Preoperative Carfilzomib and Lulizumab based desensitization prolongs graft survival in a sensitized NHP model.

Abstract

Sensitized patients are difficult to transplant due to pre-formed anti-donor immunity. We have previously reported successful desensitization using carfilzomib and belatacept in a non-human primate (NHP) model. Here we evaluated selective blockade of the co-stimulatory signal (CD28-B7) with Lulizumab, which preserves the co-inhibitory signal (CTLA4-B7). Five maximally MHC-mismatched pairs of NHPs were sensitized to each other with two sequential skin transplants. Individuals from each pair were randomized to either desensitization with once-weekly Carfilzomib (27mg/m² IV) and Lulizumab (12.5mg/kg SC) over four weeks, or no desensitization (Control). NHPs then underwent life-sustaining kidney transplantation from their previous skin donor. Rhesus-specific anti-thymocyte globulin was used as induction therapy and immunosuppression maintained with tacrolimus, mycophenolate, and methylprednisolone. Desensitized subjects demonstrated a significant reduction in donor-specific antibody, follicular helper T cells (CD4⁺PD-1⁺ICOS⁺), and proliferating B cells (CD20⁺Ki67⁺) in the lymph nodes. Interestingly, regulatory T cell (CD4⁺CD25⁺CD127^lo) frequency was maintained after desensitization in addition to increased frequency of naïve CD4 T cells (CCR7⁺CD45RA⁺) and naïve B cells (IgD⁺CD27⁻CD20⁺) in circulation. This was associated with significant prolongation in graft survival (MST = 5.8 ± 4.0 vs. 64.8 ± 36.3; p<0.05) and lower antibody-mediated rejection scores compared to control animals. However, all desensitized animals eventually developed AMR and graft failure. Desensitization with CFZ and Lulizumab improves allograft survival in allosensitized NHPs, by transient control of the germinal center and shifting of the immune system to a more naive phenotype. This regimen may translate into clinical practice to improve outcomes of highly sensitized transplant patients.

Introduction

Sensitized transplant patients have pre-formed antibodies to HLA antigens, usually from prior sensitizing events such as blood transfusion, pregnancy, or a previous transplant. These patients represent a growing challenge for the transplant community as their sensitized immune system makes it more difficult to find compatible potential donors. Indeed, only about 6.5% of sensitized patients with a panel reactive antibody (PRA) >80% receive a transplant yearly ¹. Despite the growing popularity of strategies such as kidney paired donation and recent changes to the allocation system which have demonstrably improved rates of deceased donor transplantation for sensitized recipients, the population of sensitized patients on the kidney transplant waitlist continues to grow with approximately 30% of waitlist patients being highly sensitized ^2–4. Therefore, there remains a group of difficult to transplant patients for whom desensitization may represent a suitable route to transplantation ⁵.

The challenge in transplanting such patients is effective, durable desensitization treatment. Conventional methods of desensitization have involved the use of intravenous immunoglobulin (IVIG), plasmapheresis, and less frequently rituximab to remove circulating donor specific antibody (DSA). While these methods have demonstrated acceptable short-term outcomes of 2-year patient survival of 95% and graft survival of 86%, long-term outcomes have been disappointing with higher rates of acute rejection and 5-year graft survival of 65-70% ^6–8. Recent data from the UK demonstrated no improvement in survival for sensitized recipients who were desensitized compared with remaining on the waitlist ⁹. These results have sparked interest in developing novel approaches to desensitization using pharmacological strategies to reduce circulating DSA, B cell depletion, and specific targeting of memory cell responses, which may have utility beyond the sensitized patient awaiting transplantation, and offer potential treatment options for post-transplant humoral rejection.

One such approach is targeting plasma cells using proteasome inhibitors that have been approved for use in multiple myeloma ¹⁰. Unfortunately, the use of bortezomib, a first-generation proteasome inhibitor, in desensitization regimens resulted in only modest effects on plasma cell survival and function ^11,12. Our studies in the allosensitized nonhuman primate (NHP) also revealed that desensitization with bortezomib monotherapy resulted in no improvement in allograft survival, and demonstrated rapid upstream compensation by the germinal center (GC) with consequent humoral rebound ¹³. Thus, our group proposed a ‘dual targeting’ desensitization approach that simultaneously targets both plasma cells and the germinal center for more effective desensitization. Indeed, our most recent research suggests that combining proteasome inhibition with costimulation blockade in the allosensitized recipient results in improved graft survival ¹⁴. However, while significantly prolonging graft survival, treatment with carfilzomib and belatacept failed to achieve durable desensitization and resulted in rebound of donor-specific antibody (DSA) post kidney transplantation that was associated with graft loss ¹⁵. Thus, our attention turned toward novel costimulation blockade agents with the potential to improve the efficacy and durability of our ‘dual targeting’ desensitization approach. We hypothesized, that the limited durability may be due to belatacept’s non-selective blocking of CD28 and CTLA4 signaling. CTLA4-Ig transmits a co-inhibitory signal to T cells through CTLA4, thus impacting the germinal center response ^16,17.

Lulizumab, a CD28 domain antibody antagonist, is a novel costimulation blocker that has a theoretical advantage over belatacept, in that it blocks CD28 costimulation while allowing native CTLA4 coinhibitory signaling ¹⁸. Due to its efficacy in preserving regulatory T cells, selective CD28 targeting is currently under investigation in a clinical trial of kidney allotransplantation to modulate Treg cells together with IL-6 receptor antagonist (NCT04066114). In non-sensitized NHP allotransplantation models, CD28-specific antagonists including lulizumab have demonstrated efficacy and in some cases superior results compared to belatacept ^19–23. Therefore, in this study, we hypothesized that dual targeting desensitization with carfilzomib (a second-generation proteasome inhibitor) and lulizumab would prolong renal allograft survival and prevent humoral rebound post kidney transplantation in allosensitized nonhuman primates.

Results

Desensitization with carfilzomib and lulizumab reduces donor-specific antibody levels in allosensitized nonhuman primates

To evaluate the efficacy of carfilzomib and lulizumab treatment for desensitization, we sensitized rhesus monkeys as previously described ^15,24. Briefly, naive fully MHC-mismatched rhesus macaque pairs exchanged skin transplantations twice from the same donor (swapping). To avoid bias affecting the selection of animals, we also randomized animals. Schematic representation of the Study Design showing the pairs of nonhuman primates (NHPs) sensitized to each other with skin transplant and then the individuals of each pair being randomized into two groups (control vs. desensitized; Figure 1A). Control animals did not receive pre-transplant desensitization while desensitized animals received carfilzomib (27mg/m², IV) and lulizumab (12.5mg/kg, SQ) weekly for one month before kidney transplantation. Both groups of animals received identical induction (rhATG, 20mg/kg) and maintenance immunosuppression including tacrolimus, mycophenolate mofetil (MMF), and steroids.

Figure 1. Pre-transplant desensitization with carfilzomib and lulizumab (anti-CD28dAb) reduce the level of donor-specific alloantibody.

(A) Study design and animal disposition. (B) Representative T cell flow crossmatch histogram of pre- (shaded area) and post-desensitization (clear area) for untreated control (red) and desensitized (green) animal. Pre-sensitized DSA levels are shown as the grey histogram. Post-desensitization DSA reduction is expressed as percent reduction based on pre-desensitization serum DSA level. (C) Representative flow plot reflecting the frequency of CD19+CD38+ plasma cells (gated on CD20-IgD-CD3-CD14-). NS indicates no significance.

Desensitized animals showed more significant reduction of circulating IgG DSA compared to the control animals (ΔMFI −693±79 vs −487±416, p<0.05; Figure 1B). Furthermore, we analyzed bone marrow (BM) aspirates for plasma cell (PC) frequency (CD3⁻CD14⁻CD20⁻CD19⁺CD38⁺), which revealed a trend of reduction after desensitization, which did not reach statistical significance (Figure 1C). Interestingly, the primate with the shortest survival in the treatment group is the only animal that had no reduction of BM plasma cells after desensitization with carfilzomib and lulizumab (anti-CD28dAb).

Reduced germinal center activity in allosensitized nonhuman primates after desensitization with CD28-specific blockade

To confirm the efficacy of lulizumab in controlling germinal center responses, we measured follicular helper T (Tfh) cell populations in peripheral lymph nodes (LN) before and after desensitization. We observed a significant reduction in CD4⁺PD-1⁺ICOS⁺ Tfh cells in the LN of recipients receiving desensitization with CFZ + lulizumab (p=0.02). Control primates that received no treatment showed no significant change in the LN Tfh cell populations during the same time period. When gating for CD4⁺PD-1^hiICOS⁺ Tfh cells, one treated-animal showed an increase in these activate Tfh cells and later developed anti-Class II dominant AMR (Figure 2A). We also evaluated Tfh (CD4⁺PD-1⁺ICOS⁺) cells in the peripheral blood. Interestingly, we identified a similar pattern of reduction in circulating Tfh cells in the treatment group (p<0.05; Figure 2B). In accordance with the reduction of Tfh cells, the proportion of proliferating Ki67⁺CD20⁺ B cells within the lymph nodes was also significantly reduced in those that received carfilzomib and lulizumab (Figure 2C). These flow cytometry findings correlated with lymph node histology, which demonstrated reduced germinal center activity in the lymph nodes after desensitization with carfilzomib and lulizumab (Figure 2D). Control animals without desensitzation did not show any significant reduction of Tfh cell, proliferating B cells, or germinal center response.

Figure 2. Carfilzomib and lulizumab treatment reduces follicular helper T cells (Tfh) and proliferating IgG B cells in the lymph node.

(A) Visualization of follicular helper T cells based on PD-1 and ICOS from the lymph node biopsies. GC-Tfh (PD-1^hiICOS⁺) cell did not show any difference while Tfh (PD-1⁺ICOS⁺) cells were significantly reduced in desensitized animals compared to untreated controls. (B) Visualization of circulating PD-1⁺ ICOS⁺ Tfh cells. PD-1^hiICOS⁺ GC-Tfh cell population does not exist in the peripheral blood. ICOS+PD-1+ Tfh cell were significantly reduced after desensitization. (C) Proliferated lymph node B cells before and after desensitization or time matched samples from control. Ki67⁺CD20⁺ B cells in the lymph nodes were greatly reduced after desensitization. Data represent mean ± SD of five animals before and after desensitization. (D) Representative immunohistochemistry of lymph nodes before and after desensitization (bottom panels) and time matched samples from control animal (top panels).

Lulizumab increases the proportion of circulating naive T and B cells while preserving regulatory T cells

We hypothesized that selective targeting of CD28 signal would modulate immune responses, preventing pathological maturation of effector cells and preserving regulatory T (Treg) cells. Therefore, we evaluated the kinetics of the Treg cell populations after a month of carfilzomib and lulizumab treatment. First, the frequency of B and T cells was not altered by the treatment (Supplemental figure 1). As shown in figure 3A, Treg cell populations (CD4⁺CD25⁺CD127^lo) in peripheral blood and LNs were largely unchanged (green bars). Control animals did not show any changes of Treg cell population in the same time period (red bars). Interestingly, Treg cells proportionally increased in the BM compartment after desensitization. We then compared the results from historic animals treated with carfilzomib and belatacept and observed that Treg cells were significantly reduced in all immune compartments after belatacept based desensitization (Figure 3A). We furthermore evaluated naïve and memory T cells in peripheral blood. Due to the masking effect of lulizumab on CD28 molecules, we characterized T cell subsets by CCR7/CD45RA instead of CD28/CD95. The proportion of naïve CD4 T cells (Tn; CD45RA⁺CCR7⁺) increased while central memory CD4 T cells (Tcm; CCR7⁺CD45RA⁻) decreased significantly after desensitization with carfilzomib and lulizumab (Figure 3B). Naïve and central memory CD8 T cells did not show any proportional changes; however, the absolute number of naïve CD8 T cells significantly increased after treatment (Supplemental figure 2). Control animals did not show any changes over the same time. Interestingly, B cell also showed significant increase in the naïve phenotype (IgD⁺CD27⁻) after desensitization (Figure 3C). Taken together, desensitization with carfilzomib and lulizumab showed similar immune modulation which promotes a naïve immune phenotype but does not interfere with Treg homeostasis, reducing the Treg frequency as previously seen in belatacept-based desensitization.

Figure 3. Lulizumab induces a more naive phenotype of T and B cells in peripheral blood while preserving the regulatory T cell population.

(A) Representative flow plot for regulatory T cells (CD4⁺CD25⁺CD127^lo) and their frequencies in blood, lymph node, and bone marrow. (B) Representative flow plot for CD4 and CD8 T cell subset analysis based on the expression of CCR7 and CD45RA including Naïve (Tn, CCR7⁺CD45RA⁺), central memory (Tcm, CCD7⁺CD45RA⁻), effector memory (Tem, CCR7⁻CD45RA⁻), and terminally differentiated effector memory T cells (Temra, CCR7⁻CD45RA⁺). (C) Representative flow plot for CD20⁺ B cell subset analysis based on expression of IgD and CD27 expression including naïve (IgD⁺CD27⁻), unswitched (IgD⁺CD27⁺), switched (IgD⁻CD27⁺), and exhausted (IgD⁻CD27⁻) B cells. * indicates p<0.05, NS indicates no significance.

Dual targeting desensitization with carfilzomib and lulizumab prolongs renal allograft survival in allosensitized nonhuman primates

All animals received T cell depletion with rhesus anti-thymocyte globulin (rhATG) and tacrolimus/MMF/steroid as maintenance immunosuppression. Control animals experienced early graft rejection with a MST of 5.8 ± 4.0 days while animals that received desensitization with CFZ and lulizumab showed significantly prolonged graft survival with a MST of 64.8 ± 36.3 days (Figure 4A). Kidney allograft samples either at rejection (sacrifice) or from protocol biopsies were evaluated by a renal transplant pathologist (A.B.F.) in a blinded fashion. The histologic scoring of all samples using the Banff criteria is shown in table 1. Briefly, 2 out of 5 control primates were censored from this analysis since despite of some pathological changes including focal interstitial fibrosis, tubular atrophy with associated inflammation, and mild tubular injury, they did not show histologic clear evidence of rejection. The other three monkeys showed evidence of AMR (with or without ACR). On the contrary, one out of five desensitized primates showed early allograft rejection (at POD 13) with mixed ACR and AMR while the others (4/5) did not show evidence of AMR (g+ptc=0) on one month protocol kidney biopsy. As shown in figure 4B, AMR scores calculated as g+ptc score (glomerulitis and peritubular capillaritis) were significantly lower in desensitized subjects at 4 weeks after kidney transplant compared to control kidney specimens at the time of sacrifice. Representative histology images from control animals showed moderate peritubular capillaritis (grade 2, blue arrow) with glomerulitis (grade 2, red arrow). Taken together, desensitization with carfilzomib and lulizumab prevented early antibody-mediated rejection.

Figure 4. Pre-transplant carfilzomib and lulizumab treatment significantly reduced early post-transplant antibody-mediated injury and prolonged allograft survival in sensitized NHPs.

(A) Percent graft survival of sensitized animals without desensitization (control, red line) or with desensitization (green line) after life-sustaining kidney transplantation from the same donor they were previously sensitized. (B) AMR score calculated based on Banff gradings, g and ptc. AMR score was significantly reduced in carfilzomib and lulizumab treated animals compared to controls. g, allograft glomerulitis; ptc, peritubular capillaritis. (C) Representative hematoxylin and eosin (H&E) image of control at euthanasia (POD 4) and desensitized animal at 1 month biopsy (POD27). Control animal showed moderate peritubular capillaritis (blue arrows) and glomerulitis (red arrow). * indicates p<0.05

Table 1.

Banff grading for allograft (biopsy) and other pathologic characteristics

	Case	t	v	i	ti	i-IFTA	g	ci	ct	cg	mm	cv	ah	ptc	Tubular Injury	Interstitial Plasma cells	Interstitial Neutrophils	Interstitial Eosinophils	Edema	Inclusions	PTC fibrin	Glomerular fibrin	Arteriole fibrin	C4d (N.A. = not available)	***AMR score (g+ptc)	**Diagnosis
Control	H764	3	0	3	3	3	3	1	1	1b	1	0	0	3	1	0	1	0	1	0	1	1	0	3	6	ACR, type 1B; Findings also suspicious for AMR and/or TMA
	H60L	1	0	1	1	1	2	1	1	0	0	0	0	0	1	0	0	0	1	0	0	0	0	1	2	No ACR.
	H42A	1	0	0	0	1	3	1	1	0	0	0	0	1	1	0	0	0	0	0	0	0	0	0	4	No ACR; findings also suspicious for active AMR
Desensitized (Carfilzomib + Lulizumab)	H17M	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0	No ACR. Focal interstitial fibrosis & tubular atrophy with associated inflammation are present.
	H58G	1	0	1	1	1	0	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	No ACR.
	H46G	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	0	0	0	0	0	0	0	1	0	No ACR. Proteinaceous cast/Tamm Horsfall protein collection is focally present, suggestive of urinary obstruction.
	H50R	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0	No ACR. C4d staining without evidence of rejection is of uncertain significance with regard to AMR.
	*H09W	1	0	1	1	1	1	0	0	0	0	0	0	1	1	0	0	0	1	0	0	0	0	1	2	No ACR; findings also suspicious for active AMR; neutrophil casts, suspicious for concurrent pyelonephritis

g, allograft glomerulitis; cg, allograft glomerulopathy; mm, mesangial matrix; ptc, peritubular capillary inflammation; i, interstitial inflammation; t, tubulitis; v, intimal arteritis; ci, cortical interstitial fibrosis; ct, cortical tubular atrophy; cv, vascular fibrous intimal thickening; ah, arteriolar hyalinosis; ti, total inflammation; i-IFTA, cortical inflammation; c4d, complement breakdown product that deposits in peritubular capillaries and medullary vasa recta.

^*cause biopsy

^**ACR = acute cellular rejection, AMR = antibody-mediated rejection, TMA = thrombotic microangiopathy

^***if greater than or = 2, s/o AMR

Transient effects of lulizumab in combination with carfilzomib lead to antibody rebound and return of memory cell repertoire post kidney transplantation

Despite the reduction of early AMR and conceptual benefit of lulizumab, all desensitized animals eventually experienced antibody-mediated rejection with CNI-based maintenance immunosuppression (Table 2 and Figure 5A). Animals with prolonged graft survival showed a gradual increase of anti-class I (2 of 4) and/or class II (4 of 4) DSA over time (Figure 5B). Notably, animals treated with carfilzomib and lulizumab did not show an early anamnestic response within a week as we had historically seen in carfilzomib and belatacept treated animals (Figure 5C). Peripheral blood T and B cell counts gradually returned to baseline level after induction with rhATG (Supplemental figure 3). Consistent with the rebound of DSA, Tfh cells and proliferating B cells gradually recovered within a month after transplantation in the peripheral blood (Figure 5D). In accordance with this, we observed a significant increase in lymph node Tfh cells (CD4⁺PD-1⁺ICOS⁺) one month after transplantation as well as at the time of sacrifice (Figure 5E). Additionally, the frequency of isotype switched B cells in the lymph node increased post kidney transplantation, providing indication of germinal center activity. This germinal center activity was observed despite the continuous increase of LN Tregs throughout the study period (Figure 5B). Interestingly, the recovery of GC activity matched the half-life of lulizumab, which showed only 50% receptor occupancy at 2–3 weeks post kidney transplantation on flow cytometry (Supplemental figure 4). Taken together, selectively targeting CD28 with a 4-week lulizumab-based desensitization regimen failed to induce durable control of the germinal center and long-term humoral tolerance, despite promoting Tregs. This also suggests that elevated Treg is not indicative of robust control of post-transplant humoral response and AMR.

Table 2.

Banff grading for allograft (necropsy) and other pathologic characteristics

Case	t	v	i	ti	i-IFTA	g	ci	ct	cg	mm	cv	ah	ptc	Tubular Injury	Interstitial Plasma cells	Interstitial Neutrophils	Interstitial Eosinophils	Edema	Inclusions	PTC fibrin	Glomerular fibrin	Arteriole fibrin	C4d (N.A. = not available)	**AMR score (g+ptc)	*Diagnosis
H17M	1	0	1	1	1	2	1	1	0	0	0	0	3	1	0	0	0	1	0	0	0	0	2	5	No ACR; findings also suspicious for active AMR
H58G	3	0	2	2	2	3	2	2	0	0	0	0	3	1	1	0	0	0	0	0	0	0	3	6	ACR, type 1B; Prominent proteinaceous cast & inflammatory reaction is also present, suggestive of urinary tract obstruction
H46G	1	0	0	1	2	3	1	1	0	0	0	0	3	1	0	0	0	0	0	0	0	0	3	6	No ACR; findings also suspicious for active AMR
H50R	1	0	1	1	2	3	1	1	2	3	2	0	3	1	0	0	0	0	0	0	0	0	3	6	No ACR; Findings also suspicious for AMR

g, allograft glomerulitis; cg, allograft glomerulopathy; mm, mesangial matrix; ptc, peritubular capillary inflammation; i, interstitial inflammation; t, tubulitis; v, intimal arteritis; ci, cortical interstitial fibrosis; ct, cortical tubular atrophy; cv, vascular fibrous intimal thickening; ah, arteriolar hyalinosis; ti, total inflammation; i-IFTA, cortical inflammation; c4d, complement breakdown product that deposits in peritubular capillaries and medullary vasa recta.

^*ACR = acute cellular rejection, AMR = antibody-mediated rejection, TMA = thrombotic microangiopathy

^**if greater than or = 2, s/o AMR

Figure 5. Pre-transplant desensitization with carfilzomib and lulizumab did not prevent rebound of DSA and late AMR development.

(A) AMR score (g+ptc) were significantly elevated at time of necropsy. (B) Normalized post-transplant serum DSA levels to pre-transplant time point. Serum DSA level was evaluated by T and B cell flow crossmatch. All animals showed positive B cell flow crossmatch. (C) Early post-transplant DSA kinetics. Animals with carfilzomib and belatacept treatment showed significantly elevated DSA a week after kidney transplantation while animals treated with carfilzomib and lulizumab showed no early elevation of DSA. (D) post-transplant circulating Tfh cells, proliferating B cells and plasmablasts (PB). Tfh cells and proliferating B cells were significantly increased overtime after kidney transplantation. (E) Post-transplant level of Tfh cells, isotype switched IgG⁺ B cells and Tregs in lymph nodes. Tfh and Treg cells were significantly increased after kidney transplantation in the lymph node while IgG⁺ B cells showed a strong trend of increase overtime. * indicates p<0.05, ** indicates p<0.01, **** indicates p<0.0001, NS indicates no significance.

Discussion

We have demonstrated that desensitization with carfilzomib and lulizumab (anti-CD28dAb) reduces serum DSA levels as well as Tfh cells, and proliferating B cells in the LN while promoting naïve T and B cells in sensitized rhesus monkeys. In contrast to belatacept-based desensitization, lulizumab preserved the Treg cell population during desensitization. Altogether, this lulizumab-based desensitization led to significant prolongation in kidney allograft survival in sensitized nonhuman primates.

Pharmacological desensitization became available for sensitized patients for kidney transplantation due to the plethora of novel biologics targeting components of the humoral response ²⁵. Our previous experience with desensitization in a sensitized nonhuman primate model led us to utilize costimulation blockade with proteasome inhibition for targeting germinal center responses and plasma cells, respectively ^13,14,26. Targeting the germinal center response with costimulation blockade (CTLA4-Ig and anti-CD40mAb) lowered DSA and prolonged graft survival when combined with proteasome inhibition (bortezomib or carfilzomib) in a sensitized NHP model ^14,26. However, this combination with CTLA4-Ig and anti-CD40mAb clearly represented a large immunosuppressive burden, and animals experienced significant morbidity. Our attempts to reduce this burden by using a single costimulation blocker (belatacept) with proteasome inhibition showed dramatic improvement with respect to morbidity¹⁵. However, while prolonging graft survival, belatacept-based treatment did not prevent later chronic antibody-mediated rejection (AMR). This late AMR reflects the failure of short-term belatacept, given pre-transplant to effect long-term modulation of humoral responses in sensitized animals. Therefore, in this present study, we evaluated lulizumab (anti-CD28dAb) in combination with carfilzomib in our sensitized NHP model of kidney transplantation.

Compared to belatacept, selective CD28 blockade with lulizumab theoretically provides more potent regulation of T cell activation by inhibiting CD80/86 interaction with CD28 without interfering with the negative co-inhibitory signal via CTLA4 ¹⁹. In accordance with this, a CD28 antagonist has shown benefit in preventing rejection and synergizing with other immunosuppressants in pre-clinical NHP models ^{19–21,27,28}. Notably, this selective CD28 blockade promoted regulatory T cells ¹⁹ and prevented post-transplant DSA development when combined with conventional immunosuppressive drugs including tacrolimus, rapamycin, or MMF in the NHP model ²⁰. Furthermore, in a murine model, CD28 blockade showed great efficacy in suppressing Tfh cells ^21,21. Studies with direct comparison of CD28 blockade to CTLA4-Ig demonstrate superior efficacy in inhibiting Tfh cells, GC B cells, and DSA production with selective CD28 blockade ^28,29. However, to our knowledge, lulizumab has not been tested in the sensitized setting. Therefore, we evaluated lulizumab, together with carfilzomib, to determine whether reported superiority of lulizumab over belatacept could induce more a durable desensitization in a sensitized NHP model.

In our study, we have shown that lulizumab significantly reduces the level of DSA (Figure 1) and inhibits Tfh cells and proliferating B cells in the lymph nodes (Figure 2) of sensitized rhesus primates when combined with carfilzomib. It is difficult to state whether lulizumab induced more complete suppression of Tfh or B cell proliferation since we did not directly compare it to carfilzomib and belatacept–treated animals. We observed a higher frequency of naïve T (CCR7⁺CD45RA⁺) and B (CD27⁻IgD⁺) cells after lulizumab treatment (Figure 3B and 3C), which was similar to belatacept treatment ¹⁵. However, lulizumab clearly preserved Treg cell frequency compared to belatacept-based desensitization, possibly due to maintaining Treg cell homeostasis by not interfering with the negative CTLA4 signal to Treg cells (Figure 2A).

Due to this difference, we expected more durable post-transplant humoral unresponsiveness after lulizumab and carfilzomib treatment compared to those with belatacept and carfilzomib. As shown in figure 4, the early AMR was significantly reduced compared to control group (no desensitization); however, it did not promote durable humoral unresponsiveness. All animals with desensitization showed a gradual rebound of DSA over tine ultimately leading to antibody-mediated rejection (Figure 5A). Interestingly, some animals only showed elevation of DSA against class II (Figure 5B), which is often seen in human patients ^30,31. This may possibly be due to the gradual increase of humoral immune responses including Tfh and B clonal expansion in the lymph node (GC response) after kidney transplantation (Figure 5D).

It is unclear why there is no robust difference in graft outcome even though selective inhibition of CD28/B7 signaling preserved Treg cell populations in BM, LN, and circulation compared to belatacept-based desensitization. It is possible that the role of Tregs in the sensitized recipient are not as important as in the non-sensitized setting. The reduction of Treg has been well documented with CTLA4-Ig treatment including belatacept and abatacept in many different settings ^32–37. Clearly, the Treg reduction did not alter the early risk of acute rejection after carfilzomib and belatacept desensitization ¹⁵. On the other hand, increased or maintained Treg cells in the BM niche could interfere with PC targeting since Tregs may help maintain long-lived plasma cell (LLPC) via CTLA-4 ³⁸. Thus, belatacept may provide better desensitization with carfilzomib since belatacept reduces the Treg population and also directly competes with the CTLA4 signal. However, it is important not to over interpret the superiority of belatacept in desensitization over lulizumab, since they were tested in the context of different induction agents (CD4/CD8 mAbs vs. rhATG). Furthermore, it is also important to mention that lulizumab has a significantly shorter half-life than belatacept ³⁹. As shown in Supplemental figure 4, after kidney transplantation CD28 T cell gradually re-appeared due to reduction of CD28dAb occupancy. In accordance with this, DSA started increasing after the lulizumab wash-out phase (Figure 5B). Interestingly. Treg cell population in the lymph node was maintained high during this breakthrough hurmoal response which could represent that increased Treg cannot be an indication of robust control of humoral response. It is unlikely that Treg cell at this stage lose their regulatory function. However, it is possible that elevated Treg does not elicit additional regulation on humoral response. It is also possible that regulatory Tfh (Tfr) cells are more relevant population than general Treg cells controlling rebound post-transplant humoral response. Nevertheless, this recalcitrant nature of the humoral response revealed the limitation of short-term pretransplant costimulation blockade. Therefore, our results suggest that continuous modulation of the humoral response with costimulation blockade might be necessary. Due to the characteristic preservation of Treg and lack of B cell anamnestic response, lulizumab may be a preferable maintenance agent to belatacept.

Here, we report the impact of lulizumab in combination with carfilzomib as pharmacological desensitization. The selective targeting of CD28 markedly reduced DSA, Tfh, and B cell expansion and promoted naïve T and B cells while preserving Treg cells. This led to significant prolongation in graft survival. However, despite expansion of the Treg population, we did not observe long-term stable humoral unresponsiveness. Lulizumab has however shown the ability to control the germinal center response and could have as such a role as part of a maintenance immunosuppression in sensitized transplant recipients.

Materials and Methods

Allosensitized nonhuman primate kidney transplant model

Five maximally MAMU-mismatched pairs (see supplemental Table 1) of male nonhuman primates (Macaca mulatta) were sensitized by performing swapping skin transplants of 3cm diameter patches of dorsal skin between the individuals of each pair as described previously ²⁴. Two skin transplants were performed 8 weeks apart to complete the sensitization phase. Twelve weeks after the second skin transplant, each individual in the pair was randomized by coin flip to receive either desensitization or no desensitization. Following the desensitization phase, each sensitized pair of nonhuman primates underwent swapping kidney transplantation and bilateral native nephrectomies as previously described ^15,24,26. Peripheral blood draws were performed twice weekly after kidney transplantation to monitor serum chemistries and monitor tacrolimus drug levels. Allograft survival endpoints were defined by persistent elevations in serum creatinine >4 mg/dL in combination with clinical signs of oliguria/anuria, lethargy, persistent weight loss.

Desensitization and maintenance immunosuppression regimens

Nonhuman primates randomized to receive desensitization received carfilzomib (Onyx Pharmaceuticals Inc. San Francisco, CA) 27 mg/m² intravenous and lulizumab (Bristol-Myers Squibb New York, NY) 12.5 mg/kg subcutaneous once weekly for four weeks. After kidney transplant, all nonhuman primates received induction therapy with rhesus-specific anti-thymocyte globulin (RhATG, NIH Nonhuman Primate Reagent Resource, Boston, MA) 4 mg/kg intravenous once daily for five days beginning on the day of transplant. All primates received maintenance immunosuppression with tacrolimus (Astellas Pharma Inc. Northbrook, IL) 0.05 mg/kg intramuscular twice daily beginning on the day of transplant and titrated to achieve a tacrolimus trough level of 8-12 ng/mL in peripheral blood, mycophenolate mofetil (Genentech USA Inc. San Francisco, CA) 30 mg/kg orally twice daily beginning on the day of transplant, and methylprednisolone (Pfizer Inc. New York, NY) intramuscular tapered as follows: day 0 = 15 mg/kg, day 1 = 7.5 mg/kg, day 2-3 = 3.75 mg/kg, day 4 = 1.88 mg/kg, day 5 = 0.94 mg/kg, and then 0.5 mg/kg daily thereafter. After kidney transplant, all primates received viral prophylaxis with ganciclovir (Fresenius Kabi Lake Zurich, IL) 6 mg/kg subcutaneous daily. Acute rises in creatinine suspicious for acute rejection were treated with methylprednisolone intramuscular taper as follows: 125 mg for 3 days, 75 mg for 3 days, and 25 mg for 3 days, then returned to maintenance dose of 0.5 mg/kg daily.

Flow crossmatch for donor-specific antibody determination

Weekly blood draws were obtained from all primates beginning on the day of the first skin transplant to obtain recipient serum and donor peripheral blood mononuclear cells. Donor peripheral blood mononuclear cells (0.5 x 10⁶) were stained with Fixable Blue viability stain (Invitrogen Carlsbad, CA) according to the manufacturer's protocol. Cells were washed twice with 200 μL 2% FBS in PBS and blocked with 100 μL 1:1000 dilution of Goat IgG Whole Molecule (Jackson ImmunoResearch Laboratory, Inc. West Grove, PA) for 15 minutes at 4°C. Cells were washed twice with 200 μL 2% FBS in PBS and incubated with 100 μL 1:50 dilution of recipient serum in 2% FBS in PBS for 30 minutes at 4°C. Cells were washed 5 times with 200 μL 2% FBS in PBS and then stained with FITC-conjugated polyclonal anti-monkey IgG (Seracare Life Science Inc. Milford, MA), PE-conjugated anti-CD20 (clone 2H7, BD Biosciences San Jose, CA), and PerCPCy5.5-conjugated anti-CD3 (clone SP34-2, BD Biosciences) according to manufacturers’ recommendations. Cells were washed twice with 200 μL 2% FBS in PBS and fixed with Stabilizing Fixative (BD Biosciences). Flow cytometry was performed on a BD LSRFortessa flow cytometer and analyzed using FlowJo 10 software.

Immune phenotyping by flow cytometry

Cells suspensions were prepared from peripheral blood mononuclear cells, peripheral lymph node biopsies, and bone marrow aspirates. Cells (0.5-1.5 x 10⁶) were stained with Fixable Blue viability stain (Invitrogen) according to the manufacturer's protocol. Cells were washed twice with 200 μL 2% FBS in PBS and stained for extracellular markers with fluorochrome conjugated monoclonal antibodies (mAbs) including anti-CD127, CD185 (CXCR5), CD19, CD20, CD25, CD278 (ICOS), CD279 (PD-1), CD28, CD3, CD38, CD4, CD8, CD95, IgD, IgG, IgM according to manufacturers’ recommendations. Cells were washed twice with 200 μL 2% FBS in PBS and those undergoing intracellular stains were permeabilized, stained for intracellular markers including fluorochrome conjugated mAbs against human Blimp-1, Ki67 and Foxp3 (see Supplemental table 2 for clones, fluorochrome and providers) and fixed using the Intracellular Fixation & Permeabilization Buffer Set (eBioscience San Diego, CA). Cells not undergoing intracellular staining were fixed with Stabilizing Fixative (BD Biosciences). Flow cytometry was performed on a BD Fortessa flow cytometer and analyzed using FlowJo v9 or v10 software (FlowJo LLC, Ashland, OR, USA).

Histology and immunohistochemistry

Peripheral lymph node biopsies from the axilla or groin were performed on all primates at time points prior to the first desensitization treatment, 1 week after the last desensitization treatment, and 4 weeks after kidney transplant. Peripheral lymph nodes were also obtained at necropsy. Lymph node specimens were fixed in 10% neutral buffered formalin, paraffin embedded, sectioned, and stained with anti-CD20 (Agilent, Santa Clara, CA) and anti-Ki67 (Thermo Fisher Scientific, Waltham, MA) to visualize germinal centers within the lymph node specimens. Ultrasound guided core needle biopsies of the kidney allografts were performed at 4 weeks and 10 weeks post-transplant in those that survived to these time points and renal allografts were harvested at necropsy. Kidney specimens were fixed in 10% neutral buffered formalin, paraffin embedded, sectioned, and stained with hematoxylin and eosin, periodic acid-Schiff, and polyclonal anti-human C4d (American Research Products, Waltham, MA) for histologic evaluation. Allograft histology was evaluated in a blinded fashion by a trained transplant pathologist (A.B.F) and scored according to the current Banff criteria for kidney rejection ^40–43.

Statistical analyses

Data are expressed as the mean ± SD (error bar) throughout the manuscript. Statistical analyses were performed using GraphPadPrism 8.0 (GraphPad Software, San Diego, CA, USA). Values of p <0.05 were considered to be statistically significant. Survival analysis was performed using the Kaplan–Meier method and log-rank test. Normally distributed data within the same treatment group but at different time points were evaluated using a two-tailed paired t-test. Statistical comparisons between different groups were performed with two-tailed unpaired t-test for normally distributed data.

Supplementary Material

Supplemental data

Supplemental table 1. MHC class I (MAMU-A and MAMU-B) and class II (MAMU-DRB, DQB, DPA, DPB) haplotypes from skin/kidney donor and receipts rhesus macaques.

Supplemental table 2. List of antibodies for flow cytometry

Supplemental Figure 1. T and B cell level during desensitization.

Supplemental Figure 2. Absolute number of T cell subsets after desensitization with carfilzomib and lulizumab.

Supplemental Figure 3. Repopulation of peripheral blood T and B cells after renal allotransplantation.

Supplemental Figure 4. CD28 availability on CD4 T cells after kidney transplantation.

Translational Statement.

Lulizumab, an anti-CD28 dAb, is a novel costimulation blocking agent selectively targeting the CD28-CD80/86 interaction and hence preserving the co-inhibitory signal (CTLA4-CD80/86) unlike Belatacept (CTLA4-Ig). Previous studies found that Lulizumab suppressed follicular helper T cells while preserving regulatory T cells. Here we found that short-term peri-transplant treatment with Lulizumab and Carfilzomib, the latter a proteasome inhibitor, reduced early humoral responses and AMR in highly sensitized nonhuman primate kidney transplant recipients. However, all animals showed gradual post-transplant rise of DSA, increasing follicular helper T cells and proliferated B cells, and eventual late AMR. Even though peri-transplant desensitization with proteasome inhibitor and costimulation blockade significantly prolonged graft survival, additional or continuous control of post-transplant humoral responses is necessary to avoid graft loss.

Abbreviations and Acronyms

AMR

antibody-mediated rejection

ATG

anti-thymocyte globulin

BM

bone marrow

CFZ

carfilzomib

DSA

donor-specific antibodies

FBS

fetal bovine serum

FITC

Fluorescein isothiocyanate

GC

germinal center

HLA

human leukocyte antigen

IgG

Immunoglobulin G

LLPC

long-lived plasma cell

mAb

monoclonal antibodies

MFI

mean fluorescence intensity

MHC

major histocompatibility complex

NHP

non-human primate

NIH

National Institutes of Health

PB

plasmablast

PBS

Phosphate Buffered Saline

PC

plasma cells

PE

phycoerythrin

PRA

panel reactive antibody

rhATG

rhesus-specific anti-thymocyte globulin

Tfh

follicular helper T

Treg

regulatory T cells