All washes in the AXE procedure, including those conducted during donor cell preparation and following the antibody adsorption procedure, were performed using phosphate buffered saline (PBS; Life Technologies Inc., Burlington ON, Canada). Antibody eluates were prepared using the acid eluting solution (solution I; ELU KIT^TM Plus; Immucor, Dominion Biologicals Limited; Dartmouth, Canada) and the eluate pH was neutralized (6.8–7.2 range) using the base buffering solution (Solution II; ELU KIT^TM Plus; Immucor, Dominion Biologicals Limited). The pH was verified using plastic pH indicator strips (Thermo Fisher Scientific Inc.). SAB assays were performed using LABScreen SAB kits (LS1A04 lot 13 for HLA class I and LS2A01 lots 14 and 15 for HLA class II; One Lambda, Canoga Park, CA), phycoerythrin (PE)-conjugated goat anti-human IgG (LS-AB2; One Lambda), LABScreen wash buffer (LWB; One Lambda) and 96-well V-bottom trays (Whatman Pistcataway, NJ). FCXM were set up in 96-well U-bottom BD Falcon Microplate trays (BD Biosciences) and washes were performed with flow wash buffer (FWB) composed of phosphate buffered saline (PBS; Life Technologies Inc.) and 2% (v/v) fetal calf serum (Life Technologies Inc.). Anti-CD3-PerCP (clone SK7) and anti-CD19-PE (clone SJ25C1) monoclonal antibodies were purchased from BD Biosciences (Mississauga, ON, Canada). Fluorescein (FITC) conjugated F (ab’)₂ fragment goat anti-human IgG, Fcγ specific polyclonal antibody (IgG-FITC), 1.0 mg/ml stock solution, was purchased from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA). Lymhocytes were isolated with the EasySep^TM Direct Human Total Lymphocyte Isolation Kit (EasySep^TM Direct; STEMCELL Technologies Inc., Vancouver, BC, Canada) and were treated with pronase (4.7 units/mL; Sigma-Aldrich, St Louis, MO) and DNase (11,000 units/mL; Sigma-Aldrich).

Cells for all AXE evaluation studies were obtained from acid dextrose citrate (ACD) anticoagulated whole blood samples collected from volunteers as well as live and deceased donors in accordance with the institutional assay validation protocol. Pooled positive control sera (PPC; a pool of 20 highly sensitized patient sera, cPRA >95%) were diluted to 1:128 and 1:8 in PBS for AXE protocol evaluation for class I and class II HLA antibodies, respectively. LABScreen negative control sera (One Lambda) were used in all SAB assay testing. Sera from patients awaiting solid organ transplantation were used to evaluate the AXE procedure. The ability of AXE to detect antibodies targeting native HLA epitopes was evaluated in sera containing genuine HLA reactivities that have been confirmed on cell-based crossmatches (n=6). In addition, sera with known spurious reactivity patterns (n = 3) on the SAB assay were tested to assess the assay’s specificity. All sera were treated with EDTA disodium salt solution at a final concentration of 6.0 mM (0.5 M stock solution, Cat# E7889; Sigma-Aldrich, St Louis, MO), before AXE or SAB testing.

ACD anticoagulated whole blood samples (6 ml) were centrifuged at 1800 x g for 10 min with no brake. Buffy coat layers (200 μl) were collected, placed into a 1.5 ml microfuge tube and resuspended in 1.2 ml of PBS. Cells were washed twice in PBS by centrifugation at 800 x g for 1 min and then resuspended in 1 ml of PBS.

The isolated donor cell suspension (1 buffy coat equivalent) was divided equally into two 1.5 ml microfuge tubes, centrifuged at 800 x g for 1 min and the supernatants were removed. Two hundred μl of EDTA treated test serum was added to the donor cell pellet in the first microfuge tube while PPC was added to the second tube. Serum/cell suspensions were mixed well by pipetting up and down. Tubes were placed into a 37°C heat block and incubated for 30 min. One ml of PBS was added to each tube and the cells were washed 6 times by centrifugation at 800 x g for 1 min. On the last wash 100 μl of the supernatant was collected from each tube and placed into a clean tube. This supernatant served as the last wash control to ensure no significant amount of unbound HLA antibodies were present prior to the elution procedure. Next, 50 μl of the acid eluting solution (solution I; ELU KIT^TM Plus) were added to the dry cell pellets and the tubes were gently vortexed. After a 1 min incubation at room temperature (RT), tubes were centrifuged at 800 x g for 1 min, then 40 μl of eluate was carefully removed and placed in a clean tube containing 50 μl of the base buffering solution (Solution II; ELU KIT^TM Plus) to neutralize the pH (6.8–7.2). The appropriate pH range was verified by placing 1 μl of each on the Fisherbrand Plastic pH strip. All eluates and last wash controls were immediately tested for HLA antibodies using the LABScreen SAB assay.

All SAB assays were performed using the ROB protocol as previously described (Liwski et al., 2017). Briefly, 20 μl of EDTA (6.0 mM) treated sera, eluate or last wash control samples was added to appropriate wells of 96-well V-bottom trays containing either class I or class II HLA LABSCreen single antigen beads. Trays were incubated for 15 min at RT and washed 4 times in Luminex wash buffer by centrifugation at 1800 x g for 1 min. After the last wash, 20 μl of anti-IgG-PE (1:10 dilution in PBS) secondary antibody was added to each well and the trays were incubated for 5 min at RT in the dark. Following two additional washes, beads were resuspended in 55 μl of LABScreen wash buffer and samples were acquired using Luminex 3D instruments. The results were analysed with Fusion software version 4.3.1 using the baseline MFI formula.

FCXMs were performed using the Halifaster protocol as previously described (Liwski et al., 2018). Briefly, 30 µl of test or control sera and 15 µl of purified (EasySep Direct) donor lymphocytes (1.5 × 10⁵ cells) were added to each reaction well in a 96-well U-bottom BD Falcon Microplate tray, dry vortexed, and incubated at RT for 20 min. Cells were washed three times with 200 µl FWB at 500 x g for 1 min, after which antibody cocktail (2 µl of anti-CD3-PerCP, 1 µl of anti-CD19-PE, 0.125 µl anti-IgG-FITC, made up to 50 µl with PBS) was added to the cells, dry vortexed, and incubated for 5 min at RT in the dark. Cells were washed twice more with 200 µl FWB at 500 x g for 1 min, resuspended in 150 μl FWB, and transferred into 5 ml polystyrene Falcon tubes containing an additional 250 μl of FWB before acquisition on the BD FACSCanto II flow cytometer.

AXE protocol efficiency at recovering HLA antibodies was calculated as a percentage of the SAB assay mean fluorescence intensity (MFI) seen for the HLA specificities in the eluate as compared to the original serum sample. Mean and standard deviation calculations were performed using Microsoft Excel software.

To evaluate the AXE procedure for its efficiency of recovering HLA antibodies from sera, buffy coat cells isolated from healthy donor blood samples (n = 5) were used to individually adsorb and elute HLA antibodies from the pooled positive control (PPC) sera. The PPC was generated by pooling sera from 20 highly sensitized (cPRA ≥95%) transplant candidates to ensure strong reactivity with all HLA antigens represented on the LABScreen class I and class II HLA SAB panels. For the AXE procedure evaluation, the PPC serum was diluted to 1:128 for class I HLA (Figure 1A) and 1:8 for class II HLA (Figure 2A) to test the efficiency of adsorption and elution of HLA antibodies over a broad range of mean fluorescence intensity (MFI) values. The last wash control, which is the supernatant collected after the final wash following the adsorption portion of the procedure, was used to ensure that all unbound antibodies were eliminated prior to the elution process.

Representative class I HLA SAB assay histograms of AXE evaluation studies performed using the PPC and one of the donor cells are shown in Figure 1. For ease of visualization, the HLA antigens in each SAB panel were sorted in numerical order within each locus (Figure 1). Black rectangles highlight the AXE donor cell HLA antigens (HLA-A*03:01, *29:02; B*07:02; C*07:02) with the top of each rectangle aligning with the maximum PPC MFI values of antibodies specific to the corresponding AXE donor cell antigens. The efficiency of adsorption/elution of the AXE procedure was expressed as a percentage of donor specific antibody (DSA) MFI recovered in the eluate vs. serum sample and was as follows: HLA-A3 = 73.7%, A29 = 76.9%, B7 = 66.8%, and Cw7 = 32.9% (Figure 1B). Interestingly, the efficiency of AXE when eluting antibodies targeting donor antigens was similar across a wide range of antibody MFI. For example, the AXE efficiency was comparable when eluting antibodies targeting HLA-B7 and HLA-A3/A29 donor antigens despite the significantly lower starting MFI values for HLA-B7 (MFI = 2,027) compared to HLA-A3 (MFI = 11,300) or HLA-A29 (average MFI = 12,220) in the PPC sample (Figures 1A,B). In contrast, the percent MFI recovered by AXE for antibodies binding to 3^rd party HLA antigens was variable (range: 1.01%–77.9%). For example, while HLA-A1 specific antibodies eluted at 49.5% of the PPC MFI (PPC MFI = 23,000; eluate MFI = 11,450, Figures 1A,B), HLA-B13 specific antibodies eluted at an average of only 4.7% (mean PPC MFI = 23,600; mean eluate MFI = 1,120; Figures 1A,B). This is likely related to the epitope specificity of eluted antibodies and the differential degree to which donor-specific epitopes are shared on 3^rd party HLA antigens. Importantly, the percentage of MFI remaining in the last wash control was negligible for both donor-specific (range: 0.09%–0.4%; eluate MFI range: 6.2–16.3; Figure 1C) and 3^rd party (range: 0.04%–1.7%; eluate MFI range: 5.1–25.5; Figure 1C) antigens, demonstrating that the antibodies eluted by AXE were those specifically adsorbed out by the donor cell HLA antigens.

Representative histograms for the class II HLA AXE procedure evaluation are shown in Figure 2. The percentage of DSA MFI recovered in the eluate following AXE procedure was 57.7% for HLA-DR15, 53.4% for DR51, 83.0% for DQ6 and 47.0% for DP1 (Figure 2B). Thus, the overall efficiency of AXE was excellent despite the relatively low starting MFI of DSA in the PPC sample for some of the class II HLA specificities (PPC mean MFI: DR15 = 2,930; DR51 = 1,875 and DP1 = 2,570; Figure 2A). As seen for class I HLA, the efficiency of the AXE protocol in adsorbing/eluting antibodies directed against 3^rd party class II HLA antigens was variable (range 0.4%–90.2%; Figure 2B). Importantly, HLA antibody reactivity observed in the last wash control was negligible (Figure 2C).

The overall results of AXE protocol evaluation with all five donor cells are presented in Figure 3. On average, antibodies against class I HLA antigens appeared to be adsorbed/eluted slightly better compared to those binding to class II HLA (65.1% vs. 53.2%; Figures 3A,B). Within class I, antibodies targeting HLA-A and B antigens appeared to be adsorbed/eluted more efficiently compared to those targeting HLA-C loci (71.7% for HLA-A, 68.6% for HLA-B, 38.8% for HLA-C; Figure 3A). Such locus specific differences in the efficiency of AXE protocol were also seen for class II HLA with HLA-DR specific antibodies being adsorbed/eluted most efficiently followed by HLA-DQ and HLA-DP (59.5% for HLA-DR, 44.2% for HLA-DQ, and 34.5% for HLA-DP Figure 3B). These differences in adsorption/elution efficiency are likely related to locus specific variation in HLA expression on leukocytes and the fact that a smaller percentage of leukocytes in blood express class II HLA when compared to class I. Finally, the level of HLA antibodies targeting both donor specific and 3^rd party HLA antigens detected in the last wash control samples was negligible.

To confirm that the AXE procedure selectively adsorbs/elutes HLA antibodies in an antigen specific manner, sera from sensitized patients (n = 3) were adsorbed/eluted using autologous cells (auto eluate) vs. surrogate allogeneic donor cells (allo eluate) expressing HLA antigens targeted by antibodies in sensitized patients’ sera. The eluates and the corresponding last wash controls were then tested by the SAB assay. Representative results from these experiments are shown in Figure 4 and demonstrate that while HLA antibodies cannot be adsorbed/eluted from sera using autologous cells (auto eluate; Figure 4B), allogeneic donor cells expressing HLA-A2 and B13 antigens adsorbed HLA antibodies from the same serum in a donor antigen specific fashion (allo eluate; Figure 4C). In fact, the HLA antibody reactivity pattern seen in the allogeneic donor cell eluate is explained with three eplets expressed by donor antigens: 41T (B13, 41, 44, 45, 47, 49, 50, 60 and 61; green squares; Figure 4), 62 GE (A2, B57, B58; blue squares; Figure 4) and 144TKH (A2, 68, 69; red squares; Figure 4). The adsorption/elution efficiency in this case was high (74.5% for HLA-A2 and 94% for HLA-B13 DSA; Figure 4C), despite the low starting MFI for the B13 antibody in serum (mean MFI = 640; Figure 4A). Interestingly, several relatively strong 3^rd party HLA-B and C locus antibody specificities (such as B35, B53, Cw9, Cw10 and Cw15) seen in the original serum sample were not adsorbed/eluted with the allogeneic donor cells. The eplet analysis shows that these specificities share the 94I eplet, indicating that these were genuine and not spurious reactivities. Because the target 94I eplet is not expressed on either A*02:01 or B*13:02 donor alleles, the results further illustrate the specificity of the AXE protocol as indicated by the absence of the 3^rd party specificities when adsorbed/eluted using the allogeneic donor cells. Finally, the SAB assay performed on the allogeneic donor last wash control was completely negative, demonstrating that no unbound HLA antibodies remained in the supernatant prior to the elution procedure (Figure 4D).

Non-specific antibodies directed against denatured HLA epitopes or other non-HLA targets on the single antigen beads can give positive reactions in the SAB assay but are not predicted to be adsorbed and eluted using donor cells expressing native HLA antigens. To this end, we performed the AXE protocol on three sera with known spurious reactivity patterns on the SAB test. Figure 5A depicts the class I HLA SAB assay histogram showing a moderately strong antibody reactivity pattern consistent with the 156D epitope (HLA-B8, B37, B41, B42, B*44:02, B45 and B82). This epitope was previously identified as a denatured epitope (El-Awar et al., 2009), which was also confirmed by others using acid denaturation of class I SAB (Eapen et al., 2011). As expected, AXE adsorption/elution using HLA-B*08:01 positive donor cells demonstrated no evidence of antibody binding to native HLA on the cell surface (eluate; Figure 5B). In contrast, HLA-B8 specific antibodies were adsorbed/eluted from PPC using the same donor cells with the efficiency of approximately 58% (PC serum MFI = 9,500, PC eluate MFI = 5,500; Figure 5). Another relatively common artefact seen with the LABScreen class II SAB panel is a strong reactivity with the HLA-DR53 expressing beads (Figure 6A). The AXE procedure demonstrated that these antibodies were not adsorbed/eluted using HLA-DR53 expressing donor cells (Figure 6B; eluate). The efficiency of AXE for adsorbing/eluting DR53 reactivity from PPC was 30% in this experiment (Figure 6; PC serum vs. PC eluate). Finally, Figure 7A shows a frequently observed artefact on the LABScreen class II HLA SAB assay involving DP1, DP5 and DR53 expressing beads which was shown to be non-reactive against HLA-DP1 expressing donor cells (Figure 7B; eluate), confirming that these antibodies do not bind to native HLA-DP1 antigen. Importantly, DP1 specific antibodies were adsorbed/eluted from PPC using the same donor cells with the efficiency of approximately 25% (Figure 7; PPC serum and PPC eluate).

In this section we highlight the utility of the AXE procedure in confirming the presence of low-level HLA antibodies in clinical samples where such antibodies could be difficult to distinguish from non-specific background reactivities. Figure 8A shows the class I HLA antibody SAB histogram from a highly sensitized renal transplant candidate relisted a few years after failing his first deceased donor transplant. Single antigen beads corresponding to patient’s own HLA typing and the previous transplant donor HLA typing are indicated with green and red arrows, respectively. In this case there is no clear separation of positive signals from background on the histogram. If a commonly accepted MFI threshold of 2,000 MFI is used to define unacceptable antigens (HLA-A1, A11, A23, A*24:02, A*29:01, A36, A43, A80, B44, B45, B76, B82, Cw7, Cw8 and Cw16), this patient would have an elevated cPRA of 88%. If a more stringent 1,000 MFI threshold is applied, additional positive specificities include HLA-A3, A*29:02, B46, Cw6, Cw10 and Cw12, further increasing the cPRA to 92%. When eplet analysis, a more biologically rational approach to antibody assessment, is performed using the previous donor antigens (A1, B8, B49, Cw7) to identify mismatched eplets from the first transplant (76ANT blue, 144TKR red, 163RW orange, 166DG black, 9D purple, 41T green, and 76VRN yellow rectangles), it is clear that many weak antibody specificities reacting below the 1,000 MFI cutoff could have arisen as a result of transplant related sensitization, thereby posing an immune risk to repeat transplantation. AXE studies of this serum using two distinct donor cells expressing among them all the relevant mismatched antigens and eplets (donor 1: A*01:01 A*11:01, B*08:01, C*07:01; Figure 8B and donor 2: A*03:01, A*23:01, B*49:01, C*07:01; Figure 8C) were able to adsorb/elute the entire reactivity pattern, including all specificities below 1,000 MFI in the original serum, in an antigen/eplet specific manner (Figures 8B,C). Importantly, neither the Cw2 nor Cw4 weak “autologous” reactivities seen with the SAB testing of the original serum (Figure 8A) were eluted from either donor cell (Figures 8B,C), confirming the specificity of the adsorption/elution testing.

Interestingly, the eluate from donor 1 (Figure 8B) exhibited a strong reactivity with B44, B45 and B82 antigens, which was unexpected based on the eplet analysis. The B44 and B45 specificities were explained in the original serum by the 41T eplet shared with the immunizing B49 antigen, however, this eplet was not present on any of the AXE donor one mismatched antigens. In addition, the B82 reactivity could not be explained in the original serum using eplet analysis in the context of transplant related sensitization. These findings suggest that the antibodies specific to B44, B45 and B82 antigens eluted in this AXE study are targeting a novel epitope. Indeed, the amino acid sequence alignment revealed the presence of a glycine (G) or serine (S) at position 167 that could explain these results. This polymorphic position is shared by two existing eplets: 166DG (A1, A23, A*24:02, A80 and B76) and 163LS/G (B44, B45, B76 and B82) and the adsorbing antigen from which the entire reactivity was eluted in this case was A*01:01. To confirm this novel epitope, an HLA-B*44:02 mismatched donor which carries this novel epitope was used for additional AXE studies (Figure 8D). The results confirmed that HLA-B*44:02 alone could adsorb the entire pattern of reactivity including A1, A23, A*24:02, A80, B45, B76 and B82. As expected, all specificities expressing the 41T (B*44:02 mismatch) eplet and 76VRN (C*07:02 mismatch) eplet were also adsorbed and eluted with this donor cell. Importantly, last wash AXE controls were negative with all surrogate donor cells (not shown).

Figure 9 illustrates an example where the AXE protocol could complement eplet analysis to define clinically relevant, low-level class II antibodies in a highly sensitize patient with a history of multiple pregnancies. The SABs corresponding to patient’s HLA typing are indicated with green arrows. The eplet analysis suggested that four distinct eplets, namely: 70DA (red rectangles), 55 PP (dark blue rectangles) 84DEAV (green rectangles) and 55DEE (light blue rectangles) can explain most of the reactivity in this serum. Importantly, several of the HLA-DR specificities (DR103, DRB1*04:02, DR8, DR12, DR13, DR14, DR16 and DRB5*01) accounted for by the eplet analysis are below the commonly used 2,000 and 1,000 MFI thresholds and could be interpreted as background reactivity by many HLA laboratories. However, the AXE protocol studies using cells from a donor mismatched for 3 antigens of interest (DRB1*11:01, DQB1*03:01, DQA1*05:05, DPB1*14:01) expressing the four relevant eplets clearly adsorbed and eluted the entire expected reactivity pattern including the low-level DR specific antibodies (Figure 9B). Importantly, the SAB assay performed on the last wash control (not shown) was completely negative confirming that antibodies in this serum were targeting native HLA epitopes.

In Figure 10 we show that the AXE protocol has superior sensitivity in detecting low level HLA antibodies compared to surrogate crossmatches. In this example involving a pregnancy-sensitized patient with a known weak class I HLA antibody reactivity pattern to mismatched HLA-A2 and B44 spousal antigens, flow cytometric T cell crossmatches against a surrogate donor (with HLA-A2, A33, B65) only elicited a weakly positive crossmatch result on neat serum (105 median channel shift above the negative control serum; positive T cell cutoff is >70 median channels), which was rendered negative (44 median channel shift) at a 1:2 serum dilution (Figure 10C). Similar results were seen with the B cell crossmatches (not shown), consistent with the presence of low-level class I HLA DSA. In contrast, AXE studies using the same surrogate donor cells showed a clear pattern of adsorption in a donor specific manner (HLA-A2, A33 and B65) in eluates performed with all serum concentrations from neat to a 1:4 dilution (Figure 10B). Therefore, while both FCXM and AXE confirmed the presence of low-level HLA antibody in the serum, the AXE protocol was able to verify the predicted antibody specificities correlating to the immunizing event, and appeared to exhibit greater sensitivity for detection of HLA antibodies.