Enrichment of polycyclic aromatic hydrocarbon metabolizing microorganisms on the oral mucosa of tobacco users

Donors

Donors were patients of Dental Clinics at the University of Illinois College of Dentistry or the University of Rochester. All donors provided written informed consent to participate in accordance with guidelines of the Office for the Protection of Research Subjects of the University of Illinois Chicago, with formal approval of the study protocol, 2012-1030, by the Institutional Review Board 1 of the University of Illinois Chicago and RSRB case number 00006443 by the Institutional Review Board at the University of Rochester. This study was done in full accordance with the principles of the Declaration of Helsinki. Adult patients were included as they appeared in the general dentistry clinics. Included were tobacco users who smoked at least 10 cigarettes per day for one year. Nonsmokers had not smoked in the last year. Exclusion criteria: acute periodontitis, active oral infection, usage of antibiotics in the last 30 days, and usage of mouth rinse that day. In the first study, age or gender of the patients was not recorded. In the second study, patients raged in age from 38 to 70 years old, six out of 12 smokers and were female and five out of eight nonsmokers were female. Samples were taken only from sites that appeared normal in appearance in the clinic.

Mucosal sample collection and selection

For the PAH selection procedure, swab samples were collected from the lateral border of the tongue, buccal, attached gingiva, and oral pharynx, with a single swab over 10 to 15 s, placed in PBS, vortexed and centrifuged at 5,000 xg 5′ washed with PBS, then placed in 3 mL Bushnell-Haas Broth (BHB) without glycerol. A 1 mL culture was used to inoculate 9 mL BHB with 10 µg/mL each of benzo[a]pyrene, chrysene, fluoranthrene, naphthalene, phenanthrene, and pyrene. The culture was incubated capped with aeration for 3 weeks at 37 °C in 100 ml bottles. One hundred microliters were collected and plated on BHI agar plate and incubated for 48 h at 37 °C in aerobic conditions. Samples used directly for 16S rRNA gene analysis were from gingiva or tongue and were immediately frozen in Tris-EDTA.

The second version of the PAH selection procedure was done as above with a number of differences. Patients were exclusively from the University of Rochester Dental Clinics, swabbing time was lengthened to 30 s to increase starting number of microbes and was from the lateral border of the tongue. Resuspended material (0.2 ml), after twice washing with phosphate buffered saline (PBS) was inoculated in two mL BHB with and without PAH and incubated at 37 °C for 5 weeks shaking in 15 mL tubes. Samples were purified as below and subjected to 16S rRNA gene analysis with next-generation sequencing, NGS, to identify taxa.

PAH degradation by microbes in liquid culture

C. albicans was grown overnight in Yeast Extract Peptone Dextrose (YPD broth at 30 °C with aeration). The positive control, Mycobacterium rutilum -was prepared in BHI medium and incubated overnight at 30 °C with aeration (Hennessee & Li, 2016). The overnight cultures of C. albicans and Mycobacterium rutilum were centrifuged, washed three times with BHB medium, and resuspended in BHB. Each culture was divided into nine large glass test tubes, each containing 10 mL culture in BHB medium. Pyrene or phenanthrene was added (triplicates) to a final concentration of 20 µg/mL. BHB medium with microbial cells without PAH (triplicates) and BHB medium only served as controls. Tubes were incubated with aeration for 14 days in the dark at 30 °C.

Identification of pyrene and phenanthrene metabolites

After the PAHs were incubated with the microbes, they were stored unopened at −80 °C until analyzed by GC/MS. Once the samples thawed, the cultures were spiked with 20 µg of phenanthrene d-10 or pyrene d-10 prior to extraction to allow estimation of the recovery of the unreacted pyrene and phenanthrene. After spiking with the deuterated standard, the samples were vortexed and stored overnight in a refrigerator. Each of the conical glass vials containing samples were acidified to a pH of 2 and extracted three times with three mL of MTBE (methyl-tert butyl ether, a total of nine mL). The MTBE/aqueous mixtures were centrifuged at 900 g for 10 min to separate the aqueous and MTBE layers. The MTBE extracts were combined and allowed to evaporate overnight in a standard hood and the following morning the remaining MTBE was evaporated under nitrogen. The test tubes containing the residual solid were rinsed (vortexed) with two 400 µl aliquots of MTBE. The MTBE was transferred to an autosampler vial (with a borosilicate glass insert) and evaporated to dryness in a tabletop centrifuge. This concentration and transfer step was repeated twice. Analytical procedures were designed to analyze oxidized (hydroxylated) phenanthrene and pyrene metabolites.

The samples were then derivatized (silylated) with N,O-Bis(trimethylsilyl)- trifluoroacetamide (BSTFA) with 1% trimethylsilyl chloride (TMSC) to convert the hydroxyl groups to trimethyl silyl groups. Extracts in the vials mentioned above were (10 µl BSTFA/TMCS in 200 µl CHCl₃). These samples were diluted 1/20 and 1/50 in chloroform and analyzed by GC/MS for the purpose of quantifying the unreacted PAHs. The most concentrated, undiluted samples were run as well for the purpose of identifying metabolites.

Derivatized extracts were analyzed using an Agilent 5977 Mass selective detector interfaced to an Agilent 7800 GC. A 30-meter Agilent DB5-MS column was used for the separation (Agilent, Santa Clara, CA, USA). The mobile phase was helium at a flow rate of one mL/min. The injector temperature was 240 °C. A 20-min GC gradient was used for the separation with a helium flow rate of one mL/min. The initial temperature was held at 60 °C for two minutes and then increased to 320 °C at a rate of 20°/minute and held at 320 °C for 10 min. Full scans were acquired for all analyses.

DNA extraction

Genomic DNA was extracted from swabs using the MasterPure Gram Positive DNA Purification Kit (Epicentre, Madison, WI, USA) according to manufacturer’s instructions providing both gram-positive and -negative bacteria DNA. As instructed, lysozyme was used to digest cell walls and proteinase K was used to reduce protein levels, but no ribonuclease digestion was performed.

Samples harvested from in vitro incubations, at least two colonies on BHI agar plates, were collected and sample DNA was extracted using alkaline lysis and Express Matrix purification (MP Biomedical, Santa Ana, CA, USA). Identification of clones was done after PCR amplification using consensus 16S rRNA primers.

16S-27F 5′-AGAGTTTGATCCTGGCTCAG-3′ and16S-1329R 5′-TCTACGCATTCCA CCGCTAC-3′ or, in the case of C. albicans, 28S rRNA fungal specific primers NL-1 5′-GCATATCAATAAGCGGAGGAAAAG-3′ and NL-4 5′-GGTCCGTGTTTCAAGACGG-3′ used for Sanger Sequencing (Kurtzman & Robnett, 1997). For samples harvested from in vitro incubations for the second PAH culture selection protocol, bacteria were harvested directly from culture media after centrifugation and resuspension using mechanical lysis and the Zymo Quick-DNA Fungal/Bacterial Kits as described (Zymo Research, Irvine, CA, USA) (Adami, Ang & Kim, 2021).

Characterization of microbial community structure

Microbial community structure was characterized using high-throughput sequencing of PCR amplicons generated from the V1-V3 variable regions of bacterial 16S rRNA genes. Briefly, the widely used primer sets 27F/534R, targeting the V1-V3 variable region of the 16S rRNA gene of bacteria, were used for amplification as done earlier (Adami, Ang & Kim, 2021; Yildirim et al., 2010). Alternatively, the V4 variable region of the 16S rRNA gene of bacteria was amplified and sequenced using the 515F/806R primer set (Caporaso et al., 2011). A two-stage PCR or targeted amplicon sequencing (TAS) approach was performed to generate amplicon libraries, as described previously (Bybee et al., 2011; Green, Venkatramanan & Naqib, 2015).

Sequencing was performed on an Illumina MiSeq sequencer using standard V3 chemistry with paired-end, 300 base reads or on an Illumina MiniSeq sequencer with high output reagent and 150 cycles. Fluidigm sequencing primers, targeting the CS1 and CS2 linker regions, were used to initiate sequencing. Demultiplexing of reads was performed on instrument. Library preparation was performed at the DNA Services Facility at the University of Illinois, Chicago.

Bioinformatics analysis

Raw paired-end FASTQ files were merged using the Paired-End reAd merger (PEAR) algorithm (Zhang et al., 2014). Merged data were then quality trimmed (Q20), and sequences shorter than 450 bases were removed. The remaining sequences were exported as FASTA and processed through the software package QIIME (v1.8.0) (Caporaso et al., 2010). Sequences were screened for chimeras using the USEARCH61 algorithm and putative chimeric sequences were removed from the data set (Edgar, 2010). Chimera-free samples were then pooled, and clustered into operational taxonomic units (OTU) at 97% similarity using USEARCH assigned to taxonomic levels from phylum to species.

Statistical analysis

BIOMs were used to identify taxa that were significantly differentially abundant between a priori defined groups. Differences in microbiota taxonomic abundance between the groups were tested using Welch’s t-test using the software package STAMP (Parks et al., 2014). Significance was set at P < 0.05. to examine. To compare profiles of in vitro PAH selected taxa from tongue swabs, versus that from selection without PAH, both DeSeq2 and EdgeR were used after elimination of taxa with less than 100 sequence reads in a sample.

G*Power was used for sample size calculation and effect sizes were determined as described (Chinn, 2000; Faul et al., 2007).

Survival of sampled oral mucosal bacteria after long-term PAH exposure

Studies of environmental and skin microorganisms have revealed bacteria that survive in high levels of PAHs can use these chemicals as a nutrient source (Husain, 2008; Sowada et al., 2017). We took advantage of that knowledge to investigate behavioral differences of the oral bacteria associated with long-term exposure to tobacco smoke PAHs versus minimal exposure. First, we took swab samples from the oral mucosa of six nonsmokers and 14 smokers. The washed microorganisms were than inoculated into minimal broth and allowed to incubate for 3 weeks in the presence of a cocktail of PAHs as the only carbon source. At the end of the incubation time, the cells were plated on rich medium agar plates (BHI) for 48 h to allow the detection of surviving bacteria. With 20 subjects at a power of 80% and significance level of 5% one would be able to detect a difference presuming 70% of smokers have taxa that survive the PAH assay and 10% of nonsmokers do (Faul et al., 2007).

We found that PAH selection indeed allowed the survival of several bacteria that were identified by 16S rRNA gene Sanger sequencing. For smokers’ samples, eleven out of fourteen produced colonies under PAH selection; these were identified as Acinetobacter junii, Acinetobacter baumannii, Agrobactrium tumerfaciens, Actinomyces, Bacillus pumillus, Bacillus subtilis, Kocuria rhizophila, Rhodococcus, Staphylococcus epidermidis, and Staphylococcus. In addition, multiple isolates from smokers revealed the presence of the yeast C. albicans, which was identified by 28S rRNA gene sequencing. In nonsmoker samples, one out of six grew colonies with PAH selection, which was identified as Rhodococcus. Smokers showed higher numbers of these microbes that could survive these harsh selective conditions where PAH was the only carbon source (Fisher’s Exact Test, p < 0.018). The odds ratio for being positive in the PAH selection test if the subject smoked versus did not smoke was 18.3 with a Cohen’s effect size of 1.61 (Chinn, 2000).

Differences in oral mucosal bacteria with long-term tobacco usage

Having identified oral bacteria on the oral mucosa of smokers as resistant to PAH exposure, the question remained whether these bacteria comprised a significant level of the mucosal microbiome and if they were indeed elevated on oral mucosal surfaces in smokers. Mucosal surface samples were taken from two sites, the gingiva and lateral border of the tongue. These sites were chosen because they are common sites of tobacco-associated OSCC. A total of 37 donors, 15 smokers (average age 53.6, nine male, six female) and 22 nonsmokers (average age 42.1, 12 male, 10 female) were selected. Of the eight genera identified in the earlier assay only Staphylococcus and Actinomyces were present at higher than 0.01% of oral bacteria genera at the two mucosal sites (Tables 1 and 2). Notably both taxa are found at higher levels on smoker versus nonsmoker gingiva, and trend that way on the tongue (Tables 1 and 2). Identification at the genus level does not assure identification at the species or strain level.

PAH metabolites produced by oral microorganism in vitro

Of the candidate microorganisms that showed ability to survive PAH exposure in the in vitro test, the yeast C. albicans was the most commonly identified in the assay and is well known to be at appreciable level in the oral mucosa of a subset of both smokers and nonsmokers (Darwazeh, Al-Dwairi & Al-Zwairi, 2010; Sheth et al., 2016). Two separate isolates of C. albicans obtained from two different tobacco users were incubated with PAHs as sole carbon source. In this experiment the PAHs tested were phenanthrene and pyrene. These chemicals were chosen because they can be found at relatively high levels in the oral cavity in tobacco users (Darwazeh, Al-Dwairi & Al-Zwairi, 2010; Sheth et al., 2016) and they are readily digested by many environmental microbes (Ghosal et al., 2016; Hadibarata et al., 2017; Hennessee & Li, 2016; Hesham et al., 2009; Husain, 2008; MacGillivray & Shiaris, 1993; Mallick, Chakraborty & Dutta, 2011; Mbachu, Chukwura & Mbachu, 2016; Sowada et al., 2014; Sutherland, 1992). A soil-derived strain, Mycobacterium rutilum, capable of metabolizing pyrene or phenanthrene as the sole carbon source, was used as a positive control. Comparative metabolism studies clearly showed that C. albicans does not oxidize PAHs to a significant extent while mycobacteria readily do so (Fig. 1). The data indicates that the metabolism of the phenanthrene in the mycobacterium is nearly complete. Only 1.3 ± 1.6% of the phenanthrene remains after incubation. Pyrene was more difficult for the mycobacterium to digest. Here 29 ± 6% of the starting concentration remained after two weeks of incubation. The recoveries of the PAHs from the C. albicans (76 ± 10% and 91 ± 3%) are like those from the un-incubated matrices (83 ± 7% and 84 ± 11%). The data suggests that C. albicans does not degrade PAH at detectable levels when compared to the mycobacteria.

Characterization of metabolic products: The most abundant potential phenanthrene and pyrene metabolites formed in C. albicans and Mycobacterium rutilum incubations were assumed to have hydroxyl groups (Ghosal et al., 2016; Mallick, Chakraborty & Dutta, 2011). Therefore, extracts were silylated to enhance the metabolite response in the gas chromatography/mass spectrometry (GC/MS) analysis.

Phenanthrene

The major metabolites found in the analysis of the positive control Mycobacterium rutilum incubation of phenanthrene were 9,10-dihydro-9,10-phenanthrenediol, 9,10-phenanthrenedione, and dihydroxyphenanthrene. A mass spectrum of the trimethylsilyl derivative of the 9,10-dihydrophenanthrene diol generated by the Mycobacterium is shown in Fig. 2. Analysis of chromatograms derived from extracts from oral yeast C. albicans and phenanthrene co-incubation failed to detect similar, or any, metabolites.

The molecule ion is m/z 356. Characteristic fragment ions include a loss of CH₃(m/z 341) and an ion at m/z 117 ((CH₃)₃SiCH₂CH₃⁺). We also observed a compound with a fragmentation pattern similar to the diol in Fig. 2 eluting at 9.95 min with a molecular mass two daltons less. The mass spectra suggested that this compound was the 9,10-dihydroxypyrene (Fig. 3).

The molecule ion is observed at m/z 354 and the ions at m/z 339 and m/z 117 were formed in the same way as in the spectra of the derivatized diol in Fig. 2. The 9,10-dihroxypyrene was detected in the study of mycobacteria metabolism by GC/MS as a dimethyl derivative (m/z 238) (Hennessee & Li, 2016). Figure 3 mass spectra also shows characteristic ions at m/z 264 formed by the loss of (CH₃)₃SiOH as well. In human hepatic liver (Huang et al., 2014; Matsunaga et al., 2009) and yeast (Rodriguez, Sobol & Schiestl, 2008) cells, 9,10-dihydroxyphenanthrene is in a redox equilibrium with 9,10-phenanthrenedione (Fig. 4). Under aerobic conditions, formation of the dione is favored (Huang et al., 2014). Therefore, if we find the 9,10-dihydroxyphenanthrene we should observe the dione as well. The molecular mass of the 9,10-phenanthrene dione is 208 daltons. The m/z 208 ion was observed in the mycobacterium sample (Fig. 4A) but not in the C. albicans (Fig. 4B) shown below. This observation provides more evidence that C. albicans was not oxidizing PAHs.

Pyrene

Shown are spectra from products produced by positive control Mycobacterium digestion of pyrene (Fig. 5). No similar products, nor any other products, were detectable after pyrene incubation with C. albicans. In contrast, pyrene formed both mono- and dihydroxy metabolites in the mycobacterium incubations. Two silylated dihydroxy pyrene metabolites with molecule ions at m/z 380 were observed with retention times of 10.29 and 10.50 min, respectively (Fig. 5). The mass spectra of the two compounds were similar (Fig. S1). These two compounds were most likely the silylated trans (Rt 10.29) and cis (Rt 10.50) 4,5-dihydroxy-4,5-dihydropyrene. Heitkamp et al. (1988) used HPLC, mass spectrometry, and nuclear magnetic resonance (NMR) to determine the structures of the two most abundant Mycobacterium pyrene degradation products. The trans isomer was found to be more abundant and have a shorter retention time than the cis isomer.

Two mono-hydroxypyrene metabolites were detected in the mycobacterium incubation as well. The most abundant was determined to be 1-hydroxypyrene (Fig. S2A) through co-chromatography with a standard compound. 1-hydroxypyrene is a human urinary metabolite of pyrene that is frequently monitored to assess PAH exposure in humans (Flores-Ramirez et al., 2021; Jain, 2021). The most abundant peak in the Fig. S2 mass spectra was the molecule ion at m/z 290. We also observed a compound eluting at Rt 10.84, whose mass spectra was similar to that of the silylated 1-hydroxypyrene (Fig. S2). There are three possible mono-hydroxy isomers. To our knowledge, the 2- and 4-hydroxy isomers have not been observed as metabolites in any previous study of pyrene metabolism. The abundance of the mono hydroxy isomer in Fig. S2 is about 1% of the 1-hydroxypyrene shown in Fig. S2.

Revised assay to screen for PAH metabolizers

Because there was no evidence that C. albicans was capable of digesting pyrene or phenanthrene, we repeated the selection of mucosal oral microbes for growth on PAH using a revised approach to better control for microbes that survive without metabolizing PAH. Incubation after inoculation of a mucosal swab sample into liquid culture was done for 5 weeks with PAH as sole added carbon source and then the bacterial DNA isolated directly without further subculturing on agar plates. All incubations, and not just a subset, were done side by side against a similar assay without the PAH carbon source as a negative control. Donors who were smokers were on average 55.0 years old, and six out of 12 smokers were female. For nonsmokers the average age was 54.2 and five out of eight were female.

Under these conditions, surprisingly, some tongue bacteria had the ability to survive at relatively high numbers over the 5 weeks despite lack of added nutrients. The 16S rRNA amplicon NGS sequencing of the PAH containing and control cultures lacking PAH allowed identification of the taxa that dominate the cultures. f_Enterobacteriaceae was identified at high levels in many cultures whether PAH was present or not (Table S1). Importantly, some taxa were observed at high numbers (16 to 95% of the reads) only in the presence of PAH and were not detected in the matched culture without PAH nor in any control culture without PAH (Fig. 6). Six out of 12 tongue biofilm samples fulfilled these criteria of having at least one taxon that showed specific PAH growth. Micrococcus luteus was present in samples from three donors, g_Kingella in three donors, and g_Microbacterium in one donor, all tobacco smokers. Of these, M. Luteus and Kingella enrichment was statistically significant based on DeSeq2 and/or EdgeR analysis at FDR < 0.1 with Benjamini–Hochberg correction for multiple testing. One nontobacco user out of eight tested showed a single taxon, Breviabacterium paucivorns, as enriched after PAH selection and not in the negative control, without PAH. The Fisher Exact test revealed the differences in the two groups does not reach statistical significance at p < 0.106 though it trends toward that with an odds ratio of 14 for being positive in the PAH selection test if one smokes versus does not smoke. If one were to pool the data from the earlier experiment selecting for PAH surviving microbes with these results, and compare smokers and non-smokers, N = 40, there is a difference in the smokers at p < 0.0005 , with an odds ratio of 11.4 and an effect size of 1.34.

The dependence on PAH for growth is less clear with Haemophilus parainfluenzae and Aggregatibacter. These were enriched specifically in the presence of PAH in two and three, respectively, out of the 10 tobacco user samples but were found in at least one culture grown without PAH from another donor.