The LMO2 oncogene regulates DNA replication in hematopoietic cells

Significance

Understanding how cell cycle and cell differentiation are coordinated during normal hematopoiesis will reveal molecular insights in leukemogenesis. LIM-only 2 (LMO2) is a transcriptional regulator that controls the erythroid lineage via activation of an erythroid-specific gene expression program. Here, we uncover an unexpected function for LMO2 in controlling DNA replication via protein–protein interactions with essential DNA replication enzymes. To our knowledge, this work provides the first evidence for a nontranscriptional function of LMO2 that drives the cell cycle at the expense of differentiation in the erythroid lineage and in thymocytes when misexpressed following genetic alterations. We propose that the nontranscriptional control of DNA replication uncovered here for LMO2 may be a more common function of oncogenic transcription factors than previously appreciated.

Abstract

Oncogenic transcription factors are commonly activated in acute leukemias and subvert normal gene expression networks to reprogram hematopoietic progenitors into preleukemic stem cells, as exemplified by LIM-only 2 (LMO2) in T-cell acute lymphoblastic leukemia (T-ALL). Whether or not these oncoproteins interfere with other DNA-dependent processes is largely unexplored. Here, we show that LMO2 is recruited to DNA replication origins by interaction with three essential replication enzymes: DNA polymerase delta (POLD1), DNA primase (PRIM1), and minichromosome 6 (MCM6). Furthermore, tethering LMO2 to synthetic DNA sequences is sufficient to transform these sequences into origins of replication. We next addressed the importance of LMO2 in erythroid and thymocyte development, two lineages in which cell cycle and differentiation are tightly coordinated. Lowering LMO2 levels in erythroid progenitors delays G1-S progression and arrests erythropoietin-dependent cell growth while favoring terminal differentiation. Conversely, ectopic expression in thymocytes induces DNA replication and drives these cells into cell cycle, causing differentiation blockade. Our results define a novel role for LMO2 in directly promoting DNA synthesis and G1-S progression.

More than 70% of recurring chromosomal translocations in T-cell acute lymphoblastic leukemia (T-ALL) involve transcription factors that are master regulators of cell fate. These oncogenic transcription factors determine the gene signature and leukemic cell types (1). Whether these DNA-bound factors may have additional roles beyond modulating gene expression remains unknown. LMO2, a 17-kDa protein defined by tandem zinc finger domains, is an essential nucleation factor that assembles a multipartite transcriptional regulatory complex on gene regulatory regions via direct interaction with the TAL1/SCL transcription factor, LDB1, and other DNA binding transcription factors (2–4, reviewed in refs. 5, 6). These complexes drive gene expression programs that determine hematopoietic cell fates at critical branchpoints both during embryonic development (7) and in adult hematopoietic stem cells (8, 9). Lmo2 function is essential in highly proliferative erythroid progenitors (10, reviewed in refs. 5, 6). Interestingly, Lmo2 down-regulation is required for terminal erythroid differentiation (11, 12). Because commitment to terminal differentiation is coordinated with growth arrest (13), Lmo2 may have additional molecular functions that impede this critical step marked by growth cessation.

In mouse models of T-ALL, LMO1 or LMO2 collaborates with SCL to inhibit the activity of two basic helix–loop–helix (bHLH) transcription factors that control thymocyte differentiation, E2A/TCF3 and HEB/TCF12, causing differentiation arrest (reviewed in ref. 14). However, this inhibition is not sufficient, per se, for leukemogenesis, because both TAL1 and LYL1 inhibit E proteins but require interaction with LMO1/2 to activate the transcription of a self-renewal gene network in thymocytes (15, 16) and to induce T-ALL (17, 18). Of note, downstream target genes cannot substitute for LMO1/2 to induce T-ALL, suggesting additional functions for LMO1/2.

Together, these studies underscore the dominant oncogenic properties of LMO2, as revealed by recurring retroviral integrations upstream of LMO2 in the gene therapy trial (19, 20) or by recurrent chromosomal rearrangements in T-ALL (21). As a consequence, LMO2 is misexpressed in the T lineage, where it is normally absent. In addition, LMO proteins are frequently deregulated in breast cancers (22) and neuroblastomas (23), pointing to their importance in cell transformation. In particular, in patients who eventually developed T-ALL associated with LMO2 activation after gene therapy, T-cell hyperproliferation was observed early during the preleukemic stage (19). How LMO2 affects erythroid progenitor or T-cell proliferation cannot be inferred from its downstream target genes (12, 24–28).

To understand LMO2 functions, we performed an unbiased screen for LMO2 interaction partners. We show that LMO2 associates with three replication proteins, minichromosome 6 (MCM6), DNA primase (PRIM1), and DNA polymerase delta (POLD1), and that LMO2 influences cell cycle progression and DNA replication in hematopoietic cells, indicating an unexpected function for LMO2.

Results

Identification of New LMO2 Protein–Protein Interactions in Hematopoietic Progenitors.

Lmo2 is expressed in c-Kit⁺ hematopoietic stem and progenitor cells (HSPCs) and in immature prothymocytes, but not at later stages of T-cell differentiation (29). To identify new LMO2 binding proteins in HSPCs, we constructed a cDNA library from purified murine Kit⁺Lin⁻ hematopoietic progenitors for a yeast two-hybrid screen and used LMO2 as bait. In addition to known LMO2-interacting proteins, such as LDB1, and to proteins associated with transcription, we unexpectedly identified interactions with three essential components of prereplication complexes (pre-RCs), namely, MCM6, POLD1, and PRIM1 (30) (Fig. 1A and Table S1). In comparison, a screen performed using GAL4-SCL identified only known interactions (Table S1). LMO2 interaction was specific to these three replication proteins, as confirmed by independent yeast two-hybrid assays with full-length cDNAs (Fig. 1 B and C). In addition, we identified BAZ1A/ACF1, required for DNA replication through heterochromatin (31); SetD8, for replication licensing (32); MYST2/HBO1, controlling MCM loading via ORC1 binding (33); and CCNA2-CDK1, regulating origin firing (34) (Fig. 1A). The top-ranking pathways by gene set enrichment analysis were cell cycle, DNA synthesis, and DNA replication (Fig. 1D), concurring with the view that LMO2 controls these processes via its protein partners. Finally, LMO2 coimmunoprecipitated with PRIM1, MCM6, and MYST2 in mammalian cells, and both LIM domains contributed to this interaction (Fig. 1E), whereas LDB1 binding required mostly LIM1, as expected (4). We conclude that in addition to its association with transcription regulators, LMO2 engages in protein–protein interactions with DNA replication proteins.

Fig. 1.

LMO2 interacts with DNA replication proteins. (A) New LMO2 protein–protein interactions in hematopoietic progenitors revealed by yeast two-hybrid screens using LMO2 as bait (*). (B and C) LMO2 specifically interacts with MCM6, PRIM1, and POLD1 by yeast two-hybrid assay. (D) Gene set enrichment analysis for LMO2 interaction partners. Shown are the top seven Reactome pathways significantly enriched in the LMO2 Y2H screen [false discovery rate (FDR): q < 0.001] according to the Molecular Signatures database v5.0 (PubMed identifier 16199517). (E) Both LIM domains of LMO2 are required for interaction with PRIM1, MCM6, and MYST2. GAL4-LMO2 WT or LIM1/2 mutants were cotransfected with Flag-tagged PRIM1, MCM6, or MYST2. Protein extracts were immunoprecipitated with anti-FLAG Ab, followed by Western blotting. (F) DNA replication proteins coimmunoprecipitate with LMO2 (**). Immunoprecipitation of TF-1 chromatin extracts with Abs against LMO2. Inputs (4%) and immune pellets (IP) were analyzed by IB. (G) SCL does not stably associate with DNA replication proteins. SCL or isotype control Abs were analyzed as in F (*). IB, immunoblotting; IP, immune pellet; SN, supernatant. Data shown are typical of at least two (*) or three (**) independent experiments.

Table S1.

List of proteins identified by yeast two-hybrid screening with Lmo2 or Tal1

Bait	Target (official gene symbol)	Aliases	Identification	Complete gene name	GST or IP	Nuclear
Lmo2	Bub1b	Bubr1	NM_009773	Budding uninhibited by benzimidazoles 1 homolog, beta (Saccharomyces cerevisiae)	+	Yes
Lmo2	Mcm6		NM_008567	Minichromosome maintenance deficient 6 (MIS5 homolog, Schizosaccharomyces pombe) (S. cerevisiae)	+	Yes
Lmo2	Appl2	Dip3b	NM_145220	Adaptor protein, phosphotyrosine interaction, PH domain and Leu zipper containing 2	+	Yes
Lmo2	Flii	Fliih	NM_022009	Flightless I homolog (Drosophila)	+	Yes
Lmo2	Hipk2	Stank	NM_010433	Homeodomain interacting protein kinase 2	+	Yes
Lmo2	Kars	LysRS	NM_053092	Lysyl-tRNA synthetase	+	Yes
Lmo2	Kmt2b	Mll2, Wbp7	NM_029274	Lys (K)-specific methyltransferase 2B	+	Yes
Lmo2	Ldb1	Nli, Clim2	NM_010697	LIM domain binding 1	+	Yes
Lmo2	Lrch4	Sap25	NM_146164	Leu-rich repeats and calponin homology (CH) domain containing 4	+	Yes
Lmo2	Nsun2	Misu	NM_145354	NOL1/NOP2/Sun domain family member 2	+	Yes
Lmo2	Pold1		NM_011131	Polymerase (DNA directed), delta 1, catalytic subunit	+	Yes
Lmo2	Prim1		NM_008921	DNA primase, p49 subunit	+	Yes
Lmo2	Psmc1	Rpt2	NM_008947	Protease (prosome, macropain) 26S subunit, ATPase 1	+	Yes
Lmo2	Psme3	PA28 gamma, REG gamma, Ki	NM_011192	Proteasome (prosome, macropain) activator subunit 3 (PA28 gamma, Ki)	+	Yes
Lmo2	Taf6l	Paf65a	NM_146092	TAF6-like RNA polymerase II, p300/CBP-associated factor (PCAF)-associated factor	+	Yes
Lmo2	Ndufa8		NM_026703	NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 8	+	No
Lmo2	Kat7	Myst2, Hbo1	NM_177619	K(Lys) acetyltransferase 7	−	Yes
Lmo2	Rai1	Gt1	NM_009021	Retinoic acid induced 1	−	Yes
Lmo2	Setd8	PR-SET7	NM_030241	SET domain containing (Lys methyltransferase) 8	−	Yes
Lmo2	Tab2	Map3k7ip2	NM_138667	TGF-beta activated kinase 1/MAP3K7 binding protein 2	−	Yes
Lmo2	Tubg1	Tubg	NM_134024	Tubulin, gamma 1	−	Yes
Lmo2	Uhrf1	Np95, Icbp90	NM_010931	Ubiquitin-like, containing PHD and RING finger domains, 1	−	Yes
Lmo2	Atp5c1	1700094F02Rik	NM_020615	ATP synthase, H+ transporting, mitochondrial F1 complex, gamma-polypeptide 1	−	No
Lmo2	Pck1	Pepck	NM_011044	Phosphoenolpyruvate carboxykinase 1, cytosolic	−	No
Lmo2	Zfp106		NM_011743	Zinc finger protein 106	−	No
Lmo2	Adap1	Centa1, p42(IP4)	NM_172723	ArfGAP with dual PH domains 1	N/A	Yes
Lmo2	Appbp2	PAT1	NM_025825	Amyloid-beta precursor protein (cytoplasmic tail) binding protein 2	N/A	Yes
Lmo2	Baz1a	Acf1	NM_013815	Bromodomain adjacent to zinc finger domain 1A	N/A	Yes
Lmo2	Cactin	2510012J08Rik	NM_027381	Cactin, spliceosome C complex subunit	N/A	Yes
Lmo2	Capn15	Solh	NM_015830	Calpain 15	N/A	Yes
Lmo2	Capza2	Cappa2	NM_007604	Capping protein (actin filament) muscle Z-line, alpha 2	N/A	Yes
Lmo2	Ccna2	Ccn1, Ccna	NM_009828	Cyclin A2	N/A	Yes
Lmo2	Cdk9		NM_130860	Cyclin-dependent kinase 9 (CDC2-related kinase)	N/A	Yes
Lmo2	Cxxc1	Cfp1, Cgbp	NM_028868	CXXC finger 1 (PHD domain)	N/A	Yes
Lmo2	Egln2		NM_053208	Egl-9 family hypoxia-inducible factor 2	N/A	Yes
Lmo2	Gtf3c2	TFIIIC110	NM_027901	General transcription factor IIIC, polypeptide 2, beta	N/A	Yes
Lmo2	Myl6	Mlc3	NM_010860	Myosin, light polypeptide 6, alkali, smooth muscle and nonmuscle	N/A	Yes
Lmo2	Wdr46	Bing4	NM_020603	WD repeat domain 46	N/A	Yes
Lmo2	Ylpm1	ZAP3	NM_178363	YLP motif containing 1	N/A	Yes
Lmo2	Atg2a	1810013C15Rik	NM_194348	Autophagy related 2A	N/A	No
Lmo2	B4Galnt1	GalNAc-T	NM_008080	Beta-1,4-N-acetyl-galactosaminyl transferase 1	N/A	No
Lmo2	Ckap2l	Radmis	NM_181589	Cytoskeleton associated protein 2-like	N/A	No
Lmo2	Dennd5a	Rab6ip1	NM_021494	DENN/MADD domain containing 5A	N/A	No
Lmo2	Hyou1	Orp150	NM_021395	Hypoxia up-regulated 1	N/A	No
Lmo2	Itm2b	Bri2	NM_008410	Integral membrane protein 2B	N/A	No
Lmo2	Lias		NM_024471	Lipoic acid synthetase	N/A	No
Lmo2	Mpo		NM_010824	Myeloperoxidase	N/A	No
Lmo2	Nans	Sas	NM_053179	N-acetylneuraminic acid synthase (sialic acid synthase)	N/A	No
Lmo2	Pitrm1	Ntup1	NM_145131	Pitrilysin metallopeptidase 1	N/A	No
Lmo2	Rpl17		NM_001002239	Ribosomal protein L17	N/A	No
Lmo2	Rps11		NM_013725	Ribosomal protein S11	N/A	No
Lmo2	Ypel3	Suap	NM_025347	Yippee-like 3 (Drosophila)	N/A	No
Tal1	Tcf3	E2A, E47	NM_011548	Transcription factor 3	+	Yes
Tal1	Tcf4	E2-2	NM_013685	Transcription factor 4	+	Yes
Tal1	Erg	p55; Erg-3	NM_133659	Avian erythroblastosis virus E-26 (v-ets) oncogene-related	N/A	Yes
Tal1	Fli1	Sic1; EWSR2; Fli-1; SIC-1	NM_008026	Friend leukemia integration 1	N/A	Yes
Tal1	Drg1	Nedd3; AA408859; AI132520	NM_007879	Developmentally regulated GTP binding protein 1	N/A	No
Tal1	Tusc1	TSG-9; 2200001D17Rik	NM_026954	Tumor suppressor candidate 1	N/A	Yes

Illustrated are proteins for which there were at least two independent clones per screen (analysis of 6 × 10⁶ clones from a cDNA library of Kit⁺LIN⁻ hematopoietic cells). LMO2 interacts with 52 proteins, including two Lys methyltransferases, MLL2 and SetD8, both controlling erythroid gene expression (61, 62). In contrast, TAL1 interacts with six proteins only, mostly E protein transcription factors, DRG1 as well as ERG and FLI1. IP, immunoprecipitation; N/A, not available.

LMO2 Associates with Replication Complexes.

We next addressed the question of whether LMO2 associates with the pre-RC in hematopoietic cells. Because pre-RC formation often requires a DNA template (35), we prepared DNA-containing chromatin extracts from CD34⁺Kit⁺ progenitors (TF-1 cells) (3, 13, 36, 37) to assess the interaction of LMO2 with endogenous replication proteins by immunoprecipitation. We found that POLD1, PRIM1, and MCM6, but not lamin B or beta-actin (negative controls), reproducibly and specifically coimmunoprecipitate with LMO2, but not with control Ig, in TF-1 chromatin extracts (Fig. 1F). We also detected the other members of the MCM hexameric complex (MCM2–MCM7), as well as ORC2, CDC6, CDT1, and PCNA, albeit with weaker efficiency for the latter three (Fig. 1F). In contrast, MCM6, POLD1, and PRIM1 did not coimmunoprecipitate with SCL (Fig. 1G), whereas LMO2 and LDB1 were found with SCL in this assay, as expected. Therefore, replication proteins do not stably associate with the SCL transcription complex but more specifically with LMO2.

LMO2 Binds to Origins of Replication.

LMO2 shares important transcriptional targets with SCL (15, 16, 25, 36). Nonetheless, LMO2 is rarely found at gene promoters (3%) by ChIP sequencing in multipotent progenitors, whereas SCL is more frequently observed within transcriptional start sites (28%) (24), suggesting that LMO2 has SCL-independent functions. We assessed whether LMO2 associates with DNA replication origins, using ChIP of TF-1 cells with anti-LMO2, or anti-MCM5 antibodies. We first studied seven well-characterized human DNA replication origins (38) and found that both MCM5 and LMO2 occupied four of these replication origins, c-MYC, G6PD, TOP1, and MCM4 (Fig. 2A), all mapping to early replicating G1 (ERG1) segments (39, 40) (Fig. S1A). In contrast, MCM5 did not bind the GYPA promoter, a well-defined SCL-LMO2 transcriptional target (36), whereas LMO2 occupancy was confirmed, together with SCL and GATA1, two LMO2 transcription factor partners. SCL was detected at two of the seven tested origins, although binding was 10- to 20-fold lower compared with GYPA, whereas GATA1 binding was below the detection limit (Fig. 2A). Therefore, LMO2 is recruited to well-characterized DNA replication origins with MCM5, in the absence of GATA1 and frequently of SCL.

Fig. 2.

LMO2 and MCM5 occupy DNA replication origins. (A) ChIP of well-characterized DNA replication origins in TF-1 cells with anti-LMO2, anti-MCM5, anti-SCL, and anti-GATA1 Abs. Shown are the ratio of enrichment over control Ig and background (c-Kit, −20 kb). GYPA promoter sequences were amplified as a control (36). (Left) Maps depicting the location of well-characterized DNA replication origins and PCR probes. Gene exons are illustrated by black boxes. PCR primers are shown as red bars. E, E47 consensus sites; G, GATA1/2 consensus sites. n.d., not detectable. (B) LMO2-bound replication initiation zones (gray, first column) are preferentially enriched within ERG1 segments. Forty-five replication initiation zones found in common in two studies (38, 41) were analyzed by ChIP for LMO2 binding. The replication timing of these origins according to two datasets is illustrated: study 1 [Gilbert and coworkers (42)] in leukemic patients and lymphoblastoid cell lines and study 2 [Stamatoyannopoulos and coworkers (39)] in B cells and ES cells. A higher proportion of early replicating segments (gray, last column) is found in LMO2-bound origins (P = 0.02). Mix: Some samples from study 1 showed early replication, whereas others showed late replication. (C) Diagram illustrating the synthetic DNA replication assay. Plasmid DNA was transfected into HEK293 cells and detected by PCR (primer pairs as shown). (Left) Dpn1 sensitivity distinguishes unreplicated methylated plasmid DNA (sensitive) from replicated hemi- or unmethylated DNA (resistant). (D) LMO2 fused to GAL4 stimulates the replication of a GAL4-responsive plasmid in mammalian cells. HEK293 cells were transfected with the indicated GAL-fusion genes. After Dpn1 digestion, Dpn1-resistant extrachromosomal DNA was quantified by real-time PCR (n = 2). P < 0.05. (E) LMO2-induced DNA replication depends on LMO2 tethering to DNA via GAL4 binding sites (UAS). The experiment was performed as in D. (F) LMO2 binding to artificial origins of replication is sufficient to recruit DNA replication proteins specifically. DNA capture was performed using immobilized WT or mutant UAS sequences on beads. Bound proteins were revealed by Western blotting. Data are the mean ± SD of at least two independent experiments performed in duplicate.

Fig. S1.

(A) Replication timing of DNA replication origins in leukemic and lymphoblastoid cells [study 1 (40)] and in B lymphoblastic (Bly) and ES cells [study 2, (39)]. LMO2 binding assessed by ChIP (Fig. 2A) is illustrated. (B) Subset of 45 initiation zones is identified in common by two different studies (38, 41). (C) LMO2 and MCM5 co-occupy the above initiation zones (shown in B), which are numbered according to the system of Cadoret et al. (38). Primer pairs are listed in Tables S2 and S3. Immunoprecipitated chromatin levels are indicated as the ratio of enrichment over control isotype Ig, after normalization with a negative control DNA sequence (c-Kit, −20-kb upstream sequence). Three additional background regions (bg1, bg2, and bg3) used by Karnani et al. (41) are also shown.

These results led us to assess LMO2 occupancy of replication initiation zones identified in two independent tiling array-based studies in HeLa cells, using Lambda-exonuclease digestion (38, 41) and anti-BrdU immunoprecipitation to purify origin-centered nascent DNA strands. We focused on a subset of 45 replication initiation zones that overlapped within a distance of 1 kb between the two studies (Fig. 2B and Fig. S1 B and C). In TF-1 cells, MCM5 and LMO2 were each detected on 20 and 19 replication initiation zones, respectively, nearly half of which were positive for both (Fig. 2B). We aligned these 45 replication initiation zones with ERG1 segments identified in mammalian cells (39, 42) and found that 68% of LMO2-bound zones mapped to ERG1 segments, which is almost twice the percentage of ERG1 segments found in the absence of LMO2 (Fig. 2B). Therefore, LMO2 is recruited to a significant proportion of initiation zones in TF-1 cells, corresponding, in most part, to ERG1 segments.

Tethering LMO2 to DNA Recruits Replication Proteins and Induces DNA Replication.

To determine whether LMO2 tethering to DNA was sufficient to stimulate DNA replication, we optimized a synthetic DNA replication assay in mammalian cells described by Takeda et al. (43). We assessed whether LMO2 fused to the DNA binding domain of GAL4 can drive DNA replication from GAL4 binding sites (5XUAS) in transfected 293 cells (Fig. 2C). Newly replicated DNA is hemimethylated or unmethylated and can be distinguished from transfected, bacterially derived DNA by its resistance to Dpn1 digestion (43). As expected, GAL4-ORC2 induced an approximately fivefold increase in newly synthesized DNA compared with control GAL4-VP16 (Fig. 2D). In this assay, GAL4-LMO2 reproductively directed a dose-dependent increase of newly replicated DNA, reaching a maximum of eightfold (Fig. 2D), whereas GAL4-SCL did not produce the same results. This activity was abrogated when GAL4-binding UAS sequences were mutated (Fig. 2E) or absent (pBluescript vector). We therefore conclude that LMO2 anchoring to DNA was sufficient to direct DNA replication. LMO2-driven DNA replication occurred most likely in the absence of transcription because 293T cells lack hematopoietic transcription factors that recruit LMO2 to DNA, namely SCL and GATA1/2. To determine if DNA-bound GAL4-LMO2 could recruit DNA replication proteins to the UAS template, we performed DNA capture as previously described (44), using UAS sequences immobilized on beads. LMO2 recruited POLD1, MCM5, and, to a lesser extent, CDT1 and PCNA to DNA, and this recruitment required the integrity of the GAL4 binding site (Fig. 2F). Together, our results indicate that LMO2 tethering to DNA was sufficient to nucleate the assembly of prereplication/preinitiation complexes specifically at the site of binding and to direct DNA replication.

LMO2 Expression Levels Control the Rate of DNA Synthesis and Cellular Outcome in the Erythroid Lineage.

Cell cycle is highly regulated in the erythroid lineage because ∼80% of primary proerythroblasts (E1 stage; Fig. 3A) are in S phase, exactly at the onset of erythropoietin (Epo) dependence (45), and this proportion sharply drops at the E3 and E4 stages of terminal differentiation. We observed that the E1 population segregated into cells with high and low endogenous LMO2 protein levels (Fig. 3B), corresponding to high and low proportions of cells in S/G2/M (Fig. 3C). Strikingly, LMO2 protein levels decrease from the E1^hi to E4 stage, directly correlating with a decrease in the proportion of cells in S phase (Fig. 3C), whereas GATA1 and SCL protein levels increase (Fig. 3D). This process is highly coordinated because most LMO2 partners identified here, including CCNA2, are synchronously down-regulated with Lmo2 during differentiation from proerythroblast to orthochromatic erythroblasts (Fig. S2).

Fig. 3.

LMO2 levels determine the proliferation of erythroid progenitors. (A) Diagram of erythroid differentiation according to flow cytometry profiles. (B) LMO2 levels in bone marrow erythroid progenitors (Lin⁻CD71⁺Ter119⁻) correlate with their cell cycle status by Hoechst staining: E1^low and E1^high (P ≤ 0.05). The mean fluorescence intensity (MFI) for LMO2 per cell was assessed by flow cytometry. (C) Correlation between LMO2 levels and the proportion of cells in S/G2/M during erythroid differentiation (E1 to E4). (D) Expression levels of LMO2, SCL, and GATA1 during erythroid differentiation. (E) Decreased LMO2 in erythroid progenitors reduced the fraction of cells in S phase. The shRNA Lmo2 was delivered in Ter119⁻ fetal liver cells, which were then stimulated with Epo for 2 d. The cell cycle in erythroid progenitors (E1, CD71⁺Ter119⁻) was analyzed by DAPI staining. (F) Lmo2 is required for the response of proerythroblasts to Epo. NT, nontarget (control). (G) LMO2 levels control the proliferation of erythroid progenitors in culture. (H) Decreased LMO2 abolishes the growth of erythroid progenitors (E1) in response to Epo but favors terminally differentiating erythroid cells (E4). (I) Decreased DNA synthesis induced by shRNA-mediated Lmo2 depletion. MEL cells were purified in G0/G1 and released in culture for different times in the presence of α³²P-dCTP. After electrophoresis, total DNA was quantified by autoradiography (n = 2). The slopes of the two curves were 0.48 ± 0.02 (Vector) and 0.29 ± 0.01 (shLmo2). *P < 0.0001. (J) Decreased proportion of cells in S phase after Lmo2 depletion. MEL cells expressing shLmo2 or control (Vector) were purified in G0/G1 and analyzed for cell cycle progression at different time points in culture by Hoechst staining (n = 3). All data shown are typical of at least two independent experiments.

Fig. S2.

Gene expression during the five stages of erythroid differentiation in the human and mouse. Shown are the genes encoding proteins that interact with LMO2 (Fig. 1A) within the DNA replication and cell cycle categories. RNA-sequencing data [reads assigned per kilobase of target per million mapped reads (RPKM)] were obtained from human cells (59) (A) or murine cells (60) (B).

We addressed the functional importance of LMO2 by decreasing LMO2 protein levels in erythroid progenitors via RNAi (shRNA Lmo2) (Fig. S3A). Lmo2 depletion in Ter119⁻ fetal liver erythroid progenitors decreased by twofold the proportion of cells in S phase as determined by DAPI staining and flow cytometry analysis, compared with control cells (Fig. 3E). In addition, Lmo2 depletion almost abrogated the proliferation of primary erythroid progenitors at a key commitment step marked by Epo responsiveness (Fig. 3 F and G), while enhancing the generation of mature E4 cells (CD71⁻Ter119⁺) (Fig. 3H). Therefore, high LMO2 levels are required for the Epo-dependent proliferation of CD71⁺ erythroid precursors, whereas lowering LMO2 accelerates terminal erythroid differentiation.

Fig. S3.

(A) Decreased LMO2 protein levels in MEL-expressing Lmo2-directed shRNA 867 were selected for further analysis. (B) Cells with reduced levels of LMO2 were not apoptotic. MEL cells infected as indicated were stained with the Annexin V apoptosis marker. Dead cells were stained with propidium iodide (PI). (C) Decreased proliferation of MEL cells expressing shRNA Lmo2. (D) Decreased LMO2 expression does not affect the transcriptional level of replication genes. RNA levels were normalized over HPRT and are shown as the ratio of expression levels in shRNA Lmo2 cells over control cells (empty vector) (n = 2 independent experiments performed in duplicate). Peptidylprolyl isomerase A (Ppia) served as an additional control. (E) MEL cells can be synchronized in G0/G1 by flow cytometry. (Left) Forward scatter (FSC) and side scatter (SSC) properties. (Lower Right) Cell cycle status of asynchronous MEL cells shown by DAPI staining. (Upper Right) Cell cycle status of the 10% smallest cells determined by the FSC/SSC profile. More than 90% of these cells are in G0/G1. (Right) Representative cell cycle profile of MEL cells after sorting (t = 0), and at the indicated times after release in culture. Note that when control cells were in late S phase, cells expressing shLmo2 were in early S phase. Finally, after overnight (O/N, 16 h) incubation, control cells have resumed normal cell cycle profiles, whereas Lmo2 knocked-down cells are still delayed in S phase.

Decreased LMO2 levels caused a twofold decrease in the rate of DNA synthesis monitored by the kinetics of ³²P orthophosphate incorporation into synchronized mouse erythroleukemia (MEL) cells (Fig. 3I), without causing apoptosis (Fig. S3B). Consequently, these cells failed to proliferate in culture (Fig. S3C). Decreased DNA synthesis was unlikely due to decreased expression levels for replication genes assessed by RT-quantitative PCR (Fig. S3D), consistent with transcriptome analysis of erythroid cells and lymphoid cells in which Lmo2 was knocked down or overexpressed, respectively (12, 46). Furthermore, replication genes were not occupied by the SCL–LMO2 complex in leukemic T cells (25). Together, these observations indicate a critical role for LMO2 in erythroid cell fate, via the control of DNA replication and the cell cycle that drives Epo-dependent expansion of erythroid progenitors while impeding their commitment to terminal maturation.

LMO2 Expression Regulates S-Phase Entry.

We next addressed the importance of LMO2 levels on the kinetics of cell cycle progression (Fig. 3J). Briefly, we found that viable MEL cells with low side and forward scatter properties are mostly in G0/G1 (Fig. S3E). These G0/G1 cells were purified and released in culture medium to analyze their DNA content by Hoechst staining (Fig. S3E). Decreased LMO2 (shRNA Lmo2) reproducibly caused a 2-h delay in G1/S transition, occurring at 3 h after release compared with 1 h for control cells (Vector), as well as delayed S-phase progression (Fig. 3J and Fig. S3E), indicating that LMO2 levels control S-phase entry in erythroblasts.

To mimic retroviral integration upstream of the LMO2 locus in the X-SCID gene therapy trial (19, 20, 46) and define the consequences of ectopic LMO2 expression in vivo, we delivered LMO2 in hematopoietic stem cells by retroviral infection, (Fig. 4 A–F). Upon transplantation, LMO2-transduced bone marrow cells induced T-ALL in 60% of recipient mice, with a median of 270 days (46) (Fig. 4B and Fig. S4). During the preleukemic stage (30 days), LMO2 overexpression mostly affected thymocytes and led to an accumulation of CD44⁺CD25⁻CD4⁻D8⁻ double-negative thymocyte (DN) cells and a decrease in CD4⁺CD8⁺ double-positive thymocyte (DP) cells in the thymus, reproducing the differentiation blockade at the DN stage reported in LMO2 transgenic mice (Fig. 4C), despite the fact that the retroviral vector allowed for transgene expression in all cells. Elevating LMO2 in thymocytes modified the cell cycle status of thymocyte progenitors (Fig. 4D) without affecting bone marrow stem cells (Fig. S4B), consistent with the role of LMO2 as a T-cell specific oncogene (46). Interestingly, elevating LMO2 enhanced the cell cycle status of both DN1 and DP cells, suggesting that differentiation blockade (lower number of DP cells) could be due to increased cell cycle. Furthermore, when DN1 thymocytes were synchronized in G0/G1 and released in culture, LMO2-expressing cells entered more rapidly into S phase compared with control cells (vector) (Fig. 4E), indicating that LMO2 facilitates the G1-S transition and S phase progression.

Fig. 4.

Ectopic LMO2 expression in thymocytes triggers G1/S transition, T-cell hyperproliferation, and T-ALL development. (A) Experimental strategy to deliver LMO2 in hematopoietic cells using the murine stem cell virus (MSCV) retroviral vector. BM, bone marrow. (B) LMO2 overexpression induces T-ALL in mice. Kaplan–Meier curves showing the time of leukemia onset in mice transplanted with MSCV-LMO2 or MSCV transduced cells. (C and D) LMO2 overexpression leads to an increase of the DN1 thymocyte subset but a decrease in DP cells in vivo, despite higher proportions of cycling cells in both populations. (E) LMO2 overexpression induces an increase in the percentage of DN1 cells in S phase. DN1 cells were purified in G0/G1 and analyzed for cell cycle progression by DAPI staining at different time points after coculture with MS5-DL4 stromal cells. (F) LMO2-expressing thymocytes are more sensitive to 5-FU treatment in vitro. Cells were cocultured on MS5-DL4 stromal cells and treated with 5-FU at the indicated doses. Viable Thy1⁺ cells were scored by flow cytometry. (G) Proposed model: LMO2 controls DNA replication, favors progenitor cell proliferation, and inhibits commitment to terminal differentiation in the erythroid and T-lymphoid lineages. Data shown are typical of at least two independent experiments.

Fig. S4.

(A) Murine bone marrow cells were transduced with LMO2 or the empty vector (MSCV-YFP) and purified into YFP⁺ and YFP⁻ cells. The human LMO2 transgene was revealed by PCR of genomic DNA. Murine Otx1 served as a positive control for DNA loading. (B) Cell cycle analysis of hematopoietic stem cells [Kit⁺Sca⁺Lin⁻ (KSL)] and progenitors [Kit⁺Sca⁻Lin⁻ (KL)] ectopically expressing LMO2. (C) Typical spleen from a leukemic mouse transplanted with LMO2-overexpressing cells, compared with control. (D) Immunological phenotype of LMO2-induced leukemias. Representative flow cytometry analysis of MSCV-LMO2–induced leukemia cells, compared with cells from a healthy mouse. In the thymus, an expansion of the double-negative compartment is observed. Leukemic cells express Thy1, appear as blast cells (FSC profile), and invade the bone marrow and the spleen.

The above results indicate that LMO2 overexpression in hematopoietic cells reproduced the T-cell proliferation phenotype reported for preleukemic patients in the X-SCID gene therapy trial. Actively dividing cells are more sensitive to 5-fluorouracil (5-FU), an antimetabolite that inhibits thymidylate synthase and therefore depletes the pool of dTTP. Consistent with increased DNA replication, LMO2-expressing thymocytes cocultured with MS5 stromal cells expressing DL4 were 100-fold more sensitive than WT thymocytes to 5-FU (EC₅₀ of 5 nM vs. 500 nM) (Fig. 4F).

Discussion

LMO2 has a well-established function in transcriptional regulation via direct interaction with transcription factors, mostly of the bHLH family, SCL/TAL1, TAL2, and LYL1 or GATA proteins (2–4). In this study, we revealed unexpected new functions of LMO2 in hematopoiesis and leukemogenesis through a yeast two-hybrid screen of LMO2 interaction partners in hematopoietic progenitors. Our observations indicate that LMO2 controls cell fate by directly promoting DNA replication, a hitherto unrecognized mechanism that might also account for its oncogenic properties.

Our study unravels unexpected interactions between LMO2 and three essential replication proteins, MCM6, PRIM1, and POLD1 (30), as well as chromatin-modifying enzymes that have well-characterized roles in DNA replication, BAZ1A, SETD8, MYST2, and UHRF1 (31–33, 47). More importantly, tethering LMO2 to DNA via the GAL4-DNA binding domain was sufficient to recruit MCM5 and POLD1 to DNA and to transform UAS into origins of replication, indicating that LMO2 directly controls DNA replication. Unlike other transcription complexes that have a dual role in DNA replication (48), LMO2 interaction with the RC is distinct from the well-known recruitment of LMO2 to the SCL transcription complex. Our data are in line with the observations that there is only a partial overlap between SCL and LMO2 chromatin occupancy in hematopoietic progenitors (24).

LMO2 binding to replication initiation zones reported here overlaps with early G1 replication segments described in lymphoid cells (39, 42), a possibility that may be favored by its interaction with MLL2 (Fig. 1A). Indeed, the H3K4me3 histone mark found in early replicating domains (40) can be controlled by MLL2. In eukaryotes, origins are licensed in excess, and not all licensed origins are active during a given cell cycle, which requires recruitment of DNA polymerases to initiate DNA synthesis (30). Accordingly, we show that LMO2 levels influence the rate of DNA replication in MEL cells.

Although we focused on the basic components of the pre-RC in the present study, several other proteins identified in the screen could have a regulatory function. For example, cyclin A2-CDK2 favors the onset of DNA replication at the G1-S transition and prevents rereplication during S phase (49). These possibilities would be consistent with the effects of LMO2 levels on the kinetics of G1-S transition that we observed in two cell types, delayed when Lmo2 is knocked down in murine erythroleukemic cells and, conversely, accelerated in LMO2-overexpressing DN1 thymocytes. In addition, LMO2 associates with two cell cycle checkpoint proteins, CDK9 and BUB1B (Table S1). CDK9 associates with cyclin K in replication stress response and prevents DNA damage in replicating cells (50). BUB1B is an essential component of the mitotic checkpoint. The importance of these cell cycle proteins for LMO2-induced cell proliferation remains to be addressed.

Cell cycle is a highly regulated process in the erythroid lineage. CD71⁺Ter119⁻ proerythroblasts represent a critical transitional stage marked by Epo dependency (51) and elevated DNA replication (45), both shown here to require high LMO2 protein expression. Conversely, terminal erythroid differentiation to the CD71⁻Ter119⁺ stage is necessarily coordinated with growth arrest (13, 51), which we now show to be favored by Lmo2 down-regulation, in agreement with previous work indicating that LMO2 overexpression prevents terminal erythroid differentiation (11). We conclude that LMO2 down-regulation is required for the switch to terminal erythroid differentiation due to the implication of LMO2 in DNA replication/cell cycle (Fig. 4G).

It is well established that transcription factor gene networks drive hematopoietic cell development and lineage outcome, via synergistic or antagonistic interactions (reviewed in refs. 52 and 53). It is unclear how LMO2 inhibits cell differentiation in both the erythroid (11) and T-lymphoid lineages (reviewed in ref. 53). We now propose that LMO2-dependent DNA replication in both lineages governs the switch between a proliferative state in progenitors and commitment to terminal differentiation (Fig. 4G). Failure to regulate this proliferative switch caused by ectopic LMO2 expression (19) may lead to T-ALL.

Oncogene-induced DNA replication stress (54) could lead to replication errors and ultimately cause genetic lesions, such as activating NOTCH1 mutations (55, 56), that convert preleukemic stem cells (15, 16) into leukemia initiating cells (15). Two other LIM-only proteins, LMO1 and LMO4, are also important determinants of cell cycle progression in neuroblastoma (23) and in breast cancer associated with genomic instability (22), respectively, suggesting that the mechanism(s) described here may be extended to these proteins. Emerging evidence indicates that oncogenes, such as c-MYC or HOXD13, can be part of nontranscriptional complexes involved in DNA replication (54, 57). Taken together, we propose that oncogenic transcription factors in acute leukemias and possibly other tumor types transform cells by at least two DNA-dependent mechanisms, the control of gene expression programs, as initially proposed (1), but also by deregulating DNA replication (42), with both processes being important determinants of cell fate.

Materials and Methods

Yeast Two-Hybrid.

LMO2 full-length protein was subcloned in the pGBKT7 vector and transformed in the Y187 yeast strain. The cDNA library from c-Kit⁺Lin⁻ murine bone marrow cells was constructed in the pGADT7 vector with the Matchmaker Library Construction and Screening Kit (BD Biosciences) and transformed in the AH109 yeast strain. The screen (selection of Ade⁺His⁺Leu⁺Trp⁺ colonies) was performed by yeast mating. Positive clones were isolated and sequenced. To confirm the interactions and check for specificity, full-length cDNAs for Polδ1, Prim1, Mcm2-5-6-10, Cdt1, Pcna, and Caf1 were subcloned into pGADT7 and transformed in AH109 cells.

Retroviral Gene Transfer and RNAi.

LMO2 retroviral gene transfer and mouse bone marrow transplantation were performed as described (8). For RNAi experiments, vesicular stomatitis virus (VSV-G)–producing cells were transfected with plasmids encoding shRNAs against LMO2, with a nontarget shRNA or with the empty pLKO.1 vector (Sigma). MEL cells or Lin⁻ fetal liver cells were incubated with VSV-G supernatant for 48 h and then selected with puromycin as described (8). All mice were kept under pathogen-free conditions according to institutional animal care and use guidelines. The protocols for gene transfer and transplantation in mice were approved by the Committee of Ethics and Animal Deontology of the University of Montreal.

Flow Cytometry, Cell Cycle Analysis, and Cell Sorting.

Flow cytometry, cell cycle analysis, and cell sorting are described in Supporting Information.

³²P Orthophosphate Labeling and Quantification of de Novo DNA Synthesis.

MEL cells synchronized in G0/G1 were incubated with α³²P-dCTP (100 μCi/mL; PerkinElmer) in DMEM [10 mM Hepes (pH 7.4), 10% (vol/vol) FCS] at 8 × 10⁵ cells per milliliter for the indicated times. Cells were lysed in DNA extraction buffer [80 mM Tris⋅HCl (pH 8.0), 8 mM EDTA, 100 mM NaCl, 0.5% (g/100 mL) SDS]. After proteinase K digestion and phenol/chloroform extraction, DNA was resolved on an alkaline (NaOH) agarose gel. The gel was dried on a Biodyne membrane (Pall Corporation). ³²P incorporation was visualized by autoradiography and quantitated using ImageQuant software (GE Healthcare).

Protein Extraction, Immunoprecipitation, Immunoblot, Abs, RNA, and ChIP Analysis.

Protein extraction, immunoprecipitation, immunoblot (IB), Abs, RNA, and ChIP analysis methods are fully described in Supporting Information. Primer sequences used are shown in Tables S2 and S3, and Abs used for ChIP, IB, and immunofluorescence are listed in Table S4.

Table S2.

List of primer pairs

Real-time PCR oligos (ChIP)
c-Myc Fw	ATACGTGGCAATGCGTTGCT
c-Myc Rv	GAGCCGCATGAATTAACTAC
G6PD Fw	ACTCCACAATGACCTGGA
G6PD Rv	AGAGGGTCTTAACCAATCC
Top1 Fw	CACTGCCTAGCAGAGGGGCTGGG
Top1 Rv	AGCAGTTGTGTAACAGCCTAAGTTCGC
MCM4 Fw	AAACCAGAAGTAGGCCTCGCTCGG
MCM4 Rv	GTCTGACCTGCGGAGGTAGTTTGG
Hsp70 Fw	AGACTCTGGAGAGTTCTGAGCA
Hsp70 Rv	TTTCCCTTCTGAGCCAATCACC
LaminB2 Fw	GGCTGGCATGGACTTTCATTTC
Lamin B2 Rv	CTTAGACATCCGCTTCATTAG
Dnmt1 Fw	AACTGGGTTTGGTGGCATGT
Dnmt1 Rv	ATGCAATCACGGCTCACTGT
Gpa Fw	CCACTTTCATAGCCCCAAGA
Gpa Rv	TCATGAGCTGGTTCCTGAAG
Real-time PCR oligos (RT-PCR)
LMO2 Fw	TACTTCCTGAAAGCCATCGACC
LMO2 Rv	GATCCCATTGATCTTGGTCCAC
Orc2 Fw	TCATCAACGGCTACTTTCCTGG
Orc2 Rv	GTACCCACATGACTGAGGACAT
Orc5 Fw	TGAACCTCACCTGAAGAAGGCA
Orc5 Rv	CAGTTGTCCTGGGTCTGTGTTA
MCM2 Fw	AGACCTCACAGAGCCCATCATT
MCM2 Rv	GCCACCATTAGTCAACCCTTCA
MCM3 Fw	AGCTGTGTCCTGCGTTTGTT
MCM3 Rv	ATCCAACCTTGTCATCTGCCTG
MCM4 Fw	ACCAAAGTGAGGAGCAAGTGGA
MCM4 Rv	TCGAGGGTAAGCAGAAACCATC
MCM5 Fw	AGCCGCATTGAGAAGCAACT
MCM5 Rv	TTCGGATAGCGTGCTCTGGATA
MCM6 Fw	TACTGCCGAATCTCGAACCTCA
MCM6 Rv	TCGCTTCTCTTTAGTGCCGACT
MCM7 Fw	ATTCGACCTCCTCTGGCTGATT
MCM7 Rv	TGGTGGACATAGGTGATGTGCT
Cdc6 Fw	AGTCCTCAAACCACTCTCCGAA
Cdc6 Rv	AGCAAACCAGGATCTTCTGCTG
Cdt1 Fw	ACTATGGAGGTGGTCTGTGCAA
Cdt1 Rv	AGCTTGACGTAGGTATCCGTG
Pold1 Fw	TCATCACCAAAGAGTTGACCCG
Pold1 Rv	CACCCTTAGCAGCACCAATGAT
Prim1 Fw	ACCAGTCTAGCACCGTATGTGA
Prim1 Rv	TTCACGTTGGTTTGATGGGAGC
PCNA Fw	AGCCACTCCACTGTCTCCTA
PCNA Rv	CTGGCATCTCAGGAGCAATCT
Cdc45 Fw	GTCATGTGTCCCGTCATAACCA
Cdc45 Rv	TTAAACCTGGCTGCTGTGTAGC
PPIA Fw	GTGCCAGGGTGGTGACTTTACACG
PPIA Rv	TCCCAAAGACCACATGCTTGCCA
Hprt Fw	GGCCAGACTTTGTTGGATTTG
Hprt Rv	CACAGGACTAGAACACCTGC
PCR oligos (presence and expression of transduced LMO2)
MSCV-LMO2 Fw	AAGAGCCTGGACCCTTCA
MSCV-LMO2 Rv	TCCCCTACCCGGTAGAAT
Otx1 Fw	GTCTTACCTCAAACAACCCCCA
Otx1 Rv	AAGATGTCTGGGTAGCGAGT
LMO2 Fw	AGGAACCAGTGGATGAGGTG
LMO2 Rv	CGATGGCCTTCAGGAAGTAG
S16 Fw	AGGAGCGATTTGCTGGTGTGG
S16 Rv	GCTACCAGGGCCTTTGAGATG

Fw, forward; Rv, reverse.

Table S3.

List of primer pairs

Replication zones	Forward	Reverse
Region_20	GGCCTTTGAAAATTGAGGAG	TACACTTCTACCACCCTCCCTA
Region_23a	GATAATTAACAGGCAGCGTGAG	TCTATCTACCCTTCCACCTCAA
Region_23b	GATTCTGTAGATTCTGGGCAAG	TTACACACTACTGCCACTGCAT
Region_23c	CAGCAACACTTCCCTGGATA	CTGCACTGGTAATGATGTCTCA
Region_24	ACTGGTATTGTTCTCTGAGCTAGG	GAGCTCTTATGCTCCTCTGCTA
Region_29	GGGAGGATGTAGGGGTATCG	TTCTGTGGTTAGGCAAATAGC
Region_50	AGCTCTTTCGGACTTGGACT	ACGAAGCGATTCTTACTGTCC
Region_53a	GAAGGGGGTCTTTCCTACTTTA	CAATTCTTGAGACAGTCAGTGC
Region_53b	ATCTGTTGGGGAGAAGGAAG	GAGCTATCCCAAGGATAAAGGA
Region_54	GGTTGAGTTGGGGGTATTTTCT	GAACATCTGGTTCTCAATCTGG
Region_87a	AGAGGATGGTATGTGAGGTGAG	TAGTGATGTCTGGTCATTCTGC
Region_87b	AATAGCCATGCCTGGATACAC	TAGAGGTGGGTGTAGGTTTGAG
Region_87c	GCTGAGGTAAGGAGAGATGAGA	GTTCTGTGCCCTCAACAGAT
Region_88	CTTAGTGTTTCCTGGAGAAGGA	AGGCAAGAGAAAATATGGACAC
Region_93	GGTGTGAGATCTGAGCTTGATT	ATAGTCTCTGCCTGGACCCAAC
Region_105a	ACTGCTGTGTCTCTTTCTGCTT	CCAAACATACATATGCCTACCC
Region_105b	AGGACAACGTGCTGTGAGTAAT	CAGACAGAATAGGTAGGGGTCA
Region_105c	CTCTTCATGTAGGTCAGGCTTC	TCTTCTGGTGCCTGGTAAGATA
Region_117a	CAGATAGAATCGTCTGCAACAC	GCAGGTGACTTCTGTTCTAAGC
Region_117b	TGGCTGATTACCCAGTAAGTCT	TTGATCTACAGGGCCAGTCTAC
Region_117c	CCAGTTCTATCACATTTCAGCA	AGACAGAAAGATGGAGTCTTCG
Region_126	CAGCTACTGTGGTTCTTCAAGC	TCTATCTACGTGAGGCACTCCT
Region_130a	GGACCATGATGTCAGAAGTTG	GGTATGTTAAGTGCCTCTTGGT
Region_130b	CAGACCCTCCTTAGGAAGAGA	CCTCACATCCTACTAGTCTTCCA
Region_130c	CCTGTCTCAAAGTTGTGAGGAT	AGCAAGACCCTGTCTCAAAATA
Region_151a	GACCAAAGGTCATCTGCTGTAG	CATAGTCTTTCCCATCTTGAGC
Region_151b	ACACATCTCACCAGTTTGACAC	TCTATGAGAGGCCTGTGGACTA
Region_151c	ACTCTCCATTCCCATCTTACCT	CTCAGCTATGCTGTCTTCCAGT
Region_155	GAGACTTCCCAAGGGTTACATA	CACTCACAGTCAGAAGGAAGTG
Region_162	ATGCACCTAAGCCTGTCTTGT	GGATAACAGCTGGAAGAAGGTA
Region_163	GTGGTAACTTCTGACCTGCATT	GGTGAGGGAGGAAACTAAGAAG
Region_165	GCAGTGTATTGAATCCTTGGAG	AGCAGAGCCAAGAAATCTACAG
Region_167	AGTCAGAACCCTGGTCTTTGGT	AGTGGCAGCCAGAACTTGAAAC
Region_188a	ACAACCAAAGCTATAGTAGACTGGA	CTCCACTTTTATTGGAGGAACA
Region_188b	GTGCAGTGCTCTTGAAAGAAGT	CTTAAAAGCTGCCTCAGTTACG
Region_188c	CAAGGTACTTCCCTTTATGTCG	TTCCCAGCATATACAGTTGAGG
Region_188d	AACGGCCATTATTACCATGA	ATAAGTTGGAAGCAGGAAGGTC
Region_190	GATATGACTCACCTCCTCCAAA	AGGGAGAGGAAAGGAGTAAAGA
Region_193a	CTCATGTCATCAGCCTTGAGTA	AGGTTCTGAGGCTAAACAACAG
Region_193b	GGTTGTCGATGTGAGAGCTAA	CTGTGGAGGTGTTGCTTCTATC
Region_193c	AGCTGGTGCACCTTAGTTGAT	AGTGTCTCCAGGCTTAGGAAAT
Region_194a	AGGATTGGAGGAAATCTACTGG	CCAATGGGTGAAGTAAGGTG
Region_194b	CATCAGTGTAAGCAGTCGGTAA	AAGGTACTTACCCCATCCTAGC
Region_194c	GCGAAGACATATTGCCACAG	GCTCAGGAGTTTGAAAGAAAGG
Region_202	ATAGAGCTACAGGGAGGAAAGG	TCACTCGGTAACAGGGATCTAC
Region_206	AGAGCTCAAAGCTAACAGGTTG	ACTTCCTTTGTATGAGCCAGTG
Region_207	GTCAGAGATCTGGCAGTGAATC	GTCTTTTGAAGAGTTCCACCAC
Region_208	ATGGGATTGGAGAGAAACTAGG	CAGAATTCTCCACGTCCATC
Region_209a	CGCTGACTTCAAAATGAGATTC	AGATCTGTGGTCTCTGTCTTGG
Region_209b	ATCTCAAGGAGCCAGAATCAC	CCAGTAGCCAATAGGAACAGAA
Region_209c	GCAAAGGTTGGAAAGGGAGCAA	TATCTGGCTCTTGCCTGTCTCT
Region_210	TCAAACCTAACACAGGTGGTAAC	CACAAAAGTAGTGCTGAAGATCC
Region_212	AGAATCTGTTTCGGACTTCCTC	AGGACATCAGACAGTCAAGGTT
Region_217	CTAGCAAATGAGCCTATTGACC	CTAGACTTTCAACCTGGGAATG
Region_218a	ATATTCAGTCTGGCCTCAACC	GAACAAGCAAGGGTGTCATAGT
Region_218b	CATCGCTTTATTGGCTGAGA	GAGTAGGCAGTTGAGGTTGTTG
Region_219a	GTCTCTGTTACTCTGCCAATCA	GATCAGTCTTGGGTGAAACAGT
Region_219b	ACGGACCAGTCCTACATTGAT	GAGCATCTAGAGTTTCCCTTTG
Region_219c	GAATTCCTCCAGGGCTCCTA	TTTCTTGCCCTACCCTTCTT
Region_219d	CTCCTGTTTTTCAGTTCCCAGT	GTGGAGAGGAATGGTATGGTTA
Region_219e	TCTAGCTTGGCACTTTACCTG	GGTGATCAACTGACTAGCAGGA
Region_220	GTTAACCCAGCAAGCTACAAAG	TGTCTTCTGGTAAGTCTCTGCAC
Region_221	GAACTGCTGCTCTCTTAATTCG	GTCAAAATTAGCAGGGTTCTCC
Region_225	GGCACAGGTGTACCTTGAATCT	CTGCAGTGAGCTGTGATCATGT
Region_228a	GAGCTATGAAAACCTGCTGAAC	GTGGTCACCCAAATACAGTACA
Region_228b	GCTGATTGAATAGAGAGCCACA	GAAGGGGGAAAACCAATAAACC
Region_230a	GTCCCCTATTTTCAGTGACCTA	ACCTTCATTTGGCACACAGAC
Region_230b	CTCTGCTTGGCAGTTTTCAT	CCAATTTCCCTCCCAAGTAT
Region_230c	ACCTTCAGTCCTCCTCTTTCAT	AGTTAATGGGCACAGGGATAC
Region_233a	TGTCCTCTGTCTACGGGTTACT	TTGCACAGCACAGTTAGGTATC
Region_233b	TATGTCTAAGGGAGGGGAAGAG	ATCCTAATAGGGTTAGGGCAAA
Region_233c	ACAGAGGCTGAGAGAGTTCAGA	ACAGGCTGAAGTCCTGTCTATG
Region_237	ATGCCAGACCTAGAGGATACAA	CTAGTTAAACCCTGCATGAGCA
Region_241	CTAGTGTTTGGAGCCAGATCAT	CAGAACCAGTATGGAGCAAGAT
Region_259a	CTAACCATTCTCTTGGGTTCTG	AGAAACCAGGGAAGGTACAGTT
Region_259b	TGCCTAGAACATGGAAGGTACT	AGGATGGGGAATAGGAGTAAGA
Region_264	GTGGGAAACAGCATCAAGTAGT	TGTCGAAAGTACAGCGGTTT
Region_272a	TATAGGGACCCAGCAGATTTC	GAAGCATATACGGCCTCAGTAG
Region_272b	CATTGCACCCACATTTTCTC	TCTATGAGAGTGTGAGCACTGG
Region_272c	CACCTACTTGTTCCTGCTAGATG	CCCATTCCATGAATTTCTGC
Region_bg_1	CACTGTCAAAAGTCTGTGAAGCTATT	AGGCCAAACGTTTTTCTCTTG
Region_bg_2	TAGGCGCAGCTTGTAGGACT	CAGGATTTGGGACAACTTGG
Region_bg_3	CTTGCACAATGCCTCACTCA	GAAAACACCAGCCACCAGAA

Table S4.

List of Abs

Company	Antibody
FACS Abs
BD Pharmingen	CD4 (RM4-5)
	CD25 (PC61)
	CD34 (RAM34)
	Sca1 (E13-161.7)
	CD71 (C2)
	B220 (RA3-6B2)
	CD3ε (145-2C11)
eBioscience	CD8a (Ly-2)
	CD44 (IM7)
	CD45.2 (104)
	CD16/32 (93)
	c-Kit (2B8)
	Ter-119 (11-5921)
	FCεR1 (MAR-1)
	Gr-1 (RB6-8C5)
	Thy1.2 (30-H12)
IP, ChIP, and IB Abs
Millipore	Mouse anti-Cdc6 (05-550)
	Mouse anti-CDT1 (04-1524)
Cell Signaling Technology	Rat anti-Prim1 (4725S)
Bethyl Laboratories	Rabbit anti-MCM5 (A300-195A)
Santa Cruz Biotechnologies	rat anti-GATA1 (sc-265)
	Goat anti-lamin B (sc-6217)
	Goat anti-Ldb1 (sc-11198)
	Goat anti-MCM3 (sc-9850)
	Mouse anti-MCM7 (sc-56429)
	Rabbit anti-PCNA (sc-7907)
BD Biosciences	Mouse anti-MCM2 (610700)
	Mouse anti-MCM6 (611622)
	Mouse anti-Orc2 (51-6875GR)
	Mouse anti-Pold1 (610972)
	Mouse anti-sc35 (556363)

The BTL-73 mouse anti-SCL/Tal1 antiserum was generously provided by D. Mathieu (Institut de Génétique Moléculaire, Montpellier, France).

Transient Replication Assay in Mammalian Cells.

The transient replication assay is adapted from Takeda et al. (43) and described in Supporting Information.

Flow Cytometry, Cell Cycle Analysis, and Cell Sorting

For DNA content staining, cells were incubated with the Hoechst 33342 (10 μg/mL) in DMEM (10 mM Hepes, pH 7.4) supplemented with 2% FCS for 30 min at 37 °C, and then washed and analyzed on a FACSAria cell sorter (BD Biosciences). Cell cycle was analyzed with FloJow software (www.flowjo.com). For synchronization experiments, MEL cells were purified according to cell size (forward scatter/side scatter profile) because small cells are in G0/G1 (Fig. S3E). Cells were seeded in six-well plates at 10⁵ cells per well and incubated at 37 °C for the indicated times. Cells were then harvested, fixed, and permeabilized with BD Cytofix/CytopermTM Fixation/Permeabilization Kit (BD Biosciences) before Hoechst staining (10 min). DN1 thymocytes were harvested from thymus of murine stem cell virus (MSCV) or LMO2 transplanted mice. Before cell sorting, CD4⁺ cells and CD8⁺ cells were depleted using magnetic beads (QIAGEN). The remaining fraction was stained with anti-CD44 and anti-CD25 Abs conjugated to allophycocyanin-cyanin 7 (APC-Cy7) or phycoerythrin-cyanin 7 (PE-Cy7), respectively, as well as Thy-PE, CD4–PE-Cy5, CD8-APC, and sorted using a FACSAria cell sorter. The DN1 subset was identified as the PE⁺PE-Cy5⁻APC⁻PE-Cy7⁻APC-Cy7⁺ population. DN cells were cocultured with stroma cell line OP9-DL1 in the presence of 5 ng/μL FLT3 and 5 ng/μL IL-7 (BD Biosciences).

Protein Extraction, Immunoprecipitation, ChIP, IB, Abs

TF-1 and MEL nuclear extracts were prepared as documented previously (36). Chromatin extracts of TF-1 cells were performed with or without cross-linking using formaldehyde at a final concentration of 0.5%. After 10 min, cross-linking was quenched by the addition of Gly to a final concentration of 0.125 M. Cells were harvested and washed with PBS, and protein extraction was performed by incubation with radioimmunoprecipitation (RIPA) buffer [10 mM Tris⋅HCl (pH 8.0), 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 0.1% deoxycholate]. Extracts were sonicated. For immunoprecipitation, 2 mg of protein was incubated overnight at 4 °C with 10 μg of Ab against LMO2 (AF2726; R&D Systems) in RIPA buffer. Protein complexes were immunoprecipitated by adding appropriately conjugated Pansorbin cells (Calbiochem) for 4 h at 4 °C, washed three times with RIPA buffer, and subjected to SDS/PAGE. After transfer on PVDF membranes (Millipore), proteins were visualized by IB using ECL Plus (GE Healthcare). Co-IP experiments in HEK293T were done as described in ref. 58. Abs used for ChIP and IB are listed in Tables S2 and S3.

RNA and ChIP Analysis

For ChIP, the same procedure used for TF-1 immunoprecipitation was applied, except that we used 500 μg of extracts with 1 μg of Ab. After precipitation, the immune pellet was recovered by addition of elution buffer [50 mM Tris⋅HCl (pH 8.0), 10 mM EDTA, 1% SDS], treatment with proteinase K, and phenol/chloroform extraction (36). For new replication initiation zones, we focused on 45 regions identified by Cadoret et al. (38) that showed an overlap within a 1-kb distance with 47 origins identified by Karnani et al. (41). We reasoned that these DNA regions are more likely to contain efficient origins in human cells. We designed 80 PCR primer pairs covering (i) the overlapping region when the two origins directly overlapped or (ii) each region, in addition to the “gap” region(s), when the origins were separated by a short DNA segment. After validation, these primer pairs were used to detect the presence of these origins in chromatin extracts from TF-1 cells immunoprecipitated with anti-MCM5 and anti-LMO2 Abs. For RT-PCR, total RNA was prepared from 50,000 cells as described previously (29). Primer sequences used are shown in Tables S2 and S3. Real-time quantitative PCR was done with PerfeCTaTM SYBR Green FastMixTM ROX (Quanta Biosciences) on a StepOnePlus RealTime PCR System (Applied Biosystems).

Transient Replication Assay in Mammalian Cells

This assay is adapted from Takeda et al. (43). The 5XUAS-CAT, the 5XUASmut-CAT, or the pBluescript plasmids were cotransfected in HEK293 cells with GAL4-VP16, GAL4-ORC2 (43), GAL4-LMO2, or GAL4-SCL expression vectors by calcium phosphate as described previously (36). pGEM4 was used to keep the total amount of transfected DNA equal. Cells (1–3 × 10⁶) were harvested 2–3 days later and subjected to Hirt extraction for isolation of extrachromosomal DNA. After 2–24 h at 4 °C, the extracts were centrifuged at 15,000 × g and the supernatants were extracted successively with phenol and with phenol-chloroform-isoamyl alcohol. The DNA was precipitated overnight at −80 °C with 50 μg of carrier tRNA and 0.7 vol of isopropanol. Half of the Hirt DNA was digested with Dpn1 (New England Biolabs) for 16 h at 37 °C, along with the original 5UAS-CAT or 5XUASmut-CAT plasmids used as Dpn1 digestion positive controls. One-tenth of the Dpn1 digestion was further treated with 5 units of Lambda-exonuclease in a final volume of 20 μL according to the manufacturer’s instructions. Dpn1-resistant or undigested plasmids from Hirt extraction, along with digestion control 5XUAS-CAT plasmids, were detected by quantitative PCR amplification using the forward TAC ACA TAC GAT TTA GGT GAC ACT ATA GAA and reverse GAT GAA TTC GAG CTC GGT ACC oligonucleotides. Threshold cycles (Cts) obtained from the Dpn1-digested samples were normalized with the Cts obtained with the corresponding undigested samples. Digestion control plasmids confirmed that Dpn1 followed by the Lambda-exonuclease treatment allowed us to cut 99% or more of the methylated 5XUAS-CAT plasmids. Dpn1-resistant 5XUAS-CAT plasmids obtained from GAL4-ORC2 and GAL4-LMO2 were expressed as fold above the Dpn1-resistant 5XUAS-CAT plasmids obtained from the GAL4-VP16 negative control.