SEARCH

SEARCH BY CITATION

Keywords:

  • atomic force microscopy;
  • flow cytometry;
  • microparticles;
  • platelets;
  • thrombosis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

Summary. Background: Platelet microparticles (PMPs) are a promising prognostic marker for thrombotic disorders because of their release during platelet activation. The use of flow cytometry for the enumeration of PMPs in plasma has generated controversy due to their size, which is below the stated detection limits of conventional flow cytometry instruments. The potential impact of this is an underestimation of PMP counts. Objectives/Methods: To address this possibility, we used a combination of fluorescence-activated cell sorting (FACS) and atomic force microscopy (AFM) to determine the size distribution of PMPs present in plasma from acute myocardial infarction (AMI) patients and normal volunteers, and PMPs generated by expired platelet concentrates and washed platelets treated with agonists such as thrombin and calcium ionophore (A23187). Results: According to AFM image analysis, there was no statistically significant difference in height or volume distributions in PMPs from thrombin-activated, calcium ionophore-activated, expired platelet concentrates and plasma from healthy volunteers and AMI patients. Based on volume, expired platelets released the greatest proportion of exosomes (< 1.0 × 10−22 L3 in volume) in relation to the entire PMP population (29.7%) and the smallest proportion of exosomes was observed in AMI patient plasma (1.8%). Moreover, AFM imaging revealed that PMPs from expired platelets exhibited smooth surfaces compared with other PMP types but this was not statistically significant. Conclusions: We confirm that flow cytometry is capable of analyzing PMPs from plasma by using AFM to perform nanoscale measurements of individual PMP events isolated by FACS. This method also provided the first quantitative nanoscale images of PMP ultrastructure.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

Initially termed ‘platelet dust’ by Wolfe [1] in 1967, platelet microparticles (PMPs) are vesicular fragments of platelet cell membrane released during platelet activation, as occurs in thrombin-mediated platelet activation and platelet storage lesion (PSL), in which platelet concentrates become activated and generate PMPs de novo [2–8]. While the ‘microparticle’ descriptor term refers to platelet-derived fragments < 1.0 μm in diameter, various studies have determined PMP size distributions to actually range between 50 and 500 nm [9,10]. PMPs hold clinical significance because they have been shown to exhibit thrombogenic potential in various pathophysiological contexts owing to surface expression of tissue factor [3,11–14] and PAI-1 [15]. As a result, PMP enumeration in patient plasma has emerged as a potential prognostic marker in thrombotic disorders and cancer progression [16–18].

Platelet microparticle enumeration is complicated by a plasma subpopulation of GP1b-positive exosomes, originally the alpha-granules of platelets, which are exocytosed during platelet degranulation. Exosomes range from 40 to 100 nm in diameter [19], but these markers and size range cut-offs are in conflict with recent proteomic studies of PMP subpopulations by Dean et al. [9], whereby the smallest PMP size fraction exhibited an average diameter of 130 nm with expression of both α-granule constituents and plasma membrane receptors. Furthermore, all four main PMP size fractions isolated in this study revealed distinct variations in total protein, total phospholipid and lipid/protein ratios, underscoring the heterogeneity in PMP composition and size. Indeed, platelet exosomes are considered to be platelet microparticles, with their primary distinguishing feature being their smaller diameter compared with the rest of the submicron platelet microparticle population.

While flow cytometry continues to be utilized for high-throughput enumeration of PMPs it is beset by a major technical limitation that may lead to an underestimation of PMP counts. The submicron size of PMPs and exosomes makes enumeration and sizing by flow cytometry uncertain [20] because the wavelength of incident light used in conventional flow cytometry instruments (488 nm) exceeds the reported mean diameter of PMPs present in plasma [9,10]. This technical limitation may lead to erroneous PMP enumeration by flow cytometry despite numerous reports describing the detection and enumeration of PMPs by this technique. As a consequence, the ISTH Scientific and Standardization Subcommittees have moved towards the standardization of flow cytometry (FC) instrumentation for PMP enumeration in patient plasmas by evaluating all commonly used FC instruments and fluorescent calibrated submicron beads [21] and have determined that PMP enumeration by flow cytometry is ‘feasible but dependent on the intrinsic characteristics of both the flow cytometer and the calibration strategy’ [21]. Therefore, the continued use of flow cytometry for PMP enumeration and biomarker expression on microparticles invoked our present study to determine the exact size distribution of the entire PMP range as detected by flow cytometry.

To size PMPs analyzed by flow cytometry, we employed a strategy to isolate all PMP events analyzed by a FACS instrument deemed suitable by the ISTH Scientific and Standardization Subcommittees and designed for high rates of cell sorting (MoFlo, Beckman-Coulter, Burlington, Canada; a wide FS platform) by which the diameter, height and total volume of each sorted PMP was determined by atomic force microscopy (AFM). In this study, PMPs were defined as CD41-positive events that exhibited a forward scatter below that of 1.0 μm Alexa488-labelled microspheres. We hypothesized that PMPs analyzed by flow cytometry would exhibit size distributions similar to earlier reported values of PMP size distributions using AFM [9,10]. With our method, the majority of PMP events analyzed by flow cytometry were isolated and sorted directly onto mica coverslips for imaging by atomic force microscopy [22] (Fig. 1A). Because individual PMPs were being directly deposited onto mica sheets, FITC-labelled liposomes (< 1.0 μm in diameter) were also analyzed and sorted onto mica sheets to determine if the method would result in microparticle and liposome disintegration.

image

Figure 1.  Schematic of FACS analysis and sorting of platelet microparticles (PMPs) and liposomes onto mica sheets for atomic force microscope (AFM) imaging. (A) Schematic of how PMPs and liposomes detected by FACS are sorted directly onto mica coverslips for AFM imaging. (B) FS log/SS log histoplot of 1.0-μm Alexa488-labelled microspheres. The red line indicates the FS log setpoint for setting the sort gates for events exhibiting an FS log lower than the 1.0-μm microsphere population. (C) Sort gate (grey boxes) for PMPs formed de novo by expired platelet concentrates. (D) PMPs present in acute myocardial infarction (AMI) patient plasma. (E) FITC-labelled liposomes isolated from platelet-poor plasma from normal human whole blood. (F) PMPs formed by thrombin activation of washed human platelets. (G) PMPs formed by calcium ionophore A23187 activation of washed human platelets. Events shown in C–D and F–G exhibited significant CD41a-FITC IgG binding compared with isotype controls. (H) Annexin V and CD62P surface expression on platelets over 9 days of storage. At least 30 000 CD41a-FITC-positive events were analyzed for each stain. n = 3 per group.

Download figure to PowerPoint

Atomic force microscopy is a form of scanning probe microscopy in which a cantilever controls the XYZ movements of a probe that ‘feels’ the surface of a sample in a raster-based scan, generating a three-dimensional nanoscale topographical map of the surface travelled [22,23] with an imaging resolution of 1–10000 nm [24]. The use of AFM for quantitative analysis of microparticles is not new, generating similar results produced by other analytical platforms such as dynamic light scattering and ultracentrifuge size fractionation [5,9,10,25,26]. However, due to the sensitivity and the ability to achieve nanoscale resolution by AFM, artifacts such as dust, protein and ions often complicate the task of imaging an object of interest, such as the abundance of cell debris and plasma protein during the imaging of platelet microparticles [25]. Hence, volumetric analysis of particles is marred by plasma proteins bound to the object of interest or immediately adjacent to it. We describe methodology that overcomes the contribution of artifacts present in plasma by employing a FACS isolation step to effectively separate PMPs from plasma proteins and other microparticles, thus excluding the primary source of noise during AFM imaging [27]. In order to minimize probe-sample interactions during scanning, we used tapping-mode AFM where a continuously oscillating probe is used to scan the sample [28]. AFM imaging can be performed immediately after sample isolation, which is in contrast to the lengthy sample processing steps required for electron microscopy. In this report, we describe the first nanoscale resolution images depicting the morphology and volumetrics of individual platelet microparticles isolated from plasma from normal and acute myocardial infarction patients, expired platelet concentrates and platelet microparticles generated by thrombin and calcium ionophore activation of washed human platelets.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

Blood plasma collection from AMI patients and PMP generation from normal platelets

Whole blood was collected from patients 48 h post-AMI and healthy volunteers who consented to this study as approved by the University of British Columbia and St Paul’s Hospital ethics board guidelines. Blood was collected into ACD Vacutainers (BD Biosciences, Burlington, Canada) and within 12 h was centrifuged for 20 min at 2000 × g at room temperature and the upper layer representing the platelet-poor plasma (PPP) fraction was removed to a point 2 cm from the buffy coat interface and subsequently stored at −80 °C after FACS analysis. To generate expired platelet concentrates, whole blood from healthy volunteers was drawn into ACD Vacutainers and centrifuged for 10 min at 200 × g at room temperature. The upper fraction (platelet-rich plasma, PRP) was transferred to a new tube and submitted to constant angular rocking at 22 °C for 9 days. After incubation, this PPP was centrifuged for 20 min at 2000 × g at room temperature and the upper fraction (platelet-poor plasma) was used for FACS sorting [15]. To generate washed human platelets, PRP from healthy volunteer whole blood was centrifuged for 10 min at 2000 × g at room temperature and the platelet pellet was washed twice with calcium-free Tyrodes buffer, pH 7.4 containing 0.35% BSA and 0.1 mg mL−1 apyrase. Platelets were then resuspended in calcium-free Tyrodes buffer containing 0.01 mg mL−1 apyrase [15]. To generate thrombin-activated PMPs, 2 μL of 100 mm CaCl2 and 10 units of α-thrombin (Enzyme Research Laboratories, South Bend, IN, USA) were added to 20 μL of washed platelets and left at room temperature for 30 min. After centrifugation at 5000 × g for 10 min, 5 μL of the supernatant was used for FACS sorting of PMPs generated by thrombin activation of washed human platelets. To generate calcium ionophore (A23187)-activated PMPs, 1 μL of 1 μm calcium ionophore A23187 and 2 μL of 100 mm CaCl2 were added to 20 μL of washed platelets and left at room temperature for 30 min. After centrifugation at 5000 × g for 10 min, 5 μL of the supernatant was used for FACS sorting of PMPs generated by calcium ionophore activation of washed human platelets. An ELISA kit to measure active PAI-1 in patient PPP was used, in which 5 μL of PPP diluted in 20 μL of reaction buffer was assayed in triplicate (Molecular Innovations, Ann Arbor, MI, USA). Table 1 summarizes patient data for plasmas collected in this study.

Table 1.   Normal volunteer and acute myocardial infarction patient data. Various relevant patient data parameters as recorded and provided by the clinic
Number of subjects6
Male/female6/0
Age (yr)57 ± 3.3
Smoker1/6
Complications post-analysis2/6
Patient plasma active PAI-1 (U mL−1)5.2 ± 1.1
Control plasma active PAI-1 (U mL−1) N = 91.1 ± 0.1

To determine the mechanism of PMP release by expired platelets, a 20-μL aliquot of PRP was taken at days 0, 3 and 9 of storage and performed under sterile conditions. Each aliquot (n = 3) was stained with antibodies CD41a-FITC and CD62P-PE or Annexin V-RPE (BD Biosciences). At least 30 000 CD41a-FITC-positive platelet events were analyzed.

Liposome preparation and size measurements

Liposomes were prepared by the thin film evaporation method using DOPC, TopFluor-PE, Cholesterol and DC-cholesterol (Avanti Polar Lipids, Alabaster, AL, USA) at 60:30:4:6 molar ratio. All the lipids were dissolved in chloroform. Chloroform was removed by rotary evaporation (Buchi R114C, Buchs, Switzerland) for 1 h under vacuum (10 mbar) at 37 °C. The dried lipid film was hydrated with phosphate buffer saline (PBS) at room temperature under constant agitation for 1 h. Liposomes were manually extruded (Liposofast-Basic; Avestin Inc., Ottawa, Canada) through a polycarbonate membrane filter (pore diameter 1000 nm). The total lipid amount was fixed at 6.66 mm.

The particle size of the liposomes was determined by dynamic light scattering using the Brookhaven 90Plus particle size analyzer. Liposomes were diluted to a final concentration of 0.1 mg mL−1 using PBS, prefiltered through 0.1-μm membrane filters.

Fluorescence-activated cell sorting (FACS) of PMPs and liposomes

The mean size of the liposomes as determined by dynamic light scattering was 528 ± 17 nm and the polydispersity index was 0.362 ± 0.008 (FITC-labelled, ∼0.1 mg), and were incubated in whole blood from healthy volunteers (1 mL) overnight at 37 °C. Blood was centrifuged at 1500 × g at room temperature for 20 min to generate PPP; 50 μL of this PPP was diluted in 5 mL of PBS, and FITC+ve events were gated regardless of size and sorted onto cleaved mica sheets and then stored within 50-mL Falcon tubes prior to AFM imaging. All sorted events were ‘pelleted’ directly onto the freshly cleaved mica sheet. For platelet microparticles, 10 μL of CD41a-FITC mouse IgG (BD Biosciences) was incubated with 100 μL of PPP for 30 min at room temperature and then diluted with 2 mL of PBS. To sort CD41a-positive events < 1.0 μm in diameter, neutral density filters in FACS instrument (MoFlo XDP, Beckman-Coulter/Dako-Cytomation, Burlington, Canada) were removed and forward scatter (FS) gain was increased until Alexa488-microsphere beads (1.0 μm diameter; Invitrogen, Burlington, Canada) were detected (Fig. 1B). The MoFlo instrument was used because of its high recovery and yield during sorting experiments. Events that exhibited an FS below the setpoint marked by microspheres as shown as the red line through Fig. 1(B–G), were gated as platelet microparticles (PMPs) as represented in the shaded boxes in Fig. 1(C–G). Although commercially available flow cytometry instruments do not accurately analyze and size particles with diameters smaller than the incident wavelength [20,29,30], events with fluorophore signal (FITC) can be detected below the 1.0-μm FS setpoint but these FS values are not quantitative [29].

During analysis of PMPs, the sorting compartment of the MoFlo was opened such that mica coverslips were placed directly underneath the sort stream for microparticle isolation instead of being sorted into collection tubes. With a high speed sort of PMP or liposomes (i.e. > 10 events per second) the mica coverslip was moved in a slow circular motion along the X-Y plane, resulting in a linear drop arrangement of PMPs and liposomes onto the mica coverslip. Mica coverslips with adhered PMPs and liposomes were then placed into sterile Petri dishes until imaged by AFM. Enumeration of PMPs in all samples was performed as described in [8], with the exception that the CD41a-FITC antibody was used.

Atomic force microscopy of platelet microparticles and liposomes

Mica coverslips were prepared by first mounting the entire mica stack onto a glass slide with two-sided tape. Scotch tape was applied to the top of the mica and then removed, which pulled off an entire layer of mica, leaving behind a freshly cleaved layer of mica on the glass slide. Freshly cleaved mica sheets (2 cm2; Ted Pella Inc., Redding, CA, USA) were pretreated with a dH2O wash, followed by adsorption of 10 μL of 5 mm MgCl2 onto the mica sheet for 5 min to neutralize any negative charges, followed by a dH2O wash, and then blow dried with nitrogen gas. A BioScope (Bruker AXS Inc., Madison, WI, USA) was used for Tapping Mode AFM imaging and microscopy performed in ambient air using Veeco Tapping Mode etched silicon cantilevers (k = 40 N m−1), and a tip radius of < 10 nm was used. Tapping mode scan rate of 1–3 Hz was used. XY scan directions were calibrated with a 10 × 10 μm2 grid. The z-direction was calibrated with 5 nm diameter gold particles (Ted Pella Inc.). The entire process of sorting PMP events onto mica coverslips and AFM imaging of sorted events required 2–3 h.

Atomic force microscope image acquisition and processing

All image acquisition was performed using the nanoscope 7.30 controller software (Bruker AXS Inc.). Height and amplitude images were obtained simultaneously. Images were processed by Gwyddion 2.22 (http://gwyddion.net/). All height images were processed by applying a median height correction, followed by a horizontal scar correction (if applicable), followed by a mean plane subtraction correction.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

Detection of platelet microparticles and liposomes by flow cytometry

Using the MoFlo XDP High-Speed cell sorter, Alexa488-labelled 1.0-μm microspheres were analyzed and used to set the forward scatter (FS) gate for analysis and sorting of all events < 1.0 μm in diameter (Fig. 1B). PMPs defined as being CD41a positive compared with IgG-FITC isotype control were detected by flow cytometry in aliquots of expired platelet concentrates, AMI patient plasma samples and in thrombin and calcium ionophore-activated washed human platelets, all yielding similar FS histoplots (Fig. 1C–D, F–G, respectively). FITC-labelled liposomes isolated from human whole blood were detected in a similar manner (Fig. 1E).

To determine if release of PMPs by expired platelets was by platelet apoptosis or by platelet activation, PRP was taken at Days 0, 3 and 9 and stained with the apoptosis marker, Annexin V-RPE, or the platelet activation marker, CD62P-PE. Figure 1(H) reveals that at each time-point there is a greater percentage of CD62P-positive platelets than AnnexinV-PE platelets. Day 9 platelets exhibited the highest proportion of CD62P expression, whereas Annexin V expression did not change over time. This model of expired platelets and PMP formation is consistent with other reports of platelet activation as the primary mechanism of PMP release [6].

Sorting of platelet microparticle and liposome events for atomic force microscopy

Using the sorting gates in Fig. 1(C–G), PMP and liposome events were sorted directly onto mica coverslips instead of collection tubes. Upon diverting these events onto mica coverslips, AFM imaging was immediately performed on sorted objects ‘deposited’ onto this ultraflat surface, as seen in Fig. 2(A) with PMPs generated by expired platelet concentrates and as seen in Fig. 2(B) with liposomes isolated from whole blood. With the relatively high speed of the sort (i.e. > 10 events per second) the coverslip was moved in a slow circular motion along the X-Y plane, which resulted in a linear drop arrangement of PMPs and liposomes onto the mica coverslip (Fig. 2A,B, respectively). This linear drop arrangement of PMPs allowed for straightforward identification of sorted objects while also preventing unidentified debris and plasma proteins from being erroneously imaged. This technique effectively concentrates PMPs for imaging as opposed to the low density of PMPs seen in Fig. 2(C) (yellow arrow represents a single microparticle) when plasma alone is plated onto the mica coverslip. Sorting liposomes with this method also concentrated them for imaging, which also resulted in a linear drop arrangement of liposomes onto the mica coverslip (Fig. 2B), whereas liposomes are unidentifiable from other cell-derived microparticles present in the plasma when platelet-poor plasma from whole blood supplemented with FITC-labelled liposomes was imaged (Fig. 2D).

image

Figure 2.  Comparison of AFM images of PMPs and liposomes isolated by FACS analysis and unsorted PMPs and liposomes present in plasma. (A) Sorted CD41a-positive PMPs from expired platelet concentrates as arranged onto mica by FACS. (B) Sorted FITC-labelled liposomes as arranged onto mica by FACS. (C) Unsorted human plasma sample with microparticles and cell debris of unknown origin. There are fewer PMPs observed due to lack of sorting of CD41-positive microparticles onto mica. (D) Unsorted human plasma sample containing FITC-liposomes mounted onto mica.

Download figure to PowerPoint

Nanoscale quantitation of PMP and liposome ultrastructure by AFM

The strength of AFM is its ability to probe a given object at nanometer resolution to generate a topographical perspective of the object ultrastructure. The top and bottom panels in Fig. 3(A) represent the height and amplitude channels of sorted liposomes (left panel) and PMPs isolated from normal and AMI patient plasma (middle and right panels), revealing vesicular objects with topographically active surfaces. While the height channel provides topographical distance information, the amplitude channel reports the amount of voltage required for the cantilever to maintain contact with the sample, and as a consequence generates a superficial topographical profile with a shadowing effect. Figure 3(B) describes scan line analysis of the object’s height profile as represented in the height channel of Fig. 3(A), revealing both the maximum diameter of the object, its surface granularity and mound-like contour. The right panels of Fig. 3 represent AFM images of a single liposome isolated by the same process used to isolate and image PMPs, revealing intact vesicular structures with smooth surfaces as determined by scan line analysis (right panel, Fig. 3B). Hence, the method used to isolate and image PMPs is not destructive to sorted events and can be used for ultrastructural characterization of microparticles and liposomes present in complex mixtures such as plasma and cell culture media. Fig. 3(C) consists of three-dimensional renderings of the object’s height channel, thus highlighting the surface topographies of the three PMP types, with a corresponding root mean square roughness value of each object (RMS roughness).

image

Figure 3.  AFM of liposomes and individual PMPs isolated from normal plasma and acute myocardial infarction (AMI) patient plasma. (A) Height and amplitude images (top and bottom panels, respectively) of FITC-labelled liposomes (left panels) sorted from normal plasma, sorted PMPs from normal plasma (middle panels), and sorted PMPs from AMI patient plasma (right panels). (B) Scan line analysis revealing the peak height, diameter and surface roughness of liposomes and PMPs in (A). (C) Three-dimensional rendering of the height channels of PMPs presented in (A) and RMS roughness value of the height channel.

Download figure to PowerPoint

Figure 4 describes AFM image analysis of PMPs generated by thrombin-activated (left panels) and calcium ionophore-activated (middle panels) human washed platelets and by expired platelets (right panels). These PMPs all exhibit surface granularity with portions of smooth surface, similar to PMPs from normal plasma. In contrast, PMPs from AMI patient plasma exhibited substantial surface granularity and no smooth surfaces.

image

Figure 4.  AFM of individual PMPs isolated from washed platelets activated with thrombin and calcium ionophore and from expired platelets. (A) Height and amplitude (top and bottom panels, respectively) of sorted PMPs from washed platelets activated with thrombin (left panels), PMPs from washed platelets activated with calcium ionophore A23187 (middle panels), and PMPs generated de novo from expired platelet concentrates (right panels). (B) Scan line analysis revealing the peak height, diameter and surface roughness of PMPs and liposomes in (A). (C) Three-dimensional rendering of the height channels of PMPs presented in (A) and RMS roughness value of the height channel.

Download figure to PowerPoint

PMPs generated de novo by expired platelets exhibit ‘membrane flaps’

One key observation made by these studies was that morphologically, the majority of PMPs generated by expired platelets (64.5%) exhibited very thin flap-like extensions at the base of these microparticles. In contrast, very few PMPs from the other sources exhibited these extensions (< 4% in all the other groups). Using scan line analysis, the extensions found on PMPs generated de novo from expired platelets exhibited a height of 5.28 ± 0.64 nm (Fig. 5A–C) and in particular, some of these extensions contained cable-like structures beneath them (Fig. 5E). These flat extensions varied in size between PMPs and the thickness of these flap-like extensions is similar to reported height values of bilayered phospholipid membranes [31]. Considering the agreement in height with phospholipid bilayer membranes and the relative surface homogeneity of these flat structures, these extensions were termed ‘membrane flaps’.

image

Figure 5.  Membrane flaps on PMPs and exosomes as imaged by AFM. (A) Height (left) and amplitude (right) channel images of a representative PMP from expired platelet concentrates with phospholipid membrane flaps at the base of the PMP. (B) Scan line analysis of the height image in (A) revealing the height of the flaps. (C) Three-dimensional rendering of the height channel image in (A) as shown in two different perspectives. (D) Height channels of two representative CD41a-positive exosomes from expired platelets (PSL) with volumetric data. (E) Height (left) and amplitude (right) channel images of a PMP from expired platelet concentrates exhibiting a membrane flap with cable-like structures.

Download figure to PowerPoint

Diameter, height, volumetric and roughness analyses of PMPs and liposomes

Figure 6 summarizes AFM image analysis of PMPs from liposomes (liposomes = black spots/bars, n = 26), AMI patient plasma (AMI = grey spots/bars, n = 109), normal healthy volunteer plasma (normal = white spots/bars, n = 43), from washed human platelets activated by thrombin (thrombin = red spots/bars, n = 44), from washed human platelets activated by calcium ionophore A32187 (ionophore = green spots/bars, n = 49) and from expired platelets (PSL = yellow spots/bars, n = 51). The peak height of each object (Fig. 6A), maximum XY diameter (Fig. 6B), object volume (Fig. 6C), and root mean squared roughness (Fig. 6D) was assessed in all PMP groups, with no significant differences observed in these morphological parameters amongst PMPs (mean ± SD, all P = NS, anova). Only statistically significant differences in root mean squared roughness (RMS roughness) between liposomes and any of the PMPs was observed (P < 0.05, anova, n > 20 in all groups) The majority of PMP events imaged were not spherical and were in fact ovoid and therefore the maximal XY diameter parameter was used. Furthermore, a disparity in height and diameter measurements (Fig. 6A vs. 6B) was observed in all groups. Because this technique sorts objects such as PMPs directly onto the surface of a mica coverslip, adhered microparticles undergo an ultrastructural collapse, and transform from a sphere-like object into a mound-like object. Although this compression artifact may underestimate height and diameter values, volumetrics of microparticles and liposomes are still maintained, providing useful information regarding sizing (Fig. 6C), with the majority of all objects imaged exhibiting particle volumes far below that of 1.0 × 10−18 L3.

image

Figure 6.  Quantitation of morphological parameters of sorted PMPs and liposomes. AFM height channel images were used to determine (A) peak heights of each object, (B) maximum XY diameter, (C) volume of each object and (D) root mean square roughness (RMS roughness). The grey area denotes theoretical volumes of spherical objects over a diameter range of 1.00–0.10 μm and the pink area denotes theoretical volumes of spherical objects with a diameter < 0.10 μm.

Download figure to PowerPoint

Exosomes, which are 40–100 nm in diameter, were successfully detected by flow cytometry and sized by AFM (Fig. 5D), indicating that exosomes comprise a significant fraction of total CD41-positive events in both AMI patient plasma and expired platelet concentrations. Exosomes were classified as PMPs exhibiting a volume of < 1.0 × 10−21 L3, and 1.8% of CD41-positive events in AMI patient plasma were classified as exosomes, whereas 25.6%, 29.5%, 22.4% and 29.7% of CD41-positive events in normal healthy volunteer plasma, thrombin-activated, calcium ionophore-activated and expired platelet concentrates were classified as exosomes, respectively. Our results demonstrate that patient plasma samples have a very low proportion of platelet exosomes, whereas de novo formation of PMPs such as expired platelet concentrates generate a significantly higher proportion of exosomes out of the entire PMP population. To quantitate surface roughness, the root mean squared roughness (RMS roughness) calculation was employed to analyse PMP surfaces from all groups, with liposomes exhibiting the lowest RMS roughness value. All biophysical parameters of PMPs from all groups are presented in Table 2. Table 3 provides data regarding the enumeration of PMPs as detected by flow cytometry in all plasma samples and activated human platelet preparations.

Table 2.   Biophysical parameters of platelet microparticles from various sources. Tabulated form of data presented in Fig. 6
 Liposomes (n = 26)AMI (n = 109)Normal (n = 43)Thrombin (n = 44)Ionophore (n = 49)PSL (n = 51)
Height (nm) (mean ± SD)104.4 ± 82.76160.9 ± 80.48107.2 ± 64.37156.2 ± 116.4165.9 ± 109.4141.2 ± 96.31
Max diameter (nm) (mean ± SD)830.2 ± 395.7765.9 ± 301.1553.7 ± 314.6531.4 ± 280.3443.9 ± 300.6527.6 ± 323.6
Volume (L3) (mean ± SD)4.1 × 10−20 ± 5.2 × 10−203.7 × 10−20 ± 8.5 × 10−202.8 × 10−20 ± 4.9 × 10−201.3 × 10−19 ± 3.1 × 10−191.0 × 10−19 ± 1.9 × 10−192.5 × 10−20 ± 5.4 × 10−20
RMS roughness (nm) (mean ± SD)16.6 ± 6.863.4 ± 20.256.8 ± 18.166.1 ± 31.162.9 ± 13.535.9 ± 15.5
Table 3.   Enumeration of CD41a-positive microparticles in plasma samples
 AMI (n = 6)Normal (n = 5)PSL (n = 3)Ionophore (n = 3)Thrombin (n = 3)
CD41-positive MPs uL-1 (mean ± SD)4265 ± 26232836 ± 22155070 ± 291910020 ± 31053806 ± 2905

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

To determine the sizes of PMPs that are detected by flow cytometry, we developed a technique whereby FACS and AFM were used to isolate and image individual PMPs at the nanoscale resolution. This technique allowed us to confirm that flow cytometry is capable of detecting PMPs that exhibit diameters less than one micron. PMPs from different plasma sources were evaluated, such as AMI patient plasma, normal plasma, PMPs generated de novo in expired platelet concentrates and washed platelets activated with thrombin or calcium ionophore. The height, maximum XY diameter, volume measurements and RMS roughness values of PMPs from these sources made possible by AFM are in agreement with a previous study that sized PMPs from patient plasma [9], and a significant proportion of PMPs generated by expired platelets were exosomes, whereas very few exosomes were present in patient plasma samples.

Having shown that our technique is able to size individually sorted PMPs at the nanoscale resolution, our findings show a discrepancy with a previous study that utilized AFM to image PMPs but differed in isolation of PMPs, in which mica coverslips were coated with CD41-specific antibodies in order to isolate PMPs from patient plasma [10]. It is possible that repeated washes of the mica coverslips in order to eliminate non-specific binding of other microparticles may have also eliminated larger PMPs from being imaged by AFM, which would explain the lack of PMPs with a diameter > 100 nm. According to our images, none of the PMPs imaged exhibited a peak height of 14.7 nm as reported by Yuana et al., which may represent immobilized CD41 IgG complexed with a fibrin degradation product. In contrast, our technique does not require a wash step because all sources of noise are separated from the PMP population by FACS, thus permitting the identification of PMPs and exosomes across a broad size range (50–1000 nm diameter range). More importantly, these discrepancies perhaps reveal the ineffectiveness of ELISA-based assays for PMP enumeration because of the requirement for washing steps prior to analysis. Therefore, flow cytometric analysis may be more accurate than ELISA assays in enumerating non-exosome PMP populations present within plasma because these solid-phase-based assays fail to effectively immobilize PMPs > 100 nm in diameter. Rather, the technique developed by Yuana et al. is better suited for enumerating platelet-derived exosomes. While dynamic light scattering technology has emerged as the gold standard for PMP enumeration in expired platelet concentrates [5], our AFM images of PMPs isolated from expired platelet concentrates reveal a non-spherical morphology as evidenced by phospholipid flaps at the base of PMPs. Hence, analytical settings of these devices should be corrected to allow accurate enumeration and sizing of PMPs from expired platelet concentrates.

According to our images of PMPs and liposomes, ultrastructural deformation caused by immobilization on the mica coverslip reveals that height and XY diameter alone cannot be used to determine the size distributions of PMPs. In contrast, PMPs as imaged by transmission electron microscopy remain suspended in their native state during embedment and processing [32]. To overcome this, we relied upon volumetric analysis to size PMPs, assuming that PMP immobilization will not result in changes to this parameter. Judging by images of liposomes that have no cytoskeleton, we found no irregularities on surface topographies, which suggests minimal incorporation of air within isolated microparticles. Furthermore, because each sorted event is immersed within a single droplet of sheath fluid, it is unlikely that air is incorporated within the collapsing PMP during immobilization on mica and during evaporation of sheath fluid.

Furthermore, we show that sorting of liposomes does not cause disintegration despite the physical impact of the droplets on mica during sorting. Given the variation of flow rates in FACS instruments, and variation in structural integrity of liposome formulations, flow rates can be adjusted should liposome disintegration occur. As the samples are exposed to ambient air, sorted microparticles may also exhibit deformation artifacts due to dehydration. However, in our experience, the deformation artifacts introduced by this method are minimal and do not significantly influence image interpretation.

The insights gained by this technique have provided the first nanoscale perspective of PMP populations and we find that PMPs generated de novo are ultrastructurally different from PMPs generated in vivo. Typically released during cell activation, microparticles are ‘miniature envoys with many faces’ [33], with inherent biological effects that may or may not resemble the surrogate cell. The observed heterogeneity of microparticles is not based only on size, but also in terms of surface protein accumulation, which is likely to be dependent on the physiological context of their release.

Overall, our method incorporates purification and precise image-based quantitation, which should facilitate the characterization of microparticles from diverse sources. In particular, the ability to collect detailed biophysical data at the microparticle surface due to protein binding or functionalization opens the door for more elegant analyses of microparticles from in vivo studies.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

This study was supported by CIHR MOP-84535 to J. D. Lewis, CIHR 20R90839 to T. J. Podor, PhD studentships from the Canadian Blood Services and Michael Smith Foundation for Health Research, and PDF from the Canadian Breast Cancer Foundation to H. S. Leong. We thank Y. Jiao for his expertise with atomic force microscopy, S. van Eeden for AMI plasma samples, and B. Whalen and A. Meredith for expertise with FACS.

Disclosure of Conflict of Interests

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References

The authors state that they have no conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of Conflict of Interests
  9. References