Pure culture cell survival at surface-air interfaces of desiccation-tolerant eukaryotes and prokaryotes were assessed using the large droplet method (inoculating with a single droplet rather than a spray), and heterotrophic plate counts according to Jawad et al. (1996) and Ronan et al. (2013). Arthrobacter sp., an indoor air prokaryote isolated by Ronan et al. (2013), and Cryptococcus magnus, a bentonite eukaryote isolated by Stone et al. (2016a), were grown separately overnight with agitation at room temperature. Bacteria were grown in Tryptic Soy Broth (3 g/L) and yeasts were grown in Yeast Malt Broth (10 g/L dextrose, 5 g/L peptone, 3 g/L malt extract, 3 g/L yeast extract). Media were prepared according to Atlas (2010) and chemicals purchased from Sigma–Aldrich (Oakville, ON, Canada). Both cultures were washed with sterile tap water three times (7,000 g; 5 min) and diluted in sterile tap water to a cell concentration of 10⁶ cells/mL (bacteria) and 10⁴ cells/mL (yeast). Clean glass coverslips were placed in sterile petridishes (6 per dish, 1 mm × 18 mm × 18 mm, VWR International, Mississauga, ON, Canada), inoculated with 50 μL culture per coverslip and allowed to dry in a laminar flow hood for 3 h. For each of Arthrobacter and Cryptococcus, 75 coverslips were inoculated for incubation at 30% RH, and 75 inoculated for incubation at 75% RH. This procedure was replicated, but the glass coverslips were pre-inoculated with a 50 μL droplet of hygroscopic bentonite clay (100 g/L) solution which was dried in a laminar flow hood for 3 h before inoculation with Arthrobacter or Cryptococcus and subsequent drying and incubation.

Prior to incubation at low and high RH, the cell viability after drying was assessed in triplicate (T₀). Coverslips, inoculated with Arthrobacter or Cryptococcus on glass or bentonite, were placed in separate 50 mL Falcon tubes containing 5 mL saline solution (8.9 g NaCl/L), vortexed for 1 min and dilutions plated on Tryptic Soy Agar or Yeast Malt Agar, respectively, for determination of viable cell concentrations per coverslip, as optimized in Ronan et al. (2013). This was repeated at 3 h, 24 h, 48 h and 3, 11, 15, 43, 65, and 234 days. At each time point, triplicate coverslips were assessed for (1) Arthrobacter viability at 30% RH and 75% RH, on glass and bentonite, and (2) Cryptococcus viability at 30% RH and 75% RH, on glass and bentonite. Viable cell concentrations were represented as a percentage of the cell concentration at T₀ (desiccated inoculum). Means and standard deviations of triplicate coverslips were calculated in Microsoft Excel and plotted in the Veusz plotting package¹ against time. These programs were used for all statistical analyses and figure plots in this study.

Throughout the study, RH was maintained using saturated MgCl₂ (33%) and KCl (85%) solutions (100 mL/chamber) in covered 5 L glass chambers, according to Greenspan (1977). RH was monitored with 915 MHz Wireless Temperature Stations (La Crosse Technology, Saint-Laurent, QC, Canada). Bentonite-altered chamber RH, as compared to those reported by Greenspan (1977), stabilized at 23–30% (hereafter referred to as 30%) and 75–79% (hereafter referred to as 75%), consistently returning to the controlled RH within an hour after opening and resealing.

The CO₂ generation of a mixed culture (Table 1; Stone et al., 2016a) desiccated at 30% RH and 75% RH on (1) clay vs. sand and (2) PEG vs. plastic/glass were compared during desiccation. The experiment involved a number of steps, detailed in Figure 1. Clay (Fisher, 1923) and PEG-4000 (Dontula et al., 1998) were chosen as hygroscopic substrates, whereas sand (Fisher, 1923) and glass/plastic (Dontula et al., 1998; Harbers et al., 2007) were chosen as neutral substrates.

The influence of biofilm metabolic rates on biofilm desiccation resilience (capacity to recover) was studied in biofilm flow systems, in both a desiccation-tolerant and desiccation-sensitive bacterial species. Pseudomonas aeruginosa PAO1 (desiccation-sensitive; Ronan et al., 2013) and Arthrobacter (desiccation-tolerant; Ronan et al., 2013) biofilms were grown in continuous systems and open-loop carbon dioxide generation was monitored as an indicator of biofilm activity. A schematic visualization of the experimental setup is detailed in Figure 2. Continuous flow systems were assembled according to Bester et al. (2010). Briefly, two biofilm flow systems were sterilized with ethanol (1 h) followed by a diluted (2 in 10) commercial bleach solution (3 h). The system was washed (37.5 mL/h) with sterile tap water overnight (10 h), equilibrated with 0.3 g/L TSB in tap water (1 h) and each system was inoculated identically with either Arthrobacter or Pseudomonas cells upon stopping the flow of media (0.5 mL, ~10⁶ cells/mL, grown in TSB for 14 h). After a 45 min (Pseudomonas) or 6 h (Arthrobacter) attachment period, medium flow was restarted. At 50 h (well into the stationary phase of the CO₂ profile), the medium of one biofilm was changed to sterile tap water. The two identical biofilms were continuously exposed to (a) tap water (oligotrophic) or (b) 0.3 g/L TSB in tap water (carbon-rich), respectively, for 24 h and subsequently run dry. Before running dry, 3 mL of sterile tap water was run through both biofilms at a higher rate (2.5 mL/min) to wash the TSB biofilm. Both biofilms had filter-sterilized air delivered (5 L/min) for 2 days. The desiccated biofilms were then re-exposed to TSB (0.3 g/L), with a 30 min incubation period, and the recovery respiration rate, measured with CO₂ production and normalized as a percentage of the pre-media change biofilm respiration rate, was used as an indicator of post-desiccation biofilm resilience. The CO₂ molar flow rate during biofilm desiccation was also compared between oligotrophic and carbon-rich conditions. The experiment was repeated in triplicate, for each pair of Pseudomonas and Arthrobacter biofilms. The (a) desiccation rate during desiccation and (b) resilience (post-desiccation respiration rate as a percentage of initial biofilm respiration rate) were compared between oligotrophic and carbon-rich biofilms per species, using a Student’s t-test of independent means for normally distributed data, with a confidence interval of 95%.

Following inoculation with the large-droplet technique, the survival of both the eukaryote and prokaryote was extended at bentonite-air interfaces in comparison to glass-air interfaces, at both high and low RH (Figure 3). As bentonite provides more carbon and trace elements than clean glass (Karnland et al., 2006), such an extension of survival is expected. Also, the eukaryotic isolate, C. magnus, demonstrated better viability under desiccation than the prokaryote, which was an indoor air isolate selected specifically for its reported desiccation-resistance (Ronan et al., 2013). The extended viability of Cryptococcus at both bentonite- and glass-air interfaces, as compared to Arthrobacter, confirms several reports of increased desiccation tolerance in eukaryotes (Figure 3; Brown, 1976; Morano, 2014). However, the trend of viability inhibition at high RH remained consistent throughout this study, despite the nature of both the surface and the inoculum. Thus, the conclusions of Ronan et al. (2013) were confirmed and extended to include a desiccation-tolerant eukaryote and hygroscopic clay surfaces. This also supported the long-term survival trends of natural prokaryotic and eukaryotic populations in dry bentonite reported in Stone et al. (2016a).

Microbial viability at surface-air interfaces is extended at low RH and suppressed at high RH, despite species variation (Turner and Salmonsen, 1973; Walters et al., 2005; Ronan et al., 2013). Although clays are a well-known buffer against environmental stressors, a recent study of bentonite clay microbiota showed that the suppression of microbial viability at high RH occurs within such a clay hygroscopic matrix too, suggesting that surface-bound water does not decrease the impact of RH on microbial viability. Using heterotrophic plate counts to assess the natural microbiota of bentonite after a year of dry incubation at low (30%) and high (75%) RH, both eukaryotic and prokaryotic populations were consistently significantly less viable at high (75%) RH than at low (30%) RH (Stone et al., 2016a). Ronan et al. (2013) also showed, using the large-droplet method, that the survival of a desiccation-tolerant indoor air isolate (Arthrobacter sp.) was significantly inhibited at high (75%) RH. This technique was employed in the current study to extend the above-mentioned observations, exploring the combined influence of surface hygroscopicity and water vapor on microbial longevity at surface-air interfaces. Due to its application as a sealing material in deep geological repositories for nuclear waste (Stroes-Gascoyne et al., 2011), as well as the broader environmental occurrence of clays in soil and wetland environments, Wyoming MX-80 bentonite montmorillonite clay was chosen as the primary hygroscopic microbial substrate in this work. In order to extend the work of Ronan et al. (2013) to include eukaryotes and hygroscopic surfaces, the conclusions were extended to include the same desiccation-tolerant Arthrobacter sp. air isolate, as well as a desiccation-tolerant eukaryotic bentonite isolate (C. magnus) on both glass coverslips and on glass coverslips coated with hygroscopic bentonite clay.

Spore survival and viability is often explored in desiccated conditions, but even vegetative bacteria can survive for years, challenging our understanding of bacterial physiology and metabolism (Otter et al., 2015). In addition, the metabolic state of viable but non-culturable organisms has yet to be elucidated. On this note, many of the organisms explored in this study – including the desiccation tolerant species – are vegetative non-spore formers, including Arthrobacter, Cryptococcus, and Pseudomonas, highlighting the emphasis of this study on metabolic persistence in relation to long-term viability. In contrast to survival inhibition in bentonite at high RH, the above-mentioned study (Stone et al., 2016a) also showed that the hygroscopic clay-air interface allowed for increased short-term CO₂ production of a microbial community at high (75%) RH. However, microbial metabolic CO₂ production was below the closed-loop CEMS detection limit at low (30%) RH, even at the hygroscopic clay interface. This observation led to the proposal that hygroscopic surfaces improve the access of microbial cells to water vapor at surface-air interfaces. The hypothesis designed to test this states that if a hygroscopic matrix improves microbial access to water vapor at surface-air interfaces, then both clay and PEG-4,000 at high (75%) RH will stimulate higher CO₂ production in microbial communities at surface-air interfaces than less hygroscopic surfaces such as sand and plastic. Of the eight combinations of neutral (sand and plastic) and hygroscopic (clay and PEG) surfaces inoculated with a mixed microbial community (Table 1) and incubated at high and low RH, the accumulation of CO₂ in a closed-loop CEMS (Figure 1) was only observable for microbial communities at clay-air interfaces at high (75%) RH and PEG-air interfaces at high (75%) RH (Figures 4 and 5; Table 2). For all un-inoculated controls, the production of CO₂ was approximately zero and all inoculated microcosm tubes incubated at 30% RH – irrespective of surface hygroscopicity – were not significantly higher than the controls, according to a Mann–Whitney U test for non-parametric data (Figure 5; Table 2). Any slightly negative CO₂ production rates are likely due to the matrix CO₂ absorption, which is related to pH and RH at surface-air interfaces (Stone et al., 2016b). On both sand and plastic surfaces, there was no significant difference between the means of CO₂ production (ppm/min) between samples incubated at 30% RH and 75% RH. Microbial communities desiccated on a bentonite surface and incubated at high (75%) RH exhibited a mean CO₂ production rate of 0.45 ppm/min, significantly higher than a sterile control sample (Figure 5; Table 2). The mean CO₂ production rate of the same microbial community on PEG was lower than bentonite (0.13 ppm/min), but still significantly higher than the mean sterile PEG control (Figure 5; Table 2).

The hypothesis that the combination of hygroscopic surfaces and high RH extended microbial metabolism at surface-air interfaces was thus supported, with significantly microbial CO₂ production occurring only at high RH on bentonite and PEG (Figures 4 and 5; Table 2). In both cases, only the combination of hygroscopic interfaces and high RH significantly promoted microbial metabolism during desiccation (p < 0.05).

The prolonged metabolism at bentonite-air interfaces as compared to PEG-air interfaces is again likely due to the mineralogical properties of bentonite, providing the microbes low levels of carbon and trace elements (Figures 4 and 5; Karnland et al., 2006). In addition, the microbial community comprises predominantly isolates of bentonite origin (Stone et al., 2016a), likely more adapted to bentonite than PEG. The porosity and increased surface area of bentonite and PEG, as compared to sand and plastic, are intimately related to the hygroscopic capacity of these matrices. It has been shown that laboratory mineral dissolution rates are generally exaggerated due to higher clay-surface reactive sites than natural clay deposits (Maher et al., 2006). Similarly, the activity of truly dry soil microbial populations is often difficult to extricate from the activity of microbial populations in occluded water within the soil (Stotzky and Pramer, 1972). A more fundamental link between hygroscopic surfaces, water vapor and microbial activity could be further pursued on flat surfaces coated with smooth chemical hygroscopic compounds, however, these ‘interfering’ variables of porosity and surface area are characteristic of many hygroscopic surface-air interfaces in nature and are thus more relevant to this study. Although this work did address the challenge of measuring slower growth and metabolic rates at surface-air interfaces, as opposed to the better-studied rates in liquid culture, there are elements that were not representative of in situ scenarios. High cell concentrations was the primary unnatural parameter in most of the surface-air interface analyses, likely more representative of biofilms present in the richer environments of nosocomial, wastewater or food industries – many of which are exposed to periods of desiccation – rather than organisms that land on surfaces during air or vector dispersion.

The final hypothesis expounded on the observations that microbial survival is inhibited at high RH, whereas short-term metabolic persistence is improved at high RH. The hypothesis, linking these survival and metabolism observations, proposes that if a higher persistent metabolic rate during desiccation lowers the survival of a desiccated microbial community at surface-air interfaces, then an oligotrophic biofilm will demonstrate a higher metabolic desiccation-recovery capacity (termed resilience in this study) than a metabolically active biofilm. This hypothesis was evaluated in both a desiccation-tolerant (Arthrobacter) and desiccation-sensitive (P. aeruginosa) species. In Arthrobacter and P. aeruginosa biofilms, initial growth for 2 days, starvation for 1 day and desiccation for 2 days allowed for a full recovery of both starved and metabolically active biofilms, facilitating a comparison of respiration profiles during desiccation and recovery with online CEMS (Figures 2, 7, and 8). In each experiment, both biofilms were grown on TSB well into a steady-state metabolic profile. Subsequently, one biofilm was continuously fed TSB (carbon-rich), whilst the other was changed to tap water (oligotrophic) at the “Media Change” point, for approximately 24 h (Figure 7). After washing and desiccation, the resilience (metabolic recovery from desiccation, normalized against the pre-desiccation biofilm metabolic rate) was compared between the carbon-rich and oligotrophic biofilms. Figure 7 demonstrates the CO₂ profiles of two individual replicate biofilms per experiment, whereas Figure 8 is a summary of the means of three replicates per experiment. For desiccation-tolerant Arthrobacter biofilms, CO₂ generation persisted during desiccation of a carbon-rich biofilm, whereas the oligotrophic biofilms showed greater resilience (recovery respiration rates, normalized for each individual biofilm as a percentage of the pre-desiccation respiration rates), supporting the hypothesis (Figures 7A,B). The average respiration rate during desiccation was consistently significantly lower for an oligotrophic biofilm in independent triplicate scenarios (Figure 8A; Table 3), whereas the average respiration rate post-desiccation (normalized against the pre-desiccation respiration rate) was consistently significantly higher for oligotrophic biofilms in independent triplicate systems (Figure 8B; Table 3). Significant differences, with a confidence interval of 95%, were calculated per biofilm with a Student’s t-test of independent means (Table 3). In contrast, the desiccation respiration rates and post-desiccation respiration rates showed no consistent trends in triplicate P. aeruginosa biofilms (Figures 7C,D and 8C,D; Table 3), suggesting that within these experimental parameters, metabolic activity prior to desiccation has no influence on the resilience of a desiccated P. aeruginosa biofilm.

These distinctive respiration profile responses in desiccation-tolerant versus desiccation-sensitive species indicate that the hypothesis might play a role in biofilm adaptation to desiccation, in addition to morphological and physiological adaptations. The higher CO₂ production rate of Pseudomonas than of Arthrobacter (both pre- and post-desiccation) when grown on TSB supports studies showing that the species-specific utilization of the carbon source has a distinct effect on both metabolic rate and adaptive resilience (Jackson et al., 2015). As previously suggested, “the altered morphology and/or growth patterns of bacteria growing at low humidity might be more ecologically relevant than their textbook appearance at high humidity since their natural habitats are often dry” (Lidwell and Lowbury, 1950). Longer time periods of starvation may facilitate physiological adaptation in relation to metabolic rates, and microscopy could narrow this window by direct observation of such morphological adaptation. Another valuable extension of the study would be to assess the response of mixed community biofilms to the same starvation and desiccation conditions. In a study bringing the osmolyte accumulation hypothesis into question in situ in soils (Kakumanu et al., 2013), adaptation to desiccation was more clearly linked to population diversity and dynamics than osmolyte accumulation or cell turnover, suggesting the importance of species studied. Furthermore, desiccation cycles and gradients (de Goffau et al., 2009) are representative of natural environments and could be incorporated into similar studies, since the impact of cycles such as the circadian rhythm on metabolism is extensive (Kovac et al., 2009; Trinder et al., 2015; Ho et al., 2016b).

Previous studies of microbial populations in dust showed that lower temperature decreased the thermodynamic potential for high metabolic rates, thus lowering the aging of a population and extending survival rates (Lidwell and Lowbury, 1950). Similarly, high (80–90%) RH stimulates microbial metabolism in house dust, as well as the growth of spore-bearing fungi (Korpi et al., 1997). Also, RH approaching saturation levels demonstrated some short-term increase in viable cells in dust microbial populations, prior to the typically reported long-term inhibition of viability at high RH (Lidwell and Lowbury, 1950). These authors proposed that the mechanism of inhibition at high RH was the concentration of salt ions dissolved at the hygroscopic interface. However, it is possible that metabolic persistence at high RH could mirror the negative thermodynamic influence of metabolically favorable temperatures on cell viability. Persistent metabolism could prevent desiccation-adaptive mechanisms, or higher metabolic rates during the initial stages of desiccation could decrease the total long-term nutrients and energy available to the community.

There is conflicting evidence regarding the influence of metabolic rates on the desiccation tolerance of biofilms. Some studies have shown that higher carbon fluxes lead to thicker biofilms, better suited to adaptation and stress responses (Condell et al., 2012). Others have rather emphasized adaptation due to stress exposure, claiming that the extensive morphological and physiological responses to desiccation are “not just indirect responses to stress, but rather responses of these cells to become more stress resistant” (de Goffau et al., 2009). Some of these morphological SOS responses include down-regulation of autolysin (leading to a lower growth rate) and the accompanying filamentation and decrease in cell separation, and increases in cell size, cell wall thickness and antibiotic resistance. At high RH more cell division is evidenced (de Goffau et al., 2009), supporting the proposal that persistent metabolic activity is linked to increased water availability and lower long-term viability.

Slow growth rates within biofilms have also been linked to increased antibiotic resistance (Walters et al., 2003; Otter et al., 2015). Slow growth may be sustained in biofilms by cell death (programmed or natural), predation or cannibalism. The concept of bacterial cell suicide was initially viewed with skepticism, since bacteria were considered single-cell organisms. However, increased evidence of programmed cell death within biofilms has strengthened the concept of a biofilm as a “whole organism” (Bayles, 2014). This type of terminology has also been used within medicine, referring to the gut microbiome as an “organ” (Trinder et al., 2015). Thus, viewing a biofilm as a whole organism capable of metabolic pacing increases the legitimacy of the theory that cell death and nutrient cycling could contribute to survival and metabolism at surface-air interfaces. In this scenario, an environment that stimulates persistent faster metabolism would deplete accessible nutrients faster. Similarly, predation has been shown to influence metabolic regulation in biofilms, with ciliate predation of Cryptococcus biofilms preventing stagnation and stimulating persistent metabolism (Joubert et al., 2006). Thus, slow metabolism based on cell death could be a direct adaptive response linked to community desiccation resilience. Since this SOS response of programmed cell death might be directly linked to growth rate and viability (Bayles, 2014), an interesting future study would be to compare the rates of programmed cell death and predation at high and low RH. Bester et al. (2010) demonstrated that in a saturated carbon-fed system, planktonic cell output for proliferation only costs the biofilm 1.0 ± 0.2% of the total carbon turnover. Within a liquid system, these cells contribute to biofilm spawning. However, within a desiccated biofilm – reflecting a closed nutrient system – such cell production could maintain a low level of persistent metabolism due to natural cell lysis, predation, cannibalism, or cell suicide. Regulation of the rate of this metabolism would thus be an effective adaptation to desiccation. The observation of Ronan et al. (2013) that cell viability during desiccation is greatly improved by the large droplet method as opposed to the spray method supports this contention of nutrient cycling within a population.

Diverse experimental parameters pose a challenge in the mining of hospital, food preparation and built ecology literature for support or contradiction of the results presented here. However, the survival of methicillin-resistant S. aureus after desiccation was lower at 40% RH than at 16% RH, across varying surface properties (glass, wood, plastic, cloth; Coughenour et al., 2011). Similarly, Worth Calfee and Wendling (2012) demonstrated that low temperature and RH enhanced the survival of Brucella suis on various building materials. These studies support the premise of this work, that higher RH might lead to increased microbial access to water vapor, promoting faster metabolic activity and inhibiting long-term survival by decreasing the total energy available to the community. High RH also improved the transfer efficiency of five bacterial, phage and viral species, on both porous and non-porous surfaces (Lopez et al., 2013). The authors concluded that since transfer efficiency is better under wet than dry conditions, the morphological activity necessary to promote transfer might be sustained at higher RH due to microbial access to water, also supporting the proposal of the current study that high RH promotes persistent microbial metabolism at surface-air interfaces. Similarly, airborne dispersion has been significantly and positively correlated to RH (Obbard and Fang, 2003). Again, assuming that there is an energy cost to microbial dispersion (morphological shifts that are energy expensive), this also supports the findings of this study. The antimicrobial activity of hygienic paint, containing silver nanoparticles and titanium dioxide, against Escherichia coli and methicillin-resistant S. aureus was greater at high RH than at low RH (Dominguez-Wong et al., 2014). Since low metabolic rates are correlated with higher antibiotic resistance, the hypotheses are further supported.