Particulate matter 2.5 induced inflammatory responses and metabolic alterations in human monocytes Authors: Naunpun Sangphech, Tassanee Chetwittayachan, Ganjana Lertmemongkolchai, Ian Mudway, Sakchai Laksee, Jomkhwan Meerak, Pongphan Pongpanitanont, Tanapat Palaga Published: 2026 ABSTRACT AbstractFine particulate matter 2.5 (PM2.5) is linked to rising cases of respiratory, cardiovascular, and immune-mediated diseases and increased hospitalizations in non-respiratory patients. Chiang Mai Province in Northern Thailand, faces seasonal severe PM2.5 pollution, but its impact on immune responses in human remains poorly characterized. This study aimed to investigate the association between air pollution and non-communicable diseases in Chiang Mai and to assess how PM2.5 collected in Chiang Mai (CM-PM2.5) during the peak of poor air quality in 2020 affected the innate immune responses of primary human monocytes. PM2.5, PM10, NO2, and O3 air quality index (AQI) levels concurrently increased throughout the year. Notably, PM2.5 and NO2 AQI levels were positively correlated with the incidence of chronic obstructive pulmonary disease (COPD) in Chiang Mai. Transcriptomic profiling of primary human monocytes stimulated with CM-PM2.5 (5 and 20 μg/ml) for 24 h indicated a dose-dependent effect on gene expression profiles. Alterations in pyrimidine ribonucleotide metabolism were observed at both CM-PM2.5 concentrations. Upregulation of key metabolic genes (CYP1B1, HK2, and XYLT1) and suppression of glycolytic activity were observed with treatment of 20 μg/ml of CM-PM2.5. This study provides strong evidence that exposure to PM2.5 disrupts metabolism and alters responses of human innate immune cells which may contribute to increased severity of respiratory diseases.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-57120-1.Keywords: PM2.5, Human monocytes, Metabolism, COPDSubject terms: Diseases, Immunology, Medical research Introduction Particulate matter (PM) is one of the six major air pollutants, along with carbon monoxide, lead, ground-level ozone (O3), nitrogen dioxide (NO2), and sulfur dioxide (SO2)1. A decrease in PM size increases its surface area for carrying toxic substances, leading to a greater health impact2. Fine PM, especially PM2.5, is widely recognized as a leading environmental risk factor for non-communicable diseases3. PM2.5 levels above acceptable limit are associated with increased risk of multiple diseases2. World Health Organization (WHO) redefined the annual average outdoor concentrations of PM2.5 to 5 μg/m3 in 2021. The recommended concentration for short-term (24 h) exposure is not higher than 15 µg/m34. Thailand is one of many countries that is consistently facing severe air pollution in the past decade. Thai annual average outdoor concentrations of PM2.5 are set, according to the interim target 3 of the WHO, which is 15 μg/m3, and 24 h exposure is 37.5 μg/m3. The current situation suggests an increasing burden of PM-related health problems in Thailand in the coming decade. Inhalation transports PM2.5 from upper respiratory tract to the alveoli. It deposited in different regions of the respiratory system, but mainly at the bronchiole and the alveoli. Based on the International Commission on Radiological Protection (ICRP) model, the percentage of deposition rate was about 10% with the calculated mass deposition rate of 0.42 μg⋅min− 1⋅ml− 1 at the alveoli5. The deposition rate gradually increased along with the PM2.5 concentrations. PM2.5 can carry a variety of toxic substances, triggering cellular response that is roughly categorized into two mechanisms. First, PM2.5 particles are cleared by alveolar macrophages via phagocytosis. It induces alveolar macrophage death and the release of IL-1α, leading to lung damage6,7. This process leads to recruitment of circulating monocytes to compensate the lost function8. Another mechanism is driven by variation in PM2.5 composition. Chemical components such as transition metals (Fe and Cu), induce reactive oxygen species (ROS) production, which is associated with several respiratory diseases9. Biological components, for example, lipopolysaccharides (LPS), engage pattern recognition receptors such as toll-like receptor 4 (TLR4), leading to production of proinflammatory cytokines and oxidative stress in lung resident macrophages. A recent report indicated that TNFα production in response to insoluble PM2.5 fractions, such as polycyclic aromatic hydrocarbons (PAHs), heavy metals, and microbial components, especially LPS, could be used as an indicator of air quality10,11,9. Exposure to PM2.5 has been shown to not only induce inflammation but also alter cellular metabolic status. PM2.5-treated macrophages/microglia cells were reported to increase glutamate, which is neurotoxic and implicated in brain damage12. Serum metabolites from lipid, glucose, and amino acid metabolism were altered during PM2.5 exposure, similar to several diseases that showed alteration in the metabolic profiles. It suggested that alteration in cellular metabolism during PM2.5 exposure may contribute to disease risk13,14. Impaired glucose and insulin tolerance were found in the adipose tissue of mice exposed to PM2.5 which was reversible after PM2.5 removal15. The increase in PM2.5 level was associated with metabolic syndromes such as cardiovascular disease and diabetes which have been partially attributed to PM2.5 exposure16–18. Human activity, geography, and climate contributed to the variety of PM2.5 composition. For example, SO42−, NO3−, and K+ are highly abundant in PM2.5 from the northern part of Thailand, mainly from biomass burning during the season19. PM2.5 from heavily industrialized countries tended to decrease in SO42− but increase in NO3− proportion20,21. Geography and climate are the main factors that generate biodiversity in PM2.5. Previous reports identified diverse biomaterials and microorganisms in PM2.5 such as pollen, bacteria, fungi, and viruses22. It is demonstrated that each PM2.5 is composed of a unique composition that represents its origin. Therefore, the cellular response to PM2.5 shares some common phenotype and has a unique response. Chiang-Mai is the largest city in northern Thailand with intense human activity. It experiences serious seasonal PM2.5 pollution with the hazardous level of PM2.5 for more than a decade, especially during April to May due to biomass burning (forest fires (89%), 5.4% from solid waste burning and 2.3% from agriculture residue field burning)23. The rising numbers of respiratory diseases and lung cancer have been continuously reported in relation to PM2.5 level in the region although the underlying cellular mechanisms remain poorly understood24. This study aims to examine the effect of PM2.5 collected from Chiang-Mai, Thailand (CM-PM2.5) on the primary human monocytes. We use RNA sequencing techniques to obtain an overall profile of CM-PM2.5 impact on the primary human monocytes. The cellular metabolic changes were investigated to understand the underlying mechanism that may help in part in explaining the health impact of PM2.5. Materials and methods Chiang-Mai PM2.5, PM10, NO2, and O3 AQI data These data were obtained from the Division of Air Quality Data, Air Quality and Noise Management Bureau, Pollution Control Department (https://aqicn.org/city/chiang-mai/). PM2.5 AQI data ranged from 2016 to 2024. PM10 AQI data ranged from 2014 to 2023. NO2 AQI data ranged from 2014 to 2022. O3 AQI data ranged from 2014 to May 2022. These data are in the Supplementary File 1. PM2.5 collection PM2.5 samples were collected from Chiang-Mai, Thailand during February to March 2020 by the sets of Tisch model TE-WILBUR-2.5 Low Volume Federal Reference Method (FRM) Ambient Air Particulate Samplers (Tisch Environmental) with the adjusted average flow rate at 16.7 L/min together with the PTFE (Polytetrafluoroethylene) filters with 46.2 mm diameter and 2 µm pore size. The masses of PM2.5 samples were determined by gravimetric method using a 7-digits place ultra-microbalance (Mettler Toledo: METLER UMX 2). PM2.5 extraction CM-PM2.5 was extracted using methanol as described by Kewcharoenwong et al.25. Three milliliters of HPLC-grade methanol were added to the filter in a 50 mL centrifuge tube. The tube was vortexed for 1 min and sonicated for 10 min. The supernatant was then transferred to a new tube. This extraction step was repeated two additional times using fresh methanol. The ultrasonic bath water was regularly replaced to control the consistent temperature. The filter and the combined methanol extracts were dried under nitrogen at 35 °C. The filter was reweighed as described previously to determine the amount of CM-PM2.5 extracted. The extracted CM-PM2.5 was resuspended in Milli-Q (Merck Millipore) purified water (resistivity ≥ 18.2 MΩ·cm) at a concentration of 1 mg/mL. The CM-PM2.5 solution was pretreated with Chelex-100 resin to reduce background metal contamination. The CM-PM2.5 solution was stored at − 80 °C until use. CM-PM2.5 elemental analysis Trace element analysis of PM samples was conducted using microwave-assisted acid digestion followed by inductively coupled plasma mass spectrometry (ICP-MS). CM-PM2.5 was initially dispersed in deionized water at a concentration of 1 mg/mL, following the procedure reported by Kewcharoenwong et al.25. For elemental extraction, samples were digested with a mixture of 10 mL of 65% nitric acid (HNO₃) and 2 mL of 48% hydrofluoric acid (HF) using a microwave digestion system (Berghof, SW-X). Nitric acid was employed for its strong oxidizing capability, while HF facilitated the dissolution of silicate-based components. After digestion, the solutions were filtered and diluted to a final volume of 50 mL with ultrapure water. All reagents were of analytical grade (HNO₃, Sigma-Aldrich; HF, Merck Millipore), and all solutions were prepared using ultrapure water. Samples and standards were stored in polypropylene tubes at 4 °C prior to analysis. To minimize contamination, all glassware was precleaned by soaking in 10% HNO₃ for at least 24 h and thoroughly rinsed with ultrapure water. Sample preparation was performed under a laminar flow hood. Elemental concentrations were determined using ICP-MS (Thermo Scientific, iCAP Q) operated with the manufacturer’s software. Instrumental and matrix effects were corrected using internal standardization with Indium (In, Fluka Analytical, TraceCERT) at 100 µg/L, and signal variations were found to be negligible. Procedural blanks, including reagent and filter blanks, were analyzed to assess background levels, and appropriate corrections were applied. The average daily inhalation dose calculation The ADD was calculated according to the equation described in Morakinyo et al.26 as follow: whereas C is the amount of PM2.5 in ambient air (µg/m3), whose data was acquired by Air Quality Index (AQI). InhR is the inhalation rate (m3/day). EF is the exposure frequency (days/year). ED is the exposure duration (days). InhR, EF, and ED values were followed by Morakinyo et al.26. Cell culture and primary human monocytes Healthy donor blood use was ethically approved by the Human Research Ethics Committee of Walailak University (WUEC-21-211-01 and WUEC-20-126-01) and Research Ethics Committee, National Blood Centre, Thai Red Cross Society (IRB No. 17/2563). Healthy donor blood obtained from National Blood Centre, Thai Red Cross Society. Written informed consent was obtained from all donors prior to blood donation. Samples were fully anonymized before being used for this study, and no donor-identifying information was available to the research team. All methods were carried out in accordance with relevant guidelines and regulations issued by Human Research Ethics Committee of Walailak University and Research Ethics Committee, National Blood Centre, Thai Red Cross Society. Peripheral blood mononuclear cells (PBMCs) were isolated from the buffy coat or whole blood of healthy donors by Ficoll gradient centrifugation (GE Healthcare, UK). PBMCs was used as sources for CD14 + monocytes separation by human CD14 microbeads (MACS Miltenyi Biotec, Germany). CD14 + monocytes were maintained in RPMI-1640 supplemented with 5% human serum (Sigma, USA) and antibiotics plus antimycotics (Gibco, USA). Cell viability In this study, all biological assays were conducted using the same batch of methanol-extracted CM-PM2.5, as described in the previous section. Whole (unprocessed) CM-PM2.5 particles were not directly used. CD14 + monocytes (1 × 104 cells per well, 100 µL, in a 96-well plate) were treated with CM-PM2.5 at concentrations of 0, 5, 10, and 20 µg/ml for 20 h. Ten microliters of 5 mg/ml of MTT solution were added. Cells were incubated at 37 °C in CO2 incubator for 4 h. The formazan crystal was dissolved at 0.04 N HCl in isopropanol and measured at OD540. The percentage of cell viability was calculated according to the equation Real-time PCR After the cell treatment as indicated in each experiment, total RNA was extracted using a RNeasy Micro Kit (QIAGEN, Germany). cDNA was synthesized using SuperScript™ III First-Strand Synthesis SuperMix (Invitrogen™, USA). qPCR was performed using iQTMSYBR®Green SuperMix (Bio-Rad, USA) following the manufacturer’s protocol in a StepOnePlus Real-Time PCR System (Thermofisher, USA). Forward_CYP1B1: 5’-AAGTTCTTGAGGCACTGCGA-3’, Reverse_CYP1B1: 5’-GTGATAGTGGCCGGTACGTT-3’, Forward_HK2: 5’-ATTGTCCAGTGCATCGCGGA-3’, Reverse_HK2 5’-AGGTCAAACTCCTCTCGCCG-3’, Reverse_HK2 5’-AGGTCAAACTCCTCTCGCCG3’27, Forward_ XYLT1: 5’ TACTGCCGCCACAAGTTAGG-3’ and Reverse_XYLT1: 5’-CCCACTGCACGTTCTTGTTG-3’. Forward_ β-ACTIN: 5’ACCAACTGGGACGACATGGAGAA-3’ and Reverse_β-ACTIN: 5’GTGGTGGTGAAGCTGTAGCC-3’28. The β-ACTIN mRNA level was used as a house keeping gene to calculate relative mRNA expression by 2-∆∆CT method as previously described by Livak and Schmittgen29. Library preparation and RNA sequencing CD14 + human monocytes (5 × 105 cells per well, 500 µL, in a 24-well plate) from 5 healthy individuals were stimulated with PM2.5 (5 and 20 µg/ml) for 24 h. Cells cultured in culture medium alone were used as unstimulated control. RNA was collected and extracted using an RNeasy® Mini kit (QIAGEN, Germany). Total RNA was sent to perform RNA sequencing at the Omics Sciences & Bioinformatics Center, Chulalongkorn University, Thailand. In short, the quality and quantity of sample were determined by a Bioanalyzer (Aligent2100 Bioanalyzer System, Canada). The sample library was prepared using a TruSeq mRNA library kit (Illumina, USA). RNA sequencing was performed using NextSeq500 (Illumina). Transcriptomic analysis and visualization RNA sequencing raw data (GSE237963) were submitted to Galaxy (https://usegalaxy.org) to analyze the differential expressed gene. In brief, the raw RNA sequencing datas (fastq files) were quality assessed using fastqc. Subsequently, the transcripts were trimmed to remove adapters using Trimmomatic in paired-end mode (two separate input files). The sliding window algorithm was selected for trimming, with a window size (number of bases to average across) of 4 and a required average quality score of 20. Transcripts were then quantified using Sailfish, using the GRCh38 transcriptome as the reference database. The library type was specified as inward-oriented paired-end reads (I), with unstranded sequencing (U). The input from Trimmomatic did not include a transcript-to-gene mapping file. Sequence-specific bias correction and fragment GC bias correction were not applied, and equivalence class counts generated during quasi-mapping were not retained. The down-sampling factor for GC content and fragment length probability mass function (PMF) estimation was set to 1. Strict intersection mode was not enforced. The estimated fragment length had a mean of 200 bp with a standard deviation of 80 bp. The maximum number of mappings per read was set to 200, and effective length correction was enabled. The Variational Bayesian EM algorithm was not used for optimization. Orphaned reads were not considered as valid alignments, and the unsmoothed approach to effective length correction was not applied. The maximum fragment length considered when constructing the empirical fragment length distribution was 1000 bp. Strand-aware processing was disabled, and strict library compatibility was not enforced. Dovetailing of paired-end reads was not allowed. One million fragment mappings were used to learn the sequence-specific bias model, and 10,000 uniquely aligned fragments were sampled to estimate the fragment length distribution. No Gibbs sampling or bootstrap sampling was performed. Finally, the analysis of differentially expressed genes was conducted using DESeq2. It was considered significant when the false discovery rate (FDR) was less than 0.05. A heatmap was generated using MORPHEUS (https://software.broadinstitute.org/datamorpheus), and differentially expressed genes (DEGs) were hierarchically clustered using the average linkage of one minus Pearson correlation metric. A Venn diagram was generated using Venny2.1 (http://bioinfogp.cnb.csic.es/tools/venny). Gene ontology (GO), pathway enrichment and hallmark analysis The DEGs were submitted to ShinyGO 0.77 website30. Gene identifiers were automatically converted to Ensembl gene IDs or STRING database protein IDs, and Homo sapiens was selected as the reference species using a protein-coding genome background. For GO biological process analysis, the FDR cutoff was set at < 0.05. Pathway size thresholds were set to a minimum of 2 genes and a maximum of 2000 genes. Enrichment p-values were calculated using the hypergeometric probability mass function and adjusted for multiple comparisons using the Bonferroni correction, with adjusted p values < 0.05 considered statistically significant. Redundant GO terms were removed based on similarity filtering. Genes associated with the same GO terms were grouped using hierarchical clustering and visualized as a tree plot, where dot size corresponds to statistical significance. The interactive network plot illustrates the relationships among GO terms, in which two nodes are connected if they share ≥ 20% of their genes. Darker-colored nodes indicate more significantly enriched gene sets. Node size represents the number of genes within each GO term, and edge thickness reflects the number of overlapping genes. Pathway enrichment analysis was performed using the PANTHER database via ShinyGO with the same settings described above, except that the PANTHER database was selected. For hallmark analysis, the hallmark gene sets from the MSigDB were used. The FDR cutoff was set at < 0.05, and pathway size thresholds ranged from a minimum of 2 genes to a maximum of 5000 genes. Data of PM2.5-related diseases in Thailand The data on PM2.5-related diseases in Thailand were obtained from the Division of Non-Communicable Diseases, Department of Disease Control, Ministry of Public Health in Thailand (https://www.ddc.moph.go.th/dncd/news.php?news=39911). This data included the disease incidence of asthma patients per 100,000 people in 2015, and the mortality rates of COPD, ischemic heart disease, cerebrovascular diseases, and hypertension per 100,000 people in 2021 (Supplementary File 2). The data were plotted according to the provinces of Thailand. The incidence of new cases of COPD (Supplementary File 3) and the number of COPD in various aged range (Supplementary File 4) were obtained from Chiang Mai Provincial Public Health Office (https://cmi.hdc.moph.go.th/hdc/main/index.php). The map was created using R software. The provincial boundary shapefile for Thailand was obtained through the rnaturalearth package version 1.2.0 using the ne_states(country = “Thailand”, returnclass = “sf”) function, which retrieves administrative boundary data from the Natural Earth public-domain geospatial dataset (https://www.naturalearthdata.com/) . The map was then generated using the sf version 1.1.1 and ggplot2 version 4.0.3 packages in R. The disease data were merged with the provincial boundary data by province name, and the spatial distribution was visualized using geom_sf(). Glycolytic rate assay CD14 + human monocytes (1 × 105 the cells) were stimulated with PM2.5 for 24 h at 37 °C in the 5% CO2 incubator. Before assay, cells were washed twice with PBS and incubated in non-CO2 incubator at 37 °C without CO2 for 1 h. Rotenone and antimycin A (1 μM), and 2-deoxyglucose (50 mM) were sequentially injected to measure glycolytic rate by XFe96 analyzer. The remaining cells in the well were stained with trypan blue and counted under a microscope at the end of the experiments. ECAR and OCR at basal glycolysis and compensatory glycolysis were calculated. The results were normalized with the cell number. Statistical analysis All statistical analyses were determined using GraphPad Prism software, except the RNASeq analysis. Statistical significance was determined using one-way ANOVA. A significant level was determined when the p value was less than 0.05. Results Analysis of particulate matter (PM2.5 and PM10), NO2, and O3 in Chiang-Mai, Thailand from 2014–2024. To obtain an overview of the air pollution situation in Chiang-Mai, PM2.5, PM10, NO2, and O3 AQI data were obtained from the Division of Air Quality Data, Air Quality and Noise Management Bureau, Pollution Control Department (https://aqicn.org/city/chiang-mai/). The daily PM2.5, PM10, NO2, and O3 AQI increased concurrently from January to May (Fig. 1A–D). PM2.5 AQI exceeded the daily standard (15 µg/m3/24 h) from December to May and peaked in March (Fig. 1A). PM2.5 pollution persisted for nearly half of the year in Chiang-Mai, contributing to PM2.5 related diseases and premature deaths31,2,3,32. PM10 AQI was also increased during the same period (Fig. 1B) but the toxicity was less than PM2.5, in agreement with the previous study33. NO2 levels also increased during this period and tended to decrease from 2014 to 2022, although the cause of this trend remains unclear (Fig. 1C). O3 AQI levels during January to May 2014–2016 were higher than the healthy level (Fig. 1D), which was reported to be associated with forest fires and agricultural waste burning across the border of the Thailand-Myanmar-South China34. These data support evidence that PM2.5, NO2, and O3 are indicators of air pollution because they increased in a coordinated manner. The level of these parameters were not significantly affected by the COVID-19 pandemic. One possible explanation is that their primary source was the intense forest fires, while pollution from transportation may have contributed minimally. Additionally, studying the toxicity of PM2.5 and its health effects is crucial for understanding the unique impact of different PM2.5 sources. Chiang-Mai AQI data of PM2.5, NO2, PM10, and O3. The daily AQI data of (A) PM2.5 (from 2016 to 2014) (B) NO2 (from 2014 to 2022), (C) PM10 (from 2014 to 2023), and (D) 8 h average of O3 (from 2014 to 2022) from the Division of Air Quality Data, Air Quality and Noise Management Bureau, Pollution Control Department (https://aqicn.org/city/chiang-mai/) were plotted in scatter plot. The X-axis represents the date, the left Y-axis shows AQI, and the right Y-axis indicates the corresponding pollutant concentrations. AQI values (primary y-axis) were converted and plotted on the secondary y-axis for comparison (toxicity does not scale linearly with concentration). The red line marks the recommended levels from the WHO. For PM2.5, 24 h exposure is 15 µg/m3. For PM10, 24 h exposure is 45 µg/m3. For NO2, 25 µg/m3 at 25 °C and 1 atmosphere equals 0.013 ppb, according to the UK Air Pollution Information System (APIS). For O3, an 8-h average concentration of 100 µg/m3 at 25 °C and 1 atmosphere equals 0.051 ppb. Chemical composition analysis of PM2.5 by inductively coupled plasma-mass spectrometry (ICP-MS) In this study, the PM samples were collected from the area in Chiang-Mai Province between February–March of 2020. CM-PM2.5 was extracted using a methanol extraction method prior to elemental analysis. This approach has been reported to efficiently extract inorganic and organic constituents35. CM-PM2.5 was selected for this study for cellular effects due to the annual critical situation of PM2.5 air pollution in that area36. Detailed information regarding the periods and the methods of collection are described in the Materials and Methods section. We measured elements in solutions that would later be used for cell stimulation, which represented the elements to which the cells were exposed. Elemental composition was analyzed by ICP-MS (Table 1). The top five elemental proportions in CM-PM2.5 were sodium (Na), potassium (K), aluminium (Al), magnesium (Mg), and calcium (Ca). The primary sources of PM2.5 in Chiang-Mai during haze days were biomass burning, followed by secondary sulfate, dust, and traffic37. K was a tracer for fresh biomass and may therefore reflect the source of collected PM during the severe forest fire in April–May 202138,39,21. Other components were not analyzed due to limited sample availability. Table 1Element concentrations in CM-PM2.5 from various sources determined via ICP-MS.ElementThe concentration of elements in PM2.5 (ppm) ± SDElementThe concentration of elements in PM2.5 (ppm) ± SDNa215.5224 ± 1.0365Cr0.0817 ± 0.0102K20.0744 ± 0.1621Co0.0436 ± 0.0233Al69.7575 ± 0.7636Sn0.0212 ± 0.0013Mg25.5092 ± 0.2546Se0.0168 ± 0.0039Ca28.1011 ± 1.5547As0.0182 ± 0.003Fe4.9051 ± 0.0989Mo0.0948 ± 0.0133Mn4.8347 ± 0.0529V0.0086 ± 0.0003Ni2.4154 ± 0.0387Sb0.0082 ± 0.0016Be1.986 ± 0.0726Tl0.0057 ± 0.0001Zn0.3814 ± 0.0189Te0.0004 ± 0.0002Ti0.6797 ± 0.0196Ag0.0095 ± 0.0022Cu0.3319 ± 0.0173Ga < LODLi0.3924 ± 0.0145Ge < LODPb0.0808 ± 0.0014Cd < LODRemark Limit of Detection (LOD) of Ga = 0.0000 ppb, LOD of Ge = 0.0074 ppb, LOD of Cd = 0.0030 ppb, LOD of V = 0.0634 ppb, LOD of Ag = 0.0138 ppb and LOD of Te = 0.2157 pp. Element concentrations in CM-PM2.5 from various sources determined via ICP-MS. Remark Limit of Detection (LOD) of Ga = 0.0000 ppb, LOD of Ge = 0.0074 ppb, LOD of Cd = 0.0030 ppb, LOD of V = 0.0634 ppb, LOD of Ag = 0.0138 ppb and LOD of Te = 0.2157 pp. Toxicity of PM2.5 on primary human monocytes The average daily inhalation dose (ADD) of CM-PM2.5 was calculated for the period 2012–2022 in four different age groups, using a previously described model26. The ADD was the highest in infants, followed by children and adults, ranging from 6.61 to 22.57 µg/m3 as shown in Table 2. The ADD in all ages was higher than WHO standard (1.14–3.90 µg/m3). Compared with adults, the result showed that the infant group, children of 2–5 years-old group and 6–12 years old group had 3.41, 3.11, and 1.95 times higher ADD, respectively. Table 2The average daily inhalation dose (ADD) of CM-PM2.5 from 2012 to 2022.YearYearly average PM2.5 (µg/m3)ADD (µg/m3)Infant (0–1 years)Child (2–5 years)Child (6–12 years)Adult (19–75 years)WHO standard5.003.903.552.221.14201229.0022.6420.6012.906.63201336.0028.1125.5716.028.23201434.0026.5424.1515.137.77201535.0027.3224.8615.578.00201632.0024.9822.7314.247.32201723.0017.9616.3410.235.26201827.0021.0819.1812.016.17201931.0024.2022.0213.797.09202028.0021.8619.8912.466.40202124.0018.7417.0510.685.49202219.0014.8313.508.454.34Average28.9122.5720.5312.866.61Ratio of ADD when compared to adult3.413.111.95– The average daily inhalation dose (ADD) of CM-PM2.5 from 2012 to 2022. To explore the non-toxic concentrations of PM2.5 to be used for investigating the cellular response in vitro, CM-PM2.5 toxicity to monocytes was determined by MTT assay. Primary human monocytes were treated with CM-PM2.5 at the concentrations of 0, 5, 10, and 20 µg/ml. The highest concentration tested is equivalent to approximately 106-fold higher than that described for ADD (comparable to 5 × 106 to 20 × 106 µg/m3). However, this concentration range was selected based on the exposure estimates in organs from other studies and commonly used in in vitro cell culture experiments40,12,41. The results indicated that CM-PM2.5 did not affect cell viability at the tested concentrations (0–20 µg/ml) after treatment for 24 h (Fig. 2A). Optimization of the appropriate concentration of CM-PM2.5 for stimulation. (A) The effect of CM-PM2.5 on primary human monocyte cell viability. Cells were treated with CM-PM2.5 (0, 5, 10, and 20 µg/ml) for 24 h. Cell viability was measured by MTT assay. The data was from 6 independent healthy donors. (B–F) Proinflammatory cytokine induction in CM-PM2.5–stimulated primary human monocytes. Cells were stimulated with CM-PM2.5 (5 and 20 µg/ml) or LPS (1 µg/ml) as a control stimulant for 0, 6, 12 and 24 h for RNA collection and 24 h for culture supernatant collection. (B, C) RNA was collected to examine TNFα, and IL-6 mRNA levels by qPCR. β-ACTIN was used as a housekeeping gene. (D–F) Culture supernatant was collected to examine TNFα, IL-6 and IL-1β protein levels by ELISA. The results are mean ± SD of 5–6 healthy individuals. *indicates statistically significant differences at p < 0.05 (one-way ANOVA). To examine the duration of treatment in vitro that is suitable for the transcriptomic analysis, the inflammatory response triggered by PM2.5 treatment was determined using TNFα and IL-6 as readouts. The primary human monocytes were treated with low (5 µg/ml) or high (20 µg/ml) doses of CM-PM2.5 for indicated times and the mRNA levels of TNFA were examined by qRT-PCR. The result showed that TNFA mRNA level increased at 24 h after treatment with the higher dose of PM2.5 (Fig. 2B), whereas that of IL-6 was unaltered (Fig. 2C). TNFα was significantly produced and secreted in a dose-dependent manner in the culture supernatant, as measured by ELISA (Fig. 2D). In contrast, IL-6 and IL-1β were not found to be significantly produced (Fig. 2E and F) compared to the control LPS stimulation. Therefore, CM-PM2.5 induced inflammatory cytokine TNFα production without interfering with the cell viability at the concentration of 5–20 µg/ml. Transcriptomic profiles of CM-PM2.5-treated primary human monocytes To understand how CM-PM2.5 impacts cellular response in primary human monocytes, RNA sequencing was performed. Primary human monocytes isolated from 5 healthy individuals were left untreated or stimulated with two doses, hereafter referred to as low (5 µg/ml) and high (20 µg/ml) doses for 24 h. Differentially expressed genes in untreated or CM-PM2.5-treated primary human monocytes were analyzed. In the presence of a lower dose of CM-PM2.5, a total of 770 differentially expressed genes (DEGs) (314 upregulated genes and 456 downregulated genes) in comparison to cells cultured in medium alone, was identified. On the other hand, the higher dose treatment resulted in a total of 2,242 DEGs (1,102 upregulated genes and 1,140 downregulated genes) (Fig. 3A). Both doses of CM-PM2.5 shared 715 DEGs (32.7%), indicating that the common set of genes was triggered by CM-PM2.5 in human monocytes (Fig. 3B). There were 45 (2%) and 1,415 (64.7%) DEGs uniquely altered in cells treated with low- vs. high-dose CM-PM2.5-stimulation, respectively (Fig. 3B). Taken together, the results demonstrated that CM-PM2.5 induced a substantial shared set of transcriptomic changes, regardless of the concentrations but low and high doses of CM-PM2.5 also triggered unique gene expression profiles. Transcriptomic analysis of primary human monocytes responding to CM-PM2.5. (A) Venn diagram showing the overlapping number of DEGs in 5 and 20 µg/ml CM-PM2.5-stimulated primary human monocytes. (B) Depicted the up-and down-regulated DEGs in the CM-PM2.5 (5 and 20 µg/ml)-stimulated primary human monocytes. (C–F) Gene ontology analysis in CM-PM2.5-stimulated primary human monocytes. GO was classified based on the biological function of DEGs from (2B-C) 5 and (2D-E) 20 µg/ml of CM-PM2.5-stimulated cells using the ShinyGO 0.77 website. The top 20 GO were sorted according to FDR and fold enrichment to show the relationship between enriched pathways in the interactive plot (2B, 3D). A hierarchical clustering tree was used to summarize the correlation among the selected pathways (2C, 3E). (G, H) The enrichment curated pathway was analyzed using the PANTHER database via ShinyGO 0.77 website. Enrichment pathway in DEGs from (3F) 5 µg/ml and (3G) 20 µg/ml of CM-PM2.5-stimulated cells, respectively. The dot size represents the number of DEGs in each pathway. The dot color represents the fold enrichment. Pathway enrichment analysis of the transcriptomes from CM-PM2.5-stimulated primary human monocytes To understand how CM-PM2.5 affects primary human monocyte responses, we performed GO analysis from DEGs identified in the two doses of CM-PM2.5. The biological processes of GO are shown in Fig. 3C and D for the low dose of CM-PM2.5-stimulation and Fig. 3E and F for the high dose of CM-PM2.5 stimulation. Both doses of CM-PM2.5 mainly affected GOs associated with metabolic regulation, cell death, and leukocyte differentiation and activation. Across all GO categories, the high dose of CM-PM2.5 induced more statistically significant changes than the low dose, particularly affecting intracellular protein transport and localization. Pathway enrichment analysis was performed to evaluate the notable pathways in CM-PM2.5 stimulation (Fig. 3G and H). Enriched DEGs in both the low and high doses of CM-PM2.5 treatment showed overrepresentation of the salvage and de novo pathways of ribonucleotide synthesis. These pathways, however, were more enriched at the low dose of CM-PM2.5 than in the higher dose. We used the MSigDB.Hallmark database to analyze the representative biological status of primary human monocytes upon CM-PM2.5 stimulation. The top 20 hallmarks of the cellular phenotype in response to CM-PM2.5 stimulation are shown in Fig. 4A and B. The top hallmark for both stimulation doses was TNFα signaling via NF-κB, the enrichment was greater in low dose stimulation. This may be due to the threshold effect, where the low dose of CM-PM2.5 mainly enriches some hallmarks, while it required a higher dose to activate other hallmarks42,43. Some of these hallmarks display cellular phenotypes similar to those observed in other studies, such as inflammatory response, interferon response, hypoxia, coagulation, apoptosis, and cancer44. It emphasizes that PM2.5 activated a complex response in monocytes upon CM-PM2.5 exposure. Hallmark of DEGs in CM-PM2.5–stimulated primary human monocytes. The predominant hallmark was analyzed using the hallmark.MSigDB database via ShinyGO 0.77 website. Enrichment hallmarks in DEGs from (A) 5 µg/ml and (B) 20 µg/ml of CM-PM2.5-stimulated cells were shown. The dot size represents the number of DEGs in each hallmark. The dot color represents the fold enrichment. Metabolic changes in CM-PM2.5 -treated primary human monocytes From the transcriptomic analysis, primary human monocytes exhibited altered metabolic pathway in response to CM-PM2.5. To identify the candidate metabolic-related gene(s), DEGs from low- and high-dose CM-PM2.5-treated cells were summarized in a volcano plot (Fig. 5A and B). The top twenty genes that overlap in both volcano plots of the two doses are shown as the heatmap (Fig. 5C). Although both doses increased the expression of these overlapping genes, the heatmap suggests a trend toward higher expression levels at the higher dose of CM-PM2.5 compared to the lower dose. Among these, CYP1B1, HK2, and XYLT1 were identified which encode proteins involved in metabolic regulation. They were selected for further validation by RT-qPCR. To validate the candidate metabolic-related genes expression, primary human monocytes were treated with CM-PM2.5 for 24 h. The gene expressions were examined by RT-qPCR. The result showed that CYP1B1, HK2, and XYLT1 mRNA levels were upregulated only in cells treated with the high dose of CM-PM2.5, while cells with the low-dose did not show changes in the expression levels (Fig. 5D). This result implied that the metabolic pathway was differently influenced by the two doses of CM-PM2.5 stimulation. The effect of CM-PM2.5 on candidate metabolic gene expression in primary human monocytes. (A, B) Volcano plot of CM-PM2.5-stimulated primary human monocytes. (A) 5 µg/ml and (B) 20 µg/ml of CM-PM2.5-stimulated cells. The red dot indicated DEGs with FDR < 0.05. The blue dot indicated DEGs with the top twenty FDR. (C) Heat map of the top-overlapping FDR. DEGs in CM-PM2.5-stimulated primary human monocytes. The DEGs were hierarchically clustered using the average linkage of one minus Pearson correlation metric. (D–G) CD14 + human monocytes were isolated from healthy individuals. Cells were treated with CM-PM2.5 (0, 5, and 20 µg/ml) for 24 h. HK2, CYP1B1, and XYLT1 mRNA expression were examined by qPCR. β-ACTIN was used as a housekeeping gene. The results are the mean ± SD of 5–6 healthy individuals. *indicates statistically significant differences at p < 0.05 (one-way ANOVA). Glycolytic rate in CM-PM2.5 stimulated primary human monocytes De novo and salvage pathways of nucleotides for DNA or RNA synthesis requires energy (ATP) mainly from metabolic process of glucose, glutamine, and CO245. Cells rely more on glycolysis rather than oxidative phosphorylation under hypoxic or inflammatory conditions45. According to the representative biological status of primary human monocytes upon CM-PM2.5 stimulation, both low- and high-doses of CM-PM2.5 treatment led to overrepresentation of the hypoxia hallmark, indicating that primary human monocytes were under hypoxic response upon treatment with CM-PM2.5. To examine the metabolic status of CM-PM2.5-stimulated primary human monocytes, cells were stimulated with CM-PM2.5 for 24 h before starting the Seahorse glycolytic rate assay. At basal glycolysis, the extracellular acidification rate (ECAR) remained unchanged at low doses of CM-PM2.5 but decreased at high doses compared to unstimulated cells (Fig. 6A and B). After Rot/AA infusion to inhibit oxidative phosphorylation, the low doses showed no change in ECAR, while high doses reduced ECAR. When glycolysis was inhibited with 2DG, both doses of CM-PM2.5 had no effect on ECAR. The effect of CM-PM2.5 on the glycolytic rate of primary human monocytes. CD14 + human monocytes were stimulated with CM-PM2.5 (5 and 20 µg/ml) for 24 h. Cells were measured for glycolytic rate by Seahorse XF Glycolytic Rate Assay Kit. The viable cells were stained with trypan blue and counted under a microscope at the end of the experiments. (6A-B) ECAR and (6C-D) OCR values were normalized by cell number. The glycolytic rates of 5 and 20 µg/ml of CM-PM2.5 treated cells were shown. The results were performed on three healthy individuals. *indicates statistically significant differences at p < 0.05 (one-way ANOVA). This result indicated that exposure to CM-PM2.5 impaired glycolysis in primary human monocytes. A significant decrease in the oxygen consumption rate (OCR) was observed in cells treated with a high dose of CM-PM2.5 under basal conditions; however, no significant effect was observed following Rot/AA or 2DG infusion (Fig. 6C and D). This result correlates with the hypoxia hallmark enrichment observed in cells treated with the higher dose. Taken together, this metabolic analysis reveals that exposure to high-dose of CM-PM2.5 impairs the glycolytic metabolic pathway in primary human monocytes. Incidence of non-communicable diseases by province in Thailand. Nucleotide (purine and pyrimidine) metabolism is essential for physiological processes to control cellular homeostasis, including signal transduction, carbohydrate metabolism, oxidative phosphorylation, etc. The disturbance of nucleotide metabolism causes abnormal enzymatic function, leading to nucleotide depletion or toxic substance accumulation, eventually causing diseases46. We used data available from the Division of Non-Communicable Diseases, Department of Disease Control, Ministry of Health in Thailand to generate a figure illustrating the correlation between geographical locations and the occurrence of PM2.5-related health conditions, specifically asthma, COPD, cerebral vascular disease, ischemic heart disease, and hypertension (Fig. 7A–E). In this data set, the number of asthma patients did not show a significant increase, which is consistent with previous reports indicating that an increase in PM2.5 levels decreased the hospitalization rate for asthma47. The incidence and mortality rate of PM2.5-relatd disease in Thailand. (A) The geographical figure represented the incidence of asthma patient per 100,000 people. The color key represented the desease incidence. (B–E) The geographical figure represented the mortality rate of (6B) COPD, (C) ischemic heart disease, (D) cerebrovascular diseases, and (E) hypertension per 100,000 people, respectively. The color key represented the disease mortality rate. The map was created using R software with the detail provided in materials and methods section. (F) COPD incidence rate per year from 2014 to 2024 in Chiang Mai. (G) COPD incidence rate in patients categorized by age group. (H) The number of new COPD cases each month from 2014 to 2024. (I) The cumulative number of new COPD cases each month from 2014 to 2024. (J–M) The simple linear regression between the number of new COPD cases each month and (J) PM2.5 AQI level (from 2016 to 2024), (K) NO2 AQI level (from 2014 to 2020), (L) PM10 AQI level (from 2014 to 2022), and (M) O3 AQI level (from 2014 to 2022). COPD had the highest incidence in northern Thailand where Chaing Mai is located, followed by ischemic heart disease. We investigated the number of COPD patients in Chiang-Mai from 2014 to 2024 and found that the number of COPD patients continuously increased annually (Fig. 7F). Older individuals greater than 60 years of age had a higher prevalence than other age groups (Fig. 7G), similar to a previous report that COPD prevalence increased in group with age ≥ 65 years48. Interestingly, the new case of COPD increased in January and February with the cumulative prevalence of about 22.08% and 18.58%, respectively, throughout the year (Fig. 7H-I). This period followed shortly after peak haze exposure in Chiang-Mai (started from December). However, a previous study reported that short-term exposure to PM2.5 was related to COPD hospitalization49. We performed a linear regression analysis between the number of new cases of COPD and the average PM2.5, PM10, NO2, and O3 AQI levels for each month. The result showed that PM2.5 and NO2 AQI showed significant positive correlations with new COPD cases (Fig. 7J and K), PM10 AQI and O3 AQI indeed showed no significant correlation (Fig. 7L and M). This result highlights the importance of addressing air pollution to improve health. However, it should be noted that other factors, including occupational and genetic influence, also contribute to disease pathogenesis which requires further investigation. Discussion PM2.5 exposure is used to assess the effects of PM2.5 on health and develop appropriate air pollution control plans50. Individuals are exposed to PM2.5 in varying and inconsistent doses throughout the year. This study used the annual average PM2.5 concentration (long term exposure) to estimate ADD which showed age was related to the amount of PM2.5 inhaled. Infants and children had approximately two-fold higher inhalation rates and are more vulnerable than adults36. Health risk predictions for 2020–2029 in Chiang-Mai were consistent with previous studies that showed children of all ages would suffer from air pollution and infants were at the highest risk36. Although ADD in the elderly has decreased, chronic exposure to PM2.5 remain a significant concern; each 1 µg/m3 increase is associated with a 9.2% rise in COPD mortality among elderly individuals in the United States51. However, other factors involved in disease pathogenesis should be considered. Air pollution is reported to be a risk factor for several non-communicable diseases. COPD had the highest incidence among these from the data of the Department of Disease Control, Thailand. The number of new cases of COPD was found to increase following repeated seasonal exposure to haze pollution (January and February). In this study, PM2.5 and NO2 AQI were positively associated with new COPD cases, similar to previous study that PM2.5, O3, NO2, and SO2 levels associated with COPD incidence, hospitalizations, and mortality52,53,47. In contrast to PM10, PM2.5 has smaller sizes that penetrate deeply into the alveoli and blood enter the bloodstream. Its larger surface area allows it to carry more toxic substances, making it more harmful. However, the level of NO2 AQI tend to decrease from 2014 to 2020 which contrasts with the consistently increasing trend of PM2.5 AQI. While O3 is known to exacerbate respiratory conditions, the concentration observed in this study remained within acceptable limits. Previous studies have reported that agricultural occupations involving biomass burning are associated with an increased risk of COPD54. Therefore, PM2.5 may not be the only risk factor. Combining other parameters, time points, and long-term patient monitoring will help provide a better understanding of the dynamics of the disease and how these factors interact or are interconnected. The primary sources of PM2.5 in Chiang Mai during haze days were biomass burning during haze periods and traffic emission during non-haze periods37. However, we found that Na was a predominant element in the PM2.5 used in this study. Na is typically associated with marine sources, but Chiang-Mai is located far from the ocean. We suspected that Na in CM-PM2.5 might be transported via long-range atmospheric circulation from the south (Gulf of Thailand) as previous reports indicated that PM2.5 could be transported over 1000 km in the wind55. Previous findings reported that Na level in PM2.5 from Nanjing city, China was strongly correlated with TNFα level, but not IL-6, but we did not examine this in this study56. Al was identified as one of the top five most abundant elements within the CM-PM2.5 sample used in this study. The primary sources of Al include soil, crustal dust and emissions from coal-fired power plants57. Previous toxicological studies indicated that Al exerts size-dependent cytotoxicity on macrophages, induced oxidative stress and maintain inflammation within several tissues such as intestine58,57. Furthermore, K, a tracer for biomass burning, had been linked to increased mortality in patients suffering from diabetes, dementia, cardiovascular disease, respiratory diseases in Denmark59. Given the high regional burden of cerebrovascular disease and ischemic heart disease in northern Thailand, investigating the association between biomass burning-derived K and these clinical outcomes is crucial area for further study. It is important to consider that PM2.5 is a complex matrix of biological components, physical properties, chemical components, the proportion of which are affected by extraction methods. These components often act synergistically, isolated individual components may fail to capture the real biological response60. PM2.5 induced inflammation in monocytes and macrophages is a key contributor of lung injury8,61. We found TNFα levels increased upon CM-PM2.5-stimulation in monocytes, consistent with finding from studies using PM2.5 from other sources62,63,10,9. Benzene and LPS were reported to stimulate TNFα production in microglial cell/monocytes through IKKβ/NF-κB signaling pathway64. In contrast, a high proportion of PAH inhibited TNFα secretion in THP-1 cell line63. Other components such as soil elements (Si, Al, etc.) increased proinflammatory cytokine secretion in different seasonal PM2.5 collections63. Pollution consists of toxic substances that elicit cellular responses with a threshold effect, requiring sufficiently high or cumulative exposure to trigger a significant response42,43. From the transcriptomic data, we found that the low dose CM-PM2.5 stimulation enriched genes in pyrimidine deoxyribonucleotide synthesis and coagulation cascade. In contrast, the high dose of CM-PM2.5 enriched distinct pathways including heme biosynthesis, interferon-gamma signaling pathway, and Toll receptor signaling pathway. These results implied that CM-PM2.5 modulated primary human monocyte functions in a dose-dependent manner. The core responses of primary human monocytes to the different dosages of CM-PM2.5 were the salvage and de novo pathways of ribonucleotide synthesis, oxidative stress, immunological response, and cancer-related pathways. This consistent with pathway identified from serum metabolites of volunteers exposed to short- or long-term PM2.5 exposure, where the highest contributing metabolites were purine, sphingolipid, β-alanine, and glycerophospholipid metabolisms65,66. However, serum metabolite profiles depends on multiple factors, including volunteer status, PM2.5 source, exposure time, etc.67. Short-term exposure to PM2.5 in the aging population altered pyrimidine metabolism in blood circulation similar to our study66. Pyrimidine ribonucleotide synthesis is closely linked to a cellular glucose metabolism. The upregulation of CYP1B1, HK2, and XYLT1 genes was found, but the cellular metabolism shifted from oxidative phosphorylation toward glycolysis. Additional mechanisms may contribute in the regulation of these genes, as PAH stimulation via the aryl hydrocarbon receptor (AhR) increased CYP1A1 and CYP1B1 enzymes to play a role in the detoxification process68,69,41. We observed the impairment of glycolytic rate only in the high dose stimulation. It might be possible that the signal strength activated the different pathways or epigenetic involvement in the primary human monocyte response. In non-small cell lung cancer (NSCLC), PM2.5 exposure increased glycolysis but suppressed acetyl-CoA production, lead to augment cell malignancy70. It supported the information that PM2.5 composition was unique due to the combination of biological, chemical, and physical characteristics. Changes in the metabolic profiles of immune cells can lead to impaired immune responses71. ROS levels depend on succinate and its metabolites, which are important for stabilizing HIF1α activity in an inflammatory state71. The overexpression of HK2 exacerbated the inflammatory response via increased production of pro-inflammatory cytokinesin rheumatoid arthritis and other diseases72,73. Alveolar macrophages of COPD patients exist in a high mitochondrial ROS (mtROS) environment, leading to proton leakage and insufficient energy in the cells74. It remains unclear whether inflammation or metabolic dysregulation is the initiating event. The current study has several limitations. First, the CM-PM2.5 samples were collected over a single, short period time, which may not capture the temporal and spatial variability in PM2.5 composition. Second, the PM2.5 samples were processed using methanol extraction method, which may yield a different toxicity profile compared with other extraction methods due to differences in extraction efficiency35. Third, due to limited sample availability, comprehensive chemical characterization, particularly organic compounds such as PAHs and VOCs, could not be performed. In addition, although this study identified associations between air pollutants and COPD incidence, potential confounding factors require further investigation. Spatial distributions of other diseases are presented in Fig. 7; quantitative correlation analyses were only feasible for COPD due to data availability. Finally, the in vitro experimental design using primary human monocytes may not fully reflect the complexity of in vivo exposure, where multiple cell types and environmental factors, and host conditions contribute to disease pathogenesis. Conclusion Exposure to CM-PM2.5 triggers immunometabolism alteration in primary human monocytes, characterized by inflammation and metabolic dysregulation. Notably, high doses of CM-PM2.5 distinctly suppressed the glycolytic rate, signaling cellular exhaustion. The observed epidemiological correlation between PM2.5 levels and increased respiratory disease suggests that this combination of chronic inflammation and metabolic impairment may play an important role in driving disease pathology. Additionally, the cellular response to PM2.5 is complex, and the exact mechanism by which CM-PM2.5 contributes to disease pathogenesis requires further investigation. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Acknowledgements The authors acknowledge the contributions of Catherine M Hawrylowicz, Peter Gorer Department of Immunobiology, King’s College London, UK as lead PI on the Newton Fund and for her engagement in discussion on study design and helpful critique during manuscript preparation. Author contributions The study conception, design, analysis and the first draft of the manuscript was written by [Naunpun Sangphech], [Tanapat Palaga], and [Tassanee Chetwittayachan]. Material preparation and data collection were performed by [Naunpun Sangphech], [Ganjana Lertmemongkolchai], [Jomkhwan Meerak], [Ian Mudway], [Tassanee Chetwittayachan], and [Sakchai Laksee] and [Pongphan Pongpanitanont]. The funding acquisition was acquired by [Naunpun Sangphech], [Tanapat Palaga], and [Sakchai Laksee]. All authors commented on previous versions of the manuscript. All authors read and approved of the final manuscript. Funding This work was supported by Newton Fund-Office of the Permanent Secretary (MHESI) Institutional Links 2019/20, National Research Council of Thailand (NRCT) (Grant No. 811/2563), Haze free Thailand and PM2.5 problem grant, NRCT (Grant No. N25A650419, partially support), TINToUniversity65 grant, NRCT and Chulalongkorn University (Grant No. N42A680063), and partially supported by Thailand Science Research and Innovation (TSRI) Fundamental Fund, fiscal year 2025. Declarations Competing interests The authors declare that they have no conflict of interest. Availability of data and materials RNA sequencing data that support the findings of this study have been deposited in the NCBI database with the Accession Number GSE237963. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement RNA sequencing data that support the findings of this study have been deposited in the NCBI database with the Accession Number GSE237963.