Mechanisms Underlying Antibody-Mediated Rejection

Antibodies have been a vexing problem in transplantation for over 50 years. The first major barrier that antibodies presented to successful transplantation was hyperactute rejection in presensitized patients ¹. This barrier was circumvented in 1969 when Patel and Terasaki devised a crossmatch technique that was universally implemented and essentially eliminated hyperacute rejection ². In that same year, Williams and colleagues reported that 3 of the first 4 hearts transplanted at the Medical College of Virginia were rejected acutely in spite of treatment with high doses of steroids ³. Antibodies were implicated in the rejection process because antibodies were eluted from all 3 rejected hearts. Gradually, the development of more sensitive testing for antibodies before transplantation and advances in immunosuppressive therapy decreased the incidence of acute rejection.

Currently, over 90% of cardiac transplant recipients survive longer than a year. However, each year after transplantation an increasing number of grafts are diagnosed with chronic allograft vasculopathy (CAV) that is manifested by diffuse concentric intimal proliferation and adventitial sclerosis. CAV occurs at a rate of about 5% per year after cardiac transplantation, and CAV is a leading cause of long-term mortality. ⁴ The causes of CAV are complex, and include non-immune factors such as ischemia-reperfusion injury as well as immune responses of macrophages and T lymphocytes that release mediators such as cytokines and growth factors.

CAV has also been linked to antibodies: both systemic and local antibody production have been correlated with CAV, but analysis of clinical data from large cohorts have yielded conflicting conclusions ⁵. Studies of circulating antibodies are more numerous because serial samples can be obtained throughout the course of the transplant. However, these studies are confounded by variation in assays used to test for antibodies and by the profound effects of comorbidities in transplant recipients including infection and malignancy ⁴. The case for local antibody production is supported by the observation that B cells and plasma cells are common components of CAV either as adventitial nodules, diffuse adventitial infiltrates, or neointimal infiltrates ^6–8, and microarray profiles of coronary arteries with CAV indicate that immunoglobulin genes are upregulated ^{6, 9}. However, the data from clinical specimens regarding the antigenic reactivity of antibodies produced in coronary arteries with CAV are divergent ^{6, 7}. In this context cogent experimental models can offer valuable insights.

In the current issue of Circulation, Liu et al use an in vivo experimental model in which segments of human arteries are grafted to humanized mice to gain insights into local antibody production in CAV ¹⁰. Complexes of terminal complement component C9 were deposited on the endothelial and smooth muscle cells of arteries subjected to conditions modeling ischemia-reperfusion. Prolonged ischemia also increased intimal area that contained more T cells and B cells, many of which formed conjugates.

Similar conjugates of T and B cells were reported by Liarski, et al ¹¹ using cell distance mapping in biopsies from human renal transplants diagnosed with mixed T cell and antibody-mediated rejection. The T cells conjugated to B cells in renal transplants were categorized as T follicular helper (Tfh) cells on the basis of positive staining for CD4, ICOS and PD1. These T cells also expressed high levels of IL-21.

Liu et al initially expected the T cells expanded by ischemia-reperfusion would be Tfh cells, but in vitro experiments indicated that a different subpopuation of CD4+ T cells are stimulated when human memory T cells are co-cultured with endothelial cells previously subjected to ischemia. These T cells expressed high levels of HLA-DR, ICOS and PD-1 like Tfh cells. Unlike Tfh cells, the expanded subpopulation did not express CXCR5, a homing receptor to lymphoid follicles. Instead CCR2 was expressed. CCR2 is expressed by T peripheral helper (Tph) cells that migrate into inflammatory sites.

The Tph cell subpopulation derived from co-cultures with endothelial cells exposed to ischemia expressed IFN-gamma and IL-21 but not IL-4. Importantly, in co-cultures of human endothelial cells and B cells, these Tph cells increased IgG antibody production from B cells to HLA class I and II antigens expressed on the endothelial cells. This in vitro finding coincided with the animal model, in which IgG antibodies to HLA antigens expressed on the arterial transplants were also detected. The investigators surmise that the antibodies to HLA were produced in the arterial graft because the number of B cell follicles in the spleen was very low and did not correlate with human IgG concentration in the serum.

Additional experiments exploiting the in vitro model revealed that hypoxia and reoxygenation activates NLRP3 inflammasomes in endothelial cells and production of IL-18. Several lines of experiments demonstrated that IL-18 expanded Tph cells in culture and that more IL-18 receptor (IL-18R1) was expressed on Tph than Tfh cells. These findings were confirmed in humanized mice with segmental artery grafts, in which treatment with IL-18 increased Tph cells in the circulation and in the expanded neointima of the graft. IL-18 also increased conjugates of T and B cells in the graft and antibodies to donor HLA in the circulation.

Liu et al extended their experimental data by including clinical samples from renal transplant recipients who had delayed graft function that is a known consequence of ischemic injury. These patients had increased percentages of Il-18R1 expressing Tph cells but not Tfh cells in the circulation.

These findings have important clinical implications. First, the findings provide evidence that antibodies to donor HLA can be produced locally in transplants. Two prior studies have examined the antigenic specificity of antibodies produced locally in human cardiac transplants with CAV. Huibers et al demonstrated antibodies to donor HLA antigens in eluates from 4 of 21 postmortem hearts ⁷. Two of these patients had antibodies with the same specificity in serum samples. In the second study, Zorn and co-workers produced 102 B cell clones from infiltrates of 3 cardiac transplants with CAV. These patients did not have circulating antibodies to donor HLA and the majority of these clones secreted polyreactive IgM or IgG antibodies to apoptotic cells or autoantigens ⁶. Liu et al also found that B cells produced large amounts of IgM in response to ischemia-stressed endothelial cells in vitro and in vivo. The antigenic specificity of the IgM was not tested and as they discuss, at least part of the IgM could be polyreactive. Further delineating the antigenic reactivity of the antibodies in the humanized mouse model could be an informative future experiment because CAV has been associated with autoantibodies to a range of antigens, such as vimentin, myosin, collagen V, endothelin-1 and angiotensin II type 1 receptor ¹². It should be noted that this model provides insights primarily into local antibody production because the spleens of these humanized mice contain few B cell follicles and this limits qualitative and quantitative assessment of systemic antibodies.

In connection with local antibody production is the demonstration by Liu et al that Tph cells are in the circulation of mice and humans following transplantation. These cells have been reported in the circulation and the target tissues of patients with a variety of autoimmune diseases. Furthermore, in patients with celiac disease, Tph cells upregulate PD1 and ICOS in response to gluten challenge ¹³. This suggests that monitoring Tph cells in the circulation could be of value in transplant recipients. This would require a larger prospective study with serial samples in relation to specific clinical parameters such as antibody titers and biopsy findings. Also, it is worth noting that the preliminary clinical findings were derived from renal transplant recipients. Differences between cardiac and renal transplants might be expected because chronic rejection in hearts is predominantly manifested in larger vessels, whereas chronic rejection associated with antibodies in kidneys is typified by small vessel pathology in the glomerulus and peritubular capillaries.

Finally, the evidence that Tph cells express IL-18R1 and are responsive to IL-18 not only provides mechanistic insight, but also a potential therapeutic target. To this end, Liu et al demonstrated that blocking IL-18 decreased antibody production in vitro. It would be a valuable proof of principle to extend this finding in vivo by demonstrating the effects of treating the humanized mice with blocking antibodies to IL-18. The in vitro experiments also demonstrated that IL-18 expanded Tfh cells albeit to a lesser magnitude than Tph cells. Therefore, blocking IL-18 might inhibit both local and system antibody production in vivo. Although the finding that IL-18 is a critical mediator is novel, there remains a possibility that inflammasome-dependent IL-1 may have similar functions as the authors reported previously that endocytosed membrane attack complexes induce secretion of IL-1 by endothelial cells ¹⁴.

In summary, Liu et al have provided valuable insights to guide future experimental and clinical studies. They have identified a novel subpopulation of T cells that directs B cells to produce antibodies directed against the transplanted organ. Monitoring the activation of these Tph cells may help detect responses leading to local antibody production in CAV. In addition, blocking IL-18 stimulation of T cells may be an important therapy for antibody mediated CAV.