Prostate cancer hijacks the microenvironment

Prostate cancer is difficult to treat because of molecular, cellular and clinical heterogeneity. Using single-cell RNA sequencing, a recent study reveals unexpected transcriptomic reprograming in immune cells and non-immune components of the tumour microenvironment, which may lead to viable therapeutic approaches against prostate cancer.

Despite a declining death rate over the last 30 years thanks to the advent of new treatment regimens, prostate cancer (PCa) remains the most dominant male malignancy in the world. At the initial diagnosis, PCa can be clinically localised (77%), extended to pelvic lymph nodes (11%), or found to have been metastatic (6%). The five-year survival rate for localised PCa is almost 100%, but for metastatic prostate cancer (mPCa) this drops to only 31%1. Although mPCa is initially responsive to inhibition of androgen receptor (AR) signalling and many men benefit from years of cancer control, others manifest rapid resistance. These resistant forms of mPCa lack standard-of-care therapy and have developed through heterogeneous mechanisms. They may also become AR null or undergo lineage plasticity (that is, the ability of cells to transition from one committed pathway to another), resulting in a mean survival rate of 12 months2. Therefore, to meet this clinical need and provide effective treatment opportunities for these patients, it is paramount to study cancer heterogeneity and clonal diversity alongside the development and impact of the tumour microenvironment (TME), which includes blood vessels, immune cells, fibroblasts and the extracellular matrix (ECM). In this issue of Nature Cell Biology, Chen et al.3 utilised single-cell transcriptomic analysis of 24 PCa samples to study intracellular heterogeneity and TME factors, such as tumour-infiltrating immune components. By including primary PCa, lymph node metastases, as well as non-cancerous prostate tissue, the authors made some unexpected observations regarding the TME.

Chen et al. found that basal and shared basal–luminal (‘intermediate’) cell types were the most dominant epithelial class in one high Gleason score, non-metastatic sample, while luminal was the dominant cell type for most tumours. Identifying a sample with high abundance of basal/intermediate cells is an unusual observation, contradicting at least one prior study showing nominated luminal cells as the major cell subpopulation in PCa4. Furthermore, they found that basal/intermediate cell types are transcriptionally distinct from non-transformed prostate basal cells, expressing lower amounts of basal marker genes. This observation could explain the absence of these cell markers in histopathological analysis, suggesting that basal/intermediate cells may be lurking in the background. Basal/intermediate cells were enriched for cell-cycle-related G2/M transcriptomic signatures, which correlated with depletion of luminal cell functions. This is particularly interesting because cell cycle activation (CCA) is a diagnostic predictor for disease outcome in multiple PCa patient cohorts5. Moreover, CCA correlated with activity of multiple metastasis-related transcriptional machineries in the TME across PCa samples. Thus, the authors interrogated the contribution of TME-associated factors and thereby discovered potential mechanisms underlying the immune-suppressive environment of PCa.

Notably, the authors identified a subset of tumour-associated macrophages (TAMs) that showed higher activation of osteoclast (OC)-related pathways across all samples. This is an intriguing observation, as bone is the main metastatic site for PCa, wherein OCs play a major role6. TAMs are monocytes hijacked by tumour cells via chemokine cues, which is followed by migration to the TME, inducing DNA damage in surrounding endothelial cells and producing growth factors favouring tumour outgrowth7.

Besides alterations in TAMs, Chen et al. also identified expression of KLK3 expression—also known as prostate-specific antigen (PSA)—within the tumour’s CD8+ effector T cells (the prototypical antitumour immune cells). Furthermore, KLK3-positive T cells were observed at lymph node (LN) metastasis sites. This finding is surprising, as T cells are considered AR negative and KLK3 is an AR-regulated gene. As expected, AR activity was absent in these T cells, indicating that KLK3 expression in T cells was not underlying endogenous regulation. Additionally, they found enrichment of extracellular vesicle (EV) and exosome trafficking pathways, exclusively in the KLK3-positive T cell cluster, hinting towards EV-mediated exchange from tumour cells to T cells (Fig. 1). Functional studies confirmed that KLK3 levels in T cells were elevated when exposed to AR-dependent but not AR-negative PCa cells. EVs from these cultures were examined, resulting in the exclusive detection of KLK3-containing EVs under AR-dependent cell line co-culture conditions. In addition, other prostate-specific genes were detected in KLK3-positive T cells (for example, FOLH1, encoding the prostate-specific membrane antigen (PSMA)). While these results indicate a role of exosome trafficking between cancer cells and the TME, it would have been interesting to study the nature of these EVs in more detail.

Fig. 1: Prostate cancer enables the establishment of a metastatic TME niche.

Immune and non-immune components of the TME infiltrate the prostate tumour and are then primed by the tumour (for example, by uptake of tumour-derived extracellular vesicles (EVs)) to facilitate spread to metastatic sites. a, T cells infiltrate the tumour and get primed by tumour-derived EVs to express KLK3. Primed KLK3-positive T cells migrate to lymph nodes and attract PCa cells to metastasize at these sites. b, Primed non-immune cells of the TME, such as cancer-associated fibroblasts (CAFs) and activated endothelial cells (aECs), support cancer metastasis from within the tumour by modifying the extracellular matrix (ECM). c, Osteoclast (OC)-like tumour-associated macrophages (TAMs) migrate to bone and pave the way for PCa cells to metastasize at bone sites. Figure created with

To assess whether this observed accumulation of tumour markers in T cells is PCa specific, the authors re-analysed public single-cell RNA sequencing (scRNA-seq) data for four ancillary cancer types, including lung, neck, colorectal and hepatocellular malignancies, and confirmed the expression of tumour-specific marker genes in T cells for all four cancer types. To strengthen the clinical importance of these findings, two external iliac obturator lymph nodes (LN) and primary PCa tumour samples were isolated from a high-risk patient. Previous assessment of this patient, using magnetic resonance imaging and histopathological analysis, concluded the absence of metastasis in the LNs. Accordingly, scRNA-seq analysis revealed separation of tumour cells from LN samples. Interestingly, in the left, but not right, LN, a small number of epithelial cells was identified, but no metastasis was recognised beforehand. Widespread ectopic KLK3 expression was observed across all the cell types in the left LN, with the largest population being T cells. Of note, they detected few KLK3-postive T and B cells in the right LN. Based on these findings, the researchers posited that altered immune cell gene expression in LNs occurs before induction of metastasis, indicating a pre-metastatic niche. These findings are in line with pre-clinical studies reporting that KLK3 quantification in LNs was correlated with prevalence of micrometastasis8,9. However, in this setting KLK3 expression was detected on bulk tissues and could be associated with tumour cells, but not TME. Thus, the exact mechanistic contributions of these T cells to actively promote metastasis remain to be investigated.

Besides immune cells, the non-immune compartment of the TME, which includes fibroblasts and endothelial cells (ECs), manifests increased heterogeneity in many cancers10,11. Previous studies have shown the importance of cancer-associated fibroblasts (CAF) in PCa12. Hence, Chen et al. identified three distinct subtypes of CAFs in their cohort. Surprisingly, high expression of ACTA2, a common marker for CAFs in other cancers but not in PCa13, was detected in all CAF subsets. ACTA2 expression correlates positively with the epithelial-to-mesenchymal transition (EMT) score in epithelial cells, indicating that on-going EMT in CAFs may be a possible source of ACTA2 abundance. Furthermore, scRNA-seq data analysis revealed that all CAF subtypes were associated with angiogenesis, while signatures for myofibroblastic association, cell adhesion and extracellular matrix (ECM) were more subtype specific. The authors summarise that these data are direct evidence for a shared regulatory network between CAF and non-fibroblastic lineages in the TME (Fig. 1).

ECs play a major role in cancer cell spreading by providing blood flow to tumour sites14. Based on differentially expressed gene profiles, Chen et al. identified four EC subtypes with high expression of CAF-related genes, which they termed activated ECs (aECs). By applying pseudotime analysis of the identified aEC subtypes, it was evident that, based on the underlying transcriptomic signatures, two divergent cell fate scenarios are likely to occur during disease progression to mPCa. Moreover, aECs had higher numbers of interactions with epithelial cells and substantially dysregulated ligand/receptor gene expression compared to ECs. EC clusters were mainly enriched in pathways related to immune response, whereas common pathways in aEC clusters were related to ECM–receptor signalling and focal adhesion. These findings were further validated in samples from patients with PCa and bladder cancer, which suggests a role of aECs in modifying the ECM, supporting metastatic outgrowth and suppressing immune activation (Fig. 1).

Due to the underrepresentation of immune cells in the prostate, PCa has been described as an immune ‘desert’. Consequentially, immunotherapy has not been routinely incorporated as a strategy to treat patients with PCa, and the potential for therapeutic benefit remains unclear. In this context, the current study is highly relevant, as it provides unexpected evidence that the immune component of PCa TME correlates with metastatic spread. In particular, tumour-marker-expressing T cells and TAMs with OC-like features that potentially facilitate a metastatic niche at prominent invasion sites could be therapeutically relevant.

In conclusion, Chen et al. reveal a metastasis-associated transcriptomic landscape in primary tumours of the prostate, emphasising the potential of single-cell sequencing for assessing clinical outcome and therapy response, especially when associated with rare cell populations within the tumour and tumour-associated TME. Yet, further mechanistic studies are required to confirm an active role of T cells in promoting mPCa, which currently remains speculative. Although KLK3 is expressed in distant and lymph node metastasis, it is notably lower than that in benign prostate tissue15. Presumably, KLK3 positivity derived from T cells is a rare event, and hence it remains hypothetical that T cells play a major role in orchestrating PCa metastasis. To exploit the therapeutic value of these findings, it would be important to understand how the distinct cellular repertoire of the TME actually impacts mPCa and how different TME compartments synergise with each other.


  1. 1.

    Siegel, R. L., Miller, K. D. & Jemal, A. CA Cancer J. Clin. 70, 7–30 (2020).

    Article  Google Scholar 

  2. 2.

    Metzger, A. L. et al. Prostate 79, 1452–1456 (2019).

    Article  Google Scholar 

  3. 3.

    Chen, S. et al. Nat. Cell Biol. (2021).

  4. 4.

    Wojno, K. J. & Epstein, J. I. Am. J. Surg. Pathol. 19, 251–260 (1995).

    CAS  Article  Google Scholar 

  5. 5.

    Cuzick, J. et al. Lancet Oncol. 12, 245–255 (2011).

    CAS  Article  Google Scholar 

  6. 6.

    Logothetis, C. J. & Lin, S.-H. Nat. Rev. Cancer 5, 21–28 (2005).

    CAS  Article  Google Scholar 

  7. 7.

    Sica, A. et al. Semin. Cancer Biol. 18, 349–355 (2008).

    CAS  Article  Google Scholar 

  8. 8.

    Heck, M. M. et al. Eur. Urol. 66, 222–229 (2014).

    Article  Google Scholar 

  9. 9.

    Lunger, L. et al. Prostate Cancer Prostatic Dis. (2020).

  10. 10.

    Nguyen, M. et al. Cell Rep. 25, 3884–3893.e3 (2018).

    CAS  Article  Google Scholar 

  11. 11.

    Becker, L. M. et al. Cell Rep. 31, 107701 (2020).

    CAS  Article  Google Scholar 

  12. 12.

    Vickman, R. E. et al. Prostate 80, 173–185 (2020).

    CAS  Article  Google Scholar 

  13. 13.

    Ayala, G. et al. Clin. Cancer Res. 9, 4792–4801 (2003).

    CAS  PubMed  Google Scholar 

  14. 14.

    Schaaf, M. B., Garg, A. D. & Agostinis, P. Cell Death Dis. 9, 115 (2018).

    Article  Google Scholar 

  15. 15.

    Queisser, A. et al. Mod. Pathol. 28, 138–145 (2015).

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Mark A. Rubin.

Ethics declarations

Competing interests

The authors declare no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Thienger, P., Rubin, M.A. Prostate cancer hijacks the microenvironment. Nat Cell Biol 23, 3–5 (2021).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing