Sequential development of pDCs via distinct precursor stages

The frequency of all possible marker combinations was determined for all cells within the pedigrees occurring during the 120hours of experiment time. SiglecH⁺ CCR9^low precursor cells with the annotation CCR9 on, CD11c on, Siglec H on, MHCII on or off were the most frequent cell type observed (51.2%, Fig. 2A and B). These cells contained a small percentage of cells with higher intensity of the Siglec H signal (4.2% of SiglecH⁺ CCR9^low), which was due to intracellular accumulation of the Siglec H antibody or upregulation of Siglec H expression. The second most frequent cell types were early pre-DC (CD11c on, CCR9 on or off, Siglec H off, MHCII off, 13.9%) and CCR9^high pDCs (CCR9 high, CD11c on, Siglec H on or high, MHCII on or off, 14%). CDP (all markers off, 7.8%) and cells with cDC or pre-cDC phenotype (CCR9 on or off, CD11c on, Siglec H off, MHCII on, 5.1%) were detected at lower frequencies. A clear distinction between pre-cDCs and cDCs was not possible because MHCII can already be expressed at low levels in pre-cDCs. While the CDP population was rapidly declining, the pre-DC population increased with a peak at approximately 30hours (Fig. 2B). The number of pre-DCs decreased thereafter concomitant with an increase in the SiglecH⁺ CCR9^low precursor population, which was followed immediately by an expansion of the pDC population. cDCs/pre-cDCs appeared quite early together with the SiglecH⁺ CCR9^low precursors and the population was maintained thereafter (Fig. 2B). The sequential appearance of the different populations indicates progressive development of the CDP progeny via distinct precursor stages.

Open in a separate window

Figure 2

Sequential appearance of cell types in CDP cultures.

(A) Frequency of marker combinations. Each marker can have 3 states, according to its expression level: off (black), on (grey) or high (white). Each cell is counted once, with its marker combination observed immediately before cell division. (B) Marker combinations were grouped manually into distinct cell types (right panel). The relative frequency of each cell type is shown for each time point.

Analysis of the time of onset of the marker expression in the pedigrees showed that CD11c was the earliest marker to be expressed (median 22.5hrs), followed by CCR9 (40.9hrs), Siglec H (45.1hrs) and then MHC class II (53.6hrs) (Fig. 3A). The onset of high signal intensity of CCR9 and Siglec H occurred later in the pedigrees (median 54.3 and 71.3hrs respectively) (data not shown). The sequence of time of onset of CD11c, CCR9 and SiglecH was confirmed in a second experiment (Supplementary Fig. 4A,B). The time until occurrence of the different cell populations defined by marker combinations was calculated as the weighted average of the individual waiting times within each pedigree. The median time to event was shortest for pre-DCs (22hrs) followed by SiglecH⁺ CCR9^low precursors (51hrs) and then pDCs (58hrs). The first appearance of (pre)-cDCs (CD11c⁺ MHCII⁺ Siglec H⁻ cells) was observed later than that of pre-DCs, but earlier than that of pDCs (40hrs). The small subset of SiglecH^high CCR9^low cells appeared at later time points (88hrs) (Fig. 3B). From these results we conclude, that on a population basis CD11c⁺ MHCII⁻ Siglec H⁻ early pre-DCs precede SiglecH⁺ CCR9^low cells which are followed by CCR9^high pDCs shortly thereafter. Cell phenotypes were inherited by the daughter cells in most cases, but at later time points acquisition of additional markers by the daughter cells was not synchronized and quite heterogeneous.

Open in a separate window

Figure 3

Time-dependent occurrence of cell types in CDP genealogies.

CDPs were cultured on EL08 feeder cells with Flt3L for 5 days and their progeny were observed continuously by single cell imaging and tracking. (A) The time until first occurrence of fluorescent markers in the pedigrees was determined. (B) The time until first occurrence of each cell type in the pedigrees was calculated as weighted mean. Data are shown as box plots with whiskers and outliers, median values indicated by horizontal lines. Significant differences are indicated by an asterisk (p<0.05, Kruskall-Wallis-test with Dunn’s post test).

Direct transition between cell types was analysed in the pedigrees (see example shown in Fig. 4A). The number of transitions between cell types in all pedigrees were counted (Fig. 4B, Supplementary Table 1). Transitions between CDP and pre-DC and between pre-DC and SiglecH⁺ CCR9^low cells were observed frequently. Simultaneous acquisition of CD11c, CCR9 and SiglecH signals (SiglecH⁺ CCR9^low cell phenotype) was observed (e.g. Fig. 4A. cell no. 6 and 29), but the most frequent transition to SiglecH⁺ CCR9^low cells was from pre-DCs (CD11c⁺ CCR9⁻ or CCR9⁺ phenotype). In the majority of the cases CCR9^high pDCs developed from SiglecH⁺ CCR9^low cells (exp. 1: 87%, Fig. 4B; exp. 2: 65%). Direct acquisition of the pDC phenotype by CDP or pre-DC was also observed (e.g. Fig. 4A, cell no. 28). Although the high CCR9 signal was transmitted to the daughter cells (e.g. Fig. 4A cells no. 22, 23, 28), it was not always maintained at high levels in pDCs until the end of the experiment. Calculation of the transition times between cell types (Fig. 4C) revealed that transition from SiglecH⁺ CCR9^low precursors to CCR9^high pDCs was rapid (median/IQR: 16/15hrs) compared to the transition from CDP and pre-DC to SiglecH⁺ CCR9^low precursors (median/IQR: 28/39hrs and 35/32hrs respectively). Taken together our results show that a part of the CDPs were able to give rise to pDCs within 3 days of culture and in the majority of these cases CDP progeny transited sequentially through pre-DC and SiglecH⁺ CCR9^low intermediate precursor stages of variable duration before differentiating into pDCs.

Open in a separate window

Figure 4

Frequency and duration of transitions between cell types.

(A) A representative pedigree of a single CDP and its progeny. Branches are colored according to cell type (see Fig. 4B). Apoptotic events are marked with X whereas cell losses are indicated by a question mark. (B) Cell type transition graph. Nodes represent cell types defined by marker combinations. The colour code and assigned marker combinations are shown for each cell type on the right. The number of observed transitions between two cell types defines the width of the connecting arrow. Transitions occurring less than 10 times were omitted. (C) Median waiting time spent by cell type x (rows) until transformation into cell type y (columns). Infrequent transitions (observed less than 10 times) and the main diagonal are greyed out.

Cell isolation and culture

BM cells were isolated from femora and tibiae of 6–8 weeks old mice. After red blood cell lysis (RBC Lysis Buffer, Sigma-Aldrich, St. Louis, MO, USA) cells were washed and resuspended in dendritic dell (DC) medium containing RPMI 1640, (Biochrom, Berlin, Germany), 10% (v/v) heat-inactivated fetal calf serum (PAA, Austria), 1% (v/v) non-essential amino acids (PAA), 1% (v/v) Glutamax (Thermo Fisher Scientific, Waltham, MA, USA), 1% (v/v) Sodium Pyruvate (100mM, Thermo Fisher Scientific), 1% (v/v) penicillin/streptomycin (PAA), 0.05mM β-mercaptoethanol (Sigma-Aldrich, Germany) and supplemented with 20ng/mL Flt3L (purified from supernatant of CHO-flk2 cell line kindly provided by Nicos Nicola, Walter and Eliza Hall Institute). Cells were cultured for 5 days at 37°C in 5% CO₂ to obtain FL-DCs.

Live cell imaging experiments were performed using the embryonic liver (EL) 08 stromal cell line as feeder cells⁴⁹. EL08 cells were cultured in EL08 medium (MEM-α-Glutamax (Invitrogen, Karlsruhe, Germany), 15% v/v heat inactivated fetal calf serum, 5% v/v horse serum (Stem Cell Technologies, Köln, Germany), 1% penicillin/streptomycin, 0.01mM β-mercaptoethanol) at a density of 5*10⁵/ml in 10cm dishes coated with 0.1% gelatin (Sigma Aldrich, Seelze, Germany) as described before⁴⁹. After 2–3 days of expansion, cells were detached using Trypsin/EDTA (Thermo Fisher Scientific), resuspended in EL08 medium and seeded in ibidi μ-slides (Ibidi, Martinsried, Germany) for imaging experiments.

Flow Cytometry and cell sorting

For isolation of CDPs, lineage-positive cells were depleted from BM cells by staining with FITC-conjugated antibodies against CD3, CD19, B220, Gr1, NK1.1 and CD11b in FACS buffer (PBS containing 2% FCS) and separation using anti-FITC microbeads (Miltenyi Biotech, Germany). Lineage-depleted cells were stained with FITC-conjugated antibodies against CD3, CD19, B220, Gr-1, NK1.1 (BD Biosciences), CD11b (eBioscience) and with anti-CD135-PE, anti-CD11c-PE-Cy7, anti-CD115-APC, anti-CD117-eFlour780 and anti-MHCII-eFlour450 (eBioscience) in FACS buffer (PBS, 2% FCS) containing 50% supernatant of 2.4G2 hybrioma (Fc block) for 20min on ice. To exclude dead cells from the analysis, propidium ioidide (2.5μg/ml) was added to each sample just prior to FACS analysis. CDPs were sorted as Lin⁻ Flt3⁺ CD117^low CD115⁺ CD11c⁻ MHCII⁻ cells to high purity using FACS Aria III (BD Biosciences), as shown in supplementary Fig. 1. FACS analysis of DC cultures was performed using anti-CD45.1-Pe-Cy5.5, anti-CCR9-PE, anti-CD11c-Pe-Cy7, anti-MHCII-eFluor780 (eBioscience), anti-CD45.2-eFluor450, anti-B220-BV605 (BD Biosciences) anti-SiglecH-A488 and anti-SiglecH-A647 (produced in our lab from 440c hybridoma, kindly provided by Marco Colonna, Washington University, St. Louis, USA). Samples were acquired on a FACS Gallios or Cytoflex flow cytometer (Beckman Coulter, Krefeld, Germany) and analysed using FlowJo software (Tree Star, Stanford, USA).

Live cell imaging and long-term antibody staining of CDP cultures

Prior to imaging, μ-slides (I^0.4 Luer Series, uncoated, Ibidi) were coated with 0.1% gelatin and EL08 stromal cells (2*10⁴ cells in 100μl EL08 medium) were seeded. Within 24hours, when stromal cells had reached 40–50% optical confluence, the medium was removed and 2*10³ CDPs per slide were seeded in DC medium containing 20ng/ml Flt3L for live cell imaging experiments. Higher concentrations of Flt3L had no major effect on DC subtype differentiation but increased cell density (data not shown) and were therefore less suitable for single cell tracking. Fluorescently labeled antibodies were added for in culture staining as described before³⁸. Antibodies were used at the lowest concentrations which provided sufficient fluorescent signals (0.05μg/ml anti-MHCII-eFluor450, 0.025μg/ml anti-CD11c-A488, 0.25μg/ml anti-SiglecH-A647 0.05μg/ml anti-CCR9-PE). During imaging, culture medium was not replenished and no additional antibodies or Flt3L were added. Time-lapse imaging was performed with a cell observer system (Zeiss, Oberkochen, Germany) at constant 37°C and 5% CO_2. Bright field images were taken at 2min intervals while fluorescent images were taken at 3–4h intervals with an Axiocam-HRm camera (1338×1040 pixel resolution) with 10X objective. Carl Zeiss AxioVision 4.5 Software was used in this study.

Single cell tracking and analysis of cell fates in pedigrees

Single cell tracking was performed using tracking software (TTT⁵⁰). The software allows loading of acquired phase contrast and fluorescence images with determined intervals. Individual cells were tracked and properties such as lost events, cell death, division, and fluorescent signals were logged manually by the researchers. All relevant information saved during tracking was annotated in pedigrees including fluorescent markers (off, on or high). The raw data consisted of 72 trees originating from single CDP cells. In 32 out of 72 cases (44%), CDP colonies were lost early in the CDP or pre-DC stage, because they could not be tracked further or underwent apoptosis. These colonies were excluded from further analysis. Cell fates were assigned to CDP colonies by visual analysis of annotated pedigrees (results shown in Fig. 5E). Cell types were identified by predefined marker combinations. If a pDC or cDC phenotype was observed in the pedigree, it was assigned a pDC or cDC cell fate. If a SiglecH⁺ CCR9^low phenotype was observed in a pedigree without upregulation of CCR9 it was assigned a SiglecH⁺ CCR9^low phenotype. If no marker or only CD11c or CCR9 were expressed in a pedigree it was assigned an undifferentiated cell fate.

This work is part of the thesis of E.D. E.D., A.M. and A.K. are supported by the German Research Foundation (SFB1054/TPA06, KR2199/3-2, KR2199/6-1, KR2199/9-1). T.S. is supported by the Swiss National Science Foundation. T.S. and M.E. acknowledge financial support from SystemsX.ch.