Characterization of the total and viable bacterial and fungal communities associated with the International Space Station surfaces

The International Space Station (ISS) is the largest human space platform in low Earth orbit (~ 400 km above Earth’s surface) and for the last 17 years it has been continuously inhabited by an international community of astronauts performing space research. The ISS is a hermetically sealed closed system, subjected to microgravity, radiation, elevated carbon dioxide, and the recirculation of air through HEPA filters and is considered an “extreme environment” [1, 2]. Microbes are known to survive and even thrive in extreme environments, and the microbes that are present on the ISS may have existed since the inception of the ISS while others may be introduced each time new astronauts or payloads arrive.

Since the beginning of the ISS, routine microbial monitoring of surfaces, air, and water has occurred using culture-based techniques as per the National Aeronautics and Space Administration’s (NASA) operations and maintenance requirement procedures [3]. However, culture-based analysis limits our understanding of the diversity of microbes that grow and thrive on the ISS because only a small fraction of organisms in a given environment can be cultured under standard laboratory conditions [4]. Molecular methods, such as quantitative polymerase chain reaction (qPCR) and targeted amplicon sequencing, which can identify and quantify both culturable and unculturable organisms provide a more thorough assessment of what is actually present and in what amounts [5]. However, while it has been recently shown as a proof of concept that PCR [6] and amplicon sequencing can be performed on the ISS [7, 8], microbial monitoring of the ISS with molecular-based methods is not routinely used because of the lack of simple, compact, and reliable sample processing instruments onboard the ISS. Once such devices are available, rapid, real-time microbial detection, functional analysis are possible for the long duration missions, but baseline information about the ISS environmental microbiome is still needed.

The importance of cataloging the ISS microbiome, which consists of both culturable and unculturable microbes, parallels the surge in research into the “built microbiome” here on Earth. Emerging studies on the microbiome of homes [9–11], offices, classrooms, museums [12, 13], and hospitals [5, 14, 15] have revealed an assemblage of bacteria, fungi, viruses, and protozoa unique to that indoor environment that are influenced by a variety of factors such as building design, ventilation, humidity, air pressure and flow, occupant numbers, or activities performed [16, 17]. Specific microbes in these indoor spaces have been shown to impact human health by influencing our susceptibility to allergies, infectious diseases, or sick building syndrome [18]. The influence of the indoor microbiome on human health becomes more important for astronauts during flights due to altered immunity associated with space flight [19, 20] and the lack of sophisticated medical interventions that are available on Earth.

In light of an upcoming new era of human expansion in the universe, such as future space travel to Mars, the microbiome of the closed space environment needs to be examined thoroughly to identify the types of microorganisms that can accumulate in this unique environment, how long they persist and survive, and their impact on human health and spacecraft infrastructure. For this reason, the National Research Council (NRC) Decadal Survey recommended that NASA establish a coordinated, large-scale Microbial Observatory program within the ISS platform [21]. As part of this NASA initiative, the microbial communities on ISS surfaces from eight defined locations over three flight missions, spanning 14 months, were characterized using culture-based techniques, qPCR, and amplicon sequencing of the 16S rRNA gene and internal transcribed spacer (ITS) region. Before DNA extraction, half of the sample was treated with propidium monoazide (PMA) so that the microbiome of intact/viable cells (PMA treatment) could be characterized. The PMA-untreated samples yielded information about the total microbial population (including free DNA, dead cells, cells with a compromised cell membrane, intact cells, and viable cells). PMA binds to DNA, making the DNA unavailable for amplification during PCR steps [22]. Due to its higher molecular weight and/or charge, PMA cannot penetrate into cells that have an intact cell membrane (i.e., viable) but can bind to free floating DNA or DNA inside cells with a compromised cell membrane (i.e., dead cells) [22, 23]. It is in this way that many studies have utilized PMA to distinguish between intact/viable cells and compromised/dead cells [2, 24–26].

This comprehensive analysis of the ISS microbiome was used to assess how microbial communities change over time (temporal distribution) and throughout the ISS (spatial distribution). In addition, the ISS environmental microbiome data were compared with other Earth built environmental microbiome data such as the Earth Microbiome Project [27], Hospital ([28], Qiita study 10,172), and Office microbiome ([28], Qiita study 10,423). The implementation of novel molecular techniques to monitor intact microbial populations in this unique environment opens a possibility for broadening the current surveillance practices to maintain the health of the crew and to promote advances in deep space human habitation in the future.

The cultivable microbial load from all flight samples and their distribution patterns at various locations are depicted. The average number of bacteria cultured on blood agar (BA) and R2A plates was similar between F1 and F2 but higher at F3 (though this trend was not statistically significant) (Fig. 2a). There were no statistically significant differences in the average bacterial load across the eight locations (Fig. 2b); however, the locations that exceeded 10¹⁰ CFU/m² during at least one flight sampling event were L1 (port panel next to cupola), L5 (overhead 4), L7 (lab 3 overhead), and L8 (crew quarters), with the lowest counts (less than 10⁴ CFU/m² in at least one sampling event) found at L3 (AREM) and L6 (PMM). Overall, the number of bacteria (combination of R2A and BA growth) isolated from the ISS from all 24 samples ranged from 6.7 × 10³ to 7.8 × 10¹⁰ CFU/m².

Fungi were also cultured from the ISS, ranging from 1.1 × 10⁵ to 3.1 × 10⁸ CFU/m². While, there were no statistically significant differences in fungal load over time, the highest average was found during F1 and the lowest at F2 (Fig. 2a). Similar to what was observed with bacterial counts, no differences in fungal counts were evident across the eight locations (Fig. 2b). When compared to bacteria, the fungal population was lower by 2 to 3 logs at all locations except at L6 where fungal load was 100-fold more than bacterial load (Fig. 2b). Due to the high variability between samples, there were no statistically significant differences overall in the average cultivable counts of bacteria (BA and R2A plates) from all 24 samples compared to the average fungal counts measured from the same 24 samples (P > 0.05).

Of the total bacterial and fungal isolates that grew, 133 bacterial isolates and 81 fungal isolates were identified by Sanger sequencing (16S rRNA gene for bacteria; and ITS region for fungi). The bacterial isolates belonged to three phyla: Actinobacteria, Firmicutes, and Proteobacteria. At the genus level, the most predominant genera were Staphylococcus (26% of total isolates identified), Pantoea (23%), and Bacillus (11%) and at the species level, Staphylococcus aureus (10%) and both Pantoea conspicua (9%) and Pantoea gaviniae (9%) (Additional file 1: Figure S1A). Although bacterial counts were similar across all flights (Fig. 2), only members of the family Enterobacteriaceae were cultured from F3 samples (Additional file 1: Figure S1A). S. aureus isolates were tested with the Vitek 2 system (BioMerieux, France) and found to be methicillin-sensitive; however, these isolates were resistant to penicillin, erythromycin, gentamycin, and tobramycin [28]. The whole genomes of 20 biosafety level 2 strains, isolated from these samples, have been sequenced and are publicly available [29].

The fungal population was dominated by Rhodotorula mucilaginosa belonging to the family Sporidiobolaceae (41% of the 81 examined fungal isolates) and Penicillium chrysogenum (15% of 81 fungal isolates) (Additional file 1: Figure S1B). The whole genome of one Aspergillus fumigatus strain (isolated from F1, L1 [cupola] sample) was sequenced, its virulence characterized, and this information reported elsewhere [30, 31].

The 16S rRNA gene and the ITS region were targeted in PMA-qPCR to measure intact/viable bacterial and fungal burden, respectively. The changes were not significantly different (P > 0.05), although the average number of bacterial 16S rRNA gene copies trended toward increase from F1 to F3. On the other hand, the ITS region amplicons decreased over time with F3 being statistically significantly lower than F1 (Fig. 3a). While the average bacterial (Fig. 3b) and fungal (Fig. 3c) load fluctuated across locations, there were no statistically significant differences in microbial load among different locations. Overall, bacterial loads appeared to be highest at L4 and L5 and lowest at L6, with fungal loads appearing to be highest at L1, L4, L5, and L7 and lowest at L2. The average number of bacteria present on the ISS during this study was 3.1 × 10⁹ 16S rRNA gene copy number/m² and 7.1 × 10⁸ ITS copy number/m² for fungi. A comparison between CFU and gene copy number revealed that on average, 46% of total intact/viable bacteria and 40% of intact/viable fungi could be cultured, while the remainder were viable but yet to be cultured (Additional file 1: Figure S1C).

qPCR was also performed on samples that were not treated with PMA to assess the overall microbial load, which includes intact/viable cells and compromised/dead cells. The average 16S rRNA copy number was 7.1 × 10⁹/m² and the average ITS copy number was 5.1 × 10⁸ m² in these non-PMA-treated samples (Additional file 1: Figure S1D). When the 16S rRNA gene copies were summed up for all locations and all flights, no significant difference was observed with and without PMA (Additional file 1: Figure S1D). This calculation may be affected by the artificial inflation of high copy numbers in samples treated with and without PMA. For example, the highest copy numbers of a sample with 10¹⁰ copies per m² (for example, Flight 1, Location #2) might mask the samples with 10⁷ copies per m² (for example Flight 1, Location #6). The difference in microbial populations between PMA and non-PMA-treated samples were substantial when the data from individual locations were considered. In general, ~ 0.68% (example: Flight 2; Location #3) to ~ 92.8% (example: Flight 3; Location #5) of the microbial load was present in the PMA-treated sample and therefore considered “viable” (Additional file 1: Figure S1E). Similar reduction in microbial abundance in PMA-treated samples when compared to untreated samples was reported in NASA spacecraft assembly facility (SAF) clean room floors (4 to 21%; [24]), Lunar Mars Analog Habitat floors (10 to 40%; [32]), and HEPA filter particulates of ISS and SAF (1.7 to 66.8%; [2]).

After processing the raw data from 48 samples (24 PMA samples and 24 non-PMA-treated samples) and various wipe and reagent controls, amplicon sequence variants (“ASVs”) (a higher resolution analogue of the ubiquitous “OTU”) [33] that had a cumulative sum of more reads in the controls than the cumulative sum of all the samples, were removed from the dataset. Next, ASVs that were found to be statistically significantly higher (ALDEx2 test, P < 0.05) in the control group than in the sample group were further removed from the dataset. Additional file 2: Dataset 1A summarizes the ASV read count in each sample and in each control after the above quality control measures were implemented. Next, the program “SourceTracker” was used to predict what percent of reads in the samples were unique to the samples and what percent were from “contaminating” ASVs (i.e., those were represented in a Dirichlet model trained from the control samples). Additional file 1: Figure S2B summarizes the results from SourceTracker and shows that contamination was negligible in 32 out of 48 samples and for the remaining 16 samples; the contamination was less than 6% of the total sequences. A canonical correspondence analysis (CCA) plot verifies that the bacterial communities of the controls were indeed different than those of the samples (Fig. 4). The ASV table used for downstream analysis, after the above quality control measures were implemented and after verification that the samples represented a unique microbiome, different than that of the controls, is presented in Additional file 2: Dataset S1A.

A summary of read counts, number of ASVs, and most abundant taxa in the samples are presented in Additional file 3: Table S1. Microbial populations (16S rRNA gene copies) from eight locations over the span of 14 months were calculated from PMA-treated samples (viable/intact bacteria) and non-PMA treated samples (dead/compromised bacteria). Figure 5 shows the proportion of different taxa, summarized to the family level, found in these samples and the variations in their viable populations. In both the PMA- and non-PMA-treated groups, Enterobacteriaceae dominated and made up a little over 50% of the sequences detected in all 24 samples combined, followed by Methylobacteriaceae (~ 13%) and Staphylococcaceae (~ 10%).

Of interest was whether the ISS environmental microbiome, and especially the most abundant taxa, changed over time and across locations. The taxa present in the non-PMA-treated (Fig. 6a) and PMA-treated (Fig. 6b) groups showed the same temporal progression: The relative abundances of Enterobacteriaceae was highest during F3 and lowest during F2, whereas Methylobacteriaceae was the lowest during F3 and highest during F1. Paenibacillaceae and members of the class Bacilli and order Bacillales had high relative abundances during F2, and almost negligible amounts during F1 and F3. Interestingly, F1 and F2 had higher relative abundances of sequences that could not be identified, compared to F3. Statistical analysis using ALDEx2 confirmed that the relative abundances over the three flight sessions were different for all taxa shown, expect for Paenibacillaceae in the non-PMA group (Fig. 6a) and Paenibacillacae, Staphylococcaceae, and Sphingomondales in the PMA-treated group (Fig. 6b).

Unlike the differences observed over time, no statistically significant differences in relative abundances were observed between the eight locations (Fig. 6c, d); however, some interesting trends are worth noting: Enterobacteriaceae was well represented at each location, with the highest relative abundance observed at L1, L5, and L6. Methylobacteriaceae was also highly represented across locations except for L2, L5, and L6.

The barplot in (Additional file 1: Figure S3) provides a more detailed representation of the relative abundances of family level taxa in each sample. Upon visual inspection, it appears that bacterial diversity within a sample was the lowest during Flight 3, which consisted predominately of Enterobacteriaceae, and was the highest during F2. This observation was statistically confirmed (P < 0.05) by calculating (i) Shannon’s diversity index, which measures both taxa presence and relative abundance and (ii) taxon richness, which reports the number of unique taxa in a sample (Additional file 1: Figure S4A and B). Noteworthy, both alpha diversity (measured with Shannon’s diversity index and taxa richness) and beta diversity (measured with ALDEx2) showed no differences between PMA and non-PMA treated samples across all three flights, suggesting that the DNA recovered from the ISS were from intact/viable organisms.

When the ASVs were summarized to the genus level, 121 taxa were detected, 77 of which could be assigned to known genera (Additional file 1: Figure S5). Of those 77 genera, 68% of them are known constituents of the human microbiome and the remaining 32% are found in environments such as soil and water.

Amplicon sequencing of the fungal ITS region was performed on samples collected during F1 and F2. Since F3 exhibited low abundance of cultivable fungal counts, it was not possible to generate amplicons for further sequencing. Similar to what was done with bacterial sequences, OTU counts that were higher across controls compared to samples were removed from the dataset. For one OTU, even though the cumulative read count of the controls was 33 times lower than the samples, since the count was 300,000, it was removed from the dataset. Additional file 2: Dataset S1B shows the fungal OTU table that was used for analyses after OTUs associated with controls were removed. The SourceTracker results for the mycobiome are shown in Additional file 1: Figure S6A and the total OTU read count for the sample and control wipes is presented in Additional file 1: Figure S6B.

The fungal population consisted of four genera plus members belonging to one phylum, two classes, and four families, in addition to sequences that could not be classified (Additional file 1: Figure S7). The temporal and spatial distribution of the five most relatively abundant taxa are shown in Additional file 1: Figure S8. With the exception of Sporidiobolaceae which was higher in F2 compared to F1, there were no other statistically significant differences in fungal taxa between flights.

Unlike bacteria, fungal diversity within samples (i.e., alpha diversity) did not change between Flight 1 and 2 (Additional file 1: Figure S4C and D). Similar to what was observed with bacteria, alpha and beta diversity were similar between the PMA- and non-PMA treated samples across these two flights.

Publicly available sequences of PMA-treated samples collected from two JPL clean rooms, ISS dust, ISS HEPA filters, and surface samples from an inflated Lunar Mars analogue habitat (ILMAH) were compared. As is clear from the PCoA plot shown in Additional file 1: Figure S9, the ISS surface microbiome is a unique microbiome, differing from the ISS-dust, ISS-HEPA, JPL clean rooms, and Lunar/Mars-like human-occupied habitats. In addition, PMA-untreated microbial diversity associated with ISS environments was compared to results obtained from the Earth Microbiome Project, hospital environment, and office spaces. The ISS samples grouped with the built environment data, as shown in Fig. 7. This relationship also suggests that, as expected, the environmental locations sampled on the ISS harbored microbes more similar to those on animal surfaces (e.g., skin) than to environmental soil samples. Similar pattern was seen with built environmental samples from Earth. Thus, the ISS samples resembled other Earth built environment samples, and the differences among flights (including differences in DNA extraction and PCR amplification) were relatively small compared to the differences among different sample types.

Using the unrarefied data, the unique sub-operational taxonomic units (sOTUs) present in the ISS data were also compared with other built environment datasets collected on various locations of Earth to determine whether any sOTUs appear to be unique to the ISS. For this analysis, only Flight 3 data were included since all the other built environment datasets used the same primer set. The sOTUs observed in the ISS controls were removed. There were four sOTUs that appear to be unique relative to all the built environment datasets analyzed (see Additional file 4: Table S4), although they accounted for a very small total amount of the sequence mass (~ 0.0005% of the reads). These unique sOTUs exhibited high identity to Bacteroides sp., Gottschalkia acidurici, Paenibacillus thailandensis, and Thermus thermophilus based on BLAST to nr/nt [34]. One single sOTU belonging to T. thermophilus was unique in Flight 3 samples and was not observed in the Earth Microbiome Project, nor in other built environment datasets on Earth.

The ISS environmental microbiome was characterized from eight locations throughout the ISS during three flight sampling events over a period of 14 months. This allowed the examination of temporal and spatial distribution of microbial populations on the ISS. This is the first study to utilize culture, qPCR, and amplicon sequencing to study the surfaces of the US segment and revealed a diverse intact/viable population of bacteria and fungi that changed over time but were similar across locations.

Several studies have been carried out to measure microbiological cleanliness of the ISS environment using cultivation-based approaches since the inception of this closed system [35]. Recently, several ISS surfaces in the US nodes [36] and Japanese Kibo module [37] were swabbed and targeted amplicon sequencing carried out. However, these studies did not measure the intact/viable microbiome which therefore could not be compared to culture counts nor provide an assessment for crew risk. Previous reports on the ISS intact/viable microbiomes on the ISS were only examined from air filters and debris collected via vacuum cleaner bag [2]. The ISS is a unique environment and one question that is of interest to many is how this intact/viable microbiome compares to other closed, regulated environments (Additional file 1: Figure S9 and Fig. 7). The ISS environmental microbiome resembles that of animal skin surfaces rather more than the soil microbiome. This might be due to the fact that cargo sent to the space station was cleaned thoroughly, and hence soil-associated microorganisms were not present.

The predominant organisms on ISS surfaces consisted of those that are associated with humans, with some considered opportunistic pathogens on Earth. As to whether they could cause disease in astronauts on the ISS is unknown, as it would depend on the health status of each individual and how these organisms function while in the space environment. Regardless, the detection of possible disease-causing organisms highlights the importance of further genomic and transcriptomic studies to examine how these ISS microbes function in space and how they may impact astronauts’ health. Correlating viable but opportunistic pathogens with crew member health is likely to raise too many questions about access to the crew microbiome data which is not yet publicly available, and about statistical power: because the ISS has few occupants and high turnover, identifying statistically relevant trends in crew member health that correlate with environmental microbiomes is not possible. From the time the ISS was built in 1998, as of Aug 3, 2017, 222 astronauts visited the ISS, and microbial signatures left behind by previous astronauts might interfere with the predictions. Consequently, the present ISS environmental microbial metrics could not be linked to any particular crew member. Since there were no differences in community composition and richness between PMA- and non-PMA treated samples, it suggests that the DNA analyzed from these possible opportunistic pathogens residing on the ISS environmental surfaces are indeed intact/viable and not dead organisms.

Noteworthy, approximately 46% of intact/viable bacteria and 40% of intact/viable fungi could be cultured with the culture media used during this study. This percentage is high when compared to spacecraft assembly cleanrooms on Earth where only 1 to 10% of intact/viable microorganisms can be cultured [38]. The possible explanation is that the ISS is not deprived of nutrients like spacecraft assembly cleanrooms, and while this is a hermetically sealed environment, it is exposed to microbes from astronauts (maximum six astronauts at a given time) and cargo (delivered ~ 4–6 times per year). Furthermore, no relationship was found between microbial load and sample processing time (F1: 7 days, F2: 9 days, and F3: 6 days).

Many of the organisms detected on the ISS are known to form biofilms that belong to both bacterial (Acinetobacter, Sphingomonas, Bacillus, Burkholderia, Corynebacterium, and Klebsiella) [39] and fungal (Penicillium, Aspergillus, Cryptococcus, and Rhodotorula) [40] genera. This could pose problems for astronauts if they do become infected as biofilms are known to promote resistance to antibiotics [41]. Also, biofilm formation on the ISS could decrease infrastructure stability by causing mechanical blockages, reducing heat transfer efficiency, and inducing microbial influenced corrosion [42]. Some of the microorganisms that were identified on the ISS that have been implicated in microbial-induced corrosion on Earth are Methylobacterium, Sphingomonas, Bacillus, Penicillium, and Aspergillus [43–46]; however, the role they play in corrosion aboard the ISS remains to be determined. Elucidating the potential ability to form biofilms and the magnitude of actual biofilm formation on ISS surfaces is important during long-term space missions to maintain structural stability of the crew vehicle when routine indoor maintenance cannot be as easily performed.

As expected, culture-based analysis did not detect as many genera as that with amplicon sequencing; however, its importance should not be overlooked as species level identity of ISS microbial constituents could be obtained when isolates were available. Furthermore, isolating organisms allowed us to conduct a separate study to examine the influence of microgravity and radiation on antibiotic resistance and virulence [47] and to obtain whole genome sequences of organisms that grow in space, for future comparative genomics [29]. Similar to a previous study on ISS HEPA filters, where the novel organism, Solibacllus kalamii [48]was able to be identified only through culture analysis, this study has also allowed us to detect a recently identified novel species Enterobacter bugandensis that was associated with human disease on Earth [49, 50]. A high percentage of the cultivable population represented opportunistic pathogens such as S. aureus, Staphylococcus hominis, Staphylococcus haemolyticus, P. conspicua, Acinetobacter pittii, Klebsiella quasipneumoniae, and A. fumigatus. This could have potential health impacts on astronauts, as bacteria and fungi have been shown to be transferred between surfaces and humans upon contact [51]. The scope of the present study was not to determine whether these microorganisms were more virulent or resistant to antibiotics than on Earth but the whole genome sequences have been published for the isolated biosafety level 2 microorganisms [29] and comparative genomics of these ISS isolates with Earth strains is now possible for further investigation.

Members of the family Staphylococcaceae and Enterobacteriaceae were the most predominant organisms on ISS surfaces of the US module, similar to what has been published for the Japanese module of the ISS [52], and were detected in almost every sample by both culture and amplicon sequencing (Additional file 1: Figure S10). Both are human-associated organisms, with Staphylococcaceae commonly found on the skin and in the nasal passage, and Enterobacteriaceae commonly associated with the gastrointestinal tract. These two taxa are also abundant in fitness centers [53], office buildings [54], and hospitals [5], suggesting that the ISS is similar to other built environments on Earth, in that its microbiome is shaped by human occupation [32]. On Earth, it has been observed that incoming intensive care unit (ICU) patients have a significantly higher risk of acquiring infections if the previous occupant was a carrier, despite terminal cleaning of the bed and the room [55–57]. Thus, habitation of the same area, regardless of whether individuals interact or not, may contribute to disease spread. Further studies assessing how long organisms survive on ISS surfaces and how readily they can be passed on from one individual to another in space can lead to the development of countermeasures to minimize the spread of infections from one astronaut to another during simultaneous or even separate flight missions.

Methylobacteriaceae/Methylobacterium was also dominant across the ISS and could be cultured from several samples. This is a hardy organism that can withstand harsh conditions, such as ionizing radiation and strong cleaning detergents and has previously been found in NASA spacecraft assembly clean rooms [58], hospital ICUs [59], and the MARS500 habitat [60]. Moraxallaceae, another abundant organism on the ISS, also thrives in harsh conditions, being present in higher relative abundances in spacecraft assembly cleanrooms [61], areas of the home that utilize a lot of chemicals (i.e., washing machine) [62], and deep sea sediment of inactive hydrothermal vents [63].

R. mucilaginosa was the predominant fungal isolate from the culture analysis, and belongs to the Sporidiobolaceae family which was found in high relative abundances across the ISS with amplicon sequencing. This organism can survive inside dishwashers despite high temperatures and chemical exposure [64].

Numerous studies conducted on Earth have shown that the type and amount of human activity in a particular location impacts that indoor microbiome [65–67] and while there were no differences in the average microbial load (by culture and qPCR) nor community structure (amplicon sequencing), it was clear that there were variations between sampling points across different locations. Among the eight locations sampled (Table 1), location #6 (permanent multipurpose module [PMM] port 1) exhibited low concentrations of cultivable (Fig. 2), viable (Fig. 3a), and total (data not shown) microbial burden. The PMM is a specific location within the Node 1 Nadir module of ISS (Fig. 1) to store bags intact as opposed to open and place them into racks. Minimum crew activities are expected in PMM location #6 and hence microbial abundance might be minimal compared to other locations that are heavily occupied by astronauts in a day to day activities such as observing window cupola (location #1), toilet (location #2), ARED exercise platform (location #3), dining table (location #4) performing several experiments, or sleeping quarters (location #8). In a study performed by Mayer et al. [32] in an inflated lunar/Mars analog habitat, the cultivable bacterial load was in the range of 10³–10⁵ per m² after the student crew inhabited the analog station. There were no significant changes in microbial load between the more active areas like the laboratory and other locations, while bedroom cultivable bacterial load increased toward the end of 30-day occupation. The fungal cultivable population was lower than bacterial, but it was in the range from 10² to 10⁴ per m² [68]. In contrast, the ISS results showed that cultivable microbial load were not uniform between locations and warrant more study.

In general, temporal differences were observed within the bacterial population: F2 samples had higher microbial diversity (alpha diversity) than F1 and F3 samples; only Enterobacteriaceae were cultured from F3 samples and nine out of the ten most relatively abundant family level taxa differed over the three flights. These temporal differences may be due to the different occupants onboard the ISS during each of the flight sampling session. Earth indoor microbiome studies have shown that humans shed microbes to indoor surfaces upon contact, playing a pivotal role in shaping the indoor microbiome [17]. Similarly, a study of the inflatable Lunar/Mars analog conducted here on Earth showed differences in bacterial communities between day 0 (before human occupation) and after 30 days of habitation [32] showing the effects of human occupation on indoor microbial communities in a space-like environment. Of the nine astronauts that were present aboard the ISS from F1 to F2 (2 months apart), only three were present during both flights and none of the astronauts present during F1 or F2 were on the ISS during F3 sampling. Further studies that collect microbial information from astronauts in parallel with air and surface samples would help elucidate how much of an impact astronaut have toward the ISS microbiome. Unlike bacteria, fungal communities were stable over time with no temporal differences, and this could be due to the fact that fungal and bacterial communities follow different environmental determinants [69].

It should be noted that F3 samples were sequenced separately from the F1/F2 samples and used different but similar V4 primers (see “Materials and methods” section for more details). Due to the SpaceX-7 launch failure and the uncertainty of when F3 sampling kits would be flown to the ISS for sampling, it was not possible to sequence F3 with F1/F2. However, the samples were collected in the same manner, processed identically, and the same protocol used for DNA extraction. We do not believe that the choice of primers, nor the separate sequencing runs, have influenced the differences in temporal distribution presented in this manuscript for the following reasons: (i) Additional file 5: Table S2 shows the organisms that were statistically significantly different over time and shows the efficiency of each primer pair in detecting these organisms, which are almost identical. (ii) A metagenomics analysis was performed using the same DNA samples as for amplicon sequencing and all three flights were sequenced simultaneously and without multiple displacement amplification prior to sequencing. The family level barplot for this metagenomics data in Additional file 1: Figure S11 shows the same pattern distribution of taxa, as presented in Additional file 1: Figure S3 for the amplicon sequencing. (iii) Lastly, all statistical analyses were performed with ALDEx2 which estimates per-feature technical variation within each sample using Monte-Carlo instances drawn from Dirichlet distributions. ALDEx2 uses the centered log-ratio transformation that ensures that data are scale invariant and sub-compositionally coherent meaning that all samples are numerical consistent with each other, regardless of the total sequencing read capacity at the time of sequencing [70, 71]. This ensures that the statistical results are robust and are not influenced solely by the differential detection of ASVs that can occur during different sequencing runs.

Many 16S rRNA and ITS sequences could not be identified via amplicon-targeted analyses, but a metagenomics approach recently conducted identified 318 bacterial and fungal species in these samples [72]. In addition, shotgun metagenome analysis carried out by Singh et al. [72] from the same samples revealed that reads associated with carbohydrate metabolism, amino acid derivatives and cofactors, vitamins, etc. were the highest among all three flights. Similarly, computational analyses showed that the Legionella resistome, cobalt-zinc-cadmium resistance, and multi-drug resistant resistance efflux pump were high on all flights and all locations. The shot-gun reads associated with antimicrobial resistant genes in Flight 3 increased by twofold when compared with Flights 1 and 2 which also predicted the persistence of opportunistic pathogens in Flight 3 samples [72]. Collective beta-Lactam resistance derived from the metagenome sequence analysis shows that physical (OmpF, OmpC), transformational (penicillin-binding protein), and degradational (AmpC), and MDR efflux pump (OMP, RND, MPF) mechanisms were allocated by the microorganisms on the ISS [72].

Exploring the spatial and temporal distribution of intact/viable microbial populations of closed systems such as the ISS will facilitate planning of future human habitation of Moon, Mars, and beyond. Accumulation of intact/viable microbial cells in a confined environment poses a health risk to all inhabitants. This study on bacterial and fungal load and diversity across the ISS provides a comprehensive catalog of what can be found in closed space systems and can be used to develop safety measures for NASA to meet the requirements for long-term space travel or living in space. The implications of this study are not only limited to space biology but can have significant impact on cleanrooms here on Earth such as those in the pharmaceutical and medical industries.