacco smokers (n = 25, p-value = 0.0036) as compared to healthy individuals. In addition, no significant difference was detected in smoker ESCC patients compared to non-smoker ESCC patients (p-value = 0.5700) [Fig. 4C].

Fig. 4

Association of NEAT1 expression with ESCC patients’ lifestyle status and clinicopathological characteristics compared to healthy individuals. A Age of ESCC patient. B Gender of ESCC patients. C Tobacco smoking status of ESCC patients. D Alcoholic status of ESCC patients. E Consumption of hot beverages status. F Histopathological grading of the ESCC patients. G TNM staging of the ESCC patients. Bar graphs represent the normalized expression of NEAT1 in ESCC patients compared to healthy individuals. The data are expressed as mean ± SEM where *represents p-value < 0.05, **represents p-value < 0.01, ***represents p-value < 0.001, and ****represents p-value < 0.0001 calculated using unpaired (Mann–Whitney test), paired t-tests and Chi-square test

Moreover, based on the alcoholic status of ESCC patients, NEAT1 was found to be ~ 2.31-fold downregulated (n = 15, p-value = 0.0017) in non-alcoholic patients, while no significant difference in the expression of NEAT1 was observed in ESCC patients with a history of alcohol consumption as compared to the healthy individuals (n = 19, p-value = 0.1188). Likewise, the difference was also non-significant between the two groups of ESCC patients (p-value = 0.1757) [Fig. 4D]. Next, ESCC patients with a history of hot beverage consumption (n = 21, p-value = 0.0545), as well as patients with no history of hot beverage consumption (n = 13, p-value = 0.3215), did not show any significant differences in the expression levels of NEAT1 as compare to healthy individuals. However, ~ 1.49-fold downregulation on the expression of NEAT1 was reported between the two groups of ESCC patients (p-value = 0.0055) [Fig. 4E]. Furthermore, the expression of NEAT1 was also evaluated with the numerous tumor grades of ESCC patients, and it was monitored that there was a ~ 2.88-fold downregulation in the expression of NEAT1 in patients with unknown tumor grade as compared to the patients with well-differentiated tumor (n = 23, p-value = 0.0151). However, there were no significant differences among the other groups of ESCC patients (Fig. 4F). Similar results were obtained for the two groups of ESCC patients with the TNM stage (I + II) and (III + IV), which could not show any significant differences in the expression levels of NEAT1 (p-value = 0.3723) [Fig. 4G].

By comparing the ESCC patients' lifestyles to those of healthy people, the findings presented above lead us to the conclusion that SOX2-OT and NEAT1 expression levels are influenced by lifestyle and seldom by clinicopathological characteristics.

3.4 SOX2-OT and NEAT1 target miRNAs and regulate various cancer-associated signaling pathways

As we know, cytoplasmic lncRNAs act as ceRNA by sponging miRNA activity in the human genome. In this regard, our group had recently shown two models of the molecular sponging mechanism by lncRNAs viz, unidirectional and bidirectional. Several online databases were used to screen the molecular targets of SOX2-OT and NEAT1 (see “Materials and methods”). Interestingly, we reported fifty-three miRNA targets of SOX2-OT from RAID v2.0 (http://www.rna-society.org/raid2/), fifty-seven miRNA targets from RNAInter (http://www.rnainter.org/), and hundred miRNA targets from the DIANA (https://diana.e-ce.uth.gr/home) databases. According to the DIANA, RAID v2.0, and RNA Inter databases, SOX2-OT's putative miRNA targets include fourteen common miRNAs (Fig. 5A). One hundred fifteen mRNA targets of has-miR-26a-5p predicted from Starbase (http://starbase.sysu.edu.cn/), miRBD (http://mirdb.org/), TargetScan (http://www.targetscan.org/vert_72/) and RAID (http://www.rna-society.org/raid2/) databases as shown in (Fig. 5). Additionally, GO analysis was used to evaluate the miRNA-SOX2-OT axis's associated biological, molecular, cellular, and KEGG pathway activities. In each domain, we chose the top fifteen enriched GO phrases. Four of them were shown to participate in several biological processes, including cell migration, the mitotic cell cycle, cell cycle regulation, and the cell cycle process (Fig. 5C). Similarly, cellular processes like microtubule organization and intracellular protein-containing complex are influenced by molecular processes, including protein dimerization activity, Guanyl nucleotide binding, and RNA binding (Fig. 5D). Additionally, we demonstrated that KEGG pathways in cancer were enriched for RNA degradation, the mTOR signaling pathway, and ubiquitin-mediated proteolysis (Fig. 5E).

Fig. 5

In-silico target prediction for SOX2-OT using online databases. A 14 common miRNAs were predicted by the DIANA, RAID v2.0, and RNA Inter databases to be the likely miRNA targets of SOX2-OT. B 115 mRNA targets of hsa-miR-26a-5p predicted from Starbase, miRBD, Target Scan, and RAID databases. C–E Gene Ontology molecular process, cellular process and KEGG pathway of predicted target hsa-miR-26a-5p of SOX2-OT

Next, we investigated the molecular targets of NEAT1. We observed 437 miRNA targets from the Starbase database, eight miRNA targets from the RNAInter database, and two miRNA targets from RAID v2.0. Two common miRNA targets were found to be hsa-miR-449b-5p and hsa-miR-106a-5p (Fig. 6A). 97 mRNA targets of has-miR-449b-5p were predicted from Starbase, miRBD, Target Scan, and RAID databases, as shown in Fig. 6B. In order to evaluate the linked biological, molecular, and cellular processes of the hsa-miR-449b-5p-NEAT1 axis, GO analysis was also used. The top 15 enriched GO items in each domain were chosen. Among them, hsa-miR-449b-5p and hsa-miR-106a-5p were discovered to be involved in several biological activities, such as the binding of transcription factors, the binding of kinase activity RNA, and cell adhesion molecule binding (Fig. 6C). Similarly, miR-449b-5p and miR-106a-5p also affect cellular processes such as ribonucleoprotein complex, microtubule organizing, endoplasmic reticulum, mitochondrion, etc. (Fig. 6D). Additionally, we found that miR-449b-5p and hsa-miR-106a-5p are enriched in the MAPK signaling pathway, endocytosis pathway, cell cycle, and regulation of actin cytoskeleton pathways in cancer (Fig. 6E).

Fig. 6

In-silico target prediction for NEAT1 using online databases. A 2 common miRNAs were predicted by the Starbase, RAID v2.0, and RNA Inter databases to be the likely miRNA targets of NEAT1. B 97 mRNA targets of hsa-miR-449b-5p were predicted from Starbase, miRBD, Target Scan and RAID databases. C–E Gene Ontology molecular process, cellular process and KEGG pathway of predicted target hsa-miR-449b-5p of NEAT1

Overall, the above data suggest that the downregulation of SOX2-OT and NEAT1 may be involved in the advancement of ESCC by targeting miRNAs and mRNAs by altering various pathways involved in carcinogenesis.

3.5 SOX2-OT and NEAT1 were regulated by co-expression networks of mRNAs and lncRNAs

Using co-expression networks, we found genes that are co-expressed with lncRNA and might serve as lncRNA targets. Based on this concept, we used Cytoscape and COXPRES db. Ver. 8.1 (https://coxpresdb.jp/) to visualize the network of two lncRNAs (SOX2-OT and NEAT1). Each lncRNA was found to have relationships with both mRNA and other lncRNAs in the network. Notably, seven lncRNAs SOX21 antisense divergent transcript 1 (SOX21-AS1), POU3F3 adjacent non-coding transcript 1 (PANTR1), FEZF1 antisense RNA 1 (FEZF1-AS1), long intergenic non-protein coding RNA 844 (LINC00844), sciatic injury induced lincRNA up regulator of SOX11 (SILC1), long intergenic non-protein coding RNA 1896 (LINC01896) and ARHGEF26 antisense RNA 1 (ARHGEF26-AS1) and forty four mRNAs, sex-determining region Y-box 2 (SOX2), SRY-box transcription factor 21(SOX21), sex-determining region Y-box 1 (SOX1), peripheral myelin protein 2 (PMP2), proteolipid protein 1 (PLP1), fibroblast growth factor 12 (FGF12), fatty acid binding protein 7 (FABP7), aquaporin-4 (AQP4), receptor protein-tyrosine kinase ErbB-4 (ERBB4), generalized anxiety disorder 2-item (GAD-2), glutamate metabotropic receptor 3 (GRM3) and cell adhesion molecule 2 (CADM2) etc. were anticipated to express alongside SOX2-OT. Among these, SOX2, SOX21, and SOX21-AS1 were co-expressed with SOX2-OT in the network in a direct manner. In addition, we found that several mRNAs, including SOX2, ZNF536 HESS, ASGL1, PRR18, AMER2, NKX6-2, TOX3, ASXL3, CXXC4, and ESRRG, were localised in the nucleus (as represented by purple balls) and four mRNAs were localised in the Plasma membrane (as shown by dark blue balls) [Fig. 7A].

Fig. 7

Co-expression network of SOX2-OT and NEAT1. A A total of seven lncRNAs and various mRNAs interact with SOX2-OT. B A total of seventeen lncRNAs and various mRNAs interact with NEAT1

Likewise, seventeen lncRNAs LINC00894, ASMTL antisense RNA 1 (ASMTL-AS1), LINC00893, LOC107983998, LOC105373102, LOC107984338, LOC107984754, LOC100289230, LOC101928762, PSMA3 antisense RNA 1 (PSMA3-AS1), AP1G2 antisense RNA 1 (AP1G2-AS1), LINC01004, LOC107984203, LINC00115, LOC105374298, LINC01000 and LINC02603 co-expressed with NEAT1 on the network. Moreover, various mRNAs, RNA-binding region-containing protein 3 (RNPC3), zinc finger RANBP2-type containing 2 (ZRANB2), zinc finger C3H1 domain-containing protein (ZFC3H1), arginine and glutamate-rich 1 (ARGLU1), heterogeneous nuclear ribonucleoprotein A3 (HNRNPA3), amyloid-beta A4 precursor protein-binding family B member 3 (APBB3), heat shock transcription factor 4 (HSF4), SUGP2 and protein phosphatase 1 regulatory subunit 3B (PPP1R3B) etc., nine mRNAs (RNPC3, ZRANB2, ZFC3H1, ARGLU1, HNRNPA3, APBB3, HSF4, SUGP2 and PPP1R3B) localised in the nucleus (as shown by purple balls) and one mRNA, vacuole membrane protein 1 (VMP1) were found to be localized in the plasma membrane (as shown by the dark blue ball) [Fig. 7B].

Altogether, our findings imply that dysregulation of the lncRNAs SOX2-OT and NEAT1 in ESCC simultaneously influences the expression of various other lncRNAs and mRNAs implicated in critical signalling pathways in the human biological system. By regulating the impacted gene expression along with SOX2-OT and NEAT1, this method may assist in developing a potential therapeutic strategy.