Airborne particles, bacterial and fungal communities insights of two museum exhibition halls with diverse air quality characteristics

Airborne particles, bacteria and fungi consist main determinants of indoor air quality. In this study, these characteristics were investigated in two exhibition halls of the Historical Museum of Crete, in comparison with the outdoor environment. In Zacharias Portalakis (ZP) hall, three air purifiers were operating during museum opening hours, as opposed to El Greco (EG) hall. A significant part of ultrafine (< 1 μm) particles was efficiently removed by the air purifiers in ZP hall. Airborne coarse particles PN2.5–10 were associated with human occupancy in both halls, indicating transportation from visitors and resuspension as possible sources. Average airborne bacterial concentration was also lower in ZP than in EG hall, assessed by both molecular and culture-dependent methods. The bacterial and fungal communities of both indoor halls were distinct from the outdoor counterpart. Micrococcus and Staphylococcus, in terms of bacteria, and Alternaria and Malassezia, in terms of fungi, were the most abundant genera indoors, most of them being human-related. Hierarchical clustering of indoor samples indicated that EG hall bacteria were similar to ZP hall bacteria collected on the same day, but varied between different sampling dates. This observation, together with the bacterial beta-diversity analysis, implied that both indoor halls probably shared common bacterial source(s), while the respective fungal pattern of the two indoor halls was found significantly separated. The outdoor air contribution in EG and ZP hall bacterial profile presented a comparable fluctuation pattern; this was not observed for fungi, probably indicating the complexity of potential sources for different fungal taxa.

. As humans tend to spend more time indoors than outdoors, indoor air quality (IAQ) becomes an increasingly important modifier of population health due to its impact on multiple health outcomes including allergies, respiratory and cardiovascular diseases, cancer and premature mortality (Vardoulakis et al., 2020). Studies on human exposure to indoor pollution revealed that indoor environments could be at least twice as polluted as outdoor environments (European Commission, 2003). The threats posed by a long-term exposure to indoor air pollution have become more apparent in recent years as buildings are progressively sealed against the outside environment to obtain heating and cooling energy cost savings, leading to the accumulation of indoor pollutants. The type of air pollutants that are present at public places is determined by air exchange, the specificity of a given place, the emissions of indoor materials, temperature and relative humidity fluctuations, the ventilation/purification system and the quality of the surrounding outdoor air (Śmiełowska et al., 2017). The relative importance of any single source depends on pollutant emission rate and the associated risk.
In museums, IAQ context extends beyond health considerations to include the effects of air pollutants on exposed artefacts or other exhibits. Air pollution can cause chemical damage or soiling of surfaces due to particle deposition (Chatoutsidou & Lazaridis, 2019;Godoi et al., 2006). Moreover, airborne microorganisms have been implicated in exhibited artefacts biodeterioration, resulting in high expenditure for museums. Museum airborne communities shape the biodeterioration potential through microbial deposition on the exposed surfaces Saridaki et al., 2022). Bacteria and fungi are considered the most common mediators of biological alterations on cultural heritage objects, depending on environmental factors (temperature, relative humidity, light conditions) and the nature of the underlying surface (Saridaki et al., 2022;Sterflinger & Piñar, 2013).
Human occupancy is an important factor to be considered when assessing IAQ of public places. Visitors entering the buildings are responsible for transferring particulate materials inside or resuspending the deposited dust from the floor (Lazaridis et al., 2015;Thatcher & Layton, 1995;Yoon & Brimblecombe, 2000). Clothing has been demonstrated to serve as a reservoir of microorganisms and airborne particles (Licina & Nazaroff, 2018), while visitor flow was proven to be a major contributor to soiling (Yoon & Brimblecombe, 2001). Another major driver of built environment air microbiome is outdoor air (Chen & Zhao, 2011), which is further supported by the estimated penetration efficiency of bioaerosols that may approach to 100% in naturally ventilation buildings (Nazaroff, 2014). Outdoor air microbial composition varies both temporally and geographically, which prevents a core community profile to be applied to all indoor settings to account for outdoor air impact (Prussin & Marr, 2015).
In this context, particulate matter and airborne microorganisms' control should always be the goal in order to protect valuable artefacts and reduce the drawbacks associated with health-related risks and recurrent conservation interventions. The observed variation in indoor particulate matter levels and microbial characteristics (Drougka et al., 2020;Lazaridis et al., 2015Lazaridis et al., , 2018 highlights the importance of continuous and organized monitoring in public places like museums in order to effectively characterize the microenvironments of interest. In this study, we combined particulate matter monitoring with bacterial and fungal air microbiome profiling in order to assess the main air quality determinants of two museum halls with different exhibit protection strategies. In parallel, we aimed at investigating the impact of human occupancy and air purification to indoor levels and quality characteristics of (bio)aerosols and provide insights on indoor microorganisms' diversity and outdoor air impact as source environment.

Sampling sites
Sampling took place at the Historical Museum of Crete (HMC, https:// www. histo rical-museum. gr/), in Heraklion, Crete, Greece. The museum is located on Heraklion's coastal area, at a distance of about 200 m from the sea and is housed in a neoclassical building consisted of 19 exhibition halls spanning 3 floors and 2 court yards. Indoor sampling sites, namely Zacharias Portalakis (ZP) and El Greco (EG) exhibition halls, are located on the first floor of the building (Online Resource 1, Fig. S1) and were both mechanically heated, ventilated and air-conditioned (HVAC), whereas air-purifiers (photocatalytic ionizers MC70L, Daikin, flow rate 170 m 3 /h, 0.013 kw; https:// www. daikin. eu/ en_ us/ produ cts/ mc70l. html) were present only in ZP hall (total flow rate of three photocatalytic ionizers: 510 m 3 /h; flow rate of the HVAC system: 1,920 m 3 /h; natural infiltration rate: 120 m 3 /h, Katsivela et al., 2021), working on constant mode during museum opening hours (09:00 to 17:00 every day, except Sundays). EG hall is smaller than ZP hall (48 m 2 and 58 m 2 total areas for the two halls, respectively), with 1,920 m 3 /h flow rate of the HVAC system and 110 m 3 /h natural infiltration rate. ZP hall is characterized by low artificial illumination, with wooden exhibits and canvas paintings without protective cases, while EG hall is highly illuminated during museum opening hours, with exhibits that vary, mainly comprised of organic materials, such as paper, wood, silk, wool, cotton (Online Resource 1, Fig. S1). Due to the absence of air purification system, all EG hall vulnerable exhibits lie within protective cases, except a set of paintings. Outdoor sampling site was situated at the backyard of the museum.

Aerosol monitoring
Aerosol monitoring took place in ZP and EG halls in six campaigns spanning summer and autumn months of 2018 (20-27/6, 13-16/7, 24-31/7, 28/8-5/9, 18-24/9, 2-9/10). During all monitoring campaigns, the museum daily visitors were 107 on average (min 95, max 143). Aerosol sampling was performed using an Optical Particle Sizer (OPS 3330, TSI) and a handheld aerosol monitor (DustTrak II 8532, TSI) in ZP and EG hall, respectively. The OPS operates on the principle of single particle counting via light scattering and yields particle number concentrations (PNC) in up to 16 consecutive size bins (channels) between 0.3 and 10 μm. The instrument operated with a sample air flow rate at 1 lpm and the log interval was set at 2 min. In this study, the "default TSI protocol" was used for bin selection in all of the campaigns. Measured particles were divided in three major classes according to their physical characteristics and their ability to penetrate through small openings and cracks. The sum of PNC in the first six channels is denoted as PN 0.3-1 , (midpoint diameters 0.337-1.007 μm), the following four channels as PN 1-2.5 , (midpoint diameters 1.007-2.421 μm) and the last six channels as PN 2.5-10 , (midpoint diameters 2.421-9.015 μm). OPS was supported with conversion software (Aerosol Instrument Manager, TSI) from particle number concentrations to particle mass concentrations (PMC). In this conversion, it is assumed that: a) all particles are spherical with diameter equal to the median diameter of each preselected size bin and b) all particles have the same density in each size bin. Here, a density value of 1.5 g/cm 3 was selected for all indoor particles in HMC and the corresponding mass concentrations were computed.
At EG hall, the DustTrak II (DTII) aerosol monitor was used for measuring PM 10 mass concentrations with a sample flow rate of 3 lpm and the same log interval as the OPS. Both instruments were placed 1.5 m above the ground floor.
OPS and DTII measurement data are not directly comparable, since OPS measures PNC and DTII PMC. Although OPS PNC were transformed to PMC, data are still not comparable, given the assumptions regarding the shape and the density of particles, as mentioned above. Therefore, PMC of the two instruments were only used to denote the different particulate matter profiles during working hours, which were correlated to microorganisms presence indoors.

Bioaerosol sampling and assessment of viable, cultivable airborne microorganisms
Airborne microorganisms were collected in all three sampling locations (EG /ZP hall, outdoors) using a MAS-100 NT microbial air sampler (Merck Millipore), operating at 100 lpm for 2 h. Sampling took place on six working days of the year 2018 (20/6, 9/7, 24/7, 28/8, 18/9, 8/10), during museum opening hours. Microbial DNA was collected by impaction at a 47 mm sterile mixed cellulose esters filter with 0.45 μm pore diameter (ME 25/21 ST, Whatman) attached on top of a sterile 1.5% agar plate. Upon sampling, filters for molecular analysis were placed in sterile petri dishes sealed with parafilm and stored at −20 °C until further processing. In parallel, viable, cultivable, airborne microorganisms were collected on different microbiological growth media using a second MAS-100 NT microbial air sampler. The collected air volume varied from 50 to 250 L. Two sequential repetitions were performed for each of the following categories of viable, cultivable airborne microorganisms: (i) opportunistic pathogenic, heterotrophic bacteria; (ii) autotrophic chemolithotrophic (chemoautotrophic) bacteria; (iii) acid-producing bacteria and (iv) fast-growing mesophilic fungi. The sampling methodology and the composition of the selective growth media were as described in Lazaridis et al. (2015) and Katsivela et al. (2021). The heterotrophic bacteria were cultivated in Tryptone Soy Broth (Merck, Germany) containing 1.5% (w/v) agar at 37 °C in the dark for 48 h. Incubation temperature of 37 °C was chosen for the determination of the airborne, opportunistic pathogenic, heterotrophic bacteria. The autotrophic chemolithotrophic bacteria were cultivated in minimum mineral Tris phosphate agar No. 457 (Brunner, Leibniz Institute DSMZ) without any carbon source at 37 °C for 8 days. The acid-producing bacteria were cultivated in Gluconobacter oxydans medium No. 105 (Leibniz Institute DSMZ) at 37 °C in the dark for 8 days and were counted only when clear zones were visible around the colonies. The viable mesophilic fast-growing fungi were cultivated in Malt Extract Broth (Lab M, England) containing 1.5% (w/v) agar at 20 °C in the dark for 72 h. The concentration of viable, cultivable airborne microorganisms was finally expressed as colony forming units per cubic meter of air (CFU/m 3 ). The determination of viable, cultivable, airborne heterotrophic bacteria and fungi is recommended by the Italian Ministry of Cultural Heritage and Activities (MiBAC, 1998) as well as by other researchers (Gorny et al., 2016). Threshold limit values for viable airborne heterotrophic bacteria and fungi below 750 and 150 CFU/m 3 , respectively, are proposed for the protection of museum exhibits from harmful effects of microbial agents (MiBAC, 1998). Temperature (T) and relative humidity (RH) were recorded during sampling (Online Resource 1, Table S1) using a portable air velocity meter (Veloci-Calc, TSI, Germany).

DNA extraction
Each filter was cut into three pieces of equal size and each filter piece was further processed independently. DNA was extracted from filter pieces using DNeasy PowerSoil kit (Qiagen). Each filter piece was aseptically loaded into the bead-containing tube, lysis buffer was added and the tube was heated to 65 °C for 15 min followed by 15 min of vortex shaking. The remaining steps of DNA extraction were performed according to the manufacturer instructions. In order to verify that the microorganisms detected originated from the air microbiome under investigation, three sterile filters were subjected to DNA extraction in parallel with the rest study samples. No amplification product from the blank filters was detected in subsequent stages.

Real-time polymerase chain reaction (PCR) amplification and quantification
Total bacterial DNA was quantified by real-time PCR using the universal bacterial primers and probe set described in Nadkarni et al. (2002). The reactions were performed in CFX Connect real-time PCR detection system (Biorad). Amplification was carried out in 20 μl reactions consisting of 2 μl template, 10 μl 2 × KAPA Probe Fast Universal qPCR Master Mix (Roche) and 0.2 μM final concentration of each primer and the dual-labelled probe under the following conditions: 95 °C for 3 min, then 40 cycles at 95 °C for 15 s and 60 °C for 1 min. No-template controls were included in each run. The cycle threshold of each sample was compared with a standard curve and the result was expressed as a numerical value of the target genomes in the sample. The standard curve for DNA quantification was constructed with triplicates of serial dilutions of Escherichia coli (E. coli) DSM498 strain DNA, previously quantified using Qubit dsDNA High Sensitivity Assay with Qubit 4 fluorometer (Thermo Fisher Scientific). The standard curve serial dilutions ranged from 10 1 to 10 5 E. coli genome copies (R 2 = 0.998). Real-time PCR running control, baseline correction, determination of cycle threshold (Ct) values and generation of standard curve were performed using the default settings of Biorad CFX Manager software v3.1 (Biorad). Each sampling was quantified in six replicates, two for each filter piece. Final DNA concentrations were expressed as E. coli genome equivalents (GE) per cubic meter of air, taking into account the total volume of DNA extract and air sampled.

16S rDNA and Internal Transcribed Spacer (ITS) region amplicon sequencing
For amplicon sequencing analysis of airborne bacteria, the hypervariable V3-V4 region of the 16S rDNA gene was amplified as described in Chatoutsidou et al., 2021. Fungi identification was based on internal transcribed spacer 2 (ITS2) region amplification, using the MiSeq universal primers ITS3F (5′ TCG-TCG-GCA-GCG-TCA-GAT-GTG-TAT-AAG-AGA-CAG-GCA-TCG-ATG-AAG-AAC-GCA-GC 3′) and ITS4R (5′ GTC-TCG-TGG-GCT-CGG-AGA-TGT-GTA-TAA-GAG-ACA-GTC-CTC-CGC-TTA-TTG-ATA-TGC 3′). The PCR temperature profile was 95 °C for 3 min, followed by 35 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s and a final extension step of 72 °C for 5 min. The PCR products were electrophoresed on a 1.5% agarose gel in order to examine the presence and size of the amplified fragments and were then purified by precipitation for 15 min at 37 °C using equal volume of a solution containing 20% polyethylene glycol and 2.5 M NaCl. In order to include the Illumina MiSeq indices, a second PCR was performed in 25 µl final volume containing 1 µL of the purified PCR product diluted up to 10 ng/μl, 5 µL KAPA Hifi Taq buffer (5 ×), 0.75 µL dNTPs mix (10 mM), 2.5 µL of forward index primer (10 µM), 2.5 µL of reverse index primer (10 µM), 0.5 µL of KAPA Hifi Taq polymerase (KAPA Biosystems), and 12.75 µL PCR-grade water. The temperature profile used PCR was: 95 °C for 3 min followed by 8 cycles of 95 °C for 30 s, 30 s at 55 °C, 30 s at 72 °C and a final extension step of 72 °C for 3 min. The resulting amplicons were purified using the NucleoMag NGS (Next Generation Sequencing) Clean-up and Size Selection kit (Macherey-Nagel) following the manufacturer instructions. Indexed amplicons from all samples examined were mixed in equimolar ratio (8 nM), and sequencing was performed by Macrogen using a 2 × 300 bp pair-end kit on a MiSeq platform. The datasets have been deposited to NCBI Sequence Read Archive database under BioProject PRJNA743183.

Bioinformatic analysis of amplicon sequencing data
After sequencing, bioinformatic analysis was performed using USEARCH v.11 (Edgar, 2010) and QIIME2 distribution 2019.1 (Bolyen et al., 2019). Paired-end reads were assembled and trimmed using the usearch -fastq_mergepairs option. The quality of assembled sequences was improved using -fastq_filter, followed by finding unique read sequences and abundances using -fastx_uniques option. Sequences were clustered into operational taxonomic units (OTUs) with -cluster_otus command based on 97% OTU clustering using UPARSE algorithm (Edgar, 2013). Cross-talk errors were identified and filtered with -uncross option based on UNCROSS2 algorithm (Edgar, 2018). Taxonomy was assigned with Qiime2 based on BLAST + algorithm (Camacho et al., 2009) against SILVA 138.1 release database (Quast et al., 2013) for bacterial and Unite version 8.2 database (Abarenkov et al., 2020) for fungal OTUs. Νon-bacterial/fungal OTUs were excluded from the analysis. Richness and Pielou's evenness indices of alpha diversity were calculated based on "diversity" function of "vegan" R package and plotted using "ggplot" function from "ggplot2" package. Beta diversity was analysed to evaluate the similarity of microbial communities from different sampling sites using generalized UniFrac distance (Chen et al., 2012). Beta diversity was visualized via non-metric multidimensional scaling (NMDS) plots, created using the best solution reached by monoMDS engine of "vegan" R package after 20 tries.
Outdoor air contribution was investigated using fast expectation-maximization microbial source tracking (FEAST) algorithm as similarity metric (Shenhav et al., 2019). For each sampling day, mean RA values of EG/ZP hall samples were used as "sink" and the respective outdoor mean RA values were used as "source".
The potential sources of microbial taxa were explored by assigning Environment Ontology (EnvO) terms (Buttigieg et al., 2013) to the identified sequences using Seqenv pipeline version 1.3.0 (Sinclair et al., 2016). Seqenv was run on both the 16S rRNA gene and ITS sequences using the default settings that normalize the assigned EnvO terms frequencies based on RA values of the respective OTUs in the corresponding sample.

Statistical analysis
Aerosol monitoring data analysis was performed using "openair" R package (Carslaw & Ropkins, 2012). Wilcoxon rank sum test was used to identify significant differences of alpha diversity indices and of relative abundance (RA) values between different sample groups; a P-value < 0.05 was considered indicative of statistical significance. A permutational multivariate analysis of variance (PERMANOVA) using distance matrices was calculated using "adonis" function from "vegan" R package with 999 permutations to determine significant differences between the separated groups. Benjamini-Hochberg correction was applied to calculated P-values of pairwise comparison tests for decreasing the false discovery rate. Linear regression analysis was performed using "stats" R package.

Airborne particles/mass concentration of ZP/EG halls
Particle levels in ZP and EG halls were monitored as PNC and PMC, respectively. Daily average PNC variations in ZP hall clearly depicted a decrease in particles of aerodynamic diameter lower than 1 μm (ultrafine particles) during opening hours (Fig. 1a).
Average PNC values of ultrafine particles, considering all campaigns in ZP hall, were 61 and 45 #/cm 3 , when the museum was closed and open, respectively. This demonstrates a decrease in ultrafine PNC by 26% during ZP hall occupancy. On the other hand, bigger particles showed increased PNC during museum opening hours, with average concentration values higher by 1.8 and 5.8 times for PN 1-2.5 and PN 2.5-10 particles, respectively (Fig. 1a). A sharp increase in PN 1-2.5 and PN 2.5-10 particles was observed around 09:00, which was noted in all six campaigns, but it was considerably more intense in the first and third campaign (data not shown). PNC for PN 1-2.5 and PN 2.5-10 particles began to decrease between 09:40 and 09:50 in all campaigns, but remained above the background values (closed museum) throughout the opening hours (Fig. 1a).
Particles larger than 1 μm in equivalent aerodynamic diameter (D p ) were fewer in number by an order of magnitude, but they represented a large fraction of the total particles mass. Using a density value Fig. 1 a Particle number concentration (PNC) of ZP hall, classified at different size classes: 0.3-1, 1-2.5 and 2.5-10 μm. b PM 10 particle mass concentration (PMC) of EG hall. The shaded area represents the museum opening hours of 1.5 g/cm 3 for ZP hall particles, the corresponding mass concentrations were computed (Online Resource 1, Fig. S2). Coarse particles (D p > 2.5 μm) were characterized by elevated PMC and strong variations during opening hours. The end of visiting period was combined with peak PMC for PN 1-2.5 and PN 2.5-10 particles, while PN 1-2.5 particles returned gradually to background PMC values (closed museum).
As far as EG hall is concerned, PM 10 mass concentrations for opening hours and closed museum were very close in all campaigns (Fig. 2), with average values 21 and 20 μg/m 3 , respectively, while in ZP hall the average mass concentration increased from 7.5 μg/m 3 to 12.5 μg/m 3 during opening hours. Daily variation of PM 10 particles in EG hall showed that the particle load was influenced by the presence of visitors and the consequent resuspended dust (Fig. 1b). Mass fluctuations were more intense between 09:00-17:00 and gradually the concentration reached practically steady state values during night time.

Airborne bacteria quantification
Quantification of total airborne bacteria by real-time PCR (expressed as E. coli GE per m 3 of collected air) is shown in Fig. 3a. Bacterial concentrations ranged from 419 to 2,157 E. coli GE/m 3 and from 306 to 1,530 E. coli GE/m 3 in EG and ZP hall samples, respectively, while outdoor bacterial concentration minimum and maximum values were 812 and 4,221 E. coli GE/m 3 . The estimated mean (± standard deviation) bacterial concentration was 1,090 (± 647), 797 (± 466) and 1,753 (± 1,503) E. coli GE/m 3 for EG, ZP hall and outdoor samples, respectively. Bacterial concentration in ZP hall was lower than in EG hall by 26.9% on average values.
Culture-dependent quantification of airborne, viable, cultivable microorganisms showed that bacteria were enriched indoors. The average (± standard deviation) concentrations of the opportunistic pathogenic, heterotrophic bacteria in both indoor halls (723 ± 361 CFU/m 3 and 1,084 ± 832 CFU/m 3 in ZP and EG hall, respectively) were found to be statistically higher (Wilcoxon test P-values < 0.05) than the outdoor concentration (195 ± 143 CFU/m 3 , Fig. 3b). The indoor/outdoor (I/O) ratio of the average heterotrophic bacteria values varied from 3.7 (in ZP hall) to 5.6 (in EG hall). On the contrary, the fast-growing fungi were more abundant outdoors (219 ± 289 CFU/ m 3 ), with I/O ratios 0.5 and 0.2 for ZP and EG hall average values, respectively (Fig. 3c). Comparing the two indoor sites, ZP hall heterotrophic bacteria were lower by 33.3% on average than the respective EG hall concentration, while the average fast-growing fungi level in EG hall (54 ± 38 CFU/m 3 ) was lower than in ZP hall (101 ± 140 CFU/m 3 ) by 46.8%. Τhe acid-producing bacteria (151 ± 42 CFU/m 3 in ZP hall and 143 ± 34 CFU/m 3 in EG hall) also showed enrichment indoors (I/O ratios 2.3 and 2.2 of average ZP and EG hall values, respectively, Fig. 3d). Moreover, chemoautotrophic bacteria levels were in the same range indoors (520 ± 509 CFU/m 3 in ZP hall; 612 ± 539 CFU/m 3 in EG hall), with I/O ratios 1.0 and 1.2 of average ZP and EG hall values, respectively (Fig. 3e).
In respect to the threshold limit values for the protection of the exhibited artefacts in cultural heritage and museums (MiBAC, 1998), the concentration of Fig. 2 Temporal dynamics of ZP hall PN 0.3-10 particle number concentration (PNC) and EG hall PM 10 particle mass concentration (PMC) across all sampling campaigns the viable airborne heterotrophic bacteria in ZP hall where three air purifiers were operated, followed the proposed threshold limit value (< 750 CFU/m 3 ). In addition, the HVAC system of HMC was adequate for the harmonization of fungal concentrations (< 150 CFU/m 3 ).

Airborne bacterial and fungal communities composition
The bacterial and fungal community composition and diversity of all samples collected were investigated using Illumina high throughput sequencing of 16S rDNA gene and ITS2 region amplicons, respectively. For bacteria, a total of 1,633,234 qualified paired-end reads with an average count per sample of 30,245 reads were obtained after sequencing and quality filtering. Based on a 97% species similarity, 147 bacterial OTUs classified in 10 phyla, 14 classes and 104 genera were obtained across all the samples (Online Resource 2). For fungi, the total and average number of reads per sample were 1,133,325 and 20,988, respectively. In total, 75 fungal OTUs classified in 2 phyla, 13 classes and 54 genera were identified (Online Resource 2). In total, most of the bacteria identified belonged to Proteobacteria (mean RA per sampling event 19.3%-60.9%), Actinobacteriota (18.0-47.5%) and Firmicutes (11.5-43.7%) (Online Resource 1, Fig. S3a). The rest of the phyla were Bacteroidota, Chloroflexi, Cyanobacteria, Deinococcota, Halobacterota and Verrucomicrobiota. On the other hand, fungi were represented only by two phyla, namely Ascomycota and Basidiomycota, with Ascomycota being the most abundant with mean RA ranging from 54.8 to 96.4% per sampling event (Online Resource 1, Fig. S3b).
High variation was observed among the genera abundances of different samples (Online Resource 2). Mean RA values of the 30 most abundant bacterial and fungal genera that accounted together for 76.5% and 92.1% of total abundance, respectively, are presented in Fig. 4. In terms of bacteria, outdoor samples separation in hierarchical clustering demonstrated that bacterial composition outdoors was distinct from the indoor counterparts (Fig. 4a). EG hall samples were similar to ZP hall samples collected on the same day, but varied between different sampling dates. In outdoor samples, the most abundant bacterial genus was Kocuria (mean RA 13.8 ± 2.2%). In indoor samples, the two most abundant bacterial genera were Micrococcus (13.8 ± 1.5%) and Staphylococcus (8.5 ± 0.7%), both of which were undetectable in outdoor samples.
Fungal composition of collected samples exhibited distinct characteristics as demonstrated by hierarchical clustering (Fig. 4b). While outdoor samples were mostly separated from indoor ones, EG/ZP hall samples clustering did not follow the same pattern observed for bacteria, but was rather arbitrary as far as the sampling dates or the sampling site are concerned. In most samples, Altenaria was dominant (54.7 and 28.5% mean RA in outdoor and indoor samples, respectively). Malassezia presented high abundance in indoor samples (5.2-32.0%), but very low in all outdoor samples (mean RA ≤ 0.1%). The rest of fungal genera varied considerably between different sampling sites and dates (Fig. 4b, Online Resource 2).

Differentially abundant microorganisms
The significance of bacterial and fungal RA differences between indoor/outdoor and EG/ZP hall samples was assessed by Wilcoxon rank sum test on the mean values of each sampling event. Among the 30 most abundant genera presented in Fig. 4, 16 bacterial and seven fungal ones were found to be significantly different between indoor and outdoor samples (P-values < 0.05, Table 1). On the other hand, among the same genera, only Staphylococcus presented significantly different RA between the two indoor halls (Wilcoxon rank sum test W = 33, P = 0.015).

Airborne bacterial and fungal communities diversity
Species richness and Pielou's evenness diversity index were calculated for all samples based on the number and RA of bacterial and fungal OTUs (Fig. 5) Table S2), with the most separated group being the one representing the outdoor samples (Fig. 6). On pairwise comparison level, statistically significant separation of clusters was also observed between indoor/outdoor samples both for bacteria and fungi (Benjamini-Hochberg corrected P-values = 0.0015). In terms of bacteria, EG and ZP hall samples do not form statistically different clusters, but they differentiate in terms of fungi (PERMANOVA test P-values = 0.669 and 0.016, respectively).

Bacterial/fungal communities relationship
The relationship between the airborne bacterial and fungal communities detected by molecular methods in all sampling sites was investigated by linear regression analysis and Spearman correlation test of alpha diversity indices (Online Resource 1, Fig. S5). Species richness and Pielou's evenness indices of EG hall bacterial and fungal samples were positively correlated (correlation coefficients 0.37 and 0.43, respectively, Fig. S5a, b). On the other hand, alpha diversity indices of ZP hall (Fig. S5c, d) and outdoor (Fig. S5e, f) bacterial and fungal samples were negatively correlated (correlation coefficients ranging from -0.66 to -0.2).
The relationship between the cultivable microorganisms concentrations was also explored using  S6). No correlation was found among the different groups of cultivable microorganisms in EG hall (Spearman correlation coefficients varied from -0.058 to 0.14, Fig. S6a-c). On the other hand, the heterotrophic bacteria were found positively correlated with the fast-growing fungi (r = 0.58), the acid-producing bacteria (r = 0.9) and the chemoautotrophic bacteria (r = 0.83) in ZP hall (Fig. S6d-f). Similarly, positive correlation was observed among cultivable microorganisms in outdoor samples (Fig. S6g-i).

Assessment of bioaerosol sources
The environmental habitats potentially associated with bacterial and fungal OTUs identified indoors were investigated using Seqenv pipeline. The most abundant EnvO term associated with bacteria was "soil" (17.3% mean normalized frequency in all indoor samples) (Online Resource 1, Fig. S7a). Other highly represented EnvO terms were "biofilm" and "activated sludge" (6.6 and 5.9%, respectively). On the other hand, the most frequent EnvO term for indoor fungi was "sea water" (22.2%), followed by "soil" (13.8%). Other matched terms were "garden" and "air filter" that presented different frequency between EG and ZP halls fungal samples (Fig. S7b).
Moreover, outdoor air contribution to the microbial profile of indoor halls was investigated using FEAST algorithm as similarity metric. Outdoor source contribution varied between different sampling events and ranged from 0.218 to 0.635 for bacteria (Fig. S7c) and from 0.237 to 0.798 for fungi (Fig. S7d). ZP hall bacterial profile was closer to the outdoor counterpart than the respective EG hall profile in most of the sampling events, while in terms of fungi, EG hall samples were found to be more related to the outdoor fungal communities. Outdoor source contribution showed similar fluctuation for EG/ZP hall bacteria, but distinct for EG/ZP hall fungi.

Discussion
The interplay of particulate matter and air microbiome constitutes a major determinant of air quality in indoor environments. In this study, the profile of two museum exhibition halls was investigated in terms of airborne particles dynamics and microorganisms qualitative and quantitative characteristics.
Particle levels in ZP and EG halls were monitored as PNC and PMC, respectively. The two exhibition halls were both mechanically heated, ventilated, and air-conditioned, but differed in the fact that air purification systems were present only in ZP hall. This resulted in different particle profiles during opening hours, since a significant part of particles with aerodynamic diameter below 1 μm was efficiently removed, probably by the combination of the HVAC system and the air purifiers. The operated Daikin MC70L air purifiers in ZP hall use, besides filters, an electrostatic precipitator system for particle removal. The ionization process has been reported to efficiently remove submicron particles from the air (Chen et al., 2020;Park et al., 2011;Pushpawela et al., 2017). Most of the similar studies were conducted in controlled environments, e.g. closed chambers or laboratory rooms. To our knowledge, there are no intensive Fig. 6 Non-metric multidimensional scaling plots of generalized Unifrac distances of bacterial (a) and fungal (b) samples of all sampling sites. Star plots with group mean points are indicated experimental studies of commercial air purifiers/ionizers in real life conditions, but since ultrafine particles reduction was observed only in ZP hall during working hours, it could be attributed to the use of air purifiers. Ultrafine particles can efficiently penetrate inside the showcases and finally deposit on the surfaces of the exhibited artworks (Grau-Bové & Strlič, 2013;Xiu et al., 2015). On the other hand, in both halls, airborne coarse particles were associated with the presence of visitors, indicating transportation from outdoors and resuspension as possible sources. In accordance, a sharp increase of PN 1-2.5 and PN 2.5-10 particles was observed around 09:00, which could be partly attributed to preparation and cleaning activities taking place just before and a few minutes after the opening of the museum.
PNC to PMC conversion requires particle density values for each size bin; for HMC indoor particles, no size bin sensitive density data were available. In previous studies, density values between 1 and 2 g/ cm 3 for the whole particle size distribution have been either assumed based on particle characteristics (Vu et al., 2017;Patel et al., 2020) or computed via experimental data in various indoor environments (Castro et al., 2015;Wang et al., 2020). Despite the lack of similar studies in museums' microenvironments, a density value of 1.5 g/cm 3 was selected here for all indoor HMC particles and the corresponding ZP hall mass concentrations were computed for better interpretation of particles dynamics; nevertheless, ZP and EG hall PMC values cannot be considered directly comparable. Elevated coarse particles of ZP hall and their strong variations during opening hours, were probably due to resuspended dust caused by the visitors' movement around the hall. Large particles remained airborne for relatively short time periods and were only removed if they were very close to air purifiers. Coarse particles are not usually emitted by indoor sources in museum microenvironments and their main transport mechanism is the attachment to the shoes and clothes of the people entering the museums (Camuffo et al., 1999;Mouratidou & Samara, 2004). The use of air purifiers had a clear effect on ZP hall particle profile, since ultrafine particles were dominant when the museum was closed, while opening hours were characterized by strong coarse particles fluctuations.
Bioaerosols constitute a significant part of total PM (Després et al., 2012). Nevertheless, the biotic and abiotic fraction of PM do not always share common sources (Xie et al., 2021), which explains in part why airborne microorganisms concentrations are not always in accordance with PMC or PNC, especially in cultural heritage collections (Lazaridis et al., 2015) as well as outdoors (Raisi et al., 2013). Moreover, while PM monitoring is performed on a real-time basis and may span over long periods, bioaerosol sampling usually represents a snapshot of microbial communities or reflects the microbial profile under investigation agglomerated over a period of several hours, which renders their correlation problematic. In the present study, our goal was to describe, by both molecular and culture-depended methods, the main air quality characteristics of two indoor halls in reallife conditions.
Bacterial DNA was chosen for real-time PCR quantification as an indicator of indoor air microbial load. Fungal ribosomal DNA can exhibit substantial inter-and intraspecies variation in copy number that may obscure the true fungal concentration of environmental samples (Lavrinienko et al., 2021) and was thus disqualified for real-time PCR quantification analysis. Bacterial concentration was lower in ZP than in EG hall, probably reflecting the air purifier effect. Although some variation was observed in bacterial concentrations detected by real-time PCR between different sampling days (data not shown), the overall fluctuation was of less than one logarithm as reported elsewhere (Gaüzère et al., 2014). Similarly, fluctuation of viable, cultivable microorganisms between different sampling dates (Fig. 3) was also observed, probably reflecting the variability imposed by the environmental conditions during the sampling time.
In terms of bacterial composition, Proteobacteria, Actinobacteriota and Firmicutes were the prevalent bacterial phyla as previously reported in similar air environments (Brimblecombe et al., 2004;Chen et al., 2010), as well as on different surface materials of museum artefacts at the Historical Museum of Crete (Saridaki et al., 2022). The outdoor microbial profile was clearly separated from indoor, as RA hierarchical clustering, alpha diversity indices and beta diversity analysis suggest. A group of bacterial genera consisting of Corynebacterium, Micrococcus, Staphylococcus, Streptococcus and Enhydrobacter were significantly elevated in indoor halls, while presenting minimum RA in outdoor samples. These well-documented human-associated genera may be considered as the human occupancy fingerprint in the museum and represented on average 36% of total indoor RA. Corynebacterium, Micrococcus, Staphylococcus and Streptococcus were also detected as highly abundant genera on surfaces of artefacts residing in EG/ZP halls (Saridaki et al., 2022). Similarly, Malassezia, a skin-related fungus (Gaitanis et al., 2012), was dominant in indoor air samples of this study. On the other hand, other genera like Kocuria, Blastococcus, Exiguobacterium, an Arcobacteraceae genus, Sphingomonas, Arthrobacter, Bacillus, Algitalea and Planococcus (from bacteria) and Botrytis, Podosphaera (from fungi) were found more enriched in the outdoor environment. Among them, some have been reported with differential (human/ environmental) origin (Kocuria, Bacillus, Arthrobacter), while others (Blastococcus, Exiguobacterium, Sphingomonas, Algitalea, Planococcus, Botrytis, Podosphaera) are considered as primarily environmental (Adams et al., 2013;Fujiyoshi et al., 2017 and references therein).
Several of the bacterial/fungal genera identified here in HMC indoor halls were previously detected in culture-based studies of indoor environments (Chen et al., 2010;Skóra et al., 2015), which indicates that at least a fraction of them probably represents a metabolically active community rather than airborne cell debris. The public places air microbiome is of particular importance when health-related genera are considered. Airborne fungi may affect human health through allergic reactions, infections or toxic responses. Alternaria, reported here in high RA, together with other identified genera like Aspergillus and Cladosporium, are known allergenic factors (Green et al., 2006).
Hierarchical clustering of indoor samples indicated that EG hall bacteria were similar to ZP hall bacteria collected on the same day, but varied between different sampling dates. This observation, together with the bacterial beta-diversity analysis implied that both indoor halls probably share common bacterial source(s). Interestingly, the respective fungal pattern was more complex, and the two halls beta diversity distances were significantly separated. These observations together may reveal a complexity in fungal sources for the two indoor environments or/and that fungal taxa presented distinct survival/growth rates in EG/ZP halls due to differences in microclimatic conditions, building materials or other indoor architectural characteristics. To better explore the relationship between the two co-occurring communities, as well as between the different groups of cultivable microorganisms, linear regression analysis was performed on bacterial/fungal alpha diversity indices and concentrations. The EG hall bacterial diversity indices were positively correlated with their fungal counterparts, but not the ones from ZP hall or outdoors. Similarly, ZP hall resembled outdoors in respect to the positive correlation of heterotrophic bacteria quantities with the rest cultivable microorganisms concentrations. These observations together support the diverse microbial profile of the two indoor halls.
Undoubtedly, the air purifier operation in ZP hall was a major difference from the other hall under investigation. Recently, Saw et al. (2022) reported that the location and the airflow of the air purifiers in relation to the room size can affect their performance and efficiency. Other characteristics that may potentially affect the air quality comprise differences in building or exhibited materials, architectural design, the insulation from the outdoor environment, the number of visitors and the time they spend in each hall. In this study, with the exception of the genus Staphylococcus, no statistically significant difference was observed between the microbial genera relative abundance of the two indoor halls. While the reduced bacterial concentration of ZP hall, determined by both molecular as well as culture-based quantification, could be possibly attributed to the air purifier presence, to our knowledge, there are no reports from real-life environments that specific microbial taxa are differentially affected by photocatalytic ionizers.
Outdoor samples were collected in parallel, in an attempt to estimate the outdoor air contribution in indoor microbial profile and characterize one of the most influential bioaerosol sources of built environments. Outdoor air contribution varied between different sampling dates, as well as between bacteria and fungi. Although the impact of outdoor air in EG and ZP hall bacterial profile presented a comparable fluctuation pattern, that was not the case for fungi, which may indicate the complexity of potential sources for different fungal taxa. As indoor air represents the compositional impact of several environmental habitats, the potential origin of the observed airborne taxa was explored using a text-mining approach that enabled to better place the genetic information obtained here into a wider environmental context. From this perspective, indoor bacteria were found to be mostly soil-related, whereas indoor fungi were primarily associated with sea water. Both habitats were characteristic of the coastal urban environment where the HMC lies.

Conclusions
The profile of two museum exhibition halls was investigated with regard to airborne particles dynamics and microbial communities characteristics. Indoor halls occupancy was associated with increased airborne coarse particles and human-related bacterial and fungal genera, assessed by both molecular and culture-depended methods, indicating the importance of museum visitors in this built environment properties. While the two halls shared similar bacterial communities composition, a statistically significant difference was observed in their fungal profiles. Moreover, the air purifier presence had a measurable impact on indoor air quality in terms of both airborne particles and microorganisms concentration and should be considered as a mitigation strategy of indoor environments airborne microbial burden. Biodiversity studies like the present one illuminate the microbial community composition and provide a solid base for future research perspectives towards the protection of vulnerable exhibited artefacts and occupants health.