Identification of Eosinophils in Lysed Whole BloodUsing Side Scatter and CD16 Negativity

Ramya Gopinath* and Thomas B. NutmanHelminth Immunology Section, Laboratory of Parasitic Diseases, National Institutes of Health, Bethesda, Maryland

The identification of eosinophils in lysed whole blood by flow cytometry can be problematic, since thesecells overlap significantly with the neutrophil cluster on forward scatter versus side scatter plots of wholeblood samples. Current methods can be time-consuming when running multiple samples or may compromiseyield in the interests of greater accuracy. The use of eosinophil purification techniques prior to FACS analysisor sorting as a way of ensuring purity may have unpredictable effects on eosinophil activation, leading toquestionable data interpretation. Here we describe a simple, single-step method for definition of eosinophilsutilizing their high side scatter and CD16 fluorescence negativity to differentiate them from neutrophils. Thepurity of the neutrophil and eosinophil populations sorted with this gate is close to 100% regardless of theperipheral blood eosinophil count, while the population obtained by sorting on a plot of FSC/SSC was amixture of eosinophils and neutrophils. We suggest this method as a simple, reproducible, and accurate wayof defining eosinophils by flow cytometry for analysis or sorting. Cytometry 30:313–316, 1997.r 1997 Wiley-Liss, Inc.

Key terms: CD16 negativity; eosinophils; gate; side scatter; whole blood

The use of lysed whole blood in flow cytometry allowsthe study of surface molecules on cell populations such asgranulocytes, monocytes, and lymphocytes without usingcell purification techniques that may affect expression ofthese markers (7). Due to the presence of granulescontaining major basic protein (MBP), eosinophilic cat-ionic protein (ECP), eosinophil peroxidase (EPO), andeosinophil-derived neurotoxin (EDN), eosinophils displayhigh side scatter and are found above the neutrophilcluster in plots of forward (FSC) versus side scatter (SSC)on whole blood specimens. An accurate delineation of theeosinophil cluster is difficult. Defining the eosinophilpopulation on forward and side scatter alone is problem-atic, because it often overlaps significantly with theneutrophil cluster. CD16, the FcgRIII receptor found onnatural killer cells, neutrophils, and macrophages, is in-volved in phagocytosis, ADCC, and cytokine production.In the granulocyte lineage, CD16 is expressed initially atthe metamyelocyte stage (3) and almost uniformly ex-pressed on bands and segmented neutrophils. Eosinophils,by contrast, are for the most part CD16 negative, althoughlow-level expression of this marker can be seen wheneosinophils are stimulated in vitro under certain condi-tions (6). Back-gating using CD16 fluorescence negativityor gating on the neutrophil cluster and then using CD16negativity has been frequently used as a method ofdefining the eosinophil cluster, but these involve at leasttwo steps and may be time-consuming when runningmultiple samples. Here we describe a single-step approach

that clearly isolates the eosinophil cluster both visually andby fluorescence by combining the unique features ofeosinophils, i.e., side scatter and CD16 fluorescence nega-tivity, in a single plot.

Samples of whole blood were obtained from fifteenpatients with eosinophilia of parasitic origin. Blood wascollected in EDTA Vacutainer tubes (Becton DickinsonVacutainer Systems, Mountain View, CA), and flow cytom-etry was performed on these samples as part of a study ofeosinophil activation. Anti-CD16-PE (Caltag, South SanFrancisco, CA) was used in all tubes to distinguish neutro-phils from eosinophils in combination with a set panel ofFITC-conjugated monoclonal antibodies to surface mol-ecules expressed by eosinophils.

Whole blood (50 µl) was placed in prewetted polypro-pylene tubes. Five microliters each of anti-CD16-PE andthe specific FITC-conjugated eosinophil surface markerwas added to each tube and incubated in the dark for 30min at 4°C. Between steps, cells were washed in Hank’sbalanced salt solution (HBSS) (without Ca21, Mg21, orphenol red) with 0.2% low-endotoxin BSA (Sigma Chemi-cal Co., St. Louis, MO) and 0.1% NaN3 (Sigma ChemicalCo.) and spun at 1,000 rpm for 4 min. Red cells were lysedusing 1X FACS lysing solution (Becton Dickinson Immuno-

*Correspondence to: Ramya Gopinath, Building 4, Room 126, NIAID,NIH, Bethesda, MD 20892.

E-mail: rgopinath@pop.niaid.nih.govReceived 13 May 1997; Accepted 23 July 1997

Cytometry (Communications in Clinical Cytometry) 30:313–316 (1997)

r 1997 Wiley-Liss, Inc.

cytometry Systems, San Jose, CA) twice at room tempera-ture, first for 10 min and then for 3 min with a washbetween lysing steps. After a final wash in HBSS, cells wereresuspended in 250 µl of PBS with 1% paraformaldehydeand analyzed within 24 h.

Samples were analyzed on a FACSorty flow cytometer(Becton Dickinson). Mouse IgG2a conjugated to FITC, PE,and Tricolor (UPC-10, Caltag) was used as an isotypecontrol, and compensation was adjusted using single-colorcontrols anti-CD3-FITC and -PE (UCHT1; Immunotech,Westbrook, ME). Various methods were evaluated todefine the eosinophil population; the most accurate isola-tion of the eosinophil cluster from the other granulocyteswas obtained by using a plot of SSC (X axis) versusanti-CD16 PE (Y axis). A gate was drawn around theeosinophil cluster defined in this way and applied to theplot of FSC versus SSC of the isotype control of eachsample to define quadrants. All data from tubes containingeosinophil surface markers were subsequently collectedusing this gate. Data were analyzed using CELLQuesty(Becton Dickinson) software. To demonstrate comparabil-ity of results obtained by our method of gating and aconventional two-step approach (gating on the neutrophilcluster and then separating the neutrophils and eosino-phils by CD16 fluorescence), our 15 samples were ana-lyzed using gates drawn in two ways. The results of thisanalysis are shown in Figure 1.

To confirm the purity of neutrophil and eosinophilclusters distinguished using our method, these popula-tions were sorted on venous blood drawn from a healthyvolunteer. The sample was processed as previously out-

lined except that 500 µl of whole blood were stained with20 µl each of the appropriate monoclonal antibodies andresuspended in a final volume of 1 ml PBS with 1%paraformaldehyde in order to increase the number of cellsavailable for sorting. Sorting was carried out on a BectonDickinson FACSTAR Plusy. Cells were sorted based ongates defined on a SSC versus CD16 fluorescence plot andcollected in PBS. They were then spun down and resus-pended in 100 µl RPMI with 10% fetal calf serum. Acytospin preparation was made from each population,fixed in methanol, and stained with modified Wright-Giemsa stain for examination under light microscopy. Fora comparison of the yield and purity of eosinophils gatedusing an alternative method, cells were also sorted on aFSC/SSC plot back-gated using CD16 fluorescence negativ-ity, collected in a separate tube, and processed similarly.To demonstrate that our method of defining an eosinophilgate was valid independent of the number of eosinophilspresent in a particular sample, we repeated this procedureusing blood from a patient with 31% eosinophilia at thetime of study.

The populations obtained in the sorting experimentwere subjected to post-sort analysis using the CELLQuestysoftware. The results showed that the purity of theneutrophil population obtained using our method was99.2% in the normal volunteer and 99.5% in the eosino-philic patient. The purity of the eosinophil population inthe normal volunteer was 99.1% and 100% in the eosino-philic patient. In contrast, back-gating to the FSC/SSC plotbased on CD16 negativity resulted in a sorted populationconsisting of 6.7% eosinophils and 92.7% neutrophils inthe normal individual and 70% eosinophils and 29.8%neutrophils in the eosinophilic patient. Further confirma-tion of purity was seen in the cytospin preparations (Fig.2), which visually demonstrate the purity of the popula-tions isolated by our method as opposed to the mixedpopulation of cells sorted by back-gating.

The isolation of eosinophils from whole blood speci-mens by flow cytometry takes advantage of their high sidescatter properties and the absence of the CD16 surfacemolecule that distinguishes them from neutrophils. Eosino-phils are concentrated above the neutrophil cluster inplots of FSC versus SSC but, because of variation in theirsize and the density of their cytoplasmic granules, they arefrequently found distributed throughout the neutrophilcluster. Other cell populations of interest are unlikely tofall into the eosinophil cluster defined by these param-eters. Immature neutrophils, such as promyelocytes, donot express CD16 (3) but can be distinguished fromeosinophils by virtue of their greater size and granularity.CD16 expression on neutrophils is also reduced in parox-ysmal nocturnal hemoglobinuria (PNH) (8), chronic my-eloid leukemia (CML) (3), and AIDS (2); however, sincethese conditions are relatively uncommon and usuallysymptomatic, especially at later stages, it should be pos-sible to identify those patients in which alternative meth-ods of gating on eosinophils may be needed. NK cells areCD16 positive and therefore would be clustered withneutrophils on a plot of SSC versus CD16. Basophils

FIG. 1. Comparison of the percentage of the gated population that wereeosinophils utilizing two methods of gating. The Y axis displays thepercentage obtained by plotting SSC versus CD16, and the X axisexpresses the percentage of eosinophils by first gating on the neutrophil/eosinophil cluster on a FSC/SSC plot and then separating this cluster on aplot of CD16 PE versus a FITC-conjugated monoclonal antibody.

314 GOPINATH AND NUTMAN

FIG. 2. Comparison of the purity of cell populations obtained by sortingeosinophils from the same patient based on a plot of FSC/SSC versus aplot of SSC/CD16. Top: Plot of FSC/SSC and a cytospin preparation of the

sorted population. Bottom: Plot of SSC versus CD16 PE and the cytospinpreparations of the neutrophil population (upper) and the eosinophils(lower) sorted using these gates.

315IDENTIFICATION OF EOSINOPHILS IN FLOW CYTOMETRY

normally occur in very small numbers in peripheral blood,and a combination of CD45 and IgE has been recom-mended (1, 5) for their identification. Furthermore, theyoccupy a position between the lymphocyte and monocyteclusters on a plot of FSC versus SCC, which also separatesthem definitively from eosinophils.

Attempts to gate on eosinophils solely on forward andside scatter, or even by back-gating from a fluorescenceplot, are imprecise and compromise the purity of thepopulation to be analyzed or sorted. Alternatively, restrict-ing the gate to those cells with the highest side scatter inan effort to obtain the purest population results in theartificial selection of hyperdense eosinophils and excludesa significant portion of the smaller hypodense cells.Furthermore, although most eosinophils display autofluo-rescence, we did not find that this feature offered sufficientprecision to be used alone to define the eosinophilpopulation. Efthimiadis et al. (4) have recently describedan alternative single-color immunofluorescent methodutilizing FITC to identify eosinophils, but other, moreprecise methods of gating are quite involved, often usingtwo or more plots to define the eosinophil gate. Thurau etal. (9) also used CD16 negativity and side scatter to defineeosinophils but suggested that it be used in combinationwith autofluorescence. Furthermore, they recommendedthe use of CD49d positivity in lieu of CD16 negativity; wehave found that eosinophils from patients infected withhelminths express CD49d variably, and this marker wouldbe unlikely to reliably define the entire eosinophil popula-tion.

We have found that a single plot of side scatter versusCD16 fluorescence is a highly effective and direct way ofdelineating not only the eosinophil cluster but also theneutrophil cluster clearly from whole blood. Becausethese two populations are so clearly separated, this meth-od—as demonstrated in the post-sort analysis of bloodfrom the normal volunteer—is especially useful in sampleswith a normal or low eosinophil count, as these cells mightotherwise be difficult to distinguish from the relativelyabundant neutrophil cluster. In addition, we have shownthat the numbers of eosinophils included for analysis usingour method were comparable to the numbers usinganother conventional approach, demonstrating that ourmethod, although easier, yields equal precision. Further-more, application of this gate to a plot of FL1 versus FL2shows only the population of interest, ensuring thatstatistical analysis of percent expression of surface mol-ecules and mean fluorescence intensities may be readdirectly from the CELLQuesty software without furthermathematical manipulation to normalize numbers. Addi-

tional confirmation of the purity of the populations ob-tained by sorting on a plot of SSC versus CD16 wasobtained by post-sort analyses using the CellQuesty soft-ware and by microscopic visualization.

In summary, the advantages of this method are: (1) itssimplicity—one plot and a single gate each provide areliable and reproducible way of defining the eosinophiland neutrophil populations, which may be particularlyuseful for investigators with limited experience in flowcytometry and in studies with large numbers of samples tobe analyzed; (2) equal applicability in samples with widelyranging absolute eosinophil counts; and (3) virtually 100%purity and maximal yield of eosinophil or neutrophilpopulations sorted by this method, thus obviating theneed for other cell purification techniques and attendantconcerns about unwanted activation of cells during purifi-cation procedures.

We therefore recommend the use of side scatter versusCD16 fluorescence as a simple and efficient method ofdefining the eosinophil population in samples of wholeblood for both clinical and research applications.

ACKNOWLEDGMENTS

The authors thank David Stephany, Kevin Holmes, andCalvin Eigsti from the Flow Cytometry Unit at the NationalInstitutes of Health for assistance with the sorting experi-ments, and Dr. T. Fleisher from the Department of ClinicalPathology at the Clinical Center, National Institutes ofHealth, for critical evaluation of the manuscript.

LITERATURE CITED

1. Bochner B, McKelvey AA, Schleimer RP, Hildreth JEK, MacGlashanDW: Flow cytometric methods for the analysis of human basophilsurface antigens and viability. J Immunol Methods 125:265–271, 1989.

2. Boros P, Gardos E, Bekesi GJ, Unkeless JC: Change in expression ofFcgIII (CD16) on neutrophils from human immunodeficiency virus-infected individuals. Clin Immunol Immunopathol 54:281–289, 1990.

3. Carulli G, Gianfaldoni ML, Azzara A, Papineschi F, Vanacore R,Minnucci S, Testi R, Ambrogi F: FcRIII (CD16) expression on neutro-phils from chronic myeloid leukemia. A flow cytometric study. LeukRes 16:1203–1209, 1992.

4. Efthimiadis A, Hargreave FE, Dolovich J: Use of selective binding offluorescein isothiocyanate to detect eosinophils by flow cytometry.Cytometry 26:75–76, 1996.

5. Gane P, Pecquet C, Lambin P, Abuat N, Leynadier F, Rouger P: Flowcytometric evaluation of human basophils. Cytometry 14:344–348,1993.

6. Hartnell A, Kay AB, Wardlaw AJ: IFN-g induces expression of FcgRIII(CD16) on human eosinophils. J Immunol 148:1471–1478, 1992.

7. Mawhorter SD, Stephany DA, Ottesen EA, Nutman TB: Identification ofsurface molecules associated with physiologic activation of eosino-phils. J Immunol 156:4851–4858, 1996.

8. Selvaraj P, Rosse WF, Silber R, Springer T: The major Fc receptor inblood has a phosphatidylinositol anchor and is deficient in paroxysmalnocturnal hemoglobinuria. Nature 333:565–567, 1988.

9. Thurau AM, Schulz U, Wolf V, Krug N, Schauer: Identification ofeosinophils by flow cytometry. Cytometry 23:150–158, 1996.

316 GOPINATH AND NUTMAN