Mice were kept under standardized (temperature, 22-24°C; humidity, 45-65%; 12 h light/12 h dark cycle; free access to food and water) and specific pathogen-free conditions until sacrifice. For purification of mouse HSPCs, bone marrow cells were harvested from one to two female C57BL/6N mice (8-12 weeks old; body weight, 17-21 g) which were purchased from Japan SLC, Inc. (Hamamatsu, Japan), for each experiment (n=13 in total). All animal studies were approved by the Animal Care Committees of Mie University (Tsu, Japan).

G418 (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) and puromycin (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) were used at final concentrations of 1 mg/ml and 1 µg/ml, respectively, for drug selection.

The retroviral vectors used in this study, pMYs-IRES-Neomycin^r (pMYs-IN) and pMXsU6-Kusabira Orange (KO), were previously described (36,37). The pMYs-E2A-HLF-IN vector was also described previously (36). To produce the E2A-pre-B-cell leukemia transcription factor 1 (E2A-PBX1) (38) fusion fragment, a portion of PBX1 encoding amino acid residues 89-430 (39) generated by polymerase chain reaction (PCR) using cDNA derived from K562 cells (provided by Dr. Toshio Kitamura, The Institute of Medical Science, The University of Tokyo, Tokyo, Japan), was inserted into pMYs-E2A-HLF-IN to replace that of HLF. E2A-HLF mutants were prepared as previously reported (40). Mutants lacking the AD1 domain, AD2 domain, bZIP domain, and a part of the basic region, respectively, were generated by site-directed mutagenesis using PCR with the wild-type E2A-HLF construct as a template, followed by cloning in a series of pMYs retroviral vectors (37). The E2A-HLF mutants were as follows: i) ∆AD1, which lacks 426 bp (Met-1 to Gly-142); ii) ∆AD2, which lacks the 405 bp PvuII-NaeI restriction fragment (Leu-278 to Ala-412) in the E2A transactivation region; iii) ∆bZIP, which lacks 132 bp (Try-508 to Ala-551) in the bZIP domain of HLF; and iv) ∆509-518, which lacks 30 bp (Ala-509 to Ala-518) in the basic region of HLF. For the chromatin immunoprecipitation (ChIP) analysis, E2A-HLF was fused with the FLAG epitope tag at the N-terminus in pMYs-E2A-HLF-IN. PCR for construction was performed with Phusion High-Fidelity DNA Polymerase (New England Biolabs Inc., Ipswich, MA, USA), according to the manufacturer's protocol. Each insert fragment in the plasmid was validated by DNA sequence analysis. For the reporter assay, E2A-HLF and E2A-PBX1 cDNAs were subcloned into pcDNA3.1+ expression vector (Invitrogen; Thermo Fisher Scientific, Inc.), respectively.

Mouse HSPCs were purified as described (36). In brief, bone marrow mononuclear cells (BMMNCs) were prepared from 8- to 12-week-old C57BL/6N mice. Using a MACS cell separation system (Miltenyi Biotec, Inc., Auburn, CA, USA), Lin-depleted cells were isolated from BMMNCs, and c-Kit⁺Sca⁻1⁺Lin⁻ (KSL) cells were purified from Lin-depleted cells using a FACSAria operated with FACSDiVa version 6.1.3 software (BD Biosciences, San Jose, CA, USA).

Plat-E packaging cells (41) were plated at a density of 4.5×10⁵/ml and trans-fected the following day with retroviral constructs using Polyethylenimine 'Max' (Polysciences, Inc., Warrington, PA, USA), according to the manufacturer's protocol. Retroviral supernatants of transfected Plat-E cells were harvested 48 h post-transfection following two medium changes. KSL and immortalized cells were transduced with retroviruses using RetroNectin (Takara Bio Inc., Otsu, Japan) for 48-72 h at 37°C, as previously described (36).

Myeloid immortalization assays via serial replating were performed as previously described (36). In brief, every 5-7 days, colonies were counted, followed by replating of the harvested cells (1×10⁴ cells/dish) in methylcellulose culture medium MethoCult™ M3234 (STEMCELL Technologies, Inc., Vancouver, BC, Canada) supplemented with 25 ng/ml mouse stem cell factor (SCF) (Miltenyi Biotec, Inc.), 10 ng/ml each of mouse IL-3 (Miltenyi Biotec, Inc.), human IL-6 (Miltenyi Biotec, Inc.), and mouse granulocyte macrophage-colony stimulating factor (GM-CSF) (Miltenyi Biotec, Inc.).

To evaluate the effects of knockdown of Eya2 in the immortalized cells, the Eya2-depleted cells were plated in the same methylcellulose culture medium as that used in the immortalization assays. Relative colony-forming units (CFUs) were calculated as a percentage of the colony numbers compared with the corresponding controls (normalized to 100%) in each experiment following culture for 5-7 days.

An immunophenotypical analysis was performed using a FACSCalibur and BD Cell Quest Pro version 5.2.1 software (BD Biosciences), as previously described (42).

The target sequences against the Eya2 and luciferase genes were 5′-GTGTTTCAG AGACAATCAT-3′ (shE09) and 5′-GCCTTATGCCGCCAT CTTG-3′ (shE12), and 5′-GGCTATGAAGAGATACGCC-3′ (shLuc). The shE09 and shE12 were two out of twelve sequences (shE01-shE12) designed for Eya2 knockdown, which exerted the most efficient effects (35). To design a short hairpin structure, a loop sequence (5′-CTTCAAGAGAG-3′) was used (35). Short hairpin RNA (shRNA) sequence(s) against Eya2 or luciferase were cloned into the retroviral vector pMXsU6-KO (36). Each 9 µg DNA of pMXsU6-KO derivative was transfected into Plat-E cells (41) in a 10 cm dish to produce retroviruses, as described above. The filtered culture supernatant of Plat-E cells containing retroviruses 48 h post-transfection was used for transduction of E2A-HLF-immortalized mouse KSL cells (2.5×10⁵) for 48 h at 37°C using RetroNectin (Takara Bio, Inc.) in liquid culture containing SCF, IL-3, IL-6, and GM-CSF. The shRNA-transduced cells were then subjected to sorting by KO expression on the FACSAria operated by FACSDiVa version 6.1.3 software. The sorted KO⁺ cells were cultured for 5-7 days for a colony-forming assay as previously described (36).

ChIP was performed as previously described (36). In brief, the chromatin prepared from FLAG-tagged E2A-HLF-immortalized cells was precipitated using Dynabeads anti-Mouse IgG (Invitrogen; Thermo Fisher Scientific, Inc.) preincubated with a mouse monoclonal anti-FLAG (cat. no. M2; Sigma-Aldrich; Merck KGaA), a mouse monoclonal anti-RNA polymerase II (cat. no. CTD4H8; EMD Millipore, Billerica, MA, USA), or a mouse IgG1 antibody (BioLegend, Inc., San Diego, CA, USA). The purified DNA in the precipitates was subjected to quantitative polymerase chain reaction (qPCR).

Total RNA was extracted using TRI Reagent LS (Molecular Research Center, Inc., Cincinnati, OH, USA). RT was performed with random hexamers using SuperScript II reverse transcriptase (Invitrogen; Thermo Fisher Scientific, Inc.) as previously described (36). The qPCR analyses were performed using the KOD SYBR qPCR Mix (for ChIP products; Toyobo Life Science, Osaka, Japan) or PowerSYBR^® Green PCR Master Mix (for cDNA) on a StepOnePlus™ Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.) (36). For the RT-qPCR analysis, PCR was performed for 40 cycles (94°C for 15 sec, 57°C for 15 sec and 72°C for 34 sec) in a total volume of 12 µl containing appropriately diluted cDNA. Following quantification of the expression of samples using the quantification cycle (2^−ΔΔCT) method (43) and performing normalization relative to β-2-microglobulin (B2m), the relative expression was calculated. The sequences of the primers used (Eya1, Eya2, Eya3, Eya4, and B2m) have been previously described (35,36). For the ChIP-qPCR analysis, PCR was performed for 40 cycles (98°C for 10 sec, 60°C for 10 sec and 68°C for 34 sec) in a total volume of 20 µl. When the CT values of the ChIP products had been measured and normalized to those of the corresponding input samples using the 2^−ΔΔCT method, the percentages of samples relative to the input were calculated. The primer sets (Eya2c-4) for ChIP-qPCR were described previously (35). The promoter region of hemoglobin subunit β1 was used as a negative control. RT-PCR was performed as previously described (42). In brief, to detect the transcripts of E2A-HLF, standard PCR amplification of cDNA was performed for 30 cycles (98°C for 10 sec and 68°C for 40 sec) using LA Taq (Takara Bio Inc.). The primers specific for E2A-HLF were as follows: E2A S4, 5′-GATAGAAGACCA CCTGGACGAG-3′; E2A S2, 5′-GTGAGGACTACGGCA GGGAT-3′; HLF AS1, 5′-CCAGCTCCTTCCTCAAGTCAG-3′; and HLF ASx, 5′-gaaagaattcaCAGGGGCCCGTGCCTGG-3′ (small letters, non-related sequence). To detect transcripts of B2m, PCR amplification of cDNA was performed for 26 cycles (94°C for 30 sec, 60°C for 30 sec and 72°C for 30 sec) using Quick TaqR HS DyeMix (Toyobo Life Science). The primers specific for B2m were as previously described (44). Together with 1 kb DNA ladder (New England Biolabs, Inc.) as a DNA size marker, the PCR products were electrophoresed on a 1% Tris-Acetate-EDTA agarose (Nacalai Tesque, Inc., Kyoto, Japan) gel, subsequently stained with ethidium bromide (VWR International, Radnor, PA, USA), and were visualized with an ultraviolet transilluminator system, Gel Doc 2000 (Bio-Rad Laboratories, Inc., Hercules, CA, USA), according to the manufacturer's recommendation.

The expression of transgenes in transfected Plat-E cells was examined by western blot analysis, as previously described (45). In brief, the transfected cells were harvested with lysis buffer supplemented with protease inhibitor cocktail (Sigma-Aldrich; Merck KGaA) and 2 mM phenylmethylsulfonyl fluoride. The lysates were mixed with an equal volume of 2X SDS sample buffer and boiled for 5 min. Western blot analyses of the samples obtained from the cells transfected with pMYs-IN, pMYs-E2A-HLF-IN, pMYs-E2A-PBX1-IN, pMYs-∆AD1-IN, pMYs-∆AD2-IN, pMYs-∆bZIP-IN, and pMYs-∆509-518-IN were performed using mouse monoclonal anti-E2A (Yae) (at 1:1,000 dilution for 1 h at room temperature; cat. no. sc-416; Santa Cruz Biotechnology, Inc., Dallas, TX, USA) and anti-α-tubulin (at 1:10,000 dilution for 1 h at room temperature; cat. no. T-5168; Sigma-Aldrich; Merck KGaA) antibodies, followed by reaction with a horseradish peroxidase-conjugated goat anti-mouse IgG secondary antibody (at 1:5,000 dilution for 1 h at room temperature; cat. no. 330; Medical & Biological Laboratories Co., Ltd., Nagoya, Japan).

To generate pGL3-mP, a minimal promotor (mP) sequence (derived from pGL 4.23; Promega Corporation, Madison, WI, USA) was inserted between the NheI and HindIII sites in the pGL3-Basic Vector (Promega Corporation). The putative promotor region of Eya2 (-2219 to -226, relative to the ATG of the translation initiation site) was inserted upstream of the mP sequence in pGL3-mP to generate pGL3-E2pro3. pGL3-E2pro3CSmut with the sequence CTATTCTAGT instead of CTTACCTAGT, a putative DNA-binding sequence of E2A-HLF that is similar to the consensus sequence (CS) GTTACGTAAT (9), was generated by PCR-mediated mutagenesis. All constructs for appropriate insertions were confirmed by DNA sequence analyses. Human leukemic K562 cells (2.5×10⁶) were resuspended in 400 µl K-PBS (NaCl, 30.8 mM; KCl, 120.7 mM; Na₂HPO₄, 8.1 mM; KH₂PO₄, 1.46 mM) containing reporter (6.6 µg pGL3-E2pro3 or pGL3-E2pro3CSmut), equimolar amounts of effector (pcDNA-E2A-HLF, pcDNA-E2A-PBX1 or empty vector), and 0.5 µg internal control (phTK-RL; Promega Corporation) plasmid DNA. The reporter-to-effector molar ratio of the DNA was 2:1. These cells were electroporated at 170 V and 950 µF in a 4 mm cuvette using Gene Pulser Xcell (Bio-Rad Laboratories, Inc.), and were incubated in the culture medium for 48 h. Cells were harvested post-incubation and lysed in 100 µl lysis buffer (Promega Corporation). The activity of firefly and Renilla luciferases in each lysate was measured sequentially using a Dual Luciferase Assay System (Promega Corporation) on a luminometer (TD-20/20; Turner Designs, Sunnyvale, CA, USA), according to the manufacturer's protocol.

Analyzed data are presented as mean ± standard deviation of three independent experiments. Comparisons of bar charts between two groups were performed using an unpaired one-tailed Student's t-test. As for the statistical comparisons of more than two groups, the assumption of homogeneity of variance was first tested using Bartlett's test. In cases where the variances were not equal at the 0.05 level, statistical differences were analyzed using a Friedman test as a non-parametric analysis of variance, followed by Steel's test as a post hoc multiple comparison test. All statistical tests were performed using R software version 3.1.2 (https://www.r-project.org/).

KSL cells were sorted from Lin-depleted bone marrow cells, retrovirally transduced with the E2A-HLF fusion oncogene, and subjected to a myeloid immortalization assay. E2A-HLF transduction immortalized KSL cells (Fig. 1A–C). Since it was recently revealed that Eya2 is important for inducing aberrant self-renewal activity in HSPCs (35), the expression of the Eya family genes Eya1, Eya2, Eya3, and Eya4 was analyzed in E2A-HLF-transduced colony-forming cells. An RT-qPCR analysis revealed that Eya2 was exclusively highly expressed among these genes following serial replating (Fig. 1D). Although the Eya1 expression was increased compared with that of Eya2 in the first-round colonies, the Eya1 expression was much lower compared with that of Eya2 in the third-round colonies. Eya3 and Eya4 were not highly expressed in either the first- or third-round colonies. These results suggested that an E2A-HLF fusion product induced upregulation of Eya2 in HSPCs, which may be associated with aberrant self-renewal activity in E2A-HLF-immortalized HSPCs.

To investigate the molecular mechanism underlying the upregulation of Eya2 by E2A-HLF in HSPCs, mouse KSL cells were retrovirally transduced with E2A-HLF and its mutants (Fig. 2A) and subjected to myeloid immortalization assays. The immortalized cells exhibited an immunophenotype of myeloid lineage with high expression of c-kit (Fig. 2B). Initial plating showed similar colony morphologies and reflected the transduction efficiency. In serial replating, E2A-HLF-transduced cells yielded and maintained increased numbers of dense colonies, whereas E2A-HLF mutant-transduced cells formed loose colonies that were smaller in size compared with those with E2A-HLF at the third round (Figs. 2C and D, and 3A) and failed to be immortalized (data not shown). A western blot analysis indicated similar expression of each mutant protein to that of wild-type E2A-HLF (Fig. 3B). The expression of the chimeric transcripts of E2A-HLF and E2A-HLF mutants was detected by RT-PCR of total RNA extracted from the transduced colonies at the first- and third-round plating, respectively. RT-PCR using primers E2A S4 and HLF AS1 in E2A-HLF, ∆AD1 and ∆509-518 generated 497, 497 and 467 bp products, respectively, while that using primers E2A S4 and HLF ASx in ∆bZIP generated a 426 bp product, and that using primers E2A S2 and HLF AS1 in ∆AD2 generated a 696 bp product (Fig. 3C).

∆AD1 and ∆AD2 mutants lack each transactivation domain of E2A to abolish the transactivation ability while retaining DNA-binding and dimerization abilities, while ∆bZIP lacks the basic region and the leucine zipper domain of HLF to abolish DNA-binding and dimerization abilities, and ∆509-518 lacks a part of the basic region of HLF to abolish DNA-binding ability (Fig. 2A). Wild type E2A-HLF transduction immortalized KSL cells in serial colony replating (Figs. 1C, and 4A and B). Consistent with a previous report demonstrating that the two transactivation domains of E2A and the leucine zipper dimerization domain of HLF are essential for the transforming potential of the E2A-HLF in NIH3T3 cells (12), ∆AD2- and ∆bZIP-transduced cells failed to form colonies in serial replating. In addition, ∆509-518-transduced cells also failed to form colonies (Fig. 4B). ∆AD1-transduced cells formed some colonies at the third round, although they lost the ability at the fifth round (data not shown). Transduction of E2A-PBX1, a fusion oncogene generated by t(1;19)(q23;p13) (38), also induced immortalization of KSL cells (Fig. 4B). RT-qPCR analyses demonstrated that Eya2 was highly expressed in the transduced cells with wild-type E2A-HLF or E2A-PBX1. The mutant ∆AD1- and ∆AD2-transduced cells exhibited elevated Eya2 expression, albeit to a lesser extent compared with the wild-type. However, ∆bZIP- or ∆509-518-transduced cells showed no induction of Eya2 expression (Fig. 4C). It is noteworthy that the colony-forming ability of E2A-HLF and its mutants roughly correlated with the level of Eya2 expression. These results suggested that the two AD domains of E2A are required for complete activity of Eya2 transactivation, while the DNA-binding and dimerization abilities of HLF are indispensable for Eya2 transactivation in KSL-derived cells.

To corroborate the involvement of Eya2 in myeloid immortalization by E2A-HLF, Eya2 was depleted in E2A-HLF-immortalized cells using shRNA expression (Fig. 5). The expression of E2A-HLF chimeric transcripts was not affected following shRNA transduction according to RT-PCR analysis of total RNA extracted from shRNA-transduced cells (Fig. 5B). shE09 and shE12 exhibited effective depletion of endogenous Eya2 expression in the E2A-HLF-immortalized cells (Fig. 5C), and Eya2 knockdown by shE09 and shE12 reduced the clonogenicity to ~40 and 70% of that in the control at the first round of plating, respectively (Fig. 5D). It was not possible to perform rescue experiments on Eya2-knockdown HSPCs that were immortalized with E2A-HLF by forced expression of Eya2 without the shRNA target sequence, likely due to the genotoxic stress induced by Eya2 overexpression, as the forced expression of Eya2 resulted in markedly reduced colony-forming efficiency in E2A-HLF-immortalized cells (data not shown).

It was previously observed that, in Plzf-mediated immortalization, Plzf binds to the putative Eya2 promoter region around exon 1c of Eya2 (35), in accordance with public data (46) from ChIP sequencing. Therefore, a ChIP-qPCR analysis of FLAG-tagged E2A-HLF-immortalized HSPCs with the primers around exon 1c was performed (Fig. 6). The analyses indicated increased binding of E2A-HLF to the Eya2 promoter region in the immortalized cells, accompanied by RNA pol II binding signals (Fig. 6A). These results suggested that E2A-HLF drives the aberrant expression of Eya2 in HSPCs by promoter binding. Subsequently, the effect of E2A-HLF expression on Eya2 promoter activity was confirmed using a reporter assay. The luciferase assay in K562 cells demonstrated that E2A-HLF and E2A-PBX1 activated the reporter gene expression through the Eya2 promoter region (Fig. 6C), and transactivation through the Eya2 promoter region was significantly reduced by the mutation of the putative E2A-HLF binding consensus sequence (Fig. 6B and D). Taken together, these findings suggested that E2A-HLF upregulates Eya2 transcription through the E2A-HLF binding consensus sequence.