Characterization of pig skin microbiome and appraisal as an in vivo subcutaneous injection model

Abstract

German

Spanish

French

Pig skin is commonly used in the medical industry as an injection model due to its compelling physiological affinity to human skin. However, the pig neck skin microflora remains largely uncharacterized, which may have undesirable implications for the translatability of results to humans. This study aimed to characterize pig neck skin microbiome with direct comparison with human skin microflora at emblematic injection sites to appraise its suitability as an injection model. Ten minipigs were sampled with tape strips and swabs and analysed by matrix-assisted laser desorption/ionization-time of flight mass spectrometry and 16S/ITS high throughput sequencing and confocal laser scanning microscopy. Results were directly compared with previous investigations of human injection sites. Pig skin was dominated by phyla 94.8% Firmicutes, 3% Proteobacteria, and 2.2% Actinobacteria. Staphylococcus spp. prevailed (44.4%) at the genus level with S. capitis and S. chromogenes present in all samples. Pig skin revealed populations in the 10⁴ colony-forming units (CFU)/cm² range with 3% identified as Gram-negative and increased alpha diversity (compared with 10² CFU/cm² and 10% in humans). While notable taxonomical differences on species levels were seen, pig skin encompassed 97.1% of genera found in human samples. The increased population and variation found support the pig neck as an imperfect but fidelitous subcutaneous injection model that can adequately challenge devices from a microbial standpoint.

Keywords

Immunology injection microorganism organisms and models physiology pigs species comparison techniques

Introduction

In vivo animal models remain indispensable in preclinical research for elucidating mechanisms and testing novel therapies for humans. Mammals such as rodents, rabbits and pigs are commonly used in dermatological testing, as similarities between human and mammal skin can corroborate the multifaceted and complex nature of skin dynamics and healing.^1,2 Rabbit and rodent models are inexpensive, often used for screening tests and provide sufficient, reliable insight into wound healing mechanisms.² However, substantial differences such as dense pelage, thin epidermis and dermis, and healing through wound contraction rather than re-epithelialization compromise the fidelity of these small animal models to their human counterparts.^3,4

As such, porcine models are considered the gold standard in dermatological research due to remarkable anatomical, physiological and metabolic similarities to humans.^1,3,5–7 Like humans, porcine skin is relatively hairless, with equal epidermal thickness, subcutaneous fixation and cutaneous flow pattern.^1,8 The dermal–epidermal thickness ratio ranges from 10:1 to 13:1 in both swine and humans and contains equal numbers of viable and horny layers.^9,10 Additionally, pig skin exhibits comparable epidermal turnover time to humans with similar collagen structure, elastic fibre arrangement, keratinous protein forms and lipid configuration.^3,4,8 Similarly, reepithelization is a crucial, primary component of wound healing in swine.^1,4 Finally, the size and robustness of pigs allow for long, complex studies, such as surgical training, toxicology testing and pharmacokinetic/pharmacodynamic evaluations.^4,11,12

In the medical industry, verification of injection device safety and efficacy is routinely performed on pigs due to compelling physiological affinity to human skin.^4,13,14 Specifically, the dorsal lateral pig neck serves as a high-fidelity model for subcutaneous (SC) injections, bearing similar biomechanical properties to emblematic injection sites.¹³ Moreover, this model is used to evaluate contamination and endotoxin formation during use for novel, multi-use devices. Nevertheless, while anatomical parallels have been subject to extensive investigation, similarities in skin microbiome remain largely unexplored. A few studies survey bacterial microbiota in neighbouring locations, finding more populated skin microflora with similar bacterial phyla distribution to humans, but neither assess fungi nor provide direct comparison with humans.^15–17 Similarly, these studies do not investigate typological and biogeographical distribution of microorganisms on the skin. As such, characterizing the microbial composition of pig skin with direct comparison with humans is crucial in validating the model from a microbiological standpoint. Moreover, ascertaining the ratio of Gram-negative bacteria in pig and human skin is valuable for an effective endotoxin evaluation.

This study endeavoured to elucidate pig neck skin microbiome to appraise its suitability as a model for human SC injections. Specifically, we evaluated the model’s capacity to predict device contamination rates and assess the device’s ability to remain sterile after multiple injections. The necks of 10 pigs were sampled using tape strips and swabs. Samples were analysed through traditional culture methods, high throughput sequencing (HTS), and confocal laser scanning microscopy (CLSM) for topographical visualization. Results were directly compared with identical investigations performed previously on humans at emblematic injection sites.¹⁸ We hypothesized that pig cutaneous microbiome is more populous and compositionally distinctive from human injection sites at lower taxonomical levels but contains similarities at higher taxonomical orders. These findings provide essential support to validate the pig neck as a fidelitous and pertinent SC injection model.

Methods

Ethics statement

The present study followed international, national and institutional guidelines for humane animal treatment and complied with relevant legislation from Novo Nordisk’s internal Ethical Review Council.

Animals and sampling

Two groups (Groups A and B) of 10 minipigs were included in July 2020 and December 2020, respectively, as seen in Figure 1. The Göttingen SPF minipigs were selected as the test model because of their suitability in this type of study and sample size based on ethical considerations due to the limited physical area of the pigs suitable for subcutaneous injections. Moreover, only female pigs were used as they were used for a related, more invasive study and were reused in accordance with the principles of the 3Rs in animal research (Replacement, Reduction and Refinement).

Figure 1.

Schematic of experimental design. From Group A, 10 minipigs were sampled on the pig neck skin through tape strips and swabs. The tape strips were analysed through serial dilution, cultivation, colony-forming unit determination and matrix-assisted laser desorption/ionization-time of flight identification. From Group B, 10 minipigs were sampled with tape strips, stained with 4′,6-diamidino-2-phenylindoleand analysed through confocal laser scanning microscopy.

Group A

Ten female Göttingen SPF minipigs from Ellegaard Göttingen Minipigs A/S (Dalmose, Denmark) were used. The animals were housed in accordance with EU Directive 2010/63/EU of 22 September 2010 on the protection of animals used for scientific purposes. The animals had ad libitum access to domestic quality drinking water. An SDS minipig diet (SMP (E) SQC) from Special Diets Services (Witham, Essex, UK) was offered twice daily in approximately 300 g per animal per meal. The animals had a bodyweight of at least 25 kg at arrival. A pre-treatment period of roughly three weeks (including an acclimatization period of five days) was allowed, during which the animals were being observed daily to reject animals in poor condition. All observations were recorded.

The sampling took place with pigpens provided with filtered air at 21°C ± 3°C. The minipigs were housed in floor pigpens individually (at least 2 m²) with solid floor and Aspen bedding 2HV (Tapvei Estonia OÜ, Estonia). Analyses for relevant possible contaminants were performed regularly. The animal room was cleaned and disinfected before arrival, and the pigpens were scraped free of manure once daily. The study was conducted in compliance with the Animal Welfare Act and AAALAC accreditation with relevant elements from the ISO 9000 standard and GLP (Good Laboratory Practice).

Group B

Again, 10 female Göttingen minipigs acquired from Ellegaard Göttingen Minipigs A/S (Dalmose, Denmark), housed in accordance with EU Directive 2010/63/EU of 22 September 2010 on the protection of animals used for scientific purposes. This group contained bodyweights between 25.3 kg and 33.4 kg. The minipigs had central venous catheters (Cook Medical A/S, Denmark) placed using a Seldinger technique, due to participation in other studies. Type 1 diabetes was induced in the pigs by intravenous infusion of 50 mg/kg STZ per day (Sigma Aldrich Denmark A/S, Lot no. WXBC2544V, WXBC8304V, WXBC8711V, WXBC8740V, WXBC1402V) over three consecutive days giving a total STZ dose of 3 × 50 mg/kg. The minipigs were fed restrictively and provided with 300–400 g Altromin 9023 (Altromin, Germany) twice daily. The animal housing was temperature and humidity controlled, with temperatures 22°C ± 2°C and humidity 60% ± 20%.

Sampling Group A

The skin microbiota of 10 minipigs was sampled by collecting seven layers of DS100 D’Squame Disc tape strips (Monaderm, Monaco) from the upper neck in accordance with a previously published protocol.¹⁸ Sampling was performed on both left and right sides of their necks proximal to the tattooed square areas used in a provisional injection study. All 10 pigs had a catheter in one ear. Since catheter insertion may affect local microflora, the non-catheterized side was used for the original sample. In contrast, the sample from the catheter side was saved as back-up. Individual tape strips were applied to the designated area with sterile forceps. Adhesion of the tape to the skin was secured using a D-square pressure instrument (225 g/cm²) (Monaderm, Monaco) and held for 15 s, then removed by sterile forceps. Initially, tapes 1 and 2 were discarded to remove excess dust and particles from the analysis. Tape numbers 3–6 were pooled together and placed in a sterile 5 ml cryogenic tube (ThermoFisher Scientific™, MA, USA), kept at room temperature and cultured the same day. Tape 7 was stored in a sterile 1.5 ml Eppendorf tube at –80°C for 16S rRNA and ITS amplicon HTS. All tapes were placed so that the adhesive side with the microorganism sample was facing the interior of the respective storage tubes and therefore not exposed to the plastic.

In addition, sterile swabs (ESwab™ 480C, Copan, Murrieta, CA, USA) dipped in 1 ml sterile molecular grade water (Lonza AccuGENE Molecular Biology Water, Basel, Switzerland) was used to collect the skin microbiota by applying slight pressure while rotating the swab. The swab was rubbed back and forth approximately 50 times in the marked skin area of 5 cm × 5 cm. The swabs were transferred into Eppendorf tubes and stored at –80°C.

Sampling Group B

Similarly, four tape strips (1a, 2a, 1b and 2b) were sampled pairwise in proximal vicinity on the upper right neck of the animal. Tapes 1a and 2a were stored at –80°C for back-up, tape 1b was discarded and tape 2b was processed for CLSM the same day.

Tape extraction and cultivation

The pooled tapes 3–6 had 3 ml 0.9% saline solution added, then were degassed for 5 min, followed by 5 min of sonification in an ultrasonic bath (Branson 2510, Sigma-Aldrich, St. Louis, MO, US). The fluid was then plated in a dilution series followed by incubation on (a) 5% blood agar plates (SSI Diagnostica, Hillerød, Denmark) at 37°C aerobic for 1–2 days, (b) chocolate plates (SSI Diagnostica, Hillerød, Denmark) at 37°C for 1–2 days with CO₂ (CO₂ incubator, Sanyo, Osaka, Japan) and (c) fastidious anaerobic agar plates with added 5% horse blood (SSI Diagnostica, Hillerød, Denmark) at 37°C for 2–3 days under anaerobic conditions (Concept 400, LAF technologies, Bayswater North, VIC, Australia). Each sample was serially diluted, cultured and quantified as colony-forming units (CFU) followed by identification using matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) (Bruker, Billerica, MA, USA). The bacteria were isolated and placed in cryotubes containing 600 µl LB broth (Sigma-Aldrich, St. Louis, MO, US) and 400 µl glycerol (Sigma-Aldrich, St. Louis, MO, US).

MALDI-TOF identification

Bacterial isolates were re-cultivated using the mentioned growth conditions. According to the manufacturer’s manual, single colonies were transferred to an MSP 96 target polished steel BC (Bruker, Billerica, MA, US). Each isolate was done in duplicate. For isolates that were not identified by the machine at the first attempt, 1 µl of formic acid was added to the bacteria isolate before addition of matrix. The sample IDs were then entered into the MTB Compass program and the samples analysed by the MALDI-TOF Microflex LT/SH (Bruker, Billerica, MA, USA). Reads with scores of 2.0 or over were accepted as identified on species level. Bacterial identification with lower scores was repeated up to three times until a score of 2.0 was reached or removed from the analysis.

HTS

All samples for sequencing were investigated using 16S rRNA and ITS sequencing to study the bacterial and fungal microbiome, respectively. This was performed at Clinical Microbiomics (Clinical Microbiomic A/S Fruebjergvej 3, Copenhagen, Denmark).

PCR

DNA was extracted from the samples using NucleoSpin® 96 Soil (Macherey-Nagel). Bead beating was done on a Vortex-Genie 2 horizontally at 2700 rev/min for 5 min. Positive controls (ZymoBIOMICSTM Microbial Community Standard, Zymo Research) and negative controls were included with each batch of samples. For bacteria, PCR was done using the forward primer S-D-Bact-0341-b-S-17 and reverse primer S-D-Bact-0785-a-A-21 with Illumina adapters attached.¹⁹ Illumina adapter and S-D-Bact-0341-b-S-17: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′; Illumina adapter and S-D-Bact-0785-a-A-21: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′. These universal bacterial 16S rDNA primers target the V3–V4 region. PCR programme: 98°C for 30 s, 29x (98°C for 10 s, 55°C for 20 s, 72°C for 20 s), 72°C for 5 min. Amplification products were verified on agarose gels. ITS3 (forward) and ITS4 (reverse) primers were used for fungi with Illumina adapters attached.²⁰ Illumina adapter and ITS3: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGCATCGATGAAGAACGCAGC-3′; Illumina adapter and S-D-Bact-0785-a-A-21:v5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTCCTCCGCTTATTGATATGC-3′. These universal fungi primers target the internal transcribed spacer 2 (ITS2) region between the small-subunit rRNA and large subunit rRNA genes. A low-biomass PCR programme was used: 98°C for 30 s, 29x (98°C for 10 s, 58°C for 20 s, 72°C for 20 s), 72°C for 5 min. Indices were added in a subsequent PCR using the Nextera Index Kit V2 (Illumina): 98°C for 30 s, 8× (98°C for 10 s, 55°C for 20 s, 72°C for 20 s), 72°C for 5 min. Finally, the products were run through agarose gels to confirm attachment of indices.

Normalization and sequencing

The nested PCR products were merged based on band intensity and the library cleaned with magnetic beads.

The concentration of DNA in the libraries was measured fluorometrically. Sequencing was conducted on an Illumina MiSeq desktop sequencer utilizing the MiSeq Reagent Kit V3 (Illumina) for 2 × 300 bp paired-end sequencing.

Bioinformatics analysis

An adjusted DADA2 pipeline was utilized for bioinformatics processing of the bacterial sequence into amplicon sequence variant (ASV).²¹ Using cutadapt, primers were removed from the raw reads and were filtered by removing reads without primer matches and ambiguous bases, reads shorter or longer than anticipated of sequencing cycles and lengths of primers. dada2::filterAndTrim command trimmed the 3 prime end base on sample-specific quality scores. The reads containing more than one error were removed and the remaining reads we dereplicated into unique sequences. These were denoised separately for forward and reverse reads. In this step, a less abundant sequence was assigned to closely related more abundant sequences thorough evaluation to a data-based error matrix, as the low abundant sequence is considered a sequencing error. Read pairs lacking sufficient overlap, or overlapping sections, were removed as denoised forward and reverse reads were combined. Last, an abundance table or internal abundance and sequence assessments allowed for removal of suspected chimaeras. A naïve Bayesian classifier algorithm assigned taxonomy of detected ASVs by comparison with the SILVA reference database (v. 138).

To refine the taxonomic assignment of the ASVs, in silico-obtained amplicons equivalent to the used primers from current versions of four reference databases (silva (https://www.arb-silva.de/documentation/release-138/), the genome taxonomy database (GTDB, https://gtdb.ecogenomic.org/), the ribosomal database project RDP (http://rdp.cme.msu.edu/) and the Unified Human Gastrointestinal Genome catalogue (UHGG, https://www.ebi.ac.uk/metagenomics/genomes)) were used. To improve the default assignment of the ASV, reference amplicons identical to the ASV or the amplicons with the highest sequence identity to the ASV were used. A similar method was executed for fungal sequences. We filtered out ASVs not present in at least two samples or without relative abundance of 0.8% in any sample. Taxonomic assignment of sequences was compared with the UNITE ITS database (version no 8.2; 04.02.2020, https://unite.ut.ee/repository.php).

CLSM

Tapes were processed for microscopy by placing the tape directly on a microscope cover lens with 20–50 µl 3mg 4′,6-diamidino-2-phenylindole (DAPI) staining, spread uniformly on the tape. The tape was left for 15 min and then sealed onto a microscope slide using clear nail polish. After 48 hours, these strips were analysed with an LSM710 CLSM (Zeiss, Oberkochen, Germany) with a Plan-Neofluar and 63 × /1.4 plan-apochromatic oil objectives (Zeiss) and UV light of excitation wavelength of 405 nm and an emission 410–483 nm. Images were then taken using the ZEN black 2010, v. 6.0 (Zeiss) followed by deconvolution into tiff-files using Imaris 8 × 64 version 8.1.2 (Bitplane, Zürich, Switzerland). At least three images were taken per tape as a representation for the microscopic observation.

Statistical analyses

All statistical analyses were performed in the computing environment R (v. 3.5.0; R Core Development Team, 2005).

Results

Bacterial culture and identification

The viable bacterial count from pig sampling averaged 1.5 × 10⁴ CFU/cm². Specifically, 1.5 × 10⁴ CFU/cm² were isolated from aerobic conditions, 1.5 × 10⁴ CFU/cm² from CO₂ chamber conditions and 1.4 × 10⁴ CFU/cm² from anaerobic conditions. All samples were in the 10³–10⁴ CFU/cm² range, though a notable standard deviation reflected substantial variability between pigs. As shown in Figure 2(a), the averages of each culturing condition were similar, with the most variability coming from the plates grown in the CO₂ incubator.

Figure 2.

Microbiota from cultivation of tape strips. (a) Viable bacterial specie count recovered from cultivation of tape strips from pig neck skin (n = 10), as found by each incubation method (aerobe, anaerobic, CO₂), (b) bacterial distribution of species in phyla from culture of tape strips from pig neck skin and (c) bacterial distribution of genus found in pig neck skin compared with human injection sites.

To examine the usability of pigs to predict the microbiological contamination of pen injectors, we assessed the correlation between pig neck microflora and that previously obtained at human injection sites.¹⁸ This study used an identical protocol finding human viable bacteria load to be in the 10²–10³ CFU/cm² range with an average of 3 × 10² CFU/cm² and 4 × 10² CFU/cm² for aerobic and CO₂ growth conditions and 2 × 10² CFU/cm² and 4 × 10² CFU/cm² for anaerobic conditions at the abdomen and thigh, respectively. In addition, a one-way ANOVA analysis revealed significantly higher bacteria in pig samples (p-value <0.0001). These findings indicate that pig neck skin microflora is approximately two orders of magnitude greater than human injection site bacterial counts.

Furthermore, 42/52 isolates were identified as 18 unique species from eight genera were detected from the 10 sampled pig necks. The taxonomical distribution of bacteria at the phylum level contained 94.8% Firmicutes, 3% Proteobacteria and 2.2% Actinobacteria. Staphylococcus spp. prevailed (44.4%) at the genus level with S. capitis and S. chromogenes being the most common and present on all 10 pigs. Only 3% were identified as Gram-negative, consisting of Leclercia adecarboxylata and Escherichia coli. Previous investigations in humans found 24 unique species from 12 different genera with a taxonomical distribution of 63.5% Firmicutes, 29.1% Actinobacteria and 7.8% Proteobacteria (with three unidentified from Aeromicrobium, Micrococcus and Pseudomonas spp.).¹⁸ As with the pig neck, Staphylococcus spp. predominated on the genera level (61.7%) with S. epidermidis, S. hominis and S. capitis prevailing. Of all identified bacteria, 10.2% were Gram-negative, with Mixta calida and Acinetobacter lwoffii being most populous.

As seen in Figure 2(c), human microbiota is more varied with greater diversity on a species level; however, both showed similarities at genera and phylum level, with Firmicutes and Staphyloccocus being most abundant, respectively. Actinobacteria were considerably higher in human samples (29.1% vs. 2.2%). Proteobacteria were detected in both categories, representing the Gram-negative abundances in each (7.8% in humans and 3% in pigs). While Staphylococcus spp. dominated both skin microflorae (56.6% vs. 44.4%), only four species were common for humans and pigs: S. aureus, S. capitis, S. haemolyticus and S. warneri (Figure 2(a)).

16S and ITS amplicon HTS

All pig swab (n = 10) and tape (n = 40) samples contained sufficient biological signal to pass the 3500 high-quality (HQ) threshold. Due to previously observed differences in microbial compositions of sampling technique, the subsequent analysis was performed separately for the swab and tape samples. These were similarly contrasted to previously analysed human swabs (n = 100) and tapes (n = 62) from the abdomen and thigh.¹⁸ From this analysis, the most apparent difference was that the Firmicutes:Actinobacteriota ratio was lower in human than in pig samples, as seen in Figure 3(a). For ITS sequencing, all pig swab (n = 10) and tape (n = 38) samples and human swab (n = 72) and tape (n = 18) samples tapes passed the biological signal filter 3500 HQ sequences after background removal at genus level, as seen in Figure 4(a), where green and blue shades, respectively, represent Ascomycota and Basidiomycota.

Figure 3.

16S amplicon high throughput sequencing. (a) Taxonomic profiles of all human and pig samples with more than 3500 high-quality sequences after background removal at genus level. Only the 10 most abundant genera are coloured, with the remaining pooled in the ‘Other’ category. The Firmicutes:Actinobacteriota ratio is lower in human than in pig samples. (b) Overall microbiome composition of human and pig samples as estimated by Bray–Curtis dissimilarities calculated from relative abundances at genus level, illustrated by a Principal Coordinates Analysis (PCoA). The sample type variable distinguishing human swab, pig swab, human tape and pig tape samples explained 32.7% of the overall variation based on a permutational analysis of variance and (c) Richness values and Shannon index values for tape samples were higher for pig than for human samples. Asterisks refer to the p-values of Mann–Whitney U tests (**p-value ≤0.01, ****p-value ≤0.0001), PCo1 and PCo2 refer to Principal Coordinate 1 and 2, respectively.

Figure 4.

ITS amplicon high throughput sequencing. (a) Taxonomic profiles of all human and pig samples with more than 3500 high-quality sequences after background removal at genus level. Only the 10 most abundant genera are coloured, with the remaining pooled in the ‘Other’ category. The Firmicutes:Actinobacteriota ratio is lower in human than in pig samples. (b) Overall microbiome composition of human and pig samples as estimated by Bray–Curtis dissimilarities calculated from relative abundances at genus level, illustrated by a Principal Coordinates Analysis (PCoA). The sample type variable distinguishing human swab, pig swab, human tape and pig tape samples explained 14.8% of the overall variation based on a permutational analysis of variance test and (c) Alpha diversity represented by richness and Shannon index. Asterisks refer to the Mann–Whitney U tests (****p-value ≤0.0001, *p-value ≤0.05), PCo1 and PCo2 refer to Principal Coordinate 1 and 2, respectively.

Sequencing of both tapes and swabs revealed a distinct difference in the overall microbiome composition of humans and pigs, visualized by the clear separation of samples as estimated by Bray–Curtis dissimilarities calculated from relative abundances at genus level as seen on the principal coordinates analysis in Figure 3(b). The sample type variable distinguishing human swab, pig swab, human tape and pig tape samples explained 35% of the overall variation based on a permutational analysis of variance (PERMANOVA) test. For ITS, the sample type variable distinguishing human swab, pig swab, human tape and pig tape samples explained 15.4% of the overall variation based on a PERMANOVA test, as seen in Figure 4(b). Generally, pig samples contained increased richness, particularly those attained by tape strips, as seen in Figure 3(c) and Figure 4(c), as calculated by Mann–Whitney U tests (****p-value ≤0.0001, *p-value ≤0.05). We tested for differentially abundant/prevalent taxa between human and pig samples, including only swab, only tape, or both swab and tape samples, acknowledging that these tests include some dependent samples from the same individual. As expected from the beta diversity results, many taxa differed significantly in abundance and especially in prevalence between human and pig samples. For example, the 16S analysis revealed an increased abundance of Rothia in pigs compared with humans, whereas Cutibacterium was less copious on pigs than on humans. For ITS, the genus Saccharomyces was more abundant on human than on pig skin, while the genus Vishniacozyma was more abundant on pig than on human skin.

Microscopy

The DAPI stained tape strips were analysed with CLSM using a 60x magnification, with three representative images acquired per tape. Bacteria in these microscopy images were heterogeneously distributed in larger 3D anisotropic aggregates or as single (or diplo) cells. The aggregates were scattered randomly in clusters, attached to anucleate keratinocytes of the outer epidermis. In comparison with CLMS images previously acquired from human injection sites, pig neck skin exhibited considerably higher population density, clustered in larger aggregates and often of multiple layers. Four representative images can be seen in Figure 6.

Discussion

The utilization of animals as experimental models for medicine dates back to the origin of scientific research.²² Indeed, the noteworthy structural and functional similarities between humans and other vertebrates have allowed for invaluable discoveries of physiological mechanisms and appraisal of novel therapies prior to human application. However, moral and ethical implications require careful assessment of the translatability of these findings to humans, remaining a heated point of contention in societal debate. As such, profound knowledge of animals’ biological conditions and resemblances is fundamental to designing and implementing animals’ ethical use in biomedical research. To this end, we sought to characterize the superficial skin microflora on the lateral dorsal porcine neck, draw parallels to human injection sites and validate the model’s ability to emulate contamination of injection devices during SC injections.

The metagenomic analysis revealed dominance of phyla Firmicutes, Proteobacteria and Actinobacteria for bacteria and Ascomycota and Basidiomycota for fungi. These phyla comparably prevail in human skin microflora at emblematic injection sites, though their relative abundances are slightly altered.¹⁸ However, notable differences were observed at lower taxonomical levels. Beta diversity analysis revealed numerous taxa differing significantly in abundance and prevalence between human and pig samples. Pig skin contained greater abundance of Rothia but less Cutibacterium than human injection sites. From culture, pig skin contained fewer Gram-negative bacteria than human injection sites (3% vs. 10.2%). Furthermore, pig skin samples were generally relatively homogenous (low inter-swine beta diversity), with increased alpha diversities regarding richness and Shannon index than in humans. This trend was particularly evident on tape samples.

To assess whether pig skin is a suitable model for human skin concerning the contamination risk of re-usable insulin needles, one must analyse the overlap of taxa found on human and pig skin. It is particularly relevant to describe the fraction of microorganisms only detected on human but not pig skin since the contamination risk of these taxa is potentially unaccounted for when using the pig model. The data revealed distinct compositional differences between human and pig samples. From 16S, an average of 54.3% of the bacterial signal of human samples came from ASVs that were not detected in the pig neck samples. Figure 5 portrays the proportion for the various human samples and the taxonomic profile of these ASVs at the genus level. Nonetheless, pig skin microflora contained increased richness and encompassed bacteria of the majority of genera found in human samples. In consequence, genera detected on only human samples consist of only 2.9% of the composition of the human skin microflora, as seen in Table 1 part A. For ITS, 37.4% of ASVs were unique to human sample, as seen in Table 1 part B. However, these genera constituted only 13.6% of the composition of human samples. Thus, pig skin microbiota showed increased alpha diversity and covers the microorganisms of most genera of the human skin microbiota, but species/strain compositions clearly vary between pig and human skin. Notably, the pig samples encompassed bacteria of the majority of genera found in human samples. In consequence, genera only detected on human samples consist of only 2.9% of the composition of the human skin microflora.

Figure 5.

Taxonomic profiles of microbial signals found exclusively in human samples. (a) The taxonomic profile of the human samples at genus level from 16S amplicon high throughput sequencing (HTS) are shown including only amplicon sequence variants (ASVs) that were exclusively found in human samples. These human-only ASVs covered on average 54.3% of the composition of the human samples and (b) The taxonomic profiles of the human samples at genus level from ITS amplicon HTS are shown including only ASVs that were exclusively found in human samples. These human-only ASVs covered an average of 37.4% of the composition of the human samples. Note that the coloured genera were not necessarily absent from pig samples (all pig samples included Candida and Melassezia, for example).

Table 1.

Number of taxa found in pig and human samples (swab and tape samples pooled) from (A) 16S and (B) ITS amplicon high throughput sequencing and the average proportion of the human samples that were covered by taxa that were found only in human samples.

Taxonomic level	No. of taxa (human)	No. of taxa (pig)	No. of taxa (overlap)	Mean relative abundance of human-only taxa
A
ASV	1509	990	541	54.3%
Species	847	525	401	34.4%
Genus	344	298	226	2.9%
Family	145	130	115	1.1%
Class	22	19	18	0.1%
B
ASV	484	203	161	37.4%
Species	163	75	65	10.8%
Genus	157	69	61	13.6%
Family	117	58	54	8.8%
Class	21	16	16	1.2%

ASV: amplicon sequence variant.

Figure 6.

Dispersal of bacteria on 4′,6-diamidino-2-phenylindole-stained tape strips sampled from pigs ((a) to (d)) compared with humans ((e) to (g)). Images were taken with Axio Imager.Z2, LSM710 CLSM (Zeiss, Germany) with Plan‐Neofluar and 63×/1.4 plan‐apochromatic oil objectives (Zeiss) and UV light of excitation of 405 nm and emission 410–483 nm. Tapes from pigs contained larger and more frequent multilayered aggregates than tapes from humans. Images from humans were reproduced with permission.¹⁸

Intriguingly, pigs showed greater richness and evenness than humans from the 16S/ITS data, whereas the culture analysis disclosed the opposite trend. This may be explained by the inherent bias of culturing, selecting the most abundant and aggressive microorganisms that grow readily in laboratory conditions.²³ Traditionally, microbiota characterization entailed culture-based approaches, defining phylogeny and taxonomy through phenotyping, microscopy and biochemical interactions.^24,25The advent of molecular biology has exposed much greater diversity and variation, with less than 2% of bacterial species expected to be culturable.²⁶ These advances in high-throughput sequencing have allowed for a more comprehensive understanding of skin microbiome dynamics, facilitating recognition of highly complex and variable microbial ecosystems.^27–31 Our results from cultivation reveal an overrepresentation of Staphylococcus spp. and subsequent misrepresentation of diversity in pig skin. This discrepancy underlines the importance of critical appraisal and evaluation of investigation techniques for an accurate compositional assessment.

Furthermore, pig samples showed increased bioburden compared with humans, with sufficiently biological signal for reliable metagenomic analysis. Indeed, bacteria viable counts displayed population densities in the 10⁴ CFU/cm² range compared with 10² CFU/cm² in previously surveyed human injection sites.¹⁸ Therefore, the viable bacteria count of pig neck skin is roughly two orders of magnitude greater than human injection sites. Previous analyses of other pig skin locations cited 10⁴–10⁵ CFU/cm² with less spatial heterogeneity than humans, claiming no statistical differences between the anatomical regions.⁶ Thus, pig microbiota appears less biogeographically dependent and consequently more consistent than reported in humans.^23,28,31,33 Finally, microscopic investigation revealed similar spatial distribution of bacterial aggregates on the superficial skin, though aggregate sizes were generally greater in size than those found in humans. This is similarly reflected in the increased population density.

Thus, pig neck skin microflora is significantly more populated and distinctively different from human injections sites at lower taxonomical levels. Despite the aforementioned anatomical similarities, they do diverge in vascularity and epidermal adnexal structures of the pilosebaceous system. In contrast to humans, pig skin is void of eccrine glands, possessing only apocrine sweat glands, although glands similar to eccrine appear on the snout, lips and carpal organ.^4,34,35 Furthermore, these apocrine structures open up to the skin surface directly, independent of pilary orifices.³⁴ In addition, notable lifestyle incongruences such as diet, hygienic practices and living conditions exacerbate disparities in microbial constituents. For example, a study found a considerable fraction of pig skin microflora also to be residents of pig gut microbiome, possibly due to shared community membership as well as a lifestyle that propagates faecal and skin interaction.¹⁵ Despite this, however, prominent parallels regarding taxonomical distributions on phyla levels, dominance of Staphylococcus spp. and analogous bacterial spatial distribution demonstrate certain similarities in microbial compositions. Crucially, pig skin covered 97.1% of genera detected on humans. In addition, the pig neck microflora contains higher diversity and two orders of magnitude greater microbial population than human injection sites.

Implications

The considerable compositional and population disparities between the pig neck and human injection sites render pig neck an inadequate predictor of contamination rates during SC injections in humans. However, with respect to constructing a model for bioburden tests, the increased bacterial load and diversity combined with exposure to faecal microorganisms only challenges the device further, posing as a ‘worse case’ scenario. Subsequently, the pig skin model effectively promulgates the decontamination ability of the injection system. Furthermore, humans’ skin contains a greater percentage of Gram-negative bacteria than pigs’ (10% vs. 3%), rendering the pig skin model imperfect for assessing endotoxin release. However, since pig skin contains a population two orders of magnitude larger than humans, the exposure will inherently be greater. As such, the pig model adequately tests for endotoxin formation. Nevertheless, for other, more delicate wound healing studies, the microbial counts and exact constitution might play a pivotal role in the study outcomes and should not be underestimated. For instance, in studies addressing the microbial population and diversity of chronic wounds, the differences in pig skin and human microbiota may have more severe implications.^5,36

Moreover, this study examined only healthy, young female swine of comparable genetic makeup with highly controlled living conditions, representing the animals and conditions utilized for SC tolerance studies. While multiple factors such as age, sex, genetics, diet and environmental cues significantly affect the human microbiome, the translatability of these effects in the swine microbiome could be a thrilling avenue for future studies, particularly in extension to wound healing research.^31,37

Furthermore, the sampling techniques utilized in this study surveyed only the superficial skin. While numerous studies have discovered distinctive but similar community memberships in various skin compartments of humans, the extent to which this holds for swine remains unexplored.^28,38 As mentioned previously, disparities in epidermal adnexal structures such as sweat and sebaceous glands might influence the transportation and distribution from deeper layers into the stratum corneum.^24,38 Though pig skin bears a convincing physiological affinity to human skin regarding epidermal structure and turnover, the distribution of bacteria across dermal compartments warrants further investigation for wound healing models of the future.^1,3,5

Conclusion

This study aimed to characterize the pig neck cutaneous microbiome to evaluate its faculty to emulate microbial contamination of devices during SC injections. Pig neck microflora is significantly more populous and varied than human skin and compositionally distinctive at lower taxonomical levels. Nonetheless, we observed notable similarities at the phylum level and discovered that pig neck skin microflora encompasses most human injection site genera. These findings offer vivid testimony to support the pig neck as an imperfect but rather fidelitous SC injection model that can adequately challenge devices from a microbial standpoint. The upper dorsal pig neck can thus serve as a suitable model for injection and infusion studies. Prospective research should evaluate how differences in the pig and human skin microbiome may influence the translatability of chronic wound healing models in the future.

Footnotes

Acknowledgements

The authors would like to thoroughly thank all volunteers, Dr Dorte Lindqvist Hansen and the Steno Diabetes Centre Copenhagen, Christian Andreasen for statistics, Lasse Kvich and Ida Clement Thaarup for general training in the laboratory/microscope, and the LEO foundation for funding Lene Bay.

Author contributions

Literature search and writing: SWM; figures and tables: SWM; study design: SWM, MAH, LB, TB, VPG, HB; data collection: SWM; data analysis: SWM, LB; data interpretation: SWM, LB, TB; editing and approval of final version of manuscript: SWM, LB, VPG, TB. Guarantor: SWM.

Data availability statement

The data is available to interested parties by contacting the corresponding author: tbjarnsholt@sund.ku.dk

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: SWM, VP, HB are currently employed at Novo Nordisk and MAH was recently. TB and LB are paid consultants for Novo Nordisk. No other potential conflicts of interest were reported.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Danish Diabetes Academy, which is funded by the Novo Nordisk Foundation, (grant number NNF17S0031406) and the Innovation Fund (grant number 9065-00120B).

ORCID iD

Sofia Wareham-Mathiassen

References

Masson-Meyers

, Andrade TAM, Caetan GF, et al.

Experimental models and methods for cutaneous wound healing assessment

Int J Exp Pathol 2020; 101: 21–37.

Parnell

LKS

Volk

SW.

The evolution of animal models in wound healing research: 1993–2017. Adv Wound Care (New Rochelle) 2019; 8: 692–702.

Sullivan

Eaglstein

Davis

, et al. The pig as a model for human wound healing. Wound Repair Regen 2001: 9: 66–76.

Zheng

, Tesar, DB, Benincosa, Let al. Minipig as a potential translatable model for monoclonal antibody pharmacokinetics after intravenous and subcutaneous administration. MAbs 2012; 4: 243–255.

Seaton

Hocking

Gibran

NS.

Porcine models of cutaneous wound healing. ILAR Journal 2015; 56: 127–138.

McIntyre

Peacock

Akers

, et al. Initial characterization of the pig skin bacteriome and its effect on in vitro models of wound healing. PLoS One 2016; 11: e0166176.

Pabst

The pig as a model for immunology research. Cell Tissue Res 2020; 380: 287–304.

Debeer

, Le Luduec JB, Kaiserlian Det al. Comparative histology and immunohistochemistry of porcine versus human skin. Eur J Dermatol 2013; 23: 456–466.

Rose

Vistnes

Ksander

, et al. The panniculus carnosus in the domestic pig. Plast Reconstr Surg 1977; 59: 94–97.

10.

Weinstein

GD.

Autoradiographic studies of turnover time and protein synthesis in pig epidermis. J Invest Dermatol 1965; 44: 413–419.

11.

Seaton

Hocking

Gibran

NS.

Porcine models of cutaneous wound healing. ILAR J 2015; 56: 127–138.

12.

Stricker-Krongrad

Shoemake

Bouchard

GF.

The miniature swine as a model in experimental and translational medicine. Toxicol Pathol 2016; 44: 612–623.

13.

Præstmark

Jensen

Stallknecht

, et al. Skin blood perfusion and cellular response to insertion of insulin pen needles with different diameters. J Diabetes Sci Technol 2014; 8: 752–759.

14.

Præstmark

KA, Bengtsson H, Eyving B,

, et al. Influence of hypodermic needle dimensions on subcutaneous injection delivery – a pig study of injection deposition evaluated by CT scanning, histology, and backflow. Skin Res Technol 2012; 18: 447–455.

15.

McIntyre

Peacock

Akers

, et al. Initial characterization of the pig skin bacteriome and its effect on in vitro models of wound healing. PLoS One 2016; 11: 1–18.

16.

Swe

Zakrzewski

Kelly

, et al. Scabies mites alter the skin microbiome and promote growth of opportunistic pathogens in a porcine model. PLoS Negl. Trop. Dis 2014; 8: e2897.

17.

Strube

Hansen

Rasmussen

, et al. A detailed investigation of the porcine skin and nose microbiome using universal and Staphylococcus specific primers. Sci Rep 2018; 8: 12751.

18.

Wareham-Mathiassen

, Bay L, Glenting VPet al.

Injection site microflora in persons with diabetes: Why needle reuse is not associated with increased infections?

APMIS 2022; 130: 7: 404–416.

19.

Klindworth

, Pruesse E, Schweer Tet al. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res 2013; 41: 1.

20.

Fujita

Senda

Nakaguchi

, et al. Multiplex PCR using internal transcribed spacer 1 and 2 regions for rapid detection and identification of yeast strains. J Clin Microbiol 2001; 39: 3617–3622.

21.

Callahan

, McMurdie PJ, Rosen MJet al. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 2016; 13: 581–583.

22.

Ericsson

Crim

Franklin

CL.

A brief history of animal modeling. Mo Med 2013; 110: 201.

23.

Grice

Kong

Conlan

, et al.

(2009). Topographical and temporal diversity of the human skin microbiome.

Science 2009; 324: 1190–1192.

24.

Cundell

AM.

Microbial ecology of the human skin. Microb Ecol 2018; 76: 113–120.

25.

Woo

PCY

Lau

SKP

Teng

JLL

, et al. Then and now: Use of 16S rDNA gene sequencing for bacterial identification and discovery of novel bacteria in clinical microbiology laboratories. Clin Microbiol Infect 2008; 14: 908–934.

26.

Wade

Unculturable bacteria – the uncharacterized organisms that cause oral infections. J R Soc Med 2002; 95: 81–83.

27.

Hannigan

Grice

EA.

Microbial ecology of the skin in the era of metagenomics and molecular microbiology. Cold Spring Harb Perspect Med 2013; 3: a015362.

28.

Grice

, Kong H, Renaud Get al. A diversity profile of the human skin microbiota. Genome Res 2008; 18: 1043–1050.

29.

Grice

Segre

JA.

The skin microbiome. Nat Rev Microbiol 2011; 9: 244–253.

30.

Hodkinson

Grice

EA.

Next-generation sequencing: A review of technologies and tools for wound microbiome research. Adv Wound Care 2015; 4: 50–58.

31.

Kong

HH.

Skin microbiome: Genomics-based insights into the diversity and role of skin microbes. Trends Mol Med 2011; 17: 320–328.

32.

, Byrd AL, Clay Det al. Biogeography and individuality shape function in the human skin metagenome. Nature 2014; 514: 59–64.

33.

Costello

, Lauber CL, Hamady Met al. Bacterial community variation in human body habitats across space and time. Science 2009; 326: 1694–1697.

34.

Montagna

And

Yen

JS.

The skin of the domestic pig. J Invest Dermatol 1964; 42:11–21.

35.

Meyer

Schwarz

Neurand

The skin of domestic mammals as a model for the human skin, with special reference to the domestic pig. Curr Probl Dermatol 1978; 7: 39–52.

36.

Tuttle

MS.

Association between microbial bioburden and healing outcomes in venous leg ulcers: A review of the evidence. Adv Wound Care 2015; 4: 1–11.

37.

Byrd

Belkaid

Segre

JA.

The human skin microbiome. Nature Reviews Microbiology 2018; 16: 143–155.

38.

Bay

, Bjarnes, CJ, Fritz, BGet al. Universal dermal microbiome in human skin. mBio 2020; 11: 1. e02945–19.