Gene and microRNA analysis of neutrophils from patients with polycythemia vera and essential thrombocytosis: down-regulation of micro RNA-1 and -133a

These studies were approved by institutional review boards at the NIDDK, NIH and Veterans Administration Medical Center, Washington DC. Whole blood was collected into EDTA tubes from patients with MPD, healthy subjects, and healthy subjects given G-CSF. Neutrophils isolated from the EDTA blood was used for gene expression and microRNA analysis. For MPD patients whole blood was also collected into citrate tubes and was used to isolate neutrophils for JAK V617F analysis. Blood collected in tubes without anticoagulant was used to obtain serum for protein analysis. WHO criteria was used to make the diagnosis of PV and ET [10].

Healthy subjects were given 10 micrograms/kg of G-CSF (filgrastim, Amgen, Thousand Oaks, California, USA) subcutaneously daily for 5 days. Blood was collected for analysis approximately 2 hours after the last dose of G-CSF was given.

Whole blood, 6 mL in EDTA (K2 EDTA 1.8 mg/mL, BD Vacutainer, Becton, Dickinson and Company, Franklin Lakes, NJ), was collected from healthy donors, MPD patients and donors following a course of G-CSF treatment. Percoll (Sigma, St. Louis, Missouri, USA) density gradients were used to isolate the neutrophils. Briefly, gradients were prepared by gently overlaying 63% Percoll solution on top of 72% Percoll solution, in equal volumes. Prior to overlaying the whole blood sample on the gradient, the majority of red blood cells were removed via sedimentation by diluting whole blood 1:2 with hetastarch (Hespan; 6% heta starch in 0.9% sodium chloride, B. Braun Medical Inc., Irvine, California, USA) and incubating for approximately 20 minutes at room temperature. After layering the leukocyte rich/heta starch solution on the gradient, the sample was centrifuged at 1,500 rpm for 25 minutes with no brake upon centrifuge deceleration. The neutrophil layer was harvested from the interface between the two Percoll solutions and washed twice with physiologic saline.

Flow cytometry analysis of granulocyte surface markers was performed on fresh whole blood samples. Cells were stained with monoclonal antibodies against CD177-FITC, CD15-FITC (Chemicon International, Temecula, CA), CD64-FITC, CD16-FITC, CD18-FITC, CD11b-FITC (Caltag Laboratories, Buckingham, UK) CD10-PE, CD31-PE, CD44-FITC, CD45-FITC, CD55-FITC, CD59-FITC, CD62L-FITC (eBiosciences, San Diego, CA) and incubated at 4°C for 30 minutes in the dark. Mouse IgG isotype controls were also used (Caltag Laboratories). The FACSCalibur flow cytometer and CellQuest Pro software (BD Biosciences, San Jose, CA) were used for analysis by acquiring 10,000 events and determining the viable neutrophil population by light scatter.

Isolated neutrophils were tested for JAK2 V617F by DNA sequencing. V617F mutations were identified utilizing sequence-based typing methodology. Primary amplification of the specific region of JAK2 utilized primers Jak2-1 (pf) = tgc tga aag tag gag aaa gtg cat and Jak2-2 (pr, sr) = tcc tac agt gtt ttc agt ttc aa which produced a 345bp product. After primary amplification, sequence primers Jak2-5 (sf) = agt ctt tct ttg aag cag caa and Jak2-2 (pr, sr) = tcc tac agt gtt ttc agt ttc aa were utilized for detection of the V617F mutation. Conditions included the use of 2.0 mM Mg++, 3 pmole of primer, GeneAmp 10× PCR Gold Buffer, 0.35 unit of AmpliTaq gold DNA polymerase (ABI) 5 U/ul, and 0.15 mM each of 10 mM dNTP mixture (Amersham) with Big Dye Terminator^® Cycle Sequencing kits (Applied Biosystems). Template DNA was utilized at a concentration of 40–60 ug/mL. PCR cycling parameters were 95°C for 10 minutes; 95°C for 30 seconds → 52°C for 40 seconds → 72°C for 40 seconds = 40 cycles; 72°C for 2 minutes and hold at 4°C. Sequencing reactions were run on an Applied Biosystem 3730xL DNA Analyzer and analyzed utilizing standard alignment software.

Total RNA from harvested neutrophils was extracted using Trizol reagent according to the manufacturer's instructions (Invitrogen, Carlsbad, California, USA). The quality of secondary amplified RNA was tested with the Agilent Bioanalyzer 2000 (Agilent Technologies, Waldbronn, Germany) and amplified into antisense RNA (aRNA) as previously described [11]. Also total RNA from peripheral blood mononuclear cells pooled from six normal donors was extracted and amplified into aRNA to serve as the reference. Pooled reference and test aRNA were isolated and amplified in identical conditions to avoid possible interexperimental biases. Both reference and test aRNA were directly labeled using ULS aRNA Fluorescent Labeling kit (Kreatech, Amsterdam, The Netherlands) with Cy3 for reference and Cy5 for test samples. Whole-genome human 36 K oligonucleotide arrays were printed in the Infectious Disease and Immunogenetics Section of the Department of Transfusion Medicine, Clinical Center, NIH (Bethesda, Maryland, USA) using oligonucleotides purchased from Operon (Operon, Huntsville, Alabama, USA). The Operon Human Genome Array-Ready Oligo Set version 4.0 contains 35,035 oligonucleotide probes, representing approximately 25,100 unique genes and 39,600 transcripts excluding control oligonucleotides. The design is based on the Ensembl Human Database build (NCBI-35c) with full coverage on NCBI human Refseq dataset (04/04/2005). The microarray is composed of 48 blocks and one spot is printed per probe per slide. Hybridization was carried out in a water bath at 42°C for 18 to 24 hours and the arrays were then washed and scanned on a GenePix 4000 scanner at variable photomultiplier tube to obtain optimized signal intensities with minimum (<1% spots) intensity saturation. The resulting data files were uploaded to the mAdb database http://​nciarray.​nci.​nih.​gov and further analyzed using BRBArrayTools developed by the Biometric Research Branch, National Cancer Institute http://​linus.​nci.​nih.​gov/​BRB-ArrayTools.​html.

A microRNA probe set was designed using mature antisense microRNA sequences (Sanger data base, version 9.1) consisting of 827 unique microRNAs from human, mouse, rat and virus plus two control probes. The probes were 5' amine modified and printed in duplicate on CodeLink activated slides (General Electric, GE Health, New Jersey, USA) via covalent bonding in the Immunogenetics Laboratory, DTM, CC, NIH. 4 μg total RNA isolated by using Trizol reagent (Invitrogen, Carlsbad, California) was directly labeled with miRCURY™ LNA Array Power Labeling Kit (Exiqon, Woburn, Massachusetts, USA) according to manufacture's procedure. The total RNA from an Epstein-Barr virus (EBV)-transformed lymphoblastoid cell line was used as the reference for the microRNA expression array assay. The test sample was labeled with Hy5 and the reference with Hy3. After labeling, the sample and the reference were co-hybridized to the microRNA array at room temperature overnight in the presence of blocking reagents as previously described [12] and the slides were washed and scanned by GenePix scanner Pro 4.0 (Axon, Sunnyvale, California, USA). Resulting data files were uploaded to the mAdb database http://​nciarray.​nci.​nih.​gov and further analyzed using BRBArrayTools developed by the Biometric Research Branch, National Cancer Institute http://​linus.​nci.​nih.​gov/​BRB-ArrayTools.​html.

For analysis of the gene and microRNA array data, the raw data set was filtered according to a standard procedure to exclude spots with minimum intensity that was arbitrarily set to an intensity parameter of 200 for gene expression data and 100 for microRNA array data in both fluorescence channels. Spots flagged by the analysis software and spots with diameters <20 μm for gene expression array and <10 μm for the microRNA array were excluded from the analysis.

The filtered data were normalized using median over entire array and were retrieved by the BRB ArrayTool http://​linus.​nci.​nih.​gov/​BRB-ArrayTools.​html developed at the National Cancer Institute (NCI), Biometric Research Branch, Division of Cancer Treatment and Diagnosis. Hierarchical cluster analysis was conducted on the genes or microRNA using Cluster and TreeView software [13]. For annotation of genes and functional pathways, the Database for Annotation, Visualization and Integrated Discovery (DAVID) 2007 software http://​david.​abcc.​ncifcrf.​gov/​[14] and Ingenuity Pathway Analysis software http://​www.​ingenuity.​com was used. All microRNA target prediction analysis used BRB ArrayTool microRNA targets program http://​linus.​nci.​nih.​gov/​BRB-ArrayTools.​html, TargetScan http://​www.​targetscan.​org/​ and miRBase Targets http://​microrna.​sanger.​ac.​uk.

To validate the microarray analysis, 5 genes and 2 microRNAs were selected for Quantitative PCR. Gene expressions for TNFAIP3 (Assay ID, Hs00234713_m1), NFKBIE (Assay ID, Hs00234431_m1), NFKBIA (Assay ID Hs00153283_m1), CBS (Assay ID Hs00163925_m1) and MCL1(Assay ID Hs03043899_m1) were quantified by TaqMan Gene Expression Assays (Applied Biosystems, Foster City, California, USA) according to manufacturers' protocol and normalized by GAPDH (Assay ID Hs99999905_m1) PCR amplification of target genes and quantification of the amount of PCR products were performed by ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems). Differences in expression were determined by the relative quantification method; the Ct values of the test genes were normalized to the Ct values of endogenous control GAPDH. The fold change was calculated using the equation 2^-ΔΔCt.

Differentially expressed microRNAs, miR-133a (Assay ID, 4373142) and miR-219 (Assay ID, 4373080), were measured by TaqMan microRNA Assays (Applied Biosystems, Foster City, California, USA) as previously reported [15]. The differences of expression were determined by relative quantification method; the Ct values of microRNAs were normalized to the Ct values of endogenous control RNU48 (Assay ID 4373383). The fold change was calculated using the equation 2^-ΔΔCt.

Serum samples were collected and frozen immediately, and stored at -80°C until further analysis. The serum samples were analyzed by protein expression profiling. The level of 64 soluble factors were assessed on an ELISA-based platform (Pierce Search Light Proteome Array, Boston, MA) consisting of multiplexed assays that measured up to 16 proteins per well in standard 96 well plates (Table 1). The 64 factors were selected to included hematopoietic factors, factors associated with inflammation, and those previously found to be increased in the serum of healthy subjects given G-CSF [16].

Unsupervised analysis was performed by using BRBArrayTools http://​linus.​nci.​nih.​gov/​BRB-ArrayTools.​html and the Stanford Cluster Program [17]. Class comparison analysis was performed using parametric unpaired Student's t-test to identify differentially expressed genes or microRNA among different sample groups and using different significance cutoff levels as demanded by the statistical power of each comparison. Statistical significance and adjustments for multiple test comparisons were based on univariate and multivariate permutation tests as previously described [18, 19].

Neutrophils from 6 MPD patients were studied; 4 with PV and 2 with ET. JAK2 V617F was detected in 3 of the 4 PV patients and in 1 of the 2 ET patients (Table 2). Global gene expression analyses of neutrophils from 6 subjects with MPDs were compared with 6 healthy subjects given 5 days of G-CSF and the 5 healthy subjects. Among the 17 samples and 35,000 probes in the array, 3,617 were expressed by 80% of the samples and their expression was increased by 2-fold or greater in at least one sample. Unsupervised hierarchical clustering analysis of these 3,617 genes revealed three distinct groups: the G-CSF group which included 5 of the 6 G-CSF mobilized neutrophil samples, the MPD group with 4 of the 6 MPD neutrophil samples and 2 healthy subject neutrophils, and the mixed group with 3 healthy subject, 2 MPD, and 1 G-CSF-mobilized neutrophils (Figure 1).

These results showed that the gene expression profile of MPD neutrophils differed from that of healthy subject neutrophils and G-CSF-mobilized neutrophils. Further analysis found that the expression of 1,006 genes differed among neutrophils from the MPD patients, healthy subjects, and healthy subjects given G-CSF (F-test, p ≤ 0.005). Hierarchical clustering analysis of these 1,006 genes separated the neutrophils into 3 groups; one contained neutrophils from 5 of 6 MPD patients, another included neutrophils from 5 healthy subjects and 1 MPD patient, and the third contained neutrophils from all 6 subjects given G-CSF (Figure 2). In this gene expression profile the MPD neutrophils aligned closer to the healthy subject neutrophils than the G-CSF-mobilized neutrophils. Two clusters of genes distinguished the MPD neutrophils from the healthy subject neutrophils. One cluster was made up of 17 genes whose expression was increased more in MPD neutrophils than in neutrophils from healthy subjects or healthy subjects given G-CSF (Figure 2, cluster 1) and another contained 38 genes down-regulated in MPD neutrophils but not in healthy subjects or G-CSF mobilized neutrophils (Figure 2, cluster 2). The cluster of MPD up-regulated genes included FRAT1, ZNF652, LMO4, IL10RB, and cystathionine β-synthase (CBS). FRAT1 is a regulator of the Wnt signaling pathway and is overexpressed in esophageal squamous cell carcinoma [20]. ZNF652 has a role in the suppression of breast oncogenesis and vulvar cancer [21, 22]. LMO4 is a transcription regulator and increased expression of LMO4 in pancreatic ductal adenocarcinoma is associated with a survival advantage [23]. The expression of CBS has been previously reported to be up-regulated in neutrophils from patients with MPDs [24]. Among the down-regulated genes were ribosomal proteins including 3 copies of RPL10, 2 copies of RPL3, and RPS9, RPS10P3, and RPL12P6; proteosome proteins including 3 copies of PSMD2 and PSMC; and cytochrome c oxidases COX5B and COX7A2.

To further explore the differences between MPD and G-CSF-mobilized neutrophils, the genes differentially expressed in MPD neutrophils compared to healthy subject neutrophils were identified as well as those differentially expressed in G-CSF-mobilized-neutrophils. MPD neutrophil differentially expressed genes were more likely to belong to inflammatory pathways (Figure 3A). In contrast, G-CSF-mobilized neutrophils differentially expressed genes were more likely to belong to metabolic pathways (Figure 3B).

To further characterize MPD neutrophils, we identified those differentially expressed genes whose expression was increased or decreased to the greatest fold as compared to the healthy subjects. Among the 30 genes whose expression was increased to the greatest extent in MPD neutrophils were ZNF652, CBS, LMO4, AXUD1, MCL1 and CCR1 (Table 3). AXUD1 is a regulator of the Wnt signaling pathway and is down-regulated in lung, kidney, and colon cancer [25]. MCL-1 is a member of the Bcl-2 family and is an important anti-apoptotic molecule for multiple types of hematopoietic cells [26]. CCR1 is a chemokine receptor for at least 11 different chemokines including CCL3 (MIP-1α), CCL5 (RANTES), CCL7 (MCP-3), CCL8 (MCP-2), CCL14, CCL15, CCL16 and CCL23 [27]. Among the genes down-regulated most in MPD neutrophils were neutrophil elastase 2 (ELA2) and two NF-kβ pathway genes (NFKBIA and NFKBIE) all of which are involved in inflammation (Table 4).

We used qRT-PCR to further confirm the differential expression of 3 NFKB pathway genes, NFKBIA, NFKBIE and TNFAIP3 as well as MCL1 and CBS (Figure 4). This confirmed that the expression of NFKBIA, NFKBIE, and TNFAIP3 were significantly down-regulated in both MPD and G-CSF-mobilized neutrophils compared to those from healthy subjects. The expression of CBS was significantly up-regulated in MPD neutrophils and the expression of MCL1 was up-regulated but not to a significant degree as compared to healthy subjects.

MicroRNA expression was compared among MPD, G-CSF-mobilized and healthy subject neutrophils using a microarray. Among the 827 probes, 500 remained after selecting only those expressed in >80% of samples. Unsupervised hierarchical clustering analysis of the neutrophil samples separated the samples into two groups. One group included 3 G-CSF-mobilized neutrophils and 3 healthy subject neutrophils and the second included 3 G-CSF-mobilized neutrophils, 6 MPD neutrophils and 5 normal donor neutrophils (data not shown).

Comparison of the expression of microRNA between MPD and healthy subject neutrophils found that the expression of 21 microRNA were up-regulated in MPD neutrophils and 11 were down-regulated (p < 0.05). Among the microRNA up-regulated in MPD neutrophils were 5 that were increased more than 2-fold; miR-219, miR-515-5p, miR-142-5p, miR-143, and miR-101 (Table 5). The up-regulation of miR-219 in MPD neutrophils compared to those from healthy subjects was confirmed by qRT-PCR (Figure 5). Interestingly, miR-219 has been found to be expressed in the brain and its levels exhibit circadian rhythms and are involved in the control of the suprachiasmatic nuclei (SCN), the master circadian clock in mammals [28]. The expression of 142–5p has also been found to be increased in peripheral blood leukocytes [12]. MicroRNA miR-143 has been found to be involved with cell differentiation. The differentiation of pre-adipocytes to adipocytes is associated with the increased levels of miR-143 [29]. Bruchova and colleagues have found that miR-143 is up-regulated in neutrophils from patients with polycythemia vera [30]. The expression of miR-143 is down-regulated in B cell malignancies, Burkitt's lymphoma cell lines [31], and colorectal cancer [32].

Among the microRNA down-regulated in MPD neutrophils the expression of five were decreased more than 2-fold: miR-133a, miR-504, miR-565, miR-1, and miR-216 (Table 5). The down-regulation of miR-133a in MPD neutrophils was confirmed by qRT-PCR (Figure 5). MicroRNA miR-133a and -1 are clustered on the same chromosome and are transcribed together as a single transcript [33, 34]. These two microRNA are preferentially expressed in brown adipocytes [35], cardiac, and skeletal muscle [34] and are important in the differentiation and regulation of cardiac and skeletal muscle. Little is known about miR-216, -504 and -565. Micro RNA-216 is expressed by the pancreas. A comparison of normal pancreas with 33 other tissues found that the expression of miR-216 and miR-217 and the lack of expression of miR-133a were characteristic of pancreatic tissue [36].

The levels of 64 serum proteins were compared in the 6 MPD patients and 7 healthy subjects. The levels of the 64 factors in each of the 6 MPD patients and 7 healthy controls were analyzed by supervised hierarchical clustering analysis (Figure 6). The MPD samples were characterized by 33 proteins whose levels were greater than in healthy subjects. Eleven of these were significantly increased in MPD patients compared to healthy subjects (t-tests, p < 0.05, Table 6) and included 2 chemokines (CXCL11 and CCL3), a cytokine (IL-1a), 2 matrix metalloproteinases (MMPs) (MMP-10 and MMP-13), growth factors (PDGF-BB and G-CSF) VCAM, TIMP-1, IL-6R and P-selectin.

Neutrophil expression of CD11b, CD15, CD16, CD18 and CD177 was analyzed by flow cytometry in 24 patients with MPD (11 PV and 13 ET). JAK2 V617F was detected in 13 of the 24 patients and one was homozygous (Table 7). Expression was compared to 43 healthy subjects and 27 healthy subjects who were given 5 daily doses of G-CSF.

CD15 and CD18 expression differed among MPD patients and healthy subjects, but not that of CD11b, CD16 or CD177. More neutrophils expressed CD15, Lewis-x, in people with MPD than in healthy subjects (50 ± 31% versus 21 ± 25%, p < 0.0002) (Table 7, Figure 7). This was the case for both subjects with PV and ET. The proportion of neutrophils expressing CD18 was also increased in people with MPD (73 ± 26% versus 48 ± 33%, p < 0.003), although the mean neutrophil fluorescent intensity was reduced (250 ± 81 versus 451 ± 300, p < 0.003) (Table 7, Figure 7), but was similar to G-CSF stimulated neutrophils. Both the proportion of neutrophils expressing CD177 and the mean fluorescence intensity of neutrophils were increased slightly in MPD neutrophils, but these changes were not significant.

Following G-CSF administration, the expression of CD16 and CD18 as assessed by the mean fluorescence intensity decreased (Table 7, Figure 7). In contrast, the number of neutrophils expressing CD177 and the mean fluorescence intensity of CD177 expression increased.

The expression of several other neutrophil adhesion molecules, Fc receptors and other antigens were compared in the same cohort of 6 MPD patients in whom gene and miR expression profiles and serum proteins were measured; 4 with PV and 2 with ET. The proportion of neutrophils expressing CD64 was greater in MPD patients than in healthy subjects (13 ± 9% versus 6 ± 4%, p < 0.05) but not the mean fluorescence intensity (373 ± 73 versus 201 ± 63). There was no difference in the expression of CD10, CD31, CD44, CD45, CD55, CD59, and CD62L among neutrophils from MPD patients and healthy subjects (data not shown).