Simplified Quantitation of Myeloid Dendritic Cells in Peripheral Blood Using Flow Cytometry

2000 Wiley-Liss, Inc. Cytometry 40:50 59 (2000) Simplified Quantitation of Myeloid Dendritic Cells in Peripheral Blood Using Flow Cytometry John W. Upham, 1,3 * Joachim Lundahl, 1 Hong Liang, 2 Judah A. Denburg, 1 Paul M. O Byrne, 1 and Denis P. Snider 2 * 1 Department of Medicine, McMaster University, Hamilton, Ontario, Canada 2 Department of Pathology, McMaster University, Hamilton, Ontario, Canada 3 Department of Medicine, University of Western Australia, Perth, Australia Received 3 September 1999; Revision Received 20 December 1999; Accepted 21 January 2000 Background: Recognition of the importance of dendritic cells (DC) in the initiation of T-cell dependent immune responses has led to increasing interest in methods for the identification of DC within the circulation. We sought to develop a flow cytometric method that would allow the reliable enumeration of absolute myeloid DC counts in minimally manipulated blood samples. Methods: Myeloid DC were identified by three-color staining of whole blood leukocytes as a discrete population of mononuclear cells expressing high levels of HLA-DR and CD33, yet having little or no expression of CD14 and CD16. This method was analyzed for reproducibility and variation in blood DC number during typical clinical day hours and after exercise. The new method was compared to an established commercial kit method. Results: FACS sorting of the CD33 DC showed that they morphologically resembled immature DC, and developed cytoplasmic projections typical of mature DC following overnight culture in granulocyte macrophage-colony stimulating factor (GM-CSF). Within peripheral blood, these DC were found at a mean concentration of 17.4 5.4 10 6 per liter, corresponding to 0.93 0.27% of mononuclear cells. Comparison of duplicate samples stained and analyzed in parallel showed that the intrasample variability was very low, with an intraclass correlation coefficient of 0.95. The frequency of CD33 myeloid DC and their light scatter characteristics were similar to that of CD11c myeloid cells. Four-color FACS analysis revealed complete identity of CD11c hi, HLA-DR DC with CD33, HLA-DR DC. Only rare CD33 DC coexpressed CD123 and HLA- DR. Numbers of blood myeloid DC, identified by CD33 staining, showed no significant variation during standard laboratory hours. However, their numbers rose significantly during vigorous exercise, in parallel to other blood cells. Conclusions: The method described herein is rapid, reproducible, requires only small volumes of blood, can be readily used by a clinical immunology laboratory, and requires fewer antibodies than a currently available commercial method. Cytometry 40:50 59, 2000. 2000 Wiley-Liss, Inc. Key terms: dendritic cells; whole blood; CD33; flow cytometry Dendritic cells (DC) are unmatched among antigenpresenting cells (APC) in their ability to present antigen to resting T cells, and are believed to play a crucial role in the initiation and maintenance of T-cell immunity (1). The ability to identify and enumerate DC is fundamental to understanding the role these cells may play in the pathogenesis of human disease. In this context, peripheral blood DC are an important and readily accessible cell population, well suited to the repeated sampling that may be necessary in studies of human immune responses and inflammatory diseases. Identification of DC in various tissue sites may be problematic, given that surface antigen expression, morphology, and function vary in association with the maturation or activation status of the DC population(s) being examined in a given tissue (2). This is clearly evident in the circulation: whereas myeloid DC are adept at antigen uptake, they do not exhibit typical dendritic morphology when freshly isolated and they require a period of in vitro culture before acquiring the appropriate costimulatory molecules required for optimal T-cell stimulation. Another problem hampering the identification and enumeration of circulating DC is the absence of a single DC-specific cell marker. Freshly isolated blood DC do not usually express CD1a or CD1c (3,4) and they are negative for the DC-specific antigens CD83 (4) and CMRF-44 (5). Overnight culture is required before the latter two molecules can be detected on blood myeloid DC. Thus, identification of circulating DC by flow cytometry has usually *Correspondence to: Dr. Denis Snider, HSC Room 3N26, McMaster University, 1200 Main St. West, Hamilton, Ontario L8N 3Z5, Canada. E-mail: sniderd@fhs.mcmaster.ca

depended on immunostaining directed against multiple cell surface antigens. Some investigators have identified cells that are class II MHC, yet lack specific lineageassociated antigens found on monocytes, B cells, T cells, and natural killer (NK) cells (6 8). This technique requires a substantial number of lineage-specific antibodies to identify lineage-negative cell types. The lineage-negative DC comprise between 0.1 and 2.0% of peripheral blood mononuclear cells (PBMC) and express either CD11c or CD123 (4,5,8 11). Alternatively, others have used cell sorting to purify myeloid DC populations with potent antigen-presenting cell activity from peripheral blood, characterizing these cells as CD33 (or bright ) and HLA-DR, yet CD14 (or dim) and CD16 (3,12). Although both these techniques are useful for examining functional properties of highly purified blood DC populations, they involve either extensive cell enrichment procedures or a period of in vitro culture. This may alter expression of important surface molecules (9) and make it difficult to rapidly and reliably enumerate circulating DC. Moreover, because of the scarcity of DC in the circulation, large volumes of blood are required for these isolation procedures, making them unsuitable for repeated blood sampling, when necessary over a short period of time. The large number of lineage markers used in the former method necessitates expensive multiantibody combinations for clear DC identification. We describe a method for enumerating the circulating CD33 (bright), CD14, CD16, HLA-DR myeloid DC population, by immunostaining and flow cytometry of lysed whole blood. Using this method, it is now possible to determine absolute myeloid DC counts in peripheral blood. The method is fast (less than 2 h), reproducible, requires only small volumes of blood, and can be performed using equipment available in most clinical immunology laboratories. MATERIALS AND METHODS Buffers and Solutions Except where indicated, all chemicals were purchased from Sigma Aldrich Canada (Oakville, ON). The following reagent solutions were prepared: (a) NH 4 Cl-EDTA lysing solution (154 mm NH 4 Cl, 1.5 mm KHCO 3, 0.1 mm EDTA, ph 7.2). Preliminary experiments indicated that this solution lost activity if stored at room temperature, or at 4 C, for more than 24 h. However, activity was maintained if frozen at -20 C and used immediately after thawing. Accordingly, the solution was always used fresh, or freshly thawed. (b) Washing buffer: 0.15 M phosphate-buffered saline (PBS) plus 0.1% Na azide. (c) Blocking buffer: PBS/ azide plus 2% normal mouse serum. (d) Fixation buffer: PBS plus 1% paraformaldehyde. In experiments involving the B-D DC kit, FACS lysing solution (Becton Dickinson, San Jose, CA) was used as per the manufacturer s instruction. Monoclonal Antibodies (mabs) The following directly conjugated mabs were used: mouse anti-human CD14-FITC, CD16-FITC, CD33-PE, HLA- QUANTITATION OF MYELOID DENDRITIC CELLS DR-CyChrome (Pharmingen Canada, Mississauga, ON), and the relevant isotype controls mouse IgG 1 -FITC, IgG 2a - FITC, IgG 1 -PE, IgG 1 -APC (Becton Dickinson), and IgG 2a - CyChrome (Pharmingen). Antibodies for the B-D threecolor DC kit were purchased for recommended combinations: lineage cocktail (340546), CD123-PE (340545), CD11c-PE (347637), HLA-DR-PerCP(347364), IgG 1 -PE (349043), and IgG 2a -PE (349053). CD33-APC (340474) was a gift from D. Ehman (Becton Dickinson, Mississauga, Canada). Human Subjects Blood was obtained from healthy laboratory personnel. Informed consent was obtained from each subject. Approval for the study was obtained from the Ethics Committee, McMaster University. At each time point, blood was collected into two 2.5-ml EDTA Vacutainer tubes (Becton Dickinson). One tube was sent to the clinical hematology laboratory (McMaster Health Sciences Center) for determination of total leukocyte and differential white cell counts using a Coulter Electronics (Burlington, ON) blood cell counter (model GenS). The second tube was placed on ice immediately and until sampling for immunofluorescent staining for determination of myeloid DC counts. Preliminary experiments indicated that DC numbers in whole blood were stable for up to 4 h, provided the sample was kept at 4 C. Blood was drawn from a total of 35 subjects for the initial studies to determine the percentages and total numbers of CD33 myeloid DC. For the comparison of the test method with the commercial kit method, blood was drawn from eight subjects. In one experiment, blood was collected from seven subjects between 08.00 and 09.00 h, and again 3 and 6 h later, in order to determine if there was any significant daytime variation in myeloid DC numbers. To determine the effect of vigorous exercise on numbers of circulating DC, five subjects had an initial screening exercise test to establish maximum exercise capacity and maximum heart rate. Several days later, the subjects exercised at steady state for 20 min on a cycle ergometer. This level of exercise (work rate and duration) of exercise was chosen because it produces the upper limit of aerobic exercise. Also, it has been shown previously to cause significant elevations in numbers of circulating granulocytes, lymphocytes, and monocytes (13). Blood was collected prior to exercise, during the last 60 s of the 20-min period of exercise, and again at 1 h following the end of the exercise. Preparation of Leukocytes One hundred microliters of EDTA treated blood was added to each 5-ml polystyrene tube (Falcon, Linclon Park, NJ). Red cells were lysed by addition of 3 ml of NH 4 Cl-EDTA lysing solution to each tube. The tubes were immediately vortexed and incubated at room temperature until the solution became translucent, indicating that lysis was complete (usually within 4 6 min). Cells were then washed once in ice-cold washing buffer (PBS/azide) for 7 min at 400g. The viability of leukocytes prepared in this 51

52 UPHAM ET AL. manner was always 95%, as determined by trypan blue exclusion. For the B-D kit and for the four-color experiments, 100 l of blood was e first stained with antibodies prior to treatment with FACSlyse. Fluorescence Antibody Staining for Myeloid DC Nonspecific staining was inhibited by adding 40 l of blocking buffer to each washed cell pellet and incubating on ice for 10 min prior to the addition of directly conjugated mabs. The latter were added at the following volumes: 15 l of CD14-FITC, 15 l of CD16-FITC, 17 l of CD33-PE, and 12.5 l of HLA-DR-CyChrome in a final volume of 100 l. Identical amounts of appropriate isotype control antibodies were similarly added to control cell pellets. Sample tubes were incubated on ice for a further 30 min. Finally, any residual red cells were lysed by the addition of 500 l Erythrolyse (Serotec Inc., Kidington, UK). After final washing, cells were fixed in 400 l of fixation buffer and refrigerated prior to analysis (within 24 h). In a subset of experiments, we compared our method with a commercial kit recently developed by Becton Dickinson for identification of DC subsets in peripheral blood, referred to as the B-D kit. This kit is designed to detect two distinct DC subsets, a larger population of cells expressing CD11c and a minor population of cells expressing CD123 (IL-3R ). DC are defined in this assay as HLA-DR, lineagenegative cells (lacking CD3, CD14, CD16, CD19, CD20, and CD56), and which express either CD11c or CD123. It has previously been shown that the CD123 DC subset lacks CD11c, corresponds to the plasmacytoid T-cell DC described by Grouard et al. (14), stains weakly for myeloid markers such as CD33 and CD13, and appears to migrate to lymphoid organs independently of inflammatory stimuli or exposure to foreign antigens (3,10). In contrast, CD11c, HLA-DR, lineage-negative cells lack CD123 and appear to be identical to the CD33 bright, CD14, CD16 DC described by others (3,8), and which are the focus of our studies. Blood samples were obtained from eight normal individuals. They were processed using the method described herein, along with the three-color method for the B-D kit, strictly according to the manufacturer s directions. Flow Cytometry For the CD33 DC protocol, stained and control cell preparations were analyzed using a FACScan flow cytometer equipped with a 488-nm argon laser (Becton Dickinson Instrument Systems [BDIS], Mississauga, Canada) and operated with CellQuest software. Fluorochrome compensation settings were established using leukocytes stained with anti-cd14 plus CD16-FITC alone, anti- CD33-PE alone, or anti-hla-dr-cychrome alone. Five data parameters were acquired and stored: linear forward light scatter (FSC), linear side angle light scatter (SSC), log FL-1(FITC), log FL-2(PE), and log FL-3(CyChrome). An acquisition gate was established based on FSC and SCC that included both the lymphocyte and monocyte populations (mononuclear cells), but excluded most granulocytes and debris. Each measurement contained 50,000 60,000 events within this mononuclear population. As tubes were stained in duplicate, the resulting DC counts were typically based on 900 1,000 DC, expressed as a proportion of 100 120,000 mononuclear cells. Granulocyte data were not saved, in order to keep computer files to a manageable size. Off-line analysis was subsequently performed with PCLysis software (version 1.1) as obtained from BDIS. A dot plot of FSC versus SSC was established. A region (R1) was drawn around the lymphocytes and monocytes, excluding any residual data from granulocytes and cellular debris (Fig. 1A). Next, a histogram of CD14 CD16 FITC expression was produced using data gated on R1. A second region (R2) was drawn as a histogram region including cells with negative or low CD14/16 expression, thereby excluding the well-defined populations of cells with CD14/16 expression (Fig. 1B). The final dot plot of CD33-PE versus HLA-DR-CyChrome was established by combined gating of events using R1 and R2. Within this last dot plot, a third region was drawn around the CD33 bright, HLA-DR population, termed R3 (Fig. 1C). Analysis of the data obtained by the three-color B-D kit followed the manufacturer s recommendations for off-line analysis of DC within the total white blood cell population. A region was drawn around the lymphocytes, monocytes, and granulocytes excluding cellular debris. Within this region, lineage-negative cells were identified by plotting lineage markers FL-1(FITC) against HLA-DR, FL- 3(PerCP). A second analysis gate of lineage-negative cells was used: HLA-DR CD123 (FL-2, PE) cells were identified in one tube and HLA-DR CD11c (FL-2, PE) cells were identified from a second tube stained in parallel. For some samples (N 8), in addition to the recommended acquisition protocol, data were acquired a second time using the FSC SSC gate as determined for the CD33 DC method. This allowed direct comparison of data obtained on the same cell populations using the two different staining and analysis protocols. A final set of experiments involved four-color staining for combinations of CD33-APC, HLA-DR-PerCP, CD14/16- FITC, and either CD123-PE or CD11c-PE. Antibodies were used at the concentrations stated above or as the manufacturers recommended. Samples were run on a FACSCalibur instrument with a dual laser (488 nm, 635 nm). A total of 50,000 events identified in the mononuclear gate (FSC SSC) were then collected for analysis. Data were gated as for the CD33 method, identifying mononuclear cells that did not express CD14 or CD16. Two-color plots of CD33 versus HLA-DR and CD33 versus CD11c or CD123 were then produced and separate population of CD33 or CD33 cells analyzed for expression of HLA-DR, CD11c, or CD123. Determination of CD33 DC Counts For each tube, the number of cells defined by R3 was expressed as a percentage of mononuclear cells defined by R1. The mean of duplicate tubes was calculated in order to determine the myeloid DC differential count.

QUANTITATION OF MYELOID DENDRITIC CELLS 53 FIG. 1. Identification of circulating CD33 DC by flow cytometry. A C show representative data illustrating the analysis method developed to identify myeloid DC in lysed whole blood following three-color staining using fluorescent mabs CD14-FITC plus CD16-FITC, CD33-PE, and HLA-DR-CyChrome. A sequential gating strategy was employed as outlined in Materials and Methods. A: A mononuclear cell analysis region (R1) applied to FSC SSC data acquired for exclusion of granulocytes and debris. B: The R1 gated events were then analyzed for CD14 plus CD16 staining and negative cells were gated (R2). C: Myeloid DC were thus identified as cells in R3, expressing high levels of HLA-DR and CD33, but having little or no expression of CD14 and CD16, on events gated by both R1 and R2. The number of DC in R3 was expressed as a percentage of the number of mononuclear cells in R1. D F show representative data illustrating the analysis method for identification of myeloid DC bearing CD123. A similar method identified the CD11c DC. D: A total leukocyte cell analysis region (R1) was applied to FSC SSC data acquired with exclusion of debris. E: The R1 gated events were then analyzed for expression of lineage markers (Lin 1) and negative cells gated (R2). F: CD123 HLA-DR DC were then identified as cells in region R4, on events gated by both R1 and R2. Estimates of the absolute numbers of DC in blood were calculated by multiplying the DC differential cell count by the absolute mononuclear cell count. The assumption was that lymphocytes and monocytes (as determined by the hematology cell counter) together constitute the mononuclear fraction (R1) identified by flow cytometry (Fig. 1A). FACS for Isolation of CD33 DC In some experiments, CD33, CD14, CD16, HLA- DR cells were sorted using a FACStarPLUS instrument equipped with an Enterprise 488-nm laser. The same method of staining was used to identify the CD33 DC as for the three-color analysis described above. The sorted cells were then placed in overnight culture in RPMI containing 10% fetal calf serum (FCS) and 10 ng/ml of human granulocyte macrophage-colony stimulating factor (GM- CSF; Pharmingen). Cytospin preparations of freshly isolated and cultured cells were made using a Shandon cytospin apparatus. Slides were stained with May Grunwald Giemsa and examined by light microscopy. Statistics Data are expressed as M 1 SD. Intrasample variability was determined by measurement of the intraclass correlation coefficient (ICC). Variability of counts during daytime hours, the effects of exercise, and the comparison between the test method and the B-D kit were evaluated by paired t-test. Resulting P values 0.05 were regarded as statistically significant. RESULTS Identification and Definition of CD33 Myeloid DC Previous investigators have purified CD33 (bright), CD14, CD16, HLA-DR cells from peripheral blood by FACS, and have shown that these cells possess features typical of DC (3,12). In the current study, we identified this same population of cells by immunofluorescence staining and flow cytometric FACS analysis of small quantities of lysed whole blood, as shown in Figure 1A C. By first gating out cells that were strongly positive for CD14 and CD16, a discrete cell population was identified that

54 UPHAM ET AL. FIG. 2. Morphology of FACS-sorted GM-CSF-stimulated blood DC. CD33 (bright), CD14 -, CD16 -, HLA-DR cells were purified by FACS sorting, using the same staining and gating protocol as described in Figure 1. Cytospin preparations were stained with May Grunwald Giemsa. A: Freshly isolated blood DC. B: DC after 24-h culture with GM-CSF show DC processes. strongly expressed both HLA-DR and the myeloid antigen CD33 (region R3 in Fig. 1C). In preliminary experiments, we found it necessary to use both CD14 and CD16 in the staining protocol, in order to exclude from the analysis both typical monocytes (CD14 high and CD16 ) and the minority of monocytes that coexpress both CD14 and CD16. This latter monocyte subset exhibits a lower intensity of CD14 staining than the more typical CD14 high CD16 monocytes (data not shown). In order to confirm that the CD33, CD14 -, CD16 -, HLA-DR cells were indeed DC, we purified these cells by FACS. As reported by others (3,12), freshly isolated cells were round in shape, had indented nuclei and abundant cytoplasm, but displayed none of the typical dendritic processes associated with mature DC (Fig. 2A). Those processes were readily apparent following overnight culture in the presence of serum and GM-CSF (Fig. 2B). Analysis of the CD33, CD14 -, CD16 -, HLA-DR population demonstrated that they had relatively high FSC (similar to monocytes), but low SSC (similar to lymphocytes). Data shown in Figure 3E,F indicate that those light scatter characteristics are identical to the CD11c DC that are identified by the B-D commercial kit (see below), corresponding to the myeloid DC identified by others (8,9). The exact identity of the CD33 DC with the CD11c DC described by others could only be determined by four-color FACS analysis. In addition, we also identified a second cell population of cells that were CD14 -, CD16 -, but expressed CD33 at low density and were HLA-DR (Fig. 3C). It was possible that some of these cells or those that were CD33 - and HLA-DR were among the plasmacytoid, T-cell type DC described by others (3,10,14). To answer these questions, we performed two types of fourcolor staining to identify mononuclear cells that were CD14 -, CD16 -, HLA-DR, CD33, and CD11c or CD123. Figure 4 describes typical representative data. The CD33 stain was performed with an APC conjugate of the anti- CD33 and identified a CD33 hi, HLA-DR cell similar to the CD33-PE reagent (compare Fig. 4A to Fig. 3C). In addition, a small number of CD33 lo, HLA-DR cells were identified. Staining with CD11c (Fig. 4C) indicated that all of the CD33 hi, HLA-DR cells expressed high levels of CD11c, and reciprocal gating revealed 99% identity between CD11c hi, HLA-DR cells with CD33 hi, HLA-DR cells. All CD33 hi cells expressed low levels of CD123 (Fig. 4B). Among the CD33 lo cells, the vast majority expressed high levels of CD123 and low levels of CD11c, but only rarely (on average 5%, n 3) did these cells express HLA-DR. In contrast, CD123 hi cells that were HLA-DR did not express any CD33. Thus, CD33 hi DC are identical to the CD11c DC described by others and have no expression of CD123. Evaluation of Myeloid DC Counting Protocol Having identified the CD33 myeloid DC, we then applied our simplified technique for routine accurate counting of circulating myeloid DC. Given the low frequency of these cells in the circulation, it was critical to acquire a large number of events within the mononuclear cell light scatter region in order to provide more accurate DC numbers. The resulting counts were typically based on 900 1,000 DC, expressed as a proportion of 100 120,000 mononuclear cells. DC averaged 0.93 0.27% of mononuclear cells (N 35 healthy subjects). Absolute DC counts were then determined indirectly by multiplying the percentage of CD33 DC times the sum of the lymphocyte and monocyte determined on a differential blood cell counter. The mean DC counts were 17.4 5.4 10 6 per liter. Results from duplicate samples, stained in parallel (N 35 subjects), indicated that the degree of intrasample variability was extremely low, with an ICC of 0.95 (Fig. 5). Comparison of the CD33 DC Method With a Commercial DC Kit We then compared the results obtained using the CD33 method of enumerating myeloid DC with the commercial kit. The method for the enumeration of CD123 and CD11c myeloid DC, according to the method developed by Becton Dickinson, using their three-color kit is described in Figure 1 and in the Materials and Methods section. As shown in Figure 6, there was no significant difference between the frequencies of CD33 DC ob-

QUANTITATION OF MYELOID DENDRITIC CELLS 55 FIG. 3. Comparison of CD33 and CD11c DC populations for light scatter characteristics. Representative data of peripheral blood cells from the same subject that were stained using either the test method (CD33) or the B-D kit method (CD11c) Data were analyzed as described in Materials and Methods or in the kit specifications. A,B: The respective FSC SSC profiles in the mononuclear cell gate for each staining method are shown without any gating for DC-specific markers. C: HLA-DR CD33 profile of CD14,CD16 cells gated from data in A, as per the method outline in Figure 1. R3 shows CD33 DC. The region marked? indicates the CD33 lo HLA- DR population that was frequently observed, but not defined (see Results). D: HLA- DR CD11c profile of lineage-negative cells from data in B, as per the kit method analysis procedure. R3 shows CD11c DC. E: Backgated FSC SSC profiles for cells defined by R1 (A) and R3 (C). F: Back-gated FSC SSC profile of cells defined by R1 (B) and R3 (D). tained by our test method and those obtained for the CD11c lineage- negative DC obtained using the commercial kit (P 0.33, N 8). In contrast, the frequencies of CD123 lineage-negative DC were significantly lower than that of CD33 DC (P 0.05). These results were consistent with whether the DC frequency was expressed as a fraction of the PBMC (test method) or as a fraction of the total white blood cell count (B-D kit method). In addition, the total frequency of DC calculated by the commercial method (CD11c plus CD123 ) was significantly greater than that of CD33 DC calculated by our method (P 0.05). Influence of Time of Day and Recent Exercise Two aspects of patient variability that might alter peripheral white blood cell populations are the time of day when the blood is drawn and whether or not the patient has recently exercised. It is unknown if either of these affect the myeloid DC populations in blood although other cells, especially neutrophils, are elevated after exercise (13). In order to determine whether time of blood collection might influence numbers of circulating DC, we examined blood samples collected at three time points over 7 h during normal clinic hours. For these time points, there was no evidence of significant variability in numbers of circulating numbers of DC (Fig. 7). In contrast, vigorous exercise was associated with a significant rise in circulating CD33 DC numbers (Table 1). This increase paralleled increases in granulocyte and monocyte numbers. Both myeloid DC and monocyte numbers returned to the preexercise range within 1 h, whereas granulocyte numbers remained elevated even at 1 h postexercise (Table 1).

56 UPHAM ET AL. FIG. 4. Analysis of CD33 myeloid DC for expression of CD123 and CD11c. A: Whole blood was stained with CD14/16-FITC, HLA- DR-PerCP, and control IgG-APC (left panel) or CD33-APC (right panel) and analyzed as described in Materials and Methods. CD33 hi, HLA-DR and CD33 lo, HLA-DR cells were identified within rectilinear R3 (upper) and R2 (lower). B: Additional staining with CD123-PE allowed definition of three populations of cells expressing various levels of CD33 in combination with CD123 (upper left panel). R6 (upper left rectangle) defined CD123 lo cells with high expression of CD33. These cells corresponded directly to the CD33 hi, HLA-DR DC (right panel). R5 defined CD123 hi cells that have low expression of CD33 but only a small fraction (3%) of these cells expressed HLA-DR (lower right panel). In contrast, 98% of cells defined by no CD33 expression but high CD123 expression were also HLA-DR (lower left panel). C: A similar analysis of CD33 and CD11c coexpression was performed (left panel). CD33 hi, HLA-DR cells (defined in R6) were uniformly (99%) high expressing for CD11c (right panel). Almost all CD11c lo, CD33 lo cells (R5) were negative for HLA-DR expression (lower right panel). A third population (R7) had variable levels of CD11c, no CD33, and 74% of those cells expressed HLA-DR (lower left panel). The data shown are representative of three separate experiments. DISCUSSION We have developed a simple three-color method for the identification of the major subset of myeloid DC in whole blood using commercially available reagents. When combined with direct measurements of lymphocyte and monocyte numbers in whole blood, this technique provides a rapid means to accurately enumerate myeloid DC and can be used in routine clinical immunology laboratories with the most common flow cytometers. One advantage of our method is that it requires four mabs (plus isotype controls), rather than the eight mabs (plus isotype controls) required for the lineage- negative staining method in the B-D kit. By using mab to CD33, we were able to eliminate the need for mabs against CD3, CD19, and CD56 that otherwise are needed to exclude T, B, and NK cells. Moreover, our method is reproducible (Fig. 4), with minimal tube to tube variation. Because the technique is less costly, rapid, and reproducible, and can

QUANTITATION OF MYELOID DENDRITIC CELLS 57 FIG. 5. Intrasample variability in enumeration of circulating CD33 DC. Individual blood samples were stained twice (test 1 and test 2) using the CD33 DC method and analyzed in parallel. Results are plotted test 1 versus test 2. Intrasample variability was determined by calculation of the ICC. The regression (r2) coefficient was also calculated. FIG. 7. Influence of time of day on numbers of circulating DC. Blood samples were collected from seven normal subjects between 8:00 and 9:00 am (time 0), and again 3 and 6 h later. All samples were processed and stained within 4 h of collection. No significant fluctuation in DC numbers was observed during this time. Table 1 Influence of Exercise on Myeloid DC Numbers Relative numbers of Myeloid DC (mean % SE) End exercise Recovery CD33 DC 168.8 16.8* 128.8 22.0 Monocytes 143.5 9.1* 83.5 15.2 Granulocytes 127.5 7.7* 179.3 25.7* Influence of exercise on numbers of circulating DC and other leukocytes. Subjects exercised for 20 min on a cycle ergometer, as outlined in Materials and Methods. Blood was collected immediately prior to exercise (baseline, pre-exercise), during the final 60 s of exercise (End exercise), and 1 h later (Recovery). Data shown refer to absolute numbers of leukocytes, expressed as a percentage of their numbers at baseline. *P 0.05 compared to baseline (100%). FIG. 6. Frequency of CD33 DC corresponds to that of CD11c DC detected by the B-D kit method. Blood samples were obtained from eight individuals and stained in parallel using the CD33 DC method (test: CD14-FITC, CD16-FITC, CD33-PE, HLA-DR-CyChrome) and the commercial three-color DC kit (lineage cocktail FITC, HLA-DR PerCP and either CD11c PE or CD123 PE). The data obtained using the latter kit are presented as CD11c cells, as CD123 cells, and as the total of both populations, as recommended by the manufacturer. In both cases, the monocyte plus lymphocyte gate was used rather than the total leukocyte gate to calculate percentages of myeloid DC. *P 0.05 versus test. utilize very small volumes of blood, it is possible to perform large studies that involve repeated blood sampling. In the current study, we sampled 5 ml of blood (as much as three times per day). However, it is likely that the volume of blood could be reduced to 3 ml or less, making it an attractive method for pediatric studies. Importantly, myeloid DC numbers show minimal fluctuation during normal laboratory hours, so studies can compare samples taken from various subjects any time during the day. Previous investigators have shown that CD33 (bright), CD14, CD16, HLA-DR cells purified from peripheral blood possess many features of DC (3,12). They exhibit the most potent APC activity of all blood leukocytes. Following a short period of in vitro culture, they develop typical dendrites and increase expression of various costimulatory molecules. As shown in Figure 2, we were able to confirm these previous findings, at least with respect to morphology. In addition, we determined that the CD33 DC have light scatter characteristics similar to myeloid DC previously shown to express CD11c. More importantly, this same population could be readily identified in whole blood, thereby avoiding extensive cell enrichment procedures that might influence DC phenotype or function. Using this method, CD33 DC comprised 0.93 0.27% of circulating mononuclear cells in this current study.

58 UPHAM ET AL. Others have estimated that DC comprise between 0.1% and 2.0% of the blood mononuclear cell population (3 5,8,10 12). However, most of these authors employed extensive cell separation procedures in order to enrich DC prior to staining, thereby making it difficult to reliably enumerate DC in the starting blood sample. Indeed, others have shown that significant numbers of DC may be lost during cell separation procedures (16). It may be significant that McCarthy et al. (9), who also stained for DC with minimal manipulation of the blood specimen, also found that DC averaged 0.9% of PBMC. This is remarkably similar to our own results, even though they used a different method, identifying DC as HLA-DR, but lineage negative: lacking CD3, CD14, CD16, CD19, and CD56. This same research group has recently described an alternative method for identifying and enumerating DC in whole blood (15). They suggested that DC average 3.0% or 6.6% of mononuclear cells, depending on whether monocytes are distinguished from DC on the basis of CD14 or CD64 expression, respectively. These findings are clearly at variance with their own previous results (9), the results typically obtained using the commercial B-D kit, and our own findings presented here (Fig. 6). Others have suggested that both monocytes and some DC express CD64, so this may not be an ideal marker to distinguish between these two populations. Therefore, we would argue that Macey et al. (15) may have overestimated DC numbers in peripheral blood. Nonetheless, it appears that a number of subsets of DC exist. Further studies are needed to characterize the different subpopulations of circulating DC. Preliminary data indicate that numbers of CD33 myeloidtype DC in blood decline very rapidly in the context of pulmonary inflammation (J. Upham et al., unpublished observations). In contrast, others have shown that the minor subset of circulating DC that express CD123 appears to migrate to regional lymph nodes in the absence of inflammatory stimuli (10). It will be important to compare these two DC subsets in a variety of human diseases. The risks inherent in relying on a DC differential count are obvious. Particularly for such a rare cell type, the DC differential may be influenced markedly by changes in absolute numbers of mononuclear cells. By combining our FACS analysis with parallel measures of absolute lymphocyte and monocyte counts, our method allows the indirect enumeration of absolute DC counts in peripheral blood. Future studies will address the use of calibrated counting beads to improve the accuracy of absolute DC counts, as has recently been used for enumeration of absolute CD4 T-cell counts. Numbers of circulating DC remained relatively constant over a 6-h period during normal clinic hours (Fig. 7). It is not possible to completely exclude a significant diurnal variation in DC numbers without examining further time points over a 24-h period. Thus, this method can be used for sampling repeatedly during normal daytime hours and allowing a variety of clinical research applications. In contrast, vigorous exercise is clearly associated with an elevation in circulating DC (Fig. 7), a point that future investigators should bear in mind when constructing a normal range for blood DC. It is generally agreed that the leukocyte response to exercise is mediated by a combination of increased cardiac output and catecholamine release (13). We suspect that similar mechanisms are operative with respect to DC. Moreover, our data imply the existence of a pool of marginating DC that can be readily mobilized under appropriate physiological circumstances, in much the same way as mononuclear cells. The recovery of myeloid DC numbers following exercise was similar to monocytes but is more rapid than granulocytes (Table 1). Thus, selective use of the PBMC (monocytes plus lymphocytes) count would be preferred over the total white blood cell count for accurate measures of myeloid DC if exercise, stress, or other causes of increased cardiovascular output are an issue. Otherwise, elevated granulocyte counts would skew the calculations of total DC, based on total white blood cell counts. It is clear from our experiments that the CD33 hi cell population also expresses high levels of CD11c and that there is very close agreement between the numbers of circulating myeloid DC obtained using our method, and the commercial B-D kit, for identification of CD11c DC. With light scatter characteristics and cell culture phenotype also taken into account, we must conclude that the two phenotypes describe the identical cell. Thus, our method identifies the myeloid DC that is CD33 hi, CD11c hi, HLA-DR and does not express CD123. Our method clearly shows that the CD123 hi, HLA-DR DC, described by others (10,14), do not express CD33 and are therefore distinct from the myeloid DC. This population of DC may include those with plasmacytoid features that locate within the tonsils and lymph nodes and express low levels of CD4 (14). However, based on previous work (3,12), our method appears able to identify the circulating DC with (a) potent APC activity and (b) the ability to quickly develop classical dendritic morphology. Only a very small fraction of cells expressed CD33 at low levels and CD123 with HLA-DR. It is not clear if this rare phenotype is a distinct phenotype of blood DC. An interesting large population of cells express low levels of CD33, low levels of CD11c, high levels of CD123, and no HLA-DR. This population is probably blood basophils, as defined by others (17). Future studies are clearly warranted to establish normal ranges of circulating DC in large numbers of normal individuals, and to examine the relationships between various pathological conditions and numbers of circulating DC of various subsets. Cancer patients undergoing chemotherapy or recovering from stem cell transplantation appear to have a lower number of DC in peripheral blood (11). Given that DC can migrate rapidly into the rat lung following inflammatory stimuli (18) and that experimental allergen inhalation induces a rapid fall in circulating myeloid DC in subjects with asthma (J. Upham et al., unpublished observation), it will be equally important to enumerate and study circulating DC in the context of a range of human inflammatory diseases.

ACKNOWLEDGMENTS The authors gratefully acknowledge the technical assistance provided by Lisa Hayes, Dave Tinney, and Barb Bagnorol, and the statistical advice provided by Mark Inman. Flow cytometry was performed within the Flow Cytometry Facility, Department of Pathology, McMaster University. J. Upham is supported by a Neil Hamilton Fairley Fellowship (967211) from the NHMRC (Australia). This work is supported by operating grants from MRC Canada. LITERATURE CITED 1. Steinman RM. The dendritic cell system and its role in immunogenicity. Annu Rev Immunol 1991;9:271 296. 2. Hart DNJ. Dendritic cells: unique leukocyte populations which control the primary immune response. Blood 1997;90:3245 3287. 3. Thomas R, Lipsky PE. Human peripheral blood dendritic cell subsets. Isolation and characterization of precursor and mature antigen-presenting cells. J Immunol 1994;153:4016 4028. 4. Zhou LJ, Tedder TF. Human blood dendritic cells selectively express CD83, a member of the immunoglobulin superfamily. J Immunol 1995;154:3821 3835. 5. Fearnley DB, McLellan AD, Mannering SI, Hock BD, Hart DNJ. Isolation of human blood dendritic cells using the CMRF-44 monoclonal antibody: implications for studies on antigen-presenting cell function and immunotherapy. Blood 1997;89:3708 3716. 6. Egner W, Andreesen R, Hart DNJ. Allostimulatory cells in fresh human blood: heterogeneity in antigen presenting cell populations. Transplantation 1993;56:945 950. 7. O Doherty U, Steinman RM, Peng M, Cameron PU, Gezelter S, Kopeloff I, Swiggard WJ, Pope M, Bhardwaj N. Dendritic cells freshly isolated from human blood express CD4 and mature into typical immunostimulatory dendritic cells after culture in monocyte-conditioned medium. J Exp Med 1993;178:1067 1078. 8. O Doherty U, Peng M, Gezelter S, Swiggard WJ, Betjes M, Bhardwaj N, QUANTITATION OF MYELOID DENDRITIC CELLS Steinman RM. Human blood contains two subsets of dendritic cells, one immunologically mature and the other immature. Immunology 1994;82:487 493. 9. McCarthy DA, Macey MG, Bedford PA, Knight SC, Dumonde DC, Brown KA. Adhesion molecules are upregulated on dendritic cells isolated from human blood. Immunology 1997;92:244 251. 10. Olweus J, Bitmansour A, Warnke R, Thompson PA, Carballido J, Picker LJ, Lung-Johansen F. Dendritic cell ontogeny: a human dendritic cell lineage of myeloid origin. Proc Natl Acad Sci USA 1997;94: 12551 12556. 11. Savary CA, Grazziutti ML, Melichar B, Przepiorka D, Freedman RS, Cowart RE, Cohen DM, Anaissie EJ, Woodside DG, McIntyre BW, Pierson DL, Pellis NR, Rex JH. Multidimensional flow-cytometric analysis of dendritic cells in peripheral blood of normal donors and cancer patients. Cancer Immunol Immunother 1998;45:234 240. 12. Fanger NA, Wardwell K, Shen L, Tedder TF, Guyre PM. Type I (CD64) and type II (CD32) Fc receptor-mediated phagocytosis by human blood dendritic cells. J Immunol 1996;157:541 548. 13. Hoffman-Goetz L, Pedersen BD. Exercise and the immune system: a model of the stress response? Immunol Today 1994;15:382 387. 14. Grouard G, Rissoan M-C, Filgueira L, Durand I, Banchereau J, Liu Y-J. The enigmatic plasmacytoid T cells develop into dendritic cells with interleukin (IL)-3 and CD40-ligand. J Exp Med 1997;185:1101 1111. 15. Macey MG, McCarthy DA, Vogiatzi D, Brown KA, Newland AC. Rapid flow cytometric identification of putative CD14 and CD64 dendritic cells in whole blood. Cytometry 1998;81:199 207. 16. Weissman D, Li Y, Ananworanich J, Zhou L-J, Adelsberger J, Tedder TF, Baseler M, Fauci AS. Three populations of cells with dendritic morphology exist in peripheral blood, only one of which is infectable with human immunodeficiency virus type 1. Proc Natl Acad Sci USA 1995;92:826 830. 17. Agis H, Fureder W, Bankl HC, Kundi M, Sperr WR, Willheim M, Boltz-Nitulescu G, Butterfield JH, Kishi K, Lechner K, Valent P. Comparative immunophenotypic analysis of human mast cells, blood basophils, and monocytes. Immunology 1996;87:535 543. 18. McWilliam AS, Napoli S, Marsh AM, Pemper FL, Nelson DJ, Pimm CL, Stumbles PA, Wells TN, Holt PG. Dendritic cells are recruited into the airway epithelium during the inflammatory response to a broad spectrum of stimuli. J Exp Med 1996;184:2429 2432. 59