DR ANNE - MARIE T TOBIN (Orcid ID : 0000-0003-2730-9561)

Article type : Original Article

Accepted Article
Investigation of the Skin Microbiome: Swabs vs Biopsies

S. Prast-Nielsen1, A.-M. Tobin2, K. Adamzik2, A. Powles3, L. Hugerth1,4, C. Sweeney2, B. Kirby2, L.

Engstrand1,4, L. Fry3

Affiliations: 1Department of Microbiology, Tumor and Cell Biology, Karolinska Institute, Stockholm,

Sweden; 2Department of Dermatology, St Vincent’s University Hospital, Dublin, Ireland; 3Imperial

College, London, United Kingdom; 4Centre for Translational Microbiome Research (CTMR),

Karolinska Institute, Stockholm, Sweden.

Corresponding author: Anne-Marie Tobin

E-mail: [email protected]

Abstract

Background:

Human skin is populated by diverse bacteria and there is increasing evidence that resident bacteria

play a key role initiating immune responses in cutaneous diseases such as atopic dermatitis, psoriasis

and hidradenitis suppurativa. Bacteria are present at all layers of the skin but many studies have relied

on swabs to profile the skin microbiota. As the pathogenesis of many skin conditions is dermal, we

wished to compare the microbiota obtained in swabs (surface) and biopsies (dermis).

This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/bjd.17691
This article is protected by copyright. All rights reserved.
 Using 16S rRNA gene sequencing we established the microbial profiles of skin swabs and skin

biopsies in sixteen subjects. We found differences in both diversity and taxonomic composition of the
Accepted Article
microbiome obtained from swabs and biopsies of the same individual. Several taxa were found to be

more abundant in the swabs, which displayed significantly higher community richness, but

Clostridiales and Bacteroidetes were significantly enriched in the biopsies.

The majority of published research on cutaneous microbiota has been based on skin swabs, which

represent the surface of the skin. Our study demonstrated a clear difference between the microbiome

observed in skin swabs and skin biopsies. These findings may be highly relevant in disorders such as

psoriasis where pathogenesis arises in the dermis.

Keywords: skin, microbiome, 16S rRNA gene, skin swabs, skin biopsies

Introduction

The human skin microbiome is currently the subject of intense research to elucidate its role in the

pathogenesis of skin diseases such as atopic dermatitis, psoriasis and hidradenitis suppurativa. The

interaction between bacteria and elements of the skin’s innate immune system may be the initiating

event in such inflammatory dermatoses. This mirrors similar investigations that have assessed the role

of bacteria in the induction of inflammatory diseases of the gastrointestinal tract such as Crohn’s

disease.

The emergence of high throughput next generation sequencing of the bacterial 16S rRNA gene has

facilitated microbiome research by identifying otherwise unculturable bacteria. Numerous studies of

the skin microbiome have now been carried out and emphasis has been placed on the finding that the

microbiome varies between individuals and at different sites of the body1. The majority of studies

have been performed employing swabs taken from the surface of the skin however. The first of these

studies by Gao et al. found that superficial bacteria were highly diverse with high interpersonal

This article is protected by copyright. All rights reserved.

 variation2. In subsequent work Grice et al. used three different sampling strategies in the antecubital

fossa of five subjects: swabs, skin scrapes and punch biopsies3. The authors wished to ascertain the
Accepted Article
validity of skin swabs in collecting accurate, representative samples as this collection method was

most practical and least invasive. Their results confirmed that 16S rRNA gene sequencing captures

the diversity of swabs, scrapes and biopsies. The work also concluded that similar microbial

populations were captured by each technique and that the dominant species were present in the non-

invasive swabs.

However, when we carried out our initial study on psoriasis4 we chose biopsies. Initial events in the

development of psoriasis lesions commence in the dermis and a putative antigen, streptococcal

peptidoglycan, has been found in antigen presenting cells in the dermis5. In addition, Nakatsuji and

his colleagues6 have shown that there are bacteria present in the dermis and superficial adipose

tissue6. Our results in psoriasis using biopsies showed differences in the microbiome compared to

others using swabs7,8,9,10. The lack of standard procedures impedes comparison of independent study

results. Sampling technique, sample processing, sequencing methods and analysis procedures may

influence the study results11. It is critical that the sample collected is truly representative 3,6. To

elucidate the impact of sampling technique, we carried out a study comparing the skin microbiome

obtained by swabs and biopsies from the same individuals processed and analyzed identically.

SUBJECTS AND METHODS

Collection of normal skin samples

Sixteen subjects with non-inflammatory skin disease were studied: 11 were males and the age range

was 21 – 85 (Table 1). All patients were undergoing wide elliptical excision of skin lesions and none

had psoriasis or a past or family history of the disorder. No skin cleanser was employed and no

individual subsequently developed infection in the wound. The swab was moistened with sterile

This article is protected by copyright. All rights reserved.

 buffer. The swab was not rubbed too hard, but it was important to biopsy the same site. The

microbiome may vary at different sites. The biopsy was taken down to the subcutaneous fat. But fat
Accepted Article
was removed from the specimen if present.

Patients had not received antibiotics within four weeks of the procedure. First, a swab (Puritan,

Cambridge, UK) was taken from one end of the ellipse. Next, a 2 mm punch biopsy was taken from

the same site, employing the local anaesthetic 2% lignocaine with 2% adrenaline 1:200,000. The swab

and biopsy specimens were then placed into buffer (Master Pure Yeast Cell Lysis solution, Cambio,

Cambridge UK) in sterile tubes, snap frozen and stored at -80°C. All the specimens were taken from

the trunk and limbs: eight from dry and eight from sebaceous sites. All the subjects gave their

informed consent. The study was approved by St. Vincent’s Hospital Dublin Ethical Committee and

conducted according to the Declaration of Helsinki Principles. An aliquot of Master Pure Yeast Cell

Lysis solution was placed in a sterile tube and processed identically to the skin samples serving as a

negative control.

DNA extraction, library preparation and 16S rRNA gene sequencing

A total volume of 180 μl lysozyme buffer (20 mM Tris-HCL, pH 8,0, 2 mM EDTA, 1,2% Triton-X

and 20 mg/ml lysozyme, Sigma-Aldrich, Germany) was added to the samples and incubated at 37 °C

for 1 hour. Bacterial DNA extraction was then carried out using the DNeasy Blood and Tissue kit

(Qiagen, Germany) with proteinase K and AL buffer treatment for 16 h at 56 °C following the

manufacturer´s instruction. A final volume of 50 μl of elution buffer was used for recovery of

bacterial DNA from spin columns following extraction.

Sequencing libraries were prepared according to Illumina´s protocol (Preparing 16S Ribosomal RNA

Gene Amplicons for the Illumina MiSeq System). 16S rRNA gene sequencing (region V3-V4) was

performed using the 341f-805r primers described by Hugerth et al.11. After the initial amplification a

second PCR was performed to attach Illumina adapters as well as barcodes that allows for

multiplexing11. All samples were pooled equimolarly and sequenced in one MiSeq (Illumina) run.

This article is protected by copyright. All rights reserved.

 16S rRNA gene sequencing analysis

Read preprocessing, clustering and taxonomic identification

Accepted Article
Cutadapt 1.9.1 was used to remove all sequences lacking the amplification primers 13. The primer

sequences, all bases with a Phred score below 15 and all reads with less than 120 bp left after

trimming were also discarded. The resulting reads were merged with vsearch v2.3.0 and reads failing

to merge or producing merging products shorter than 380 bp, longer than 520 bp or with more than

three expected errors were excluded14. All unique full-length sequences occurring at a frequency

higher than 10-6 in the dataset, were submitted to Unoise2 for denoising and clustering at 100%

identity14. All the merged reads were then mapped back to the accepted centroids and assigned to the

OTU with the highest identity, at a minimum of 98%. In case of equally high identity matches, the

most abundant centroid was selected. Taxonomy was assigned using the SILVA database (Release

128_SSURef_Nr99)15(Quast et al. 2013). The taxonomic annotation of the database was manually

curated to remove uninformative species-level information (e.g. “soil metagenome”) as well as

obvious misclassifications (such as eukaryotic species classified as bacteria). For the classification

levels of strain, species, genus, class and phylum, identity cutoffs were set at 100%, 99%, 97%, 95%

and 90%, respectively. The top hit above each identity threshold was identified and all hits with

scores at least 95% of it and still above the cut-off were selected. The last common ancestor of these

taxa was then selected and truncated at the appropriate taxonomic level. Subsequently, the most

complete classification for each OTU was selected. Since the curated database does not include strain

information, this classification level is in practice not done, but kept to ensuring preference for exact

matches over 99%.

Scripts for performing this analysis are available at https://github.com/ctmrbio/Amplicon_workflows.

This article is protected by copyright. All rights reserved.

 Downstream analysis

The resulting OTU table was submitted to QIIME 1.9.1.17. Alpha diversity indices were calculated
Accepted Article
using the alpha_diversity.py command. The Shapiro-Wilk test of normality was used in R 3.3.2 to

estimate the distribution of alpha diversities followed by a paired t-test. A barplot summarizing

taxonomies at phylumn level was produced with core_diversity_analyses.py at a rarefaction level of

21,227 reads per sample. Statistical analysis of otu abundance at all taxonomic levels was performed

using the Wilcoxon's signed rank test for paired data in R. P-values for taxa with a false discovery rate

(FDR) < 0.1 are summarized in Table 2. The table presented is a summary of six statistical tests

performed at six different taxa levels, i.e. phylum (25), class (62), order (139), family (240), genus

(577) and species (1060). The numbers in thebrackets represent the number of different taxa at each

level and they influence

the multiple correction. Multiple correction was performed for all p-values at

each taxa level and therefore, different FDR values may result from similar p-

values depending on whether multiple correction was performed for e.g. 25 taxa

or 577. Hierarchical pie charts were produced with Krona Tools 2.718. NMDS analysis was performed

in R 3.3.2 using the vegan 2.4-3 library.

RESULTS AND DISCUSSION:

RESULTS:

Sequencing data set

Thirty-three sequencing samples (two of each subject and one negative control) resulted in 1,028,171

pre-processed and quality filtered reads with a minimum of 21,227 reads, a maximum of 42,521 and a

mean of 32,130 reads per sample clustering into a total of 3,822 observations.

This article is protected by copyright. All rights reserved.

 Diversity analysis

For alpha diversity (within communities), we calculated the number of observed OTUs, the Chao1
Accepted Article
index taking rare species, i.e. singletons and doubletons, into account as well as the Shannon index for

evenness. The Shapiro-Wilk test of normality demonstrated a normal distribution of all indices with

all p-values > 0.05. Thus, a paired t-test was performed demonstrating significantly higher community

richness for swabs compared to biopsies (Figure 1, left and middle panel, **p < 0.01). The Shannon

diversity index, which takes into account both richness and abundance, was not significantly different

between swabs and biopsies (right panel) suggesting that swabs are richer in species but these are

more evenly abundant in our dataset.

Beta diversity (between communities), estimated by Nonmetric Multidimensional Scaling using Bray-

Curtis indices, showed clear clustering of samples by sampling procedure (Figure 2). The clustering

was evident for both samples from dry and sebaceous skin.

Analysis of microbial composition

Microbial communities obtained from biopsies and swabs from the same individual were different

(Figure 3 at genus level and Figure 4 at species level as representative examples). Volunteer number

16 showed high abundance of Staphylococcus aureus in the swab (21%) (Figure 4). This

predominance could not be recaptured in the biopsy sample. Interpersonal variation in both biopsies

and swabs were evident at phylum level (Figure 5). Using a Wilcoxon's signed rank test for paired

data statistically significant differences emerged at various taxonomic levels (Table 2). The order

Clostridiales was the most abundant taxon enriched in the biopsies. These are obligate anaerobes

suggesting that the composition of the microbiota in the different skin layers is partly driven by the

oxygen tolerance of that taxon.

This article is protected by copyright. All rights reserved.

 DISCUSSION

This study demonstrates that there is a difference in the microbiome in swabs and biopsies taken from
Accepted Article
the same site and from the same individual. A variation in the normal skin microbiome dependent on

the anatomical site has been demonstrated1. However, this study has been on swabs and therefore only

revealing the surface microbiome. Swabs were deemed satisfactory following the study by Grice et

al., in which they compared the microbiome obtained by three different techniques: swabs, scrapes

and biopsies3. They captured similar microbial populations irrespective of the sampling technique

employed. From then on investigators have used swabs as the method of choice for their studies given

the invasive nature of skin biopsies. The dataset of this study, however, comprised only 5,373

sequences clustering into 113 OTUs at > 97% similarity. Differences between sampling techniques

were detected but with a maximum of four reads per site-specific OTU regarded as rare. Today,

almost ten years later, high throughput next generation sequencing enables us to go much deeper. Our

dataset comprised of over one million quality filtered reads resulting in almost 4000 OTUs. We now

find significant differences between swabs and biopsies that may have been difficult to detect with so

few reads as in the study from 2008.

In 2013 Nakatsuji et al. demonstrated the presence of bacteria in the dermis and superficial adipose

layer6. Hence, depending on the research question asked, swabs may not be sufficient for studying the

skin microbiome. Our results support this view.

In our initial study on normal skin and psoriasis we chose biopsies rather than swabs to study the

microbiome4. We decided on this approach because the pathogenesis of psoriasis commences in the

dermis. CD4 activated lymphocytes first appear in the dermis and then migrate to the epidermis. In

addition, one of the proposed antigens which may trigger psoriasis, streptococcal peptidoglycan, is

found in the dermis5.

This article is protected by copyright. All rights reserved.

 There have been at least five studies on the microbiome in psoriasis: four employing only swabs and

one biopsies7-10,4. Although there were similarities in the results seen in biopsies and swabs there were
Accepted Article
also differences both between biopsies and swabs in psoriasis and control samples4. Such differences

in microbial community composition depending on the sampling technique has recently been

demonstrated for gut luminal or mucosal sampling(Zmora et al. 2018). It therefore seemed important

to compare the microbiome of swabs and biopsies in normal skin.

Our results have shown differences in the microbiome obtained from either swabs or biopsies from

the same individual in both alpha and beta diversity. Statistical analysis revealed a significant increase

in the abundance of Clostridiales in biopsy samples. Presumably, conditions prove more favorable for

this group of obligate anaerobes in the lower parts of the skin as oxygen levels decrease. The phylum

Bacteroidetes was also significantly increased in biopsies. This phylum comprises both aerobic and

anaerobic bacteria but no significant enrichment in biopsies could be found at lower taxonomic levels

within Bacteroidetes. Another taxon found increased in biopsy samples compared to swabs was

Rhodococcus erythropolis (family Nocardiaceae). However, Rhodococcus has been identified as a

contaminant of DNA extraction kit reagents and ultrapure water systems, which may lead to its

erroneous appearance in microbiota or metagenomic datasets19. In our negative control it was detected

at an absolute abundance of 0.2%, which is lower than in most biopsies and higher than in most swabs

but this finding needs to be treated carefully. All other significant changes summarized in Table 2

were due to enrichment in the swabs compared to biopsies. The majority of these taxa are aerobic or

oxygen tolerant.

The difference in the microbiome between swabs and biopsies is also demonstrated by the increased

relative abundance of Staphylococcus aureus in swab versus biopsy samples (Figure 4). This was,

however, just a trend in our dataset and not statistically significant with the small number of subjects

included in this study. Nevertheless, the higher abundance of S. aureus in swabs may have

This article is protected by copyright. All rights reserved.

 implications for microbiome studies of skin diseases such as atopic dermatitis where this species is

involved20.
Accepted Article
In conclusion, we have shown that there is a difference in the microbiome between swabs and

biopsies in normal skin. Thus, we believe it is important to carefully consider sampling technique

when designing microbiome studies for skin disease. Whereas it may seem easier to motivate patients

to donate swab samples than small punch biopsies the decision to collect biopsies, swabs or both

should depend not only on technical feasibility but also on the scientific question asked and the

dermatosis studied.

CONFLICTS OF INTEREST:

● The study was approved by St. Vincent’s Hospital Dublin Ethical Committee and conducted

according to the Declaration of Helsinki Principles. All patients gave informed consent.

● All parties give consent for publication.

● The datasets generated and analyzed during the current study are available in the NCBI
BioSample repository (https://www.ncbi.nlm.nih.gov/biosample) under the accession number
PRJNA393029.

● The authors declare that they have no competing interests.

● This project was funded by the Skin Research Foundation

● SPN, LE, LH carried out 16sRNA gene sequencing, bioinformatics and statistical analyses

and contributed to writing manuscript. AMT, KA and BK, contributed to recruitment of

patients and writing of manuscript. AP, LE and LF, designed the study and contributed to

authorship of manuscript. SCM organized collection, storage and shipping of samples. All

authors read and approved the final manuscript.

This article is protected by copyright. All rights reserved.

 ACKNOWLEDGEMENTS:

We wish to thank our patients for kindly contributing their skin samples for this study. We
Accepted Article
would also like to acknowledge support from Science for Life Laboratory, the Swedish

national infrastructure SNISS, and Uppmax for providing assistance in massive parallel

sequencing and computational infrastructure.

Bibliography

1. Grice, EA, Kong, HH, Conlan, S, et al. (2009). Topographical and temporal diversity of the
human skin microbiome. Science 324: 1190–2.
2. Gao, Z, Perez-Perez, GI, Chen, Y, et al. (2010). Quantitation of major human cutaneous
bacterial and fungal populations. J Clin Microbiol 48: 3575–81.
3. Grice, EA, Kong, HH, Renaud, G, et al. (2008). A diversity profile of the human skin
microbiota. Genome Res 18: 1043–50.
4. Fahlén, A, Engstrand, L, Baker, BS, et al. (2012). Comparison of bacterial microbiota in skin
biopsies from normal and psoriatic skin. Arch Dermatol
5. Baker, BS, Laman, JD, Powles, A, et al. (2006). Peptidoglycan and peptidoglycan-specific Th1
cells in psoriatic skin lesions. J Pathol 209: 174–81.
6. Nakatsuji, T, Chiang, H-I, Jiang, SB, et al. (2013). The microbiome extends to subepidermal
compartments of normal skin. Nat Commun 4: 1431
7. Gao, Z, Tseng, C, Strober, BE, et al. (2008). Substantial alterations of the cutaneous bacterial
biota in psoriatic lesions. PLoS ONE 3: e2
8. Alekseyenko, AV, Perez-Perez, GI, De Souza, A, et al. (2013). Community differentiation of
the cutaneous microbiota in psoriasis. Microbiome 1:
9. Statnikov, A, Alekseyenko, AV, Li, Z, et al. (2013). Microbiomic signatures of psoriasis:
feasibility and methodology comparison. Sci Rep 3: 2620.
10. Martin, R, Henley, JB, Sarrazin, P, et al. (2015). Skin Microbiome in Patients With Psoriasis
Before and After Balneotherapy at the Thermal Care Center of La Roche-Posay. J Drugs
Dermatol 14: 1400–5.
11. Kong, HH, Andersson, B, Clavel, T, et al. (2017). Performing skin microbiome research: A
method to the madness. J Invest Dermatol 137: 561–8.
12. Hugerth, LW, Wefer, HA, Lundin, S, et al. (2014). DegePrime, a program for degenerate
primer design for broad-taxonomic-range PCR in microbial ecology studies. Appl Environ
Microbiol 80: 5116–23.
13. Martin, M (2011). Cutadapt removes adapter sequences from high-throughput sequencing
reads. EMBnet j 17: 10.
14. Rognes, T, Flouri, T, Nichols, B, et al. (2016). VSEARCH: a versatile open source tool for
metagenomics. PeerJ 4: e2584.
15. Edgar, RC (2016). UNOISE2: improved error-correction for Illumina 16S and ITS amplicon
sequencing. BioRxiv.

This article is protected by copyright. All rights reserved.

 16. Quast, C, Pruesse, E, Yilmaz, P, et al. (2013). The SILVA ribosomal RNA gene database
project: improved data processing and web-based tools. Nucleic Acids Res 41: D590–6.
17. Caporaso, JG, Kuczynski, J, Stombaugh, J, et al. (2010). QIIME allows analysis of high-
Accepted Article
throughput community sequencing data. Nat Methods 7: 335–6.
18. Ondov, BD, Bergman, NH, Phillippy, AM (2011). Interactive metagenomic visualization in a
Web browser. BMC Bioinformatics 12: 385.
19. Leyden, JJ, Marples, RR, Kligman, AM (1974). Staphylococcus aureus in the lesions of atopic
dermatitis. Br J Dermatol 90: 525–30.
20. Salter, S, Cox, MJ, Turek, EM, et al. (2014). Reagent contamination can critically impact
sequence-based microbiome analyses. BioRxiv.

Quast, C, Pruesse, E, Yilmaz, P, et al. (2013). The SILVA ribosomal RNA gene database project:
improved data processing and web-based tools. Nucleic Acids Res 41: D590-6.

Zmora, N, Zilberman-Schapira, G, Suez, J, et al. (2018). Personalized Gut Mucosal Colonization

Resistance to Empiric Probiotics Is Associated with Unique Host and Microbiome Features.
Cell 174: 1388–1405.e21.

This article is protected by copyright. All rights reserved.

 TABLES:

Patient Year of Birth Gender Skin Site Diagnosis

Accepted Article
1 1925 M dry BCC
2 1981 F dry Melanoma
3 1947 M sebaceous Seb Keratosis
4 1993 F sebaceous Naevus
5 1946 M dry Naevus
6 1979 F dry Atypical neavus
7 1963 F sebaceous BCC
8 1963 M sebaceous BCC
9 1974 M dry Naevus
10 1943 M sebaceous Seb Keratosis
11 1954 M dry Bowens
12 1981 M sebaceous Naevus
13 1937 M sebaceous SCC
14 1948 M dry Naevus
15 1950 F sebaceous Naevus
16 1946 M dry Reexision of SCC, all clear
Table 1: Patient information. Each individual provided one swab and one biopsy. BCC (basal cell
carcinoma), Seb (seborrhoeic), SCC (squamous cell carcinoma)

This article is protected by copyright. All rights reserved.

 Biopsy Swab P value FDR Taxonomy
(mean) (mean)
0.0002 0.0035 0.0006 0.03 Bacteria; Actinobacteria; Actinobacteria; Actinomycetales
Accepted Article
0.0002 0.0035 0.0006 0.03 Bacteria; Actinobacteria; Actinobacteria; Actinomycetales; Actinomycetaceae

0.0002 0.0030 0.0008 0.07 Bacteria; Actinobacteria; Actinobacteria; Actinomycetales; Actinomycetaceae;

Actinomyces
0.0158 0.0017 0.0006 0.03 Bacteria; Actinobacteria; Actinobacteria; Corynebacteriales; Nocardiaceae

0.0157 0.0006 0.0001 0.02 Bacteria; Actinobacteria; Actinobacteria; Corynebacteriales; Nocardiaceae;

Rhodococcus
0.0155 0.0002 0.0007 0.09 Bacteria; Actinobacteria; Actinobacteria; Corynebacteriales; Nocardiaceae;
Rhodococcus; Rhodococcus_erythropolis
0.0101 0.0391 0.0000 0.03 Bacteria; Actinobacteria; Actinobacteria; Micrococcales; Micrococcaceae;
Micrococcus; Micrococcus_luteus
0.0004 0.0038 0.0004 0.05 Bacteria; Actinobacteria; Actinobacteria; Micrococcales; Micrococcaceae; Rothia

0.0004 0.0034 0.0006 0.09 Bacteria; Actinobacteria; Actinobacteria; Micrococcales; Micrococcaceae; Rothia;
Other
0.0003 0.0082 0.0001 0.01 Bacteria; Actinobacteria; Actinobacteria; Other

0.0003 0.0082 0.0001 0.01 Bacteria; Actinobacteria; Actinobacteria; Other; Other

0.0003 0.0082 0.0001 0.02 Bacteria; Actinobacteria; Actinobacteria; Other; Other; Other

0.0045 0.0256 0.0001 0.03 Bacteria; Actinobacteria; Actinobacteria; Other; Other; Other; Other

0.0711 0.0153 0.0017 0.04 Bacteria; Bacteroidetes

0.0000 0.0012 0.0009 0.07 Bacteria; Bacteroidetes; Bacteroidia; Bacteroidales; Prevotellaceae; Prevotella_7

0.0008 0.0036 0.0001 0.03 Bacteria; Firmicutes; Bacilli; Bacillales; Staphylococcaceae; Staphylococcus;
Staphylococcus_pasteuri
0.0004 0.0056 0.0003 0.02 Bacteria; Firmicutes; Bacilli; Other

0.0004 0.0056 0.0003 0.03 Bacteria; Firmicutes; Bacilli; Other; Other

0.0004 0.0056 0.0003 0.04 Bacteria; Firmicutes; Bacilli; Other; Other; Other

0.0045 0.0364 0.0003 0.06 Bacteria; Firmicutes; Bacilli; Other; Other; Other; Other

0.2056 0.0476 0.0021 0.07 Bacteria; Firmicutes; Clostridia; Clostridiales

0.0000 0.0030 0.0007 0.09 Bacteria; Firmicutes; Negativicutes; Selenomonadales; Veillonellaceae; Veillonella;
Other
0.0001 0.0010 0.0026 0.08 Bacteria; Fusobacteria; Fusobacteriia; Fusobacteriales; Leptotrichiaceae

0.0174 0.1030 0.0002 0.03 Bacteria; Proteobacteria; Gammaproteobacteria; Pseudomonadales; Moraxellaceae;

Acinetobacter
0.0052 0.0650 0.0003 0.06 Bacteria; Proteobacteria; Gammaproteobacteria; Pseudomonadales; Moraxellaceae;
Acinetobacter; Other
0.0004 0.0050 0.0025 0.07 Bacteria; Proteobacteria; Gammaproteobacteria; Vibrionales

0.0004 0.0050 0.0025 0.08 Bacteria; Proteobacteria; Gammaproteobacteria; Vibrionales; Vibrionaceae

Table 2: Statistically significant differences (Wilcoxon's signed rank test for paired data) exist at
several bacterial taxa levels between skin biopsies and skin swabs.

This article is protected by copyright. All rights reserved.

 Figure 1. Alpha diversity plots showing significant differences between biopsies and
swabs in the number of observed OTUs and species richness (Chao1) but not in community
diversity (Shannon). Data are presented as mean and IQR. Paired t-test with **p < 0.01.
Accepted Article
Figure 2: NMDS analysis showing clustering by sampling procedure where biopsies are in
plum and swabs in turquoise. The lighter shades represent sebaceous skin; the darker
shades represent dry skin. The white dot is the negative control, which is separated in the
NMDS from the remaining samples but shows more similarity to the biopsies than swabs.

Figure 3: Taxonomic Krona plots of volunteer 2, genus level. A biopsy and swab from
the same individual showing different microbial taxa with swabs showing increased diversity.

Figure 4: Taxonomic Krona plots of volunteer 16, species level. Biopsy and swab
samples from the same healthy volunteer showing Staphylococcus aureus to be more
abundant in swab (21%) than biopsy sample (2%).

Figure 5: Taxonomic profiles at phylum level. Bacteroidetes were increased in the

biopsies, which was supported by statistical analysis summarized in Table 2.

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

 Figure 3
Biopsy 2
Accepted Article

Swab 2

This article is protected by copyright. All rights reserved.

 Figure 4
Biopsy 16
Accepted Article

Swab 16

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.