Particulate matter from indoor environments of classroom induced higher cytotoxicity and leakiness in human microvascular endothelial cells in comparison with those collected from corridor.

AN0122561518;hyb01may.17;2018Jul06.12:50;v2.2.500

Particulate matter from indoor environments of classroom induced higher cytotoxicity and leakiness in human microvascular endothelial cells in comparison with those collected from corridor. 

We investigated the physicochemical properties (size, shape, elemental composition, and endotoxin) of size resolved particulate matter (PM) collected from the indoor and corridor environments of classrooms. A comparative hazard profiling of these PM was conducted using human microvascular endothelial cells (HMVEC). Oxidative stress‐dependent cytotoxicity responses were assessed using quantitative reverse transcriptase polymerase chain reaction (RT‐PCR) and high content screening (HCS), and disruption of monolayer cell integrity was assessed using fluorescence microscopy and transwell assay. Scanning electron microscopy (SEM) coupled with energy‐dispersive X‐ray spectroscopy (EDX) analysis showed differences in the morphology and elemental composition of PM of different sizes and origins. While the total mass of PM collected from indoor environment was lower in comparison with those collected from the corridor, the endotoxin content was substantially higher in indoor PM (e.g., ninefold higher endotoxin level in indoor PM8.1‐20). The ability to induce oxidative stress‐mediated cytotoxicity and leakiness in cell monolayer were higher for indoor PM compared to those collected from the corridor. In conclusion, this comparative analysis suggested that indoor PM is relatively more hazardous to the endothelial system possibly because of higher endotoxin content.

cardiovascular; corridor environment; cytotoxicity; endothelium; indoor environment; particulate matter

Practical Implications

Air quality of classrooms has been a subject matter for many studies because of its relevance to the health status of students who spend considerable amount of time inside classrooms. Understanding the relationship between composition of PM collected from student occupied classroom environments and biological mechanisms underpinning cardiovascular diseases is important in predicting the health consequences and in building engineering controls to mitigate health risks. Relatively higher endotoxin content in indoor PM and its effect on viability and integrity of primary endothelial cells imply possible health risks of PM originating from classroom environments.

Introduction

Epidemiological and toxicological studies have unraveled the increasing mortality and morbidity associated with cardiovascular diseases owing to the exposure to airborne particulate matter (PM).[1] , [2] , [3] , [4] , [5] As evidenced, the increase in cardiovascular mortality is not only related to chronic exposure but also to short‐term elevation of PM content in ambient air.[6] , [7] The biological mechanism underpinning PM exposure to cardiovascular diseases involves direct impact of PM that trespass pulmonary epithelium into circulation and indirect effect resulting from systemic oxidative stress and inflammation.[1] , [8] Certain components such as soluble metals and organic compounds present in inhaled PM of size 2.5 micron and below could enter into the systemic circulation from the respiratory epithelium, leading to interaction with endothelial cells to have negative impact on the function of endothelial system.[9] , [3] PM‐induced dysfunctional endothelial responses such as loss of nitric oxide (NO) generation leading to atherosclerotic plaque formation and endothelial cell leakiness, which manifest as edema of the alveoli and other areas of vasculature leading to life‐threatening conditions, are shown by previous studies.[2] , [10]

Generally, PM is composed of a complex mixture of elemental carbon (soot), organic carbon compounds such as polycyclic aromatic compounds (PAH), inorganic compounds such as metallic compounds and acids (e.g., sulfates and nitrates), and biological entities such as endotoxins.[11] The relative abundance of these constituents varies with the size and origin of PM which in turn is reflected in their toxic potential.[12] , [13] Higher proportion of redox‐active organic molecules in fine and ultrafine PM samples increases their potential to cause oxidative stress and toxicity in comparison with coarse particles.[8] The redox‐active organic and inorganic constituents catalyze the formation of reactive oxygen species (ROS) which cause oxidative stress responses in cells. In addition to redox‐active organic molecules, biological constituents such as endotoxin from outer membrane of Gram‐negative bacteria could also influence the toxic potential of PM.[14] Endotoxin level could vary with the size of the PM [14] as well as the environment from which the PM is collected.[15] Relatively, higher endotoxin levels are detected in PM samples collected from indoor environments in comparison with those collected from the outdoor environments.[15]

PM collected from indoor environments of classrooms has been a subject matter of many studies due to its relevance to the health status of students who spend considerable amount of time inside the classrooms.[16] While such studies have contributed to the knowledge on the health hazards of PM from indoor environments, studies addressing the relationship between physicochemical properties of PM originating from classroom environment to their biological effects are scarce. Of those studies conducted so far, very little attention has gone into studying the potential effect of PM on the endothelial system.

Here, we report the physicochemical properties and hazard potential of size resolved PM samples collected from classroom and the corridor environments. We chose primary endothelial cells as an in vitro model in this study due to its relevance to physiological and pathological roles in the human body. Most importantly, endothelial cells that line the lung microvasculature create the first biological barrier between air and the rest of the body.[10]

Materials and methods

Sample collection and preparation

Three sampling campaigns were carried out in a tertiary education institution in Singapore in September 2014, December 2014, and July 2016. PM samples from the indoor and the corridor environments of four different classrooms were collected using commercially available particle collector—Naneum Nano ID Selecttm (Naneum Ltd, Canterbury, UK). Figure S1 depicts the sampling spots and the sampling machine used. Each sampling was performed for a total period of 35 hours during the school days when the classrooms were occupied with students. The classrooms were located at the ground floor of the school building, and the floors (area of 350 m2) were carpeted. While the total seating capacity was 130 (plastic chairs), approximately 60 students (17‐19 years old) attended successive classes that ran throughout the sampling duration. The classrooms were air‐conditioned and mechanically ventilated, with a temperature of 20°C and relative humidity of 60%. Air‐handling unit was of constant supply air volume (CAV) system with an air velocity of 2.6 m/s. Fresh air was taken from the corridor, and the filtration was carried out with pre‐filters and soft bag filters. The school premise had low to medium traffic flow. For PM collection from the indoor environment, the sampling device was placed away from the windows and doors (to avoid direct effect from corridor environments) and positioned at the back of the classroom.

Size differentiated particles were collected onto pre‐weighed, clean‐dry glass slides in the following size ranges: 20‐35 μm (stage 1, PM20‐35), 8.1‐20 μm (stage 2, PM8.1‐20), 4.0‐8.1 μm (stage 3, PM4.0‐8.1), 2.0‐4.0 μm (stage 4, PM2.0‐4.0), 1.0‐2.0 μm (stage 5, PM1.0‐2.0), 0.5‐1.0 μm (stage 6, PM0.5‐1.0), and 0.25‐0.5 μm (stage 7, PM0.25‐0.5). Immediately after the completion of sampling, the glass slides containing PM were transferred to petri dishes, sealed, and placed in humidity‐controlled storage boxes. The mass of PM samples collected onto individual glass slides was determined within 24 hours of sampling using a microbalance (XS105, Mettler Toledo, Singapore) in a temperature/humidity‐controlled environment (25°C, RH 55%). PM samples were retrieved from the glass slides using dimethyl sulfoxide (DMSO) so as to obtain 1 mg/mL stock solution. The mixtures were then transferred to micro‐centrifuge tubes and were stored in 4°C before using them (~1 month) for additional assays. For endotoxin assay and cellular studies, aliquots of stock solution were retrieved and freeze‐dried immediately before adding EndoGRO™ medium to prepare the first working concentration of PM. It should be noted that, freeze‐drying may reduce the volatile component of PM but may not affect the physical dimension of PM as the hydrodynamic diameter measured using dynamic light scattering showed decreasing agglomeration size along the decreasing primary particle size (data not shown).

PM characterization

The morphology (shape and size) of PM samples was determined using an environmental scanning electron microscope (ESEM, VP‐SEM SU1510, Hitachi Ltd, Tokyo, Japan). For ESEM, the samples were analyzed directly from the glass slides without sputter coating within 1 week of sample collection. The elemental compositions of the samples were analyzed with energy‐dispersive X‐ray spectroscopy (EDX) coupled with the field emission scanning electron microscopy (FESEM, JSM‐6701F, JOEL USA Inc, Peabody, MA, USA). PM samples were sputter coated with platinum before FESEM‐EDX analysis for semi‐quantitative estimation of the elemental composition of PM samples.

Endotoxin assay

The endotoxin levels in PM samples were determined using ToxinSensor™ Chromogenic LAL endotoxin assay kit (Genscript, Piscataway, NJ, USA). The protocol was slightly modified from the manufacture's instruction to run the analysis in a multiwell format. For this, 30 μL of PM samples from stock solution was mixed with 70 μL of deionized water before freezing and subsequently subjected to freeze‐drying. Thirty microliters of LAL (Limulus amebocyte lysate) reagent water was added to the freeze‐dried samples and sonicated for 30 s at 80 W. Subsequently, 10 μL of PM samples (freeze‐dried and reconstituted) and the standards (positive and negative as supplied in the kit) were dispensed into individual wells of a sterile 96‐well plate in triplicate and was mixed with 10 μL of reconstituted LAL. The plate was incubated at 37°C for 60 minutes. After incubation, 10 μL of reconstituted chromogenic substrate solution was added into each well and incubated at the same condition for another 6 min. Subsequently, 50 μL of reconstituted color stabilizer was added into the wells and the absorbance was read at 545 nm. Sample blanks were prepared by replacing the chromogenic substrate with 10 μL deionized water. Absorbance reading of the blank was subtracted from that of PM sample reading. The levels of endotoxin in PM samples were determined by substituting the absorbance values of sample into the equation obtained from the standard curve (obtained using incremental levels of endotoxin provided in the assay kit).

Cell culture and PM solution preparation

Human microvascular endothelial cells (HMVEC) (Life Technologies Holdings Pte Ltd, Singapore) were cultured in EndoGro™‐endothelial cell growth medium (Merck Millipore Division, Merck Pte Ltd, Singapore). Cells grown to 80% confluency in T75 flask were used for cytotoxicity assessment in Passage 10‐12. Automated liquid handing system (Nimbus, Hamilton, NJ, USA) was used to obtain eight incremental concentrations of PM (0.49‐30 μg/mL, corresponding to 0.56‐34.5 mg human exposure[12] in EndoGRO™ for cell exposure studies.

Cell Viability

We conducted assays for cell viability in HMVEC using resazurin reagent. Resazurin is an electron acceptor (from electron transport chain) which when reduced forms a fluorescent product, resofurin. Cell viability can be assessed depending upon the fluorescence intensity resulting from the formation of resofurin.[17] Cells grown in 96‐well plates were added with cell culture media containing PM and were incubated for 24 hours. Later, each well received 100 μL of resazurin solution and the fluorescence intensity was measured (excitation 560 nm and emission 590 nm) after 30 minutes of incubation. Sample blanks were prepared by adding resazurin solution with different concentrations of PM in cell culture media. The fluorescence intensity was recorded at the same wavelength. The blank reading was subtracted from the sample reading to avoid interference from PM during cell viability assay. The data are expressed as % negative control.

To verify the role of endotoxin in determining the cytotoxicity of PM samples, we conducted resazurin assay in cells treated with PM samples added with Polymyxin B. Polymyxin B is an antibiotic that partially neutralizes the effect of endotoxin.[18] For the assay, PM samples collected from classroom that showed highest level of endotoxin were mixed with 10 μg/mL of Polymyxin B in EndoGRO™. PM samples with and without Polymyxin B were added to HMVEC cells, and the resazurin assay was conducted as detailed above.

Multiparametric cytotoxicity assessment using high content screening

PM‐induced oxidative stress responses were assessed using a previously reported high content screening (HCS) strategy[12] , [19] , [20] , [21] , [22] ; Cellular responses such as altered nuclear area, intracellular hydrogen peroxide, mitochondrial superoxide generation, mitochondrial depolarization, intracellular calcium flux, and cell membrane damage were measured using Hoechst 33342, DH2‐DCFH‐DA, MitoSox™, JC1, Fluo‐4™, and propidium iodide (PI) (Molecular Probes, Life Technologies Holdings Pte Ltd, Singapore), respectively. Wavelength compatible fluorescent probes were mixed to prepare three dye cocktails: Hoechst 33342+DH2‐DCFH‐DA+MitoSox™, Hoechst 33342+JC1, Hoechst 33342+Fluo‐4+PI.

For the purpose of conducting HCS, 5000 cells in 50 μL EndoGRO were transferred to three 384‐well plates (Greiner, Monroe, NC, USA) using a microplate dispenser system (ELX 406; Biotek, Winooski, VT, USA). After incubating the plates for 24 hours, the culture medium was replaced with 50 μL of PM samples at eight incremental doses (twofold dilutions starting with 30 μg/mL) and plates were incubated for another 24 hours. Subsequently, each plate received 25 μL of PBS containing a dye cocktail. Fluorescence images (blue, green, and red) of cells from these plates were acquired using 10× objective of the Image‐XpressMicro XLS Widefield High Content Screening System (Molecular Devices, Singapore). Images were analyzed by Meta‐Xpress software (Molecular Devices) to score the percentage of cells positive for each cytotoxicity signal. Hoechst 33342 dye was used to count the total number of cells using the following blue channel settings: minimum width=3 μm (~3 pixels), maximum width=10 μm (~7 pixels), threshold intensity above background=50 gray levels. The area of the Hoechst 33342 stained region was measured to measure the nuclear area. The corresponding red channel settings for PI and MitoSox Red were the following: minimum width=5 μm (~6 pixels), maximum width=30 μm (~22 pixels), threshold intensity above background=250 gray levels and that for JC1 (green channel) was minimum width=5 μm (~6 pixels), maximum width=30 μm (~22 pixels), threshold intensity above background=750 gray levels.

Cells with fluorescence intensity above the threshold levels were scored positive for that cytotoxicity response, and the percentage (%) of cells positive for particular cytotoxicity response was generated by the software based on total number of cells. Strictly standard mean difference (SSMD) values were calculated from quadruplicate raw data (% cells positive for each cytotoxicity signal). A heat map was constructed based on the SSMD values to summarize the multiparametric data wherein a red signal indicated SSMD value ≥3 and green representing 1 and shades in between for intermediate SSMD values.

Measurement of nitric oxide

Production of NO was measured using Griess Reagent Kit (Promega, Singapore). For this assay, 50 μL of supernatant from cells treated with 30 μg/mL of PM samples was added to the wells of a 96‐well plate in triplicate. Subsequently, the wells were added with 50 μL of the sulfanilamide solution. After 10 minutes of incubation at room temperature, 50 μL of the NED solution was added into all the wells. The mixture was incubated for another 10 minutes and the absorbance was measured using a plate reader and compared against the standard reference curve.

Immunofluorescence staining

The integrity of endothelial cells monolayer was visualized with an immunofluorescence assay as reported previously.[23] Briefly, 40 000 HMVECs were seeded in the eight‐well chamber slide and were incubated under standard cell culture conditions overnight to form a confluent monolayer of endothelial cells. Exposure to PM was carried out by replenishing the growth medium with PM supplemented medium. Following the treatment, the media was removed and the treated cells were washed thrice with chilled PBS. Thereafter, the cells were fixed with the addition of 4% paraformaldehyde for 15 minutes, permeabilized for 15 min by adding permeabilization solution (0.2% Triton X‐100 in PBS), and were blocked with blocking solution (2% BSA, 0.1% Triton X‐100 in PBS). Following that, the fixed cells were incubated overnight at 4°C with anti‐VE‐cadherin antibody solution (1:200; Cell Signaling Technology, Danvers, MA, USA). The cells were then washed thrice with PBS and were incubated for 1 hour with Alexa‐488‐conjugated secondary antibody solution (1:400, Life Technologies). Antibody solutions used in the study were prepared in PBS solution containing 0.2% BSA. Finally, these cells were added with the ProLong® gold antifade with DAPI stain (Life Technologies), and the images were captured using Leica epifluorescence inverted microscope (DMI6000; Leica Microsystems (SEA) Pte Ltd, Singapore) with 40× magnification objective lens.

To ascertain the involvement of ROS, the endothelial cells were pre‐treated with N‐acetylcysteine (final concentration 1 mmol/L; Sigma‐Aldrich, Singapore) for 2 hours prior to PM treatment. Subsequently, the solution was removed and replaced with 30 μg/mL of PM solutions supplemented with 1 mmol/L NAC. The cells were treated for 6 hours, and at the end of incubation period, the cells were further processed as previously described.

Transwell permeability assay

The degree of endothelial cells leakiness was measured with transwell permeability assay as previously described.[23] Briefly, 20 000 of HMVECs were cultured on the 6.5‐mm transwell insert (with polycarbonate filter, 0.4 μm pore, Corning, Singapore) for overnight to achieve confluent monolayer. The monolayer endothelial cells were then treated with PM suspensions supplemented with 1 mg/mL of FITC‐dextran (MW 40 000; Sigma‐Aldrich). Following the treatment, 100 μL of solution at the basolateral compartment was taken and the amount of FITC‐dextran penetrating the insert was analyzed with microplate reader (H4FM; Biotek). The fluorescence signal was quantified at excitation and emission wavelength of 492 and 520 nm, respectively. The reading from the treated samples was normalized against the control, and their fold change was reported.

To ascertain the involvement of ROS, monolayer of endothelial cells was pre‐treated with N‐acetylcysteine (final concentration 1 mmol/L, Sigma‐Aldrich) for 2 hours prior to PM treatment. Following that, the solution was removed and replaced with 30 μg/mL of PM solution supplemented with 1 mmol/L NAC and 1 mg/mL of FITC‐dextran. The cells were then incubated for 6 hours. At the end of the incubation period, the sample from the basolateral compartment was taken and analyzed as previously described.

Quantitative real‐time polymerase chain reaction (qRT‐PCR)

HMVEC cells were seeded at 10 000 cells/cm2 in 24‐well culture plates (Greiner, NC, USA) and were allowed to grow for 24 hours. These cells were then exposed to 30 μg/mL of selective PM (PM8.1‐20, PM2.0‐4.0 and PM0.25‐0.5) for 6 and 24 hours. RNeasy Mini Kit (Qiagen Singapore Pte Ltd, Singapore) was employed to run qRT‐PCR using EXPRESS One‐Step SYBR® GreenER™ Kits (Life Technology). The expression of targeted genes was normalized to GAPDH and β‐actin wherein the primers for housekeeping and specific genes were selected from primer bank (http://pga.mgh.harvard.edu/primerbank/) as listed below.

Statistical analysis

All experiments were repeated thrice with at least triplicate samples in each run. Values are presented as the mean±SD. Student's t test was used for significance analysis of comparisons between two groups. One‐way analysis of variance (ANOVA) was applied for the comparisons among multiple values. “P value” <.05 was considered statistically significant.

Results

Differential mass, size, morphological and elemental properties of corridor, and indoor PM

Figure [NaN] A shows the total mass of collected PM and calculated concentrations in corridor and indoor air. In general, PM concentrations in the corridor air were higher than those from the indoor air. For example, the concentration of PM4.0‐8.1 in the corridor air is 30 μg/m3, which is almost three times that from the indoor air (10.24 μg/m3). The ratio of indoor/corridor PM concentration ranged from 0.3 to 0.7. Furthermore, it is observed that the majority of the mass contribution, 60%‐70% of the total amount of PM, was from the larger PM samples collected on to stages 1‐3 (Figure [NaN] B).

Deposition of size‐segregated PM onto glass slides slotted into successive stages of the sampling machine is evident from the ESEM images in Figure [NaN] C. The size of primary particles decreased consistently from stages 1 to 7. In both PM samples, the larger particles were found to be of irregular shape while the smaller particles were granular and spherical. Interestingly, square tabular‐shaped crystals were found in corridor and indoor PM0.25‐0.5 (stage 7) and membranous and amorphous structures were observed in indoor PM4.0‐8.1, PM2.0‐4.0, and PM1.0‐2.0 (stages 3‐5).

Table [NaN] summarizes semi‐quantitative estimation of the elemental composition of PM samples. EDX peaks corresponding to Si, Ca, and Al were the most prominent ones in the larger PM samples (PM8.1‐20 and PM2.0‐4.0). Minor peaks corresponding to Na, Mg, K, Cl, and Fe were also detected. Fe was comparatively higher in PM2.0‐4.0. Notably, amorphous‐shaped particles observed in PM0.25‐0.5 showed the presence of C and S as the major elemental components. However, the EDX analysis of the crystalline structures in PM0.25‐0.5 revealed composition of Ca, S, and K. The EDX analysis of the flaky‐shaped particles in stage 4 of indoor PM showed composition of C, O, Si, Ca, and Na.

ESEM‐EDX analysis of selected corridor/indoor PM samples. The table shows the atomic percentage of major elements

Additional data on the mass concentration, ESEM images, and EDX analysis of PM samples collected from other classrooms are given in supplementary information (Figure S2 and Table S1). While the mass concentration varied, similar morphology and chemical composition were observed across each size range of PM samples collected from indoor environments of different classrooms.

Indoor PM showed relatively higher endotoxin level

PM samples collected from indoor and corridor showed the presence of endotoxin (Figure [NaN] ). The endotoxin level was found to be relatively higher in indoor samples in comparison with those present in the corridor samples. For example, endotoxin content in indoor PM8.1‐20 was 1.1 EU/m3 which was nine times higher than that in the corridor sample of same size range. Generally, the endotoxin level was observed to be relatively higher in bigger PM samples (PM8.1‐20 and PM2.0‐4.0) in comparison with PM0.25‐0.5 (Figure S3).

PM showed dose‐dependent decline in cell viability

Resazurin assay was used to assess the viability of HMVECs exposed to corridor and indoor PM samples for 24 hours. We observed a dose‐dependent decrease in viability of cells exposed to PM samples (Figure [NaN] A). About 40%‐60% of cell death was observed at the highest tested dose of PM (30 μg/mL). Significantly higher cell death was observed for indoor PM2.0‐4.0 in comparison with corridor PM2.0‐4.0 at 30 μg/mL concentration.

Addition of Polymyxin B was observed to reduce the cytotoxicity of PM samples (Figure [NaN] C). For example, viability of cells exposed to 30 μg/mL of indoor PM8.1‐20, was increased from 45% to 93% when PM samples were mixed with Polymyxin B. Other size fractions also showed a similar trend.

High content screening showed size‐ and dose‐dependent cytotoxicity of ...

Figure [NaN] B shows the heat map of the SSMD values for the cytotoxicity responses assessed in PM exposed cells. We observed that both corridor and indoor PM lead to intracellular ROS production, [Ca2+]i flux, depolarization of mitochondria, and cell death (PI uptake). There was a noticeable difference in the cytotoxic potential of corridor and indoor PM. For example, significant effects were found on the intracellular ROS production (DCF), mitochondrial superoxide generation (MitoSox), mitochondrial depolarization (JC1), and [Ca2+]i flux for HMVECs treated with indoor PM.

NO production

Figure [NaN] summarizes the amounts of NO measured in cells exposed to PM collected from indoor and corridor environments. PM collected from corridor induced significantly higher amount of NO when compared to the indoor PM. For instance, the NO generation in cells treated with 30 μg/mL of corridor PM2.0‐4.0 is 89% higher than that of indoor PM2.0‐4.0.

Corridor and indoor PM‐induced leakiness in the monolayer of endothelial cells

We investigated the effect of PM samples on the endothelial cell function. Endothelial cells’ main role is to form physical barrier that serves to regulate the solutes transport across the blood vessel. As such, the barrier integrity of the endothelial beds holds significant importance in the overall functioning of the endothelial cells barrier. Thus, we sought to understand whether the corridor and indoor PM could exert disruption on the endothelial cells barrier. Corridor and indoor PM samples induced the formation of intercellular gaps between the endothelial cells (Figure [NaN] A) which otherwise forms intact monolayer. In addition to this visual observation, the extent of endothelial cells monolayer disruption was quantitatively assessed with transwell permeability assay.

We found that PM sample induced dose‐ and time‐dependent disruption of the endothelial cells monolayer, as evidenced by the increase in FITC‐dextran that was found in the basolateral compartment of the transwell insert. Dose‐dependent increase in FITC‐dextran penetration was observed following the introduction of PM samples with concentration ranging from 3.75 to 30 μg/mL onto the endothelial cell monolayer for 6 hours (Figure [NaN] B). Approximately 1.7‐ to 1.8‐fold increase in FITC‐dextran penetration level was detected after the monolayer cells were exposed to corridor PM, while the indoor airborne PM were found to induce leakiness on the monolayer by 1.6‐ to 2.1‐fold change compared to the untreated control (Figure [NaN] B).

PM samples induced the disruption of the endothelial cells barrier in a time‐dependent manner (Figure [NaN] C). Leakiness on the endothelial cell barrier was detected as early as 1 hours after the exposure and persisted up to 24 hours. The leakiness reached approximately 2.6‐ to threefold change following 24‐hours exposure to both corridor and indoor PM (Figure [NaN] C). The detected leakiness at 24‐hours exposure corresponded well with our previous cytotoxicity data that showed significant amount of cell death. This is expected as cells shrink when they die. The cell shrinkage leads to the disengagement of endothelial cells from each other and finally results with the disruption of the monolayer barrier.

Hierarchical oxidative stress responses and pro‐inflammation induced by PM

Quantitative RT‐PCR was applied to explore the oxidative stress and pro‐inflammation responses caused by corridor and indoor PM in the transcriptional level. As shown in Figure [NaN] , antioxidant enzyme HO‐1 was generally shown to be upregulated after 6 hours of PM exposure. A significant upregulation of a pro‐inflammatory marker of IL‐8 was found after 24 hours. NF‐κB signaling pathway was possibly activated, as suggested by the upregulation of NF‐κB1 and IL‐8 expression level following the cells exposure to PM samples. PM exposure to HMVECs also stimulated the upregulation of another pro‐inflammatory marker of ICAM1 expression by 24 hours. In contrast, a decrease in cell membrane markers (especially at the cellular junction sites) of PECAM1 and VE‐cadherin was observed after 24 hours of PM exposure.

The endothelial cells barrier leakiness is modulated by oxidative stress

Cumulative evidence suggested the involvement of ROS and oxidative stress in biological responses elicited by PM.[24] In addition to that, we observed that both corridor and indoor PM induced oxidative stress response when the cells were exposed to them for 24 hours. Thus, we sought to determine whether the disruption on the endothelial cells barrier was initiated by ROS.

We studied the involvement of ROS in the disruption of endothelial cells barrier by employing N‐acetylcysteine (NAC)—a ROS scavenger. We first introduced the NAC and then followed with exposure to PM. We observed that the NAC significantly rescued the leaky endothelial cells barrier back to the intact endothelial cells barrier (Figure [NaN] A). Consistent to our immunofluorescence data, we observed the significant reduction of the endothelial cells leakiness in the transwell permeability assay (Figure [NaN] B). The NAC addition was found to reduce approximately 70%‐90% of leakiness, bringing back the FITC‐dextran penetration level close to the basal level. This observation suggested that ROS is involved in the disruption of endothelial cells barrier.

Discussion

In view of the cardiopulmonary toxicity posed by PM, we compared the potential of size resolved PM collected from indoor and corridor environments of classrooms to cause oxidative stress pathway and endothelial leakage in human microvascular endothelial cells. Elemental composition and morphology of PM samples corresponding to each size fraction were conserved across the indoor environments of different classrooms. In contrast to PM samples from corridor, those collected from the indoor environments showed relatively higher endotoxin level and abundance of membranous structures in PM4.0‐8.1 and PM2.0‐4.0 fractions. PM induced dose‐dependent increase in cellular responses that are embedded in the oxidative stress pathway such as intracellular ROS, calcium flux, mitochondrial depolarization, and cell death. The oxidative stress‐dependent cytotoxicity was markedly higher in PM collected from indoor environment in comparison with those collected from the corridor. Similarly, PM showed dose‐dependent increase in the induction of endothelial cell leakage. Our data suggested the possible adverse effect of PM originating from indoor environments of classrooms to the endothelial system.

Our studies showed that the abundance of PM in the indoor air was relatively lower which concurred with observations from previously reported studies.[25] Consequently, the calculated indoor–corridor ratio of PM ranged from 0.3 to 0.7 depending on the size fraction. Remarkably, a systematic increase in this ratio was noticed along the decreasing particle size suggesting a rise in the relative abundance of smaller particles in the indoor air. Studies have shown that, the type of ventilation system affects the indoor–outdoor ratio of PM concentration in the air.[13] The indoor environment of the classrooms was mechanically ventilated and the outside air entered the room only through designated inlets. Thus, while the centralized air‐conditioning system helped to reduce the concentration of PM in the indoor environment, our studies suggested that the filtration was more effective for larger particles. The presence of Si, Ca, and Al in larger PM suggested siliceous or clay‐like compositions, possibly originating from walkway and road dust.[26] The source of amorphous PM0.25‐0.5 may be attributed to the fossil fuel combustion (from industrial sources and motor vehicles),[26] while the crystalline PM0.25‐0.5 suggested the formation of secondary crystals from the continuous deposition of salts from the air pollutants.[27]

Cellular models that focus on molecular pathways underlying PM toxicity are valuable tools for identifying mechanisms of toxicity and relative risk assessment of PM. Oxidative stress pathway is one of the well understood cellular mechanisms underlying the pathology of PM [28] . A high content screening strategy that quantitatively evaluates a panel of cytotoxicity responses embedded in oxidative stress pathway such as intracellular ROS, mitochondrial health, calcium flux, and cell membrane damage has been shown useful for the relative risk assessment of size resolved airborne PM.[12] Likewise, we observed the capability of PM samples to induce oxidative stress and cell death in HMVECs. Noticeably, induction of intracellular hydrogen peroxide (H2O2) and mitochondrial superoxide production was more prominent in PM samples collected from the indoor environment in comparison with those collected from the corridor.

The higher redox activity of indoor samples despite lower level of metal elements suggested contribution from other components. PM of size 10 and 2 μm showed the presence of membranous substance, suggesting their biological origin. Previous studies have shown the possibility of endotoxin in biological aerosol produced by occupants.[29] Therefore, we suspected that the higher oxidative stress response in indoor PM is caused by higher level of endotoxin. Our investigations showed relatively higher level of endotoxin in PM samples collected from indoor environments (Figures [NaN] and S3). Among the different size fractions tested, larger PM induced higher cell death which was consistent with the relatively higher endotoxin level in these PM. Additional evidence for the involvement of endotoxin in determining the cytotoxicity of indoor PM was shown by the relative increase in viability of cells exposed to PM samples mixed with Polymyxin B in comparison with cells exposed to PM samples alone (Figure [NaN] C). The endotoxin, also known as lipopolysaccharides (LPS), originates from the outer cell wall of Gram‐negative bacteria (e.g. Escherichia coli) and has been reported to trigger inflammatory responses in lungs.[30] , [31]

One of the main functions of the endothelial cells is to provide barrier in the blood vessels that regulates the solute transfer to and from the blood circulation. Increase in endothelial leakiness has been identified in various lung pathological conditions such as asthma, acute respiratory distress syndrome (ARDS), and acute lung injury (ALI).[32] As such, investigation of endothelial barrier leakiness is pivotal in understanding the possible role of PM in disrupting the integrity of the endothelial barrier. Our studies showed that PM is capable of inducing dose‐ and time‐dependent increase in endothelial barrier leakiness. Activation of inflammatory responses due to the presence of endotoxin in PM could be one of the possible causes of the leakiness observed on the endothelial cells barrier. This is supported by the increased expression of ICAM1 which normally occurs on inflamed vasculature.[33] Elevated expression of these proteins facilitates the leukocyte rolling movement and allows the leukocyte to transmigrate through the leaky intercellular gaps between the endothelial cells.[33] Similarly,[34] reported higher expression of ICAM1 gene after brain endothelial cells were exposed to TiO2 nanoparticles causing leakiness of endothelial cell barrier. The role of ROS as second messengers in the opening of gap junctions cannot be discarded.

ROS which comprises of short‐lived species such as H2O2, O2˙−, and NO have been observed to modulate various intracellular processes, including the opening of intercellular gaps between the endothelial cells leading to leakiness of the endothelial barrier.[35] , [36] , [37] , [38] The involvement of ROS in PM induction of endothelial barrier leakiness was confirmed by adding the NAC, an antioxidant, and a precursor to glutathione‐ an intracellular reducing agent. The cells subjected to NAC showed significant reduction in endothelial barrier leakiness, despite the differences in the intracellular production of H2O2 and O2˙−following the indoor and corridor PM exposure of endothelial cells (Figure [NaN] B). Nevertheless, both the indoor and corridor PM induced similar level of endothelial cell barrier leakiness (Figure [NaN] ), suggesting the involvement of other ROS species.

Our studies showed that corridor PM induced significantly higher amount of intracellular NO when compared to the indoor PM (Figure [NaN] ). The difference in poly aromatic hydrocarbon (PAHs) content in these particles (Table S2) is suspected as the possible cause of difference in inducing NO. It is possible that the increase in intracellular NO production in cells, compensated the lack of H2O2 and O2˙− intracellular production and modulate the endothelial cells leakiness by the corridor PM. Another possibility is the superoxide‐dependent formation of peroxynitrite by indoor PM, which could also induce endothelial cells leakiness.[39] Thus, while different ROS/RNS species seems to modulate the endothelial cells leakiness, these species shared a common target in their intracellular modulation of the leakiness. One of such target is the junction proteins (e.g., VE‐cadherin and PECAM‐1) which are responsible to maintain the integrity of the endothelial barrier. These proteins have been reported to be internalized and degraded following the cue from the intracellular ROS.[37] Further, the possible role of ROS in downregulating the expression of VE‐cadherin and PECAM‐1 genes is evident from the q‐PCR result that showed upregulation of IL‐8, HO‐1 but downregulation of VE‐cadherin and PECAM‐1, after 24‐hours exposure. In addition to junction proteins, ROS modulate the endothelial barrier leakiness through the remodeling of cellular cytoskeleton. For instance, iron oxide nanoparticle has been reported to cause tubulin network remodeling through the Akt pathway in microvascular endothelial cell which in turn leads to leakiness on the endothelial barrier.[40] Similarly, Wang et al. suggested that ROS generated by PM could activate MAPK signaling that phosphorylate HSP27. The phosphorylated HSP27 mediates the synthesis of stress fibers leading to the generation of paracellular gaps.[32]

Conclusion

While the adverse health impact of outdoor air pollution is widely appreciated, biological mechanisms underpinning health consequences from exposure to indoor PM are largely unknown. Our studies showed that, in comparison with PM collected from corridor, those collected from the indoor environments of classroom (student occupied) are more potent in inducing oxidative stress‐mediated cytotoxicity and endothelial cell leakiness in primary endothelial cells. Control experiments suggested endotoxins as a major determinant of the cytotoxicity of indoor PM. While this study adds on to the existing knowledge of biological mechanisms linking the composition of PM to cardiovascular toxicity, further studies are warranted in understanding the source of indoor PM in student occupied classrooms and possible mitigation strategies.

Acknowledgements

SG acknowledges MOF‐RF grant—“National Indoor Environmental Quality (IEQ) Risk Assessment Programme”—MOF‐RF grant N/RPF00/60042/LHAD/LSLB01/N83010/201300/0 for the support of this study. This work is also partially supported by Singapore Ministry of Education Tier 2 grant (R279000418112 to DTL) and World Future Foundation (Award to MIS).

Graph: Mass and morphology of PM samples from corridor and indoor. (A) Mass concentration of PM. Total mass of PM collected onto individual glass slides was determined within 24 h of sampling using a microbalance in a temperature/humidity‐controlled environment. The mass concentration of PM was calculated from the total mass of PM collected and total volume of air sampled during 35 h. (B) Mass proportion was calculated for each PM size range based on the total mass of the particles and mass of PM collected in the specific size range. (C) ESEM images of size resolved PM samples collected onto glass substrate. ESEM images were obtained by directly observing the glass slides containing PM using ESEM without drying and sputter coating. Scale bar=20 μm

Graph: Endotoxin levels in selected PM samples collected from corridor and indoor environments. PM samples collected onto glass slides were retrieved using DMSO , and the stock solutions (1 mg/ mL ) were stored in −20°C. Aliquots of stock solutions required for preparing working concentrations were taken into clean and sterile Eppendorff tubes and were added with equal amount of deionized water. The resulting solution was subjected to freeze‐drying to remove DMSO. PM remaining in the tubes was used for determining endotoxin level using the endotoxin assay kit according to the manufacturer's instructions. Data are means±SD, n=3. Student's t test, P <.05, *significant against the corridor samples

Graph: Size‐ and dose‐dependent cytotoxicity of selected PM samples. (A) PM samples collected onto glass slides were retrieved using DMSO , and the stock solutions (1 mg/ mL ) were stored in −20°C. Aliquot of stock solutions required for preparing the first working concentrations (15 μL) were taken into clean and sterile Eppendorff tubes and were added with equal amount of deionized water. The resulting solution was subjected to freeze‐drying to remove DMSO. PM remaining in the tubes was added with 500 μL Endo GRO ‐ cell culture medium and probe sonicated for 15 s to disperse the particles. The first working concentration of PM samples was subjected to serial twofold dilutions before adding 100 μL to HMVEC cells grown in 96‐well plates. After 24 h of incubation at 37°C, the wells containing PM ‐treated cells were added with 100 μL of resazurin reagent dissolved in the cell culture medium. Fluorescence intensity was read using a plate reader after 1 h of incubation, and the percentage viability of cells was calculated based on reading from negative control. Data are means±SD, n=3. Student's t test, P <.05, *significant against untreated control. (B) HMVEC cells grown in 384‐well plates were subjected to increasing concentrations of particles and were added with wavelength compatible dye cocktails to measure cytotoxicity responses. Cellular images acquired by the high content microscope were analyzed by the software to calculate percentage of cells positive for specific response. The raw data were transformed to obtain strictly standardized mean deviation ( SSMD ) displayed as heat map where the scale of the heat map—green to red—indicates SSMD values of 0‐3. SSMD value greater than three indicates statistically significant difference from the negative control. (C) The freeze‐dried PM samples were added with 10 μg/ mL of Polymyxin B in endothelial growth media and probe sonicated for 15 s to disperse the particles. The cell viability was measured using resazurin reagent, as detailed in 3A

Graph: NO generation in cells exposed to PM samples. Supernatant solution of cells treated with increasing concentrations of PM samples was used for measuring NO. The NO content was measured using Griess Reagent Kit, according to manufactures instruction. The absorbance was measured using a plate reader and compared against the standard reference curve for obtaining μmol/L concentration of NO produced by cells. Data are means±SD, n=3. Student's t test, P <.05, *significant against untreated control

Graph: Corridor/Indoor PM induced leakiness on the endothelial cells barrier. (A) Immunofluorescence staining shows intercellular gaps between endothelial cells (read arrowhead) following PM treatment (30 μg/ mL , 6 h). VE ‐cadherin was visualized with immunofluorescence (green), and the nucleus was stained with DAPI (blue). Scale bar: 50 μm. (B) Quantitative transwell assay shows the dose‐dependent leakiness of the endothelial cells barrier after 6‐h treatment of airborne PM s (3.75, 7.5, 15, and 30 μg/ mL ). Data are means±SD, n=3. Student's t test, P <.05, *significant against untreated control. (C) Quantitative transwell assay shows the time‐dependent leakiness of the endothelial cells barrier after being exposed to 30 μg/ mL of airborne PM s (1, 3, 6, 12, and 24 h). Data are means±SD, n=3. Student's t test, P <.05, *significant against untreated control

Graph: qRT ‐ PCR result on the expression of oxidative stress induced genes expression in HMVEC s subjected to corridor and indoor PM. Effect of PM on the expression of HO ‐1, TNF ‐a, IL ‐8, NF κB1, ICAM , PECAM , and VE ‐Cad in HMVEC s. PM8.1‐20 , PM2.0‐4.0 , and PM0.25‐0.5 from both outdoor and indoor environment were exposed to HMVEC s at 30 μg/ mL for 6 and 24 h. After thorough homogenization of cells, total RNA was isolated using RN easy Mini Kit (Qiagen) to run qRT ‐ PCR using EXPRESS One‐Step SYBR® Green ER™ Kits (Life Technology). Relative Quantification values are normalized to the control without PM treatment. *Statistically significant from negative control group P <.05

Graph: The endothelial cells barrier leakiness is modulated by reactive oxygen species ( ROS ). (A) Immunofluorescence staining shows fewer occurrences of the intercellular gaps between the endothelial cells with the addition of 1 mmol/L NAC. VE ‐cadherin was visualized with immunofluorescence (green), and the nucleus was stained with DAPI (blue). Scale bar 50 μm. (B) Quantitative transwell assay measured significant reduction of leakiness on the endothelial cells barrier when NAC was. Data are means±SD, n=3. Student's t test, P <.05, #significant against cells without NAC treatment

Graph

By M. L. Chua; M. I. Setyawati; H. Li; C. H. Y. Fang; S. Gurusamy; F. T. L. Teoh; D. T. Leong and S. George