home Research Household air microbial community resembles that of household surface and occupant’s skin? Maybe, maybe not…

Household air microbial community resembles that of household surface and occupant’s skin? Maybe, maybe not…

Household air microbial community resembles that of household surface and occupant’s skin? Maybe, maybe not…

By Marcus Leung (Twitter @leungmarcus)

Indoor microbial communities, such as those found in indoor air and indoor surfaces, have been shown to be closely connected to microbial assemblages on humans. Also, re-suspension of house dust from surfaces following routine residential activities may provide another source of microorganisms in the air. Thus, it appears that there exists a close connection between the skin, surface and air communities within a residential household. However, we are not aware of previous works concurrently analyzing all three ecosystems within a household. We believe that attempts to unravel their connections will increase our understanding in improving occupants’ health and living conditions by changing indoor building design and other characteristics.

Our research group at the School of Energy and Environment at the City University of Hong Kong has been focusing on the relationships between microbiomes of residential surfaces, air, and its occupants’ skin. Here we have summarized parts of our latest work (not all observations from manuscript are featured here) characterizing household air and surface microbiomes, as well as skin microbiomes of occupants of households throughout Hong Kong. More importantly, we hypothesized that microbiomes on skin and residential surfaces disperse into the air, and generate a “microbial fingerprint” of occupant skin microbiomes. If this is the case, we should expect to see that:

  1. indoor-air communities consists of a sizable portion of skin-associated microbial members
  2. indoor-air communities are household-specific and resembles occupants’ skin and household surface microbiomes
  3. source tracking of communities within households would show that occupant skin and household surfaces are major sources of air microbiomes

(Note – these have been color coded and then the text below has been highlighted in the matching color when commenting on these).

We collected a total of 76 air samples (four air samples per household: living room, kitchen, toilet, bedroom), 152 surface samples (eight surfaces per household: remote control, TV screen, kitchen ventilator, fridge door seal, toilet flush button, shower curtain, bed headboard, bed blanket), and 200 skin swab samples (five skin swabs per individual: forehead, left/right forearm, and left/right palm. Microbial community for each sample was determined by targeting the V4 region of the bacterial 16S rDNA and sequencing was performed on the Illumina MiSeq platform. Following quality filtering, chimera, chloroplast, mitochondria, and dataset singleton removal, high-quality sequence reads were clustered into operational taxonomic units (OTUs) by the UPARSE pipeline, and taxonomic information was assigned to them using Greengenes database as a reference.

Our results show that some of the most abundant genera in air were indeed those commonly associated with the skin, consistent with our first statement above. Also, more OTUs were shared between skin and surface communities than any other pairwise comparisons, and surfaces that were frequently touched also contained higher abundances of skin-associated genera.

UniFrac (abundance-weighted) analysis reveals that skin, surface, and air microbiomes clustered by households. However, when determining whether communities between ecosystems (air/surface, surface/skin, skin/air) were more similar within households than between households (by weighted UniFrac), microbiomes of household air and their occupants were no more similar than that of household air and non-occupants, which does not support the second statement above. While SourceTracking predicted with high accuracy, that skin and surfaces could be potential sources of each others’ microbiomes, it was rather weak in predicting surface and air microbiomes as potential sources for each other, and actually scored far worse in allocating residing occupants’ microbiomes as potential sources of respective air microbiomes, regardless of household occupancy. From these and other observations, we were not able to provide strong support for our hypothesis, that members of the skin and surface communities leave a specific “microbial fingerprint” on the air of residences.

Based on the results pertaining to statements ii) and iii), why did we still see abundant skin-associated genera in air samples? Also, what is driving the observed household-specific clustering of microbial communities? For the first question, we hypothesize that outdoor air, which has been shown to strongly influence indoor air, as a possible alternative source of microbial communities. Therefore, we took outdoor bioaerosol data from our previous work on the Hong Kong subway system, and treated them as sources for the residential indoor air community “sinks” using SourceTracker. Indeed, the software predicted that over 50% of the microbial communities originated from outdoor air. For the second question, we postulate that human sources other than the skin (e.g. oral/nasal cavities and gut) contribute to the observed communities. We classified genera based on their association with gut, oral, and skin environments, and re-performed ANOSIM test separately for these three environments. Results show that while all three tests revealed significant community clustering according to household, the clustering magnitude, as expressed by ANOSIM R value, was strongest for oral, followed by gut and skin.

These results may suggest that previous assessments on the roles of skin microbiomes on indoor microbial assemblages may be overestimated. However, we must stress that the households in Hong Kong, which are notorious for its small sizes, may be subjected to other emission sources. Also, additional analyses, such as controlling for ventilation methods and household occupancy levels, should fine-tune the roles of these factors on residential bioaerosols. Nonetheless, we encourage increased focus of other human body parts (oral/nasal cavities and gut) in regards to their relationships with indoor microbiomes, as their proportions of contributions to indoor microbial communities may vary depending on different residential settings.



2 thoughts on “Household air microbial community resembles that of household surface and occupant’s skin? Maybe, maybe not…

  1. Nice work! Thanks for starting to put things in perspective.
    It has been known for centuries that the indoor air is composed of microbial elements fro occupants, from outdoor, and also from the building microbiome itself. Humans are just one factor in the whole picture. Denying this fact ise egocentric, and little supported scientifically. New technologies applied to building ecology are mainly providing a light version of a well-established reality.
    A common mistake that molecular biologists make during building surveys is to only sample from visible surfaces. Buildings are more than what we see. Most active microbial communities (the building microbiome) grow in either hidden building structures (walls, ceiling and floor structures) or in naturally moist constructions as sinks, pipes, showers, crawlspaces, etc. A great deal of microbes that you cannot account for in your paper are probably spread by air leakages from hidden growth in moist hidden structures.

    1. Thank you for your contribution! Interesting point and I agree! I personally think in general, a greater microbial community characterisation of additional surfaces (both visible+hidden ones) will come naturally as we (as a scientific community) start to propose other non-human factors in shaping indoor microbiomes. In fact, some of the homes we sampled are distinctively designed with wooded tiles, such that cracks (some as wide as 1/8 of an inch) can form between individual wooden pieces over time.

Leave a Reply

%d bloggers like this: