Introduction

Innate immunity is presumed to be the evolutionary descendant of ancient invertebrate defense mechanisms against foreign pathogens [1]. In vertebrates, it is the first line of defense, before adaptive immune responses, the other main immune strategy, have time to establish. Dr. Charles Janeway initially proposed that conserved portions of pathogen invaders, pathogen-associated molecular patterns (PAMPs), are first sensed by pattern recognition receptors (PRRs) on the host cells, resulting in activation of the innate and adaptive immune responses [2]. Later, Dr. Polly Matzinger proposed that the host is much more concerned with sensing “danger” or “damage” signals, and does so by recognizing cell-derived damage-associated molecular patterns (DAMPs) generated during cellular damage [3]. This concept is particularly relevant to organ transplantation, as transplantation necessitates a period of tissue hypoxia when the donor organ is procured, transported, and before it is re-implanted in the recipient. This period leads to ischemia-reperfusion injuries (IRI) characterized by a dysregulation of cellular metabolism and subsequent production of DAMPs (e.g., heat-shock proteins, uric acid, ATP, DNA fragments) from the donor organ. More recently, recognition of the “non-self” major histocompatibility complex class I (MHC I) molecules by several receptors on innate cells such as macrophages and NK cells has also been shown to contribute to transplant immunity. These observations point to the possibilities of alloantigen-specificity and immune memory of innate immune responses previously thought to be largely non-antigen-specific and naïve. Like adaptive immune responses, these alloantigen-specific, memory innate immune responses are also expected to raise the immune barrier of transplantation.

While it is not possible to fully detail all pathways by which the innate immune system contributes to alloimmunity, it is the goal of this review to provide a comprehensive overview of the current understanding of how the innate immune system may respond to a transplanted organ. Specifically, we will describe known signals by which the innate immune system is activated following transplantation; how it develops “memory” through training; and lastly, the “effectors” of the innate immune system (cell populations and factors) implicated in alloimmunity.

Naive Innate Immune Response in Transplant

One of the most well-researched receptors for PAMPs and DAMPs is the Toll-like receptor (TLR) family members. TLRs are PRRs expressed on a wide variety of cell types including macrophages (Mϕs), dendritic cells (DCs), and epithelial cells; and activation of TLRs by inflammatory products results in intracellular recruitment of the myeloid differentiation primary response protein 88 (MyD88) in most cases. While TLRs may function through many pathways, the ultimate end-product is cytokine and chemokine production, activation of DCs and other innate immune cells, and downstream activation of the adaptive immune system [4,5,6]. MyD88 protein deficiency in a mouse model of heart and skin transplantation results in prolonged graft survival and acceptance of MHC-minor mismatched allografts [7, 8]. On the other hand, TLR activation and signaling leads to failure of tolerance induction [8,9,10].

Although IRI occurs in many situations (e.g., trauma, syngeneic transplant), DAMPs alone are not sufficient to initiate alloimmunity. There are other pathways of allo-recognition that host innate cells utilize when encountering donor tissue - the most well-known is the MHC molecules [4]. The MHC molecules have been initially studied in the context of adaptive immunity, but multiple groups have recently shown that in a T and B cell deficient RAG-/- mouse background, allogeneic tissues elicit an immune response not seen by syngeneic tissues [11,12,13]. Additionally, this host innate response to allogeneic tissues may persist for weeks after transplantation. Specifically, Oberbarnscheidt et al demonstrated that compared to syngeneic grafts, in allogeneic grafts, monocyte-derived dendritic cells (Mo-DCs) are more prevalent, more mature, and produce more IL-12 for weeks after transplantation [13].

A potential underlying mechanism of this innate cell response is the class I MHC – paired immunoglobulin-like receptor (PIR) axis. The polymorphic PIR proteins are found on mast cells, B cells, myeloid cells, and DCs in mice and capable of recognizing MHC class I molecules (MHC-I). PIR-B protein activation results in transmission of an inhibitory signal while PIR-A activation transmits an excitatory one [14]. Activation of PIR-A with MHC-I in a mouse model of allogeneic tissue transplant led to activation and proliferation of recipient DCs with an increased interferon-gamma production [15]. Self MHC-I also interacts with the killer cell immunoglobulin-like receptor (KIR) expressed by natural killer (NK) cells. This self-recognition induces NK cell tolerance to the target cell [16]. On the other hand, lack of such self-recognition, referred to as the “missing self” signal, often resulting from downregulation of MHC-I on some tumors or viral-infected cells, can trigger NK activation to kill the MHC-I deficient cells [17]. In the event the donor allogeneic tissue MHC-I expression does not match that of the host, an imbalance of NK cell stimulation versus inhibition can occur, resulting in NK cell activation and donor tissue injury [18].

Non-MHC allo-determinants implicated in innate immunity include the CD47 – signal regulatory protein α (SIRPα) pathway [19,20,21]. SIRPα is a member of a glycoprotein family expressed on neurons, myeloid, myocyte, epithelial, and endothelial cells and acts as an inhibitory receptor for its protein ligand CD47, which is also found on numerous cell types and is thought to distinguish live from dying cells. Binding of CD47 on the allogeneic cell inhibits the phagocytotic process by the SIRPα-expressing Mϕs. Simply put, the cognate binding of CD47-SIRPα results in a signal for the SIRPα-expressing Mϕs to NOT phagocytose (i.e., “eat”) the CD47-expressing allogeneic cell. Therefore, cognate SIRPα-CD47 interaction has been colloquially referred to as the “don’t eat me” signal [22]. However, amino acid polymorphisms in the SIRPα molecule, specifically in the portion that recognizes CD47, can affect the binding affinity of CD47 to SIRPα. It was shown by Dai et al. that transplantation of an allograft expressing a mismatched SIRPα variant compared to the recipient results in dysregulation of the SIRPα-CD47 binding kinetics, recipient monocyte activation, and differentiation of monocytes into DCs [23]. Furthermore, Pengam et al demonstrated that disruption of the SIRPα-CD47 axis on SIRPα-expressing myeloid-derived suppressor cells (MDSCs) leads to loss of tolerance and graft rejection. This process is thought to be mediated through a loss of MSDCs and an increase in Mϕ-recruiting chemokines promoting the inflammatory M1 Mϕ pathway [24].

Memory Innate Immune Response in Transplant

While it is commonly accepted that innate cells serve as one of the first lines of immunological defense, there is a growing appreciation about the plethora of options by which these cells recognize danger or invaders. Additionally, the field is beginning to understand that while initial exposure to allogeneic cells is sufficient to stimulate innate immune cells, repeated exposures can elicit a stronger and more rapid response. This response is reminiscent of the “immunological memory” that canonically characterizes the adaptive system. Memory is typically long lasting, sometimes years, and is antigen-specific; therefore stimulation with a third-party antigen will not elicit the same response. A growing pool of evidence suggests that the innate system possesses a similar system which has been recently termed “trained immunity” [25]. The trained immunity phenomenon was first identified in invertebrates in 2003, in a species of crustaceans found to develop immunologic memory after infection with a parasitic tapeworm, preventing reinfection on a second exposure [26]. While the crustacean host response was not permanent and faded with time, it was specific and more powerful with repeated parasitic exposures. A similar phenomenon of innate immune memory has been recently described in transplantation [27]. Mechanisms of trained immunity range from epigenetic reprogramming to clonal expansion and differentiation. Intrinsically, trained immune cells experience both epigenetic histone and metabolic changes. From an epigenetic perspective, changes in histone methylation or acetylation can lead to the exposure of DNA promoter region of pro-inflammatory cytokines that become more readily transcribable [28, 29]. These epigenetic signatures result in a rapid upregulation of IL-6, IL-1β, and TNF to protect the host against re-infection [30]. Trained innate immune cells can also alter their metabolic pathways to better meet the energy demands of an activated state. This occurs through an upregulation of the aerobic glycolysis and oxidative phosphorylation pathways exhibited by trained versus untrained Mϕs [31]. Implicated in this metabolic shift is the dectin-1-AKT-hypoxia inducible factor 1α (HIF-1α) pathway that leads to activation of the mammalian target of rapamycin (mTOR). Thus, inhibition of the mTOR pathway with rapamycin or metformin interferes with Mϕ training, resulting in inhibition of innate immune memory [28].

In addition to epigenetic and metabolic imprints, there is increasing evidence suggesting that repeated alloantigen exposure can also induce specific innate immune memory formation through signaling pathways discussed previously in this review. Dai et al. demonstrated that mice which were first immunized with allogeneic splenocytes and later underwent a same-donor allogeneic bone marrow plug graft, exhibited significantly greater infiltration of recipient monocyte-derived dendritic cells (Mo-DCs) than unimmunized mice. This pattern did not occur when the immunized mice were exposed to third-party allogeneic splenocytes. One important note is the temporality of the innate system training—the effect of immunization was lost at day 49. The group subsequently found that this effect was lost by either blockade or complete disruption of the PIR-A proteins, suggesting that this pathway could be a potential therapeutic target for inhibiting innate immune memory [32].

Innate Cell Populations and Soluble Mediators Implicated in Transplant Immunity

Connecting the above-reviewed pathways to cellular effectors that determine transplant clinical outcomes, the remaining of this review will summarize the current state of knowledge of specific innate cell populations and soluble mediators implicated.

Macrophages

Mϕs have long been known to constitute a main immune cell population in the transplanted allograft during episodes of acute rejection [33]; however, their precise roles in alloimmunity and allograft health are largely unknown. This is most likely due to the enormous plasticity of these cells that confers them with the ability to adapt to and assume different functionality cued by their resident microenvironment. Adding to this complexity, the transplanted organ typically carries with it donor resident Mϕs which, upon vascular anastomosis, interact with host circulating precursors of Mϕs, and play a critical role in their trafficking and retention in the allograft, and in their subsequent differentiation to Mϕs [34].

Not surprisingly, on the timeline of a solid organ transplant, Mϕs appear to play different roles at different stages: 1. peri-transplant stage in which donor Mϕs promote recipient monocyte graft infiltration and graft retention; 2. early post-transplant stage in which recipient graft-infiltrating monocytes differentiate to graft resident Mϕs; 3. late post-transplant stage in which graft resident Mϕs may exit homeostasis in accordance with environmental disturbances and subsequently participate in allograft inflammation, repair, and/or fibrosis. Better understanding of mechanisms of Mϕ function at each stage would provide more precise therapeutic targets with fewer side effects.

Specifically, during the peri-transplant stage, donor tissue-resident Mϕs play a critical role in recruiting recipient circulating Mϕ precursors to the graft in a CCL8-CCR8 dependent manner [34]. CCL8 is an inducible chemoattractant produced by donor Mϕs upon transplantation. The signals for its induction are less clear but are at least in part driven by IFN-γ produced upon innate immune recognition of alloantigens upon transplantation [13, 35]. Consequently, removal of donor Mϕs prior to transplantation or blocking CCL8-CCR8 signaling immediately following anastomosis can both significantly reduce recipient myeloid cell infiltration and consequently promote better allograft function in short-term. Next, once infiltrated to the allograft from the circulation, recipient monocytes take residence and quickly differentiate to inflammatory Mϕs in an Axl-dependent manner [36, 37]. Axl is one of the TAM (standing for Tyro3, Axl and MerTK) receptor tyrosine kinase family members. It is conventionally known to be responsible for non-inflammatory clearance of apoptotic cells (a process known as “efferocytosis”). However, its role in monocyte to Mϕ differentiation is unknown with respect to the ligand(s) it is sensing (or the lack of) and/or its downstream signaling requirements. Despite the many unknowns, inhibiting the tyrosine kinase function of the Axl receptor with a highly specific Axl inhibitor results in an inhibition of recipient monocyte to Mϕ differentiation, and prolongs allograft survival [38]. Lastly, once residing in the allograft as tissue resident Mϕs, these cells may respond to allograft injuries and exit homeostasis, for instance after ischemia, to promote inflammation, its resolution, and tissue repair. However, such responses could range from adaptive to maladaptive, achieving different end results as they pertain to long-term allograft health. One such factor in determining resident Mϕ responses to ischemic injury is Allograft Inflammatory Factor-1 (AIF-1), an intracellular calcium binding protein, whose function directly or indirectly alter Mϕ transcription factors and inflammatory cytokine production [39, 40]. Importantly, conventional M1 versus M2 subtyping of Mϕs represents their pro-inflammatory versus pro-repair/pro-fibrotic functional phenotype, which are likely the two extremes of a continuum. Unilaterally targeting one aspect of macrophage function may promote the other aspect of their function. For instance, inhibiting Mϕ pro-inflammatory function often promotes their pro-fibrotic function, which may ultimately result in compromised long-term allograft function even though it mitigates immediate allograft injury. Interestingly, AIF-1 appears to control the association of these two aspects of Mϕs, such that genetic ablation of AIF-1 [41] results in a significantly more desirable Mϕ phenotype that is both anti-inflammatory and anti-fibrotic. Therefore, targeting AIF-1 is a promising strategy to guide Mϕ responses to ischemic injuries during homeostasis towards an anti-inflammatory and anti-fibrotic one to benefit long-term allograft health and function.

NK Cells

The notion that NK cells are not implicated in allograft rejection has been recently challenged, most poignantly in their roles in antibody-mediated rejection (AMR) [42]. Transcriptionally, a selective enrichment of NK cell signature has been demonstrated in human kidney allograft biopsies during acute AMR [43]. At a cellular level, NK cells have been found to localize to the proximity of endothelial cells in the vasculature and contribute to vascular injuries. How NK cells mechanistically contribute to allograft endothelial damage has been recently examined. Specifically, NK cells are the only immune cell population that abundantly express the Fcγ receptor 3A (FcγR3A) which binds to immunoglobulin subtype G1 and G3 in an identical manner as complements [44, 45]. In the context of transplantation, this affinity confers NK cells with the ability to bind to donor-specific antibody (DSA)-opsonized graft endothelial cells. This binding and subsequently signaling lead to activation of NK cells resulting in: (1) NK cell production of cytotoxic molecules capable of direct damage of endothelial cells; (2) production of chemokines by activated NK cells and damaged endothelial cells to recruit inflammatory monocytes; (3) production of chemokines by recruited monocytes and damaged endothelial cells to further recruit NK cells. These interactions form a positive feedback loop fueling inflammation around vascular endothelial cells, ultimately leading to pathological features characteristic of AMR [45].

There are two other mechanisms for NK cell activation that originate from the cell’s ability to recognize the presence of, of lack of, class I MHC via a set of KIRs. The first mechanism is the original observation that NK cells recognize “missing self” (or absence of class I MHC). In this pathway, NK cell interaction with a target cell expressing appropriate class I MHC results in an inhibition of target cell cytolysis. This requires the target cell to express a class I MHC molecule that can engage the NK cell inhibitory receptor; if this class I molecule is lost through infection or malignant transformation, the inhibitory signal is not delivered, and the target cell is lysed.

In addition to “missing self” recognition, NK cell KIRs are also capable of recognition of “non-self” [46]. For example, the NK cell KIR3D receptor is capable of distinguishing between the class I MHC Bw4-Bw6 epitope based on a single amino acid substitution at position 80 [47]. A distinguishing point in this second pathway is that the NK cell for target cell lysis is an activating one, rather than inhibitory [48]. This mode of NK cell activation is highly relevant to transplantation as a complete mismatch of MHC-I molecules between the donor and the recipient is rather common, but amino acid or allelic differences are likely [49] Ultimately both pathways result in processes like FcγR3A-mediated NK-cell activation: degranulation, endothelial damage, chemokine release, and further immune cell recruitment. It is important to note that these two modes of NK cell activation are not mutually exclusive; on the contrary, they may act synergistically to maximally augment NK cell activation and graft endothelial damage [50].

γδ T Cells

γδ T cells differs from conventional αβ T cells in the composition of their T cell receptors (TCR). Typically, αβ TCR T cells respond to antigens in an MHC-dependent manner, whereas γδ TCR T cells respond largely in an MHC-independent manner. The antigens to which the γδ TCRs recognize are largely unknown with a few exceptions which include phosphoantigens, CD1 family, the stress-inducible MHC class I chain-related proteins MIC-A/B, annexin A2, UL16 binding proteins (ULBPs) and endothelial protein C receptor. Once activated, γδ T cells execute effector function in a number of different ways: (1) they can exert direct cytotoxicity by perforin- and granzyme-dependent mechanisms; (2) they can be induced to produce pro-inflammatory cytokines [51] such as IL-17, IFN-γ, and TNF-α, as well as chemokines that promote further chemotaxis of inflammatory cells; (3) they are also known to promote maturation of several cell types such as DCs to become mature antigen-presenting cells [52], leading to downstream αβ T cell priming and possibly antibody production. Relevant to transplantation, during IRI, release of stress-induced molecules activates γδ T cells and promotes their effector functions, including cytokine production and inflammatory immune cell chemotaxis. However, most data supporting this view come from animal models of transplantation. In humans, rejection of the kidney [53], heart [54], and liver allografts has been associated with an enrichment of γδ T cells in the graft and the periphery, although the exact role these cells in mediating the allograft rejection is not clear.

It is important to note that γδ T cells, like NK cells, also express cell surface FcγR3A (CD16). This expression confers the γδ T cells with the ability to interact with antibody-antigen complexes. In the context of transplantation, this means that γδ T cells may interact with DSAs binding to donor organ cells (e.g., endothelial cells in the donor organ vasculature), become activated and execute several effector functions, ultimately contributing to AMR [55]. This pathway is particularly prominent in recipients with cytomegalovirus (CMV) infection, as CMV infection is known to induce a significant upregulation of CD16 on γδ T cells [56].

Confounding investigations of roles of γδ T cells in transplantation are several unique characteristics of this unconventional T cell population: (1) numerous subsets of these cells based on different combinations of the γ and δ subunits of the TCR likely respond to different yet-to-be-identified ligands; (2) the specific effector function of these cells differs significantly based on their local tissue environment. Therefore, naïve versus memory response of this innate immune population in transplant immunity remains to be uncovered.

Complement Proteins

Complement proteins have been conventionally viewed as the effector arm for antibodies. However, recent investigations reveal that their functions are far beyond this initial understanding [57]. In transplantation, they promote IRI associated with organ procurement and cold storage during transportation and contribute to subsequent delayed graft function (DGF); they interact with coagulation pathways to promote thrombosis and exacerbate vascular inflammation; and most recently, they are found to directly activate B cells through B cell-surface complement receptors and promote B cell maturation and antibody-producing capacity, thereby contributing to AMR.

The speculation of their direct role in kidney IRI was initially based on the unexpected discovery of complement gene expression by renal tubuloepithelial cells, particularly in association with inflammatory renal tubular diseases [57]. Such local complement production can be induced by a number of kidney injuries, with IRI being the most relevant to transplantation [58]. Importantly, such locally produced complement, compared to circulating complement produced by the liver, have a greater impact on immune-mediated pathology. Activation of locally produced complement proteins is thought to be initiated by recognition of carbohydrate antigens on ischemic tissues by the lectin pathway gene products such as Collectin-11 (CL-11) and mannan-binding lectin (MBL). Of the three compliment activation pathways, namely the classical, alternative, and lection pathways, it is thought that the lectin pathway combined with amplification by the alternative pathway are the most predominant in IRI, and result in the ultimately acute tubular damage consequent to IRI.

Circulating complement proteins contribute to immune-mediated pathology in two additional ways. First, complement pathways and coagulation pathways interact and mutually activate each other, forming a positive feedback loop [59, 60]. Specifically, complement can induce the expression of tissue factors in several cell populations to initiate the extrinsic coagulation pathway. In addition, MBL-associated serine proteases can cleave prothrombin, fibrinogen, and factor XIII of the coagulation pathways, leading to their activation and subsequent thromboinflammation. Second, complement proteins can directly signal in B cells via complement receptors on the surface of B cells [61]. This signaling capacity requires a coordinated shift of expression of complement regulators on the surface of B cells [62]. Complement signaling combined with the canonical co-stimulation from helper T cells (carrying donor specificities) induce B cell affinity maturation and promote their differentiation into DSA-producing plasma cells, which eventually leads to AMR.

Therefore, complement proteins are another innate immune component that sits at the interface between innate and adaptive alloimmunity and directly impacts transplant outcome.

Summary

The prevailing of view of adaptive immunity being the primary transplant immunity will likely need to be revised as data continuously emerge to demonstrate the critical importance of innate immunity in transplantation. It is hoped that this review has provided a comprehensive and balanced view of the various components of the innate immune system in transplantation based on currently available data. Further studies in this nascent area are urgently needed for identifying precise and efficacious therapeutic targets to benefit long-term transplant allograft health and function.