Search
Close this search box.

Enzymatic Conversion Of Human Blood Group A Kidneys To Universal Blood Group O – Nature Communications – Renal.PlatoHealth.ai

Donor demographics

In total, 18 donor kidneys from nine donors were declined for human transplantation and consented for research and recruited to this study between July 2022 and May 2023. All kidneys were blood group A, and the donor details are summarised in Table 1. In total, six kidneys (three biological pairs) were assigned to each of three research cohorts, namely, normothermic machine perfusion (NMP cohort), hypothermic machine perfusion (HMP cohort), or hypothermic machine perfusion followed by ABO-incompatible reperfusion (HMP-ABOi cohort).

Table 1 Donor demographics for the pairs of kidneys in the NMP, HMP, and HMP-ABOi cohorts

Enzymatic blood group A antigen removal during normothermic machine perfusion

Previous studies have shown FpGalNAc deacetylase and FpGalactosaminidase, herein referred to as FpGalNAc DeAc and FpGalNase, can remove blood group A antigens from human lungs during normothermic perfusion4. We sought to investigate A antigen removal in human kidneys by perfusing three biological pairs of blood group A kidneys for 6 h using an acellular perfusate at approximately 37 °C. These kidneys were designated the ‘NMP cohort’ and the demographics of the kidney donors are shown in Table 1, with all three pairs of kidneys retrieved from DCD donors. Within each pair, one kidney was treated with 1 mg/L of each of FpGalNAc DeAc and FpGalNase administered via the arterial cannula, and the contralateral kidney acted as a control with no enzyme addition. Treated kidneys were termed ‘ABOe’ kidneys.

Biopsies taken during perfusion were analysed by immunofluorescence microscopy for the presence of the A and H antigens (Fig. 2a). Quantification of A antigen removal and H antigen emergence during perfusion are shown in Fig. 2b for six fields of view from the three kidneys per group (n = 18 images per group total). A summary of normalised A and H antigen expression quantification is shown in Tables 2 and 3, respectively.

Fig. 2: Blood group A antigen removal during normothermic machine perfusion (NMP).
figure 2

a Immunofluorescence images of blood group A antigen removal during NMP in control (left column) or treated (right column) type A human kidney cortical biopsies. At each timepoint (pre-treatment, 1 h after enzyme addition, or 6 h after enzyme addition), a composite image of blood group A antigens (magenta) and H antigens (green) is shown. An 8× zoom inlay of a representative peritubular capillary (blue box) is shown in the bottom left of each panel (white box). Scale bar represents 100 μm. b Quantification of A and H antigen expression at 1 h, 2 h, 4 h, and 6 h after enzyme addition in control (blue) and treated (red) kidneys. Antigen expression is normalised to expression level in the pre-perfusion biopsies for n = 3 kidney pairs. Error bars show mean ± 95% confidence intervals. Renal blood flow (RBF; panel c), mean arterial pressure (MAP; panel d) and urine output (e) during perfusion are shown for control and treated kidneys. Error bars show mean ± standard deviation. f The concentration of perfusate and urine NGAL after 6 h perfusion in control and treated kidneys. Error bars show mean ± standard deviation. NMP normothermic machine perfusion, NGAL neutrophil gelatinase-associated lipocalin.

Table 2 Quantification of A antigen expression in control vs treated kidneys during NMP or HMP
Table 3 Quantification of H antigen expression in control vs treated kidneys during NMP or HMP

Pre-treatment sections show strong peritubular capillary staining for the A antigen, with some mild H antigen expression on the vascular surface (Fig. 2a). After 1 h, blood group A antigen expression in ABOe kidneys dropped to 56% of the pre-perfusion expression level, which was significantly lower than the control group (p = 0.0056). Blood group A antigen expression reached a low of 19% after 2 h in ABOe kidneys, with no further decrease observed up to 6 h. This was significantly lower than control kidneys which showed no decrease in antigen expression during perfusion.

In contrast, H antigen expression increased by 4.85-fold after 1 h FpGalNAc DeAc and FpGalNase treatment in the ABOe group, in line with the conversion of the A antigen structure to that of the H antigen (Fig. 1b), with no further change at 6 h. H antigen expression in the control group showed no increase over the course of perfusion.

Effect of FpGalNAc DeAc and FpGalNase treatment on renal function during NMP

Functional perfusion parameters were assessed continuously during perfusion for signs of treatment-related kidney injury. Renal blood flow (RBF) and mean arterial pressure (MAP) were stable throughout the perfusion, with no difference between control and treated kidneys across the perfusion period of 6 h (Fig. 2c, d). After correction for individual kidney weight, mean RBF at 6 h was 230 ± 76 ml·min−1 100 g−1 (mean ± SD) for control kidneys compared to 249 ± 33 ml·min−1 100 g−1 for those treated with FpGalNAc DeAc and FpGalNase (p = 0.7500). MAP was also comparable between the groups (missing value prevents statistical testing). Urine output was also measured hourly during perfusion, with no significant difference between mean total urine output across the 6 h of perfusion (control: 46 ± 10 ml; treated: 37 ± 28 ml; p = 0.7500; Fig. 2e).

To evaluate potential histological signs of damage induced by the FpGalNAc DeAc and FpGalNase treatment, haematoxylin and eosin (H&E) stained biopsies were assessed pre- and post-treatment and showed no difference in injury grading between control and ABOe kidneys. Mild acute tubular injury was noted in pre-perfusion as well as post-perfusion samples in all cases, with vacuolation, fibrosis, cell shedding, and tubular flattening highlighted as signs of ischaemic injury attributed to the period of cold ischaemia after organ retrieval (Supplementary Fig. 1). No signs of acute vascular damage were indicated in treated or control samples.

Furthermore, the kidney damage marker NGAL was assessed after 6 h perfusion, with no significant difference found in perfusate (control group = 158 ± 70 ng/ml; treated = 205 ± 114 ng/ml; p = 0.5000) or urinary NGAL concentration between control and treated groups (control group = 133 ± 85 ng/ml; treated = 181 ± 123 ng/ml; p = 0.2500; Fig. 2f).

Enzymatic blood group A antigen removal during hypothermic machine perfusion

After successfully achieving antigen removal during NMP, we investigated the possibility of using the enzymes during hypothermic machine perfusion (HMP) which is the simplest perfusion strategy to integrate into clinical practice. We first aimed to examine the efficacy of antigen removal during HMP over a 24 h perfusion period using a LifePort Kidney Transporter at approximately 4 °C. Six kidney pairs were perfused, where one kidney per pair was treated as before with FpGalNAc DeAc and FpGalNase, and the other kidney acted as a control. Three kidney pairs were perfused for the full period of 24 h (HMP cohort), and a further three pairs for 6 h (HMP-ABOi cohort). Biopsies taken at various timepoints (pre, 1 h, 6 h, 12 h, 24 h for HMP cohort and pre, 1–6 h for HMP-ABOi cohort) were analysed by immunofluorescence microscopy (Fig. 3a; Supplementary Fig. 2).

Fig. 3: Blood group A antigen removal during hypothermic machine perfusion (HMP).
figure 3

a Immunofluorescence images of blood group A antigen removal during HMP in control (left column) or treated (right column) type A human kidney cortical biopsies. At each timepoint (pre-treatment, 1 h, 2 h or 6 h after enzyme addition) a composite image of blood group A antigens (magenta) and H antigens (green) is shown. An 8× zoom inlay of a representative peritubular capillary (blue box) is shown in the bottom left of each panel (white box). Images are representative of a minimum of n = 3 kidney pairs. Scale bar represents 100 μm. b Quantification of A and H antigen expression at 1–6 h, 12 h and 24 h after enzyme addition in control (blue) and treated (red) kidneys. Antigen expression is normalised to expression levels in the pre-biopsies. Error bars show mean ± 95% confidence intervals. Renal blood flow (RBF; panel c) and mean arterial pressure (MAP; panel d) during perfusion are shown for control and treated kidneys. Error bars show mean ± standard deviation. HMP hypothermic machine perfusion.

No significant change in blood group A antigen expression was found after 1 h of FpGalNAc DeAc and FpGalNase perfusion compared to untreated controls (p = 0.2856; Fig. 3b; Table 2). A significant decrease in expression compared to the control group was noted after 2 h (p = 0.0015), with a maximal antigen decrease to approximately 20% expression level, which was maintained up to 24 h (Fig. 3b).

H antigen expression increased by 3.36-fold after 1 h in the ABOe group compared to the slight increase in the control group (1.87-fold; p = 0.0204; Table 3). H antigen expression reached peak expression at 4 h, decreasing up to 24 h although was significantly higher than the control group at all timepoints. The increase in H antigen expression in ABOe kidneys was markedly higher during HMP than in the NMP group, reaching an approximate maximum of 35.0-fold after 4 h HMP, compared to the maximal 4.85-fold increase in the NMP group.

Effect of FpGalNAc DeAc and FpGalNase treatment on renal function during HMP

Histological assessment of H&E-stained serial biopsies again showed signs of acute tubular injury in pre- and post-perfusion samples, but no signs of additional vascular injury or progressing ischaemic changes up to 24 h perfusion (Supplementary Fig. 3). Assessment of NGAL concentration in the circulating perfusate after 6 h of machine perfusion of control and treated kidneys showed no significant difference (control = 30 ± 28 ng/ml; treated = 23 ± 13 ng/ml; p = 0.5625; Supplementary Fig. 4). These values were notably lower than the NMP cohort due to the suppressed metabolism in hypothermic vs normothermic conditions.

Functional parameters were continuously assessed, with no significant difference observed between control and treated kidneys in terms of RBF at 6 h (p = 0.9375), or 24 h (p = 0.5000; Fig. 3c). Similarly, no significant difference was observed for MAP at 6 h (p = 0.2500) or 24 h (p > 0.9999; Fig. 3d). During HMP, the cold perfusion solution is continuously recirculated from the central chamber containing the submerged kidney and so urine output is not monitored.

Comparison of blood group A antigen removal during NMP vs HMP

A direct comparison of the kinetics of A antigen removal in kidneys treated during NMP vs HMP is summarised in the heat map in Fig. 4, where both strategies reached comparable levels of antigen removal after approximately 2 h. As enzyme treatment during HMP is the most clinically translatable strategy at present, we proceeded with further experiments using kidneys perfused during HMP for 6 h to ensure full conversion.

Fig. 4: Heat map summary comparing blood group A antigen loss in NMP vs HMP for FpGalNAc DeAc and FpGalNase-treated kidneys.
figure 4

Y-axis indicates the hours on perfusion post-enzyme addition. Colour gradient represents percentage blood group A antigen expression from 100% (purple) to 0% (white) normalised to pre-perfusion levels (0 h). Data in each column are the mean expression level from n = 3 pairs except 1 h and 6 h timepoints in the HMP column which are derived from n = 6 kidneys. NMP normothermic machine perfusion, HMP hypothermic machine perfusion.

ABOe kidneys do not bind circulating antibodies in an ABOi reperfusion model

After demonstrating that FpGalNAc DeAc and FpGalNase can remove blood group antigens during NMP and HMP, we sought to assess whether ABOe kidneys could withstand the immunological challenge of ABO-incompatible conditions in a potential recipient. We established an ex vivo model of an A to O transplant in the three pairs of human kidneys that underwent only 6 h HMP (HMP-ABOi cohort). Each kidney underwent 6 h HMP with or without FpGalNAc DeAc and FpGalNase treatment, as outlined previously, and were subsequently reperfused with perfusate containing anti-A and -B antibodies at normothermic temperature (37 °C). Briefly, each pair of kidneys was removed from the LifePort after the HMP phase, placed on ice, and flushed with Ringer’s solution before undergoing a 4 h reperfusion phase with a type O red blood cell-based solution supplemented with 10% human AB serum. Human type AB serum contains no anti-A or anti-B antibodies, so to induce ABOi conditions, purified and concentrated mouse monoclonal anti-A and -B IgM antibody was manually added after 20 min of perfusion to a set final titre of 1:128 of anti-A and 1:128 anti-B. Human serum was used instead of fresh-frozen plasma (FFP) to prevent the presence of anticoagulants used in plasma collection from inhibiting complement activity, although this meant our circuit contained no coagulation factors.

Haemodynamic parameters during the ABOi reperfusion phase were similar in control and treated kidneys (Fig. 5). After 4 h perfusion, no significant difference was observed between incompatible control kidneys (ABOi) and enzyme-treated (ABOe) in terms of the RBF (p = 0.2500; Fig. 5a) or MAP (p = 0.7500; Fig. 5b) which is to be expected as our model did not include coagulation factors. Total urine output was numerically higher in the ABOe group, but this did not reach significance (ABOi = 42 ± 15 ml; ABOe = 74 ± 36 ml; p = 0.5000; Fig. 5c). Estimated glomerular filtration rate (eGFR) was also comparable between ABOi and ABOe kidneys during reperfusion (Fig. 5d). The kidney injury marker NGAL was measured in the perfusate and urine after 4 h reperfusion with no significant difference observed between groups (perfusate: p = 0.5000; urine: p = 0.2500; Supplementary Fig. 5a). H&E injury grading showed no evidence of increased vascular injury or ischaemic damage in the ABOi group (Supplementary Fig. 5b).

Fig. 5: Haemodynamic parameters during ABO-incompatible reperfusion.
figure 5

Measurements of renal blood flow (RBF; a), mean arterial pressure (MAP; b), urine output (c), and eGFR (d) for control and treated kidneys during 4 h ABOi reperfusion for n = 3 kidney pairs. Error bars show mean ± standard deviation. eGFR estimated glomerular filtration rate, ABOi ABO-incompatible.

To assess whether the circulating antibody in the reperfusion phase could bind the vascular surface of enzyme-treated ABOe kidneys, biopsies were stained with fluorescently tagged goat anti-mouse-IgM (Fig. 6a). In ABOi kidneys, mouse IgM deposits were found on the surface of the peritubular capillaries in post-reperfusion biopsies, indicating binding in situ during active perfusion. Endothelial IgM deposition was significantly higher in all ABOi kidneys compared to their ABOe pairs where minimal binding was observed (p < 0.0001; Fig. 6b). This indicates that ABOe kidneys do not bind circulating anti-A antibodies in our model of ABO-incompatible reperfusion.

Fig. 6: Anti-A binding during ABO-incompatible reperfusion.
figure 6

a Immunofluorescence images of cortical kidney biopsies pre-reperfusion and 4 h post-reperfusion. Anti-IgM staining is shown in yellow with a representative capillary selected (blue box) for an 8× zoom inlay (white box). Capillaries in IgM negative sections were selected based on anti-H staining (separate channel, not shown). Scale bar shows 100 μm. b Quantification of IgM staining in control vs treated kidneys. Six fields of view were selected per kidney (n = 3) for 18 images total per group. Data were compared between control and treated groups with a two-tailed Wilcoxon matched-pairs signed rank test. c Representative histogram of IgM-stained RBCs in perfusate samples taken before (pre), immediately after (0 h), and 1–4 h after antibody addition in one control (ABOi) and one treated (ABOe) kidney from a single biological pair. MFI of all kidneys at each timepoint is shown to the right per histogram. d MFI values at all time points normalised to 0 h MFI for n = 3 control and treated kidney pairs. In all cases, error bars show mean ± standard deviation. ****p < 0.0001. A.U. arbitrary units, RBCs red blood cells, MFI median fluorescence intensity.

To further analyse antibody-antigen interaction in ABOi and ABOe kidneys, the relative amount of anti-A in the perfusate over time was assessed with a red blood cell flow cytometry assay. Briefly, samples of perfusate containing antibody were incubated with fixed human type A red blood cells (RBCs) which were then washed and incubated with a fluorescent anti-mouse-IgM antibody (AF647). The median fluorescence intensity (MFI) of serial samples was normalised to the MFI of samples taken immediately after antibody addition. Anti-mouse-IgM MFI decreased steadily during perfusion in the ABOi control group but remained constant in the ABOe group (Fig. 6c, d). Overall circulating antibody levels reached a low of 28.4% ± 7.4% after 4 h perfusion in ABOi kidneys compared to 101.3% ± 21.6% in ABOe kidneys. The findings are consistent with circulating antibody levels decreasing as anti-A antibodies bound free antigens on the vascular endothelium. ABOe kidneys retained sufficiently few blood group A surface antigens to have any effect on circulating anti-A antibody levels after 4 h, showing a promising feasible strategy of immune evasion of ABOi grafts in potential transplant recipients.

Activation of the classical complement pathway in ABOi but not ABOe kidneys

As a feature of acute antibody-mediated rejection is activation of the classical complement pathway, activation components of the complement cascade were assessed in the perfusate and tissue of control (ABOi) and ABOe kidneys during reperfusion. The classical complement pathway is initially activated by IgM or IgG1 or IgG3 subclasses of IgG immunoglobulins binding their antigen on the cell surface and engaging the C1q-r2s2 complex9. A detailed summary of complement activation is demonstrated in Fig. 7a. The presence of C1q, C4d, and C5b-9 (membrane attack complex) in the tissue was assessed by immunofluorescence microscopy while the concentration of the anaphylatoxin C5a and soluble C5b-9 (sC5b-9) were analysed in the perfusate and urine.

Fig. 7: Activation of the classical complement cascade during ABOi reperfusion.
figure 7

a Schematic overview of the activation of the complement cascade, showing activation of the classical activation pathway (top panel), and terminal activation pathway (bottom panel). b Representative immunofluorescence images of kidney cortical biopsies after 4 h ABO-incompatible reperfusion in control (ABOi) and treated (ABOe) kidneys stained for C1q (top row; grey), C4d (middle row; grey), and C5b-9 (bottom row; grey) for n = 3 biological replicates. An 8× zoom inlay of a representative peritubular capillary (blue box) is shown in the bottom left of each panel (white box). Composite images also show IgM (yellow) and H antigen (green) staining. Scale bar represents 100 μm.

C1qA, a subunit of the C1q protein, was observed to bind the microvasculature in ABOi but not in ABOe tissue and directly colocalised with the presence of anti-A mouse IgM (Fig. 7b). Further activation components C4d and C5b-9 also showed discrete localisation to the microvasculature in ABOi but not ABOe tissue. There was no difference in the concentration of released soluble activation factors (C5a and sC5b-9) in the perfusate or urine between the two conditions at any timepoint, however (Supplementary Fig. 6). This may be due to the subtle changes in soluble component concentrations being masked by acute, non-sustained complement activation responses due to ischaemia reperfusion injury10. These results show firstly that the binding of mouse anti-A IgM in this model of ABOi reperfusion is capable of engaging C1q binding to initiate activation of the classical complement cascade. Secondly, we can conclude that ABOe kidneys which do not bind circulating anti-A antibodies can avoid complement activation up to 4 h after reperfusion, demonstrating a key method of potentially evading hyperacute antibody-mediated rejection in a transplant setting.