Influence of oil, dispersant, and pressure on microbial communities from the Gulf of Mexico

The Deepwater Horizon incident in the Gulf of Mexico in 2010 released an unprecedented amount of petroleum hydrocarbons 1500 meters below the sea surface. Few studies have considered the influence of hydrostatic pressure on bacterial community development and activity during such spills. The goal of this study was to investigate the response of indigenous sediment microbial communities to the combination of increased pressure, hydrocarbons and dispersant. Deep-sea sediment samples collected from the northern Gulf of Mexico were incubated at atmospheric pressure (0.1 MPa) and at elevated pressure (10 MPa), with and without the addition of crude oil and dispersant. After incubations at 4 °C for 7 days, Colwellia and Psychrobium were highly abundant in all samples. Pressure differentially impacted members of the Alteromonadales. The influences of pressure on the composition of bacterial communities were most pronounced when dispersant was added to the incubations. Moritella and Thalassotalea were greatly stimulated by the addition of dispersant, suggesting their roles in dispersant biodegradation. However, Moritella was negatively impacted by increasing pressure. The presence of dispersant was shown to decrease the relative abundance of a known hydrocarbon degrader, Cycloclasticus, while increasing pressure increased its relative abundance. This study highlights the significant influence of pressure on the development of microbial communities in the presence of oil and dispersant during oil spills and related response strategies in the deep sea.

The explosion of the Deepwater Horizon (DWH) oil drilling platform in 2010 led to the release of 700,000 metric tons of crude oil into the Gulf of Mexico at the water depth of 1500 m 1,2 . The subsurface release of oil formed a persistent plume spanning 1000 and 1200 m depth 3,4 and analysis of deep sediment cores collected near the blowout location shortly after the spill indicated that some of this oil was ultimately deposited on the sea floor 5,6 . The deposition of these hydrocarbons were from marine oil snow sedimentation and flocculent accumulation (MOSSFA) events, where crude oil compounds were attached to sinking of particles 7 , and from the direct contact of the deep plume with the continental shelf, referred to as the bathtub-ring hypothesis 6 . Unlike the archaeal community, the bacterial community exhibited a measured response to the massive input of hydrocarbons from the DWH event 8 . Marine oil-degrading bacteria responded with increased abundances in the presence of crude oil. Half-lives of dispersed oil in aerobic marine waters varied from days to months and were influenced by various factors including pre-adaptation of microbial communities to hydrocarbons and the availability of nutrients essential for microbial growth and biodegradation (nitrogen, phosphorous, or iron) 9 . It is estimated that the portion of oil degraded by the bacterial community was as high as 61% 5 . A number of studies investigated the successions of the bacterial community compositions in the water column and in the sediments [10][11][12][13] and found that these successions were driven by the hydrocarbons input from the DWH spill 10,12,14 .
One of the response strategies employed during the spill was the subsea application of dispersant, Corexit EC9500A, which was directly injected at the wellhead during the spill 1 . The effectiveness of dispersant on marine oil biodegradation is a subject of debate 15,16 . One study suggested that dispersant inhibited hydrocarbon-degrading Marinobacter, but stimulated dispersant-degrading Colwellia 16 . A similar study suggested that dispersant enhanced oil biodegradation 15 .
The deep sea is a unique environment, where the hydrostatic pressure increases linearly with depth (1 MPa per 100 m). Previous studies on the influence of pressure on oil biodegradation [17][18][19][20] , showed that pressure as low as 5 MPa impaired growth and activity of hydrocarbon-degrading Alcanivorax borkumensis 18 and the growth of naphthalene degrading Sphingobium yanoikuyae was inhibited at 8.8 MPa 21 . A recent study demonstrated a 4% decrease in n-alkane biodegradation for every 1 MPa increase in hydrostatic pressure 22 . In general, investigations on the impact of dispersant on hydrocarbon biodegradation were performed at ambient pressures, overlooking the effect of pressure 15,[23][24][25][26][27] . More recently, growth of Rhodococcus isolated from deep-sea sediment has been shown to be impaired in the presence of dispersant at 15 MPa 28 . However, biodegradation studies with crude oil and dispersants at increased pressures have not yet been conducted on benthic bacterial communities. In order to understand how each of these factors influenced deep-sea bacterial communities, investigations under controlled ex-situ conditions were conducted. Here we report how the interactions between crude oil, pressure, and dispersant application changed deep benthic microbial communities around the DWH region.

Results
Analysis of microbial diversity. Alpha diversity. The alpha diversity was determined for each treatment and pressure scenario before and after incubation. After 7 days of incubation for all treatments and pressures, there was a decline in alpha diversity (Fig. 1). The control, without addition of oil or dispersant at ambient pressure, lost more than half of the diversity after incubation. Pressure did not affect the alpha diversity of control groups (no oil or dispersant added). The number of observed species, the Shannon diversity index and the Simpson diversity index of the control group were not significantly different at 0.1 MPa and 10 MPa (Fig. 1). The addition of oil and/or dispersant led to decreases in alpha diversities at both 0.1 MPa and 10 MPa. Observed numbers of species in the incubation with dispersant were lower at 10 MPa than at 0.1 MPa. Alpha diversities of the sediment communities treated with oil at 0.1 MPa were significantly different from those at 10 MPa. Observed numbers of species, Simpson indices, and Shannon indices of 10 MPa incubations were lower than those of 0.1 MPa incubations (Fig. 1). Although the observed species of the communities incubated with oil and dispersant at 0.1 MPa and 10 MPa were not different, the presence of oil consistently resulted in statistically significant decreases in alpha diversity when the communities incubated at 0.1 MPa and 10 MPa were compared.
Microbial community structure changes in response to pressure. The phylum Proteobacteria increased in relative abundance in controls and all treatments after the incubation. Planctomycetes was present in all treatments with a maximum contribution of 2.5% of the total number. Bacteroidetes, Acidobacteria and Chloroflexi contributed to less than 0.1% of the population in all treatments after incubation. Gammaproteobacteria was the dominant class of Proteobacteria comprising more than 90% of the community in controls and in all treatments after incubation ( Supplementary Fig. S2). Among the Gammaproteobacteria, Alteromonadales and Oceanospirillales were the    most abundant bacterial orders. Colwelliaceae and Shewanellaceae were the two major families detected in all incubations (Fig. 4). Pressure differentially affected bacterial taxa across taxonomic groups. Sequences related to the genus Moritella (Moritellaceae) responded negatively to pressure in the presence of dispersant. In the control and oil-treated groups, Moritella's relative abundances reached up to 3.5% and 3.9% at 0.1 MPa, and 0.5% and 0.7% at 10 MPa, respectively. In dispersant-treated incubations, sequences related to the genus Moritella were more abundant at 0.1 MPa (19.0-28.9%) than at 10 MPa (3.7-10.2%). In the presence of both oil and dispersant, the genus's relative abundance was also higher at 0.1 MPa (3.0-24.0%) than at 10 MPa (0.1-4.2%) (Fig. 5A).
Thalassotalea did not exhibit a significant response to pressure application with a dispersant-only treatment. The relative abundances were 6.4-11.0% at 0.1 MPa, and 6.2-8.1% at 10 MPa. However, Thalassotalea did respond to an increase in pressure when both oil and dispersant were present, in which the relative abundances were higher at 0.1 MPa (8.0-13.0%) than at 10 MPa (1.3-5.6%) (Fig. 5B), showing a decrease with increasing pressure.
Sequences belonging to the genus Cycloclasticus (Cycloclasticaceae) showed an increase in relative abundance with an increase in pressure to 10 MPa in all treatments. In the control groups, the maximum relative abundance at 10 MPa was three times of that at 0.1 MPa. In the oil-only treatment, the maximum relative abundance at 10 MPa was approximately double that at 0.1 MPa (1.1% at 10 MPa and 0.6% at 0.1 MPa). However, in the dispersant-only treatment, the maximum relative abundances of Cycloclasticus were 0.3% at 10 MPa and 0.07% at 0.1 MPa. However, in the presence of both oil and dispersant, the maximum relative abundance were 0.7% at 10 MPa and 0.17% at 0.1 MPa (Fig. 5C).
Differential abundance analysis. The control incubations at both 0.1 MPa and 10 MPa were dominated by the genera Colwellia (Colwelliaceae) and Psychrobium (Shewanellaceae). In order to identify the amplicon sequence variants (ASVs) that responded solely to pressure, differential abundance analysis was performed. This analysis resulted in 7 ASVs were differentially abundant at 10 MPa; 3 of which belonged to the genus Motiliproteus (Nitrincolaceae, Oceanospirillales). The other 4 ASVs were classified as uncultured Rhodobacteraceae (Rhodobacterales), Psychrobium (Shewanellaceae), Colwellia (Colwelliaceae), and Cycloclasticus (Cycloclasticaceae). Comparisons between the control incubations at 10 MPa and the oil incubations at 10 MPa were conducted in order to identify changes that were due to hydrocarbon addition at high pressure. Five ASVs, which were differentially abundant in incubation with oil at 10 MPa, belonged to the genus Psychrobium. There were 2 Colwellia ASVs and 1 Thalassotalea ASV that increased in abundance due to oil addition at 10 MPa.
The ASVs that were higher in abundance in the treated groups were classified as treatment responders. Subsequently, all of the treatment responders were subjected to Venn diagram analysis (Supplementary Fig. S1). In order to identify the ASVs that responded to oil and dispersant at 0.1 and 10 MPa, the microbial communities of treatment groups were first compared to control groups at 0.1 MPa or 10 MPa to identify those ASVs that did not respond specifically to the treatment. Those ASVs that were upregulated by oil were classified as oil responders. The oil responders dominantly consisted of the genus Psychrobium at both 0.  www.nature.com/scientificreports www.nature.com/scientificreports/ dispersant treatment were categorized as "dispersed oil" responders. They contained mostly sequences related to the order Oceanospirillales including Cobetia, Amphritea and Motiliproteus. However, there was no sequence that was specifically enriched by dispersed oil at 10 MPa (Supplementary Table S1).

Discussion
In this study, microbial communities incubated with dispersant and/or crude oil at two different pressures, 0.1 MPa and 10 MPa, exhibited different responses, suggesting that pressure influenced the succession of microbial communities. The effect of pressure was more conspicuous in the presence of external substrates (dispersant and crude oil) suggesting that hydrostatic pressure is an important parameter that influences how microbial communities respond to an oil spill.
The increase in the relative abundance of Colwellia and Psychrobium in all treatments, including the control, (Figs. 1 and 3) suggested that these genera favored the incubation conditions and were able to thrive on the dissolved organic material that was present in the sample without an additional hydrocarbon source. The members of Alteromonadales, including Colwellia and Psychrobium, have previously been shown to be enriched in unamended long-term incubations at high pressure and low temperature 29 , and an addition of dissolved organic material could increase growth of Colwellia and Psychrobium 30 .
We used differential abundance analysis to detect variants that responded specifically to crude oil or dispersant. The results showed that 6 ASVs belonging to the genus Colwellia were dispersant responders at 0.1 MPa. The role of Colwellia in the pelagic community has been studied extensively, and it has been suggested that Colwellia increased in abundance in response to the spill 10,31 . Subsequent microcosm studies involving the addition of dispersant in seawater showed an increase in the abundance of Colwellia, indicating that Colwellia is a dispersant degrader 16 . Researchers have been able to isolate Colwellia from the water column in the deep sea, and have shown that the organism is capable of oil and dispersant degradation 24,25 . While there have been studies showing www.nature.com/scientificreports www.nature.com/scientificreports/ the dominance of Colwellia in incubations of deep-sea sediments from Faroe-Shetland Channel with oil 26 , much less is known about their role in deep benthic communities than in the water column. The 16S rRNA gene data indicated that the most heavily oiled sediments were enriched in a Colwellia species, which was closely related to a Colwellia clone from the DWH deep-sea plume 32 . Our study suggests that benthic Colwellia plays a role in dispersant degradation.
In this study, ASVs of the genus Psychrobium was enriched at both ambient and higher pressures with the addition of oil, and thus, has been defined as an oil responder. The variant that responded to oil at 10 MPa shared 99.07% of sequence identity to an uncultured bacterial clone found in oil-impacted surficial sediment at approximately 1500 meters depth from the Gulf of Mexico 33 . Members of Shewanellaceae have been shown to be piezophilic 34 and hydrocarbonoclastic 35 . However, the genus Psychrobium has not previously been linked to oil degradation. Detailed physiological studies on the preferred hydrocarbons substrates and the biodegradation performance under pressure, of Psychrobium could further the understanding on the fate of deposited hydrocarbons in the deep biosphere.
The genus Moritella has been identified as a petroleum hydrocarbons degrader in previous studies 26,36,37 and to increase in relative abundance in response to naphthalene in a seawater microcosm study 36 . In studies using deep-sea sediments, the genus Moritella was enriched in the incubations with hydrocarbons mixture containing napththalene and Superdispersant-25 26,37 . In this study, the dispersant-only treatment, using Corexit 9500 A, strongly promoted Moritella. This suggested that Moritella played a role in the degradation of the dispersant components. Interestingly, several Moritella species have been classified as piezophiles, and have optimal growth pressures above 20 MPa 38,39 , but the relative abundance of Moritella was consistently lower at 10 MPa than at 0.1 MPa. This reflects the impact of pressure on dispersant biodegradation and may explain a why the degradation of the dispersant slowed down at plume depth after the oil spill 40 .
In addition to Moritella, Thalassotalea also increased in relative abundance in incubations with dispersant. Thalassotalea has been isolated from deep marine environments and is able to grow in the presence oil 41,42 . The major ASV in this study had 99.77% identity to a sequence of a clone found in oil-impacted sediments at the depth of about 1500 m sampled in September 2010 in the Gulf of Mexico 33 . This study is the first to show that Thalassotalea may participate in dispersant degradation. Intriguingly, pressure did not exert as strong an effect on Thalassotalea as on Moritella. The differential effects of pressure on two prominent dispersant responders suggests a different ecophysiology of the two genera. It has been previously reported that the addition of Corexit 9500 increased the number of heterotrophic bacteria in the marine microbial consortium 43 , but the identities of the heterotrophs were not elucidated. Our study demonstrated that Moritella and Thalassotalea were stimulated by Corexit 9500 A. Thus, Moritella and Thalassotalea could serve as microbial markers of dispersant application in deep-sea sediment communities.
Cycloclasticus is a known obligate aromatic hydrocarbon degrader 44 . The major ASV of Cycloclasticus in this study had 97.89% identity to a clone detected in oil-impacted surficial sediment at 1560 m depth of the gulf of Mexico 33 . Consistently higher relative abundance at 10 MPa of Cycloclasticus across all treatments suggested that this ASV might be a piezophilic hydrocarbon degrader. Furthermore, this ASV was completely absent in the presence of dispersant at 0.1 MPa (Fig. 5C). Our results suggest that the addition of dispersant may inhibit hydrocarbon degraders growth, as it has been reported for Marinobacter 16 . In contrast to the previous study showing that dispersant exacerbated the inhibitory effect of dispersant on hydrocarbon degrading Rhodococcus 28 , pressure attenuated the negative impact of dispersant on piezophilic Cycloclasticus. This emphasized the pivotal role of hydrostatic pressure as an important factor controlling the fate of oil in the deep sea.

Conclusion
It is shown that pressure has a significant influence on microbial communities in deep-sea sediments with the addition of oil and dispersant. Hydrostatic pressure differentially impacted microorganisms across different microbial taxa responding to different substrates. Pressure negatively impacted Moritella, which responded to dispersant. The slow rate of dispersant biodegradation in the deepwater 40 and the persistence of the dispersant in the environment 45 might be explained by the evidence that dispersant degraders are inhibited at high hydrostatic pressure. Therefore, in determining the ultimate fate of the dispersant that is applied in the deep biosphere, the increased pressure and its impact on biodegradation must be considered among the most important environmental factors. Dispersant has been shown to have an inhibitory effect on hydrocarbon degraders 16,28 . The persistence of dispersant due to hydrostatic pressure could further limit oil biodegradation. Therefore, understanding the interplay between high pressure, dispersant, and oil biodegradation, is critical to assess the overall effectiveness and impacts of subsea dispersant application.  Table S2). The sediment and water samples were stored and shipped to Hamburg University of Technology at 4 °C and atmospheric pressure. Equal wet weights of surficial sediments (0-1 cm) from 5 sites were pooled in a sterile container. Bottom water from 3 sites (DWH01, DSH10, and SW01) were filtered through 0.22 µm filter (Corning, USA) and added to the pooled sediments to make a sediment slurry of 50 mg of sediment per ml. Incubations containing only 5 ml of the slurry served as controls. Experimental treatments included the addition of the following added in sterile 10 ml glass vials: (1)  www.nature.com/scientificreports www.nature.com/scientificreports/ comparable to the pressure where the subsea plume deposited hydrocarbons during the Deepwater Horizon spill 6 . Pressure of 10 MPa was achieved by continuously pressurizing with nitrogen gas into the headspace containing atmospheric air in order to keep the aerobic condition as previously described 21,22 . The slurries were incubated at 4 °C with stirring at 200 rpm by means of a magnetic stirring bar for 7 days. DNA extraction and 16S rRNA amplicon sequencing. Total DNA was extracted from 2 ml of sediment slurry with PowerSoil DNA Isolation Kit (QIAGEN, Germany) according to manufacturer's protocol. Paired ended amplicon sequencing of V3 and V4 variable regions of the 16S rRNA gene was performed at LGC Genomics (Germany) facility on the Illumina MiSeq platform using 341 F (CCTACGGGNGGCWGCAG) 46 and 785 R (GACTACHVGGGTATCTAAKCC) 47 primers. The resulting sequences were truncated, quality-filtered, denoised, chimera-filtered and merged with DADA2 48 as implemented in QIIME2 (2018.4 Release) 49 . The taxonomy was assigned to ASVs with the naive-Bayes classifier 50 . The classifier was trained with SILVA132 released database 51 trimmed to V3-V4 region 52,53 . Statistical analysis. The resulting sequences were exported into and further processed in the Phyloseq package 54 in R 55 . The samples were rarefied to even sequence depth across samples to a minimum number where the rarefaction curves were constant (8181 sequences per sample). The sequence abundances were transformed into relative abundances by dividing by total reads. The relative abundances were plotted with ggplot2 56 . The statistical analyses of alpha diversity and beta diversity were calculated using Phyloseq. The differences between the communities was determined using the function adonis in the package Vegan 57 . The permutational multivariate analysis of variance (PERMANOVA) using Bray-Curtis matrix were carried out with 999 permutations. Differential abundant taxa between pressures were analyzed using Calour with nominal discrete false discovery rate of q < 0.1 58 . Each treatment (oil/dispersant/both oil and dispersant) was first compared to the controls at the same pressure to identify ASVs that responded to the treatment. The 6 resulting groups of ASVs were subjected to Venn diagram analysis using InteractiVenn 59 to identify the ASVs that responded to oil, dispersant, or dispersed oil at 0.1 MPa or at 10 MPa.

Data availability
Sequences from this study are available through the European Nucleotide Archive under project PRJEB33386 (ERP116173).