key: cord-0274920-wqceoqw1
authors: Bischofberger, Anna M.; Hall, Alex R.
title: Community Composition of Bacteria Isolated from Swiss Banknotes Varies Depending on Collection Environment
date: 2021-12-06
journal: bioRxiv
DOI: 10.1101/2021.12.01.470750
sha: e2f7de9e578a3a2163b81be011d4d65fed056c8f
doc_id: 274920
cord_uid: wqceoqw1

Humans interact constantly with surfaces and associated microbial communities in the environment. The factors shaping the composition of these communities are poorly understood: some proposed explanations emphasize the influence of local habitat conditions (niche-based explanations), while others point to geographic structure and the distance among sampled locations (dispersal-based explanations). However, the relative roles of these different drivers for microbial community assembly on human-associated surfaces are not clear. Here, we used a combination of sampling, sequencing (16S rRNA) and culturing to show that the composition of banknote-associated bacterial communities varies depending on the local collection environment. Using banknotes collected from various locations and types of shops across Switzerland, we found taxonomic diversity dominated by families such as Pseudomonadaceae, Staphylococcaceae and Streptococcaceae, but with banknote samples from particular types of shops (especially butcher shops) having distinct community structure. By contrast, we found no evidence of geographic structure: similarity of community composition did not decrease with increasing distance among sampled locations. These results show that microbial communities associated with banknotes, one of the most commonly encountered and exchanged human-associated surfaces, can reflect the local environmental conditions (in this case, the type of shop), and the signal for this type of variation was stronger than that for geographic structure among the locations sampled here.

Humans encounter microbial communities constantly and we now know that they contribute, 33 among other things, to food and drink quality (Wareing, of America (USA)) with the following settings: cycle nr = 2, strength = 5.5, time = 45s, interval 157 = 0. After transferring the supernatant to clean 2mL collection tubes, we added 250µL Solution 158 C2, vortexed the tubes for 5s at the highest amplitude and incubated them at 4°C for 5min. 159

After centrifugation at 13'000rpm for 1min at 4°C, we transferred 600µL supernatant into clean 160 2mL collection tubes and added 200µL Solution C3. We vortexed tubes briefly at the highest 161 amplitude and incubated them at 4°C for 5min, followed by centrifugation at 13'000 rpm for 162 1min at 4°C. Avoiding the pellet, we transferred 750µL supernatant into clean 2mL collection 163 tubes, added 1200µL well-mixed Solution C4 and vortexed tubes for 5s at the highest 164 amplitude. We loaded 675µL sample onto MB spin Columns, centrifuged them at 13'000rpm 165 for 1min at 4°C and discarded the flow-through. We repeated this until the entire sample was 166 processed. After adding 500µL Solution C5, we centrifuged samples at 13'000rpm for 1min at 167 4°C and discarded the flow-through, followed by centrifugation at 13'000rpm at 4°C for 30s. 168 We placed the MB Spin Columns into clean 2mL collection tubes and added 70µL Solution 169 C6 to the centre of the filter membrane. Following incubation at 65°C for 5min and 170 centrifugation at 13'000rpm at 4°C for 1min, we discarded the MB Spin Columns and stored 171 We quantified the obtained DNA using the Quant-iT TM dsDNA BR Assay Kit (product nr. 9 Diego, USA) after library preparation with the Nextera XT 96 Index kit v2, Set D (Illumina,  181 San Diego, USA). The sequences of the four sets of primers used during library preparation 182 (limited cycle PCR) are listed in supplementary table 1.  183 If not noted otherwise, raw sequence data were processed with USEARCH 184 (v11.0.667_i86linux64; Edgar (2010)). After end trimming reads ( To quantify within-sample diversity, we used the alpha() function in the microbiome 215 package (version 1.16.0) on filtered data to calculate several different diversity indices 216 (Shannon diversity index; Gini-Simpson index; observed richness; Chao1 index). The Shannon 217 diversity index H takes into account both richness and evenness; the Gini-Simpson index gives 218 the probabilty that two random draws from the same sample (with replacement) are not from 219 the same type (here, the same family); the observed richness counts the number of unique 220 species (ZOTUs) present in the sample; Chao1 index is a non-parametric estimator of species 221 richness that assumes Poisson distribution of the data (Chao, 1984; Chao and Bunge, 2002) . 222

We used analysis of variance (ANOVA) to test whether within-sample diversity varied among 223 locations or shop types (aov() function in the stats package (R version 4.1.3)). 224

Before analysing between-sample diversity and to account for different library sizes (Weiss 225 et al., 2017) , we normalized filtered data with a regularized-logarithm (rlog() function in the 226 DESeq2 package (version 1.34.0)), after comparing the two available transformations offered recommended for smaller datasets and for cases with large ranges of sequencing depths across 230 samples (Love et al., 2014) ; (2) it performed better in the meanSdPlot generated. 231

As a first test of whether samples from different locations and shop types differed in terms 232 of bacterial community structure, we used permutational multivariate analysis of variance 233 (PERMANOVA; Anderson, 2001; Zapala and Schork, 2006) with the adonis2() function in the 234 vegan package (version 2.5-7), with filtered-and-normalized data and using Bray-Curtis 235 dissimilarity (vegdist() function) to measure differences among sampled communities. Next, 236 to visualise any clustering among samples in terms of community composition, we performed 237 ordination of filtered-and-normalized data with the ordinate() function in the phyloseq package 238 (version 1.38.0), with non-parametric multidmensional scaling (method = "NMDS") and Bray-239

Curtis dissimilarity (distance = "bray") as input parameters. We used NMDS as the ordination 240 method because we obtained a stress value of 0.11 for our data with NMDS ordination; stress 241 values <0.2 generally indicate good fit (Clarke, 1993). We visualised ordination results with 242 the plot_ordination() function in the phyloseq package (version 1.38.0). We used a third type 243 of analysis as an additional test for geographic structuring of microbial community 244 composition, by conducting a Mantel test. This allowed us to account explicitly for the physical 245 distance between individual sampling locations (shops), rather than testing for average 246 differences among locations (as above in PERMANOVA). We did this using the mantel.test() 247 function in the ape package (version 5.5), with the vegdist() function in the vegan package 248 (version 2.5-7) for Bray-Curtis dissimilarity of filtered-and-normalized data. We used the 249 geodist() function in the geodist package (version 0.0.7) to calculate geographical distance (in 250 metres) between individual sampling locations. We used the same functions when repeating 251 the analysis for sample sets from individual shop types; here and elsewhere, we accounted for

The core microbiome is defined as the species present above a certain threshold and shared 254 among a set of samples (Salonen et al., 2012; Shade and Handelsman, 2012) . We determined 255 the core microbiome, based on ZOTUs, for each shop type as follows: Using filtered-and-256 normalized relative abundance data (transform() function in the microbiome package (version 257 1.16.0; "compositional" as input for transform argument) and subsetting data by shop type, we family Sphingomonadaaceae), Staphylococcus (n = 1; family Staphylococcaceae), Veillonella 339 (n = 1; family Veillonellaceae). This is consistent with our family-level analysis (Fig. 1) , where 340 some of the families to which these butcher-associated genera belong were over-represented 341 among butcher samples (e.g., Vibrionaceae, Lactobacillaceae). 342 A comparison between core ZOTUs unique to butcher samples and NMDS loadings (Fig.  343 3) provides further information about the taxa that drive the among-shop-type variation: the 344

Veillonella core ZOTU and one Streptococcus core ZOTU unique to butchers were among the 345 ten ZOTUs with the highest positive loadings for NMDS1; the Lactococcus and three 346

Photobacterium core ZOTUs unique to butchers were among the ten ZOTUs with the highest 347 positive loadings for NMDS2 (again consistent with families Vibrionaceae and 348

We found no evidence that samples collected in different locations (towns) harboured Fig. 1B Fig. 4) . The exception to this trend was Enterococcus faecalis (3/157 colonies or 2%), 381 which belongs to the Enterococcaceae family. This was not among the most abundant families 382 in our sequence data (Fig. 1B) , and we only found ZOTUs categorized as Enterococcaceae 383 prior to filtering the data, indicating it was present but at lower abundances than suggested by 384 colony formation data. For most of the other key taxa identified by sequence data that we did 385 not find on agar plates, prior information about these taxa suggests the growth conditions 386 (Chromatic TM MH agar; 37°C; 24h aerobic incubation) are not conducive to colony formation. 387 Therefore, the absence of these taxa from our cultured samples is probably because they do not 397 grow well in these conditions, and does not necessarily contradict our finding them in the 398 sequence data. 399

As above for our community-level sequence data, the relative abundances of different taxa 400 inferred from colony counts on chromatic agar varied among different shop types was collected is surprising: this suggests money has a microbiome reflecting its local 416 environment, even though we assume banknotes are only transiently exposed to each 417 habitat/shop type. By extension, we expect individual banknotes to have a dynamic 418 microbiome that changes during their "lifecycle" as they pass from one habitat to another. By 419 contrast, we found no evidence for geographic structuring, unlike some previous studies with 420 However if anything we would expect any effect of pandemic-related hygiene measures to make our conclusions conservative: when hygiene standards are less stringent we might expect 502 even more exchange of microbiota between the local environment and banknotes. A further 503 limitation of our study is the spatial scale and number of shop types. While our study design 504 revealed significant variation among shop types, the lack of geographic structure across 505

Switzerland may differ from that across larger spatial scales or across different currency types. 506

In summary, we found that the bacterial communities associated with Swiss banknotes can 507 vary depending on the environment in which the banknotes were collected and that, by contrast, 508 geographic distance between sampled locations does not predict variation of the associated 509 microbial communities. Our results also show that the most abundant taxa are not only present 510

in the form of cell debris and genetic material, but as viable cells. This raises the question of 511 banknotes potentially being involved in transmission of bacteria between people and the 512 environment, and/or of genes, such as antibiotic resistance genes, among bacteria co-occurring 513 on the same banknotes. Therefore, a key avenue for future work building on the link between 514 microbial community structure on banknotes and the local environment (shop type) identified 515 here is to ask whether this translates to variable infection risk or horizontal gene transfer. 

Leuconostocaceae Family, Second Edi

Moderated estimation of fold change and dispersion for RNA-seq 617 data with DESeq2

The Limiting Similarity, Convergence, and Divergence of Coexisting 619 Species

Phylogenomic analyses of the Staphylococcaceae 621 family suggest the reclassification of five species within the genus Staphylococcus as heterotypic 622 synonyms, the promotion of five subspecies to novel species, the taxonomic reassignment of five 623

Filthy lucre: A 625 metagenomic pilot study of microbes found on circulating

Moraxella catarrhalis, a human respiratory tract pathogen

The Halophilic World of Lourens Baas Becking

Bacterial diversity in typical Italian 632 salami at different ripening stages as revealed by high-throughput sequencing of 16S rRNA amplicons

The genus Lactobacillus

Influences of Ingredients and 637 Bakers on the Bacteria and Fungi in Sourdough Starters and Bread

Vibrio species pathogenic for human in seafood products consumed in France using real-time PCR

The adult intestinal core microbiota is determined 644 by analysis depth and health status

Characterisation of lactic acid bacteria isolated from 646 naturally fermented Greek dry salami

Upon the occurrence of bacteria and minute algae on the surface of paper money

Beyond the Venn diagram: The hunt for a core microbiome

Prevalence of salmonella, yersinia, aeromonas, campylobacter, and cold-growing escherichia coli on 653 freshly dressed lamb carcasses

Human skin microbiome: Impact of intrinsic and extrinsic factors on 656 skin microbiota

Estimating and mapping ecological

Survival of influenza virus 665 on banknotes

A "missing" family of classical orthogonal polynomials

Journal of Physics A: Mathematical and 668 Theoretical

Dirty money: A matter 670 of bacterial survival, adherence, and toxicity

Ecological role of Lactobacilli in the gastrointestinal tract: Implications for fundamental and 672 biomedical research

Microbiology of soft drinks and fruit juices

Exploring the Phylogenetic Structure of Ecological Communities: An Example for Rain Forest 676 Trees

Normalization and 678 microbial differential abundance strategies depend upon data characteristics

Geographic barriers isolate endemic populations of 680 hyperthermophilic archaea

Geographic barriers isolate endemic populations of 682 hyperthermophilic archaea

Multivariate regression analysis of distance matrices for testing 684 associations between gene expression patterns and related variables

Spatial scaling of functional gene diversity across 687 various microbial taxa

Stochastic assembly leads to alternative 689 communities with distinct functions in a bioreactor microbial community

Stochastic Community Assembly: Does It Matter in Microbial Ecology

The sequencing data have been deposited in the European Nucleotide Archive under the study 694 accession number PRJEB45390

Other data presented in this study will be made available at the Dryad Digital

Benefits Generated: Benefits from this research accrue from the sharing of our data and 699 results on public databases as described above