C1q as a potential tolerogenic therapeutic in transplantation

Abstract

In 1963, Lepow and colleagues resolved C1, the first component of the classical pathway, into three components, which they named C1q, C1r, and C1s. All three of these components were demonstrated to be involved in causing hemolysis in vitro. For over 30 years after that seminal discovery, the primary function attributed to C1q was as part of the C1 complex that initiated the classical pathway of the complement cascade. Then, a series of papers reported that isolated C1q could bind to apoptotic cells and facilitate their clearance by macrophages. Since then, rheumatologists have recognized that C1q is an important pattern recognition receptor (PRR) that diverts autoantigen containing extracellular vesicles from immune recognition. This critical function of C1q as a regulator of immune recognition has been largely overlooked in transplantation. Now that extracellular vesicles released from transplants have been identified as a major agent of immune recognition, it is logical to consider the potential impact of C1q on modulating the delivery of allogeneic extracellular vesicles to antigen presenting cells. This concept has clinical implications in the possible use of C1q or a derivative as a biological therapeutic to down-modulate immune responses to transplants.

The pro-inflammatory functions of complement during organ recovery, ischemia-reperfusion, and antibody-mediated rejection (AMR) have been recognized for a long time.¹ As a result, numerous therapeutic inhibitors of complement have been developed and tested in transplant recipients. Inhibitors of the terminal complement component C5, and more recently, the first component of the classical pathway C1, have been tested most extensively.^2–7 These inhibitors have been applied “off label” with some success but without knowing the complete spectrum of effects.

Targeting C1 is complex because C1 is a composite of molecules—C1q that binds to a range of substrates and provides a scaffold for the proteolytic tetramer of two C1r and two C1s components. These components impart disparate functions to C1. The function of C1 that was first defined and continues to be most widely recognized is its capacity to bind antibodies and initiate the classical pathway of complement.⁸ In this capacity C1q binds to antibody and the enzymatic components of C1 (C1r and C1s) initiate the complement cascade leading to the inflammatory complement split products that recruit and activate neutrophils and macrophages characteristic of AMR (Figure 1). The clinical implications of complement activation through C1q have been accentuated in transplantation by the association of poor outcomes with in vitro assays for donor-specific antibodies that bind C1q or that activate C4 and C3. In vitro experiments have identified several parameters that determine C1q binding to antibodies including subclass, glycosylation status, and concentration. Effective binding of C1q requires clusters of IgG antibodies bound to the membrane of target cells. Consequently, when antibodies are tested in serial dilutions, C1q binding declines precipitously.

FIGURE 1.

Divergent functions of C1q as a pattern recognition receptor (PRR) involved in clearing extracellular vesicles (EV) by noninflammatory macrophages and as an anchoring scaffold for C1r and C1s proteolytic enzymes to initiate the classical pathway of complement and generate pro-inflammatory split products such as C5a as well as the membrane attack complex (C5b-9)

Unlike other complement components, C1q is not produced by hepatocytes but primarily by myeloid cells (monocytes, macrophages, and immature dendritic cells). As a result, C1q is produced and secreted in tissues at the site of inflammation as well as circulating in plasma. Although C1q is secreted and circulates as a soluble protein, some C1q is associated with receptors on the membranes of monocytes and macrophages. The amount of membrane associated C1q can be regulated and is increased during inflammation by IFNγ.

The activation of the classical pathway of complement is controlled endogenously by circulating C1 inhibitor (C1inh), a serine protease inhibitor or serpin. C1inh contains pseudo substrates for C1r and C1s that irreversibly bind these enzymatic components and remove them from the C1 complex leaving C1q intact. However, isolated C1q is not impotent. Extensive evidence indicates C1q functions as a pattern recognition receptor (PRR) that binds apoptotic cells and mediates a noninflammatory clearance by macrophages.^9–11 The PRR function of C1q is prototypical because C1q precursors evolved in invertebrate species long before antibodies and the terminal complement components.¹² In species that lack antibodies, the C1q precursor functions as a PRR for pathogens and injured tissues rather than as an initiator of the complement cascade.

The PRR function of C1q has not become vestigial with evolution. In fact, C1q deficient patients and mice do not clear apoptotic cells efficiently and develop florid autoimmunity to the contents of the apoptotic bodies.^13–15 The strong correlation between C1q deficiency and a severe lupus-like disease impelled the investigation of tolerogenic functions of C1q. In vitro experiments demonstrated that C1q binds directly to apoptotic cells and mediated uptake of apoptotic bodies by macrophages.^10,11 These findings spawned the theory that C1q acting as a PRR diverted autoantigens contained in apoptotic bodies from immune recognition. This theory was further supported by the phenotypic analysis of C1q knockout mice that demonstrated a failure to clear apoptotic bodies combined with the production of antinuclear antibodies and proliferative glomerulonephritis.¹³ However, it is now recognized that the link between autoimmunity and C1q is more complex because deficiencies of similar molecules with PRR function such as mannose-binding lectin (MBL) result in accumulation of apoptotic bodies but not autoimmune disease. More recent studies have demonstrated that C1q also inhibits maturation of monocytes into dendritic cells and metabolically regulates T cells.¹⁶ Furthermore, when presented with apoptotic bodies in the context of C1q, macrophages produce IL-10 and TGFβ, express elevated PD-L1 and PD-L2, and suppressed surface CD40.⁹ Similarly, in renal cell cancer, local C1q secreted by tumor associated macrophages is associated with increased expression of the C1q receptor LAIR1 and the check point ligand PD-L2 on the macrophages.¹⁷ It is not known whether this is the result of C1q interacting with EV released from the cancer cells or from direct interaction with the cancer. Taken together, these findings in autoimmunity and tumor immunology suggest that C1q could have greater impact on immune responses to transplants than other candidate PRR such as MBL.

Understanding the range of PRR functions performed by C1q has evolved as the structure of C1q has become more completely defined. C1q is the product of three genes (C1q A, B, and C) each of which encodes a protein with a collagen-like tail and a globular head. The A, B, and C chains assemble into trimers, and six trimers form hexamers (Figure 2). This large composite molecule (460 kD) forms multiple interactive sites. Crystallographic models have indicated that the binding site for antibodies is on the outer surface of the globular heads, whereas apoptotic cells are bound on the inner surface of the globular heads.¹⁸ Multiple receptors on macrophages and dendritic cells have been found to engage sites in the globular heads and the collagen tails of C1q. Of particular relevance to the clearance of apoptotic cells are CD91 (the α-2-macroglobulin receptor) and the C-type lectin SIGN-R1. CD91 on human monocytes binds sites on both the collagen tail and globular heads of C1q and induces phagocytosis of apoptotic cells.¹⁹ SIGN-R1 binds the globular head of C1q and localizes C1q opsonized apoptotic cells to marginal zone macrophages in the spleen of mice.²⁰ The binding sites for C1r and C1s are between the hinge region of the collagen tails and the globular heads.²¹ Therefore, the association of C1r and C1s is thought to restrict the flexibility of C1q and possibly its binding to apoptotic cells.¹⁸

FIGURE 2.

Structure of C1q: the trimer subcomponent (left) and the complete hexamer (right) with binding sites for apoptotic cells, antibodies, and the proteolytic tetramer of two C1r and two C1s molecules indicated

Although C1 has been targeted in transplant patients in an attempt to limit the activation of the classical complement cascade, remarkably little is known about the effects of C1q as a PRR in transplantation. A few clinical and preclinical studies have used C1inh or antibodies to C1s (BIVV009) to “disarm” the C1 complex of the C1r/C1s enzymatic components during AMR.^2,5 One small clinical trial found that blocking C1s effectively abrogated the activation of C4 in late AMR but was less effective in diminishing tissue injury or improving graft function.² C1inh has had variable benefits when added to other treatments for AMR,^5–7 but both clinical trials and animal experiments suggest that C1inh could be more effective in limiting the consequences of ischemia-reperfusion.^3,4,22–24

In a recent clinical trial, injury caused by delayed graft function was diminished and 3-year graft survival was increased by treatment with C1inh in the operating room before graft reperfusion and 1 day later.^3,4 This approach was extended in a recent preclinical study by treating nonhuman primate kidney donors with C1inh for 20 h after inducing brain death and before the harvested organ was stored for 44 h on ice in UW solution.²² This resulted in decreased graft injury in the first week after transplantation as measured by decreased serum creatinine and urinary lipocalin 2 (Lcn2 or Ngal) and slightly improved graft survival. Similar findings have been reported in a porcine model of warm ischemia²³ and a rat model of brain-dead kidney donors.²⁴ However, the interpretation of all these findings has been limited to assessing the blockade of complement activation either through the classical or the lectin pathway. Although treatment with C1inh preserves C1q function, no consideration or more importantly experimental data have been pursued about the potential anti-inflammatory effects of complement. Specifically, no measurements of clearance of apoptotic bodies or altered T and B cell priming related to C1q have been reported in transplantation.

The results with C1inh invite the obvious question: Does C1inh work because it truncates the complement cascade and decreases production of downstream inflammatory mediators or does C1inh work because it leaves C1q intact to modulate macrophages and other cells that express C1q receptors? Of course, these are not mutually exclusive.

Appreciation of the PRR function of C1q has clinical implications. During reperfusion of transplants, C1q could function as a PRR to opsonize apoptotic cells and extracellular vesicles (EV) for noninflammatory clearance by macrophages. In this capacity, C1q induces macrophages to produce IL-10 and TGFβ, whereas later complement split products, especially C5a, induce the production of iNOS, TNFα, and IL-1β.⁹

While the binding of C1q to apoptotic cells is well documented, the potentially equally critical interaction of C1q with EV that express autoantigens and MHC antigens has not been tested. This includes the highly immunogenic “apoptotic exosome-like vesicles” (ApoExo) that have been reported by Marie-Josée Hébert and co-workers to contain fragments of perlecan and other extra cellular matrix molecules.²⁵ Apoptotic bodies and EV including ApoExo and microvesicles that are formed by budding from the outer cell membrane exteriorize phosphatidylserine as detected by annexin V binding. Phosphatidylserine is a major ligand for C1q as demonstrated by the co-localization of C1q and annexin V on apoptotic bodes.²⁶ Therefore, we propose that the clearance kinetics of phosphatidylserine-expressing vesicles is dependent in large part on the concentrations of C1q. Moreover, we expect that other EV may also be bound by C1q because C1q has a wide range of ligands in addition to phosphatidylserine.²⁷

We recognize that exosomes have been demonstrated to be immunogenic in vivo. However, larger EV have also been reported to express high levels of MHC antigens and costimulatory molecules.²⁸ These larger ECV stimulate T cell responses in vitro.²⁹ Yet little is known about these EV in vivo where C1q could critically modulate their immunogenicity. The quantities and qualities of apoptotic bodies and EV released at different times after reperfusion are not well-defined.

Other variables that remain to be defined include local and systemic mediators and cells that influence the immunogenicity or tolerogenicity of exosomes and EV. It is known that the physiological events causing the release of EV can alter the contents and response to EV. For example, stress-induced apoptosis causes endothelial cells to release apoptotic bodies in vitro that contain mature IL-1α. Intraperitoneal injection of these stress-induced apoptotic bodies elicits a neutrophilic inflammatory response.³⁰ In the context of transplantation, Prunevieille et al³¹ reported that intraperitoneal injection of allogeneic exosomes in adjuvant elicits a vigorous T cell response and accelerates skin graft rejection, but neither of these responses to exosomes occurred in the absence of adjuvant. Adjuvants are designed to recruit and activate macrophages, and this is likely key to the increased immunogenicity of exosomes in this model because the investigators found that allogeneic exosomes were recognized by T cells, but no response was elicited unless the exosomes were first captured by antigen-presenting cells. Antigen-presenting cells also appear to determine responses to exosomes in studies of pregnancy and liver transplantation where dendritic cells that co-express allogeneic MHC molecules with PD-L1 induce tolerance.^32,33

Another unknown is the effect of transplantation on circulating C1q. One recent clinical study reported that over 20% of recipients had decreased levels of C3 at the time of and 1 month after renal transplantation,³⁴ but C1q was not measured. Limited data are available concerning the regulation of C1q production. One study found that IL-6 increased, whereas IL-1 and IFNγ decreased C1q production by peritoneal macrophages in vitro.³⁵ All of these dynamics of C1q metabolism deserve further investigation both in deceased donors and transplant recipients.

Moreover, it is not known whether C1q is depleted by binding to apoptotic bodies and EV. Normally, C1q is present in small amounts (80–100 ug/ml in normal plasma), and this could limit its capacity to clear immunogenic apoptotic cells and EV in donors and recipients. It is known that C1q is decreased during flares of lupus activity. In some cases this is correlated with autoantibodies to C1q, but C1q consumption by immune complexes is another mechanism that depletes C1q levels. In transplantation, amounts of C1q available for clearing apoptotic cells and EV could be exceeded in situations such as kidneys with prolonged ischemia or from expanded criteria deceased donors. Adding C1q and C1inh to the perfusate prior to storing the grafts on ice or during normothermic perfusion as well as at various times after transplantation could be an effective therapeutic not only because of the limited amounts of endogenous C1q but also because most circulating C1q is complexed with the proteolytic tetramer of two C1r and two C1s molecules. Because C1r and C1s restrict the flexibility of C1q, these components limit the capacity of C1q to bind apoptotic cells.^18,36 For this reason, C1inh could increase the effectiveness of intrinsic C1q in the circulation by freeing C1q from C1r and C1s. Moreover, administration of isolated C1q would compete favorably with circulating whole C1 to clear EV and have the added benefit of not activating the complement cascade.

More sophisticated knowledge of structural aspects of C1q has permitted strategic modifications for therapeutic applications. Mutations in recombinant human C1q resulted in variants that do not form complexes with C1r and C1s. As a result, these variants do not initiate the classical pathway of complement but retain the PRR functions in vitro.³⁷ However, the half-life and immunogenicity of these C1q variants have not been tested in vivo. Additionally, the identification of different binding sites for antibodies and apoptotic cells on the globular heads of C1q suggests that molecular modifications could be devised that abrogate interactions with antibodies but retain pattern recognition. Similarly, modifications might be designed that enhance the interaction of C1q with selected receptors on macrophages.

In summary, the concept that C1q functions as a PRR that removes potentially immunogenic apoptotic bodies and EV during reperfusion challenges the conventional view that complement is purely pro-inflammatory in transplants. It provides an additional mechanistic explanation for the new clinical finding that C1inh mitigates ischemia-reperfusion injury in renal transplant recipients. Importantly, if C1q is demonstrated to be beneficial, then C1q or a modified C1q could be an effective new therapeutic biologic to decrease the immunogenicity of transplants.

Abbreviations:

C1inh

C1 inhibitor

EV

extracellular vesicles

MBL

mannose-binding lectin

PRR

pattern recognition receptor