Dlk1 in Normal and Abnormal Hematopoiesis

S Sakajiri1, J O’Kelly1, D Yin1, CW Miller1, WK Hofmann2, K Oshimi3, L-Y Shih4, K-H Kim5, HS Sul5, CH Jensen6, B Teisner6, N Kawamata1 and HP Koefﬂer1

1Division of Hematology/Oncology, Cedars-Sinai Medical Center, UCLA, Los Angeles, CA, USA; 2Department of Hematology/ Oncology, Johann-Wolfgang Goethe University Frankfurt/Main, Frankfurt/Main, Germany; 3Division of Hematology, Juntendo University School of Medicine, Tokyo, Japan; 4Division of Hematology/Oncology, Chang Gung Memorial Hospital and Chang Gung Univeristy, Taiwan; 5Department of Nutritional Sciences and Toxicology, University of California, Berkeley, CA, USA; and 6Department of Immunology and Microbiology, University of Southern Denmark, Odense, Denmark

Dlk1 (Pref-1) is a transmembrane and secreted protein, which is Dlk1 is highly expressed in neuroblastoma and small-cell lung a member of the epidermal growth factor-like family, homo- carcinoma and has been suggested to play an important role in logous to Notch/Delta/Serrate. We have found by real-time RT- 1 PCR that Dlk1 mRNA levels were high in CD34 þ cells in 10 of tumorigenesis of these cells. Neuroblastoma cell lines treated 12 MDS samples compared with CD34 þ cells from 11 normals. with agents that induced differentiation towards the chromafﬁn Also, Dlk1 mRNA was elevated in mononuclear, low density lineage increased their expression of Dlk1, while compounds bone marrow cells from 11/38 MDS patients, 5/11 AML M6 and that induced differentiation towards the neural phenotype 2/4 AML M7 samples. Furthermore, 5/6 erythroleukemia and 2/2 decreased Dlk1 levels.1 megakaryocytic leukemia cell lines highly expressed Dlk1 The role of Dlk1 in hematopoiesis has not been clearly mRNA. Levels of Dlk1 mRNA markedly increased during deﬁned.5,6 Stromal cells of the bone marrow can produce Dlk1, megakaryocytic differentiation of both CMK megakaryoblasts 5 as well as normal CD34 þ hematopoietic stem cells. High serum which may promote hematopoiesis. Soluble Dlk1-IgG Fc levels of Dlk1 occurred in RA (4/10) and essential thrombo- chimeric protein inhibited the clonal growth of lineage-marker cythemia (2/10) patients. Functional studies showed that forced negative (Lin-) bone marrow cells stimulated by GM-CSF, G-CSF, expression of Dlk1 enhanced proliferation of K562 cells or M-CSF in the presence of stem cell factor (SCF).6 It may be growing in 1% fetal bovine serum. Analysis of hematopoiesis involved in stromal cell-pre-B cell interactions.5 Also, T cells in of Dlk1 knockout mice suggested that Dlk1 contributed to 6 granulocyte, megakaryocyte and B-cell clonogenic growth and the developing thymus express Dlk1 on their cell surface; and was needed for generation of splenic B-cells. In summary, Dlk1 in fetal thymus organ cultures, thymocyte cellularity is increased 6 is overexpressed in selected samples of MDS (especially RA by exogenous dimeric Dlk1 fusion proteins and it is expressed and RAEB) and AML (particularly M6, M7), and it appears to be at higher levels in the thymus during fetal development than in associated with normal development of megakaryocytes and B adult mice.7 Microarray data by others and ourselves found that cells. Dlk1 mRNA was markedly upregulated in CD34 þ hematopietic Leukemia (2005) 19, 1404–1410. doi:10.1038/sj.leu.2403832; marrow stem cells from patients with low-risk MDS compared published online 16 June 2005 8,9 Keywords: MDS; megakaryocyte; B-cell development; with normal individuals. The present studies show that Dlk1 erythroleukemia; real-time PCR is dysregulated in selected cases of MDS and AML, and Dlk1 contributes to normal hematopoiesis and B lymphocyte development. Introduction

Dlk1 (Pref-1, FA1) is a transmembrane and secreted protein.1 It is a member of the epidermal growth factor-like family with homologies to Notch/Delta/Serrate, containing a signal peptide, Materials and methods followed by six epidermal growth factor-like repeats, a 1 transmembrane domain, and a short intracellular tail. Unlike Cell culture and chemical reagents other delta-family members, Dlk1 lacks the DSL motif, which may be crucial for the interaction with the Notch family of 1 Cell lines generously provided include NB4 (acute promyelo- molecules, suggesting that Dlk1 may exert its activity cytic leukemia (AML-M3) cell line from Dr Lanotte (Hopital independent of the Notch receptors. Dlk1 is present in animals 2 Saint-Louis, Paris, France)); KCL22 (chronic myelogenous from birds to mammals and is an imprinted gene. The protein is leukemia cell line, Dr I Miyoshi (Kochi University, Kochi, expressed in a variety of fetal and selected adult tissues and is Japan)); CMK (megakaryocytic leukemia, Dr T Sato (Chiba thought to participate in embryonic growth,1 hematopoiesis,1 3 University, Chiba, Japan)); HEL, OCIM-I and HEL-R (erythroid and wound healing. It is downregulated during adipocyte leukemia, Dr T Papayannopoulou (University of Washington, differentiation; addition of recombinant soluble Dlk1 inhibits Seattle, WA, USA)); and TF-1 (erythroid leukemia, Dr K adipocyte differentiation of 3T3-L1 cells1 and antisense Dlk1 4 Kitamura (Tokyo University, Tokyo, Japan)). KG-1 (early AML increases adipocyte differentiation of Balb 3T3. In the grand- cell line) was established by our group. All other cells were ular epithelial cells of the developing pancreas, this protein 1 obtained from the American Type Culture Collection (Rockville, decreases in an age-dependent manner. MD, USA) and cultured as described. K562 cells were treated with either 40 mM hemin (Sigma, St Louis, MO, USA) for 48 h to Correspondence: Dr N Kawamata, Division of Hematology/Oncology, induce erythroid differentiation as monitored by hemoglobin Cedars-Sinai Medical Center, UCLA School of Medicine, 8700 Beverly 10 concentrations as described or 50 nM 12-O-tetradecanoyl- Blvd., Los Angeles, CA 90048, USA; Fax: þ 1 310 423 0443; E-mail: [email protected] phorbol-acetate (TPA) (Sigma, St Louis, MO, USA) for p5 days Received 20 October 2004; accepted 15 April 2005; published online to induce megakaryocytic differentiation, monitored by expres- 16 June 2005 sion levels of Integrin b3.11 Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1405 Brefeldin A and concanamycin A were purchased from Sigma previously described.8 The expression level of GAPDH, as (St Louis, MO, USA). K562 cells (1 Â 106 cells/ml) were cultured determined by GeneChip assay, was required to be greater than with either drug at the indicated concentrations for 4 h. 5000 (raw data) and was measured as ‘present’ (Affymetrix Call) in all of the samples. The experiments were performed in triplicates for each of the time points. Patient and normal samples

Samples were from 60 newly diagnosed cases of MDS and 31 Cell transfection and Western blot AML patients. Normal bone marrow CD34 þ cells were obtained from 11 healthy donors. CD34 þ cells from MDS and The expression vector pcDNA-Dlk1 was constructed by placing normal bone marrow were obtained from the mononuclear cell full-length human Dlk1 cDNA into the pcDNA3.1 vector fraction (Ficoll density separation) followed by immunomag- (Invitrogen). The constructs were transfected into K562 cells netic bead selection with monoclonal murine antihuman CD34 by electroporation and transfectants were selected for their antibodies using the Auto MACs automated separation system G418 resistance (800 mg/ml). Dlk1 cDNA sequence was (Miltenyi Biotec, Mo¨nchengladbach, Germany). Yield and subcloned into the pBI vector (Clontech Laboratories, Palo Alto, purity of the positively selected CD34 þ cells were evaluated CA, USA) to generate pBI-Dlk1. Empty vector or pBI-Dlk1 were by ﬂow cytometry (FACScan) (Becton Dickinson, Heidelberg, transfected with pTK-Hyg plasmids into Tet-Off K562 cells. Germany). Written informed consent was obtained in keeping Transfectants were selected for their resistance to G418 (800 mg/ with institutional policies. ml) and hygromycin (200 mg/ml). The Dlk1-expressing clones AML mononuclear cells were separated using gradient were detected by Western blot analysis. Western blot analysis centrifugation with Ficoll-Paque (Pharmacia, Uppsala, Sweden). was performed as previously described.14 Anti-Dlk1 polyclonal At the time of diagnosis, the leukemic cells generally antibody (C-19) (Santa-Cruz, Santa-Cruz, CA, USA) and anti- represented 60–99% of the mononuclear cell fraction. GAPDH antibody (Reseach Diagnostic, Flanders, NJ, USA) were used. Real-time RT-PCR assay Cell proliferation in liquid culture and clonogenic Total RNA was isolated from cell lines and clinical samples assays using Trizol reagent (Gibco, BRL) according to the standard protocol. Total RNA (1 mg) was processed directly to cDNA by Viable cells were counted using the trypan blue dye exclusion reverse transcription with Superscript II (Life Technologies, Inc.) method beginning in 24-well plates at 1 Â 105 cells/well; and according to the manufacturer’s protocol in a total volume of because FBS can contain Dlk1,15 experiments were performed 20 ml and used for real-time RT-PCR, the details are provided in with both 10 and 1% FBS. Each experiment was performed in Supplementary Information. triplicate, and results represent the mean7s.d. of three Level of mRNA of each gene was evaluated as a ratio to the experiments. level of ribosomal 18S RNA to standardize the quantity of Dlk1 knockout (KO) mice16 as well as age 6–8-weeks-old cDNAs of each sample. Furthermore, levels of Dlk1 mRNA from control mice were analyzed. Peripheral blood was obtained by the various samples were quantiﬁed as relative values to those heart puncture. Mononuclear cells from both bone marrow and present in K562 cells; the relative value of Dlk1 transcripts spleens were separated by Ficoll Hypaque density centrifugation in K562 cells as measured by real-time RT-PCR, was regarded (Amersham Pharmacia, Uppsala, Sweden) and studied for their as 100. myeloid, erythroid, and megakaryocyte clonogenic growth as recommended by Stem Cell Technologies Inc. (Vancouver, CA, USA) and as previously reported.5 Details of the cell culture ELISA experiments are provided in Supplementary information.

Using prospectively collected serum samples from 121 normal individuals, 16 MDS, four myelofibrosis, 10 polycythemia rubra Flow cytometric analysis vera, and 10 essential thrombocythemia patients, serum Dlk1 was quantified using the sandwich ELISA technique based on Studies were performed as previously described using B220/ polyclonal anti-FA1 (anti-Dlk1) antibodies purified by immuno- CD45R (RA3-6B2), CD43 (Ly-48), CD4 (L3T4), CD8a (Ly-2), specific affinity chromatography as previously described.12 The CD19, murine IgM (R6–60.2), and CD34 (RAM34) (Pharmingen) normal values for serum Dlk1 are based on the concentrations in monoclonal antibodies conjugated with either fluorescein the serum samples from healthy donors. isothiocyanate (FITC), phycoerythrin (PE), or peridinin chlor- ophyll protein (PerCP). Gene expression analysis by oligonucleotide microarrays Results

CD34 þ bone marrow cells from healthy individuals were MDS and AML cells overexpress Dlk1 cultured in vitro with TPO to induce megakaryocytic differentiation as previously described.13 The cells were harvested on Dlk1 mRNA expression in CD34 þ hematopoietic stem cells was days 0, 4, 7, and 11 for expression proﬁling. Total RNA was examined in individuals with MDS using real-time RT-PCR. extracted and used for RNA expression analysis using HG- Level of Dlk1 mRNA in K562 cells was high and was used as an U133A microarray (Affymetrix Inc.) and analyzed by the arbitrary standard (100%) for quantiﬁcation of Dlk1 mRNA in Microarray Analysis Suite 5.0 (Affymetrix, Inc.) and GeneSpring other samples. Dlk1 mRNA was nearly undetectable in CD34 þ software version 4.2 (Silicon Genetics, San Carlos, CA, USA) as cells from 11 normal individuals, but levels were high in six of

Leukemia Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1406 seven (85%) RA and four of five (80%) RAEB samples was confirmed by treatment of K562 cells with either brefeldin A (Figure 1a). Also, expression of Dlk1 mRNA in light density, or coneanamycin A, which inhibits protein transportation mononuclear bone marrow cells was nearly undetectable in the to the outside of the cells. Levels of Dlk1 protein in K562 eleven normal samples, but was easily measurable in six of 21 cells markedly increased as shown by Western blot (data RA (28%), 0/4 RARS, 2/13 RAEB (15%), 3/4 RAEB-T (75%), 0/2 not shown). CMMoL, 2/4 overt leukemias from MDS (50%), 1/4 AML To examine the effect of Dlk1 on cellular proliferation, we M1(25%), 0/2 AML M2, 1/1 AML M3 (100%), 0/4 AML M4, generated K562 cells stably expressing Dlk1. Three stable clones 0/1 AML M4Eo, 0/2 AML M5, 5/11 AML M6 (45%), 2/4 AML M7 expressing Dlk1 and three control clones (Neo) were isolated (50%), and 1/1(100%) AML M2 transformed from polycythemia (Figure 3a). Each grew similarly in the presence of 10% FBS, but vera (PV) (Figure 1b, these cases are different from those the Dlk1 expressing clones grew faster than controls in a low analyzed in Figure 1a). concentration of FBS (1%) (Figure 3a). Further, to confirm the proliferative activity of Dlk1 in K562 cells, we generated a tetracycline-inducible clone, pBI Tet Dlk1-cl5. In the absence of Dlk1 mRNA was expressed in erythroid and doxycycline, expression of Dlk1 was induced; and withdrawal megakaryocytic cell lines and levels increased during of this antibiotic from the culture media, shut off the expression normal megakaryocyte differentiation of Dlk1 (Figure 3b). Induced expression of Dlk1 in K562 cells again enhanced their proliferation in culture media containing Human erythroid leukemia cell lines (K562, HEL, HEL-R, TF-1) 1% FBS (Figure 3b). To elucidate the mechanisms of the and megakaryoblastic leukemia cell lines (CMK and Meg 01) proliferative activity of Dlk1, we analyzed the levels of cell had high Dlk1 mRNA expression compared to nearly undetect- cycle regulatory proteins in these cell lines. However, no able levels in myeloid leukemia lines except KG-1 cells (CD34 þ difference was detected in levels of Rb, cyclin D1, and p21WAF1 cells established from an individual with erythroleukemia) proteins between stable Dlk1 expressing clones of K562 and (Figure 1c). control clones (data not shown). Dlk1 levels markedly decreased (98%) during erythroid differentiation of K562 cells (Figure 1d). In contrast, Dlk1 expression prominently increased (E40-fold) during megakar- Hematopoiesis in Dlk1 KO mice yocyte differentiation of CMK cells (Figure 1e). Also, as normal CD34 þ hematopoietic stem cells differentiated to megakaryo- In order to understand further the function of Dlk1, hematopoi- cytes, their levels of Dlk1 mRNA increased (Figure 1f). esis was studied in Dlk1 KO mice. Analysis of peripheral blood of Dlk1 KO mice showed that they had hematocrits (38% vs 44%, KO vs wild type (WT)), white blood cell counts (7.3 vs Serum levels of Dlk1 are high in some RA patients 8.4 Â 103/ml, KO vs WT) and platelet counts (604 vs 805 Â 103/ ml, KO vs WT); however, P-values were not significantly Since Dlk1 is both a membrane and secreted protein, we different between Dlk1 KO and WT mice. Overall bone marrow measured levels of Dlk1 protein in the sera of patients cellularity, differential counts, percentage of red cell precursors with MDS, myeloproliferative disorders and normal indivi- (TER119 þ ), and common hematopoietic stem cells (CD34 þ ) duals (Figure 2). Whereas levels of Dlk1 protein in serum from were similar in the Dlk1 KO and WT mice (data on request). almost all normal individuals were less than p50 ng/ml, high The progenitor cells in the marrows of Dlk1 KO and WT mice levels were detected in four of 10 patients with RA (40%) were assessed by a variety of in vitro clonogenic assays using and two of 10 individuals with essential thrombocythemia growth factors specific for the various cell lineages (see (20%) (Figure 2). Materials and methods). Dlk1 KO mice had significantly more megakaryocyte colony-forming cell (CFU-MK) colonies (mean 53.877.7) than the WT mice (mean 43.273.3) (Po0.001) Dlk1 is a secreted protein that causes cellular (Figure 4a) significantly more granulocyte colony-forming cell proliferation of K562 cells (CFU-G) colonies (mean 32.5710.7) than WT mice (mean 17.575.0) when the marrows were grown with G-CSF alone Although K562 cells prominently expressed Dlk1 mRNA (Po0.01) (Figure 4b), and significantly fewer pre-B lymphocyte (Figure 1c), low expression of Dlk1 protein was detected colonies (mean 8.675.1) than WT mice (mean 20.776.1) by Western blotting in these cells (data not shown). We (Po0.01) (Figure 4c). The number of bone marrow granulocyte- hypothesized that most of the Dlk1 protein generated in macrophage precursors (CFU-GM and CFU-M), and erythroid these hematopoietic cells was secreted; and therefore, it was precursors (BFU-E and CFU-E) were comparable between the not detected in the cellular lysates by Western blot. This Dlk1 KO and WT mice (data not shown).

Figure 1 Cellular and serum levels of Dlk1 in MDS, AML, myeloproliferative disorders, and myeloid leukemia cell lines. (a) Quantitation of Dlk1 mRNA in CD34 þ hematopietic stem cells from MDS (seven RA and five RAEB patients) and 11 normal individuals measured by real-time RT- PCR. Levels of Dlk1 are indicated relative to those of K562 cells (relative value of K562, 100). (b) Quantitation of Dlk1 mRNA in the low density, mononuclear bone marrow cells from 60 MDS (RA, RARS, RAEB, RAEB-T, CMMoL), 31 AML (FAB M1-M7), and 11 healthy volunteers. Levels of Dlk1 are indicated relative to those of K562 cells (regarded as 100). (c) Quantitation of Dlk1 mRNA in leukemia cell lines. Levels of Dlk1 are indicated relative to those in K562 cells (regarded as 100). (d) Dlk1 mRNA levels during hemin induced differentiation of K562 cells towards erythrocytes. Levels of Dlk1 mRNA (upper panel) and hemoglobin (Hgb) concentration in K562 cell lysates confirming that these cells differentiated towards erythrocytes (lower panel). (Hgb concentration displayed as a % of total cellular protein). (e) Dlk1 mRNA levels as CMK megakaryoblasts differentiated towards megakaryocytes (TPA, 50 nM) (upper panel). Lower panel shows levels of the megakaryocyte-related integrin b3 mRNA in CMK, confirming their differentiation toward megakaryocytes. (f) Expression of Dlk1 during normal megakaryocyte development. CD34 þ bone marrow cells from healthy individuals were stimulated in vitro with thrombopoietin to induce megakaryocyte differentiation. Cells expressing the megakaryocyte cell surface marker, CD61, were isolated and their Dlk1 gene expression was measured by DNA microarray. (values of Dlk1 expressed as a fraction of GAPDH).

Leukemia Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1407 ac 1300

ryoblasts

egaka

A Erythroid M 800

mRN

A 100

400 RN

Dlk1

100

-1

K562 Normal (11) RA(7) RAEB (5) 22

NB4

KG1 HEL

TF-1

eg 01

K562 U937

HL60

CD34

THP-1

HEL-R

KCL

OCIM

Kasumi

Norm b 1200

900

A 400

mRN #2

Dlk1

#1 100

) M3 M2 K562 mia(4) M1(4) M2(2) M4(4)M4EoM5(2) M7(4) RA(21) M6(11) RARS(4) PV RAEB(13) RAEBT(4) Normal (11 CMMoL(2)

Overt leuke

d % e %

K562 (Hemin) 100 CMK (TPA) 100

5 Dlk1 mRNA Dlk1 mRNA Dlk1 Levels 0 12 24 36 48 024 48 96 hrs hrs 04711 % % Megakaryocyte Development (days) K562 (Hemin) 100 CMK (TPA) 5 3 mRNA β

Hemoglobin/Protein Integrin

02448 0489624 hrs hrs

Leukemia Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1408 Effect of recombinant murine (m) Dlk1 protein on clonal mice was dramatic at 6 weeks, no difference in their lymphoctye growth of bone marrow cells from Dlk1 KO mice showed populations was detected between the Dlk1 WT and KO mice a mean 2376.0 (7s.d.) CFU-G colonies in cultures with (data not shown). In contrast, whereas the population of B220 þ no added mDlk1 and 1476.7 CFU-G colonies with mDlk1 sIgM þ mature B cells in the spleens of Dlk1 WT mice was (5 ng/ml) (p ¼ 0.048) (Figure 4d). In contrast, a mean 1471.4 51% at the 16th week, this population was only 33% at the pre-B lymphocyte colonies formed from bone marrows 16th week in the Dlk1 KO mice (Figure 5c). Also, the popula- of Dlk1 KO mice; addition of mDlk1 (5 ng/ml) increased the tion of B220 þ CD43À pre-B lymphocytes increased gradually number of pre-B lymphocyte colonies (2776.3) (Po0.01) (62% at 16 weeks) in WT mice, but this population reached (Figure 4e). only 45% at 16 weeks in Dlk1 KO mice (Figure 5d). No Splenic weights of Dlk1 KO mice were less than those in WT difference was detected in T lymphocyte populations (CD3, mice at p8 weeks after birth (Figure 5a). For example, mean CD4, CD8) between Dlk1 KO and WT mice at any age (data weight of spleens at 6 weeks was 157738 (7s.d.) mg in WT not shown). mice, and 5274.2 mg in Dlk1 KO mice. This difference was also reﬂected in splenic sizes (Figure 5b). Splenic sizes and weights were similar by 16 weeks of age (Figure 5a). Since Dlk1 Discussion KO mice had fewer pre-B lymphocyte colonies compared with WT mice as measured by clonogenic assay of bone marrow In this study, we have found high expression of Dlk1 mRNA in cells, we analyzed B-cell development in the spleens. Although AML and MDS patients. Interestingly, samples from two patients the difference in the size of spleens between Dlk1 WT and KO with RA and RAEB-T (patients #1 and #2, respectively, in

250

200

150

100

Serum Dlk1 (ng/ml) 50

0 RA (10) RAEB (6) MF (4) PV (10) ET (10) Normal (102)

Figure 2 Dlk1 serum levels in MDS, myeloproliferative disorders and normal individuals: Dlk1 levels were determined by ELISA. MF, myeloﬁbrosis; PV, polycythemia vera; and ET, essential thrombocythemia. Numbers in parenthesis indicate the number of samples examined.

a b WT Neo-cl 1 Doxycyclin (-) 1 10 ng/ml

WT Neo-cl 1 Neo-cl 2 Neo-cl 3 Dlk1-cl 17 Dlk1-cl 18 Dlk1-cl 19 Neo-cl 2 Dlk1-cl 5 tet (-) Neo-cl 3 Dlk1 Dlk1 Dlk1-cl 17 Dlk1-cl 5 tet (+) Dlk1-cl 18 Neo-cl 1 tet (-) GAPDH Dlk1-cl 19 GAPDH Neo-cl 1 tet (+)

10% FBS 1% FBS 1% FBS 16 16 20

14 14 18 16 12 12 14 cells/ml) cells/ml)

cells/ml) 10

10 5 5

5 12 8 8 10

6 6 8 6 4 4 Cell number (x10 Cell number (x10 Cell number (x10 4 2 2 2 0 0 0 012345 012345 0 1 2 3 4 Time in Culture (Days) Time in Culture (Days) Time in Culture (Days)

Figure 3 Effect of Dlk1 protein on proliferation of K562 cells. (a) Upper: expression of Dlk1 protein in pcDNA-Dlk1 stably transfected K562 clones (Dlk1 -cl 17, -cl 18, -cl 19), but not in K562 clones expressing the neo gene alone (Neo-cl 1-3). Lower: overexpression of Dlk1 (pcDNA- Dlk1)-stimulated proliferation of K562 clones (cl-17, -18, -19) cultured in low serum conditions (1% FBS). Control Neo-cl 1, -cl 2, -cl 3 grew similarly to wild-type (WT) K562 cells. Panel b: Expression of Dlk1 in tet-off inducible K562 clone. Upper: Dlk1 expression was conﬁrmed by Western blot either without or with doxycycline (1 or 10 ng/ml, 24 h). Lower: overexpression of Dlk1 (Dlk1-cl5 tet(À): tet-off Dlk1 expression vector)-enhanced proliferation of K562 cells under low serum conditions (1% FBS). Controls Dlk1-cl5 tet( þ ); Neo-cl1 tet(À); Neo-cl1 tet( þ ) grew at a slower rate. FBS, fetal bovine serum. tet(+): culture with 1 ng/ml doxycyclin.

Leukemia Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1409 Figure 1b) had very high levels of Dlk1 (490% of the level normal megakaryocytes. Mechanism of overexpression of Dlk1 found in K562 cells) and very low levels of platelets (1.0 and mRNA in MDS and AML is unclear and requires further study. 1.5 Â 104/ml, respectively). Since we have only two cases with Since Dlk1 is an imprinted gene and Dlk1 is transcribed from very high levels of Dlk1, we cannot conclude an association only the paternal allele in normal human,2 loss of imprinting between platelet production and Dlk1 levels, but this possible might occur in selected MDS and AML samples associated with association deserves further attention. Our data suggest that overexpression of this gene. expression of Dlk1 is associated with CD34 þ MDS cells, early Forced expression of Dlk1 in K562 cells enhanced their erythroid and megakaryoblastic AML cells as well as maturing proliferation. High expression of Dlk1 in MDS and AML could provide them with a growth advantage over their normal hematopoietic counterparts. Mechanism by which Dlk1 en- CFU-Mk (IL-3, IL-6, IL-11, TPO) hances cellular proliferation is unclear at this time. a 70 Dlk1 KO mice display growth retardation, obesity, blephar- 60 p<0.01 50 ophimosis, skeletal malformation, and increased serum lipid 16 40 metabolites. Similar symptoms occur in humans with maternal 30 uniparental disomy for chromosome 14 (mUDP14).17 Lack of 20 Dlk1 may be responsible for some of the symptoms observed in colony number 10 humans with mUPD14. Dlk1 is an imprinted gene residing on 0 WT n=4 KO n=5 chromosome 14, and the gene is transcribed only from the paternal allele. Since patients with mUPD14 disease have b 50 CFU-G (G-CSF) disomy of the maternal allele of chromosome 14, Dlk1 will not 45 17 40 be transcribed in these patients. Hematopoietic abnormalities 35 p<0.01 in individuals with mUDP14 have not been emphasized. 30 Detailed analysis of hematopoiesis in the Dlk1 KO mice showed 25 20 enhanced CFU-G and CFU-MK clonal growth, but decreased 15 pre-B lymphocyte colony formation. Addition of the soluble

colony number 10 5 form of Dlk1 to bone marrow cells from Dlk1 KO mice norma- 0 WT n=4 KO n=5 lized the numbers of CFU-G and pre-B lymphocyte colony numbers. Despite impairment of clonogenic growth of several c 30 pre-B lymphocyte CFU (IL-7) lineages, no signiﬁcant difference in the peripheral blood cell 25 counts were detected in the Dlk1 KO mice. A compensating p<0.01 mechanism might allow hematopoetic normality. Perhaps Dlk1 20 plays a more important role in embryonic hematopoiesis and 15 becomes less important with development. By analogy, Notch1 10 conditional KO mice have a normal peripheral blood proﬁle colony number 5 except for the T-cell compartment, but closer scrutiny shows 18 0 hematopoietic abnormalities of these mice. WT n=4 KO n=5 At 6–8 weeks of age, spleens of Dlk1 KO mice were smaller than those in WT mice. However, no abnormalities of CFU-G (G-CSF) lymphocyte contents were noted in their spleens. The cause of d p=0.048 these small spleens might be nutritional, since Dlk1 KO mice 16 30 have nutritional abnormalities and are small at birth. Impair- ment of B-cell development was noted in spleens of Dlk1 KO mice at 16 weeks of age; and as mentioned above, their clonogenic growth of pre-B cells from the bone marrow of Dlk1 was impaired. Our data suggest that Dlk1, especially the soluble

colony number 10 form of Dlk1, may play an important role in B lymphocyte development. These data buttress the experiments suggesting that Dlk1 is important for B-cell development in the presence of IL7 and stromal cells.1 WT n=4 KO n=2 KO+Dlk1 n=2

e pre-B lymphocyte (IL-7) Figure 4 Hematopoietic clonogenic growth of bone marrow cells p<0.01 from Dlk1 KO mice. (a–c) Bone marrow cells collected from WT and Dlk1 KO mice were stimulated with a variety of cytokines in soft gel 30 plates to examine hematopoiesis in these animals. Data represent mean colony numbers7s.d. of triplicate cultures containing bone marrow cells stimulated with appropriate growth factors (details are provided in Supplementary Information). WT: bone marrow cells from Dlk1 þ / þ mice. KO: bone marrow cells from Dlk1À/À mice. n: numbers of mice examined; CFU-MK, megakaryocyte colony-forming cell; CFU-G, granulocyte colony-forming cell. (d, e) Effect of puriﬁed colony number 10 Dlk1 protein on clonogenic growth of hematopoietic cells from Dlk1 WT and KO mice. (d) CFU-G.; (e) pre-B lymphocytes. Mean colony numbers7s.d. in triplicate cultures containing bone marrow cells from Dlk1 KO or WT mice stimulated with G-CSF (10 ng/ml) or IL-7 (10 ng/ ml) either with or without mDlk1 protein (5 ng/ml). WT: Dlk1 Wild- WT n=4 KO n=2 KO+Dlk1 n=2 type mice; KO: Dlk1 deletional mice. n numbers of mice examined.

Leukemia Dlk1 in normal and abnormal hematopoiesis S Sakajiri et al 1410 a References WT 1 Laborda J. The role of the epidermal growth factor-like protein dlk KO 250 in cell differentiation. Histol Histopathol 2000; 15: 119–129. 200 2 Schmidt JV, Matteson PG, Jones BK, Guan XJ, Tilghman SM. The 150 Dlk1 and Gtl2 genes are linked and reciprocally imprinted. Genes Dev 2000; 14: 1997–2002. 100 3 Samulewicz SJ, Seitz A, Clark L, Heber-Katz E. Expression of 50 preadipocyte factor-1(Pref-1), a delta-like protein, in healing

spleen weight (mg) 0 mouse ears. Wound Repair and Regeneration 2002; 10: 215–221. 6816 4 Garcces C, Ruiz-Hidalgo MJ, Bonvini E, Goldstein J, Laborda J. Weeks after birth Adipocyte differentiation is modulated by secreted delta-like (dlk) variants and requires the expression of membrane-associated dlk. b Spleens (6 weeks) Differentiation 1999; 64: 103–114. 5 Moore KA, Pytowski B, Witte L, Hicklin D, Lemischka IR. KO Hematopoietic activity of a stromal cell transmembrane protein containing epidermal growth factor-like repeat motifs. Proc Natl Acad Sci USA 1997; 94: 4011–4016. WT 6 Ohno N, Izawa A, Hattori M, Kageyama R, Sudo T. Dlk inhibits stem cell factor-induced colony formation of murine hematopoietic progenitors: Hes-1-independent effect. Stem Cells 2001; 19: 71–79. 7 Kaneta M, Osawa M, Sudo K, Nakauchi H, Farr AG, Takahama Y. c % mature B (B220+, sIgM+) A role for pref-1 and HES-1 in thymocyte development. J Immunol 2000; 164: 256–264. 8 Hofmann WK, de Vos S, Komor M, Hoelzer D, Wachsman W, + 50 Koeffler HP. Characterization of gene expression of CD34 cells WT from normal and myelodysplastic bone marrow. Blood 2002; 100: KO 3553–3560. 9 Miyazato A, Ueno S, Ohmine K, Ueda M, Yoshida K, Yamashita Y Et al. Identification of myelodysplastic syndrome-specific genes by DNA microarray analysis with purified hematopoietic stem cell 0 fraction. Blood 2001; 98: 422–427. 6816 10 Conscience JF, Miller RA, Henry J, Ruddle FH. Acetylcholinester- Weeks ase, carbonic anhydrase and catalase activity in Friend erythro- leukemic cells, non-erythroid mouse cell lines and their somatic d % pre B (B220+, CD43-) hybrids. Exp Cell Res 1977; 105: 401–412. 11 Shattil SJ, Kashiwagi H, Pampori N. Integrin signaling: the platelet paradigm. Blood 1998; 91: 2645–2657. 50 12 Jensen CH, Krogh TN, Stoving RK, Holmskov U, Teisner B. Fetal WT antigen 1 (FA1), a circulating member of the epidermal growth KO factor (EGF) superfamily: ELISA development, physiology and metabolism in relation to renal function. Clin Chim Acta 1997; 268: 1–20. 13 Hofmann WK, Kalina U, Wagner S, Seipelt G, Ries C, Hoelzer D 0 et al. Characterization of defective megakaryocytic development 6816 in patients with myelodysplastic syndromes. Exp Hematol 1999; Weeks 27: 395–400. Figure 5 Spleen sizes in Dlk1 WT and KO mice, and lymphocyte 14 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning,, 2nd edn, analysis of spleens from Dlk1 WT and KO mice. (a) Splenic weights Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, (mean7s.d. of four samples) at different ages from Dlk1 WT and KO 1989; 3: 18.47–18.76. mice. (b) Spleens from the 6-week-old Dlk1 wild type (WT)and Dlk1 15 Minoshima Y, Taniguchi Y, Tanaka K, Yamada T, Sasaki Y. deletional (KO) mice. (c, d) Splenocytes from Dlk1 WT and KO mice Molecular cloning, expression analysis, promoter characterization, were analyzed by two-color flow cytometry using anti-B220 and anti- and chromosomal localization of the bovine PREF1 gene. Anim sIgM antibodies, or anti-B220 and anti-CD43 antibodies. WT: Dlk1 Genet 2001; 32: 333–339. WT mice; KO: Dlk1 KO mice. 16 Moon YS, Smas CM, Lee K, Villena JA, Kim KH, Yun EJ et al. Mice lacking paternally expressed Pref-1/Dlk1 display growth retardation and accelerated adiposity. Mol Cell Biol 2002; 22: 5585–5592. High expression of Notch1 is frequently detected in AML;19 17 Fokstuen S, Ginsburg C, Zachmann M, Schinzel A. Maternal and as mentioned, Notch1 conditional KO mice have impaired uniparental disomy 14 as a cause of intrauterine growth retardation T-cell development.20 Taken together, the data suggest that and early onset of puberty. J Pediatr 1999; 134: 689–695. 18 Maillard I, He Y, Pear WS. From the yolk sac to the spleen: new Notch1 and Dlk1 enhance cellular proliferation and play a role roles for Notch in regulating hematopoiesis. Immunity 2003; 18: in lineage-specific development of lymphocytes. Recent data 587–589. suggest that Notch1 is also critical for hematopoietic stem cell 19 Tohda S, Nara N. Expression of Notch1 and Jagged1 proteins generation;21 the role of Dlk1 in normal and abnormal in acute myeloid leukemia cells. Leuk Lymphoma 2001; 42: hematopiesis requires additional studies. 467–472. 20 Radtke F, Wilson A, Stark G, Bauer M, van Meerwijk J, MacDonald HR et al. Deficient T cell fate specification in mice with an Supplementary Information induced inactivation of Notch1. Immunity 1999; 10: 547–558. 21 Kumano K, Chiba S, Kunisato A, Sata M, Saito T, Nakagami- Yamaguchi E et al. Notch1 but not Notch2 is essential for Supplementary Information accompanies the paper on the generating hematopoietic stem cells from endothelial cells. Leukemia website (http://www.nature.com/leu). Immunity 2003; 18: 699–711.

Leukemia