PTPN2 deficiency along with activation of nuclear Akt predict endocrine resistance in breast cancer

Purpose The protein tyrosine phosphatase, non-receptor type 2 (PTNP2) regulates receptor tyrosine kinase signalling, preventing downstream activation of intracellular pathways like the PI3K/Akt pathway. The gene encoding the protein is located on chromosome 18p11; the 18p region is commonly deleted in breast cancer. In this study, we aimed to evaluate PTPN2 protein expression in a large breast cancer cohort, its possible associations to PTPN2 gene copy loss, Akt activation, and the potential use as a clinical marker in breast cancer. Methods PTPN2 protein expression was analysed by immunohistochemistry in 664 node-negative breast tumours from patients enrolled in a randomised tamoxifen trial. DNA was available for 146 patients, PTPN2 gene copy number was determined by real-time PCR. Results PTPN2 gene loss was detected in 17.8% of the tumours. Low PTPN2 protein expression was associated with higher levels of nuclear-activated Akt (pAkt-n). Low PTPN2 as well as the combination variable low PTPN2/high pAkt-n could be used as predictive markers of poor tamoxifen response. Conclusion PTPN2 negatively regulates Akt signalling and loss of PTPN2 protein along with increased pAkt-n is a new potential clinical marker of endocrine treatment efficacy, which may allow for further tailored patient therapies.


Introduction
Anti-oestrogen treatment significantly reduces the recurrence and death rates in women with oestrogen receptor (ER)-positive breast cancer. Endocrine therapy is a welltolerated treatment to which most ER-positive tumours respond, however, around 30% of the ER-positive tumours show de novo or acquired resistance to the treatment. A commonly suggested mechanism to this resistance is the crosstalk between ER and growth factor signalling pathways, specifically the receptor tyrosine kinase (RTK)/PI3K/Akt/ mTOR axis (Musgrove and Sutherland 2009;Miller 2013). RTK signalling consists of complex networks of proteins with numerous feedback mechanisms. Protein tyrosine phosphatases (PTP) negatively regulate RTK signalling by dephosphorylation of tyrosine residues. Genetic and/or epigenetic alterations resulting in deregulation of PTP function have been shown to contribute to the development of several diseases, including cancer (Bussieres-Marmen et al. 2014;He et al. 2014;Julien et al. 2011).

3
One such PTP is protein tyrosine phosphatase, nonreceptor 2 (PTPN2). PTPN2 was first found in T-cells and is, therefore, also known as T-cell PTP (TCPTP) (Mosinger et al. 1992). The gene encoding PTPN2 is located in the chromosomal region 18p. This region is commonly deleted in breast cancer and associated with poor outcome (Addou-Klouche et al. 2010;Climent et al. 2002;Karlsson et al. 2015). Alternative splicing produces two main isoforms, the original 48.5 kDa (TC48) and a 45 kDa (TC45) isoform. TC48 contains a hydrophobic C-terminus, mainly localising it to the endoplasmic reticulum. The shorter TC45 is primarily targeted to the nucleus in resting cells but can enter the cytoplasm upon growth factor stimuli (Tiganis 2013). PTPN2 is ubiquitously expressed and has been shown to regulate receptor tyrosine kinase signalling, thereby preventing downstream activation of intracellular pathways, amongst others the PI3K/Akt pathway (Klingler-Hoffmann et al. 2001;Tiganis et al. 1999). PTPN2-regulated receptors include the epidermal growth factor receptor (EGFR), the insulin receptor (IR), the vascular endothelial growth factor receptor (VEGFR) and Met (Galic et al. 2003(Galic et al. , 2005Mattila et al. 2008;Omerovic et al. 2010;Tiganis et al. 1998Tiganis et al. , 1999Sangwan et al. 2008). Due to its involvement in the regulation of these oncoproteins, PTPN2 has been suggested to be a tumour suppressor.
This study aimed to evaluate PTPN2 protein expression in a large breast cancer patient material, its possible associations with PTPN2 loss, Akt activation, and the potential use as a new clinical marker in breast cancer.

Patient material
The cohort consisted of post-menopausal breast cancer patients enrolled in a randomised adjuvant trial between November 1976 and April 1990. Study design and longterm follow-up data have been previously reported in detail (Rutqvist et al. 2007). Briefly, breast cancer patients with a tumour diameter of ≤ 30 mm and no lymph node involvement were included in the cohort. The patients were randomised to receive post-operative tamoxifen for 2 years or no endocrine treatment. The women who were recurrence-free after 2 years of tamoxifen treatment and who consented were randomised to three additional years of tamoxifen or no further treatment. All patients were primarily treated with a modified radical mastectomy. Tumour tissues were formalin-fixed paraffin-embedded and stored at room temperature. Tumour tissue material in the form of tissue microarrays (TMA) was still available from 664 patients and high-quality DNA could be prepared from 146 patients (Fig. 1). Tumour characteristics and treatment were comparable with the original cohort (Bostner et al. 2010). The local ethics board at the Karolinska Institute, Stockholm, Sweden, approved retrospective studies of biomarkers.

Clinicopathological variables and biomarkers
The status of ER and progesterone receptor (PR) were previously analysed by immunohistochemistry (IHC) (Jerevall et al. 2011). The cut-off level for both ER and PR positivity was > 10% stained nuclei. When IHC data was not available, ER status determined at the time of diagnosis was used with the cut-off of 0.05 fmol/µg DNA (Rutqvist et al. 2007;Rutqvist and Johansson 2006). Isoelectric focusing and IHC data have been shown to be comparable in this cohort (Khoshnoud et al. 2011). Nottingham histological grade and HER2 protein expression were evaluated retrospectively (Jansson et al. 2009;Jerevall et al. 2011). Staining and grading of Akt, phosphorylated at S473 (pAkt), have been described previously (Bostner et al. 2013). In this study, cytoplasmic pAkt staining is referred to as pAkt-cyt. pAkt expression that was stronger in the nucleus than in the cytoplasm is referred to as pAkt-n > cyt.   (Karlsson et al. 2015). Briefly, 10-25 ng total DNA was added to a 10 µL reaction with 1x Taq Man Fast Universal PCR master mix (Applied Biosystems) and 0.1 µM primer and probe for PTPN2 or the endogenous control Amyloid Precursor Protein (APP). PTPN2 gene quantification was performed with the Comparative Ct method using DNA from the cell line MCF7 as the calibrator sample on each plate. Samples were run in triplicates and standard deviations < 0.3 were required for inclusion in further analysis. With this criterion, PTPN2 status was obtained for 146 patients. Primers and probes sequences were as follows:

Immunohistochemistry
Protein expression of PTPN2 in the available tumours was evaluated with immunohistochemistry. First, tissue microarrays (TMAs) were created; triplicates of core needle biopsies from paraffin-embedded tissues were re-embedded in new paraffin blocks. The blocks were cut into 4 µM sections and mounted on frost-coated slides. Deparaffinisation, rehydration and antigen retrieval of the slides was performed with the PT link system (Dako, Glostrup, Denmark)  Whole-slide images were generated with the Aperio ScanScope AT at 200x magnification (Leica Biosystems, Wetzlar, Germany) and staining was evaluated with the Imagescope software (Leica Biosystems). Two independent observers performed grading. PTPN2 was assessable in 664 tumours and cytoplasmic staining was graded as negative, weak, moderate or strong ( Fig. 2a-d). These groups were dichotomised for further analyses into a low group, comprised of the negative and weak staining, and a high group including moderate and strong staining. Protein specificity of the PTPN2 antibody was validated with four different siRNAs against PTPN2, to wit: Hs_PTPN2_9 (siRNA9), Hs_PTPN2_10 (siRNA10), Hs_PTPN2_15 (siRNA15), Hs_PTPN2_16 (siRNA16) (Qiagen). MCF7 cells were transfected with 10 µM siRNA using Dharmafect 1 (Dharmacon, Thermo Fisher Scientific, Waltham, MA, USA) as a transfection agent and cells were incubated for 48 h with siRNA. Western blot was performed to visualise specificity (Fig. 2e).

Statistical analysis
Spearman rank order correlation test was used to determine the association between PTPN2 gene copy number and protein expression levels. The relationships between PTPN2 gene copy number, protein expression, and clinical variables were assessed by the Chi-square test or Chi-square test for trend, when appropriate. The product-limit method was used for estimation of cumulative probabilities of distance recurrence-free survival (DRFS). Differences in survival between groups were tested with the log-rank test. Analysis of distant recurrence rates, as well as interaction tests, were performed with Cox proportional hazard regression. All the procedures were comprised in STATISTICA 12 (Statsoft, Inc, Tulsa, OK, USA). The criterion for statistical significance was P < 0.05.

Results
The

Correlations to clinical variables and pAkt expression
The associations between PTPN2 and clinicopathological parameters were further assessed. PTPN2 protein expression correlated with ER positivity in the tumours (P = 0.0066, Table 1) and borderline associated with PR expression (P = 0.058). Furthermore, it was found to be correlated with pAkt-cyt (P < 0.0001, Table 1) and inversely correlated with pAkt-n > cyt (P = 0.006, Table 1).
PTPN2 gene deletion was not significantly associated to any of the clinical variables in the analysis.

PTPN2 and nuclear pAkt predict tamoxifen benefit
Patients with tumours expressing low levels of PTPN2 had no significant benefit from tamoxifen treatment (P = 0.14), whereas the group with high protein expression did have benefit (P = 0.00005, interaction test P = 0.11; Table 2; Fig. 3a, b). Restricted to patients with grade 2  or 3 tumours, the interaction test reached significance (P = 0.043, Table 2; Fig. 3c, d).
We previously showed that pAkt-n was borderline significant as a predictive factor of tamoxifen response in this cohort (Bostner et al. 2013). Because of the implications of PTPN2 as a regulator of Akt signalling, a combination variable was created including PTPN2 and nuclear pAkt protein levels. Tamoxifen treatment was associated with a strongly reduced risk of distant recurrence in the group of patients with ER-positive tumour and high PTPN2 concurrent with low pAkt expression, whereas no significant benefit from tamoxifen could be seen in the group with low PTPN2 and/or high nuclear pAkt (interaction test P = 0.044, Table 2; Fig. 4a, b). This predictive value of PTPN2 and nuclear pAkt was also evident when restricting the analysis to the group of patients with histologically grade 2-3 tumours (interaction test P = 0.019, Table 2; Fig. 4c, d).

Discussion
Few studies have explored the role of PTPN2 in breast cancer; therefore, we aimed to evaluate the clinical value of PTPN2 in a large breast cancer cohort.
PTPN2 gene copy loss could be detected in 17.8% of the cases, which is in agreement with our previous study on a post-menopausal breast cancer cohort (Karlsson et al. 2015). Low PTPN2 protein expression was detected in 53.3% of the cases. We found a trend to correlation between PTPN2 gene deletion and expression levels of the corresponding protein. While, to our knowledge, there are no studies looking at the correlation between gene deletion and protein expressions levels, PTPN2 gene deletion has previously been associated with low corresponding mRNA levels (Addou-Klouche et al. 2010;Karlsson et al. 2015). Whether genomic loss of PTPN2 leads to decreased expression of its corresponding protein is still unclear. Like previous studies, loss of PTPN2 was most common in the ER-negative subgroup (Shields et al. 2013;Karlsson et al. 2015). Low PTPN2 protein expression was associated with poor response to tamoxifen in the group of patients with tumours histologically graded as 2 or 3, suggesting a need for other types of treatment in this group. Patients with NHG 1 tumours with low PTPN2 tended to have a higher risk for distant recurrence, indicating that PTPN2 loss might have a prognostic value in patients with NHG1, which is normally associated with good prognosis.
We previously found a correlation between PTPN2 gene deletion and high levels of phosphorylated Akt in breast cancer patients (Karlsson et al. 2015). Lee and colleagues showed significantly higher levels of phosphorylated Akt in PTPN2 knockout mice (Lee et al. 2017). In the present study, we provide further indications that PTPN2 regulates Akt signalling by showing that low PTPN2 protein expression was associated with increased nuclear pAkt levels. Increased levels of phosphorylated Akt in the nucleus have been shown to be associated with poor response to tamoxifen in breast cancer patients (Bostner et al. 2013) and the oestrogen receptor has been shown in vitro to be a direct substrate of Akt phosphorylation (Campbell et al. 2001). Interestingly, Akt activation and translocation to the nucleus have been shown to be promoted by the oncogene T-cell leukaemia/lymphoma 1B (TCL1), in turn, activated by oestrogen signalling (Pekarsky et al. 2000;Badve et al. 2010). When low PTPN2 protein Tam (n=187) No Tam (n=169) Fig. 4 Predictive value for tamoxifen response of PTPN2 protein expression and nuclear phosphorylated Akt (pAkt-n) expression. Distant recurrence-free survival (DRFS) for breast cancer patients treated with tamoxifen (Tam) vs no tamoxifen in relation to low PTPN2 protein expression and high pAkt-n expression (oestrogen receptor-positive (ER+), (a), high PTPN2 protein expression and/or pAkt-n expression low in ER + tumours (b), PTPN2 expression low and pAkt-n high in ER + tumours with grade 2 or 3 (c), and PTPN2 expression high and pAkt-n low in ER + tumours, grade 2 to 3 (d). P values were estimated with the log rank test was analysed in combination with high nuclear expression of phosphorylated Akt, the combination variable was a strong predictor of tamoxifen resistance amongst all patients with ER-positive breast cancer. In summary, this study demonstrates that PTPN2 negatively regulates Akt signalling and that loss of PTPN2 protein along with increased nuclear pAkt may be a new potential clinical marker of endocrine treatment benefit, which may allow for further tailored patient therapies.