Emerging role of PTEN loss in evasion of the immune response to tumours

Abstract

Mutations in PTEN activate the phosphoinositide 3-kinase (PI3K) signalling network, leading to many of the characteristic phenotypic changes of cancer. However, the primary effects of this gene on oncogenesis through control of the PI3K–AKT–mammalian target of rapamycin (mTOR) pathway might not be the only avenue by which PTEN affects tumour progression. PTEN has been shown to regulate the antiviral interferon network and thus alter how cancer cells communicate with and are targeted by immune cells. An active, T cell-infiltrated microenvironment is critical for immunotherapy success, which is also influenced by mutations in DNA damage repair pathways and the overall mutational burden of the tumour. As PTEN has a role in the maintenance of genomic integrity, it is likely that a loss of PTEN affects the immune response at two different levels and might therefore be instrumental in mediating failed responses to immunotherapy. In this review, we summarise findings that demonstrate how the loss of PTEN function elicits specific changes in the immune response in several types of cancer. We also discuss ongoing clinical trials that illustrate the potential utility of PTEN as a predictive biomarker for immune checkpoint blockade therapies.

Background

Mutations in specific genes, including activating mutations in oncogenes and loss-of-function mutations in tumour-suppressor genes, are critical for cancer development.^1–3 The inactivation or deletion of tumour-suppressor genes not only facilitates tumour development and progression, thereby keeping these cells in a continuous proliferative state,⁴ but might also influence immune responses.^5,6 For instance, somatic inactivation or deletion of the gene that encodes the tumour-suppressor phosphatase and tensin homologue (PTEN)⁷ leads to the activation of phosphoinositide 3-kinase (PI3K) and subsequent downstream signalling to AKT and mammalian target of rapamycin (mTOR),⁸ a serine/threonine kinase that regulates cell growth, survival and proliferation.⁹ PTEN loss is also linked to aggressive cancer phenotypes.¹⁰ However, new studies have shown that, in addition to its established role in cancer progression, PTEN deficiency can lead to an immunosuppressive tumour microenvironment (TME)^11–14 that is unfavourable for effective antitumour immune responses.¹⁵

Immune and tumour cells often communicate with each other by secretion of factors such as cytokines, chemokines, interleukins and extracellular vesicles.¹⁶ These signalling molecules can enhance or inhibit the activity of cytolytic cells, primarily represented by natural killer (NK) and CD8⁺ T cells.^17,18 When CD8⁺ T cells directly recognise tumour antigens as non-self or are presented with tumour antigens through dendritic cells and macrophages, they target and kill the malignant cells.¹⁹ On the other hand, cytolytic cells are suppressed by cell-to-cell contact with regulatory T (Treg) cells and other immunosuppressive cells, and this is one of the several immune evasion mechanisms that cancers use to suppress antitumour immune responses.²⁰

Immunotherapy does not target tumour cells directly: it acts by boosting the natural propensity of the immune system to recognise and respond to the presence of tumours.²¹ The principle of immune checkpoint inhibition involves the use of specific agents that block the suppressive interactions between a developing tumour and the defensive immune system of the patient.²² Immune checkpoint proteins such as programmed death ligand 1 (PD-L1), for example, downregulate the immune system and promote self-tolerance by suppressing T cell inflammatory activity against tumours, so that blocking the expression of checkpoint proteins with immune checkpoint inhibitors (ICIs) usually restores the capacity of the immune system to recognise tumour cells and kill them.^23–25

Therapeutic response of individuals treated with ICIs is highly heterogeneous across tumour types, with some patients showing durable responses in previously incurable and highly aggressive cancers, but the majority of patients still have no appreciable benefit with these new drugs.²⁶ Most mechanisms of resistance to ICIs involve functional changes occurring in the TME as well as the acquisition of intrinsic tumour-cell resistance.²⁷ Tumour cell-intrinsic somatic mutations function at two different levels: not only do they promote higher levels of cell proliferation and cancer growth but they also activate mechanisms that allow tumours to avoid immune system attack.^28,29 Recent findings have shown that the overall pattern of somatic mutations and genomic changes in tumours can be associated with resistance to immunotherapy in several different types of cancer,^30,31 indicating that somatic point mutations, copy number alterations, epigenetic changes and dysregulation of gene expression can profoundly influence the interactions between immune and cancer cells.^29,32 However, presently, the various molecular determinants of resistance to immunotherapy are complex and poorly characterised.

In this review, we discuss the most recent findings on PTEN in genomic stability, cancer immunogenicity, immune cell infiltration and the immune response across different cancers and summarise the emerging role of PTEN as a predictive biomarker for the use of ICIs. The techniques for detecting PTEN deficiency in tumours are now highly reproducible (see Table 1), which increases the potential utility of this biomarker for disease stratification³³ and for predicting the chances of a successful response to immunotherapy.

Table 1.

Laboratory tests for the detection of PTEN loss in tumours.

Technique	Advantages	Disadvantages
Immunohistochemistry (IHC)	Sensitive, rapid and semiquantitative detection. Excellent morphological correlation. Can be used as the main assay to screen for PTEN loss (with FISH as reflex test if required)	Need to use validated method120 and an established IHC antibody. Detection of heterogeneity or ambiguous results requires confirmatory FISH of adjacent section
Fluorescence in situ hybridisation (FISH)	Accurate. Good correlation of findings with morphology. Detects genomic heterogeneity and hemizygous deletions	More laborious and costly than IHC. Recommended to perform initial screening by IHC, followed by FISH analysis in cases that are ambiguous or indeterminate by IHC120
Quantitative PCR techniques, MLPA	Fast, sensitive detection of clonal gene copy number changes	Cellular or genetic heterogeneity not detected. No morphologic correlation
Sequencing-based gene dosage analysis and detection of point mutation	Highly sensitive for detection of somatic point mutations, partial deletions and indels	Large PTEN deletions and copy number heterogeneity not easily detected

The functions of PTEN

PTEN, one of the most commonly somatically mutated or deleted genes in cancer,⁷ is located at the 10q23.31 cytoband and consists of 9 exons that encode the full-length molecule of 403 amino acids.^34,35

Cytoplasmic functions of PTEN

In the cytoplasm, the PTEN protein acts as a dual-specificity phosphatase and a direct antagonist of PI3K signalling by converting the second messenger phosphoinositol-(3,4,5)-trisphosphate into phosphoinositol-(4,5)-bisphosphate, which inhibits downstream signalling pathways³⁶ that would otherwise normally mediate downstream activation of the AKT protein. Phospho-AKT activates different substrates, such as mTOR,³⁷ a serine/threonine kinase that regulates cell growth, survival and proliferation. Somatic PTEN inactivation in tumours causes total loss of PTEN function that disrupts not only its catalytic phosphatase activities but also its regulatory control of the PI3K pathways. PTEN loss leads to downstream changes that govern important cellular processes crucial to cancer progression, including survival, proliferation, energy metabolism and changes in cellular architecture.

Nuclear functions of PTEN

In addition to its cytoplasmic functions in regulating cell growth and proliferation, PTEN regulates genome integrity and the stability of DNA repair in the nucleus.³⁸ Pten-null mice show increased genomic and chromosomal instability, leading to centromere disruption, chromosomal translocations and spontaneous DNA double-stranded breaks that appear to occur independently of the PI3K–AKT–mTOR pathway. The protective role of PTEN in the genome is also supported by the finding that the PTEN.R189X mutation (which lacks the C-terminal region that is responsible for binding to the centromere protein CENP-C) leads to a significant increase in chromosomal aberrations, with transfected PTEN189 cells developing high levels of aneuploidy. This mechanism of protection is directly regulated by the interaction of PTEN with CENP-C, as well as by PTEN regulating the expression of Rad51 and its influence on the double-stranded break repair machinery.³⁹ In glioblastoma, DNA repair is attenuated after cell exposure to ionising irradiation when nuclear PTEN is phosphorylated at position 240.⁴⁰ The phosphorylated PTEN binds to chromatin and recruits RAD51 to promote DNA repair. In prostate cancer, PTEN also binds and promotes the degradation of the DNA-binding factor chromodomain helicase DNA-binding protein 1 (CHD1); PTEN deficiency leads to CHD1 protein stabilisation, which then engages the H3K4me3 epigenetic mark to activate transcription of downstream pro-tumorigenic and pro-inflammatory tumour necrosis factor α (TNFα)/nuclear factor κB (NF-κB) gene networks.⁴¹ NF-κB is a complex transcription system governing a diverse set of response genes mediating inflammatory and stress responses. The effects of PTEN on NF-κB transcription are also regulated by a translational variant of PTEN, called PTEN-L. The specific effects of loss of PTEN and PTEN-L on NF-κB-related immune responses are discussed below.

PTEN also regulates cellular senescence (i.e. irreversible arrest of cellular proliferation) mechanisms in cells that have lost proliferative capacity (reviewed in ref. ⁴²). Loss of PTEN induces cellular senescence as a failsafe mechanism to defend against tumorigenesis.⁴³ In prostate cancer, PTEN loss has been shown to activate p53-dependent cell senescence by mTOR kinase binding to p53.^43,44 Concomitant loss of p53 allows cells to override the cytostatic effects of PTEN-induced senescence. Nuclear PTEN also interacts with the anaphase-promoting complex (APC) and regulates cellular senescence through APC–cadherin 1-mediated protein degradation.⁴⁵ Moreover, senescence activation in PTEN-deficient cells is also associated with cytokine secretion that leads to an immunosuppressive TME.⁴⁶

The various established PTEN signalling pathways associated with PI3K, genome stability and emerging features related to immune responses are summarised in Fig. 1.

Fig. 1. PTEN functions in the cytoplasm and nucleus of cells.

Tumour-suppressor functions: The PI3K–AKT–mTOR pathway is negatively regulated by PTEN in the cytoplasm through the dephosphorylation of phosphatidylinositol (3,4,5)-trisphosphate (PIP₃) to phosphatidylinositol (4,5)-bisphosphate (PIP₂). Increased activation of PI3K–AKT–mTOR leads to abnormal cell growth and proliferation. Cell senescence: PTEN also regulates cell senescence through the PI3K–AKT–mTOR network: the mTOR complex directly phosphorylates p53 that promotes the accumulation of p21. Consequently, p21 induces cell senescence. Nuclear PTEN also interacts with the anaphase-promoting complex (APC) and regulates cellular senescence through an APC–cadherin 1 complex. In this manner, PTEN loss promotes cell senescence as a failsafe against tumorigenesis. Immune and inflammatory response: PTEN negatively regulates the nuclear factor κB (NF-κB) signalling pathway through chromodomain helicase DNA-binding protein 1 (CHD1), which is ubiquitinated (Ub) in the presence of PTEN and thus is unable to promote the transcription of NF-κB genes in the nucleus. On the other hand, PTEN-L promotes the nuclear import of p65, which consequently induces the transcription of NF-κB genes. The presence of cytoplasmic DNA—as a consequence of genomic instability—activates the STING pathway, which phosphorylates interferon-regulatory factor 3 (IRF3). PTEN and PTEN-L are required for the migration of IRF3 into the nucleus, where this transcription factor mediates the immune response by promoting the expression of type I interferon (IFN) genes, such as interleukin (IL)-6 and chemokine (C-X-C motif) ligand 1 (CXCL1). DNA integrity: The concomitant presence of DNA damage repair (DDR) gene mutations or genome instability leads to double-stranded breaks (DSBs) in the DNA, which often causes self-DNA to migrate into the cytoplasm. Such genomic changes are also controlled by PTEN, since this tumour suppressor regulates cell cycle checkpoints, maintains centrosome stability and is involved in DNA repair.

PTEN inactivation, genomic instability and cancer immunogenicity

As outlined above, PTEN normally regulates several mechanisms that maintain genome stability in the nucleus, so PTEN-deficient cancers are known to exhibit high rates of chromosome rearrangements^47,48 that are usually associated with increased mutational load.⁴⁹ To better understand the consequences of aneuploidy in tumours, >10,000 tumours from The Cancer Genome Atlas were investigated in two studies.^49,50 Both reports found that high levels of aneuploidy (defined as a whole chromosome or chromosome arm imbalances) were associated with decreased immune response levels, as determined by a low level of lymphocyte infiltration⁴⁹ and a high expression of anti-inflammatory cytokines.⁵⁰ However, this observation—that tumours with high aneuploidy levels are infiltrated by leukocytes to a lesser extent—seems to contradict the concept that increased levels of genomic change (higher mutational load) are often associated with increased immunogenicity.³² Tumours with increased mutational burden are more likely to produce neoantigens, which can be recognised as non-self and elicit an immune response. Increased immunogenicity may lead to a higher immune-cell abundance in the TME, improved patient survival and a better response to ICIs.^51,52 As PTEN inactivation promotes higher rates of genomic instability,⁴⁷ it would be generally expected that PTEN-deficient tumours would be pro-inflammatory, having a greater mutational burden and exhibiting increased immunogenicity in the TME. However, for many of the aneuploid tumours, it seems that immunoresistance may be able to suppress the inflammatory antigenic effects of higher mutational loads and increased genomic change⁴⁹ (see Fig. 2).

Fig. 2. Pro- and anti-inflammatory effects of PTEN deficiency on the immune response of cancer.

PTEN loss is associated with cytokine and chemokine signalling that creates an immunosuppressive microenvironment. The TME becomes populated with immune cells that suppress the antitumour response, such as myeloid-derived suppressor cells (MDSCs), regulatory (Treg) cells and M2 macrophages (left side). PTEN-deficient tumours have also been linked with higher indoleamine 2,3-dioxygenase 1 (IDO1) and PD-L1 expression, which are known to reduce the activity of cytotoxic immune cells capable of killing cancer. In contrast, PTEN deficiency may also result in pro-inflammatory effects due to loss of the various nuclear functions of PTEN (right side). For example, there are higher levels of genomic instability, which may result in tumours that produce neoantigens. Neoantigens can elicit an immune response and activate CD8⁺ T cells. However, to counteract the effects of neoantigens, it is likely that tumours with high levels of genomic instability are able to suppress host immune responses to counter pro-inflammatory activity. TAM tumour-infiltrating macrophage.

Emerging evidence indicates that, in addition to growth-promoting PI3K alterations, PTEN loss is also associated with an immunosuppressive tumour state^{11,47,53–55} that favours cancer progression.^53,54,56,57 For instance, in a murine model of T cell acute lymphoblastic leukaemia, Pten deficiency was reported to promote cell survival in a previously unsupportive TME.⁵⁸ As discussed above, PTEN loss triggers senescence as a failsafe protection mechanism against cancer onset. Depending on the genetic changes already present in the tumour, PTEN-induced senescence can elicit changes in the cytokine network that result in an antitumour immune response in the TME⁴⁶ (see Figs. 2 and 3).

Fig. 3. Proposed model of the TME of PTEN-deficient and PTEN-intact cancer.

Tumour cells harbouring PTEN loss are linked to a highly immunosuppressive environment mediated by myeloid-derived suppressor cells (MDSCs), regulatory (Treg) cells and M2 macrophages. In this model, the various changes caused by PTEN loss (shown schematically in Fig. 2) are thought to interact to suppress antitumour responses. There are likely to also be uncharacterised tumour-specific differences in immune response. For example, there is no consensus for the relationship between PTEN loss and CD8⁺ T cell density (for more details, see Table 2). TAM tumour-infiltrating macrophage, PD-L1 programmed death-ligand 1, IDO1 indoleamine-pyrrole 2,3-dioxygenase.

PTEN and the immune response

Several reports have shown that mutated oncogenes and tumour-suppressor genes have secondary functions in regulating the immune response in the TME in addition to their primary role in cell proliferation and survival in cancer (reviewed in ref. ⁵). In addition to its tumour-suppressor functions, PTEN normally influences the innate immune response by dephosphorylating Ser⁵⁹ of interferon regulatory transcription factor 3 (IRF3) to enable its nuclear migration.¹³ Once in the nucleus, IRF3 promotes the transcription of type I interferon (IFN)-response genes.¹³ Type I IFN responses promote the expression of several genes that (i) control the response against infectious agents, (ii) modulate antigen presentation and cytokine expression and (iii) activate the T cell- and B cell-mediated immune response.⁶⁰ Therefore, the activation of the type I IFN network is key in balancing the adaptive and innate immune responses. Both PTEN and PTEN-L have been shown to influence the transcription of type I IFN-regulated genes during viral infections by regulating IRF3 activation. PTEN-L regulates the import of IRF3 and the NF-κB component p65 into the nucleus, where both transcription factors activate type I IFN response and NF-κB-regulated genes, respectively.¹⁴ NF-κB regulates both immune and inflammatory responses and tumour-cell proliferation and apoptosis.⁶¹ PTEN-mediated enhancement of the type I IFN response is expected to promote the recruitment and activation of antitumour T cells and NK cells in addition to stimulating antigen presentation by dendritic cells.⁶² It is known that the hyperactivation of type I IFN responses is strongly linked to a better response to immunotherapy and increased activation of antigen presentation by tumour cells (reviewed in ref. ⁶³). As a result, PTEN-deficient tumours might show impaired activation of both the type I IFN and the NF-κB pathways, which could be highly favourable for tumour progression because of an immunosuppressed TME.

The host antitumour immune response can also be triggered by cancer-specific genomic alterations and mutations.³² For instance, as a consequence of double-stranded DNA breaks caused by the presence of inactivating mutations in DNA damage repair (DDR) genes, there might be increased levels of DNA fragments in the cytoplasm of tumour cells.⁶⁴ These increased levels of DNA fragments might activate the cytosolic double-stranded DNA-sensing cGAS-STING pathway, which promotes the phosphorylation of IRF3 (see Fig. 1), thereby promoting its retention in the cytoplasm. In this manner, PTEN deficiency in combination with DDR-mutated tumours might further attenuate the levels of immune response activation due to the lack of transcription of IFN-regulated genes by IRF3.⁶⁵ Cao and colleagues¹⁴ showed that cells expressing only PTEN-L synthesised the highest levels of C-X-C motif chemokine ligand (CXCL) 10 and the pro-inflammatory cytokines interleukin (IL)-6 and CXCL1. In addition, Wang and colleagues⁶⁶ demonstrated that the knockdown of PTEN-L significantly reduced the expression of TNFα and IL-6, which both have a pro-inflammatory role in mediating immune response. In this context, loss of PTEN and its secreted variant PTEN-L in tumour cells is likely to increase the activity of immunosuppressive cytokines on the immune cells in the TME. Interestingly, another three translational variants of PTEN exist; however, little is known about their effects in the immune response (reviewed in ref. ⁶⁷). Thus the emerging role of PTEN in mediating the immune response is complex and involves multiple pathways that collectively repress antitumour responses in the TME (shown schematically in Figs. 2 and 3).

PTEN is also known to play a crucial role in the regulation of signalling in immune cells.¹² Pten deletion of specific immune cell subsets in mice cause defects in T cells,⁶⁸ Treg cells⁶⁹ and B cells.⁷⁰ Myeloid cell-specific PTEN deficiency and increased PI3K levels reduce inflammation, increase macrophage phagocytic ability and facilitate resistance to infection.⁷¹ These findings are consistent with the early observation that PTEN deficiency was associated with chronic inflammation and autoimmunity.⁷² Since inflammation and immunity are known key factors in tumour progression,⁷³ it is important to understand how PTEN mediates responses in the TME.

PTEN deficiency and the immune response in cancer

The impact of PTEN loss and PI3K activity on T cell-mediated antitumour responses was investigated in a preclinical model of melanoma. Inhibition of PTEN-null tumours with a PI3K inhibitor led to in vitro killing by T cells and restoration of antitumour control of immune checkpoint inhibition using anti-PD-L1 and anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA4)³⁰ antibodies. However, the authors demonstrated that PD-L1 and major histocompatibility complex class I expression do not constitute the primary mechanism of immunosuppression mediated by PTEN loss in melanoma. Their xenografting experiments suggested that the lack of response to ICIs could have resulted from a higher expression of vascular endothelial growth factor (VEGF) and increased secretion of inhibitory cytokines by tumour cells lacking PTEN.

PTEN loss was also found to be significantly associated with altered macrophage densities and CXCL8 expression in a study conducted using 28 radical prostatectomy-derived specimens.⁷⁴ CXCL8 is a pro-inflammatory chemokine that binds to the C-X-C chemokine receptor type 1 CXCR1 and CXCR2. Together, the CXCL8–CXCR1/2 axis mediates tumour initiation and progression, in addition to promoting neovascularisation of the TME.⁷⁵ Haploinsufficiency of Pten combined with oncogenic Kras mutations was associated with NF-κB pathway activation in a mouse model of pancreatic ductal adenocarcinoma. The authors of this study also showed that stimulation of the NF-κB network led to TME remodelling through the infiltration of immune cells with pro-tumorigenic properties.⁷⁶

Immune checkpoint inhibition is most often met with little or no success in prostate cancer, owing to low levels of immune-cell infiltration. In Pten-null prostate conditional mice, tumours are strongly immunosuppressed as a result of the downregulation of the tyrosine-protein phosphatase PTPN11 and consequent Janus kinase 2 (Jak2)–signal transducer and activator of transcription 3 (Stat3) pathway activation.⁴⁶ In another study conducted with Pten-deficient mice, additional ablation of the growth-regulatory kinase extracellular signal-regulated kinase 5 (ERK5) promoted T cell infiltration into prostate tumours through Ccl5 and Cxcl10 production.⁷⁷ Thus the concomitant loss of ERK5 was able to restore immune functions caused by PTEN deficiency leading to tumours with higher T cell infiltration. These data provide important clues about why human prostate cancer does not always benefit from current immunotherapeutic approaches.

PTEN loss has also been linked to changes in the secretome (that is, cell-secreted molecules, such as chemokines and ILs) so that its role in mediating immune responses in the TME might also be indirect. In non-small cell lung cancer, CXCR4 expression was shown to be dependent on the PI3K–AKT–mTOR signalling pathway, which is directly controlled by PTEN.⁷⁸ The loss of Pten in stromal fibroblasts of mouse mammary glands led to the altered production of cytokines and chemokines in the TME, increased macrophage recruitment and enhanced vascularisation, which together accelerated the initiation, progression and malignant transformation of mammary epithelial tumours.⁷⁹ Another study conducted with murine prostate epithelial cells and human prostate cancer cell lines demonstrated that the loss of PTEN enhanced the expression of CXCR4 and CXCL12.⁸⁰ Moreover, in prostate cancer cell lines and prostatic tissue, Maxwell and colleagues⁸¹ demonstrated that the expression and secretion of the pro-inflammatory chemokine CXCL8 were also associated with PTEN inactivation.

A recently published study conducted in mice demonstrated that the diversity of genomic changes in prostate cancer can determine the composition of immune cells within the TME.⁸² The authors showed that tumours derived from Trp53^−/−Pten^−/− tumours expressed high levels of Cxcl17, which recruited cells with immunophenotypes of monocytic myeloid-derived suppressor cells (MDSCs) and monocytes. Moreover, these tumours exhibited a Treg cell-mediated immunosuppressive signature that was linked to the tumour-promoting effect of MDSCs. The details of the role of MDSCs in PTEN-deficient tumours are discussed below. On the other hand, simultaneous ablation of Pten and Pml in mice was related to an ‘immune desert’ phenotype, characterised by low T cell densities and immune checkpoint expression. In summary, changes in Pten and the interplay with other genes studied in mouse models of prostate cancer have been shown to collectively shape the immune-cell content and expression of immunosuppressive markers in the TME.

PTEN also undergoes germline-inactivating mutations that are responsible for various human syndromes.⁸³ The PTEN hamartoma tumour syndrome is a hereditary tumour syndrome that predisposes patients to benign and malignant breast, thyroid, renal cell and endometrial tumours.⁸⁴ A recently published study demonstrated that the innate immune cells from patients with PTEN hamartoma tumour syndrome produce high levels of lactate.⁸⁵ Lactate confers a pro-inflammatory phenotype of monocytes in the TME of PTEN-deficient thyroid cancer. Interestingly, when PTEN-deficient thyroid cancer cells were exposed to the immunomodulatory effects of metformin (which is used for treating Type 2 diabetes), there was a more pro-inflammatory phenotype in the TME.⁸⁵ Such a phenotype reflects a more immunogenic TME, which presents increased infiltration of immune cells together with the synthesis of cytokines and chemokines. By contrast, treatment with rapamycin (an mTOR blocker) led to reduced pro-inflammatory cytokine expression. This study provided evidence that the PI3K–AKT–mTOR network directly influences the immune response within the TME of PTEN-deficient tumours. Indeed, mTOR activation directly promotes immune-cell differentiation (reviewed in ref. ⁸⁶). It is thus critical to confirm the multiple roles of PTEN and its related signalling pathways in the control of the inflammatory and immune responses.

PTEN loss and immune cell infiltration

The levels of immunoreactive CD8⁺ T cells can be helpful in evaluating overall immune responses, but the immune-cell landscape of tumours that harbour PTEN somatic mutations is complex and highly heterogeneous, with some studies showing that PTEN deficiency is linked to a high CD8⁺ T cell density^53,87 while other studies have shown the inverse correlation^88–90 (Table 2).

Table 2.

Associations between PTEN deficiency, immune cell composition and immune checkpoint expression in cancer.

Tumour type	CD8+	Treg	MDSC	TAM	Immune checkpoints	Sample size	Studies in murine models	Studies with human patients
Breast cancer	↑					836 patients	—	53
Breast cancer cell lines					↑ PD-L1	836 patients	—	53
Colorectal cancer					↑ PD-L1	404 patients	—	121
Colorectal cancer			↑			145 patients	—	122
Endometrial	NA					382 patients	—	123
Endometrial				↑		3 mice per group	—	94
Glioblastoma	↑ Ap.					26 patients	—	116
Glioblastoma				↑		66 patients	—	22
Glioblastoma				↑		32 patients	—	98
Glioma					↑ PD-L1	10 cell lines	—	54
Gastric and breast cancer				↑ M2		12 patients	—	124
HNSCC	↓	↑				5 mice per group	88	—
LCNC, SCLC					NA—PD-L1	189 patients	—	125
LSCC					↑ PD-L1	5 mice per group	126	—
LSCC					↓ PD-L1	102 patients	—	127
LUAD					↑ PD-L1	ND	55	—
LUADa				↑ M2		13 mice	95	—
Melanoma	↓	↑				3 mice per group	89	—
Melanoma	↓					135 patients	—	90
Melanomab	↑			↑ M2		4 mice per group	96	—
Melanoma cell lines					↑ PD-L1	33 patients	—	112
Prostate cancer				↑		70 patients	—	74
Prostate cancer		↑			↑ IDO1	91 patients	—	128
Prostate cancer	↑	↑				312 patients	—	87
Prostate cancer			↑			3 mice per group	57	—
Prostate cancerc			↑PMN			4 mice per group	82	—
Prostate cancer			↑			3 mice per group	102	—
Prostate cancer			↑			3 mice per group	100	—
Prostate cancer					NA—PD-L1	20 patients	—	114
Thyroid		↑	↑	↑ M2		8 mice per group	129	—
Uterine leiomyosarcoma					↓ PD1	1 patient	—	107

This is a summary of the literature of the effects of PTEN loss in various tumours based on studies of human cancer and mouse models. The arrows indicate that there is a significant association between PTEN loss, immune cell density and checkpoint expression: ↑ indicates higher density of immune cells or higher expression of immune checkpoints in PTEN-deficient tumours; ↓ indicates that there is lower cell density and expression of immune checkpoints in PTEN-deficient tumours.

LUAD lung adenocarcinoma, SCC squamous cell carcinoma, LSCC lung squamous cell carcinoma, LCNC large-cell neuroendocrine cancer, SCLC small-cell lung cancer, TAM tumour-associated macrophage, NA no significant association observed, ND not described, HNSCC head and neck squamous cell carcinoma. CD8⁺ CD8⁺ T cell, Treg regulatory T cell, MDSC myeloid-derived suppressor cell, TAM tumour-infiltrating macrophage, PD-L1 programmed death ligand 1, PD1 programmed death protein 1, PMN polymorphonuclear MDSC. Ap. apoptosis.

^aPten^D5/D5;Kras^Lox/+;CCSP^Cre/+ mice.

^bBrafV600E;Pten^−/− mice.

^cPten^pc−/−;Zbtb7^pc−/− mice.

A high degree of tumour infiltration by CD8⁺ T cells is strongly linked to a better prognosis in several types of cancer.⁹¹ It is known that immune-cell recruitment and activation relies on tissue-specific mechanisms, which vary across different tumour types⁹² and tissues, and might even present immune-cell contents that are distinct for different lesions from a single patient.⁹³ Thus it is expected that PTEN deficiency in tumour cells could elicit effects on the TME that are tumour-type specific or even tissue specific. Interestingly, several reports have shown that tumours harbouring PTEN deficiency are more likely to present with an increased M2 macrophage cell density.^22,94–96 The presence of M2 macrophages in tumours is linked to favourable growth and invasive tumour features through the synthesis of anti-inflammatory cytokines and the inhibition of antigen presentation and cytolytic cell activation and growth.⁹⁷

Interestingly, PTEN-deficient tumours often exhibit high densities of Treg cells in the TME (Table 2). Treg cells have an immunosuppressive effect in the TME by inactivating the priming and effector activities of CD4⁺ and CD8⁺ T cells.⁵⁹ Moreover, several reports have confirmed the associations between the density of Treg cells and poor outcome in different types of tumour.⁹¹ A recently published report on glioblastoma showed that PTEN deficiency was linked to high macrophage densities.²² Interestingly, the authors of this study found that the density of CD8⁺ T cells increased in PTEN wild-type tumours after PD1 blockade therapy. Conversely, PTEN-deficient tumours did not show a significant increase in CD8⁺ T cell density after therapy. Another new study on the impact of PTEN loss in glioblastoma demonstrated increased infiltration of macrophages via the yes-associated protein 1–lysyl oxidase b1 (LOX-b1)–integrin–PYK2 axis.⁹⁸ The authors showed that LOX expression activated specific pathways in macrophages that promoted their recruitment into the TME where they secrete the growth factor osteopontin (SPP1).

Studies with mice have shown that PTEN deficiency can increase tumour-infiltrating MDSC densities in the TME (Table 2). MDSCs comprise a heterogeneous population of myeloid progenitors and immature granulocytes, macrophages and dendritic cells. Mechanistically, MDSCs are highly immunosuppressive cells that suppress CD8⁺ T cell activity through the secretion of reactive oxygen species, peroxynitrite and nitric oxide (reviewed in ref. ⁹⁹). MDSCs are also thought to sustain tumour growth by protecting proliferating tumour cells from senescence.¹⁰⁰ Owing to the ability of MDSCs to modulate the antitumour immune response, the presence of high densities of MDSCs correlates with a poor outcome in several tumours.^91,101 In a preclinical prostate cancer, PTEN loss results in tumours with high expression of the cytokines colony-stimulating factor 1 and Il1b, which leads to an expansion of infiltrating MDSCs.⁵⁷ The development of castration-resistant prostate cancer—a condition in which tumours are insensitive to androgen-deprivation therapy—was found to be associated with the presence of high densities of tumour-infiltrating MDSCs and high levels of IL-23, which can activate the androgen receptor pathway in prostate tumour cells to promote cell survival and proliferation.¹⁰² As PTEN loss occurs in 40–50% of castration-resistant prostate cancers, it is critical to understand how an immunosuppressive and tumour-tolerant TME is related to increased IL-23 and infiltration of MDSCs in such advanced cancers.

Although PTEN deficiency appears to shape the TME of cancers by mediating the secretion of signalling molecules, one can speculate that these effects could also be an indirect secondary consequence of PI3K–AKT pathway dysregulation. It is known that the PI3K pathway has a role in regulating the antitumour immune response and immune-cell differentiation.¹⁰³ Reports on PI3K blockade indicate the associations between this signalling network and macrophages, CD8⁺ T cells and Treg cells (reviewed in ref. ¹⁰⁴). However, PTEN inactivation is sometimes reported to occur together with downstream PIK3CA mutations.⁸ This finding suggests that some of the somatic mutations of the PTEN gene are redundant but are selected for because of the loss of other regulatory functions that are independent of its role in controlling PI3K activity.

PTEN deficiency and immune checkpoint expression

The most effective current immunotherapeutic interventions, such as PD-L1 inhibitors, are able to restore T cell activity against tumours.¹⁰⁵ Unfortunately, some patients relapse after an initial response from immune-blockade therapies, suggesting that acquired tumour-specific alterations might trigger resistance to immunotherapies.⁵ As discussed above, the accumulation of somatic mutations (and thus high levels of neoantigens) and pathway dysregulation of tumours might lead to an altered immune cell composition in the TME that strongly influences therapy responses.¹⁰⁵

As a result of innate immune resistance, the constitutive activation of oncogenic pathways can promote the synthesis of PD-L1 in tumour cells independently of the immune-cell state in the TME. For instance, in murine models of lung cancer, activation of the AKT–mTOR pathway promoted the expression of PD-L1.⁵⁵ Functional studies with lung cancer cell lines also demonstrated that tumour-infiltrating macrophages might promote PD-L1 synthesis by secreting IFN-γ in a JAK–STAT3-dependent manner.¹⁰⁶ Interestingly, PTEN-deficient melanomas and sarcomas express high levels of VEGFA and STAT3, while also showing resistance to ICIs.^30,107 If PTEN loss occurs, the upregulation of angiogenesis, as mediated by VEGFA and STAT3, may also influence the trafficking of immune cells in the TME.^108,109 It is known that the tumour blood vessels can be modulated by both inflammatory triggers and the types of infiltrating immune cell subsets in the TME. To infiltrate into the tumour and its TME, immune cells must enter the tumour vasculature, adhere to the endothelium and migrate across the blood vessel wall. For these reasons, there is interest in combining anti-angiogenic agents with ICIs to potentially improve immunotherapy responses (reviewed in ref. ¹¹⁰).

Several studies in breast cancer, colorectal cancer and glioma have shown that cancer cell intrinsic PD-L1 expression increases as a result of loss of PTEN. This finding suggests that the context of pre-existing tumour immune states could be informative for more precise use of immunotherapy (Table 2). Through in silico analyses of The Cancer Genome Atlas sarcoma specimens, it was shown that the presence of an oncogenic mutation and a heterozygous deletion in the remaining PTEN allele led to a lower expression of CD8⁺ T cell markers (granzyme A [GZMA], perforin 1 [PRF1], and cluster of differentiation 8A [CD8A]) and PD1 and interferon γ (IFNG) genes.¹⁰⁷ This observation indicates that complete inactivation of PTEN is associated with a reduced CD8⁺ T cell density and activation within the sarcoma TME. In murine lung cancer models, activation of the PI3K–AKT–mTOR cascade resulting from PTEN deficiency was associated with the overexpression of PD-L1.¹¹¹ This finding indicates that PTEN loss and consequent overactivation of the PI3K–AKT–mTOR induces PD-L1 expression. The authors also showed that tumour growth was inhibited, the number of tumour-infiltrating T cells was increased and the density of Treg cells decreased with the combined treatment of mTOR inhibition and anti-PD1 agents.

In PTEN-deficient breast cancer cell lines, Mittendorf and colleagues⁵³ observed overexpression of PD1 and PD-L1, which was downregulated after PI3K pathway inhibition. Moreover, PD-L1 overexpression induced by PTEN loss in tumour cells was associated with low T cell proliferation rates.⁵³ Similarly, the absence of PTEN expression in melanoma cell lines was associated with the upregulation of PD-L1 expression.¹¹² In gliomas, PTEN loss was significantly associated with PD-L1 overexpression.⁵⁴ PTEN mutations have been linked to anti-PD1 therapy resistance in glioblastomas.²² Indeed, PTEN-intact glioblastomas showed a high density of CD3⁺CD8⁺ cells after treatment with PD1 inhibitors, but PTEN-deficient tumours did not exhibit this effect.

In a murine model of castration-resistant prostate cancer, the combination of deletions of Pten, transformation-related protein 53 (Trp53) and Smad4 led to resistance to anti-PD1 and anti-CTLA4 antibodies.¹¹³ Interestingly, PTEN deficiency alone was not associated with PD-L1 expression in prostate cancer. Indeed, PD-L1 expression in radical prostatectomy-derived prostate tumours is rare.¹¹⁴ Collectively, these observations demonstrate that knowledge of the status of PTEN might be useful in deciding which patients and which specific tumour types might benefit from current immunotherapeutic approaches.

Clinical trials investigating the influence of PTEN on the response to immunotherapy

As PTEN functions determine how tumour cells react to the immune response in the TME, this gene might be a useful biomarker for determining how patients will respond to therapy. A Phase 2 trial showed that patients with PTEN-deficient castration-resistant prostate cancer had improved radiographic progression-free survival when treated with a combination of the hormone therapy drug abiraterone and an AKT blocker.¹¹⁵ Although PTEN biomarker testing has been a consideration for standard therapeutic responses, only a limited number of clinical trials plan to determine PTEN status in relation to the clinical response of tumours with ICIs (more details in Table 3). This suggests that PTEN might have been overlooked in trial designs, despite the compelling associations between this pathway and immunotherapy response. However, several downstream markers of the PTEN signalling pathway, such as AKT and PI3K, are being investigated in conjunction with checkpoint blockade therapies.

Table 3.

Clinical trials investigating immune response biomarkers and downstream effectors of PTEN–PI3K–AKT–mTOR pathway using checkpoint blockade therapies.

PTEN-associated mechanism	Tumour type	Drug	Study details	Trial number
PTEN loss and phospho-AKT	Non-small cell lung carcinoma (NSCLC)	AZD6244 (KRAS mutant patients) Erlotinib (wild-type KRAS patients) AZD6244+Erlotinib	Phospho-ERK (p-ERK), phospho-protein kinase B (p-AKT) and PTEN expression will be determined. PD1 expression will be investigated in Tregs and CD8+ T cells	NCT01229150
PTEN loss and AKT	Advanced or metastatic solid tumour malignancies	AZD5363 (AKT blockade) Durvalumab (anti-PD1)	Investigate the links between mutations in Akt/PIK3CA/PTEN pathway and response to AZD5363+Olaparib+Durvalumab. To understand the role of Tregs in improving response to Durvalumab. Determine the role of AZD5363 as an immunomodulator	NCT03772561
Phospho-AKT	Stage I–IV oral and oropharyngeal squamous cell carcinoma	Metformin hydrochloride/pioglitazone hydrochloride extended-release tablet	PD1 and PD-L1 expression will be compared between patients before and after treatment. IHC will be performed with (p)AKT, pAMPK, pS6 and tumour-infiltrating immune cells (CD8, IFNγ, Treg and CD68)	NCT02917629
Phospho-AKT	NSCLC Squamous cell adenocarcinoma	AZD5363 (AKT inhibitor) Durvalumab (anti-PD-L1) Other drugs (ZD4547; Vistusertib; Palbociclib; Crizotinib; Selumetinib; Docetaxel; Osimertinib; Sitravatinib)	Multi-drug and genetic testing in a multi-arm Phase 2 trial. No genomic or expression tests.	NCT02664935
Phospho-AKT	Metastatic breast cancer	MEDI4736 (anti-PD-L1) AZD5363 (AKT inhibitor) Other drugs	DNA will be investigated by NGS and microarray	NCT02299999
Phospho-AKT	NSCLC	MEDI4736 (anti-PD-L1) AZD5363 (AKT inhibitor)	DNA will be investigated by NGS and microarray.	NCT02117167
PI3K inhibition	Unresectable or metastatic microsatellite-stable solid tumour along with microsatellite-stable colon cancer Colon cancer	Copanlisib (PI3K inhibitor) Nivolumab (anti-PD1)	Phase 1/2 study of PI3K inhibition (Copanlisib) and anti-PD1 (Nivolumab) in refractory mismatch-repair proficient (MSS) colorectal tumours. No genomic or expression tests	NCT03711058
PI3K inhibition	Classical Hodgkin lymphoma	Tenalisib Pembrolizumab	Phase 1/2 study to investigate the safety and efficacy of RP6530 (PI3Kδ/γ dual inhibitor) in combination with an anti-PD1 therapy (pembrolizumab). No genomic or expression tests	NCT03471351
PI3K inhibition	Metastatic NSCLC	Abemaciclib	NGS for 245 genes, NanoString nCounter including immune signature and IHC with PD-L1 in patients treated with PI3K inhibitor and PD1/PD-L1 inhibitors	NCT03356587
PI3K inhibition	Advanced solid tumours	Itacitinib Epacadostat INCB050465	JAK inhibitor with JAK1 selectivity (Itacitinib) in combination with an IDO1 inhibitor (epacadostat; INCB024360; Group A) and Itacitinib in combination with a PI3Kδ inhibitor (INCB050465; Group B)	NCT02559492

These clinical trials and their associated biomarker studies may provide more information of the impact of PTEN/PI3K on responses to various drugs and checkpoint inhibitors in different solid tumours.

IHC immunohistochemistry, NGS next-generation sequencing, pAMPK phosphorylated AMP-activated protein kinase, Treg regulatory T cell.

Interestingly, inhibition of components of the PI3K–AKT–mTOR network in PTEN-deficient primary cultures of gliomas led to a decrease in T cell death.¹¹⁶ The effects of restoration of PI3K control suggests that PTEN loss in tumours may be enhancing the adaptive immune response against cancer cells.¹¹⁷ Concordantly, ablation of the PI3K pathway led to a marked enhancement of the antitumour functions of Toll-like receptor ligands in different murine models of cancer.¹¹⁸ The outcome of several ongoing trials investigating genomic alterations and potential links between PTEN loss and response to ICIs (Table 3) could be very informative. The trials (NCT02299999, NCT02117167 and NCT03772561) will involve transcriptome analysis, so it will be possible to determine whether genomic changes such as PTEN loss, combined with other mutations, are involved in the differential response to immunotherapy. In this manner, determining PTEN deficiency in tumours might be useful for future biomarker-guided combination drug trials.

Concluding remarks

Various lines of evidence suggest that loss of PTEN has a crucial role in the development of an immunosuppressive cancer phenotype for some tumours. As restoring the function of PTEN is presently not feasible,¹¹⁹ the most obvious way to counter the effects of PTEN loss would be to inhibit PI3K signalling. However, as PTEN and PTEN-L appear to mediate immune responses independently of PI3K, future therapeutic approaches should target other downstream pathways and signalling molecules that directly control TME immune responses. For instance, blockade of the JAK–STAT3 pathway in PTEN-null prostate cancers induced the TME to become more immunogenic and to be infiltrated by increased numbers of T cells.⁴⁶ In this way, tumours could be primed to respond to anti-PD1 or anti-PD-L1 therapy as the immune cells in the TME can be induced to better recognise tumour cells. Other approaches, such as anti-angiogenic therapies, might also benefit patients before exposure to immunotherapeutics.¹¹ Interestingly, PTEN-deficient glioblastomas overexpressed the CD44 cell-surface adhesion receptor and had a more compact tumour-cell phenotype that could exclude vascularisation and immune cells from the TME than wild-type glioblastomas,²² rendering them less likely to respond to ICI. More studies are required to determine the mechanistic links between PTEN, angiogenesis and the response to anti-PD1/PD-L1. As many patients experience relapse after immunotherapy, it is critical to characterise new tumour-specific genomic biomarkers and the molecular signatures of response that will be informative for immune-based treatments.

Author contributions

All authors contributed equally to this review.

Ethics approval and consent to participate

Not applicable.

Consent to publish

Not applicable.

Data availability

Not applicable.

Competing interests

The authors declare no competing interests.

Funding information

This work was supported by São Paulo Research Foundation (awards #2015/22785-5, #2017/08614-9, and #2015/09111-5) and CNPq (award 306864/2014-2).