As early as 1975, in some experiments carried out in vitro, a phenomenon was noticed: some lymphocytes of an undefined type from the normal mouse spleen selectively fought against Moloney leukemia cells spontaneously (1). In 1979, the same cell functions were also described in healthy humans (2, 3). The phenomenon was described as “natural cytotoxicity,” and the lymphocytes were ultimately named “natural killer” (NK) cells.

Initially, NK cells were believed to act as just an “annoying” background of cytolytic activity in several cell lineages. In 1986, as a result of the discovery of several cell surface markers and the confirmation of natural cytotoxicity, NK cells were determined to be a new lineage of lymphocytes (4, 5). The first 30 years of research into NK cells were focused mainly on descriptions of the functions and the identification of single surface markers. The overall progress into research of NK cells was slow and lagged behind that of most other types of immune cells.

Over the past decade, developments in “-omics”-scale technology, such as analyses of gene expression as well as quantification of proteins and metabolites, have enriched our understanding of the complex biologic processes of NK cells. This understanding includes their phylogeny, developmental programs, plasticity, and immune reactivity for controlling viral infections and malignancy at the molecular level. More importantly, NK cells have recently attracted attention for their therapeutic prospects in cellular immunotherapy due to technical progress that has helped immunologists and clinicians gain a better understanding of NK cells (6).

The English-language neologism “-omics” contains several specific molecular levels, such as “genomics” (the sequence and expression of DNA), “transcriptomics” (DNA transcription into RNA), “proteomics” (RNA translation into proteins), or “metabolomics” (metabolites). These methods generate large data sets, which are often referred to as “-omics” data. The expansion in “-omics” methods is due to mainly tremendous advancements in technology through approaches such as next-generation sequencing (NGS) and mass spectrometry (MS).

In this review, we highlight the “-omics”-scale data that have assisted research of NK cells (Figure 1), including methods that examine their phenotypes, transcriptional signatures, and effector functions in various biologic processes or niches. This approach can help to constantly update the “road map” of gene expression that forms a more comprehensive regulation network and provides new strategies to manipulate NK cells for the treatment of human disease.

Aside from metabolomics, most “-omics” technologies are usually considered to be based on genomics, which arose largely from the deciphering of the complete human genome (7, 8) and mouse genome (9) in the early 2000s, undertakings that marked a new milestone in the life sciences. “Omics” technology presents a panoramic view of the unbiased molecular determinants of NK cells not only from development to an exhaustive process but also for multiple responses to effector function in different environments. To analyze NK cells using different levels of “-omics” approaches, very different biotechnologies are applied in each case.

Thanks to the tireless efforts of immunologists, the study of NK cells has made great progress, and we have a more extensive understanding of NK cells. However, due to their complexity, three major research questions regarding NK cells remain: (i) understanding of NK cells from multiple perspectives (what is a NK cell?); (ii) the origin and development of NK cells (where do NK cells come from?); and whether NK cells can be transformed and applied (where are NK cells going?) (Figure 2).

These problems are complicated, but they can be divided into different levels and solved using a single “-omics” or a combination of multiple “-omics” (“multi-omics”) (Figure 2). “Omics” analyses are based on selecting the “-omics” technology and making the corresponding programs according to the experimental target.

First, in the study of the characteristics of NK cells, molecular profiles (e.g., transcriptome, proteome) were often analyzed comparatively among different cell types (e.g., NK cell subsets, NK cells, and ILCs) by microarray (14), RNA-seq (19), and CyTOF (25) (Table 1). Those studies revealed the unique profiles of gene expression or protein expression of different types of NK cells, but did not detail the key molecular mechanisms or carry out integration of data analyses to identify new regulatory elements. Furthermore, the “multi-omics” analysis of RNA-seq and mass spectrometric can also greatly improve the reliability of data and compensate for the shortage of data repeatability of small samples. But in terms of the project by Rieckmann and colleagues, it contains only the most classic human NK cell subsets: CD56^bright and CD56^dim, although it is the credible and ambitious resource (26). In addition, the diversity of NK cell receptors has been a problem for researchers (54), especially the KIR (human) or Ly49 (mouse) families. “Multi-omics” could be used to study their regulomes by combining ATAC-seq with RNA-seq (23).

Second, innate immunity is a protective mechanism present in many types in plants and animals and even in prokaryotes (5, 55). The phylogeny of NK cells is not well understood. NK-like cells and some important receptor families related to the receptors of NK cells in mammals have been confirmed in bony fishes, amphibians, reptiles, and birds (56–58). In addition, with the rapid development of NGS technology and after mapping of the human and mouse genomes, a great deal of species-level genome sequencing has been completed. These big data sets of genomics could help decipher the phylogeny of NK cells according to the cluster of characteristic genes that contain multiple aspects of NK cells.

Third, in the study of disease-related NK cells, “multi-omics” that contain transcriptome, proteome, and even metabolome is an effective means of research (Figure 2). Paired single-cell analyses by scRNA-seq, RNA-seq, and CyTOF in combination have been used to describe the immune environment in lung cancer tissues and showed that the number of NK cells is severely reduced and impaired during the progression of lung cancer (59). It has been suggested that tumor immunotherapy of NK cells may be effective only in the early stage of lung cancer, but a new therapeutic target or possible methods are lacking.

From NFIL3, the first relatively specific TF (60, 61) reported, to Forkhead box protein O1 (FOXO1), the first negative TF (62) reported, the past decade has seen a sharp increase in research of the transcriptional regulation of NK cell development.

Nuclear factor, interleukin 3 regulated is a crucial regulator for the early development of NK cells and commitment to the NK lineage because Nfil3^−/− mice exhibit impaired production of NK cells at the transition of NK precursor cells to immature NK cells in the bone marrow (60). NFIL3 acts in the positive feedback loop of the IL-15 receptor (CD122) (63) by determining the expression of the downstream TFs Id2 and eomesodermin (EOMES) directly (60, 64, 65). Although several TFs have roles in NK cell development, not only Eomes but also T-bet regulate the development and function of NK cells (66). T-bet is known to be the critical TF of interferon (IFN)-γ production downstream of the IL-12 pathway and drives the development of T-helper 1 cells (67). With regarding to NK cells in the bone marrow, Tbx21^−/− mice can block the production of NK cells at the transition from stage III (CD27⁺CD11b⁺) to stage IV (CD27⁻CD11b⁺) (68). Many target genes of T-bet and EOMES necessary for the appropriate development of NK cells and selective regulation of effector functions have been identified, such as Ifn-γ, Granzyme B, Perforin, Blimp1, and S1p5 (68–70). T-bet and EOMES synergize the transcriptional regulation of cytotoxic factors in NK cells (66). Because T-bet is so important, several recent studies have focused on the negative factors or checkpoints for T-bet. FOXO1 downregulates T-bet expression (62) or mothers against decapentaplegic homolog 3 (SMAD3) downregulates NFIL3 expression (71) to impair the maturation and function of NK cells. Although those studies have used various “-omics” technologies and gene knockout mice, they have not described the entire transcriptional regulatory network of NK cell development due to a lack of research on posttranscriptional regulation.

It is also becoming evident that the development and functions of NK cells are not only regulated by TFs but are also influenced by posttranscriptional regulation through non-coding RNAs (ncRNAs) (72). Recent studies have shown that ncRNAs, including miRNAs, that are short ncRNAs (19–26 nt) and long ncRNAs (>200 nt), are also important for the development and function of NK cells (73, 74). Microarray analyses have been used to screen miRNAs in different NK cells from different tissues and shown that miR-483-3p decreases the cytotoxicity of NK cells due to inhibition of activated signal transducer and activator of transcription 5 by insulin-like growth factor 1 (16). Studies have also shown that miR-362-5p facilitates the function of NK cells by downregulating deubiquitinating enzyme CYLD expression (17). A similar experimental approach was used to analyze long ncRNA differences in NK cells from different tissues, and a novel long ncRNA, lnc-CD56, was identified, which positively regulates CD56 in human NK cells (18).

Although some progress has been made, research in this area is relatively scarce. More importantly, the transcriptional regulation program of NK cell development is a “cat’s cradle” of networks performing at multiple levels. Thus, as with any single-factor analysis, understanding the molecular program of NK cell development completely is challenging. “Multi-omics” can help (i) predict and analyze new regulatory elements and (ii) better understand the molecular mechanisms of transcriptional regulation in NK cell development (Figure 2).

Traditionally, NK cells have been thought to be a homogenous population derived from the bone marrow and which circulate throughout peripheral tissues. In recent years, studies have shown that NK cells constitute various unique subsets with different phenotypes and functions (45, 75, 76).

From the perspective of NK cells in the liver, in 2013, Tian’s group is the first to identify CD49a⁺DX5⁻ NK cells as trNK cells. Through a comprehensive transcriptome obtained via microarray and fluorescence-activated cell sorting analyses (32, 77–80), they suggested that lineages of trNK cells in the liver may be different from cNK cells in the spleen. The discovery of liver trNK cells has rejuvenated scholars and will lead to trNK research in other areas (81). In 2014, Yokoyama’s group showed that CD49a⁺ trNK cells are present not only in the liver but also in the skin and uterine tissue (19). RNA-seq and multiple TF gene deficiencies in mice were used to provide more complete evidence to answer why trNK cells are different to cNK cells, especially in terms of TFs requirements. Their data confirmed the notion that the development of trNK cells in the liver is independent of GATA-3 or NFIL3 but dependent on T-bet (19). After that discovery, it was revealed that a “T-bet⁺ Eomes⁻ CD49a⁺ NK cell subsets” was present in the human liver (82). However, CD49a⁺ NK cells in the human liver may be present in variable quantities.

Cytometry by time of flight can be used to discover and define unique cell populations even if a specific marker for a given subset is not used. By using CyTOF and humanized mice, Yokoyama and colleagues showed that CD49e⁻ is a characteristic marker of trNK cells (25). In addition, several research teams have also reported that the trNK cells observed in the uterus, kidney (83), and salivary glands (84, 85) are different from cNK cells in terms of origin, development, and function using “-omics” analysis. Thanks to progress in two-photon microscopy, the discovery of extramedullary hematopoiesis in the liver, spleen (86), and even lungs (87) has better defined the origin of trNK cells.

Thanks to “-omics” analysis, the emergence of NK cell diversity based on tissue specificity or the production of different cytokines and the recently identified ILCs have led to a new nomenclature that assigns cNK cells into ILC1s (88, 89). Current studies on ILCs are dependent mainly on a mouse model, and technical limitations (e.g., multicolor FCM requires at least eight fluorescence channels; spectral overlap) have hampered adequate characterization of human ILCs. CyTOF provided considerable help to Simoni and colleagues in profiling ILCs from human tissues. Surprisingly, they showed that ILC1s (gating strategy: CD45⁺Lin⁻CD94⁻CD127⁺ CRTH2⁻c-Kit⁻NKp44⁻) were undetectable in human tissues, and an intraepithelial ILC1-like population not restricted to mucosal tissues and which displayed similarity to NK cells was found (90). Bernink et al. showed that NK cells could be distinguished from ILC1s because NK cells highly expressed EOMES, perforin, and granzyme B along with a lack of cell surface expression of CD127 and CD49a (91). A more interesting finding was that NK cells and ILC1s had more closely overlapping gene expression on phenotypes and functional programs (15). Although NK cells have similar functions to ILCs, they may be derived from distinct progenitors and have different requirements for EOMES and T-bet (92). Those studies seem to suggest that these two cell types represent only a subset of the broad NK lineage (93).

The emerging knowledge of the diversity of ILC2s and ILC3s is important (90, 93). The diversity of ILCs is a very complicated and confusing problem. A more optimized “multi-omics” analysis uncovered the veil of the diversity of NK cells and allowed us to better understand how cell diversity affects their functions in different tissues in physiologic and pathologic conditions (Figure 2).

Natural killer cells spontaneously kill cells that are deemed to be “dangerous” to the host, including tumor cells (1) and viruses (94). NK cells have been valuable for fighting against cancer, and researchers are now close to a big breakthrough: NK cells may be able to identify and rapidly kill tumor cells without damaging healthy cells or risking the “storm” of pro-inflammatory cytokines caused by activated T cells (6).

Recently, due to the wider applications of “-omics” analysis, including scRNA-seq and CyTOF, the classifications and descriptions of NK cell subsets have reached a new level (95). Using CyTOF combined with analyses rooted in epidemiology and population genetics, it not only showed that haplotypes with −21M HLA-B rarely encode the KIR ligands Bw4⁺HLA-B and C2⁺HLA-C KIR but also showed that stepwise addition of each KIR ligand associated with NK cells helped to “educate” and recognize the responses of CD94:NKG2A and HLA-E (96, 97). In fact, those findings suggest new ways to dissect the numerous clinical associations with HLA class-I molecules and are important for the clinical application of NK cells (98). Romee and colleagues investigated the potential of memory-like NK cells in cancer therapy. Through CyTOF, SPADE analysis was used in memory-like NK cells pre-activated by IL-12, IL-15, and IL-18. Results showed that these cells were effective against leukemia targets regardless of KIR–KIR ligand interactions (48). In a study by Miller and colleagues in 2005, this new treatment strategy using pre-activation elicited greater progress than direct transfer of NK cells in inducing the remission of AML (99). As an “off-the-shelf” therapy, on 20 March 2017, the US Food and Drug Administration granted a designation of “orphan drug” to the NantKwest Company for activated NK cell therapy for patients diagnosed with malignant Merkel cell carcinoma. We believe that massive “-omics” data will provide more information to immunologists for developing more accurate and effective NK cell therapy for tumor immunotherapy.

“Omics” analysis is also widely used in antiviral studies using NK cells. Memory-like NK cells have been induced in viral-infected mice (100–102), but the formation mechanism of the pool of memory-like NK cells is not clear. Results of transcriptome and DNA methylation analyses have shown that the formation and maintenance of memory-like NK cells is dependent on the epigenetic changes associated with functional changes (103) and antibody-dependent expansion (104). Moreover, there is insufficient evidence for a correlation between the diversity and the function of NK cells. CyTOF has been used to assess changes in NK cell diversity during human immunodeficiency virus (HIV) infection. Results showed that an increase in NK cell diversity could reduce the ability of expansion and degranulation though promotion of the secretion of cytokines, which resulted in an increased risk of HIV infection (24). Recently, Aguilar et al. identified the viral ligand m12 for NK1.1 (105) receptors through protein structure-related big data analysis (106). That study has elicited considerable progress in the study of NK cells and has important implications for immunotherapy.

In addition, human cytomegalovirus (HCMV) infection has been shown to be related to some autoimmune diseases (ADs) (107–109) and regulatory NK cells (110–114). An HCMV-induced autoantibody was identified from AD patients using phage display technology and provided a clear intrinsic connection between reduced numbers of CD56^bright NK cells caused by autoantibodies and AD (115). In addition, during a successful pregnancy, NK cells act as crucial regulatory cells, producing IFN-γ to suppress Th17-mediated inflammation at the maternal–fetal interface (116). However, the regulation of NK cells has backfired in insulin resistance; experimental data show that stimulated NK cells are linked to obesity-induced adipose stress and lead to increased numbers of pro-inflammatory macrophages and exacerbate insulin resistance (117, 118). Irrespective of their use in the treatment of tumors, viral infections, or ADs, NK cells will usher in breakthroughs due to advancements in “-omics” technologies (Figure 2).

Natural killer cells are more complicated than originally thought. Due to technical limitations, for a long time, the study of NK cells lagged behind those of T cells and B cells. Reviewing the timeline of studies of NK cells, breakthroughs have been in parallel with advances in “-omics” technology. Such advances have not only been translated into new powerful tools but have also rejuvenated research into NK cells. “Omics” technology can provide an overwhelming amount of information in one experiment. Massive amounts of information can give immunologists richer clues and more ample data to better answer questions that remain regarding the biology of NK cells and further enhance understanding of NK cells (Figure 2). Moreover, “-omics” technology is a golden opportunity to accelerate the process of exploring the basic research of NK cells and developing NK cell-mediated immunotherapy to combat various diseases.

YZ collated data and wrote the review. XX collected data from online databases. ZT and HW conceived and edited the review.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.