The site of the study was the Intensive Care Unit at Karolinska University Hospital, Huddinge, Stockholm, Sweden.
Correspondence to: F. Mobarrez, Karolinska Institutet, Department of Clinical Sciences, Danderyd Hospital, Division of Cardiovascular Medicine, Stockholm, Sweden. E-mail: email@example.com
Microparticles (MPs) are small membrane-bound vesicles that arise from activated and dying cells and promote inflammation and thrombosis. To characterize the in vivo release of MPs, we used flow cytometry to measure MPs in the blood of 15 healthy volunteers administered bacterial endotoxin (lipopolysaccharide or LPS) in the presence of a low dose of hydrocortisone with or without inhaled nitric oxide. MPs, defined as particles less than 1.0 μm in size, were assessed following labelling for CD42a, CD14 and CD62E or CD144 antibodies to identify MPs from platelets (PMPs), monocytes (MMPs) and endothelial cells (EMPs). In addition, PMPs and MMPs were labelled with anti-HMGB1 and stained with SYTO13 to assess nuclear acid content. Administration of LPS led to an increase in the numbers of PMPs, MMPs and EMPs as defined by CD62E, as well as the number of MMPs and PMPs staining with anti-HMGB1 and SYTO13. Inhalation of NO did not influence these findings. Together, these studies show that LPS can increase levels of blood MPs and influence phenotype, including nuclear content. As such, particles may be a source of HMGB1 and other nuclear molecules in the blood during inflammation.
Inflammation arises in response to a wide variety of stimuli including products of infecting micro-organism (PAMPs or pathogen associated molecular patterns) as well as molecules released during cell activation, injury or death (DAMPs or damage associated molecular patterns). DAMPs include nuclear and cytoplasmic molecules that undergo translocation or extracellular expression to display immune cell activity. PAMPs and DAMPs can stimulate similar receptors called pattern recognition receptors (PRRs) of which the toll-like receptors (TLRs) have the best-characterized roles during infection [1-3].
Among PAMPs, endotoxin or bacterial lipopolysaccharide (LPS) is one of the most powerful stimulants of immune cell activation. LPS can induce the production of a wide variety of cytokines and other pro-inflammatory mediators as well as activating immune cell populations via TLR4. Depending on the intensity of stimulation in the whole organism, LPS can lead to shock and tissue injury, including the elaboration of DAMPs . Indeed, DAMPs may be important mediators of late events in sepsis and provide target for therapy.
In addition to inducing cytokine production, LPS can elicit the release of microparticles from cells either as a consequence of activation or death . MPs are small membrane-bound vesicles that are released from cells by a blebbing process. Blebbing occurs with both activation and cell death and can lead to particle release by a process dependent on Rho-associated kinases (ROCK) [6, 7]. Importantly, MPs contain a wide array of membrane, nuclear and cytoplasmic components from their cell of origin, including cytokines, growth factors and proteases [8, 9]. Nucleic acids present in particles include both DNA and RNA although the content may vary depending on whether MPs arise during activation or cell death . Because of their content of cellular molecules, MPs can mediate communication among cells and promote both inflammation and thrombosis [11, 12]. In vitro, LPS as well as other PAMPs such as poly (I:C), a ligand of TLR3, can induce MP production in a process that depends upon nitric oxide (NO) . Furthermore, elevated levels of monocyte-derived microparticles (MMPs) occur in the blood of healthy volunteers eight hours after LPS administration .
Among other mediators induced by LPS, high-mobility group box 1 (HMGB1) is a non-histone nuclear protein that serves as a prototype alarmin or DAMP [15, 16] In response to pro-inflammatory stimuli including cytokines and TLR ligands, HMGB1 is translocated from the nucleus to the cytoplasm of cells and released extracellularly. HMGB1 can act alone or in combination with other molecules such as cytokines and LPS, although its immunological effects vary with redox status [17, 18]. HMGB1 levels are increased in sepsis, shock and other inflammatory disease, with beneficial effects of HMGB1 blockade in animal models pointing to a direct role in pathogenesis [16, 19]. While MPs and HMGB1 can be induced by similar stimuli and show increased expression in similar disease states [20, 21], the exposure of HMGB1 on MPs in the blood has not been previously studied.
In studies reported herein, we investigated the effects of LPS administration on the generation of MPs in the blood of healthy human subjects. For this purpose, we used samples obtained in a prior study on the possible effects of inhaled nitric oxide (iNO) in combination with hydrocortisone on the cytokine production induced in vivo by LPS . Blood samples were collected before, during and 24 h after the LPS injection. Using those samples, we assayed particle number and phenotype by flow cytometry, using a phycoerythrin (PE) labelled anti-HMGB1 antibody and SYTO 13 staining to assess the content of nuclear molecules. SYTO 13 is a SYTO dye that has preference for double-stranded DNA although it can bind both DNA and RNA with a high fluorescent yield; this binding allows measurement of MPs on the basis of their nucleic acid content and facilitates the detection of particles from nucleated cells [23, 24].
As results of these studies show, LPS can cause major changes in the profile of MPs in the blood, including increases in levels of MPs from platelets, monocytes and endothelium. Furthermore, LPS administration increased the number of particles containing nuclear molecules as indicated by staining with anti-HMGB1 and SYTO13. Together, these results demonstrate effects of LPS on number and phenotype of MPs in the blood and suggest that HMGB1 exposure on MPs represent a useful biomarker for translocation of nuclear molecules during inflammation.
Materials and methods
The design of the study has been presented previously  and is also shown in Fig. 1. The study had a randomized, double-blind, cross-over design with regard to the effects of iNO or placebo (inhaled N2). A time period of at least 6 weeks separated the experiments. Information regarding type of treatment (iNO or placebo) was kept in sealed and numbered envelopes. To blind the treatment, a separate person broke the sealed envelopes and handled the inhalation (iNO or placebo). All volunteers received LPS and glucocorticoids at both study occasions. Fifteen healthy volunteers (five females and 10 males; mean age 24.5 ± 3.5 years) participated. The volunteers were without any medication for 2 months prior to the experiment, including contraceptives. Female participants had to perform a pregnancy test prior to the experiment (Clearblue® urine pregnancy test Unipath Ltd, Bedford, UK). One male volunteer only completed one of the two experiments (iNO) due to personal reasons.
The volunteers arrived at the laboratory after obtaining a light breakfast without caffeine or nicotine. They were then placed in a reclining position, and two intravenous lines were inserted: an 18 gauge in the antecubital fossa for blood sampling, and a 20 gauge for infusions (BD Venflon™, Franklin Lakes, NJ, USA) in the opposite arm. After a 30-min rest (at time −2.5 h), baseline blood samples were obtained. Nitrogen dioxide (NO2) was constantly measured (INOvent®) in inspired air as recommended by the manufacturer. The gas was mixed and administered through a nasal cannula (SALTER LABS, Arvin, CA, USA). After baseline samples (−2.5 h), treatment with iNO or placebo gas was initiated, lasting for 7.5 h. Two hours after the initiation of iNO or placebo inhalation, LPS was infused during 5 min (at time −0.5 h), followed by a flush of 20 ml saline. Hydrocortisone was injected intravenously 30 min after the LPS infusion (at time 0 h). The site of the study was the intensive care unit at Karolinska University Hospital, Huddinge, Stockholm, Sweden. Written informed consent was obtained from each volunteer before entering the study. The protocol was approved of by the Human Research Ethics Committee at Karolinska Institutet, Stockholm. Data from this study regarding plasma cytokine and clinical responses have been reported previously .
LPS (endotoxin; 2 ng/kg BW, Lot nr G3E069; United States Pharmacopeia, Rockville, MD, USA) was added to sterile water and mixed for 10 min by ultrasound (Bransonic 3510; Bransonic Ultrasonic Corp, Danbury, CT, USA) before administration. Hydrocortisone sodium succinate (2 mg/kg BW; Solu-Cortef ®, Pfizer, Sweden) was used as the glucocorticoid. NO (post-dilution target of 80 ppm NO in inspired gas; INOmax ®, INO Therapeutics, 1060 Allendale Dr. Port Allen, LA 70767, USA) in an oxygen flow of 3 l per minute or placebo (nitrogen, N2, INO Therapeutics) was administered through the INOvent® delivery system, Datex Ohmeda, Inc., Madison, WI, USA.
Blood samples at −2.5, −0.5, 2 and 5 h were collected through an intravenous line (BD Venflon™). Blood samples at 24 h were obtained through a clean puncture of the antecubital vein with no or minimal stasis and after 15–20 min of rest. All samples were collected in test tubes containing 1/10 of 0.129 m sodium citrate. Platelet-poor plasma (PPP) was obtained after centrifugation at 2000g for 10 min at 4 °C and then frozen at −80 °C until analysis.
Assay of microparticles
The previously frozen PPP samples were thawed and centrifuged at 2000g for 20 min at room temperature (RT). The supernatant was then recentrifuged, at 13,000g for 2 min at RT to remove any cells or cell debris prior to flow cytometric analysis. Twenty μL of the supernatant was incubated for 20 min in dark with phalloidin-Alexa-660 (Invitrogen, Paisley, UK, in order to exclude cell membrane fragments) , lactadherin-FITC (Haematologic Technologies, Essex Junction, VT, USA), CD42a-PE (Glycoprotein IX, BD, Clone Alma-16), CD14-PC7 (Beckman Coulter, Dublin, Ireland) and CD62E-APC (Becton Dickinson Immunocytometry Systems, Franklin Lakes, NJ, USA). In addition to CD62E, EMPs were identified through labelling with CD144-APC (AH diagnostics, Stockholm, Sweden). MPs were also stained with anti-HMGB1-PE (R&D Systems, Minneapolis, MN, USA).
MPs were measured by flow cytometry on a Beckman Gallios™ instrument (Brea, CA, USA). The threshold was set on FS, and the MP gate was determined using Megamix beads (BioCytex, Marseille, France); these beads are a mix of beads with diameters of 0.5 μm, 0.9 μm and 3.0 μm, respectively. MPs were defined as particles less than 1.0 μm in size. Conjugate isotype-matched immunoglobulin (IgG1-FITC, IgG1-PE, IgG1-APC and IgG1- PC7) with no reactivity against human antigens was used as a negative control to define the background noise of the cytometric analysis (Fig. 2). The concentration of MPs was calculated by means of the following formula: (MP counted × standard beads ⁄ L) ⁄ standard beads counted, (FlowCount, Beckman Coulter, Dublin, Ireland).
Determination of the nucleic acid content of MPs
To measure the content of nucleic acids in MPs, we modified our flow cytometry method. MPs were identified through gating according to size (forward versus side scatter) and SYTO 13 binding instead of phosphatidylserine (PS) exposure (lactadherin binding). The latter approach was used due to incompatibility between lactadherin-FITC and the SYTO 13 dye. As described above, separate plasma samples were thawed and centrifuged at 2000g for 20 min at RT. The supernatant was then recentrifuged, at 13,000g for 2 min at RT. Twenty μl of the supernatant was incubated in dark for 20 min with 250 nm SYTO 13 (Molecular Probes, Eugene, OR, USA), anti-HMGB1-PE (R&D Systems) together with either CD14-PC7 (Beckman Coulter) or CD41-PC7 (Beckman Coulter); CD41 had to be used instead of CD42a in these experiments due to lack of anti-CD42a antibodies with compatible labelling. MPs were measured with a Beckman Gallios™ flow cytometer as described above. Results are presented as concentrations of MPs (Fig. 5) and as mean fluorescence intensity (MFI; Fig. 6).
Statistical calculations were performed using spss (20.0; SPSS Inc., Chicago, IL, USA) and graphpad prism (5.0c; GraphPad Software Inc, La Jolla, CA, USA) software. Skewed data were logarithmically transformed prior to analysis and a P-value <0.05 was considered to indicate a statistically significant difference. All the statistical analyses were performed between −0.5 h (at LPS administration) and 24 h. To evaluate the effect of treatment (iNO or placebo; factor 1) and changes over time (the effect of LPS; factor 2), a two-factor repeated-measures analysis of variance (ANOVA) was performed (Figs. 3-5). Differences in MFI were assessed between two types of particles using a two within-factors repeated-measurements ANOVA. The two within-factors consisted of the following: (1) type of particle; and (2) effect over time, and the interaction type of particle and time (Fig. 6).
Samples for these analyses came from a prior study to determine whether glucocorticoids, alone or in combination with inhaled nitric oxide, could modulate the response to LPS in normal human subjects. In that study, the effects of LPS were manifest as all participants, after the infusion, reported varying degrees of malaise and other symptoms (e.g. headache, fever and fatigue) appearing within 60–90 min and lasting 3–4 h. Furthermore, white blood cell count, TNF-α, IL-6, IL-8 and IL-10 increased significantly over time, with a normalization of the values at 24 h . As the subjects receiving LPS showed a robust inflammatory response despite the treatment with glucocorticoids with or without inhaled NO, we think these samples provide an accurate reflection of the response to LPS. We have therefore used them to characterize MP production. Of note, there was no difference between iNO/glucocorticoid and placebo/glucocorticoid treatments in levels of any of the analytes measured , suggesting that iNO did not affect the responses measured.
Flow cytometric assay of microparticles
In the current study, the number and phenotype of MPs in blood were first determined by flow cytometry. Using approaches described previously to assure detection of particles rather than cell fragments , we found that the percentage of particles in the MP gate positive for phalloidin was 10.6 ± 6.9% (mean ± SD). These findings indicate acceptable sample handling with low numbers of cell fragments in plasma samples. As results showed further, the total number of PS+ MPs (lactadherin positive) and cell-specific MPs (CD42a, CD62E and CD14 positive particles, i.e. platelet- [PMPs], endothelial- [EMPs] and monocyte-derived microparticles [MMPs], respectively) increased significantly over time after LPS infusion (P < 0.001, Fig. 3). Of note, EMPs defined by CD144 (VE-Cadherin) did not increase significantly (Fig. 3D).
Phenotypic properties of MPs were next assessed. As LPS can induce HMGB1 translocation into the extracellullar space as well as MP release from macrophages/monocytes [14, 26, 27], we determined whether MPs from monocytes could expose HMGB1. As data in Fig. 4 indicate, HMGB1+ MMPs also increased significantly over time (P < 0.001, Fig. 4). In the sample obtained at 5.5 h, we observed the highest concentration of MMPs positive for HMGB1, which was approximately 10% of all PS+ MPs. These results are notable because assay of these samples by ELISA failed to show an elevation in soluble HMGB1 .
We measured HMGB1 exposure on platelets and monocyte MPs together with nucleic acid content (SYTO 13 binding) to assess any coordinate release of nucleic molecules. As HMGB1 binds to DNA, we also investigated whether SYTO13+ MMPs or PMPs could expose HMGB1. As results of these experiments showed, both the total number of SYTO 13+ and SYTO 13+ MMPs and PMPs increased significantly over time (Fig. 5A,B,D). A similar pattern was observed when SYTO13+ HMGB1+ exposure was measured on MMPs and PMPs, respectively (Fig. 5C,E). These findings indicate that PMPs, like MMPs, both expose HMGB1. Furthermore, as shown in Fig. 5, we observed that the number of SYTO13+ PMPs or SYTO13+ HMGB1+ PMPs was higher compared with the number of SYTO13+ MMPs or SYTO13+ HMGB1+ MMPs (around 1.5 times and 10 times higher, respectively). When SYTO 13 data are expressed as MFI (Fig. 6A), we found significantly higher levels of SYTO13 binding in MMPs than PMPs (P < 0.001). No significant differences were observed in HMGB1 exposure (expressed as MFI) on SYTO13+ HMGB1+ particles of the two types, that is, MMPs and PMPs (Fig. 6B, P = 0.9).
Results of these studies provide new insights into the effects of LPS on the generation and properties of circulating MPs. Thus, in plasma of normal human subjects exposed to LPS in the presence of hydrocortisone, we found significant increases in the number of PMPs (CD42a+), EMPs (CD62E+) and MMPs (CD14+). We also found a significant increase in the number of PMPs and MMPs containing nucleic acids (as indicated by SYTO13 binding) and exposing HMGB1. Although the dose of LPS in these experiments was comparatively low, it was nevertheless sufficient to induce MP production from several cell populations, indicative of a significant system effect.
As described previously, the original purpose of this study was to determine the effects of a combination of inhaled NO and glucocorticoid on the response to LPS, testing the hypothesis that inhaled NO could potentiate the glucocorticoids action by up-regulating the glucocorticoid receptor . Because of this design, all subjects received hydrocortisone. Clear-cut elevations in particles occurred in both experimental settings, indicating that MPs are increased by LPS exposure despite the interventions. While we cannot exclude greater effects on MP production in the absence of glucocorticoids, we feel that our findings depict well MP formation by LPS, especially in view of evidence of increased cytokine production as described previously .
In this study, we measured the number of EMPs using CD62E (E-selectin) or CD144 (VE-Cadherin) as markers, observing an increase only with CD62E+ MPs (Fig. 3E). While CD144 has been proposed as a marker for vascular endothelial cells, its use may provide less sensitive assays than staining for CD62E, which may be more specific for activated endothelial cells [29, 30]. CD144 is weakly expressed on EMPs in healthy individuals although elevated levels of circulating CD144+ EMPs occur in patients with the antiphospholipid syndrome . The increase in CD62E+ MPs is in accordance with a previous study which showed activation of the vascular endothelium following LPS administration as reflected by elevated plasma levels of von Willebrand antigen . Studies on EMP expression in septic patients have produced varying results, however [33, 34]. In the present study, the number of MPs in plasma observed is higher than those demonstrated in previous studies in humans on the effects of LPS . In our assays, we used lactadherin instead of annexin V to detect MPs. Lactadherin is a protein of approximately 50 kDa that can bind to PS in a calcium-independent manner. This compound provides more sensitive detection of PS exposure than annexin V, which is commonly used in flow cytometry studies . The use of lactadherin and a more advanced flow cytometer (Beckman Gallios™) may explain the higher numbers of MPs than previously reported .
In addition to showing increases in particle numbers with LPS, our findings indicate changes in their phenotypic properties, including an increase in HMGB1+ MMPs. HMGB1 is an abundant nuclear protein that is expressed widely by cells that respond to LPS, including endothelial cells, leucocytes and platelets . During cell activation, HMGB1 can translocate from the nucleus to the cytoplasm for eventual transport into the extracellular space. HMGB1 can also exit from cells during both necrosis and apoptosis [37-39]. Once in the extracellular space, HMGB1 is immunologically active, mediating downstream effects of LPS either alone or in association with LPS and cytokines . However, in the present study, we cannot determine whether HMGB1 exposed on MMPs and PMPs is immunologically active.
As shown in these studies, blood contains a particle-associated form of HMGB1 with an increase in HMGB1-positive particles from both platelet and monocytes occurring at 5.5 h. While studies on HMGB1 have focused on its origin in nucleated cells, HMGB1 also occurs in platelets where it was originally identified as amphoterin . In addition to activation, platelets, despite the absence of a nucleus, can undergo apoptosis [40, 41], suggesting that HMGB1 could be translocated into particles.
The presence of HMGB1 on particles is especially notable in view of our prior studies analysing plasma levels of HMGB1 in these samples. These studies used a commercial ELISA kit to assess HMGB1 levels in plasma although at different time points (1.5, 3.5 and 24 h after LPS administration) . These assays did not show significant elevations of HMGB1. The difference between ELISA and flow cytometry results could reflect differences in the sensitivity and specificity of the assays. HMGB1 can bind directly to PS  and consequently could bind to PS exposed on MPs of various cell origins. Our flow cytometry method detects HMGB1 exposed on particles and can allow assessment of the cellular origin of the particles with HMGB1 (i.e. platelets, monocytes or endothelial cells). The ELISA technique, on the other hand, measures both free- (soluble) and particle-bound HMGB1 in plasma with the risk of disturbing matrix effects. It is also possible that the peak in HMGB1 exposure on particles occurring at 5.5 h was missed at the 3.5 h sampling. Of note, studies on HMGB1 in patients with trauma or acute inflammation (appendicitis) show elevations within hours of hospital admittance [42, 43].
Studies using staining with SYTO13 provide additional information concerning the origin of HMGB1 (and potentially other nuclear proteins) in the MPs. SYTO13 binds both DNA and RNA with high fluorescent yield, although it has preference for double-stranded DNA. As particles contain both DNA and RNA, SYTO13 binding facilitates detection of particles, especially small particles that may be missed by light scatter . As shown previously , the binding of SYTO13 by MPs generated from apoptotic cells in vitro reflects the presence of both RNA and DNA that likely occurred during translocation into blebs. As LPS can cause apoptosis, an increase in SYTO13+ MPs or SYTO13+ HMGB1+ MPs could result from cell death induced by LPS.
As described in the Results section, we observed a higher number of SYTO13+ or both SYTO13+ HMGB1+ PMPs compared with MMPs (Fig. 5). When data on SYTO13+ PMPs and MMPs are expressed as MFI, we found significantly higher levels of SYTO13 binding in MMPs compared with PMPs (Fig. 6A, P < 0.001). This finding is not unexpected as monocytes, in contrast to platelets, are nucleated cells and should contain more DNA and RNA. The amount of HMGB1 present on SYTO13+ HMGB1+ PMPs did not differ from the corresponding type of MMPs (Fig. 6B, P = 0.9). These findings could result from a number of processes. Thus, translocation of HMGB1 to PMPs could differ in extent compared with that of MMPs. Alternatively, HMGB1 released by platelets could associate with the plasma membrane during activation via binding to exposed PS and sulfoglycolipids . In addition, HMGB1 released from other cell types, such as monocytes or endothelial cells, could bind to activated platelets. Although we measured HMGB1 on PMPs, we cannot be sure of the cellular origin of the HMGB1 molecules found on these particles. It is thus possible that the higher number of SYTO13+ HMGB1+ in the PMP population relates to the higher concentration of circulating PS+ PMPs compared with PS+ MMP (Fig. 3B,C), with the PS involved in capturing and binding extracellular HMGB1 in the blood.
In summary, we have demonstrated that LPS administration to healthy volunteers can increase circulating levels of MPs and influence the phenotype and nuclear content of the MPs. As such, these findings suggest that measurement of particle-bound HMGB1, especially in conjunction with staining for DNA/RNA, may be a sensitive biomarker to assess the effects of LPS (and other stimuli) on translocation of nuclear molecules during immune activation and/or cell death.
The authors are very grateful to Inga Hellström for technical assistance. NO (INOmax®) and the INOvent® was kindly provided by Ylva Sjöquist, Ikaria, Inc. Author CF wishes to disclose financial interest in the clinical use of iNO.
These studies were supported by a VA Merit Review grant, NIH grant AI056363, and by grants from the Swedish Heart and Lung foundation, Karolinska Institutet, and through the regional agreement on medical training and clinical research (ALF) between Stockholm County Council and Karolinska Institutet. Minor financial support was also obtained from CF Research & Consulting AB (Stockholm).
C.F wishes to disclose financial interest in the clinical use of inhaled nitric oxide (iNO). He has participated in patent applications for the clinical use of iNO as well as acted as a consultant on a part-time basis as clinical expert in the development of iNO as therapy with marketing authorization in the European Union. All other authors declare no competing financial interests.