AUTHORS
Daniele Colazzo, Roberto Merlo, MS2, Danilo Tarantino, PhD3, Alberto Caregari4, Enrico Gaffo, PhD5, Manjola Balliu, PhD6, Alessio Enderti, MSC7, Paola Guglielmelli, MD, PhD8, Niccolò Bartalucci, PhD9, Stefania Bortoluzzi, PhD10, Alessandro Vannucchi, MD11
AFFILIATIONS
1 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
2 Department of Oncology Surgery and Gastroenterology, Department of Biology, University of Padova, Padova, Italy
3 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
4 Department of Oncology Surgery and Gastroenterology, Department of Biology, University of Padova, Padova, Italy
5 Department of Oncology Surgery and Gastroenterology, Department of Biology, University of Padova, Padova, Italy
6 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
7 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
8 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
9 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
10 Department of Oncology Surgery and Gastroenterology, Department of Biology, University of Padova, Padova, Italy
11 Department of Experimental and Clinical Medicine, University of Florence, Centro di Ricerca e Innovazione Malattie Mieloproliferative (CRIMM), AOU Careggi, Florence, Italy
ABSTRACT
Background. Ruxolitinib (RUX) is a cornerstone of therapy for patients with myelofibrosis (MF). However, approximately half of responding patients lose clinical benefit within 3–5 years with subsequent poor outcomes. Reliable biomarkers predicting RUX resistance (RR) and loss of spleen response are lacking. Circular RNAs (circRNAs) have emerged as key regulators of chemoresistance in different hematologic malignancies, but their role in MF and RR remains unexplored.
Aim. To characterize the MPN circRNAome and identify RR biomarkers using in vitro models and patient samples.
Methods. RR was modelled in vitro by chronic exposure of parental SET2 cells (SET2-P) to pre-determined RUX IC50 (0.25µM), generating RR SET2 cells (SET2-R). SET2-R cell line exhibited <20% cell number reduction following RUX treatment compared with SET2-P. After circRNA enrichment, the full-length circRNAome of different SET2 cell lines was characterized by long-read sequencing (LRS), detected and annotated using CIRI-long. DESeq2 has been used to test for differential expression. Selected circRNAs were further analyzed in SET2 cell lines and in longitudinal samples (purified granulocytes) of MF patients treated with ruxolitinib (n=8): at diagnosis (D), at best spleen response (S), at loss of spleen response (R), and, in 5 patients, at additional time points between S and R timepoints (pre-R). Spleen response, and loss of it, were defined according to revised IWG-MRT criteria. In vitro silencing was used to functionally validate one circRNA compared with non-silencing control cells (ctrl).
Results. SET2-R cells cultured in the presence of RUX (0.25 µM) exhibited significantly attenuated reduction in cell number (-14.8%, P<0.001) compared with SET2-P cells exposed to the same concentration of RUX (-54.2%, P<0.001), and in cell proliferation (-16.3%±4.6% versus 50.0%±4.6%, respectively (P=0.0002). RUX-induced G1 cell-cycle arrest observed in SET2-P (+14.4% from baseline; P<0.0001) was reversed in SET2-R (-8.3% from baseline; P=0.0477), which also displayed lower levels of Annexin-V+ cells (2.59% vs 7.74%). Overall, LRS identified 12,992 circRNAs whose expression profiles discriminated SET2-R from untreated and RUX-treated parental SET2 cells by unsupervised analysis. SET2-R showed a dramatic shift in the circRNAome compared with SET2-P with 4,843 differentially expressed circRNAs, of which 3,489 (72%) were upregulated. A core set of 13 most differentially expressed circRNAs derived from RR-associated signature, including four distinct RUNX1-derived isoforms spanning exons 4 to 8, circSH2B3, circPPIB, and circCDYL, was validated by qRT-PCR. None of the corresponding linear transcripts mirrored the circRNA expression patterns: 5 remained unchanged and 8 displayed discordant expression. Consistent with findings in SET2 cells, the previously unreported circRUNX1(6–8) (Log2Fold Change (LFC) range 0.6 -3.0; P=0.021), circSUGP1 (LFC range 0.03 -1.6; P=0.022), and circSH2B3 (LFC range 0.6 -4.3; P=0.041) resulted downregulated in longitudinal patient samples at spleen response compared with pre-treatment. Remarkably, circRUNX1(6–8) expression levels resulted significantly increased at RR onset (LFC range:0.88–4.27; P=0.0029) in all patients. Analysis of 4 longitudinal samples per patient (n=5) showed that circRUNX1(6–8) levels increased from spleen response timepoint as early as 24 months before occurrence of spleen response loss (mean LFC:1.88 vs S; P=0.0375), with further increase 12 months before RR (mean LFC:2.48 vs S; P=0.0112). Functional investigation showed that circRUNX1(6-8) silencing in SET2-R (-40.3±3.74% vs ctrl P=0.0048) significantly impaired proliferation (-30.8%±13.1% vs ctrl; P=0.0001) and restored G1 cell-cycle arrest (+6.8% vs ctrl, P=0.038) counteracting the resistant phenotype (S phase: -7% vs ctrl, P=0.018). Of note, among the predicted target genes of the identified circRUNX1 isoforms, IGF1 (LFC:-0.89; P=0.0002), HMGA2 (LFC:-0.63; P<0.0001), and BCL2 (LFC:0.26; P=0.0129) were downregulated, while CSF3 was upregulated (LFC:1.36; P<0.0001) following circRUNX1(6–8) knockdown in SET2 cells.
Conclusion. We established an in vitro RR SET2 model and characterized the circRNAome by LRS; findings in SET2-R cells were validated in longitudinal patient samples. circRUNX1(6–8) was identified as a non-invasive biomarker predicting loss of spleen response and might represent a promising therapeutic target for fighting the JAK2 inhibitor resistance.
DISCLOSURES
Daniele Colazzo: No relevant conflicts of interest to declare.