Genomic variations in patients with myelodysplastic syndrome and karyotypes without numerical or structural changes

Myelodysplastic syndrome (MDS) is an onco-hematologic disease with distinct levels of peripheral blood cytopenias, dysplasias in cell differentiation and various forms of chromosomal and cytogenomic alterations. In this study, the Chromosomal Microarray Analysis (CMA) was performed in patients with primary MDS without numerical and/or structural chromosomal alterations in karyotypes. A total of 17 patients was evaluated by GTG banding and eight patients showed no numerical and/or structural alterations. Then, the CMA was carried out and identified gains and losses CNVs and long continuous stretches of homozygosity (LCSHs). They were mapped on chromosomes 1, 2, 3, 4, 5, 6, 7, 9, 10, 12, 14, 16, 17, 18, 19, 20, 21, X, and Y. Ninety-one genes that have already been implicated in molecular pathways important for cell viability were selected and in-silico expression analyses demonstrated 28 genes differentially expressed in mesenchymal stromal cells of patients. Alterations in these genes may be related to the inactivation of suppressor genes or the activation of oncogenes contributing to the evolution and malignization of MDS. CMA provided additional information in patients without visible changes in the karyotype and our findings could contribute with additional information to improve the prognostic and personalized stratification for patients.

Cytogenetic analysis. Karyotypes from metaphase with a 450-band resolution were obtained through cell culture in bone marrow tissue samples from patients with primary MDS. The chromosomes were banded by the GTG method (following standard procedures), modified, and optimized in accordance with previous protocols 27 . Short-term cultures (48 h) were established by incubating 1.0 mL of cell suspension in 5 mL RPMI1640 (Gibco Life Science, USA) supplemented with 1 mL of fetal bovine serum (Gibco Life Science, USA) and 100 μL L-glutamine (Gibco Life Science, USA) at 37 °C and in 5% CO 2 . To achieve metaphase, mitotic divisions were interrupted by adding 100 μL of colchicine solution (10 μg/mL) in water (Gibco Life Science, USA) to the culture system for 30 min prior to harvesting the cell suspension. Subsequently, the cells were incubated in a hypotonic solution of KCl (0.075 M) (Merck KGaA, Germany) for 30 min at 37 °C and in 5% CO 2 . Following hypotonization, the cells were fixed in methanol solution/glacial acetic acid (3:1), that was added by drip. The cells were mixed, centrifuged, and resuspended in fresh fixative solution. The fixation step was repeated three to four times. The fixed cells were dripped onto microscopy slides, subjected to GTG chromosome banding with the enzyme Trypsin (Invitrogen Life Technologies, USA), and stained in 4% Giemsa solution (Gibco Life Science, USA). Trypsin enzyme was diluted in phosphate-buffered saline (Invitrogen Life Technologies, USA). Twenty metaphases were analyzed per sample. The metaphase images were captured with the aid of an Axio-Imager 2 microscope (Carl Zeiss, Germany) connected to the blade-sweeping platform Metafer4 (MetaSystems Corporation, Germany). Chromosomal analyses were performed by the software IKAROS (Version 5.0, Meta-Systems Corporation, Germany). SNP array analysis. The platform GeneChip CytoScan HD array (Thermo Fisher Scientific, USA) was used for the identification of CNVs and LCSHs in the genome of 8/17 patients with primary MDS, who presented karyotypes without numerical or structural alterations. The genomic DNA of bone marrow tissue was extracted using the DNA GenomicPrep Mini Kit (GE Healthcare, USA) following the manufacturer's instructions. A total of 250 ng of DNA from each sample was digested with the restriction enzyme NspI, following the manufacturer's recommendations (Thermo Fisher Scientific, USA). Once digested, the samples were connected to the adapters, and then a universal primer was used to amplify the fragments using restriction fragment length polymorphism (RFLP) through polymerase chain reaction (PCR). PCR conditions were optimized to amplify fragments of sizes ranging from 150 to 2000 bp in length. The fragments were confirmed on 2% agarose gel in 1X TBE, subjected to a constant electric field of 10 V/cm for 1 h, and stained in an aqueous solution of ethidium bromide (5 mg/mL). The gel image was captured by the GelDoc XR + Imaging System (Bio-Rad Laboratories, USA).
The amplified products were purified with magnetic beads and quantified on the spectrophotometer NanoVue Plus (GE Healthcare, USA). Next, the amplified products were fragmented into 50-to 200-pb segments, purified, confirmed on a 4% agarose gel in TBE 1X, and separated in a constant electric field of 10 V/cm/h. The bands were revealed in an aqueous solution of ethidium bromide (5 mg/mL). The gel image was captured by the video documentation system GelDoc XR + Imaging System (Bio-Rad Laboratories, USA). DNA fragments were tagged In the ChAS, analysis parameters were fixed at 50 markers and a size of 100 kpb to detect gains and at 25 markers and a size of 100 kpb to detect losses. The filter was fixed to ≥ 3 Mb to detect the LCSHs. In the study, the parameters for CNV analysis followed both standard criteria based on manufacture's recommendations and on the LCSHs analysis previously reported 28,29 . All analyzed chips met the quality control metrics, being MAPD and SNP-QC defined at ≤ 0.25 and ≥ 15, respectively. CNVs, LCSHs, and genes were identified, selected, and separated based on their association with molecular factors of MDS. The databases Online Mendelian Inheritance in Man (OMIM) and RefSeq were accessed by software R to support data mining, analysis, and interpretation of the cytogenomic information obtained by CMA.
Expression analysis through public databases. Expression analysis was performed on public databases of patients with primary MDS. An NCBI GEO dataset file (https ://www.ncbi.nlm.nih.gov/geo/) was downloaded. The study (accession number GSE61853) was based on the analysis of gene expression in mesenchymal tissues of bone marrow stromal cells of patients with MDS and normal controls. The transcriptome was investigated with the aid of the platform Thermo Fisher Scientific Human Gene 1.0 ST Array (USA). Based on the global expression profile of bone marrow mesenchymal stem cells (BM MSCs), the top 250 genes were ranked based on P-value in the study GSE61853 where smallest P-value were the most significant. Afterwards, we compared the genes selected in the CNVs and LCSHs regions with the top 250 genes from the database. The logFC values were used to define which genes were considered to positively or negatively regulated in the individuals with MDS involved in this study.

Results
The average age of the 17 patients was 56  years. The small sample group is a result of the reduced incidence of primary MDS in the population. Of the 17 cases, in 6/17, the results of karyotype with banding GTG demonstrated numerical or structural changes, being + 8, + 15, + 17, + 21, − X, t(4;11)(q21;q23) add(1)(q?), and polyploidy clones; in 3/17, the karyotype was not informative, given that there was no clonal expansion, so these cases were removed from subsequent analyses. In 8/17 patients, the karyotype showed no numerical and/or structural chromosomal changes, and their results were previously reported and were not included in the current analysis 30 . With assistance from CMA, 41 genomic variants were identified in 7/8 patients with no visible changes in karyotype, including 11/41 variations in the number of copies per gain, 2/41 regions with genomic loss, and 28/41 long continuous extensions of homozygosis, mapped on chromosomes 1, 2, 3,4,5,6,7,9,10,12,14,16,17,18,19,20,21, X, and Y (Figs. 1 and 2). Ninety-one genes that have already been implicated in relevant molecular pathways for cell viability were selected, including 9/91 gains and 4/91 losses, and 78/91 in regions with LCSHs.
The cytogenetic risk of patients with karyotypes without numerical or structural alterations was rated as good, based on the International Prognostic Scoring System-Revised, while the risk of prognostic transformation of MDS ranged from very low to low. Only one patient died, while the others remained stable at the end of the study. Table 1  Among the selected genes, 28/91 are differentially expressed by up-regulation or down-regulation in the mesenchymal stromal cells of bone marrow tissue from patients with MDS and karyotypes without numerical or structural changes. The value of logFC was used to define which of the 91 selected genes were regulated positively or negatively in the MDS patients included in this study. Table 2 shows the biological functions and expression patterns of the 28 genes at diagnosis, in addition to the logFC value.
Some recurrent genomic variations have been identified in patients in the present sample group. The LCSHs in cytoband 19q13.32 with sizes of 13.64 Mb and 5.48 Mb were identified in patients ONCO003 and ONCO005, respectively. In this region, the genes ERCC2, SIX5, SYMPK, STRN4, BBC3, BAX, GYS1, CD37, FLT3LG, BCL2L12, and IL4I1, which were common among these patients, were selected. That said, the genes SIX5, SYMPK, and FLT3LG were differentially expressed by up-regulation or down-regulation. The LCSHs in cytoband 10q26.12 with 3.9 Mb and 10.76 Mb were identified in the patients ONCO003 and ONCO012, respectively, and the gene C10orf88 was identified to be differentially expressed by down-regulation. CNVs of genomic gain involving cytoband 17p11.2 with different sizes were also identified. In patient ONCO002 with

Discussion
In the present study, patient karyotypes without numerical and/or structural alterations were classified according to cytogenetic risk as good, whereas risk prediction for primary MDS ranged from very low to low. These results confirmed that cytogenetic risk is an important informative variable for inferring prognosis because, according to the clinical outcomes, only one patient died, while the others remained stable at the end of the study. The CMA analysis of patients with primary MDS and no karyotype changes visible under the microscope also provided useful information for better understanding of the etiological aspects of this hematological disease 3,8,10,11 . CNVs and LCSHs that are potentially associated with the etiology of MDS were identified in 87.5% (7/8) of the patients included in this study, of which 91 genes were selected that had already been implicated in molecular pathways important for cell metabolism and viability. According to the Cancer Genomics Consortium (CGC), CMA shows efficiency in detecting genomic variations in primary MDS in 10 to 80% of patient karyotypes without numerical or structural chromosomal changes 13 . Despite recent advances in diagnostic tools, MDS continues   www.nature.com/scientificreports/ to show variability in its clinical course and response to treatment 1,3,31,32 . In this scenario, the identification of genomic variants associated with primary MDS could prove widely useful, offering additional information for the biological understanding and prognostic classification of this disease 13,15,33,34 . The imbalances, characterized by gains and losses in the genome, represented 31.7% (13/41) of the genomic variations identified. Therefore, gene copy gains and losses may result in an increase or decrease/absence of any functional transcript. In this context, alterations in gene dosages play a determining role in the quantity of the expressed product and, consequently, down-regulate important molecular pathways of hematopoietic progenitor cells (HPCs), which are associated with the etiological processes of MDS 35,36 .
Expression analysis via a public database indicated that 30.8% (28/91) of the genes in this study located in regions with CNVs and LCSHs are up-regulated or down-regulated in the mesenchymal stromal cells of bone marrow tissue in patients with MDS, when compared with those of controls. Additionally, alterations in gene expression may be related to the inactivation of suppressor genes or the activation of oncogenes 26 . These altered molecular biological mechanisms may interfere with cell survival and correlate with the expansion of cells with ineffective hematopoietic functions, which may contribute to the possibility of the evolution and malignization of clinical MDS. Table 1. Clinical features, cytogenetic and cytogenomic results, and risk estimates (according to the International Prognostic Scoring System-Revised) of a cohort of patients with myelodysplastic syndrome in addition to clinical outcomes. F female, M male, NAF no alterations found, LCSH long continuous stretches of homozygosity. *ISCN2016: an international system for human cytogenomic chromosome nomenclature. **IPSS-R: risk estimates, according to the International Prognostic Scoring System-revised ***No cytogenomic rearrangements and no genes were identified using the CMA in the analyzed region. # Estimated score according to hemoglobin levels, absolute neutrophil count, platelet count, percentage of blasts in the bone marrow, and cytogenetic risk. www.nature.com/scientificreports/ Of the altered genes, 32.1% (9/28) were down-regulated and 67.9% (19/28) were up-regulated. The main functional pathways in which these genes are involved have been identified; in this context, the investigation only described the biological processes that may be affected by altered gene expression. This study reports that the molecular pathways associated with gene expression control, immune response, and tumor proliferation and suppression were the most commonly deregulated. The relationship between RNA polymerase and the etiology of MDS is not well-defined 26 . One can infer that RNA polymerase dysregulation would alter the normal transcription of critical genes, such as tumor suppressor genes (GSTs), because RNA polymerase plays a role in modulating DNA transcription.
Identified CNVs harbor 17.9% (5/28) of genes with positive or negative regulation in the mesenchymal stromal cells of bone marrow tissue in patients with MDS. In the CNV of genomic loss in 20q, the RALY and ADA genes were found to be up-regulated and SLC2A10 and RAB5IF were down-regulated. The ADA and RALY genes are associated with immune response pathways and embryonic development, respectively, while the genes SLC2A10 and RAB5IF correlate with glucose homeostasis pathways and protein interaction factor, respectively; these genes are expressed in the mesenchymal stromal cells of normal bone marrow tissue.
According to reports previously published 37,38 , loss of CNVs lead to haploinsufficiency and can inactivate GSTs. In addition, the RALY and ADA genes were up-regulated in lost CNVs, suggesting compensation for protein products. Therefore, in microdeletion events in the genome, varied molecular mechanisms can occur and be associated with the pathophysiology of MDS. Thus, CNVs of genomic loss represent important markers with biological significance, providing additional information for understanding the mechanisms and molecular pathways related to the etiology and transformation potential of MDS 36 .
The MAP2K3 oncogene, located in the CNV with genomic gain in 17p with 0.29 Mb, shows altered expression with positive regulation. The protein encoded by this gene is a dual-specificity kinase that belongs to the family of map-kinases and is activated by mitogenic and environmental stress 38 . Changes in the pathways involving these kinases can deregulate molecular processes related to differentiation, cell cycle survival, and control, leading to the process of apoptosis in patients with MDS. Imbalances caused by positive gene expression, when compared with negative gene expression, may be less unfavorable for cellular biological pathways 37,38 . The present study Table 2. Biological functions and expression patterns at diagnosis of (28/91) selected genes in MDS patients with karyotypes without numerical or structural changes. *The biological functions of the genes were identified via RefSeq (NCBI Reference Sequence Database: http://www.genom e.jp/dbget -bin/www_bget?ds:H0148 1; Kegg Disease: Myelodysplastic syndrome-Genome Net). **For the analysis of expression through public databases of MDS patients, one GEO Dataset from NCBI (https ://www.ncbi.nlm.nih.gov/geo/) was downloaded. The transcriptome was based on the study with accession number GSE61853 in which the gene expression patterns of mesenchymal bone marrow stromal cells from patients with MDS and normal controls were analyzed. The value of logFC was used to define which genes were up-regulated or down-regulated in MDS patients with karyotypes without numerical or structural changes. www.nature.com/scientificreports/ confirms this statement, since about 2/3 of the genes are up-regulated and 7/8 patients with no visible changes in the karyotype exhibited stable clinical outcomes at the end of the study. All three gain CNVs involving the 17p11.2 cytoband with different sizes, being 0.29 Mb, 0.13 Mb, and 0.15 Mb, are significant. Events involving chromosome 17p were considered important due to the proximity of the cytoband 17p13.1, which houses the GST TP53. The TP53 protein participates in the regulation and expression of several target genes, inducing cell cycle control, apoptosis signaling, cell senescence, and DNA repair 39,40 . Mutations in one or both alleles of TP53 can deregulate several molecular pathways involved in about 50% of human cancers and 20% of onco-hematologic malignancies, including MDS. Therefore, according to the risk prognosis and clinical results, these CNVs with a 17p gain may have a negative effect on hematological evolution and patient survival, as evidenced by the patient who died during the course of this study. The other CNVs with genomic gain in 2q, 17p, 17q, 21q and Yq, the CNV of genomic loss in 4q, and the LCSHs in 4q, 18q and Xq presented in Table 1 were selected as relevant due to the high instability inherent in the genomes of MDS patients, which is the basis for the accumulation of mutations in this disease 24 . Expression per positive regulation of the ADA gene, inserted in a region with genomic loss in 20q, and of the FLT3LG gene in 19q and KLHL5 in 4p, located in regions within LCSHs, may be associated with changes in interferon signaling pathways (immune response) 41 . These deregulated pathways in the medullary microenvironment in patients with MDS demonstrate that these factors participate in inducing the inflammatory state and immunological disorders, which alters hematopoiesis, thereby causing intramedullary apoptosis along with abnormal HPC differentiation and maturation.
LCSHs are abnormalities that represent 68.3% (28/41) of the genomic variations identified in the present study. Of the selected genes, 85.7% (78/91) were inserted in regions with LCSHs and associated with biological regulatory functions. Additionally, 82.1% (23/28) of up-regulated or down-regulated genes in mesenchymal stromal cells in the medullary microenvironment of patients with MDS were located in regions with loss of heterozygosity. The pathological potential of LCSHs emerges as a mechanism of clonal alteration in a proportion of somatic cells with uniparental disomy (UPD) 19,20,42,43 . In this context, recurring areas of LCSHs are strongly associated with the loss of the wild allele or an entire region, which also leads to the hypothesis of haploinsufficiency. Thus, these events indicate changes in biological processes that are important to the cells and signal a predisposition to hematological malignancies 19,20 . The terminal segmental LCSHS identified in 6q, 14q, 18q, 19q, and 21q presented in Fig. 2 indicates that break-induced replication (BIR) can be a dominant mechanism by which a cell duplicates a somatically acquired event, such as a mutation, microdeletion, or epigenetically suppressed region, and consequently becomes homozygous for this segment. BIR seems to be a common repair mechanism in replication, but only the reduplication of a region that contains a genetic or epigenetic alteration gives the cell a growth advantage, allowing for its dominance and clonal selection. Regions with LCSHs involving 3p, 4p, 10p, 12q, and 16p were identified only in the patient who died. The UCHL1, TMEM33, and OCIAD2 genes inserted in 4p, the HACD1 gene in 10p, and the POC1B gene in 12q, which encode neural transmission functions, transmembrane protein, tumor suppressor, signaling molecule, and centrosome, respectively, were down-regulated. Five out of nine selected and down-regulated genes were located in the LCSHs identified in the patient who died. Thus, unlike up-regulation, down-regulation events are potentially more damaging to important molecular pathways and fundamental cellular functions, as observed in this patient during the hematological evolution of the disease 44 .
The MX1 gene, located in an LCSH region at 21q, was the most up-regulated, being 2.74-times more expressed in bone marrow mesenchymal stromal cells in patients with MDS. This gene encodes a protein that metabolizes the guanosine triphosphate (GTP) that participates in the antiviral cellular response. The protein encoded by this MX1 gene is induced by the type I and II interferon pathways. In this study, the patients with genomic variations involving 21q were classified as having low and very low predicted risk, respectively, with stable clinical outcomes. These findings contradict previously published 9 reports that changes involving chromosome 21 have been observed in more advanced cases of MDS, with a more aggressive evolution and rapid leukemic transformation 1,11 . On the other hand, the UCHL1 gene inserted in an LCSH region with 4p was the most downregulated, being 1.12-times less expressed in the mesenchymal stromal cells of the medullary tissue of the MDS patient who died. The gene UCHL1 provides information for the production of the enzyme Ubiquitin Carboxyl-Terminal Esterase L1, which is involved in the cellular machinery that breaks down unnecessary substances 45 . In cells, damaged or excess proteins are marked with ubiquitin molecules so that the ubiquitin-proteasome system can act as a cellular quality control system.
In patients with MDS, persistent and refractory cytopenias can be induced by cytokines and associated with T-cell mediated myelosuppression 46 . Alongside this, changes in the medullary microenvironment inhibit the growth of HPCs through the high secretion of interferon-gamma (IFN-γ), TNF, and Interleukin-6 (IL-6) 14,47 . An increase in the number of stem cells could offset the phenotypic consequences of the simultaneous repression of differentiation genes, creating a pre-leukemic hematopoiesis with little or no morphological abnormalities 48 . As such, the association between dysplastic cell characteristics and the increase in total genomic changes is an observation that suggests a significant parallel trend. Also, the more morphological dysplasia a marrow tissue sample presents, the more extensive the underlying genomic changes can be 48 . Both events, CNVs and LCSHs, contributed significantly to the correlation between the identified genomic variations and the clinical and laboratory phenotypes of patients with MDS in this study, highlighting the usefulness of matrix platforms containing markers for SNPs.
In only 12.5% (1/8) of the patients, the CMA did not reveal genomic variations. It is important to note that this result without changes does not imply that there could not be some kind of deleterious mutation in the genes associated with the molecular pathways associated with the MDS phenotypes. Thus, it was not possible to suggest a genetic cause for this disease, since the CMA resolution was not comprehensive enough to identify changes in the genome. This study highlights that balanced translocations are not detectable by CMA, which is a disadvantage of this cytogenomic analysis technique 49 www.nature.com/scientificreports/ genomic technologies, such as Next Generation Exome Sequencing (NGS), may be a differential approach to identify point mutations that may be associated with MDS, in which the etiology is multifactorial and extremely heterogeneous 51 . The limitation of this study was the small size of the sample group. This limitation is typical of studies involving patients with primary MDS, which have reduced casuistry due to the low incidence of the disease in the population. The present findings from CNVs and LCSHs can contribute to subsequent studies of the adequate characterization of the MDS phenotype and bring additional information to improve the quality of prognostic and personalized stratification for the patient, in addition to the possibility of identifying potentially more effective therapeutic treatments in the near future 13,16 . The interpretation and association of CNVs that result in primary MDS is still challenging, mainly due to the unclear effects of variations in the genome that can be influenced by incomplete penetrance, variable expressiveness, and epistasis 40,52,53 . In this context, further investigations are needed in larger cohorts, including patients with greater heterogeneity of clinical outcomes and prognostic risks.

Conclusion
This study identified and reported CNVs and LCSHs that may be associated with primary MDS in the group of participants. Ninety-one genes involved in important molecular pathways of metabolism and cell viability were selected. From the analysis of the cytogenomic findings, when compared to information from public genomic databases, CNVs and LCSHs were related to bone marrow failure, which was characterized by the expansion of cells with ineffective hematopoietic functions. CMA is a robust and efficient cytogenomic technique for expanding the resolution of the genomic analysis of the medullary cells of primary MDS patients, allowing the identification of unusual and clinically significant genomic variations in patients without visible changes in the karyotype. Finally, 30.8% (28/91) of the selected genes showed altered expression and 2/3 of these genes showed positive regulation when compared to public data on gene expression in mesenchymal stromal cells of bone marrow tissue in primary MDS. Changes in gene expression may be related to the etiology and progression of primary MDS, with a resulting effect on the relative risk and clinical outcome of this onco-hematological disease. www.nature.com/scientificreports/