CRLF2 over-expression is a poor prognostic marker in children with high risk T-cell acute lymphoblastic leukemia

Pediatric T-ALL patients have a worse outcome compared to BCP-ALL patients and they could benefit from new prognostic marker identification. Alteration of CRLF2 gene, a hallmark correlated with poor outcome in BCP-ALL, has not been reported in T-ALL. We analyzed CRLF2 expression in 212 T-ALL pediatric patients enrolled in AIEOP-BFM ALL2000 study in Italian and German centers. Seventeen out of 120 (14.2%) Italian patients presented CRLF2 mRNA expression 5 times higher than the median (CRLF2-high); they had a significantly inferior event-free survival (41.2%±11.9 vs. 68.9%±4.6, p=0.006) and overall survival (47.1%±12.1 vs. 73.8%±4.3, p=0.009) and an increased cumulative incidence of relapse/resistance (52.9%±12.1 vs. 26.2%±4.3, p=0.007) compared to CRLF2-low patients. The prognostic value of CRLF2 over-expression was validated in the German cohort. Of note, CRLF2 over-expression was associated with poor prognosis in the high risk (HR) subgroup where CRLF2-high patients were more frequently allocated. Interestingly, although in T-ALL CRLF2 protein was localized mainly in the cytoplasm, in CRLF2-high blasts we found a trend towards a stronger TSLP-induced pSTAT5 response, sensitive to the JAK inhibitor Ruxolitinib. In conclusion, CRLF2 over-expression is a poor prognostic marker identifying a subset of HR T-ALL patients that could benefit from alternative therapy, potentially targeting the CRLF2 pathway.


CRLF2 rearrangements
The presence of the P2RY8-CRLF2 fusion transcript in the AIEOP and BFM-G cohort was investigated by RQ-PCR, as previously described. [3] In particular, the UPL System was used, with primers designed in the first exon of P2RY8 (5'-GCTACTTCTGCCGCTGCTT-3') and in the first exon of CRLF2 (5'-GCAGAAAGACGGCAGCTC-3') with the UPL probe n. 28 (Roche UPL cat. n. 04687604001).
IGH@-CRLF2 translocationor any other rearrangement involving CRLF2 was searched in CRLF2 over-expressed AIEOP patient, for which fixed cells from BM at diagnosis were available, by Fluorescence in situ hybridization (FISH) on interphase nuclei using CRLF2 Breakapart Probe (Cytocell Ltd, Cambridge, UK). Analyses were carried out using Zeiss Axio Imager Z2 fluorescent microscope (Carl Zeiss AG Corporate, Oberkochen, Germany) and ISIS software (MetaSystems GmbH, Altlussheim, Germany). For each case 150/200 interphase nuclei were scored.

Other genetic aberrations
High Resolution Melting (HRM) analysis was performed to identify JAK2 mutations in exon 16, as previously described [4].

Cell culture
The human T-ALL cell lines LOUCY (a kind gift of DSMZ, Germany), MOLT-4, CCRF-CEM and Jurkat were cultured in RPMI medium with 10-20% bovine calf serum.

Immunofluorescence-analyses
Cells were fixed in 4% paraformaldehyde, resuspended in PBS, and let to dry out at RT on gelatincoated slides. After re-hydration, samples were treated briefly with 0,1 M Glycin in PBS (pH 7.4) followed by 0.3% Triton X-100 buffer. Cells were incubated O/N at 4°C with primary Abs, (goat anti-human CRLF2 antibody AF981, R&D Systems and rabbit anti-Calnexin, Stressgen, Victoria, Canada) washed and incubated for 1h at RT with secondary Abs (488-Donkey anti-goat, Abcam, Cambridge, UK and 555-Donkey anti-rabbit, Immunological Sciences, Rome, Italy) and 594-WGA (Life Technologies, Carlsbad, CA, USA). For nuclei staining slides were treated with Toto-3 iodide 642/660 (Life Technologies). Confocal microscopy was carried out on a Radiance 2100 microscope (Biorad Laboratories, Hercules, CA, USA) equipped with a Red laser diode and Kripon/Argon laser.

Gene-expression and gene set enrichment analysis
An independent cohort was used to perform Gene expression profiling (GEP) analysis. RNA samples of 100 T-ALL (AIEOP ALL study cohort, diagnosed from 2000 to 2006) were processed according to Affymetrix protocols as previously described. [7] GeneChip Human Genome U133 Plus 2.0 array were used and microarray data (.CEL files) were generated from raw signals using integrated microarray Affymetrix software. Microarray data, normalized by the justRMA algorithm, were analyzed by R-Bioconductor (Version 2.15.3). The expression values of CRLF2 probe 208303_s_t were analyzed in the 100 T-ALL specimens using the same cut-off values as previously established for CRLF2 RQ-PCR expression values. Using all gene expression data of .CEL files the 15% of patients with highest CRLF2 expression were compared with the 15% of patients with lowest CRLF2 expression. Differentially expressed probes between the two groups (CRLF2-high vs. CRLF2-low) groups were obtained using Wilcoxon T-test and local false discovery rate (lfdr) was used to correct the p-value. A lfdr <0.05 was considered significant for probe sets differentially expressed between compared groups.
Gene set enrichment analysis (GSEA) was run on the differentially expressed genes resulting from the Wilcoxon test in order to explore presence of specific pathways and oncogenic signature defined directly from microarray gene expression data.