Acquired uniparental disomy of chromosome 9p is a frequent stem cell defect in polycythemia vera

Experimental Hematology 30 (2002) 229–236

Acquired uniparental disomy of chromosome 9p is a frequent stem cell defect in polycythemia vera Robert Kralovicsa,b, Yongli Guana, and Josef T. Prchala a Department of Medicine, Baylor College of Medicine, Houston, Tex., USA; German Cancer Research Center (DKFZ), Molecular Oncology and Hematology, Heidelberg, Germany

b

(Received 24 October 2001; revised 19 November 2001; accepted 27 November 2001)

Objectives. Clonal stem cell proliferation and increased erythrocyte mass are hallmarks of the myeloproliferative disorder polycythemia vera (PV). The molecular basis of PV is unknown. Methods. We carried out a genome-wide screening for loss of heterozygosity (LOH) and analyzed candidate genes within the LOH loci. Results. Three genomic regions were identified on chromosomes 9p, 10q, and 11q. The presence of these LOHs in both myeloid and lymphoid cells indicated their stem cell origin. The 9pLOH prevalence is 33% and is the most frequent chromosomal lesion described in PV so far. We report that the 9pLOH is due to mitotic recombination and therefore remains undetectable by cytogenetic analysis. Nineteen candidate genes were selected within the 9pLOH region for sequencing and expression analysis. No mutations were found in these genes; however, unexpectedly, increased expression of the transcription factor NFI-B was detected in granulocytes and CD34 cells in PV with 9pLOH. Since a member of the NFI gene family (NFI-X) was reported to result in TGF- resistance when overexpressed in vitro (TGF- is a known inhibitor of hematopoiesis), we transfected the NFI-B gene to the mouse 32D cell line. We found that overexpression of the NFI-B gene confers TGF- resistance in vitro. Conclusions. We characterized a new region on chromosome 9p frequently involved in LOH in PV. Analysis of genes within this 9pLOH region revealed increased expression of the NFI-B gene. Our in vitro studies suggest that TGF- resistance may be the physiologic mechanism of clonal stem cell expansion in PV. © 2002 International Society for Experimental Hematology. Published by Elsevier Science Inc.

Myeloproliferative disorders are acquired clonal diseases involving the pluripotent stem cell. The somatic mutations in myeloproliferative disorders lead to clonal expansion of the myeloid lineages with typically one lineage dominating in each type of myeloproliferative disorder. In polycythemia vera (PV), the expansion of the erythroid lineage results in high erythrocyte mass, although thrombocythemia, high neutrophil counts, and splenomegaly are also present in a significant proportion of PV cases. The molecular lesion responsible for the PV disease phenotype is unknown. Clonality studies performed using G6PD isoenzyme expression demonstrated that PV is an acquired defect of the pluripotent stem cell [1]. The stem cell acquiring the PV mutation

Offprint requests to: Josef T. Prchal, M.D., Department of Medicine, Baylor College of Medicine, One Baylor Plaza 802E, Houston, TX 77030; E-mail: [email protected]

expands, and subsequently the terminally differentiated myeloid cells are all (or virtually all) derived from the PV stem cell clone. The contribution of the normal stem cell population to the production of circulating myeloid cells may in some patients become undetectable using sensitive clonality assays based on X-chromosome inactivation. B lymphocytes are often derived from the PV stem cell clone but T lymphocytes were always polyclonal, i.e., derived from the non-PV clone, in all subjects studied [2]. Cytogenetic lesions are present in about 14% of PV cases at diagnosis, but none of these lesions is specific for PV and are shared among myeloproliferative disorders [3]. The most common cytogenetic abnormalities (present in about 8% of PV cases) are deletions involving chromosome 20q. Trisomies 1,8,9, and deletions on chromosomes 13q and 5q occur less frequently (0.5–3% of the PV cases) [4]. Abnormal hematopoietic progenitor cell responses to cytokines have been consistently demonstrated in PV. The erythroid progenitors in PV exhibit erythropoietin-indepen-

0301-472X/02 $–see front matter. Copyright © 2002 International Society for Experimental Hematology. Published by Elsevier Science Inc. PII S0301-472X(01)0 0 7 8 9 - 5

230

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

dent colony formation in the absence of exogenous erythropoietin (EPO) in serum-containing cultures [5,6]. In serumfree culture systems, hypersensitivity to other cytokines has also been reported. The abnormal responsiveness of PV progenitors to cytokines promoted a number of studies focusing on cytokine receptor signaling and cytokine-dependent survival of PV progenitor cells. Quantitative abnormalities were described in PV myeloid cells, such as elevated phosphatase activity in erythroid progenitors [7], decreased SHP-1 phosphatase expression in PV erythroid cells [8], decreased thrombopoietin receptor expression in PV platelets [9], and increased expression of bcl-xL in PV erythroid cells [10]. Changes in transcriptional regulation of the PRV-1 [11] and the CDKN2A [12] genes were also observed in PV granulocytes and cultured erythroid cells, respectively. Our previous clinical and genetic studies of rare familial clustering of PV suggested that loss of heterozygosity (LOH) might play a role in the PV pathogenesis [13]. To identify mutations of genes (or genomic regions) that cause the PV phenotype, we screened the whole genome for LOH without the bias of previously characterized cytogenetic lesions observed in myeloproliferative disorders. In this study, we identified three new genomic loci that undergo changes in the PV stem cells. The genes mapping to these regions are potential candidates involved in the hyperproliferative nature of the PV hematopoietic cells.

Materials and methods Clinical material All PV subjects in this study had a classical PV phenotype including acquired history of elevation of hemoglobin (in most with documented elevated red cell mass), increased platelet counts, splenomegaly, low serum EPO levels, and normal arterial blood oxygen saturation. Their peripheral blood progenitors formed burst-forming unit erythroid (BFU-E) colonies in the absence of EPO (greater than 10% of maximum number grown at 3 U of EPO). Further, all females heterozygous for one of the three X-chromosome exonic polymorphisms (IDS, G6PD, and p55) expressed only a single allelic transcript of these clonality markers [14,15] in their myeloid cells. In addition, six subjects with essential thrombocythemia were included in this study that had clonal myeloid cells in the X-chromosome inactivation pattern assay and were negative for EPO-independent erythroid colony growth. The clinical material was obtained using approved IRB protocols initially at the University of Alabama at Birmingham, and more recently at the Baylor College of Medicine. Cell separations, DNA, and RNA isolation Peripheral blood mononuclear cells and crude granulocytes were prepared by standard protocols using Histopaque (Sigma, St. Louis, MO, USA) gradient centrifugation [15]. The crude granulocytes were resuspended in phosphate-buffered saline (PBS) without calcium and magnesium (Life Technologies, Rockville, MD, USA) supplemented with 0.5% bovine serum albumin (BSA) and 2 mM EDTA and further purified by magnetic selection for CD15 cells. Labeling and magnetic separation of CD15 cells

was carried out according to the manufacturer’s recommendations (MACS CD15 labeling kit, and MS column separation, Miltenyi Biotech, Auburn, CA, USA). Mononuclear cells were washed in PBS and the contaminating erythrocytes were removed by lysis in erythrocyte lysis buffer as described in crude granulocyte isolation. The erythrocyte-depleted mononuclear cells were used for magnetic separation of CD34, glycophorin A, and T lymphocytes using magnetic labeling and separation according to the manufacturer’s protocol (Miltenyi Biotech, Auburn, CA, USA). The purity (measured by flow cytometry) of CD15 and T cells was 98–100%, and of CD34 and glycophorin A 80–90%, respectively. In some subjects skin fibroblasts were obtained by skin biopsy and grown using standard procedures. Isolation of DNA from the isolated cell populations was carried out using a standard proteinase K/phenol extraction protocol. The TRI reagent (MRC Inc., Cincinnati, OH, USA) was used for granulocyte RNA isolation and the RNeasy mini kit (Qiagen, Santa Clarita, CA, USA) for RNA isolations from CD34, T, and glycophorin A cells following the manufacturer’s protocol. Genome-wide screening for LOH Fluorescent dye (6-FAM, NED, and HEX)–labeled primers used for microsatellite polymerase chain reaction (PCR) were part of the MD-10 linkage mapping set (Applied Biosystems, Foster City, CA, USA). Only the autosomal panels were used for LOH screening that consisted of 382 markers providing 10 cM marker spacing throughout the entire genome. PCR was performed in a 96-well format in 10.6 L reaction volume using 3 pM of primer mix, and 20 ng genomic DNA per reaction. The PCR was performed as recommended be the manufacturer. Pools of PCR products were analyzed using the ABI Prizm 377 DNA Sequencer (Applied Biosystems, Foster City, CA, USA). The data obtained from the gels were analyzed using the GeneScan (version 3.1.2) and Genotyper (version 2.5) software packages (Applied Biosystems, Foster City, CA, USA). In each subject, genotypes obtained from clonal granulocyte DNA were compared to genotypes from either fibroblast or T-cell DNA serving as a polyclonal control (a total of 764 genotypes were scored per subject). Only heterozygous markers in fibroblast or T-cell DNA were informative for the LOH analysis. Allelic loss of 95% or more detected in granulocyte DNA was considered as LOH. Quantitative genomic hybridization High-molecular-weight granulocyte genomic DNA samples were mechanically shredded using the QIAshredder spin column (Qiagen, Santa Clarita, CA, USA). Three g of DNA was denatured in 0.4 M NaOH at 100C for 10 minutes. The denatured DNA was alkaline vacuum transferred using the Bio-Dot SF slot blotting apparatus (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s protocol onto Zeta-Probe GT nylon membrane (Bio-Rad Laboratories). SNAPC3 and GAPDH fulllength cDNA probes were labeled by [32P]dCTP (Amersham Pharmacia Biotech, Piscataway, NJ, USA) using the DECAprime II labeling kit (Ambion, Austin, TX, USA). Hybridization was performed at high stringency overnight in ULTRAhyb hybridization solution (Ambion) following the manufacturer’s protocol. The GAPDH hybridization was used to verify the amount of blotted DNA. Quantitative analysis of hybridization signal was performed by densitometry of autoradiograms using the Fluor-S MultiImager with the QuantityOne software (Bio-Rad Laboratories).

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

Transcriptional clonality assay using the X-chromosome marker IDS We used a transcriptional clonality assay based on X-chromosome inactivation in females informative (heterozygous) for exonic polymorphisms in the IDS genes. The genotypes of the subjects were determined using allele-specific PCR as previously described [15]. In heterozygous females, the transcriptional analysis of the alleles was performed with end-labeled forward primer as described elsewhere. Sequence analysis The cDNA sequences of candidate genes were obtained from GenBank. Primers derived from the cDNA sequences were designed to amplify the protein coding sequences of the candidate genes. For cDNA sequencing, granulocyte RNA from a 9pLOH-positive patient was used. The CDKN2A gene (3 exons), CDKN2B gene (2 exons), MPDZ gene (46 exons), and the NFI-B gene (9 exons) were sequenced using genomic DNA. For genomic DNA sequencing, granulocyte DNA of a 9pLOH-positive PV patient was used as template. Gene structures of the MPDZ and NFI-B genes were unknown; therefore, we performed exon mapping by cDNA sequence alignment against the HTGS database available at NCBI (http://www.ncbi.nlm.nih.gov). Primer sequences used for sequencing and gene expression analysis are available upon request from the authors. PCR products were gel purified and sequencing reactions were performed using the BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) and the resulting sequencing products were analyzed with the ABI Prizm 377 DNA Sequencer (Applied Biosystems, Foster City, CA, USA). Gene expression studies Hematopoietic cell expression of genes within the 9pLOH region was screened using the Access RT-PCR System (Promega, Madison, WI, USA) with 100 ng/reaction normal CD34 cell total RNA and 10 pM/reaction primers. All reactions were performed in 25 L reaction volumes. Semiquantitative reverse transcriptase polymerase chain reaction (RT-PCR) analysis of the selected candidate genes was performed as follows. 300 ng of total RNA (quantified by spectrophotometry) was reverse transcribed using the SuperScript II reverse transcriptase (Life Technologies) using 0.5 g random primers (Life Technologies, Rockville, MD, USA). For each gene, the cycle number for linear amplification was determined using 1 L of RT reaction/PCR reaction. RT-PCR products were analyzed on standard agarose gels. Quantitative multiplex RT-PCR analysis of NFI-B and HPRT genes was performed using the following primers: NFIB-F 5-CGAACATGGCACGAAAGAGA-3, NFIB-R 5TGGACTGGATGGGTCATAAG-3, HPRT-F 5-GCTGAGGA TTTGGAAAGGGT, and HPRT-R 5-GCGATGTCAATAGGAC TCCA-3. Optimalization of primer ratio, template amount, and cycle number were performed as described elsewhere [16]. The ratios of the HPRT and NFI-B amplification products were measured by densitometry using the Flour-S MultiImager with the QuantityOne software (Bio-Rad Laboratories, Hercules, CA, USA). Transfection of 32D cells with NFI-B and TGF- inhibition assay The NFI-B cDNA (splice variant B2) was cloned into the mammalian expression vector pcDNA3.1/V5-His TOPO (Invitrogen, Carlsbad, CA, USA). The mouse interleukin-3 (IL-3)–dependent cell line 32D (ATCC) was grown in RPMI 1640 with glutamine and HEPES in the presence 10% fetal bovine serum (Life Technologies, Rockville, MD, USA) and 1 ng/mL mouse IL-3 (R&D Systems, Minneapolis, MN, USA). The functional studies were per-

231

formed using the MACSelect 4.1 transient expression system (Miltenyi Biotech, Auburn, CA, USA). Electroporation and isolation of transfectants by magnetic sorting were carried out as described in the manufacturer’s protocol. Vector-transfected and NFI-B–transfected cells were assayed for TGF-1 inhibition using the cell proliferation kit II XTT (Roche Molecular Biochemicals, Indianapolis, IN, USA) in a microplate format. A TGF-1 (R&D Systems, Minneapolis, MN, USA) concentration range between 0.05–10 ng/mL was used in the presence of 1 ng/mL mouse IL-3. Spectrophotometric measurement of proliferation was performed using the Model 550 microplate reader (Bio-Rad Laboratories, Hercules, CA, USA).

Results Genome-wide screening for LOH To identify genomic regions with LOH, we compared DNA isolated from clonal granulocytes and polyclonal T cells or skin fibroblasts. Six subjects with PV had undergone the entire genome-wide LOH screening. We analyzed 382 autosomal microsatellite markers per DNA sample to cover the genome at 10 cM density. We identified LOH on the short arm of chromosome 9, and on the long arms of chromosomes 10 and 11 (Fig. 1). The sizes of the LOH regions were 40 cM on 9p, and 80 cM on 10q and 11q. To estimate the frequency of LOH in these regions, we screened a total of 20 PV and 6 clonal essential thrombocythemia (ET) subjects. The 9pLOH was present in 6 of 20 (33%) PV subjects and was not detected in any clonal ET subjects. Each of the 10q and 11q LOH was observed in only a single PV subject (5%), and not in any clonal ET. The 11q LOH was found in a subject that also had a LOH on chromosome 9p. Stem cell origin of LOH Pluripotent stem cell defects are defined as acquired mutations occurring in a cell that gives rise to both myeloid and lymphoid progeny. To demonstrate that the LOH on chro-

Figure 1. Genome-wide screening for LOH in six PV subjects. Three chromosomal regions involved in LOH were detected and are shown in black (uninformative markers are not shown).

232

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

mosomes 9p, 10q, and 11q are stem cell lesions, we analyzed the presence of LOH in both myeloid and lymphoid cells of PV subjects. As shown in Figure 2A, the LOH is detectable in the myeloid lineages as well as in 30% (9p, 11q LOH) and 60% (10q LOH) of T lymphocytes. This finding indicates that the LOH detected on 9p, 10q, and 11q are clonal stem cell lesions. In addition to clonality analysis using LOH, we also analyzed the clonality of the same cell isolates using X-chromosome inactivation pattern assay (Fig. 2B). We found no significant differences between clonality data obtained by LOH analysis and by X-chromosome inactivation pattern analysis. These data demonstrate that virtually all clonally derived cells were positive for the presence of LOH. Interestingly, in the subject positive for both 9p and 11q LOH we observed about 15% of CD34 early hematopoietic progenitor cells that were not clonal (i.e., not derived from the PV stem cell clone) in either of the LOH assays or in the IDS clonality assay. This is may be due to lymphoid contamination of the CD34 cell isolate or

Figure 2. Stem cell origin of LOH. (A) The presence of the 9p, 10q, and 11q LOH was studied in isolated populations of myeloid and lymphoid cells (CD34: early hematoipoietic progenitors; GPA: glycophorin A erythroid cells; CD15: granulocytes; T cells: T lymphocytes; FIB: fibroblasts). The LOH is detectable in both myeloid cells and in a proportion of T lymphocytes, indicating that the LOH originated in the pluripotent stem cell. The allele involved in the LOH is indicated by an arrow. (B) A 9pLOH female PV subject was also examined for clonality using the transcriptional X-chromosome inactivation pattern assay. The granulocytes (CD15), platelets (PLT), and a vast majority of CD34 cells were clonal, confirming the findings based on the LOH clonality analysis. In both assays, the T lymphocytes (T) were polyclonal.

due to a small population of normal polyclonal CD34 cells without LOH in this subject. CD15 granulocytes isolated to homogeneity were 100% clonal and all were positive for the LOH. We could not detect CD34 cells without LOH in the subject with the 10q LOH. The LOH on chromosome 9p is due to mitotic recombination Since the 9pLOH occurred with highest frequency, we focused our studies primarily on this genomic region. Cytogenetic studies of the subjects with detectable 9pLOH did not indicate any terminal losses or deletions of the entire short arm of chromosome 9 (data not shown), suggesting that the LOH may be due to mitotic recombination. To confirm this hypothesis, we analyzed the copy number of genes mapping to the 9p minimal LOH region by quantitative genomic hybridization. DNA samples isolated from granulocytes of five 9pLOH subjects were hybridized with a probe derived from the SNAPC3 gene that is localized within the LOH region. Two copies of the SNAPC3 gene were detected in all five analyzed 9pLOH subjects as well as in three normal individuals (Fig. 3). These results indicate that the presence of the 9pLOH in the studied PV subjects resulted from mitotic recombination events rather than interstitial or terminal deletions, in which case only a single copy of the gene would be present. The LOH mechanism by mitotic recombination is further supported by the presence of LOH extending to the 9p telomere in all the analyzed cases (see below).

Figure 3. Detection of mitotic recombination on chromosome 9p. Genomic hybridization of granulocyte DNA of five 9pLOH PV subjects and three normal controls with a probe derived from the 9p LOH region (SNAPC3 gene). The SNAPC3 hybridization signal was normalized against the hybridization signal obtained using the GAPDH control probe. The hybridization intensity of PV samples did not differ from the control DNA samples, suggesting that no loss of chromosomal material is present within the 9p LOH region. This finding indicates that LOH on chromosome 9p is due to mitotic recombination and is not a result of deletions.

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

Mapping of genes in the minimal LOH region on chromosome 9p We identified six PV subjects with LOH on chromosome 9p. In five out of six cases, the breakpoint of the mitotic recombination occurred between markers D9S1817 and D9S273. In one subject, however, the breakpoint occurred between markers D9S161 and D9S1817. This delineated a minimal LOH region located between the 9p telomere and marker D9S1817. Based on the physical map of the short arm of chromosome 9 available at NCBI, there are 85 genes and ESTs mapped within the 9p telomere and marker D9S1817 (data as of 4/26/2001). To reduce the number of candidate genes, we performed qualitative RT-PCR analysis to eliminate genes that are not expressed in normal CD34 hematopoietic progenitor cells. Genes with housekeeping functions as well as genes without obvious regulatory functions were also eliminated. Using the above criteria, 19 likely candidate genes were selected for sequencing and gene expression studies (Fig. 4). Mutational and expression analysis of genes within the minimal 9pLOH region In order to identify the target gene within the minimal 9pLOH region, we performed sequence analysis in one of the 9pLOH PV subjects using granulocyte RNA or DNA as template. Only the protein coding regions of the 19 candidate genes were sequenced. No mutations were detected except for known polymorphisms. Since LOH by mitotic recombination may alter methylation patterns or amplify mutations affecting RNA stability, we also performed a semiquantitative RT-PCR analysis of gene expression of the candidate genes. Granulocyte RNA of the same PV subject, as used in the gene sequencing, was used for the RT-PCR screening parallel with granulocyte RNA of a normal individual. A summary of semiquantitative expression analysis of the 19 candidate genes (a total of 20 transcripts) is pro-

233

vided in Table 1. None of the genes was found to have deficient RNA levels in PV. However, we unexpectedly detected significantly elevated expression of the transcription factor gene NFI-B in the PV granulocyte RNA. To assess the expression level more accurately, we developed a semiquantitative RT-PCR detection of the NFI-B gene using the HPRT gene expression as an internal control. We analyzed the NFI-B expression in two PV subjects without and two PV subjects with the 9pLOH and found that the increased NFI-B expression was present only in the 9pLOH subjects. PV subjects without detectable 9pLOH had comparable levels of the NFI-B transcript to that of the control cells (data not shown). Expression of NFI-B, CDKN2A, and MPDZ genes in CD34 cells The choice to use granulocyte RNA for sequencing and gene expression analysis was dictated by ready availability of granulocytes and their relative abundance in peripheral blood among all clonal PV cells. The use of CD34 early progenitor cells was limited due to their low abundance in peripheral blood samples (IRB permission for bone marrow samples, where these cells are more abundant, was not granted). In addition, the transcripts of the tumor suppressor gene CDKN2A and the MPDZ gene, listed among the candidate genes, could not be detected in granulocyte RNA samples. Based on these considerations, we only examined the expression of the NFI-B, CDKN2A, and MPDZ genes in CD34 cells using semiquantitative RT-PCR analysis. The two transcripts of the CDKN2A gene (p16INK4A and ARF) transcribed from different promoters were analyzed in separate reactions. As shown in Table 1, the increased expression of NFI-B is detectable also in the CD34 cells (in addition to granulocytes) in a 9pLOH PV subject. The CDKN2A and MPDZ gene transcripts had normal levels in CD34 cells (data not shown).

Figure 4. Mapping of the minimal LOH region. Six PV subjects were characterized positive for the presence of LOH on chromosome 9p. Alignment of the LOH regions in these subjects mapped the minimal LOH region between the 9p telomere and marker D9S1817. Nineteen candidate genes were selected based on expression in hematopoietic cells and elimination of housekeeping genes.

234

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

Table 1. Summary of expression analysis of candidate genes within the minimal 9p LOH region CD34 cells

Granulocytes

JAK2 MDS030 FLJ20639 PRO0365 MPDZ NFI-B SNAPC3 PSIP1 RAGA FLJ20060 FLJ20686 MLLT3 MTAP LOC51198 CDKN2A-p16 CDKN2A-ARF CDKN2B LOC65591 BAG1 DNAJA1

Control

9pLOHPV

                   

                   

Control

9pLOHPV

 

  Figure 5. Overexpression of the NFI-B gene results in TGF-1 resistance. The 32D mouse cell line was transfected with the NFI-B cDNA construct. Transfection with the cloning vector was used as a control. The inhibition of proliferation by TGF-1 was measured by the XTT proliferation assay, and was expressed as the percentage of maximum proliferation. The NFIB–transfected cells had increased resistance to TGF-1 inhibition compared to vector-transfected cells.

 

 

Overexpression of NFI-B results in decreased sensitivity to TGF-1 inhibition The NFI-X gene (a member of the NFI gene family) was reported to confer TGF-1 resistance when overexpressed in vitro [17]. To examine if the NFI-B gene might exhibit a similar function, we transfected the NFI-B to the mouse IL-3–dependent cell line 32D. The proliferation of vectortransfected and NFI-B–transfected cells was measured in the presence of IL-3 and variable TGF-1 concentrations. As shown in Figure 5, the NFI-B–transfected cells exhibited decreased sensitivity to TGF-1 inhibition when compared to the vector-transfected cells. Discussion Excessive clonal expansion is the major phenotypic feature of the PV stem cell clone. To characterize genomic regions associated with this stem cell behavior, we screened the entire genome for LOH at high resolution using microsatellite analysis. We identified three genomic loci involved in LOH on chromosomes 9p, 10q, and 11q. The LOH on chromosome 9p occurred with the highest frequency. It was detected in 6 of 20 (33%) PV subjects and was absent in all clonal essential thrombocythemia subjects studied. These results establish 9pLOH as the most frequent chromosomal abnormality associated with PV described to date (followed by 20q deletions present in 8% of PV cases). We suspected mitotic recombination as the cause of LOH on chromosome 9p because 1) cytogenetic studies of the PV subjects with the 9pLOH did not reveal any losses of 9p, and 2)

the chromosomal region involved in LOH extended to the 9p telomere. Therefore, we examined the number of 9p copies in five 9pLOH subjects and found no detectable losses of chromosomal material, indicating that mitotic recombination is the mechanism of LOH in all the analyzed subjects. As a consequence, uniparental disomy is present in these subjects involving the entire short arm of chromosome 9. Unfortunately, none of the parents of the 9pLOH subjects were available for the analysis of parental origin of the disomic chromosomes. The 9pLOH due to mitotic recombination in PV is not detectable by cytogenetic analysis, fluorescence in situ hybridization, and comparative genomic hybridization. The B lymphocytes have been previously reported to be part of the PV clone [18]. X-chromosome inactivation studies always found T lymphocytes to be polyclonal in PV; however, these studies could not rule out that a minor proportion of these long-lived cells may be generated from the PV clone. Using 9p, 10q, and 11q LOH as clonality markers, we could for the first time examine this issue in a quantitative assay and confirm that the PV stem cell clone also generates T lymphocytes, thus establishing unequivocally the pluripotent stem cell origin of the PV defect. In addition to the LOH on chromosome 9p, we also detected two loci on chromosomes 10q and 11q involved in LOH at lower frequencies. This finding may reflect the heterogeneity of the acquired genetic lesions leading to clonal stem cell expansion in PV, although these loci should play a minor role in the development of typical PV phenotype. However, we cannot rule out the possibility that these isolated LOHs represent a random occurrence of phenotypically benign LOH in the same cell that also acquired a phenotypically relevant mutation and that the subsequent amplification of the mutant clone resulted in coamplification of these LOHs. The tumor suppressor gene PTEN is the most frequent target of LOH on chromosome 10q. We examined the PTEN expression level in CD34 cells and sequenced the entire PTEN gene in the 10q LOH PV subject. No abnormalities of the PTEN gene were found (data not shown).

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

The short arm of chromosome 9 has been shown to be one of the most mutated genomic regions in cancer. Frequent point mutations, homozygous deletions, and hypermethylation of the tumor suppressor genes CDKN2A and CDKN2B were characterized in a wide range of tumors. Hypermethylation of the CDKN2B gene (p15INK4B) had been documented in myelodysplastic syndrome, suggesting the involvement of deficient CDKN2B expression in clonal hematopoiesis [19]. We did not detect any CDKN2B mutations in 9pLOH PV subjects or abnormal CDKN2B expression in PV subjects with or without 9pLOH. The gene expression analysis performed in our study relied on uncultured cell populations isolated directly from peripheral blood samples. This may account for our inability to confirm the increased p16INK4A transcription in granulocytes and CD34 cells as previously reported for in vitro–expanded PV erythroid cells [12]. Although the CDKN2A and CDKN2B genes are the primary tumor suppressor loci on 9p, alternative 9p loci associated with neoplastic transformation had been proposed [20]. The role of the NFI-B gene in these alternative transformation mechanisms remains to be studied. Although we studied 19 candidate genes selected from the 9pLOH region, we cannot rule out defects in the genes that were excluded from our candidate gene list or genes that were not yet characterized within or mapped to the 9pLOH region. The detailed analysis of the role of 9p in the pathogenesis of PV will require further studies since only 25% of genes of this region were covered in this study. Our goal in this study was to identify genomic regions in PV subjects that satisfy the criteria for a clonal stem cell defect. Consequently, we focused on rapid screening for potential gene abnormalities within the identified regions rather than examining each candidate PV gene for expression in a comprehensive manner in a large set of PV samples. We identified increased expression of the NFI-B gene in several PV subjects positive for the presence of the 9pLOH. PV subjects without the 9pLOH had normal NFI-B levels, although it remains possible that in these subjects other functionally relevant changes other than increased transcript of this gene may be present. The frequency of increased NFI-B expression in PV as well as its possible correlation with the presence of the mitotic recombination on 9p remains to be determined in a larger set of PV subjects. These studies should be performed in not readily available CD34 cells wherein a mutation resulting in PV is expected to exert its disease-causing pathology. At this point, the association between the increased NFI-B expression and the occurrence of mitotic recombination resulting in LOH is unclear. Mutations of the NFI-B gene (copied into both of the NFI-B alleles by mitotic recombination) may be responsible for the increased NFI-B expression. Such mutations may be present either in the NFI-B promoter region, resulting in its increased transcription, or in the 7 kb-long 3-untranslated region (3-UTR), causing increased RNA stability. Sequence analysis of these regions

235

has not been performed as yet due to high GC content of the promoter and unavailability of the 3-UTR sequence of the NFI-B mRNA in GenBank. It is also possible that the increased NFI-B transcription is a result of epigenetic changes on chromosome 9p that are amplified by the mitotic recombination. Genome-wide hypomethylation and changes in chromatin structure occur frequently in cancer. Loss of methylation occurring in one of the alleles can duplicate into both alleles after a mitotic recombination. This would erase the methylation pattern of both alleles and in turn affect transcription of the locus. Genomic imprinting of genes (established by selective methylation of either the maternal or the paternal allele) can be altered by mitotic recombination, resulting in either loss of imprinting or imprinting of both alleles. Although there are no genes on chromosome 9p known to be imprinted, we cannot rule out such a possibility for the NFI-B gene. Interestingly, heterozygous animal for the knockout of the NFI-A gene (an NFI family member) had different survival depending on the parental origin of the knockout gene, suggesting the possibility of paternal imprinting [21]. It remains to be seen if NFI-B is the only gene affected by the uniparental disomy on 9p. NFI-B is a member of the nuclear factor I (NFI) transcription factor family (also known as CTF or CAAT box transcription factor). In vertebrates, the NFI family consists of four genes (NFI-A, NFI-B, NFI-C, and NFI-X) binding as homo- or heterodimers to the TTGGC(N5)GCCAA consensus sequence [22]. Since a large number of cellular and viral gene promoters contain the NFI binding site, at this point it is impossible to predict the phenotypic outcome of increased NFI-B expression. Interestingly, overexpression of one of the NFI genes (NFI-X) was shown to result in TGF-1 resistance in vitro [17]. TGF- is an important negative regulator of hematopoiesis. Given the structural homology and functional redundancy of the NFI genes, other NFI genes seemed likely to play a role in TGF-–induced cell-cycle arrest. When we transfected the mouse 32D cell line with the NFI-B construct we detected TGF-1 resistance in the transfected cells. Although TGF-1 resistance (or decreased sensitivity to TGF-1) would be an interesting explanation of clonal stem cell expansion in PV, it remains to be determined if the increased expression of NFI-B results in decreased sensitivity of PV cells to TGF-. In fact, it is known that some cancer cells exhibit TGF- resistance and produce TGF- by an autocrine mechanism that leads to inhibition of the normal cell milieu. If such a mechanism occurs in PV it may explain the expansion of the PV clone, and the yetmysterious suppression of non-PV hematopoiesis by the PV clone. This hypothesis is now being examined.

Acknowledgments The work was supported by the Myeloproliferative Disorders Foundation.

236

R. Kralovics et al./Experimental Hematology 30 (2002) 229–236

References 1. Adamson JW, Fialkow PJ, Murphy S, Prchal JF, Steinmann L (1976) Polycythemia vera: stem-cell and probable clonal origin of the disease. N Engl J Med 295:913 2. Prchal JT, Prchal JF, Belickova M, et al. (1996) Clonal stability of blood cell lineages indicated by X-chromosomal transcriptional polymorphism. J Exp Med 183:561 3. Diez-Martin JL, Graham DL, Petitt RM, Dewald GW (1991) Chromosome studies in 104 patients with polycythemia vera. Mayo Clin Proc 66:287 4. Bench AJ, Nacheva EP, Champion KM, Green AR (1998) Molecular genetics and cytogenetics of myeloproliferative disorders. Baillieres Clin Haematol 11:819 5. Prchal JF, Axelrad AA (1974) Letter: Bone-marrow responses in polycythemia vera. N Engl J Med 290:1382 6. Prchal JF, Adamson JW, Murphy S, Steinmann L, Fialkow PJ (1978) Polycythemia vera. The in vitro response of normal and abnormal stem cell lines to erythropoietin. J Clin Invest 61:1044 7. Sui X, Krantz SB, Zhao Z (1997) Identification of increased protein tyrosine phosphatase activity in polycythemia vera erythroid progenitor cells. Blood 90:651 8. Wickrema A, Chen F, Namin F, et al. (1999) Defective expression of the SHP-1 phosphatase in polycythemia vera. Exp Hematol 27:1124 9. Moliterno AR, Hankins WD, Spivak JL (1998) Impaired expression of the thrombopoietin receptor by platelets from patients with polycythemia vera (see comments). N Engl J Med 338:572 10. Silva M, Richard C, Benito A, Sanz C, Olalla I, Fernandez-Luna JL (1998) Expression of bcl-x in erythroid precursors from patients with polycythemia vera. N Engl J Med 338:564 11. Temerinac S, Klippel S, Strunck E, et al. (2000) Cloning of PRV-1, a novel member of the uPAR receptor superfamily, which is overexpressed in polycythemia rubra vera. Blood 95:2569

12. Dai R, Krantz SB (2001) Increased expression of the INK4a/ARF locus in polycythemia vera. Blood 97:3424 13. Kralovics R, Castilos FA, Prchal JT (1999) Familial polycythemia vera: mode of inheritance, clonality, and genetic analysis of candidate genes and chromosomal regions. Blood 94:113a 14. Liu Y, Phelan J, Go RC, Prchal JF, Prchal JT (1997) Rapid determination of clonality by detection of two closely linked X-chromosome exonic polymorphisms using allele-specific PCR. J Clin Invest 99: 1984 15. Gregg XT, Kralovics R, Prchal JT (2000) A polymorphism of the X-linked gene IDS increases the number of females informative for transcriptional clonality assays. Am J Hematol 63:184 16. Wei Q, Guan Y, Cheng L, et al. (1997) Expression of five selected human mismatch repair genes simultaneously detected in normal and cancer cell lines by a nonradioactive multiplex reverse transcription polymerase chain reaction. Pathobiology 65:293 17. Sun P, Dong P, Dai K, Hannon GJ, Beach D (1998) p53-independent role of MDM2 in TGF-1 resistance. Science 282:2270 18. Prchal JT, Guan YL, Prchal JF, Barany F (1993) Transcriptional analysis of the active X-chromosome in normal and clonal hematopoiesis. Blood 81:269 19. Uchida T, Kinoshita T, Nagai H, et al. (1997) Hypermethylation of the p15INK4B gene in myelodysplastic syndromes. Blood 90:1403 20. An HX, Claas A, Savelyeva L, et al. (1999) Two regions of deletion in 9p23-24 in sporadic breast cancer. Cancer Res 59:3941 21. das Neves L, Duchala CS, Godinho F, et al. (1999) Disruption of the murine nuclear factor I-A gene (Nfia) results in perinatal lethality, hydrocephalus, and agenesis of the corpus callosum. Proc Natl Acad Sci U S A 96:11946 22. Gronostajski RM (2000) Roles of the NFI/CTF gene family in transcription and development. Gene 249:31