Nanopore RNA-Seq to HLA genotype and correlate donor HLA expression with flow cytometric crossmatch results - Eric Weimer

Eric opened his talk by exploring the development and value of the crossmatch test to prevent immediate graft loss in kidney transplantation, the main transplant that occurs in the United States. Paul Terasaki’s group postulated that pre-formed antibodies were responsible for immediate failure of kidney transplants, and as a result they developed the crossmatch test using prospective recipient serum and donor cells. This test has been cited nearly 1,000 times and forms the basis of pre-transplant testing in the US today - every patient now gets a crossmatch test done prior to receiving an organ.

To determine donor compatibility, crossmatch tests performed have included: the complement-dependent cytotoxicity (CDC) crossmatch test, flow cytometric crossmatch (FCXM), and virtual crossmatch (VXM). The CDC crossmatch was the first approach implemented, from which they moved to flow cytometric matching, and now virtual crossmatching is typically used. The significant benefit of virtual crossmatching means that we know a patient's antibody reactivity before a potential donor even comes up, making the transplant process more efficient, which is vital for a process with such tight time constraints.

In addition to crossmatching, donor HLA typing is also performed using qPCR prior to organ matching. The antibody profile as well as donor HLA type can therefore be used for matching with a suitable donor.

Eric next explained how lymphocyte HLA expression, as determined by FCXM, has been found to vary by both donor and tissue, and that a reduced level of expression impacts the outcome of the FCXM crossmatch test. By extension, this therefore impacts if the transplant will proceed. He noted that because B lymphocytes express both HLA class I and class II molecules, these are the cell types investigated, in comparison to T lymphocytes, which express only HLA class I.

Why investigate HLA expression based on molecular measurements, rather than flow-cytometric measurements?

Eric explained that, firstly, not a lot of monoclonal antibodies are available to cover all the HLA alleles possible, so it is not a good option to determine lymphocyte HLA expression. Secondly, flow cytometry is time inefficient compared to molecular-based methods. He did caution however that transcript levels do not necessarily equate to protein levels (e.g. due to post-translational modification). Nonetheless, an RNA-based assay would enable you to do two things at once - genotype HLA and measure HLA expression levels.

The question is, can this be done quickly each for deceased donors? Eric said that there is only a 6-7 hour period from receipt of the specimen to reporting donor compatibility, and this includes all analysis; time is therefore crucial in crossmatching.

In a proof-of-concept study, Eric and his team performed nanopore whole-transcriptome sequencing to measure HLA genotyping and expression, simultaneously. This workflow involved magnetic negative separation of lymphocytes from blood of healthy donors, magnetic mRNA isolation, RNA clean-up and concentration, and then library preparation using the PCR-cDNA sequencing kit and PCR barcoding kit, followed by 18 hours sequencing on the MinION. Athlon was used for HLA genotyping and the CLC Genomics Workbench was used for alignment and normalisation. On average, 1.8 million reads were obtained per sample, of which 84.7% mapped to the genome; a mean of 0.11% of the reads were specific for HLA class I.

Only expression of the HLA class I locus was measured; HLA class II expression was too low because HLA class II-expressing B-cells were too small a proportion of the total lymphocyte population to measure, which is mostly comprised of T-cells.

Eric identified that HLA class I expression varied among different individuals and at different time points, but remained relatively consistent within an individual. HLA-B had the highest expression, followed by HLA-A, and lastly HLA-C, which is consistent with previous reports of transcript levels measured for HLA class I. Eric said that he anticipated seeing these results reflected in protein-level data.

The Athlon HLA genotyping data showed accurate genotyping for HLA classes A and B, but inaccurate HLA-C genotyping, which correlated with the low transcript levels of the alleles affected (HLA-C*07:01 or HLA-C*07:02). They were, however, able to downsample to only 250 reads for accurate genotyping.

As "the whole point of this work was to incorporate this into some sort of virtual crossmatch assessment", they next investigated the correlation of expression with flow cytometric crossmatch results. Eric displayed these data, saying that they found that there was high correlation between them - if there was low RNA expression, there was a low crossmatch outcome.

Eric concluded that these data are a great first step in determining that we can use this method with crossmatch donors going forward.