A Whole Genome In Vivo Crispr Screen in Primary ALL Predicts Leukaemic Relapse

Although paediatric acute lymphoblastic leukaemia (ALL) has a favourable prognosis, a number of cases will invariably relapse. One of the major problems associated with relapse is drug resistance, in particular to glucocorticoids, the mainstay of ALL treatment. Examining the underlying mechanisms is complicated by clonal heterogeneity within a patient and the potential impact of the leukaemic niche.

To address mechanisms of drug resistance in a patient-relevant setting, we performed a genome-wide in vivo CRISPR screen in primary ALL material. To that end, we took advantage of primografted material from patient L707, who initially presented with a Dexamethasone (DEX) sensitive t(17;19) ALL, but relapsed 5 months after initial diagnosis. We transduced DEX sensitive presentation cells with the full genome GeCKOv2 CRISPR library, before transplantation into immunodeficient NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice. Mice were subsequently treated with DEX by oral gavage (15mg/kg for 5 weeks, 10mg/kg thereafter).  DNA from several engrafted sites in the mouse was extracted and PCR amplified before being sequenced on the Illumina HiSeq2500. Changes in pool complexity were analysed using MaGEcK software to determine which sgRNAs were significantly enriched or depleted. By far the most significantly enriched sgRNAs were those targeting NR3C1, the gene encoding the glucocorticoid receptor. In addition, two of the top five significantly depleted sgRNAs targeted the Plexins, PLXNA1 and PLXND1. Whilst PLXNA1 is expressed at low levels, PLXND1 is highly expressed and has been linked to dexamethasone resistance.

Notably, the matched relapsed material from L707 was highly DEX resistant both in tissue culture and when transplanted into NSG mice.  SNP 6.0 analysis revealed a 5q deletion in the relapse, spanning 5 genes including NR3C1. Whole genome sequencing showed this was comprised of 2 deletions both targeting NR3C1, with different breakpoints for each allele. The differential gene expression between the L707 presentation and relapse established that NR3C1 was the most significant of all the genes lost at relapse, based on gene set enrichment analysis (GSEA). This contrasts with many ALL cases, where one of the downstream effectors of apoptosis is lost as opposed to NR3C1. Growth of the relapse material in vivo and in vitro was slower than the presentation in a competitive situation, but with DEX treatment the relapse phenotype began to emerge with a small percentage of cells showing a heterozygous deletion of NR3C1. These combined data strongly suggest that the NR3C1 deletion is the main driver of DEX resistance in the L707 relapse. Moreover, it proves that our in vivo CRISPR screen predicted the leukaemic relapse.

These results confirm NR3C1 deletion as a driver in glucocorticoid resistance and demonstrate the power of in vivo CRISPR screens to predict mechanisms of gain of drug resistance and subsequent relapse. The parallels that can be drawn between the relapse and the CRISPR screen are striking, giving the indication that the progression from presentation to relapse may follow the same path in a patient derived xenograft setting as it did in the patient.