High Insta High Instability Of The CSF3R Gene In Severe Congenital Neutropenia Patients As Judged By Multiple CSF3R Mutations In The Majority Of Patients

It has been reported by us and others that approx. 30 % of all patients with congenital neutropenia (CN) acquire CSF3R mutations in their life time. More than 80 % of the CN patients who develop myeloid leukemia (CN/AML) harbour CSF3R mutations. This suggests that they are the first hit in leukemogenesis. However, detecting sequence changes e.g. by Sanger sequencing reveals only mutations presented in more than 20 % of the cells due to its technical detection limit. Therefore, we asked whether there is a systematic underestimation of cell clones harbouring CSF3R mutations, which might have been traditionally overlooked. We applied the deep-sequencing technology (SOLID 5500xl) to identify CSF3R mutations in myeloid cells from 158 patients with different types of neutropenia (86 severe congenital neutropenia (CN) patients with known inherited mutations (ELANE, HAX1, G6PC3), 21 cyclic neutropenia (CyN) patients, 28 patients with severe chronic neutropenia with so far unknown inherited mutations, 11 patients with SBDS-associated neutropenia) as well as a group of 12 healthy individuals. All neutropenia patients were treated with G-CSF and notably 21 CN patients developed leukemia or MDS. Deep sequencing data were processed according to our custom NGS pipeline (annotation of sequences and prediction of damaging effects on the coding sequence by Polyphen2, removal of known dbSNP variants, and accepting significant Phred-scores at the variant calling stage). Overall the read numbers ranged between 18 and 128069 (median 716), while only variants with at least two percent of the reads were accepted for further consideration (the statistically significant limit is between one and two percent of all reads). All together, we detected 92 CSF3R mutations in 42 CN patients leading to 49 distinct amino acid exchanges (38 missense and 11 stop-codon mutations). The frequency of the mutant alleles ranged from 2 to 96 %. In contrast, in CyN only five out of 21 patients harbour CSF3R mutations; interestingly, two of them in isoform IV of CSF3R (p.P752T). Most notably, whereas 18 patients displayed only one CSF3R mutation, 24 individuals had more than one CSF3R mutation (2-10 mutations, in total 74 mutations). During follow up of some patients, we could demonstrate that the number of mutations increased over time. The majority of mutations were located in the cytoplasmatic region (aa 651-831) of CSF3R, while 15 patients presented mutations within the extracellular region of CSF3R. Intriguingly, in 16 patients we detected 23 non-sense mutations, where 20 of these are stop-codon mutations affecting glutamine (Q) 768, 770, 776, and 781. This suggests that this part of CSF3R is highly instable. In two patients who did not respond to Filgrastim treatment, we detected a stop codon at aa 546 and 547, respectively, affecting the Fibronectin type-III like part of the CSF3R. Twelve patients who developed leukemia (CN/AML) had more than one CSF3R mutations (two to ten) , whereas eight with CN/AML harbored only one mutation.

None of the healthy controls, only three neutropenia patients with unknown inheritance, and only one SBDS patient revealed mutations in CSF3R. Taken together, this data suggests that CSF3R is highly prone to genetic instability in severe congenital neutropenia, because more than one mutation in half of the patients was observed and various CSF3R mutations during the course of life accumulated. Once a cell clone harboring CSF3R mutation obtains a second hit (e.g. RUNX1 mutation), they are prone to undergo leukemic transformation.