Enhancement of red blood cell transfusion compatibility using CRISPR-mediated erythroblast gene editing

Abstract

Regular blood transfusion is the cornerstone of care for patients with red blood cell (RBC) disorders such as thalassaemia or sickle-cell disease. With repeated transfusion, alloimmunisation often occurs due to incompatibility at the level of minor blood group antigens. We use CRISPR-mediated genome editing of an immortalised human erythroblast cell line (BEL-A) to generate multiple enucleation competent cell lines deficient in individual blood groups. Edits are combined to generate a single cell line deficient in multiple antigens responsible for the most common transfusion incompatibilities: ABO (Bombay phenotype), Rh (Rh_null), Kell (K₀), Duffy (Fy_null), GPB (S-s-U-). These cells can be differentiated to generate deformable reticulocytes, illustrating the capacity for coexistence of multiple rare blood group antigen null phenotypes. This study provides the first proof-of-principle demonstration of combinatorial CRISPR-mediated blood group gene editing to generate customisable or multi-compatible RBCs for diagnostic reagents or recipients with complicated matching requirements.

Keywords: CRISPR; BEL‐A; erythroid; transfusion; universal donor.

Figure 1. Flow cytometric confirmation of individual blood group knockouts in BEL‐A‐derived reticulocytes

BEL‐A blood group knockout lines were created using lentiviral CRISPR–Cas9. Knockout lines were single cell sorted into clonal sub‐lines which were differentiated for 14 days. Enucleated reticulocytes were identified based on negativity for Hoechst 33342. Expression levels of targeted blood group antigens in knockout lines overlay with IgG controls indicating complete protein knockouts. RHAG knockout was screened with both LA1818 (anti‐RhAG) and BRIC 69 (anti‐Rh) in order to confirm RHAG knockout and Rh_null phenotype.

Figure EV1. Flow cytometric analysis of major erythrocyte membrane proteins in individual blood group knockout BEL‐A reticulocytes

No unexpected alterations in expression of band 3, GPA, GPC, RhAG or Rh proteins compared to untransduced BEL‐A controls were observed. As expected, CD47 expression was reduced in the RhAG knockout line due to disruption of the Rh subcomplex.

Figure 2. Characterisation of 5× KO BEL‐A reticulocyte phenotype

The 5× KO BEL‐A cell line was created using lentiviral CRISPR–Cas9 targeted to five blood group genes: KEL, RHAG, ACKR1, FUT1 and GYPB. Knockout cells were sorted into clonal sub‐lines which were differentiated for 14 days to generate reticulocytes for analysis.

1. Hoechst‐negative untransduced BEL‐A‐derived reticulocytes are positively labelled by antibodies to indicated blood groups/antigens. 5× KO reticulocytes labelled with antibodies to targeted blood group proteins/antigens are completely deficient in expression.

2. Representative cytospin images illustrate similar morphology of leukofiltered reticulocytes derived from untransduced and 5× KO BEL‐A cell lines.

3. Indirect antiglobulin test using column agglutination of BEL‐A reticulocytes. Absence of GPB, H antigen, Duffy, Kell and Rh in 5× KO reticulocytes is supported by IAT tests with anti‐U, anti‐H, anti‐Fy3, anti‐Ku and anti‐Rh29 antibodies, respectively. In contrast to the untransduced control cells, 5× KO cells did not agglutinate upon exposure to any of the tested antibodies and cell pellets were observed at the bottom of the microtubules in all tests.

4. Deformability index of untransduced control and 5× KO BEL‐A cell line‐derived reticulocytes determined using an Automated Rheoscope Cell Analyser. Untransduced BEL‐A control n = 11, 5× KO n = 5. Error bars indicate standard deviation.

5. Scatter plot depicting relative protein abundance of membrane and cytoskeletal proteins in reticulocytes derived from 5× KO compared to untransduced BEL‐A cells as identified by TMT labelling and mass spectrometry. Data were categorised to identify membrane and cytoskeletal proteins using Proteome Discoverer 2.1. Log₂ fold ratios are based on the mean of two technical replicates. Data were filtered using a FDR of 1% with exclusion of proteins for which only a single peptide was detected.

Figure EV2. Extended serological analysis of 5× KO BEL‐A reticulocytes

Gel card indirect antiglobulin tests support the absence of GPB, H antigen, Kell, Rh and Duffy in 5× KO cells using anti‐U, anti‐H, anti‐Ku, anti‐Rh29 and anti‐Fy3, respectively. Cellular controls included unedited BEL‐A reticulocytes, positive control RBCs and negative control RBCs. AB serum controls were performed with both anti‐mouse and anti‐human secondary antibodies.

Figure EV3. Flow cytometric analysis of major erythrocyte membrane proteins in 5× knockout BEL‐A reticulocytes

No alteration in levels of band 3, GPA or GPC confirms the absence of gross membrane disruption. As expected, CD47 expression was reduced due to disruption of the Rh subcomplex.

Figure EV4. Quantitative proteomics of 5× KO and untransduced BEL‐A reticulocytes

Scatter plot depicting relative protein abundance of all proteins detected in reticulocytes derived from 5× KO compared to control BEL‐A cells as identified by TMT labelling and mass spectrometry. Log₂ fold ratios are based on the mean of two technical replicates. Data were filtered using a FDR of 1% with exclusion of proteins for which only a single peptide was detected. The full proteomic data set is included in Dataset EV1.