Microbial Assessment in A Rare Norwegian Book Collection: A One Health Approach to Cultural Heritage

Microbial contamination poses a threat to both the preservation of library and archival collections and the health of staff and users. This study investigated the microbial communities and potential health risks associated with the UNESCO-classified Norwegian Sea Trade Archive (NST Archive) collection exhibiting visible microbial colonization and staff health concerns. Dust samples from book surfaces and the storage environment were analysed using culturing methods, qPCR, Next Generation Sequencing, and mycotoxin, cytotoxicity, and azole resistance assays. Penicillium sp., Aspergillus sp., and Cladosporium sp. were the most common fungi identified, with some potentially toxic species like Stachybotrys sp., Toxicladosporium sp., and Aspergillus section Fumigati. Fungal resistance to azoles was not detected. Only one mycotoxin, sterigmatocystin, was found in a heavily contaminated book. Dust extracts from books exhibited moderate to high cytotoxicity on human lung cells, suggesting a potential respiratory risk. The collection had higher contamination levels compared to the storage environment, likely due to improved storage conditions. Even though overall low contamination levels were obtained, these might be underestimated due to the presence of salt (from cod preservation) that could have interfered with the analyses. This study underlines the importance of monitoring microbial communities and implementing proper storage measures to safeguard cultural heritage and staff well-being.


Introduc�on
Collec�ons in libraries and archives predominantly consist of paper materials, which due to their organic composi�on and hygroscopic behaviour are prone to microbial coloniza�on.By degrading paper-based materials, microorganisms can also contribute to the dis�nc�ve musty odour o�en associated with historical libraries [1], [2].To mi�gate the growth of mould, libraries and archives implement strategies such as maintaining efficient ven�la�on and keeping rela�ve humidity levels below 60%.However, challenges arise when climate control systems fail or cannot adequately handle sudden spikes in temperature and humidity.Emergencies like leaks or floods can also lead to microbial outbreaks and even if the immediate issue is resolved, the environment stabilized, and collec�ons cleaned, fungal spores may s�ll linger within the paper fibres.Several studies have so far iden�fied causa�ve agents for microbial biodeteriora�on on paper, but not considering the health impact of such contamina�on [3].Under sound environmental condi�ons and for immunocompetent hosts, most fungi are harmless.However, 19% of the species so far iden�fied in libraries and archives can cause various health effects [4].Besides spores and fungal remains, such as mycelia, which can cause allergic reac�ons, fungi also excrete exotoxins during their growth -mycotoxins -which can cause allergies, asthma other health-related issues among staff and employers [3], [5], [6], [7].
The Norwegian Sea Trade Archive (NSTA), housed in the University Library of Bergen, Norway (ULB) documents the ac�vity of private companies that traded stock and salted dry cod fish from the 16th un�l the middle of the 20th century [8].This unique collec�on, composed of 2311 items, mostly accoun�ng books, is included in UNESCO´s Memory of the World Register due to its cultural, historical, and economic significance.Historically, the fish was stored on the ground floor of the wooden buildings located along the quay at Bergen Port while the accoun�ng was done on the first floor.Because hea�ng the building was prohibited due to fear of fire, the environment was humid and cold leading to biological deteriora�on of the collec�on at that �me.Adding to this, in 2016 the HVAC system in the NSTA-ULB storage where this collec�on was housed was out of order and the environment became again ideal for microbial development.The books also exhibit soiling and a strong codfish smell.In the last decades, library workers have been repor�ng various skin, eye, and respiratory symptoms from contact with the collec�onsymptoms that could be associated with occupa�onal exposure to chemical contaminants but also to moulds, mycotoxins, and endotoxins (toxins of bacterial origin).As a result, access to this important collec�on is currently very restricted.
Aiming to understand what the cause of the manifested symptoms could be and characterize the microbial contamina�on around and inside the books of this important collec�on, we have joined exposure science and cultural heritage conserva�on, in a One Health approach [9].Using passive sampling, culture-based methods, and molecular tools, we aimed to iden�fy and quan�fy the microorganisms in both the books and the storage environment, evaluate the resistance of iden�fied species to azole-based fungicides, assess the presence of mycotoxins, and determine the poten�al health effects due to exposure through cytotoxicity analyses.

Sampling campaign
Sampling of the books was performed by vacuuming the surface of the pages and covers with a museumgrade vacuum cleaner (Muntz museum vacuum cleaner 555 MU with HEPA filter), having 8 cm squares of filter paper (coffee filters nº 4, Auchan, France) between the hose and the nozzle to capture the aspirated par�cles (FP samples).Before sampling, each filter paper was sterilized under UV radia�on for 1 hour inside a biological safety cabinet and kept in sterile bags [10].Sterilized coffee filter pouches, uncut, were also used to wrap the paper squares a�er sampling so that the collected dust would not electrosta�cally adhere to the sterile plas�c bags.Eight books in total (n=8) were sampled: four books showing clear signs of fungal coloniza�on and another four with no visual evidence of such deteriora�on.From each visually fungal-affected book (n=4), two samples were collected, for comparison: a first sample, focusing on the localized fungal colonies (samples FP×A -where × is the sample number), and a second sampling encompassing the rest of the book (samples FPxB), which summed up to a total of 12 samples collected from the books.With the vacuum s�ll on, so that the dust would not fall off, the nozzle was disconnected, and the filter paper sample was put inside the coffee filter pouch (with tweezers) and then placed inside a sterile bag.Both the vacuum cleaner hose and the tweezers were disinfected with 70% ethanol between each sample collec�on.One filter paper inside the coffee filter pouch was le� unused and served as a control sample.
To analyse the storage environment, electrosta�c dust collectors (EDCs), which are simple electrosta�c cloths (Swiffer), were used to perform a passive sampling [10].The EDCs, cut to 95 × 130 mm, were taped (Scotch tape) to the interior of a prin�ng paper bifolio (210 ×145 mm).The prepared EDCs samplers were sterilized the same way as the filter papers, by 1 hour UV exposure and stored in sterilized bags un�l applica�on.11 EDCs were placed in the storage room, and one was kept in the sterile bag and used as a control.
The EDCs were distributed uniformly over the storage area, placed open over shelves or cabinets, at c. 1.5 m high, to collect airborne dust, which is the one that contributes to human exposure by inhala�on.They were secured in place with scotch tape to prevent accidental moving and a warning sign "Do not touch" (in Norwegian) was placed next to them.EDCs were maintained in place for 31 days and then collected into individual sterile bags and shipped for analysis.The collec�on of the EDCs involved only the folding of the paper bifolio, avoiding direct contact with the interior.
All samples were analysed according to the diagram presented in Figure 1.
The same extracts were also employed for molecular detec�on of the selected fungal targets (Aspergillus sec�ons Circumdati, Flavi, Fumigati and Nidulantes) following the previously published procedures [12].Fungal DNA was extracted using the ZR Fungal/Bacterial DNA MiniPrep Kit (Zymo Research, Irvine, USA) following the producer's instruc�ons, and molecular iden�fica�on was accomplished by Real-Time PCR (qPCR) using the CFX-Connect PCR System (Bio-Rad).For each amplified gene, a non-template control and a posi�ve control (DNA obtained from reference strains kindly provided by the Mycology laboratory of the Na�onal Ins�tute of Health Dr Ricardo Jorge) were employed.

Iden�fica�on of microorganisms by NGS
The extracted DNA was used also to perform environmental metagenomics.The DNA concentra�on was determined by fluorometry with Quantus™ Fluorometer ONE dsDNA quan�fica�on kit (Promega, Madison, USA), according to the manufacturer's instruc�ons.Microbial communi�es were characterized according to published protocols [14], [14], [15].Bacterial communi�es were characterized by Illumina Sequencing technology for the 16S rRNA V3-V4 region.Metagenomic DNA was amplified for the hypervariable regions with specific primers and further re-amplified in a limited-cycle PCR reac�on to atach a sequencing adaptor and dual indexes.The prokaryo�c popula�on was characterized using the 16S V3 forward primer 341F 5ʹ-CCTACGGGNGGCWGCAG-3' and 16S V4 reverse primer 805R 5ʹ-GACTACHVGGGTATCTAATCC-3' [16], [17].Besides, the 16S target-specific sequences, the primers also contained adaptor sequences allowing uniform amplifica�on of the library with high complexity ready for downstream NGS sequencing on Illumina Miseq.The hypervariable regions were amplified for each sample by PCR, in a LifeEco Thermal Cycler (Bioer Technology, China), for a total volume of 25 µL, containing 10 µL of Bioline My Taq HS Mix, 5 µL of each primer (1 mM) and 5 µL of DNA.The PCR program consisted of 1 min of denatura�on at 95°C, followed by 35 cycles of denatura�on at 95°C for 15 seconds, annealing at 55°C for 15 seconds and polymeriza�on at 72°C for 10 seconds, and a final extension at 72°C for 2 min.For the eucaryo�c communi�es, the ITS3 of the nuclear ribosomal RNA genes was amplified using the following primers: (5'-CATCGATGAAGAACGCTG-3', ITS3_001R 5'-TCCTSCGCTTATTGATATGC -3'.The hypervariable regions were amplified for each sample by PCR, in a LifeEco Thermal Cycler (Bioer Technology, China), for a total volume of 25 µL, containing 10µL of Bioline My Taq HS Mix, 3.5µL of pool primer (10 mM) and 5µL of DNA.The PCR program consisted of 2 min of denatura�on at 95°C, followed by 35 cycles of denatura�on at 95°C for 30 seconds, annealing at 55°C for 30 seconds and polymeriza�on at 72°C for 20 seconds, and a final extension at 72°C for 2 min.Nega�ve controls were included for all amplifica�on reac�ons.The amplifica�on products were detected by electrophoresis in a 2% (w/v) agarose gel with a 100 bp DNA ladder and the gel was stained with Green Premium and visualized under UV light in a Bio-Rad Molecular Imager ® Gel Doc™ XR+ Imaging System.The amplified fragments were purified using the High Prep TM PCR Cleanup System according to the manufacturer's instruc�ons.Next, dual indexes and Illumina sequencing adapters were atached to both ends using the Illumina Nextera XT Index Kit (Illumina, San Diego, CA, USA), using 25 µL of 2X KAPA HiFi HotStart Ready Mix, 10 µL of H 2 O RNase Free, 5 µL of Illumina Nextera XT Index Primers 1 (N7XX), 5 µL of Illumina Nextera XT Index Primers 2 (N5XX) and 5 µL of amplicon PCR product purified, for a total of 50 µL.The PCR index program consisted of a 3-minute denatura�on step at 95°C, followed by 8 cycles of amplifica�on: denatura�on at 95°C for 30 seconds, annealing at 55°C for 30 seconds, and polymeriza�on at 72°C for 30 seconds, and a final extension at 72°C for 5 minutes.The metagenomic libraries/ Index PCR products were detected by electrophoresis in a 2% (w/v) agarose gel with a 100 bp DNA ladder.The amplicon products were subsequently purified using the HighPrep™ PCR Cleanup System, according to the manufacturer's instruc�ons.The library concentra�on was determined by fluorometry with Quantus™ Fluorometer ONE dsDNA quan�fica�on kit (Promega, Madison, USA), according to the manufacturer's instruc�ons.Libraries were normalized and pooled to 4 nM.Pooled libraries were denatured and diluted to a final concentra�on of 10 pM with a 15% PhiX (Illumina) control.Sequencing was performed using the MiSeq Reagent Nano Kit V2 in the Illumina MiSeq System.Samples sequencing was performed using a 2 × 250 paired-end (PE)configura�on; image analysis and base calling were conducted by the MiSeq Control So�ware (MCS) directly on the MiSeq instrument (Illumina, San Diego, CA, USA).The forward and reverse reads were merged by overlapping paired-end reads using the AdapterRemoval v2.1.5[18] so�ware with default parameters.The QIIME package v1.8.0 [19] was used for Opera�onal Taxonomic Units (OTU) genera�on, taxonomic iden�fica�on, and sample diversity and richness indexes calcula�on.Sample IDs were assigned to the merged reads and converted to fasta format (split_libraries_fastq.py, QIIME).Chimeric merged reads were detected and removed using UCHIME [20] against the Greengenes v13.8 database [21] for V3-V4 samples.OTUs were selected at a 97% similarity threshold using the open reference strategy.First, merged reads were pre-filtered by removing sequences with a similarity lower than 60% against Greengenes v13.8 databases.The remaining merged reads were then clustered at 97% similarity against the same databases listed above.Merged reads that did not cluster in the previous step were again clustered inOTU at 97% similarity.OTUs with less than two reads were removed from the OTU table.A representa�ve sequence of each OTU was then selected for taxonomy assignment.

Analysis of mycotoxins
Sample prepara�on and chromatographic analysis of mycotoxins followed the procedure outlined in Viegas et al. [22].In summary, 0.10 g of dust collected from the books were subjected to vigorous shaking for 60 minutes, using 3.0 ml of an acetonitrile/water/ace�c acid mixture (79/20/1; v/v/v).A�er 5-minute centrifuga�on at 5000 rpm, 2ml of the extract was evaporated to dryness under a stream of nitrogen and then recons�tuted in a 400 μl of methanol/water mixture (2/8; v/v) and centrifuged again for 30 min at 14500 rpm.Thus, the sample dilu�on factor was 6.
The detec�on of mycotoxins was conducted using a high-performance liquid chromatograph (HPLC) system, specifically the Nexera model from Shimadzu (Kyoto, Japan), coupled with a mass spectrometry detector, the 5500 QTrap from Sciex (Foster City, USA).Mycotoxins were separated by chromatography on a Gemini C18 column (150 × 4.6 mm, 5 μm) manufactured by Phenomenex in Torrance, CA, USA.The flow rate was set at 1 ml/min, and a 5 μL injec�on volume was employed.
Two dis�nct mobile phases were u�lized: Phase A, comprising methanol/water/ace�c acid in a ra�o of 10/89/1 (v/v/v), and Phase B, consis�ng of methanol/water/ace�c acid in a ra�o of 97/2/1 (v/v/v).Both mobile phases were supplemented with 5 mmol/L of ammonium acetate.The chromatographic gradient proceeded as follows: ini�al elu�on with 0% B up to 2.0 minutes, followed by a linear increase to 50% B from 2.0 to 5.0 minutes, further ramping up to 100% B from 5.0 to 14.0 minutes, maintaining 100% B un�l 18.0 minutes, and ul�mately returning to the ini�al 0% B composi�on by 22.5 minutes.
Tandem mass spectrometry analysis was conducted in the scheduled mul�ple reac�on monitoring (sMRM) mode for both nega�ve and posi�ve polari�es within a single chromatographic run.The electrospray ioniza�on (ESI) source parameters were set as follows: a curtain gas at 30 psi, collision gas at a medium level, ion spray voltage at -4500 V (nega�ve polarity) and 5500 V (posi�ve polarity), ion source temperature maintained at 550°C, ion source gas1 at 80 psi, and ion source gas2 at 80 psi.Table S1 (Supplementary material) shows the instrument se�ngs op�mized for product ions of each compound.The Analyst 1.6.2so�ware (Sciex, Foster City, CA) was used for data acquisi�on and processing.
Cells were seeded on a 96-well micro�ter plate and incubated with 100 µl of the prepared sample dilu�ons per well for 48 hours at 37 °C in a humidified atmosphere with 5% CO2.Subsequently, MTT (3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) solu�on (20 μL) was added, and plates were incubated for another 4 h.The supernatant was then removed, and 100 μL dimethyl sulfoxide (DMSO) was added to each well.The forma�on of formazan was measured by spectrophotometric absorbance using an ELISA microplate reader (ELISA LEDETECT 96, Biomed Dr Wieser GmbH, Salzburg, Austria) at a wavelength of 510 nm (=maximum absorp�on wavelength of formazan deriva�ves).The lowest sample concentra�on dropping absorp�on to <50% of cell metabolic ac�vity (IC50) was defined as the threshold toxicity level.

Figure 3. Fungal contamination (MEA and DG18) median values from EDC (CFU.m-2.day -1 ) and filters (CFU.m -2 ).
There was a prevalence of Penicillium sp.followed by Cladosporium sp. in both EDCs and filter papers (MEA culture medium).Using the lower water ac�vity medium (DG18), there was a higher prevalence of Cladosporium sp. on EDCs, whereas in filter papers Aspergillus sp. was the most prevalent genera (Table 1).
Concerning an�fungal resistance, although Penicillium sp.(9.37x100 CFU.m -2 .day)and Cladosporium sp.(3.12x100 CFU.m -2 .day)from EDCs, and Aspergillus sec�on Nidulantes (3.12x100 CFU.m -2 .day)from filter papers were able to grow on the control SDA plates, no fungal growth was observed in the media supplemented with the four an�fungal agents.The Aspergillus sec�ons targeted by qPCR were not detected in the analysed samples.

Next Generation Sequencing: Bacteria and Fungi
Despite normal levels of extracted DNA, quan�fied in the pre-treatment phase, low amplifica�on rates were obtained.The number of OTUs for Procaryota (bacteria) in all samples was 553 and for Eucaryotes (fungi) it was 67, and their distribu�on is presented in Figures 5 and 6, respec�vely.
The procaryo�c diversity profile obtained for all samples was very similar and the number of reads obtained was both low and very similar amongst the samples and when comparing the samples (EDC1 to EDC12; PF 1 to PF8B) to the respec�ve controls (Figure 5).Nevertheless, the profile delivered a predominance of Ac�nobacteria, Proteobacteria and Cyanobacteria in all samples followed by the Firmicutes and other less represented phyla.Within these, it is possible to iden�fy Euryarchaeota, or Archae, known for their ability to thrive in extreme environments such as heavily salted ones [23].All sequences regarding the Procariota were deposited in NCBI under Bioproject PRJNA1071534.
Regarding Eucaryota, a sharp presence of Ac�nobacteria followed by Basidiomycota is observed in all samples (Figure 6).Unlike the results obtained for the Procaryota, however, one Eucaryo�c sample -PF6B -showed a dissimilar result, substan�ally different from the controls and the remaining samples (Figure 7).It was the case of a heavily contaminated book, in which the Ascomycota phylum accounts for 98% of the contamina�on.In terms of Class distribu�on, Dothidomycetes account for 43%, followed by Euro�omycetes (20%), Saccharomycetes (17%) and Sordariomycetes (11.4%).The Order distribu�on presents Botryosphaeriales (20%) followed by Eurotyales (19%), Capnodiales (18%), Saccharomycetales (17%) and Hypocreales (13%).When analysing Figure 7 it is possible to verify the presence of relevant genera in sample VF6B such as Stachybotrys sp., Toxicocladosporium sp. and Aspergillus, sp., all of them with possible health implica�ons [24], [25].The remaining samples (controls included) present a somewhat similar distribu�on (see figures 6 and 7) but the possibility that this could be the result of cross-contamina�on from sample FP6B cannot be discarded because this method is very sensi�ve and amplifies minute amounts of DNA through several cycles of amplifica�on.So, despite careful manipula�on of the samples, more research is needed to confirm this hypothesis.
Figure 7. Relative abundance of fungal genera in all samples.The lowest abundances were grouped under the name "Others".

Mycotoxins
Only one vacuum filter sample (FP6B -the same one showing a different fungal profile) presented posi�ve results.The mycotoxin detected was sterigmatocys�n (<LOD 2.2 ng).None of the EDC samples presented detectable contamina�on.

Cytotoxicity
The effect of vacuumed dust from books and EDCs' contaminants on cell viability was assessed by employing the MTT test on swine kidney (SK) and human lung epithelial (A549) cells (Table 2).These cells are relevant in vitro models for toxicological assessment of human exposure to biological contaminants [26], [27].
In A549 cells, filter papers ranged from low (25%) to high (42%) cytotoxicity, whereas EDCs showed low cytotoxicity (25%) only.No cytotoxicity was observed in swine kidney cells for any sampling method.

Discussion
The sampling and analy�cal techniques proposed in this study have been successfully tested by our team in se�ngs where microbiological contamina�on needed to be addressed for its possible impacts on human health [28], [29].It was, therefore, the first choice when assessing a se�ng where undiagnosed ailments keep affec�ng not only the well-being of the conservators-restorers that handle this collec�on but also pose a risk to the general public who wishes to consult this historic documenta�on.The collec�on -or part of it -could, therefore, be seen as an i�nerant hazard as symptoms seem to appear whenever it is manipulated.
The fact that part of this collec�on is visually contaminated by microorganisms sustained the hypothesis of a causa�ve biological agent.The results, however, did not confirm a high burden of contamina�on nor a great biological diversity in either the environment (EDCs) or the filter papers containing the par�culate collected from the books.This was true for both the tradi�onal culturing approaches and the DNA/PCRbased methodologies.
When looking for reasons for this occurrence -especially when contamina�on is so visible in the booksone cannot avoid wondering if the environment itself could have played a role since all the sampled documenta�on (and possibly the environment on which it rests) might s�ll contain salt, a permanent presence wherever cod is preserved.This being the case, then the microbial communi�es are probably adapted to high osmo�c pressure and salty environments and conven�onal analy�cal protocols do not consider specific environments because these tend to be rare and specific.By not supplemen�ng the culture media with NaCl and by not foreseeing its presence as a possible contamina�on agent for DNAbased strategies, we may have missed relevant informa�on.Thus, the obtained results may be underes�mated.
DNA is ubiquitous in DNA extrac�on kits and other laboratory reagents and the results obtained from samples containing a low microbial biomass (or low quan��es of successfully amplified target DNA) can be hampered by these contamina�on levels [30], [31].While the low number of reads was expected in the controls (given the amount of DNA encountered a�er quan�fica�on) the lack of amplifica�on within all the samples was surprising and may be related to the presence of PCR contaminants (salt is a possibility) especially within the sampled books.Salt is a known inhibitor of in vitro DNA amplifica�on techniques and might have hindered our efforts to obtain the full spectra for bacterial and fungal contamina�on.
The molecular biology protocols for both bacteria and fungi, despite their limita�ons, do point to the presence of halophilic bacteria, and Archaeobacteria both reinforcing the existence of an extreme environment.
It is less likely that the environment (assessed through EDCs) would be as affected by salt contamina�on.
Given the care presently being put on achieving proper storage condi�ons (temperature and rela�ve humidity control, protec�on of the shelves with contaminated books), and the constant precipita�on levels in Bergen (which are expected to lower the number of spores entering the archival storages) the obtained low contamina�on levels may, in fact, depict a low contaminated environment, as shown by both the metagenomics and the classic culturing methods used.
As for the sampled books, they show, on average, higher levels of contamina�on.Our results show thatas previously concluded by other studies [3] -the most common genera found in archives are the environmental fungi Penicillium sp., Aspergillus sp. and Cladosporium sp.Many of the species belonging to these genera are celluloly�c and monitoring is warranted to iden�fy possible surges in quan�ty and, being that the case, proceed to a new round of iden�fica�on.The metagenomic approach (for sample FP6B) confirmed the presence of Penicillium sp., Cladosporium sp., Aspergillus sp., and added Toxicladosporium sp. and Stachybotris sp. to the list of relevant genera.The highly toxic Toxicladosporium irritans was also found in the university library of Coimbra [6] and Aspergillus fumigatus was detected in several archives in Poland and Portugal [6], [7].Stachybotrys sp. is a black toxic mould associated with sick building syndrome [32] and has already been found both in the air and on paper in archives in Lithuania, Italy, Spain and Colombia [3], [5], [33], [34], [35].
Concerning the results obtained from culture-based methods we should not neglect the fact that assessing the viability of pathogenic and poten�ally pathogenic microorganisms is crucial when considering poten�al health effects and, therefore, reach a detailed risk assessment and iden�fy the most relevant risk management measures to implement [36].Gram-nega�ve bacteria were observed in the present analysis, and we should ponder the presence of several pathogenic bacteria that can be a threat to human health, such as E.coli, Klebsiella pneumoniae or Pseudomonas aeruginosa, just to name a few [37].In addi�on, and besides the more common fungal species, in the EDC and filter paper samples, we isolated Aspergillus sec�ons, all with toxigenic poten�al.In the filter samples we isolated Aspergillus sec�on Fumigati, listed by WHO as of cri�cal priority and proposed as a surrogate of harmful fungal contamina�on in different indoor environments [11], [38], [39].
Although species with toxigenic poten�al were isolated, mycotoxin contamina�on was low: only one filter paper sample (FP6B) presented contamina�on by a single mycotoxin -Sterigmatocys�n, produced by Aspergillus sp.Nevertheless, this scenario can change due to many aspects that influence mycotoxin produc�on such as the fungal species present and environmental condi�ons (e.g.temperature, humidity and availability of nutrients) [40].
Cytotoxicity was only observed in lung epithelial cells, being higher in the filter papers than in the EDCs.Since filter papers presented the highest bacterial loads and Aspergillus sp.prevalence, it can be hypothesized that the differences observed in cytotoxicity might be par�ally atributed to these contaminants.Moreover, Fumigati sec�on (which was iden�fied in filter papers but not in EDCs) is reported to have a cytotoxic effect on macrophages, due to the produc�on of gliotoxin [41].Besides gliotoxin, other toxins from Aspergillus sec�on Fumigati such as trypacidin, were also reported to be cytotoxic for lung cells [42].Several other studies corroborate that Aspergillus sec�on Fumigati present the highest cytotoxicity among Aspergillus species [43], [44], [45], [46], [47].Although not assessed in this study, the effect of par�culate mater and/or vola�le organic compounds cannot be excluded.
The fungal species tested for azole-based fungicides resistance, failed to grow on supplemented media, revealing no an�fungal resistance to the tested compounds.However, further studies in the scope of environmental surveillance regarding an�fungal resistance should be in place, including cultural heritage se�ngs.The changing climate has boosted the spread and acquisi�on of fungal diseases, leading to increased dispersion of fungi [48], forcing a higher use of azole-based fungicides and triggering the acquired azole resistance and the poten�al pathogenic fungi for humans, as well the toxigenic poten�al [49].

Conclusions
Under a One Health approach, we aimed to characterize the microbial burden in the books and storage environment of the NSTA-ULB, which due to microbial contamina�on is being kept from the public un�l further treatment deems it secure for handling.This approach was pivotal to ascertaining the risk of health effects, recommending appropriate measures in terms of protec�ve gear and disinfec�on, and also increasing our ability to effec�vely control and remediate the biodeteriora�on of the historic and cultural assets present in this Archive.
The analysis of the storage environment in the NSTA-ULB revealed low levels of contamina�on, low cytotoxicity, and no mycotoxins, which could be related to the environmental condi�ons atained in this se�ng, besides the climate condi�ons of Bergen.
The collec�on itself presented a higher contamina�on burden, as would be expected, given the visible (in some cases intense) microbial coloniza�on exhibited by the collec�on books.Besides the most common environmental fungi Penicillium sp., Aspergillus sp. and Cladosporium sp., which carry biodeteriora�on poten�al, we have also iden�fied Toxicladosporium sp., Stachybotris sp. and Aspergillus sec�on Fumigati all with high toxigenic poten�al.Also, the cytotoxicity on lung cells suggests a poten�al health risk for staff handling the collec�on.These results may explain why in the clean and stable environment of the University Library of Bergen the staff con�nues to experience health issues.This would jus�fy the need to eliminate microbial remains as much as possible from the affected materials.
The obtained results can, however, be underrepresen�ng the real scenario, as, despite the visually evident microbial coloniza�on on part of the studied books, we did not obtain a high burden of contamina�on nor a great biological diversity, either through culturing or DNA/PCR based protocols.Only one of the samples showed a substan�ally different fungal profile from the rest of the samples and the control samples, and it was also on that same sample that a mycotoxin was iden�fied.These results can point out poten�al chemical contamina�on of the samples with salt (salt is a known inhibitor of in vitro DNA amplifica�on techniques), due to the salt-rich environment of the premises where the Norwegian Sea Trade archive was originally kept.These results contribute to the growing realiza�on that cultural heritage objects can be considered extreme environments, to which the colonizing microorganisms are well adapted, and to which specific isola�on and iden�fica�on methods need to be applied.

Figure 1 .
Figure 1.Diagram of employed analyses for each type of sample.

Figure 4 .
Figure 4. Aspergillus sections distribution on MEA and DG18 from EDC and paper filters.No Aspergillus section was detected on MEA for the paper filters.

Table 1 .
Fungal distribution on MEA and DG18 from EDC and vacuum filter samples.

Table 2 .
Distribution of threshold toxicity (IC50) among human lung epithelial (A549) and swine kidney (SK) cells on EDCs and filter paper (FP) samples.