Skip to main content
Regenerative Therapy logoLink to Regenerative Therapy
. 2025 Nov 1;30:991–998. doi: 10.1016/j.reth.2025.10.019

Microbiome-based profiles of airborne bacteria to support microbial risk assessment in cleanroom environments

Mitsuru Mizuno a,⁎, Yusuke Ogata b, Yuto Nishihara c, Miwako Nishio c, Hisako Katano a, Ichiro Sekiya a
PMCID: PMC12616117  PMID: 41244152

Abstract

Introduction

Maintaining aseptic conditions is essential for cell product processing, as sterilization cannot be applied to living cells. Conventional environmental monitoring relies on particle counts and culture-based colony-forming unit measurements. These indicators fail to capture much of the diversity and provenance of airborne microbes because many taxa are nonculturable or require growth conditions not supported by standard culture media. Therefore, comprehensive DNA-based microbiome analysis is critical for evaluating microbial risks that conventional methods may overlook; however, such studies remain limited in cleanroom settings. This study aimed to comprehensively visualize the structure of airborne microbial communities in cleanroom environments and clarify microbial risks that cannot be fully captured by particle counts or culture-based methods.

Methods

We collected airborne bacterial DNA from cleanrooms with environmental Grades B, C, and D using a high-volume air sampler. The DNA was extracted and analyzed via 16S rRNA gene amplicon sequencing targeting the V3–V4 regions. Bioinformatic analysis was performed using the QIIME2 pipeline, and microbial diversity was assessed using alpha and beta diversity indices. Abundant taxa were categorized based on their likely origin (environment- or skin-derived), and their distributions were examined in relation to facility management practices.

Results

Analysis revealed the consistent detection of skin-associated bacteria, such as Cutibacterium and Corynebacterium, and environmental bacteria, including Bacillus and Paracoccus, across all cleanroom grades. Alpha- and beta-diversities exhibited no significant differences among the grades. However, temporary and irregular increases in skin-derived bacteria indicated operator-related non-persistent contamination. This interpretation was supported by skewness and kurtosis analyses, which indicated occasional but noticeable shifts in microbial abundance, particularly in high-grade cleanroom environments.

Conclusions

This study demonstrates the limitations of conventional culture-based monitoring and underscores the value of DNA-based approaches for characterizing airborne microbial communities in cleanrooms. The detection of temporary increases in skin-associated bacteria indicates that operator-related contamination can occur even under stringent environmental conditions. These findings support the development of integrated monitoring strategies that can capture both the composition and temporal fluctuations of airborne microbiota to enhance microbial risk assessment.

Keywords: Airborne microbiome, Cleanroom monitoring, 16S rRNA genome sequencing, Skin-derived bacteria, Cell products

Highlights

  • •

    Airborne bacterial communities in cleanrooms profiled by 16S rRNA gene sequencing.

  • •

    Standard culture-based media missed some salt-tolerant or non-culturable microbes.

  • •

    Skin- and environment-derived taxa coexisted in all cleanroom grades.

  • •

    Alpha And beta diversities show no significant grade effect on airborne communities.

  • •

    Skewness and kurtosis helped detect irregular microbial events in clean environment.

1. Introduction

Cell product processing requires the management of environmental microorganisms to ensure product sterility [[1], [2], [3], [4]]. This is particularly critical in processes involving living cells, as product sterilization is not feasible, thereby necessitating the minimization of microbial contamination risks within the manufacturing environment. Consequently, operations must be conducted in clean environments, such as high-efficiency particulate air–filtered cleanrooms and biosafety cabinets [[5], [6], [7], [8], [9]]. Cleanroom environmental monitoring has conventionally relied on particle counts and culture-based enumeration of colony-forming units (CFUs) [[10], [11], [12]]. However, these approaches have a limited capacity to comprehensively capture the overall profile, origins, and dissemination pathways of airborne microorganisms.

Conventional culture-based methods detect only a small subset of culturable bacterial species that can grow on standard media, such as soybean casein digest (SCD) agar, which is commonly used for environmental monitoring. Consequently, these methods may fail to reflect the true diversity of microbial communities. Cell culture media used in regenerative medicine, for example, are typically enriched with nutrients such as fetal bovine serum (FBS) and various growth factors to support cell maintenance and proliferation [13,14]. These components may also support the growth of microbial taxa overlooked by standard monitoring. In recent years, attention has been paid to the relationship between airborne particle counts and the presence of microbes, although this correlation has yet to be quantitatively elucidated [15,16]. Furthermore, even within the same cleanroom, microorganisms can originate from diverse sources, including operators, building materials, and equipment surfaces, rendering it challenging to distinguish their origins using conventional monitoring methods. Therefore, a comprehensive analysis of microbial communities in indoor air and the development of quantitative evaluation methods are necessary, regardless of the cleanroom grade or cleanliness level. Although such techniques have been explored, the available information remains limited [16,17].

This study aimed to comprehensively visualize the structure of airborne microbial communities in cleanroom environments and clarify microbial risks not fully captured by particle counts or culture-based methods. Specifically, we subjected microbial DNA recovered from air samples of different environmental Grades (B, C, and D) to 16S rRNA gene-based metagenomic analysis to quantitatively assess community composition and diversity. We further examined the frequencies and presumed origins of the detected taxa to characterize the spatial patterns of microbial distribution. This study provides a scientific foundation for understanding microbial risks in clean environments and offers baseline data for discussing the limitations of current environmental management practices.

2. Materials and methods

2.1. Airborne bacteria sampling in general laboratory environment (GE)

Air was sampled in a GE using an AirPrep ACD220 electret filter air sampler (Innovaprep Inc., MO, USA) operated at an inflow rate of 200 L/min for 120 min. Microorganisms collected on the filter were eluted using an Innovaprep PBS Elution Kit following the manufacturer's instructions. The recovered bacteria were double-stained with acridine orange for live cells and propidium iodide for dead/apoptotic cells using a live/dead assay kit (Logos Biosystems, South Korea) and then observed under a fluorescence microscope.

2.2. Recovery of airborne bacteria on agar plates in GE

Four types of agar media were prepared: SCD (Becton Dickinson and Company (BD), NJ, USA), sheep blood (Shimadzu Diagnostics, Tokyo, Japan), marine (BD), and Dulbecco's Modified Eagle Medium (DMEM)-based. The marine agar medium was prepared by dissolving 55.1 g in 1 L of distilled water, autoclaving at 121 °C for 15 min, and dispensing 10 mL into Petri dishes for gelation at room temperature. The DMEM agar was prepared by dissolving double-strength DMEM (Shimadzu Diagnostics) in distilled water, supplemented with l-glutamine and 20 % FBS (Thermo Fisher Scientific, MA, USA), and then filter-sterilized. Double-strength agar (Fujifilm Wako, Osaka, Japan) was autoclaved separately at 121 °C for 15 min and cooled to approximately 60 °C. Equal volumes of DMEM and agar solutions were mixed to yield a final concentration of 1 × , supplemented with 10 % FBS, poured into dishes, and allowed to gel at room temperature.

A Biosamp air sampler (Midori Anzen Co., Ltd., Tokyo, Japan) was used to collect 1000 L of air from the GE for each agar type. Four plates (one for each agar type) constituted one set, and sampling was repeated on multiple days. The plates were incubated aerobically at 32.5 °C. After three days, the colonies formed were visually inspected and counted as CFUs. Representative macroscopic images of the colonies were captured using a digital camera (EOS 60D; Canon, Tokyo, Japan).

2.3. Airborne bacterial DNA sampling in cleanrooms

To evaluate airborne microbiota in cleanrooms, air samples were collected from the cell-processing facility of the Institute of Science Tokyo Hospital (established in 2015) using an AirPrep ACD220 electret filter air sampler. Samples were collected from Grade D (n = 10), Grade C (n = 15), and Grade B (n = 11) environments, following the European Union Good Manufacturing Practice classification, and from a GE (positive control, n = 1). In addition, DNA was extracted from six samples that served as negative controls (NC, n = 5), without air sampling. The sampling locations are shown in Fig. 1a. Personnel wore clean laboratory coats or equivalent attire in Grade D areas. In Grade C and B areas, personnel wore disposable low-lint nonwoven cleanroom clothing (IsoClean; DuPont de Nemours, Wilmington, DE, USA). Goggles in any grade were not worn.

Fig. 1.

Fig. 1

Cleanroom air sampling scheme and representative detection of airborne bacteria. (a) Sampling points and cleanrooms of Grades B, C, and D. Parenthetical symbols indicate pressure relative to the normal room pressure. (b) Airborne bacteria in a general laboratory environment. (c) Numbers of colony-forming units (CFUs) from a general environment cultured on various agar media, including soybean casein digest (SCD), sheep blood, marine, and Dulbecco's Modified Eagle Medium (DMEM) containing 10 % fetal bovine serum (FBS). Data are presented as the median with the interquartile range.

The sampler was operated continuously for 2 h at the maximum flow rate (200 L/min), resulting in a total air volume of 24 000 L per sample and positioned at the center of the room. This volume was determined based on the equipment's maximum intake capacity and preliminary testing, which showed that shorter durations resulted in insufficient sequencing depth for diversity analysis. Between samples, the sampler was cleaned with 70 % ethanol and UV-irradiated for 15 min, with its exterior wiped with 0.1 % sodium hypochlorite to avoid cross-contamination. Bacteria trapped on the filter were recovered using an Innovaprep Tris Elution Kit, followed by DNA extraction.

2.4. DNA extraction for 16S rRNA metagenomic analysis

Eluted airborne microorganisms (∼8 mL) were treated with MycoCollect reagent (Shimadzu Diagnostics) following the manufacturer's instructions, gently vortexed, and centrifuged (12 000 g, 10 min, 4 °C). The pellet was resuspended in 600-μL Tris-EDTA (TE) buffer (Nippon Gene, Tokyo, Japan) containing 7.5-mg lysozyme (Sigma-Aldrich, Darmstadt, Germany) and incubated at 37 °C for 30 min with gentle shaking. Subsequently, 70 μL of 10 % sodium dodecyl sulfate solution (final ∼1 %) and 7 μL of recombinant proteinase K (Fujifilm Wako) were added, and the samples were incubated at 55 °C for 30 min. Equal volumes (0.7 mL) of phenol/chloroform/isoamyl alcohol (Sigma-Aldrich) were added, mixed for 30 min at room temperature, and centrifuged (12 000 g for 10 min). The aqueous phase was transferred to a new tube, supplemented with 80 μL of 3 M sodium acetate (final ∼0.3 M), mixed gently, and precipitated with an ethanol precipitation aid (Nippon Gene) and 0.7 mL isopropanol (Fujifilm Wako) on ice for more than 10 min. The pellet obtained after centrifugation (12 000 g, 10 min, 4 °C) was washed with 1 mL of 75 % ethanol, centrifuged again, vacuum-dried, resuspended in 20-μL TE buffer, and stored at 4 °C until analysis.

2.5. Preparation, sequencing, and analysis of 16S rRNA gene amplicon library

The V3–V4 hypervariable regions of the 16S rRNA gene were amplified using Illumina overhang adapter primers:

341F (5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG) and.

805R (5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC).

First-round polymerase chain reaction (PCR) was performed in 25-μL reactions containing KAPA HiFi HotStart ReadyMix (Kapa Biosystems, Roche, Switzerland) for 35 cycles (95 °C, 3 min; 35 cycles of 95 °C, 30 s; 55 °C, 30 s; 72 °C, 30 s; final extension 72 °C, 5 min). Amplicons were purified using AMPure XP beads (Beckman Coulter, Brea, CA, USA) and quantified using a Qubit fluorometer (Thermo Fisher Scientific). Index PCR was performed using the Nextera XT Index Kit (Illumina, CA, USA), followed by purification. Library size (∼600 bp) was confirmed using an Agilent Bioanalyzer, and concentrations were normalized for pooling. The pooled library was spiked with 40 % PhiX control (Illumina) and sequenced on the Illumina NextSeq platform (2 × 300 bp paired-end reads) using the NextSeq Reagent Kit v3, following the manufacturer's protocols.

2.6. Bioinformatic analysis

Raw reads were processed using the QIIME2 software (version 2024.2). Demultiplexed paired-end reads were trimmed and denoised using DADA2 to eliminate chimeras and correct sequencing errors. Amplicon sequence variants (ASVs) were inferred without clustering, and taxonomic assignment was performed using a pre-trained Naïve Bayes classifier against the SILVA reference database at 99 % similarity. ASV abundance tables were constructed for downstream analysis, and samples with read depths below the minimum threshold were excluded to ensure data quality.

Each dataset was evaluated after removing Escherichia coli reads, which were considered to be reagent-derived. These reads were excluded based on their consistent presence in NC without air sampling across all sequencing runs, in which no environmental DNA was expected. Therefore, we interpreted their occurrence as reagent-derived contamination rather than environmental signals. Furthermore, E. coli was not detected on SCD agar in our facility's long-term environmental monitoring. Based on these findings, we excluded E. coli from the downstream microbiome analysis to avoid overestimating its environmental relevance.

The data obtained before removal are shown in Fig. S1. Subsequently, taxonomic groups with at least 5 % of the total reads in each sample were extracted and visualized at the genus or family level. Taxonomic groups were color-coded according to species, and samples were sorted by grade (Fig. 2). As a classification rule, species with a read count proportion of 5 % or more were extracted, while those with less than 5 % were defined as “others.” Similarly, species with a read count proportion of 5 % or more but less than 10 % and appearing only once were defined as “others.” When data for a higher taxonomic group were available, they were combined with those of another higher taxonomic group. For example, if Halomonadaceae and Halomonas are present, they are classified as Halomonadaceae. Closely related species that could not be distinguished using the 16S sequences used in this study were represented by abbreviated names, with the names of closely related species listed in the legends. Specifically, Allorhizobium–Neorhizobium–Pararhizobium–Rhizobium is denoted as ANPR.

Fig. 2.

Fig. 2

Microbiome in cleanrooms of Grades B, C, and D. The figure illustrates the composition of the airborne bacterial communities in Grades B, C, and D. The category “others” includes taxa representing <5 % of the total reads per sample, or taxa with a relative abundance of 5–10 % detected at only one sampling location. Taxa without species-level identification or classified only to the family level are indicated at the family level. Closely related species indistinguishable by 16S rRNA sequences are grouped, with ANPR referring to Allorhizobium–Neorhizobium–Pararhizobium–Rhizobium. Sequences corresponding to Escherichia coli, presumed to be reagent-derived, were excluded from the main analysis.

2.7. Alpha diversity analysis

Alpha diversity metrics—observed OTU richness, Shannon index, and Faith phylogenetic diversity (PD)—were calculated using QIIME2 for cleanroom samples (Grades B, C, and D) and NC. Positive controls (GE) were used for calculations and excluded from the graph because n = 1. Distributions were visualized using min-to-max plots in Prism 10 (GraphPad Software, LLC, La Jolla, CA, USA). Group differences were tested using the Kruskal–Wallis test, with p < 0.05 considered significant.

2.8. Beta diversity analysis

Beta diversity was assessed using unweighted and weighted UniFrac distances calculated using QIIME2. Principal coordinate analysis (PCoA) was used for visualization [18]. Statistical differences between the cleanroom grades (Grades B, C, and D) were evaluated using PERMANOVA (adonis2 function, vegan package, R v4.4.1) with 999 permutations, excluding the negative and positive controls. Pairwise comparisons between grades were performed using the pairwise.perm.manova function (RVAideMemoire package) with false discovery rate correction.

2.9. Comparative analysis of top five taxa per grade

The five most frequently occurring genera in each grade were extracted and compared. Ranking was based on occurrence counts per grade. Genera were categorized into environmental bacteria and skin commensals based on the literature [19,20], and their relative abundances were visualized as heat maps and bar charts.

2.10. Statistical analysis

All statistical analyses were performed using Prism 10 and R software. Statistical tests and presentation formats are described in the corresponding figure legends. Statistical significance was set at p < 0.05.

3. Results

3.1. Cleanroom air sampling scheme and representative detection of airborne bacteria

Sampling points were distributed throughout the cleanroom facility to capture airborne microorganisms across the different environmental grades (Fig. 1a). Two sampling sites were located in Grade B areas, including the cell processing room and second gowning area. Three sampling sites were positioned in Grade C zones, namely, the supply room, first gowning area, and corridor. One sampling site, representing a Grade D environment, was established at the entrance. The personnel flow, indicated by arrows in Fig. 1a, shows the transition pathway from the entrance through Grade D and C areas into the Grade B clean zones. This arrangement enabled a systematic evaluation of airborne bacterial communities across distinct cleanliness levels, from the highly controlled cell processing environment (Grade B) to the less stringent entrance area (Grade D).

Airborne bacteria captured on AirPrep filters were visualized using live/dead staining and imaged via fluorescence microscopy. Bright-field, green (viable), and red (non-viable) channels were merged, confirming the coexistence of live and dead cells in the sampled air (Fig. 1b).

Culturable bacteria were assessed on four media: SCD, sheep blood, marine, and DMEM. Representative colonies and CFU bar plots were summarized for comparison (Fig. 1c). CFU yields and colony types differed among media, indicating that culture recovery depends on the nutrient formulation.

3.2. Microbiome in cleanrooms of grades B, C, and D

The taxonomic composition of airborne microbiota was profiled across cleanroom Grades B, C, and D (Fig. 2a–c). In these analyses, reads assigned to E. coli, presumed to originate from DNA extraction or PCR reagents, were excluded (Fig. 2 and S1). Although E. coli can be readily cultured and may appear on environmental agar plates, it has not been identified during facility environmental monitoring to date and was therefore omitted from the primary analysis. The results that include E. coli are presented in Fig. S1.

Taxa with a relative abundance of at least 5 % in each sample were visualized at the family or genus level, with color coding as described in Section 2. The detected taxa included Staphylococcus and Koculia, which are human commensal or opportunistic pathogens that likely originate from the skin or mucosal surfaces. Environmental bacteria such as Sphingomonas and Methylobacterium, which are often associated with water systems, air-handling filters, and building materials, exhibited a high relative abundance in some samples.

The presence of these groups indicates that microbial ingress and colonization are not completely eliminated, even in cleanrooms. Understanding these spatial distributions provides important insights into airborne dissemination routes and potential sources of colonization. Fig. 2 provides a visual overview of the dominant taxa across samples and grades, enabling an intuitive comparison of microbiome patterns in controlled environments.

3.3. Alpha diversity analysis

Alpha diversity in Grades B, C, D, and NC samples was compared using the observed OTU richness (Fig. 3a), Shannon index (Fig. 3b), and Faith PD (Fig. 3c).

Fig. 3.

Fig. 3

Alpha and beta diversity of airborne microbial communities in a cleanroom. Intergroup comparisons using (a) observed operational taxonomic unit (OTU) richness, (b) Shannon index, and (c) Faith phylogenetic diversity (Faith PD), where the p-value was calculated using the Kruskal–Wallis test. Principal coordinate analysis (PCoA) based on the (d) weighted UniFrac distance and (e) unweighted UniFrac distance, where the p-value was calculated using the permutational multivariate analysis of variance (PERMANOVA) test.

The OTU richness values in Grades B, C, D, and NC samples were in the ranges of 40–110, 40–309, 29–218, and 26–123, respectively, showing considerable overlap among grades (Fig. 3a). The Shannon index values also overlapped: 3.05–5.54 (B), 3.43–6.76 (C), 4.19–5.43 (D), and 3.12–4.80 (NC) (Fig. 3b). The Faith PD values were in the ranges 3.69–10.32 (B), 3.51–12.42 (C), 3.32–19.10 (D), and 4.46–12.49 (NC) (Fig. 3c). All comparisons yielded p-values >0.05.

These results indicate that, in terms of alpha diversity, the environmental grade does not significantly influence the diversity of airborne bacterial communities.

3.4. Beta diversity analysis

Beta diversity was evaluated using weighted and unweighted UniFrac distances and visualized using PCoA. PERMANOVA was applied to Grades B, C, and D samples (excluding the negative and positive controls).

Weighted UniFrac analysis showed that 13.1 % of the variance was explained by the environmental grade (R2 = 0.131, p = 0.060), indicating no statistically significant difference but a slight trend in compositional variation by grade (Fig. 3d). In particular, Grade B samples tended to cluster closer to NC samples, although the differences were not significant. Unweighted UniFrac analysis revealed no significant differences (R2 = 0.030, p = 0.725), indicating that the presence/absence-based compositions were similar across the grades (Fig. 3e).

Sample similarity appeared to be highly influenced by individual sampling events, rather than by grade, without noticeable clustering over time or experimental background. These findings show that transient operational factors, such as personnel movement, may contribute to the observed community structures. Overall, the beta diversity results indicate that cleanroom grade has a limited impact on the airborne microbiome structure.

3.5. Comparison of top five genera per location and evaluation of environmental versus skin-derived taxa

Across the grades, the five most frequent genera were identified, revealing 12 genera commonly found in cleanrooms. Based on the literature, these bacteria were categorized as environmental bacteria: Alcanivorax, ANPR, Enhydrobacter, Halomonadaceae, Paracoccus, Pseudomonas, Roseomonas, Sphingomonas, and skin commensals (Corynebacterium, Cutibacterium, Kocuria, and Streptococcus).

The relative abundances of these taxa were visualized in a heatmap (Fig. 4a), and the mean values per grade were plotted (Fig. 4b). Enhydrobacter, including Moraxella, was the most frequently detected taxon in all environments.

Fig. 4.

Fig. 4

Relative abundance profiles of airborne bacterial genera across cleanroom environments. (a) Heatmap of the top five most abundant bacterial genera in each environmental grade (B, C, and D), with higher relative abundance shown in red. Each square represents the mean relative abundance for a genus in a sample. Asterisks (∗) indicate skin commensals (Cutibacterium, Corynebacterium, Kocuria, and Streptococcus). Taxa indistinguishable by 16S rRNA sequencing are abbreviated as ANPR. (b) Bar graph showing the mean relative abundance of each extracted genus across grades. Comparisons of (c) skewness and (d) kurtosis values between environmental and skin commensals in each grade. The p-values were calculated using the Mann–Whitney test, with p < 0.05 considered significant (“ns” indicates not significant). Note that statistical analysis was performed within each grade only, and no intergrade comparisons were conducted.

Statistical analysis revealed that, in Grades B and C, skin commensals exhibited significantly higher skewness and kurtosis than environmental bacteria, indicating that skin-derived taxa are more likely to be accidently detected as outliers. In contrast, environmental bacteria were more consistently detected. In Grade D, no significant difference was observed, implying that in highly clean environments, the relative impact of skin commensals may increase episodically.

4. Discussion

This study aimed to comprehensively visualize airborne bacterial communities in cleanroom environments by applying 16S rRNA gene-based metagenomic analysis to airborne nucleic acids collected from different environmental grades. This approach revealed microbial diversity and community structures that could not be fully captured by conventional particle counts or culture-based methods, thereby providing new insights into microbial risk assessments in controlled environments. By characterizing community structures and performing quantitative evaluations, we enhanced the understanding of cleanliness management and operator-associated contamination risk. These findings provide a basis for the development of more robust monitoring strategies.

Live/dead staining distinguished viable from non-viable cells, while culture assays across multiple agar media quantified the culturable fraction of airborne bacteria. Fluorescence microscopy enabled a direct confirmation of airborne bacteria, demonstrating that a portion of the airborne microorganisms was non-viable. This highlights the need for risk assessments that consider microbial viability, rather than presence alone. Although direct microscopic visualization of airborne bacteria is rare [21], such images provide ground-truth evidence of airborne cells—including non-viable fractions—and aid in the interpretation of culture and 16S results. Culture-based quantification revealed marked differences in CFU counts depending on the agar medium, underscoring the medium dependence of environmental monitoring. Specialized media, such as sheep blood agar and marine agar, supported the preferential growth of specific taxa that were not recovered on standard SCD agar. For instance, sheep blood agar facilitates the recovery of nutritionally fastidious environmental bacteria, such as Paracoccus and Roseomonas [[22], [23], [24]], both of which were frequently detected in this study. These taxa may have been underrepresented in standard monitoring methods. Similarly, marine agar, with its high salt concentration, enables the growth of halotolerant taxa such as Alcanivorax and Halomonadaceae [25,26], which are difficult to recover on SCD agar. Notably, such bacteria may survive in the nutrient-rich cell culture media (e.g., DMEM with 10 % FBS) used in manufacturing processes, posing a potential risk of undetected persistence. However, it remains unclear whether these bacteria directly contribute to contamination or compromise product safety. Our findings highlight their detectability outside standard protocols, but further investigation is required to determine their relevance to manufacturing risks.

The 16S rRNA metagenomic analysis characterized the taxonomic composition of airborne microorganisms and their relative abundances across spatially distinct cleanroom grades. Reagent-derived nucleic acids were excluded from the analyses to ensure taxonomic reliability. Results revealed the repeated detection of human skin commensals such as Cutibacterium and Corynebacterium, indicating the introduction of microorganisms from the operators. Environmental taxa, including soil-derived Bacillus and water-system-associated Methylobacterium and Paracoccus, were also frequently observed. This demonstrates that microbial ingress in cleanrooms originates from multiple niches. A notable finding was the detection of many genera common to multiple environmental grades, indicating that spatial zoning may not guarantee a distinct separation of microbial communities. Similar genera have been reported in culture-based monitoring studies [12,27], highlighting the overlap between culture-dependent and molecular approaches. Intragrade variability in taxa and relative abundances further indicates that transient and dynamic factors, such as localized airflow patterns, work tasks, and equipment operation, can influence microbial communities.

Alpha diversity analysis using multiple indices (observed OTU richness, Shannon index, and Faith PD) revealed no significant stratification among the cleanroom grades, indicating that the overall microbial diversity is not substantially influenced by environmental grade [28]. Beta diversity analysis using both unweighted and weighted UniFrac distances showed that the airborne microbial communities did not form distinct clusters based on grade. Although the weighted UniFrac exhibited a weak compositional trend (R2 = 0.131, p = 0.060), the differences were not statistically significant [29]. In some cases, Grade B communities resembled those of NCs, possibly reflecting extremely low microbial loads. However, the within-grade variability was substantial, and some samples diverged markedly from others. These findings show that airborne microbial communities are shaped predominantly by transient, stochastic processes rather than by physical environmental segmentation, rendering stable prediction and control challenging. Although human activities, such as environmental monitoring, may be the primary cause of variability, a broader array of background factors, beyond those addressed by current cleanliness standards, may also be necessary. Accordingly, developing a revised microbial risk-assessment framework that incorporates these diverse and underrecognized influences is essential for improving the reliability of microbial risk prediction and control strategies.

Genus-level profiling identified the five most frequent genera in each grade and categorized them as environmental bacteria or skin commensals. Many of these genera were detected across multiple grades, indicating that some taxa were widespread regardless of spatial segregation. Environmental taxa such as Enhydrobacter, Paracoccus, and Sphingomonas were relatively stable across the grades, whereas skin commensals such as Cutibacterium, Corynebacterium, and Streptococcus exhibited higher variability, particularly in Grades B and C. Statistical analysis revealed that skin commensals exhibited significantly greater skewness and kurtosis in Grades B and C, signifying that their detection was primarily incidental and associated with transient contamination events, such as operator movement. In Grade D environments, the distribution patterns of environmental and skin commensals were similar, indicating that in highly clean environments, the relative influence of skin-derived contamination events may increase. This implies that microbial risk assessments should consider both the origin and dispersal characteristics of taxa. Furthermore, metrics such as skewness and kurtosis could serve as supplemental indicators for evaluating facility operational status and contamination risk.

This study has four main limitations. First, we used amplicon sequencing targeting the V3–V4 region of the 16S rRNA gene, which limits the taxonomic resolution of closely related species and strain-level identification [30]. Moreover, because DNA-based detection includes non-viable cells, it was not possible to confirm the absence of reagent-derived E. coli nucleic acids or to distinguish viable pathogens from their non-viable counterparts [[31], [32], [33]]. Second, sampling was limited in time, frequency, and spatial coverage, rendering it insufficient to fully capture the associations between microbial communities and operational factors, such as daily variation, work content, and HVAC operation. A direct demonstration of causal links with human activity would require extended, multisite longitudinal sampling integrated with worklogs and particle dynamics. Third, the classification of taxa as environmental or skin commensals was based on literature-derived functional attributes, limiting the definitive assignment of the origins of all detected taxa. For example, although Pseudomonas was classified as environmental, it is also abundant in human-associated environments [34,35]. Finally, the present statistical analyses focused on diversity metrics, whereas the actual risk is determined by microbial functional traits such as pathogenicity, antimicrobial resistance, and biofilm formation. Future studies should incorporate functional metagenomics or transcriptomics to assess risk-relevant attributes.

Molecular microbiome profiling in controlled cleanrooms uncovered airborne risks not detectable by conventional culture-based surveillance. Although no noticeable differences in diversity were observed among the environmental grades, both skin commensals and environmental taxa were found to be universally airborne across spaces, with their distribution influenced by stochastic and localized events. These findings demonstrate the utility of molecular microbiological methods as a complementary tool for cleanliness evaluation, providing a framework for improving the accuracy and situational relevance of airborne microbial monitoring in controlled environments.

Authors' contributions

MM: Sample acquisition, data analysis, interpretation, statistical analysis, and manuscript drafting. YO: Sequencing experiments and bioinformatic analysis. YN: Organizing and arranging data. MN Providing critical advice. MM and IS: Manuscript revision for important intellectual content. All the authors have read and approved the final manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We thank Saeri Kimbara and Tomoko Oono for their assistance with experimental sampling, and Chiaki Okumura and Asuka Asami for their laboratory management. This study was supported by Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (Grant Number JP 24K13435 awarded to Mitsuru Mizuno.

Footnotes

Peer review under responsibility of the Japanese Society for Regenerative Medicine.

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.reth.2025.10.019.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Multimedia component 1
mmc1.pdf (88.5KB, pdf)
Multimedia component 2
mmc2.docx (20.8KB, docx)

Data availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

References

  • 1.Mizuno M., Suda D., Matsumura C., Sekiya I. Changeover method for biosafety cabinets using ozone gas. PLoS One. 2025;20 doi: 10.1371/journal.pone.0318006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Mizuno M., Abe K., Kakimoto T., Hasebe H., Kagi N., Sekiya I. Operator-derived particles and falling bacteria in biosafety cabinets. Regen Ther. 2024;25:264–272. doi: 10.1016/j.reth.2024.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Mizuno M., Kimbara S., Ichise H., Ishikawa N., Nishihara Y., Nishio M., et al. Cleaning methods for biosafety cabinet to eliminate residual mycoplasmas, viruses, and endotoxins after changeover. Regen Ther. 2025;28:73–80. doi: 10.1016/j.reth.2024.11.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Sugahara Y., Mizuno M., Miyashita N., Katano H., Sekiya I. Operator stress factors and cell contamination risks in cell processing facilities: an online survey-based analysis. Regen Ther. 2025;29:397–403. doi: 10.1016/j.reth.2025.03.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Mizuno M., Abe K., Kakimoto T., Hasebe H., Sekiya I. Decreasing electricity costs of clean room for cell products during non-operation. Regen Ther. 2024;26:1117–1123. doi: 10.1016/j.reth.2024.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Mizuno M., Abe K., Kakimoto T., Yano K., Ota Y., Tomita K., et al. Volatile organic compounds and ionic substances contamination in cell processing facilities during rest period; a preliminary assessment of exposure to cell processing operators. Regen Ther. 2023;24:211–218. doi: 10.1016/j.reth.2023.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Mizuno M., Matsuda J., Watanabe K., Shimizu N., Sekiya I. Effect of disinfectants and manual wiping for processing the cell product changeover in a biosafety cabinet. Regen Ther. 2023;22:169–175. doi: 10.1016/j.reth.2023.01.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Mizuno M., Sugahara Y., Iwayama D., Miyashita N., Katano H., Sekiya I. Stress and motivation of cell processing operators: a pilot study of an online questionnaire survey. Regen Ther. 2022;21:547–552. doi: 10.1016/j.reth.2022.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Mizuno M., Yori K., Takeuchi T., Yamamoto T., Ishikawa N., Kobayashi M., et al. Evaluation of cleaning methods for change-over after the processing of cell products to avoid cross-contamination risk. Regen Ther. 2024;26:489–495. doi: 10.1016/j.reth.2024.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Mizuno M., Endo K., Katano H., Tsuji A., Kojima N., Watanabe K., et al. The environmental risk assessment of cell-processing facilities for cell therapy in a Japanese academic institution. PLoS One. 2020;15 doi: 10.1371/journal.pone.0236600. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Mizutani M., Samejima H., Terunuma H., Kino-Oka M. Experience of contamination during autologous cell manufacturing in cell processing facility under the Japanese medical practitioners act and the Medical Care Act. Regen Ther. 2016;5:25–30. doi: 10.1016/j.reth.2016.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Morandi F., Della Lastra M., Bandettini R., Tripodi G., Zara F., Airoldi I. Microbes identified from monitoring cell manipulations in 5-year life of the Cell Factory G. Gaslini. Regen Ther. 2024;27:234–243. doi: 10.1016/j.reth.2024.03.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Mizuno M., Maeda Y., Sanami S., Matsuzaki T., Yoshikawa H.Y., Ozeki N., et al. Noninvasive total counting of cultured cells using a home-use scanner with a pattern sheet. iScience. 2024;27 doi: 10.1016/j.isci.2024.109170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Mizuno M., Matsuzaki T., Ozeki N., Katano H., Koga H., Takebe T., et al. Cell membrane fluidity and ROS resistance define DMSO tolerance of cryopreserved synovial MSCs and HUVECs. Stem Cell Res Ther. 2022;13:177. doi: 10.1186/s13287-022-02850-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Raval J.S., Koch E., Donnenberg A.D. Real-time monitoring of non-viable airborne particles correlates with airborne colonies and represents an acceptable surrogate for daily assessment of cell-processing cleanroom performance. Cytotherapy. 2012;14:1144–1150. doi: 10.3109/14653249.2012.698728. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Kawai M., Ichijo T., Takahashi Y., Noguchi M., Katayama H., Cho O., et al. Culture independent approach reveals domination of human-oriented microbes in a pharmaceutical manufacturing facility. Eur J Pharmaceut Sci. 2019;137 doi: 10.1016/j.ejps.2019.104973. [DOI] [PubMed] [Google Scholar]
  • 17.Lian J., Ma X., Li X., Xia L. The environmental microbial retrieving assessment of cell-processing facilities for cell therapy in a hospital laboratory. Microbiol Spectr. 2024;12 doi: 10.1128/spectrum.01257-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Bolyen E., Rideout J.R., Dillon M.R., Bokulich N.A., Abnet C.C., Al-Ghalith G.A., et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat Biotechnol. 2019;37:852–857. doi: 10.1038/s41587-019-0209-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Byrd A.L., Belkaid Y., Segre J.A. The human skin microbiome. Nat Rev Microbiol. 2018;16:143–155. doi: 10.1038/nrmicro.2017.157. [DOI] [PubMed] [Google Scholar]
  • 20.Gilbert J.A., Hartmann E.M. The indoors microbiome and human health. Nat Rev Microbiol. 2024;22:742–755. doi: 10.1038/s41579-024-01077-3. [DOI] [PubMed] [Google Scholar]
  • 21.Yamaguchi N., Ichijo T., Sakotani A., Baba T., Nasu M. Global dispersion of bacterial cells on Asian dust. Sci Rep. 2012;2:525. doi: 10.1038/srep00525. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Kämpfer P., Busse H.J., Galatis H., Criscuolo A., Clermont D., Bizet C., et al. Paracoccus haematequi sp. Nov., isolated from horse blood. Int J Syst Evol Microbiol. 2019;69:1682–1688. doi: 10.1099/ijsem.0.003376. [DOI] [PubMed] [Google Scholar]
  • 23.Dong K., Pu J., Yang J., Zhou G., Gao Y., Kang Z., et al. Whole-genome sequencing of Paracoccus species isolated from the healthy human eye and description of paracoccus shanxieyensis sp. Int J Syst Evol Microbiol. 2023;73 doi: 10.1099/ijsem.0.005838. [DOI] [PubMed] [Google Scholar]
  • 24.Dé I., Rolston K.V., Han X.Y. Clinical significance of Roseomonas species isolated from catheter and blood samples: analysis of 36 cases in patients with cancer. Clin Infect Dis. 2004;38:1579–1584. doi: 10.1086/420824. [DOI] [PubMed] [Google Scholar]
  • 25.Korsakova E.S., Anan’ina L.N., Nazarov A.V., Bachurin B.A., Plotnikova E.G. Diversity of bacteria of the family Halomonadaceae at the mining area of the Verkhnekamsk salt deposit. Microbiology. 2013;82:249–252. doi: 10.1134/S0026261713020070. [DOI] [Google Scholar]
  • 26.Scoma A., Boon N. Osmotic stress confers enhanced cell integrity to hydrostatic pressure but impairs growth in Alcanivorax borkumensis SK2. Front Microbiol. 2016;7:729. doi: 10.3389/fmicb.2016.00729. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Mizuno M., Yori K., Takeuchi T., Yamaguchi T., Watanabe K., Tomaru Y., et al. Cross-contamination risk and decontamination during changeover after cell-product processing. Regen Ther. 2023;22:30–38. doi: 10.1016/j.reth.2022.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Checinska A., Probst A.J., Vaishampayan P., White J.R., Kumar D., Stepanov V.G., et al. Microbiomes of the dust particles collected from the International space station and spacecraft assembly facilities. Microbiome. 2015;3:50. doi: 10.1186/s40168-015-0116-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Leung M.H., Wilkins D., Li E.K., Kong F.K., Lee P.K. Indoor-air microbiome in an urban subway network: diversity and dynamics. Appl Environ Microbiol. 2014;80:6760–6770. doi: 10.1128/AEM.02244-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Won S., Cho S., Kim H. rRNA operon improves species-level classification of bacteria and microbial community analysis compared to 16S rRNA. Microbiol Spectr. 2024;12 doi: 10.1128/spectrum.00931-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Li R., Tun H.M., Jahan M., Zhang Z., Kumar A., Dilantha Fernando WG., et al. Comparison of DNA-, PMA-, and RNA-Based 16S rRNA Illumina sequencing for detection of live bacteria in water. Sci Rep. 2017;7:5752. doi: 10.1038/s41598-017-02516-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Salter S.J., Cox M.J., Turek E.M., Calus S.T., Cookson W.O., Moffatt M.F., et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 2014;12:87. doi: 10.1186/s12915-014-0087-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Silkie S.S., Tolcher M.P., Nelson K.L. Reagent decontamination to eliminate false-positives in Escherichia coli qPCR. J Microbiol Methods. 2008;72:275–282. doi: 10.1016/j.mimet.2007.12.011. [DOI] [PubMed] [Google Scholar]
  • 34.Crone S., Vives-Flórez M., Kvich L., Saunders A.M., Malone M., Nicolaisen M.H., et al. The environmental occurrence of Pseudomonas aeruginosa. APMIS. 2020;128:220–231. doi: 10.1111/apm.13010. [DOI] [PubMed] [Google Scholar]
  • 35.Ambreetha S., Singh V. Genetic and environmental determinants of surface adaptations in Pseudomonas aeruginosa. Microbiol. 2023;169 doi: 10.1099/mic.0.001335. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.pdf (88.5KB, pdf)
Multimedia component 2
mmc2.docx (20.8KB, docx)

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.


Articles from Regenerative Therapy are provided here courtesy of Japanese Society for Regenerative Medicine

RESOURCES