Characterization of Chemical and Microbial Species from Size-Segregated Indoor and Outdoor Particulate Samples

The respirable particles in both outdoor and indoor air contain several different components that are considered to have adverse health effects; e.g., polycyclic aromatic hydrocarbons (PAHs), various metals and microbial species. In this study, size segregated particle samples were collected for chemical, microbial and toxicological analyses from the indoor and outdoor air during each season of the year. The indoor sampling was carried out in a new, detached house with a novel sampling approach. The inorganic species accounted for 8–43% of the total respirable particles. The highest fine particle metal concentrations, both outdoors and indoors, were observed during summer, when the air quality was affected by wildfire smoke plumes, while in coarse particles the total metal concentrations were the highest during the spring, due to the high contribution from mineral dust. The PAH concentrations were 1.3 to 4.8 times higher in outdoor than in indoor air, and they were clearly the highest during winter, most probably due to residential heating, which is a major PAH source. PAHs with four rings had the largest contribution to the total PAHs. Microbial DNA was observed in all size classes, but the highest concentrations were measured in the coarse (PM2.5–10) fraction. The microbial concentrations were higher in the indoor air samples during winter, while in the outdoor ones during summer.


INTRODUCTION
It is well established that exposure to urban air particles is consistently associated with increased cardiorespiratory mortality and morbidity (WHO, 2003).However, outdoor exposure data alone do not reflect the true exposure to total airborne particles since most people spend the majority (> 80%) of their time indoors (Myers and Maynard, 2005).Indoor exposure data concerning non-infectious bioaerosols both at home and work environments have highlighted their critical importance as contributors many of today's most relevant public health problems, e.g., respiratory symptoms, asthma, and clusters of autoimmune diseases.The indoor exposure involves particles both of outdoor and indoor origin.This view is supported by data showing that the concentrations and characteristics of respirable particles indoors are largely affected by both ambient air quality and by indoor sources including numerous indoor activities (Jones et al., 2000;Chen and Hildemann, 2009).However, the causal relationship between the health outcomes and human exposure has proved difficult to clarify due to the difficulties in determining the true exposure and true dose.In ambient air, the respirable particle concentrations are influenced by meteorological conditions, long-range transport of pollutants, local primary particle sources and new particle formation in the air.Important indoor sources for particulate matter (PM) are household cleaning, cooking, smoking, and heating e.g., by wood combustion (Kamens et al., 1991, Jones et al., 2000, Ward and Noonan, 2008).Furthermore, resuspension of sedimented dust makes a contribution to the indoor air coarse particle concentrations and human activities increase substantially their concentrations (Raunemaa et al., 1989, Thatcher and Layton, 1995, Jones et al., 2000).Although various human activities typically lead to even higher indoor particle mass concentrations during daytime than outdoor concentrations, in most cases the ambient aerosol has a considerable effect on the indoor air quality due to the extensive infiltration of outdoor pollutants (Komarnicki, 2005;Chen and Zhao, 2011).The infiltration rate of outdoor particles into the indoor air is particle size dependent and largely affected by the ventilation system and the ventilation conditions (Chen and Zhao, 2011).Coarse particles are easily filtrated or become deposited on surfaces during the house air intake while fine particles have a high infiltration rate (Abt et al., 2000;Jones et al., 2000).On the other hand, the ultrafine particles have also low infiltration rates due to diffusion driven deposition during air intake (Chen and Zhao, 2011).
Both indoor and outdoor particles are complex mixtures of various chemical components.The coarse particles often contain mineral species from soil but also organic aerosol particles of biological origin (Thatcher and Layton, 1995;Hinds, 1999;Layton and Beamer, 2009).Instead, the fine particles typically contain elemental carbon (soot) from combustion sources, different metals and salts, secondary inorganic components and a large variety of organic compounds of both natural and anthropogenic origin (e.g., Jones et al., 2000;Adgate et al., 2002;Sillanpää et al., 2006).From the health point of view, the most important are considered to be the carcinogenic organic compounds (such as polycyclic aromatic compounds PAH), airborne microbes, transition metals (Lighty et al., 2000) and insoluble minerals (Schwarze et al., 2006).The penetration of these species in respiratory tract is strongly particle size dependent and therefore a size dependent characterization of aerosol particles is essential in the assessment of their potential health outcomes.
Microbes are present in both indoor and outdoor air; Cheng et al. (2009) reported that fungal spores contribute to 0.1-1.2% of organic carbon in airborne particulate matter.The main contributors to the outdoor air microbial content are soil, plant litter and phylloplane in addition to local sources, such as composts, waste-water plants, and animal feces (Bowers et al., 2011).The microbial composition of the outdoor air not only exhibits prominent seasonal variation (Bowers et al., 2011), but varies greatly even on a daily basis (Fierer et al., 2008).Brodie et al. (2007) claimed that meteorogical factors were a stronger predictor of microbial communities in urban aerosols than geographical location.The microbial composition of the indoor air is influenced by both indoor and outdoor sources (Burge et al., 2002, Shelton et al., 2002).Indoor sources include mainly skin bacteria and yeasts shed by human and pets (Sciple et al., 1967).The outdoor air contribution depends on the ventilation type of the building, i.e., filtration of the incoming air reduces the microbial concentrations (Salonen et al., 2007).Furthermore, the type, age and heating system of the building as well as the indoor temperature can influence indoor microbial levels (Chew et al., 2003;Bartlett, et al., 2004;Wu et al., 2005).
The aim of this study was to characterize respirable particles simultaneously in indoor and outdoor air in order to understand differences in their health-related properties.The size-segregated particle samples were collected simultaneously indoors and outdoors during each season of a year in a brand new detached house with its inhabitants and the samples were characterized for chemical, microbial and toxicological properties.This paper describes the sampling campaigns and the results from chemical and microbial analyses, including a variety of metals, inorganic ions, PAH compounds as well as fungal and bacterial concentrations and flora.The same particulate samples were analysed for toxicological properties in a cell line and the results are published by Happo et al. (2013).Since the collection of sufficient size-segregated particulate material (preferably at least 40 mg) for such extensive analyses is problematic directly from indoor air, a new sampling approach with high volume impactor (HVCI) collections from the house ventilation exhaust was utilized for the first time.

Sampling Sites and Study Design
The PM sampling was carried out in a new single family house in a suburban area of the Finnish city of Kuopio with around 95000 inhabitants.The house has a wooden frame with wooden covering with total area of 140 m 2 (containing five rooms and a kitchen).The ventilation system is arranged via a mechanical exhaust and air supply (Ilto 440) with one air intake which distributes intake air into all rooms.The air intake includes a EU7 class air filter.The air exchange rate in the house was set to 0.64 (when windows and doors are closed).The house is connected to the district heating network.In addition, there is a heat recovery system in the mechanical ventilation and one wood-log fired masonry heater which has been installed in the living room of the house.
The local pollutant sources in the neighborhood include houses with domestic-scale biomass-fired combustion appliances, nearby streets with very low traffic volumes, starting of the car on the yard, earthworks in the neighboring lots and point sources.The possible in-house PM sources include typical household tasks and activities, e.g., frying food, sauna heating (an electric stove), cleaning and heating of the masonry heater.The kitchen was equipped with a hood in order to remove the fumes formed during cooking.The residents (two adults, one child and three cats) moved into the house at the end of December 2009.The residents were asked to keep a diary of their activities during the PM sampling campaigns in order to identify any potential exceptional inhouse aerosol sources that could have influenced the particle concentrations and their characteristics.
The collection of sufficient amount (preferably at least 40 mg) of particulate material from each of the particle size class for the chemical, microbial and toxicological analyses is challenging.With relatively clean indoor air this requires a high volume sampling which easily affects the "natural" flow dynamics of the house.To solve this problem the PM samples were drawn for the high volume impactor from the main air exhaust line before the heat exchanger (Fig. 1).The sampled air was returned back to the exhaust line to avoid changes in the ventilation rate.The impactor used is a modified configuration of the Harvard high-volume cascade impactor (HVCI) (Sillanpää et al., 2003) that enables a high efficiency size-segregated collection of particulate mass for toxicological, chemical and microbial analyses.The modal pattern of both ambient and indoor aerosol mass size distribution was considered in the selection of the particulate size ranges for the HVCI (Sillanpää et al. 2003): coarse (PM 2.5-10 ; aerodynamic diameter 2.5-10 μm), inter-modal size range (PM 1-2.5 ; diameter 1-2.5 μm consisting, to a varying degree, of parts of coarse and accumulation mode particles), accumulation (PM 0.2-1 ; diameter 0.2-1 μm), and so-called ultrafine (PM 0.2 ; diameter < 0.2 μm) particles.In the three uppermost size ranges, particles were collected onto a polyurethane foam (PUF), while the collection of particles in PM 0.2 size range was on a Fluoropore Membrane Filter (3.0 µm FSLW) (Millipore, Ireland).The sampling instruments were placed in the attic in a separate room not connected to the house ventilation system.The sampling of the outdoor air was carried out in a transportable sampling station, which was located approx.50 meters north west from the house in a backyard area (no streets around).The sampling air flow rates with both indoor and outdoor measurements were adjusted to 850 L/min.The flow rate was checked daily by measuring the pressure drop over the impaction stage 2 and adjusted if needed.Each fraction was collected simultaneously both indoors and outdoors.

Sampling Campaigns
Table 1 shows the sampling periods of the PM collection.The sampling campaigns were performed during each season in 2010 after the residents had moved into the house.During the campaigns, PM samples were collected simultaneously from both indoor and outdoor of the house for 2-3 weeks.The meteorological conditions during the campaigns were followed by a weather transmitter (Vaisala WXT520) which was located on the roof of the house on top of a 2-meter mast.The measured weather parameters were temperature, relative humidity, barometric pressure, wind speed and direction, and liquid precipitation.During the winter campaign, a clear effect of residential wood combustion on both indoor and outdoor air aerosols was expected.During the summer campaign, the ambient air quality in the Eastern Finland was affected by long distance transported smoke from wildfires burning in Russia (Portin et al., 2012).

Sample Weighing and Extraction
The sampling substrates were washed with methanol and dried after which they were allowed to stabilize at room temperature overnight before weighing in an analytical balance, having a sensitivity of 0.01 mg (Mettler Toledo XP 105DR, Mettler Instrumente AG, Zurich, Switzerland).After the sampling of the particles from outdoor and indoor air, the weighing procedure was repeated to calculate the collected mass; the frozen samples were allowed to stabilize first in the weighing room for a few hours in closed containers and overnight in open containers before weighing.The relative humidity, temperature and barometric pressure in the weighing room were recorded and the filter masses were corrected for any change in the buoyancy of air.
The HVCI samples of each campaign were extracted and pooled together with separate particulate size ranges.However, the samples in PM 1-2.5 and PM 0.2-1 size ranges were combined to form one pooled PM 0.2-2.5 sample to obtain a sufficient particulate mass for all analyses (chemical, microbial, toxicological).The size-segregated particulate samples were prepared for the chemical, microbial and toxicological analyses using previously validated procedures (Jalava et al., 2005(Jalava et al., , 2006)).In brief, the sampled PUF-strips or PTFE filters were placed in 50-mL glass tubes and filled with methanol (J.T. Baker HPLC grade, Deventer, The Netherlands).The samples were extracted 2 × 30 min in a waterbath sonicator at a temperature not exceeding 35°C.Subsequently, the methanol extracts from the samples or the blank filters were concentrated in a rotary evaporator.

Chemical Characterization of PM Samples
The particle samples were analysed for elemental composition (Ag, Al, As, B, Ba, Be, Bi, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Rb, Sb, Se, Si, Sr, Th, Ti, Tl, U, V and Zn) in an inductively coupled plasma-mass spectrometer (ICP-MS) and for water soluble anions (SO 4 , NO 3 , Cl, PO 4 , Br and F) by ion chromatography (IC).The analyses were carried out by Labtium ltd (Espoo, Finland).In the ICP-MS analysis, the samples were first dissolved in nitric acid in a glass tube and transferred thereafter into a polypropylene tube, where fluoric acid was added.
Analyses of polycyclic aromatic hydrocarbon (PAH) compounds were conducted in the Fine Particle and Aerosol Technology Laboratory (Department of Environmental Science, University of Eastern Finland) by using a gas-chromatograph mass spectrometer (6890N GC, equipped with 5973 inert Mass Selective Detector, Agilent Technologies) and a HP-17-MS column for the separation of the compounds.The equipment was operated in the selected ion monitoring (SIM) mode.A total of 30 PAH compounds were analysed from samples extracts.The samples were extracted and the analysis was carried out as described by Lamberg et al. (2011, Supplemental content).An external standard mixture comprising of 30 commercially available PAH compounds as well as the internal standard were used for identification and quantification of PAH compounds in particle samples.The quantification limit was 0.1 ng/mg.The sum of all detected PAH compounds, that of the known genotoxic PAH compounds (WHO, 1998) and that of the PAH recommended to be measured from the outdoor air PM 10 in Europe (Directive 2004/107/EC) were calculated (see Table 3 for the compounds).Blank samples were prepared by inserting and removing substrates from the impactors and extracting the filters similarly as all the other particulate samples.The blank samples were analysed for inorganic species and PAH.Finally, blank corrections were made for ICP-MS results whereas in the ion chromatography and PAH analyses the blank concentrations were close to detection limit and therefore negligible.

Microbial Characterization of the PM Samples
DNA was isolated from the samples using NucleoMag 96 Plant (Macherey-Nagel, Duren, Germany) DNA isolation kit and the KingFisher mL automated DNA isolation station (Thermo Fisher Scientific).Briefly, the dried samples were dissolved in 500 µL of MC1 buffer from the DNA-isolation kit, transferred into sterile 2 mL screw cap tubes (Simport Plactics, Canada) containing 0.5 g glass beads (G-1277, Sigma-Aldrich, Germany).A total of 2 × 106 conidia of Geotrichum candidum in 0.5% Tween 80 were added to the tubes as the internal control.The tubes were shaken 1 min in Bead-Beater (Biospec Products, USA) at maximum speed to disrupt the cells and subsequently incubated 30 min at 56°C.After clearing the lysate by centrifugation at 5600 xg, the supernatant was transferred into the Squarewell block and the DNA was purified using the automated KingFisher mL magnetic particle processor according to the kit manufacturer's instructions.Purified DNA samples were stored at -80°C until analysis.Concentrations of two fungal and two bacterial groups and all bacteria were determined using qPCR.The qPCR assays applied have been previously published and were designed to detect the fungal genera Penicillium/Aspergillus/ Paecilomyces variotii (PenAsp), the fungal species Cladosporium cladosporioides (Clad1), the bacterial genera Mycobacterium spp.(Myco) and Streptomyces spp.(Strep) and the universal bacterial assay (Haugland et al., 2004;Rintala and Nevalainen, 2006;Kärkkäinen et al., 2010;Torvinen et al., 2010).The internal standard was monitored using the assay described by Haugland et al. (2004).The qPCR reactions were prepared in 96 well optical reaction plates (Applied Biosystems, USA) and contained "Universal Master Mix" -a proprietary mixture of AmpliTaq Gold™ DNA polymerase, AmpErase® UNG, dNTPs, passive reference dye and optimized buffer components (Applied Biosystems), the appropriate primers and probes in concentrations given in the original publications, bovine serum albumin (BSA) at a final concentration 0.2 mg/mL (New England BioLabs, USA) and nuclease free water, and then 2 µL of sample DNA were added.The qPCR cycling was done using either Rotor-Gene RG-3000 instrument (Corbett Life Science, Australia) or ABI Prism 7000 (Applied Biosystems) instrument under the cycling conditions given in the original publications.Finally, the results were calculated using the comparative threshold method described (Haugland et al., 2004).

Statistical Analyses
All the detected chemical and microbial concentrations were analyzed with Spearman's rank correlation (two-tailed) to examine the relationships between constituents.Correlation coefficients (rho) between the different variables were regarded as being statistically significant at *p < 0.05 and **p < 0.01 levels.All the data were analyzed by using IBM SPSS statistics 19.0 (IBM®, New York, NY).

Seasonal Characteristics of the Sampling Campaigns
Table 1 shows the measured weather parameters that reflect the overall meteorological conditions during the campaigns.During the winter campaign, the weather was typical for late winter and the conditions reflect the presence of an inversion layer with poor mixing.The average daytime mixing layer height, based on model calculations by the European Centre for Medium-Range Weather Forecasts (ECMWF), was 619 m.There was snow on the ground during the whole campaign.The spring campaign was held in the late spring, with no snow on the ground.During that period, there were heavy rain showers which cleaned the air effectively of aerosols and pollen.During the summer campaign, the weather was dry and warm, and the average mixing layer height was relatively high (1242 m).The ambient air quality in eastern Finland was at that time affected by long distance transported smoke from several wildfires burning in European Russia (Konovalov et al., 2011).There were two intense smoke episodes (29.7.2010 and 8.8.2010) with PM 10 concentrations 3-6 times higher than at normal conditions detected at all air quality measurement stations in Kuopio (Portin et al. 2012).During the autumn campaign the weather was typical for late autumn, except for the rainfall, which was unusually low for the autumn.
The reported indoor activities during each season included regular cleaning and cooking of food.No exceptional activities were reported.However, during the winter campaign, the masonry heater was heated with wood logs during 7 days which may have acted as an important indoor source.With respect to local outdoor sources, earthworks in the garden around the house were reported during spring and summer.Since the outdoor measurement station was located further from the yard (20 m) this source is likely to have a greater effect on indoor values.

Size Segregated Particle Concentrations
The collected indoor and outdoor air volumes ranged from 17188 to 25718 m 3 , yielding size segregated particulate masses between 19 and 168 mg.The lowest particulate masses were obtained for the PM 0.2 fraction.The average outdoor particle concentrations during different seasons were from 8 to 15 μg/m 3 for PM 10 and 4 to 10 μg/m 3 for PM 2.5 (Table 2).They mainly correspond to relatively good air quality and are similar to those observed in earlier studies in clean rural and suburban areas as well as in Kuopio (Monn et al., 1997;Jones et al., 2000;Sillanpää et al., 2003;Lazaridis et al., 2006;Portin et al., 2012) and are generally lower than those observed in European urban environments (Gotschi et al., 2002;Komarnicki, 2005;Pennanen et al., 2007;Diapouli et al., 2011).The highest concentrations were observed during the summer campaign when the air quality was affected by wildfires burning in Russia (Portin et al., 2012).The PM 10 concentrations were generally dominated by coarse (PM 2.5-10 ) and intermodal (PM 1-2.5 ) size fractions (Fig. 2).During summer, clearly elevated concentrations were found not only in the intermodal and coarse size fractions but also in the ultrafine particle fraction.This is most likely related to the wildfire smoke plumes that were found to increase substantially both ultrafine particle concentrations and coarse particle concentrations all around the Kuopio region (Portin et al., 2012).In addition, the weather was exceptionally dry which probably elevated the coarse mineral dust concentrations.
The average indoor PM 10 and PM 2.5 concentrations during different seasons were 10-13 μg/m 3 and 4-10 μg/m 3 respectively (Table 2).These correspond to 76-122% of the outdoor concentrations and are generally among the lowest concentrations usually observed in indoor air studies (Monn et al., 1997;Gotschi et al., 2002;Lazaridis et al., 2006;Chen and Hildemann, 2009) which are a result of low outdoor concentrations but also indicate the relatively small impact from indoor sources.In addition, the filtration of the house intake air may decrease the contribution of coarse particles from outdoor aerosols when compared to naturally ventilated buildings.Similarly to the outdoor concentrations, the particle size distribution was dominated by coarse and intermodal size ranges and the highest concentrations were observed during the summer campaign (Fig. 2).

Inorganic Compounds
The concentrations of the analysed species in size classified particle samples are shown in Table 3 (outdoor) and Table 4 (indoor) and the compositions of various particle size fractions in Fig. 3.The analysed inorganic species accounted for 8-43% of the total particulate matter.The highest inorganic fractions were observed in the PM 0.2-2.5 samples of winter and summer outdoor aerosols (38% and 41%, respectively), in the wintertime indoor ultrafine sample (43%) and in the spring time outdoor coarse particle sample (39%).
In general, the compound-specific particle size distributions can be used to estimate the possible sources of the species.For example, so-called accumulation mode particles are mainly aged aerosols which are often thought to be attributable to long-range transport while the ultrafine fraction is thought to indicate a near-source.Thus, the measured PM 0.2 fraction is likely to contain aerosol particles from near-sources while PM 0.2-2.5 can be thought to contain more aged aerosols.However, for example particles above size of 1 μm may also contain coarse-type of particles from near-sources.Thus, the compound specific differences in different size fractions may reflect changes in the particle sources, but they cannot be considered as "pure" samples of distinct sources.
The coarse particle samples contained mainly a variety of different elements typical for soil minerals (Si, Ca, Mg, Al, Fe, K, Na) (Tables 2 and 3).Positive correlations between these species point to a similar source for most of these species (Supplemental information).The mineral dust concentrations were at their highest during spring and summer and clearly lowest during winter which is due to the ice and snow cover.During spring, the mineral concentrations are typically high due to street dust episodes (Hosiokangas et al., 1999).During summer, the weather was exceptionally dry, which was found to elevate the coarse particle concentrations also in several other air quality measurements stations in Kuopio.
In all of the PM 0.2-2.5 and PM 0.2 samples, the most dominant inorganic compound was sulphate (Tables 2 and  3) which is in agreement with many other ambient air studies (Sillanpää et al., 2006;Remoundaki et al., 2013).
The amount of sulphate was in clear excess in relation to the analysed metal cations which indicates that sulphates were most likely present as ammonium sulphate (Sillanpää et al., 2005) or as some other volatile form, e.g., sulphuric acid which is present in heavy fuel oil and diesel engine emissions (Maricq, 2007;Sippula et al., 2009b) or bound with organic anions e.g., as observed in aged biomass smoke (Gao et al., 2003).It is likely that the majority of the PM 0.2-2.5 and PM 0.2 inorganic fractions are secondary aerosols formed in the atmosphere but also near sources (e.g., combustion processes) and aerosol formation indoors cannot be excluded.The highest sulphate concentrations, both indoor and outdoor, were found during summer.In addition, the PM 0.2-2.5 and PM 0.2 fractions contained significant fractions of nitrate and several metals.
The seasonal variation in some of the analysed metals is shown in Fig. 4. The concentrations of K, Zn and Pb in PM 0.2-2.5 and PM 0.2 samples displayed the highest levels outdoors during summer and during winter.In addition, the concentrations of Ni and Fe were clearly highest in the summer outdoor samples.The elevated concentrations of these five metals during summer were likely to be attributable to the wildfire smoke plumes, especially K, Zn and Pb are known to be efficiently released in fine particles during biomass combustion (Sippula et al., 2007, Sippula et al., 2009a) while during the winter the main emission sources were probably biomass combustion for heating in the area.This is supported by the data from winter time      ultrafine fractions showing clearly the highest fractions of K (Fig. 3) which has been shown to be a good indicator of the biomass combustion in ambient air (Sillanpää et al., 2005).In contrast, in coarse particle samples, the highest metals concentrations (e.g., Fe, Al, Si, Ca, K, Na, Zn, V, Ni) were observed in spring outdoor samples which can be explained by the street dust suspended during spring.

Polycyclic Aromatic Hydrocarbons
The concentrations of analysed PAHs in the PM 10 particulate mass ranged from 71 ppm mass (summer indoor) to 1487 ppm mass (winter outdoor).The outdoor concentration of total analysed PAH was 3-10 fold higher in winter when compared to the other seasons and the outside temperature was found to display a trend with most of the PAH compounds outdoors (Figs. 5 and 6).Similar trends were found also for the known genotoxic PAHs and for the PAHs according to the EU Directive 2004/107/EC (Fig. 6).It must be noted that the winter, spring and autumn campaigns lack data on PAH content in the ultrafine fraction due to the fact that the sample amount was too small.However, based on analyses made from other campaigns, the PAH contents in the ultrafine fraction are very low compared to total PAH in PM 10 .In line with outdoor PAH concentrations, the highest indoor PAH concentrations were found during the coldest seasons (Fig. 6).However, the indoor total PAH, genotox PAH and EU PAH concentrations were only 21-78%, 26-59% and 26-48% of those measured outdoors, respectively.Among of the analysed PAHs, 8-31% were lower molecular weight compounds (two and three rings) with the highest fraction (39-54%) representing the 4-ring compounds.The most abundant PAH compounds found in the outdoor winter sample were Fluoranthene, Pyrene, Phenantrene, Benzo(a)anthracene and Benzo(a)pyrene (BaP) which are typical byproducts of combustion processes, e.g., residential heating and traffic (Bari et al., 2011;Oliveira et al., 2011;Kaivosoja et al., 2012).The observed wintertime PAH concentrations are comparable to other studies from areas with high influence from residential wood combustion (Hellen, et al., 2008;Bari et al., 2011).During winter, the outdoor BaP concentration in PM 0.2 -10 was 1.05 ng/m 3 , exceeding the annual target value given by EU (Directive 2004/107/EC) while during other seasons, its concentrations were clearly lower, between 0.1-0.34ng/m 3 .
Since the total analysed PAH as well as many individual PAHs seem to have a connection with outside temperature, the high PAH concentrations during winter are likely to have originated from near sources due to residential heating which has been observed also in other studies carried out in areas where residential heating is an important pollution source (Sanderson et al., 2004).Furthermore, the mixing layer height (Table 1) also correlates positively with the outside temperature and thus, there are two possible reasons for the higher PAH concentrations in cold weather; higher fuel consumption in small heating units and less extensive mixing of pollutants in the ambient air.Third, also photo-degradation of the PAH compounds plays an  important role (Jang and McDow, 1995;Marr et al., 2006).In summer, the outdoor PAH concentrations were clearly lower than during winter, despite of the presence of the wildfire smoke plumes.The low concentrations of PAHs in the summer samples are likely to be explained by photooxidation and degradation of the PAHs in the plume since the residence time from the wildfire area was estimated to be 1-3 days (Portin et al., 2012).The resistance of PAHs to sunlight varies largely, for example BaP is known to decompose rapidly (Esteve et al., 2004).The diagnostic ratio BeP/(BaP + BeP) displayed clearly lower values in the summer season (0.58 and 0.55 for outdoors and indoors respectively) than during other seasons (0.62-0.71) indicating greater aging of the PAHs due to photolysis in summer (Tobiszewski, 2012).Furthermore, in the outdoor samples the sampling temperature also can affect the amount of lower molecular weight PAHs in the sample since these compounds exist in both particulate and gas-phase and at higher temperatures less of these compounds will be collected on the impactor substrates.
The major part of the PAH compounds were found in the PM 0.2-2.5 fraction.However, the distribution in different size fractions seemed to be dependent on the vapour pressure of the compound.The heavier compounds were mainly found in the PM 0.2-2.5 and PM 0.2 fraction while the lighter compounds were present also in coarse particles but almost absent in ultrafine particles.This might be due to several reasons.First, low molecular weight PAHs have been found to be easily transferred to the coarse particle fraction in ambient air (Benner et al., 1989).Second, a measurement artifact can be caused by adsorption of the gaseous PAH compounds in the used coarse particle substrates (Saarnio et al., 2008).

Microbial Concentrations
The size-segregated concentrations of microbial species are shown in Tables 3 and 4. The measured fungal genera Penicillium and Aspergillus (PenAsp) are the most common fungal species in indoor environments and the species Cladosporium cladosporioides (Cclad) is a common outdoor air fungus (Miller et al., 1988;Ren et al., 1999).The bacterial genera Streptomyces (Strep) and Mycobacterium (Myco) are common in soil and thus, while they are present in the outdoor air, they have also been commonly found in indoor environments (Rintala et al., 2004, Torvinen et al., 2010).
The concentrations of fungal and bacterial species both in indoor and outdoor air varied greatly depending on the microbe, season and size fraction (Tables 3 and 4).The highest concentrations were measured in the coarse (PM 2.5-10 ) fraction, with some exceptions regarding the bacterial genus Streptomyces; its concentration peaked in the PM 2.5 fraction in the outdoor air samples and in the indoor air spring sample.This is an expected result, since the spore sizes of the measured fungi lie in the size range of the coarse (PM 2.5-10 ) fraction (Samson et al., 2010).Bacterial spores and cells are usually smaller, ranging from 1-5 µm.However, both fungi and bacteria can form spore chains or cell clusters or they can become attached to larger particles, such as clay or other soil particles, plant fragments or skin scales shed from human and animals (Tham and Zuraimi, 2005).All of the measured microbial groups were sporadically detected in the smallest size fraction (PM 0.2 ), which indicates that microbial fragments smaller than intact cells or spores are continually present in the indoor and outdoor air.This is supported by the findings of Gorny et al. (2002) who showed that cell fragments considerably smaller than spores can be released from fungal contaminated surfaces.Our results further demonstrate that these fragments can also include DNA.It is conceivable that microbial DNA were found in all size fractions, since the qPCR method detects microbial DNA, which can be inside of spores or cells or hyphal fragments or even smaller fragments of disrupted cells.
The microbial concentrations measured ranged from 1-80 CE/m 3 in the coarse fraction, 0.3-60 CE/m 3 in the intermodal fraction and from below the detection limit to 1 CE/m 3 in the ultrafine fraction.For comparison, Meklin et al. (2007) reported average C. cladosporioides concentrations of 103 and 165 CE/m 3 in filter samples collected from indoor and outdoor air of US residential homes, respectively.Furthermore, in the study of Kaarakainen et al. (2008), outdoor air samples collected in a rural and an urban area in Finland contained about 101 to 104 CE/m 3 of C. cladosporioides, between 103 to 104 CE/m 3 of Penicillium/ Aspergillus and from the detection limit to102 CE/m 3 of Streptomyces spp.As far as we are aware, this is the first paper to describe microbial concentrations from parallel size-fractionated high-volume impactor samples from indoor and outdoor air.Therefore the present results cannot be directly compared to previous data which has originated from analysis of filters representing all the fractions together.Thus, the somewhat lower microbial concentrations detected in the present study compared to previous data is most probably due to the methodological differences in sample collection and processing.In addition, the sampling time of the filter samples was shorter than for the high volume impactor samples collected in this study and thus, more prone to temporal variation.
A seasonal variation was observed for all microbial groups measured, and they also exhibited differences between outdoor and indoor air.The outdoor concentrations peaked in summer for all microbes in the PM 2.5-10 fraction.In the PM 0.2-2.5 fraction, the highest concentrations were measured in spring (Cclad, Myco, Bact), summer (Strep) or autumn (PenAsp).The lowest concentrations were measured in winter for both of these size fractions.This finding is in line with previous culture-based studies indicating that viable microbial concentrations in outdoor air are at their lowest in winter (Reponen et al., 1992).This view is also partly supported by culture-independent studies measuring both viable and non-viable microbes which showed lowest concentrations of Streptomyces spp.and C. cladosporioides in outdoor air in winter, but not for the Penicillium/ Aspergillus group (Kaarakainen et al., 2008).Moreover, other culture-independent studies have shown that the microbial content of the outdoor air can vary greatly even on a dayto-day basis as the aerosols are transported with the air currents (Fierer et al., 2008;Fahlgren et al., 2010).
In contrast to the outdoor air, in the indoor air PM 2.5-10 the concentrations of all bacteria and Mycobacterium spp.did not reveal any major seasonal variation.The concentrations of the fungi were highest in spring or summer and Streptomyces spp. in winter.In the PM 0.2-2.5 fraction all but Cclad concentrations were the highest during winter while C. cladosporioides concentrations did not display any seasonal variation.The indoor PM 0.2 fraction did not reveal seasonal patterns in the measured concentrations, and the overall measured levels were low, with many of the samples being below the detection limit.

Indoor-Outdoor Relationships
I/O ratios were calculated for different particle mass size fractions and for chemical and microbial constituents (Table 5,Figs. 4 and 5, see Supplementary information for more details).In the evaluation of the ratios, one must note that these represent average values for the whole campaign and no time resolved data on indoor/outdoor concentrations are available.Therefore, the identification of single sources is in most cases not possible but the major influencing factors may be identified from the data.
The coarse particle (PM 2.5-10 ) I/O ratios were between 0.64-1.84,with the lowest value being encountered during summer.The coarse particle fractions were rich in mineral species typical for earth dust and the highest concentrations of these species (e.g., Si, Ca, Al, Fe, K, Mg, Zn) were observed during spring in both indoor and outdoor air, indicating that these minerals had been transported from outside.However, the transport of these species is not necessarily infiltration of the aerosol but might be also due to resuspension of deposited material, e.g., on shoes and clothes.Due to this phenomenon also the element specific I/O ratios vary extensively, and are for some elements higher than unity.
The I/O ratios for the size fraction PM 0.2-2.5 were similar during all seasons, ranging between 0.68 and 0.84.The I/O ratios of the total inorganic fraction were 0.36-0.51and for sulphate (which is the main inorganic component) 0.34-0.50.In each of these cases, the lowest value was found in the wintertime which is consistent with the fact that during winter the windows are not kept open for long periods.The fine particulate sulphate has been proposed to be used as a marker for outdoor contribution to indoor fine particles (Sarnat, 2002) which would indicate approximately 30-50% contribution of outdoor air for indoor aerosols in the respective particle size range.An examination of the other single elements shows large differences in I/O ratios.Mycobacterium, Bact = all bacteria na = not analysed nd = not detected due to concentrations below detection limit However, ratios higher than unity were observed only for some elements (Cu, Mg, Cl, F, K) and the concentrations of these species were relatively small.
The I/O ratios for the PM 0.2 fraction were between 1.11 and 1.85 and for the PM 0.2 sulphate between 1.11 and 2.38.These values indicate an indoor source for ultrafine particles including also a source for sulphate which is traditionally thought to be solely of outdoor origin, but has been also earlier observed in excess in indoor air (Adgate et al., 2007).Since the I/O ratios above unity for ultrafine sulphate were observed during every season, it is unlikely that they are being caused by some random sources, but unequivocal identification of this source would require more studies.
The concentration of ultrafine particles is determined by the competition between new particle formation and loss processes.The main loss processes are deposition and scavenging onto the larger particles.In the case when a fraction of the larger particles are lost during outdoorindoor transport, resulting in a value below unity for the I/O ratio for larger particles, this can cause the coagulation sink indoors to be less than outdoors.At the same time, the precursors for new particle formation may not be lost as effectively (or there can be additional sources indoors).This may also explain the higher concentrations of ultrafines in indoors compared to the outdoor values.
During the winter campaign, the fireplace was used in the house seven times but no exceptionally high I/O ratios of wood combustion markers, such as K and Zn, were detected in the ultrafine fraction, nor were high I/O ratios of PAHs observed.This is in line with observations of Lévesque et al. (2001) stating that wood-burning appliances inside the house often only make a very small contribution to the indoor air pollutants.
The I/O ratios for PAHs were mainly below unity and the observed dependencies of outdoor temperature with both outdoor and indoor PAHs (Fig. 6) point to outdoor air as being the main source of indoor PAHs.However, in the summer campaign, I/O ratios were at their highest, possibly due to the fact that windows were kept open.Furthermore, I/O ratios clearly above unity were observed for light PAH compounds indicating indoor-sources, such as frying, or near-sources in the yard.The I/O ratios of the particle size classified PAHs vary extensively but this is likely to be caused by volatility of these compounds i.e., they are present both in particulate and gaseous phases and can be easily transported between different particle size classes.
The indoor/outdoor ratio was above unity in the winter PM 2.5-10 and PM 0.2-2.5 fractions for all microbes.In the summer, the ratio was < 1 for all microbes in both of these size fractions.During other seasons and in the PM 0.2 fraction, the results were mixed (Table 5).The higher ratio in winter indicates that all the microbes also had indoor sources, such as the occupants and their pets, firewood or household activities.Since the family moved into the house before the winter campaign, it is also possible that the microbes were carried into the building with the furniture and other belongings and were gradually removed by cleaning.The probable reason for the low I/O ratios in the ultrafine fraction is the fact that ultrafine particles are not similarly deposited and resuspended as coarse particles, and therefore they are not as greatly affected by human activities (Thatcher and Layton, 1995;Chen and Hildemann, 2009).

CONCLUSIONS
High volume impactor sampling simultaneously from outdoor and indoor air was successfully carried out to provide particle size segregated samples for chemical, microbial and toxicological analyses.The high volume sampling system that is earlier used in outdoor air studies was found to be suitable also for collection of relatively large amounts of particulate mass from indoor air; it is specially suited for houses with mechanical ventilation.This makes it possible to study the relationships between particulate composition and their toxicological properties also in indoor air.In the present study, the PM 10 particle mass concentrations indoors and outdoors were nearly at the same level but there were clear differences in their chemical and microbial contents between particle size ranges, seasons as well as between indoor and outdoor air.
The coarse particles contained the highest amount of metals that are typical for soil dust as well as the highest microbe concentrations.The soil dust metals were at highest levels both outdoor and indoor during warm and dry seasons (spring, summer).
The microbial outdoor concentrations were also the highest during warm seasons while indoor concentrations did not show as clear variations; in particular the total bacterial load and mycobacteria remained constant throughout the year indicating that these microbes have indoor sources.The PM 0.2 fraction contained only small amounts of microbial DNA fragments; the most prevalent microbial group being the bacteria in general.
The fine particles showed clear differences in respect to metal concentrations.Metals typically released during biomass burning (K, Zn, Pb, Fe) were elevated during the summer which is consistent with the wildfire smoke episodes occurring during this campaign.In contrast, in the ultrafine fraction (PM 0.2 ) the highest K concentration both outdoors and indoors were observed in the winter samples indicating contribution from nearby biomass combustion sources.The PAH concentrations were lowest during summer both outdoors and indoors showing that the observed smoke plume did not significantly contribute to the PAH concentrations, probably due to photodegradation of the compounds.In contrast, the PAH concentrations were the highest during the winter and both the outdoor and indoor concentrations seem to be connected with the outside temperature, indicating residential heating as a major source of PAHs in the studied site.
During each campaign, PM 0.2 concentrations indoors were higher than those outdoors.In contrast, in each campaign the particle size fraction PM 0.2-2.5 in indoor air was about 70-80% to that measured outdoors.The calculated indoor/outdoor ratios indicate that the majority of the PAH species are of outdoor origin, while majority of the microbes are of indoor origin in winter but of outdoor origin in summer and more or less mixed during the other two seasons.With inorganic species, extensive variation in the element specific I/O ratios was observed, but in total their concentrations were lower indoors than outdoors.Overall, the study shows that there are considerable differences in the contents of health-relevant species of indoor air and outdoor particles.

Fig. 1 .
Fig. 1.Arrangement of the particle sampling by a high volume impactor from the house ventilation exhaust.
to the EU Directive 2004/107/EC

Fig. 3 .
Fig. 3. Concentrations of inorganic species in different particle mass size fractions.

Fig. 4 .
Fig. 4. Concentrations and indoor/outdoor ratios of selected metals in different particle size fractions.

Fig. 5 .
Fig. 5. Concentrations and indoor/outdoor ratios of PAH compounds in PM 10 of different molecular sizes and compound groups.*Analyses from ultrafine fraction are missing in the winter, spring and autumn samples from outdoor air.

Fig. 6 .
Fig. 6.Concentrations of PAHs indoors and outdoors as the function of outside temperature.*Analyses from ultrafine fraction are missing in the winter, spring and autumn samples from outdoor air.

Table 1 .
Measurement campaigns, sampling volumes and respective meteorological conditions including the minimum (min), average (ave) and maximum (max) of the temperature (T) and relative humidity (RH), the cumulative rainfall (RF), the average daytime mixing layer height (MLH) and the prevailing wind directions (PWD).

Table 2 .
Average concentrations of particle mass size fractions and indoor/outdoor ratios.

Table 3 .
Concentrations of chemical and microbial constituents in outdoor particle size fractions.Total PM and chemical species are given in ng/m 3

Table 4 .
Concentrations of chemical and microbial constituents in indoor particle size fractions.Total PM and chemical species are given in ng/m 3