Introduction

Cancer is a dynamic and complex disease, and the development, progression and treatment of cancer require the involvement of the entire organism. Tumour formation is the result of the interaction of cancer cells with infiltrating immune cells, stromal cells, blood vessels, the extracellular matrix (ECM), secretory products (cytokines, chemokines, metabolites) and specific environmental conditions (e.g. hypoxia) [77]. NK cells appear to regulate the TME in a dynamic fashion [78].

Notably, the production of prostaglandin E2 (PGE2) in cancer cells can reduce the accumulation of cDCs by impairing NK cell viability and chemokine production, causing immune evasion [71]. Similarly, the production of IL-6 and IL-10 in cancer cells and immune cells in the TME also causes DC dysfunction [79]. Upregulation of CTLA-4 expression on NK cells has a negative impact on DC maturation in human NSCLC [80]. Studies have also shown that DCs may impair NK cell function [81].

In conclusion, the crosstalk between NK cells and DCs is critical for regulating antitumour immunity and may be a promising target for effective antitumour therapy [82]. For example, in an orthotopic liver tumour model, CD47 blockade enhanced antitumour efficacy by triggering the DC- NK cell axis [184]. Although cancer cells can respond to extracellular acidosis and redox reactions by regulating the expression of monocarboxylate transporters (MCT) to adapt to survival and proliferation, exposure to a high lactate environment damages the effector function of NK cells. Preliminary evidence suggests that lactate impairs NK cell function and survival, thereby causing immune evasion and tumour progression [185]. Similarly, the accumulation of lactate in the TME leads to mitochondrial dysfunction and apoptosis of liver-resident NK cells, and the depletion of liver-resident NK cells leads to CRC liver metastases [186]. Tumour-derived lactate can inhibit the cytolytic function of NK cells through direct or indirect pathways involving decreased expression of NK cell perforin, granzyme, and NKp46 and increased abundance of MDSCs [187]. Lactate accumulated in the TME has been considered a potential antitumor target, but this discovery faces significant challenges in translating into clinical therapy [188]. Comprehensive analysis of lactate metabolic signalling and the interaction of lactate with other components in TME, especially NK cells, may be a promising strategy to overcome the limitations of immunotherapy, and more studies are needed.

Innervated niche

The "innervated niche" is an emerging concept that aims to describe a specific biological niche "governed by nerves" that is established by the crosstalk between nerves and tumours. There are four main types of cancer-nerve interaction: electrochemical interactions, paracrine interactions, systemic neural-cancer interactions, and cancer therapy-nervous system interactions [189]. Neurological modulation of the immune system and cancer immunotherapy effects on the nervous system represent different mechanisms of neural-cancer crosstalk.

Environmental enrichment (EE) is an established stress model for studying neurogenesis and brain plasticity. Increasing evidence shows that EE can affect the phenotype and function of NK cells. For example, EE positively regulates the maturation, proliferation, cytotoxicity, and tumour infiltration capacity of NK cells, and mice exposed to EE exhibited an antitumour phenotype [190]. EE exposure selectively upregulates hypothalamic brain-derived neurotrophic factor (BDNF) and other brain mediators, induces sympathetic nervous system (SNS) and hypothalamic–pituitary–adrenal axis (HPA) activation, and directly targets NK cells. Moreover, EE acts on adipose tissue and other endocrine organs, which regulate NK cells by secreting leptin, adiponectin, cytokines, and hormones [191]. These mechanisms contribute to the role of NK cells as key mediators in EE-tumour crosstalk and emphasise the significance of a positive stress response and positive emotions for improving immune function and possible antitumour effects (Fig. 4). In contrast, negative stress responses (forced swimming, abdominal surgery) induce inhibition of NK cell activity, which is a major mediator of tumour progression in stress patterns [192]. In population studies, psychological stressors (such as divorce) have also been associated with reduced NK cell activity and increased risk of cancer [193].

Fig. 4
figure 4

Mechanisms of innervated niche crosstalk with NK cells. Positive stressors enhance the antitumor effects of NK cells via the hypothalamic–pituitary–adrenal axis or the sympathetic nervous system. Likewise, NK cells have positive effects on the nervous system. IFN-γ, Interferon-γ. IL-10/17, Interleukin-10/17. BDNF, Brain-derived neurotrophic factor. Adapted from “Hypothalamic-Pituitary-Organ Axis with Cellular Effect (Layout)”, by BioRender.com (2022). Retrieved from https://app.biorender.com/biorender-templates

Full size image

The effects of NK cells on the nervous system are also worth exploring. Under the recruitment of chemokines (CX3CL1, CCL2, and CXCL10) secreted by central nervous system (CNS) resident cells (microglia, astrocytes, and neurons), circulating NK cells enter the CNS through the blood–brain barrier (BBB) and the choroid plexus, settle in the CNS parenchyma, and become a small proportion of the CNS immune cells [194]. There is growing evidence that NK cells ameliorate a variety of neurodegenerative diseases, including Alzheimer’s, Parkinson’s, and multiple sclerosis. Degradation and internalization of neurotoxic aggregates such as alpha-synuclein (α-syn) protein by NK cells [195], activation of microglia with neuroprotective effects by IFN-γ secreted by NK cells [196], suppression of myelin-reactive Th17 cells [197], and inhibition of inflammatory cells by IL-10 are among the mechanisms involved. NK cells have demonstrated neurotoxic effects via mechanisms such as IFN-γ-mediated cytotoxic T cells, DC recruitment and priming, inflammatory cells promotion via cytokines such as granulocyte–macrophage colony stimulating factor (GM-CSF), and direct release of perforin and granzyme [198]. These findings highlight the elusive dual role of NK cells in CNS homeostasis. It is unclear whether NK cells contribute to the CNS disease pathogenesis or outcome, and this dual role is more likely to be related to individual disease stages and CNS-resident NK cell phenotypes. Overall, the immunomodulatory role of NK cells in mediating cancer-neural crosstalk is critical for cancer progression, and more research is needed to further characterise neural-NK cell-cancer regulatory networks.

Mechanical microenvironment

The TME, as a dynamic and heterogeneous integrated environment, consists mainly of biochemical and mechanical microenvironments. In recent years, researchers have begun to focus on the regulation of the mechanical microenvironment in the TME. Recent studies have shown that the biological behaviour of NK cells in the TME is influenced by mechanical signals, including adhesion, migration, tissue infiltration, and cytotoxic functions [199].

The infiltration of NK cells is accompanied by dramatic changes in the mechanical properties of the stroma and target cells, including cancer cell solid stress, ECM stiffness, topography, and fluid stress [200]. Altered mechanical properties affect the cytotoxic effects of NK cells. It was observed that the softness or hardness of the substrate did not affect the extent of NK cell degranulation; however, exposure to hard substrates enhanced the cytotoxic effect of NK cells, suggesting that mechanosensitivity mediates the cytotoxic effect of NK cells [201]. The decreased rigidity of invasive cells leads to the inability of NK cells to fully respond, which represents a novel mechanism by which cancer cells evade NK cell surveillance [201].

Dynamic actin-mediated mechanical signals are essential to maintain the effector function of NK cells. Evidence suggests that conformational changes in SH2-domain-containing protein tyrosine phosphatase-1 (SHP-1) mediated by actomyosin retrograde flow (ARF) regulate NK cell cytotoxicity. NK cell-activating receptors cause rapid actin flow, which prevents SHP-1 from binding to the actin network, and SHP-1 maintains an inactive conformation and initiates cancer cell death. Conversely, inhibitory receptors slow actin flow, and SHP-1/actin binding activates the SHP-1 conformation, thereby inhibiting NK cell activation [202]. Similarly, actin associated with cytoskeletal conformational rearrangements is actively involved in the formation of lytic IS in NK cells [203]. In conclusion, an understanding of NK cell mechanical signal regulation is necessary for the construction of in vitro mechanical microenvironment models, which are essential for evaluating future NK cell therapies.

Microbial microenvironment

The microbial microenvironment, which has been an underestimated part of the TME for a long time, has now been shown to form a multidimensional tumour-immune-microbe biological network and to be a potential area for improving tumour immunotherapy [204]. Cumulative data suggest that multiple probiotics in the gut enhance NK cell activity and function by promoting the expression of inflammatory cytokines by NK cells [205]. The microbe-mediated immunomodulatory effects of NK cells are also being investigated in the TME.

Microbial regulation of NK cells can alter tumour progression. In a carcinogen-induced in situ mouse model of CRC, colonisation of Helicobacter hepaticus (Hep) resulted in increased infiltration of NK cells and reduced tumour load [206]. Interestingly, Lam et al. found that microbiota could switch the tumorigenic preferences of the microenvironment. A favourable microbiota (intact microbiota, high-fibre diet, responder patients) triggers the production of IFN-I by intratumoral monocytes and regulates macrophage polarisation as well as DC-NK cell crosstalk to form an antitumour microenvironment. In contrast, when the microbiota are adversely disrupted, the monocyte-IFN-I-NK cell-DC cascade is disrupted, and macrophages display the protumour phenotype, creating a protumour microenvironment [207].

The regulation of NK cells by microbiota can affect tumour metastasis. It has been observed that the growth of melanoma cells in bone triggers the proliferation of intestinal NK and Th1 cells and their homing to tumour-bearing bone to inhibit melanoma bone metastasis. This trigger is microbially dependent, and it is weakened by microbiome depletion, thus increasing the progression of bone metastasis [208]. Similarly, Yin et al. found that Fusobacterium nucleatum promotes CRC liver metastasis by increasing the accumulation of MDSCs and Tregs in the liver of a CRC mouse model and reducing the infiltration of NK cells to inhibit the immune niche of the liver [209]. Consistently, the results of Gur et al. showed that Fusobacterium nucleatum adhering to various cancer cells inhibited NK cell cytotoxicity by binding to the NK cell inhibitory receptor TIGIT [210].

Although the beneficial effect of microbiota on NK cell activity still needs further exploration, it represents a new immunomodulatory mechanism.

NK cells as potential participants in tumour metastasis: interaction of NK cells with the metastatic niche

Metastasis is a multistage process that comprises three phases: dissemination, dormancy, and colonization. Metastasis is initiated and maintained by a subpopulation of cancer cells with a stem cell-like phenotype and immune evasion properties, called metastasis-initiating cells (MICs). During propagation and dormancy, MICs are in dynamic balance with host immunity. Once immune surveillance fails, metastasis and organ colonisation will occur [211]. The primary tumour and distant premetastatic sites form a metastasis-friendly microenvironment, termed the premetastatic niche (PMN), which has been recognised as an important factor in immune surveillance failure [212]. NK cells have long been characterised as powerful mediators of cell death that directly eliminate cancer cells and inhibit tumour metastasis. However, recent studies have shown that NK cells can exhibit a prometastatic state [213]. Understanding how PMN promote the prometastatic state of NK cells will aid in the refinement and optimization of NK cell antitumour strategies.

Hematogenous metastasis is the main route of metastasis for most tumours. Primary cancer cells invade and intravasate into new capillaries, shedding circulating tumour cells (CTCs), which travel from the circulation into a new host parenchyma to complete the metastasis of the primary tumour to regional and distal sites. In the circulation, cancer cells must overcome shear stress, oxidative stress, and immune cell attack [211]. PMN remodel the vascular state to aid CTC colonisation during this process. A single CTC in the intravasation cycle is often cleared by NK cells, while CTC clusters are protected from NK cell attack by platelet coats and platelet/neutrophil-associated clusters [214, 215]. This observation underscores the unique role of platelets in protecting cancer cells from NK cell attack, and a significant decrease in NK cell-mediated tumour cell survival in Galphaq (a G protein essential for platelet activation) deficient mice has also been observed [216]. Lo et al. reported the mechanism by which CTC cluster resistance to NK cell immunosurveillance in a follow-up study. The metastatic advantage of CTC clusters is associated with their alteration of cell adhesion and epithelial-mesenchymal properties [217].

Metastatic dormancy is the balance between cancer cell cycle arrest, attempted proliferation, and colonization. Dormant cancer cells adapt to niches and establish complex interactions with immune cells. Massagué summarised that dormant MIC metastasis necessitates not only overcoming systemic immunity of NK cells, resident immunity, and reactive stroma, but also adapting to the phenotype required for dormancy, which includes immune suppression adaptations such as perivascular niches, immune evasive dormancy, TGF-β inhibition, metabolic adaptation, and reactive stroma blockers [211]. It has been observed that MICs can initiate a dormancy programme themselves by autocrine signalling of DKK1 to inhibit WNT signalling and bring MICs into a quiescent state. Quiescent MICs downregulate the expression of NK cell-activating ligands to evade NK cell-mediated clearance [218]. In a latent MICs model, re-entry of latent MIC cells into a proliferative state triggered the expression of NKG2D, and latent MIC cells were cleared by NK cells. Similarly, in a NK cell depletion model, depletion of NK cells to allow progressive growth of metastatic lesions was observed [219]. Interestingly, NK cells can "chew off" a portion of the cell membrane carrying a surface molecule (such as PD-1) of the donor cell membrane from an antigen-presenting cell through trogocytosis, thus entering a dormant state and relieving the anticancer activity [220]. In addition, Correia et al. found in dormant disseminated tumour cells (DTCs) that decreased numbers of NK cells in the liver caused a large proportion of cancer cells to metastasize to the liver and that amplification of NK cells with IL-15 caused the cancer cells to enter a dormant state where metastasis was prevented [221].

Although the interaction between NK cells and the metastatic niche complicates the exploration of how cancer cells in dormancy evade NK cell-mediated cell death, the role of NK cells in maintaining tumour dormancy is gaining attention and represents a promising prospect for antimetastatic drug development.

Current therapeutic strategies for overcoming the TME

CAR-NK

The concept of chimeric antigen receptors (CAR) was pioneered by Gross and his colleagues in 1989. After decades of development, CAR-T cell therapy has made many breakthroughs, and the design of CAR has been gradually optimized. Compared with T cells, NK cells are more advantageous in the development of a CAR engineering platform due to their broad-spectrum antitumour activity, relative safety, broad accessibility, and allogeneic use [7]. Based on the unique advantages of CAR-NK technology, many preclinical and clinical trials have been conducted to evaluate the therapeutic value, safety, and manufacturing process of CAR-NK technology and to actively summarise the shortcomings and challenges of CAR-NK cell therapy.

Various strategies have been developed to enable CAR-NK cell therapy to deregulate the suppressive TME. In recent years, combinations of cytokines have been widely used to stimulate NK cell proliferation and improve persistence, such as the simultaneous expression of IL-15 and other molecules on CAR-NK cells [222]. CAR design targeting TME products (e.g., lactate, adenosine, ROS, etc.) and the hypoxic microenvironment represents an innovative CAR construction concept that may be a new weapon to enhance CAR-NK immunotherapy in solid tumours. For example, a genetic modification strategy that involves terminating the release of immunosuppressive metabolites (e.g., ROS) in solid tumours was used to enhance the resistance and efficacy of CAR-NK cells against solid tumours [223]. A CAR incorporating the oxygen-sensitive structural domain of HIF-1α (HIF-CAR) can generate CAR constructs responsive to hypoxic environments [224]. Depolarized or repolarized immunosuppressive cells such as CAFs, TAMs, and MDSCs have also shown convincing efficacy in preclinical models of CAR-T cells [225,226,227]. It is also an attractive option to enhance CAR-NK immunotherapy in the future.

NK cell engagers

Unlike CAR-NK cell therapy, which expresses CAR on NK cells to mediate the targeted killing of tumour cells, NK cell engagers (NKCEs) consist of a single-chain fragment variable (scFv) of antibodies against NK cell activation receptors (e.g., CD16, NKp30, NKp46, and NKG2D) and one (Bispecific killer engagers, BiKEs) scFv of antibodies against different tumour antigens [228]. This "two-pronged" strategy allows the effective binding of NK cells and cancer cells together, forming IS and conferring specific killing activity to NK cells [229]. The introduction of more scFvs or the insertion of IL-15 as a linker, assembled into tri-or even tetra-specific killer cell engagers (TriKEs and TetraKEs), further enhanced NK cell proliferation and survival [230]. In preclinical studies, NKCEs have been used effectively in CD33+ acute myeloid leukaemia [231] and myelodysplastic syndromes [232], epidermal growth factor receptor (EGFR) in multiple cancers [233], B cell maturation antigen (BCMA) in multiple myeloma [234], etc.

Although most NKCEs are currently in the preclinical stage and their safety and efficacy need further evaluation, it is undeniable that they are emerging as highly promising tumour therapeutic strategies and have the potential to overcome the suppressive TME due to their target specificity [235]. It has been observed that 161,533 TriKE exhibits excellent antitumour activity and promotes NK cell persistence in vivo [236]. Clinical trials applying NKCEs to a variety of hematologic malignancies have demonstrated potential antitumour effects (NCT01221571, NCT01221571) or are recruiting (NCT03214666, NCT04101331).

Immune checkpoint inhibitors

Antitumour immunity of NK cells is a "gambling" process of dynamic balance between NK cells and cancer cells, in which the complex interactions of multiple activating or inhibiting receptors or ligands form the basis of NK cell recognition and activation [237]. Cancer cells exploit NK cells' inhibitory receptors for immune escape, and immune checkpoint inhibitors (ICIs) reactivate NK cells by relieving this inhibition.

A variety of NK cell checkpoint receptors are becoming popular targets for tumour immunotherapy [238] and have demonstrated success in terms of safety, tolerability, and survival [239, 240]. However, the efficacy of immune checkpoints in the TME in patients with solid tumours needs further evaluation, and single checkpoint receptor blockade may not be sufficient to completely rescue NK cells within the TME, which express multiple immune checkpoint ligands [241]. Multiple combination strategies need to be tried to overcome the TME, such as ICIs combined with surgery, chemotherapy, radiotherapy, targeted therapy, and other therapies, which are gradually becoming a popular research direction.

Nanoparticles

Accumulated evidence has shown the great potential of nanoparticles in enhancing NK cell-mediated antitumour immunity, making it possible to unlock numerous innovative approaches to inhibit the microenvironment in the TME. For example, nanoscale liposomes and nanoemulsion system could be used to deliver TGF-β inhibitors and modify TME [132, 242]. Nanoparticle strategies have also been developed to improve the homing and infiltration properties of NK cells, such as liposomes loaded with TUSC2 or nanocomposite microspheres encapsulated with IFN-γ to induce a significant increase in NK cell infiltration [243, 244]. The application of an external magnetic field also helps to guide NK cells bound to magnetic nanoparticles into the tumour, and increase NK cell infiltration [245]. There are also cases of applying nanocarriers to silence NK cell inhibitory signals to release NK cell activity and using multi-targeted nano-junction platforms to promote NK cell recruitment and activation [246, 247].

Small molecules

Small molecules' ability to easily cross cell membranes to access intracellular targets makes them more permeable to the TME, giving them an inherent advantage in overcoming the TME. Zhong et al. reviewed the current application of small molecules in interrupting the hypoxic, acidic, and inflammatory environments in the TME as well as the aberrant ECM network, stating that small molecules can be an attractive strategy for targeting TME to enhance tumour therapy [248].

It has been proven that a variety of small molecules can regulate the antitumour function of NK cells by changing the balance of NK cell activation and inhibition signals or by participating in the expansion, activation, differentiation, and maturation of NK cells [249]. In conclusion, the application of NK cell immunotherapy using small molecules is worthy of anticipation and further research.

Outlooks and directions

In the previous chapters, we discussed the interaction between NK cells and specialised TME, which formed an intricate crosstalk network. Given that targeting TME-NK cell crosstalk represents a promising direction for tumour therapy, a number of questions have been raised.

First, higher-resolution methods are needed to characterise the heterogeneous subsets, spatial distribution, and functional status of NK cells in TME (Fig. 5). Single-cell RNA sequencing (scRNA-seq) makes it possible to characterise specific cell populations and cell subset states, and is a powerful tool for dissecting heterogeneous TME [250, 251]. An unanswered question, however, is why scRNA-seq does not fully map the spatial location of cells in involuted TME. The technology that combines scRNA-seq with spatial transcriptomics (ST) emerges as a solution to this problem [252]. ST can realise the spatial visualisation of transcriptome, and it is worth looking forward to applying it to the TME cell-to-cell crosstalk. To surmount the limitations of single omics analysis, scRNA-seq methods combined with multi-omics (transcriptomics, proteomics, epigenomics, and metabolomics) methods were developed [253]. Integrative analysis of single-cell multi-omics data reveals cellular heterogeneity in the TME across multiple molecular dimensions, providing more novel insights into the identification of specific cell subsets, molecular features, and underlying mechanisms that mediate changes in cellular functional state.

Fig. 5
figure 5

Directions and outlooks for overcoming inhibitory microenvironments. ST, Spatial transcriptomics. Adapted from “Components of Tissue Engineering”, by BioRender.com (2022). Retrieved from https://app.biorender.com/biorender-templates

Full size image

Second, in vitro research models need to be established to simulate and decipher specific interactions between specializsed TME and NK cells (Fig. 5). Microfluidic in vitro models have been shown to be a simple alternative tool for studying cancer-NK cell interactions, with the benefits of real-time monitoring of NK cell infiltration and simulating the direct contact between NK cells and cancer cells [254, 255]. Three-dimensional (3D) organoids and 3D bioprinting have advantages in displaying multicellular structures and complex positions of cells in the TME and have been developed for 3D reconstruction of complex microenvironments [200, 256, 257]. The above 3D bionic system allows for the in vitro reconstruct for the complex microenvironment, which is a dynamic crosstalk with NK cells.

Last but not least, it is necessary to visualise the functional status of NK cells at different clinical stages and under different experimental characteristics (Fig. 5). In the panoramic analysis of TME and the characterization of cell spatial location, traditional protein techniques, including immunohistochemistry and immunofluorescence, have exposed some limitations. Spatial protein techniques, including multiplex immunohistochemistry (mIHC), multiplex immunofluorescence (mIF), cytometry by time of flight (CyTOF), and multiplex digital spatial profiling (DSP), allow for the simultaneous assessment of multiple protein markers and contribute to the acquisition of a more comprehensive microenvironment map [258]. Live or intravital imaging techniques, such as confocal microscopy, two-photon microscopy, and fluorescence-based biosensors, have shown attractive potential in subcellular dynamic tracking [259, 260], allowing 3D monitoring and visualisation of NK cell interactions in the TME.

Conclusion

NK cells naturally recognise self and non-self, and kill cancer cells through multiple lytic pathways, representing a powerful tool for cancer immunotherapy. Recognizing that TME is often an important driver of NK cell dysfunction as well as the immune escape of cancer cells, a deep understanding of the logic of TME-NK cell crosstalk is necessary to propose new therapeutic strategies.

Here, we systematically discuss TME-mediated changes in NK cell functions from different classifications of TME based on multiple perspectives. Despite the progress made, our characterization of the functional state of NK cells in TME is poorly understood or even lacking. It is difficult to define whether NK cells are friends or foes. NK cells cooperate with or fight against other participants in the TME, and together they form a complex TME ecosystem. Just as NK cells can shift from a tumour killer to a pro-tumour metastatic state, a more plausible explanation is that NK cells switch roles at different stages of cancer. Despite representing therapeutic potential, lack of evidence from clinical trials remains the greatest challenge in targeting TME-NK cell crosstalk.