Derivation of functional mature neutrophils from human embryonic stem cells

Embryonic stem (ES) cells can self-renew and differentiate into cells derived from all 3 germ layers (ie, ectoderm, endoderm, and mesoderm). Both mouse and human ES cells give rise to mature blood cells of granulocyte/macrophage, erythroid, and megakaryoid lineages in vitro. For blood cell induction from ES cells, the majority of investigators use a coculturing system with mouse stromal cells such as S17¹  or OP9.^2,3  Embryoid body (EB) formation is also a commonly used method to obtain starting materials for further culture.^4-6  Cell surface antigens, such as CD45 and CD34, and colony-forming ability are used as blood cell markers. Hemangioblasts, which have the capacity to differentiate into both endothelial and blood cells, have also been produced.^7-9  Only a few studies, however, have achieved specific and effective induction of mature blood cells from ES cells, particularly human ES cells (hESCs).¹⁰ 

Human ESC-derived blood cells are potentially useful as a replacement for donation-based blood for transfusion in clinical settings, for drug discovery screening, and for monitoring drug efficacy and toxicity. The current blood donation system for transfusion is incapable of providing enough granulocytes for patients with life-threatening neutropenia, although granulocyte transfusion could have a potentially significant benefit for a certain population of severely neutropenic patients.^11,12  Given the large amount of neutrophils required for transfusion,¹³  hESC-derived neutrophils might be a unique solution for this treatment demand. Therefore, the development of a highly effective method of neutrophil differentiation from hESCs is an important step for both clinical application of hESCs and granulocyte transfusion medicine.

The lack of an effective method for obtaining hESC-derived neutrophils with purity sufficient for functional analysis, however, has hampered progress in this field. Once neutrophils with a high purity can be generated from hESCs, it will be important to compare their activities with those of neutrophils produced in vivo, particularly given the fact that hESCs rarely give rise to hematopoietic stem cells in vitro,¹⁴  and thus, that hESC-derived neutrophils might not be a progeny of hematopoietic stem cells. Here, we developed an effective method of deriving mature neutrophils from hESCs through EB formation and subsequent coculture with OP9, and analyzed their morphologic and phenotypic characteristics. We then performed functional analyses of hESC-derived neutrophils in vitro, focusing on superoxide production, phagocytosis, bactericidal activity, and chemotaxis, in comparison with peripheral blood neutrophils (PB-Neu's) obtained from healthy subjects.

In all experiments using hESCs, we used KhES-3¹⁵  cells (a kind gift from Dr Nakatsuji; Kyoto University, Kyoto, Japan), which were maintained as previously described.¹⁶  Briefly, KhES-3 colonies were cultured on irradiated mouse embryonic fibroblasts in Dulbecco modified Eagle medium/F12 (Invitrogen, Carlsbad, CA) supplemented with 20% KNOCKOUT serum replacer (Invitrogen) and 2.5 ng/mL human basic fibroblast growth factor (Invitrogen). The culture medium was replaced daily with fresh medium. Colonies were passaged onto new mouse embryonic fibroblasts every 6 days. To induce the formation of EBs, KhES-3 colonies were picked up using collagenase, and cultured in suspension in nonserum stem cell medium that we previously used in a hematopoietic stem cell expansion protocol.¹⁷  After 24 hours, the colonies formed EBs, which were collected and cultured further for 17 days in Iscove modified Dulbecco medium (IMDM; Invitrogen) containing 15% fetal bovine serum (FBS), 1% nonessential amino acid (Invitrogen), 2 mM l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, and 0.1 mM 2-mercaptoethanol (ME) supplemented with cytokines (25 ng/mL bone morphogenic protein-4 [R&D Systems, Minneapolis, MN], 50 ng/mL stem cell factor [SCF; R&D Systems], 50 ng/mL Flt-3 ligand [R&D Systems], 50 ng/mL interleukin-6 [IL-6]/IL-6 receptor fusion protein [FP6; Kyowa Hakko Kirin, Tokyo, Japan], and 20 ng/mL thrombopoietin [TPO; Kyowa Hakko Kirin]).

OP9 cells (a kind gift from Dr Nakano; Osaka University, Osaka, Japan) were irradiated with 20 Gy and plated onto gelatin-coated 6-well tissue culture plates at a density of 1.5 × 10⁵/well. The next day, the EBs (incubated for 18 days after the initiation of suspension culture) were trypsinized and disrupted into single cells. Cells were suspended in the progenitor expansion medium (IMDM supplemented with 10% FBS, 10% horse serum [StemCell Technologies, Vancouver, BC], 5% protein-free hybridoma medium [Invitrogen], 0.1 mM 2-ME, 100 U/mL penicillin, 100 μg/mL streptomycin, 100 ng/mL SCF, Flt-3 ligand, FP6, and 10 ng/mL TPO and IL-3 [R&D Systems]) and plated onto the irradiated OP9 cells (day 0). Each well contained up to 5 × 10⁵ EB-derived cells. The culture medium was replaced with fresh medium on day 4.

On day 7 of the progenitor expansion phase, floating cells were collected, suspended with terminal differentiation medium (IMDM supplemented with 10% FBS, 0.1 mM 2-ME, 100 U/mL penicillin, 100 μg/mL streptomycin, and 50 ng/mL granulocyte colony-stimulating factor [G-CSF; Kyowa Hakko Kirin]), and transferred onto the newly irradiated OP9 cells. The culture medium was replaced with fresh medium on day 10. This terminal differentiation phase culture was continued for 6 or 7 days.

Human peripheral blood and bone marrow cells were obtained from healthy adult donors after obtaining informed consent in accordance with the Declaration of Helsinki. The institutional review board of the University of Tsukuba approved the use of peripheral blood neutrophils in this research. PB-Neu's were prepared as previously described.¹⁸  The purity of the neutrophils was greater than 90%, with the remaining cells mainly eosinophils. Neutrophils were suspended in Hanks balanced salt solution (HBSS; Invitrogen) containing 0.5% bovine serum albumin (BSA) and placed at 4°C. In some experiments, peripheral blood mononuclear cells (PB-MNCs) were collected from the intermediate layer after centrifugation with Lymphoprep (Axis-shield, Oslo, Norway). Bone marrow cells were directly centrifuged with Lymphoprep, and only mononuclear cells were collected. Bone marrow mononuclear cells (BM-MNCs) were used immediately for RNA extraction.

The morphology and granule characteristics of hESC-derived cells at the indicated days were assessed by Wright-Giemsa staining, comparing them with normal PB-Neu's. Myeloperoxidase and alkaline-phosphatase staining was performed using the New PO-K staining kit and alkaline phosphatase staining kit (MUTO PURE CHEMICALS, Tokyo, Japan). The prepared slides were inspected using an Olympus BX51 microscope equipped with a 100 × /1.30 UPlan objective lens (Olympus, Tokyo, Japan). Images were acquired with an HC-2500 digital camera and Photograb-2500 software (Fujifilm, Tokyo, Japan).

After 13 or 14 days culture, the population contained predominantly morphologically mature neutrophils, and was defined as hESC-Neu's. The hESC-Neu's and PB-Neu's were fixed in 2% paraformaldehyde/2.5% glutaraldehyde in 0.1 M phosphate buffered saline (PBS; Invitrogen) for at least 12 hours, and then postfixed in 1% osmium tetroxide in 0.1 M PBS for 2 hours. After fixation, samples were dehydrated in a graded ethanol series, cleared with propylene oxide, and embedded in Epon. Thin sections of cured samples were stained with uranyl acetate and Reynolds lead citrate. The sections were inspected using a transmission electron microscope, H7000 (Hitachi, Tokyo, Japan).

Total RNA was obtained from hESC-derived cells of indicated culture days, PB-Neu's, PB-MNC's, and BM-MNC's using an RNeasy mini kit (QIAGEN, Hilden, Germany), and cDNA was synthesized from each RNA sample using a random primer and SuperScript III (Invitrogen) following the manufacturer's protocol. Semiquantitative polymerase chain reaction (PCR) was performed as previously described.¹⁹  The sequence information of gene-specific primers used in reverse transcription (RT)–PCR and the PCR conditions is available upon request.

Surface antigens of hESC-derived cells harvested on the indicated days were analyzed by flow cytometry using fluorescence-activated cell sorting (FACS) Aria (Becton Dickinson Immunocytometry Systems, San Jose, CA). Fc receptors on the cells were blocked with PBS containing 2% FBS and FcR Blocking Reagent (Miltenyi Biotec, Bergisch Gladbach, Germany). Antigens were stained with either fluorescein isothiocyanate (FITC)–conjugated antihuman CD13, CD64, CD11b (Beckman Coulter, Fullerton, CA), or CD14 (BD Pharmingen, San Diego, CA) antibodies; phycoerythrin-conjugated antihuman CD16, CD32, CD33 (Beckman Coulter), CD11b, or CD45 (BD Pharmingen) antibodies; or allophycocyanin-conjugated antihuman CD15, CD117 (BD Pharmingen), CD34, or CD133 (Myltenyi Biotec) antibodies. The negative range was determined by referencing the fluorescence of isotype controls. Dead cells were detected using 7-amino-actinomycin D (Via-Probe; BD Pharmingen).

Neutrophils (hESC-Neu's and PB-Neu's) were suspended in IMDM with 0.5% BSA and incubated in 6-well plates at 37°C with 5% CO₂, with or without 50 ng/mL G-CSF. At the indicated time, neutrophils were harvested, stained with FITC-conjugated Annexin V and propidium iodide (PI) using an Annexin V-FITC Kit (Beckman Coulter), and analyzed by FACS Aria. Cells negative for both Annexin V and PI were judged as live cells.

Because the functions of hESC-Neu's are modified by G-CSF in the culture medium, we stimulated hESC-Neu's and PB-Neu's (PB-Neu(G+)'s) for 15 minutes at 37°C with 50 ng/mL G-CSF in the reaction medium. As a control, PB-Neu's without G-CSF stimulation (PB-Neu(G−)'s) were prepared. hESC-Neu's, PB-Neu(G+)'s, and PB-Neu(G−)'s were used for functional assays directly without changing the medium.

Neutrophil production of reactive oxygen species was detected by flow cytometry using dihydrorhodamine123 (DHR; Sigma-Aldrich, St Louis, MO) as described previously.^20-22  Briefly, 1 × 10⁵ hESC-Neu's, PB-Neu(G+)'s, or PB-Neu(G−)'s were suspended in 400 μL of the reaction medium (HBSS containing 0.5% BSA) per tube, and 3 tubes were prepared of each sample. Catalase (Sigma-Aldrich) at a final concentration of 1000 U/mL, 1.8 μL 29 mM DHR, and 100 μL 3.2 μM phorbol myristate acetate (PMA; Sigma-Aldrich) were added to 1 of the 3 tubes; either no DHR or only DHR was added in the other 2 tubes as controls. Reaction medium was added to bring the final volume up to 500 μL. After 15-minute reaction at 37°C, the samples were washed twice with ice-cold reaction medium, and suspended in 200 μL reaction medium. Rhodamine fluorescence from the oxidized DHR was detected using FACS Aria.

Phagocytosis and NBT reduction were visualized in a single set of experiments. Autoclaved Baker yeast was suspended in 0.5% NBT solution (0.5% NBT [Sigma-Aldrich] and 0.85% sodium chloride in distilled water) at a density of 1 × 10⁸/mL. A 5-μL aliquot of this yeast suspension was added to hESC-Neu's, PB-Neu(G+)'s, and PB-Neu(G−)'s at 2.5 × 10⁵ in 50 μL FBS. After 1 hour at 37°C, the samples were washed and stained with 1% safranin-O (MUTO PURE CHEMCALS) for 5 minutes. The samples were then washed twice and suspended in 100 μL PBS. A small aliquot of each sample was placed onto a glass slide and topped with a cover glass, and the number of ingested yeast cells and their change in color from brown to purple or black were examined using a microscope. Ingested yeast cells that changed color in the cells were counted as NBT-reaction positive, whereas those that were ingested but did not change color were counted as NBT-reaction negative. The phagocytosis rate was calculated as the percentage of neutrophils that contained one or more NBT-positive yeast cells. The phagocytosis score was calculated as the total number of positive yeast cells in 100 neutrophils. Only morphologically determined neutrophils were scored, excluding contaminating cells such as macrophages, the percentage of which was less than 15% of the total cells.

The bacterial killing assay was performed using Escherichia coli ATCC25922 as previously described²³  with some modifications. Briefly, 1 × 10⁸ colony-forming units (CFUs) of exponentially growing bacteria were suspended in 1 mL HEPES-buffered saline with 10% human AB serum (MP Biomedicals, Irvine, CA) and opsonized at 37°C for 30 minutes. Neutrophils (hESC-Neu's, PB-Neu(G−)'s, and PB-Neu(G+)'s) were suspended in HEPES-buffered saline with 40% human AB serum at a concentration of 5 × 10⁶/mL. The opsonized E coli was added to the suspension of hESC-Neu's and PB-Neu's, at a neutrophil/bacteria ratio of 2:1, or control medium. After 1-hour incubation, 50 μL of samples with and without neutrophils were diluted in 2.5 mL alkalinized water (pH 11) for lysis of neutrophils. Samples were further diluted with PBS, and duplicate aliquots were added to molten tryptic soy broth with 1.5% agar kept at 42°C, rapidly mixed, and plated on dishes. The CFUs were counted after overnight incubation.

Chemotactic ability was determined using a modified Boyden chamber method.²⁴  Briefly, 700 μL of the reaction medium (HBSS containing 0.5% BSA) with or without 10⁻⁷ M formyl-Met-Leu-Phe (fMLP; Sigma-Aldrich) was placed into each well of a 24-well plate, and the cell culture insert (3.0-μm pores; Falcon; Becton Dickinson, Franklin Lakes, NJ) was gently placed into each well to divide the well into upper and lower sections. Neutrophils were suspended in the reaction medium at 2.5 × 10⁶/mL and 200 μL cell suspension was added to the upper well, allowing the neutrophils to migrate from the upper to the lower side of the membrane for 90 minutes at 37°C. After incubation, the membranes were washed, fixed with methanol, stained with Carrazi hematoxylin (MUTO PURE CHEMICALS), and mounted on the slide glass. The number of neutrophils that migrated through the membrane from the upper to the lower side was counted using a microscope with a high-power lens (× 400) in 3 fields: 2 near the edge and 1 on the center. Only mature neutrophils were counted.

Results are expressed as mean plus or minus SD. Statistical significance was determined using a 2-tailed Student t test. Results were considered significant when P values were less than .05.

After initiating the suspension culture of EB-derived cells, small clusters of round-shaped cells appeared on the OP9 stromal layer around day 4. The morphology of the day-7 cells visualized with Wright-Giemsa staining suggested that they were myeloblasts and promyelocytes. On days 9 and 11, myelocytes and metamyelocytes were predominant, and on day 13 or 14, 70% to 80% of the cells appeared to be stab and segmented neutrophils (Figure 1A), with approximately 90% of the granulocytes at the metamyelocyte stage or later (Table 1). This finding indicated that hESC-derived cells differentiated into mature neutrophils by a process similar to physiologic granulopoiesis. The remaining cells appeared to be macrophages or monocytes, and cells of other lineages, such as erythroid or lymphoid cells, were not observed at any time during the culture. The number of total cells peaked around days 9 to 11, with an average 2.9-fold increase (range; 0.5- to 10.0-fold in 23 independent cultures) compared with the initial EB-derived cell number. The final yield of the cells on day 13 or 14 was 1.7-fold (range; 0.1- to 8.8-fold in 28 independent cultures). We attempted to further purify the hESC-derived mature neutrophils from the “hESC-Neu” population using density gradient methods, but higher purification could not be achieved without massively reducing the cell yield. We therefore used hESC-Neu's in the subsequent experiments.

Most (97.3% ± 1.5%) of the hESC-derived mature neutrophils defined by Wright-Giemsa staining were positive for myeloperoxidase, and the alkaline-phosphatase score of hESC-Neu's was 284 plus or minus 8.6 (Figure 1B). Under transmission electron microscopy, segmented nuclei and round cytoplasmic granules of hESC-Neu's appeared very similar to those in PB-Neu's (Figure 1C).

Some myeloid cell lines, such as HL-60, have abnormal biosynthesis of secondary granule proteins.^25,26  Thus, it is important to verify the biosynthesis of secondary granule proteins in hESC-Neu's. The levels of lactoferrin mRNA in hESC-derived cells at different stages were compared with those in PB-Neu's and BM-MNCs by semiquantitative RT-PCR (Figure 1D). Lactoferrin biosynthesis begins at the myelocyte stage and terminates by the beginning of the band stage.^25,27  Lactoferrin mRNA was not detected in PB-Neu's from some donors, but was detected in PB-Neu's from others. Human ESC–derived cells at various stages as well as BM-MNCs expressed lactoferrin mRNA. The expression level of lactoferrin mRNA in the hESC-derived cells was highest at day 10 of the induction culture and declined on days 13 and 14. These findings are consistent with the documented pattern of lactoferrin biosynthesis.

Surface antigen expression at each level of differentiation of hESC-derived cells was analyzed by flow cytometry (Figure 2). From days 7 to 13, the common blood cell antigen CD45 was expressed in almost all the cells. CD34, CD117, and CD133, cell surface markers on normal immature hematopoietic cells, were detected in a small population of the cells on day 7, but disappeared by day 10. Common myeloid antigens CD33 and CD15 were also highly expressed, whereas CD11b expression increased during the course of maturation. CD13 is also a common myeloid antigen, but its expression was observed in less than 20% of the cells on day 7 and did not subsequently increase. CD16 (Fcγ receptor (FcγR) III), which is expressed in neutrophils as well as natural killer cells, macrophages, and a small subset of monocytes,²⁸  was already expressed by day 7, and increased with maturation. This expression pattern of CD16 is consistent with that during normal neutrophil differentiation, although the proportion of CD16⁺ cells was lower than that of morphology-defined mature neutrophils on day 13. The ratio of CD32 (FcγRII)–positive cells increased as the differentiation stage advanced, and eventually reached 90%. CD64 (FcγRI) expression was greater than 80%, peaking on day 10, and the high percentage was maintained through day 13. CD14 was expressed in 20% to 25% of the cells on days 10 and 13.

In normal peripheral blood, both neutrophils and monocytes express CD15 and CD11b. In addition, mature neutrophils express CD16, whereas monocytes express CD14.^28,29  Detailed analysis on day 13 revealed that approximately 70% of CD15⁺ and CD11b⁺ cells were CD16⁺, and almost all CD15⁺ and CD16⁺ cells expressed CD11b (Figure 2Bi,ii). Given that 70% to 80% of the cells on day 13 were morphologically mature neutrophils (Table 1), it is likely that the majority of hESC-Neu's had CD15, CD11b, and CD16 expression patterns similar to PB-Neu's, although some hESC-Neu's did not express CD15 or CD16, particularly CD16.

CD32 is broadly expressed on myeloid cells, whereas CD64 is expressed only on monocytes but not on neutrophils in the peripheral blood.²⁸  In the bone marrow, CD64 expression is observed in a small population of myeloblasts, peaks at the promyelocyte, myelocyte, and metamyelocyte stages, and then diminishes, although a small proportion of the stab neutrophils still express CD64.^30,31  We confirmed that virtually no PB-Neu's expressed CD64 (data not shown). In contrast, almost all CD15⁺ and CD16⁺ hESC-Neu's expressed CD64 on day 13, indicating that both stab and segmented hESC-Neu's expressed CD64, because segmented neutrophils represented more than 50% of the cells on day 13 (Figure 2Biii; Table 1). Nearly 50% of CD15⁺ and CD16⁺ cells were weakly positive for CD14, in contrast to the negative expression of CD14 in steady-state PB-Neu's (Figure 2Biv). This aberrant expression of CD64 and CD14 in hESC-Neu's is similar to their positive expression on some of the neutrophils harvested from healthy donors who received G-CSF administration^32,33  and the neutrophils derived from bone marrow CD34⁺ cells in vitro by G-CSF stimulation.³¹ 

In the steady state, PB-Neu's have a short life span of approximately 24 hours, but this can be prolonged by G-CSF stimulation.³⁴  Some hESC-Neu's were already apoptotic at the time of harvest and therefore we focused on the nonapoptotic fraction of hESC-Neu's (Figure 3). In contrast to the PB-Neu's, which underwent apoptosis within 6 hours without G-CSF, consistent with previous reports,³⁴  a proportion of apoptotic cells among hESC-Neu's in the medium without G-CSF did not increase for up to 6 hours after the start of the culture. In addition, there were no differences between the cultures with and without G-CSF for up to 6 hours. After 6 hours, however, there was a more rapid decrease in nonapoptotic cells in hESC-Neu's without G-CSF than in hESC-Neu's with G-CSF, which resulted in a lower number of viable cells than hESC-Neu's with G-CSF at 24 hours, although the number of viable cells of hESC-Neu's without G-CSF was still higher than that of PB-Neu's without G-CSF.

Oxidative burst is an essential function of neutrophils when killing microorganisms, but an inappropriate burst sometime causes injury to the host tissue. We assessed the ability to convert DHR to rhodamine in hESC-Neu's and PB-Neu's using flow cytometry.²⁰  Because G-CSF, which could substantially affect the result, was used during the culture, we compared hESC-Neu's with PB-Neu(G+)'s and PB-Neu(G−)'s as described in “G-CSF stimulation prior to assessing neutrophil function.” When DHR was added to the neutrophil suspensions, rhodamine-specific fluorescence was detected in hESC-Neu's, and in PB-Neu(G−)'s and PB-Neu(G+)'s without PMA stimulation, indicating basal superoxide production without PMA stimulation in each neutrophil preparation (Figure 4). PMA stimulation increased rhodamine mean fluorescence intensity in hESC-Neu's, but to a lesser extent than in PB-Neu(G−)'s and PB-Neu(G+)'s. Consequently, the mean rhodamine fluorescence intensity after PMA stimulation was similar in hESC-Neu's, PB-Neu(G−)'s, and PB-Neu(G+)'s, suggesting that the maximum superoxide production is comparable between hESC-Neu's and PB-Neu's.

Neutrophils protect against infectious microorganisms by phagocytosing and subsequently killing them. These functions of hESC-Neu's and PB-Neu's were evaluated in an experimental system using NBT-coated yeast. Under the microscope, mature neutrophils could be easily distinguished from contaminating macrophages by the unique shape of their nuclei after 1% safranin-O staining (Figure 5A). NBT-coated yeast that had not been ingested had a red-brown color that began to change to purple or black, beginning at the periphery, and eventually became completely black, because the NBT coating on the yeast was reduced by neutrophils after phagocytosis. Thus, neutrophils that had phagocytosis and NBT-reducing ability could be easily identified. hESC-Neu's had a slightly lower phagocytosis rate than PB-Neu(G−)'s and PB-Neu(G+)'s (Figure 5B). The phagocytosis score, however, was not significantly different between hESC-Neu's and PB-Neu(G−)'s and PB-Neu(G+)'s (Figure 5C). The cells on day 8 of the culture, most of which were morphologically myeloblasts and promyelocytes, were rarely observed to phagocytose the yeast or reduce NBT if they had ingested the yeast, indicating that we observed phagocytosis and NBT reduction that was specific to mature neutrophils.

Because the hESC-Neu's had sufficient phagocytosing ability and superoxide production, we next investigated whether hESC-Neu's can kill bacteria. The bactericidal activity of hESC-Neu's and PB-Neu's was compared using E coli. When incubated with hESC-Neu's and PB-Neu(G−)'s and PB-Neu(G+)'s, the numbers of CFUs were similarly reduced to approximately 40% that of the control, indicating comparable bactericidal activity against E coli between hESC-Neu's and PB-Neu's (Figure 5D).

We compared chemotaxis of hESC-Neu's and PB-Neu's using a modified Boyden chamber method. After incubation with or without fMLP in the lower well, neutrophils had migrated from the upper side to the lower side of the membrane. Neutrophil migration without fMLP in the lower well was considered random migration. The number of neutrophils that migrated randomly was not significantly different between hESC-Neu's and PB-Neu(G−)'s, but PB-Neu(G+)'s showed significantly more random migration than the others (Figure 5E). The number of migrated cells increased in hESC-Neu's, PB-Neu(G−)'s, and PB-Neu(G+)'s when fMLP was added in the lower well. The increase in cell migration induced by chemotaxis to fMLP was calculated by subtracting the number of randomly migrated cells without fMLP from that of migrated cells with fMLP. There were no significant differences between hESC-Neu's and PB-Neu(G−)'s or PB-Neu(G+)'s in the net fMLP-induced chemotaxis.

We developed a specific and effective method for deriving mature neutrophils from hESCs, making it possible to analyze hESC-derived neutrophils in detail. hESC-derived neutrophils had characteristics similar to steady-state peripheral blood mature neutrophils in morphology and essential functions, although there were some differences in surface antigen expression.

Unfortunately, attempts to further purify the hESC-derived mature neutrophils from the hESC-Neu population by density gradient methods led to a massive reduction in cell yield. In the flow cytometric analysis, the mean intensity of hESC-Neu's in forward scatter was higher than that of PB-Neu's (data not shown), indicating that the size of morphologically mature neutrophils, comprising 70% to 80% of the hESC-Neu population, was larger than that of PB-Neu's. This finding indicates that the density of morphologically mature neutrophils in the hESC-Neu population was lower than that of PB-Neu's, which made it difficult to separate hESC-Neu's from other contaminating cells.

In this culture, we observed morphologically defined myeloblasts, promyelocytes, myelocytes, metamyelocytes, and, eventually, mature stab and segmented neutrophils, in this order, during the 13-day culture, which is similar to the granulocyte maturation process in bone marrow. The surface antigen expression pattern during differentiation was similar to that during normal granulopoiesis, with CD34 and CD117 expression on immature cells, and an increase in CD16 expression as differentiation advanced. Most hESC-Neu's expressed CD16, CD15, CD11b, CD33, and CD45. This pattern is consistent with normal PB-Neu's, but the percentage of CD16-expressing cells was lower than that of mature neutrophils determined by morphology. The lower CD16 expression level is documented in neutrophils derived in vitro from bone marrow CD34⁺ cells by stimulation with G-CSF, and is considered to be the effect of G-CSF on myeloid progenitors.³¹  G-CSF also induces CD64 and CD14 expression on mature neutrophils,^31,35  and these effects are also observed in vivo when G-CSF is administered to healthy volunteers.^32,33  Therefore, the G-CSF present in the culture from day 7 may have affected the progenitors and led to the relatively low expression of CD16 on hESC-Neu's and aberrant expression of CD64 and CD14 on CD15⁺ and CD16⁺ hESC-Neu's.

In the apoptosis assay, some hESC-Neu's were already apoptotic at the time of the harvest from the induction culture, but the proportion of apoptotic cells among hESC-Neu's in the medium without G-CSF did not increase for up to 6 hours after the start of the culture. There are 2 possible reasons for the difference in the rate of apoptosis. First, the hESC-Neu's were more heterogeneous than the PB-Neu's, as they comprised cells at different stages from incompletely differentiated cells such as metamyelocytes to maturation-completed and aged neutrophils. Relatively immature cells or unaged mature neutrophils in the hESC-Neu population might have a longer lifespan than PB-Neu's. Second, the effect of G-CSF used in the induction culture might continue even after the washout.

In the chemotaxis assay, the random migration of hESC-Neu's was almost the same as that of PB-Neu(G−)'s, but lower than that of PB-Neu(G+)'s, although hESC-Neu's were stimulated by G-CSF before the assay. The effect of G-CSF on the random migration of neutrophils is controversial; random migration increases in vitro when neutrophils are stimulated by G-CSF,³⁶  whereas neutrophils obtained from G-CSF–treated patients with nonmyeloid malignancies show decreased random migration and chemotaxis.^37,38  Our in vitro experiment with PB-Neu(G+)'s and PB-Neu(G−)'s replicated the former result. Nevertheless, hESC-Neu's showed relatively low random migration despite stimulation with G-CSF, while maintaining almost normal fMLP-induced chemotaxis. One possible reason for these differences might be the continuous stimulation by G-CSF; hESC-Neu's were stimulated from the myeloblast stage, and thus, it was expected that the characteristics of the hESC-Neu's were more similar to those of neutrophils from G-CSF–stimulated donors rather than to normal mature neutrophils.

The low yield of hESC-Neu's is a major obstacle to their functional analysis in animals, and further, to their potential use in drug screening and clinical applications. The number of hESC-Neu's produced was less than twice that of the input EB-derived cells. Recently, erythroid progenitor cell lines that could differentiate into functional mature red blood cells both in vitro and in vivo were established from mouse ESCs.³⁹  In that report, the starting number of ESCs required to establish one progenitor line was 5 × 10⁵, and transplantation of 2 × 10⁷ cells of the progenitor line could ameliorate anemia in mice by increasing the red blood cell count. Similar methods could be considered in the granulopoiesis from hESCs. Another potential method is to use more immature or proliferation-competent cells than the cells with which we initiated the induction culture. One candidate may be hematopoietic progenitors that emerge in saclike structures derived from ESCs. In a report using cynomolgus monkey ESCs,⁴⁰  EBs were created and subsequently subjected to adherent culture on a gelatin-coated dish. After 2 weeks, saclike structures emerged that contained hematopoietic precursors at various stages of myeloid lineage. The authors mentioned the possible existence of hemangioblasts, because endothelial cells could be produced from those precursors under different conditions. Others have also reported similar saclike structures containing hematopoietic precursors created from hESCs.¹⁰  In this paper, megakaryocytes were created from the inner cells, which were positive for hematoendothelial markers, such as CD34, CD31, vascular endothelial growth factor-receptor 2, and vascular endothelial-cadherin. These similar findings suggest that the cells in the saclike structures contain cells that are more immature than our EB-derived cells, and that the precursors inside the saclike structures have greater proliferation potency than our EB-derived cells. Because neither paper directly demonstrated the efficiency of mature blood cell production from monkey or human ES cells, however, the efficiency of producing neutrophils from our EB-derived cells should be compared with that from the saclike structure–derived cells.

We thank Dr Nakatsuji for providing the KhES-3, and Dr Nakano for providing the OP9 cells. We are grateful to Kyowa Hakko Kirin for providing TPO, FP6, and G-CSF, and to Kyokuto Pharmaceutical Industrial for the nonserum medium used in the EB formation. We also thank S. Ichimura for hESC culture.

This work was supported in part by a Grant-in-aid from the Japan Society of Promotion of Sciences (KAKENHI nos. 17390274, 18013012, 19390258, and 20015010); Research on Pharmaceutical and Medical Safety, Health and Labor Sciences Research Grants from the Ministry of Health, Labor and Welfare of Japan (H16-Iyaku-32); grants from the Astellas Foundation for Research on Metabolic Disorders; the Uehara Memorial Foundation; and the Sagawa Foundation for Promotion of Cancer Research (S.C.); and the Project for Realization of Regenerative Medicine (S.O.).

Contribution: Y.Y. and T.S. performed the experiments; K.H. designed the NBT-coated yeast cell-phagocytosis test; M.S.-Y., and K.K. assisted with interpretation of experiments and provided insightful comments; Y.Y. interpreted the data, made the figures, and wrote the paper; T.T., M.K., and S.O. advised on experimental design; S.C. provided critical reading of the paper; T.S. and S.C. designed the research.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Shigeru Chiba, Department of Clinical and Experimental Hematology, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8575, Japan; e-mail: schiba-tky@umin.net.