Reprogramming human B cells into induced pluripotent stem cells and its enhancement by C/EBPα

Abstract

B cells have been shown to be refractory to reprogramming and B-cell-derived induced pluripotent stem cells (iPSC) have only been generated from murine B cells engineered to carry doxycycline-inducible Oct4, Sox2, Klf4 and Myc (OSKM) cassette in every tissue and from EBV/SV40LT-immortalized lymphoblastoid cell lines. Here, we show for the first time that freshly isolated non-cultured human cord blood (CB)- and peripheral blood (PB)-derived CD19+CD20+ B cells can be reprogrammed to iPSCs carrying complete VDJH immunoglobulin (Ig) gene monoclonal rearrangements using non-integrative tetracistronic, but not monocistronic, OSKM-expressing Sendai Virus. Co-expression of C/EBPα with OSKM facilitates iPSC generation from both CB- and PB-derived B cells. We also demonstrate that myeloid cells are much easier to reprogram than B and T lymphocytes. Differentiation potential back into the cell type of their origin of B-cell-, T-cell-, myeloid- and fibroblast-iPSCs is not skewed, suggesting that their differentiation does not seem influenced by ‘epigenetic memory’. Our data reflect the actual cell-autonomous reprogramming capacity of human primary B cells because biased reprogramming was avoided by using freshly isolated primary cells, not exposed to cytokine cocktails favoring proliferation, differentiation or survival. The ability to reprogram CB/PB-derived primary human B cells offers an unprecedented opportunity for studying developmental B lymphopoiesis and modeling B-cell malignancies.

Introduction

Induced pluripotent stem cells (iPSCs) provide a unique platform to explore donor/patient-specific somatic cells for regenerative medicine, drug screening and disease modeling.¹ Although the most common source for human iPSC derivation is skin dermal fibroblasts, human iPSCs have been generated from a variety of somatic tissues including keratinocytes,² mesenchymal stem cells³ and hematopoietic stem/progenitor cells.^{4, 5, 6, 7} Generation of iPSC from human hematopoietic cells is an attractive option because they can be generated from peripheral blood (PB) cells, which are easily accessible through noninvasive methods, and from cord blood (CB) cells, which are young cells carrying minimal somatic mutations stored as large collections in public CB banks.⁸ To date, human iPSCs have been generated from CD34+ hematopoietic stem/progenitor cells^{4, 6, 7, 8} and also from T cells and myeloid cells.^{8, 9, 10} However, whether iPSCs may be induced from human normal or leukemic B cells remains a mystery.^{6, 11} Reprogramming CB/PB-derived B cells to pluripotency (iPSCs) will offer a valuable in vitro system to study cellular, molecular and epigenetic events underlying the physiology of B-cell lymphopoiesis and the pathogenesis of B-cell malignancies. B cells have been shown to be refractory to reprogramming,^{9, 10, 12} and B-cell-derived iPSCs have only been generated from B cells of mice engineered to carry doxycycline-inducible Oct4, Sox2, Klf4 and Myc (OSKM) lentiviruses in every tissue,¹² and from EBV/SV40LT-transformed lymphoblastoid cell lines.^{11, 13} Importantly, recent work has revealed that mouse B cells can be reprogrammed into iPSCs with high efficiency when the cells were pulsed with the CCAAT/enhancer binding protein-α (C/EBPα) before of their exposure to the reprogramming factors OSKM.¹⁴ This strongly suggests that biological rather than technical barriers underlie the inability to reprogram B cells to pluripotency. Here, we show for the first time that freshly isolated non-cultured human B cells, derived from both CB and PB, can be reprogrammed into iPSCs carrying complete VDJH immunoglobulin (Ig) gene monoclonal rearrangements, using non-integrative polycistronic Sendai Virus (SeV).^{15, 16} We also demonstrate that transient co-expression of C/EBPα with OSKM increases B-cell iPSC generation. Differentiation of B-cell-, T-cell-, myeloid- and fibroblast-iPSCs into B-cell, T-cell and myeloid cell fate revealed that iPSC differentiation potential does not seem influenced by the residual ‘epigenetic memory’ of the cell type of origin.

Materias and methods

CB and PB collection and isolation of CD19+ and CD19− cell populations

Independent umbilical CB samples (n=10) from healthy newborns and PB samples (n=5) from healthy donors (age 30–40 years old) were obtained from the Barcelona Blood Bank upon approval by our local Ethics and Biozahard Board Committee (ABR/JFJ/S-23). Mononuclear cells were isolated using Ficoll-Hypaque (GE Healthcare, Stockholm, Sweden). After lysing the red blood cells (Lysis solution, Cytognos, Salamanca, Spain), mononuclear cells were stained with anti-human CD19-biotin (eBioscience, San Diego, CA, USA) and magnetic-activated cell sorting (MACS)-enriched (purity>80%). MACS-enriched CD19+ cells were cultured overnight in RPMI+20% fetal bovine serum (FBS) supplemented with hIL7 (10 ng/ml; Preprotech, London, UK), and then further fluorescence-activated cell sorting (FACS)-purified for CD20 (anti-CD20-APC, BD Biosiences, San Jose, CA, USA) with the FACSAria Fusion flow sorter using ‘low-recovery high-purity’ sorting settings.¹⁷ Post sorting B-cell purity was consistently higher than 99.5% (Figure 1a).

Figure 1

Generation and characterization of mature B-cell iPSCs. (a) Representative flow cytometry plots of CD19+ and CD19− populations before and after FACS sorting (n=15). Of the sorted B cells, 99.6% were CD19+CD20+ (not shown). (b) Genome structure of the OSKM-miR-302-expressing SeV vector. (c) Schematic drawing of the strategy used to derive iPSCs from B cells. (d) Complete VDJH monoclonal rearrangements were present in all CB-derived B-cell iPSCs. Top panel shows polyclonal complete VDJH rearrangements from normal CB. Bottom panel shows a complete VDJH monoclonal rearrangement in a representative CB-derived B-cell iPSC. (e) Representative electropherogram of a B-cell iPSC showing the junction region of the PCR product sequence shown in (d) confirming a complete VDJH monoclonal rearrangement (IGHV1-69*01/IGHD3-9*01/IGHJ4*02). (f) Morphology and alkaline phosphatase staining of a representative B-cell iPSC. (g) Immunostaining for the pluripotency markers OCT4, NANOG, SOX2 and TRA-1-60 in a representative B-cell iPSC. (h) qRT-PCR for the pluripotency transcription factors OCT4, SOX2, NANOG and REX1. (i) Representative qRT-PCR demonstrating SeV elimination after 10 passages. The inset is a representative anti-SeV immunostaining showing a high infection (~85%) with SeV of CD19+ B cells 2 days after infection. (j) Diploid karyotype of a representative B-cell iPSC at p15. (k) Pyrosequencing revealing demethylation of OCT4 and NANOG promoters in representative B-cell and fibroblast iPSCs. (l) Teratoma analysis revealing three germ layer differentiation of a representative B-cell iPSC. α-sm actin, FoxA2 and β-III tubulin are mesoderm-, endoderm- and ectoderm-specific markers.

OSKM and C/EBPα Sendai Vector transduction of CB- and PB-derived CD19+ and CD19− cells

Immediately after sorting, between 1 × 10⁵ and 1 × 10⁶ fresh, non-stimulated CB/PB-derived CD19+/CD20+ and CB CD19− cells were infected with the tetracistronic defective and persistent SeV encoding OCT3/4, KLF4, SOX2 and c-MYC factors with the miR-302 target sequence (miROSKM-SeV) (Figure 1b) (multiplicity of infection=3) alone or in combination with the inducible C/EBPαER-SeV (multiplicity of infection=20) (custom-made from DNAVEC, Tsukuba, Japan) for 3 h at 37 ºC. Tetracistronic SeV was developed, generated, reconstituted and tittered as previously described in detail.^{15, 18} B cells co-infected with miROSKM-SeV and C/EBPαER-SeV were then half-split and plated onto irradiated mouse embryonic fibroblasts in RPMI+20% FBS+hIL7, and 24 h later, β-estradiol (or vehicle) was added to the medium at a final concentration of 100 nM to trigger C/EBPα activity in the nucleus of the infected cells.¹⁴ Then, the cells were changed to human embryonic stem cell (hESC) medium/MEF-conditioned media supplemented with 8 ng/ml basic fibroblast growth factor (Miltenyi, Bergisch Gladbach, Germany), generated as previously described.^{19, 20} β-Estradiol was maintained for two further days to ensure proper nuclear expression of C/EBPα. This medium was changed every other day and maintained until the iPSC colonies were picked. To avoid cell loss during medium changes, floating cells were harvested, centrifuged and plated back on top of the feeders. Identification of the first emerging hiPSC colonies (~12 days after SeV infection) and passage onto fresh feeders was performed mechanically as previously described.^{6, 21} By day 21, iPSC clones were Tra-1–60-stained to determine the reprogramming efficiency. The generated iPSC clones were maintained on irradiated mouse embryonic fibroblasts in hESC media^{6, 21} supplemented with 8ng/ml of basic fibroblast growth factor. The hESC media was changed every other day, and the cells were passaged using 1 mg/ml collagenase IV every 7–9 days.

iPSC characterization

Established iPSCs were fully characterized upon confirmation that they were transgene-independent. SeV elimination was determined by qRT-PCR as described.^{15, 18} Expression of pluripotency markers was performed by immunostaining (alkaline phosphatase, OCT4, NANOG, SOX2 and TRA-1–60) and quantitative RT-PCR (OCT4, NANOG, SOX2, REX1) using the antibodies and primers previously described (Supplementary Table S1).^{22, 23} G-banding karyotype, in vivo teratoma formation and immunocytochemistry for α-smooth muscle actin, FOXA2 and β-III tubulin, was performed as it has been extensively described by our group.^{21, 24, 25}

Bisulfite pyrosequencing of OCT4 and NANOG promoters

Bisulfite modification of genomic DNA was performed with the EZ DNA Methylation-Gold kit (Zymo Research, Irvine, CA, USA) following the manufacturer's instructions. The set of primers for PCR amplification and sequencing of NANOG and OCT4 were designed using the software PyroMark Assay Design (version 2.0.01.15; Qiagen, Hilden, Germany): Forward-NANOG (5′-IndexTermTAT TGG GAT TAT AGG GGT GGG TTA-3′), Reverse-NANOG (5′-[Btn]- IndexTermCCC AAC AAC AAA TAC TTC TAA ATT CAC-3′), and sequencing primer S-NANOG (5′-IndexTermATA GGG GTG GGT TAT-3′); Forward-OCT4_prox (5′- IndexTermGGG GTT AGA GGT TAA GGT TAG TG-3′), Reverse-OCT4_prox (5′-[Btn]- IndexTermACC CCC CTA ACC CAT CAC-3′), and sequencing primer S-OCT4_prox (5′-IndexTermGGG GTT GAG TAG TTT-3′); Forward-OCT4_dist (5′- IndexTermTTT TTG TGG GGG ATT TGT ATT GA-3′), Reverse-OCT4_dist (5′-[Btn]- IndexTermAAA CTA CTC AAC CCC TCT CT-3′), and sequencing primer S-OCT4_dist (5′-IndexTermATT TGT ATT GAG GTT TTG GA-3′).²⁶ Primer sequences were designed to hybridize with CpG-free sites to ensure methylation-independent amplification. PCR was performed with primers biotinylated to convert the PCR product to single-stranded DNA templates, using the VacuumPrep Tool (Qiagen). After PCR amplification, pyrosequencing reactions and methylation quantification were performed using PyroMark Q24 reagents, equipment and software, according to manufacturer’s instructions.

TCR and Ig gene monoclonal rearrangements

To determine the myeloid, T-cell or B-cell origin of the established iPSCs, TCR and Ig gene monoclonal rearrangements were analyzed. Genomic DNA from iPSCs and normal PB (polyclonal control) was isolated using standard methods. For Ig gene monoclonal rearrangements, complete VDJH and incomplete DJH rearrangements were amplified and identified using the BIOMED-2 framework 1-3 strategy.²⁷ For amplification of complete VDJH rearrangements, a set of family-specific primers of the FR1 and FR2 regions and one JH consensus primer were used in two multiplexed PCR reactions. Amplification of incomplete DJH rearrangements was performed in two different reactions using family-specific primers for DH1 to DH6 families, respectively, together with the consensus JH primer. The monoclonal nature of the rearrangements was analyzed by the identification of single amplification peaks by GeneScanning analysis following described criteria.^{28, 29} All products were sequenced as previously described.²⁸ Monoclonal PCR products were purified with ExoSap (USB Co, Cleveland, OH, USA) and directly sequenced in an 3500XL Genetic Analyzer (Applied Biosystems, Carlsbad, CA, USA) using BigDye Terminators with the v1.1 Cycle Sequencing kit (Applied Biosystems).²⁸ TCR rearrangements were also amplified and identified using also the BIOMED-2 strategy. TCR gamma (TCRγ) gene rearrangements were analyzed using the TCRγ primer set that identifies the majority of monoclonal T-cell populations.²⁷

Myeloid, B-cell, T-cell re-differentiation and spontaneous differentiation assays

Undifferentiated B-cell-, T-cell-, myeloid- and fibroblast-iPSCs were re-differentiated into myeloid cells, T cells and B cells following protocols extensively established. Briefly, for myeloid differentiation, confluent iPSCs were transferred into low-attachment plates and allowed to form embryoid bodies (hEBs) in differentiation media supplemented with hematopoietic cytokines (300 ng/ml stem cell factor (SCF), 10 ng/ml interleukin-3, 10 ng/ml interleukin-6 and 50 ng/ml granulocyte colony stimulating factor, 25 ng/ml BMP-4). At day 15 of hEB development, a single cell suspension was achieved using collagenase B followed by 10-min treatment with cell dissociation buffer, and by gentle pipetting and stained with anti-CD31-FITC, anti-CD45-APC, anti-CD34-PE and 7AAD (BD Biosciences). Live cells (7AAD-) were analyzed using a FACS Canto-II flow cytometer (BD Biosciences). Hematopoietic cells were identified as CD34+CD31+CD45− (hemogenic progenitors) and CD45+ cells. The myeloid clonogenic potential was determined by colony-forming unit assay by plating 50 000 d15 hEB cells into methylcellulose H4435 (Stem Cell Technologies, Vancouver, BC, Canada). Cells were incubated at 37 °C/5% CO₂ humidified atmosphere, and colonies counted after 14 days using standard morphological criteria.^{17, 19, 30, 31, 32, 33, 34}

For B- and T-cell differentiation, hiPSCs were harvested by 40 min collagenase IV/Dispase (Invitrogen, Carlsbad, CA, USA) pretreatment and seeded onto over-confluent OP9 stroma on gelatinized 10-cm dishes. Culture media consisted of αMEM supplemented with 10% FBS, 1% PenStrep, 100 μM monothioglycerol and 50 μg/ml ascorbic acid. After 10 days of coculture onto OP9 stroma, cells were harvested by 30 min digestion with collagenase IV (200 U/ml) followed by 15 min with TrypLE (Invitrogen) and CD34+ cells were MACS-purified using CD34 microbeads (Miltenyi) following the manufacturer's instructions. To induce B-cell differentiation, MS5 stroma was used.³⁵ Isolated iPSC-derived CD34+ cells were cocultured onto confluent MS5 monolayers at a density of ~5 × 10⁴ cells/10-cm dish in αMEM supplemented with 10% FBS, 1%PenStrep, interleukin-7 (20ng/ml), SCF (50 ng/ml) and Flt3L (50 ng/ml) (all from R&D Systems, Minneapolis, MN, USA). MS5/CD34+ cocultures were maintained for 30 days and then harvested by collagenase IV/trypLE digestion and stained with CD29-FITC, CD19-PE and CD45-APC (Miltenyi).³⁵ To induce T-cell differentiation, purified iPSC-CD34+ cells were cocultured on confluent monolayers of OP9-DLL4 in αMEM supplemented with 10% FBS, 1%PenStrep, 100 μM monothioglycerol, 50 μg/ml ascorbic acid, interleukin-7 (5 ng/ml), SCF (10 ng/ml) and Flt3L (5 ng/ml). OP9-DLL4/CD34 cocultures were split every 6–8 days and maintained for 30 days. Cocultures were then harvested by collagenases IV/trypLE digestion and stained with CD5-APC and CD45-PerCP-Cy5.5 (BD Biosciences).³⁶

For spontaneous differentiation, iPSCs/ESCs were harvested with Collagenase IV (1 mg/ml) and cell clumps were transferred into low-attachment plates for EB formation. EBs were maintained for 6 days in differentiation media (KO-DMEM plus 20% KO-serum replacement, 0.1 mM β-mercaptoethanol, 1 mM L-glutamine, 1% PenStrep, 1 mM sodium pyruvate and 1 × non-essential amino acids). Medium was changed every 2 days and EBs were collected after 6 days for RNA extraction and qRT-PCR. Primers used are shown in Supplementary Table S1.

Results

Generation of iPSC from CB- and PB-derived human primary B cells

In 10 independent experiments, between 1 × 10⁵ and 1 × 10⁶ highly purified (>99.5%) CB-derived CD19+/CD20+ B cells and CD19− non-B cells were infected with non-integrative tetracistronic SeV encoding OSKM with the miR-302 target sequence (Figures 1a and b).^{9, 15} Infected cells were cultured on irradiated mouse embryonic fibroblasts in MEF-conditioned media until the emergence of iPSC-like colonies which were plucked and expanded for further 15 days on irradiated mouse embryonic fibroblasts in hESC media into stable iPSC with bona fide hESC morphology (Figure 1c). From CB, a total of 154 (efficiency: 0.051%) and 33 (efficiency: 0.0013%) iPSC clones were generated from the CD19− and CD19+ cell fractions, respectively (Table 1), indicating that the CD19− fraction is reprogrammed into pluripotency with ~40-fold higher efficiency than the CD19+ cell fraction. Lymphocyte development involves sequential DNA genetic rearrangements of the T-cell receptor or Ig loci.^{37, 38} Thus, to investigate whether the iPSCs were derived from B, T or myeloid (non-B non-T) cells, we analyzed both TCR and Ig gene rearrangements using the BIOMED-2 strategy.²⁷ Three out of the 154 iPSCs generated from the CB-derived CD19− fraction tested positive for TCR Vγ-Jγ rearrangement (Supplementary Figure S1a) while the remaining 151 iPSCs were negative for both TCR and Ig rearrangements (Supplementary Figure S1a and Figure 1d) indicating that 151 iPSCs have originated from myeloid cells (0.05% efficiency) and 3 iPSCs from T cells (0.001% efficiency) (Table 1). Upon gradient centrifugation of mononuclear cells and CD19 MACS-depletion, the ratio of T-cell to myeloid cell was 1:1 (data not shown), confirming that myeloid cells are much easier to reprogram than T cells.

Table 1 iPSC induction efficiency from cord blood-derived B cells, T cells and myeloid cells

Within the CB-derived CD19+ fraction, 27 out of the 33 iPSCs tested positive for complete VDJH Ig monoclonal rearrangement (Figures 1d and e and Table 1) demonstrating that these iPSCs have originated from terminally differentiated CD19+ B cells, although at a low efficiency, 0.001%. B-cell iPSCs were consistently generated from all the individual CB assayed. The remaining six iPSCs tested negative for both Ig and TCR rearrangements. Because FACS purity was >99% (Figure 1a), these six iPSC lines seem the result of contaminating myeloid cells (Table 1). We then characterized several CB-derived B-cell iPSCs (Figure 1), T-cell iPSCs (Supplementary Figure S1) and myeloid iPSC (Supplementary Figure S2). They displayed hESC-like morphology (Figure 1f and Supplementary Figures S1b and S2a) and were positive for alkaline phosphatase (Figure 1f and Supplementary Figures S1b and S2a) and NANOG, OCT4, TRA-1-60 and SOX2 (Figure 1g and Supplementary Figures S1c and S2b). By qRT-PCR, all iPSCs expressed the pluripotency factors NANOG, OCT4, SOX2 and REX1 (Figure 1h and Supplementary Figures S1d and S2c), and by passage 10, they have completely eliminated SeV (Figure 1i and Supplementary Figures S1e and S2d), indicating they are bona fide transgene-free iPSCs. All iPSCs were karyotypically normal (Figure 1j and Supplementary Figures S1f and S2e), and consistent with the activation of endogenous pluripotency genes, reprogramming was accompanied by an extensive loss of CpG methylation at the OCT4 and NANOG promoters (Figure 1k and Supplementary Figures S1g and S2f).The most rigorous pluripotency test of human iPSC is the formation of teratomas in immunedeficient mice.²⁵ On subcutaneous injection into NSG mice, all iPSCs generated well-differentiated teratomas representing the three germ layers (Figure 1l and Supplementary Figures S1h and S2g).

For PB-derived mature B cells (n=5), highly purified (>99.5%) CD19+/CD20+ B cells were reprogrammed using the conditions and miROSKM-SeV as described above for CB B cells (Figures 1a and b). Under basal conditions, 10 iPSC clones were generated resulting in a reprogramming efficiency of 0.001%, very similar to that of CB B cells (Table 2 and Figure 2b). These PB B-cell iPSCs displayed hESC-like morphology (Figure 2c), had complete VDJH Ig monoclonal rearrangement (Figure 2d) and expressed (at both RNA and protein level) the pluripotency factors NANOG, OCT4, SOX2, REX1 and Tra-1-60 (Figures 2e and f).

Table 2 iPSC induction efficiency from peripheral blood-derived B cells in the presence or absence of C/EBPα

Figure 2

C/EBPα expression facilitates OSKM-mediated CB and PB B-cell reprogramming. (a) Schematic overview of the strategy used to derive iPSCs from CB and PB B cells in the presence of C/EBPα. (b) C/EBPα expression induces a 2.5-fold increase in iPSC generation (measured as Tra-1–16+ colonies) from PB B cells. Data are shown relative to colony number in the absence of C/EBPα. Error bars indicate s.d. (n=3). (c) Representative morphology of PB B-cell iPSCs and PB Bα-cell iPSCs. (d) Complete VDJH monoclonal rearrangements were present in both PB B-cell and PB Bα-cell derived iPSCs. (e) Representative immunostaining for the pluripotency markers OCT4-SOX2 and TRA-1-60-NANOG in PB B-cell iPSC and PB Bα-cell iPSC. (f) qRT-PCR for the pluripotency transcription factors OCT4, SOX2, NANOG and REX1 in PB B-cell iPSC and PB Bα-cell iPSC. PB-mononuclear cells were used as negative controls. Error bars indicate s.d. (n=3). (g) C/EBPα expression induces a 2.5-fold increase in iPSC generation (measured as Tra-1-60+ colonies) from CB B cells. Data are shown relative to colony number in the absence of C/EBPα. Error bars indicate s.d. (n=3). (h) qRT-PCR showing the exogenous expression of C/EBPα in B cells upon infection with the C/EBPαER-SeV. (i) qRT-PCR for the pluripotency transcription factors OCT4, SOX2, NANOG and REX1 in CB Bα-cell iPSC. CB-mononuclear cells were used as negative controls. Error bars indicate s.d. (n=3). (j) Representative morphology and immunostaining for the pluripotency markers OCT4, NANOG, SOX2 and TRA-1-60 in a CB Bα-cell iPSCs.

Transient expression of C/EBPα facilitates OSKM-mediated CB and PB B-cell reprogramming

We recently demonstrated that mouse B cells can be reprogrammed into iPSCs with high efficiency when the cells were pulsed with the myeloid transcription factor C/EBPα before exposure to the reprogramming factors OSKM.¹⁴ We thus hypothesized that C/EBPα may function as a ‘path breaker’ facilitating human B-cell reprogramming. CB and PB mature B cells were reprogrammed basically as described above but they were now co-infected with miROSKM-SeV and C/EBPαER-SeV and then maintained the first 72 h in the presence of β-estradiol (or vehicle) to trigger C/EBPα activity in the nucleus of the infected cells¹⁴ (Figure 2a). C/EBPα pulse increased ~twofold the efficiency of OSKM-mediated PB B-cell reprogramming (Figure 2b and Table 2). C/EBPα-mediated PB B-cell iPSCs (Bα-cell iPSCs) were characterized for morphology (Figure 2c), complete VDJH Ig monoclonal rearrangement (Figure 2d), and RNA and protein expression of the pluripotency factors NANOG, OCT4, SOX2, REX1 and Tra-1–60 (Figures 2e and f), resulting indistinguishable from those B-cell iPSCs generated in the absence of C/EBPα. Similarly, co-expression of C/EBPα and OSKM also enhanced ~2.5-fold the reprogramming efficiency of B cells from CB (Figure 2g). CB-derived B-cell iPSCs and Bα-cell iPSCs were also morphologically, molecularly and phenotypically indistinguible (Figures 2h and j). Addition of sodium butyrate, an HDAC inhibitor previously reported to facilitate iPSC generation,³⁹ did not display an effect on OSKM-mediated B-cell reprogramming when used either alone or in combination with C/EBPα (Table 2).

The differentiation potential of B-cell-, T-cell- and myeloid-iPSCs is not influenced by the cell type of origin

There is controversy on whether iPSCs retain residual 'epigenetic memory' of the donor tissue/cell type from which they were derived and display skewed differentiation potential.⁴⁰ Here, we next tested and quantified whether the differentiation potential of B-cell-, T-cell-, myeloid- and fibroblast-iPSCs is determined by the cell type of origin. Myeloid differentiation from the indicated B-cell, T-cell, myeloid- and fibro-iPSCs (n=3) was assessed by FACS (CD45+CD33+) and colony-forming unit potential using a hEB formation system^{17, 19, 30, 31, 32, 33, 34} (Figure 3a). As compared with B-cell-, T-cell- and fibroblast-iPSCs, myeloid-iPSCs displayed a slight trend towards the generation of more hematopoietic cells (P-value>0.1; Figure 2b) with higher colony-forming unit potential (P-value>0.1; Figure 2c). Among all the different iPSCs tested, the levels of B-cell (CD45+CD19+) differentiation on MS5-iPSC coculture (Figures 2d and e) and T-cell (CD45+CD5+) differentiation on OP9-DLL4-iPSC coculture (Figures 2d and f) ranged between 0.2–1% and 20–60%, respectively, with no trend of predisposed differentiation (‘epigenetic memory’) towards the cell type from which they were derived. Next, we evaluated more globally the developmental potential of the different B-cell-, Bα-cell-, T-cell- and myeloid-iPSCs. Upon hEB spontaneous differentiation, qRT-PCR was performed for mesoderm- (CXCR4, PDGRFα, PDGFRβ), ectoderm- (MAP2, PAX6) and endoderm-specific (SOX17, AFP, FOXA2) master factors (Supplementary Figure S3a). All iPSCs readily differentiated into tissues representing the three germ layers with no evident difference depending the cell type from which the iPSCs were derived (Supplementary Figure S3b). Altogether, these data indicate that re-differentiation back into the cell type of origin of B-cell-, T-cell-, myeloid- and fibroblast-iPSCs is not skewed, suggesting that their differentiation potential does not seem influenced by ‘epigenetic memory’.

Figure 3

Myeloid, B-cell and T-cell differentiation of B-cell, T-cell, myeloid- and fibroblast-iPSCs. (a) Schema of hiPSC hematopoietic differentiation system based on hEB formation and time points for FACS and colony-forming unit (CFU) analysis. (b) Hematopoietic potential of the different iPSCs indicated. Inset is a representative FACS analysis of hematopoietic cells (black dots). HEP stands for hemato-endothelial progenitors. Error bars indicate s.d. (n=3). (c) Myeloid potential measured as clonogenic capacity of the different iPSCs indicated. Error bars indicate s.d. (n=3). Right panels show representative CFU colonies (top) and flow cytometry (bottom) of the CFUs confirming the myeloid phenotype (CD33+). (d) Schema of hiPSC B-cell (OP9+MS5 cocultures) and T-cell (OP9+OP9-DLL4 cocultures) differentiation systems and time points for FACS analysis. (e) B-cell potential measured by FACS (CD45+CD19+) of the different iPSCs indicated. Error bars indicate s.d. (n=3). (f) T-cell potential measured by FACS (CD45+CD5+) of the different iPSCs indicated. Error bars indicate s.d. (n=3).

Discussion

To date, human primary B cells have been shown to be refractory to reprogramming.^{9, 10, 12} Here, we show for the first time that human CB/PB-derived B lymphocytes can be induced to pluripotency using a non-integrative tetracistronic miROSKM-SeV, and that the B-cell reprogramming process is enhanced by co-expression of OSKM with the ‘path breaker’ C/EBPα. Our systematic study using CB and PB demonstrates that miROSKM-SeV allows the generation of iPSC from primary myeloid, T-cell and B cells. Myeloid cells are the easiest to reprogram in contrast to a previous report describing higher reprogramming efficiency of pre-stimulated T lymphocytes as compared with myeloid cells.¹⁰ Our data reflect the actual intrinsic cell-autonomous reprogramming capacity of myeloid, B cells and T cells because we have avoided unbiased or skewed reprogramming by using freshly isolated uncultured primary cells, not exposed to hematopoietic cytokines favoring proliferation or differentiation. The efficiency of SeV infection was similar (~85%) in CD19+ and CD19− cells, indicating that SeV-mediated reprogramming efficiency is not associated to viral infection efficiency. Importantly, however, the tetracistronic miROSKM-SeV system was compared side-by-side with the monocistronic SeV system (four genes in four different plasmids). Only B-cell iPSCs were obtained when the tetracistronic miROSKM-SeV was used. In contrast, reprogramming efficiency of myeloid cells was similar upon delivering the OSKM factors in one or several vectors (data not shown), confirming that the delivery of all the reprogramming factors in one genome may be a requirement for B-cell reprogramming, likely by increasing the likelihood of simultaneous expression of the four factors within the same cell.

We found efficiencies of iPSC induction for myeloid and T cells similar to previous reports.^{7, 9, 10} Chou et al.⁴¹ found that by using an episomal reprogramming plasmid, the efficiency of iPSC generation is higher in CB than in PB. However, using miROSKM-SeV, we here report that CB- and PB-derived B cells are reprogrammed at a very similar efficiency, likely reflecting that potential reprogramming barriers such as more mature phenotype and lower proliferation index of activated circulating B lymphocytes may be circumvented with the use of a robust tetracistronic miROSKM-SeV. Despite successful reprogramming of CB/PB-derived B cells, the efficiency remains very low. Reprogramming involves a concomitant silencing of the transcriptional program specific of the somatic cell and activation of a pluripotent transcriptome, whereas mature B cells depend upon survival signals delivered to the cells by their antigen receptor (BCR).⁴² Thus, if the Ig/BCR locus rapidly switches off upon ectopic OSKM expression, then the B cells will probably not survive long enough to allow for the functional phenotypic conversion to pluripotency to occur, probably representing a major barrier in B-cell reprogramming.

C/EBPα is a master myeloid transcription factor that induces transdifferentiation of B cells into macrophages.⁴³ Earlier work in mouse showed that co-expression of C/EBPα with OSKM increases between 15- and 100-fold the reprogramming efficiency of murine B cells.^{12, 14} We reasoned that C/EBPα may also function as a ‘path breaker’ in human B cells, thus facilitating the reprogramming of human primary B cells. We demonstrated that C/EBPα increased 2.5-fold the reprogramming efficiency of both CB and PB B cells. The C/EBPα-mediated enhancement of B-cell iPSC generation is not as strong as that observed with mouse B cells,^{12, 14} likely reflecting species-specific differences in B-cell developmental stages. However, technical differences cannot be ruled out because the SeV biology target cells have to be simultaneously infected with both miROSKM-SeV and C/EBPαER-SeV, and the miROSKM-SeV used here was not inducible thus preventing us to perform an initial C/EBPα priming followed by OKSM induction. CB/PB B-cell iPSCs obtained upon C/EBPα exposure were characterized and resulted indistinguishable from those B-cell iPSCs generated in the absence of C/EBPα. Whether C/EBPα poises human B cells for rapid reprogramming by inducing the expression of the dioxygenase TET2 which, in turn, makes the chromatin of pluripotency genes more accessible by promoting demethylation of such genes by conversion of 5mC to 5hmC requires further investigation.

It has been suggested that iPSCs may retain a residual 'epigenetic memory' of the donor tissue/cell type from which they are derived, and therefore the donor cell type can influence the epigenome and differentiation potential of hiPSCs.^{40, 44} However, almost all of these studies have compared the differentiation potential of iPSCs developed from tissues/cell types with different developmental origin. Here, we analyzed whether the differentiation potential of iPSCs derived from distinct blood cell types (B cells, T cells and myeloid cells) from the same tissue (CB) and same donors is determined by the cell type of origin. Using robust direct differentiation protocols,^{30, 32, 35, 36} B-cell-, T-cell- and myeloid-iPSCs were re-differentiation into myeloid cells, B cells and T cells. In addition, more global developmental potential into mesoderm, ectoderm and endoderm was also assessed upon spontaneous hEB differentiation. Overall, no 'epigenetic memory'/predisposed differentiation towards the cell type of their origin was found, indicating that the developmental differences between CB-derived myeloid, B cells and T cells are not instructive enough as to influence the differentiation potential of the resulting hiPSCs. Residual ‘epigenetic memory’ is thought to be progressively lost upon cell passaging. In this study, all the iPSCs used for differentiation assays were at passage 20–30, when no traces of miROSKM-SeV were detected, and thus some degree of predisposed differentiation/’epigenetic memory’ at early passage cannot be ruled out. In addition, it should be taken into consideration that a limited number of iPSC lines from distinct healthy individuals were used in this study and therefore these data must be validated in a larger cohort of blood cell type-derived iPSCs from a larger variety of donors.

In sum, generation of iPSC from human hematopoietic cells from PB and CB is an attractive option because they are easily accessible through non-invasive methods, and in the case of CB, cells are young and harbor minimal somatic mutations. Reprogramming healthy donor and/or patient-specific PB-derived B cells to pluripotency will offer a valuable in vitro system to study cellular, molecular and epigenetic events underlying the physiology of B-cell lymphopoiesis and the pathogenesis of B-cell malignancies. This study should encourage further strategies to increase the reprogramming efficiency of normal and leukemic B cells for downstream applications such as disease modeling, drug screening and developmental immunology.