Skin Aerosolization Predominance in a Pig Farm

Abstract

Bacterial genera present in several areas of a pig farm were analyzed using high-throughput sequencing techniques. Samples were collected from the skin and feces of animals, as well as from surfaces, water, and air. The analyses revealed a strong correlation between air and skin samples, supporting the idea that bacterial growth on skin is potentially a mechanism of aerosolization and airborne transport. A water–air transmission route also appears to be present, although the direction of the transport mechanism cannot yet be determined. Other potential routes, such as contact with surfaces or feces, seem to be less efficient.

1. Introduction

Sunlight beams in a room reveal a hidden world of airborne particles, fibers, and dust continuously suspended in still air. Fluid dynamics explains how particles smaller than a micron can remain floating in still air for hours, while submicrometric particles can even remain for days in perfectly still air. However, a small thermal imbalance or feeble airflow, for instance, caused by the nearby movement of an animal or a person, will provide an ascending thrust on non-bound micrometric and submicrometric particles, making them soar. The smaller a particle is, the easier it will be aerosolized and the longer it will remain airborne. The viscosity of air makes it a stable medium for airborne matter. However, this world of airborne matter smaller than the wavelength of visible light remains invisible to the human eye.

Airborne transmission of pathogens is a well-documented route of infection in both humans and animals [1]. While sneezing and coughing have classically and intuitively been identified as powerful mechanisms for aerosolization, other routine actions, such as breathing, singing, or speaking, have also been associated with aerosolization of pathogens like Mycobacterium tuberculosis [2] or SARS-CoV-2 [3,4].

In livestock production systems, particularly in pig farms, understanding the dynamics of airborne microbial transport is critical for preventing the spread of respiratory diseases. HE-FARM project, funded by Horizon Europe, aims to experimentally measure and prevent the efficiency of different transport mechanisms, particularly focusing on airborne transmission of pathogens, in the “farm-to-fork” sector.

A key component of this project involves characterizing the microbiome of the air within pig farms, specifically in the present work, and identifying the main sources contributing to its composition.

According to the logic of fomite transport of microorganisms, it was hypothesized that the determination of airborne microbiome profile could be a good reflection of the microbiota present in feces (intestine), pigsty surfaces, and skin of animals. Eventually, the aeromicrobiome (microorganisms present in air) was expected to be the result of aerosolization mechanisms acting on microbiomes. For instance, water splashing or water spraying may aerosolize water microbiome; defecation may release gut-associated microorganisms into the air; and movement of animals could mobilize microbes from surfaces, feces, or water.

2. Materials and Methods

2.1. Selection of Farm

The experimental design and the sampling of this study were performed on a pig farm located in Cuenca (Spain). This is an intensive pig production farm which specifically includes the maternity and weaning stages. This farm houses both sows and weaning piglets. Samples were collected from three different areas, including the maternity area, the service area where the sows are artificially inseminated (cubricontrol), and the weaning area where piglets are separated from the sows immediately after lactation.

Although this study included only non-invasive sampling of microorganisms from the skin and feces from the rectum of animals, it did not require ethical approval, as it did not involve any procedures that could cause harm or distress to the animals. However, all sampling was conducted in accordance with good animal handling practices, under the supervision of trained personnel, and with the explicit consent of the farm owner. Care was taken to minimize any disruption to the animals and farm operations during the sampling process.

2.2. Sampling Methodology

The characterization of the microorganisms present in the farm studied in HE-FARM included sampling from different sources, including air, surfaces, animal skin, feces from the rectum, and water.

Samples from surface, skin, and feces were taken using conventional sterile swabs. Surface samples were collected from various areas of the enclosure housing the pigs, including the bars of the cage, the surrounding walls, and other surfaces frequently contacted by both sows and piglets. Skin swabs were obtained from the dorsal area of the animals, and fecal samples were collected directly from the rectum to ensure sample integrity and avoid environmental contamination. Surface samples were then introduced into a PBS solution, and samples taken from the animals were kept in 1 mL of stabilization solution, DNA/RNA shield (Zymo Research, Irvine, CA, USA). Water samples were collected by transferring a small fraction of water into a sterile recipient for further analysis. Water samples were collected from the drinking troughs used by the pigs.

Air samples were taken using two different methodologies. The first one consisted of a filtration system through polytetrafluorethylene (PTFE) narrow-pore films connected to a vacuum pump. PTFE filters can collect particles with high efficiency, maintaining the recovery efficiency of viable virus [5]. This filter was placed in the corridor of the farm stages at a height of around 1.8 m (Figure 1). Each PTFE filter sampling lasted for 2 h, and the filters were collected in tubes containing 2 mL of phosphate-buffered saline (PBS) and were kept at 4 °C until sample processing.

The second technology is the Bioaerosol Fast Sampler (BIAFTS) (Counterfog S.L., Madrid, Spain), a device that collects airborne particles into a liquid medium. Counterfog technology uses dynamic aerosol aggregation, creating dynamic fog cones of liquid composed of nanometric droplets, which, when projected, capture agents transported in the air [4,6]. This device was placed on the floor of the farm stages (Figure 1). Each air sample was collected in 2 min and was ready to be analyzed in the laboratory. The air samples were kept refrigerated at 4 °C until DNA extraction. These two methodologies, based on basically different physical principles, have proven to be effective bioaerosol samplers, even for the smallest viruses [4]. Good correlation between species sampled by both types of technologies supports their effectiveness [4].

2.3. Sample Processing

All samples were processed in the laboratory. The different volumes of supplemented PBS containing the air samples from BIAFTS were first concentrated using Amicon^® Ultra-15 50 K (Merck, Darmstadt, Alemania) up to approximately 2 mL. A total of 350 µL was mixed with one volume of Lysis buffer (Promega, Madison, WI, USA) for DNA extraction. On the other hand, the PTFE samples were vigorously vortexed, and 350 µL was diluted 1:1 in Lysis buffer.

DNA was isolated from all samples using the ZymoBIOMICS DNA Miniprep Kit (Zymo Research, Irvine, CA, USA) following the manufacturer’s instructions, with modifications for air and water samples due to their liquid nature. These modifications consisted of concentrating the sample by evaporation to a final volume of 250 µL before starting the DNA extraction protocol.

Table 1. Sample distribution by sample type across the three farm rooms.

Total	Weaning	Maternity	Insemination
Animal
10	5	5	-	Piglet Feces
10	5	5	-	Piglet Skin
15	-	5	10	Sow Feces
15	-	5	10	Sow Skin
Air
3	-	2	1	BIAFTS
3	2	1	-	Filter
4	-	3	1	Surface
3	-	2	1	Water
63	12	28	23	Total

shows the distribution of the 63 samples across the three rooms of the farm and classified by sample type, where 28 corresponded to the maternity room, 23 to the insemination room, and 12 to the weaning room.

2.4. Metagenomics Analysis

A total of 67 samples were obtained from the farm, out of which sufficient DNA was extracted from 63 samples for 16S sequencing.

The V3-V4 region of the 16S rRNA gene was sequenced using the primers “CCTAYGGGRBGCASCAG” and “GGACTACNNGGGTATCTAAT” by an external service (Novogene, Cambridge, UK). Sequencing libraries were generated using TruSeq^® DNA PCR-Free Sample Preparation Kit (Illumina, San Diego, CA, USA), according to the manufacturer’s instructions, and index codes were added. Library quality was assessed using the Qubit@ 2.0 Fluorometer (Thermo Scientific, Waltham, MA, USA) and the Agilent Bioanalyzer 2100 system (Agilent, Santa Clara, CA, USA). Finally, sequencing was performed on an Illumina NovaSeq platform, generating 250 bp paired-end reads. The sequencing of the 16S region was performed to assess the microbial composition of the collected samples.

Raw reads from the 63 samples were processed and quality filtered using Qiime2 (version 2022.11) [7]. Sequences shorter than 400 bp were removed to ensure high-quality data. Then, the filtered sequences were processed into Amplicon Sequence Variants (ASVs) using the DADA 2 tool [8]. Taxonomic classification was conducted using the Greengenes2 classifier “https://greengenes2.ucsd.edu/ (accessed on 30 August 2024)”), generating the corresponding taxonomy for each ASV.

The further analyses were carried out in RStudio (version 4.4.1) [9], using the phyloseq package [10]. In the first place, the abundance of genera by samples was calculated, and Beta-Diversity through Principal Coordinates Analyses based on Bray–Curtis dissimilarities and Analyses of Similarities (ANOSIM) [11] were assessed. Beta diversity refers to the variation in species composition between different environments or samples. It measures how similar or different communities are across habitats and helps to understand patterns of species turnover or replacement across spatial or ecological gradients.

3. Results

3.1. Maternity

Figure 2 shows the ten most abundant bacterial genera in the maternity area. There is a general correspondence between the two air samples collected using BIAFTS and the one using a PTFE filter, except for the Psychrobacter genus, which is abundant in the two BIAFTS samples, but does not appear in the PTFE filter sample. Psychrobacter is also present in a water sample, but not detected elsewhere.

Terrisporobacter was present in the air for all kinds of samplers and on the skin of both sows and piglets, as well as on pigsty surfaces. It was found in the feces of the sows, but only in some piglets. It was not detected in the water.

Lactobacillus was present in all samples from air, skin, and surfaces; however, it was detected in only 40% of fecal samples from both sows and piglets.

Pseudomonas was predominantly detected in the water samples, with one sample showing an overwhelming dominance of this genus. In contrast, Pseudomonas showed low abundance in surface samples, as well as in other sample types, where it was represented by only a few ASVs and was not among the most prevalent bacterial genera.

Escherichia was moderately present across all sample types; however, it became predominant in 20% of piglet fecal samples.

Prevotella, Clostridium, Faecousia, and Treponema were ubiquitous across air, skin, feces, and surfaces but were absent in water samples. Their relative abundances vary depending on the sample type. For instance, the relative abundance of Clostridium was lower than that of Prevotella, in general, on skin and surfaces, and this difference was even more evident in feces. In contrast, their abundances in the air were quite similar.

As shown in Figure 3, microbial composition clusters strongly by sample type. In particular, there was a clear correspondence between the microbial communities in the PTFE filter sample, one of the surfaces, and those found on the skin of both sows and piglets, as evidenced by their close spatial proximity in the PCoA plot. Most skin samples were grouped around PCo1 ≈ −0.2 and PCo2 ≈ 0.0 coordinates, while this air sample was also located in a similar position, suggesting shared microbial profiles. Notably, BIAFTS samples slightly deviate from the main skin cluster, moving toward the other surface sample in the PCo1 coordinate, likely due to the similar composition of the samples. ANOSIM analyses showed a significant difference between sample types (p = 0.001).

Fecal samples showed more variability than skin samples. Those formed a consistent cluster, indicating that the skin microbiome has its own profile when it is not in contact with fecal material.

3.2. Insemination Area (“Cubricontrol”)

The relative abundances of the ten most abundant genera in the insemination area are shown in Figure 4. The microbial composition of the skin appears relatively uniform and closely resembles that of the air samples (BIAFTS). On the contrary, the water sample displayed a markedly different microbial profile. For instance, Epilithonimonas accounted for over 60% of the relative abundance in water, but it was absent from feces, skin, and air samples. According to the data this genus only marginally appeared in the water sample from the maternity area (represented by only a few ASVs and not among the most prevalent bacterial genera in this area).

Corynebacterium was abundant (around 20% of total counts) in all samples from skin, surface, and air, yet it was completely absent from feces. On the other hand, Porphyromonas was consistently present in feces, with great variability between individual sows, but was absent from air or the pigsty surface, and only marginally present on skin.

Both Clostridium and Prevotella were detected on skin, on the pigsty surface, in feces, and in the air, and were also observed in the maternity area. However, in the insemination area, Clostridium was more abundant than Prevotella on the skin of sows, pigsty surface, and in air samples, while in feces, the opposite trend was observed. It is remarkable that this tendency was inverted in the sow skin samples collected in the maternity area.

The analysis of Principal Coordinates based on the Bray–Curtis dissimilarity matrix of the samples collected in the insemination room, shown in Figure 5, shows that samples from sow skin and BIAFTS air clustered closely together, indicating a high degree of similarity in their microbial profiles. Otherwise, fecal samples formed a distinct cluster, clearly separated from skin and air samples, which suggested a different microbial composition. PERMANOVA ANOSIM analyses showed a significant difference between sample types (p = 0.001).

The surface sample from the pigsty was separated along PCo2 from skin and air, but remained relatively close on the PCo1 axis, suggesting partial similarity. Meanwhile, the water sample appeared as an outlier with two fecal samples, which are relatively different from the rest, and this could be due to the similarity in the microbial composition of the remaining taxonomic composition, even if those were not the most abundant.

3.3. Weaning Area

In this area, air samples were collected using only PTFE filters, once again showing a microbiome composition quite similar to that of piglets’ skin. Figure 6 shows an increase in Lactobacillus and a reduction in Terrisporobacter in air samples compared to piglet skin.

Parabacteroides was highly prevalent, with an abundance larger than 30% in one of the five fecal samples, but was marginal or simply absent in the remaining fecal samples. It was, indeed, marginally present in all air and skin samples.

Treponema, which was relatively abundant in the feces of both sows and piglets, was found in a small percentage (a few ASVs) in only two fecal samples (out of five) in this area. It seems that fecal samples containing Treponema did not contain Parabacteroides.

Escherichia was not among the top ten genera in this area; however, as shown in Figure 7, it was detected in air filter samples and one fecal sample. It appeared only marginally in skin samples but was absent from the other fecal samples.

Prevotella largely dominated the fecal microbiota in the weaning area, except in those samples where Parabacteroides was present.

Streptococcus was found on skin, in air, and in fecal samples, lacking Parabacteroides. It was not detected in fecal samples where Parabacteroides was present, reinforcing the apparent negative association between these genera.

Finally, principal coordinates analyses in this area revealed that skin samples formed a cluster, while the fecal samples were scattered. Additionally, the air samples were located nearby, although slightly separated along both PCo1 and PCo2. On the other hand, ANOSIM analyses did not show a significant difference between sample types (p = 0.182).

Finally, the heatmap shown in Figure 8 shows a different clustering of microbiome profiles by room. There are only two genera, Terrisporobacter and Prevotella, present in the three areas, suggesting a core microbiota influenced by a shared environment. On the other hand, other genera showed room-specific patterns, pointing out environmental differences, in this case, animals with different ages (piglets or sows) and animal density in the rooms.

4. Discussion

The present work shows, contrary to what was expected, that the strongest correlation of aeromicrobiome is found with the skin of the animals. This is evidence for an effective mechanism for aerosolization: the skin cell ejection due to the natural renewal of skin cells. The outermost cells of the skin are continuously ejected, becoming airborne. In this work, we show evidence that bacteria on the skin of the animals are also ejected and become airborne in this process. Ejection of skin cells is clearly an effective defense mechanism preventing penetration of microorganisms through the skin; however, as a side effect, it provides an effective mechanism for aerosolization and, therefore, transport from animal to animal.

In addition to the veterinary relevance of the results of the present study, there is a clear repercussion on medical sciences and transmission of infectious diseases. In fact, pig skin is known to be the most similar to human skin [12]. Moreover, the results can be reasonably extrapolated to all animal species, provided they have skin. Even more, the epithelium of the whole respiratory system would also be an effective emitter of bio-aerosols, affecting the transmission of respiratory diseases.

The main focus and concern of the HE-FARM project and of the sampling performed in this work is air microbiota; in other words, air samples and the microorganisms present in the air that can be easily transported from one part of the farm to another.

The present study shows that the microbiota of skin samples predominates in the airborne microbiome profile of this farm. Actually, this pattern was consistently observed across maternity, insemination, and weaning areas, where genera identified in skin samples were more similar to those found in air samples than those found in fecal or water samples.

This overlap in microbial composition observed between air and skin samples supports the idea that skin shedding and animal movement would drive microbial aerosolization. These results agree with previous findings by Krämer et al. [13,14], who found that the nasal microbiota of pig farmers is similar to the airborne microbiome profile in barns.

Although fecal samples contained a diverse microbial community, our results point out that their direct contribution to the airborne microbiome was limited compared to that of skin. While some genera commonly associated with the gastrointestinal tract, such as Clostridium, Prevotella, and Treponema, were found in both feces and air samples, their relative abundance and distribution patterns differed notably. In particular, these genera were typically more dominant and variable in feces, whereas in air, they appeared at lower, more uniform levels. Our results partially align with Yan et al. [15], who detected gut-associated genera in bioaerosols from pig barns. However, unlike Yan et al., who inferred fecal origin without direct comparison to host-associated microbiota, our work included direct sampling from rectal content and skin. This allowed us to distinguish that skin microbiota more closely resembled airborne communities than fecal microbiota.

Regarding the similarity between air and water samples, only Psychrobacter was predominant in both the water samples and the two BIAFTS air samples in the maternity area. However, it was only marginally present in the skin samples from the same room. This suggests a potential water–air–skin transport mechanism. Nevertheless, the directionality of this mechanism (from water to air and skin, from skin to air and water, or otherwise) remains unclear. In the insemination area, the case is different since this genus was marginally and proportionally present in water, air, and skin samples.

The differences in aerosolization observed here are also consistent with the findings of Marin et al. [16], who proposed that abundant taxa in slurry or water do not necessarily aerosolize efficiently, particularly under good hygiene conditions.

In summary, the predominance of skin-associated microorganisms in the aeromicrobiome, compared to those of fecal or water origin, may be explained by several biological and physical factors. Unlike feces, which are collected directly from the rectum and, thus, remain less exposed to air, the skin is in constant contact with the surrounding environment. Animal movement and surface contact further facilitate the aerosolization of skin-associated microbes. Additionally, although fecal matter comes into contact with the air after defecation and may contribute to microbial dispersion, the proportion of fecal-derived microbiota identified in airborne samples was lower than that of skin origin. Regarding the water–air route, while we acknowledge its potential role in microbial aerosolization, our current dataset does not include sufficient sampling to explore this mechanism in depth, and the dynamics behind it remain unclear.

Abundances of genera in the air are quite similar to those on the skin of animals in the respective area. Skin microbiota seems to evolve on piglets from maternity to weaning, as well as on sows from insemination to maternity. In all cases, the composition of bacterial genera on the skin of animals is uniform within the area. In other words, animals that share the same air environment also seem to share the same microbiota.

Finally, the fecal microbiota of piglets seems to be more diverse during the maternity phase but tends to become more uniform in the weaning area. This may reflect different developmental states and, eventually, suggests that the process of microbial homogenization in feces occurs more slowly than in the air–skin system.

Some limitations of this study relate to the low number of air, water, or surface samples in certain areas, due, in part, to some samples not meeting the quality criteria required for metagenomic analysis. This limited sampling reduces the ability to fully assess the role of water and surfaces in microbial aerosolization. Future studies could benefit from more extensive sampling across different farm zones, also considering seasonal variations, temperature, and ambient humidity. Additionally, investigating these mechanisms in other types of farms, such as extensive (open range) systems, would be valuable. In such open environments, microbial communities may disperse more widely due to air currents, resulting in greater dilution and additional influencing factors that should be considered.

5. Conclusions

A first conclusion from this study is that skin desquamation is a powerful mechanism of aerosolization and bacterial dispersion. The bacterial genera detected in air samples are largely shaped by those present on animals’ skin. A water-to-air transmission route also seems to be present, although the directionality of this mechanism remains unclear. Other routes, such as contact with surfaces or even feces, seem to be less efficient in this sense.

The relative abundance of bacterial genera on the animal’s skin seems to be consistent within each area but differs across areas. This suggests that animals within the same area share a common skin microbiota, likely influenced by continuous microbial exchange through the air they share.