Confounding effects of platelets on flow cytometric analysis and cell-sorting experiments using blood-derived cells




Flow cytometric analysis and cell-sorting of peripheral blood leukocytes is commonplace; however, platelet contamination is typically ignored during immunophenotypic analysis and sorting of blood-derived cells.


Red blood cells, platelets, T & B lymphocytes, monocytes, and granulocytes were sorted from rat blood preparations. Presort enrichment was performed by differential centrifugation for all cell types. Additionally, leukocyte samples were prepared by ammonium chloride lysis of red blood cells.


Unless proper precautions were taken, significant numbers of platelets were sorted along with (nonplatelet) cells of interest. The amount of platelet contamination varied greatly from experiment to experiment with the highest level of leukocyte-platelet association observed in the neutrophil/granulocyte population in samples prepared using ammonium chloride-based red blood cell-lysing solution.


Addition of an immunophenotypic marker for platelet identification is a simple, yet prudent, measure to help evaluate the impact of platelets on immunophenotypic staining when performing flow cytometric analysis or sorting of blood-derived cells and should become a routine practice. Platelet presence in postsort fractions can be due to free platelets as well as target cell-associated platelets and both sources of contamination must be addressed. © 2005 Wiley-Liss, Inc.

Flow cytometry-based immunophenotypic analysis of blood typically implies only pan-leukocyte or peripheral blood mononuclear cell (PBMC) analysis. Pan-leukocyte enrichment can be accomplished simply by lysis of red blood cells (RBCs). PBMC analysis requires differential centrifugation to enrich mononuclear leukocytes (lymphocytes, monocytes, and natural killer cells) and deplete granulocytes, mainly neutrophils. In general, however, platelets are not considered while staining or performing the actual phenotypic analysis of enriched leukocyte preparations. Consequently, most investigators are likely unaware of the presence of platelets in blood preparations and therefore do not consider effects that platelets might have on results and data interpretation.

Results presented herein suggest that platelet contamination of leukocyte preparations is commonplace. In this context, platelet contamination is defined as the presence of platelets in enriched preparations of (nonplatelet) blood-derived cells. Proper instrument setup and sample preparation can help to identify and minimize platelet contamination during flow cytometric analysis and cell-sorting of blood-derived cells. Platelet contamination may be in the form of free platelets/microparticles, platelet aggregates, or platelet–leukocyte aggregates. Strategies to recognize and minimize platelet contamination in blood samples prepared for flow cytometry and cell-sorting applications will be discussed and justified.


Blood Collection and Hematology

Eight- to 12-week-old male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were killed by CO2 asphyxiation and exsanguination. Blood (6–10 mL per rat) was collected from the caudal vena cava into a 10 mL syringe with an 18-gauge needle containing 200 μL of 10% w/v disodium EDTA (VWR Scientific Products, West Chester, PA). Syringes were rocked at room temperature (RT) until processed. Blood was then either processed to enrich the population of interest or red cell lysis was performed (see later). Differential cell counts were performed on EDTA-blood and enriched fractions with an Advia 120 (Bayer, Tarrytown, NY) hematology instrument.

RBC Lysis for Pan-Leukoctye Enrichment

EDTA-blood was added to 15-mL conical tubes in 500 μL aliquots. Ice-cold ammonium chloride (10 mL) lysis buffer (150 mM NH4Cl, 10 mM NaHCO3, and 1 mM disodium EDTA) was added and the tubes were capped and immediately vortexed for 10 s. Tubes remained at RT for a maximum of 20 min. Samples were centrifuged at 120g for 5 min at RT and washed twice with 10 mL PBS by centrifuging in the same manner. Cell pellets were resuspended in phosphate-buffered saline (PBS) and combined in a single tube for counting, and then aliquotted into 12 × 75 mm polystyrene tubes. Samples were centrifuged as before and then cell pellets were stained, according to the leukocyte procedure described later.

Mononuclear Cell Enrichment

EDTA-blood (6–10 mL) was added to a 50-mL conical tube. Cell culture medium (15 mL; RPMI) was added to the blood and mixed. A pipette was used to underlay 15 mL of Ficoll-Paque™ PLUS (Amersham Pharmacia Biotech, Uppsal, Sweden) with minimal mixing, followed by centrifugation at 400g (no brake) for 30 min at RT. Mononuclear cell layer was removed and transferred to 12 × 75 mm polystyrene tube(s), was washed twice with 2 mL PBS by centrifugation at 200g for 5 min at RT, and then cell pellets were stained, as described later.

Granulocyte Enrichment

EDTA-blood (6–10 mL) was added to a 50-mL conical tube; 15 mL of cell culture medium (RPMI) was added to the blood and mixed. A pipette was used to underlay 15 mL of Ficoll-Paque™ PLUS (Amersham Pharmacia Biotech, Uppsal, Sweden) with minimal mixing, followed by centrifugation at 400g (no brake) for 30 min at RT. Contents above the red pellet were removed and discarded; 20 mL of PBS and 6 mL of 3% Dextran T-500 (Amersham Pharmacia Biotech) were added and contents mixed well. Preparation was incubated at 37°C for 90 min. Supernatant was transferred to a 50-mL conical tube and centrifuged at 200g for 10 min at RT. Supernatant was discarded and pellet was washed twice with PBS by centrifuging at 200g for 10 min at RT. Residual RBCs were lysed with ice-cold sterile water. Pellet was resuspended in minimal volume of PBS and transferred to a 12 × 75 mm polystyrene tube then stained, as described later.

Platelet Enrichment

EDTA-blood (2 mL) was added to a 12 × 75 mm polystyrene tube and centrifuged (no brake) at 150g for 20 min at RT. Platelet-rich plasma (100–500 μL) was removed and added to a clean 12 × 75 mm polystyrene tube. PBS (2 mL) containing 1% EDTA (GIBCO BRL, Grand Island, NY) was added, tubes were centrifuged at 1,200g for 5 min at RT, supernatants were removed, and then cell pellets were stained as described later.

RBC Enrichment

EDTA-blood (100 μL) was added to a 12 × 75 mm polystyrene tube and diluted with of 4 mL PBS. After mixing, 100 μL was aliquoted to 12 × 75 mm polystyrene tubes and diluted further with 2 mL PBS, then centrifuged at 1,200g for 1 min in a benchtop centrifuge (serofuge) at RT. Supernatant was removed, 2 mL PBS added, and tubes were centrifuged as before. Supernatant was removed, pellet was resuspended in 2 mL PBS, and cells were counted with an Advia 120. Cells (1 × 108) were aliquoted to a 12 × 75 mm polystyrene tube, centrifuged at 400g for 5 min, supernatants removed, and then cell pellets were stained, as described later.

Antibodies and Fluorescent Reporters

CD45R-FITC (clone HIS24, IgG2b), CD11b/c-PE and -APC (clone OX-42, IgG2a), CD45-CyChrome™ (clone OX-1, IgG1), CD42d-PE-Cy7 (clone RPM.4, IgG2a), CD3-APC (clone 1F4, IgM) and erythroid-biotin (clone HIS49, IgM) mouse anti-rat antibodies, Streptavidin(SA)-APC-Cy7, and appropriate isotype-matched controls were purchased from BD Biosciences, San Diego, CA. Aminostilbamidine, methanesulfonate (ASBMS) was purchased from Molecular Probes, Eugene, OR (Table 1).

Table 1. Flow Cytometry Staining Reagents
ReagentIsotypeCloneManufacturerPopulation identified
  • a

    Aminostilbamidine, methanesulfonate.

  • b

    Not applicable. ASBMS is not an antibody.

  • c

    Cells with compromised plasma membranes.

Anti-CD45RMouse IgG2bHIS24BD BiosciencesB cells
Anti-CD11b/cMouse IgG2aOX-42BD BiosciencesMonocytes, granulocytes/neutrophils
Anti-CD45Mouse IgG1OX-1BD BiosciencesLeukocytes
Anti-CD42dMouse IgG2aRPM.4BD BiosciencesPlatelets
Anti-CD3Mouse IgMIF4BD BiosciencesT cells
Anti-erythroidMouse IgMHIS49BD BiosciencesErythroid (RBCs)
ASBMSaN/AbN/AbMolecular ProbesDead cellsc

Staining Procedures

Leukocyte preparations (1 × 108 cells) were simultaneously incubated with 4 μg each CD45R, CD11b/c, CD45, CD42d, CD3, and erythroid-biotin antibodies in a total volume of 100 μL PBS on ice for 30 min. Cells were washed once with PBS and then resuspended in PBS. SA-APC-Cy7 (4 μg) was added in a volume of 100 μL PBS, cells were incubated on ice for 30 min, washed with PBS, and then resuspended in PBS containing 10 μM ASBMS at a concentration of 5 × 107 cells/mL. Cells were incubated at RT for 10 min then placed on ice until analysis/sorting. Platelet preparations (1 × 108 cells) were stained in a similar manner with the following exceptions. Only CD42d, erythroid-biotin, SA-APC-Cy7, and ASBMS reagents were employed. Cells were maintained at RT at all times and centrifuged at 1,200g. RBC preparations (1 × 108 cells) were stained only with CD42d and ASBMS and RBCs identified as CD42d-/low forward scatter cells. Cells were maintained at RT at all times and centrifuged at 400g.

Sorter Setup and Optimization

Analysis and cell-sorting were performed on a FACSVantage (BD Biosciences, San Jose, CA) equipped with digital electronics and using BD FACSDiva software for data acquisition (BD Biosciences). A sheath pressure of 44 psi and a 70 μm nozzle tip were used for all sorts. Three-beam alignment and optimization of all fluorescence detectors was done prior to each experiment, using standardized particles (URFP, Spherotech, Libertyville, IL). Drop drive frequency and amplitude were optimized to create smooth droplet breakoff and stable side streams. A dual threshold of forward and side laser scatter was necessary to resolve free platelets. With the drop drive engaged, threshold values were lowered for both scatter parameters to locate noise. A combination of adjusting threshold values and laser scatter detectors' sensitivities was used to resolve platelets from noise. This process was repeated with sort logic engaged.

Flow Cytometry/Cell-Sorting

Cell samples (5–10 × 107 cells/mL) were filtered through 50 μm nylon mesh then analyzed/sorted with a FACSVantage cell-sorter equipped with 350, 488, and 633 nm laser excitation wavelengths. FITC, PE, PECy5, PECy7, APC, APCCy7, and ASBMS fluorescences were collected through 530/30, 575/26, 710/20, 780/60, 660/20, 780/60, and 530/30 bandpass filters, respectively. All sorted fractions were rerun to assess purity. To better visualize free platelets and RBCs, forward laser scatter was displayed on a logarithmic scale in all sorts; forward and side laser scatter were both displayed logarithmically during platelet and RBC sorts. Gating strategies for sorting are described in Figures 3–5.

Figure 3.

Gating strategy for cell-sorting of RBCs from rat blood preparations. Use of light scatter and anti-CD42d to identify platelets negates need for RBC positive identification. (a) All events displayed. “intact cells” gate defines main (RBC) population. Leukocytes have much greater laser light scatter and are excluded on this basis. (b) “Intact cells” displayed. “Singlets” gate defines events with low FSC-W, presumably single-cell events. (c) “Singlets” displayed. “RBCs” gate defines events that are negative for CD42d. (d) Heirarchical gating tree and population statistics for representative rat blood red blood cell preparation. (e) All events displayed. “CD42d+” gate defines CD42d+ events (platelets). Most platelets (blue events) are excluded by “intact cells” gate in (a) based simply on laser light scatter. The remainder is excluded by “RBCs” gate in (c).

Figure 4.

Gating strategies for cell-sorting of monocytes, granulocytes, T lymphocytes, and B lymphocytes from rat blood preparations. Data shown describes mononuclear fraction prepared by differential centrifugation. Leukocyte identification strategy was identical for all leukocyte preparations. Titles on a–h indicate population/gate displayed therein. (a) All events displayed. “Intact cells” gate defines main cell population. (b) “Intact cells” displayed. “Live cells” gate defines cells that exclude the dead cell discriminator ASBMS. (c) “Live cells” displayed. “Singlets” gate defines events with low FSC-W, presumably single-cell events. (d) “Singlets” displayed. “Platelet-free” gate defines events that are negative for CD42d. (e) “Platelet-free” displayed. CD4S-PE-Cy5-A is plotted with APC-Cy7-A for ease of gating only; cells were not stained with APC-Cy7. “Leukocytes” gate defines CD45+ cells. (f) “Leukocytes” displayed. “Myeloid” gate defines CD45+/CDI lb+ cells. “Lymphoid” gate defines CD45+/CD11b-cells. (g) “Myeloid” displayed. “Monocytes” gate defines low side scatter (SSC-A) myeloid cells. “Granulocytes” gate defines high SSC-A cells. (h) “Lyniphoid” displayed. “B cells” gate defines CD3-/CD45R+ cells. “T cells” gate defines CD3+ICD45R-cells.

Figure 5.

Gating strategy for cell-sorting of platelets from rat blood preparations. (a) All events displayed. “Intact cells” gate defines main cell population. (b) “Intact cells” displayed. “Singlets” gate defines events with low FSC-W, presumably single-cell events. (c) “Singlets” displayed. “Platelet” gate defines events that are erythroid-ICD42d+. (d) Heirarchical gating tree and population statistics for representative rat blood platelet preparation.


Flow Cytometry and Cell-Sorting

Using a monoclonal antibody against platelet glycoprotein V (CD42d), a large platelet component was frequently observed in lysed-whole blood preparations with a cluster of free platelets (i.e. not leukocyte associated) visualized near the forward/side laser scatter threshold cutoff (P1, Fig. 1a). A percentage of lymphocyte, monocyte, and granulocyte populations, defined by immunophenotype, as well as characteristic laser scatter, also stained positive for CD42d, indicating platelet–leukocyte association (Table 2). Although very good results (i.e. low platelet–leukocyte adhesion) were obtained on occasion with lysed whole blood preparations (Table 2, Experiment 2), differential centrifugation more consistently gave lower incidence of platelet–leukocyte association.

Figure 1.

Low laser light scatter events are platelets. When displaying flow cytometric data obtained from blood leukocytes on a forward versus side laser light scatter dot plot, events in the lower left corner are typically inferred to be “red cells, dead cells, and debris.” However, immunophenotypic analysis of lysed whole rat blood with an anti-CD42d antibody reveals that the vast majority of these events are platelets. (a) P1 defines a cluster of events below the lymphocyte cluster that represent 41.6% of total events. (b) P1 (92.3%) events are platelets.

Table 2. Platelet-Leukocyte Association
Leukocyte population% CD42d positive events
Lysed whole bloodDifferential centrifugation
Experiment 1Experiment 2Experiment 3Experiment 4Experiment 5Experiment 6
  1. Assessment of platelet-leukocyte association following blood preparation by red blood cell lysis or differential centrifugation. Positive staining for anti-CD42d-PECy7 on leukocytes indicates platelet association. No attempt was made to determine the approximate number of platelets associated per leukocyte. The percent of platelet-associated leukocytes (i.e. platelet–leukocyte aggregates) is reported for each leukocyte population assayed. In experiments 1–3, all four leukocyte populations were analyzed in a single lysed whole blood preparation. In experiments 4–6 a granulocyte fraction and a mononuclear fraction was obtained by differential centrifugation. Granulocyte data is from granulocyte fractions and monocyte, B cell and T cell data is from mononuclear fractions.

B cells1.81.433.
T cells0.60.327.

To be certain that free platelets were properly resolved in leukocyte and red cell preparations and not sorted along with target cells, platelets were isolated as described earlier and sorted. A dual forward/side scatter threshold (500/1,000) was necessary to reduce noise, as platelet scatter signals were only slightly above noise, especially when the drop drive was engaged. Even so, electronic noise that overlapped the platelet population was apparent in the scatter channels if laser scatter threshold values were too low. At the instrument settings described, with drop drive engaged but sort logic off, there was a small percentage of events that impinged on the platelet population (red events, Fig. 2a). These events did not express CD42d (P1, Fig. 2c) and had a high forward scatter width (FSC-W) signal (red events, Fig. 2e). These few events (<5%) were likely due to electronic noise, but were of no concern due to low incidence and ease of removal by either the “Singlets” gate based on FSC-W (Fig. 2b) or the “Platelets” gate based on CD42d expression (Fig. 2c). However, when the sort logic was engaged (i.e. deflection pulses generated), a large (≥40%) population of events impinged on the platelet population (red events, Fig. 2f). These events were easily distinguishable from platelets based on CD42d expression (P1, Fig. 2b), but had very similar FSC-W (red events, Figs. 2g and 2j). Sorting at these settings was untenable and required an increase of both forward scatter and side scatter voltage and threshold settings. Once these adjustments were made, sorting was efficient and postsort purity was high (≥99%).

Figure 2.

Electronic noise impinges on platelet population if low laser light scatter signal amplification and threshold levels are used. Side and forward scatter detector voltages were 500 and 390, respectively. Dual threshold of side (1000) and forward (500) scatter was used. Sort logic was off in a–e and on in f–j. Platelets (CD42d+) are blue and noise is red in all the figures. (a) and (f) Forward versus side scatter dot plot with all events displayed. (b) and (g) Histogram overlays of platelets and noise. (c) and (h) Gate “P1” defines nonplatelet (noise) events. (d) and (i) Heirarchical gating tree and population statistics. (e) and (j) Plot of FSC-A versus width (FSCW) parameters. A certain amount of electronic noise (red) is seen overlapping the platelet population (blue) whether the sort logic is off (a) or on (f). However, the amount of noise “events” is small (3.8%) with the sort logic off (P1, d) but high (40.4%) with the sort logic on (P1, i). This high amount of noise greatly reduces sorting efficiency and remains upon reanalysis. This problem can be alleviated by increasing forward and side scatter detector voltages and threshold values. Effectively, platelets are moved higher on scale but noise remains in place and is then removed by higher threshold values.

Red cells isolated as described were sorted using the optimized settings determined by platelet sorting. Two main populations were resolved based on FSC-A and SSC-A (Fig. 3a), with the majority of the platelets (blue) falling below the RBCs and excluded by the “intact cells” gate. However, RBCs were better represented by gating first on laser scatter and then excluding CD42d positive events (Fig. 3c). Note that RBCs were sorted only by negative selection. The anti-rat erythroid antibody most readily available is of an IgM isotype and leads to rapid agglutination of red cells when stained at the concentrations (5–10 × 107) desirable for high-speed sorting. Leukocytes were excluded on the basis of laser scatter. Very high postsort purity (≥99.8%) was obtained using this strategy. Additionally, thiazole orange could be included to sort reticulocytes exclusively (data not shown).

The major blood leukocyte populations were sorted using the gating strategy described in Figure 4. The same strategy was applied regardless of leukocyte enrichment procedure, although not all cell populations were sorted from all preparations. That is, mononuclear cells were not sorted from granulocyte preparations and granulocytes were not sorted from mononuclear preparations. In most cases, multiple populations were sorted simultaneously by sorting in two directions. Postsort purity was reasonably high (≥95%) for all populations sorted using this strategy. A comparison of the amount of leukocyte–platelet association seen in the different leukocyte enrichment procedures used is summarized in Table 2.


It is clear that some finite amount of platelets will be analyzed/sorted along with leukocytes isolated from peripheral blood unless proper precautions are taken to exclude them. This is not a trivial matter because platelets have been shown to be transcriptionally active (1) and may express genes (mRNA) and proteins normally associated with peripheral blood leukocytes. In fact, transcript profiling of platelets by microarray has demonstrated the presence of numerous “leukocyte gene” transcripts (2). Reports also describe platelet expression of leukocytic immunomodulatory proteins such as CD154 (3), chemokine receptors (4), and inflammatory cytokines (5). The first problem this presents is that platelets directly associated with cells of interest can confound immunophenotyping results, as platelets can be directly associated with other blood-derived cells by coincidence, specific- or nonspecific-binding during flow cytometric analysis. This could lead to an instance where a cell population is purported to express a protein that it does not, but that in fact is expressed by associated platelets or platelet fragments (6). In addition, like monocytes and polymorphonuclear leukoctyes, platelets express Fc receptors capable of binding IgG (7) and therefore have potential to be nonspecifically labeled with monoclonal antibodies of this isotype. Therefore, platelet-associated cells need to be resolved and, perhaps, excluded from analysis/sorting in the same manner that other cell aggregates are excluded. The second issue is specific to sorting applications and is more insidious. Free platelets falling below instrument threshold, typically based on a laser scatter parameter, are unaccounted for, will be sorted along with cells of interest, and will go completely undetected. In addition to expressing genes and producing cytokines common to leukocytes, platelets can affect leukocyte functions, such as cytokine secretion and adhesion molecule expression, by direct cell–cell contact (8, 9). Therefore, if platelets are sorted along with cells of interest, this can potentially affect experimental results obtained from sorted cells where leukocyte function or gene expression analyses are performed.

Given that there are several explanations for platelet contamination, all must be addressed to obtain the best results. First, coincidence occurs when two, or more, cells reach the interrogation point (i.e. laser intercept) at nearly the same instant. Coincidence will be most problematic in samples where the ratio of platelets to nonplatelets is high. High cell concentrations (5–10 × 108 cells/mL) often used during high speed sorting can also increase coincidence frequency. Simple depletion of platelets or enrichment of target cells will be helpful to reduce coincidence. As described in Materials and Methods, during wash steps platelets were centrifuged at a gravitational force multiple 10 times (1,200g) that of leukocytes. In fact, platelets do not pellet well at 120g and this alone serves to deplete platelets in lysed whole blood preparations. This phenomenon can also be exploited after differential centrifugation of granulocytes, wherein platelets are retained with granulocytes in the same density gradient band. Multiple wash steps can greatly reduce the number of platelets in a sample. Second, there is always potential for nonspecific adherence of platelets to target cells. The addition of protein, such as bovine serum albumin (BSA), to the wash/stain buffer is commonly used to aid in reduction of nonspecific protein adhesion (e.g. nonspecific antibody binding, cell aggregation) and may therefore aid in reduction of nonspecific binding of platelets to other cells. However, it appears unlikely that this is the major source of platelet contamination and therefore BSA was not used in the current study.

The third cause of platelet contamination defined, and seemingly the most problematic, is specific leukocyte–platelet adhesion wherein the activation or proadhesive status of platelets and leukocytes contribute to the formation of platelet–leukocyte aggregates. These interactions are more difficult to attenuate than nonspecific aggregation and therefore require that great care is taken to minimize activation/priming of platelets and leukocytes. To this end, it is important that proper venipuncture technique is practiced to minimize shear stress and vessel trauma. Shear stress and release of tissue factor from the puncture site can lead to platelet activation and/or initiation of the clotting cascade, as can insufficient anticoagulant in the collection apparatus. The current study employed 10% w/v disodium EDTA as anticoagulant, but many other anticoagulants are available and may be superior depending on experimental design and endpoints measured. The effects of anticoagulants on platelets, leukocytes, and other relevant experimental parameters have been discussed elsewhere (10–14).

Results demonstrated that platelet–leukocyte aggregates were more prevalent in samples prepared by lysis of RBCs versus differential centrifugation (Table 2). This could be partly due to activation of platelets by red cell lysis, a phenomenon that has been reported previously (15). There is also a body of evidence that indicates that temperatures below normal human body temperature, in some reports by even a few degrees Celsius, can lead to platelet priming, activation, degranulation, and/or hypersensitivity to shear stress (16–19). Other factors that can influence platelet activation and may therefore have bearing on platelet contamination of leukocyte preparations include discarding a portion of the blood collected, time elapsed before analysis, centrifugation, fixation (or lack thereof), and storage conditions of blood before sample preparation (15, 20). Nevertheless, leukoctye–platelet interactions do occur in vivo (21, 22) and may persist ex vivo, and therefore, a small number of aggregates may be observed regardless of steps taken to minimize this.

When sorting, the three sources of platelet contamination described earlier must be addressed in addition to a fourth, free platelets. Because of their small size, platelets produce a forward laser scatter signal well below leukocytes, and likely below forward scatter threshold value, when using instrument settings typical of leukocyte analysis. Conventional wisdom dictates that events on a dot plot of forward versus side laser scatter that appear below the lymphocyte cluster are “unlysed red cells, dead cells, or cellular debris.” However, use of anti-CD42d demonstrated that the majority of these low scatter events are in fact platelets (Fig. 1). Consequently, if these events are below threshold, not only will they be sorted along with cells of interest but also the operator will be unaware of their existence. This is true even if a phenotypic marker such as anti-CD42d is used to identify platelets. The practice of raising the (forward laser scatter) threshold value to ignore events below lymphocytes may be acceptable for analysis if these events truly are of no interest. However, this is a misguided approach for dealing with unwanted cellular events in sorting applications; adjusting the instrument to ignore these events does not remove them from the sample. It is at least as important to resolve all contaminating cells as it is to resolve target cells to obtain a high purity postsort sample. If platelet contamination in various forms is not addressed, platelets will undoubtedly be present in sorted fractions, resulting in potentially misleading data. First, free platelets that are below threshold and sorted along with cells of interest may become visible upon reanalysis of the sorted sample due to formation of platelet aggregates or association with other cells upon reanalysis, for instance. This result would likely be misinterpreted as poor sorter performance. Also, as alluded to earlier, platelets present in postsort samples can directly or indirectly impact cell function and gene expression analyses. Even though the RNA content of platelets is small compared to leukocytes, the contribution from platelets can be significant if they are abundant or if using a technique that is sensitive enough to detect low copy number genes (e.g. TaqMan®).

Sorter setup and optimization are critical for resolution of free platelets from noise in both forward and side laser scatter channels. The following discussion is specific to a digitally enhanced FACSVantage (DiVa) sorter, but may also be applicable to other sorters with similar capabilities. The normal approach of using a single threshold parameter (forward laser scatter) and displaying laser scatter (forward and side) on a linear scale does not work well when trying to minimize platelet contamination from sorts. A dual threshold of forward and side laser scatter is recommended. Displaying forward scatter on a logarithmic scale is necessary to truly resolve platelets and leukocytes simultaneously and ensure that platelets are above threshold, but makes it difficult to resolve individual leukocyte subsets. Unfortunately, this is unavoidable with current instrument configuration and display capabilities. Replacing the forward scatter diode with a photomultiplier tube may improve dynamic range and will be investigated. Displaying both scatter parameters on a logarithmic scale is helpful when trying to resolve platelets from RBCs (Fig. 3) and the majority of leukocytes should be off scale at the high end. If threshold values are too low (≤500), electronic noise may overlap the platelet laser scatter signals when sorting logic is engaged as was the case with the instrument used for this study (Figs. 2f–2j). This can be resolved by increasing both the voltage to the scatter detectors and the dual-scatter threshold values.

Numerous examples of how platelets can confound data interpretation from leukocytes have been provided but analysis/sorting experiments involving other blood-derived cells can also be impacted. In fact, platelet contamination can also perturb data collected from hematopoietic stem cells, RBCs, circulating endothelial cells (CECs), and endothelial progenitor cells (EPCs) for similar reasons. For instance, (human) platelets express the hematopoietic stem cell antigen CD34 (23). Platelet-endothelial cell adhesion molecule (PECAM), also known as CD31, is expressed on platelets as well as leukocyte subsets (24), hematopoietic progenitors (25, 26), and stem cells (27). The type B scavenger receptor (CD36) is identical to human platelet GPIV (28) and is also expressed on leukocyte subsets, endothelium, and RBCs (29). Furthermore, as was the case with leukocytes, platelets can specifically adhere to endothelial cells resulting in modulation of function (30, 31) and platelet–CEC aggregates may be encountered during flow cytometric analysis. Therefore, it is deemed especially important to resolve and exclude platelets when sorting rare blood cell populations such as hematopoietic stem cells/progenitors and CECs/EPCs.

Given their small size and potentially large number in leukocyte preparations, it is difficult to completely ablate platelet contamination, but following the suggestions given herein will greatly reduce the extent to which they impact experiments with blood-derived cells. Minimally, platelet activation/aggregation/clotting should be avoided; free platelets should be resolved above threshold, a platelet immunophenotypic maker should be included, and postsort analysis should be performed. There are many antiplatelet antibodies available for humans and rodents. Note that calcium-chelating anticoagulants (e.g. EDTA, sodium citrate, etc.) may impede the binding of certain anti-bodies or dissociate surface glycoproteins, leading to poor immunostaining of platelets. Investigators should confirm that blood processing does not interfere with antibody reporters. Anti-CD42d worked well with rat blood under the conditions described in this article.

In the past, platelets were regarded merely as anucleate particles necessary for hemostasis. In the current context they are described as “contamination” and mostly a nuisance. But platelets are now regarded as complex multifunctional cells with immunomodulatory capabilities (3, 32) and a peer-reviewed scientific journal bears their name. Even though flow cytometric analysis of platelets and leukocytes are commonplace, combination analyses are not. This is true despite the fact that analyses specific for leukocyte–platelet aggregates have been described (21, 22, 33, 34) as informative of platelet activation status and data suggests that platelet-associated leukocytes may represent a functionally distinct subclass (35). Therefore, it seems that platelet–leukocyte interaction should be of interest to platelet researchers and traditional immunologists alike. Perhaps it is time for platelets to be included in routine immunophenotypic/flow cytometric analysis of blood preparations. At the very least, platelets should be resolved to ensure that they are not influencing experimental results from other blood-derived cells of interest.


The authors thank Nancy Lee for hematology support; Laura Storck and Dana Pietrzak for expert technical assistance in blood sample preparation for sorting experiments; Mark Kukuruga for instrument configuration advice; Tom Covatta for compilation and formatting of figures; and Dr. Padma Narayanan and Dr. Lester Schwartz for manuscript review.