SHP-1: the next checkpoint target for cancer immunotherapy?

Immunotherapy has ushered in a new era in cancer treatment. Both the success of immune checkpoint inhibition strategies, and the limitations, which include non-responsiveness of some patients, as well as toxicity, has led to a search for new checkpoint targets. At the same time, the rise of cell-based immunotherapy, and an improved range of techniques for genetic modification, has expanded the range of possible targets to include intracellular checkpoints such as Src homology 2 domain-containing protein tyrosine phosphatase 1 (SHP-1). In this brief review, the potential of SHP-1 in the context of current immunotherapy strategies will be discussed.

Until the start of the 21st century, all cancer treatment strategies focused on targeting and directly killing cancer cells. However, greater understanding of the regulation of T-lymphocytes in the late 1980s and 1990s led to an entirely new strategy for tumour treatment; exploiting T-cell regulatory molecules to ‘arm’ the immune system in order to clear tumours. The first of these checkpoint inhibitors to reach the clinic was an anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) antibody, ipilimumab, which first demonstrated effectiveness in the treatment of melanoma in 2008 [1,2]. This was closely followed by therapies targeting programmed death receptor-1 (PD-1) [3], the ligand for which, programmed death ligand-1 (PD-L1), is widely expressed by tumour cells [4,5]. These strategies have been recently and comprehensively reviewed elsewhere [6,7], so will not be discussed in further detail here; but their importance in signalling a sea-change in cancer therapy should not be underestimated.

Although checkpoint inhibition seeks to improve the ability of endogenous T-cells to clear tumours, adoptive transfer can take one of two approaches; ex-vivo expansion of a patients' own tumour-infiltrating lymphocytes (TILs) which are then infused back into the patient [8], or generation of T-cells genetically modified to target the tumour, either through introduction of tumour-specific T-cell receptors (TCRs) or chimeric antigen receptors (CARs) [9,10], which replace the antigen recognition domain of a TCR with the epitope binding moiety of an antibody [11]. The former strategy suffers from the same limitation as checkpoint inhibition; it relies upon the existence of endogenous T-cells specific for the tumour. As tumours develop from normal tissue, many of their antigens are recognized as ‘self’, and those that are not are generally poorly immunogenic [12]. Mutations during tumorigenesis give rise to ‘neoantigens’; novel antigens that can be targeted by the immune system [13]. Incidence of neoantigens is associated with improved response to checkpoint inhibitor therapy [14]. Unfortunately, neoantigens are not equally distributed across cancer types [15], meaning that either checkpoint inhibition or adoptive transfer of endogenous TILs is unlikely to offer clinical benefit to patients with low-neoantigen malignancies, which include most haematological malignancies. In contrast, the greatest success to date with CAR-T-cell therapy has been with chronic lymphoid leukaemia, as circulating cancer cells may be targeted by their expression of CD19 [16]. Like any other cell-based therapy, CAR-T-cells are subject to suppression by the tumour microenvironment, and also carry the additional risk of on-target, off-tumour toxicity, including normal B-cells expressing CD19. To address these limitations, researchers are examining all aspects of CAR design, from receptor affinity [17] to adding additional properties to CAR-T-cells, such as cytokine production or release of neutralizing scFvs directed against checkpoint inhibitors in so-called ‘armoured CAR-T-cells’ [18].

SHP-1 [protein tyrosine phosphatase, non-receptor type 6 (PTPN6)] is expressed by all mature haematopoietic lineages and at low levels, in a different isoform, by endothelial cells [19]. There is 95% homology between human and mouse SHP-1, making it amenable for study in pre-clinical mouse models [20]. SHP-1 consists of three domains; the N-terminal Src homology-2 (SH2) domain, the C-terminal SH2 domain, and the C-terminal catalytic protein tyrosine phosphatase (PTP) domain [21]. The N-terminal SH2 domain is auto-inhibitory; binding to the PTP domain until the C-terminal SH2 domain binds to a phosphopeptide ligand, allowing a conformational change and the release of autoinhibition [21]. Maximal phosphatase activity is achieved only when both SH2 domains are engaged [22]. Given this requirement, it is likely that SHP-1 interacts with proteins of the inhibitory-receptor superfamily (IRS) containing immunoreceptor tyrosine-based inhibitory motifs (ITIMs) (I/V/LxYxxL/V) within their cytoplasmic tails [23]. It has been shown that SHP-1 constitutively interacts with ITIM-containing leucocyte-associated immunoglobulin receptor-1 (LAIR-1) [24], what is less clear is whether it directly interacts with PD-1, which also contains a cytoplasmic ITIM domain [25]. Studies in human CD4 T-cells and JURKAT cells have demonstrated co-immunoprecipitation of SHP-1 and PD-1 [26,27], however, a recent study in human CD8 T-cells found that SHP-1 and PD-1 acted independently to inhibit T-cell activation; with PD-1 preferentially inhibiting T-cells with the highest affinity TCRs, while SHP-1-mediated inhibition increased incrementally as TCR affinity increased [28]. Furthermore, only SHP-2 has been demonstrated to interact directly with PD-1 in activated T-cells [29]. CTLA-4 does not contain any ITIMs, but does have cytosolic tyrosines that could represent potential binding sites for SHP-1, however, although other PTPs have been shown to associate with these cytosolic tyrosines, there is no direct evidence for SHP-1 interaction with CTLA-4 [30]. To date, no combinatorial studies of SHP-1 inhibition together with PD-1 or CTLA-4 inhibition have been conducted, however, the studies discussed above, in particular the work by Hebeisen et al. [28], suggest that such combinations are more likely to be synergistic than redundant in their anti-tumour effects.

Other than LAIR-1, little is known for certain about SHP-1 binding partners in T-cells, and there is similar debate regarding its substrates, although zeta-chain associated protein kinase 70 (Zap70) [31], lymphocyte-specific protein tyrosine kinase (Lck) [32], phosphoinositide 3-kinase (PI3K) [33], Vav [34] and TCRζ [35] are all strongly implicated [36] (Figure 1). However, the functional effect of SHP-1, or, rather, its absence, on T-cells is better understood. In the absence of SHP-1, CD8 T-cells form more stable and durable synapses with antigen presenting cells (APCs) [37]. This leads to reduced activation thresholds and increased proliferation [38], which is beneficial for any kind of adoptive transfer strategy for two reasons: firstly, numbers of T-cells available for transfer are often limited, especially where genetic modification is involved; and, secondly it is known that the balance of regulatory T-cells to effector T-cells is key in tumour progression [39], so any modification that can bias towards increased effector T-cells is likely to improve treatment efficacy (Figure 2). It is worth noting that SHP-1 has also been shown to be inhibitory to T regulatory cells [40], and therefore inhibition of SHP-1 in these cells leads to increased suppressor function. As with CD8 T-cells, this effect is attributed to increases in TCR–APC conjugate formation and duration. Specific deletion of SHP-1 in all CD4 T-cells via a floxed Shp1^fl/fl CD4-cre system in mice demonstrated a key role for SHP-1 in negatively regulating the responsiveness of CD4 T-cells to interleukin-4 signalling, and therefore maintenance of a TH1 phenotype [41]. Deletion of SHP-1 in other haematopoietic lineages in mouse models, such as B-cells, neutrophils and dendritic cells, is associated with a variety of pathologies [42–45], although SHP-1^−/− CD 8 T-cells have not been linked to any pathological effects, to date.

In 1965, a spontaneous recessive mutation was observed among the mice in Jackson Laboratories, and was given the name ‘motheaten’ due to the marked skin lesions observed on homozygous animals [46]. Motheaten mice die at 3–4 weeks of age, but in 1985 a new mutant mouse was described that had a mutation in the motheaten locus, but survived up to 9 weeks of age; this mouse was dubbed ‘motheaten viable’ [47]. It was not until 1993 that these mutations were associated with a haematopoietic phosphatase [48,49], later named SHP-1 by consensus. The motheaten mouse suffers a range of pathologies, including myeloid-driven skin lesions, interstitial pneumonitis (usually fatal), and a range of haematological abnormalities; polyclonal activation of B-cells, decreased NK cell activity, haemolytic anaemia, decreased dermal dendritic cells, as well as the previously described hyperproliferative T-cells [50,51]. The short lifespan of these animals and the range of multifactorial immunopathologies make them difficult to use effectively in the study of T-cell function. However, the extent of immune dysregulation in these animals indicates the importance of SHP-1 in the regulation of the immune system, and further suggests that specific targeting of SHP-1 in individual cell populations might be a safer approach in patients, rather than global inhibition, as in anti-CTLA-4 and anti-PD-1 therapies.

A number of strategies to exploit the benefits of SHP-1 abrogation have been attempted to date. In pre-clinical studies, adoptive transfer of SHP-1 knockout T-cells has been shown to be beneficial in a model of leukaemia [52], whereas two phase I clinical trials have been taken place to assess the safety of using systemic treatment with sodium stibogluconate (SSG), a licensed treatment for leishmaniasis that is also an active-site inhibitor of SHP-1 and the related SHP-2, as a cancer therapy [53,54]. A further pre-clinical study has looked at developing a new, orally-bioavailable (SSG must be infused intravenously) SHP-1 inhibitor; a small-molecule, aromatic compound denoted as tyrosine phosphatase inhibitor 1 (TPI-1) by the authors [55]. In this study, TPI-1 was found to be ∼58 times as effective as SSG in vitro, and elicited an anti-tumour effect against 4-day established B16 melanomas in vivo, where SSG failed to have any effect. NSC-87877 is a small molecule competitive inhibitor of SHP-2, which is also inhibitory to SHP-1 [56] and is being explored as an anti-tumour agent, however this is due to its inhibitory effects on dual specificity protein phosphatase 26 (DUSP 26), which is overexpressed in neuroblastoma, rather than as a result of SHP-1 inhibition [57]. Suramin is another anti-parasitic agent that has been found to mediate active-site inhibition of a range of PTPs, and is therefore being investigated as an anti-tumour agent, however, its wide spectrum of target PTPs puts it beyond the scope of this review [58]. Historically, active-site-directed inhibitors of PTPs have been challenging due to the problem of creating cell membrane-permeable yet highly negatively charged compounds, however, recently, a cryptic allosteric inhibition site has been successfully targeted in SHP-2 [59], which represents a new strategy for PTP inhibition that might improve the clinical applicability of PTP inhibition.

In the study by Stromnes et al. [52], the authors used an Lck-driven cre to knockout floxed SHP-1 in mature T-cells. This system was used in preference to the SHP-1^null motheaten mouse, in order to avoid any confounding influence of other aberrantly activated SHP-1^null immune cells [60] on the maturation of the T-cells. In order to mimic clinical adoptive transfer strategies, T-cells were subject to three rounds of in vitro antigen stimulation prior to transfer. Although this system might appear to fail to take advantage of the increased antigen-dependent proliferation of naive SHP-1^null T-cells described by Sathish et al. [37,61], the authors observed increased proliferation of transferred effector T-cells in response to tumour in vivo, reduced apoptosis and improved survival of SHP-1^−/− T-cells, and, ultimately improved clearance of leukaemia. This demonstrates that abrogation of SHP-1 is beneficial in effector T-cells, not just in naive T-cells, and therefore knocking out SHP-1 in in vitro-activated, genetically modified T-cells would still add value to adoptive transfer strategies.

To date, although carried out in cancer patients, clinical trials of small-molecule SHP-1 inhibitors remain restricted to phase I dosing studies, and therefore anti-tumour effects, although measured, were not the primary purpose of the studies. In the event, no clinically measurable anti-tumour effects were observed in either study [53,54]. Although not the purpose, this is disappointing and does bring into question the effectiveness of SSG administration as an anti-cancer strategy. No phase II studies of small-molecule SHP-1 inhibition have been completed. Evaluation of toxicity of SSG was somewhat limited in both studies due to the combination of SSG with interferon and/or chemotherapy, and therefore where severe and/or life threatening adverse effects were observed (in up to 68% of patients), it was difficult to establish which treatment was responsible. Dose-limiting toxicities observed included pancreatitis, bone marrow suppression, fatigue, lipase elevation and gastrointestinal upset. Not observed was the fatal cardiac toxicity seen in 5–7% of leishmaniasis patients treated with SSG [62]. Both studies concluded that SSG treatment was well tolerated.

Interestingly, especially when considering global SHP-1 inhibition with agents such as SSG or TPI-1, SHP-1 expression is altered in a range of malignancies; up-regulated in breast and ovarian cancers [63,64], and gene-silenced in lymphomas, leukaemias and colorectal cancers [65–67].

The disappointing performance of SSG/TPI-1 as an anti-cancer agent in both the pre-clinical and clinical studies described above suggests that the adoptive transfer approach of Stromnes et al. [52] might be the most promising avenue for exploitation of SHP-1 inhibition for tumour therapy. The cytosolic nature of SHP-1, and the difficulty in identifying inhibitors that will not target SHP-2 and other PTPs, means that genetic manipulation would be the best strategy for translational studies. There are currently a range of different techniques available for genetic manipulation that have been utilized in various adoptive transfer and CAR-T-cell approaches. A recent study used zinc finger nucleases via RNA electroporation to knockout PD-1 in TILs on a clinical scale in order to treat metastatic melanoma [68], however, limited success meant only in vitro evaluation of the modified cells was possible. In our own lab, we are currently investigating a zinc finger nuclease approach for ablating SHP-1 in human CD8 T-cells for tumour therapy. In the past, lenti- and retrovirally mediated gene transfer strategies have been popular, but difficulties with transduction of T-cells has led to electroporation of either DNA or RNA becoming the method of choice. CAR-T-cell therapies have optimized a number of genetic modification approaches, including the Sleeping Beauty transposon system [69], clustered regularly interspaced short palindromic repeats (CRISPR) [70] and transcription activator-like effector nucleases (TALEN) [71]. These approaches are reviewed in more detail elsewhere [72,73]. However, the range of clinically applicable gene transfer techniques available today mean that the additional knockout of a molecule like SHP-1 from T-cells already undergoing genetic modification becomes a much more straightforward proposition, making it more likely that the beneficial anti-cancer properties of SHP-1^−/− T-cells can be exploited in the clinic in the near future.

This work was supported by The Wellcome Trust [grant number 094511/Z/10/Z (to H.A.W., J.M., A.A.)]; the Medical Research Council UK [grant number MR/L008742/1 (to H.A.W., S.W., A.A.)]; and S.W. was supported by a Medical Research Council UK PhD studentship.

CAR

chimeric antigen receptor

CTLA-4

cytotoxic T-lymphocyte-associated protein 4, CD152

LAIR-1

leucocyte-associated immunoglobulin receptor-1

Lck

lymphocyte-specific protein tyrosine kinase

PD-1

programmed cell death protein 1, CD279

PTP

protein tyrosine phosphatase

PTPN6

protein tyrosine phosphatase, non-receptor type 6

SH2

Src homology 2 domain

SHP-1

Src homology 2 domain-containing protein tyrosine phosphatase 1, PTPN6

SSG

sodium stibogluconate

TALEN

transcription activator-like effector nucleases

TCR

T-cell receptor

TIL

tumour-infiltrating lymphocyte

Zap70

zeta-chain associated protein kinase 70