Cytornetry (Communications in Clinical Cytometry) 26:172-177 (1996) Establishing a Pure Lymphocyte Gate for Subset nalysis by Flow Cytometry Joseph M. Homtinovich, Sara D. Sparks, and Karen P. Mann Clinical Flow Cytometry, Duke University Medical Center, Durham, North Carolina Development of a more cost-effective and efficient method of performing lymphocyte subset analysis is of continuing importance in clinical flow cytometry laboratories. Current two-color methods utilize forward and right angle light scatter and multiple tubes per sample and are thereby liable to gate contamination. Methods using three-color analysis with CD45 vs. right angle light scatter (RN gating cannot always exclude a contaminating nonlymphoid population. We have established a two tube approach to directly measure total T cells, T suppressor, and T helper subsets, total B cells and total natural killer cells. The technique involves staining of whole blood with a mixture of five monoclonal antibodies conjugated to three fluorochromes: CD4+ CD19 fluorescein isothiocyanate (FITC), + 3 phycoerythrin (PE), CD45 peridin chlorophyll alpha protein (PerCP), CD8+CD16 FITC, +3 PE, and CD45 PerCP. nalysis is performed using a single laser flow cytometer. This method has equivalent recovery to and improved purity of the lymphocyte gate when compared to well-established methods. These antibody combinations additionally allow clear separation of lymphocytes from other leukocytes and debris as well as separation of the T cell helper and suppressor subsets, natural killer cells and B lymphocytes. We additionally provide preliminary data that an accurate lymphocyte subset analysis can be performed on a single tube containing five antibodies (CD4+ CD19 FITC, + 3 PE, and CD45 PerCP), although some measurements are performed deductively. @ 1996 Wiley-Liss. Inc. Key terms: Whole blood, monoclonal antibody, fluorescence intensity, antigen specificity, three-color immunophenotyping, lymphocyte subset analysis Conventional flow cytometry methods employ techniques based on two-color fluorescence and scatter gating from multiple tubes to identify lymphocytes and their subsets (5). lthough this approach is adequate in many samples, gate purity can vary from tube to tube, leading to inaccurate results. n alternative method uses threecolor analysis with CD45 PerCP as a third color and allows easy exclusion of red blood cells and debris. This technique uses a gating strategy based on bright CD45 positive cells and low right angle light scatter (RLS) to identify lymphocytes (6). This method, however, does not have an internal mechanism to identify monocytes and granulocytes, and although the majority of granulocytes and monocytes have distinct CD45 expression and RLS characteristics, some cannot be distinguished from lymphocytes based on these criteria. The approach described in this article, which we have called puregate, avoids these drawbacks by employing a gating strategy designed to isolate lymphocytes from contaminating sources using fluorescence intensity and antigen specificity to differentiate between lymphocytes and non-lymphocytes and T cells and non-t cells using multiparameter flow cytometry with five antibodies labeled with three fluorochromes in a single tube. This method is similar to one proposed by Horan et al. ( 1986) and improved upon by Carayon et al. ( 199 1 ), in which cells in a mixture are classified according to their staining characteristics (4,l). This article describes the puregate method and compares it with the alternative methods mentioned above. MTERILS ND METHODS Specimens used in this study were from those received for lymphocyte subset analysis in the clinical flow cytometry laboratory at Duke University Medical Center (Durham, NC). Specimen preparation was based on standard whole blood lysis procedure as outlined from FCS brand lysing solution (Becton-Dickinson, San Jose, C). ppropriate amounts of blood, approximately 5. X lo5 white blood cells were stained with titered amounts of monoclonal antibodies (Table 1 ). ppropriate titration was determined using Quantum Received for publication pril 16, 1995; accepted July 5, 1995. 1996 Wiley-Liss, Inc.
PURE LYMPHOCYTE GTE FOR SUBSET NLYSIS 173 Table 1 Monoclonal ntibody Combinations and mounts Used" Two-color method Tube FITCb PEb 1 CD45 CD14 2 IgG 1 lgg2 3 CD4 4 CD8 5 CD19 6 CD16 +CD56 CD45/RLS method Tube FITCb PEb PerCP 1 CD4 CD45(2.5pI) 2 CD8 CD45(2.5pl) 3 CD19 CD45(2.5pI) 4 CD16 + CD56 CD45(2.5~1) Puregate method Tube FlTC PE PerCP 1 CD4(1pI) +CD19(5pI) CDS(1OpI) + 3(1pI) CD45(2.5pI) 2 CD8(1pI)+CD16(5pl) (1pl)+3(1~I) CD45(2.5p1) Recovery/purity method Tube FlTC PE PerCP 1 CD14(5pl) (1pI) +3(1pl) CD45(2.5pl) all antibodies were obtained from Becton-Dickinson, San Jose, C except CD19 FlTC (MC, Inc., Westbrook, M) and CD14 FlTC (Coulter Corporation, Hialeah, FL). blopl of Simultest reagent (Becton-Dickinson) combining FlTC and PE conjugated antibodies were used. Simply Cellular Beads (Flow Cytometry Standards Corporation, San Juan, PR). Non-specific binding of antibodies was determined using the negative control microbead and subsequent titering was performed to reduce the negative population below the first log of fluorescence. Total saturation was confirmed by the presence of four distinct positive histogram peaks. Several combinations of multiple monoclonals with the same fluorochrome were examined to determine which combination produced maximal separation of the cell populations (data not shown). The antibody combinations were then used to determine whether quantum yield and/or binding affinity were affected. Peak channels of normal peripheral blood stained with PE alone and +3 PE were measured. Positive lymphocyte populations demonstrated minimal interference with peak channels of 842 and 835, respectively. Similar measurements were made with the combinations of CD4 and CDl9 FITC, and CD8 and CDl6 FITC with +3 PE and CD45 percp added to each tube. CD4 alone and CD4+CD19 produced peak channels of 587 and 592, respectfully, Erom isolated T lymphocytes. CD19 alone and CD4 + CD19 produced peak channels of 488 and 494 from isolated non-t lymphocytes. CD8 alone and CD8+CDl6 produced peak channels of 7 and 716 from isolated T lymphocytes. CDl6 alone and CD8+CDl6 produced peak channels of 592 and 532 from isolated non-t lymphocytes. MI samples were incubated at room temperature for 15 min and then lysed for 1 min with FCS lysing solution. The cells were then centrifuged, washed with phosphate buffered saline and resuspended in.5 ml of.5% formaldehyde. Samples were stored at 4 C overnight and acquired the following morning. ll samples were acquired using a FCSCan instrument (Becton-Dickinson) compensated for three-color analysis. List mode data from the six-tube two-color method were acquired and analyzed with Simulset software (Becton-Dickinson). Lymphocyte purity and recovery were automatically determined and subsequent corrections performed. n isotype-matched negative control was used for marker placement and positive lymphocyte subsets were determined. List mode data for both three-color methods were acquired and analyzed with CELLQuest software (Becton-Dickinson). gate was set on positive cells using CD45 vs. RLS display during acquisition for both three-color methods to eliminate any debris that might be present and 1, events were acquired. The CD45 vs. RLS method analysis was set up to display CD45 vs. RLS and FLI vs. FL2. gate was set on bright CD45 low RLS events corresponding to the lymphocyte population. Display of dual parameters (FLI vs. FL2) of gated lymphocytes provided subset percentages. The puregate method analysis required the display of + 3 vs. RLS and FL1 vs. FL2. gate was set on + 3 negativeow RLS and + 3 bright/ low RLS populations, corresponding to non-t and T cell lymphocytes, respectively. Display of dual parameters (FLl vs. FL2) of both gated populations corresponds to lymphocyte and subsequent subset percentages. Isotypematched controls for the three-color methods were not used since distinct separation of positive and negative populations are easily identified.
174 HORVTINOVICH ET L. Calculation of recovery and purity was performed as outlined by the Centers for Disease Control (CDC) (2) and as follows. large region (region 1 ) is set around the lymphocyte population (see Fig. 1) which includes all lymphocytes and contaminating cells and debris. Figure 1B displays FI.1 vs. FL3 of region 1. Quadrant 4 contains the total number of lymphocytes gated (CD45 + CD14- ) which is used to determine recovery of lymphocytes in the various gating strategies displayed. Purity is determined by dividing the number of lymphocytes in quadrant 4 by the total number of events from the gated lymphocyte region and multiplying by 1. RESULTS We compared three methods of identifying the lymphoid population to determine which produced the best recovery and purity (Fig. 1). This principle is based on differential antigen expression of leukocytes with CD45 and CD14 antibodies as outlined by the CDC (2 and see Materials and Methods). single peripheral blood specimen was stained with CD14 FITC, + 3 PE, and CD45 PerCY. Figure l,b demonstrates the gate used to calculate total lymphocytes and allow determination of the recovery and purity of the gating strategies of interest. Figure 1C,D demonstrates use of conventional FLS and RLS gating on the same sample to identify the lymphocyte population. In this example there is 22% contamination by monocytes (CD45 + CD14 + ) and debris (CD45 - ). lthough this contamination can be compensated for mathematically, tube to tube variation may cause inaccurate results. Figure le,f demonstrates the use of CD45 vs. RLS gating to identify the lymphocyte population. This type of gate forces one to choose between high inclusivity and low purity or low inclusivity and high purity (3), and it can be difficult, if not impossible to exclude all of the monocytes. In this example there is 13% contamination by monocytes, although debris is easily excluded. Figure 1G,H demonstrates the puregate strategy which combines a mixture of two antibodies conjugated to the same fluorochrome ( + 3 PE) in conjunction with CD14 FITC and CD45 PerCP. This strategy allows the exclusion of debris and red blood cells during acquisition using a display of CD45 and RLS. The combination of and 3 allows easy separation of the monocytes with low side scatter from the lymphocytes and shows less than 1% gate contamination. Table 2 compares the recovery and purity of these techniques on six peripheral blood samples (including the one described above, sample 4). ll three methods showed similar rates of recovery although the recovery varied from sample to sample. The puregate method, however, had consistently better purity than the other methods. Five out of six samples achieved greater than 99% purity, while the alternate methods varied in gate purity and achieved less than 99% purity. It is interesting to speculate that the variability of recovery in the pure- s - R L Q S.:. 2 4 6 81 26 2 46 4 6b 6 86 81 16 FLS cd45 PeCP C g - R L S 26 4 6 8 I6 FLS E..I. CD45 PeCP F FIG. 1. Gating strategies used for lymphocyte recovery and purity from a peripheral blood sample stained with CD14 FITC, +3 PE, and CD45 PerCP in a single tube. : Traditional display of FLS vs. RLS used to determine lymphocyte recovery. large region (R1) is set around the lymphocyte population which includes monocytes and debris. B: Display of CD45 vs. CD14 of the gated region R1. Total number of lymphocytes is displayed in quadrant 4 (bright CD45 positive, CD14 negative). C: Same display used in diagram with a region (R2) set around the lymphocyte population. D: Display of CD45 vs. CD14 of the gated region R2. Recovery is determined by dividing the number of events in quadrant 4 by the number of lymphocytes determined in diagram B x 1. Purity is determined by dividing the number of events in quadrant 4 by the total number of gated events x 1. E: Display of CD45 vs. RLS. region (R3) is set around bright CD45 positive, low RLS cells corresponding to the lymphocyte population. F Display of CD45 vs. CD14 of the gated region R3. 6: Puregate display of + 3 vs. RLS. region is set around non-t lymphocytes (R4) and T lymphocytes (R5). H: Display of CD45 vs. CD14 of both gated regions R4 and R5 in diagram G. 2 46 6 86 I CD45 PeCP 2 4 6 8 1 t3 PE H CD45 PeCP
PURE LYMPHOCYTE GTE FOR SUBSET NLYSIS 175 Table 2 % Recovery and Purity nalysis Sample FLS vs. RLS CD45 vs. RLS Puregate Recovery 1 96.5 98. 96. 2 99.6 99.4 98.7 3 99.1 99. 99.4 4 96.3 93.2 98.1 5 98.4 99.1 97.5 6 99. 96.7 99.2 Mean 98.2 97.6 98.2 Purity 1 96.4 78.2 97.5 2 89.7 97.4 99.4 3 9.6 95.4 99.4 4 78.4 86.6 99.2 5 89.8 96.4 99.4 6 84.5 98.3 99.9 Mean 88.2 92.1 99.1 gate strategy may be due to the presence of 3+ CD14- cells in the CD45 + CD14- gate used for calculation; however, this was not directly measured. lthough this initial step demonstrated that this gating strategy provides equivalent recovery and improved purity when compared to conventional gating strategies, we wished to use this method to generate a more detailed analysis of lymphocyte subsets. Figure 2 shows the puregate strategy used to isolate lymphocytes from a peripheral blood sample. The novelty of the puregate method is the combining of two regions to achieve a pure cell-type population. The combination of and 3 PE allows the separation of non-t lymphocytes from non-lymphocytes (i.e., monocytes and granulocytes) and T-lymphocytes from non-lymphocytes. Combining the two isolated regions gives a pure population of lymphocytes free of debris, monocytes, and granulocytes. Clear separation of monocytes and basophils from lymphocytes with this technique can be seen with display of CD45 vs. RLS identifying these populations (Fig. 2B). By adding FITCconjugated CD4 and CD19, the percentage of T helper and B cell subsets can be determined (Fig. 2C). Similarly, suppressor T and natural killer cells can be determined by the addition of FITC-conjugated CD8 and CDl6 (Fig. 2D). lternatively, suppressor T and natural killer cells can be determined deductively on a single tube (see be- low). Twenty-seven peripheral blood samples were analyzed using a single tube containing CD4 + CDl9 FITC/ + 3 PEKD45 PerCP with the puregate method. In all cases the populations were easily distinguished. In 1 of these patients, a second tube containing CD8+ CDL6 FITC/ + 3 PEKD45 PerCP was added and parallel analysis was performed using both Simultest reagents and Simulset software and a three-color analysis with gating by CD45 vs. RLS. Using all three methods, the sum of the helper and suppressor T cells equaled the averaged total T cell percentage within 25% with a maximum variability of < 1%. Lymphosum calculations from the 1 parallel samples ranged from 89-14% using all three methods. The lowest value, which fell outside the acceptable range (9-1 1% ) was determined using Simulset software and reagents. Lymphosum calculations are not meaningful on the samples from the remaining 17 samples, as the subsets are determined in a single tube and therefore, by definition, sum to 1%. Correlation of determination of total T cells, helper T cells, suppressor T cells, B cells, and natural killer cells using the puregate method with the alternate methods is summarized in Table 3. In all samples the puregate method showed good correlation for the total T cells, T cell subsets, and B cells. It consistently underestimated the number of natural killer cells, however, this was most likely because CDl6 was used as a natural killer marker and not a combination of CDl6 + 56 as is used in the other methodologies. Better correlation of natural killer cells was achieved using deductive measurements (see below). Because the puregate method provides a pure population, it should be possible to determine the percentage of natural killer and suppressor T cells by deductive measurement. The clear separation of B cells ( + 3 negativekd4 + CD 19 positive) from non-t cells ( + 3 negativekd4 + CD19 negative) and helper T cells ( + 3 positivdcd4 + CD19 positive) from T cells ( + 3 positivekd4 + CD19 negative) makes this idea plausible. Preliminary results show good correlation for direct measurement of natural killer cells and suppressor T cells vs. deductive measurements using the puregate method (Table 3), however, more testing is needed before clinical adaptation can be expected. DISCUSSION We have suggested an alternate approach to identify a pure lymphocyte population using a combination of CD45 vs. RLS gating and separation of myeloid from lymphoid population by use of multiple antibody combinations. This approach is shown to have a comparable degree of lymphocyte recovery and improved lymphocyte purity when compared to gating by CD45 and RLS while avoiding the problems of tube to tube variation which can be seen using two-color methods. This approach allows the direct measurement of lymphocyte subsets including total T cells, T helper and suppressor cells, total B cells, and natural killer cells in 2 tubes. Initial data suggests that one can additionally determine all of these subsets in a single tube, although some of these are measured deductively. lthough we tested several antibody combinations, we did not make an exhaustive search of all possible combinations. The antibody combinations we suggest are commercially available directly conjugated to the appropriate fluorochrome and are, therefore, easily available to a clinical laboratory. It is possible, however, that alternative combinations exist which would provide similar or even improved results. It is important, however, to appropriately titrate reagents and demonstrate that they can be used in combination and do not affect quantum yield or binding efficiency. n additional advantage to this method is the distinct
176 HORVTINOVICH ET L. B +CO33 PE cd3+cd33 PE FIG. 2. Puregate strategy used to isolate lymphocyte subsets. : Display of +3 PE vs. RLS of peripheral blood sample. The combination of +3 PE allows the separation of non-t lymphocytes (red) and T lymphocytes (green) from non-lymphocytes. Notice clear separation of monocytes with low RLS (cyan) and basophils (black) with lymphocyte populations in comparison with CD45 vs. RLS in diagram 8. B: Display of CD45 vs. RLS of same sample in diagram. Identifi- cation of cell populations in diagram are evident with corresponding colored populations. C and D Display of FL1 vs. FL2 of both gated lymphocyte regions (R1 and R2) isolated using the puregate method. FlTC conjugated CD4 and CD19 identify T helper and B cell subsets, while FlTC conjugated CD8 and CD16 identify suppressor T and natural killer subsets. Table 3 Statistical naylsis of bsolute Lymphocyte Subsets Demonstrating Correlation by Puregate vs. Sirnulset and Puregate vs. CD45/RLS Methods (n = 1) Lymphocyte Suppressor Suppressor Natural Natural subset Total T cells Helper T cells T cells T cellsa B cells killer cells killer cellsa Simulset Corr. Coef..998.989.979.985.982.893.969 R squared.995.979.96.97.965.799.94 CD45/RLS Corr. Coef..998.995.997.997.985.853.962 R Squared.997.99.994.994.971.728.927 Values represent deductive measurements determined using Puregate single tube with CD4 and CD19 FITC. separation of the populations of interest. This separation should allow development of computer software which would permit increased automation of the analysis process. In summary, a five-monoclonal antibody assay for the simultaneous quantification of T cells, helper T cells, suppressor T cells, and B cells in one tube using a single-laser flow cytometer is highly desirable in the clinical laboratory for reasons of cost and efficiency. The puregate method could make this a reality. LITERTURE CITED 1. Carayon P, Bord, Raymond M: Simultaneous identification of eight leukocyte subsets of human peripheral blood using three-colour im-
PURE LYMPHOCYTE GTE FOR SUBSET NLYSIS 177 munofluorescence flow cytometric analysis. J lmmunol Methods 138:257-264, 199 1. 2. Centers for Disease Control: 1994 revised guidelines for the performance of CD4 + T-cell determinations in persons with human immunodeficiency virus (HIV) infection. Morbid Mortal Wkly Rep 43 (RR-3): 1-2 1, 1994. 3. Givan L Flow Cytometry: First Principles. John Wiley & Sons, Inc., New York, 1992. 4. Horan PK, Slezak SE, Poste G: Improved flow cytometric analysis of leukocyte subsets: simultaneous identification of five cell subsets using two-color immunofluorescence. Proc Natl cad Sci US 83: 8361-8365, 1986. 5. Loken MR, Brosnan JM, Bach B, ult K: Establishing optimal lymphocyte gates for immunophenotyping by flow cytometry. Cytometry 11:453-459, 199. 6. Nicholson JK, Jones BM, Hubbard M: CD4 T-lymphocyte determinations on whole blood specimens using a single-tube three-color assay. Cytometry 14685-689, 1993.