1. IntroductionNatural killer (NK) cells are key effectors of the innate immunity, which cooperate with adaptive immunity in the protection from microbial infections, viruses, fungi, and cancer. Innate immune cells have been initially considered unable to identify and eliminate microbes without pre-sensitization; however, several studies clearly demonstrated that innate immunity recognizes microbial-associated or pathogen-associated molecular patterns (PAMPs) through their pattern recognition receptors (PRRs), including Toll-like receptors (TLRs), NOD-like receptors (NLRs), C-type lectin receptors (CLRs), and RIG-I-like receptors (RLRs) [1,2]. Toll receptors were first described in the mid-1990s as essential molecules for embryonic patterning in drosophila that also play a role in antifungal immunity [3]. Subsequently a homologous family of Toll receptors, the so-called TLRs, were found in mammals [4]. TLRs are currently the most well-defined PRRs with respect to PAMP recognition and induction of innate immune response [5].TLRs are type I transmembrane proteins with a N-terminal leucine-rich repeat (LRRs) ectodomains that mediate PAMP recognition, a transmembrane domain, and a C-terminal Toll-interleukin 1 (IL-1) receptor (TIR) domain necessary for signal transduction [6]. TIR domains are characteristic of many adaptor proteins that interact homo-typically with the TIR domains of TLRs and IL-1 receptors as the first step in the signaling cascade. Interestingly, homologs of TIR domains are also present in some plant proteins where they confer resistance to pathogens [7]; this suggests that the TIR domain represents a very ancient motif that served an immune function before the divergence of plants and animals. A typical stretch of around 20 hydrophobic residues composes the TLR transmembrane domain. In particular, endosomal TLRs transmembrane domains include the UNC93B protein that recognizes nucleic acid PAMPs and directs these TLRs to endocytic compartments [8]. The N-terminal ectodomains (ECDs) of TLRs are glycoproteins with 550–800 amino acid residues [9].The TLR family comprises 10 members (TLR1/TLR10) in humans and 12 members in mice (TLR1-TLR9 and TLR11-TLR13), TLR1–TLR9 being conserved in both species [10]. The TLR genes are dispersed throughout the human genome: those encoding TLR1 and TLR6 map to human chromosome 4p14, TLR2 and TLR3 to 4q31.3–q35, TLR4 to 9q32–q33, TLR5 to 1q33.3–q42, TLR7 and TLR8 to Xp22, and TLR9 to 3p21.3.Based on their cellular localizations, human TLRs can be divided in cell surface receptors (TLR1, 2, 4, 5, 6 and 10), which are primarily designated to recognize extracellular macromolecular ligands from bacteria and fungi, and endosomal (TLR3, 7, 8 and 9), recognizing cell ligands that require internalization to generate a signal as microbial DNA or RNA [11]. In particular, the TLR2-TLR1 heterodimer recognizes triacyl lipopeptides from Gram-negative bacteria and mycoplasma, whereas the TLR2-TLR6 heterodimer recognizes diacyl lipopeptides from Gram-positive bacteria and mycoplasma [12,13]. TLR4 responds to lipopolysaccharide, a surface structure of Gram-negative bacteria that can cause septic shock, whereas TLR5 binds the flagellin in bacterial flagella [14,15]. Among endosomal TLRs, TLR3 recognizes double-stranded RNA (dsRNA) produced by a number of replicating viruses [16,17,18]. By contrast, TLR7 and TLR8 recognize single-stranded RNA (ssRNA) derived from RNA of viruses, such as vesicular stomatitis virus, HIV, influenza A and some silencing RNAs [19,20,21]. Lastly, TLR9 recognizes unmethylated 2′-deoxyribo CpG DNA motifs in bacteria and viruses [5].In addition to responding to PAMPs, TLRs respond to danger-associated molecular patterns (DAMPs), also called alarmins, and trigger inflammatory responses. Alarmins are produced as a result of cell death and injury or by tumor cells [22,23]. Thus, TLRs are critical sensors for immunosurveillance against tumors. The tumor microenvironment (TME) is rich in molecules potentially able to activate TLR signaling in local antigen presenting cells (APCs) to improve anti-tumor T cell responses, such as heat shock proteins, high mobility group proteins, DNA from necrotic cells, and hyaluronic acid [24,25]. However, besides their role in inducing anti-tumor response, tumor cells may activate negative regulatory circuits critical for normal homeostasis of the immune system through TLRs by inducing and maintaining immune tolerance to cancer. The most active negative regulators include extracellular decoy receptors (soluble TLRs), transmembrane suppressive receptors, several miRNAs, and intracellular inhibitors [26]. After ligand engagement, TLRs form homodimers or heterodimers and undergo conformational changes to recruit downstream adaptor proteins. Examples are the TLR3 homodimer that recruits the TIR domain-containing adaptor inducing IFN-β (TRIF), the homodimer TLR9 and the heterodimer TLR7/8, both recruiting myeloid differentiation primary response gene 88 (MyD88). Both pathways lead to activation of transcription factors, such as NF-kB, AP1, CREB, IRF3/7, which regulate gene expression of type I and III IFN, inflammatory cytokines/chemokines, costimulatory and adhesion molecules, and antimicrobial mediators [27]. TLR expression and function have been widely studied in APCs, but some reports provided clear evidence that they may be the first-line defense also on NK cells against bacterial, viral, and fungal pathogens. In addition, TLR ligands can activate NK cells directly or indirectly with accessory cells through cytokines or cell-to-cell contact [27,28,29], therefore, they have the potential to stimulate immunological effector function of NK cells for cancer immunotherapy and infectious diseases. In this review, we will discuss the features of endosomal TLRs in human NK cells and the role of their agonists in immunotherapy. 2. TLR Expression in NK CellsControversial observations have been reported on the expression of TLRs (especially of endosomal TLRs) on human NK cells, probably due to no specific detecting antibodies. However, the use of quantitative PCR primers specifically discriminating among TLR family members allowed to detect mRNA generating reliable data about the presence of each TLR in different cell types [11,27,30,31].Data report that TLR family members are expressed in immune cells, but also in a variety of other cells, including vascular endothelial cells, adipocytes, cardiac myocytes, and intestinal epithelial cells. The expression level of TLRs among immune cells is variable [32]. During the last two decades, the group of Alessandro and Lorenzo Moretta highlighted and repeatedly confirmed the presence of all endosomal TLRs on NK cells by different methods (mRNA expression, Western Blot analysis, and confocal microscopy) performed on purified NK cells, NK92 cell lines, or NK cell clones [33,34,35,36,37]. Recently, the presence of endosomal TLRs has been directly demonstrated at the protein level and functionally through the activation of purified NK cells by specific ligands/agonists, particularly that of TLR8 [38]. NK cells express all endosomal TLRs independently of their state of activation at different levels. Despite differences among endosomal TLRs, their expression has been detected both in different donors and in NK cell clones derived from the same individual [39]. As mentioned, TLR3 is overall highly expressed, followed by TLR7 and TLR8 moderately expressed and TLR9 expressed at low or undetectable level [27,30,31].Additionally, studies demonstrate that mRNA expression of different TLRs is modulated upon different conditions. Examples are the overexpression of TLR3 in monocytes of Kaposi’s sarcoma patients [40] or the overexpression of TLR3, 4, 7, and 9 in tumor cells of patients affected by esophageal squamous cell carcinoma [41]. TLRs expression can be altered upon stimuli, such as granulocyte-macrophage colony-stimulating factor (GM-CSF), which induces the downregulation of TLR1, 2 and 4 in human monocytes, or IL12, able to increase the expression of TLR2 and TLR4 in mast cells [42]. Interestingly, it has been demonstrated that engagement of certain TLRs leads to the regulation of other TLRs expression. In particular, stimulation with TLR8 selective ligands induces both the upregulation of TLR2 and the downregulation of TLR7 and TLR9 in monocytes and in macrophages, respectively [32]. Our data demonstrate that mutual regulation of TLR expression exists also in NK cells. Similar to monocytes and macrophages, the main regulator of other TLRs expression is TLR8, whose triggering in NK cells induces the upregulation of TLR7 (unpublished data). Finally, recent evidence indicates that certain TLR functions differ based on the NK cell subset. In particular, our recent study demonstrated that TLR7 and TLR8 are both consistently expressed on freshly isolated CD56bright and CD56dim NK cell subsets, but only TLR8 agonists activate exclusively CD56bright and not CD56dim NK cell subsets [38]. Discovering how TLRs interact each other, regulating their own expression, could be useful in order to trigger TLR sequentially. The increased function and proliferation of NK cells induced in vitro by TLR ligands can be exploited to improve the infiltration of NK cells if they are administered intratumorally. The local therapy with TLR agonists is considered at present as a novel potential strategy to treat solid tumors. Furthermore, these compounds allow easy and quick amplification in vitro of NK cells which can be used as “off the shelf” cells to be adoptively administered in several tumors. 5. Clinical Trials with Endosomal TLR Agonists Stimulating NK Cells At present, 62 clinical trials (active, recruiting, completed, or terminated) on the TLR agonist employment for CIT are registered, and most of them involve the endosomal TLR agonists in combination with other treatments (chemotherapy, radiotherapy, or mAbs treatment)(https://clinicaltrials.gov). The VTX-2337 molecule (USAN: motolimod) has been used in 11 clinical trials against solid tumors, such as ovarian and squamous cell carcinoma of the head and neck (SCCHN). In particular, the studies of Chow L. and Dietsch G. N were focused on VTX-2337 in combination with cetuximab for the treatment of SCCHN patients in two phase II clinical trials (NCT01836029, NCT01334177); in this context, they observed an increased NK cell-mediated ADCC [88,107]. Other studies employed MGN1703 for treatment of colorectal carcinoma in combination with chemotherapy treatment. They evaluated especially the frequencies of NK and NKT cells as positive biomarkers in colorectal carcinoma patients (NCT01208194) [94]. Therapeutic advances in childhood leukemia (TACL) and lymphoma phase I consortium performed a pilot study on three patients with minimal residual disease (MRD) positive acute leukemia (NCT01743807) by using GNKG168 as a TLR9 agonist. They demonstrated that the SIGIRR, IL1RL1, CCR8, IL7R, CD8B, and CD3D genes are downregulated after GNKG168 treatment. In particular, IL7R decrease could promote CD56bright NK cell subset suppression of graft-versus-host disease (GvHD) in acute leukemia patients [95]. Nevertheless, most of clinical trials didn’t achieve the desired therapeutic effect and provoked different adverse reactions. For example, MEDI9197 used as an adjuvant for advanced solid tumor treatment (NCT02556463), induced side effects such as cytokine release syndrome [104]. The description of clinical trials using endosomal TLR agonists in human tumors has been summarized in some complete, recent reviews [11,82,108,109,110]. 6. Novel Approaches of CIT with TLR-Activated NK CellsTargeted drug delivery can significantly influence the efficacy of the molecule in terms of pharmacokinetics and bioavailability [111]. In the past decades, diverse drug-delivery technologies, including nano- and microparticles, co-crystals, and microneedles have been developed to maximize therapeutic efficacy and minimize unwanted side effects of therapeutics. In particular, vaccine development is often accompanied by adjuvants that promote strong T cell responses by prolonging antigen presentation by DCs and by activating NK cells [112]. Aluminum salts are the most widely applied adjuvants in human vaccines since they are safe and well tolerated [113]. The micrometer-sized aluminum aggregates were transformed into nano-sized vaccine carriers by shielding their positive charges with a polyethylene glycol (PEG)-containing polymer [114]. Internalization of these molecules is highly dependent on scavenger receptor A-mediated endocytosis [114]. Polymeric nanoparticles (NPs) and liposomes have been investigated to encapsulate payloads, including drugs, proteins, vectors, and nucleic acids [115]. Moreover, NPs tend to accumulate in tumors and not in normal tissue as a consequence of leaky tumor vasculature and damaged lymphatic drainage [116]. In the last few years, nanomaterials have been designed to boost NK cell CIT. In particular, most NPs were developed to modify the immunosuppressive TME [117] so as to improve the recruitment of NK cells in tumor sites [101,118] and to restore NK cells’ function by increasing their proliferation, cytotoxicity and cytokine production [119]. Many studies assessed NP loaded with TLR agonists, alone or in combination with other molecules [120]. For example, PLGA-NP co-loaded with indocyanine green (ICG) fluorescent dye and R848 (PLGA-ICG-R848) was proposed for treatment of prostate cancer, where it induced an increase of cytotoxic activity of NK cells [80]. Kim, H. et al. used a pH-responsive polymeric NP wherein they encapsulated a TLR7/8 agonist [121]. The treatment with this compound prolonged NK cell activation and improved in vivo cytotoxicity much more than using a soluble agonist. In addition, TLR7/8 agonist-loaded NPs potentiated NK cell mediated ADCC in combination with cetuximab in melanoma model [122]. Other groups used a virus-like particle (VLP) CMP-001 as a ligand for TLR9. The VLP is composed by CpG-A ODN incapsulated in a Qβ bacteriophage capsid that protects the ligand from a rapid degradation. CMP-001 was employed in gastrointestinal and pancreaticobiliary cancer, thus improving NK cell infiltration and activity in vivo [123]. Importantly, CMP-001 can directly trigger TLR9 on NK cells, improving the anti-PD1 mAb treatment in melanoma mouse models [96]. Perry, J. L. et al. utilized nanoparticle replication in nonwetting templates (PRINT) to deliver TLR9 agonists into murine lungs. They observed not only an indirect stimulation of NK cells mediated by DC and macrophages, but also a direct interaction between PRINT-CpG and NK cells, suggesting a direct TLR9 triggering of these cells [124].The employment of NP-based strategies for NK cell immunotherapy starts to be explored, and the majority of recent data are in favor of the combination of endosomal TLR agonists and NPs [82,125,126,127,128]. 7. Conclusions and Future PerspectivesTLRs are expressed on tumor cells and immune cells of the TME. Endosomal TLRs, which recognize exogeneous (pathogens) and endogenous DNA and RNA, are mainly expressed on APC, while their expression on NK cells has been long debated in the past. Recently, we definitively demonstrated that the four endosomal TLRs (TLR3, 7, 8, and 9) were consistently expressed on freshly isolated CD56bright and CD56dim NK cell subsets through different genetic and biochemical techniques. By checking TLR7 and TLR8 for protein expression and endosomal localization, we observed that TLR8, but not TLR7, mostly localizes at the late endosome level. In addition, taking advantage of synthetic specific agonists of human TLRs, we demonstrated that CD56bright, and not the CD56dim NK cell subset, is responsive to TLR8 engagement, while TLR7 is not [38]. Thus, these TLR ligands may be considered as new relevant immunotherapeutic adjuvants to increase treatment efficacy and improve cancer patient outcomes. Indeed, several in vitro and in vivo results suggest that TLR8 agonists deeply modify TME, since they may: i. reverse the suppressive functions of human tumor-derived CD4+ or CD8+αβ T cells, and γδ T cells [85,129,130]; ii. induce the switch from M2 to M1 profile of intra-tumor macrophages [131]; iii. enhance apoptosis of MDSC [132]; iv. prevent T cell senescence/exhaustion [133]; and v. induce in vivo the metabolic control of CD4+ T regulatory cells in the solid tumor TME (ovarian cancer) [134]. Several data on murine models and clinical trials strongly suggest that locally infused TLR7/8 agonists delayed the tumor growth and induced tumor regression [81,107,135,136,137]. Thus, TLR8 agonists can be considered a novel strategy able to promote anti-tumor immunity activating NK cells, which, in turn, prime robust and sustained adaptive immune response against the tumor.

However, many more studies are necessary to better understand the role of TLRs locally administered in cancer TME, taking into account several variables, such as TLR expression on tumor cells, mutagenesis, roles of TLR adaptors, and many other related mechanisms. In addition, it is important to underline that each cell population has different TLR expression and responds differently to TLR agonists.

Whereas TLR ligand may be used in DC vaccines which are widely assayed for CIT, immunotherapy with TLR-activated NK cells is still under investigation. In recent years, nanomedicine has offered new strategies for CIT. One of the advantage of NPs is that they can be designed to various sizes, shapes, and functions. They can be modified to target specific components in TME, molecules on innate immune cell surfaces, or loaded with various drugs or adjuvants, thus achieving targeted delivery and simultaneous delivery of therapeutic agents. Nano- and micro-particle systems bound to TLR agonists allow to use a lesser dose of adjuvants and increase anti-tumor immunity compared to traditional treatment. For instance, the development of biodegradable PLGA-PEG NPs, a delivery vehicle for local, slow and sustained release of Poly (I:C) and R848, are potent candidates to treat solid tumors resistant to first-line therapies [80]. Despite this, modified NPs are already used to target DC. Many questions about their toxicity and immunogenicity still remain open. NPs loaded with TLR agonists could be proposed to selectively trigger NK cells in adoptive settings. As previously reported, the triggering of TLR8 is a good strategy to active CD56bright NK cells [38]. Thus, a novel therapeutic strategy could use NPs loaded with TLR8 agonists as an innovative procedure to activate and expand in vitro circulating or intra-tumoral CD56bright NK cells, which will be later used for adoptive therapy in autologous or allogenic recipients (Figure 1).