Professional antigen presenting cells (pAPCs) such as macrophages are one of the major sources of IL-12 [38, 39], including in the TME. The effect of IL-12 on group 1 ILCs depends on the subtype and location. During Toxoplasma gondii intestinal infections, ILC1s exhibit a greater response to IL-12 stimulation than NK cells, due to a higher constitutive expression of the IL-12 receptor beta 1 (IL-12Rb1) subunit [11]. IL-12 acts through binding to the IL-12 receptor, resulting in the activation of the janus kinase (JAK) 2-signal transducer and activator of transcription (STAT) 4 signalling pathway [40, 41]. STAT4 promotes the expression of IL-12 target genes, in particular IFNγ resulting in the downstream elicitation of group 1 ILC effector function [42, 43].

It has been shown that ILC1-derived IFNγ reciprocally regulates macrophage activation [22]. Tumour-associated macrophages (TAMs) are found in solid tumours and classically exhibit one of two distinct phenotypes designated proinflammatory (formerly M1 macrophages) and alternatively activated (formerly M2) macrophages, reflecting their alignment towards type 1 and type 2 immunity respectively. The type 1 immune cytokine IFNγ promotes macrophage polarisation towards the anti-tumour proinflammatory IL-12-producing phenotype [44, 45], while stimulation by the type 2 cytokines IL-4 and IL-13 results in polarisation towards the tumour-promoting alternatively activated phenotype [46, 47]. This may result in a positive feedback loop where proinflammatory macrophage-derived IL-12 stimulates ILC1s to produce IFNγ, which in turn reinforces macrophage function (Figure 1). Mechanistically, proinflammatory macrophages have been shown to directly elicit cytotoxicity against tumour cells through releasing nitric oxide (NO) and reactive oxygen species (ROS), and via antibody-dependent cell-mediated cytotoxicity by recognising antibodies bound to tumour cells [48].

In the azoxymethane/dextran sodium sulphate (AOM/DSS) model of colitis-associated cancer (CAC), reverse transcription polymerase chain reaction (RT-PCR) showed there to be reduced IFNγ expression and flow cytometry showed there to be reduced ILC1 levels [22]. Additionally, ILC1 frequency positively correlated with proinflammatory macrophage populations in CRC tissue while negatively associating with alternatively activated macrophages [22]. This polarisation was blocked by the administration of anti-IFNγ antibodies, confirming that IFNγ is the key mediator of this process. These findings were corroborated by independent groups and indicate that ILC1s may control the ratio between proinflammatory and alternatively activated macrophages thereby promoting anti-tumour immunity. The protective effect of proinflammatory macrophages against CRC is further shown by others where stimulation of alternatively activated towards classical proinflammatory TAM polarisation through consuming a ketogenic diet can prevent the progression of CRC [49]. Furthermore, the significance of macrophage subtype ratio has also been demonstrated in lung cancer [50]. In peripheral blood samples from patients with lung adenocarcinoma compared to controls, there was an increase in alternatively activated macrophages while the proportion of proinflammatory macrophages decreased. Critically, the mRNA of both proinflammatory macrophages and importantly ILC1 cytokines were reduced while that of alternatively activated macrophages increased.

IFNγ-based treatment or adoptive lymphocyte transfer of ILC1s may thus provide another way of regulating the anti-tumoural macrophagic response (Figure 1). However, while promising, there are logistical challenges associated with ILC1 adoptive cell transfer, the most pertinent being to generate the sufficient numbers of ILC1s required to achieve clinical efficacy. ILC1s, like most ILC subsets, are relatively few in number compared to adaptive lymphocytes such as T cells, and therefore in vitro expansion of sort-purified ILC1s ex vivo or de novo generation of ILC1s in vitro may be required. In mice, recent studies have repeatedly demonstrated that adoptive transfer of flow cytometry sort-purified ILC1s can achieve efficacy experimentally, suggesting that ILC1s are able to retain their function ex vivo and after transfer, and that sufficient numbers to reach in vivo efficacy can be achieved [51, 52]. Indeed, recent studies have also demonstrated that bona fide human ILC1s can be generated and expanded in vitro from CD34⁺ haematopoietic stem cells (HSCs) through a combination of factors including IL-15 [53], which theoretically will provide sufficient numbers for human adoptive transfer therapies. The relative clinical efficacies of in vitro-generated and ex vivo ILC1s, along with the optimal cell numbers required will need to be established in future clinical trials.

A comprehensive list of ongoing clinical trials involving IL-12 and IFNγ has been summarised recently [54]. Due to the toxicity associated with high levels of systemic IL-12 [55], more advanced methods of cytokine delivery have been developed in an attempt to improve anti-tumour efficacy and reduce associated cytotoxicity. One study utilised genetically engineered myeloid cells to deliver IL-12 to metastatic sites, which reversed immune suppression and reduced both primary and metastatic tumour burden in mice [56]. Other recent developments include engineered cytokines in the form of membrane-anchored IL-12 expressed on T cells made possible through retroviral gene transfer, which allowed targeted delivery of IL-12 to tumours while reducing systemic exposure and toxicity [57]. Similarly, antibody-cytokine fusions containing an antibody component that targets specific tumour antigens have also demonstrated promising efficacy in preclinical studies [58]. Small molecule inhibitors have also been proposed to temporarily inhibit cytokine pathways to limit treatment-associated toxicities [59]. Future strategies targeting this axis may involve concomitant adoptive transfer of ILC1s together with IL-12 administration which may show potentiation in anti-tumour efficacy.

ILC1s have also been shown to exert anti-tumoural properties against haematological cancers such as in AML [20]. Excessive differentiation of leukaemia stem cells (LSCs) into leukaemia progenitor cells results in the propagation of AML [60]. At low frequencies, ILC1s may prevent such differentiation and instead divert LSCs into non-leukemic cells. While high frequencies of ILC1s can directly induce apoptosis of LSCs [20] through ILC1-derived IFNγ in a JAK-STAT or phosphatidylinositol-3-kinase (PI3K)/AKT pathway-dependent manner. Additionally, murine AML ILC1s produced less IFNγ than those from healthy mice, suggesting that the antileukemic function of ILC1s is impaired in AML. Thus, targeting ILC1s as a cell-based source of IFNγ, in addition to their other anti-AML functions, may be a suitable immunotherapy approach for treating AML.

However, recent studies have found that in melanoma and lung cancer, ILC1s may show an exhausted phenotype characterised by programmed cell death protein 1 (PD-1) and TNF-related apoptosis-inducing ligand (TRAIL) expression, similar to exhausted T cells found in many cancer types [61]. Accordingly, despite an increase in the total ILC1 population representing a build-up of cells, there is an overall reduction in ILC1-derived IFNγ secretion, indicating a similar impairment in function akin to exhausted T cells [62]. While this allows solid tumours to escape ILC1-derived IFNγ-mediated killing, there are exciting prospects of utilising checkpoint inhibitors, in particular PD-1 blockers to restore ILC1 function and IFNγ production. Indeed, blocking other inhibitory receptors on ILC1s for example cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), T cell Ig and ITIM domain (TIGIT) and T cell immunoglobulin and mucin domain-containing protein 3 (TIM3) may similarly enhance ILC1 anti-tumour activity [63, 64]. However, it is important to realise that ILC2s also express inhibitory checkpoint receptors such as PD-1, and therefore systematic administration of immune checkpoint inhibitors may simultaneously precipitate pro-tumoural functions of ILC2s particularly in cancer types such as CRC and lung cancer where ILC2s are found to promote tumorigenesis [4, 5, 65, 66]. Indeed, the lack of efficacy of immune checkpoint inhibitors in the majority of CRC patients as seen in clinical trials may potentially be due to concomitant stimulation of pro-tumoural ILC2s offsetting the beneficial activation of anti-tumoural type 1 lymphocytes.

Nevertheless, the revolutionary success of checkpoint inhibitors observed in the clinic to date across many cancer types may in part be due to revitalisation of ILC1s in addition to T cells, and highlights PD-1⁺ ILC1s as potential biomarkers in the future to stratify patient likelihood of response to immune checkpoint inhibitor treatment.

Conversely, ILC1s may promote tumorigenesis depending on the disease stage and models used, typically under the circumstances of chronic inflammation. As alluded to previously, cancer progression is often associated with impaired ILC1 anti-tumoural function. Furthering this, independent groups have found that group 1 ILCs may indirectly promote cancer through inducing inflammation, one of the “hallmarks of cancer”, and predispose to cancer development [92, 93]. Crohn’s disease (CD) is one of the autoimmune inflammatory bowel diseases characterised by inflammation throughout the alimentary tract, and the associated chronic inflammation increases the risk of CRC in CD patients [94]. CD is known to be predominantly caused by a deranged type 1 immune response, and multiple studies have found ILC1s to play critical roles in the pathogenesis of CD driving inflammation and disease progression [25, 95–97]. Accordingly, ILC1 frequency is significantly elevated in the inflamed mucosa of CD patients [75]. Analysis of publicly available human CD scRNAseq databases allowed deciphering of the individual cell types’ contribution to the upregulation of CD-related genes [25]. Strikingly, ILC1s demonstrated the strongest induction of CD-related genes, and the particular genes involved were also highly enriched in both CAC and sporadic CRC suggesting that ILC1s play a critical role in intestinal inflammation and subsequent CRC development. Similarly, others have found inflammatory ILC1s to be increased at the precancerous stage of cutaneous squamous cell carcinoma (cSCC) contributing to tumour development and progression [26].

Mechanistically, ILC1s produce both IFNγ and TNFα. While IFNγ is largely anti-tumoural, TNFα can promote tumorigenesis for example through inducing angiogenesis [98–102]. Therefore, the ratios between the different cytokines ILC1s secrete and the overall tumour inflammatory milieu may contribute to determining their net effect on cancer development and progression. Within group 1 ILCs, NK cell populations exhibited a higher ratio of IFNγ⁺ cells to TNFα⁺ cells compared to ILC1s [36]. This provides an explanation for NK cells having greater anti-tumoural properties than ILC1s outside direct cytotoxicity. The involvement of ILC1s and TNFα in tumour development is further demonstrated through the observation that skin precancerous papilloma-associated ILC1 populations have the greatest ratio of TNFα⁺ cells to IFNγ⁺ cells, resulting in net pro-tumoural effects [26]. The reduction in IFNγ expression correlated with increased expression of inhibitory checkpoint receptors on ILC1s including PD-1 and TIM3 in both mice and human cSCC. Potential immunotherapeutic strategies to overcome this may involve combination treatment of immune checkpoint inhibitors to revitalise ILC1 function, together with cytokines such as IL-12 and IL-15 that stimulate ILC1 IFNγ production.

ILCs show plasticity and are able to polarise towards different subsets depending on the overall cytokine milieu [1, 103, 104]. In viral infections and patients with chronic obstructive pulmonary disease (COPD), ILC2s have been documented to convert to an ILC1-like phenotype characterised by T-bet expression and IFNg production in response to the cytokines IL-1b, IL-12 and IL-18 [105, 106]. Critically, established tumours are able to co-opt ILC plasticity mechanisms to achieve immune evasion (Figure 2). IL-13⁺IFNγ⁺ ILC2s have been identified in patients with CD, suggesting that plasticity of ILC2s towards an IFNγ-producing ILC1-like phenotype may contribute to CAC development [107]. In a study on human hepatocellular carcinoma (HCC), a combination of bulk RNA sequencing, flow cytometry and scRNAseq of ILCs from non-tumour liver margin and tumour core was performed to investigate the role of cytokines in modifying ILC composition and the impact this had on prognosis [108]. The tests identified NK cells in the non-tumour tissue that could transition into tumour ILC1s and NK-like ILC3s when influenced by TGFβ. The non-cytotoxic ILC1s that the NK cells transformed into were unable to control tumour growth or metastasis. In addition to causing plasticity between group 1 ILCs, TGFβ may also cause group 1 ILCs to acquire angiogenic abilities, as has been observed in non-small cell lung cancer (NSCLC) where the CD56⁺CD16^– NK cell subset was associated with increased vascular endothelial growth factor (VEGF) expression [109]. In that study, supernatants from the NSCLC NK cells induced endothelial cell chemotaxis and the formation of capillary-like structures in vitro, facilitating cancer metastasis formation. Thus, by facilitating the secretion of TGFβ in its microenvironment, tumours utilise the plastic nature of ILCs to evade surveillance and destruction [36]. Taken together, the plasticity of group 1 ILCs may complicate the prospects of analysing them for diagnostic purposes or targeting them in immunotherapies. For instance, adoptive transfer of NK cells or activated anti-tumoural ILC1s may show limited effector function upon encountering TGFβ in the TME thereby limiting treatment efficacy. Concomitant blockade of TGFβ may be required and various immunotherapeutic strategies are actively being developed to target the TGFβ signalling pathway [110].

Alternatively, in cases where ILC1s exert anti-tumoural functions, cancers may inhibit ILC1 activity by promoting the conversion of ILC1s into ILC3s and this is associated with a poor prognosis [111]. Analyses of ILC subsets and cytokine expression revealed there to be low percentages of ILC1s and high frequencies of ILC3s in pulmonary squamous cell carcinomas (SqCC) but not in adenocarcinomas. This contrast is likely related to the lack of IL-23 expression by adenocarcinoma cells while, tumour cells from SqCCs produced IL-23 that promoted the conversion of ILC1s to ILC3s which produced IL-17, the latter capable of facilitating the recruitment of myeloid-derived suppressor cells (MDSC) and tumour-associated neutrophils (TANs) to the TME [112, 113] and inhibiting the CD8⁺ T cell response [114]. Similarly, in HCC, it has been shown that IL-23 expression promoted the differentiation of ILC1s into natural cytotoxicity receptor (NCR)-negative ILC3s characterised by a lack of NKp46 expression [114]. In addition to contributing to HCC progression through IL-17 expression, these ILC3s were found to directly regulate T-cell responses through expression of major histocompatibility complex (MHC) class II, which reduced CD8⁺ T-cell proliferation and directly induced apoptosis.

Thus, in order to achieve maximal efficacy with group 1 ILC-based immunotherapies, focussing on stimulating ILC1 effector function alone is unlikely to be sufficient. Strategies to maintain the anti-tumoural group 1 ILC phenotype and prevent plasticity in the TME, for example through concomitant blockade of TGFβ and IL-23, will be equally crucial.