High expression of AGR2 in lung cancer is predictive of poor survival

Lung cancer is the leading cause of cancer related deaths in the United States and many parts of the world. According to the American Cancer Society approximately 224,210 individuals in the United States will be diagnosed with the disease and 159,260 (71 %) will die of it in 2014 [1]. Most of these are classified as non-small cell lung cancers (NSCLC), and include adenocarcinomas, squamous carcinomas, large cell carcinomas as well as less commonly occurring histopathologic patterns such as adenosquamous carcinoma. The search continues for bio-markers which can delineate outcomes related to long term survival or response to treatments.

Anterior gradient 2 (AGR2) is expressed by many solid tumor types and is known as an adenocarcinoma antigen [2]. In prostate cancer, expression of AGR2 is increased in the tumor epithelial cells; normal luminal cells have no detectable expression [3, 4]. In bladder cancer, expression of AGR2 is absent in ~75 % cases; normal urothelial cells have moderate AGR2 expression (in comparison to that in prostate cancer cells) [5]. Expression in primary prostate cancer is associated with lower grade disease (and longer biochemical recurrence-free survival) [6, 7]. Breast cancer also shows AGR2 expression [8]. Like prostate cancer, high AGR2 expression is associated with low grade and low expression with high grade. Besides differential expression in tumor grades, AGR2 expression is found preferentially in the adenocarcinoma type than the squamous cell carcinoma type of NSCLC [9]. In esophageal cancer, AGR2 can be used to improve distinguishing adenocarcinoma (99 % positive) from squamous cell carcinoma (37 % positive, typically in focal areas) in some circumstances [10]. However, oral squamous cell carcinoma has recently been reported to express AGR2 [11].

From the foregoing description, it would appear that cancer expression of AGR2 is more associated with diseases that have a lower potential of progression. This contrasts with the molecular functioning of AGR2 in cancer cells. Cell lines transfected with AGR2 produced metastasis in a xenograft model [12], showed gain of anchorage-independent growth and promoted tumor growth [13]. In gastric cancer cells, AGR2 expression was found up-regulated in a cancer cell subline with high metastatic potential for invasion to lymph nodes [14]. Although AGR2 positive primary prostate tumors were associated with better clinical outcome than AGR2 negative tumors, many distal prostate cancer metastases in bone and visceral organs showed strong immunostaining for AGR2 [7]. In ovarian cancer, high AGR2 expression was reported in 12 % high-grade serous type, which showed significantly lower overall survival [15]. Thus, AGR2 has a prognostic value for some cancers, but this is variable.

The widespread expression of AGR2 in human cancer underscores its potential importance in tumor biology. In some organs like the prostate and pancreas, AGR2 is absent in normal epithelial cells and present in cancer cells, while in others like the bladder, AGR2 is present in normal urothelial cells but absent in cancer cells of a majority of tumors. AGR2 expression being a feature of low grade cancer appears to contradict the metastasis promoting property of AGR2. This may be due more to our lack of understanding about the precise molecular functioning of AGR2 in cancer. In this paper, we describe AGR2 expression in NSCLC and adjacent non-neoplastic lung tissue. We also show it has prognostic value in younger NSCLC patients.

AGR2 is classified as a member of the protein disulfide isomerase (PDI) family of proteins on the basis of amino acid sequence homology. It is localized to the endoplasmic reticulum (ER) as well as secreted [3]. It has also been found in the nucleus and on the cell surface. PDI enzymes contain CXXC domains, which function in oxidation/reduction reactions and isomerization of disulfide bridges. Several PDI family members including AGR2 contain a CXXS motif, and exhibit a lower enzymatic activity than those with CXXC domains [2]. In normal tissues, AGR2 is strongly expressed in lung, trachea, GI tract, and tissues that contain mucin secreting cells or have endocrine functions [2]. Like the bladder urothelium, AGR2 is expressed in the normal bronchial epithelium. Unlike urothelial carcinoma cells, where AGR2 expression is frequently absent, the lung cancer cells retain AGR2. Like prostate, pancreatic, and breast cancer, tumor expression decreases somewhat with increasing grade. With this common trend, we expected cancer expression of AGR2 to be correlated with better patient survival as was demonstrated for prostate cancer [6]. However, this was not the case for lung cancer, in particular, for patients younger than 65. Instead, like high-grade serous ovarian cancer, higher cancer AGR2 expression is correlated with poorer survival. A possible explanation is that AGR2 functions differently in the cancer cell types, or that other differentially expressed genes affect AGR2 function. Depending on the cancer cell types, AGR2 may play roles in differentiation (as in low grade prostate tumors), cell growth and migration (as in advanced prostate cancer) [13].

AGR2 transcription is reported to be regulated by both estrogen and androgen receptors. Upregulation of AGR2 was seen in response to estradiol [26]. AGR2 was identified as being differentially expressed in estrogen receptor (ERα) positive breast cancer cell lines. Treatment of breast cancer patients with Letrazole resulted in decreased AGR2 transcription [26]. Interestingly, tamoxifen enhanced AGR2 expression [27], and AGR2 was found to be a predictor of poorer prognosis in tamoxifen-treated breast cancers [28]. Binding sites for FOXA1, FOXA2, hepatic nuclear factor 1, NF-κB and SOX9 were identified in the promoter region of AGR2 [26]. MAPK/ERK signaling was shown to regulate AGR2 expression in tumor cell lines in response to physiological stress from serum and oxygen depletion [29]. Vanderlaag et al. showed AGR2 could regulate cell growth, cell cycle progression, and survival in breast cancer cells through modulation of cyclin D1, ERα and survivin [30]. In addition, AGR2 can negatively regulate the p53 transcriptional pathway [31]. Dong et al. reported that binding of EGFR to AGR2 in the ER was necessary for transport of EGFR to the cell surface for EGFR signaling [32]. Huang et al. suggested various proteins and pathways influenced by AGR2 that may affect tumor metastasis including MAST1, SOAT2, POU2AF1 and IFI6 [33]. These studies demonstrate the multi-faceted functional aspect of this protein. Therefore, depending on the precise interaction in particular cell types AGR2 can have different effects on tumor cell behavior. Further studies on the molecular biology of AGR2 in cancer as well as normal cell types are needed.

Our study results differed in some aspects with those in the literature. In studying AGR2 in lung cancers, Fritzsche et al. [9] found no association of AGR2 expression and survival in 95 cases of NSCLC. Their conclusion was based on a smaller number of patients without the cohort dichotomization along age 65. When we considered our entire NSCLC population, we also did not find a significant predictive value for AGR2. The predictive value became significant only when younger patients were considered. Their paper, as well as that of Pizzi et al. [34], reported stronger staining in adenocarcinomas vs. squamous carcinomas, similar to our results. Inoue et al. [35] reported AGR2 expression in almost all adenocarcinomas, 96 % squamous carcinomas and 3/3 large cell carcinomas, as well as AGR2 RNA in positive lymph nodes. This contrasts with the general lower expression in squamous carcinomas and large cell carcinomas in our series. Chung et al. [36] reported that AGR2 was up-regulated almost 10-fold in lung adenocarcinoma, and negative expression (seen in 6 % of cases) was associated with poor survival. We did not observe AGR2 up-regulation in cancer cells vs. normal cells. Chung et al. [37], however, also reported a significantly higher mean serum AGR2 level even for stage I lung adenocarcinoma (AUC = 0.858), and positive serum AGR2 to be associated with recurrence after surgery (P = 0.025) and poor prognosis. Park et al. showed AGR2 to be present within the ER of intestinal secretory epithelial cells, to interact with MUC2 and to be a necessary requirement for mucus production and protection from colitis [38]. That adenocarcinomas with areas of mucinous histologic patterns in our study had higher AGR2 levels on average fits with a role for AGR2 in mucin production.

AGR2 could potentially become a viable serum biomarker for lung cancer [37]. In healthy people, the serum level of AGR2 is non-detectable or very low. This was borne out by mass spectrometry proteomics [39] and query of the PeptideAtlas database [40] that showed very low AGR2 signature peptide counts reported to date.

Normal lung epithelial cells express AGR2 and most NSCLC cells retain this marker. For NSCLC in younger patients, higher tumor AGR2 expression is correlated with a poorer survival in our cohort. Although not as strong a finding, patients with stage I cancer also showed poorer survival with higher levels of AGR2, which could suggest a deleterious role for AGR2 in lung cancer.