Heterogeneity of porcine bone marrow-derived dendritic cells induced by GM-CSF

In vitro generation of dendritic cells (DCs) is advantageous for overcoming the low frequency of primary DCs and the difficulty of applying isolation techniques for studying DC immunobiology. The culture of bone marrow cells with granulocyte-macrophage colony-stimulating factor (GM-CSF) has been used extensively to generate bone marrow-derived dendritic cells (BMDCs). Studies have reported the heterogeneity of cells grown in murine GM-CSF culture based on the levels of MHCII expression. Although porcine DCs are generated by this classical method, the exact characteristics of the BMDC population have not yet been defined. In this study, we discriminated GM-CSF-grown BMDCs from gnotobiotic miniature pigs according to several criteria including morphology, phenotype, gene expression pattern and function. We showed that porcine BMDCs were heterogeneous cells that differentially expressed MHCII. MHCIIhigh cells displayed more representative of DC-like morphology and phenotype, including costimulatory molecules, as well as they showed a superior T cell priming capacity as compared to MHCIIlow cell. Our data showed that the difference in MHCIIhigh and MHCIIlow cell populations involved distinct maturation states rather than the presence of different cell types. Overall, characterization of porcine BMDC cultures provides important information about this widely used cellular model.


Introduction
Dendritic cells (DCs) are components of the immune system that can present antigens to T cells [1]. Conventional DCs (cDCs) provide signals for T cell activation and differentiation, and are therefore regarded as professional antigen-presenting cells (APCs) of the immune system [2]. However, study of these essential cells has been complicated by the low frequency of DC populations in blood and tissue. For this reason, the biology of DCs has been studied in cells grown in vitro from hematopoietic precursors, in the presence of growth factors [3]. Besides, in vitro generated DCs have been designated as cell-based vaccines for immunotherapy [4]. Bone marrow cells (BMCs) have been cultured with granulocyte-macrophage colonystimulating factor (GM-CSF), a cytokine involved in the development and homeostasis of a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 mononuclear phagocytes, to generate bone marrow-derived dendritic cells (BMDCs) that resemble tissue DC [5,6].
In bone marrow cultures induced by GM-CSF, CD11c + MHCII + cells have been assumed to be the source of pure BMDCs, whereas macrophages are thought to be adherent cells [3,7]. However, the studies reported that this classical method produces heterogeneous populations of murine myeloid cells in non-adherent populations and loosely adherent populations [8][9][10]. The study suggested that MHCII high cells, which were previously shown to be DCs and MHCII low cells, closely resemble macrophages in the murine GM-CSF-derived heterogeneous population. Other studies suggested that MHCII low cells contain immature DCs, which further upregulate MHCII on their surface, indicating maturation in mice [11,12]. The porcine immune system is similar to the human immune system with respect to DC biology [13,14]. The gnotobiotic miniature pig is the best model to study immunology, including immune cell ontogeny, microbial infection, and xenotransplantation [15][16][17][18]. To study porcine DC biology, in vitro differentiated DCs have been widely used [19], especially BMCs are cultured with GM-CSF for generation of BMDCs likewise other species [20]. The nonadherent cells have been considered as pure BMDCs and are characterized by expression of the surface molecules, CD1, CD16, CD80/86, CD172a, and MHC class II [21]. However, it is unclear whether porcine BMDCs are heterogeneous like murine BMDCs.
In this study, BMCs were isolated from gnotobiotic miniature pigs and cultured with GM-CSF to generate DCs. We classified GM-CSF-grown porcine BMDCs into MHCII high cells and MHCII low cells, in a similar manner as murine BMDCs. These two populations from non-adherent cells were characterized according to their morphology, phenotype, gene expression profile, and function. On the basis of these characteristics, we showed that non-adherent cells isolated from GM-CSF-grown BMC cultures were heterogeneous in terms of their levels of MHCII expression. Therefore, these findings of GM-CSF-derived porcine BMDCs could lead to improvements in our understanding of the porcine immune system.

Animals
Gnotobiotic miniature pigs were kept under absolute barrier contained facility at the Bioorgan Research Center of Konkuk University, Seoul, Republic of Korea [22]. Animal experiments were carried out based on the National Institutes of Health guidelines for the care and use of laboratory animals. The study was conducted after obtaining approval from the Institutional Animal Care and Use Committee (IACUC) of Konkuk University (KU16168). In this study, three, 3-week-old piglets were used: K8082-1, K8082-2, and K8083-4. The animals were sacrificed using CO 2 according to IACUC guidelines, and then the humerus, tibia, and femur were collected to isolate BMCs.

Cell preparation
The BMDCs were generated using a previously described method with some modifications [20]. The BMCs were cultured for 10 days at a density of 5 × 10 5 cells/mL in RPMI-1640 medium (Gibco, Gaithersburg, MD, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco), 100 U/mL penicillin, 100 μg/mL streptomycin, 1 mM minimal essential medium, non-essential amino acids (Gibco), and 100 ng/mL of porcine GM-CSF (R&D Systems, Madrid, Spain). GM-CSF was additionally supplemented on days 2, 4, and 6. Differentiated cells were obtained from the non-adherent cell population after day 10. The cells were incubated at 37˚C in a humidified atmosphere of 5% CO 2 in air.
To isolate MHCII high and MHCII low populations from BMDCs and c-kit + hematopoietic stem cells (HSCs) from BMCs, the cells were sorted by a FACSAria™ instrument (Becton Dickinson). Flow cytometry analysis was conducted using FlowJo software (https://www.flowjo. com/).

RNA sequencing
Total RNA was extracted from sorted cell subsets including c-kit + HSC, MHCII high , and MHCII low cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). In order to construct cDNA libraries with the TruSeq RNA library kit (illumine, San Diego, CA, USA), 1ug of total RNA was used. The protocol consisted of polyA-selected RNA extraction, RNA fragmentation, random hexamer primed reverse transcription and 100nt paired-end sequencing by Illumina HiSeq2500 (illumine, San Diego, CA, USA). The libraries were quantified using qPCR according to the qPCR Quantification Protocol Guide and qualified using an Agilent Technologies 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). We processed reads from the sequencer and aligned them to the Sus scrofa using Tophat v2.0.13 [23]. Transcript assembly and abundance estimation using Cufflinks v2.2.1 [24].
The transcript-level relative transcript abundances were measured in FPKM (Fragments Per Kilobase of exon per Million fragments mapped) using Cufflinks. We performed the statistical analysis to find differentially expressed genes (DEG). Filtered data were log2-transformed and subjected to quantile normalization. For DEG set, Hierarchical clustering analysis was performed using complete linkage and Euclidean distance as a measure of similarity. Geneenrichment and functional annotation analysis for significant gene list was performed using Gene Ontology (http://geneontology.org/). We used multidimensional scaling (MDS) method to visualize the similarities among samples. We applied to the Euclidean distance as the measure of the dissimilarity.

Real-time polymerase chain reaction
The cDNA was reverse-transcribed from total RNA using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. Synthesized cDNA was denatured at 95˚C for 10 min and amplified using SYBR Premix Ex Taq II (Takara, Kusatsu, Japan) on an Applied Biosystems 7500 Real-Time PCR System cycler, with 40 cycles of 95˚C for 15 s and 60˚C for 1 min. All data were acquired as ΔCt values and automatically converted to double delta Ct (ΔΔCt) values by 7500 software (Applied Biosystems). The value of 2−ΔΔCt was calculated to obtain expression fold-change data. Primers specific for CD86, CD40, IFR4, CCR7, FcεR1α, CSF1R, CD163, and CD117 (S1 Table) were used.

Mixed lymphocyte reaction
For the preparation of allogeneic T cells, splenocytes were isolated and incubated for 2 h to remove attached cells. Floating splenocytes were harvested and labelled with carboxyfluorescein succinimidyl ester (CFSE; Invitrogen). CFSE-labelled allogenic splenocytes were co-cultured with APCs (MHCII high and MHCII low cells) for 5 days. Then, 10 5 splenocytes were mixed with APCs according to the desired APC:splenocyte ratio (1:2, 1:6, 1:18, 1:54 and 1:162) in 96-well U-bottom plates. For gating proliferating population, CFSE-stained splenocytes was checked their proliferation at 24h and confirmed without proliferation (data not shown). T cell proliferation was examined using flow cytometry and analyzed by FlowJo software.

Phagocytosis assay
The cells were seeded at 2 × 10 5 cells and incubated with latex beads coated with FITC-labelled rabbit IgG (Cayman Chemical, Ann Arbor, MI, USA) for 30 min, 1 h, 2 h, 3 h, and 4 h in 96-well U-bottom plates. To distinguish cells that were phagocytosed from those simply binding to the beads at the surface, a short (1-2 min) incubation with Trypan Blue dye quenching solution, followed by a wash with assay buffer, was used to quench the surface FITC fluorescence. Phagocytosed cells were detected using flow cytometry and analyzed by FlowJo software.

Heterogeneity of the GM-CSF-derived BMDCs
To generate BMDCs in vitro, BMCs were obtained from 3-week-old gnotobiotic miniature pigs and cultured with GM-CSF supplementation. To enrich BMDCs, we harvested nonadherent cells from the GM-CSF culture and confirmed the MHCII expression of these cells. The non-adherent cells were comprised of two distinct populations (MHCII high and MHCII low ) based on the MHCII expression ( Fig 1A). Adherent cells were mainly composed of MHCII low population (S1 Fig). We employed FACS sorting to purify MHCII high and MHCII low populations that could not be distinguished by adhesion properties. In the two populations, differences in morphology were observed; the MHCII high cells had a more dendritic morphology, and showed cluster formation, relative to the MHCII low cells (Fig 1B). Thus, MHCII high cells were more representative of DC-like morphology than MHCII low cells. From these results, two populations were isolated from non-adherent cells from porcine BMDC cultures, based on different expression levels of MHCII, in a similar manner to murine cells.

Surface marker expression levels of MHCII high and MHCII low cells
Because there were differences in MHCII expression and morphology, MHCII high and MHCII low cells were sorted to confirm their different phenotypes (Fig 2). We examined CD86, CD1, CD16, CD11b/CD18, CD172a, CD14, and CD163 to clearly define distinct populations. The MHCII high and MHCII low cells expressed CD172a and CD14, indicating that they both differentiated into myeloid lineages. We observed high expression of the porcine DC markers, CD86, CD1, and CD16, in MHCII high cells, and low expression in MHCII low cells. Complement receptor CD11b/CD18 and scavenger receptor CD163, which are highly expressed on activated myeloid cells, were found to be highly expressed on MHCII high cells, but they showed very low expression on MHCII low cells. The immature DC phenotype involved intermediate or low expression of MHCII and costimulatory molecules such as CD86, together with high expression of CD14. These results suggested that MHCII low cells resembled immature DCs, and MHCII high cells underwent spontaneous maturation and expressed higher amounts of the same markers.

Gene expression patterns of MHCII high and MHCII low cells
For gene expression analysis, we sorted c-kit + HSCs from BMCs and the two populations (MHCII high and MHCII low cells) described in Fig 1A, followed by mRNA sequencing. Using gene hierarchical cluster mapping and MDS analysis, each cell type from the three different piglets clustered together, confirming that the transcription profiles of each cell type were similar (Fig 3A and 3B). We then confirmed that 368 genes were differentially expressed between MHCII high and MHCII low cells. However, the MHCII high and MHCII low cell populations were close together, suggest that these they are not the completely separated as distinct cell type unlike what has been described for mice and humans [8,9]. We therefore hypothesized that MHCII high and MHCII low cell populations were comprised of BMDCs in distinct maturation states, as opposed to different cell types being present.
To further explore this possibility, DC-related gene expression patterns were investigated in each cell type. Transcriptome analysis revealed that MHCII (SLA-DR) and costimulatory molecules (CD40, CD80, CD83, and CD86) were highly expressed in MHCII high cells that were induced during BMDC differentiation (Fig 3C). In addition, receptors (CD163, MRC1, FCER1A, FCGR2B, and TLRs) involved in the DC innate immune response were also highly expressed in MHCII high cells. MHCII low cells expressed more CD34, CD59, CD177, Sox4, and Foxo1, which showed the highest expression levels in HSCs. These data showed that MHCIIhigh cells highly expressed genes related to the DC signature and innate immune response, whereas, genes enriched in HSCs were found to be expressed more in MHCII low cells than MHCII high cells.
To determine the difference between MHCII high and MHCII low cells, DC-related gene expression patterns were validated by qPCR. As expected, the CD86 and CD40 costimulatory molecules were expressed at high levels by MHCII high cells, and to a lesser extent by MHCII low cells (Fig 3D). In addition, the IRF4 transcription factor, which controls the development of BMDCs falling within the mature gate, was highly expressed on MHCII high cells compared to MHCII low cells. MHCII high cells also expressed higher amounts of CCR7, FcεR1α, CD163, and CSF1R, whereas MHCII low cells showed higher expression of CD117. Consistent with these results, activated DC-related genes were highly expressed on MHCII high cells compared to MHCII low cells. Therefore, we assumed that the two populations (MHCII high and MHCII low cells) had cell-to-cell variations that were the result of different states of BMDC maturation.

Functions of MHCII high and MHCII low cells
To examine the ability of DCs that can stimulate T cells as professional APCs, the mixed lymphocyte reaction was conducted using allogeneic splenocytes co-cultured with sorted MHCII high and MHCII low cells (Fig 4). When sufficient APCs were supplied to expand splenocytes (the APC:splenocyte ratio was 1:2~1:18), the MHCII high and MHCII low cells displayed comparable ability to stimulate T cells. There was no significant difference in the T cell proliferation ability of MHCII high and MHCII low cells. In contrast, when given a lower number of APCs to stimulate T cells (APC:splenocyte ratio, 1:54~1:162), MHCII low cells were inferior in terms of proliferating T cells compared to MHCII high cells. Accordingly, MHCII high cells showed superior T cell priming capacity compared to MHCII low cells.
DCs are mononuclear phagocytes; therefore, to analyze phagocytic abilities, MHCII high cells and MHCII low cells were cultured with phagocytic beads (Fig 5). MHCII high cells had more uptake of phagocytic beads during 3 h (Fig 5A; 50% of MHCII high cells and 37% of MHCII low cells). When cells were incubated with phagocytic beads, MHCII low cells were significantly less efficient at phagocytosis, as expected ( Fig 5B). Together, these results indicated that MHCII high cells are more functionally activated DCs than MHCII low cells, because they had superior T cell-priming ability and phagocytic ability.

Discussion
The in vitro generation of DCs in culture is advantageous for studying DC biology. In particular, GM-CSF, a hematopoietic growth factor, has been used to supplement BM cultures to generate CD11c + MHCII + cells, which are often termed BMDCs [5]. From the GM-CSF BM culture, DCs have been enriched from non-adherent cells, whereas adherent cells are thought to be macrophages. In addition, the studies of the discrimination of murine BM cultures showed the heterogeneity of GM-CSF-derived non-adherent cells and loosely adherent cells [8,9]. They suggested that the MHCII high cell population, considered as a DC and MHCII low cell population, actually corresponded to macrophages from murine BM cultures. However, another study showed that GM-CSF culture induced differentiation towards immature and mature cDC2s, which were shown to be efficient at promoting Th17, as well as Th2 immune responses, in a non-adherent population [11].
The pig has been considered an important large animal model, and gnotobiotic miniature pigs are probably the best model for studying immunology [18]. The porcine immune system resembles the human immune system with respect to DC biology, because their gene expression signature for cDC2 is close to the human counterpart [14,25,26]. The classical protocols for generating in vitro DCs in humans and mice are similar to the porcine method. Although GM-CSF-generated porcine BMDCs have been widely used, the heterogeneity of the cells has not been defined. Considering that murine GM-CSF cultures often provide two populations based on MHCII expression level, we discriminated GM-CSF-grown BMDCs from gnotobiotic miniature pigs based on several criteria.
In this study, we noted heterogeneity in the non-adherent cells from gnotobiotic miniature pigs according to their MHCII expression levels (MHCII high and MHCII low cells). It has been reported that cells developing in porcine GM-CSF culture were also heterogeneous, as in murine cultures. Although both populations showed DC-like morphology, MHCII high cells had a more dendritic morphology, and showed cluster formation, relative to MHCII low cells. The phenotype analysis showed that MHCII high cells displayed a DC-like phenotype that involved CD86 + , CD1 + , CD16 + , CD11b/CD18 + , CD172a + , CD14 low , and CD163 + . MHCII low cells also expressed these DC markers; however, they had low expression levels of CD86, CD1, CD16, CD11b/CD18, and CD163, and higher CD14 expression which is downregulated during DC maturation [27]. According to our results, MHCII low cells appear to represent an immature DC phenotype with low expression of MHCII and costimulatory molecules, such as CD86.
In accordance with morphology and phenotype analysis, transcriptome analysis confirmed heterogeneity in BMDC maturation: DC-related genes were highly expressed in MHCII high cells, including costimulatory molecules and innate immune receptors, whereas MHCII low cells showed higher levels of genes mainly expressed on HSCs. The higher expression levels of IRF4 and CCR7 in MHCII high cells supported BMDCs being within the mature gate, as well as the development of subsets into cDC2s [28,29]. In accordance with, BMDCs under the influence of GM-CSF appeared closer to cDC2s [27]. The MDS analysis revealed the differences between MHCII high and MHCII low cells. It also showed that the few difference between the two cell populations involved the maturation state rather than being due to the presence of distinct cell types. One sample from the MHCII low cell population (MHCII low _3) was more close to the MHCII high population in the MDS analysis. It is possible that variations were due to differences between individual samples, or to factors such as variable culture conditions. In further studies, it should be possible to identify their closest relatives in vivo by transcriptome analysis of in vitro-generated BMDCs from gnotobiotic miniature pigs. Furthermore, our RNA-sequencing data may provide information relevant to the investigation of porcine HSCs and BMDCs.
In addition, MHCII high cells expanded into T cells and phagocytized beads more efficiently than MHCII low cells, with similar gene ontology enrichment of antigen presentation and innate immune receptors. Accordingly, porcine GM-CSF culture preferentially differentiated BMCs into immature (MHCII low cells) and mature (MHCII high cells) DCs.
On the basis of morphological, phenotypical, and gene expression criteria, we classified cell two populations based on MHCII expression; we suggest that the MHCII high and MHCII low populations can be best-classified by their maturation stage. Therefore, this study might lead to a better understanding of the function of DCs and provides useful information for future studies using porcine BMDCs.