B. ovatus cultivation

A lyophilized Bacteroides ovatus strain ATCC 8483 received from American Type Culture Collection was first suspended in inoculating fluid (Biolog Inc. 72401), followed by inoculation in anaerobic medium [27] with a modified glucose content of 0.5% (w/v). After the cell suspension reached an optical density at 600 nm (OD₆₀₀) of 1, 0.5 ml of the suspension was collected and subsequently inoculated into a serum bottle with 60 ml of anaerobic medium. B. ovatus cells were harvested after 14.5 h of incubation at 37°C by centrifugation at 15,557 g for 16 min at 20°C. The pellets were washed three times with 10 mM sterilized phosphate buffer solution at pH 8.2 by sequential centrifugation at 15,557 g before irradiation experiments.

Extracellular AOM preparation

Bulk microalgae solution of the species S. costatum LB 2308 propagated in Erdschreiber’s medium was purchased from The Culture Collection of Algae at the University of Texas at Austin (UTEX). The bulk solution was subject to centrifugation at 2740 g for 10 min, and the supernatant containing organic cell exudates was collected. The supernatant was filtered through a cellulose acetate membrane of 1.2 μm pore size (EPS®, Inc.) to remove algae cells and debris. The filtered supernatant was subjected to dialysis using 3.5 k Dalton MWCO membrane (Thermo Scientific SnakeSkin, 88245) against deionized water. The dialysis concentrate, which is referred to as the extracellular AOM, was harvested when the solution conductivity was below 10 μS/cm. The extracellular AOM after dialysis had total organic carbon (TOC) of 1.94 mg C/L, specific UV absorbance at 254 nm SUVA of 1.03, pH of 6.18, and salinity < 1 parts per thousand at 27°C.

Experimental setup

Eighty ml glass beakers (Pyrex) were used as microcosms, which were wrapped with black tape (3 M) to prevent the reflected and deteriorated simulated sunlight screened by the beaker from reaching the liquid inside the microcosm. Mixing was facilitated by a Variomag electronic stirrer set at 130 rpm. Seawater temperatures recorded near King Abdullah University of Science and Technology (KAUST) were 20°C in February 2013 and 36°C in May 2013. Therefore, experiments in this temperature range were conducted. Temperature of the reactors was regulated by a water bath set at 27°C. A reactor covered with a piece of aluminum foil was used as the dark control.

Solar exposure was conducted in an Atlas Suntest® XLS + photosimulator (Chicago, IL) equipped with a xenon arc lamp. A 280 nm cutoff filter (Newport, MA) was placed on top of reactors to prevent liquid loss through evaporation. A 320 nm cutoff filter (Newport, MA) was used to determine the significance of UVB for B. ovatus inactivation. An ILT950 spectroradiometer (International light technologies, MA) was used to measure the irradiance from 280 to 700 nm inside the solar simulator. The xenon arc lamp was set at an irradiance of 400 Wm^-2 with a UV special glass filter, which delivers 3248 μW/cm² of UV irradiance received by the reactors. In comparison, the UV irradiance on the University of Illinois campus in Urbana during a clear day at 12 pm on April 22^nd was 3899 μW/cm², and on October 21^st a total UV irradiance of 2618 μW/cm² was measured at around 2 pm on the KAUST campus during a clear day in Jeddah, Saudi Arabia. To convert the irradiance (μW/cm²) to fluence (J/cm²), cumulative irradiance over 280-700 nm wavelength for a specified duration of time was calculated.

A final cell concentration of 10⁹-10¹⁰ cells/ml was used for all solar exposure experiments. For experiments with low salinity and without AOM, 10 mM sterilized phosphate buffer solution (Fisher Scientific Inc.) was prepared to resuspend the washed B. ovatus cell pellet to 60 ml. Samples were taken intermittently after the initiation of the exposure to the simulated sunlight. 1.2 ml of aliquot was drawn out of each microcosm and stored at -20°C for a day before DNA extraction using the Ultraclean Soil DNA Isolation Kit (Mo-Bio, CA) following the manufacturer’s instructions. Two batches of samples were collected from a given reactor, one was treated with PMA to determine intact cell count and the other was not. Both batches were subsequently enumerated with qPCR assay. For reactors with AOM, natural organic matter stock solution was added to the beaker, and 500 mM sterilized phosphate buffer (Fisher Scientific Inc.) was used to adjust the pH to 8.2. Ionic strength was then compensated with sodium chloride to 10 mM. For selected experiments, 10 mg/L of NaNO₂ was used as synthetic sensitizer of · OH. Preliminary experiments on Escherichia coli to optimize the majority of the workflow were done in triplicates. Experiments done on B. ovatus were conducted five times as a result of the power analysis (1–β = 80%).

To mimic the high salinity condition that B. ovatus may experience upon being released into the sea, solar exposure experiments were also conducted in artificial seawater, containing 32 g/L of sodium chloride, 0.8 g/L of potassium chloride, 1.3 g/L of calcium chloride, and 6.1 g/L of magnesium chloride. This ionic concentration was selected to match salinity of Red Sea water at 39.5 parts per thousand collected on the KAUST campus [28]. Experiments conducted with this artificial seawater were referred to as high salinity, while experiments conducted with solutions of 10 mM ionic strength were referred to as low salinity.

Propidium monoazide (PMA) treatment

PMA (Biotium Inc., CA) was used to count cells with intact membrane. This chemical binds to double stranded DNA in both the extracellular environment and in cells with compromised membrane, preventing subsequent qPCR amplification as previously described in other studies [6, 17]. 2 mM of PMA stock solution was prepared using 20% DMSO as the solvent and stored in the dark at -20°C until used. For each PMA treatment, 63 μl of the stock solution was added into the 1.2 ml sample to achieve 100 μM of the final PMA concentration. The samples containing PMA were incubated for 5 min on a rotary shaker in the dark, followed by exposure to two 500 W halogen lamps for 6 min at a distance of 20 cm to photoactivate the dye. During the exposure, the samples were placed on top of the aluminum foil with a tray of ice beneath them to prevent samples from overheating.

Quantitative PCR (qPCR)

Amplification and detection of the targeted 16S rRNA gene of B. ovatus were used to enumerate both intact and/or compromised cells. 384-well clear optical reaction plates (Applied Biosystems®, CA) were used in an ABI PRISM 7500 HT real time quantitative PCR machine (Applied Biosystems®, CA). A MicroAmp optical adhesive film (Applied Biosystems®, CA) was used to cover each 384-well clear optical reaction plate to prevent evaporation and contamination. Each 15 μl qPCR reaction mixture contained 2 μl of DNA template and 13 μl of qPCR mastermix. The mastermix consisted of a mixture of Power SYBR Green (Applied Biosystems®, CA), 0.2 μM Bac32F and Bac303R primers (IDT, IA), and molecular biology grade water (Thermo Scientific, MA). Triplicate analysis was done for each sample, and a standard curve was generated with serial dilution of Promega pGEM®-T Easy Vector (Madison, WI) harboring the entire 16S rRNA gene from B. ovatus. The thermal cycling parameters include an initial 2 min at 50°C, followed by 10 min at 95°C, and 40 cycles of 15 s at 95°C and 1 min at 60°C. A dissociation cycle of 15 s at 95°C, followed by 15 s at 60°C, ramping up to 95°C for 15 s was done to assess for non-specific amplification. Two samples that contained all the qPCR reagents except for the DNA samples were prepared as negative controls to check for PCR contamination. The detection limit for the Bac32 F and Bac303R primers is 29 copies per reaction, with amplification efficiency for the set of primers ranging from 90-94%.

Decay rate calculations

B. ovatus cells with intact membrane will be referred as “intact cells” and cells with compromised membrane will be referred as “compromised cells” below. Pseudo-first decay model was used to fit the data and k _obs , the pseudo-first decay rate constant, was obtained by linear regression fitting. Light screening correction was done on k _obs to account for the light absorption and scattering in the solution [29]. Besides the inactivation rate constant, the half-life of B. ovatus decay was also calculated. Paired sample t-test was used to determine if the treatment group was significantly different from the dark control group. One-way ANOVA test was done to determine if k _obs or half-life among more than three treatment groups were significantly different from each other. Power 1-β = 80% was used for all statistical tests.

Influence of UV irradiation on B. ovatus decay

For the full spectrum irradiation experiment at 27°C comprising UVB, UVA, and visible light exposure, 4 log₁₀ of decay was observed over 6 h of experiment, with a pseudo-first decay rate constant (k _obs ) significantly different from zero (p < 0.05, α = 0.05, Table 1, Figure 1). The total B. ovatus cells, including both intact and compromised, decreased 25% over 6 h of exposure to full solar spectrum. When UVB was blocked and B. ovatus cells were exposed to 6 h of simulated sunlight containing only UVA and visible light, 1.5 log₁₀ of decay was observed. The presence of UVB clearly led to faster B. ovatus decay, reflecting in an additional 2.5 log₁₀ of decay over 6 h of exposure. As a result, to reveal a more prominent decay trend in subsequent experiments, full spectrum irradiation containing UVB, UVA, and visible light was used. In low salinity water irradiated by full spectrum simulated sunlight, the half-life calculated based on the pseudo-first decay model for intact cells was 0.08 h, or 4.8 min. This half-life is significantly shorter than 9 h, which was determined for cells kept in the dark. For the dark control, no significant growth or decay was observed (Figure 1).

Condition	k obs in hr-1for low salinity water (half-life in h)	k obs in hr-1for high salinity water (half-life in h)
Phosphate buffer irradiated by UVA and visible light	4.1 ± 2.0	Not available
	(0.21 ± 0.12)
Phosphate buffer irradiated by full spectrum	9.1 ± 1.0	3.3± 0.4
	(0.08 ± 0.009)	(0.21 ± 0.02)
Phosphate buffer dark control	0.08 ± 0.04	-0.01 ± 0.04
	(9.17 ± 3.28)	(4% increase)*
Phosphate buffer and NaNO2 irradiated by full spectrum	6.7 ± 1.7	4.2 ± 0.3
	(0.11 ± 0.02)	(0.17 ± 0.01)
Phosphate buffer and NaNO2 dark control	0.06 ± 0.04	0.02 ± 0.01
	(12.84 ± 7.18)	(34.12 ± 17.5)
AOM solution irradiated by full spectrum	2.1 ± 0.3	4.0 ± 1.2
	(0.34 ± 0.05)	(0.19 ± 0.04)
AOM solution dark control	-0.02 ± 0.04	0.04 ± 0.01
	(7% increase)*	(16.18 ± 0.19)
NaNO2 AOM solution irradiated by full spectrum	4.7 ± 1.9	7.0 ± 2.0
	(0.17 ± 0.05)	(0.1 ± 0.03)
NaNO2 AOM solution dark control	-0.05 ± 0.08	-0.03 ± 0.02
	(36% increase)*	(9% increase)*

*Half-life could not be calculated due to the increase in biomass, therefore the percentage of growth was calculated. Data were corrected for light screening. Kinetic data are also plotted in Additional file 1: Figure S1.

Influence of exogenous · OH and algae organic matter on B. ovatus decay in low salinity water

The role of exogenously produced · OH was studied because these highly reactive radicals can non-specifically attack and oxidize cellular components [30]. Production of 0.77 ± 0.08 fM · OH in low salinity water was observed from solutions containing 0.14 mM NaNO₂ exposed to full solar spectrum irradiation at 27°C. The decay rate constants (k _obs ) measured based on the intact cells were not significantly different for solutions with and without NaNO₂ in the low salinity water (p > 0.1, α = 0.1, Table 1, Figure 2). These similar k _obs values suggested that the presence of 0.77 ± 0.08 fM · OH did not significantly impact the inactivation of B. ovatus in low salinity water.

The solution containing AOM has a half-life of 0.34 h, the longest among the data recorded for low salinity irradiated experiments. For the solution containing NaNO₂, half-life of 0.11 h was significantly shorter than those obtained for solutions containing AOM but not significantly different than those obtained for sensitizer-free solutions (p > 0.05, α = 0.05, Table 1). When solutions were kept in the dark, long half-life values of more than 12 h were observed. In addition, 7% increase in B. ovatus biomass was observed for solutions containing AOM kept in the dark.

Influence of salinity on B. ovatus decay

For experiments without exogenous sensitizers and experiments with only NaNO₂ as the sensitizer, the decay rate constants of B. ovatus in high salinity solutions under full spectrum simulated sunlight were significantly smaller than that at lower salinity (p < 0.05, α = 0.05, Figure 3, Additional file 1: Figure S1b). Less than 1 log₁₀ of B. ovatus decay was observed over 6 h of exposure, with the total cell count changed insignificantly throughout the experiment. Similar to the findings with low salinity experiments, 0.14 mM NaNO₂ in AOM-free high salinity solutions did not significantly change B. ovatus decay rate constants (p > 0.05, α = 0.05, Table 1). For AOM-containing solutions, high salinity did not lead to a significant decrease in k _obs values (p > 0.05, α = 0.05, Table 1). In both cases, 1 log₁₀ decay of intact B. ovatus was observed, with the total cell count decreased slightly over time. In solutions containing both 0.14 mM NaNO₂ and AOM, and only AOM, high salinity led to faster B. ovatus decay compared with corresponding low salinity settings (7.01 h^-1 vs.4.71 h^-1; 4.03 h^-1 vs.2.06 h^-1, Table 1), despite similar · OH concentrations (p > 0.05, α = 0.05, Table 2). Mixing 0.14 mM NaNO₂ and AOM in a high salinity setting yielded an unexpected result, with B. ovatus decaying faster than when either 0.14 mM NaNO₂ or AOM was present alone in the high salinity environment (p < 0.05, α = 0.05, Table 1, Additional file 1: Figure S1c). However, this synergistic effect was not observed for low salinity solutions.

For high salinity solutions irradiated by full spectrum simulated sunlight, a half-life of 0.21 h was recorded, which was longer than that in low salinity solutions (p < 0.05, α = 0.05, Table 1). The half-life values were statistically the same for high salinity solutions with and without NaNO₂ (p > 0.05, α = 0.05). The presence of both NaNO₂ and AOM in high salinity solutions led to the observed half-life of 0.1 h, which was significantly shorter than when either NaNO₂ or AOM was present alone in the high salinity environment (p < 0.05, α = 0.05, Table 1). When both NaNO₂ and AOM were present, high salinity half-life was also shorter than those obtained from the low salinity (p < 0.05, α = 0.05, Table 1, Figure 4). For all dark controls in high salinity experiments, the decay kinetics of total B. ovatus cells had a zero slope (p > 0.05, α = 0.05), indicating that salinity alone did not exert a significant impact on B. ovatus persistence in the dark. Irradiation experiments in high salinity water with UVB cut-off was not conducted because UVB irradiation was already identified as an important decay mechanism for B. ovatus.

Mechanisms of B. ovatus decay

From the dark control data for low and high salinity conditions, PMA-qPCR did not detect changes in the copy numbers of total cells (i.e., B. ovatus with both intact and compromised membrane) and intact cells, suggesting that solution conditions alone did not lead to cell decay or membrane damage. This may be explained by the fact that though B. ovatus are strict anaerobes, they possess superoxide dismutase, which helps to fight oxidative stress to a certain level [31, 32]. The oxidative stress created by dissolved oxygen for dark controls happened to be within the capacity of superoxide dismutase, enabling B. ovatus to remain intact in the dark control. However, for irradiated samples, total cells decayed much less than intact cells, suggesting that irradiation damaged cell membranes. This is in agreement with past studies that reported damage to viral protein capsid by medium-pressure UV and endogenous inactivation of E. coli by UVB [33, 34].

In addition, bacteria with larger genome sizes were observed to be more susceptible to UV damage, presumably because larger genomes offered more base per sites for UV damage [35]. The size of the B. ovatus constructed from 32 scaffolds was approximately 6.5 Mbp, and was larger than the genome of Pseudomonas aeruginosa 01 (i.e., 6.3 Mbp) that was evaluated by McKinney and Pruden [35]. P. aeruginosa 01 had the largest genome compared to methicillin-resistant S. aureus, vancomycin-resistant enterococci, and E. coli SMS-3-5, and was determined to be most susceptible to UV damage. As such, it is likely that B. ovatus is equally, if not more, susceptible to UV radiation than P. aeruginosa 01.

The statistically similar inactivation rate constants with and without extracellular · OH produced by NaNO₂, both in low or high salinity settings, suggest that exogenous inactivation was much less significant than endogenous inactivation mechanism for B. ovatus. Specifically, the presence of exogenous · OH produced by 0.14 mM of NaNO₂ did not promote significant B. ovatus decay. In surface water, nitrite is usually converted to nitrate during nitrification. Only in places with incomplete nitrification can nitrite concentrations of up to 66 μM be found [36]. In seawater, the concentration of nitrite has been found to be even lower, and accounts for a concentration of 10.5 ± 2 nM [37]. Thus, it is likely that the persistence of B. ovatus in water bodies would not be controlled by environmentally-relevant concentration of · OH produced by nitrite or nitrate.

The presence of AOM in the extracellular environment slowed down the decay of B. ovatus in the low salinity setting, suggesting that AOM may protect B. ovatus from sunlight irradiation. When B. ovatus is discharged into the low salinity water environment where AOM is also present on a sunny day, 50% of the total intact cells will be lost within 20 min, persisting longer than those conditions without AOM (i.e., 5 min). Our observations suggest that in pockets of freshwater directly exposed to discharge from wastewater treatment plants, it is anticipated that the AOM would be higher than 1.94 mg C/L and would facilitate a longer persistence of B. ovatus. A possible explanation to account for a longer persistence of B. ovatus in low salinity water with AOM may be due to the partial coverage of the cell surface where the endogenous sensitizers are located by AOM. The solar irradiation that reaches the cell to produce endogenous reactive oxygen species could be attenuated by this layer of AOM, resulting in less endogenously produced · OH that can inactivate the cells.

In contrast, the presence of AOM in extracellular environment did not slow down the decay of B. ovatus in high salinity water. This can possibly be accounted for by the partial scavenging of endogenously produced or exogenously produced · OH in the presence of chloride ions [38]. Therefore, when B. ovatus is released into seawater with AOM, B. ovatus cells have a half-life of 11.4 min, which is essentially the same as the half-life for the high salinity water without AOM (i.e., 12.6 min). A synergistic effect of AOM, · OH, and high salt concentration led to a faster decay of B. ovatus in high salinity environment than that without NaNO₂ (p < 0.05, α = 0.05, Table 1). However, this synergistic effect only translates to a difference of 4.2 min in the half-life of B. ovatus, which might not be apparent when performing actual fecal monitoring efforts on-site.

Implication for use as fecal indicator

Sunlight irradiation, in particular UVB, played an important role in the endogenous inactivation of intact B. ovatus cells. To illustrate, the half-life of B. ovatus ranged from 9 to 13 h in low salinity and 16 to 34 h in high salinity solutions when sunlight irradiation was absent. This meant that if B. ovatus were to be released into either fresh or seawater environment in the evening or night-time, at least 50% of the cells would persist and remain detected by the next morning even though the contamination event may not be recent. On the contrary, if B. ovatus were to be released into fresh or seawater environment in the daytime, the cells would only persist for up to 6 h, and a positive detection of B. ovatus would be indicative of the occurrence of fresh fecal contamination. Our observations, therefore, provide a possible explanation to account for the different test outcomes when monitoring the conventional fecal indicators at different times of the day [39, 40].

At the same fluence value exerted by the solar simulator, B. ovatus cells with intact membrane decayed much faster than the B. ovatus total DNA (Figure 3). The half-life of total DNA is from 40-100 times longer than that of the intact cells. This observation agrees with a previous study by Walters et al. [14], showing that a Bacteroidales human-specific DNA marker decayed slower than cells of the members in the Bacteroidales upon exposure to natural sunlight. This would imply that the currently proposed method by US-EPA to determine fecal pollution, which is based on the qPCR enumeration of the total DNA of the Bacteroidales, might still be limited in its efficacy to determine recent fecal contamination events.

A limitation of this study is that the high decay rates of B. ovatus may be attributed to the constant exposure to irradiance from the solar simulator, and may not represent the diurnal sunlight exposure in the natural environment. To illustrate, an earlier study did not observe a significant impact on the persistence of the species in the Bacteroidales in seawater under diurnal sunlight [4]. In the natural environment, temperature fluctuations, insufficient solar radiation transmittance, as well as possible absence of solar UVB due to the possible overcast weather may occur and prolong the persistence of the members in the Bacteroidales. Regardless, this study yielded similar B. ovatus decay trend at the early stage of the solar irradiation experiment (<6 h) as that obtained from a microcosm exposed to diurnal sunlight [17]. This suggests that the decay kinetic data obtained in this study remains useful in predicting the persistence of intact B. ovatus cell biomass after its release into the water environment. The decay kinetic data would allow subsequent calculation of the original amount of biomass upon release, and therefore assist in using this bacterial species as a general fecal indicator.