Characterization of human peritoneal dendritic cell precursors and their involvement in peritonitis

Abstract

Scattered evidence suggests that the human peritoneal cavity contains cells of the dendritic cell (DC) lineage but their characterization is missing. Here, we report that the peritoneal cavity of normal subjects and of stable patients on peritoneal dialysis (PD) contains a population of CD14⁺ cells that can differentiate into DCs or macrophages. Within this pool, we characterized a CD14⁺CD4⁺ cell subset (2·2% of the peritoneal cells) fulfilling the definition of myeloid DC precursors or pre-DC1 cells. These cells expressed high levels of HLA-DR, CD13, CD33, and CD86, and low levels of CD40, CD80, CD83, CD123, CD209, TLR-2 and TLR-4. These cells retained CD14 expression until late stages of differentiation, despite concomitant up-regulation of DC-SIGN (CD209), CD1a, CD80 and CD40. Peritoneal pre-DC1 cells had endocytic capacity that was down-regulated upon LPS/IFN-γ stimulation, were more potent allo-stimulators than peritoneal CD14⁺CD4^–/lo cells and monocyte-derived macrophages, and induced Th1 cytokine responses. More importantly, the number of peritoneal pre-DC1 cells increased during PD-associated peritonitis, with a different profile for Gram positive and Gram negative peritonitis, suggesting that these cells participate in the induction of peritoneal adaptive immune responses, and may be responsible for the bias towards Th1 responses during peritonitis.

Introduction

Dendritic cells (DC) are a heterogeneous population of professional antigen-presenting cells (APC) that regulate the balance between immunity and tolerance [1]. This is achieved through a complex interplay between their differentiation profile and their migratory properties. Starting in the bone marrow, nonproliferating precursors of DC migrate into peripheral tissues to become immature DC whose main function is to capture antigen by macropinocytosis, receptor-mediated endocytosis, and/or phagocytosis [2]. In the presence of pathogen-related antigens and/or inflammatory stimuli, DC mature and this results in a down-regulation of antigen uptake [3] and concomitant up-regulation of adhesion molecules and chemokine receptors that allow them to migrate to the T cell areas of draining lymph nodes. Once in the lymph nodes, DC interactions with CD40L-bearing cells result in the completion of DC maturation that is characterized by increased antigen presenting capacity due to the up-regulation of cell surface expression of MHC class II and costimulatory molecules [4].

Characterization of DC is difficult as no single, specific marker for discrete stages of DC differentiation exist. By default, DC are characterized phenotypically by high expression of MHC class II molecules and costimulatory molecules, morphologically by large membrane protrusions called dendrites, and functionally by their potent allo-stimulatory capacity. Much of what is known about human DC have come from the isolation and study of blood DC or from the in vitro generation of DC from either CD34⁺ bone-marrow precursors or blood monocytes [5,6]. This has led to the identification of two main DC subsets: one subset derived from myeloid precursors and another derived from lymphoid precursors [7,8]. However, characterization of these same subsets for human DC in nonlymphoid peripheral tissues is still missing. This is partly due to the need of high cell numbers to perform differentiation studies to discriminate between myeloid-lineage DC and other myeloid-derived cells such as monocytes/macrophages [9,10]. These types of studies become more challenging when assessing precursors of the myeloid-lineage of DC or pre-DCs and DC ‘intermediates’ [7,8].

To examine the biology of human DC in situ, we focused on peritoneal cells obtained from the peritoneal lavages of individuals undergoing elective laparascopy or from the peritoneal dialysis (PD) effluents of patients with end-stage renal disease (ESRD) [11]. The majority of these cells have been traditionally identified as macrophages and lymphocytes [12]. However, scattered evidence suggests that DC are present in the human peritoneal cavity and may constitute around 6% of the peritoneal cells isolated from PD effluents [13]. Here, we report that the peritoneal cavity of normal individuals and of patients on PD contains a discrete CD14⁺CD4⁺ subset of cells with the features of myeloid-derived DC precursors. These cells participate in local adaptive immune responses particularly to Gram-positive bacteria, and impart a Th1 profile of T cell responses. In addition, we demonstrate that the entire peritoneal pool of CD14⁺ mononuclear cells have the capacity to differentiate into DC or macrophages in vitro, under appropriate conditions, indicating that these cells are not fully differentiated despite their location in peripheral tissues. These data provide a mechanistic explanation for the predominant Th1 environment observed during episodes of PD-related peritonitis [14,15].

Materials and methods

Patient samples

Patients were recruited from the London Health Sciences Centre Peritoneal Dialysis Unit (London, Ontario, Canada). Two litre, long duration (6–12 h) dwell PD effluent bags (n = 234) were collected from 95 noninfected patients over the course of 24 months. Patients were on automated or cycler PD. Patient demographics are listed in Table 1 and are typical of current PD populations [16]. For some experiments, patient blood samples were also obtained. In addition, peritoneal lavage with saline solution was collected from three control subjects, with normal renal function, undergoing elective laparoscopy. Patient participation in the study was voluntary and research protocols were approved by the Office of Research Ethics at the University of Western Ontario.

Table 1.

Patient demographics

Parameter	Stable PD patients (n = 92)	PD-associated peritonitis patients (n = 8)
Mean age (years)	56·2 ± 17·76	65·50 ± 15·79
Sex
Male/Female	59/33	5/3
Primary Kidney Disease
Diabetes	31	0
Glomerulonephritis	20	1
Hypertension	10	2
Other	31	5
Mean time on PD (Months)	18·61 ± 22·65	24·75 ± 21·01
Number of previous peritonitis  episodes	0·59 ± 0·938	1·875 ± 1·356
Etiology of peritonitis (gram positive/gram negative)	N/A*	4/4

^*N/A: not applicable.

Diagnosis of PD-associated peritonitis was based on the presence of abdominal pain, a cloudy PD effluent with a leucocyte count above 100 × 10⁶/l of effluent, and a positive microbiological culture. During episodes of PD-associated peritonitis, 8 patients had their long duration dwell effluents collected daily. Patient demographics for this subsample are listed in Table 1.

Cell isolation and cultures

Cells were isolated from PD effluents by centrifugation at 426 g for 15 min at 4°C [12]. Peripheral blood mononuclear cells (PBMC) from PD patients and control donors were prepared by Ficoll-paque gradient [17]. Peripheral blood lymphocytes (PBL) were obtained by plastic adherence, and purified monocytes were obtained from normal donor PBMC by negative selection using magnetic column separation (Stemsep or MACS). Cells from each preparation were washed with phosphate buffered saline (PBS) and resuspended in complete culture RPMI 1640 medium supplemented with 1%l-glutamine, 1% penicillin-streptomycin, 1% HEPES, and 10% FCS. Monocyte-derived DC or macrophages were generated from PBMC, purified monocytes, or peritoneal cells by culturing with GM-CSF and IL-4 (1000 U/ml each) or M-CSF alone (1000 U/ml), respectively, in complete culture media for up to 10 days. Culture media and cytokines were replenished every 2–3 days. DC maturation was induced with LPS (100 ng/ml) and IFN-γ (10 ng/ml) for 24–48 h.

Flow cytometry

Peritoneal cells (5 × 10⁵) were incubated with normal human serum before staining with the following directly labelled mouse antihuman monoclonal antibodies (mAbs): anti-CD3 (UCHT1), anti-CD4 (RPA-T4), anti-CD8 (RPA-T8), anti-CD19 (HIB19), anti-CD33 (WM-53), anti-CD40 (5C3), anti-CD56 (MEM188), anti-CD83 (HB15e), anti-CD86 (IT2·2), anti-TLR2 (TLR2·1), and anti-TLR4 (HTA125), all purchased from eBioscience (San Diego, CA). Anti-CD1a (HI149), anti-CD11c (B-Ly6), anti-CD13 (WM-15), anti-CD14 (M5E2), anti-CD16 (3G8), anti-CD80 (L307·4), anti-HLA-DR (G46-6), anti-CD209 (DCN46), and anti-CD123 (9F5) antibodies were purchased from BD (San Diego, CA, USA). Two and four-colour flow cytometric analyses were performed with appropriate isotype-matched controls using PE-, APC-, Cy-Chrome or PerCP-, and FITC-conjugated antibodies. Cells were fixed with 2% paraformaldehyde and analysed by flow cytometry (Becton Dickinson, Mountain View, CA, USA). Analyses were performed using CellQuest computer software (Becton Dickinson) [18,19].

Isolation of CD14⁺CD4⁺ cells

Peritoneal effluent cells, isolated as described above, were washed, incubated with normal human serum, and subsequently surface-labelled with FITC or PE-conjugated anti-CD4 and APC-conjugated anti-CD14 antibodies. Labelled cells were sorted to obtain CD14⁺CD4⁺ and CD14⁺CD4^–/lo cells using a Fluorescence-Activated Cell Sorter (FACS) (FacsVantage, Becton-Dickinson).

Mixed lymphocyte reaction

Sorted, irradiated (5000 rad) CD14⁺CD4⁺ cells, CD14⁺CD4^–/lo cells, in vitro monocyte-derived DC and monocyte-derived macrophages, or PBMC were used as stimulators. Autologous and allogeneic PBL from PD patients or controls were used as responders. Cells were plated in 96-well, round-bottom plates at stimulator:responder ratios of 1 : 2, 1 : 20, 1 : 50, and 1 : 100 in a total volume of 200 µl at 5% CO₂ and 37°C for 5 days. Responder proliferation was measured by adding 1 µCi/well of ³H-thymidine for 18 h on day 5. Cells were harvested and ³H-thymidine uptake was measured using a microbeta counter. Results were expressed as mean c.p.m ± SEM of triplicate samples.

Confocal microscopy

Unfractionated peritoneal cells and mononuclear cells (1 × 10⁶) were stained with FITC-labelled anti-CD4 and PE-labelled anti-CD14 and fixed with 4% paraformaldehyde on ice. Next, they were plated on poly L-lysine-coated 35 mm glass bottom microwell plates (MatTek) for 30 min. Images were captured using a confocal microscope (Carl Zeiss, Inc.) and analysed with the LSM 510 software program (Carl Zeiss, Inc.; Microsoft).

Morphology

Unfractionated peritoneal cells left untreated or cultured in the presence of GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) or M-CSF (1000 U/ml) (without or with stimulation with LPS (100 ng/ml) and IFN-γ (10 ng/ml) for 24 h), were plated on poly L-lysine-coated 35 mm glass bottom microwell plates. Light phase images were acquired using a confocal microscope (Carl Zeiss, Inc.). To analyse cellular and nuclear morphology, cytospun preparations of 1 × 10⁵ cells were stained with Giemsa and images were collected using a light microscope.

Endocytosis

Unfractionated peritoneal cells, either left untreated or incubated in the presence of IL-4 (1000 U/ml) and GM-CSF (1000 U/ml) or M-CSF (1000 U/ml), followed by stimulation with LPS (100 ng/ml) and IFN-γ (10 ng/ml) for 24 h, and patient PBMC (5 × 10⁵) were incubated for 30 min at 0°C (control) or 37°C with FITC-labelled dextran (Mr = 40000, Molecular Probes Inc., Eugene, OR) at 1 mg/ml. Cells were then incubated with anti-fluorescein rabbit IgG Fab fragments (Molecular Probes) in order to quench any surface-bound FITC-dextran molecules. Cells were subsequently stained with anti-CD14 and anti-CD4 antibodies and analysed by flow cytometry (FACScalibur, BD). Analyses were performed using CellQuest computer software.

Cytokine production

FACS-sorted CD14⁺CD4⁺ cells, CD14⁺CD4^–/lo cells, or MACS sorted blood monocytes were used as stimulators. Autologous and allogeneic PBL from PD patients or healthy donors were used as responders. Cells were plated in 96-well, round-bottom plates at stimulator:responder ratios of 1 : 25, 1 : 50, and 1 : 100 in a total volume of 200 µl at 5% CO₂ and 37°C for 48 h. PBL stimulated with PMA (100 ng/ml) plus Ionomycin (2 µg/ml) served as positive controls. Cytokine production (IFN-γ as a Th1 cytokine, and IL-4 as a Th2 cytokine) was assessed in culture supernatants by ELISA for IFN-γ (BD) and IL-4 (eBioscience). Each supernatant was assayed in triplicate.

Statistics

Analysis of variance (anova) followed by Tukey Post Test and Student's t-test were performed using GraphPad Prism Software (GraphPad Software Inc.). A difference between groups was considered significant when P ≤ 0·05. Ex vivo experiments were performed at least three independent times.

Results

The pool of CD14⁺ cells from the peritoneal cavity contains dendritic cell precursors

Long duration dwell (>8 h) PD effluents from stable patients contained an average of 5 × 10⁶ cells/2l (Fig. 1a). The majority of cells isolated from these effluents expressed CD14 (39·26%, n = 21) or CD3 (21%, n = 9) (Fig. 1b) consistent with previous reports showing that macrophages and T lymphocytes are the two major cell types present in PD effluents [11]. However, two-colour FACS analysis (Fig. 1c) and confocal microscopy of peritoneal cells (Fig. 1d) revealed the presence of a distinct CD14⁺ subset that also expressed high levels of CD4, in contrast to the conventional CD14⁺ peritoneal macrophage pool that expresses low levels or no CD4. The CD14⁺CD4⁺ subset represents on average 2·24% of the cells in PD effluents, or 1·7 × 10⁵ cells/2l exchange (n = 43). Of interest, an equivalent CD14⁺CD4⁺ population was detected in peritoneal lavages from subjects undergoing elective laparoscopy, indicating that the presence of such population is not unique to patients on PD.

Fig. 1.

Two subsets of CD14⁺ cells are found in PD effluents. (a) Total number of unfractionated cells from 76 long dwelling, 2 L PD effluents (mean number of cells = 4·7 × 10⁶/2 l; range 1·5 × 10⁵−31 × 10⁶/2 l). (b) Lineage of unfractionated cells as determined by expression of CD14 (myelo-monocytic cells), CD3 (T cells), CD19 (B cells), CD16 (neutrophils), CD56 (NK cells), and lineage negative HLA-DR⁺ (DC). Isotype-matched irrelevant antibodies were used as negative controls for staining with each antibody. Results are graphed as mean and standard deviation of multiple stainings and are representative of 232 fluids from 92 patients. (c) Two colour FACS profile of CD14 and CD4 expression on cells isolated from PD effluents. (d) Confocal microscopy analysis of CD14 and CD4 expression on PD effluent cells (indicated by arrows). Cells were collected, stained at 4°C (1 × 10⁶/plate) and fixed before analysis. Results are representative of stains from 92 stable PD patients. Isotype-matched, irrelevant antibodies were used as controls and are shown in the bottom left panel.

As monocytes and macrophages down-regulate CD4 expression upon migration into tissue and activation [20,21], the higher level of CD4 staining on CD14⁺ peritoneal cells led us to hypothesize that this cell population may represent a discrete dendritic cell precursor stage. To test this hypothesis, we first performed four-colour flow cytometric analyses to characterize the phenotype of these cells (Table 2). The results of these studies showed that a higher proportion of the CD14⁺CD4⁺ cells expressed the costimulatory molecules CD80 (15·45%, n = 15, P < 0·01) and CD40 (24·75%, n = 23, P < 0·05) in comparison to peritoneal macrophages, defined as CD14⁺CD4^–/lo cells (2·07% and 9·82%, respectively). Although both the CD14⁺CD4⁺ cells and the CD14⁺CD4^–/lo cells expressed CD86 (82·09% and 62·25%, n = 6) and HLA-DR (90·97% and 81·02%, n = 25, P < 0·05), the cell surface density (i.e. MFI) of CD40, CD80, CD86 and HLA-DR was approximately two fold higher on the CD14⁺CD4⁺ population (Table 2). The CD14⁺CD4⁺ cell pool also expressed low levels of the IL-3Rα chain (CD123) (18·88%, n = 17, P < 0·05), and levels of CD13 and CD33 comparable to the peritoneal macrophages (CD14⁺CD4^–/lo cells). Together, these results indicate that, although the two CD14⁺ populations most likely arise from the same monocyte precursor pool, the CD14⁺CD4⁺ cells have a phenotype that is more consistent with that of a pre-DC1 population as previously defined for blood DC precursors [7]. In addition, we found that a small percentage of pre-DC1 cells expressed CD83 (12·53%, n = 4) and DC-SIGN (CD209) (12·46%, n = 15, P < 0·05), further supporting that some of the cells were maturing in vivo in the peritoneal cavity [22,23].

Table 2.

Phenotype of CD14⁺CD4⁺ and CD14⁺CD4^-/lo peritoneal cells^*.

	% of positive cells for each marker			Fold difference
Marker	n	CD14+ CD4+	CD14+ CD4–/lo	P-value	CD14+CD4+/ CD14+CD4–/lo
CD14	25	100	100	NS†	1·0
CD4	25	100	100	NS	5·1
CD40	23	24·75	9·82	0·0172	2·2
CD80	15	15·45	2·07	0·0011	2·6
CD86	6	82·09	62·25	NS	1·8
HLA-DR	25	90·97	81·02	0·0356	2·0
CD123	17	18·88	10·49	0·0298	1·5
CD209	15	12·46	4·95	0·0412	1·7
CD83	4	12·53	3·27	NS	1·9
CD33	4	77·79	68·42	NS	1·4
CD13	4	79·75	67·79	NS	1·3
TLR2	4	11·57	5·36	NS	1·2
TLR4	4	9·04	3·62	0·0283	1·4

^*Whole peritoneal cells from stable PD patients were analysed by four-colour flow cytometry. Two populations were gated based on their expression of CD14 and CD4 (CD14⁺CD4⁺ and CD14⁺CD4^–/lo) and analysed for expression of CD40, CD80, CD86, HLA-DR, the IL-3Rα chain (CD123), DC-SIGN (CD209), CD83, the myeloid markers CD33 and CD13, and the Toll-like receptors 2 and 4. Results are expressed as mean percentages and comparisons were made using Student's t-test. The mean fluorescence intensity (MFI) of these markers was also analysed and expressed as a fold difference (CD14⁺CD4⁺/CD14⁺CD4^–/lo).

^†NS: not significant.

It is of interest to note that some of the pre-DC1 cells were interacting with CD4⁺ T cells in the effluents since staining with an antibody against the CD3ɛ chain of the T cell receptor revealed the presence of CD3ɛ⁺ cells within the pool of CD14⁺CD4⁺ cells. Such doublets were resistant to disruption with EDTA. It is of interest to point out that interactions between CD14⁺ cells and CD8⁺ T cells were observed at lower levels despite an inverted CD4:CD8 T cell ratio in the effluents of PD patients ([24] and M.L.M and J.M. personal observation).

The total number of CD14⁺CD4⁺ cells increases during PD-associated peritonitis

As mentioned above, in the absence of infection, CD14⁺ cells and T lymphocytes comprise the two largest populations of immune cells present in the peritoneal cavity. However, at the onset of infection, the population proportions change dramatically as millions of polymorphonuclear cells (PMN) (i.e. neutrophils) are recruited to the peritoneal cavity (Fig. 2a, b top panel). This initial innate response is characterized by the presence of increased numbers of macrophages, in addition to PMN, and an increase in the production of IL-1, IL-6, IL-8, TNF-α and TGF-β [25–27]. By the third and fourth days of infection, the percentage of CD14⁺ cells and T cells increases as the percentage of PMN decreases, indicative of the transition from a predominant innate response to a predominant adaptive response (Fig. 2b).

Fig. 2.

The total number of CD14+ CD4+ cells increases during PD-associated peritonitis. Long dwelling PD effluents from 8 consecutive patients diagnosed with PD-associated peritonitis were collected each day after diagnosis. (a) Total cell count from the effluents from the first four days of clinical peritonitis and (b) The percentage polymorphonuclear cells (PMN), CD14⁺ cells and T cells in whole cell preparations during the course of PD-associated peritonitis caused by Gram positive (▪) and Gram negative (▴) organisms. (c) Average number of CD14⁺CD4⁺ cells under basal conditions and during the first four days of PD-associated peritonitis. (d) Time course of increase in CD14⁺CD4⁺ cell numbers between PD- associated peritonitis episodes caused by gram positive (n = 4) and gram negative (n = 4) organisms. *P < 0·05.

Next, we examined the changes in CD14⁺CD4⁺ cellularity during PD-associated peritonitis. We found that the number of peritoneal pre-DC1 cells increased significantly during the course of PD-associated peritonitis (Fig. 2c). In comparison to the levels found in the effluents of noninfected PD patients, CD14⁺CD4⁺ numbers increased on average 29 fold over baseline on the first day of a clinical peritonitis (Fig. 2c). The number of CD14⁺CD4⁺ cells returned to basal levels by the fourth day of infection. In addition, the rate of increase of CD14⁺CD4⁺ cell number was dependent on the causal pathogen, occurring earlier in patients infected with Gram-positive organisms (n = 4; S. epidermidis (3), E. faecalis (1)) than in patients infected with Gram-negative organisms (n = 4; P. aeruginosa (2), Citrobacter (1), P. agglomerans (1)) (Fig. 2d).

Peritoneal pre-DC1 cells endocytose FITC-Dextran, are potent allo-stimulators, and imprint a Th1 cytokine response

Precursors of DC, including pre-DC1 cells, have a high endocytic capacity, and are not as potent allo-stimulators as mature DC but more potent than macrophages [10]. Therefore, to corroborate our claim that the CD14⁺CD4⁺ cells are pre-DC1 cells, we incubated peritoneal cells with FITC-labelled dextran, a molecule that binds the macrophage mannose receptor (MMR), and measured the endocytosis of this molecule by flow cytometry [3]. We found that CD14⁺CD4⁺ cells had a similar capacity to endocytose via the MMR as blood monocytes and peritoneal macrophages (CD14⁺CD4^–/lo) (Fig. 3). At 0°C, receptor-mediated endocytosis was inhibited, thus serving as a negative control whereby any positive staining at this temperature is indicative of residual, nonquenched cell-surface bound dextran molecules.

Fig. 3.

Endocytic capacity of CD14⁺CD4⁺ and CD14⁺CD4^–/lo peritoneal cells. Unfractionated peritoneal cells (5 × 10⁵/group) isolated from the effluents of stable PD patients (n = 4) were incubated with FITC-labelled dextran for 30 min at 0°C (□) or 37°C (▪). Cells were then stained with anti-CD14 and anti-CD4 monoclonal antibodies to discriminate between the two CD14⁺ populations found in these peritoneal dialysis effluents, and analysed by FACS. Results are plotted as a percentage of cells staining positive for FITC-dextran at 37°C in comparison to those staining positive at 0°C. Peripheral blood monocytes from the same PD patients (n = 2) were used as positive controls. *P < 0·05.

Next, we tested the allo-stimulatory capacity of peritoneal CD14⁺CD4⁺ cells on PBL. We found that, at a ratio of 1 : 50 (stimulators:responders), CD14⁺CD4⁺ cells were significantly more potent stimulators of T cell proliferation than either CD14⁺CD4^–/lo cells or allogeneic PBMC (Fig. 4a). In addition, peritoneal CD14⁺CD4⁺ cells induced a proliferative response that was comparable to that induced by immature monocyte-derived DC (Mo-DC), but not as potent as mature Mo-DC (Fig. 4b). For comparison, both the CD14⁺CD4⁺ cells (P < 0·05) and immature Mo-DC (P < 0·05) induced responses that were significantly greater than those induced by either the CD14⁺CD4^–/lo or the monocyte-derived macrophages (Mo-Mac), and as expected, the CD14⁺CD4^–/lo cells induced a response that was comparable to that induced by monocyte-derived macrophages (Mo-Mac). No significant differences were seen below a ratio of 1 : 50.

Fig. 4.

CD14⁺CD4⁺ peritoneal cells are more potent allo-stimulators than peritoneal CD14⁺CD4^–/lo cells or macrophages. (a) Sorted CD14⁺CD4⁺ or CD14⁺CD4^–/lo cells from PD effluents were incubated with autologous or allogeneic PBL (150 000 cells/well) at a ratio of 1 : 50 (Stimulator: Responder) in a mixed lymphocyte reaction, and cell proliferation measured by ³H-thymidine incorporation. Control experiments included basal PBMC proliferation and PBL response to allogeneic peripheral blood monocytes as stimulators. Results are plotted as the mean c.p.m. ± SEM of triplicate wells and are representative of 4 separate experiments. (b) Allo-stimulatory capacity of CD14⁺CD4⁺ and CD14⁺CD4^–/lo peritoneal cells was compared to the allo-stimulatory capacity of in vitro generated immature and mature monocyte-derived DC and macrophages at a ratio of 1 : 50. Results are representative of 2 separate experiments. *P < 0·05, **P < 0·01, and ***P < 0·001. Auto, autologous; Allo, allogeneic; iMo-DC, immature monocyte-derived dendritic cell; mMo-DC, mature monocyte-derived dendritic cell; Mφ, macrophages. Different stimulator and responder cells were used for experiments shown in Fig. 4a,b.

Since pre-DC1 have been linked to the development of Th1 responses [28,29], we assessed the T cell cytokine profile induced by the peritoneal CD14⁺CD4⁺. We found that allogeneic T lymphocytes cultured with either peritoneal CD14⁺CD4⁺ or CD14⁺CD4^–/lo cells responded with production of IFN-γ, and a high IFN-γ to IL-4 ratio (318 fold and 1130 fold, respectively) compared to the response of the same T lymphocytes to polyclonal mitogens (PMA and Ionomycin) (49 fold) (Fig. 5). Monocytes, when used as stimulators, induced a similar ratio (IFN-γ/IL-4) as that seen for PMA plus Ionomycin (data not shown). This indicates that peritoneal CD14⁺ cells not only induce a potent proliferative response, but also polarize T lymphocytes to secrete Th1 cytokines. Therefore, based on phenotype, endocytosis, allostimulatory potential, and Th1 polarization, we concluded that peritoneal CD14⁺CD4⁺ cells represent pre-DC1 cells.

Fig. 5.

CD14⁺ peritoneal cells induce a profile of Th1 cytokine production by allogeneic T lymphocytes. Sorted CD14⁺CD4⁺ or CD14⁺CD4^–/lo cells from PD effluents were incubated with autologous or allogeneic peripheral blood lymphocytes (PBL) (150 000 cells/well) at a ratio of 1 : 50 (Stimulator: Responder) in a mixed lymphocyte reaction for 48 h, and IFN-γ and IL-4 production were measured by ELISA. Control experiments included PBL cytokine production in response to PMA (100 ng/ml) and Ionomycin (2 µg/ml), or allogeneic peripheral blood monocytes as stimulators (data not shown). Results are plotted as the mean pg/ml ± SD of triplicate wells (top and middle panels) and as a ratio of IFN-γ to IL-4 (bottom panel), and are representative of 4 separate experiments. PMA, phorbol-myristate acetate; IONO, ionomycin.

Peritoneal CD14⁺ cells retain the capacity to differentiate into DC or macrophages

As the peritoneal CD14⁺CD4⁺ cells phenotypically and functionally resemble myeloid DC precursors [7], we examined whether similar cells expressing CD14 and CD4 could be generated in vitro using blood monocytes. We found that 24 h after culture in GM-CSF and IL-4, the pattern of CD14 and CD4 expression on blood monocytes was highly similar to that seen for peritoneal cells in vivo(Fig. 6a, n = 3). Such a response under these culture conditions required the presence of lymphocytes, and was not seen when highly purified monocytes (<5% lymphocytes) were used for this culture or when blood monocytes were cultured with M-CSF. Therefore, an equivalent CD14⁺CD4⁺ cell stage can be generated from blood monocytes, indicating that not only are the peritoneal CD14⁺CD4⁺ cells derived from blood monocytes, but that their state of differentiation is equivalent to a 24 h in vitro culture in DC-inducing conditions.

Fig. 6.

In vitro differentiated DC from peritoneal cells but not peripheral blood monocytes retain CD14 expression. (a) CD14 and CD4 expression on blood monocytes from a healthy donor (left panel) and on monocytes cultured in vitro in the presence of lymphocytes (65%) for 24 h with either GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) (middle panel) or M-CSF (1000 U/ml) (right panel) at a density of 5 × 10⁵ cells/ml. A two-colour stain for CD14 and CD4 expression by peritoneal cells is shown as control. (b) Monocytes from either a healthy donor (top row) or a stable PD patient (middle row), and peritoneal cells from a stable PD patient (bottom row) were cultured in the presence of GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) (DC-inducing conditions) or M-CSF (1000 U/ml) (macrophage-inducing conditions) for 6 days. Cells were collected and analysed by FACS for expression of CD14 and CD4 was assessed by flow cytometry. Results are representative of 3 separate experiments.

Next, we assessed the differentiation potential of the peritoneal CD14⁺ pool (containing both CD4⁺ and CD4^–/lo cells). Two possibilities were considered. One is that, once migrated to the peritoneal cavity, blood monocytes would become either immature DC or macrophages, and thus differentiate separately. Alternatively, one could argue that monocyte-derived cells remain uncommitted after migration to the peritoneal cavity and can differentiate into DC or macrophages under appropriate culture conditions. We distinguished between these two possibilities by culturing peritoneal CD14⁺ cells under conditions leading to DC differentiation (with GM-CSF and IL-4) or to macrophage differentiation (with M-CSF). Separate cultures for each CD14⁺ peritoneal population was not possible due to limitations in CD14⁺CD4⁺ cell number after cell sorting. Peripheral blood monocytes from the same patients were used as controls. It is important to note that in comparison to healthy donors, PD patient blood monocytes had on average lower surface expression of CD4 even after treatment with GM-CSF and IL-4. This finding is not surprising as such a reduction of CD4 on monocytes has been reported previously for patients with impaired immune responsiveness [30]. We observed that, after 6 days in culture with GM-CSF and IL-4, blood monocytes had become DC and all had lost surface expression of CD14 (Fig. 6b). In contrast, half of the peritoneal cells still expressed CD14 after 6 days, and 15% after 9 days of culture under these conditions.

In order to determine if this decrease in CD14 expression after 9 days of culture was a reflection of differentiation into DC, we compared the surface expression of HLA-DR, CD40, CD80, CD16, CD1a, and DC-SIGN (CD209) on peritoneal cells cultured in the presence of GM-CSF and IL-4 for 9 days with the expression of these markers on Mo-DC (Fig. 7). In the absence of exogenous stimuli, the surface phenotype of CD14⁺ peritoneal cells, with respect to the markers listed above, was remarkably similar to that seen on blood monocytes [31,32]. The only noticeable difference between these two populations was the expression of HLA-DR, which was higher on peritoneal cells. After 9 days in culture with GM-CSF and IL-4, we found that the decreased surface expression of CD14 correlated with a concomitant up-regulation of expression of CD40, CD80, CD1a and DC-SIGN (Fig. 7, bottom row). However, even at this time point, the surface levels of CD1a were still significantly lower in comparison to CD1a expression on monocyte derived-DC, suggesting a delayed differentiation response to GM-CSF and IL-4 [33]. When the same pool of peritoneal cells were cultured in the presence of M-CSF for the same amount of time, much like what was seen for monocytes cultured in the same way, the resulting cells retained CD14 and did not up-regulate expression of CD1a or DC-SIGN (Fig. 7, 3rd and 4th rows). Such a unique differentiation profile of peritoneal CD14⁺ cells was not due to the uremic state of these patients as their blood monocytes retain the ability to down-regulate CD14 when cultured with GM-CSF and IL-4 (Fig. 6b).

Fig. 7.

Phenotype of in vitro differentiated DC from peritoneal CD14⁺ cells. Expression of HLA-DR, CD40, CD80, CD16, CD1a and DC-SIGN (CD209) was assessed by flow cytometry on untreated peripheral blood monocytes from a healthy donor (top row) and peritoneal cells from a stable PD patient (second row), on monocytes and peritoneal cells cultured at a density of 5 × 10⁵ cells/ml in the presence of M-CSF (1000 U/ml) for 9 days (third and fourth row, respectively), and on monocytes and peritoneal cells cultured in the presence of GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) for 9 days (fifth and sixth rows, respectively). Dotted lines represent isotype-matched controls. Results are representative of 4 separate experiments. MO-MAC, monocyte-derived macrophages; PD-DERIVED MACROPHAGES, peritoneal dialysis-derived macrophages (denotes macrophages derived in vitro from peritoneal CD14⁺ cells); MO-DC, monocyte-derived dendritic cells; PD-DERIVED DC, peritoneal dialysis-derived dendritic cells (denotes dendritic cells derived in vitro from peritoneal CD14⁺ cells).

The differentiation potential of peritoneal cells into mature DC was further documented by assessing the down-regulation of their antigen uptake ability before and after stimulation with LPS and IFN-γ (Fig. 8) [3]. As expected, nonstimulated peritoneal cells cultured with either GM-CSF and IL-4 or M-CSF had high endocytic capacity as measured by FITC-labelled dextran uptake. However, after stimulation with LPS and IFN-γ, the GM-CSF and IL-4 cultured cells down-regulated their ability to take up FITC-labelled dextran molecules, while M-CSF-cultured peritoneal cells uniformly retained high endocytic capacity. This result confirmed that the peritoneal cells cultured with GM-CSF and IL-4 differentiate into DC that can subsequently mature upon stimulation with LPS and IFN-γ.

Fig. 8.

In vitro differentiated DC from peritoneal cells down-regulate endocytic capacity after stimulation. Peritoneal cells from a stable PD patient were cultured in the presence of M-CSF (1000 U/ml) (top two panels) or GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) (bottom two panels) for 9 days, and then either left untreated (left column panels) or stimulated with LPS (100 ng/ml) and IFN-γ (10 ng/ml) for 24 h (right column panels). Cells were then analysed for endocytic capacity by incubating with FITC-dextran and measuring uptake by flow cytometry (thick solid line). (---) represents isotype-matched control and (······) represents surface FITC-dextran staining at 0°C. Results are representative of 3 separate experiments.

Peritoneal pre-DC1 cells acquire a morphology characteristic of DC upon differentiation

In concert with the phenotypic and functional changes seen in the peritoneal cells cultured for 9 days with GM-CSF and IL-4 or M-CSF, the morphology of the peritoneal CD14⁺ cells also changed over the course of culture under these conditions (Fig. 9). In the absence of cytokines, growth factors and stimulation, peritoneal CD14⁺ cells appeared round by light microscopy and Giemsa staining (Figs 9a,b and 2a). After incubation for 9 days in the presence of GM-CSF and IL-4, some cells remained round but became larger with a smaller and more rounded nucleus (Fig. 9c,d), while others had a morphology characteristic of mature DC with well-defined, sharp membrane protrusions (Fig. 9e,f). For comparison, peritoneal CD14⁺ cells cultured with M-CSF developed a macrophage-like appearance, with large vacuoles and cell spreading, consistent with the phenotype and endocytic capacity seen upon culturing with this growth factor (Fig. 9g,h).

Fig. 9.

Morphology of untreated or differentiated CD14⁺ peritoneal cells. Cellular morphology was analysed by light phase microscopy (a,c,e,g) and Giemsa-stained cytospun cells (b,d,f,h). Briefly, CD14⁺ peritoneal cells from stable PD patient were left untreated (a,b) or cultured in the presence of GM-CSF (1000 U/ml) and IL-4 (1000 U/ml) (c–f) or M-CSF (1000 U/ml) (g,h) for 9 days.

Discussion

Although the peritoneal cavity has traditionally been considered a source of macrophages following local induction of inflammation [34,35], the precise characterization of the mononuclear cells present in the cavity under basal conditions remains unclear. Since some of these cells do not have the phenotypic features of macrophages, a few studies have concluded that the cavity contains a small population of DC representing about 6% of the total cell numbers [13,36]. In this manuscript, we demonstrate for the first time that the peritoneal cavity of both normal human subjects and patients in PD contains a population of CD14⁺CD4⁺ cells that phenotypically, morphologically and functionally fulfills the definition of myeloid DC precursors or pre-DC1. Such a cell population may be generated from peritoneal CD14⁺ cells that retain differentiation potential along both the DC lineage as well as the macrophage lineage. More importantly, we show that peritoneal pre-DC1 cells increase in number during the course of local infections (e.g. PD-associated peritonitis), this occurring more rapidly in infections caused by Gram positive bacteria. The relevance of our findings is twofold. First, the pre-DC1 cell population may play a critical role in the coordination of local immunity along the cell-mediated, Th1-type immune responses as seen in peritonitis. Second, the presence of this population provides a unique experimental setting to study DC biology especially as it relates to in vivo and in situ differentiation of these cells.

Previous studies have shown that PD-related peritonitis correlates with increased levels of IFN-γ levels in the peritoneal effluent, implying that the predominant adaptive immune response to peritoneal infection is cell-mediated [14,15]. DC play a critical role in the initiation of such a response by uptaking antigens from the aetiological pathogen and subsequently migrating to nearby secondary lymphoid tissues to activate T cells. Two DC-dependent factors may affect the type of T cell response induced: one is the location/microenvironment in which the DC originated, and the other is the lineage of the DC as previously described in mice [37,38]. In particular, DC of myeloid lineage or DC1 have been shown to preferentially induce a Th1 response by secreting the Th1-promoting cytokine IL-12 [28]. Thus, our data lead us to conclude that the population of pre-DC1 cells present in the peritoneal cavity, which induced large amounts of IFN-γ release from T cells in vitro, are responsible for the local bias towards cell-mediated, Th1-type immune responses in vivo [4,39]. Current studies are underway to analyse the role that pre-DC1 cells play during PD-related peritonitis as our findings show that the time course with regards to the increase in the CD14⁺CD4⁺ pool is highly dependent on the causal pathogen. In response to Gram positive organisms, the number of pre-DC1 increase dramatically on the first day of infection, which is in contrast to infections caused by Gram negative organisms in which the increase in this pool is delayed until the second day of infection. Although the relevance of this finding is currently unknown, it can be postulated that this profile is a reflection of cellular responsiveness to different pathogen-associated molecular patterns between these two types of microorganisms, and correlates with Toll-like receptor expression [40].

We have shown here that the peritoneal pool of CD14⁺ cells is not terminally differentiated since in vitro culture of these cells with standard polarizing conditions resulted in the generation of DC or macrophages. The interesting aspect of this finding is not that the peritoneal CD14⁺ population of PD patients can become DC after culture, as this has been shown previously in mice (14), but that they have a unique differentiation pathway as illustrated by the fact that these cells do not uniformly down-regulate CD14 expression as peripheral blood monocytes do during their differentiation to DC [41]. After 6 days of culture, peripheral blood monocytes are completely negative for CD14, while two populations of cells, one CD14^–/lo and another CD14⁺, are clearly distinguishable after peritoneal cells are cultured with GM-CSF and IL-4. The possibility that the primary disease responsible for the end-stage renal failure or the uremic state of the patients are responsible for such unique differentiation profile was ruled out by the fact that blood monocytes from the same patients had identical differentiation profiles as monocytes from normal volunteers. Although lactate-buffered PD solutions may inhibit the differentiation and maturation of monocyte-derived DC in vitro [33], this effect is an unlikely cause for our observations since an equivalent CD14⁺CD4⁺ pre-DC1 population was observed in the peritoneal cavity of normal subjects.

Differentiation of monocyte-derived DC in the peritoneal cavity may be driven by specific environmental cues, which can induce the maintenance of CD14 expression. This hypothesis is consistent with the observations from two other groups, both using mouse models, reporting CD14 expression on fully differentiated DC [42] or late DC progenitors [43]. One may argue that local antigens cause a low degree of inflammation and subsequent production of cytokines such as IL-1 or TNF-α, which are implicated in the induction of CD14 expression [44,45]. Similar factors may be operational in other mucosa as suggested by the presence of CD14⁺ DC in mucosal tissues and thymus, all of these being sites of high level of host–pathogen interaction [46–48]. Differentiation of pre-DC1 cells in situ may require direct contact between these cells and CD4⁺ T lymphocytes, through high affinity interactions since these doublets are difficult to disrupt [49]. The selective advantage of CD14 expression by DC in these locations remains to be determined.