AKR1C2 acts as a targetable oncogene in esophageal squamous cell carcinoma via activating PI3K/AKT signaling pathway

Abstract The aldo‐keto reductases family 1 member C2 (AKR1C2) has critical roles in the tumorigenesis and progression of malignant tumours. However, it was also discovered to have ambiguous functions in multiple cancers and till present, its clinical significance and molecular mechanism in oesophageal squamous cell carcinoma (ESCC) has been unclear. The aim of this study was to explore the role of AKR1C2 in the tumorigenesis of ESCC. Here, we showed that AKR1C2 expression was found to be up‐regulated in ESCC tissues and was significantly associated with pathological stage, lymph node metastasis and worse outcomes. Functional assays demonstrated that an ectopic expression of AKR1C2 in ESCC cells resulted in increased proliferation, migration and cisplatin resistance, while knockdown led to inversing effects. Bioinformation analyses and mechanistic studies demonstrated that AKR1C2 activated the PI3K/AKT signalling pathway, furthermore, the inhibitor of PI3K or the selective inhibitor of AKR1C2 enzyme activity could reverse the aggressiveness and showed synergistic antitumour effect when combined with cisplatin, both in vitro and in vivo. In conclusion, Our findings revealed that AKR1C2 could function as an oncogene by activating the PI3K/AKT pathway, as a novel prognostic biomarker and/or as a potential therapeutic target to ESCC.


| INTRODUC TI ON
Oesophageal carcinoma (EC) is the ninth most common aggressive malignancy worldwide and is ranked as the sixth leading cause of cancer-related mortality. 1 China has a high incidence of EC, accounting for more than 50% of its global morbidity and mortality. More than 90% of EC patients in China are classified as oesophageal squamous cell carcinoma (ESCC). 2 The outcome of EC is poor, with a dismal 5-year overall survival (OS) rate at ~20%-30%, 3,4 due to its poorly understood the underlying pathological molecular mechanisms. 3,5,6 At present, many medical scientists are trying to find effective targets for the treatment of ESCC, but the targeted therapy trials have thus far been disappointing. EGFR is one of the most investigated molecular targets in the field of EC, but the results of clinical studies did not yield a significant survival benefit for EC, including ESCC. [7][8][9] Therefore, there is no effective targeted drug for the treatment of EC so far, and effective biomarkers selection is urgently warranted. In addition, platinum-based chemotherapy regimens are the main chemotherapy regimens for EC; however, it poorly responses to chemotherapy, 10 and the underlying mechanism is unclear.
In the present study, we demonstrated that AKR1C2 was highly expressed in ESCC patients' tissues and ESCC cell lines. Its high expression was associated with poor prognosis of ESCC patients. Our findings further indicated that AKR1C2 could act as an oncogene, was associated with cisplatin resistance and might be a potential therapeutic target for ESCC treatment.

| Cell lines and culture
Cells used in the present study were one immortalized normal oesophageal cell line (NE1) and seven ESCC cell lines (KYSE30, KYSE180, KYSE410, KYSE510, KYSE520, EC18, EC109), which were kindly gifted by professor Guan (Department of Clinical Oncology, the University of Hong Kong). All cell lines were free of mycoplasma infection. They were cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco) and 1% penicillin streptomycin mixed solution in 5% CO 2 at 37°C.

| Real-time PCR
Total RNA was extracted with TRIzol (Invitrogen) following the manufacturer's instructions. Reverse transcription was performed using the PrimeScript RT Master Mix Kit (TAKARA). RT-qPCR was carried out using SYBR Green Master Mix (YEASEN), followed by Roche 96/384 holes Real-Time PCR system (Roche). GAPDH was used as an internal control. The primers used are listed in Table S1.

| Immunohistochemical (IHC) staining
Immunohistochemical was performed according to previously described standard method. 29 Antibodies were used to against AKR1C2 (Abcam), pAKT S473 (Affinity), Ki-67 (Cell Signaling Technology), cleaved-caspase 3 (SAB). IHC staining was reviewed and scored separately by two independent pathologists who were blinded to the patients' characteristics and scoring discrepancies were resolved by and 81972785 K E Y W O R D S AKR1C2, cisplatin resistance, combination therapy, oesophageal squamous cell carcinoma, PI3K/AKT signalling pathway mutual consensus. The scoring procedures and method were similar to previously described. 29 The median scores of AKR1C2 (score = 8) were used as the cut-off value to classify the patients into a high expression (score ≥ 8) or low-expression (score < 8) group.

| Lentiviral-mediated knockdown assay
For AKR1C2 knockdown assays, one lentiviral constructs containing short hairpin RNAs (shRNA) specifically targeting AKR1C2 were purchased from GeneCopoeia and were packaged in the 293T cells.
Virus-containing supernatants were collected and stably transfected into KYSE180 and EC109 cells. Empty vector transfected cells were used as controls and selected stably transduced cells by 2.5 mg/ mL puromycin (Sigma) for 7 days. The sequences of shRNA against AKR1C2 is listed in Table S3.

| Function studies and drug treatment experiments
Protocols for cell proliferation, migration assays and drug treatment assays are provided in Appendix S1.

| Animal experiment
The experiments with mice were approved by the Research Animal Resource Center of Sun Yat-Sen University (Approval number: L102042019060D). Protocols for establishing human ESCC xenografts and drug treatment assays are provided in Appendix S1.

| Statistical analysis
Statistical analyses were performed using GraphPad Prism 8, SPSS version 21.0 (IBM). The relationship between AKR1C2 expression and clinicopathological characteristics were analysed using the Pearson chi-square test. Kaplan-Meier analysis with log-rank test was applied for survival analysis. Univariate and multivariate Cox proportional hazard regression models were used to evaluate the survival hazard with a forward stepwise procedure. For functional assays analysis, when the data were normally distributed, the comparison between two groups was conducted using the Student's t test. P < .05 was considered as statistically significant.

| AKR1C2 expression was distinctively upregulated in ESCC
The expression of the top 20 up-regulated genes was shown in  Figure 1C). Consistently, the protein expression of AKR1C2 was elevated in 11 paired samples, including the three pairs of RNA-sequencing samples ( Figure 1D).
In addition, data from TCGA ( Figure S1A) and Oncomine databases ( Figure S1B,C) further confirmed that AKR1C2 was elevated in ESCC.

| AKR1C2 was associated with advanced clinicopathological features and worse survival as an independent prognosis factor
To further investigate the frequency of AKR1C2 up-regulation in Furthermore, AKR1C2 expression was gradually increased from stage I to stage Ⅳ ( Figure 2A). Consistently, AKR1C2 expression was found to gradual increase expression from pN0 to pN3 in ESCC patients ( Figure 2B).
According to the cut-off value (median IHC score = 8) of AKR1C2 staining, patients were classified into a low (score < 8) or high (score ≥ 8) AKR1C2 expression group. Next, we explored the to evaluate the clinicopathological features between the two groups in the training cohort. As shown in Table 1, strong associations were observed between AKR1C2 expression to pT stage (P = .005), pN stage (P = .000), pathological stage (P = .000) and vascular invasion (P = .003). The characteristics of the patients in the validation cohorts were attached in Table S2.
To explore the association between AKR1C2 expression and the prognosis of ESCC patients, we evaluated the correlation between AKR1C2 expression and clinical outcomes. For the training cohort, the median OS in the high and low AKR1C2 expression groups were 41.8 and 69.7 months, respectively. As shown in Figure 2C, a high expression of AKR1C2 was correlated to a poorer prognosis in patients with ESCC (P < .0001) and was further confirmed in the validation cohort (P = .04; Figure 2D). Univariate and multivariate analyses (Cox proportional hazards regression model) performed in the training cohort showed that age, BMI, pathological stage, vascular invasion, as well as AKR1C2 expression were independent prognostic factors for OS (Table 2).

| AKR1C2 promotes ESCC cells proliferation in vitro and in vivo
To explore the role of AKR1C2 in ESCC, we identified the AKR1C2 expression level in ESCC cell lines. As shown in Figure 3A Then, colony formation assays were performed and showed that the knockdown of AKR1C2 significantly suppressed the proliferation of KYSE410 and EC109 cells as compared to the control groups ( Figure 3D). By contrast, the overexpression of AKR1C2 promoted the proliferation of KYSE30 cell ( Figure 3E). Results from MTS assays further confirmed these findings ( Figure 3F).
To further investigate whether AKR1C2 affects cell proliferation in vivo, the shAKR1C2 mediated knockdown in KYSE180 and EC109 cells and overexpressed plasmid mediated in KYSE30 cell were inoculated into BALB/C nude mice. As shown in Figure 3G, the knockdown of AKR1C2 could significantly inhibit the in vivo tumour growth while overexpressed AKR1C2 could promote them. These results were further confirmed by IHC experiments ( Figure 5E).

| AKR1C2 promotes ESCC cells migration by inducing epithelial-mesenchymal transition (EMT)
To explore the biological function of AKR1C2 in ESCC, migration assays were performed in KYSE410 and EC109 cells which were knockdown by siRNA. The results revealed that the knockdown of AKR1C2 significantly inhibited cells migration as compared to scramble control groups. However, ectopic overexpression of AKR1C2 in KYSE30 cell had the opposite effect ( Figure 4A). Woundhealing assays confirmed the effect of AKR1C2 on cell migration as it was found that cells motility were repressed when AKR1C2 was silenced but promoted when AKR1C2 was overexpressed ( Figure 4B).
The EMT markers were detected by Western blotting to investigate the mechanism of migration mediated by AKR1C2. As shown in the Figure 4C, compared with the control groups, the epithelial markers (E-cadherin) were increased and the mesenchymal markers (N-cadherin, vimentin) were reduced in the knockdown groups.

| PI3K/AKT signalling pathway is regulated by AKR1C2
The results of RNA-sequencing followed by KEGG analysis of three pairs matched tissues indicated that PI3K/AKT signalling pathway was involved in the ESCC tumorigenesis ( Figure 5A) and was validated by six pairs of matched tissues by Western blotting ( Figure 5B). To further screen the signalling pathway affected by the AKR1C2 in ESCC, KYSE30 cell transfected with AKR1C2 plasmid or empty control plasmid were performed with RNA-sequencing. Next, the KEGG pathways were analysed and the canonical PI3K/AKT signalling pathway signalling was found to be significantly enriched ( Figure 5C). Moreover, the results of Western blotting confirmed that, compared with the control groups, phosphorylated AKT (pAKT) was reduced in the knockdown cells and increased in the overexpressed cell ( Figure 5D).
This was further confirmed in ex vivo xenograft tumour samples by IHC which exhibited positive correlation with Ki-67 ( Figure 5E).
Consistently, AKR1C2 expression also showed positive correction with pAKT in the paraffin-embed ESCC tissues (n = 47, r = .5669, P < .001; Figure 5F). Furthermore, LY294002, an inhibitor of PI3K, was applied to detect the effect of PI3K/AKT signalling pathway on AKR1C2-mediated cell migration by migration assays. As shown in Figure 5G, LY294002 could significantly attenuate cell migration. Moreover, the pAKT level could be restrained by LY294002 accompanied with the increased E-cadherin and reduced Vimentin in the AKR1C2 overexpression cell. (Figure S3).

| AKR1C2 overexpression resulted in cisplatin resistance in ESCC
Platinum drugs are the basic drugs used as chemotherapy for locally advanced or advanced EC but it has been found that most of the patients had decreased sensitivity to cisplatin. To study whether AKR1C2 was involved in the reported platinum resistance, the IC50 of cisplatin was determined in six ESCC cell lines. Our findings showed that the cisplatin IC50 was positively correlated with AKR1C2 expression in ESCC cell lines (r = .8115, P = .05; Figure 6A).
Next, we found that overexpressing AKR1C2 attenuated the sensitivity to cisplatin while knocking it down had a reversed effect ( Figure 6B). KEGG pathway analysis also showed that AKR1C2 had potential involvement in the platinum drug resistance ( Figure 5B).
Based on our above findings that AKR1C2 mediated cisplatin resistance and activated the PI3K/AKT signalling pathway, we attempted to combine cisplatin and LY294002 to treat indicated cells to explore whether this combination therapy would have a synergistic antitumour effect. After drugs were added to the indicated cells for 72 hours, cell viability was tested by MTS. As shown in Figure 6C and S4A, the combinational use of cisplatin plus LY294002 had a more obvious inhibition effect in the AKR1C2 overexpression cell as compared to the vector control group and showed the dose-dependent effect of LY294002. Furthermore, Western blotting was performed to examine the levels of pAKT and apoptosis markers (cleaved-caspase3 and cleaved-PARP). As shown in Figure 6D, in the combination treatment groups, pAKT was inhibited and accompanied with an increase in apoptosis markers compared to the single drug groups. This phenomenon was more distinctive in the AKR1C2 overexpression cell and also showed the LY294002 dose-dependent effect.
Next, to validate the above findings, we chose two parental ESCC cell lines, namely KYSE30 L and KYSE180 H , to which the phosphoinositide (PI)/Annexin V-FITC apoptosis assays were performed. As shown in Figures S4B and 6E

TA B L E 1 (Continued)
Then, a KYSE180 H xenograft model was established in nude mice to investigate the in vivo effects of the above findings. The results showed that the tumour volume was much smaller in the combination therapy groups as compared to those treated with the single drug (cisplatin or LY294002) groups ( Figure 6F). Additionally, the results were further evaluated by IHC staining of c-caspase3 in ex vivo tumour samples ( Figure 6G).

| AKR1C2 can be a potential therapeutic target
KYSE30 cell transfected with AKR1C2 plasmid or empty control plasmid was performed with RNA-sequencing, following GO analysis results inferred that AKR1C2 overexpression accompanied with the functional changes of inositol metabolism ( Figure S5A), and the KEGG pathway analysis of EC109 cell suggested that the steroid and steroid hormone biosynthesis maybe change when AKR1C2 was knocked down ( Figure S5B). Ursodeoxycholic acid (UDCA), a selective inhibitor of AKR1C2 enzymatic activity, was used to treat indicated cells followed by MTS ( Figure 7A) and migration assays ( Figure 7B). The results indicated that UDCA could inhibit the proliferation and migration in AKR1C2-overexpressed cells but not the control group. Further, the AKR1C2 inducing EMT was found to be reversed by UDCA ( Figure S5C) Figure 7C).
AKR1C2 is the metabolic enzyme of steroid hormone DHT, which could recognize and bind with AR to activate the DHT-AR pathway.
Therefore, we explored the AR protein level in ESCC clinical samples. Our results showed that that AR was barely expressed in the eight pairs of matched ESCC tissues ( Figure 7D).
In light of the findings observed in this present study, a proposed model illustrating the oncogenic effect of AKR1C2 in the pathogenesis of ESCC is provided in Figure 7E. suggest that AKR1C2 may act as an oncogene in ESCC and may be a potential therapeutic target.

| D ISCUSS I ON
In additional, the PI3K/AKT pathway has been described as a major pathway frequently activated in human ESCC. 30   dose-dependent manner, especially in the AKR1C2 overexpression cells. Partly consist with our findings, several studied have also found that AKR1C2 regulated cisplatin resistance in other cancers 22,27,33 but the resistance mechanism was sti eously injected into the right flank of nude mice for 30 days.  27,39 Taken together, these results demonstrated that UDCA could reverse the aggressiveness mediated by AKR1C2, at least in vitro and may be a potential therapeutic target.
Combining the above results, we speculate that the role of AKR1C2 in tumours depends on AR, which has been reported as a pivotal drug target in breast cancer, [40][41][42] when acting in an AR-dependent manner, it inhibits the DHT-AR signalling pathway by metabolizing DHT 11,23 ; On the contrary, when acting in an AR-independent manner, AKR1C2 metabolizes DHT into 3α-diol, which has been reported can activate the PI3K/AKT signalling pathway. [43][44][45][46][47][48][49] In addition, previous studies reported that P) could be chemical synthesized or biosynthesized by diol in vitro. 50,51 As ESCC tissues hardly expresses AR ( Figure 7D), thus, it is reasonable to infer that the metabolites 3α-diol may be reused to biosynthesis PI, known as a precursor of PI3K, 52 following by activating the PI3K/AKT pathway in an AR-independent manner, [43][44][45][46] However, there is insufficient evidence that whether 3α-diol could biosynthesis PI directory in vivo. Collectively, these results suggest that for AR negative AKR1C2 positive (AR − /AKR1C2 + ) ESCC, AKR1C2 mediates activation of PI3K/AKT pathway may be an alternative to DHT-AR pathway, which uses androgen metabolites to activate a classic tumour signalling pathway. However, further study is still required to confirm these findings.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of the study are available from the corresponding author on reasonable request.