In Vitro Multi-Species Oral Biofilms Grown in Presence of H2O2 Production-Affecting Substrates Show Health-Associated Alterations in Composition, Metabolism and Virulence.


 Modulation of the commensal oral microbiota is a promising preventive or therapeutic strategy for oral health and can for instance be achieved by increasing the abundance and/or activity of certain species. This study evaluated whether 10 selected substrates could modulate in vitro multi-species oral biofilms towards a more health-associated state. These substrates were chosen based on the possibility that they could stimulate H2O2 production by certain commensal species and/or increase their abundance, as previously reported or as hypothesized based on known bacterial H2O2 pathways. Biofilms grown in presence of the substrates at a clinically relevant concentration of 1%(w/v) often showed increased abundances of commensal species and decreased abundances of periodontal pathogens. Furthermore, most biofilms also showed an altered metabolic profile. Effects on the expression of a selection of virulence genes were substrate-dependent, but often a decreased expression of certain genes could be observed. In conclusion, this study found that a selection of substrates chosen for their hypothesized beneficial effects on the commensal oral microbiota were able to modulate in vitro multi-species oral biofilms towards a more health-associated state. These modulatory effects were found to be substrate-dependent.


ABSTRACT 25
Modulation of the commensal oral microbiota is a promising preventive or therapeutic strategy 26 for oral health and can for instance be achieved by increasing the abundance and/or activity 27 of certain species. This study evaluated whether 10 selected substrates could modulate in vitro 28 multi-species oral biofilms towards a more health-associated state. These substrates were 29 chosen based on the possibility that they could stimulate H2O2 production by certain 30 commensal species and/or increase their abundance, as previously reported or as 31 hypothesized based on known bacterial H2O2 pathways. Biofilms grown in presence of the 32 substrates at a clinically relevant concentration of 1%(w/v) often showed increased abundances 33 of commensal species and decreased abundances of periodontal pathogens. Furthermore, 34 most biofilms also showed an altered metabolic profile. Effects on the expression of a selection 35 of virulence genes were substrate-dependent, but often a decreased expression of certain 36 genes could be observed. In conclusion, this study found that a selection of substrates chosen 37 for their hypothesized beneficial effects on the commensal oral microbiota were able to 38 modulate in vitro multi-species oral biofilms towards a more health-associated state. These 39 modulatory effects were found to be substrate-dependent. 40

INTRODUCTION 41
The health status of the oral cavity is determined by a variety of external factors such as oral 42 hygiene, diet and lifestyle 1,2 . Simultaneously, also the intrinsic characteristics of an individual 43 such as age, genetic predisposition and systemic diseases are known to strongly influence 44 one's predisposition to develop oral illnesses such as periodontal diseases [3][4][5] . Eventually, the 45 oral health status will be determined by the presence or absence of a complex, calibrated 46 interplay between the host, its environment and its commensal oral microbiota 6 . Knowledge on 47 the latter has increased exponentially over the past decades, leading to the understanding that 48 despite the existence of inter-and intra-individual variability, each person possesses a so-49 called 'core oral microbiome' that plays a crucial role in maintaining the homeostatic, symbiotic 50 relationship between the oral microbiota and its host 7-9 . 51 In oral health, an individual's oral microbiota mainly consists of a few hundred bacterial species 52 that often organize into robust, highly specialized oral biofilms (dental plaque) 6,8,9 . Within these 53 biofilms, an optimal microenvironment allows oral bacteria to flourish while they are sheltered 54 from external aggressors and stress but where they are also in close contact with each 55 other 10,11 . As a result, complex inter-and intra-species interactions occur that help shape the 56 homeostatic balances. These interactions are diverse and mediated through for instance 57 metabolic cross-feeding and the production of substances (e.g. hydrogen peroxide (H2O2)) that 58 (in)directly affect the function and survival of nearby species [12][13][14] . In this way, the commensal 59 oral microbiota can deal with potentially disease-provoking disruptions while maintaining its 60 core composition and functions 15 . However, once these disruptions surpass a certain threshold 61 and/or when the function of the commensal oral microbiota is impaired, an imbalanced 62 relationship between host and microbiota or within the microbiota eventually leads to dysbiosis 63 and the onset and progression of oral diseases 6,15,16 . 64 Up until today, prevention and treatment of oral diseases are mainly accomplished by 65 mechanical strategies (i.e. plaque removal) in combination with adjunctive antimicrobial 66 therapy (i.e. antibiotics or antiseptics) 17-21 . However, these approaches also come with certain 67 disadvantages, such as the often aspecific removal and killing of both pathogenic and 68 24.2±7.2%, respectively. Saccharin also increased cariogenic pathogens abundance 134 (4.2±2.3%). For the other set of substrates, the control biofilms consisted of 31.7±9.3% 135 commensal species, 68.3±9.3% periopathogens and 0.1±0.0 cariogenic pathogens (Table 1) F. nucleatum and P. gingivalis) was made of which the expression profiles were analysed to 164 determine the effects of each substrate on multi-species biofilm virulence (Table 3). 165 Noteworthy is that significant changes in virulence gene expression relative to the control 166 condition were only considered to be biologically relevant when there was >1.5-fold 167 upregulation or >2-fold downregulation and that only such changes were considered. 168 Altogether, as can be seen based on the color scale used in Table 3, more substrate-gene 169 combinations showed at least a tendency towards decreased virulence gene expression than 170 combinations showing at least a tendency towards increased virulence gene expression. For 171 D-sorbitol, lactic acid, sodium fumarate, D-(-)-arabinose, 6/10 genes showed a tendency 172 towards downregulated expression, for sodium succinate and sodium acetate this were 5/10 173 genes, for saccharin, sodium L-lactate and sodium pyruvate 4/10 genes and for potassium 174 acetate 3/10 genes. 175 For five substrates, A. actinomycetemcomitans virulence gene expression was found to be 176 downregulated 3.3-to 100-fold (Table 3). D-(+)-sorbitol downregulated apaH, cagE and orf859 177 expression (3.6-, 100-and 5.9-fold). Lactic acid, sodium fumarate, D-(-)-arabinose and sodium 178 acetate downregulated orf859 expression with 3.3-, 4.0-, 5.9-and 3.4-fold. On the other hand, 179 pgA expression was upregulated for D-sorbitol, saccharin and sodium pyruvate (3.6-, 3.8-and 180 3.2-fold). For F. nucleatum, hemin receptor gene expression was downregulated 2.3-to 9.1-181 fold for D-sorbitol, lactic acid, potassium acetate and sodium acetate ( Research on novel preventive and therapeutic interventions for oral health is rapidly 194 evolving, with one of the focuses lying on the modulation of the commensal oral microbiota as 195 a 'pro-microbial' approach. Such modulation can for instance be achieved by increasing the 196 abundance and/or activity of certain species, eventually resulting in a more balanced oral 197 microbiota. This study evaluated whether 10 selected substrates could modulate in vitro multi-198 species oral biofilms towards a more health-associated microbiological composition, an altered 199 metabolic activity and a decreased virulence gene expression profile. The selection of the 200 evaluated substrates was based on the possibility that they could stimulate H2O2 production 201 by certain commensal species and/or increase their abundance, which has been described in 202 literature or was hypothesized based on known bacterial H2O2 pathways. Biofilm growth in 203 presence of the substrates at a clinically relevant concentration of 1%(w/v) often resulted in a 204 microbiological composition with increased abundances of commensal species and decreased 205 abundances of periodontal pathogens. Furthermore, most substrate conditions also altered the 206 metabolic profiles of these biofilms. The effects on virulence gene expression, based on a 207 selection of 10 important virulence genes of 3 periodontal pathogens, were highly substrate-208 dependent, but for several substrates a decreased expression of certain genes could be 209 observed. Altogether, this study provides novel findings on oral biofilm modulation by 10 210 substrates selected for their possible effects on the activity and/or abundance of certain 211 commensal oral bacteria. To our knowledge, this work is the first one to simultaneously 212 investigate the modulatory effects of these specific substrates on the microbiological 213 composition, metabolic and virulence profiles of complex, in vitro multi-species oral biofilms. 214 The substrates included in this study were selected based on previous findings in 215 literature and/or their involvement in bacterial H2O2 production pathways. Arabinose, sorbitol 216 and saccharin have been shown to stimulate H2O2 production by S. oralis and S. sanguinis 36 . 217 Pyruvate and lactate are two substrates with a central role in the two major oral streptococcal 218 H2O2 production pathways 14,37,38 . Furthermore, it is known that lactic acid might serve as a 219 substrate for certain streptococcal species resulting in the production of H2O2 38,39 . Potassium 220 acetate was shown to slightly affect H2O2 production by S. gordonii 40 , which led to the 221 hypothesis this could also be the case for sodium acetate. Finally, the conversion of fumarate 222 to succinate is known to yield H2O2 as a by-product, and succinate can in turn be re-converted 223 to fumarate 41 . The selected substrates thus have a clear link with H2O2-producing commensal 224 oral species, but the current study merely focused on the effects of these substrates on multi-225 species oral biofilms rather than on the potential underlying mechanisms of these effects. The 226 rationale for this is that accurate determination of H2O2 production within complex oral biofilms 227 has not been achieved yet. Due to diffusion restrictions, the effects of H2O2 can be very 228 localized and take mainly place within the biofilm 42,43 . Furthermore, determining the effects of 229 the substrates on certain aspects of a complex multi-species community could be considered 230 to be more relevant than merely investigating their mode of action in a simpler setting. Since with a compositional effect, changes in abundances were generally observed for one or more 248 of the above-mentioned species. For instance, sodium lactate, sodium pyruvate and potassium 249 acetate all resulted in increased S. oralis numbers while simultaneously also a decrease in one 250 or two periodontal pathogens like P. gingivalis, P. intermedia and F. nucleatum was observed. 251 On the other hand, sorbitol was found to increase S. mitis and S. sanguinis numbers while also 252 decreasing A. actinomycetemcomitans and P. gingivalis numbers. The effects of other 253 substrates like saccharin or lactic acid were generally limited to decreases in periopathogens. 254 However, this does not automatically imply that such substrates have no effects on the activity 255 of commensal species, as saccharin and lactic acid have been shown to increase H2O2 256 production by certain oral streptococci 36,38,39 . Altogether, the majority of the substrates tested 257 in this study shifted the biofilm composition towards a more health-associated one. It can be 258 hypothesized that, besides increasing the abundance of certain commensals, this could also 259 be mediated by stimulating the activity of these species. 260 Insights into the metabolic profile of oral communities can provide valuable information 261 on the role they play in oral health or disease. In periodontal disease, inflammophilic species 262 characterized by asaccharolytic and proteolytic metabolisms are enriched in abundance and 263 show increased activity 15,50,51 . This eventually provides for a reciprocally reinforced feedback 264 loop between inflammation and dysbiosis, allowing such species to thrive and which acts as 265 an important disease driver 15 . Species like Porphyromonas, Prevotella and Fusobacterium are 266 characterized by such metabolic profiles through which peptides and amino acids are 267 converted into organic acids like formate, acetate, propionate and butyrate 50,52 . In this study, it 268 was remarkable that most of the substrate conditions showed decreases in butyrate 269 production. Although butyrate is known to play a protective role in the gut, butyrate production 270 in the oral cavity is known to be associated with periodontal inflammation 53-55 . Therefore, the 271 observed decreased butyrate levels can be considered as a favourable metabolic change. 272 Similar findings on decreased butyrate levels were previously reported in an in vitro study 273 identifying potential prebiotic substrates for oral health 46 . However, commensal species like 274 Actinomyces and Streptococcus have a saccharolytic metabolism, leading to the production of 275 lactate, acetate and formate 52 . Given that several substrate conditions showed increases in 276 one or two streptococcal species, one would expect to observe an increase in lactate levels, 277 although this was not the case. This can be explained by the complexity of multi-species 278 biofilms, which are characterized by a wide variety of interspecies interactions and metabolic 279 cross-feeding 11,52,56-58 . Lactate produced by streptococci forms a nutritional source for 280 Actinomyces and Veillonella species, which results in the production of formate, acetate and 281 propionate (Veillonella spp.) or acetate (Actinomyces spp.) 52,56,57 . Formate has been shown to 282 have an inverse relationship with the severity of periodontal disease 53 , and in the current study, 283 it was increased in the sodium pyruvate condition. On the other hand, in some studies, it has 284 also been associated with undesired effects on oral epithelial cells in vitro, which is also the 285 case for acetate and propionate 59 . However, given the entanglement of metabolic pathways 286 within complex multi-species oral biofilms, it is difficult to fully interpret the impact of all 287 metabolic shifts observed in this study. biofilms were also previously observed 46 , and this could be explained as a response to external 302 stress, something reported for other Aggregatibacter species 68 . For F. nucleatum, 303 downregulation was often observed for the gene encoding a hemin receptor, which is highly 304 immunogenic and plays an important role in hemin uptake 65 . Apart from sorbitol and lactic acid, 305 most conditions showed upregulated ABC transporter permease gene expression. As its gene 306 product is involved in membrane transport 65 , this could also be a response to the induced 307 environmental changes. For P. gingivalis the effects were also diverse. Most substrates led to 308 decreased rgpA expression, a gingipain gene encoding an arginine-specific cysteine protease 309 involved in several processes such as disturbance of host defense systems and tissue 310 degradation 67 . FimA and kgp expression, encoding a fimbrilin involved in attachment to oral 311 surfaces and a gingipain gene encoding a lysine-specific cysteine protease, respectively 67 , 312 were sometimes downregulated and sometimes upregulated, depending on the substrate. 313 Altogether, the effects of the substrates on the virulence profiles of the biofilms were 314 found to be highly diverse. Nevertheless, it is important to look at the overall effect of the 315 substrates, since oral diseases are caused by the concerted virulence, (metabolic) function 316 and composition of synergistic polymicrobial biofilms 6,16 . From that point of view, most 317 substrates had beneficial modulatory effects on at least one, and often two or all three of these 318 aspects. To conclude, future research should look into some of the limitations and aspects that 319 were not addressed in the current study. For instance, a broader selection of virulence genes 320 could provide further insight into changes in virulence, and also evaluating the effects on the 321 inflammatory potential of the biofilms towards oral cells could be of interest. Furthermore, now 322 the effects of the substrates on a complex multi-species biofilm have been established, the 323 underlying mechanisms of these effects should be investigated. Given the rationale for the 324 selection of the substrates, this should first focus on the influence they might have on 325 streptococcal H2O2 production. In conclusion, this study found that a selection of substrates 326 chosen for their hypothesized beneficial effects on the abundance and/or activity of commensal 327 oral bacteria were able to modulate in vitro multi-species oral biofilms towards a more health-328 associated state. More specifically, biofilms grown in presence of the substrates at a clinically 329 relevant concentration often showed a beneficial shift in microbiological composition, an 330 altered metabolic profile and sometimes a decreased virulence, the latter of which was highly 331 The substrates used in this study were selected based on the following two criteria: (1) shown 357 in literature to (possibly) stimulate H2O2 production by a limited number of oral bacterial 358 species; and/or (2) (in)direct involvement in known pathways of oral bacterial H2O2 production. 359 All substrates were dissolved in BHI-2 without mucin at a concentration of 2%(w/v), followed by 360 pH adjustment to 7.4 and filter sterilization. For the biofilm experiments, one volume of this 361 was supplemented with one volume of sterile BHI-2 with double-concentrated mucin (2 x), 362 yielding sterile BHI-2 solutions (with 1 x mucin) with a final substrate concentration of 1%(w/v). 363 Following substrates were selected for this study: D-(-)-arabinose, lactic acid, potassium 364 acetate, saccharin, sodium fumarate, sodium L-lactate, sodium pyruvate, sodium succinate (all 365 Sigma-Aldrich Co, St. Louis, USA), sodium acetate and D-sorbitol (both VWR, Radnor, USA). 366

Multi-species biofilm formation assays, DNA extraction and quantification 367
Biofilms were grown horizontally on Calcium Deficient Hydroxyapatite (CAD-HA) disks 368 (Hitemco Medical, Old Bethpage, USA) on the bottom of a 24-well plate in presence of a 369 substrate. Samples from the bioreactor-derived multi-species community were diluted 1:5 in 370 fresh BHI-2 with 2 x mucin, after which 1 mL was added to each well containing a HA disk. 371 Equal volumes (1 mL) of 2%(w/v) substrate solutions in BHI-2 without mucin were added to the 372 bacterial suspensions (final multi-species community dilution of 1:10, final substrate 373 concentration of 1%(w/v) in BHI-2). As a negative control, BHI-2 without substrate 374 supplementation was used. Biofilms were allowed to grow for 48 h under micro-aerophilic (6% 375 O2, 7% CO2, 7% H2, 80% N2) conditions (170 rpm, 37°C). All experiments were repeated on 376 three different days. After 48 h, biofilms were gently washed with phosphate buffered saline 377 (PBS, pH 7.4) to detach non-adherent cells, after which remaining biofilms were disrupted by 378 trypsinization and bacterial cells were harvested as described before 30 . DNA from only living 379 bacteria was extracted using a previously described propidium monoazide (PMA) treatment 30 . 380 Bacterial numbers were determined using a quantitative polymerase chain reaction (qPCR) 381 assay as described by Slomka et al. 30 , whereas species-specific primers and probes were 382 listed by Herrero et al. 69 . 383

Organic acid analysis of multi-species biofilm supernatants 384
Concentrations of lactate, acetate, formate, propionate and butyrate in the filter sterilized 385 supernatant of the multi-species biofilm assays were determined with a 761 Compact Ion 386 Chromatograph (Metrohm, Switzerland) with a Metrosep Organic acids 250/7.8 column and a 387 Metrosep Organic acids Guard/4.6 guard column, with the eluent consisting of 1 mM H2SO4 at 388 a flow rate of 0.8 mL min -1 . Organic acid production/consumption was calculated as the organic 389 acid concentrations detected in the filter sterilized supernatants, minus the concentrations of 390 those organic acids detected in sterile BHI-2 with or without supplemented substrate. 391

RNA extraction and virulence gene expression analysis 392
Biofilm-coated disks were dip-rinsed in PBS (pH 7.4) to remove unattached cells, followed by 393 bacterial RNA extraction as described previously 60  (confidence level of 95%) followed by Dunnett's correction for simultaneous hypothesis testing. 411 Changes in absolute bacterial abundances expressed as the difference between the value of 412 the control condition and the value of the substrate condition were analysed through a two-413 tailed, one sample t test to detect differences significantly different from 0 (no difference 414 between control condition and substrate condition

Table 1 Changes in relative composition of multi-species biofilms grown in presence of H2O2 production-affecting substrates
Multi-species biofilms were grown during two series of experiments (upper part and lower part) in the absence (=control; BHI-2 medium) or presence of the different substrates dissolved in BHI-2 medium at a concentration of 1%(w/v). Relative abundances of commensals, periopathogens and cariogenic pathogens are shown as mean ± SD (n = 3) and expressed in %genome equivalents per millilitre (%Geq/mL). Statistically significant changes in comparison with the control condition are shown in bold and are marked with '*' (P < 0.05, ANOVA + Dunnett's correction for simultaneous hypothesis testing).

Table 2 Organic acid production/consumption by multi-species biofilms grown in presence of H2O2 production-affecting substrates
Multi-species biofilms were grown during two series of experiments (upper part and lower part) in the absence (=control; BHI-2 medium) or presence of the different substrates dissolved in BHI-2 medium at a concentration of 1%(w/v). Organic acid production/consumption (shown as mean ± SD (n = 3) and expressed in mg/L) was calculated as the organic acid concentrations detected in the filter sterilized supernatants, minus the concentrations of those organic acids detected in sterile BHI-2 with or without supplemented substrate. Values preceded by a negative sign ('-') indicate organic acid consumption (net decrease), whereas all other values indicate organic acid production (net increase). Statistically significant changes in comparison with the control condition are shown in bold and are marked with '*' (P < 0.05, ANOVA + Dunnett's correction for simultaneous hypothesis testing). OA: organic acid.

Table 3 Changes in virulence gene expression of multi-species biofilms grown in presence of H2O2 production-affecting substrates
Multi-species biofilms were grown during two series of experiments in the absence (=control; BHI-2 medium) or presence of the different substrates dissolved in BHI-2 medium at a concentration of 1%(w/v). Changes in the expression of a selection of virulence genes from three periodontal pathogens present in the multi-species biofilms were determined. Fold changes in virulence gene expression were calculated with the 2^-ΔΔCt method and were determined relative to the control (BHI-2). Data are shown as geometric mean and C.I. (n = 3) of the 2^-ΔΔCt values. Values between 0 and 1 indicate downregulation relative to the control, values >1 indicate upregulation relative to the control. Statistically significantly different fold changes relative to the control with a value <0.5 (more than 2-fold downregulated) or >1.5 (more than 1.5-fold upregulated) are considered biologically relevant and are shown in bold (P < 0.05, ANOVA + Dunnett's correction for simultaneous hypothesis testing). The color