JAK2V617F and MPLW515L/K are myeloproliferative disorder (MPD)-associated mutations. We genotyped 552 individual hematopoietic colonies obtained by CD34+ cell culture from 16 affected patients (13 JAK2V617F and 3 MPLW515L/K) to determine (a) the proportion of colonies harboring a particular mutation in the presence or absence of cytokines, (b) the lineage distribution of endogenous colonies for each mutation, and (c) the differences (if any) in the pattern of mutation among the various MPDs, as established by genotyping of individual colonies. Genotyping analysis revealed cohabitation of mutation-negative and mutation-positive endogenous colonies in polycythemia vera as well as other MPDs. Culture of progenitor cells harboring MPLW515L/K yielded virtually no endogenous erythroid colonies in contrast to JAK2V617F-harboring progenitor cells. The mutation pattern (i.e., relative distribution of homozygous, heterozygous, or wild-type colonies) was not a distinguishing feature among the MPDs, and MPLW515 mutations were detected in B and/or T lymphocytes in all three patients tested. These observations suggest that clonal myelopoiesis antedates acquisition of JAK2V617F or MPLW515L/K mutations and that the latter is acquired in a lympho-myeloid progenitor cell.
Disclosure of potential conflicts of interest is found at the end of this article.
The recently described JAK2V617F  and MPLW515L  somatic gain-of-function mutations are thought to play a key role in the pathogenesis of BCR-ABL-negative myeloproliferative disorders (MPD). JAK2V617F is found in virtually all polycythemia vera (PV) cases and roughly 50% of primary myelofibrosis (PMF) or essential thrombocythemia (ET) cases . In contrast, MPLW515 mutations (MPLW515L and MPLW515K) occur less frequently and appear to be restricted to PMF (5%) or ET (1%) . JAK2V617F occurs at the hematopoietic stem cell level , and its expression in vivo in a murine bone marrow transplant assay results in a phenotype resembling PV [6, , –9], in contrast to MPLW515L expression in a similar assay, which results in a PMF-like phenotype .
Indirect evidence, based on studies of mutation prevalence within clonal granulocytes [3, 10, 11] as well as familial MPD studies [12, –14], suggests that JAK2V617F may be acquired as a secondary genetic change, especially in ET and PMF. Other studies, however, suggest through erythropoietin-independent (endogenous) erythroid colony (EEC) analysis that JAK2V617F is a prerequisite for EEC growth and that progenitors homozygous for JAK2V617F occur in virtually all patients with PV but not ET [15, 16]. Similar information regarding the prevalence and pattern of MPLW515 mutations in hematopoietic colonies is presently not available. In the current study, we applied mutation analysis to single hematopoietic colonies derived from MPD patients harboring either JAK2V617F or MPLW515 mutations, in the context of both endogenous- and cytokine-supported cultures, in order to address these issues.
Materials and Methods
Patient Accrual and Sample Collection
The current study was approved by our institutional review board. All patients provided verbal and written informed consent, and research was carried out according to the principles of the Declaration of Helsinki. Patients were prospectively accrued to the study if found to carry either JAK2V617F or MPLW515 mutations, and diseases were classified according to World Health Organization criteria .
Peripheral blood-derived CD34+ cells and T and B lymphocytes were purified as previously described . Enriched CD34+ cells (≥95% pure) were plated in MethoCult medium (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com). Clonogenic assays were set up in duplicate in the presence of serum, either with (catalog number H4435) or without (catalog number H4230) cytokines, per manufacturer's guidelines. We plated 1 × 103 and 2–2.5 × 103 CD34+ cells per plate for cytokine-supported and cytokine-independent colony growth for MPD patients, respectively. For healthy controls, we plated 5 × 103 CD34+ cells per plate with and without added cytokines. The plates were incubated at 37°C in a 5% carbon dioxide and 95% air mixture for up to 14 days without (endogenous colonies) or up to 10 days with cytokines, and colonies were scored using standard morphological criteria. Per convention, “endogenous” colony assays were considered positive when at least one colony (i.e., colony forming unit-erythroid, burst forming unit-erythroid, or colony forming unit-granulocyte/monocyte) was detected in the absence of cytokines . Individual well separated colonies were harvested for DNA sequencing as previously described .
Immunostaining and Fluorescence In Situ Hybridization Analysis
Cytopreps were made from pooled endogenous colonies and fixed in 4° acetone for 5 minutes. Antibodies used were α-CD14 and α-CD235a (both from DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com) or α-CD71 and α-CD38 (both from Novocastra Ltd., Newcastle upon Tyne, U.K., http://www.novocastra.co.uk). The slides were incubated in phosphate-buffered saline (PBS)/EDTA for 1.5 minutes then rinsed with tap water for 2 minutes. Staining was continued on the DAKO Autostainer Plus instrument (DakoCytomation). All antibodies were diluted (1:20) in Antibody Diluent (DakoCytomation) and slides processed per the labeled streptavidin-biotin (LSAB) 2 protocol; briefly, slides were sequentially rinsed in PBS and water, blocked in 3% hydrogen peroxide plus 0.1% sodium azide for 10 minutes, and rinsed again with PBS and water. Slides were then incubated in Protein Block (DakoCytomation) for 10 minutes, pulse air-dried, and incubated with primary antibody for 30 minutes, then rinsed in PBS and incubated with secondary antibody for 10 minutes, followed by 10 minutes with a peroxidase-based visualization kit (LSAB 2 HRP-Linked System [DakoCytomation]). After one more rinse in PBS and water, the slides were stained with substrate Vector Nova Red Chromagen (DakoCytomation) twice for 5 minutes each. Finally, the slides were removed from the Autostainer, rinsed in water, and counterstained in hematoxylin for 1 minute. Cytoprep slides were prepared from pooled EEC cells, and fluorescence in situ hybridization (FISH) analysis was performed with relevant probes as previously described .
Genotyping of Mutant Alleles
Individual hematopoietic colonies were genotyped by DNA sequencing as previously described for JAK2V617F  and MPLW515 . For lymphocyte genotyping, T and B cells were purified from peripheral blood by magnetic cell sorting, and DNA was isolated for sequencing.
We initially ascertained whether growth of endogenous colonies from CD34+ cells was MPD-specific. We plated peripheral blood mononuclear cells (PBMCs) as well as CD34+ cells from two healthy controls and 2 MPD patients in the presence or absence of cytokines and/or serum (Table 1). Growth of endogenous colonies was observed to be restricted to MPD patients; no colonies were obtained from healthy controls, despite plating 2–2.5-fold more CD34+ cells from these individuals as compared with the MPD patients.
Table Table 1.. Comparison of colony growth from PBMCs and CD34+ cells under various conditions of cytokine and/or serum support
Since EEC derived from CD34+ cells may be relatively small, we confirmed the lineage identity of these colonies by immunostaining. We pooled cells from several colonies identified as EEC by morphology—virtually all cells stained strongly positive for the erythroid-specific markers Glycophorin A (CD235a) or Transferrin receptor (CD71) but not for the myeloid marker CD14 (supplemental online Fig. 1). In contrast, cells from similarly pooled myeloid colonies stained positive for CD14 but not for the erythroid markers. We excluded possible growth of contaminating lymphocytes under these culture conditions by immunostaining cells for the lymphoid-marker CD38—as expected, cells from pooled EEC and myeloid colonies did not stain for CD38 (supplemental online Fig. 1).
Since both MPD patients were known to harbor specific cytogenetic abnormalities in the bone marrow, we used FISH to interrogate cells from pooled EEC for these abnormalities. For the PV patient, 80% of cells harbored the del5q cytogenetic abnormality (Fig. 1). Similarly, a majority (62%) of cells from the post-PV myelofibrosis patient harbored the expected chromosome 9 abnormalities (44% and 18% had trisomy 9 and der(1;9), respectively). These data confirm that most EEC derived from CD34+ cells of MPD patients are derived from the abnormal MPD clone.
We subsequently studied 16 additional MPD patients for hematopoietic colony growth from CD34+ cells—552 individual erythroid or myeloid colonies were genotyped for presence of JAK2V617F (13 patients) or MPLW515 mutations (three patients; two with MPLW515K and one with MPLW515L) (Table 2). In 10 of the 16 patients, including seven with JAK2V617F and three with MPLW515 mutations, genotyping of endogenous colonies was performed (Table 2). In the remaining six patients, endogenous colonies were not obtained, thus precluding a similar analysis. In the former group, although most EEC harbored JAK2V617F, we found a minor population of mutation-negative EEC to coexist with mutation-positive colonies for three of seven PV or ET patients (cases 1, 6, and 11) (Table 2). When CD34+ cells harboring MPLW515 mutations were plated, virtually no EEC growth was observed, in contrast to JAK2V617F (data not shown). The majority of endogenous myeloid colonies was, however, found to carry MPLW515L/K mutations in two of the three PMF patients (Table 2).
Table Table 2.. Results of single colony genotyping for JAK2V617F and MPLW515 mutations
JAK2V617F-homozygous colonies were identified in six of nine PV patients (cases 1–3 and 6–8), in both of 2 ET patients (cases 11 and 12), and in the lone PMF patient (case 10) (Table 2). Three PV patients (cases 4, 5, and 9), however, exhibited only JAK2V617F-heterozygous colonies. Colonies homozygous for a MPLW515 mutation were found in two of three PMF patients along with coexisting heterozygous colonies (cases 14 and 15). Thus, it appears that, at least at the individual colony level, the mutation pattern (i.e., relative distribution of homozygous, heterozygous, or wild-type colonies) does not distinguish among PV, ET, or PMF.
Lastly, we screened for presence of MPLW515 mutant alleles within both B and T lymphocytes (cases 15 and 16) or T lymphocytes alone (case 14) isolated from peripheral blood. Lymphocyte subsets were 90%–94% pure (supplemental online Fig. 2). We found MPLW515K within both B and T lymphocytes of case 16 (mutant allele ≥ 50%; ≫CD34+ cells) and in T cells of case 14 (mutant allele ∼50%; ≅CD34+ cells) (Fig. 2). Furthermore, the MPLW515L allele appeared to be present within B lymphocytes but not in T lymphocytes of case 15.
We used CD34+ progenitor cells as opposed to PBMCs for the colony assays for several reasons. First, colonies obtained from CD34+ cells are more likely to reflect truly endogenous growth, because EEC obtained from PBMCs may be partially dependent on the presence of contaminating cytokine-secreting cells (i.e., endothelial cells, stromal cells, macrophages, lymphocytes, etc.) . Second, individual colonies can be picked in the relative absence of background contaminating cells, which is essential for the accurate genotyping of individual hematopoietic colonies. Since the procedure for growing endogenous colonies from CD34+ cells is not standardized among laboratories, we initially demonstrated that growth of such colonies was relatively MPD-specific and that most EEC were derived from the abnormal MPD clone. Analysis of CD34+ cell-derived colonies may not, however, be possible in all MPD patients—some exhibit scant or absent endogenous colony growth, underscoring that the assay may not reflect the in vivo biological characteristics of every MPD patient.
The coexistence of mutation-negative and mutation-positive endogenous colonies represents important additional evidence for the acquisition of JAK2V617F and MPLW515 mutations as a secondary event within an abnormal clone that already harbors an unknown mutation causing or contributing to the MPD. Indirect evidence for this model stems from correlation between granulocyte clonality as determined by X-chromosome inactivation pattern analysis and JAK2V617F mutational frequency in informative female patients [3, 10, 11] as well as from studies of familial MPD [12, –14]. Additional support for this model comes from the recent identification of alternative disease-promoting mutations in JAK2V617F-negative cases (e.g., JAK2 exon 12 mutations, MPLW515L/K) . Other activating alleles of JAK kinases have also recently been described in acute megakaryoblastic leukemia cell lines, namely JAK2T875N  and JAK3A572V . Although these particular alleles have not been detected in patients as yet, they illustrate the potential for additional activating alleles involving JAK family members. Furthermore, PBMC-derived EEC that carry only wild-type JAK2 have recently been identified in PV patients by quantitative allele-specific PCR, which further supports the aforementioned model . In contrast, a single study found at least one JAK2V617F allele in every EEC examined from PV patients . A possible explanation for the discrepant results in this latter study may relate to methodological issues, including the particular assay used for genotyping. Another difference between the current study and that by Scott et al. , which is unrelated to methodology, is our observation that the mutation pattern at the single colony level (i.e., relative distribution of homozygous, heterozygous, or wild-type colonies) does not distinguish among PV, ET, or PMF.
The virtual absence of EEC growth when plating CD34+ cells from patients with MPLW515L/K mutations suggests that the circulating CD34+ cell pool in these patients has a greater proportion of myeloid-committed progenitors as compared with patients harboring JAK2V617F. The observation that a greater proportion of EEC (relative to cytokine-supported erythroid colonies) harbors JAK2V617F suggests that exogenously added erythropoietin at least partially overcomes the intrinsic bias conferred by JAK2V617F toward erythroid differentiation.
Finally, we were able to demonstrate the presence of MPL515 mutations within lymphocyte subsets of MPD patients by DNA sequencing (i.e., nonquantitatively). The virtually uniform occurrence of the mutant allele in the lymphocyte compartment, in contrast to JAK2V617F [1, 13, 15, 18, 27, 28], suggests that MPL515 mutations may be acquired in an ontogenically more primitive progenitor cell than JAK2V617F, although this needs to be confirmed by additional studies.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
This work was partially supported by a grant from the Myeloproliferative Disorders Foundation, Chicago. All authors approved the final draft of the paper. A.P. and A.T. wrote the paper. A.P., D.G.G., and A.T. participated in conception and design of the study. A.P., T.L.L., C.F., D.G.G., R.P.K., and A.T. conducted the experiments and/or analyzed the data. R.A.M. and W.J.H. contributed vital reagents and helped collect data.