Immunogenic Dendritic Cell Generation from Pluripotent Stem Cells by Ectopic Expression of Runx3

Pluripotent stem cells (PSCs), including embryonic stem (ES) cells and induced PSCs cells, provide an inexhaustible source for cell replacement and adoptive immune cell therapy because of their unlimited self-renewal activity and broad differentiation capacity. PSC-derived functional cells can be generated through directed differentiation using well-defined protocols (1). However, it is still challenging to steer the differentiation of PSCs to adult-like cells because the end products often represent embryonic-type or immature cells with limited activity. For example, human ES cell–derived RBCs readily expressed the embryonic and fetal globins (ε and γ), but the adult β-globin protein was barely detected in these cells (2). Similarly, the gene expression signature of PSC-derived insulin-producing cells mimicked that of fetal pancreatic tissue rather than the adult β cells (3). In addition, immaturity of the sarcoplasmic reticulum and diminished inotropic response to hormonal stimuli were detected in murine ES cell– or induced PSC–derived cardiac cells (4). These findings suggest that embryonic developmental programs are readily activated in PSC-derived ex vivo–differentiated cells; however, these regulatory networks usually do not guarantee the production of fully active mature cells. For proper maturation, further steps are needed that are unknown or missing from the existing standard in vitro–differentiation protocols.

Cell differentiation and commitment are governed by lineage-determining and stimulus-activated transcription factors (5). These master regulators are gradually induced during the embryonic development, and distinct factors modulate cell fate specification during the various stages of differentiation. In this study, we examined the transcription factor network in stem cell–derived differentiated immune cells. GM-CSF–dependent dendritic cell (DC) development was selected as a differentiation model because this cell type can be generated from adult stem cells and ES cells (6–13). Moreover, numerous transcription factors were described that control the commitment and specification of DCs (14, 15). In addition, in vitro–generated DCs were applied in cell therapy–based clinical trials to provoke anticancer immune responses (16). Therefore, ex vivo DC manufacturing has an immediate biotechnological application. DC generation from PSCs is a promising approach; however, PSC-derived DCs often exhibited a suboptimal T cell–activation capacity (6, 11, 13). Consistent with this notion, in this study we found that ES cell–derived DCs represented less mature cells compared with adult stem cell–derived DCs. Furthermore, our gene expression analysis revealed that three DC-affiliated transcription factors (IRF4, SPI-B, and RUNX3) were poorly expressed in ES cell–derived DCs (ES-DCs). Remarkably, improved DC maturation with enhanced chemotactic activity was detected on Runx3-instructed ES-DCs, suggesting that reintroduction of a missing transcription factor can greatly enhance the immunogenicity of PSC-derived APCs.

Mouse ES cells were maintained on a mitomycin C (Merck Millipore, Darmstadt, Germany)–treated mouse embryonic fibroblast layer in knockout DMEM (Life Technologies, Carlsbad, CA) with 15% FBS qualified for ES cells (Biochrom, Cambridge, U.K.) and 1000 U/ml leukemia inhibitory factor (Merck Millipore). ES-DCs were differentiated with a GM-CSF–dependent OP9 coculture method, as described (11) with minor modifications. In brief, OP9 cell density was set to 100,000 in a T25 flask 1 d before the experiment. Coculture was started by adding 100,000 harvested ES cells to the OP9 stromal cell layers; cells were cocultured for 5 d in α-MEM containing 20% FBS (Life Technologies). Half of the medium was replaced at day 3. At day 5, cells were harvested and reseeded onto fresh OP9 layers. OP9 cell density was set to 40,000 cells per well in a six-well plate 1 d before the experiment. A total of 200,000 5-d differentiated ES cells was added to the OP9 layers. Cells were cultured in α-MEM/20% FBS medium containing 50 ng/ml GM-CSF (PeproTech, Rocky Hill, NJ) and 50 μM 2-ME (Sigma, St. Louis, MO) for six additional days. At day 11, floating and loosely adherent cells were harvested and further cultured for 8 d in RPMI 1640 medium (Sigma) containing GM-CSF (50 ng/ml) and 2-ME (50 μM) without OP9 cells. In the case of inducible ES cell–derived cells (Irf4, Spi-B, or Runx3), 10-d differentiated CD45⁺ cells were sorted and further cultured for 9 d in RPMI 1640 medium containing GM-CSF (50 ng/ml) and 2-ME (50 μM). To induce DC maturation, medium was replaced with fresh RPMI 1640 on day 18, and ES-DCs were treated with 100 ng/ml LPS.

Bone marrow (BM) cells were obtained from 12-wk-old male C57BL/6 or 129S1 mice. The femurs and tibiae were removed, cleaned of all connective tissue, and placed on ice in 2 ml of PBS. The ends of each femur and tibia were clipped to expose the marrow. For BM-derived DC (BM-DC) differentiation, 500,000 freshly isolated BM cells were cultured for 9 d in RPMI 1640 medium containing 10% FBS (Life Technologies), GM-CSF (50 ng/ml), and 2-ME (50 μM) in six-well tissue culture plates. Half of the medium was replaced every 3 d. To induce DC maturation, medium was replaced with fresh RPMI 1640 on day 8, and BM-DCs were treated with 100 ng/ml LPS.

RNA was prepared from mouse splenic cells with TRI Reagent (MRC, Cincinnati, OH), and cDNA was generated with a Transcriptor High Fidelity cDNA Synthesis Kit (Roche, Basel, Switzerland). The coding sequences of mouse Spi-B and Irf4 were amplified from splenic cell–derived cDNA using a FastStart High Fidelity PCR System (Roche), and the PCR products were subcloned into the pDONR221 plasmid. The Runx3-carrying Gateway master vector (pENTR223.1; clone ID: FLH481113.01X) was purchased from the DNASU plasmid depository. The obtained Gateway entry clones were recombined to a modified p2Lox plasmid (17, 18) containing Gateway destination sequences using the Gateway Cloning System (Life Technologies). For ES cell transfection, 5 μg of the targeting constructs was electroporated into ZX1 (19) ES cells with the Neon Transfection System (Life Technologies). Inducible cassette-exchange recombination was used to insert the selected open reading frames into a euchromatic site on the X chromosome (17, 18). ES cell colonies were selected in 300 μg/ml G418 containing ES cell medium, picked on day 8, and expanded. At least five independent colonies per transgene were expanded and characterized.

Splenic T cells were isolated and purified from male BALB/c mice using a Pan T Cell Isolation Kit II (Miltenyi Biotec, Bergisch Gladbach, Germany). The purified T cells were used as responders. For allogeneic MLRs, 10³ or 10⁴ ES-DCs as stimulators were cocultured with 10⁵ responders in wells of 96-well round-bottom culture plates for 5 d. BrdU was added during the last 12 h of the culture. At the end of the culture, half of the cells were centrifuged onto a 96-well plate, and the incorporation of BrdU was measured with a BrdU Cell Proliferation Assay Kit (Merck Millipore), according to the manufacturer’s recommendations.

RNA was isolated from ES cell– or BM-derived cells with TRI Reagent (MRC), from which cDNA were reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Life Technologies). Quantitative PCR was performed using a real-time PCR system (LightCycler 480; Roche), as described previously (20). In brief, 40 PCR cycles were run at 95°C for 12 s and 60°C for 30 s using TaqMan Gene Expression Assay primer-probe sets (Life Technologies). TaqMan assay IDs are shown in Supplemental Table I. The comparative Cycle threshold method was used to quantify transcripts, and the expression level was normalized to β-actin. All PCR reactions were done in triplicate with one control sample in which reverse transcriptase was omitted during cDNA synthesis.

Whole-cell extract of 15 μg of protein was separated by electrophoresis in 10% polyacrylamide gel and then transferred to a polyvinylidene difluoride membrane (Merck Millipore). Membranes were probed with anti-RUNX3 (GTX12343; GeneTex, Irvine, CA) or anti-IRF4 (sc-6059; Santa Cruz Biotechnology, Dallas, TX) polyclonal Abs and then reprobed with anti-GAPDH mAb (AM4300; Thermo Fisher Scientific, Waltham, MA).

Supernatants collected from ES-DCs or BM-DCs were evaluated for production of cytokines (IL-1β, IL-6, and TNF-α) using DuoSet ELISA kits (R&D Systems, Minneapolis, MN), according to the manufacturer’s instructions. Detection limits for the kits are 15.6 pg/ml (IL-6 and IL-1β) and 31.25 pg/ml (TNF-α). For cytokine analysis, cell culture medium was replaced with fresh medium 24 h before harvesting the ES-DC or BM-DC supernatants.

FITC–dextran 70 (Sigma; average molecular mass 70 kDa) was used to measure mannose receptor–mediated endocytosis. Immature ES-DCs or BM-DCs were incubated with 1 mg/ml FITC–dextran (1.5 × 10⁵ DCs in 24-well plate) for 1 h at 37°C (control at 0°C), and uptake was analyzed by flow cytometry. Phagocytosis was assessed by the cellular uptake of latex beads (Sigma; carboxylate modified, mean diameter 1 μm). Immature ES-DCs or BM-DCs were incubated with latex beads (1.5 × 10⁵ DCs + 1.2 × 10⁶ latex beads in a 24 well-plate) for 2 h at 37°C (control at 0°C), and uptake was quantified by flow cytometry.

DC migration toward CCL19 and CCL21 was assessed using Transwell migration assays. Chemotaxis was tested using 24-well plates and inserts with 5-μm pores (Corning-Sigma). A total of 600 μl of RPMI 1640 media with or without chemokines CCL21 (500 ng/ml; R&D Systems) and CCL19 (500 ng/ml; R&D Systems) was placed in the lower chamber. Thereafter, 2 × 10⁵ LPS-treated ES-DCs or BM-DCs were placed in the Transwell inserts (volume 100 μl) and allowed to migrate through a polycarbonate mesh at 37°C. After 4 h, cells that migrated to the lower chamber were collected, and cell number was determined by acquiring events for a fixed time period (3 min) using a constant flow rate with flow cytometry.

Cells were analyzed and sorted with a FACSAria III (BD Biosciences, San Diego, CA). Live cells were gated based on a forward scatter (FSC)/side scatter profile to eliminate dead cells or cell debris. For FACS sorting, ∼150,000 Flk1⁺ or 150,000–500,000 CD45⁺ cells were sorted. The conjugated Abs CD45-FITC (30-F11), CD11b-PE (M1/70), CD135-PE (A2 F10.1), CD11c-allophycocyanin (HL3), MHC2-FITC (I-A/I-E; 2G9), CD80-allophycocyanin (16-10A1), and CD86-allophycocyanin (GL1) were obtained from BD Biosciences. F4/80–Alexa Fluor 488 (BM8) Ab was purchased from eBioscience (San Diego, CA), and CD115-PE (AFS98) was from BioLegend (San Diego, CA). Data analysis was performed with BD FACSDiva 6.1.3 software (BD Biosciences).

All data are presented as mean ± SD of the mean. Significant differences between mean values were evaluated using a two-tailed unpaired Student t test.

To accelerate our understanding of DC differentiation, we investigated the efficiency of DC generation from mouse ES cells and assessed the phenotype of the obtained products. There are a few protocols for directed DC differentiation from murine PSCs (11–13, 21, 22). We applied a GM-CSF–driven OP9 coculture method that was established by Senju et al. (11). In brief, mouse ES cells were cocultured on OP9 stromal cells for 5 d; thereafter, the obtained differentiated cells were transferred to new OP9 cell layers and cultured for an additional 6 d in the presence of GM-CSF. At day 11, the loosely adhered and floating cells were transferred to new tissue culture flasks and cultivated for 8 d in the presence of GM-CSF using feeder cell–free tissue culture conditions. To characterize these ex vivo–generated immune cells, we compared the cell surface expression of ES-DCs with BM-DCs. Fig. 1A showed that CD45 and CD11b were readily detected on both cell types; however, we consistently obtained lower expression of CD11c on ES-DCs. Next, we investigated the expression of CD135 and CD115, receptors for Flt3L and M-CSF, which are necessary for the differentiation of conventional DCs and monocyte-derived cells, respectively (23). Importantly, ES-DCs and BM-DCs contained <1% CD135⁺ (FLT3) cells. In contrast, 40–60% of ES-DCs and BM-DCs expressed CD115, a monocyte marker (Fig. 1B). These data suggest that ES- and BM-derived cells resemble the CD11b⁺ monocyte-derived inflammatory DC type (24); these monocytic cells are ontogenetically unrelated to FLT3-dependent conventional DCs. In addition, our flow cytometric analyses revealed that both cell types had a rather similar immune phenotype, although ES-DCs tend to express less CD11c.

Next, we investigated the maturation/activation capacity of these ex vivo–differentiated cells. Unexpectedly, maturation markers (CD80 and MHC class II [MHCII]) were moderately induced upon LPS administration in ES-DCs: <25% of the cells were double positive for MHCII/CD80. In contrast, >50% of the BM-DCs were double positive for MHCII/CD80 (Fig. 1C). In addition, we found that LPS-exposed ES-DCs exhibited a heterogeneous CD86 expression; in contrast, the majority of LPS-activated BM-DCs were CD86⁺. These results suggest that ES-DCs represent a distinct subset of myeloid cells that possess an impaired maturation capacity. It is worth mentioning that in our study ES-DCs were derived from the E14 ES cell line (genetic background 129/Ola); in contrast, BM cells were isolated from C57BL/6 animals. To prove that the altered maturation capacity and CD11c expression are independent of the genetic background, ES-DCs derived from the B6 IA2 ES cell line (genetic background C57BL/6) were also investigated. Again, impaired expression of CD11c/MHCII and CD86 was detected (Supplemental Fig. 1), demonstrating that ES-DCs tend to express less CD11c and represent less mature cells. To further characterize these ex vivo–generated DC subsets, we investigated the transcription factor constitution of these cells.

Tissue-specific transcription factors are important regulators of cellular function; in addition, some of them have a direct role in lineage commitment (25). To characterize the ES-DC–specific transcription factor profile, we quantified the RNA expression of 17 DC/macrophage-specific transcription factors (Batf3, Bcl-6, Egr1, Egr2, Id2, Ikzf1, Irf2, Irf4, Irf8, Maf, Mafb, Relb, Runx3, Pu.1/Sfpi, Tcf4, Spi-B, and Zbtb46) in ES-DCs and their progenitors. These selected genes had well-established roles in regulating DC or macrophage development (14, 15).

Expression data were extracted from the starting pluripotent ES cells and the 5-, 11-, and 19-d differentiated cells. Of note, upon differentiation, the emerging cells at days 5 and 11 represent a mixed population; therefore, the putative DC progenitors were purified by cell sorting (Supplemental Fig. 2A). Our quantitative mRNA transcript analysis indicated that more than half of the investigated transcription factors (Egr1, Egr2, Id2, Ikzf1, Irf2, Irf8, Maf, Pu.1/Sfpi1, and Tcf4) exhibited a relatively high expression in the 19-d differentiated ES-DCs (Supplemental Fig. 2B). Remarkably, Id2, Irf8, and Pu.1 were the most abundant transcripts among the tested genes. In contrast, seven transcription factors were barely detected in ES-DCs (Batf3, Bcl6, Irf4, Mafb, Relb, Spi-B, and Zbtb46). Interestingly, those factors that were firmly detected at day 19 usually exhibited detectable transcript levels at day 11, implying that the expression of these DC-specific transcription factors was established during the early stage of myeloid development. Altogether, these RNA profiling data indicated that several classical DC/macrophage-specific transcription factors were expressed in ES-DCs; however, other DC-specific genes were barely detected in these cells. It is important to note that our transcription factor selection included conventional and plasmacytoid DC (pDC)-specific genes (e.g., Batf3, Spi-B, and Tcf4). Therefore, the low transcript level of these factors was predictable and expected in this FLT3-independent GM-CSF–driven DC-differentiation model.

To further define the gene expression signature, mRNA levels of the selected transcription factors in ES-DCs were directly compared with BM-DCs. Interestingly, among the highly expressed genes only Maf showed altered expression in ES-DCs; the transcript levels of the other eight genes were similar in both cell types (Fig. 2A). These results suggests that a similar set of transcription factors participated in the differentiation and maintenance of GM-CSF–dependent myeloid DCs, regardless of cell origin.

In contrast to the abundantly expressed transcripts, overt alterations were encountered when mRNA levels of the moderately expressed genes were examined. Five genes exhibited lower expression in ES-DCs than in BM-derived cells (Fig. 2B). Of note, the E14 mouse cell line (genetic background 129/OLA) was used for ES-DC generation, whereas BM-DCs were derived from C57BL/6 animals. To prove that the observed alterations are independent of the genetic background, DCs were differentiated from additional stem cell sources. ES-DCs were derived from R1 (genetic background 129X1/SvJ × 129S1) ES cells, and BM-DCs were obtained from 129S1 animals. Importantly, four genes showed significantly altered expression in this independent sample set: Spi-B, Irf4, and Runx3 had lower expression in ES-DCs than in BM-DCs, whereas higher Maf transcript levels were detected in ES-DCs (Fig. 2C).

Next, we evaluated the protein expression of Irf4 and Runx3, whose RNA transcripts exhibited a relatively high expression in BM-DCs. In agreement with the RNA profiles, we observed detectable levels of IRF4 and RUNX3 proteins from BM-DCs samples, but these proteins were barely detected in ES-DCs (Fig. 2D, 2E). In conclusion, most of the tested myeloid/DC-specific transcription factors exhibited similar expression patterns in ES- and BM-derived DCs. However, three transcription factors were underexpressed in ES-DCs, suggesting that upregulation of these factors might modify the differentiation and function of these immune cells.

In light of this altered gene expression, we probed the effects of the underexpressed transcription factors (Spi-B, Irf4, and Runx3) in developing ES-DCs using isogenic expression screening. We applied a genetically modified mouse ES cell line (ZX1), into which a gene of interest can be inserted by Cre-mediated recombination (17–19). In these cells, the investigated genes can be induced in response to tetracycline or doxycycline (Tet-on system). This inducible system was used previously to modulate ES cell–derived blood, skeletal muscle, or thyroid development via induction of lineage-specific transcription factors (26–28).

To evaluate the early effects of Irf4, Spi-B, or Runx3 on myeloid blood cell development, the inducible ES cell clones were differentiated for 11 d with the OP9 coculture method. The selected transgenes were turned on during the second stage of differentiation (days 5–11). Two independent ES cell clones per transgene were tested in all cases. Unexpectedly, much fewer CD45/CD11b double-positive cells were detected upon the forced expression of Irf4 (Fig. 3A), suggesting that this transcription factor exerted a detrimental effect on the early stage of ES-DC development. In contrast, more CD45/CD11b⁺ cells were differentiated upon activation of Spi-B or Runx3 (Fig. 3B, 3C). These results imply that ectopic expression of Spi-B or Runx3 enhances the GM-CSF–driven myeloid differentiation in ES cell–derived progenitors.

We next examined the effects of these three factors on fully differentiated ES-DCs. In this set of transgenic experiments, transcription factors were turned on between days 5 and 19. In addition, cells were treated with LPS at day 18 to stimulate DC maturation/activation. It is important to mention that, in contrast to the previous experiments (Figs. 1, 2), CD45⁺ cells were sorted at day 10, and these purified myeloid cells were differentiated further until day 19. We found that this sorting step was necessary to recover the Runx3-expressing hematopoietic progenitors because these cells tend to adhere firmly to the OP9 layers at days 10–11. First, we tested the impact of Irf4 on the 19-d differentiated ES-DCs. Notably, only a few live cells were detected from 200,000 CD45⁺ sorted cells in the presence of doxycycline (data not shown). The poor yield precluded a detailed phenotypic analysis, and this result indicates that Irf4 exerts a general inhibitory effect on ES-DC development.

We then assessed the phenotype of the Spi-B– and Runx3-programmed APCs. To evaluate the maturation of the final differentiation products, cell surface expression of CD80 and MHCII were determined in the presence or absence of LPS. As expected, augmented CD80/MHCII⁺ cell formation was obtained upon LPS stimulation. However, overexpression of Spi-B failed to modify the maturation capacity of ES-DCs, because the frequency of MHCII/CD80⁺ cells was unaltered in the presence of doxycycline (Supplemental Fig. 3). In striking contrast, an elevated percentage of MHCII/CD80⁺ cells was detected in Runx3-instructed DCs upon LPS treatment (Fig. 4). Moreover, we observed a distinct MHCII-expressing subpopulation of ES-DCs in the presence of RUNX3, even without LPS administration (Fig. 4A). These results suggested that the sustained expression of Runx3 potentiates the maturation capacity of ES-DCs; however, this MHCII/CD80 expression level was still lower compared with BM-DCs (Fig. 4B). To further address the maturation ability of LPS-dependent DCs, we examined the expression of CD86, a well-established DC maturation marker, on Runx3-instructed ES-DCs. In line with MHCII/CD80 upregulation, more CD86⁺ cells were obtained in LPS-treated Runx3-activated ES-DCs (Fig. 5). This finding confirmed that Runx3 positively regulates the maturation/activation capacity of ex vivo–differentiated ES-DCs.

Finally, we examined the functional characteristics of Runx3-instructed ES-DCs. First, we assessed the endocytic activity of these transgenic cells by two methods: engulfment of latex beads for detection of phagocytosis and internalization of FITC-dextran, which is mainly taken up by mannose receptor–mediated endocytosis. Our flow cytometric analysis revealed that immature (non–LPS-treated) ES-DCs efficiently accumulated latex beads after 2 h of incubation and that Runx3 induction did not influence the latex beads uptake capacity of these cells (Fig. 6A). Furthermore, BM-DCs possessed a comparable phagocytic activity (latex beads uptake capacity). In contrast, induction of Runx3 negatively modulated the FITC-dextran uptake capacity of ES-DCs; furthermore, BM-DCs also exhibited lower FITC-dextran uptake compared with control ES-DCs (Fig. 6A). These results suggest that immature ES-DCs had a profound phagocytic- and receptor-mediated endocytic activity that was not modified or negatively regulated, respectively, by Runx3.

We next tested the cytokine-production capacity of ES-DCs using ELISAs. Remarkably, LPS-treated ES-DCs released high amounts of IL-6 and TNF-α; in addition, IL-1β was readily detectable in these mature cells (Fig. 6B). Of note, a similar cytokine production potential was obtained with or without doxycycline treatment (only a nonsignificant reduction in IL-1β secretion was observed in Runx3-activated ES-DCs). Furthermore, comparable cytokine levels were detected in supernatants derived from BM-DCs. These findings indicate that our GM-CSF–driven inflammatory DC subsets have a strong and universal ability to secrete proinflammatory cytokines. In addition, we conclude that, in contrast to the well-established surface maturation markers, proinflammatory cytokine production by ES-DCs was not altered by Runx3.

Next, we evaluated the CCL19/CCL21-dependent migratory ability of mature ES-DCs using Transwell-migration assays. After 4 h of incubation, transmigrated cells were enumerated by flow cytometry. Interestingly, ES-DCs migrated very poorly in the absence of doxycycline treatment, even in the presence of chemokines (CCL19 plus CCL21). In contrast, Runx3-activated ES-DCs transmigrated efficiently in the presence of chemokines (Fig. 7). To compare the potency of ES- and BM-derived cells, we also assayed the migration potential of BM-DCs. Interestingly, these cells exhibited a stronger migration, even in the absence of chemokines. We conclude that Runx3-activated ES-DCs possess an enhanced migratory capacity although it is still inferior to that of their BM-derived counterparts. Finally, the T cell–activation capacity of ES-DCs was evaluated by MLRs using DCs as stimulators. Consistent with the enhanced maturation capacity of Runx3-expressing cells, we observed an elevated T cell proliferation upon doxycycline treatment. Of note, we also tried to test the T cell–priming capacity of BM-DC samples; however, some of the absorbance values were out of the recommended linear range. In agreement with the elevated expression of MHCII (Fig. 4B), this observation suggests that BM-DCs possess an even higher T cell–activation capacity.

Collectively, these results indicate that Runx3-programmed ES-DCs exhibited an elevated migratory capacity and possessed a superior T cell–priming activity, although this potential is still lower compared with the abilities of BM-DCs. These functional data, together with the phenotypic profile, support the conclusion that enforced expression of Runx3 improves ES-DC immunogenicity.

Novel generation of DC vaccines must build on the increased knowledge of cell manipulation, including the efficient production of various DC subsets ex vivo. GM-CSF–instructed BM-DCs, which were used in this study as control cells, are considered monocytic inflammatory DCs (29). Consistent with previous findings (11, 12, 21), we confirmed that ES-DCs resembled this nonconventional inflammatory DC subset, because the final differentiation products were CD11b⁺, and they efficiently produced proinflammatory cytokines upon LPS treatment. However, we also observed apparent differences between ES-DCs and BM-DCs. For example, lower expression of CD11c was detected in ES-DCs; in addition, ES-DCs exhibited impaired expression of several maturation-specific markers upon stimulation with LPS. It is worth mentioning that a similar phenotype was obtained whether DCs were differentiated from 129/Ola (E14) or C57BL/6 (B6 IA2) animal-derived ES cells, suggesting that these alterations are independent of the genetic background of the starting mouse pluripotent cells.

Of note, most previous studies suggested that ES-DCs are functionally equivalent to adult stem cell–derived DCs. In contrast, we observed that ES-DCs exhibited a distinct phenotype, and they had an impaired maturation ability compared with BM-derived cells. Importantly, we tested several ES cell lines, and we always observed an impaired maturation capacity (without transgene induction). It is possible that minor differences in the protocols have an impact on the final cell products. Moreover, in earlier studies (6, 11, 13), ES-DCs often exhibited a lower T cell–activation potential compared with the adult counterparts, suggesting that impaired immunity is a general characteristic of PSC-derived DCs.

The prominently distinct phenotype of ES-DC prompted us to examine the gene expression signature of DC-specific transcription factors in both ES- and BM-DCs. A handful of DC/macrophage-affiliated transcription factors was recently identified and characterized (14, 15, 30); however, the contribution of these master regulators to the development of PSC-derived DCs has not been investigated. Our focused analysis revealed that the expression of three DC-affiliated transcription factors (Spi-B, Runx3, and Irf4) was impaired in mouse ES-DCs compared with BM-DCs. These factors were already implicated in DC development. Irf4 is a prominent transcriptional regulator of CD11b⁺ conventional DCs (31–34). In addition, GM-CSF–driven BM-DC differentiation is Irf4 dependent (35). Unexpectedly, our analysis failed to support the central role of this factor in ES cell–derived cells, because the baseline expression of this gene was rather low in ES-DCs. In addition, we observed impaired myeloid blood cell and DC development upon the forced expression of this transgene in ES-DC progenitors. It is worth mentioning that Irf8 is an abundant transcript in ES-DCs; therefore, our observations raised the possibility that Irf4 induction might interfere with the putative Irf8-driven ES-DC developmental program.

In contrast to Irf4, Spi-B exerted a positive effect on myeloid blood cell development. Enhanced production of CD45/CD11b⁺ cells was obtained after the upregulation of this transcription factor during the early stage of differentiation. Interestingly, it was reported that Spi-B is highly expressed in pDCs, and gene-silencing studies revealed that this factor was necessary for human pDC development (36, 37). In this study, we failed to obtain plasmacytoid characteristics after the overexpression of Spi-B in ES-DCs (data not shown). This finding is consistent with a previous gain-of-function study that suggested that Spi-B itself was unable to promote pDC differentiation in the presence of Id2 in human thymic progenitor cells (38). In addition to pDC development, Spi-B was implicated in the early primitive myeloid differentiation in Xenopus (39). Congruent with this observation, our results indicate that this factor positively modulates early myeloid DC development in ES cell–derived progenitors; however, we found that the fully differentiated ES-DC phenotype remained unaltered in the presence of this transcription factor.

Our gain-of-function analysis uncovered that Runx3 also exerted a positive effect on early myeloid development. More importantly, this transcription factor enhanced the maturation capacity of ES-DCs, because we obtained a higher percentage of MHCII/CD80⁺ cells in the presence of Runx3. Moreover, Runx3-instructed ES-DCs exhibited an increased migratory capacity and a superior T cell–priming activity. It was reported that Runx3 is expressed in mature BM-DCs and mediates their response to TGF-β (40). Moreover, Runx3 stimulated the differentiation of a specific epidermal DC subset (Langerhans cells) and rescued BM-derived Langerhans cell development, even in the absence of Pu.1 (41). In addition, it was recently described that Runx3 was strongly expressed in CD11b⁺ splenic DCs, and cell-specific gene-targeting models uncovered that the CD11b⁺ DC development was compromised in Runx3-deficient animals (42). Furthermore, it was suggested that MHCII expression is directly modulated by Runx3 (42). Our results are consistent with this finding because increased MHCII positivity and expression were observed in ES-DCs in the presence of this transcription factor. However, the Runx3-dependent effect is not confined to the regulation of MHCII in our ES cell–derived APCs; we also observed augmented expression of CD86, suggesting that Runx3 generally improves the maturation capacity of ES-DCs.

Notably, although Runx3 endowed ES-DCs with enhanced immunogenicity, BM-DCs still contained a higher percentage of MHCII/CD80⁺ and CD86⁺ cells, and they exhibited a stronger migratory capacity than Runx3-induced ES-DCs. These findings indicated that Runx3-instructed ES-DCs remained inferior compared with BM-DCs. These results inspire us to probe additional factors, along with Runx3, to further boost the immunogenicity of Runx3-instructed PSC-derived immune cells in the future.

In conclusion, our phenotypic analyses revealed that ES-DCs had compromised migratory and maturation potentials. However, overexpression of Runx3 was sufficient to impart ES-DCs with enhanced maturation, chemotactic, and T cell–activation capacities. These findings suggest that lineage-determining transcription factor–based cellular programming is a rational strategy to improve DC development from ES cell–derived progenitors. We expect that a similar gene-induction strategy will be applicable to human cells, and these results pave the way for the future clinical application of PSC-derived DCs for adoptive immune cell therapy.

We thank Iren Mezo for excellent technical help and Michael Kyba (University of Minnesota, Minneapolis, MN) for providing the OP9 cells, the mouse ES cell lines (ZX1, E14, and R1), and the p2Lox plasmid construct.

The authors have no financial conflicts of interest.