The use of flow cytometry to assess granulocyte, phagocytosis, and oxidative burst was first described in 1994 as a two-color experiment involving isolated polymorphonuclear (PMN) cells and labeled bacteria . Since this initial description, the use of flow cytometry to assess granulocyte function has become routine [2-10]. While flow cytometry techniques are commonly used in a variety of disease models, the information that can be obtained is limited by current flow cytometer technology. One means by which to expand the information that can be obtained is to combine image-based analysis with flow cytometry. To our knowledge, only a single method has been published demonstrating the use of image-based cytometry to assess granulocyte function . Even this single study was limited by the use of a first generation imaging cytometry (Amnis IS100) that lacked many of the advances of current generation instruments. In addition to method limitations associated with the type of analytical cytometer, assay considerations designated by the investigator may also effect what can be reported following a granulocyte measurement.
For the measurement of granulocyte function, the important assay design considerations are the type of target bacteria used, the length of the assay incubation, and the method used to identify granulocytes. The type of bacteria (gram-positive, gram-negative, etc.) is known to influence the degree of changes in granulocyte phagocytosis. Both Escherichia coli (E. coli, gram-negative) and Staphylococcus aureus (S. aureus, gram-positive) have countermeasures to limit phagocytosis; however, since these are different mechanisms, the observed granulocyte phagocytosis response is also likely to differ. S. aureus also produces catalase and superoxide dismutase that blunt granulocyte oxidative burst response, meaning that when using S. aureus as opposed to E. coli, the oxidative burst response is likely to differ. The majority of studies use only a single assay incubation time, rather than a series of incubation durations. Studies that have used a series of assay incubations are able to better describe the time course of observed changes in granulocyte function [11, 12]. A final limitation of existing methods is the reliance on forward scatter (FSC) and side scatter (SSC) to identify granulocytes [1-6, 8, 10-12], rather than using definitive cell-surface markers like CD66b and/or CD45. The one drawback associated with the use of immunophenotypic markers is that their expression may be altered during granulocyte activation; however, definitive markers allow more accurate ID of cell types, ensuring that the granulocyte population is free of monocytes or other phagocytic leukocytes.
The purpose of this technical note was to describe a novel, image-based flow cytometry approach that divides activated granulocytes into three subsets based on using phagocytosis and oxidative burst. Within this description we compared the effect of different types of bacteria targets (E. coli and S. aureus), assay incubation times (5–140 min), and granulocyte gating strategies (scatter vs. CD markers). One significant advantage of image-based flow cytometers is that they may allow investigators to rapidly develop and deploy complex functional assays, such as the ones described in this report. With the recent advent of lower cost, bench-top image-based flow cytometers, the findings presented in the current technical note are of even greater consequence.
All sample collection procedures described in this report were conducted in accordance with the declaration of Helsinki and approved by the UNT institutional review board (IRB). Subjects gave written consent for collection of their blood and were screened to ensure that they were apparently healthy, of normal body weight, and disease free. After an overnight fast (>8 h), subjects reported to the laboratory and sat quietly in a chair for 30 min prior to blood collection. Venous blood was collected from a peripheral arm vein into an evacuated, heparinized tube (BD-Vacutainer; Franklin Lakes, NJ). Blood samples were held at room temperature and processed within 1 h of collection. All experiments described in this report were replicated months to provide evidence that results were consistent and repeatable. The inter-assay CV of the replications was within the measurement daily measurement variability of the FlowSight (<5%).
Prior to testing, after consulting the available literature, reagent specific titrations were completed to determine optimal working concentrations. A stock of reagents was prepared and kept frozen at −20°C until needed for the assay. E. coli (P35361; Life Technologies, Grand Island, NY) and S. aureus (A10010; Life Technologies) bioparticles labeled with pHrodo Red were diluted to a stock concentration of 1 mg/ml in sterile PBS. Bioparticles were titered to a final concentration of 0.15 mg/ml in the assay. Dihydroethidium (DHE; D7008; Sigma-Aldrich; St. Louis, MO) and N-ethylmaleimide (4259; Sigma-Aldrich) were used at final concentrations of 10 µg/ml and 15 mM, respectively. DHE is converted to fluorescent ethidium bromide in the presence of oxygen-free radicals (i.e., the oxidative burst) and N-ethylmaleimide was used to prevent additional phagocytosis following the specified assay incubation. It was confirmed that the DHE titer was optimal by using imaging to provide validation that the nuclear DNA was not stained. CD66b-APC (clone#G10F5; eBioscience; San Diego, CA; DF = 1:100) and CD45-APCeFluor780 (clone#2D1; eBioscience; DF = 1:100) were used to identify granulocytes (CD66b+/CD45+) and exclude mononuclear cells (CD66b−/CD45+). 7AAD (EMD Millipore; Billerica, MA; DF = 1:20) was used to stain nucleated cells after the cells were fixed. 7AAD was included to ensure that the granulocytes were not contaminated with granulocyte microparticles (GMP, CD66b+/7AAD−). In preliminary analysis, we found that our granulocyte populations were consistently >99% positive for 7AAD and thus not contaminated with GMPs (data not shown). Information concerning the dye specific laser excitation wavelength and FlowSight emission channel can be found in Table 1.
Table 1. Dye combinations used for determination of activated granulocytes
It is important to note that while the emission wavelengths of Channels 11 and 12 overlap with Channels 5 and 6, because of the unique design of the FlowSight, there is minimal spectral overlap. The spectral overlap that does exist can be corrected during compensation.
Aliquots of sodium heparin-treated whole blood (100 µl) were added to individual 1.2 ml tubes. Blood samples were tested with two types of bioparticles (S. aureus and E. coli; 20 µl per bioparticle) 10 assay incubation lengths [0 (negative control), 5, 10, 20, 40, 60, 80, 100, 120, and 140 min]. DHE (40 µl) was also added to each tube to stain for the oxidative burst. Assay incubations used a 37°C water bath by starting with the 140 min incubation and then adding additional tubes at the specified times. All tubes were kept on ice in the dark until placement in the 37°C water bath. After the 37°C incubation period, all assay tubes were placed on ice and kept in the dark for all subsequent processing steps. N-ethylmaleimide (15 µl) was added immediately after the incubation to block additional phagocytosis and contain phagocytized bioparticles within the cell. After a 30 min incubation, cells were washed with PBS (750 µl; Sigma-Aldrich) and stained with antibodies (10 µl each diluted antibody) against CD66b-APC and CD45-APCeFluor780 for 60 min (eBioscience). White blood cells (WBCs) were fixed with a commercially available fixative/red blood cell (RBC) lysis solution (750 µl; 1-Step Fix/Lyse Solution; eBioscience; Catalog # 00-5333-57). Fix/Lyse solution was purchased as a 10× concentrate and diluted to 1× in sterile water prior to use in assay. For lysis, 1× Fix/Lyse was used at a 1:8.5 ratio of blood to Fix/Lyse. After a final wash, supernant was decanted and the cell fractions were resuspended in PBS (50 µl) and 7AAD (10 µl) was added to each tube to stain the nucleus of each WBC. Samples were covered with foil and stored in the refrigerator prior to acquisition for a maximum of 24 h. A minimum of 3,000 granulocyte (CD66b+) events were acquired on an ASSIST calibrated Millipore-Amnis FlowSight (EMD Millipore-Amnis, Seattle, WA) equipped with blue–green (488 nm, 60 mW), red (642 nm, 100 mW) and side scatter (SSC; 785 nm, 12 mW) solid-state lasers. The FlowSight instrument used in this method was equipped with a quantitative imaging upgrade that included a 488 nm laser power doubler and increased image resolution (40× vs. standard 20×). Raw image files (.rif) were acquired and adjusted for spectral overlap postacquisition using Amnis IDEAS software (Millipore-Amnis; v.5.0.983.0).
Analysis of Data
Along with the above sample treatment, prior to the assay, single color control tubes were collected with no bright field illumination and no SSC laser. These were used to create a compensation matrix in IDEAS software. The matrix was applied to .rif files to generate a compensated image file (.cif) for each sample. Data was subsequently analyzed using two different approaches:
Traditional gating technique: Identification of the percentage of granulocytes positive for phagocytosis and/or oxidative burst.
Image-based gating technique: Image-based gating to separate activated granulocytes into three subsets based on phagocytosis and oxidative burst.
Traditional Gating Technique
Figure 1 demonstrates an approach for phagocytosis gating consistent with previously published methods [3, 8, 9]. Please note that the background plots presented in Figure 1 should be considered representative of the background response across all assay incubation times. Single cell events were identified using a dot plot of bright field (BF) aspect ratio versus BF area (Fig. 1A). Aspect ratio is a feature value calculated by dividing the height of the cell by the width. On the FlowSight single cell events tend to have an aspect ratio between 0.7 and 1.0. A second plot was generated to separate granulocytes (CD66b+/CD45+) from mononuclear cells (CD66b−/CD45+; Fig. 1B). On the basis of the granulocytes, additional plots were created to identify the percentage of granulocytes that were positive for a given bioparticle relative to the negative control (Fig. 1C). A second set of plots from the granulocyte gate was generated to identify percentage of granulocytes positive for the oxidative burst relative to the negative control (Fig. 1D). A third set of granulocyte plots was used to identify the percentage of granulocytes positive for bioparticles and oxidative burst relative to the negative control (Fig. 1E). Representative images for the various populations of granulocytes are also presented in Figures 1F–1I. Visual inspection of individual, activated granulocytes images revealed what appeared to be different degrees of activation. These subsets were not obvious when observing dot plots, which appeared to present a homogenous population. It was this observation that prompted our lab to segregate activated granulocytes into different subsets based on their relative activation.
Image-Based Gating Technique
Figure 2 demonstrates the image-based gating approach that we used to divide activated granulocytes into three subsets. Similar BF aspect ratio versus BF area (Fig. 1A) and CD66b versus CD45 (Fig. 1B) plots were generated to identify single cell events and granulocytes, respectively. With single granulocyte events defined, plots of E. coli (Fig. 2A), S. aureus (Fig. 2F), versus CD66b were generated identify granulocytes that were positive for bioparticle uptake. We then examined the images of individual activated granulocytes to set gates on dot plots that allowed separation of activated granulocytes. Inactivated granulocytes were placed into a fourth region and confirmed via the negative control (Figs. 2B, 2G; MFI: 81.9–2064.8). It is worth noting that the negative control did not correctly identify all negative cells. Using only dot plots, some of the granulocytes in the “activated” area were in fact negative. Thus, care should be taken when exclusively using negative controls to separate active and inactive granulocytes.
Activated region one (Figs. 2C, 2H; MFI: 2064.9–4024.0) contained granulocytes with very low function (low bioparticle phagocytosis, no oxidative burst, and no colocalization of phagocytosis and oxidative burst). Activated region two (Figs. 2D, 2I; MFI: 4024.0–19750.0) contained granulocytes with greater function (medium bioparticle phagocytosis, low oxidative burst, and no colocalization of phagocytosis and oxidative burst). Activated region three (Fig. 2E, 2J; MFI: >19750.0) contained granulocytes with high function (high bioparticle phagocytosis, medium oxidative burst, and colocalization of phagocytosis and oxidative burst). Activated region three granulocytes demonstrated the highest function, as evidenced by colocalization of both phagocytized bioparticles and oxidative burst. Longer incubations (100–140 min) resulted in a shift of activated granulocytes from region one to regions two and three. To our knowledge, our lab is the first to propose the use of image-based gating to classify active granulocytes based on phagocytosis, oxidative burst, and colocalization of phagocytosis and oxidative burst.
Effect of Bioparticle Type and Assay Incubation Time
Change in granulocyte phagocytosis response as a function of bioparticle type and assay incubation time is presented in Figure 3. Figure 3 also demonstrates the changes in the percentage of granulocytes falling within each defined region in Figures 3C and 3F. The bioparticles we selected for this experiment represent the ones most commonly used in the literature and are commercially available providing access to others who may be interested in using our technique [1, 6, 8, 11, 12]. As anticipated, the observed time course of the phagocytosis response was slightly different for each bioparticle type. The fastest phagocytosis response was observed at 10-min with S. aureus; however, the maximal oxidative burst response for S. aureus did not occur until 120 min (Fig. 3C). The slowest maximal phagocytosis response was observed for E. coli at 20 min, and maximal oxidative burst was not observed even at the 140 min incubation with E. coli (Fig. 3F). Beyond maximum responses, the general time course of change in granulocyte function appears to change at a different rate as a function of bioparticle type. This later finding underscores the importance of carefully selected and testing a given bioparticle in a targeted experiment.
On the basis of the literature, the most common method when using a flow cytometry based phagocytosis assay is to use a negative control (no incubation) and a single time point incubation between 20 and 40 min [3-6, 9]. A handful of studies have used more than a time course incubation lasting up to 40 min to reveal how different diseases or treatments influenced granulocyte function [11, 12]. On the basis of the method presented in this technical note, it appears that assay incubations of up to 40 min may be sufficient for detecting changes in phagocytosis when using S. aureus, but not E. coli. Also, we observed that colocalization of bioparticles and oxidative burst did not occur until 120 min of incubation, meaning that incubations shorter than 120 min may not be sufficient to fully detect changes in granulocyte oxidative burst. Care should be taken when selecting bioparticle type and assay incubation time to ensure the desired granulocyte functional measure can be appropriately measured. Failure to safe guard the method design may result in false or misleading findings with regard to granulocyte function.
SUMMARY OF TECHNICAL NOTE
In this technical note, we described the advantages and additional information that can be gained when utilizing benchtop, image-based flow cytometry to simultaneously measure granulocyte phagocytosis and oxidative burst. Specifically, we found that image-based analysis allowed us to divide activated granulocytes into three subsets. We also found that when comparing different types of bacteria, S. aureus yielded the best speed of phagocytosis response, while E. coli yielded the best overall oxidative burst response. With respect to assay incubation time, we learned that in general incubations of as little as 20 min are sufficient to detect phagocytosis, but it takes a minimum of 120 min to elicit a maximal oxidative burst response.
Our goal of preparing this technical note was to provide a detailed method for how to utilize image-based flow cytometry to assess granulocyte function via changes in phagocytosis and oxidative burst. It is an objective of our lab to advance the field of flow cytometry methodology and it is our hope that the information presented in this technical report will be useful to others interested in using a similar approach. Future studies should seek to apply our image-based granulocyte function assay in different disease states and/or experimental treatments.