miR-15a targets the HSP90 co-chaperone Morgana in chronic myeloid leukemia

Morgana is a ubiquitous HSP90 co-chaperone protein coded by the CHORDC1 gene. Morgana heterozygous mice develop with age a myeloid malignancy resembling human atypical myeloid leukemia (aCML), now renamed MDS/MPN with neutrophilia. Patients affected by this pathology exhibit low Morgana levels in the bone marrow (BM), suggesting that Morgana downregulation plays a causative role in the human malignancy. A decrease in Morgana expression levels is also evident in the BM of a subgroup of Philadelphia-positive (Ph+) chronic myeloid leukemia (CML) patients showing resistance or an incomplete response to imatinib. Despite the relevance of these data, the mechanism through which Morgana expression is downregulated in patients’ bone marrow remains unclear. In this study, we investigated the possibility that Morgana expression is regulated by miRNAs and we demonstrated that Morgana is under the control of four miRNAs (miR-15a/b and miR-26a/b) and that miR-15a may account for Morgana downregulation in CML patients.

myeloid leukemia 27 , recently renamed MDS/MPN with neutrophilia 28,29 .The molecular basis behind the onset of this disease is complex and not completely understood.Indeed, BM cells are characterized by non-recurrent karyotypic abnormalities and mutations in several genes have been identified 30 .No standard of care is available for these patients and hematopoietic stem cell transplantation is the best option for eligible patients with appropriate donors.Otherwise, patients can be treated with hypomethylating agents, pegylated-interferon-α, hydroxyurea or erythropoiesis-stimulating agents.Despite these efforts, the median overall survival is between 12 and 25 months 10,[31][32][33] .
Our previous research demonstrated that Morgana is expressed at very low levels in the BM cells of patients affected by MDS/MPN with neutrophilia and is robustly downregulated in 30% of Ph + CML patients in chronic phase 11 .Ph + CML and MDS/MPN with neutrophilia are both characterized by ROCK hyperactivation.In Ph + CML, ROCK is activated by BCR-ABL 16 and is further sustained by low Morgana levels 11 .In MDS/MPN with neutrophilia the decrease in Morgana expression may represent the main mechanism for ROCK overactivation 11 .Of note, treatment with fasudil, a ROCK inhibitor in clinical use, induced apoptosis in hematopoietic BM stem cell from Morgana heterozygous diseased mice and restored the efficacy of imatinib in BM cells from Ph + CML patients expressing low Morgana 11 .Analysis of exon-sequencing data of 16 patients affected by MDS/MPN with neutrophilia 11 and search on the TCGA database on myeloproliferative neoplasms did not reveal mutations in the CHORDC1 gene.Thus, despite the clear indications supporting the role of Morgana in CML, the molecular mechanism behind the drop of its expression in patients' BM cells is still unknown.
Here, we investigated the possibility that Morgana expression levels are controlled by microRNAs (miRNAs).miRNAs are small non-coding single-stranded RNAs 34 that play a pivotal role in regulating gene expression, controlling a variety of cellular processes such as proliferation, survival, differentiation and metabolism and contributing to several human diseases 35 .miRNAs are involved in cancer initiation and progression both as oncogenes and tumor suppressors 36 .miRNAs can simultaneously control thousands of genes and regulate all the hallmarks of cancer, including tumor growth, apoptosis resistance, angiogenesis, invasion and metastasis formation [37][38][39][40] .We found that Morgana expression is regulated by miR-15 and miR-26 families and that miR-15a negatively correlates with Morgana expression in a cohort of MDS/MPN with neutrophilia and CML patients.Our findings suggest that, at least in a subgroup of patients, miR-15a may be responsible for the drop in Morgana expression and thus for the onset of myeloid neoplasms and resistance to therapy.

CHORDC1 locus deletion does not account for Morgana downregulation in CML patients
As in CML cells are often present genomic abnormalities 41,42 , we hypothesized that deletion of Morgana coding gene (CHORDC1) could account for the low Morgana levels detected in the BM of patients affected by MDS/ MPN with neutrophilia and in a subgroup of CML patients.We generated a fluorescent probe annealing to the CHORDC1 gene, located at 11q14.3 and we performed a FISH analysis on BM cells from patients affected by Ph + CML, with normal (four patients) and low (five patients) Morgana levels.Patient BMs were considered to express low Morgana when CHORDC1 mRNA level was less than half of the mean level observed in normal BMs used as controls (n = 10), as defined previously 11 (Supplementary Fig. S1a).In all the cases analyzed the CHORDC1 probe showed clearly visible signals (Supplementary Fig. S1b), excluding the possibility of Morgana locus deletion.FISH analysis was also performed on BM cells from three MDS/MPN with neutrophilia cases expressing very low Morgana levels (Supplementary Fig. S1c), but also in these patients CHORDC1 gene deletion was not observed.This data indicates that Morgana locus deletion is not the major mechanism accounting for Morgana downregulation in CML patients.

Morgana is a predicted target for miR-15a/b and miR-26a/b
Several mechanisms are described to control the expression of oncosuppressors and a number of studies have shown that miRNAs and their dysregulation are implicated in the pathogenesis of hematological malignancies [43][44][45] .Thus, in order to identify miRNAs targeting Morgana mRNA we interrogated four miRNA target prediction databases (PicTar 46 , MirMap 47 , miRDB 48 , TargetScan 49 ) (Supplementary Table S1).Among the miRNAs common to all the databases, miR-15a/b, miR-26a/b and mir-137 are the only predicted miRNAs with a conserved binding site on the CHORDC1 transcript in vertebrates.Among them, we chose to validate miR-15a/b for their well-documented role in hematological malignancies [50][51][52] and miR-26a/b for having been previously described as able to target Morgana 53,54 (Fig. 1a).Prediction databases identified one binding site for miR-15a/b and two binding sites for miR-26a/b in CHORDC1 3'UTR (Fig. 1b).We decided to validate the ability of these miRNAs to interfere with Morgana expression.

miR-15a/b and miR-26a/b regulate Morgana mRNA and protein levels
We transfected HEK293T cells, which express significant amounts of endogenous Morgana, with precursors for miR-15a and miR-15b and miR-26a and miR-26b.Quantitative real-time PCR analysis showed that pre-miRNA transfection upregulated miRNA levels (Fig. 2a) and that both 15a/b and 26a/b pre-miRNAs significantly reduced CHORDC1 mRNA expression (Fig. 2b).Then, we analyzed Morgana protein levels in HEK293T cells and we found that transfections with pre-miR-15a/b and pre-miR-26a/b reduced Morgana expression by up to 50% (Fig. 2c,d).To obtain cell lines stably overexpressing miR-15 and miR-26, we infected HEK293T cells with pLemiR vectors containing pre-miR-15a or pre-miR-26a as representatives of the two miRNA families.We achieved a 5-tenfold overexpression of the miRNAs and we detected a significant downregulation of Morgana mRNA and protein (Fig. 2e-g).This modulation has a functional impact, since the phosphorylation of LIMK, a well-known ROCK substrate, increased in cells overexpressing miR-15a and miR-26a (Fig. 2g).
The direct effect of miR-26 on CHORDC1 3'UTR has already been demonstrated by luciferase reporter assay and confirmed by disrupting miR-26b-mediated attenuation of luciferase activity upon mutation of the miR-26a/b seed sequences on CHORDC1 3'UTR by other groups 53 .To test the specificity of miR-15a/b in regulating Morgana expression, we cloned CHORDC1 3'UTR into a firefly luciferase reporter plasmid and mutations of the predicted seed-matching sites for both miR-15a and miR-15b were generated (Fig. 3a).Both wild-type and mutant constructs were transfected into HEK293T cells with pre-miR-15a, pre-miR-15b and a control pre-miR.Luciferase activity was significantly reduced when the wild-type CHORDC1 3'UTR was co-transfected with both pre-miR-15a and pre-miR-15b, compared to the negative control (Fig. 3b,c).This effect depends on a direct interaction between miR-15a and miR-15b and their seed sequences in CHORDC1 3'UTR, indeed their mutations completely abrogated the reduction in luciferase activity (Fig. 3b,c).Altogether, these data indicate that miR-15a/b and miR-26a/b can regulate Morgana at both mRNA and protein levels and may be responsible for Morgana downregulation in CML patients.

miRNA 15a levels anti-correlate with Morgana expression in CML
To evaluate the presence of an anti-correlative relationship between Morgana and miR-15 and miR-26 families in hematological malignancies, we performed a bioinformatic analysis exploiting the Cancer Genome Atlas (TCGA) dataset.The results support our experimental findings, highlighting negative correlations between the amount of the Morgana transcript (CHORDC1) and miR-26a (Fig. 4a) and miR-15b (Fig. 4b) in 47 samples of Diffuse Large B-cell Lymphoma (DLBCL) and miR-15b in 210 Chronic Lymphocytic Leukemia (CLL) samples (Fig. 4c).These results indicate that miR-15 and miR-26 families may regulate Morgana expression levels in human hematological malignancies.To verify if the expression of miR-15 and 26 correlates with Morgana downregulation in CML patients' BM (Supplementary Table S2), we tested, by quantitative real-time PCR, both miRNAs and CHORDC1 transcript expression in a cohort of 31 CML patients in chronic phase (12 Ph-and 19 Ph+).We found that miR-15a significantly anti-correlates with Morgana expression (Fig. 5a), while miR-15b and miR-26a/b did not show significant correlation (Fig. 5b-d).Indeed, miR-15a was significantly overexpressed in patients affected by CML and MDS/MPN with neutrophilia with low Morgana expression levels (Fig. 5e), while miR-15b and miR-26a/b levels were not different between Morgana normal and Morgana low groups (Fig. 5f-h).Of note, the expression levels of BCR-ABL and miR-15a significantly correlated, opening the intriguing possibility that BCR-ABL could potentiate miR-15a expression (Fig. 5i, Supplementary Fig. S2a-c).These results suggest that in CML, Morgana levels may depend on miR-15a expression.

Discussion
Morgana haploinsufficiency causes the onset of a lethal myeloproliferative neoplasm in the mouse, resembling human CML and MDS/MPN with neutrophilia.This evidence indicates a causative role of Morgana downregulation in the onset of the mouse pathology.The fact that BMs from MDS/MPN with neutrophilia patients express very low Morgana levels is suggestive of equal relevance in human pathology.Morgana expression levels are also relevant in Ph + CML, where its underexpression confers imatinib resistance by inducing ROCK hyperactivation.Indeed, the combined treatment with imatinib and a ROCK inhibitor restored an apoptotic response in BM cells from CML patients with low Morgana 11 .However, the causes of Morgana downregulation in patients' BM cells remain elusive.In this study, we addressed this issue and found that the CHORDC1 locus is not deleted in the BM from 5 CML and 3 MDS/MPN with neutrophilia patients expressing very low Morgana levels.This suggests that deletion is not the predominant mechanism through which leukemia cells suppress Morgana expression.Given that miRNA upregulation has been often described in cancer cells to promote oncosuppressor silencing 39 , we decided to investigate the relevance of miRNAs in regulating Morgana expression.We selected miR-15 and   53,54 and miR-15 and miR-16 55 .Bioinformatic analysis between Morgana expression and miR-15 and miR-26 families on the TCGA datasets revealed a significant inverse correlation with miRNAs 26a, 16a, and 15b in Diffuse Large B-cell Lymphoma and Chronic Lymphocytic Leukemia.These results suggest that these miRNAs have the potential to regulate Morgana expression in human hematological neoplasms.miRNA-26 family has aroused interest for its role in cancer where it regulates genes involved in proliferation, survival, metabolism and invasion 56 .miRNA-26a has been found to be downregulated in melanoma 57 , breast 58 , bladder 50 and prostate cancer 53 , but strongly overexpressed in glioma 59 and ovarian cancer 60 .miR-26 inhibits cell viability and invasiveness in melanoma 61 , cell aggressiveness in bladder cancer 62 and cell proliferation in ovarian cancer 63 and promotes cell sensitivity to doxorubicin in hepatocellular carcinoma cells 64 .In the hematological context, miR-26 is downregulated in Burkitt and Follicular lymphoma 65 .Furthermore, the ectopic expression of miR-26 in Diffuse Large B-cell Lymphoma (DBCL) cells decreased cell proliferation and increased apoptosis 66 .
The miR-15 family consists of three bicistronic clusters: miR-15a/16-1, miR-15b/16-2 and miR-497/195, the latter not required for hematopoietic cell function and with restricted relevance for embryonic development and healthy lifespan 67,68 .miR-15 and miR-16 were the first miRNAs to be associated with cancer acting mainly as oncosuppressors 69 .Indeed, the miR-15a/16-1 cluster located at chromosome 13q14.3 is frequently deleted or downregulated in B cell malignancies, such as chronic lymphocytic leukemias (CLLs) 69 , multiple myeloma 70 and in a fraction of Acute Myeloid Leukemia (AML) patients 71,72 .These miRNAs have been described to induce apoptosis, through the inhibition of prosurvival genes as Bcl-2 and directly or indirectly to deregulate genes involved in cell cycle progression 55,73 .
Our specific interest was to evaluate if these miRNAs are responsible for Morgana downregulation in the BM of CML patients.Our analysis on a cohort of 12 MDS/MPN with neutrophilia and 19 Ph + CML patients indicates that only miR-15a negatively correlates with Morgana expression, suggesting its relevance in regulating its levels in this specific context.Instead, miRNAs 15b, as well as miRNAs 26a and b, exhibit no significant anti-correlation.
Intriguingly, the loss of expression of miR-15/16 and the consequent overexpression of their targets have been described as critical features in the transition from chronic phase to blast crisis in CML patients 74,75 .Protein downregulation through miRNA expression is a reversible mechanism that allows adaptation to different phases of the pathology.It is possible to speculate that the overexpression of miR-15, and consequently low Morgana levels, favors the onset of the disease and sustains the chronic phase, by inducing ROCK overactivation and therefore myeloid cell proliferation and genomic instability 3,11,16 .At the opposite, during the progression to blast crisis, malignant cells may benefit more from Morgana overexpression, as well as, other miR-15/16 targets, such as Bmi-1, ROR1, and Bcl-2 74 .Indeed, Morgana is able to act also as a proto-oncogene as it is frequently found overexpressed in different types of tumors, where it sustains AKT and NF-κB activity 9,12 .Unfortunately, we were unable to analyze Morgana expression levels in patients during blast crisis, as these samples are extremely rare due to the success of TKI therapy in Ph + CML and the rarity of MDS/MPN with neutrophilia.
It is conceivable that several mechanisms may regulate Morgana expression levels and that regulation must be cell type-dependent and may vary among patients and stages of the pathology.Thus, miR-15 and miR-26 family members are likely to be only some of the players involved.To clarify the mechanism by which Morgana is downregulated in CML patients remains a crucial point to understand the onset and progression of this pathology and, consequently, to choose the best therapy to avoid resistance and design therapeutic interventions for MDS/ MPN with neutrophilia patients, for which overall survival remains very poor 10,[31][32][33] .

Fluorescence in situ hybridization (FISH)
We produced the FISH probe for Morgana from BAC RP11-1129K7 which covers the region 11q14.3 of the CHORDC1 gene.Cells containing BAC RP11-1129K7 (Invitrogen) were expanded overnight in LB medium supplemented with chloramphenicol.The BAC was extracted using Midiprep Kit JET Star 2.0, according to manufacturers' instructions.Briefly, DNA was extracted using a lysis buffer, precipitated and purified by using isopropanol and cold ethanol 75%.BAC DNA was quantified using a NanoDrop ND-1000 Spectrophotometer and 250 ng was labeled with biotin-14-dCTP and random primers from the BioPrime® DNA Labeling System, Invitrogen.The biotinylated probe was purified using illustra MicroSpin G-50 Columns, GE Healthcare.It was then precipitated and preannealed with 1 µl of Salmon Sperm DNA Solution (Life Technologies) and 30 µl of Human Cot-1 DNA® (Life Technologies).The probe was precipitated with 8 µl of sodium acetate 3 M and 160 µl of ethanol at − 20 °C overnight, then it was washed by cold ethanol 75% and dried.The probe was resuspended in the LSI Buffer (Abbott).The centromere-specific probe (CEP11 Spectrum Green, Vysis Abbott) for chromosome 11 was used as technical control.10 µµl of solution composed by 1 µl of Morgana probe, 1 µl of CEP11 and 8 µl of LSI buffer (Abbott) were used on each chromosome spread slide.Chromosome spreads in metaphase and interphase were obtained by culturing cells with 0.2 µg/ml of Colcemid (Invitrogen) overnight or for 2 h.Cells were incubated with a hypotonic solution of potassium chloride 0.075 M at 37 °C for 20 min and then a cold fixative solution of methanol: acetic acid (3:1) was added.Chromosome preparations were spread on slides and dried by passages in ethanol series.The hybridization program was set to 75 °C for 5 min and 37 °C for 20 h (ThermoBrite, Abbott Molecular).After washing in 2 × SSC 0.3% NP40 solution at 73.5 °C and saturation, the slides were incubated with streptavidin red-conjugated antibody for 30 min.DNA was labelled with DAPI.All metaphases and interphases on the slide were observed using an ApoTome ZEISS microscope and Carl Zeiss™ AxioVision Rel.4.8.2Software (https:// www.micro-shop.zeiss.com/).

Real time-PCR
Total RNA from cells or human BM samples was isolated using TRIzol® Reagent (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer's instructions.All RNA quantifications were performed using the NanoDrop-1000 spectrophotometer (Nanodrop, Wilmington, DE).RNA was reverse transcribed using Applied Biosystem kits.qRT-PCRs for detection of miRNAs were performed with the indicated TaqMan® MicroRNA Assays (Applied Biosystems, Foster City, CA) on 10 ng total RNA according to the manufacturer's instructions.Gene expression analysis of CHORDC1 was performed using the TaqMan Gene Expression Assay (Applied BioSystems).Quantitative normalization was performed on the expression of U44snoRNA for miRNAs and on 18S for CHORDC1.The relative expression level between samples was calculated using the comparative delta CT (threshold cycle number) method (2 −ΔΔCT ) with a control sample as reference point.As defined previously, patient BMs were considered to express low Morgana when CHORDC1 mRNA level was less than half of the mean of the level in normal BMs (n = 10).We demonstrated that using this threshold we obtained 90% correlation with low Morgana level as defined by immunohistochemistry analysis 11 .

Mutagenesis
The 3'UTR of the human CHORDC1/Morgana gene was obtained by PCR using genomic DNA as template and the primer pair F (5′-AGT GGG AGA TGG AAG GAA GG-3′) and R (5′-ACA AAA GAA AGA TGA GAG GCACA-3′) with the addition at the 5′ of the SpeI (on F) and HindII (on R) recognition sequence.The PCR product was cloned into pGEM vector (Promega, Madison, WI).The sequence was mutated at the miR-15a/b binding site with a four-base substitution by using the QuikChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA).The mutagenized sequence was then sequenced, excised using SpeI and HindIII and subcloned into the pMIR-REPORT™ System (Applied Biosystem, Foster City, CA).

Luciferase assay
The 3′UTR of CHORDC1, previously generated by PCR amplification of the full length 3′UTR from a plasmid containing the human cDNA sequence of CHORDC1, was inserted into the pMIR REPORT™ luciferase vector (Ambion, Austin, TX).miR-15a and miR-15b binding site in the 3′UTR was mutagenized using the QuickChange Site-Directed Mutagenesis kit (Stratagene, Cedar Creek, TX) according to the manufacturer's instructions.6.5 × 10 4 cells were co-transfected with 50 ng of the pMIR REPORT™ (Ambion, Austin, TX), Firefly Luciferase constructs containing WT or specific miR-15a/b binding site-mutated sequences, 20 ng of pRL-TK Renilla Luciferase as a transfection efficiency control (Promega, Madison, WI) and 75 nM of the indicated pre-miR using Lipofectamine TM 2000 (Invitrogen Life Technologies, Carlsbad CA).Lysates were collected 48 h after transfection and Firefly and Renilla Luciferase activities were measured with a Dual-Luciferase Reporter System (Promega, Madison, WI).

Patient samples
Primary leukemia cells were obtained from the BM of patients with MDS/MPN with neutrophilia and Ph + CML in chronic phase after appropriate informed consent and Institutional Review Board approval (Institutional Ethics Committee Approval # 81/2011).All samples were collected at diagnosis before initiation of treatment.The research was performed in accordance with relevant guidelines and regulations.BM samples were processed for RNA extraction and quantitative Real Time-PCR.

Correlation analysis
For CLL study, two GEO series (belonging to the same GSE51529 GEO superSeries) have been analyzed: GSE51527-MicroRNA Profile and GSE51528-Gene Expression.This superSeries of microarray experiments contains the gene and miRNA paired expression profiles of purified B-cell chronic lymphocytic leukemia (B-CLL) cells obtained from 210 patients (Binet stage A); both single series contain a higher number of patients, but some of them cannot be matched.TCGA diffuse large B-cell lymphoma data have been selected to analyze Morgana-miRNA correlation.The common samples between the gene and miRNA datasets are 47 (RNA_Seq data are annotated to Human Mar.2006 (NCBI36/hg18).Computational analyses were performed by GenoBi-ToUs Genomics and Bioinformatics Service (University of Turin).

Statistical analysis
For statistical analyses, significance was tested using a two-tailed Student's t-test or, when required, one-or twoway ANOVA with Bonferroni's correction.A minimum value of p < 0.05 was considered to be statistically significant.Statistical analyses were performed using GraphPad Prism version 8.0.2 (https:// www.graph pad.com/).

Figure 2 .
Figure 2. miR-15a/b and miR-26a/b regulate Morgana mRNA and protein levels.(a) miRNA expression levels at different time points were measured by qRT-PCR upon transfection of the indicated pre-miRs into HEK293T cells.(b) CHORDC1 (Morgana coding gene) mRNA expression level was assessed by qRT-PCR after transfection of the indicated pre-miRs into HEK293T cells at different time points.(c, d) Western Blot (top) and quantification graphs (bottom) assessing Morgana protein expression levels at different time points after pre-miR transfection into HEK293T cells.HSP90 was used as loading control.(e) Relative microRNA expression levels after infection of HEK293T cells with an empty pLemiR, or pLemiR containing miR-15a or miR-26a as assessed by qRT-PCR.(f) CHORDC1 mRNA expression levels measured after infection of HEK293T cells with an empty pLemiR, or pLemiR containing miR-15a or miR-26a.(g) Western blot analysis of HSP90, Morgana, p-LIMK1 (with relative quantification) and LIMK1 on HEK293T cells infected with an empty pLemiR, or pLemiR containing miR-15a or miR-26a.Original blots are presented in Supplementary Information.Data are the results of at least three independent experiments, bars in graphs represent mean ± S.E.M. (*P < 0.05; **P < 0.01;***P < 0.001).

Figure 4 .
Figure 4. miR-26a and miR-15b negatively correlate with Morgana expression in human hematological malignancies in the TCGA database.Bioinformatic analysis indicated an anti-correlation among the indicated microRNAs and Morgana expression levels (CHORDC1 mRNA) in DLBCL (a-b) and CLL (c) subsets of TCGA datasets.

Figure 5 .
Figure 5. miRNA-15a levels anti-correlate with Morgana expression in patients affected by Ph + and Ph− neoplasms.(a, b, c, d) Correlation between Morgana expression (CHORDC1 mRNA assessed by qRT-PCR) and miRNA expression levels in a cohort of MDS/MPN with neutrophilia (Ph−) and Ph + CML (Ph+).(e, f, g, h) miRNA expression levels in a cohort of MDS/MPN with neutrophilia (Ph-) and Ph + CML patients (Ph+) with normal and low Morgana expression.(i) Correlation between BCR-ABL expression (International Scale) and miR-15a levels in Ph + CML patients.Bars in graph represent mean ± S.E.M (**P < 0.01).