**1. Introduction**

Exposure to indoor pollutants are strongly associated with increased morbidity and mortality, mainly among school children who spend most of their time in the classrooms [1]. Reports have shown that exposure to high concentration of particles, nitrogen dioxide (NO2), carbon dioxide (CO2), ozone (O3), volatile organic compounds (VOCs), and fibres induce persistent airway inflammation, which is mediated by the immune system [2,3]. Biomarkers, such as fractional exhaled nitric oxide (FeNO), cytokines, chemokines, lipid mediators, enzymes, adhesion molecules, and other growth factors, have been considered as the indicators of allergic airway inflammation [4]. The indicators that underlie

the complex molecular pathways that regulate inflammation have not been fully elucidated; however, numerous studies have reported that the process certainly involves the activation of eosinophils and neutrophils [5,6]. They play roles in innate host defence via effector mechanisms, including degranulation, DNA traps, and cytolysis, following the activation cascaded by diverse mediators [7].

Extensive investigation of the mediators that are implicated in allergy, lung inflammation, and asthma has been documented but there was insufficient evidence available for the multi-dimensional characters of the activation and degranulation markers expression of CD11b, CD35, CD63, and CD66b on eosinophils and neutrophils. Activation markers, such as integrin Mac-1 (CD11b/CD18), L-selectin (CD62L), CBRM1/5, ICAM-1 (CD54), PD-L1 (CD274), PSGL-1 (CD162), FcγRII (CD32), CD16, CD44, and CD69, were shown to be upregulated on circulating or sputum granulocytes from asthma patients [8,9]. Response to allergens and environmental stimuli among asthma patients had a detrimental outcome and was characterized by the infiltration of different granule contents from eosinophils and neutrophils following activation [10]. Eosinophils and neutrophils have four distinct granules, namely azurophilic, secondary, tertiary, and secretary, which are formed throughout development in the bone marrow [11]. Several models have demonstrated the link between degranulation of these granules and pollutants exposure at different levels of stimuli with the expression of CD11c, CD63, and CD66b [12,13]. At present, biomarkers help tailor the managemen<sup>t</sup> of respiratory illness and studies have highlighted precision therapy approaches based on disease mechanisms by targeting the cytokines and chemokines [14,15]. Recently, advanced statistical analysis approaches, such as hierarchical clustering and latent class analyses, have renewed the interest of researchers in the investigation of airway translational inflammation phenotypes [16,17]. This new vision helps the researchers to better understand and precisely describe the inflammatory phenotypes [18,19].

This study aimed to investigate the concentrations and sources of indoor pollutants in classrooms located in the urban and suburban areas of Hulu Langat, Selangor, Malaysia. Moreover, it aimed to predict the toxicodynamic effects of indoor pollutants in the classrooms towards marker expression on the eosinophils and neutrophils in sputum samples by using chemometric analysis techniques.

#### **2. Materials and Methods**

#### *2.1. Study Population*

The study population was randomly sampled from eight secondary schools in Hulu Langat, Selangor, Malaysia. The researchers of this study targeted school children at the age of 14 of which they were randomly selected from four classrooms in each school. The total number of students who received their guardian's consent and was thus recruited in the study was 470. Among them, only 50 (10.6%) students were diagnosed with asthma by a doctor based on the survey questionnaire. Another 46 students out of the remaining 420 students were randomly selected as a potential control group. The control group was selected among students who produced an adequate sputum cell count from the same class and school. In total, 96 students were included in the final study group. Students with a history of smoking in the last 12 months and students who received antibiotic treatments in the past four weeks were excluded from this study. The school areas were classified as urban and suburban by the Ministry of Education, Malaysia, based on the locale classification of the ecological measures. The data collection from the clinical assessments and indoor air monitoring were carried out at the same time in August until November 2018 and February 2019.

The researchers used the questionnaire adopted from the European Community Respiratory Health Survey (ECRHS) and International Study of Asthma and Allergies in Childhood (ISAAC) that are inclusive of questions on doctor-diagnosed asthma, allergies, and respiratory symptoms. The self-administered questionnaire was distributed to the selected students in the same week of the technical measurements. Subsequently, the researchers went through the questionnaires during a face-to-face interview with the students to clarify any uncertainty. The doctor-diagnosed asthma in this study is defined as having asthma medication, asthma attacks, and wheezes with breathlessness

in the last 12 months, which were diagnosed by the physician [20,21]. This information was verified during face-to-face interviews and telephone calls with the students' respective guardians.

#### *2.2. FeNO Assessment and Allergy Skin Test*

FeNO measurement was performed by a chemiluminescence analyser (NIOX VERO, Circassia, Sweden) as recommended by the American Thoracic Society/European Respiratory Society (ATS/ERS) [22]. The detection limits and accuracy for this device are 5–300 ppb and ±5 ppb, respectively. The students were asked to inhale deeply through the mouthpiece attached to the patient filter and then slowly exhale for about six to ten seconds at a constant flow rate (50 mL/s)—the single-breath technique. This process was repeated at least twice to ge<sup>t</sup> an average value. Students were instructed to refrain from eating and drinking for one hour before the FeNO assessment.

An allergy skin prick test was performed on the volar side of the forearm alongside *Dermatophagoides pteronyssinus* (Derp1) (house dust mite), *Dermatophagoides farina* (Derf1) (house dust mite), *Cladosporium herbarium* (fungi), *Alternaria alternate*(fungi), and *Felis domesticus*(cat) allergens in liquid form (Prick-Test Diagnostic, ALK-Abelló, Madrid, Spain). The same amount of histamine (10 mg/mL) and normal saline were used as the positive and negative controls, respectively. The reaction was measured after 15 min of which the wheal diameter was recorded. The allergen's wheal diameter of 3 mm was considered as a positive control. Atopy was defined as a significant positive skin test reaction to at least one of the applied allergens [23].

#### *2.3. Sputum Induction and Processing*

Sputum induction was conducted by the inhalation of a nebulised, sterile mixture of 4.5% sodium chloride (hypertonic) and salbutamol 200 μg, followed by the coughing and expectoration of airway secretions. For nebulisation, an ultrasound nebuliser (Model CUN60 Citizen System Japan Co. Ltd., Tokyo, Japan) was was used as recommended [24] with a mouthpiece that fitted an output of ∼1 mL·min−<sup>1</sup> to achieve successful sampling. The induced sputum samples collected from the respondents were kept in an icebox and further processed within two hours by using flow cytometry. The method of processing sputum samples has been previously described [25]. In short, the sputum sample was diluted with freshly prepared phosphate buffer saline and gently vortexed at room temperature for homogenisation. These steps were repeated thrice. Subsequently, the sputum samples were centrifuged at 800× *g* for 10 min. Next, cytospin slides of sputum cells were stained with May–Grunwald–Giemsa for the cell differentiation count. Samples with >80% of squamous epithelial cells were excluded for the flow cytometry analysis.

#### *2.4. Flow Cytometry*

The processed sputum sample at a concentration of 1 × 10<sup>6</sup> cells in 100 μL was resuspended in 1 mL of stain buffer (FBS) (BD PharmingenTM, San Diego, CA, USA) and washed twice in cold stain buffer by centrifugation at 200× *g* for 5 min. Subsequently, the supernatant was removed, and the cell pellet was resuspended in 300 μL of stain buffer. A volume of 100 μL aliquots of the cell suspension were transferred into three different sterile polypropylene round-bottom tubes, and monoclonal antibodies and isotype controls were added to the cells according to the manufacturer protocol. The cells were immunostained with antibodies for neutrophil and eosinophil surface markers of PE-Cy 7 Mouse Anti-Human CD11b/Mac-1, BB515 Mouse Anti-Human CD35, PE Mouse Anti-Human CD63, APC-H7 Mouse Anti-Human CD16, PerCP-Cy 5.5 CD41a, and Alexa Fluor 647 Mouse Anti-Human CD66b purchased from BD Biosciences, US. The isotype controls of PE-Cy 7 Mouse IgG1, BB515 Mouse IgG1 K, PE Mouse IgG1 K, APC-H7 Mouse IgG1 K, PerCP-Cy 5.5 Mouse IgG1 K, and Alexa Fluor 647 Mouse IgM K (BD Biosciences, US) were added to the cells. The cell samples were incubated for 15 min at room temperature in the dark. Next, the cell samples were washed twice in 1 mL of stain buffer and centrifuged at 300× *g* for 5 min. The supernatant was removed, and the cells were loosened up by tapping the tube. Subsequently, the cells were carefully resuspended in 500 μL of the stain buffer. Samples without antibodies and isotype were used as the controls. Hereafter, the cells were analysed by using the BD FACSCanto II flow cytometry instrument (BD Bioscience, US) equipped with blue, red, and violet lasers. Compensation was set to account for spectral overlap between the four fluorescent channels. The gating region was set so that less than 1% of the samples were stained with negative controls. Data were processed by using FlowJo software (Version 10.1r1) [26,27].

The activation of eosinophils was measured by using CD11b (integrin αM), CD35 (CR1), and CD66b (CEACAM-8). Meanwhile, CD11b was used as an activation marker for neutrophils. The degranulation activity was measured by the expression of CD11b and CD63 for eosinophils as a marker for cyctalloid (specific/secondary) and secretory (sombrero) vesicles, respectively. For neutrophils, the CD11b, CD35, CD63, and CD66b markers were used as tertiary, secretory, azurophilic, and specific granule expression markers in the degranulation activities [28]. Isotype controls were used as the positive control and to address the background produced by non-specific antibody binding, whereas the Anti-Mouse Ig K Negative Control Compensation Particle Set was used to optimise the fluorescence compensation settings [29] (Figure 1).

**Figure 1.** Gating strategy to identify activated sputum granulocytes. ( **A**) A forward-/side-scatter (FSC/SSC) gate to identify the granulocytes. (**B**) The activated granulocytes were based on CD11b+. (**C**) CD41a was gated with SSC to confirm that the activated granulocytes were neutrophils and eosinophils. ( **D**) Subsequently, the eosinophils (blue) and neutrophils (green) were separated based on the expression of CD16− and CD16+ on SSC, respectively. The activation markers of CD11b, CD35, CD63, and CD66b for eosinophils (**E**,**F**) and neutrophils ( **G**,**H**) were based on the negative gates of the isotype control for all antibodies.

#### *2.5. Assessment of Indoor Air Quality in Classrooms and Building Inspection*

Indoor pollutants and physical parameters, including temperature (◦C), relative humidity (%), and carbon dioxide (ppm), were measured in the classrooms during learning session within an hour by using a Q-TrakTM IAQ monitor (Model 7565 TSI Incorporated, Shoreview, MN, USA) with the average log interval values over one minute. The accuracy of this device on temperature, relative humidity, and CO2 are ±0.6 ◦C, ±3%, and ±50 ppm, respectively. The sampling for the particles was measured by using a Dust-Trak monitor (Model 8532 TSI Incorporated, Shoreview, MN, USA) at a sampling rate of 1.7 L/min with a resolution of 0.001 mg/m<sup>3</sup> and detection limit of 0.001–150 mg/m3. The PPM FormaldemeterTM htV-M (PPM Technology Ltd, Wales, UK) with accuracy of 10% at 2 ppm was used to measure the concentration of formaldehyde. In each school, a total of four hours of measurements

were taken from four randomly selected classrooms for a period of an hour each during the learning session, as has been previously described [30–32]. The instruments used were placed one metre from the ground in the centre of the classrooms. All instruments used were calibrated regularly. The NO2 (μg/m3) concentration was measured by using a diffusion sampler (IVL, Goteborg, Sweden) for a period of a week. The sampler was place at height of approximately 2–3 m above the ground and returned to the IVL Swedish Environmental Research Institute Laboratory (Goteborg, Sweden), an accredited laboratory for further analysis. This measurement technique provides an average concentration of NO2 in the air during a week, with a limit of detection (LOD) of 0.5 μg/m<sup>3</sup> and a 10% (at the 95% confidence level) measurement uncertainty [33]. A building inspection was carried out before the indoor air quality assessments were obtained. Details on the building information, floor furnish, furniture, and type of ventilation system were noted [34].

#### *2.6. Ethical Statement*

The Ethics Committee for Research Involving Human Subjects Universiti Putra Malaysia (JKEUPM) has approved this study (JKEUPM-2018-189) and each of the students was given a written consent form for their guardian's approval.

#### *2.7. Data Analysis*

The descriptive analysis was carried out by Mann–Whitney tests using the Statistical Package for the Social Sciences (SPSS) 25.0. The differences in biomarker expression between the doctor-diagnosed asthmatic children and healthy children were made by using GraphPad Prism 8 for Windows. Subsequently, a principle component analysis (PCA) and agglomerative hierarchical clustering (AHC) were applied to explore the association and pattern recognition between the biomarker expression and the concentrations of indoor pollutants. The final prediction models were generated by logistic regression analysis in which the models' performance was based on the coefficient of determinant, overall accuracy, sensitivity, specificity, and the area under the curve (AUC) of the receiver operator characteristics (ROC) [35]. The researchers used the standardized data of the indoor pollutants and biomarkers in the chemometric and regression analyses. The multivariate analysis was carried out by using the Statistical Package XLSTAT Evaluation 2019.2.3 (Addinsoft, New York, US).
