Expression and correlation of COX-2 and NUCB1 in colorectal adenocarcinoma

Introduction

Bevacizumab is commonly applied to patients with middle-to-advanced colorectal adenocarcinoma with metastasis. By binding to vascular endothelial growth factor-A (VEGF-A), it blocks VEGF-A-induced angiogenesis, consequently inhibiting tumor metastasis and prolonging patients’ survival (Midgley & Kerr, 2005). However, bevacizumab is prone to drug resistance, which brings great difficulties to cancer treatment (Midgley & Kerr, 2005; Hurwitz et al., 2004). Many studies have shown that interfering with the expression of VEGF can affect the efficacy of bevacizumab (Feng et al., 2020). COX-2 is one of the influencing factors. Our preliminary clinical data once again confirmed the correlation between COX-2 and bevacizumab resistance. COX-2-induced VEGF overexpression may be a mechanism of bevacizumab resistance (Rao et al., 2004). Nucleobindin1 (NUCB1) was first identified in systemic lupus erythematosus (SLE) and was characterized by a variety of functional binding domains, including DNA binding domains, nuclear localization signals, etc., which gave it many functions (Sinha, Pattnaik & Aradhyam, 2019). Structurally, the supposed binding domain of NUCB1 can bind with COX-2, and they can be colocalized (Leclerc et al., 2008). Functionally, the addition of hrNUCB to purified hrCOX-2 showed increased PGE2 synthesis, suggesting that NUCB enhances COX-2 activity (Leclerc et al., 2008). Therefore, it is reasonable to speculate that NUCB1 and COX-2 are closely related. Immunohistochemistry was applied to detect the expression of COX-2 and NUCB1 in colorectal adenocarcinoma and adjacent tissues. By analyzing their relationship with clinicopathologic parameters, as well as their correlation, it is possible to explore a potential pathway for the occurrence and development of colorectal cancer. It may pave the way for exploring the mechanism of COX-2-induced bevacizumab resistance.

Materials and Methods

Results

Bioinformatics analysis

The UALCAN database showed that the expression of COX-2 is up-regulated in cancer tissues, especially in stages II and III (Figs. 1A and 1B, P < 0.01), suggesting that COX-2 may be a cancer-promoting factor in colorectal adenocarcinoma. The STRING database provided clues to the correlated molecules of COX-2. Figure 2A only showed a part of them. After preliminary screening, six molecules were identified, namely APC, CAV-1, CTNNB1, MMP7, NUCB1, and DUSP1. Owing to the fact that the correlation between NUCB1 and COX-2 in colorectal adenocarcinoma has not been reported, we finally locked on NUCB1 (Fig. 2B predicts the correlation between COX-2 and NUCB1). The UALCAN database showed that the expression of NUCB1 was decreased in cancer tissues of stages I to IV, suggesting that NUCB1 was a negative regulatory molecule, probably (Figs. 1C and 1D, P < 0.01).

RFS, OS, and PPS times were predicted by the Kaplan–Meier Plotter database (Fig. 3). In comparison to low expression, COX-2 high expression had a longer RFS time (P < 0.05). NUCB1 was the contrary (P < 0.05), though. OS time of COX-2 and NUCB1 had the same trends (P < 0.05). The PPS time of COX-2 high expression was shorter than that of low expression (P < 0.05). However, the PPS time of NUCB1 was not statistically significant.

The expression of COX-2 and NUCB1 in colorectal cancer was predicted using the Human Protein Atlas (HPA) (Fig. 4). In healthy colon tissue, NUCB1 was highly expressed while COX-2 was not present. COX-2 and NUCB1 expression ranged from low to high in cancerous tissue.

The expression of COX-2 and NUCB1 in immunohistochemistry

COX-2 and NUCB1 were mainly located in the cytoplasm. In 56 cases of cancer tissues, 39 showed different levels of COX-2, with a positive rate of 69.64%, whereas only 16 cases and 28.57% were in the adjacent cancer tissues correspondingly. The chi-square test is X² = 19.85 as well as P < 0.005, confirming that COX-2 was significantly upregulated in colorectal adenocarcinoma. There were 36 cases with a positive NUCB1, accompanied by a positive expression rate of 64.29%, corresponding to 11 cases and 19.64% in the cancerous adjacent tissues. The chi-square test result was X² = 24.79, with a P < 0.005, indicating that NUCB1 was markedly increased in colorectal adenocarcinoma (As shown in Fig. 5 and Table 3).

Table 3:

Expressions of COX-2 and NUCB1 in colorectal adenocarcinoma.

	n	COX-2				Positive rate	P	NUCB1				Positive rate	P
		–	+	++	+++			–	+	++	+++
Cancer tissue	56	17	10	12	17	69.64%	P < 0.005	20	17	8	11	64.29%	P < 0.005
Adjacent tissue	56	40	6	3	7	28.57%		45	8	2	1	19.64%

DOI: 10.7717/peerj.15774/table-3

Discussion

COX-2 is nearly not present in normal tissues but can be induced by growth factors, cytokines, TNF-α, and vascular activator peptides to participate in tumor growth, metastasis, invasion, and angiogenesis (Wang & Dubois, 2006; Zhan et al., 2004). Through bioinformatics analysis, we discovered that COX-2 in colorectal cancer was significantly up-regulated, especially in stages II and III. And COX-2 high expression had the highest probability in terms of RFS time but opposite in terms of OS and PPS time. COX-2 was negative in normal colon tissue and presented to varying degrees as positive in cancer tissue. These results suggest that COX-2 may be a potential promoter factor. According to immunohistochemistry, the positive rate of COX-2 in cancer tissues was 69.64%, which was a lot higher than the negative rate of 28.57% in healthy tissues nearby (P < 0.05). It validates previous bioinformatics analysis and basically conforms to the erstwhile experimental results of approximately 85% of colorectal cancer expressing COX-2 and about 20% of normal colorectal tissues (Eberhart et al., 1994; Ogino et al., 2008; Zhang & Sun, 2002). In this research, COX-2 was closely correlated with TNM staging, differentiation degree, and lymphatic metastases. And the proportion of COX-2 expression increased with TNM staging. It gives an insight into the real possibility that COX-2 may provide advantageous conditions for the growth of cancer cells and aggravate their invasiveness and migration. Owing to the decreasing differentiation degree, the positive rate of COX-2 also decreased, indicating that COX-2 may be involved in the differentiation process of colorectal cancer. In TNM stage IV and distant metastases, the expression rate of COX-2 reached 100%. Despite the fact that there are so few advanced cases, it is possible to speculate that COX-2 is almost exclusively expressed in advanced, high-grade colorectal cancer. It was consistent with previous research results: COX-2 was associated with the advanced stage and high proliferation activity of colorectal cancer (Zhang & Sun, 2002). Therefore, COX-2 may be a contributing factor and a vital target for diagnosis and treatment. In addition, there was no statistical difference in COX-2 and age, gender, distant metastases, or serosal invasion, which may be attributed to an insufficient sample. We will continue to expand the sample and attempt to perfect the relevant results. Not only in colorectal cancer, but also in several animal models, selective COX-2 inhibitors can greatly weaken cancer growth. COX-2 may be a crucial factor in the occurrence and deterioration of cancer. Some evidence has been provided that exogenous induction of COX-2 in human oral epithelial keratinocytes can trigger the proliferation of the keratinocytes and secretion of cyclin D1, taking a further step toward malignant transformation (Wang et al., 2019). COX-2 is a key enzyme to generate PGE2, which can enhance the proliferation of ovarian cancer cells by binding to the corresponding receptor and then catalyzing the phosphorylation of NF-κB/P65 (Zhang et al., 2019). In light of this evidence, we will continue to detect COX-2 mRNA and protein expression at the cellular and mouse levels to verify the function of COX-2 in colorectal cancer and go a step further toward exploring whether COX-2 is an important cause of bevacizumab resistance.

Through the retrieval of the STRING database, we initially screened out six molecules, namely APC, CAV-1, CTNNB1, MMP7, NUCB1, and DUSP1, and then searched and analyzed them by PubMed. We found that there were few relevant studies to NUCB1, such as expression level and function. The relationship between NUCB1 and COX-2 has also only been mentioned in vitro. Accordingly, we chose NUCB1 as the object of study. NUCB1 was first discovered in systemic lupus erythematosus (SLE), which bound to nucleosome ladder DNA to induce autoimmunity and thymus apoptosis (Miura et al., 1992; Kanai et al., 1995). The most typical feature of NUCB1 is its multiple functional domains: DNA binding sites, heterodimerization domains, two EF motifs, Ca2+ binding sites, nuclear localization signals, leucine zipper regions, and non-classical domains, including G-protein binding domains and cyclooxygenase binding domains (Miura et al., 1992). These endow NUCB1 with important implications in a variety of cellular processes, including autoimmunity, intracellular signaling, osteogenesis, inflammation, and cancer. For example, the DNA binding domain of NUCB1 can combine with the E-box sequence of the Cripto promoter to activate many intracellular signaling pathways and trigger the biological behavior of cancer cells (Sinha, Pattnaik & Aradhyam, 2019). The non-classical binding domain of NUCB1 binds to cyclooxygenase, which has been confirmed by yeast two-hybrid systems and immunoprecipitation experiments (Ballif et al., 1996). This gives us confidence to continue our research. Through bioinformatics, NUCB1 in colorectal cancer tissues was reduced compared to normal tissue, both at the mRNA and protein levels. However, the OS time of NUCB1 high expression was shorter. These results are contradictory. In this study, compared to 19.64% in adjacent tissues, the positive rate of NUCB1 in cancer tissues was significantly increased to 64.29% (P < 0.05), which is consistent with the research results of Sinha, Pattnaik & Aradhyam (2019). In addition, NUCB1 was significantly relevant to TNM stage, differentiation degree, and lymphatic metastases, suggesting that NUCB1 may induce the proliferation and metastasis of colorectal cancer. The higher the differentiation degree of colorectal cancer, the higher the positive rate of NUCB1, indicating that NUCB1 may be involved in the differentiation of colorectal cancer. The Spearman correlation test showed that the correlation coefficient between NUCB1 and COX-2 is r_s = 0.6312 and P < 0.001, revealing a significant positive correlation between NUCB1 and COX-2. This study did not show any statistical difference between NUCB1 and gender, age, tumor site, serosal invasion, or distant metastases. Interestingly, NUCB1, along with COX-2, was highly expressed in TNM stage IV and distant metastases. It is rational to speculate that the combination of NUCB1 and COX-2 induces the progression of colorectal cancer. Next, we will continue to improve cell experiments to further demonstrate the expression and function of NUCB1 in colorectal cancer cells, and explore whether the combination of NUCB1 and COX-2 induces bevacizumab resistance.

Supplemental Information