Tissue immunostaining of candidate prognostic proteins in metastatic and non-metastatic prostate cancer

Prostate cancer (PCa) lacks specific markers capable of distinguishing aggressive tumors from those with indolent behavior. Therefore, the aim of this study was to evaluate the immunostaining of candidate proteins (PTEN, AKT, TRPM8, and NKX3.1) through the immunohistochemistry technique (IHC) on patients with metastatic and non-metastatic PCa. Tissues from 60 patients were divided into three groups categorized according to prognostic parameters: better prognosis (n = 20), worse prognosis (n = 23), and metastatic (n = 17). Immunostaining was analyzed by a pathologist and staining classifications were considered according to signal intensity: (0) no staining, (+) weak, and (++ and +++) intermediate to strong. AKT protein was associated (p = 0.012) and correlated (p = 0.014; Tau = − 0.288) with the prognostic groups. The immunostaining for TRPM8 (p = 0.010) and NKX3.1 (p = 0.003) proteins differed between malignant tumor and non-tumoral adjacent tissue as well as for proteins in cellular locations (nucleus and cytoplasm). TRPM8 was independently associated with the ISUP grade ≥ 4 (p = 0.024; OR = 8.373; 95% CI = 1.319–53.164). The NKX3.1 showed positive and predominantly strong immunostaining in all patients in both tumoral and non-tumoral adjacent tissues. All metastatic samples had positive immunostaining, with strong intensity for NKX3.1 (p = 0.021; Tau = − 0.302). In the non-metastatic group, this strong protein staining was not observed in any patients. This study confirmed that NKX3.1 is highly specific for prostate tissue and indicated that NKX3.1, AKT, and TRPM8 may be candidate markers for prostate cancer prognosis.


Introduction
According to the World Health Organization (WHO), prostate cancer (PCa) is the fourth most incident cancer, with approximately 1.4 million new cases worldwide. In men, it is the second most common type after lung cancer (Global Cancer Observatory 2020; Culp et al. 2020;Sung et al. 2021) and the fifth leading cause of death (Bray et al. 2018;INCA 2019;Sung et al. 2021). For each year of the 2020-2022 triennium, 625,000 new cases of cancer are estimated in Brazil (INCA 2019), with prostate cancer being the second most common type (Global Cancer Observatory 2020; INCA 2019). Currently, in Brazil, digital rectal examinations and prostate-specific antigen (PSA) measurements are used as screening methodologies for PCa, and patients with abnormalities in the exam and/or PSA dosages above 10 ng/ mL are referred for a transrectal ultrasound-guided needle biopsy (Sociedade Brasileira de Urologia 2018; Porcaro et al. 2019;Vendrami et al. 2019).
In addition, it is known that PSA is an excellent marker for identifying prostatic alterations; however, it is not specific and exclusive to malignant alterations (Vendrami et al. 2019;Lomas and Ahmed 2020). In this context, the search for specific biomarkers that could become potential molecular markers for PCa, capable of predicting clinical and pathological complications in the patient, is important and is under study. Immunohistochemistry (IHC) is a widely used tool in clinical routine to confirm diagnoses with tissue markers (Giannico et al. 2017;Orakpoghenor et al. 2018;Comperát 2019).
Some molecules are already being studied by IHC as they are considered potential candidates as markers of PCa, among them those involved in the cell survival pathway, phosphatidylinositol-3-kinase/serine-threonine kinase/mammalian target of the rapamycin complex (PI3K/AKT/mTOR), in addition to transient melastatin 8 (TRPM8) and NK3 homeobox 1 (NKX3.1) stand out, as they play an important role in prostate carcinogenesis.
It is known that the PI3K/AKT/mTOR signaling pathway is one of the pathways that is most dysregulated in cancer (Koundouros and Poulogiannis 2018), and an aberrant expression of this pathway has already been demonstrated in studies in the early and late phases of PCa (Taylor et al. 2010;Sreenivasulu et al. 2018). Studies also show that deletion of the PTEN tumor suppressor gene in PCa is very common, being very present in metastatic and castration-resistant tumors (Robinson et al. 2015;Wozniak et al. 2017;Jamaspishvili et al. 2018).
The TRPM8 channel is a homotetramer formed by subunits, showing 8 putative glycosylation sites and an immunogenic epitope. It is highly expressed in the prostate, as its ion channel functions as a testosterone receptor, suggesting a role in the regulation of androgenic responses (Asuthkar et al. 2015b). Furthermore, evidence demonstrates that it has an important role in the development and progression of neoplasms, especially in PCa, being overexpressed in malignant tumor tissue compared to nonmalignant tissue. This protein is present in PCa refractory hormone and with a high Gleason score (Yee 2015).
The tumor suppressor gene NKX3.1 is a member of the NK family of homeobox genes, participating in cell specification and organogenesis processes in several species. In humans, this gene is primarily related to normal prostate development. Its loss of expression leads to defects in prostate protein secretion and ductal morphogenesis and contributes to prostate carcinogenesis (Abate-Shen et al. 2008).
Therefore, the current study aimed to evaluate tissue immunostaining of the tumor suppressor proteins PTEN and NKX3.1, of oncogenic AKT protein, involved in a cell survival pathway, in addition to the testosterone receptor TRPM8 in samples from patients with metastatic and nonmetastatic PCa in the search for candidate markers for prostate cancer.

Study group and sample characterization
In this retrospective longitudinal study, 60 prostatic paraffin-embedded samples of malignant and respective adjacent non-tumor tissues were evaluated. Samples were randomly selected from male patients with a confirmed diagnosis of PCa after radical prostatectomy, at Hospital do Câncer de Londrina (HCL), between the years of 2006 and 2016. Of this amount, 51 samples are radical prostatectomy (RP) products; however, in the metastatic group, 5 samples came from biopsy and 3 from transurethral resection (TUR); in the group with the worse prognosis, 1 sample was from TUR.
The study was approved by the Research Ethics Committee Involving Human Beings of the State University of Londrina-Brazil, under number 176/2013. Patients participated voluntarily and signed a free and informed consent form and answered a modified personal questionnaire based on Carrano and Natarajan (1988).
Histopathological data were obtained from medical records, which were used, together with the guidelines of the National Comprehensive Cancer Network (NCCN version 4.2019), for the classification of patients into three experimental groups: (1) PCa with a better prognosis (n = 20); (2) PCa with a worse prognosis (n = 23); and (3) metastatic PCa (n = 17). Patients with a ISUP grade ≤ 2 (3 + 4), staging ≤ T2b, and PSA ≤ 10 ng/mL were considered to have a better prognosis PCa. Patients with a ISUP grade ≥ 3 (4 + 3), staging ≥ T3a, and PSA ≥ 20 ng/mL were considered to have worse prognosis PCa. Patients with metastasis were classified according to the presence of lymph node invasion and/ or distant metastasis and/or positive bone scintigraphy. A table containing clinical and pathological characteristics of all patients was included as Online Resource 1.
The sample's protein profiles were compared between the metastatic versus non-metastatic group (with better and worse prognosis), as well as their malignant and adjacent non-tumor tissues. All samples in the present study were from biopsy and radical prostatectomy, without neoadjuvant chemotherapy.

Histopathological analysis
Tissues obtained from the biopsy were stained with hematoxylin and eosin to confirm the clinical diagnosis of PCa and to verify the presence of tumor and adjacent non-tumor tissue for further analysis and comparison of immunostaining of proteins in the tissues. This step was performed by pathologists from the HCL. The histopathological classification used was based on international standards established by the WHO, such as ISUP grade (European Association of Urology 2022) and clinical staging determined by the Tumor/Node/Metastasis (TNM) system, following the recommendations of the AJCC (American Joint Committee on Cancer).

Immunohistochemistry
Experiments were carried out according to Guembarovski et al. (2018) with modifications, regarding antigenic retrieval and the background blocker. Formalin-fixed paraffin-embedded tissue samples of metastatic and non-metastatic malignant and adjacent non-tumor tissues were obtained. Cuts were made, 5-6 µm thick, and fixed on silanized StarFrost ® slides (Knittel glass, ALE).
Negative controls were performed to verify the specificity of the primary antibody in all slide batteries, where it was replaced by phosphate-buffered saline (PBS). The secondary antibody kit (mouse/rabbit detection kit HRP/DAB ABC, Abcam, Cambridge, MA, USA) was used according to the manufacturer's instructions and as a chromogen, the Pierce ™ DAB substrate kit (Thermo Fisher Scientific, Rockford, IL, USA), using concentrated DAB ([2x]) for NKX3.1 and ([4x]) for the other antibodies (AKT, PTEN, and TRPM8), based on the manufacturer's protocol.
Immunostaining for protein profiles in experimental groups was analyzed by an experienced pathologist. In the analysis of adjacent non-tumor tissue, only normal glands and areas of benign hyperplasia were considered, excluding areas of atrophy. The classifications were considered according to staining signal strength: (0) no staining, (+) weak, and (++ and +++) strong, according to Figs. 1 and 2.

Statistical analysis
The comparison of the mean ages of the experimental groups was performed using the Student's t test. To compare the staining in tumor and adjacent non-tumor tissues for each protein evaluated, the McNemar test for related samples was used.
Kendall's Tau test was used to analyze the correlations between the protein's immunostaining by IHC and clinicopathological parameters, and logistic regression was performed for the variables that showed significance to verify whether they were independently associated with protein staining. To analyze the interaction between protein tags, the Kendall Tau correlation test was also performed.
Some data were omitted from the statistical analyzes due to lack of information contained in patient records and because some samples had worn paraffin blocks; therefore, some patients had no tumor tissue (2) or adjacent non-tumor tissue (3) on the same slide for comparison.
Photomicrograph of weak intensity immunostaining using the immunohistochemistry technique for PTEN, AKT, and TRPM8 proteins, evaluated in tumor tissue samples and adjacent non-tumor tissue from patients with PCa. Arrows point to weak immunostaining of proteins. Letters represent the evaluated proteins, being a (negative control), b (PTEN), c (AKT), and d (TRPM8). 40 × magnification. Source: the author himself.
Photomicrograph of strong intensity immunostaining using the protein immunohistochemistry technique for NKX3.1, evaluated in tumor tissue samples and adjacent non-tumor tissue from patients with PCa. Arrows point to strong immunostaining of NKX3.1. Letters represent the evaluated proteins, being a (negative control) and b (NKX3.1) 0.40× magnification. Source: author himself.

PTEN
PTEN protein did not show differences in immunostaining between tumor and adjacent non-tumor tissues (p = 0.647) or in the cellular locations (cytoplasm: p = 0.195) and (nucleus: p = 0.587) ( Table 1). PTEN immunostaining did not demonstrate any significant association or correlation with the prognostic groups, the prognostic parameters, or biochemical recurrence and metastasis (Online Resource 8).

AKT
AKT protein was also not expressed differently in the tissues of the same patients (p = 0.552) or in relation to cellular locations: cytoplasm (p = 0.194) and nucleus (p = 0.526) (Table 1). Furthermore, it was associated (p = 0.012) and correlated (p = 0.014; Tau = − 0,288) with the prognostic groups: better and worse prognosis and metastasis (Online Resource 9).

TRPM8
TRPM8 protein was expressed differently in tumor and adjacent non-tumor tissue (p = 0.010), with higher immunostaining in malignant tumor, even if of low intensity; the same result was observed for cellular locations of the immunostaining nucleus (p = 0.012), in both tissues (Table 1). In addition, protein immunostaining was associated with the ISUP grade parameter (p = 0.039) and perineural invasion (p = 0.020) (Online Resource 10).

NKX3.1
NKX3.1 protein was expressed differently between tumor and adjacent non-tumor tissue (p = 0.003), with higher immunostaining and strong intensity in malignant tumor tissue when compared to the adjacent non-tumor tissue; the same result was observed for cellular locations of the immunostaining: cytoplasm (p = 0.003) and nucleus (p = 0.008), in both tissues (Table 1).
There was no significant association between NKX3.1 immunostaining in relation to prognostic groups, prognostic parameters, or biochemical recurrence and metastasis. But it was observed that all patients in the metastatic group (17/17, 100%) presented positive and strong immunostaining, while in the non-metastatic group although strong immunostaining was also verified, this result was not observed in all patients

Multinomial and binary logistic regression
To verify whether clinical-pathological variables that showed statistical significance were independently associated with proteins' (PTEN, AKT, TRPM8 and NKX3.1) tumor staining, a multinomial or binary logistic regression analysis was used. TRPM8 was independently associated with ISUP grade (p = 0.024; OR = 8.373; 95% CI = 1.319-53.164), with the tumor immunostaining being a risk factor in patients with ISUP grade ≥ 4. In the perineural invasion parameter, TRPM8 immunostaining was associated, but not independently. AKT on the other hand was associated with prognostic groups (better and worse prognosis and metastatic), but not independently, as shown in Table 3.

Protein interaction
To assess whether the immunostaining results are related to each other, an interaction analysis was performed, considering the staining only in the tumor tissue for all possible protein combinations. Significant interactions were observed between PTEN and AKT (p < 0.001; χ 2 < 0.001) and PTEN and TRPM8 proteins (p = 0.014; χ 2 = 0.095) (Table 4).

Discussion
The evaluation of four proteins immunostaining by IHC (PTEN, AKT, NKX3.1, and TRPM8) in malignant tumors and adjacent non-tumor tissues of patients with prostate cancer indicated that TRPM8 protein was differentially expressed, with higher immunostaining in malignant tumor tissue. However, our most prominent result was the fact that NKX3.1 immunostaining was observed in all samples in the present study, both in tumor tissue and in the adjacent tissue non-tumor, indicating high specificity for prostate tissue. In addition, there was strong tumor immunostaining for NKX3.1 in all patients of the metastatic group. Certain results obtained from biopsies may be inconclusive, as the biological material collected is insufficient, with few atypical glands for analysis, and a repeat examination may be necessary. Therefore, the use of complementary techniques and the use of new biomarkers are extremely important. A widely used and highly relevant tool is the IHC technique (Kristiansen 2018;Orakpoghenor et al. 2018), which allows the identification of the presence or absence of certain proteins in specific tissues, and the staining intensity is generally used as the gold standard (Jamaspishvili 2018).
It is known that age is one of the main risk factors linked to PCa (Vaidyanathan et al. 2016;Junior et al. 2016;Tse et al. 2018;INCA 2019); therefore, aging increases the risk of developing the disease and its possible aggravation. The significative results obtained in this study comparing the mean ages of patients in the better prognosis and metastatic groups (p = 0.010), in addition to the result of the comparison between the metastatic versus non-metastatic groups (better and worse prognosis) (p = 0.010), confirm that PCa is a disease of advanced age, and that late diagnosis may be associated with a more severe condition of the disease, such as the development of metastases. The genomic deletion of PTEN is very common in PCa, as it is the most lost tumor suppressor in the early stages of the disease (Lotan et al. 2011;Jamaspishvili et al. 2018;Hamid et al. 2019). Consequently, an increase in AKT expression would possibly occur, as these proteins participate in the same signaling pathway (Kurose et al. 2001). Low or absent expression of PTEN was expected in the worse prognosis and metastatic groups; on the other hand, the AKT protein should be more expressed in tumor tissue in these groups of patients. However, we verified that the immunostaining intensities of PTEN and AKT were similar, showing a similar pattern of immunostaining in many of the samples, in the tumor tissue PTEN: 30/59 (50.8%) and AKT: 30/59 (50.8%) and in the adjacent non-tumor tissue PTEN: 27/58 (46.5%) and AKT: 36/58 (62.1%). Our data are not distinct from the literature, as according to the database The Human Protein Atlas (2021a), the PTEN protein has predominantly cytoplasmic immunostaining, and the prostate tissue has low expression of this protein. The AKT protein, on the other hand, presents nuclear immunostaining and in normal prostate tissue, its expression is median (The Human Protein Atlas 2021b).
Tumoral immunostaining of PTEN did not show any significant association with the groups, prognostic parameters, or recurrence and metastasis. AKT immunostaining, on the other hand, was associated and correlated with prognostic groups. Since this is an oncogenic protein, its presence may favor tumor growth.
The literature data demonstrate that PTEN and AKT are part of the same cell survival pathway, PI3K/AKT/mTOR, where PI3K acts as an agonist of the pathway converting PIP2 into PIP3, favoring the binding of AKT to PIP3, which will mediate, through activation of proteins, cell growth, proliferation, survival, and migration (Gonçalves et al. 2018). PTEN protein, on the other hand, acts as a direct antagonist of the pathway, converting PIP3 into PIP2, therefore, having a role as a lipid phosphatase (Gonçalves et al. 2018;Jamaspishvili et al. 2018). This role of antagonist (PTEN) and agonist (AKT) of the cell survival pathway was confirmed in our protein interaction analysis, in which it was observed that strong AKT intensity immunostaining was present when absent or weak intensity immunostaining of PTEN was observed. Asuthkar et al. (2015aAsuthkar et al. ( , b, c, 2017 suggested that TRPM8 is a key element in the testosterone-induced response pathway and that its activity can significantly contribute as an anti-tumor defense mechanism, serving as a new therapeutic target. The immunostaining of the TRPM8 protein, in general, was quite weak in the samples of the present study, in the tumor tissue: 42/60 (70.0%) and in the adjacent non-tumor tissue TRPM8: 39/59 (66.1%). In addition, it was more present in tumor tissue compared to adjacent non-tumor tissue and a significant difference was observed in immunostaining of this protein in the cellular location. According to the literature data, this protein is highly expressed in the prostate, both in tumor-free individuals and in patients with PCa (Asuthkar 2017). Our results do not support this study, but the fact that this protein was more marked in the tumor tissue than in the adjacent tissue non-tumor suggests the need for future studies with new sample groups, to verify whether it may have any correlation with the malignant change. Asuthkar et al. (2015b) found that TRPM8 protein and testosterone are directly involved in localized interactions in the plasma membrane of cells in the periphery of the prostate and in the plasma membrane of the endoplasmic reticulum of cells in the lumen. Furthermore, the authors suggested that testosterone-induced TRPM8 might be an important regulator of Ca 2 homeostasis and the cell cycle in prostate cells. Although TRPM8 mRNA levels increase during prostate tumor progression (Tsaveler et al. 2001), protein levels are not proportionally equal. Asuthkar et al. (2015c) found that TRPM8 is redirected to degradation in PCa, while protein recovery effectively suppresses tumor cell growth. This fact could explain the low expression of this protein in the samples of the present study.
One of the main prognostic factors described in the literature is the histological grade, with the Gleason score being the most used (Cambruzzi et al. 2010). This is a very important prognostic parameter in the assessment of tumor progression and aggressiveness (Löbler et al. 2012). TRPM8 was associated with perineural invasion (p = 0.020), but not with the groups or biochemical recurrence and metastasis; however, it was independently associated with the ISUP grade ≥ 4 (p = 0.024; OR = 8.373; 95% CI = 1.319-53.164). Yu et al. (2014) and Yee et al. (2015) also demonstrated an association between TRPM8 and Gleason score, associated the immunostaining of this protein with a high score.
Another interesting result was the interaction observed between PTEN and TRPM8 proteins, in which it was verified that the presence of one protein is associated with the presence of the other. However, when analyzing the interaction through the String Database (2021), we found no direct interaction between these proteins, but that TRPM8 can interact with other peripheral proteins of the PI3K/AKT/ mTOR and PTEN signaling pathway, such as PPP1CA, PPP1CB, and PPP1CC, for example, which are catalytic serine/threonine-protein phosphatase subunits, and associate with several regulatory proteins to form highly specific holoenzymes, aiming to dephosphorylate hundreds of biological targets.
NKX3.1 protein presented the most intense immunostaining in the tissues of the patients with PCa in the present study, with a significant difference being observed in the tumor tissue (48/59; 81.3%) versus adjacent non-tumor tissue (29/57; 50.9%). Differences in immunostaining regarding cellular locations (nucleus and cytoplasm) were also observed. These results corroborate the study of Gurel et al. (2010), which clearly showed the nuclear immunostaining of NKX3.1 present in practically all analyzed samples, presenting a high pattern of nuclear staining and a high rate of positivity in metastases. According to the database The Human Protein Atlas (2021c), NKX3.1 presents immunostaining both in the nucleus and in the cytoplasm, with nuclear staining being the most evident. In addition, this protein has a high expression rate in the prostate tissue.
NKX3.1 is an androgen-regulated homeobox gene, and its expression is almost exclusively restricted to the prostate (He et al. 1997). In adulthood, NKX3.1 stimulates the repair of DNA damage induced by transcription, through interaction with topoisomerase I, and this interaction is mediated by its homeodomain (Bowen et al. 2007;Puc et al. 2015). Therefore, NKX3.1 is a prostate-specific transcription factor/pioneer, which functions to specify prostate development in addition to its tumor suppressor role (Griffin et al. 2022).
It is already well established that clinicopathological parameters used when studying neoplasms are extremely important, as the data help to confirm and classify patients, and can help guide more effective therapy. The prognosis of PCa is fundamentally related to some histopathological data, such as topography/laterality, tumor volume/size, histological type, degree of differentiation, presence of capsular to extraprostatic neoplastic invasion, state of the surgical margins, and the presence of metastases in regional or distant lymph nodes (Cambruzzi et al. 2010). Within this context, PTEN and TRPM8 proteins were not significantly correlated with prognostic parameters, biochemical recidive or metastatic cases evaluated in our samples. However, AKT was positively correlated with prognostic groups, but not with biochemical recidive or metastatic cases evaluated in our samples.
NKX3.1 immunostaining was negatively correlated with the metastatic group, where all patients had intense immunostaining. Previous studies found that NKX3.1 expression was strongly present in metastatic PCa samples, showing high sensitivity for the prostate, and even when associated with PSA (Kristiansen 2017), PSMA , or HOXB13 (Abouhashem and Salah 2020), they were considered good markers for detecting metastases of prostatic origin. In addition to these studies, the International Society of Urological Pathology (ISUP) (Epstein et al. 2016) has already indicated the NKX3.1 protein as an excellent biomarker of prostate origin in PCa metastases, being highly specific for this tissue.

Conclusions
AKT and TRPM8 may be prognostic markers for prostate cancer that deserve future investigation. Our results also confirm the NKX3.1 specificity for prostate tissue and that it can be used to identify a primary site of metastasis. This marker needs further investigation for its validation as a candidate for early prediction of this phenomenon, given its immunostaining profile in metastatic patients.
Acknowledgements All the authors would like to thank the Hospital do Câncer de Londrina and Angela Navarro Gordan for providing the samples for this study.
Author contributions ÉRP participated in the study design and acquisition of data, experimental procedures, performed the statistical analysis and interpretation, and drafted the manuscript. ALF and LCLP participated in the collection of samples and medical records, and also participated in the study design and immunohistochemical reactions. CAM participated in the study design and experimental procedures. AFMLG participated in the histopathological assays for the selection of tumor tissues and adjacent non-tumor tissues and performed the immunohistochemical analysis of the samples. KBdO participated in the statistical analysis and data interpretation. PEF made the sample collection possible. IMdSC participated in the design of the study and reviewed the manuscript for important intellectual content. RLG participated in the design of the study, interpretation of data, and gave final approval of the version to be published. All the authors read and approved the final manuscript.