Figures
Abstract
In this study we sought to identify genetic factors associated with the presenting white blood cell (WBC) count in B-precursor acute lymphoblastic leukemia (BP-ALL). Using ETV6-RUNX1-positive BP-ALL patient samples, a homogeneous subtype, we identified 16 differentially expressed genes based on the presenting WBC count (< 50,000/cumm vs > 50,000). We further confirmed that IL1R1, BCAR3, KCNH2, PIR, and ZDHHC23 were differentially expressed in a larger cohort of ETV6-RUNX1-negative BP-ALL patient samples. Statistical analysis demonstrated that expression levels of these genes could accurately categorize high and low WBC count subjects using two independent patient sets, representing positive and negative ETV6-RUNX1 cases. Further studies in leukemia cell line models will better delineate the role of these genes in regulating the white blood cell count and potentially identify new therapeutic targets.
Citation: Edwards H, Rubenstein M, Dombkowski AA, Caldwell JT, Chu R, Xavier AC, et al. (2016) Gene Signature of High White Blood Cell Count in B-Precursor Acute Lymphoblastic Leukemia. PLoS ONE 11(8): e0161539. https://doi.org/10.1371/journal.pone.0161539
Editor: Obul Reddy Bandapalli, German Cancer Research Center (DKFZ), GERMANY
Received: April 12, 2016; Accepted: August 8, 2016; Published: August 18, 2016
Copyright: © 2016 Edwards et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The gene expression microarray data has been deposited in the GEO repository, accession number GSE63370.
Funding: This study was supported by grants from the National Cancer Institute (R01 CA120772), Center for Urban Responses to Environmental Stressors [CURES: Wayne State University] P30 Center grant (P30 ES020957), Karmanos Cancer Institute, Kids Without Cancer, Elana Fund, the Herrick Foundation, Justin’s Gift Charity, The Buric Family, Sehn Family Foundation, Dale Meyer Memorial Endowment for Leukemia Research, Children’s Hospital of Michigan Foundation, Ginopolis-Karmanos Endowment, Decerchio/Guisewite Family, and the Ring Screw Textron Endowed Chair for Pediatric Cancer Research. JTC is a predoctoral trainee supported by T32 CA009531 from the National Cancer Institute. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer. Over the last 30 years, the 5-year survival rate has significantly improved, increasing from 57% to 90% [1]. The introduction of risk-stratified therapy has contributed significantly to the improved survival [2]. For B-precursor ALL (BP-ALL), the National Cancer Institute has defined two risk groups based on age and presenting white blood cell (WBC) count, which continue to remain important prognostic factors today [2]. The standard risk group consists of patients with a presenting WBC count < 50,000/cumm and between 1 and 10 years of age, whereas the high risk group has a WBC count ≥ 50,000/cumm and/or age ≥ 10 years [3]. To date, factors regulating the presenting WBC count and its association with prognosis in BP-ALL remain unknown. At diagnosis, there is essentially complete replacement of normal hematopoiesis with blast cells and there are no differentiating morphologic features between patients with low and high WBC count (Fig 1a). Based on this, it is hard to envision why, for example, a one log difference in peripheral WBC (e.g. 10,000/cumm vs 100,000/cumm) can have such a major impact on prognosis. Thus, identifying genetic factors associated with the regulation of WBCs in the peripheral circulation may improve our understanding of this disease and potentially lead to the development of new therapies.
(a) Diagnostic bone marrow aspiration samples from a patient with low (left side) and high (right side) WBC count at initial presentation were Wright-Giesma stained. (b) Total RNAs were isolated from 15 ETV6-RUNX1-positive patient samples. Aminyoallyl-aRNA was produced using 100 ng total RNA. Gene expression microarray was performed using the Agilent 60-mer oligo array (Human Gene Expression V2, 8X60K). Slides were scanned with the Agilent dual laser scanner with SureScan High Resolution Technology. Tiff images were analyzed using Agilent’s feature extraction software version 11.0.1.1 to obtain fluorescent intensities for each spot on the array. Probes that were not detected significantly above background in at least 75% of the samples for at least any one category (low or high WBC) were filtered out, leaving 28,973 probes. Differentially expressed genes were identified by P-values ≤ 0.05, which were calculated using a moderated t-test with Westfall and Young multiple test corrections. Principal component analysis using the 26 differentially expressed features was performed. Data points are labeled with WBC count (x 103/cumm).
BP-ALL is complex and heterogeneous which could complicate the identification of genetic factors linked to the peripheral WBC count. Therefore, we selected the largest molecular subgroup, ETV6-RUNX1-positive BP-ALL accounting for ~25% of pediatric ALL cases, to identify genes linked to the peripheral WBC count. Although these cases are usually associated with low-risk features, up to 25% of cases have a presenting WBC count > 50,000/cumm. We found that IL1R1 (Il1r1 interleukin 1 receptor, type I), BCAR3 (breast cancer anti-estrogen resistance 3), KCNH2 (potassium channel, voltage gated related subfamily H, member 2), PIR (pirin), and ZDHHC23 (zinc finger, DHHC-type containing 23) were differentially expressed. These findings were further confirmed in a larger cohort of samples from patients with ETV6-RUNX1-negative BP-ALL, thus identifying a unique gene expression signature for high WBC in BP-ALL.
Methods
Clinical Samples
Diagnostic bone marrow samples with de novo ALL were obtained from the Children’s Hospital of Michigan leukemia cell bank. Mononuclear cells were purified by standard Ficoll-Hypaque density centrifugation. Written informed consent was provided by the parent or legal guardian according to the Declaration of Helsinki. Sample handling and data analysis protocols were approved by the Human Investigation Committee of the Wayne State University School of Medicine.
Gene Expression Microarray Analysis
Total RNAs were extracted using TRIzol according to the manufacturer’s instructions (Life Technologies, Carlsbad, CA). Aminoallyl-aRNA was produced using TargetAMP 1-Round Aminoallyl-aRNA Amplification Kit 101 (Epicentre, Madison, WI) and Agilent Spike-in Controls for one color microarrays according to the manufacturer’s protocol (Agilent Technologies, Santa Clara, CA). Three μg of each aminoallyl-aRNA sample was incubated with Alexa Fluor 555 (Life Technologies) for 30 min at room temperature and then run through the RNeasy Mini Elute column (Qiagen, Valencia, CA) to remove unincorporated dye. The samples were prepared for hybridization following the Agilent “One-Color Microarray-Based GE Analysis” protocol. Six hundred ng of Alexa Fluor 555 labeled aminoallyl-aRNA was used to hybridize to the Agilent 60-mer oligo array (Human Gene Expression V2, 8X60K). The data has been deposited in the GEO repository, accession number GSE63370.
Quantification of Gene Expression by Real-time RT-PCR
cDNAs were prepared, as previously described [4]. Transcripts were quantitated using TaqMan probes (Life Technologies) and a LightCycler LC480 real-time PCR machine (Roche Diagnostics, Indianapolis, IN), based on the manufacturer’s instructions.
Protein Interaction Network Analysis
Protein interaction analysis was performed using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins, string-db.org) (PMID 25352553). Data settings in the program were: active interaction sources = all; minimum required interaction score = 0.150; max number of interactors, first shell = 20, second shell = 10. Gene ontology (GO) analysis was performed within STRING using the GO biological process.
Results and Discussion
In this study, we examined the gene expression profile of 15 diagnostic ETV6-RUNX1-positive primary patient samples (age < 10 years) by gene expression microarray, comparing patients with low vs high WBC count (Table 1). We found 26 differentially expressed features (probes, P-values ≤ 0.05). Principal component analysis revealed sample clustering based on the combined profiles of these features, with high WBC samples clustering on the right and low WBC on the left (Fig 1b). These results demonstrated that the samples could be separated by the expression profiles of the 26 differentially expressed features. The probe IDs were then cross-referenced with Entrez Gene IDs, which provided 16 gene IDs. Validation of the microarray results were performed using real-time RT-PCR. When comparing low and high WBC samples, BCAR3 (P = 0.0127), IL1R1 (P = 0.0193), KCNH2 (P = 0.0193), PIR (P = 0.0080), ZDHHC23 (P = 0.0193), SLC5A11 (P = 0.0047), JAKMIP1 (P = 0.0047), and P2RX7 (P = 0.0047) expression levels were significantly different (Fig 2). To extend our findings, we measured the expression levels of these genes in 60 randomly selected ETV6-RUNX1-negative pediatric BP-ALL samples (Table 2). Expression of BCAR3 (P = 0.0028), IL1R1 (P = 0.0046), KCNH2 (P < 0.0001), PIR (P = 0.0009), and ZDHHC23 (P = 0.0170) were significantly different between the low and high WBC samples (Fig 3). Then we used logistic regression to create a predictive model from these five genes measured by RT-PCR in the 60 ETV6-RUNX1-negative pediatric BP-ALL cases and obtained a predictive model with a goodness of fit P-value < 0.0001. Furthermore, the predictive power of the model was evident from receiver operating characteristic (ROC) curve analysis. The area under the curve was 0.88 (data not shown), indicating a very good predictive accuracy in classifying samples as either high or low WBC based only on the expression pattern of the five genes.
Total RNAs from ETV6-RUNX1-positive patient samples were isolated. cDNA was prepared using 1 μg total RNA and transcript levels were determined by real-time RT-PCR. Transcript levels were normalized to RPL13a and relative expression levels were calculated using the comparative Ct method (comparing all samples to the expression levels in the ETV6-RUNX1-positive REH cell line). The horizontal lines indicate the median. The P-values were calculated using the Mann-Whitney U-test and GraphPad Prism 5.0 software.
Total RNAs from ETV6-RUNX1-negative patient samples were isolated. cDNA was prepared using 1 μg total RNA and transcript levels were determined by real-time RT-PCR. Transcript levels were normalized to RPL13a and relative expression levels were calculated using the comparative Ct method (comparing all samples to the expression levels in the ETV6-RUNX1-positive REH cell line). The horizontal lines indicate the median. The P-values were calculated using the Mann-Whitney U-test and GraphPad Prism 5.0 software.
Our results suggest that expression levels of BCAR3, IL1R1, KCNH2, PIR, and ZDHHC23 make up the gene signature associated with high WBC count BP-ALL. The roles these genes play in the regulation of WBCs in the peripheral blood circulation and/or their association with prognosis remain to be determined. Pirin is an iron-dependent redox coregulator of NF-KB [5]. In acute myeloid leukemia, downregulation of Pirin contributes to myeloid differentiation arrest [6]. KCNH2 (hERG1; Kv11.1) is expressed in multiple cancers and controls cell proliferation and apoptosis [7, 8]. BCAR3 has been shown to regulate breast cancer cell migration, proliferation, and adhesion [9]. IL1R1 is the receptor protein for IL1α and IL1β, which are involved in B cell proliferation [10]. ZDHHC23 is involved in regulating expression of calcium-activated potassium channels [11].
We investigated protein-protein interactions and functional associations between the five proteins using STRING (PMID 25352553). The analysis revealed that there are no known direct interactions between the five proteins. To investigate indirect interactions between the five proteins we also considered the top 30 additional proteins from the STRING database having primary and secondary interactions with the five target proteins. The overall interactions between this set of 35 proteins (nodes) forms a connected graph with all nodes reachable, including the five target proteins. This result suggests that the five proteins may have common functional associations. Gene ontology analysis of the 35 proteins revealed that this set is significantly enriched for proteins involved in “immune response-activating signal transduction” (FDR P = 1.1e-13). Further studies to characterize the impact of dysregulation of these five genes on immune response signaling are warranted.
In conclusion, we identified, for the first time, genes linked to the peripheral WBC count in BP-ALL; BCAR3, IL1R1, KCNH2, PIR, and ZDHHC23 expression make up the gene signature of high WBC count BP-ALL. Statistical analysis demonstrated that expression levels of the genes identified in this study could accurately categorize high and low WBC subjects using two independent patient sets, representing positive and negative ETV6-RUNX1 cases. Our results, although in a small sample set, suggest that these genes may be involved in the underlying cellular mechanism leading to differences in WBC and patient prognosis and warrants further investigation in a prospective study. It would be interesting to determine if expression of these genes correlate with clinical outcome and establish them as independent prognostic markers. However, due to the patients being treated on various clinical trials over the past 15 years, the relationship with clinical outcome is not available. Further studies are in progress using leukemia cell line models to identify the role of these genes either individually or in combination in regulating WBCs in the peripheral circulation, which may aid in the development of prognostic factors or identify cellular pathways to target with new therapies.
Author Contributions
- Conceptualization: JWT.
- Formal analysis: HE MR AAD JTC RC ACX RT MN LHM YG JWT.
- Investigation: HE MR AAD.
- Writing - original draft: HE MR AAD JTC RC ACX RT MN LHM YG JWT.
- Writing - review & editing: HE MR AAD JTC RC ACX RT MN LHM YG JWT.
References
- 1. Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent cancer statistics, 2014. CA: a cancer journal for clinicians. 2014;64(2):83–103. pmid:24488779.
- 2. Teachey DT, Hunger SP. Predicting relapse risk in childhood acute lymphoblastic leukaemia. British journal of haematology. 2013;162(5):606–20. pmid:23808872.
- 3. Smith M, Arthur D, Camitta B, Carroll AJ, Crist W, Gaynon P, et al. Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 1996;14(1):18–24. pmid:8558195.
- 4. Xie C, Drenberg C, Edwards H, Caldwell JT, Chen W, Inaba H, et al. Panobinostat enhances cytarabine and daunorubicin sensitivities in AML cells through suppressing the expression of BRCA1, CHK1, and Rad51. PloS one. 2013;8(11):e79106. pmid:24244429; PubMed Central PMCID: PMC3823972.
- 5. Liu F, Rehmani I, Esaki S, Fu R, Chen L, de Serrano V, et al. Pirin is an iron-dependent redox regulator of NF-kappaB. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(24):9722–7. Epub 2013/05/30. pmid:23716661; PubMed Central PMCID: PMC3683729.
- 6. Licciulli S, Cambiaghi V, Scafetta G, Gruszka AM, Alcalay M. Pirin downregulation is a feature of AML and leads to impairment of terminal myeloid differentiation. Leukemia. 2010;24(2):429–37. Epub 2009/12/17. pmid:20010624.
- 7. Cherubini A, Hofmann G, Pillozzi S, Guasti L, Crociani O, Cilia E, et al. Human ether-a-go-go-related gene 1 channels are physically linked to beta1 integrins and modulate adhesion-dependent signaling. Molecular biology of the cell. 2005;16(6):2972–83. Epub 2005/04/01. pmid:15800067; PubMed Central PMCID: PMC1142440.
- 8. Arcangeli A. Expression and role of hERG channels in cancer cells. Novartis Foundation symposium. 2005;266:225–32; discussion 32–4. Epub 2005/07/30. pmid:16050271.
- 9. Schrecengost RS, Riggins RB, Thomas KS, Guerrero MS, Bouton AH. Breast Cancer Antiestrogen Resistance-3 Expression Regulates Breast Cancer Cell Migration through Promotion of p130Cas Membrane Localization and Membrane Ruffling. Cancer Research. 2007;67(13):6174–82. pmid:17616674
- 10. Sims JE, Smith DE. The IL-1 family: regulators of immunity. Nature reviews Immunology. 2010;10(2):89–102. pmid:20081871.
- 11. Tian L, McClafferty H, Knaus HG, Ruth P, Shipston MJ. Distinct acyl protein transferases and thioesterases control surface expression of calcium-activated potassium channels. The Journal of biological chemistry. 2012;287(18):14718–25. Epub 2012/03/09. pmid:22399288; PubMed Central PMCID: PMC3340283.