PLASMA CELL BIOLOGY: FOUNDATIONS FOR TARGETED THERAPEUTIC DEVELOPMENT IN TRANSPLANTATION

SUMMARY

Solid organ transplantation is a lifesaving procedure for patients with end-stage organ disease. Over the past 70 years, tremendous progress has been made in solid organ transplantation, particularly in T-cell targeted immunosuppression and organ allocation systems. However, humoral alloimmune responses remain a major challenge to progress. Patients with pre-existing antibodies to human leukocyte antigen (HLA) are at significant disadvantages in regard to receiving a well-matched organ, moreover those who develop anti-HLA antibodies after transplantation face a significant foreshortening of renal allograft survival. Historical therapies to desensitize patients prior to transplantation or to treat posttransplant AMR have had limited effectiveness, likely because they do not significantly reduce antibody levels as plasma cells, the source of antibody production, remain largely unaffected. Herein, we will discuss the significance of plasma cells in transplantation, aspects of their biology as potential therapeutic targets, clinical challenges in developing strategies to target plasma cells in transplantation, and lastly, novel approaches that have potential to advance the field.

1 |. INTRODUCTION

Since the first successful solid organ transplant in 1954, immunosuppressive drug development has focused heavily on suppressing or depleting T-cells¹. This was due to seminal observations by Medawar and colleagues that thymus-educated cells were required for organ rejection while passive transfer of antibodies was unable to induce accelerated allograft rejection in naïve recipients ^2–8. In the subsequent decades, advances in T-cell targeted immunosuppression effectively controlled cellular alloimmune responses, such that antibody-mediated rejection (AMR) is now arguably the predominant mechanism for alloimmune allograft loss ⁹. While the presence of donor specific antibodies (DSA) has been acknowledged for much of this history, their causative versus consequential role in rejection was hotly debated through the early 2000s when Paul Terasaki published his humoral theory of transplantation ¹⁰. In 2009, the Cincinnati group published seminal data on the effects of AMR, including coining of the term, mixed acute rejection (MAR), in which humoral components of rejection were described as a major factor in determining renal allograft survival following acute rejection ⁹. Over the following decade, there has been an increasing appreciation for the causative role of antibodies in long-term allograft rejection and in particular, allograft loss. Indeed, while T-cell mediated rejection episodes are typically effectively treated by high dose corticosteroids and T-cell depleting anti-lymphocyte antibody preparations, treatment options for AMR remain limited. While early AMR and early MAR are effectively treated with current agents, late AMR and late MAR are associated with marked foreshortening of renal allograft survival ^{9, 11–15}. In addition to reducing allograft longevity, allosensitization also complicates repeat transplantation in terms of histocompatibility matching with future donors and rejection risk, resulting in prolonged waiting times ^{16, 17}. As such, there is a critical need to develop humoral-targeted therapies to desensitize patients prior to transplantation and to prevent and treat AMR after transplantation. To date, most therapeutic approaches have focused on clearing antibodies from the serum or depleting B-cells, but these approaches fail to eliminate plasma cells, which are the source of antibody, and therefore long-term successes are infrequent. Successful treatment regimens will ideally target multiple stages of the humoral response, including sensitization, memory, and allograft injury. This review will cover the genesis of allospecific plasma cells, their role in allograft injury, strategies for therapeutic targeting of plasma cells, and lastly, application of newer technologies that will enable deeper basic biologic insights, thereby providing the necessary foundations for moving the field forward. In the following discussion, we will focus primarily on renal transplantation as there is an abundance of literature in this area and the findings are likely applicable across other solid organs as well.

2 |. HUMORAL ALLOSENSITIZATION

Sensitization to human leukocyte antigens (HLA) is a critical development in kidney transplant candidates and recipients as it complicates tissue matching prior to transplantation and is a major pathophysiological event in allograft rejection. In this section, we will review HLA and non-HLA antigens and sensitizing events.

2.1. Human Leukocyte Antigen

Polymorphisms exist for many proteins, however, the most polymorphic proteins in most species are major histocompatibility complex (MHC) proteins, that are termed human leukocyte antigens (HLA) in human and histocompatibility-2 (H2) in mice. There are 12 classical HLA loci and over 25,000 alleles ¹⁸. A result of this high degree of polymorphism is the significant risk for developing antibodies against the broad array of HLA antigens that are encountered in transplant recipients. This led to recognition of the importance of “matching” HLA antigens in transplantation over 60 years ago ¹⁹.

HLA Class I molecules are present on the surface of all nucleated cells and platelets and display peptides from endogenous proteins to CD8+ T-cells. HLA Class II molecules are expressed predominately by antigen presenting cells (APCs) such as dendritic cells and B-cells and present peptides from exogenous proteins to CD4+ T-cells. HLA Class II is also expressed on endothelial cells through an interferon-γ (IFNγ) dependent mechanism ^{20, 21}. HLA antigen recognition differs for T cells and B-cells: whereas T cells recognize peptides (8 – 15 amino acids) that are presented in the context of MHC by APCs, B-cells recognize short amino acid sequences (epitopes) within native proteins. Epitopes can be further broken down into “eplets” which are the 2–5 amino acid length sequences that antibodies bind ²². Eplets can be shared between a few to several different HLA molecules, and each HLA molecule may possess multiple eplets across its external surface. The sharing of eplets between unique HLA proteins explains in part why a few to several antibodies can result in patients being sensitized to over 99% of potential donors, making finding a compatible donor statistically highly unlikely.

In the early 2000s, Rene Duquesnoy published an algorithm called HLAMatchMaker that predicts eplets based on structural modeling of antibody-epitope interactions from crystalized antibody complexes ²². Several studies have found using the number of eplet mismatches or “epitope load”, instead of antigenic mismatches, is a better predictor of which patients will develop DSA, AMR, and eventual allograft failure ^23–27. For instance, a DQ7 mismatch with a DQ6/DQ8 recipient has only one eplet mismatch whereas one with a DQ2/DQ2 recipient has 16 eplet mismatches and a greater likelihood of developing DSA and subsequent AMR ²⁷. Thus, HLA molecules are a major target of alloantibodies and eplet mismatches play a significant role in AMR as well as organ availability.

2.2. Non-HLA antigens

While anti-HLA antibodies are the most important and well-recognized cause of humoral rejection, studies showing AMR in recipients of HLA-identical sibling kidneys have highlighted the importance of non-HLA antigens in rejection ^28–30. Most non-HLA antigens are expressed on endothelial cells as they are the interface between allograft and host immune cells ^{31, 32}. Non-HLA antibodies come in two flavors – those directed against polymorphic proteins and those directed against self-proteins. Non-synonymous single nucleotide polymorphisms (SNPs) result in polymorphic proteins with the potential to be seen as non-self by the recipient ^{33, 34}. In cases of a male donor and female recipient, proteins encoded by the Y-chromosome, although not highly polymorphic, can also be seen as non-self ³⁵. Other genes lack SNPs in protein-coding regions but contain SNPs that lead to alterations in the degree of protein expression and the potential for auto- and alloantibody formation. As well, cellular damage during transplantation or acute rejection can lead to the presentation of self-antigens in an inflammatory context and break self-tolerance.

One of the more important non-HLA antigens in transplantation is the angiotensin II type I receptor (AT1R) ^{36, 37}. Antibody binding to G-protein coupled AT1R receptors causes receptor signaling and in the case of AT1R, can result in endothelial activation and hypertension ^{36, 38, 39}. A prospective study from the Paris Transplant Group published in 2019 found patients with anti-AT1R antibodies had a higher incidence of AMR, an increased risk of allograft loss, and a distinct phenotype that included hypertension, endothelial activation, arterial inflammation, and an absence of complement deposition ³⁹. Moreover, patients with both anti-AT1R antibodies and anti-HLA antibodies had significantly worse allograft survival compared with patients with either antibody alone. Thus, antibodies to non-HLA molecules can drive AMR and patients are often screened for the present of AT1R antibodies prior to transplantation.

2.3. Blood transfusion

Red blood cell (RBC) transfusion is indicated for a variety of reasons including hemorrhage from trauma, surgery, or pregnancy; sepsis; and insufficient erythropoiesis from malignancy or renal disease ⁴⁰. Patients with end stage renal disease (ESRD) often develop anemia from the kidney’s inability to make sufficient erythropoietin. Despite the advent of recombinant erythropoietin, 3–4% of ESRD patients on dialysis require transfusion ⁴¹. Blood transfusions in patients awaiting transplantation, regardless of leukoreduction, have a rate of allosensitization of 20–30% ^42–44. In patients previously allosensitized, such as multiparous women, the rate of transfusion associated sensitization increases to approximately 50% ⁴². Sensitization through blood transfusion occurs through passenger leukocytes expressing surface HLA, RBC-bound HLA, and soluble HLA in the blood ^45–47.

2.4. Pregnancy

Despite the tolerogenic setting of pregnancy that promotes “acceptance” of the fetal allograft, pregnancy remains a significant source of HLA sensitization. Approximately 54% of primiparous and 74% of multiparous women develop antibodies to the father’s HLA ^{48, 49}. Starting at week eight of pregnancy. As early as the first trimester, anti-HLA antibodies can be detected, primarily bound to fetal-derived soluble HLA in the blood ⁵⁰. Detection of alloantibodies increases over the first 90 days postpartum but subsequently decreases over time with roughly 50% of women having undetectable alloantibodies after two years postpartum ⁵¹. Taken together, allosensitization by pregnancy can be a significant barrier to transplantation. Indeed, women receive a living donor kidney significantly less often than men and wait longer to find a matched organ, both of which contribute to poorer long-term outcomes ^{52, 53}. Understanding mechanisms of allosensitization during pregnancy may help mitigate the sex disparities in transplantation.

A 2021 study from Chong’s group found pregnancy induces a tolerogenic cellular response to paternal-derived fetal antigens but a sensitized humoral response in a mouse model of pregnancy allosensitization ⁵⁴. As shown previously, most allospecific T-cells generated during pregnancy were regulatory, and the conventional T-cells that did develop had an anergic phenotype and were incapable of IFNγ production ^{55, 56}. Conversely, fetus-specific antibody titers increased throughout pregnancy and with subsequent pregnancies.

To evaluate the influence of pregnancy-induced allosensitization on allograft acceptance under costimulatory blockade (CoB), postpartum mice were treated with CoB and transplanted with a heart from their offspring. In virgin mice not treated with CoB, hearts reject around day 11 post-transplant. With CoB treatment, at 60 days post-transplant, virgin mice had 100% allograft survival whereas postpartum mice had 30% allograft survival. Interestingly, the resistance to CoB appeared to be due to B-cells rather than T-cells as the loss of B-cells restored CoB-driven allograft acceptance. Conversely, adoptive transfer of antibodies or secretory-immunoglobulin knockout B-cells from postpartum mice was sufficient to abrogate CoB-mediated tolerance in virgin mice. These results led to the hypothesis that while allosensitized B-cells do not affect the development of T-cell tolerance during pregnancy, in the context of transplantation, allospecific B-cells override T-cell tolerance either through antigen presentation or secretion of inflammatory cytokines. The authors suggested that if true in humans, humoral desensitization in multiparous women could lead to improved acceptance of allografts from offspring.

2.5. Transplantation

In addition to blood transfusion and pregnancy, organ and tissue transplantation can also obviously result in HLA sensitization. In transplantation, patients can develop de novo DSA which are associated with chronic AMR and worse outcomes than the pathologies associated with preexisting DSA (including from previous transplant). It is not clear why de novo DSA responses are more deleterious to allograft survival than are early naïve or anamnestic DSA responses, however, it is possible that patients who develop de novo DSA late after transplantation have allograft injury that proceeds over long periods between physician appointments, therefore late DSA often have long periods in which they are unrecognized.

While our knowledge about mechanisms of B- and T-cell response to pathogens is fairly well developed, we understand relatively little about how B- and T-cells become sensitized to alloantigens. Here we will discuss what is known regarding humoral allosensitization but also focus on critical unanswered questions.

Our current understanding of B-cell allosensitization is that B-cells encounter alloantigens through several mechanisms: 1) recipient follicular dendritic cells and subcapsular sinus macrophages capture allograft-released soluble HLA or HLA-containing extracellular vesicles and present it intact to allospecific B-cells in allograft-draining secondary lymphoid tissues (gSLT) ^57–59; 2) recipient APCs traffic through the allograft and return to gSLTs with alloantigens and present them intact to B-cells ⁶⁰; 3) donor APCs migrate from the allograft to gSLTs and B-cells directly interact with their surface HLA ⁶¹. The contribution of donor APCs to B-cell allosensitization in murine models is debated. A 2021 publication from Morelli’s group demonstrated donor APCs could not be found in gSLT in allograft recipients but were present in syngeneic recipients only after the time point when the allogeneic grafts were rejected. In human transplantation however, allograft rejection does typically not occur as early as in the murine model leading us to question if donor APC trafficking to recipient gSLTs remains relevant to human B-cell allosensitization. It’s also possible the interventions that circumvent hyperacute rejection (HLA matching, immunosuppression, “desensitization”) delay B-cell allosensitization long enough that donor APCs are dead or killed by allograft infiltrating recipient cells before B-cell allosensitization occurs. This would depend on factors unique to each patient including degree of lymphodepletion, medication adherence, and drug metabolism. As well, donor APCs may traffic to recipient gSLTs but are killed by recipient natural killer cells before direct allorecognition by recipient T- and B-cells can occur ⁶². In other organs such as lung, donor APCs’ relevance to allosensitzation is more apparent as donor alveolar macrophages can survive for the life of the allograft and are associated with DSA formation ⁶³. Similarly, there are long lived kidney resident macrophages however, they are particularly sensitive to ischemia/reperfusion injury and their contribution to B-cell allosensitization remains unclear ⁶⁴.

An area of transplant allosensitzation that is particularly intriguing is that most de novo DSA are against Class II and HLA Class II mismatch is associated with the development of DSA, AMR, and allograft loss^{23, 65–68}. There are several proposed hypotheses to explain such phenomena:

• Class II is primarily expressed on APCs, which are presumed to be replaced by host APCs in the allograft by the time that HLA antibodies can be detected in recipient circulation. This leaves donor Class II expression to endothelial cells. A subset of endothelial cells express Class II at baseline but expression significantly increases during periods of inflammation when IFNγ levels are increased ²¹. Thus, Class II-specific B-cells may only encounter their antigen in the context of inflammation where Class II-specific CD4+ T-cells are also active and can encounter Class II both directly on the endothelium and indirectly through recipient APC presentation of donor HLA peptides. Conversely, an explanation for reduced association between Class I-specific antibodies and severity of rejection is that anti-Class I B-cells initially “see” their antigen during immunosuppression where they lack T-cell help and a fulminant response fails to occur.

• Building on the above scenario, if there’s a stronger activation of CD4+ T-cells because they can encounter Class II HLA antigens both directly and indirectly, perhaps they activate recipient dendritic cells more robustly, subsequently licensing them to induce stronger CD8+ T-cell responses ⁶⁹. This could account for the association of AMR severity and Class II if Class II-associated AMR is primarily mixed rejection with a significant amount of CD8 mediated damage ^{70, 71}.

Understanding mechanisms of transplant allosensitization can improve selection of maintenance immunosuppression to prevent de novo allosensitization.

3 |. ANTIBODY MEDIATED REJECTION

The humoral immune response can cause organ rejection and loss in both antibody-dependent and independent mechanisms. Here, we will review antibody-dependent mechanisms as antibody-independent mechanisms are outside the scope of this review but are nicely covered in recent reviews ^72–74. This section will briefly review antibody mediated allograft injury, acute and chronic AMR, and approaches to desensitize patients and treat AMR.

3.1. Mechanisms of antibody mediated allograft injury

Endothelial damage is a defining component of AMR. Donor specific antibodies (HLA and non-HLA antibodies) directly interact with the allograft endothelium as it is the interface between recipient and allograft. Antibodies cause direct and indirect injury to the allograft by binding epithelium and inducing endothelial cell activation, complement activation, and Fc receptor (FcR) mediated leukocyte activation and recruitment. Antibodies against Class I and class II HLA induce unique signaling events, which have been described in detail ⁷⁵. Endothelial cells become activated when anti-HLA antibodies bind to endothelial surface HLA, with overall effects being a function of several factors including anti-HLA antibody concentration; class I and class II HLA expression density; complement activation; and FcR engagement on macrophages, NK cells, and neutrophils ⁷⁶. Conceptually, it is important to understand that complement is activated by antibodies via the classical pathway, where optimal C1q activation occurs when antibodies saturate HLA molecules, such that each of the six globular heads of C1q are engaged by HLA bound antibodies. However, under conditions in which complement is suboptimally activated, sublytic complement activation can result in distinct intracellular signaling via noncanonical NFKb pathways ^77–79. Depending on the specific antigen, endothelial cell activation can result in proliferation, expression of inflammatory mediators such as IL-6, and adhesion factors that attract leukocytes into the allograft (e.g. ligation of HLA Class I) ⁸⁰. In the case of agonistic binding to receptors like AT1R, chronic activation can lead to intimal remodeling and vascular pathologies like hypertension ⁸¹.

Optimal C1q engagement by the complement binding region of antibodies is mediated when anti-HLA antibodies are bound to cell surface HLA molecules (particularly if HLA molecules are highly expressed) and arranged in a hexagonal pattern. Once bound to cell surface molecules on the endothelium, antibody subclasses IgG1 and IgG3 in particular, bind the complement component C1q, and initiate the classical complement cascade ^{82, 83}. This results in recruitment of leukocytes which further induce inflammation and allograft damage as well as direct lysis of the targeted cell through complement dependent cytotoxicity. Furthermore, clinical studies have demonstrated that the presence of complement-fixing DSA portends worse allograft damage and survival ⁸⁴. Historically, deposition of the complement product, C4d, in allograft microvasculature was a requirement for diagnosing AMR, however in the mid 2010s adequate evidence for C4d-negative AMR had accumulated that Banff diagnostic guidelines were updated to include C4d negative AMR.

Lastly, antibody binding to alloantigens on the surface of donor cells (primarily endothelium) results in the recruitment of leukocytes like NK cells, monocytes, neutrophils, and CD8+ T-cells ^85–87. In particular, NK cells express Fc receptors that bind the Fc portion of antibodies bound to the vascular endothelium, inducing NK cell degranulation and donor cell lysis via antibody-dependent cellular cytotoxicity or release of proinflammatory cytokines that cause further endothelial cell activation and damage to the allograft ⁸⁸.

Based on the observed clinical associations with antibody specificities (e.g., anti-Class II, anti-AT1R, anti-collagen), there are likely different AMR phenotypes with unique etiologies and optimal treatment and prevention strategies. For example, AMR associated with anti-AT1R antibodies presents with features of hypertension, arterial inflammation, and a lack of C4d deposition despite the presence of IgG1 and IgG3 antibody subclasses and responds to treatment with the AT1R antagonist, losartan ³⁶. Pathologically, AMR presents with a spectrum of histologic features and patterns ranging from mild AMR with acute tubular injury but without C4d deposition and with low level microvascular inflammation (peritubular capillaritis and glomerular capillaritis) to moderate forms in which C4d deposition is prominent with significant microvascular inflammation and interstitial mixed inflammatory infiltrates ⁸⁹. Severe AMR can present with more extensive features of moderate AMR, but includes vascular necrosis and vascular thrombosis and infarction, which, if unabated, can lead to rapid allograft loss. Future research further classifying alloantibodies, their associated phenotypes, potential mechanistic etiologies, and outcomes will improve our approach to patient risk stratification, treatment, and prognosis.

3.2. Acute AMR

Approximately a third of the kidney transplant waitlist has pre-existing antibodies to HLA antigens, resulting from prior exposure to blood transfusions, pregnancy or prior organ transplants ⁹⁰. The degree of sensitization is quantified by calculated panel reactive antibodies or cPRA ⁹¹. cPRA is calculated by first determining the specificity of HLA antibodies using solid phase assay techniques such as single HLA antigen beads ⁹². The frequencies of HLA antigens in the population against which the patient has antibodies determines the cPRA. The higher a patient’s cPRA, the lower the statistical chance of finding an organ donor match which in turn, results in longer wait times. The 8-year survival rate for sensitized patients who remain on dialysis without a match is 43.9% and is 62.9% for those who eventually find a match ⁹³.

Dating back approximately 20 years, desensitization was attempted prior to transplantation using a combination of intravenous immunoglobulin (IVIG), plasmapheresis, and subsequently with addition of anti-CD20 antibody depletion with rituximab ^{94, 95}. However, upon transplantation with a mismatched organ, remaining allospecific memory B-cells encountering cognate antigen have the potential to proliferate and differentiate into DSA-secreting plasmablasts. Such anamnestic responses could result in rapid and marked increases in DSA resulting in acute AMR. Notably, hyperacute rejection, typically due to preexisting high levels of anti-ABO or anti-HLA antibodies is uncommonly encountered at present due to highly sensitive and reliable histocompatibility testing techniques. Regardless of the mechanism of sensitization, pre-existing antibodies and immune memory remain a significant hurdle for transplantation.

3.3. Chronic AMR

De novo DSA are defined as not resulting from prior HLA sensitization and typically develop several weeks or later following transplantation. Estimates of the prevalence of de novo DSA versus pre-existing DSA have been limited historically by the sensitivity of detection methods as preexisting DSA can be present at very low or undetectable levels, particularly when the original sensitization event occurred years or decades prior, such as with prior pregnancies. However, with the advent of the highly sensitive single antigen bead assay, DSA detection has now become quite sensitive and more reliable. Despite these advances, an approach is needed for detection of prior humoral HLA sensitization in the absence of detectable circulating anti-HLA antibody, a process where memory B-cells capable of producing anti-HLA antibodies can be reliably identified ^{96, 97}.

Approximately 20% of kidney transplant recipients develop de novo DSA by 5 years post-transplantation ^{67, 98}. De novo anti-HLA antibodies generally occur late following transplantation and tend to be class II specific (usually HLA DR or DQ), often because of insufficient immunosuppression from patient non-adherence or physician-directed immunosuppression reduction. Another important contributing factor is the choice of immunosuppressive agents ^{99, 100}. In addition to directly suppressing T-cell responses, many immunosuppressive agents, including calcineurin inhibitors, directly and indirectly affect B-cells which should be considered when selecting a regimen, particularly for patients at high risk of sensitization.

3.4. Current approaches to desensitization and AMR

Recognition that preexisting DSA were a significant risk factor for hyperacute and acute AMR led to strategies aimed at decreasing DSA titers and antibody function beginning in the 1980s. Historically, the first IVIG-based approach for desensitization of HLA sensitized kidney transplant candidates prior to transplantation consisted of IVIG and plasmapheresis, an approach that derived from its application in patients with autoimmune neurologic diseases dating back more than two decades prior to its application in transplantation ^{95, 101–104}. Later, pan B-cell depletion with the anti-CD20 antibody rituximab was added to plasmapheresis and IVIG ⁹⁵. IVIG is a blood product, derived from plasma in which the immunoglobulin containing fraction from thousands of blood and plasma donors are pooled. IVIG contains immunoglobulins of all subtypes against a vast array of pathogens, environmental antigens, alloantigens, and idiotypic antibodies (antibodies that bind to the variable region of other antibodies) ¹⁰⁵.

Among the primary mechanisms proposed to explain how IVIG quells antibody mediated allograft damage are: 1) anti-idiotypic antibodies bind and block the function of DSA; 2) antibodies bind and saturate available Fc receptors, limiting ADCC and Fc-mediated leukocyte activation; 3) antibodies bind the neonatal Fc receptor to inhibit antibody recycling; 4) Fc-dependent upregulation of the inhibitory Fc receptor, FcRb IIb; and 5) complement scavenging ¹⁰⁶.

The questionable benefit of rituximab in desensitization and AMR treatment is largely due to the lack of sufficiently powered, randomized, controlled desensitization trials that have been conducted to date. A recent systematic review found that, while not conclusive, acute AMR tended to show a benefit from rituximab whereas chronic AMR did not ¹⁰⁷. This is consistent with the role of rituximab in depleting memory B-cells that would normally proliferate and differentiate into plasma cells during an anamnestic AMR ¹⁰⁸. While B-cell proliferation also occurs during the primary de novo response, by the time patients are diagnosed with chronic AMR, the contribution from memory B-cells is likely less significant. B-cell depletion with rituximab during plasma cell targeted therapies with proteasome inhibitors may also be used to mitigate the potential role of memory B-cell proliferation and differentiation that can repopulate the plasma cell niche and drive a rebound of DSA titers.

The above therapies, while insufficient in the long term, have allowed a number of patients to undergo transplantation. To extend the life of transplanted organs there is a critical need for combination therapies that provide bona fide desensitization i.e., desensitization where anti-HLA antibodies are eliminated prior to transplantation (including high level preexisting HLA antibodies) and do not rebound post transplantation, a goal that remains to be achieved.

4 |. PLASMA CELL BIOLOGY

Research over the past several decades has uncovered factors critical for plasma cell survival and has begun to identify potential therapeutic targets to reduce or eliminate plasma cells. Here we will focus on some of the elements, their respective therapies, and the data on their effectiveness in transplantation which should complement more mechanistic details on plasma cell homeostasis and targeting strategies in transplantation as recently reviewed ^109–115.

Primary humoral responses comprise robust production of antigen-specific plasmablasts however, only a subset survive (generally <10%) and become long-lived plasma cells (LLPCs) ¹¹⁶. The prevailing dogma in plasma cell biology is that for plasma cells to survive, they must traffic to specialized niches and receive survival signals. The molecular and cellular details of this process are still not clear but are an area of active research. The plasma cell niche hypothesis is supported by experiments showing that removal of plasma cells from the bone marrow results in their rapid cell death, unless cultured with bone marrow stromal cell culture supernatant and/or exogenous survival factors ¹¹⁷. Strikingly, antibody responses to vaccination have been estimated to have a half-life of over 3000 years for some vaccines, while others such as tetanus have half-lives of roughly 20 years ¹¹⁸. It has long been held that the majority of these LLPCs reside in the bone marrow, though many, primarily IgA-producing, reside in the intestine, where they likely rely on a distinct set of homeostatic survival mechanisms ¹¹⁹. Plasma cells can also be found in areas of inflammation like in the joint spaces of rheumatoid arthritis patients, kidneys of patients with lupus nephritis ¹²⁰, or in ectopic lymphoid-like structures in solid organ allografts ^121–126. Plasma cells are not infrequently found in rejecting kidney allografts, but their relevance in terms of antigen specificity and allograft injury is not known.

4.1. Homing to the niche

Newly generated plasma cells are recruited to the bone marrow by CXCL12 gradients from bone marrow hematopoietic and stromal cells. Bone marrow plasma cells (BMPCs) continue to express CXCR4 but whether they retain chemoattraction to CXCL12 once in the marrow is debated. Murine studies from Hauser et al. found ~70% of plasma cells isolated from the bone marrow six days post immunization migrated toward CXCL12 gradients in vitro but those isolated at day 12 did not despite 77% of them expressing CXCR4 ¹²⁷. Human studies from Nakayama et al. analyzing chemokine expression on BMPCs in steady state (i.e., not following immunization) found high expression of CXCR4 and CXCR6 and intermediate expression of CCR10. The ligands for each chemokine were confirmed to be expressed highly in the bone marrow and induced chemoattraction in vitro. Additionally, in vitro exposure to each of the respective ligands induced VLA-4 dependent adhesion to fibronectin, factors important to plasma cell maintenance in the niche ¹²⁸. Recognizing the potential context-dependent role of chemokine receptors in plasma cell biology is important for the interpretation of studies that attempt plasma cell deletion through chemokine targeting.

Regardless of CXCL12/CXCR4-driven chemoattraction, CXCL12/CXCR4 signaling indirectly supports plasma cell survival and niche retention. Although the mechanisms for survival and retention are not well characterized in plasma cells, in other immune cells, CXCL12 signaling supports survival in part through regulation of Bcl-2 family members ^{129, 130}. CXCL12 also induces the expression of the adhesion factor very late antigen 4 (VLA-4) that, together with lymphocyte function-associated antigen 1 (LFA-1) anchors plasma cells into the niche through interaction with vascular cell adhesion molecule 1 (VCAM-1) and intercellular adhesion molecule 1 (ICAM-1) expressed on mesenchymal stromal cells (MSCs). Fibronectin, also secreted by MSCs, interacts with VLA-4 and CD44 expressed on plasma cells and is important for their differentiation, antibody secretion, and survival in vitro ^{117, 131}. The critical importance of life-long antibody production supports the plausibility of multiple redundant mechanisms having evolved to maintain plasma cells.

Targeting chemokine receptors and adhesion factors to deplete plasma cells is an attractive idea given the importance of plasma cell survival niches. Additionally, while other cell types including hematopoietic stem cells (HSCs) and neutrophils use the same adhesion factors and chemokine gradients, their survival is not dependent on niche residence thus rendering treatment potentially somewhat selective.

Plerixafor is a drug that interrupts the interaction of CXCL12 with CXCR4 and was originally developed to inhibit HIV infection of CXCR4⁺ T-cells ¹³². It has since been used to mobilize HSCs from the marrow for HSC transplantation ¹³³. Given the reliance on this interaction for plasma cell homing and maintenance in the bone marrow, we and others have hypothesized that plerixafor would mobilize plasma cells to the periphery and induce death by growth factor deprivation in a subset of plasma cells while potentially sensitizing others to drug-induced cell death by agents such as proteasome inhibitors.

To this end, our group recently concluded a clinical trial evaluating the ability of plerixafor to mobilize plasma cells into the peripheral blood and sensitize them to proteasome inhibitor treatment for plasma cell depletion in highly sensitized patients awaiting transplantation ¹³⁴. Our data show plerixafor mobilizes plasma cells into peripheral blood; however, the degree of mobilization varies significantly between patients, a phenomenon also observed when plerixafor is used for CD34⁺ stem cell mobilization ¹³⁵. In addition, it is possible that plasma cells may be mobilized out of their protective niches, but remain in the bone marrow, a condition we refer to as “micromobilization”. Quantitating plasma cell mobilization is challenging, in part because if plasma cell mobilization induces apoptosis, the cells may be only transiently detectable in the peripheral blood. This possibility is supported by the significant decrease in the population of bone marrow plasma cells that we observed in a patient that did not show significant increases in peripheral blood plasma cells. In the study, we obtained bone marrow aspirates at varying times in relation to plerixafor and bortezomib administration. Bone marrow aspirates were obtained after four doses of plerixafor alone, after four doses of plerixafor and one dose of bortezomib, and after six doses of plerixafor and two doses of bortezomib. In all four patient groups, we noted greater than a 50% decrease in the number of BMPCs.

In summary, what we know from studies of HLA sensitized dialysis patients awaiting transplantation is that plerixafor and/or bortezomib administration results in: 1) macromobilization of BMPCs into the peripheral blood with evidence of apoptosis in mobilized plasma cells and 2) significant loss of plasma cells in the bone marrow

Our data stand somewhat in contrast to a 2017 mouse study by Moore et al which failed to show a robust effect of plerixafor on plasma cell mobilization or sensitization to proteasome inhibitors ¹³⁶. The authors found maximal mobilization of plasma cells into the peripheral blood one hour after treatment with return to baseline by 24 hours. While there was a significant increase from baseline, the estimated number of plasma cells per mL of blood (almost the entire blood volume of a mouse) was only 500,000 cells and there was no detectable loss of plasma cells in the marrow or spleen in response to plerixafor alone. Further, the authors concluded the mobilized plasma cells were originating from the spleen as splenectomy one hour prior to treatment abrogated mobilization. They also attempted to combine plerixafor with bortezomib but were unable to observe additional plasma cell depletion beyond what they see with bortezomib alone in either the bone marrow or spleen.

A potential explanation for the lack of synergy between plerixafor and bortezomib treatments is that mobilizing plasma cells from their niche results in metabolic alterations that are unable to support the degree of antibody production that normally renders plasma cells acutely dependent on a functional proteasome. This hypothesis is supported by data from Lee’s group using an in vitro culture system designed to support BMPC survival ¹¹⁷. After fluorescence activated cell sorting, they found plasma cells were in a stunned state defined by limited antibody secretion for the initial 24 hours after sorting. However, when cultured in supernatants from mesenchymal stromal cell cultures, they regained their secretory capacity over the subsequent 24 hours in culture ¹¹⁷.

Consistent with our observations, a 2018 study from Cheng et al. used a mouse model of spontaneous systemic lupus erythematosus (SLE; model NZB/W) and found that plerixafor treatment significantly decreased the number of plasma cells in the bone marrow and spleen ¹³⁷. Using BrdU labeling during treatment, the authors distinguished the response of newly formed (BrdU+) and “long-lived” (BrdU-) plasma cells and found the effects of plerixafor to be more pronounced on the long-lived population which is expected given their association with CXCL12⁺ cells in the bone marrow. The finding with the greatest clinical potential from the Cheng et al. study was their long-term use of plerixafor after initial depletion of plasma cells with the proteasome inhibitor bortezomib. While the combination of bortezomib with plerixafor gave minimally better depletion than bortezomib alone, continued dosing of plerixafor for the subsequent four and a half months significantly decreased repopulation of the bone marrow plasma cell population by memory B-cells differentiating into plasmablasts or ongoing humoral responses, which has been a significant problem in plasma cell depleting desensitization and AMR regimens to date ^138–140.

In contrast to the Moore et al study that treated with 1mg/kg and tested up to 5mg/kg of plerixafor, Cheng et al. used a dose of 20mg/kg. This difference in dosing is a potential explanation for the disparate results however, another potential explanation is the differences in the plasma cell dynamics of the two mouse models. In the Moore et al. study, mice were unsensitized (i.e., “naive”) and in the Cheng et al. study the mice spontaneously developed the antibody-mediated disease lupus. With the lupus model, plasma cells are constantly being generated and could be more sensitive to CXCL12/CXCR4 chemoattraction to and survival signaling in the niche.

Targeting the CXCL12/CXCR4 axis as a plasma cell depletion approach has mixed results in mice, yet promising results in humans that warrant further exploration in carefully designed clinical trials. Effects of CXCR4 blockade on AMR, which has not yet been studied, may be even more impactful by limiting the ability of plasmablasts to enter the niche and receive survival signals. In this same vein, CXCR4 blockade could potentially be used as a prevention strategy for de novo DSA development by inhibiting the establishment of DSA-secreting LLPCs in the bone marrow.

Targeting adhesion factors that directly maintain plasma cells in the niche and support their survival is another attractive complementary approach to plasma cell depletion. The adhesion factors lymphocyte function-associated antigen 1 (LFA-1) and very late antigen 4 (VLA-4) are highly expressed on BMPCs and their ligands, intercellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion molecule 1 (VCAM-1), are highly expressed on bone marrow stromal cells. In addition to their role in physically anchoring plasma cells to the niche, signaling through VLA-4 and LFA-1 promotes survival through phosphatidylinositol-3-kinase (PI3K) activation and suppression of the transcription factor Forkhead-Box-Protein O1/3 (FOXO1/3) as well as inhibition of caspases 3 and 7 ¹⁴¹. Indeed, a 2008 publication from DiLillo et al. found treatment with an anti-LFA-1/VLA-4 monoclonal antibody depleted 70% of bone marrow plasma cells ¹⁴². However, upon cessation of treatment, the BMPC population was re-seeded by differentiating memory B-cells as evidenced by the lack of repopulation with concurrent anti-CD20 treatment. Unfortunately, clinical desensitization studies with proteasome inhibitors following rituximab therapy have demonstrated substantial antibody rebound following cessation of therapy, indicating that B-cell depletion in humans is not sufficient to prevent rebound, and possibly BMPC repopulation. These discrepant observations may lie in the partial depletion of memory B-cell populations seen with rituximab in humans ¹⁴³.

4.2. Survival signals

In addition to contact-dependent survival signals, mesenchymal and hematopoietic cells support plasma cell survival through secreted factors, most notably a proliferation induced ligand (APRIL) and interleukin 6 (IL-6) as shown using gene-targeted knockout (KO) mice.

APRIL is a member of the tumor necrosis factor superfamily, along with B-cell activating factor (BAFF), both of which signal through B-cell maturation antigen (BCMA) and transmembrane activator and calcium-modulator and cyclophilin ligand interactor (TACI), receptors that are highly expressed on plasma cells. APRIL has a greater affinity for BCMA than BAFF. BAFF additionally signals through the BAFF receptor but it is expressed at lower levels on plasma cells ¹⁴⁴. BAFF is important for plasmablast differentiation but unlike APRIL, it is not critical for long-term maintenance of plasma cells ¹⁴⁵. While the molecular mechanisms of APRIL-driven survival are incompletely defined, signaling through BCMA leads to nuclear factor kappa B (NFκB) signaling and inhibition of caspase 12 as well as expression of the anti-apoptotic proteins Mcl-1, Bcl-2, and Bcl-x_L ^{141, 146}.

Given the importance of APRIL and BAFF signaling for plasma cell survival, inhibiting these cytokines has been attempted to deplete plasma cells. Moreover, BAFF signaling is critical for B-cell survival thus targeting BAFF could help decrease repopulation of the plasma cell compartment by memory B-cells. In transplantation, a rapid and marked elevation in serum BAFF levels is noted after B-cell depletion including following alemtuzumab (anti-CD52, targeting T- and B-cells) induction ¹⁴⁷ and also following B-cell depletion with rituximab (anti-CD20) ¹⁴⁸. This is attributed to a decrease in the consumption of BAFF and is hypothesized to contribute to repopulation of B-cells, plasma cells, and DSA thereby likely contributing to antibody “rebound” following depletion-based antihumoral therapies (e.g. with proteasome inhibitors) ¹⁴⁹.

Atacicept is a fusion protein containing the extracellular TACI domain and the Fc portion of human Ig and acts as a decoy receptor by binding APRIL and BAFF to prevent their interaction with TACI and BCMA on the surface of plasma cells and B-cells ¹⁵⁰. To study the role of BAFF in this context and the potential for atacicept to decrease de novo DSA formation, AMR development, and increase allograft survival, Knechtle and colleagues treated non-human primates (NHPs) with atacicept in their model of kidney transplant AMR ¹⁵¹. Transplanted NHPs were treated with a T-cell depleting drug regimen that leads to de novo DSA and antibody mediated allograft injury: anti-CD3, alefacept (LFA3-Ig), tacrolimus, and methylprednisolone with the experimental group additionally receiving atacicept. They found a reduction in early DSA formation during treatment but thereafter, DSA titers increased similar to control animals. The delayed DSA formation was associated with decreased AMR and extended allograft survival which was not statistically significant, though the study was underpowered, given the use of NHPs.

Ultimately, signals emanating from BAFF-R, TACI, and BCMA (along with other cytokine receptors) drive plasma cell survival by modulating the balance of pro- and anti-apoptotic Bcl-2 family members. As plasma cells differentiate from B-cells, their dependence also switches from Bcl-2/Bcl-xL driven survival to Mcl-1-driven survival. Indeed, treatment with a Bcl-2/BclxL specific inhibitor only minimally depleted plasma cells, while genetic loss of Mcl-1 led to a substantial loss of plasma cells ^{146, 152–154}. Thus, it is possible that therapeutic approaches that target Mcl-1 could drive substantial loss of plasma cells. However, one word of caution is that Mcl-1 acts as an essential survival factor in almost all cells as it interacts with the downstream mediators Bax and Bak to prevent apoptosis ¹⁵⁵. Thus, inhibition of Mcl-1 may drive the death of many cell types and will likely not be selective for plasma cells. In addition, Mcl-1 levels are largely controlled post-translationally, so potential synergy by combining Mcl-1 inhibition with proteasome inhibition may have untoward effects on Mcl-1 turnover.

Interleukin 6 (IL-6) was identified in the 1980s by Toshio Hirano, Tadamitsu Kishimoto, and colleagues as a T-cell derived driver of B-cells into antibody secreting cells ¹⁵⁶. Since its discovery, IL-6 has come to be appreciated as a pleiotropic cytokine involved in inflammatory pathologies, response to infection, metabolism, and other homeostatic functions ¹⁵⁷.

The importance of IL-6 for plasma cell differentiation and survival was initially identified from in vitro experiments showing addition of recombinant IL-6 in culture drives peripheral blood plasmablast differentiation into mature plasma cells similar to those in the bone marrow ¹⁵⁸. A key source of IL-6 is likely bone marrow stromal cells as addition of anti-IL-6 to stromal cell culture supernatant or the use of supernatant from IL-6 deficient mice, significantly reduces the survival effects of these supernatants ^158–160. In vivo, IL-6 is primarily critical in early humoral responses by driving B-cell proliferation and differentiation from immature B-cells to plasmablasts in germinal center reactions. IL-6 secreted by plasmablasts feeds back to drive T follicular helper cell differentiation which supports ongoing humoral immune responses ¹⁶¹. Mice deficient in IL-6 have a delayed primary antibody response but eventually accumulate equal numbers of antigen specific plasma cells, antibody titers, and secondary responses similar to wildtype controls ^{159, 162, 163}. An explanation for the disparate in vivo and in vitro results is redundancy that does not exist in stromal cell supernatants, for instance cell-cell contact, hypoxia, or signaling through CXCR4. These data illustrate the challenges in in vitro modeling of the BM niche – that in vivo data are not always recapitulated in in vitro experiments. Nonetheless, these data suggest IL-6 is critical in driving B-cell maturation and survival to the point of plasma cell generation but is not required for long term plasma cell survival in the marrow. Thus, from the standpoint of targeting the humoral arm of alloimmunity, IL-6 directed therapies will likely be most beneficial for thwarting de novo DSA formation and anamnestic responses and may have limited influence on chronic AMR and chronically sensitized states driven by LLPCs, unless combined with other plasma cell targeted therapies.

In patients with kidney disease, IL-6 levels are often elevated from kidney damage, hemodialysis, and the underlying etiology including SLE and diabetes ^{164, 165}, reviewed in references ^{166, 167}. IL-6 is also elevated immediately following surgery and in patients undergoing acute rejection ^168–170. Given the association of elevated IL-6 with AMR and other antibody mediated diseases, targeting IL-6 has been attempted by both IL-6 neutralization and IL-6R blockade. Tocilizumab (TCZ) is a monoclonal antibody (mAb) against the IL-6 receptor (IL-6R) and inhibits classical and trans IL-6 signaling. TCZ has shown promise in treating complicated rheumatoid arthritis and is currently in clinical trials for pre-transplant desensitization and AMR treatment.

In 2015, a pilot desensitization trial using TCZ in highly sensitized patients awaiting kidney transplant showed promising results in highly sensitized patients awaiting kidney transplant who had not responded to at least two attempts using the standard desensitization regimen of IVIG and rituximab, ± plasmapheresis ¹⁷¹. Patients were treated with a single dose of IVIG followed by monthly TCZ for six months. The mean wait-time for transplantation decreased three-fold from the first round of desensitization. Five of the eight patients remaining in the study were transplanted and continued TCZ monthly for six months in addition to standard immunosuppression. While results must be cautiously interpreted given the small sample size, open-label use, and lack of a placebo control, the study found modest decreases in DSA at the time of transplant which continued to decrease after transplantation. No immunodominant DSA (the most abundant DSA at diagnosis) were detected at 12 months post-transplant (i.e., six months after TCZ cessation) except for one patient who had two weak DSAs. One patient had AMR at 12 months post-transplantation with detection of antibodies to AT1R but no anti-HLA DSA. Thus, although promising, more work is necessary to determine if targeting IL-6R signaling could desensitize patients to the point where they can accept an allograft without AMR.

A 2017 single center open-label case study used TCZ to treat chronic AMR ¹⁷². Thirty-six kidney transplant patients with AMR unresponsive to standard treatment with IVIG and rituximab ± plasmapheresis were treated monthly with TCZ for 6–25 months and followed for up to 8 years (mean 3.26 years). The authors found an allograft survival probability of 80% at 6 years post AMR diagnosis which is promising compared to other estimates of <40% allograft survival probability at 6 years post de novo AMR diagnosis ¹⁷³. Importantly, the four patients who lost their allografts had discontinued TCZ treatment six months prior for medical or insurance reasons. Because inhibition of the IL-6 receptor leads to accumulation of IL-6, it’s possible that upon treatment cessation or during non-adherence, the abundant IL-6 now free to bind its receptor could support a humoral response to the allograft, resulting in de novo DSA formation. The pleiotropic function of IL-6 could also drive rejection through its action on other immune cells and direct endothelial activation in the allograft. Understanding the cellular and molecular outcomes following anti-IL-6R treatment cessation will be critical to evaluating its safety and long-term effectiveness. As well, interrogating patient bone marrow to determine whether IL-6R neutralization drives functional defects in plasma cells or whether it drives their deletion is important to our mechanistic understanding.

The authors of this study subsequently published a retrospective analysis determining serum antibody titers before and after TCZ treatment compared to controls who received standard AMR treatment ¹⁷⁴. They found TCZ reduced total IgG, IgG subclasses 1 and 3, and anti-HLA antibodies (total IgG and IgG3) whereas standard IVIG-based treatment did not. The reduction in IgG isotypes 1 and 3 is clinically important as they are capable of binding C1q and Fc receptors. ⁸⁴.

A caveat of interpretation for these case studies is that patients were previously treated with the B-cell depleting agent, rituximab, amongst other medications including corticosteroids. To address this, a case study published in 2020 treated 15 patients who had chronic AMR with TCZ. None of the patients had received any prior AMR treatment, all had anti-Class II iDSA, and most had severe transplant glomerulopathy and microvascular inflammation at the time of diagnosis. At 6-month protocol biopsy, patients had significant improvement of inflammation and no progression of transplant glomerulopathy or other pathologic features. One patient experienced allograft loss 20.5 months after starting TCZ treatment. Immunodominant DSA were significantly decreased after six months of treatment, however, most remained above the 10,000 median fluorescence intensity cutoff on the single antigen bead assay which is associated with poorer allograft survival probability ¹⁷⁵. AT1R antibodies also decreased by approximately half after six months of treatment and completely disappeared in six patients. Overall, the study found TCZ to be beneficial as a first line treatment for AMR and that pre-treatment with rituximab did not appear to be critical for its effect. This study highlights an important consideration in developing AMR treatments. While a multi-target approach is necessary to ameliorate ongoing damage and prevent future damage, thoughtful selection of agents is critical in an already-immunosuppressed population.

Another approach to target the IL-6 axis is by directly inhibiting IL-6 using a monoclonal antibody against the cytokine. The proposed benefits of targeting IL-6 directly are 1) mAb binding to IL-6 should lead to its clearance and avoid cytokine signaling rebound; 2) the treatment doesn’t induce ADCC or receptor cross-linking; and 3) the anti-IL-6 mAb, Clazakizumab, is the longest acting treatment to target IL-6 signaling with a half-life of 30 days ^{176, 177}. Notably, multiple factors determine if antibody-cytokine complexes get cleared or recycled. Because degradation through FcγR-mediated endocytosis depends on many factors, in particular the complex size, cytokine-directed antibodies will actually prolong the cytokine half-life if the complex isn’t favorable for degradation. Monomeric complexes are more often recycled through the neonatal FcR whereas larger complexes are cleared through the FcγR. Methods developed to overcome cytokine signaling rebound by increasing clearance include using a polyclonal mixture of anti-cytokine antibodies and using bispecific antibodies that bind to two different cytokine epitopes ¹⁷⁸.

Results from a phase II randomized pilot study (N=20) using clazakizumab to treat chronic AMR were published in early 2021 ^{176, 177}. The study found all patients treated with clazakizumab had a modest reduction in iDSA and when DSA titers were interpolated from dilution experiments, the decrease in DSA was more significant. The traditional measurement of DSA by MFI in single antigen bead assays does not directly measure serum titers but is also influenced by antibody binding avidity and affinity for the antigen on the bead thus also calculating dilution-based titers is more informative ¹⁷⁷. Although no effect of clazakizumab was observed at 11 weeks, by 51 weeks, 38% of patients had a negative histological AMR score. Notably, 25% of patients in the treatment arm experienced serious infectious events highlighting the need for careful selection and monitoring of patients. There was also a 20% incidence of colonic diverticular bowel perforation, an observation that if confirmed may substantially hinder the future development of clazakizumab ¹⁷⁷. A phase III multicenter randomized controlled trial of Clazakizumab to treat chronic AMR is underway and completion is estimated for 2028 (NCT03744910).

4.3. Proteostasis

B-cell differentiation into plasma cells, requires substantial expansion of rough ER and Golgi apparatus in order to acquire the capacity needed to synthesize and secrete thousands of antibodies per second ¹⁷⁹. This massive expansion of protein synthetic capacity is driven by spliced X-box binding protein 1 (sXBP-1), a key transcription factor in the unfolded protein response (UPR). As protein synthesis ramps up, misfolded proteins accumulate because of protein folding errors. These misfolded proteins are toxic to the interior of the ER, as they expose hydrophilic residues that are normally internalized in a properly folded molecule. While the true error rate of protein synthesis (i.e. protein misfolding) remains unknown, even if a conservative estimate of 0.01% of all proteins having an error in misfolding, assuming 5,000 antibodies per second, that results in 30 misfolded proteins every minute ^{179, 180}. The UPR is a highly complex physiologic response that utilizes several approaches for dealing with misfolded proteins. These approaches include increased production of foldases and chaperones in an attempt to fix the misfolded proteins, reduction of protein synthesis to reduce the rate of misfolded protein generation, and induction of degradation pathways including the ubiquitin proteasome system (UPS) and autophagy lysosomal pathway. In conjunction with the UPS, cells induce the ER associated degradation (ERAD) pathway to facilitate the transport of misfolded proteins out of the ER and target them to for degradation by the proteasome through ubiquitination (as reviewed ¹⁸¹). If these measures are insufficient to restore proteostasis, the UPR will eventually induce apoptosis.

During the humoral response to infection or vaccination, germinal center resident B-cells undergo a fate decision to become either a memory B-cell or an antibody-secreting plasmablast. Plasmablasts are critical to a rapid, abundant production of antibody to quell the infection but to limit the humoral response, most of them die ¹⁸². Cenci’s group has proposed that the imbalance in proteasome capacity vs load is a major mechanism that limits the humoral response ^{183, 184}. This is based on the observation of a paradoxical reduction in proteasome capacity when plasmablasts are at the height of antibody secretion (i.e., at the time of peak need in proteasome capacity) ¹⁸⁵. How a subset of cells survives this imbalance or whether the surviving subset never experiences the imbalance is currently unknown. There are several non-mutually exclusive potential mechanisms that favor plasmablast survival including: 1) induction of autophagy that provides additional protein degradation capacity; 2) metabolic changes that increase proteasomal capacity or decrease protein synthesis; or 3) changes in Bcl-2 family members that raise the cell’s threshold for apoptosis. Importantly, each of these scenarios could be driven by factors in the bone marrow niche including hypoxia, cell contact, and APRIL.

Autophagy is essential in maintaining proteostasis in plasma cells. Cenci’s group has also demonstrated that plasma cells require autophagy for long term survival via “ERphagy” that regulates ER mass, and therefore protein synthetic capacity ¹⁸⁶. In the absence of ER mass regulation via autophagy, plasma cells have an increased ER mass and therefore greater antibody synthetic capacity. An important corollary consideration to ER stress is the concomitant oxidative reductive and energy stress that accompanies immunoglobulin production. Immunoglobulin synthesis requires considerable folding and formation of several cysteine bonds to achieve proper tertiary conformation. If the energy state of the plasma cell is not sustainable, the cell will eventually succumb to death due to energy imbalance and oxidative/reductive stress ^{187, 188}.

The autophagy lysosomal system and ubiquitin proteasome system were long thought to be two distinct processes involved in proteostasis; however, evidence has accumulated indicating that they are interrelated processes ^{189, 190}. First demonstrated in Drosophila and subsequently mammalian cells including neurons and cardiomyocytes, autophagy compensates for proteasome dysfunction by degrading aggregates of ubiquitinated proteins in a process known as “aggrephagy” ^190–192. Importantly, aggrephagy has been demonstrated to mediated proteasome inhibitor resistance in some multiple myeloma lines ¹⁹³. However, to our knowledge, the ability of autophagy to compensate for proteasome dysfunction in nonmalignant plasma cells has not yet been demonstrated. Given the importance of autophagy in cell maintenance and the induction of autophagy related genes during the UPR, we expect that plasma cells would employ aggrephagy during proteasome dysfunction. It’s also possible that ERphagy (or a combination of both) continues and consequently decreases ER stress but does not directly compensate for the proteasome. Taken together, targeting autophagy is another potential strategy to deplete plasma cells, either in combination with proteasome inhibitors or alone.

Proteasome inhibitors were discovered in the 1990s upon realization that the proteasome was a major mechanism of catabolism ¹⁹⁴. Subsequently, proteasome inhibitors were found to effectively treat the plasma cell cancer, multiple myeloma, and other hematologic malignancies ¹⁹⁵. Our group, in a search for agents that may be useful in targeting plasma cells to treat antibody mediated diseases, including transplant rejection, reasoned that proteasome inhibitors, because of their known effects on myeloma and also normal lymphocyte populations, were the best available candidates to target plasma cells therapeutically. Our group was the first to employ plasma cell targeted therapies in humans (i.e., nontransformed plasma cells) (which was also the first species in which plasma cell targeting was employed therapeutically), using the proteasome inhibitor bortezomib ^{9, 138, 196, 197}. In this initial experience, six kidney transplant recipients experiencing IVIG-refractory late AMR or late mixed acute rejection were treated with a bortezomib-based regimen that including rituximab and plasmapheresis ¹⁹⁶. Each patient experienced a rapid decline in DSA levels and also improvements in allograft inflammation, including a patient with vascular endotheliitis. Importantly, as experience with bortezomib increased, its limitation by peripheral neurotoxicity was recognized ¹⁹⁸. This was further explored in an iterative-design desensitization trial, where bortezomib exposure and dosing density was incrementally increased between treatment groups ¹³⁸. In this trial, we concluded that bortezomib possessed considerable potency in reducing anti-HLA antibody levels in a dose dependent manner. Given the neuropathy limitation of bortezomib, further improvements were needed to optimize effectiveness using bortezomib-based combination therapy, or alternatively more potent and less toxic proteasome inhibitors. Therefore, we initiated a desensitization trial with carfilzomib, which is an irreversible proteasome inhibitor. This trial, which is ongoing, has demonstrated increased potency and reduced toxicity compared to bortezomib ^{139, 199, 200}. Our group found >65% depletion of bone marrow plasma cells following carfilzomib treatment. RNA-sequencing analysis of bone marrow plasma cells before and after treatment, revealed an increase in the expression of immunoproteasome subunits in cells surviving treatment, suggesting proteasome plasticity as a potential resistance mechanism ^{139, 200}.

Studies from other groups have been less promising with proteasome inhibitors. As an example the randomized trial, Bortezomib in Late Antibody Mediated Kidney Transplant Rejection (BORTEJECT) ²⁰¹ demonstrated a lesser effect than the trials conducted at the University of Cincinnati. The BORTEJECT trial (N=44) evaluated the effectiveness of bortezomib as a single agent therapy (in addition to maintenance immunosuppression) on patients with chronic AMR. In this trial, the authors found no significant difference in renal function, DSA, or histological features of AMR. Several potential explanations exist for the lack of detection of a treatment effect. First, patients were treated with bortezomib monotherapy, without concomitant corticosteroid dosing, which is thought to potentiate proteasome inhibitor treatment in the myeloma field (ESW). Second, the two bortezomib cycles were often separated in time by several weeks or more, and the interval between cycles was not precisely defined, which likely led to marked reduction in dosing density in a number of patients. Finally, the authors did not employ dilutional analysis to determine whether HLA single antigen beads were saturated, which could have precluded detection of significant reductions in Anti-HLA antibody levels.

According to Cenci’s proteasome load/proteasome capacity ratio, BMPCs with low immunoglobulin synthesis rates should have low ER stress, and therefore may not be susceptible to terminal UPR induction by proteasome inhibitors. Further study of the sensitivity of individual plasma cells to proteasome inhibitors is needed to gain insight into this issue. In addition, studies of other compensatory mechanisms such as autophagy are needed to sort out how plasma cells adapt and cope with acute ER stress.

Lastly, it is important to highlight that drugs developed to target specific pathways such as proteasome inhibitors may also disturb the expression of other survival factors. In a rat model of allograft transplantation, Li et al. found treatment with the immunoproteasome-specific inhibitor, ONX-0914, was associated with decreased plasma cell expression of VLA-4 and LFA-1, CXCR4, IL-6R, and BCMA ²⁰². A significant reduction in the expression of APRIL and IL-6 protein in the bone marrow was associated with decreased eosinophil counts. The authors also found decreased Mcl-1 protein in the plasma cells of ONX-0914-treated rats which they attributed to decreased BCMA signaling. This work highlights the importance of studying not only the plasma cell response to treatment, but also the effect on the plasma cell microenvironment, i.e., the cellular components of the BMPC niche that are critical to plasma cell survival.

5 |. CHALLENGES IN CLINICAL TRANSPLANTATION

Plasma cell depletion regimens for desensitization and AMR treatment are relatively simple in design, however, clinical implementation is more complex. This section will place plasma cell targeting regimens in a clinical context as some considerations are unique to transplantation.

5.1. Concurrent immunosuppression, co-morbidities, nephrotoxicity

While insufficient immunosuppression can underlie the development of de novo DSA, patients undergoing AMR treatment remain on multiple immunosuppressants that require consideration when plasma cell targeted therapies are implemented. As producers of neutralizing antibodies, plasma cells are the backbone of most successful vaccines to date and if depleted, patients may experience an increased vulnerability to infection. This vulnerability has been emphasized with the coronavirus disease 19 (COVID-19) pandemic. Recent data show only 15% of transplant recipients who received two doses of SARS-CoV-2 mRNA vaccine mounted an antibody response after both doses, 39% developed antibodies only after the second dose and to a lesser degree, and 46% failed to mount any antibody response ^203–205. Importantly, the use of antimetabolite immunosuppressants and belatacept (CTLA4-Ig) portended worse antibody responses. These data contrast with the 95% seroconversion rate found in immunocompetent recipients of the mRNA vaccines during clinical trials ²⁰⁶. As plasma cell targeted therapies progress and plasma cell depletion becomes more efficient, hypogammaglobulinemia may become a limiting factor as a result of increased infection risk. To date, although plasma cell depletion efficiencies as high as 80% or more have been achieved, infection risk (other than herpes infections which are known to be associated with proteasome inhibitors) have not been treatment limiting.

The ability to target antigen specific plasma cells would represent an important achievement. A potential approach to label and deplete antigen specific plasma cells was described by Cheng et al. in 2020 with proof of concept demonstrated in an OVA-immunized mouse model ²⁰⁷. The approach uses an antibody to a highly expressed plasma cell surface protein (e.g., CD138) that is conjugated to the antigen, in this case OVA. The anti-CD138-OVA antibodies coat the surface of all plasma cells. For those that secrete anti-OVA antibodies, the anti-OVA antibodies will bind to the OVA on the surface of the anti-CD138-OVA coated plasma cells and cause complement dependent cytotoxicity and/or ADCC. With the degree of polymorphism of HLA, translating this to human allosensitization would be difficult but likely an improvement over depleting plasma cells non-selectively.

In addition to immunosuppression, renal transplant recipients typically have underlying co-morbidities including SLE or diabetes. The effect of these ongoing diseases and their required treatments must also be considered, especially the antibody mediated diseases like SLE.

For desensitization, patients with end stage renal disease are undergoing intermittent dialysis, which can affect pharmacokinetics of individual drugs, which must be taken into account. As well, for both pre- and post-transplant patients, drug metabolism and nephrotoxicity are important considerations.

5.2. Ongoing antigen exposure

The immunobiology of B-cell and T-cell responses derives largely from studies in animal models that are often well defined. In contrast, clinical trials occur in patients who may be undergoing treatments such as chronic immunosuppression and or dialysis treatments that are not often duplicated in animal models. For example, a patient with a rejected kidney may return to dialysis, yet there is long term chronic ongoing inflammation in the rejected kidney. How infection, drug metabolism, fluctuations in immunosuppression, or other factors affect the immunogenicity of the donor tissue or the immune system’s “perception” of the organ on the cellular and molecular level are largely unknown and not replicated in animal models. Understanding how these factors contribute to the response to therapies is critical to developing treatments with long-term effectiveness. For instance, while depletion of DSA and allospecific memory B-cells may be achievable with a relatively short course of treatment to desensitize a patient prior to transplantation, the issue of continued antigen exposure in AMR patients makes this a much greater challenge and likely requires the long-term use of anti-humoral therapies.

5.3. Antibody Rebound

A significant barrier to long term effectiveness of plasma cell targeting is the reappearance of antibody, a phenomenon commonly referred to as antibody “rebound”^{138, 139}. To date, most clinical studies on the effectiveness of plasma cell targeting therapies for desensitization or AMR treatment rely on serum DSA as their readout for the effect on plasma cells. Rebound is typically described as the observation that DSA serum titers will nadir around the time of cessation of plasma cell targeted therapies, only to increase thereafter, reaching levels that may be below or near the baseline levels prior to treatment. Knechtle’s group found that single agent bortezomib for desensitization of NHPs significantly depleted BMPCs but also induced humoral compensation as noted by naïve and memory B-cell proliferation in secondary lymphoid tissues ²⁰⁸.

Other non-mutually exclusive potential explanations for antibody rebound following cessation of plasma cell targeted therapies, particularly proteasome inhibitors, include 1) reacquisition of antibody production and secretory capacity in remaining plasma cells; and 2) plasma cell repopulation via the bone marrow plasma cell population or via extramedullary sources via antigen-independent or antigen-dependent mechanisms ²⁰⁸. Thus, the phenomenon of rebound will be a significant hurdle that needs to be addressed in any successful plasma cell targeted therapy.

6 |. TECHNIQUES TOWARD PROGRESS

Here, we briefly summarize several techniques that we believe will move the field forward. A major limitation to progress in desensitization and AMR treatment are the difficulties in performing sufficiently powered randomized controlled trials given the variability in patient treatment and generally small sample sizes. However, we can carefully choose methodologies that will maximize the amount of information we gain from biological samples in any study design. Because mechanistic studies are limited in humans, both in vitro studies using mouse and human cells along with in vivo animal models (from mouse to NHPs) are critical for uncovering biological responses to potential new drug therapies.

6.1. Sampling human bone marrow

For several decades, endpoints in desensitization and AMR trials have been serum DSA levels, renal function (estimated glomerular filtration rate (eGFR)), and allograft histology (AMR patients). These parameters are valuable to our overall understanding of the therapy’s effect on disease outcome, but they provide little information regarding mechanisms favoring a response or a lack of response to the therapy. For instance, in a NHP study using bortezomib as a single agent therapy for desensitization, no change in DSA was observed but bone marrow aspirates revealed a significant depletion of BMPCs ²⁰⁸. Likewise, in human studies that lack bone marrow sampling, no loss of DSA could be interpreted as the drug having no impact on plasma cells, when a sampling of the bone marrow may reveal the opposite. Thus, careful monitoring of DSA after substantial plasmapheresis as well as an investigation of cells in their target organs will reveal more mechanistic information from which to make future decisions based on the science.

In addition to aiding in understanding whether agents deplete BMPCs, bone marrow sampling before and after plasma cell targeting treatments enables more sophisticated analyses such as single cell RNA sequencing to analyze heterogeneity in plasma cell populations and identify changes in gene expression patterns before and after treatment, thereby potentially providing insight into compensatory mechanisms induced by plasma cell targeting. Combining single cell RNA sequencing with assay for transposase-accessible chromatin (ATAC) sequencing to look at chromatin landscape expands our view of how different treatments affect gene regulatory networks.

Spatial transcriptomics is a new technique that delivers gene expression data in the context of tissue structure. Given the importance of the niche in plasma cell identity and survival, appreciating changes in organization and pockets of gene expression may enhance our understanding of plasma cell niche constitution, the dynamics of plasma cell niche residence, and effects on other niche-resident cell populations. These data could eventually be leveraged to develop drug delivery systems that target specific areas of the niche. For instance, if spatial transcriptomics analysis of bone marrow biopsies revealed plasma cells surviving treatment were concentrated in a particular area of the marrow and had a unique hypoxia gene signature, hypoxia-targeted drug delivery systems could be attempted. We are currently working on developing spatial transcriptomics on bone marrow biopsies before and after proteasome inhibitor therapy. We also think that application of spatial transcriptomics will be particularly informative for AMR/MAR biopsies. The ability to connect histological findings and gene expression, particularly for the endothelium, will advance our understanding of AMR/MAR phenotypes and inform potential treatment strategies.

Importantly, deciphering the biology of plasma cell responses to therapy will necessarily need to go beyond gene expression analyses and examine protein expression. For example, the response to proteasome inhibition in plasma cells may largely occur at the post-translational level (e.g., autophagy, UPR activation) which is missed by only looking at mRNA and chromatin accessibility. A major challenge in all plasma cell research is obtaining enough cells to conduct deep proteomics research. In this vein, Myers et al. published a method in 2019 that uses a fluorescence activated cell sorting-based proteomics workflow using liquid chromatography-mass spectrometry to quantify over 7,000 proteins from only 150,000 cells ²⁰⁹. Unlike other low-input proteomic approaches, this one does not require highly specialized equipment or expertise and is thus amenable to most academic research institutions.

Lastly, the small sample sizes of human desensitization and AMR trials combined with interpatient variability make it difficult to confidently draw conclusions about drug responses. However, a major benefit of generating these large datasets is that we can perform analyses that leverage patient characteristics (e.g., co-morbidities, drug regimens) to define phenotypes of responses to treatment that can eventually be used to identify patients who are most likely to benefit from the evaluated treatment.

6.2. Mechanistic studies in animals

Combination therapy targeting each “step” of the humoral response will be critical for long term success of treatments. However, when testing drugs in desensitization or AMR trials, the combination of additional immunosuppressants and variations between therapies, makes it impossible to confidently attribute effects to a single drug. Modeling desensitization and AMR in NHP have been valuable in evaluating new approaches, and also in evaluating approaches that have been tried in humans by modifying human protocols(as reviewed in ¹⁰⁹). However, NHPs are expensive, and studies are often limited to small numbers of subjects, availability of reagents is limited, and genetic manipulation is currently not possible. Thus, combining what we learn from NHP models and human models with our capabilities in mouse for high-throughput evaluation of individual drugs and combinations on the humoral response to transplantation, plasma cell depletion, and DSA is critical.

An exciting development for murine studies is the recent generation and characterization of the first plasma cell-specific cre recombinase transgenic mouse. These mice express a tamoxifen-inducible cre recombinase (estrogen receptor T2 cre) under control of the joining chain (J-chain) promoter. The J-chain multimerizes IgM and IgA but is expressed in most plasma cells across all isotypes. Importantly, its expression is significantly greater in plasma cells than any B-cell subsets including germinal center B-cells (40-fold difference) ²¹⁰. Preceding these mice, most studies evaluating various genes and pathways important for plasma cell survival were done either through global knockouts or genetic manipulation at the level of B-cells. Obviously, there are huge caveats to interpreting of the role of a particular gene in plasma cells when it is knocked out prior to plasma cell genesis. Thus, we are anxiously awaiting data derived from studies utilizing J-chain cre mice to delete various genes at specific time points in plasma cell development.

7 |. CONCLUSION

Throughout the several decade history of transplantation, the importance of antibody mediated responses and preexisting anti-donor antibodies has been recognized. Notable advances in identifying and quantifying HLA antibodies, histologic examination, and therapeutic approaches have enabled clinical improvements in outcomes. However, much remains to be accomplished, and it is likely that murine models and advanced genomics and proteomic techniques will provide useful tools to enable targeting of more specific populations, including plasma cells.