Optimized quantification of lymphocyte subsets by use of CD7 and CD33

Authors


Abstract

Identification and quantification of lymphocyte subsets is based on the expression of specific cell surface antigens. As only a minority of these structures is lineage-restricted gating strategies were established, which should permit to include a maximum of lymphocytes, to reach a high purity within the gate and to avoid specific loss of subsets. Two problems remain: First, the incomplete removal of nonlymphoid cells when CD14 is used to exclude them from the lymphocyte gate. Second, the lack of a restricted marker to identify NK cells that are usually defined as CD3-/CD16/56+ lymphocytes, though contaminating monocyte subsets share the expression of CD16, respectively, CD56. This study demonstrates the advantage of CD33 over CD14 at the creation of a pure lymphocyte gate, because CD33 enables the exclusion of all monocyte subpopulations as well as basophils and granulocytes. Independent of the applied NK cell marker mean purity was significantly higher, when CD33 was used (P < 0.001). For the identification of NK cells, CD7 was compared with CD16/56 and with single stained CD56. CD7 and CD16/56 exhibited as equivalent in various CD33 settings (P ≥ 0.173), whereas the mean proportion of CD56+ NK cells was significantly lower (P ≤ 0.008). Use of CD14 entailed a significantly higher amount of CD3-/CD16/56+ cells than of CD3-/CD7+ cells (P = 0.008) because of the remaining CD14-/CD16+ monocytes. As CD7 is restricted to T cells and NK cells in peripheral blood, misclassification of contaminating monocytes is avoided and CD7+ NK cells can be identified by lack of CD3. Applying this new selection of mAbs, we reached a mean purity of ≥99.50% within the revised lymphocyte gate. As gates can be set very broadly, high inclusivity and high purity are not mutually exclusive. We propose the adoption of CD7 and CD33 for the quantification of lymphocyte subsets. © 2013 International Society for Advancement of Cytometry

Identification and quantification of lymphocyte subsets is a main application of flow cytometry ( 1). The discrimination between lymphocytes and other types of leucocytes of the peripheral blood, such as granulocytes, monocytes, eosinophils, and basophils, is based on the expression of specific cell surface antigens (2). As only a minority of them is lineage-restricted, gating strategies were established aiming to include a maximum of lymphocytes and to reach a high purity within the gate (1, 3).

Lymphocyte gating makes use of the differences between the main leucocyte populations in forward and side scatter characteristics (FSC/SSC) ( 1), as well as in the expression levels of the pan-leucocyte antigen CD45 (3). Well-established two-color flow cytometric applications are back gating CD45+++/CD14- lymphocytes in a light scatter gate and subsequently correct the results for each tube depending on the proportion of nonlymphoid cells or debris within the FSC/SSC gate (1). In multicolor settings with three or more fluorochromes, CD45 versus side scatter (CD45/SSC) gating is recommended (3) and used most often (4). Despite these strategies lymphocyte gates remain contaminated by a variable proportion of monocytes and basophils. Often a trade-off of high purity versus high inclusivity, which is required to prevent specific loss of lymphocyte subsets (1), becomes unavoidable.

Lymphocyte subsets include T cells (which are subclassified into T helper cells and T effector cells), B cells, and natural killer (NK) cells. The T cells are characterized by the expression of CD3 while B cells express CD19, which are both lineage-restricted. In contrast, there is no specific antigen to determine the proportion of NK cells ( 3, 5). This subset can be detected only by combining general properties of a lymphocyte population (strong expression of CD45, low forward and side scatter signals) with a certain pattern of antigen expression. NK cells are usually defined as CD3-/CD16+ and/or CD56+ lymphocytes (1) (CD3-/CD16/56+), they do not express CD3, CD19, or CD14.

The vast majority of monocytes (about 90% of all) ( 6, 7) expresses high levels of CD14 and is thus excluded from the lymphocyte gate. The side scatter signals of monocyte cells are usually higher than those of the lymphocytes, but there are overlaps (8) just as in the expression levels of CD45. In addition, there exists a subpopulation with low expression of CD14 and high expression of CD16 (6, 7, 9), which may remain in the lymphocyte gate. Literature claims that monoclonal antibodies (mAbs) against CD16 do not label NK cells specifically in many cases (3). This is why the previously mentioned monocyte population may be misclassified as belonging to the NK cell subset. Despite this most studies on reference values of peripheral blood lymphocyte subsets use CD16 and 56 marked with the same fluorochrome (CD16/56) in combination with CD3 to determine the relative amount of NK cells (10–13). Misclassification may be avoided by use of only CD56, in contrast this of course entails a variable loss of CD16+/CD56- NK cells. Even basophils cannot be removed from the lymphocyte gate by the typical strategy based on CD14. Although basophils are distinctly granulated, they have the same light scattering properties as lymphocytes (1) (low forward and side scatter signals), CD45 levels are slightly diminished (1, 14), CD14 is not expressed at all (15).

In our study, we compared a modified selection of cell surface markers to identify monocytes, basophils, and NK cells with standard protocols. The novel approach allows for the removal of almost all nonlymphoid cells, as well as for the inclusion of a maximum of lymphocytes resulting in a mean purity of about 99.5% within such a revised lymphocyte gate.

MATERIALS AND METHODS

Experiment Overview

Seven combinations of mAbs were compared using a five-color (5Cl) flow cytometric application with regard to their suitability to identify and to quantify lymphocyte subsets in peripheral blood (n = 20). Removal of nonlymphoid cells was done by using CD14, respectively CD33, NK cells were identified by CD7, CD16/56, or CD56. The panel of mAbs is shown in Table 1. In addition the coexpressions of CD4, CD7, and CD16/56 within the CD3- population of the lymphocyte gate were evaluated. A further tube contained CD4 and CD203c to identify monocytes and basophils ( 16) within the lymphocyte gate, respectively, within the CD33dim population with low SSC.

Table 1. Panel of mAbs
PANELFITCPEECDPE-CY5PE-CY7
  1. Applied volume was 3 μl per mAb except for ready-to-use CD3/CD16 + 56 (10 μl) and CD203c (10 μl). The following mAbs were died with PBS: CD3 (1 + 1), CD4 (1 + 2), CD19 (1 + 2) and CD45 (1+4). All mAbs Beckman Coulter, Inc., Brea, CA USA, except CD7 PE and CD3/CD16 + 56, which are distributed by BD Biosciences, Franklin Lakes, NJ USA.

115 μl37453319
219 μl316 + 56453319
315 μl356453319
415 μl37451419
519 μl316 + 56451419
615 μl356451419
715 μl33345719
819 μl316 + 564574
922 μl3203c45334

Flow Sample and Specimen Description

All 20 peripheral blood samples were obtained from adults (m/f: 9/11, 21a–83a), who had given informed consent for analysis of lymphocyte subsets on a routine basis as part of treatment contract. Leucocytes and lymphocytes were within normal adult ranges (4.0–10.0 G/l, respectively 1.0–3.5 G/l) in all cases. We applied a standard lyse-wash method, staining 5 × 105 white blood cells with 15, 19, or 22 μl (corresponding to Table 1, panel of mAbs) of a combination of five mAbs per tube. Fluorochromes were FITC/PE/ECD/PE-CY5/PE-CY7. After incubation for 15 min at room temperature and lysis of erythrocytes (OptiLyse C, Beckman Coulter, Brea, CA), samples were centrifuged for 5 min at 400g. The supernatant was removed and the cells were resuspended in phosphate buffered solution (PBS) containing 0.2% of natrium azide.

Instrument and Data Analysis Details

Samples were acquired with a 5Cl single-laser flow cytometer (Cytomics FC500, Beckman Coulter, Inc., Brea, CA) using CXP acquisition software (CXP software distributed by Beckman Coulter, Brea, CA). Compensation was accomplished automatically by use of Flow-Set™ Fluorospheres (Beckman Coulter, Inc., Brea, CA) and single-stained tubes containing CD45 in FITC, respectively, in PE, ECD, PE-CY5, or PE-CY7 and subsequent verification by a combination of CD4/CD8/CD45/CD3/CD19 (Beckman Coulter, Brea, CA). Fine tuning was done manually. Threshold was set on FSC (100). Flow cytometer performance was checked using Flow-Check™ Fluorospheres (Beckman Coulter, Brea, CA). Data analysis was performed by CXP analysis software, which is based on continuous gating (demonstrated in Fig. 1). This concept can be explained as follows: If a region A is created within the whole of events, it may serve as a gate A, which is restricted to events with the property A. When a region B is subsequently set within gate A, all events within B are sharing the properties A and B. A new gate B = A AND B is automatically defined by a Boolean operation and restricted to events with both properties. A further region C within gate B causes gate C = A AND B AND C and so on. For each approach a basic protocol consisting of a fixed configuration of dot plots, regions (color coding distinct populations) and crosslines was applied. Exact positions of regions and crosslines were adjusted for each list mode file. Arithmetic operations were subsequently achieved by means of an Excel sheet (Microsoft), because the analysis software allows for exporting final data of single-file or multifile analysis to this spreadsheet program.

Figure 1.

(a) Gating strategy for quantification of lymphocyte subsets. Black full lines stand for continuous gating, dotted lines stand for back gating of color coded populations within CD45/SSC. (b) The same gating pathway was used for all combinations of CD14, CD33, CD7, CD16+56, and CD56. In Figures 1a and 1b, the new selection of mAbs (CD33/CD7) is compared with the commonly used combination of CD14/CD16+56, corresponding dot plots with different configuration of mAbs are highlighted. In this example purity is higher for CD33/CD7 (99.7% vs. 97.6%), whereas the proportion of NK cells is higher for CD14/CD16+56 (19.2% vs. 18.7%). Stricter gating of lymphocytes in the CD14/CD16+56 setting results in a lower proportion of NK cells (18.4%, dot plots 5c–8c) at concurrent increase of purity (98.5%). This may be due to the exclusion of CD16+ monocytes. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Gating Strategies

The gating strategy applied to tube 1–7 (Table 1, panel of mAbs) is described in Figure 1. We calculated the proportions of T cells, B cells, and NK cells within a revised lymphocyte gate after elimination of nonlymphoid cells. As the quantification of all individual subsets was done within each single tube, purity could therefore be defined as the sum of T cells, B cells, and NK cells. The evaluation of coexpressions of CD7, CD16/56, and CD4 (tube 8) within the CD3- population of the lymphocyte gate is described in Figure 2. Monocytes and basophils could be easily identified by the expression of CD4, respectively, CD203c ( 16) within the CD33dim population with low SSC (tube 9, Fig. 3) as well as within the lymphocyte gate.)

Figure 2.

Co-expression of CD7/CD4 and CD16+56/CD4. Whereas CD3-/CD7+ cells exhibit as CD16+56+/CD4-, CD3-/CD16+56+ cells divide into CD7+/CD4- and CD7-/CD4dim subsets. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Figure 3.

Identification of CD33dim/SSClow cells. The CD33dim subset with low side scatter divides into CD4dim monocytes and CD203c+ basophils. Back gating in CD45/SSC exhibits different expression levels of CD45, both populations can be found within a broad lymphocyte gate. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Figure 4.

Main results. (a) Comparison of lymphocyte gate purities. Highest purity was reached by use of combinations containing CD7/CD33 respectively CD16+56/CD33, but there was a significant advantage over CD56 in the CD33 setting (n = 20). Differences were moderate in all cases except of case 18 referring to an increased proportion of CD16+/CD56- NK cells. An unusually low expression of CD33 was observed for case 20 resulting in lower gate purity because of incomplete removal of both monocytes and basophils (Fig. 5). When CD14 was used to eliminate non-lymphoid cells from the lymphocyte gate, purity was significantly higher for CD16+56/CD14 in comparison with CD7/CD14 and CD56/CD14 due to remaining CD14-/CD16+ monocytes. (b) NK cells in corresponding settings. For all NK cell markers lymphocyte gate purity was significantly lower, if nonlymphoid cells were excluded by use of CD14 instead of CD33. As data were not scaled to 100%, even a lower percentage of NK cells was expected for the CD14 setting. Whereas this could be verified for use of CD7, mean values of NK cells were slightly higher for CD56/CD14 versus CD56/CD33 and distinctly higher for CD16+56/CD14 versus CD16+56/CD33 due to remaining misclassified CD14-monocyte subsets with expression of CD16 respectively CD56. Also, this fact is reflected in the difference of purities of CD16+56/CD14 and CD7/14. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Figure 5.

Case 20. Unusually low expression of CD33 results in contamination of the revised lymphocyte gate by CD4dim monocytes and CD203c+ basophils. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Statistics

Statistical analysis was done by SPSS Statistics 17.0 (IBM). If the P-value was lower than the level of significance (α = 0.05), results were regarded as significant. Kolmogorov-Smirnov (KS) test was used to test for Gaussian distribution. As all data were normally distributed, means were compared by t-test for paired samples.

RESULTS

Main results are graphically demonstrated in Figure 4. Figure 4a shows the comparison of seven combinations of CD14, CD33, CD7, CD16/56, and CD56 with reference to gate purity. Independent of the applied NK cell marker mean purity was significantly higher, when CD33 was used to remove nonlymphoid cells (P < 0.001). For the identification of NK cells CD7 and CD16/56 exhibited as equivalent (P ≥ 0.173), but there was an advantage over CD56 PE (P ≤ 0.008). Mean values of NK cells were 13.86% (CD7 PE), 14.11% (CD7 PE-CY5), 13.93% (CD16/56 PE), and 13.20% (CD56 PE). Within the combinations containing CD14 the use of CD16/56 caused significantly higher mean gate purity (P < 0.001) and a significantly higher mean proportion of NK cells (P ≤ 0.008). Mean values were 13.67% (CD7 PE), 14.46% (CD16/56 PE), and 13.31% (CD56 PE). Results of the corresponding combinations of NK cell markers with CD33, respectively, CD14 are compared in Figure 4b.

When the coexpressions of CD4, CD7, and CD16/56 were evaluated within the CD3- subset of a broad (nonrevised) lymphocyte gate, CD7+ cells did not show relevant signal levels of CD4, whereas CD16+56+ cells divided into two populations with expression of either CD7+/CD4- or CD7-/CD4dim (Fig. 2). The CD33dim/SSClow population clearly divided into CD203c+ basophils (mean 33.9% of the population) and CD4dim monocytes (mean 58.3%) (Fig. 3) and proved to be part of a broad lymphocyte gate.

DISCUSSION

Remaining problems in the quantification of lymphocyte subsets are first the incomplete removal of nonlymphoid cells from the lymphocyte gate, second the lack of a restricted marker to identify NK cells ( 3, 5). Although the vast majority of monocytes can be easily excluded from the lymphocyte gate due to high levels of CD14 (6, 7), the discrimination of subsets with low expression or absence of CD14 is not as well defined. Although commonly used, we do not regard CD14 as the most suitable for lymphocyte gating, because basophils do not express CD14 at all. In contrast, basophils and monocytes show a distinct expression of CD33 (14, 15). This membrane protein belongs to the family of sialic acid-binding immunoglobulin-like lectins (siglecs) (17, 18) and is usually classified as “classic myeloid antigen” (19, 20). Previously it was believed to be restricted to the myelo-monocyte lineage (17) until a lymphoid CD33 protein with lower molecular weight was discovered on activated T cells and NK cells in vitro (18): Fresh peripheral blood lymphocytes were tested negative for CD33 and subsequently cultured in the presence of alloantigens and IL-2 for up to 4 weeks. Under these conditions activated T cells and NK cells exhibited an increasing proportion of CD33+ subpopulations. Despite this, we propose that this finding does not play a role in routinely obtained and processed peripheral blood. Evaluating CD33 versus side scatter in the leucocyte gate, we detected the granulocytes with medium expression levels of CD33 and the monocyte population with strong expression of CD33 and further a third population with constantly diminished expression of CD33 (9) and low side scatter signals. When it was back gated in CD45/SSC, the CD33dim population divided into two subsets with an essential difference in the expression levels of CD45. We could prove that CD45+++ cells were CD4dim monocytes, whereas the CD45dim subpopulation exhibited as basophils, which could be identified by the expression of CD203c (16) (Fig. 3). An obvious advantage of CD33 over CD14 is the exclusion of basophils, all monocyte subpopulations and any remaining granulocytes from the lymphocyte gate resulting in very high gate purity. An unusually low expression of CD33 was observed for case 20 resulting in lower gate purity due to incomplete removal of both monocytes and basophils (Fig. 5). Despite this, gate purity was still better than for use of CD14, because in addition to the subset with comparatively higher expression of CD33 even part of the population with low side scatter signals and diminished expression of CD33 could be excluded, which is usually CD14-.

The second problem in lymphocyte gating affects the identification of NK cells, which are usually defined as CD3-/CD16/56+ lymphocytes ( 1), sharing the expression of CD16 (6, 7, 9), respectively, CD56 (3, 5, 21) with subsets of T cells and monocytes. Previous studies suggest that mAbs against CD16 do not label NK cells specifically in many cases (3), so the CD16+ monocyte population may be misclassified as belonging to the NK cell subset. We could demonstrate this by detecting the highest mean proportion of NK cells of all combinations of mAbs at concurrent use of CD14 and CD16/56. In contrast, single use of CD56 may take a variable loss of CD16+/CD56- NK cells, which we could confirm too. The mean proportion of NK cells was significantly lower in the CD33 setting, when CD56 was used to identify them. The resulting differences in purity were moderate in all cases except of case 18 (Fig. 4). Beyond that 2 to 10% of CD56+ cells without expression of CD3, CD19, and CD14 do not express any typical NK cell associated receptor. The authors of a recent study (5) on functionally distinct subsets of human NK cells and monocyte/dendritic like cells describe this population as exhibiting a pattern of myeloid associated markers and identified it as a mix of monocytes and dendritic cells. The discrimination between CD56+ NK cells and CD56+ monocytes was done by the use of CD7. This membrane protein is a single-domain molecule of the immunoglobulin superfamily. Although it is expressed in early stages both of lymphocyte and myeloid development ( 22), CD7 is restricted to T cells and NK cells in peripheral blood (23, 24). In our study, we demonstrated the equivalence of CD7 and CD16/56 in reference to the identification of NK cells. Whereas the expression levels of CD7 are not consistent in T cells and a subset of them does not express CD7 at all, it seems to be strongly positive for almost all NK cells. Besides, we could prove the lack of any relevant coexpression of CD7 and CD4 on CD3- lymphocytes. We consider CD7 to be the perfect option for the identification of NK cells, as those can be easily distinguished from T cells by the absence of CD3. In addition, the relative amount of CD7- T cells contains useful information, because CD7 is absent on malignant T cells in many cases (22).

Use of CD14 and CD16/56 requires tight gating to minimize contamination with basophils and CD16+/CD14- monocytes. In contrast, applying CD33 instead of CD14 for the exclusion of nonlymphoid cells and of CD7 instead of CD16/56 or CD56 for NK cell identification allows for creating a much broader lymphocyte gate. In multicolor settings it may be useful to stain T cells and B cells in the leucocyte gate. Back gating of the color coded populations in CD45/SSC helps to adjust the lymphocyte gate optimally, especially if these cells exhibit unusual high side scatter signals. Use of CD33/CD7 causes a mean purity of about 99.5% within the revised lymphocyte gate. It avoids contamination by misclassified monocyte subpopulations as well as specific loss of lymphocyte subsets. As high inclusivity and high purity can be achieved concurrently, the comparability of reference values may be increased. We propose the adoption of CD7 and CD33 for the quantification of lymphocyte subsets.

Ancillary