PAX5 Loss Compromises CD22 Protein Levels and Responses to Inotuzumab Ozogamicin in B-Cell Acute Lymphoblastic Leukemia

One of the main causes of immunotherapy failures in B cell acute lymphoblastic leukemia (B-ALL) is either complete loss of the cognate epitope (frequently seen with CD19-directed CAR T cells) or the decreased epitope density. The latter scenario is exemplified by CD22low relapses reported in B-ALL patients treated with either CD22-directed CAR T cells or the antibody-drug conjugate inotuzumab ozogamicin. Previously published work identified frequent skipping of CD22 first coding exon as one rate-limiting factor, but little is known about possible involvement of transcriptional mechanisms. Here we set out to identify key regulators of CD22 surface expression in an unbiased way using a genome-wide CRISPR/Cas9 screen performed on commonly used Nalm-6 and Reh B-ALL cell lines.

These two cell lines were transduced with the Brunello library targeting 19,114 genes with a total of 77,441 sgRNAs (4 sgRNAs per gene) + 1000 non-targeting sgRNAs. sgRNA-expressing cells were tracked by green fluorescent protein (GFP) expression. Then they were sorted into CD22-bright (top 10%) and CD22-dim (bottom 10%) bins. Approximately 2-3 million cells from each bin were collected, and genomic DNA extraction was performed. CRISPR mutagenesis libraries were obtained by PCR amplification and subsequently subjected to Next Generation Sequencing. The sgRNA counts were normalized using the non-targeting controls present in the library as implemented in the Model-based Analysis of Genome-wide CRISPR/Cas9 Knockout (MAGeCK) algorithm, which was also used to identify significantly enriched and depleted guides. High-confidence hits in each cell line were identified, and the intersection between the two lists was chosen for further consideration. Topping the list of CD22 activators were three transcription factors known to be involved in B-lymphoid development: PAX5, EBF, and RUNX1.

To determine the pathophysiological relevance of these findings to human disease, we analyzed a previously generated RNA-seq dataset corresponding to B-ALL samples from patients on the Children's Oncology Group AALL1621 phase II clinical trial (NCT02981628) of inotuzumab. Specifically, we performed regression analysis of PAX5/EBF/RUNX1 vs CD22 mRNA expression levels. We observed the strongest (r=0.513) and the most significant correlation (p=0.0212 per ANOVA) between PAX5 and CD22 levels, suggesting that the former is the key positive transcriptional regulator of the latter in at least in immunotherapy-naïve B-ALL. Furthermore, this trend also held in patients pre-treated with inotuzumab ozogamicin (the smaller NCI pediatric B-ALL cohort; r=0.335).

As the PAX5 gene is frequently deleted or inactivated in B-ALL, we determined the effects of acute PAX5 loss on CD22 protein levels. To this end, we transfected Nalm-6 cells with Cas9 ribonucleoprotein particles containing a PAX5-targeting sgRNA, selected for genome-edited cells, and validated gene knockout using RT-PCR and Western blotting. To quantify the number of CD22 molecules/cell, we employed flow cytometry with Quantum™ Simply Cellular® (QSC) microspheres. We reproducibly observed ~3-fold reduction in surface CD22 in PAX5 sgRNA-transfected cells compared to their control guide-transfected counterparts. To determine whether PAX5 regulates CD22 at the transcriptional or pre-mRNA splicing level, we amplified the entire CD22 cDNA using gene-specific primers and subjected the PCR products to Oxford Nanopore Technologies long-read sequencing. We observed no change in the inclusion of the first coding exon, suggesting that, as anticipated, splicing is not the mechanism by which PAX5 regulates CD22 protein output.

Finally, we treated PAX5 KO Nalm-6 cells with inotuzumab ozogamicin at varying concentrations to generate dose response curves. We reproducibly observed ~7-fold increase in half maximal inhibitory concentration (IC50) of this drug in PAX5-null cells. Collectively, these findings implicate PAX5 as an important determinant of sensitivity to inotuzumab ozogamicin and PAX5 loss-of-function mutations - as possible biomarkers of therapeutic resistance.

No relevant conflicts of interest to declare.