Aberrant expression of AKR1B1 indicates poor prognosis and promotes gastric cancer progression by regulating the AKT-mTOR pathway

Specimen collection

Our study recruited 115 patients with primary GC and paired contiguous normal stomach tissues after undergoing radical gastrectomy from the First Affiliated Hospital of Wannan Medical College during 2015–2016. Each subject participating in our study signed written informed consent forms. The agreement followed the ethical guidelines set out in the Declaration of Helsinki. The Affiliated Suzhou Hospital of Nanjing Medical University (ID: 2021240) and the institutional review board of the First Affiliated Hospital of Wannan Medical College (ID: 202248) approved this study.

Immunohistochemistry

We fixed the GC specimens and paired contiguous normal stomach tissues with formalin. Then, they were buried in paraffin and cut into 5-μm slices for immunohistochemical staining as referred in a previous study [13]. Sections were cultured at indoor temperature for 2 hours using 1:100 diluted anti-AKR1B1 antibody (1:100, Bioss, China). Two blinded researchers evaluated the staining scores to analyze immunohistochemistry (IHC) results. IHC score was decided by the multiple of the intensity (0 as negative; 1 as weak; 2 as moderate; 3 as strong) and extent (0 as 0–5%; 1 as 6–25%; 2 as 26–50%; 3 as 51–75%; 4 as >75%) score. The final point of 0 was considered −; 1–4 considered +; 5–8 considered ++; 9–12 considered +++. In our research, ++ or +++ was deemed to be a positive protein expression, and – or + was considered to be negative.

Cell cultures

The included human GC cell strains were acquired from the Nanjing Medical University (Nanjing, China). They were sprouted in RPMI medium 1640 (Hyclone, USA) with 10% FBS (Gibco, USA), 100 units/ml penicillin G sodium, and 100 μg/mL streptomycin sulfate (Gibco, USA). The clones were cultivated in 5% CO₂ at 37°C.

Transfection with shRNA

These cells were cultured until 80% fusion and transfected through a lentivirus AKR1B1-shRNA technology (Human AKR1B1-targeting shRNA were procured from GenePharma, China) according to the company’s manuals [14]. The cells after transfection were screened using G418 (Roche, Switzerland). Additionally, cell lines with a stable AKR1B1 knockdown were picked for subsequent research.

Protein preparation and western-blot test

The whole proteins were extracted with RIPA lysis buffer comprising proteinase and phosphatase inhibitors (Sigma, USA). The whole protein was separated with SDS-PAGE and shifted to a PVDF membrane which was sealed using 5% skimmed milk and probed using antibodies at 4°C all night long. We visualized the protein bands using chemiluminescence analysis and quantified them with ImageJ (NIH, USA). Antibodies comprised in western-blot were anti-AKR1B1 (1:1000, Bioss, China), anti-p-AKT (1:1000, Cell Signaling Technology, USA), anti-p-S6K1 (1:1000, Cell Signaling Technology, USA), and anti-β-actin (1:1000, Bioss, China) antibodies.

Colony-forming test

About a thousand cells were cultivated in every 6 pore plates for 10 days. The poietic cell lines were fixed by methanol and dyed with 0.1% crystal violet solution. Finally, they were calculated using an optical microscope (Nikon, Japan) equipped with a digicam (Nikon, Japan).

Transwell test

Cells were inoculated into the upper cavities at a density of 10,000 cells/200 μl in a serum free medium. Each well in the lower cavities was full of complete medium of 800 μl. After incubation for twelve hours, the filters were fixed using 4% paraformaldehyde and stained by 0.1% crystal violet, and the mean value of 5 random fields in each sample was determined.

Cell viability assay

The GC cell viability was measured with the Cell Counting Kit-8 (CCK-8) test (Abmole Bioscience, USA) in 96-well plates with 2000 cells/well based on the manufacturer’s protocols.

Subcutaneous xenograft

SPF male BALB/c naked mice (four weeks, approximately 19 g) were stochastically split into groups, each group of 5. After one week of adaptation, 10⁷ AKR1B1-KD or NC-shRNA AGS cells were injected to the right dorsal subcutaneous area of the mice on day 0. Then we measured the weight of mice fortnightly, and the dimensions of their subcutaneous neoplasms was determined. Four weeks later, the subcutaneous neoplasms were peeled and weighed. Ethical clearance by the Animal Ethics Committee of the Affiliated Suzhou Hospital of Nanjing Medical University (Suzhou, China) has been obtained for all the experiments in this study.

Statistical analysis

The data were shown as mean ± S.D. of a minimum of three stand-alone experiments. Data were analyzed using SPSS 22.0, GraphPad Prism 8, and R software (version 3.6.1). Two groups were compared using the unpaired and two-tailed t-test or Mann–Whitney U test. The IHC results were analysed using Chi-squared or Fisher's exact tests. P < 0.05 was determined statistically significant.

Data availability

Data will be made available on request.

Increased AKR1B1 expression in GC tissues and their association with clinicopathological indicators

Firstly, the AKR1B1 expression was detected in 115 GC tissues and contiguous normal stomach tissues using IHC. The IHC points were decided based on the intensity and extent of the staining (Figure 1A). The positive AKR1B1 expression levels among normal and cancer gastric cancer tissues were 43.5% and 26.1%, respectively. Based on the IHC scores, there was a high expression of AKR1B1 in GC tissues in contrast to contiguous normal stomach tissues (P = 0.003, Figure 1B). What’s more, AKR1B1 expression was observably elevated in patients with lymphaden infiltration (P < 0.001, Figure 1C) and TNM classification III (P < 0.001, Figure 1D).

Figure 1. AKR1B1 expression in GC and adjacent normal tissues. (A) The representative IHC images depicting in situ AKR1B1 expression in Case 1#, 2#, 3# GC and adjacent normal tissues (scale bar = 100 μm). (B–E) The IHC scores of AKR1B1 in (B) GC vs. normal tissues, (C) Tumors with or without lymph node metastasis, and (D) TNM stage I–II vs. stage III. (E) The overall survival analysis of AKR1B1^pos vs. AKR1B1^neg GC patients. (F) In the subgroup of TNM staging I–II, the overall survival analysis of AKR1B1^pos vs. AKR1B1^neg GC patients. (G) The overall survival analysis of AKR1B1^pos vs. AKR1B1^neg GC patients in the subgroup of TNM staging III.

The staining results were divided into negative and positive based on the IHC score to determine the AKR1B1 expression and its prospective influence in GC tissues. The pertinence between AKR1B1 expression within GC tissues and the clinical pathological indexes of GC patients was assessed. The results suggested AKR1B1 expression was observably correlated with age (P = 0.044), vascular invasion (P = 0.030), neural infiltration (P = 0.017), lymphaden metastasis (P < 0.001), and the TNM classification (P < 0.001, Table 1). In contrast, no pertinence could be detected between AKR1B1 expression and different clinical pathological indexes (P > 0.05, Table 1).

Aberrant expression of AKR1B1 in GC tissues indicates low survival rate

Depending on the close connections between AKR1B1 and clinicopathological indicators, the influence of AKR1B1 expression level was further assessed in GC tissues, affecting the overall survival rate of sufferers. The patients were classified based on AKR1B1^pos (AKR1B1 positive) and AKR1B1^neg (AKR1B1 negative) expressions. The survival rate of AKR1B1^pos sufferers was obviously lower than AKR1B1^neg sufferers (P < 0.001, Figure 1E). Further subgroup analysis in the light of TNM classification indicated that high AKR1B1 expression determined the low survival rate of sufferers in the stage I-II subgroup (P = 0.014, Figure 1F). Nevertheless, AKR1B1 expression was found to be statistically unrelated to overall survival rate in the stage III subgroup (P = 0.610, Figure 1G). Thus, AKR1B1 expression could only predict the prognosis of early-stage GC patients instead of an advanced stage.

A subgroup survival analysis was performed according to additional clinical pathological indexes to clarify the predictive effect of AKR1B1 expression on survival across different clinicopathological subgroups. The survival rate of AKR1B1^pos patients was markedly lower than AKR1B1^neg patients (P < 0.05, Figure 2), irrespective of gender, age, differentiated degree, the size of tumor, and vascular and perineural infiltration. Intriguingly, the AKR1B1 expression level did not significantly affect the tumor invasion depth T1-2 (P = 0.050) or patients with TNM stage III (P = 0.618). Conversely, increased AKR1B1 expression was related to worse prognosis in GC patients with the tumor invasion depth T3-4 (P = 0.021) or TNM stage I-II (P = 0.018). Meanwhile, high AKR1B1 expression could not predict a significant effect on survival, irrespective of whether they belonged to lymph node (P = 0.097) or no lymphaden (P = 0.232) metastases. Consequently, the AKR1B1 expression level can be detected in GC tissues to evaluate patient survival depending on the clinicopathological indicators.

Figure 2. The overall survival analysis of AKR1B1^pos vs. AKR1B1^neg GC patients in subgroups has been demarcated based on individual clinicopathological indicators.

The single factor analysis of Cox’s proportional risk model proved that differentiation degree, vascular invasion, neural invasion, tumor invasion depth, lymphaden metastasis, and AKR1B1 expression (P < 0.05, Table 2) are prognosis factors affecting GC patient survival. Lymph node metastasis acted as the independent factor in the multivariate analysis. Moreover, other indicators, such as AKR1B1 expression, were important in prognosis. By nomograms, we proved the pivotal part of the AKR1B1 expression level in GC to predict post-operative patient survival (Figure 3). The nomograms could forecast the overall survival rate of GC sufferers at 3 and 5 years. The negative or positive AKR1B1 expression could significantly affect the total points, demonstrating the pivotal part of AKR1B1 in forecasting GC patient survival (Figure 3).

Figure 3. The nomograms for predicting 3- and 5-year overall survival of GC patients. The points of each variable were obtained using a vertical line between each variable and the point scale. The predicted survival rate was connected with the total points through a vertical line drawn from the total points scale to the overall survival.

Ectopic AKR1B1 regulates the AKT-mTOR pathway to facilitate GC cell migration and proliferation

To gain mechanistic insights into the AKR1B1 effects on GC cell migration and proliferation and the prospective regulatory mechanism of the AKT-mTOR pathway, we firstly retrieved AKR1B1 expression in GC clones through the Cancer Cell Line Encyclopedia (CCLE) (Figure 4A, 4B). Whereafter, AKR1B1 expression was detected in GC clones using western blotting, consistent with CCLE platform data (Figure 4C, 4D). Therefore, AGS and MKN45 were selected with high expression of AKR1B1 for transfection with AKR1B1-shRNA and establishment of AKR1B1 knockdown cell lines (AKR1B1-KD). The fluorescence microscope images of AGS and MKN45 cells with control-shRNA and AKR1B1-shRNA transfection were detected to evaluate the knockdown effect (Figure 4E). We also investigated the effect of AKR1B1-KD GC cells with immunoblot, and AKR1B1 knockdown was evident in GC clones (Figure 4F, 4G).

Figure 4. AKR1B1-KD cells were constructed, and the regulatory relationship between AKR1B1 and AKT-mTOR pathway was evaluated. (A) The AKR1B1 expression in GC cell lines was searched from the CCLE platform. (B) The CCLE platform helped select AKR1B1 expression in four GC cell lines. (C) AKR1B1 expression in four GC cell lines was detected using western blotting. (D) The immunoblot result of AKR1B1 expression was semi-quantified in four GC cell lines with ImageJ. (E) The white and fluorescence microscope images of AGS and MKN45 cells were transfected with control-shRNA (NC) and AKR1B1-shRNA (KD). (F) Western blot indicates the AKR1B1 protein levels in NC and AKR1B1-KD AGS and MKN45. (G) The immunoblot results were semi-quantified using ImageJ. (H) The GEPIA platform was used to determine the correlation analysis of AKR1B1 and AKT or mTOR gene expression levels in GC patients from TCGA datasets. (I) Western blot reveals the AKR1B1, p-AKT, p-S6K1, and β-actin protein levels in AKR1B1-KD GC cells. (J) The immunoblot results of AGS were semi-quantified using ImageJ. Abbreviations: CCLE: Cancer Cell Line Encyclopedia; NC: negative control; KD: AKR1B1-shRNA. ^**P < 0.01, ^***P < 0.001.

The prospective relevance between AKR1B1 and the AKT-mTOR pathway was analyzed in TCGA datasets using the GEPIA database. The expression degree of AKR1B1 in GC tissues had a tight connection with the expression levels of AKT (P = 0.006) and mTOR (P = 0.014, Figure 4H). The expression of phosphorylated S6K1 in cancers depicted the AKT-mTOR pathway activation status as a key downstream target. This pathway is crucial in the migration, proliferation, apoptosis, drug resistance, and various abilities of cancer cells [15, 16]. Consequently, we explored the relationship between AKR1B1 expression and phosphorylated AKT and S6K1 using AKR1B1-KD GC cell lines. The outcome was inhibited in AKR1B1-KD GC clones, p-AKT, and p-S6K1 levels (Figure 4I, 4J).

The biology function of AKR1B1 was analysed by AGS and MKN45 knockdown structures. We investigated the multiplication capacity of AKR1B1-KD cells through the CCK-8 and colony-forming assays. Compared with negative controls, the hyperplasia of AKR1B1-KD cells was more markedly repressed (Figure 5A). Agree with these results, the AKR1B1-KD cells similarly exhibited a poor colony-forming capacity (Figure 5B, 5C). Then, the migration capacity alteration of the AKR1B1-KD cells was evaluated using a transwell assay, which was markedly suppressed (Figure 5D, 5E).

Figure 5. AKR1B1 enhances the proliferation and migration ability of GC cells. (A) The CCK-8 assay determined the proliferation ability of AGS and MKN45 transfecting with AKR1B1-shRNA. (B) The colony formation capacity of AGS and MKN45 transfected with AKR1B1-shRNA was assessed. (C) The colony-forming units were counted. (D) The migration capacity of AGS and MKN45 transfected with AKR1B1-shRNA was assessed. (E) The number of migrating cells was counted. Abbreviations: NC: negative control; KD: AKR1B1-shRNA. ^*P < 0.05, ^**P < 0.01.

AKR1B1 promoting tumorigenesis and deteriorating nutrition condition in vivo

We built a xenograft murine model using AGS cell lines to demonstrate the AKR1B1 expression level influences on neoplasm growth and nutrition in mice. In contrast to the NC group, the nutrition condition of mice was markedly improved in the AKR1B1-KD group (Figure 6A). Simultaneously, the growth of subcutaneous neoplasms was markedly lowered in the AKR1B1-KD group than that in the NC group (Figure 6B, 6C). The mouse weight was observably influenced by the growth of subcutaneous neoplasms. Additionally, AKR1B1 expression had a vital impact on GC cell hyperplasia and the nutrition condition of mice. Therefore, the weights of tumor masses were investigated to calculate the net body weight after deducting the respective tumor weights. Observably inhibiting the proliferative capacity of GC cells in the AKR1B1-KD group appeared as reduction of tumor weight and reverse of mouse weight without tumor weight compared with the control group (Figure 6D, 6E).

Figure 6. Reduced AKR1B1 inhibits GC tumor growth in vivo. (A) During the experiment, the body weight of mice was recorded twice a week. (B) The tumor volume of mice was recorded twice a week during the experiment. (C) The representative images of subcutaneous tumors harvested from NC and AKR1B1-KD groups. (D) The weights of tumor masses. (E) Net body weight after subtracting the respective tumor weights. (F) AKR1B1 expression was detected in subcutaneous tumors using western blotting. (G) The immunoblot result of AKR1B1 expression in subcutaneous tumors was semi-quantified using ImageJ. (H) The association analysis between the AKR1B1 expression levels in tumor tissues and weight. (I) The stratification of mice in Cluster 1 and Cluster 2 depends on the tumor weight and AKR1B1 expression levels. (J) The percentage of NC and AKR1B1-KD mice in each cluster. n = 5 for each group. Abbreviations: NC: negative control; KD: AKR1B1-shRNA. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Furthermore, the AKR1B1 protein expression was detected in subcutaneous tumors with immunoblot. Thus, AKR1B1 expression was more significantly inhibited in the AKR1B1-KD group than in the control group (Figure 6F, 6G). AKR1B1 protein levels were prominently correlated in situ with the subcutaneous neoplasm weight (Figure 6H). Then, the cluster analysis was performed based on the neoplasm weight and AKR1B1 protein levels. The influence of AKR1B1 expression degree was evaluated on the neoplasm growth in mice (Figure 6I). The results indicated that 20% of the NC group mice and 100% of the AKR1B1-KD group mice belonged to Cluster 1. Moreover, 80% of the NC group mice belonged to Cluster 2 (Figure 6J).