NRP1 expression in human tissues and immune cell subsets mirrors that found in mouse models. The most notable exception is T_reg cells. NRP1 was initially considered a marker for tT_reg (or natural) cells in mice,37 in part due to its apparent coexpression with Helios.78–80 Indeed, the murine Nrp1 gene is a direct target of Foxp3-mediated transcriptional regulation, demonstrated by ectopic expression and chromatin immunoprecipitation experiments.81–83 However, subsequent investigation revealed that NRP1 is not expressed by human peripheral T_reg cells in blood or lymph nodes.84 Instead, healthy donor T_reg cells upregulate NRP1 on in vitro activation,84 indicating that immune processes may regulate NRP1 expression in vivo. Though NRP1 regulation may have species-specific determinants, results discussed below suggest that its impact on T_reg cell phenotype and function remains conserved.

In the context of cancer, T_reg cell expression of NRP1 potentiates immune suppression through at least two parallel pathways: T_reg cell recruitment to the tumor by acting as a coreceptor for VEGF,85 and maintaining tumor-specific T_reg cell stability via Semaphorin-4A (Sema4a) ligation.38 39 Initial analysis of the effects of T cell-restricted Nrp1 deletion in tumors utilized Cd4 ^Cre Nrp1 ^L/L mice. Though bulk Nrp1-deficient T_reg cells retained equal in vitro suppressive function, the proportion and function of intratumoral CD8⁺ T cells was dramatically increased. This led to enhanced tumor-free survival and reduced tumor growth kinetics in both implantable and spontaneous tumor models. As NRP1 is known to act as a VEGFR2 coreceptor in the context of angiogenesis,86 87 it was demonstrated that NRP1⁺ T_reg cells migrated along a VEGF gradient in vitro. Additionally, in vivo reduction in tumor growth in Cd4 ^Cre Nrp1 ^L/L mice could be recapitulated in wildtype mice by administering Vegf ^–/– fibrosarcoma tumors subcutaneously in contrast to VEGF replete tumors. These observations highlight the importance of T_reg cell chemotaxis in response to VEGF as a critical component of T_reg cell-established immune suppression in tumor models.

A complimentary finding from our group revealed that beyond cell trafficking, NRP1 was essential for maintaining intratumoral T_reg cell function and phenotype.38 Disruption of the NRP1 pathway, either by antibody blockade or by T_reg cell-specific genetic deletion (via Foxp3 ^Cre Nrp1 ^L/L), impeded intratumoral T_reg cell suppressive function, thereby restoring antitumor immunity, without permitting off-target peripheral autoimmunity. Compared with their Nrp1 ^+/+ wild-type counterparts, Nrp1 ^–/– intratumoral T_reg cells had decreased Bcl2 and increased active caspase-3, indicating dysregulation of survival pathways. Furthermore, Nrp1 ^–/– intratumoral T_reg cells downregulated activation markers (including Helios, IL-10, ICOS, CD73) and adopted characteristic T helper lineage markers (such as Tbet, CXCR3, IRF-4, and RORγt). Our group showed that NRP1 localized to the immunologic synapse during T cell activation to restrain Akt (aka protein kinase B or PKB) activity through phosphatase and tensin homolog (PTEN), thereby relieving Akt-mediated antagonism on Foxo1/3 to stabilize T_reg cell function. Follow-up work demonstrated that although intratumoral Nrp1-deficient T_reg cells retain Foxp3 expression, their loss of suppressive function is potentiated by adoption of a pro-inflammatory phenotype, marked by increased production of IFNγ.39 Of further interest, IFNγ production by Nrp1 ^–/– T_reg cells initiated dysfunction of neighboring NRP1⁺ T_reg cells, in a process termed infectious fragility. The clinical significance of this finding was demonstrated by the requirement for T_reg cell sensitivity to IFNγ in order to mediate tumor clearance on anti-PD1 immunotherapy in the MC38 tumor model. Indeed, mice harboring T_reg cells that lack the IFNγ receptor were insensitive to anti-PD1. These findings provide substantial clinical rationale for further investigation of NRP1 antagonism as a therapeutic agent.88 89 In fact a recent report detailed how NRP1 therapeutic blockade mirrors the effects of T_reg cell-restricted NRP1 genetic deletion in murine models, both in vitro and in vivo.90

Although T_reg cell expression of NRP1 differs significantly between mice and humans, numerous studies have reported increased NRP1⁺ T_reg cells in patients with cancer.39 90–95 This T_reg cell phenotype was anecdotally reported in peripheral blood of pancreatic adenocarcinoma and liver metastases from colorectal cancer.91 In chronic lymphocytic leukemia, elevated NRP1 on B cells and T_reg cells was observed in patient blood that is decreased following thalidomide treatment, linking antiangiogenic therapies to alleviation of antitumor suppression.92 It has also been reported that there is an enrichment of NRP1⁺ T_reg cells in tumor-draining lymph nodes (TDLN) of patients with cervical cancer, particularly TDLN where tumor metastases are established.93 94 Furthermore, human NRP1⁺ T_reg cells are more functionally suppressive and both FOXP3 and glucocorticoid-induced TNFR-related protein (GITR) expression are increased in NRP1⁺ T_reg cells. This observation matches recent findings from our group. Interestingly, NRP1 antagonism at the VEGF-binding domain reduced human intratumoral T_reg cell suppression by approximately 20% by inducing NRP1 protein internalization.90 For human T_reg cells, NRP1 expression may also reflect exposure to ongoing inflammation as enrichment of NRP1⁺ T_reg cells has also been reported in the synovial fluid of patients with rheumatoid arthritis.96 It is also worth noting that the frequency of NRP1⁺ T_reg cells decreases following therapeutic intervention, both pharmacologic and surgical.91–93 Lastly, our group reported elevated NRP1⁺ T_reg cells in the peripheral blood and tumors of treatment-naive melanoma and patients with head and neck squamous cell carcinoma.39 The prevalence of intratumoral NRP1⁺ T_reg cells negatively correlated with disease-free survival in both cohorts. Together, these findings support the notion that NRP1⁺ T_reg cells are functionally enhanced in cancer. Furthermore, increased NRP1⁺ T_reg cells in peripheral blood was observed in patients with cancer.39 92 This finding is unique among IRs, which tend to have minimal expression in blood samples and may suggest that NRP1 expression on circulating T_reg cells could serve as a pretreatment or on-treatment prognostic biomarker for clinical outcomes.97

In contrast to the constitutive expression of NRP1 on thymically derived murine T_reg cells, its expression on naive CD8⁺ T cells is undetectable (both mouse and human) and is only induced on T cell activation. Transcriptional upregulation of NRP1 by CD8⁺ T cells was first documented in an early molecular and functional profiling of CD8⁺ T cell differentiation using LCMV-specific (P14) TCR transgenic T cells,98 where Nrp1 transcription peaked at the effector CD8⁺ T cells and the effector-to-memory transition phases. Nrp1 upregulation coincided with a group of genes encoding proteins involved in T cell migration and adhesion, such as CCR5, CD44, and p-selectin glycoprotein ligand 1 (PSGL-1). This raises the question of whether NRP1 also modulates CD8⁺ T cell migration, as it does in neuronal or endothelial cells. Consistent with this finding, our group observed upregulation of NRP1 expression (both gene transcription and protein level) on polyclonal intratumoral effector CD8⁺ T cells as well as activated tumor-antigen specific CD8⁺ T cells. Therefore, TCR engagement seems to be necessary to drive NRP1 expression in CD8⁺ T cells, a feature shared by most known T cell coreceptors. However, despite the observed upregulation, the functional role for NRP1 during the early priming of CD8⁺ T cells is unknown.

Some early observations have suggested NRP1 may be an IR-like molecule in CD8⁺ T cells. It was first found highly induced on a subset of immunosuppressive intestinal CD8⁺ T cells (the Foxp3⁺ CD8⁺ T_reg cells), along with molecules known to be associated with CD4⁺ T_reg cells such as PD1 and CD103. These CD8⁺ T_reg cells may contribute to maintaining intestinal homeostasis in vivo by down-modulating effector functions of T cells.99 Consistently, in a later report using Gag-specific (TCR^Gag) CD8⁺ T cells to understand cell intrinsic mechanisms regulating CD8⁺ T cell tolerance versus immunity,100 it was determined that NRP1 was preferentially expressed on tolerant, self-reactive CD8⁺ T cells, mirroring PD1, LAG3 and CTLA4, although NRP1 was dispensable for tolerance. Additional evidence suggested that NRP1 may have a role in T cell dysfunction, a term used to describe T cells that are anergized or exhausted as a result of lacking costimulation or persistent antigen exposure. T cell dysfunction is phenotypically characterized by high IR coexpression and reduced effector marker expression,101 and it was found that NRP1 belongs to a core transcriptional signature of 174 genes shared by all aforementioned T cell dysfunctional states.102

Indeed, a recent report indicated that CD8⁺ T cell NRP1 expression in mice and humans is exclusive to a subset of intratumoral CD8⁺ T cells marked by high expression of PD1, whereas NRP1 is minimally detected on the PD1^neg intratumoral CD8⁺ T cells.40 Compared with the NRP1^–PD1^– and NRP1^–PD1⁺ counterparts, the NRP1⁺PD1^hi cells exhibited higher expression of classical IRs (eg, LAG3, TIM3, TIGIT, 2B4), as well as markers related to cell proliferation (eg, Ki67) and cytotoxicity (eg, Granzyme B). They also express higher levels of exhaustion-associated transcription factors, such as NFATc1, TOX, Blimp1 and IRF4, but decreased levels of genes associated with cell survival (Bcl2) and memory/exhaustion precursor cells (TCF1). This is highly reminiscent of ‘terminally exhausted’ CD8⁺ T cells that have been defined in both chronic viral infection and tumor models.103 Importantly, NRP1 was functionally involved in the terminal exhaustion of intratumoral CD8⁺ T cells, rather than a mere consequence of this dysfunctional state. Specifically, the CD8⁺ T cells recovered from B16F10 tumors treated with a neutralizing anti-NRP1 antibody, expressed higher levels of Perforin and Granzyme B, the key molecules mediating the cytotoxic activity of CD8⁺ T cells, and exhibited enhanced cell killing towards autologous tumor cells ex vivo.40 In the same study, in vivo NRP1 blockade led to reduced tumor growth, which further synergized with PD1 blockade, although such synergy was not observed in terms of enhanced cytotoxic activity ex vivo. The differences between in vivo and ex vivo settings may be due to enhanced tumor recruitment of recently activated CD8⁺ T cells with anti-NRP1 treatment through blockade of Sema3A–NRP1 axis. Importantly, there might also be the contribution from blocking NRP1 on cell types (eg, T_reg cells) other than CD8⁺ T cells under an in vivo anti-NRP1 regimen. Indeed, the latter seems to be supported by our study utilizing a CD8⁺ T cell-specific Nrp1-deficient mouse strain (E8I ^Cre Nrp1 ^L/L) in which primary tumor growth was similar to wild-type counterparts. Nevertheless, with the E8I ^Cre Nrp1 ^L/L mice we also observed clear synergy between genetic ablation of Nrp1 and PD1 blockade in the MC38 colon adenocarcinoma model, further supporting the notion that NRP1 contributes to, although indirectly, the defective CD8⁺ T cell-mediated primary tumor control.104

In fact, the most striking phenotype of the E8I ^Cre Nrp1 ^L/L mice is their improved protection against secondary tumors in a mouse model of postsurgical tumor immunity.104 This observation suggested a primary role for NRP1 in CD8⁺ T cells may be it contributes to the defective CD8⁺ T cell-mediated immunological memory against tumors. Of clinical significance, this model resembles control of disease relapse in patients with cancer. Further investigation revealed that Nrp1 ^–/– intratumoral CD8⁺ T cells were more capable of sustaining a stem cell-like, memory/exhaustion T cell progenitor phenotype, as opposed to the irreversible differentiation of a terminally exhausted T cell phenotype. Consequently, a larger pool of tumor-specific memory CD8⁺ T cells are formed from these Nrp1 ^–/– memory precursors, a scenario which is not reported when blocking any of the other IRs, either by genetic deficiency or by antibody blockade.

In contrast to effector CD8⁺ T cells, the reported expression of NRP1 on memory CD8⁺ T cells varies substantially between studies. During acute viral infection, Nrp1 gene transcription was downregulated (although higher than naive cells) on long-lived memory cells compared with T effectors.98 Such downregulation of NRP1 expression was even more striking on tumor-specific memory CD8⁺ T cells, wherein Nrp1 transcription decreased to baseline level in naïve cells (unpublished data). Contrary to these observations, one report described NRP1 as a surface marker for liver-primed memory CD8⁺ T cells generated through liver sinusoidal endothelial cells cross-presenting antigen under non-inflammatory conditions, while it was functionally dispensable for these memory CD8⁺ T cells.105 The discrepancies between these studies may be explained in part by different in vivo conditions (eg, inflammatory milieu and the type of antigen) under which functional memory is generated. Thus, the expression of NRP1 on memory CD8⁺ T cells (likely the same for other IRs) is dictated by the environmental milieu during memory formation.

In summary, the evolving biology of NRP1 in CD8⁺ T cells contributes to a hypothesis that independent sets of IRs, or immune ‘checkpoints’, seem to exist, which respectively control the effector versus memory functions of intratumoral CD8⁺ T cells. Further mechanistic elucidation of this hypothesis is critical, in the hope of improving the durability of T cell-targeted immunotherapeutics, including immune checkpoint blockade and adoptive T cell transfers.

Clinical investigation of NRP1 has predominantly focused on its contributions to tumor angiogenesis by acting as a coreceptor with VEGFR2, rather than its immune regulatory function through binding semaphorins or other ligands. Initial pharmacokinetic studies of an anti-NRP1 monoclonal antibody (MNRP1685A) demonstrated potent inhibition of the human VEGF pathway with additive efficacy in combination with anti-VEGFA, bevacizumab.106 However, subsequent safety analysis in Phase I studies revealed high levels of adverse events, most notably grade 2 or 3 proteinuria (protein accumulation in the urine) in over 50% of subjects in one study, that ultimately terminated therapeutic investigation of this agent.107–109 A recent investigation of an anti-VEGFR2 agent, ramucirumab, found marked impacts on the T_reg cell compartment, leaving room for speculation about whether NRP1 function and expression contributed to the observations.110

To date, only one anti-NRP1 monoclonal antibody is being clinically assessed for its inhibitory effect of T_reg cell function. ASP1948 (human IgG4, Astellas Pharma Inc) in combination with Nivolumab is under Phase Ib evaluation (NCT03565445) for patients with advanced solid tumors. Results from this trial are expected in 2022 and will be critical to further assess the clinical significance and potential of targeting NRP1. Immune monitoring of T_reg cell and CD8⁺ T cell alterations, even in the patient periphery, could be informative for therapeutic response and mechanism of action.

In addition to this, insight into the potential outcomes of NRP1-directed trials may be gleaned from examining how the semaphorin family, which is the primary group of NRP1 ligands implicated in its immune regulation, is being therapeutically targeted. To this end, although the primary ligand for NRP1 on T cells, Sema4A, has yet to be targeted in the clinic, preclinical evaluation of agents against Sema3A (a known NRP1 ligand) have yielded promising results. Sema3A is considered a vasculature normalizing factor through its interactions with NRPs complexed with plexins. While multiple studies have shown that Sema3A function inhibits tumor cell growth and migration,111 112 one recent study in glioblastoma suggests that neutralization of Sema3A impedes tumor growth in patient-derived xenograft (PDX) models.113 The discrepancies between these findings remains to be reconciled; however, it is possible that differences in tumor vasculature requirements and immune infiltration across tumor types could determine whether Sema3A has pro-tumor or antitumor function. In fact, Sema3A intrinsically regulates T cell activation114 115 and thus therapeutic blockade of this interaction may mediate enhanced antitumor immunity.

Whereas prior small molecule or peptide antagonists of NRP1 primarily targeted the b1 domain interaction with VEGF-A to reduce tumor cell migration and angiogenesis,77 116 117 novel candidates also appear to intrinsically regulate T_reg cell function in vitro.118 This combinatorial action may permit lower dosing regimens to mitigate potential side effects; however, the therapeutic efficacy and safety profile of the lead candidate (EG01377) has yet to be evaluated with preclinical in vivo models. Although human T_reg cells selectively express NRP1 in the context of cancer, making it an attractive tumor-specific immune target, NRP1 is constitutively expressed by subsets of both hematopoietic and non-hematopoietic cells including pDCs and endothelial cells. The impact of EG01377 on the function of these cell types in steady state and disease will further shape its clinical applicability.

A growing body of literature demonstrates that NRP1 is a unique immune modulator in cancer immunotherapy (figure 3). In the TME, NRP1 intrinsically regulates both T_reg cell and CD8⁺ T cell function to collectively impede antitumor immunity and is expressed concurrently with multiple IRs. NRP1 benefits the function of intratumoral T_reg cells by both facilitating their recruitment to the tumor bed and enforcing their functional stability amidst ongoing inflammation. Patients with cancer have a higher abundance of NRP1⁺ T_reg cells across malignancies and therapeutic intervention is associated with decreased NRP1 expression in peripheral T_reg cells. Though of relatively minor consequence to CD8⁺ T cell effector function, the impact of NRP1 on memory formation and durable response is distinct compared with the function of other IRs. These novel characteristics may translate into non-overlapping clinical efficacy with existing standard of care checkpoint inhibitors, thereby providing informed rationale for therapeutic combinations.

As the opportunity rises to improve NRP1-targeted cancer therapy by focusing on its immunoregulatory roles, key questions remain regarding fundamental NRP1 biology and translational application.

In conclusion, the translation of NRP1 biology into viable therapeutic interventions for patients with cancer holds substantial future promise, particularly in combination with immunotherapies that do not directly target T_reg cell function or CD8⁺ T cell memory formation. However, a more holistic view of NRP1 in the complete TME, namely tumor, stroma and immune cells, is necessary to design the most optimal clinical strategies surrounding this promising next generation immune modulator.