Circulating microparticles (MPs) are small (<1 μm in diameter) membrane fragments shed from cells following a host of physiologic and pathophysiologic stimuli. MPs are normally found in small numbers (relative to other circulating cells) in the vascular space, where they participate in complex intercellular signaling pathways (1).The number of circulating MPs greatly increases in numerous disease states, including cardiovascular disorders (1, 2), infectious diseases (3), cancer (4), and hemolysis (5, 6). MPs are currently under intense study, with pathologic, prognostic, or even protective functions being reported (1–6). For example, red blood cells (RBC) may shed MPs as a protective means to dispose of irreversibly altered hemoglobin or bound immunoglobulin, preventing otherwise healthy RBCs from premature degradation (7). However, RBC MPs also house potentially harmful macromolecules, like hemoglobin that can scavenge nitric oxide (8), or procoagulant (9) or immune-modulating phospholipids (10). More functions of, and even ways to modulate these transient bioeffectors will undoubtedly be uncovered in the near future.
Flow cytometry is the preferred method for examining MPs. However, caution is required, as the small size of MPs relative to the wavelength of interrogating light results in considerable artifact (11, 12). Thus, markers are required to identify flow cytometry events as true MPs (11–13). In particular, fluorescently labeled annexin V, which binds to phosphatidylserine (PS), a phospholipid normally on the inner cellular leaflet of the plasma membrane but which is externalized on MPs, has been used for this purpose (14). As millimolar calcium ion concentrations are required for annexin V-PS binding (15), many studies of MPs have either (1) isolated MPs from biological fluids, followed by resuspension in phosphate-buffered solutions and addition of a calcium salt (16–20); or (2) added calcium salt directly to plasma of patients with impaired phosphate excretion (and elevated plasma phosphate) (21). As calcium-phosphate has limited solubility, it is possible that calcium-phosphate precipitation can occur in phosphate-containing fluids when Ca++ is added (22, 23).
The purpose of this study was to examine if the formation of calcium-phosphate microprecipitates interferes with flow cytometry determination of MPs. Calcium-phosphate microprecipitates were compared with MPs obtained from expired donor blood, a robust source of red cell-derived MPs. Our results indicate that calcium-phosphate microprecipitates and MPs overlap in light scatter (and therefore, size); and that fluorescently labeled antibodies may bind nonspecifically to microprecipitates. Thus, experiments examining MPs should be carefully designed (1) to avoid microprecipitate formation or (2) to confirm the cellular basis of the MP by a secondary method.
Materials and Methods
All buffers were sterile-filtered using 0.20 μm syringe filters (Corning, Tewksbury, MA) and either kept on ice or at room temperature for 30+ min prior to mixing and examination with flow cytometry. Dulbecco's phosphate buffered saline (containing no Ca++ but 9.5 mM phosphate) at pH 7.3–7.4 (PBS) and Hank's balanced salt solution (HBSS) at pH 7.4 (which contains 0.77 mM phosphate and 1.3 mM CaCl2, HBSS) were from Invitrogen (Life Technologies, Grand Island, NY). Normal saline (NS, 0.9% w/v) adjusted to pH 5–7.4 (NS) and Calcium chloride were from Sigma-Aldrich (St. Louis, MO) or Fischer Scientific (Pittsburgh, PA) when protocols were independently repeated by a second researcher. Tris-HCl for Tris-buffered saline (50 mM Tris with 150 mM NaCl) at pH 7.4–7.6 (TBS) was also from Fischer. CaCl2 was mixed in purified water at 100 mM final stock concentration. Expired (>42 day-old) leukocyte- and platelet-reduced donor blood units (N = 4) in adenine-citrate-dextrose and Solution 1 were obtained from the BloodCenter of Wisconsin (Milwaukee, WI) as a known source of MPs (6). Platelet-rich plasma was obtained from a healthy, consented donor in accordance with IRB-approved protocols. Platelet-rich plasma was isolated by centrifugation of citrate-anticoagulated whole blood at 200g for 10 min, followed by 1500g for 15 min at room temperature to avoid cold-induced platelet activation. Potassium EDTA (Fischer) was diluted 6% w/v in purified water, and was added to achieve 0.1% final concentration at the peak of the Ca++-PBS titration (5 mM CaCl2). This concentration of EDTA is used clinically to anticoagulate whole blood. Beads with known nominal diameters were from Sigma (0.55 μm), Solulink (San Diego, CA) (0.8 μm), Pierce (1.0 μm), New England Biolabs (Ipswich, MA) (2.0 μm), and Bang's Laboratories (Fishers, IN) (3.0 μm). PE-labeled mouse antihuman PECAM-1 (CD31) and rat antimouse IgG1 isotype control antibodies were from BD Pharmingen (San Jose, CA) and used at 0.1 μg/ml. Annexin V-PE was also from BD Pharmingen, and used at 1:100 (per manufacturer's recommendation). APC-labeled antimouse IgG1 isotype from eBioscience (San Diego, CA) was used at 1:200. Streptavidin Cy3 was used at 1:50 per the manufacturer (Invitrogen). Lactadherin-FITC (Haematologic Technologies, Essex Junction, VT) was used per manufacturer's instructions as a Ca++-independent PS-probe to confirm PS staining on biological samples.
Flow cytometry was initially performed on an Accuri C6 and subsequently on a LSRII for confirmation and comparison (BDIS,); the Accuri C6 was chosen initially, as its sheath fluid is distilled water, while the LSRII sheath fluid is a proprietary phosphate-buffered solution. Experiments titrating Ca++ into the various buffers (Fig. 2A) were carried out independently by two researchers. The Accuri C6 was run at the default “slow” flow rate of 11 μl/min, with FSC-H and SSC-H threshold at 10000. The threshold was empirically determined to provide an acceptable minimal level of background (a maximum 100 events/minute on sterile-filtered distilled water), as a smaller threshold resulted in increased noise (data not shown). The background “noise” can be seen in Figure 2 as roughly 1–2 events per microliter in sterile-filtered saline. The events per microliter rate reported by the cytometer was checked against beads with a known concentration, and was consistently on the same order of magnitude (data not shown). For experiments using side scatter to approximate size, no SSC-H threshold was applied. All samples were vortexed immediately prior to flow cytometry examination. Samples were acquired in at least three separate triplicates for 30 sec or 10,000 events (at minimum); 50,000 events (at minimum) were recorded of samples containing RBCs. Flow cytometry analysis was primarily done using C-Flow Plus (BD Biosciences). Singlet populations were gated based on the smallest forward scatter width by height (Fig. 1F). A standard curve for estimating the size based on the side scatter area of singlet microprecipitates was established by a linear fit of the various beads, as the forward scatter at these small sizes is not accurate for estimating size (24). This was repeated on the LSRII; however, the sample intake of the LSRII may contaminate samples with phosphate from the sheath fluid drip during normal operation. The settings of the LSRII were as follows: FSC voltage: 550; SSC voltage: 355; FSC threshold: 1500, which resulted in a background of <1 event/sec on sterile-filtered water; slow flow rate. Samples were collected for 30 sec or 15,000 events. Flow cytometry analysis from the LSRII was done using Cyflogic (CyFlo, Turky, Finland).
For microscopic examination of the microprecipitates, PBS with increasing concentrations of Ca++ was examined directly under the microscope. Additionally, aged blood was washed three times with sterile NS (Sigma). Washed RBCs were then diluted 1:500–1:1000 into sterile-filtered PBS with 0, 2.5, or 5 mM CaCl2 added, and the solution was placed on glass slides. Cells along with microprecipitates were immediately examined with microscopy. Brightfield images were acquired on a Nikon Eclipse using a 40× objective with 200 ms exposure using NIS-Elements software (Nikon, Melville, NY). Image adjustment and analysis was performed using ImageJ (US NIH, http://imagej.nih.gov/ij/, Bethesda, MD). When performing image analysis, particles with less than 10 or more than 10,000 pixels were excluded (this arbitrary range encompassed both the precipitates and RBCs).
As a way to overcome the potential for precipitate formation, EDTA was used in the presence of annexin V with increasing Ca++. This was compared against Lactadherin, which is another PS-binding protein that does not require Ca++. RBC MPs in the supernatant of aged blood were isolated by gentle centrifugation (325g for 10–12 min to remove RBCs). The MP-containing supernatant was then diluted 1:100 in sterile-filtered PBS. Annexin V at 1:100 with or without 0.1% EDTA was added to label RBC MPs. Ca++ was titrated into the solutions as they were examined with flow cytometry.
Statistical analysis was done using Student's 2-tailed t-test when comparing two groups or ANOVA for comparison of two or more groups with differing amounts of CaCl2. For data displayed on a logarithmic scale, data were first log-transformed prior to applying the t-test. An alpha of 0.05 was used to determine significance. Data are presented as mean ± standard error of mean unless explicitly stated.
MPs are challenging to study using flow cytometry (11–13). To ensure correct threshold and gate settings, we used a series of steps to empirically define a correct MP gate. Figure 1A shows the forward scatter signal height by width for distilled and autoclaved water with no threshold on the Accuri C6. We applied a minimum threshold resulting in 1–2 events per microliter as an acceptable background (Fig. 1B). Figure 1C shows beads with a 0.55 μm nominal diameter without the threshold. The beads are clearly seen with the threshold applied (Fig. 1D), and a few residual beads can be seen after repeated filtration of the beads with a 0.20 μm pore-diameter filter (Fig. 1E). This ensures the minimum size range of the threshold is below the International Society of Thrombosis and Hemostasis (ISTH) recommendation of 0.5–0.9 μm for platelet-derived MPs (25). The upper limit of what we considered a MP was established by the FSC-H of 1 μm beads (Fig. 1F). The minimum threshold was maintained in subsequent experiments; the upper limit was re-established in each set of experiments using the same beads when a MP-gate was needed.
Calcium-Phosphate Microprecipitates and MPs Light-Scatter Signal
Shown in Figures 2A and 2B are representative dot plots using the 1 μm forward-scatter gate (representing the upper limit of MPs) based on the FSC-H intensity of commercial beads (as shown in Fig. 1). All dot plots display data from 5 μl of sample or 10,000 events from different samples or conditions.
There was no change in the number of detected events after the addition of CaCl2 to sterile-filtered NS (Fig. 2A). However, there was a concentration-dependent increase in flow cytometry events after the addition of Ca++ to PBS. This increase in total events in PBS is striking, being ∼200-fold more at 2.5 and ∼7500-fold more at 5 mM CaCl2 added than the 0 mM CaCl2 control. These flow cytometry events essentially disappeared with the addition of EDTA to chelate Ca++ (Figs. 2B and 2C). Figure 2D compares RBCs and RBC-derived MPs from an expired donor unit, platelet rich plasma, and various commercial beads. Comparison of Figures 2B and 2D shows clear overlap between the light scatter of microprecipitates and MPs derived from biological sources.
Buffer, Instrument, Time, and Temperature Effects on Microprecipitate Formation
CaCl2 was also titrated into other commonly used buffers to assess the potential of calcium-dependent microprecipitates. HBSS and TBS were used along with NS as a negative control. Detection was performed using two different cytometers to examine machine-specific effects (Figs. 3A and 3B).
As shown in Figure 3A generated using the Accuri C6 cytometer, there was minimal increase in the number of events as a function of Ca++ concentration in TBS and HBSS (from a minimum of 4 to a maximum of 15 events/μl, indicating a 3 to 4-fold increase of baseline). Titration into PBS gave similar results to Figure 2C. With the addition of the chelator, EDTA, to the 5 mM CaCl2 solutions, there was no difference in the number of events in TBS and HBSS; however, as before, there was a marked drop in the number of total events per microliter of calcium-containing PBS (P < 0.001).
Figure 3B shows similar experiments using the LSRII. Again, there is a Ca++ concentration-dependent increase in the number of events. However, substantially more events were observed at lower Ca++ concentrations, and microprecipitates were also observed in nonphosphate buffers (for example, there was a 10-fold increase above baseline in the number of events in HBSS with only 1 mM CaCl2 added; adding 2.5 mM CaCl2 resulted in a 100-fold increase). We observed that the LSRII intake drips phosphate-containing sheath fluid into the sample when positioning the sample; consequently there was an unknown amount of phosphate in all samples detected using the LSRII. This result highlights the fact that instrument factors can also be responsible for microprecipitate formation, and not simply buffer choice.
While forward scatter is used to determine cellular size, it does not accurately represent size of the submicron particles (24); however, side-scatter has been used for this purpose (26). To estimate the size of the microprecipitates based on side scatter, a linear standard curve was established using multiple types of beads with known nominal diameters. Multiplets were excluded based on their disproportionate height to width (see Fig. 1F).
There was no difference between the size of singlets from 2.5 and 5 mM CaCl2 conditions tested in either the Accuri C6 or the LSRII (P > 0.2). The estimated diameter of the microprecipitates using the Accuri C6 was 0.76 ± 0.12 μm (Fig. 3C) compared with the 0.84 ± 0.13 μm using the LSRII (Fig. 3D). From the standard curves, the apparent diameter of these microprecipitate singlets (an estimated 0.64–0.97 μm) fell into the commonly accepted MP range between 0.1 and 1 μm, and the 0.5–0.9 μm platelet-MP window recommended by the ISTH (25).
The time required for microprecipitate formation was examined at room temperature. The time from which CaCl2 was added to PBS until the appearance of microprecipitates generally was less than the sample handling and cytometer lag time. Figure 3E shows a representative real-time experiment. In this series, data were collected at 30 seconds intervals, following the addition of 2.5 mM CaCl2 and immediate replacement of the sample into the cytometer. This process was repeated again with an additional 2.5 mM CaCl2 and then the chelator. Figure 3F shows the precipitate formation rate for different Ca++ concentrations in PBS. The formation and EDTA-dependent dissolution of microprecipitates occurs relatively quickly.
Calcium-phosphate solubility is temperature-dependent (22). To assess this temperature dependence, experiments were carried out keeping all tubes and buffers on ice or at room temperature. In the solutions containing phosphate (PBS and HBSS), there was no significant difference in the mean total events per microliter between the two temperatures across the range of CaCl2 added (data not shown).
Microprecipitate Interaction with Fluorescently Labeled Antibodies
Since antibodies can interact nonspecifically via charged residues (27), the fluorescent signal of the calcium-phosphate precipitates were examined in the presence or absence of fluorescently labeled antibodies (Fig. 4).
In the absence of fluorescently labeled antibodies, there was no significant alteration in the median fluorescent signal with increasing Ca++ (Fig. 4A). In contrast, when antihuman IgG1 PE-labeled antibodies were added, there was a Ca++-dependent increase in the average median fluorescent intensity (MFI) of the microprecipitates (Figs. 4B and 4C).
The above procedure was repeated using different fluorophore-labeled probes, including annexin V, APC-labeled anti-IgG1, PE-labeled anti-CD31 (PECAM-1), and streptavidin Cy3 (Fig. 4D). This phenomenon may be protein-specific, since there was a significant increase in MFI signal with the addition of 5 mM CaCl2 using annexin V, anti-IgG1, and anti-PECAM-1, but not streptavidin. FITC-labeled streptavidin yielded similar results as Cy3-labeled (data not shown).
The MFI at 0 mM CaCl2, was largely variable, since the events were essentially background. Thus, comparisons were carried out across the different buffers and calcium concentrations using a two-way ANOVA. There was a significant difference between the MFI of microprecipitates with annexin V and the isotype antibody (P < 0.05).
Microprecipitates in MP-Containing Biological Fluids
RBCs naturally shed MPs as they age or are damaged; the concentration of MPs in donor blood accumulates over time (6). Thus, to examine the potential overlap between microprecipitates and a known source of MPs, donor blood aged beyond expiration (42 days) was diluted in PBS and examined with flow cytometry.
As shown in Figures 5A and 5C, there was an increase in MP-gated events normalized to RBCs with the addition of 5 mM CaCl2, similar to the order-of-magnitude increase in buffer-only experiments (refer to Fig. 2).
In the presence of PE-labeled antihuman PECAM-1 (CD31) antibodies, there was also an accompanying increase in the number of PECAM-1+ events at 5 mM CaCl2 despite the WBC- and platelet-reduced nature of the biological fluid (Figs. 6B and 6D). This was analogous to the increase seen in buffer-only experiments (refer to Fig. 5).
With the addition of EDTA to chelate Ca++, the concentration of total and PECAM-1 positive MP-gated events again decreased to near baseline, similar to the result seen in buffer-only experiments (refer to Fig. 2).
Microscopy was used to confirm the findings, and particle analysis was used to determine automatically the area of the microprecipitates. Representative images of RBCs in PBS with 0, 2.5, and 5 mM CaCl2 added are shown in the upper portion of Figures 5E, 5F, and 5G, respectively. An automatic particle identification algorithm was used, and the particle outlines are shown directly below. The microprecipitates are identifiable as birefringent, subcellular structures. Using a particle analysis algorithm, the area of the microprecipitates was determined.
Analysis of the calcium-phosphate microprecipitates revealed that while there was a clear difference in numbers, the size of a microprecipitates was similar between those formed either in 2.5 or 5 mM CaCl2 in PBS (P = 0.93, Fig. 5H). The geometric mean area of the microprecipitates was 1.67 ± 0.51 μm2 in the conditions tested. Assuming a grossly spherical shape, the diameter as measured via this method is 0.53 ± 0.16 μm, overlapping with the range of that suggested by flow cytometry, as well as the range commonly used to define a MP.
Annexin V-PS Binding in the Presence of EDTA
Annexin V has a high affinity for PS/Ca++ as revealed by a dissociation constant (Kd) of 7 nM (28) or lower (29); whereas EDTA is a relatively weak Ca++ chelator, with an approximate Kd of 50 nM (30). Thus, to overcome the potential for precipitate formation, annexin V binding to PS and calcium was examined in the presence of EDTA.
Annexin V was added at 1:100 to a 1:1000 dilution of RBC MPs in PBS. The events per microliter significantly increase with the addition of 5 mM CaCl2; however, there was an inconsistent change in the annexin V MFI signal. With the addition of 0.1% EDTA, the number of events per microliter significantly decreased, returning to baseline. There was also a significant increase in annexin V MFI compared with baseline. For comparison and assurance of surface PS on the MPs, a more sensitive and calcium-independent PS probe, lactadherin (31), was also used in identical Ca++-free conditions (Fig. 6C).
The results of this study suggest that, when analyzing MPs with flow cytometry using annexin V (and hence, adding Ca++ as an essential cofactor), one must consider microprecipitates in phosphate-containing samples or buffers. Such precipitates closely mimicked MPs in light scatter; and bound nonspecifically to fluorescently labeled proteins, potentially leading to false-positive cytometry events.
When applied to a biological system with MPs, the addition of Ca++ appeared to cause the formation of PECAM-1 antigen-positive MPs. However, this population was essentially abolished after the subsequent addition of a Ca++ chelator. Additionally, the differences in affinity for Ca++ between annexin V and the relatively weak Ca++-chelator EDTA were used to differentiate MP from microprecipitate. This represents one way to differentiate MPs from microprecipitate. An alternative way to distinguish MP from artifact is the complementary use of microscopy to confirm flow cytometry findings. Microscopy provides qualitative data to compliment quantitative findings obtained with flow cytometry.
While the use of annexin V staining is routinely used for cell-surface PS staining, it is not entirely sensitive (31, 32) nor specific (33) for surface-exposed PS as its binding requires Ca++, multiple PS molecules, and the lipid headgroups around the PS molecules not to interfere with the electrostatic PS-Ca++ bridge (29). On physiologic membranes, the primary phospholipid surrounding PS that enables annexin V binding is phosphatidylethanolamine (PE) (31). Additionally, concerns have been raised about the addition of Ca++ to the effects of down-stream assays (34, 35). This report highlights the utility of calcium-independent phospholipid probes, such as lactadherin for specifically targeting PS (36) or duramycin for specifically targeting PE (37), for the detection of MPs.
Interestingly, the size of the microprecipitates did not differ between 2.5 and 5 mM CaCl2 in PBS. The size of a crystal is determined by numerous factors. Solute concentration is a vital factor, but the interplay of concentration, nucleation rate, and growth rate determine the ultimate size of a crystal (22). In direct visualization, it can be seen that the average singlet microprecipitate size was similar between the 2.5 and 5 mM, but the precipitates were much more numerous in the 5 mM calcium chloride/PBS solution.
There are numerous reports of increasing amounts of circulating MPs in patients with end-stage renal disease (16, 17, 21, 38). The contribution of artifactual versus actual microprecipitates in such studies need to be carefully addressed, since calcium-phosphate crystals have been implicated in vascular pathologies (23, 39). Patients with end-stage renal disease typically have altered mineral homeostasis (i.e., hyperphosphatemia and hypercalcemia). The results of this study additionally point out the potential utility of flow cytometry for microcrystal analysis, as distinguishing MPs from microprecipitates can easily be done with the addition of a chelating agent. This provides an exciting alternative approach for elucidating the effects of microcrystals in human diseases.
In summary, when designing experiments involving MPs, researchers need to carefully consider factors increasing the potential for microprecipitate formation—including reagents, buffers, instruments, and endogenous crystals in the samples of interest.