The Quantity of G-Quadruplex in Bacterial Genome affects G-Quadruplex Ligand Sensitivity in Hypertension associated Tongue Coating Microbiota

Background Recently, more and more attention has been paid to the role of oral microbiota in hypertension. It was reported that the disorder of tongue coating microbiota would signicantly increase systolic blood pressure, yet the characteristics of tongue coating microbiota of hypertensives remain unknown. Microbiota is regulated by genes, G-quadruplex (G4) is a secondary structure of nucleic acid, which plays an important role in the regulation of microbial biologic features. there are many G4 ligands in oral administration, but how G4 ligands affect the bacterial biological phenotype needs further exploration. We used metagenomics for analysis of 58 subjects including 23 healthy subjects and 35 hypertensives, and bioinformatics technology for detecting G4 characteristic sequences, nally veried by biological and chemical experiments. We found that Actinomyces decreased signicantly in the hypertension group with the highest average maximum putative G-quadruplex forming sequences (PQS) and GC quantity. We also screened out two species with signicantly different abundance between two groups, Actinomyces odontolyticus and Acinetobacter baumannii. A. odontolyticus had higher GC% and frequency of PQS per 1000bp in the genome, which led to differential inhibition of bacterial growth, metabolism, biolm formation by G4 ligand, sanguinarine. Major Facilitator Superfamily (MFS) was found to be involved in these biological phenomena, we found that sanguinarine could bind and stabilize the G4 structure related to MFS and further inhibited the expression of MFS. G4 sequences bacterial affect G4 ligand sensitivity in hypertension-associated odontolyticus A. G4 ligand bind and the G4 characteristic sequence of the MFS to inhibit the and then inhibit the biolm


Introduction
Hypertension is one of the cardiovascular diseases that seriously threaten people's health. The number of disability-adjusted life years and deaths caused by high systolic blood pressure in China ranks the rst in the world [1,2] , hypertension is also considered to be the most serious risk factor of death in China [3] . It has been suggested that aberrant gut microbiota would contribute to the pathogenesis of hypertension, which emphasized the importance of bacterial homeostasis [4] .
Oral microbiota, as the second-largest microbial community in the human body, also plays an important role in cardiovascular disease, especially hypertension. It has been reported that frequent use of broad-spectrum bactericidal mouthwash may cause the disorder of tongue coating microbiota, and decrease the production of nitric oxide, which results in a signi cant increase in systolic blood pressure [5] . It demonstrated that the homeostasis of tongue coating microbiota is closely related to the regulation of blood pressure. However, the current research on tongue coating microbiota is mostly limited to 16S rRNA technology. Therefore, further study of the characteristics of tongue coating microbiota in patients with hypertension, especially those hypertension-associated species with metabolic functions, is helpful to explore the effect of oral microbiota in hypertension.
Human oral microbiota is an intricate ecosystem, in healthy individuals, oral microbiota maintains oral homeostasis and is usually in dynamic balance, but external stimulus (such as exposure to drugs, host lifestyle) [6,7] and genomic diversity will disturb this balance. It is reported that the genomic sequence and structure determine the expression of genes. G-quadruplex (G4) is a special nucleic acid secondary structure which is formed by several G-tetrads, accumulating evidence indicates that G4 serves as a regulatory role in the biological process [8] , the stabilization of G4 may be linked to genome instability and diseases, such as cancers [9,10] and cardiovascular diseases [11] . For bacteria, the analysis showed that G4 was widespread in bacterial genomes [12] , and was crucial for bacterial DNA replication, transcription [13][14][15] and translation [16] . The formation of G4 inhibits the expression of related genes [17] and further affects the biological phenotype of bacteria [18] . It should be noted that many G4 ligands are applied locally in the oral cavity, for example, oral mouthwash contains sanguinarine [19] , herbal toothpaste contains nitidine, oral local anti-in ammatory drugs contain berberine [20] . Nevertheless, the effect of G4 ligands on different species is still unclear.
Therefore, we utilized metagenomics to explore the characteristics of tongue coating microbiota in patients with hypertension, took bioinformatics analysis to detect G4 sequences of the oral bacterial genome, and veri ed the biological effects of G4 ligand, sanguinarine on hypertension associated strains with different G4 characteristics in the genome.

Materials And Methods
Study subjects 58 subjects were recruited in this study from a cohort of Jixian, Tianjin. They were divided into two groups according to the global hypertension practice guidelines [21] , 23 subjects in the control group with SBP 140 mmHg and DBP 90 mmHg, 35 subjects in the hypertension group with SBP ≥ 140 mmHg or DBP ≥ 90 mmHg, the blood pressure was measured in a sitting posture with medical electronic sphygmomanometer after 5 minutes rest, and the mid-arm was at heart level, it took 3 times with an interval of 1min and used the average data.
This study was approved by the ethical committee of Peking University Third Hospital, and the informed consent were obtained from all subjects before the study. The exclusion criteria of subjects were as follows: A history of drug treatment (such as antihypertensive drugs, lipid-lowering drugs, anticoagulants, etc.); Suffering from other cardiovascular diseases (such as coronary heart disease), diabetes and kidney diseases; Suffering from gastrointestinal diseases and cancers; Suffering from oral diseases (such as dental caries, etc.); A history of antibiotics or mouthwash treatment within three months; Suffering from respiratory infection within one month; Tongue coating sample collection and DNA extraction Tongue coating was collected using oral sterile swabs before breakfast, the tongue was scraped from the root of the tongue to the tip 10 times, the swab was snapped into DNA preservation solution and stored at -80 ℃ in 4h. Lysozyme (10 mg/mL, 100 μL/sample) was pretreated before DNA extraction, and DNA was extracted from each tongue coating sample following the protocol of QIAamp DNA Mini Kit (Qiagen, Germany).

Metagenomic sequencing and Library construction
The DNA libraries were constructed following the Illumina TruSeq DNA Sample Prep v2 Guide (Illumina, Inc., San Diego, CA, USA), and all samples were subjected to 150 bp paired-end sequencing on an Hiseq X-ten platform (Illumina, Inc., San Diego, CA, USA).

Quality control of Illumina reads
Illumina raw reads should be screened as following criteria: Removing reads containing adaptor contamination, more than three ambiguous N bases, low quality (Q<20) bases or less than 60% of highquality bases; Abandoning the host genomes and selecting the bacterial genomes with SOAPaligner (version 2.21) [22] for further analysis.
Microbial relative abundance pro ling Bacterial data were aligned to the NCBI database (National Center for Biological Information, http://www.ncbi.nlm.nih.gov) for detection of known microbiota by SOAPaligner 2.21 [22] , and then classi ed as Kindom, Phylum, Class, Order, Family, Genus, Species to count classi cation and taxonomic relative abundance, the total abundance of each species in a single sample is 1.

Genome assembly, gene prediction and gene catalog construction
The assembly of reads was carried out with a series of k-mer (51, 55, 59, 63) by SOAPdenovo (version 2.04) [23] , and contigs longer than 500bp were kept for further analysis. Software MetaGeneMark [24] was used to predict open reading frames (ORFs) not less than 100bp. CD-HIT (version 4.5.7) [25] was used for pairwise comparison of predicted ORFs which were used for construction of a non-redundant gene catalog set [26] .
Gene functional annotation and functional pro ling BLAST (Version 2.2.28+) was used to align the non-redundant gene catalog set with KEGG (Kyoto Encyclopedia of Genes and Genomes) and eggNOG (evolutionary genealogy of genes) database for the annotation information. And then the relative abundance of all orthologous genes was accumulated to generate the relative abundance of each KO/NOG.

Bacterial strains
This study contained two strains, A. baumannii (JCM 6841) was purchased from China General Microbiological Culture Collection Center (CGMCC) which was grown in LB broth at 37°C, A. odontolyticus (ATCC 17929) was purchased from American Type Culture Collection (ATCC) which was grown in BHI broth at 37°C.

Bioinformatics Analysis
According to the ve common bacteria of tongue coating, including Actinomyces, Streptococcus, Veillonella, Neisseria, and Prevotella, and ten differential strains with high relative abundance respectively enriched in two groups, all complete bacterial genomic DNA sequences was downloaded from eHOMD. If the strain had several complete genomes, the largest genome was selected for G4 analysis. To further analyze the G4 sequence of A. baumannii and A. odontolyticus which were considered as hypertensionassociated bacteria, 244 complete genomes of A. baumannii and 2 genomes of A. odontolyticus were downloaded from National Center for Biotechnology Information.
Putative G4 sequences were predicted by G4RNA screener [27] (http://gitlabs cottgroup.med.usherbrooke.ca/J-Michel/g4rna_screener). The sliding window of 30 nucleotides (nt) moving with steps of 15 nt was used to search potential G4s. The thresholds were as follows: cGcC score >4.5, G4Hunter score >0.9, G4NN score > 0.5. Putative G4 sequences that met the scoring threshold and had gene annotation function remained. Circular dichroism and thermal stability studies CD spectroscopy was recorded on a J-815 CD spectrometer (JASCO, Japan) at 20°C, over 220-330 nm with the scanning speed at 100 nm/min. Each sample was measured three times. DNAs with 5nM nal concentrations were prepared in 30mM Tris-HCl and 0-150 mM KCl. The mixtures were then heated at 95°C for 15 min and gradually cooled to room temperature. For CD thermal stability studies, spectra were recorded with a temperature range from 20 to 95 °C, and the temperature rising rate was 3°C/min.

ESI mass spectrometry (ESI-MS)
The ESI-MS experiments were performed on Bruker SolariX-XR mass spectrometer (Bruker, Billerica, MA, USA) with an ESI source. All samples were tested in a negative ion mode, with a capillary voltage of 2.7 kV, and the samples were injected with the ow rate at 120 μL/h. To test the binding properties, the samples were prepared in the solutions with 25% CH3OH to have better e ciency, the sanguinarine and DNA sequences were at 4:1 ratio.

Bacterial growth assays
To monitor the effect of sanguinarine on A. odontolyticus and A. baumannii, the bacterial suspension containing agents (5, 10, 20 µg/mL sanguinarine) was grown under the anoxic condition at 37℃, the control was free of sanguinarine. At designated time points (0, 2, 4, 6,10,12,14 h), the cell broths were measured the OD with a microplate reader at 570 nm.

Bio lm formation assay
A. odontolyticus (10 6 CFU/mL) was inoculated in culture dishes with transparent bottom to form bio lms, containing sanguinarine (5 µg/mL, 10 µg/mL or 20 µg/mL) at 37℃ and in anaerobic condition. The control was free of sanguinarine. After 24h culture, the dishes were washed two times with PBS. Subsequently, 1.5 μL SYTO 9 dye from LIVE/DEAD BacLight Bacterial Viability Kit were diluted in 1mL PBS and then the diluent was added into each dish. All wells were incubated for 30 min in the dark. After the staining, the wells were washed twice with PBS and each well was added 1mL 4% paraformaldehyde for 15 min for xation. The images were visualized by laser confocal microscope (Carl Zeiss, LSM-780, Germany) at wavelengths of 480 nm excitation and 500 nm emission.
LC-MS/MS for quantitation of TMA Overnight cultured A. odontolyticus (OD570 of 0.15) was diluted into BHI broth (1:100 dilutions) supplemented with 10 μg/mL sanguinarine and 10-4 mM γ-Butyrobetaine at 37°C under anoxic condition for 24h. Samples (20 µL bacterial suspension) were mixed with 80 µL of 10 µM d9-TMA in methanol to precipitate protein. LC-MS/MS analysis was conducted on a Q TRAP5500 mass spectrometer. The following settings were selected: curtain gas, 20 psi; source temperature, 600 °C; gas 1 and gas 2, 35 and 50 psi; spray voltage, 4.5 ESI+ kV; collision activation parameter, medium. Supernatants (10 µl) were analyzed by injection onto a silica column (Luna 5u Silica 100A, 2.0*150 mm) at a ow rate of 0.5 mL/min. The temperature was set at 35°C. Mobile phases A consisted of 1‰Propionic Acid in LC-MS grade water and Mobile phases B consisted of 1‰FA in MeOH. And the Gradient (B %) was 2% for 1 min, 95% for 11 min, 2% 6.5 min, and stop at 7 min. The internal standard d9-TMA was used for quanti cation.
Biolog ECO microplates assay A. odontolyticus were prepared in BHI solid broth for biolog assays. Monoclonal colonies were mixed in IF-A liquid until reaching 90-98% turbidity. Then the IF-A liquid (150 μL) was added to ECO microplates, which were cultured at 37°C in the anaerobic condition. After 48h, the ECO microplates were read at absorption 590 nm using a BIOLOG microplate reader (MOLECULAR DEVICES, United States). And AWCD was used to test the carbon source utilization capability of A. odontolyticus [28] .

Statistics.
R software was used to perform all statistical analyses in metagenomics. Wilcox rank-sum test was used to calculate the signi cance of different taxonomic (phylum, class, order, family, genus, species). Other data were expressed as mean ± SEM. Statistical analysis between two groups used the Student's t-test. Statistical analysis was performed on SPSS 23.0.

Results
The presence and frequency of G4 Sequences in tongue coating microbiota To identify the characteristics of tongue coating microbiota in hypertensive patients, 58 tongue coating samples were collected from the hypertension cohort in Jixian county of Tianjin for metagenomics detection. According to the diagnostic criteria of hypertension [21] , 58 subjects were divided into control group (n=23) and hypertension group (n=35). The systolic and diastolic blood pressure of the hypertension group were signi cantly higher than those of the control group, but there was no signi cant difference in other variables (including age, BMI, fasting blood glucose, blood lipids) ( Table 1).
To explore the difference of microbial composition on phylum, class, order, family and genus levels, 58 tongue coating samples were used for DNA extraction, and all samples were subjected to 150 bp pairedend sequencing on the Illumina platform, average raw data per sample is 14.00 ± 1.90 Gb, average clean data per sample is 7.01 ± 3.43 Gb (Fig. S1, Table S1), all clean data were used for analysis. Figure 1a and Figure S2a-d gave an overview of the relative abundances, Bacteroidetes, Bacteroidea, Bacteroidales, Prevotellaceae, Prevotella showed the highest relative abundance, yet with no signi cant difference between the two groups. The ve most abundant genus were Prevotella, Neisseria, Veillonella, Streptococcus, Actinomyces (Fig. 1a), in both groups, however, only actinomyces have a signi cant change (Fig. 1b). In addition, the relative abundance of Actinomycetales, Actinomycetaceae also showed a signi cant decrease in the hypertension group compared to controls on order and family level (Fig. S2).
We analyzed the quantity of putative G-quadruplex forming sequences (PQS) in 33 oral bacterial species including 8 species of Actinomyces, 3 of Neisseria, 6 of Prevotella, 15 of Streptococcus and 1 of Veillonella. The mean length of the ve species varied from 2.13 Gb to 3.14 Gb, and the mean GC% varied from 38.63% to 68.75%. Using G4 screener to analyze the genomes of ve species, we found that the most abundant species was Actinomyces with an average quantity of 5144 PQS, the minimum was Veillonella with an average quantity of 130 PQS (Fig. 1c, Table S1). We further analyzed the relationship between PQS frequencies per 1000 bp and GC%, we found that high PQS frequencies corresponded with high GC%, in ve most abundant oral bacterial species, Actinomyces had the highest GC content and PQS frequency (Fig. 1d, Table S2).

The characteristics of G4 Sequences of hypertension-associated species
Wilcoxon rank-sum test was used to calculate P values for the differences of tongue-coating microbiota between the control and hypertension groups. We found that 147 species were signi cantly enriched in the control group and 42 species signi cantly enriched in the hypertension group (Table S3). Respectively, twenty of them with higher relative abundance were further shown (Fig. 2a, b). Most importantly, the relative abundance of A. odontolyticus decreased in the hypertension group, which was considered as an effective nitrate-reducing species [29] , and the relative abundance of A. baumannii increased in the hypertension group which produced TMA, then TMA could be oxidized into TMAO which promoted agerelated endothelial dysfunction via oxidative stress [30] . Meanwhile, the heatmap showed that A. odontolyticus was negatively correlated with systolic blood pressure, and A. baumannii was positively correlated with systolic blood pressure (Fig. 2d). Therefore, A. odontolyticus and A. baumannii were considered hypertension-associated bacteria because of their metabolic function.
Furthermore, we analyzed the characteristics of genomic G4 of 20 enriched species, only those with complete genomic sequences from expanded Human Oral Microbiome Database (eHOMD) were analyzed, including A. ondontolyticus, Bi dobacterium longum and Eubacterium sulci of the control group, Porphyromonas gingivicanis and A. baumannii of hypertension group. The results showed that A. ondontolyticus had 2647 PQS, signi cantly higher than A. baumannii with 304 PQS (Fig. 2c). Further, we analyzed 244 complete genomes of A. baumannii and 2 complete genomes of A. odontolyticus from NCBI, similarly, A. odontolyticus had higher GC% and frequency of PQS per 1000bp (Fig. 2e, f, Table S3).

Differential inhibition of bacterial growth and metabolism by G4 ligand sanguinarine
Growth curves were measured to assess the impact of different concentrations of G-quadruplex ligands on the growth of A. ondontolyticus and A. baumannii. As shown in Fig. 3a and c, 5 μg/mL sanguinarine slightly reduced the growth of A. ondontolyticus during the phase 8-16h, and 20 μg/mL sanguinarine inhibited the growth completely, yet 20 μg/mL sanguinarine had no effect on the growth of A. baumannii. The results of nitidine were similar to sanguinarine ( Figure S3a, b). Meanwhile, sanguinarine demonstrated different potency over the same concentration for bacterial metabolism. Sanguinarine dose-dependently inhibited A. ondontolyticus nitrite metabolism (Fig. 3b), but TMA metabolism wasn't affected at the concentration of 10 or 20 μg/mL for A. baumannii (Fig. 3d).
Inhibition of bio lm formation and carbon source metabolism of A. odontolyticus by sanguinarine Except for the inhibition of bacterial growth and nitrate metabolism, sanguinarine was also found to dose-dependently reduce bio lm formation of A. odontolyticus, 10 and 20 μg/mL sanguinarine markedly reduced bio lm densities (Fig. 4a). Based on the metagenomic analysis, Fig. 4b and c showed the microbial functions of tongue coating samples, mainly associated with metabolism, especially carbohydrate and amino acid metabolism. ECO microplate contains 31 different carbon sources, using carbon sources would make a color reaction, so average well color development (AWCD) is used to re ect metabolic functions of microorganisms [28] . At 48h, the AWCD was signi cantly inhibited with 10 μg/mL sanguinarine, which indicated that sanguinarine in uenced the carbon source metabolic capability of A. odontolyticus (Fig. 4d).
The bind and stabilization of G4 structure by sanguinarine inhibits the expression of MFS The MFS family is one of the largest group of secondary membrane transporters and is present in bacteria and mammals [31,32] . MFS transporters contribute to small molecule transport especially uptake of sugars, and are related to bacterial survival, bacterial communication and bio lm formation [33] . The MFS family is important for biological phenotypes of bacteria, we further analyzed the characteristic of G4 sequences in the genome of A. odontolyticus, we found that there were 26 G4 sequences annotated as MFS (Table S5). We carried out two G4 sequences with the highest score (AO-13) and the lowest score (AO-19).
Circular dichroism (CD) spectroscopy was used to determine G-quadruplex con guration, as shown in Fig. 5a, the values at 260nm increased with the increasing concentration of KCl, which indicated that KCl induced the formation of a parallel G-quadruplex, yet the sequence (AO-19) showed KCl independent which formed an unconventional structure (mixed-type conformation) with a positive peak at 275 nm and a negative peak at 240 nm (Fig. 5b). In ESI-MS spectrum, the solution concluded 50 mM NH 4 OAc and sanguinarine (C sequence : C sanguinarine =1:4), the peaks of G-quadruplex complex ions with one, two and three ligands ([S+L] 5-, [S+2L] 5-, [S+3L] 5-) emerged with the [S] 5as the base peak which demonstrated that there were two binding sites (Fig. 5c), as for the sequence of AO-19 (Fig. 5d), it showed four binding sites. In CD melting assay, the Tm value increased from 61.33℃ to 74.34℃ of AO-13 sequence (Fig. 5e) and from 59.51℃ to 72.23℃ of AO-19 sequence (Fig. 5f) with 100 mM KCl. These results proved that sanguinarine could promote the formation and stability of the G-quadruplex of MFS related sequence . Furthermore, the expression of MFS was signi cantly inhibited with 10 μg/mL sanguinarine (Fig. 5g, h).

Discussion
Previous studies have shown that the disorder of tongue coating microbiota caused an increase in blood pressure [34] , but limited studies have reported the correlation in hypertensives. And several external factors have an impact on bacterial relative abundance, such as diet, medication, it should be noted that many G4 ligands are applied locally in the oral cavity, such as sanguinarine, however, it remains unknown about the effect of G4 ligands on hypertension associated tongue coating microbiota.
Here, we collected 58 tongue coating samples which were subsequently detected by metagenomics. All subjects were from the cohort of Tianjin to lower the in uence of diet. Subjects who suffered from cardiovascular diseases, oral diseases, digestive diseases such as gastritis [35] and pancreatitis [36] were excluded, and all of them didn't receive any treatment, including antihypertensive drugs, antibiotics. All the exclusion criteria were set to reduce the impact of external factors. In the present study, we found the ve most abundant tongue coating genus were Prevotella, Neisseria, Veillonella, Streptococcus and Actinomyces. Actinomyces were enriched in the control group which was in contrast to gut microbiota [4] , probably because the environment from the oral to the colon is inconsistent [37] . Meanwhile, we analyzed the number of G-rich sequences by G4 screener, we found that G-rich sequences were common in the genomes of tongue coating genus, and the quantities of G-rich sequences varied greatly among different genus which is consistent with a previous study [12] . Particularly, Actinomyces had the highest GC content and PQS frequency among the ve genus.
According to the metabolic functions, we screened out two hypertension-associated bacteria, A. odontolyticus and A. baumannii. Concomitant with the characteristics of G-rich sequences of Actinomyces, we observed that A. odontolyticus had higher GC% and frequency of PQS per 1000bp than A. baumannii. However, it was still unknown whether the quantity of G-rich sequences would affect the sensitivity of G4 ligands. In this study, sanguinarine was chosen as a typical G4 ligand, 10 µg/mL sanguinarine could signi cantly inhibit the growth and nitrate metabolism of A. odontolyticus, but had no impact on the growth and TMA metabolism on A. baumannii. The results of nitidine were similar, nitidine was regarded as a G4 ligand for promoting the formation of G4 structure in telomere [38] . It has been reported that [39,40] sanguinarine could intercalate DNA and thus inhibit bacterial replication and transcription, but it wasn't enough to explain the differential inhibition. Therefore, the number of genomic G4 may affect the response of bacteria to G4 ligands, but it calls for further veri cation.
To further explore the antibacterial mechanism of sanguinarine, the results demonstrated that sanguinarine inhibited the growth, nitrate metabolism, formation of bio lm and carbon source metabolism of A. odontolyticus. It has been reported that MFS transporters are related to the transport of small carbon source molecules such as sugars and amino acids, as well as bio lm formation [41] and bacterial virulence [42] . Furthermore, chemical experiments were performed to verify that sanguinarine could bind and stabilize the G4 sequence of the MFS gene to inhibit the expression of the MFS gene, thereby inhibiting the biological phenotype of A. odontolyticus. However, the effect of sanguinarine on other G4 sequences still needs to be further veri ed, which is the limitation of this study.

Conclusions
This study indicated the characteristics of genomic G4 sequences of hypertension-associated tongue coating microbiota, the quantities of G4 sequences was signi cantly different among the ve most abundant genus, and G4 ligands sensitivity differed between A.odontolyticus and A.baumannii with signi cantly different genomic G4 quantities. Moreover, sanguinarine, which was identi ed as one of the G4 ligands, could affect the biological phenotype of A.odontolyticus by promoting the formation and stability of G4 in the MFS gene. Our ndings point out that it is necessary to pay more attention to the role of G4 ligands in oral local drug use.