microorganisms

Article
Healthy Oral Lifestyle Behaviours Are Associated with
Favourable Composition and Function of the Oral Microbiota
Shirleen Hallang 1 , Anders Esberg 2, * , Simon Haworth 1,2,3 and Ingegerd Johansson 2

1 Faculty of Health Sciences, Bristol Dental School, University of Bristol, Bristol BS1 2LY, UK;
[email protected] (S.H.); [email protected] (S.H.)
2 Department of Odontology, Umeå University, 901 87 Umeå, Sweden; [email protected]
3 Medical Research Council Integrative Epidemiology Unit, Department of Population Health Sciences,
Bristol Medical School, University of Bristol, Bristol BS8 2BN, UK
* Correspondence: [email protected]

Abstract: Modifiable lifestyle interventions may influence dental disease by shifting the composition
of the oral microbiota. This study aimed to test whether lifestyle traits are associated with oral
microbiota composition and function. Swedish volunteers, aged 16 to 79 years, completed a lifestyle
traits questionnaire including lifestyle characteristics and oral health behaviours. Bacterial 16S rDNA
amplicons were sequenced and classified into genera and species, using salivary DNA. Microbiota
functions were predicted using Phylogenetic Investigation of Communities by Reconstruction of
Unobserved States and the KO Database of Molecular Functions by ortholog annotation. Tests for
association used partial least squares and linear regression analysis with correction for multiple





 testing. The main analysis included 401 participants and 229 common bacterial species (found
in ≥10% of the participants). The overall microbiota composition was strongly associated with
Citation: Hallang, S.; Esberg, A.;
questions “do you think caries is a disease?” and “do you use floss or a toothpick?”. Enriched relative
Haworth, S.; Johansson, I. Healthy
Oral Lifestyle Behaviours Are
abundance of Actinomyces, Campylobacter, Dialister, Fusobacterium, Peptidophaga and Scardovia genera
Associated with Favourable (all p < 0.05 after adjustment for multiple testing), and functional profiles showing enrichment of
Composition and Function of the carbohydrate related functions, were found in participants who answered “no” to these questions.
Oral Microbiota. Microorganisms 2021, Socio-demographic traits and other oral hygiene behaviours were also associated. Healthier oral
9, 1674. https://doi.org/10.3390/ microbiota composition and predicted functions are found in those with favourable oral health
microorganisms9081674 behaviours. Modifiable risk factors could be prioritized for possible interventions.

Academic Editor: Georgios Keywords: oral behaviour; lifestyle; oral microbiome

N. Belibasakis

Received: 18 June 2021

Accepted: 2 August 2021
1. Introduction
Published: 6 August 2021
Healthy humans are colonized by a wide range of commensal organisms which form
Publisher’s Note: MDPI stays neutral
niche-specific microbiotas. Eubiosis and dysbiosis of these microbiotas are thought to be
with regard to jurisdictional claims in
relevant to an increasing range of health outcomes [1,2]. Lifestyle interventions which
published maps and institutional affil- modulate bacteria or other microorganisms are, therefore, one possible way to influence
iations. disease, but there are few examples where this is used as a major treatment modality in
clinical practice.
An exception is in the management of dental diseases, where modulation of the oral
microbiota is a main form of prevention for dental caries and periodontitis. Here, the aim is
Copyright: © 2021 by the authors.
to shift the characteristics of the oral microbiota from dysbiosis to an eubiotic state through
Licensee MDPI, Basel, Switzerland.
a combination of interventions. These interventions include both changes to lifestyle (oral
This article is an open access article
hygiene behaviours, dietary habits, etc.) as well as topical use of antimicrobial agents,
distributed under the terms and such as chlorhexidine in selected cases [1,3]. Thus, dental caries and periodontitis act as
conditions of the Creative Commons model examples of conditions where lifestyle interventions aim to change the microbiota
Attribution (CC BY) license (https:// composition or function.
creativecommons.org/licenses/by/ While these interventions are effective in reducing clinical disease burden [3,4], it is
4.0/). not altogether clear which lifestyle factors are most influential in shifting the microbiota

Microorganisms 2021, 9, 1674. https://doi.org/10.3390/microorganisms9081674 https://www.mdpi.com/journal/microorganisms

Microorganisms 2021, 9, 1674 2 of 14

characteristics, nor which species and microbiota functions are most strongly associated
with these behaviours. For example, flossing has been reported to be associated with
altered measures of microbial diversity in some studies [5] but not in others [6], and other
studies report only small effects of lifestyle choices [7]. Likewise, several reviews have
discussed lifestyle interventions and their effects on single bacterial species or a limited
number of species, including modifications to diet, use of tobacco, alcohol consumption
and oral hygiene behaviours [1,8], however there are few studies which assess the entire
oral microbiota and fewer still have adopted a systematic approach to assess for lifestyle
associations with microbiota function [5,7].
It would, therefore, be useful to systematically screen lifestyle factors for association
with oral microbiota composition and function, which was the aim of the present study.
The results of this may highlight the most relevant lifestyle factors which could be targeted
in trials to improve oral health, as well as provide a model to prioritize interventions for
other microbiota-related health outcomes.

2. Materials and Methods

2.1. Study Cohort
The study included 427 participants aged between 16 and 79 years who lived in the
northern part of Sweden. Participants <20 years were recruited as they attended their
annual dental check-up, and participants aged (≥20 years) were volunteers who responded
to a request for study participants. The exclusion criteria were cognitive disability, severe
illness, antibiotic treatment within 3 months and inability to communicate in Swedish
or English.

2.2. Questionnaire Information

The participants completed a questionnaire on living conditions, tobacco use, medical
status and medications, lifestyle traits including lifestyle characteristics and oral health
behaviours. The core questions were taken from a questionnaire used in the Västerbotten
Intervention Programme [9], but were supplemented with additional questions specifically
about oral health behaviours and attitudes reflecting oral disease risk factors. The questions
and response options are presented in Table S1. Habitual food intake over the latest
year was recorded using a semi-quantitative food frequency questionnaire (FFQ) with
93 questions for food items or food aggregates. The response options were “never”, “less
than once a month”, “1–3 times per month”, “once a week”, “2–3 times a week”, “4–6 times
a week”, “once a day”,”2–3 times a day” and “4 or more times a day”. Participants were
asked to estimate portion size using example photographs and select a photograph which
matched their regular portion size. For foods which form natural portion sizes (such as an
egg) weights were taken from the food database at the Swedish National Food Agency [10].
Reported intakes were transformed to intakes per day. Daily intake of energy, sucrose
and sugar (sucrose plus the monosaccharides glucose and fructose) were calculated using
weights from the Swedish National Food Agency [10]. FFQ assessed intakes have been
validated against 10 repeated 24-hour recalls [11]. Overall diet quality was summarized
using a healthy eating index [12].

2.3. Microbiota Analysis

Participants were asked not to brush their teeth on the morning of saliva collection
and not to eat or drink for 1 h before saliva collection. Stimulated saliva was then collected
for 3 min while the participants chewed on a 1g piece of paraffin wax. Saliva samples were
stored at −80 ◦ C until used.
DNA was extracted from saliva, a mock community, and a negative control (ultra-pure
water). Bacterial 16S rDNA amplicons were generated from the v3—v4 hypervariable
region using PCR with fusion primers with 341F (ACGGGAGGCAGCAG) forward and
806R (GGACTACHVGGGTWTCTAAT) reverse primers as described by Caporaso [13].
Equimolar 16S rDNA amplicon libraries were pooled and purified using AMPure XP beads
Microorganisms 2021, 9, 1674 3 of 14

(Beckman Coulter, Stockholm, Sweden) and sequenced using the Illumina Miseq platform.
The samples were spiked with 5% PhiX (Illumina, Stockholm, Sweden). Each run included
two mock samples and two negative controls alongside the test samples.
Sequence reads were de-multiplexed using deML [14], and cleaned using DADA2 in
the QIIME2 next-generation microbiome bioinformatics platform [15–17]. During clean-
ing pair-end reads were fused, primers, ambiguous, chimeric and PhiX sequences were
removed, and amplicon sequence variants (ASVs) were retained. These ASVs were then
classified against the expanded Human Oral Microbiome Database (eHOMD) [18,19]. ASVs
with at least 2 reads and 98.5% identity with a named species or unnamed phylotype in
eHOMD were retained, and those with the same Human Microbial Taxon (HMT) ID num-
ber were aggregated. The HMT aggregated taxa were standardized to the level of the
sample with fewest reads (19,700 reads), and then transformed using inverse hyperbolic
sine transformation. This transformation was selected because zero values are common in
species-level abundance variables, and this method provides values for variables contain-
ing zeros.
The term “species” is used in the remainder of the text to refer to both species and
unnamed phylotypes for simplicity.

2.4. Prediction of Oral Microbiota Functions from the 16S rRNA Gene Sequences
Predicted molecular functions of the oral microbiota were generated from the obtained
16S rRNA sequences using Phylogenetic Investigation of Communities by Reconstruction
of Unobserved States (PICRUSt2) [20] and the KO Database of Molecular Functions by
ortholog annotation (KEGG orthologues, KO, within QIIME2) [21]. A closed reference
feature table was created using the Greengenes database version 13_5 [22] which is trained
against PICRUSt2. Core diversity metrics were estimated in QIIME2, and a KEGG KO
feature table was exported for downstream analyses.

2.5. Data Handling and Statistical Analysis

Demographic information was summarized as means with standard deviation (sd)
for 95% confidence intervals (CI), or by reporting a proportion (%). Nutrient intakes were
adjusted for sex, body mass index (BMI) and estimated energy intake using generalized
linear modelling. For microbiota comparisons, Shannon and Simpson alpha diversity
measures were calculated using QIIME2. All tests were two-sided and p-values < 0.05 were
considered statistically significant unless false discovery rate correction at FDR 0.05 was
applied to account for multiple testing.
PERMANOVA, using Paleontological Statistics (PAST4) [23] with 9999 permutations
and FDR-corrected p-values, was used to compare groups based on distance measures.
Multivariate analyses used partial least squares (PLS) analysis to identify association
between lifestyle traits and measures of microbiota composition and function. Separate
models were fitted for abundance measures (relative abundance, restricted to species
detected in ≥10% participants) and predicted functions (restricted to functions with a non-
zero predicted level in ≥10% participants). Models were fitted using SIMCA p+ version 15.0
(Sartorius Stedim Data Analytics AB, Malmö, Sweden) and variables were scaled to unit
variance. Cross-validation was performed using a K-fold method, with systematic removal
of every 7th observation and prediction of the remaining observations (Q2 -values). Results
of overall model fit and separation were displayed in score loading plots. Importance
attributable to different lifestyle traits was plotted as bar plots of the variance explained
(R2 ) and variance predicted (Q2 ) by each lifestyle trait in the fitted and cross-validated
models. Volcano plots, based on the VIP-value (metric summarizing the importance of
each variable in driving the observed group separation) and p(corr) (a loading scaled as
a correlation coefficient) between the model and original data were used to illustrate the
distribution of genus/species or most influential functional pathways. For genus/species,
variables were considered influential if they had VIP >1.5 and p(corr) <−0.50 or >0.50, and
for predicted functions VIP >1.9 and p(corr) <−0.65 or >0.65.
Microorganisms 2021, 9, 1674 4 of 14

Regression models were carried out using relative species-level abundance for species
detected in ≥10% participants. Models were fitted for each bacterial species in relation to
each lifestyle trait using linear regression and included adjustment for age, sex and edu-
cational level. p-values were adjusted for multiple testing using the Benjamini-Hochberg
method with false discovery rate set to 0.05. As a sensitivity analysis, species detected
in <10% of participants were modelled using logistic regression, with adjustment for
covariates and multiple testing as described above.

3. Results
3.1. Study Group and General Sequencing Results
In total, 427 participants met the inclusion criteria, of which 17 were excluded for miss-
ing questionnaire information and nine did not have saliva available for DNA extraction,
leaving 401 participants in the final study group. Of these, 62.3% were females, 51.6% were
below 20 years, and mean (95% CI) BMI was 23.1 (22.8, 23.4), with 24.4% having a BMI ≥ 25
(overweight or obese). In this group, 6.4% reported being current or past smokers and
12.9% were current or former users of Swedish snus (snuff) (Table 1). Further characteristics
of the study group are shown in Tables 1 and 2.

Table 1. General and lifestyle characteristics of the participants in the study group.

Variable in Analysis and

General and Lifestyle Characteristics Value
Short Label in Figures
Women, % 62.3
Age, years, mean (95% CI) 28.9 (27.4, 30.5)
<20 years, % 51.6
BMI, mean (95% CI) 23.1 (22.8. 23.4)
Overweight/obese (BMI ≥25), % 24.4 overweight
Highest educational level for age a , % 47.4 education
How do you assess your health?, %
good 83.3
not so good 14.5
How was your health last month?, %
good 89.8
not so good 10.2
Were you ill the week before sampling?, %
yes 15.6
no 84.4
Do you take any medicine?, %
yes 24.2
no 75.8
Smoking, % smoke
present/ex-smoker 6.5
never smoked 93.5
Snuff use, % snuff
present/ex- user 13.0
never used 87.0
Physical activity at work, % work-load
non-heavy work 75.4
heavy work 24.6
Physical activity at leisure time, % leisure-time
<1 time per week 33.4
≥1 time per week 66.6
When was the last time you ate or drunk?, % time-eat
≤2 h ago 70.9
>2 h ago 29.1
Sugar intake, g/day, mean (95% CI) 56.2 (54.3, 58.0) sugar
Sucrose intake, g/day, mean (95% CI) 31.0 (29.9, 32.3) sucrose
Heathy diet score, mean (95% CI) 11.9 (11.5, 12.3) diet-score
a The question is phrased “What is the highest education you have completed”. This is then coded to reflect whether
the participant has the highest educational level which is possible for their age. Highest level refers to upper high
school, college or university depending on age.
Microorganisms 2021, 9, 1674 5 of 14

Table 2. Oral health behaviour-related characteristics of the participants in the study group.

Variable in Analysis and

Oral Health Behaviours Value, %
Short Label in Figures
Tooth brushing < 1 per day 10.3 brush
Do you use floss or a toothpick, yes? 51.9 floss
Do your gums bleed on brushing, yes? 21.0 bleeding
Do you use a fluoridated toothpaste, yes? 75.1 fluoride toothpaste
Do you use extra fluoride, yes? 15.6 extra-fluoride
Do you use any mouth rinse, yes? 25.6 rinse
Do you think cavities is a disease, yes? 37.8 caries-a-disease

After quality filtering and chimera removal, 20,600,607 sequences remained for the
401 participants and were clustered into amplicon sequence variants (ASVs) which rep-
resented 116 genera and 466 species or phylotypes with 2 or more reads when classified
against the eHOMD at ≥98.5% identity. The mean (sd) number of reads/sample was 51,373
(17,268) and the negative controls yielded (mean (sd) 141 (52)) reads. The species in the
mock communities were correctly identified. For the present paper species/phylotypes
detected in ≥10% of the participants were considered for all lifestyle traits in the main
analyses (229 common species), and uncommon species (detected in <10%) were only
included in species-by-species regression sensitivity modelling. A full list of common
species/phylotypes with their relative abundances are shown in Table S2.

3.2. Association between Self-Reported Lifestyle Characteristics and Oral Health Behaviours and
Overall Oral Microbiota
To explore the potential effect of self-reported lifestyle traits including oral health
behaviours and lifestyle characteristics on overall oral microbiota composition, orthog-
onal partial least square regression analysis with all selected lifestyle traits (n = 17, see
Tables 1 and 2) was carried out. The model illustrates the relationship between and among
the bacterial genera/species and the lifestyles traits. The overall model indicated that
characteristics associated with higher education, healthy lifestyle, good oral hygiene and
use of fluoride products clustered together, seen at the left of Figure 1A. Lifestyle traits that
were significantly influential in the model were level of education and physical workload,
and with regard to oral health behaviours, tooth-brushing frequency, flossing, use of extra
fluoride and whether the participant thought that caries was a disease, were also influential
(PCV-ANOVA < 0.05, Figure 1B).

3.3. Lifestyle Associations with Single Bacterial Species or Phylotypes

The two lifestyle traits most strongly associated with overall microbiota composition
were “do you think caries is a disease?” and “do you use floss or a toothpick?”. To understand
which bacterial genera or species were driving this association, we used orthogonal partial
least square regression discrimination analysis to evaluate these traits in further detail.
The models showed evidence for association at the species level (R2 = 47.2%, Q2 = 23.0%,
(PCV-ANOVA = 1.8 × 10−17 ) and (R2 = 35.5%, Q2 = 19.7%, PCV-ANOVA = 1.7 × 10−17 ), re-
spectively (Figure 2A,C). The models indicated, to some extent, overlapping enrichment
of species belonging to the genera Actinomyces, Campylobacter, Dialister, Fusobacterium,
Peptidophaga and Scardovia to associate with answering “no”, whereas answering “yes”
associated with enrichment of Veillonella in both models (Figure 2B,D).
To further evaluate species associations with oral health behaviours and lifestyle char-
acteristics, we performed sensitivity analysis to adjust for age, sex and highest educational
level for all 17 lifestyle traits and the same 229 common species, i.e., present in at least 10%
of the participants, included in the multivariate analysis. After adjustment for multiple
testing, abundance of 21 species were associated with three or more lifestyle traits and
nine lifestyle traits were associated with an abundance of five or more common species
(Figure 3). Results were concordant with the main analysis, where “do you think caries is a
Microorganisms 2021, 9, 1674 6 of 14

disease?” was the strongest associated oral health trait. Sensitivity analysis of uncommon
Microorganisms 2021, 9, x FOR PEER REVIEW
species did not find any association passing correction for multiple testing. Results 6 of are
15
presented in full in Tables S3 and S4.

Figure
Figure 1.
1. Lifestyle
Lifestyle characteristics
characteristicsandand oral
oral health
health behaviours
behaviours associated
associated with
with oral
oral microbiota
microbiota composition.
composition. Associations
Associations
between
betweenlifestyle
lifestyle characteristics and oral
characteristics and oralhealth
healthbehaviours,
behaviours,and
andsaliva
salivamicrobiota
microbiota composition
composition were
were evaluated
evaluated by by orthog-
orthogonal
onal partial least square regression analysis, where the oral health behaviours and lifestyle characteristics were introduced
partial least square regression analysis, where the oral health behaviours and lifestyle characteristics were introduced
as dependent variables (y) and bacterial genus and species as independent variables (x). (A) A loading score plot of the
as dependent variables (y) and bacterial genus and species as independent variables (x). (A) A loading score plot of the
generated model that illustrates the correlation between the participants’ lifestyle characteristics and oral health behav-
generated
iours model
and the that illustrates
composition the correlation
of their between
oral microbiota. (B)the
Barparticipants’
plot show thelifestyle characteristics
R2 value and oralthe
which indicates health behaviours
fraction of the
and the composition of their oral microbiota. (B) Bar plot show the R2 value which indicates the fraction of the original
original data explained by the model and the Q parameter indicates the fraction of the original data predicted by the 7-
2

datacross-validation
fold explained by the model
model. * indicates Q2 parameter
and the significant indicates
models the fraction
for indicated ofCV-ANOVA
trait (P the original data
* < 0.05, ** <predicted
0.01, *** <by the 7-fold
0.001).
cross-validation model. * indicates significant models for indicated trait (PCV-ANOVA * < 0.05, ** < 0.01, *** < 0.001).
3.3. Lifestyle Associations with Single Bacterial Species or Phylotypes.
3.4. The
Lifestyle
two Characteristics
lifestyle traits and
most Oral Healthassociated
strongly Behaviourswith
Associated
overallwith an Overall
microbiota Change in
composition
Predicted Microbiota Functions
were “do you think caries is a disease?” and “do you use floss or a toothpick?”. To understand
whichGiven the genera
bacterial associations between
or species were lifestyle traits
driving this and oral microbiota
association, composition,
we used orthogonal we
partial
hypothesized that there might be associations between lifestyle and healthy
least square regression discrimination analysis to evaluate these traits in further detail.or unhealthy
microbiota
The models functions. Microbiota
showed evidence functions were
for association at thepredicted using(Rorthologue
species level 2 = 47.2%, Q annotation
2 = 23.0%,
(KEGG orthologues, KOs) and then modelled in relation to lifestyle traits.
(PCV-ANOVA = 1.8 × 10 ) and (R = 35.5%, Q = 19.7%, PCV-ANOVA = 1.7 × 10 ), respectively
−17 2 2 −17 The model
used functions with a prevalence of at least 10% among the participants
(Figure 2A,C). The models indicated, to some extent, overlapping enrichment of species (10,140 functions
included out of the total number of 10,543 predicted functions). The overall model in-
belonging to the genera Actinomyces, Campylobacter, Dialister, Fusobacterium, Peptidophaga
dicated that lifestyle traits associated with unhealthy oral health behaviours along with
and Scardovia to associate with answering “no”, whereas answering “yes” associated with
unhealthy diet, physical work, lower educational level and not thinking that caries is a
enrichment of Veillonella in both models (Figure 2B,D).
disease clustered to the right (Figure 4A). Six lifestyle traits had significant association with
predicted functions; “physical activity at work”, “tooth brushing < 1 per day”, “do you use floss
or a toothpick?”, “do you use a fluoridated toothpaste?”, “do you use extra fluoride?” and “do you
think caries is a disease?” (Figure 4B) (PCV-ANOVA < 0.05).
Microorganisms 2021, 9, 1674 7 of 14
Microorganisms 2021, 9, x FOR PEER REVIEW 7 of 15

Figure2.2. Identification
Figure Identificationofofgenera
generaandand
species which
species are most
which strongly
are most associated
strongly with two
associated oraltwo
with health behaviours.
oral Orthog-
health behaviours.
onal partial least square regression discrimination analysis was used to evaluate the association between participants an-
Orthogonal partial least square regression discrimination analysis was used to evaluate the association between participants
swer to the questions (A,B) “do you think caries is a disease?” and (C,D) “do you use floss or a toothpick?” and the oral micro-
answer to the questions (A,B) “do you think caries is a disease?” and (C,D) “do you use floss or a toothpick?” and the oral
biota composition. The model score plots (A,C) illustrate the observed group separation (each participant is represented
microbiota composition.
by a dot) based on theirThe model score
microbiota plots(B,D)
profile. (A,C)Inillustrate
order tothe observed
identify the group
subsetseparation
of bacterial(each participantimportant
species/genus is represented
for
by
the observed group separation, selection based on a combination of Variable Influence in Projection (VIP) and p for
a dot) based on their microbiota profile. (B,D) In order to identify the subset of bacterial species/genus important (PLS the
observed group
correlation separation,
coefficient) wasselection
performed.basedVIPonisa combination
a metric thatof Variable Influence
summarizes in Projection
the importance (VIP)
of each and p (PLS
variable correlation
in driving the
observed group
coefficient) separationVIP
was performed. andisp(corr) is athat
a metric loading scaled as
summarizes thea importance
correlation coefficient (ranging
of each variable from −1.0
in driving theto 1.0) between
observed group
the model and
separation and original
p(corr) isdata. Inclusion
a loading criteria
scaled as awere set to VIP
correlation >1.6 and(ranging
coefficient p(corr) <−0.5
from or−>0.5. (g_)
1.0 to indicates
1.0) betweengenus level, (s_)
the model and
species level.
original data. Inclusion criteria were set to VIP >1.6 and p(corr) <−0.5 or >0.5. (g_) indicates genus level, (s_) species level.

To further
Having evaluate
found a globalspecies associations
shift in predicted with oral based
functions health on
behaviours and lifestyle
lifestyle traits, we next
tried to identify functions important for the observed group separation for the twoedu-
characteristics, we performed sensitivity analysis to adjust for age, sex and highest most
cational level
prominent for alltraits,
lifestyle 17 lifestyle
“do youtraits
thinkand the
caries is same 229 common
a disease?” species,
and “do you i.e.,orpresent
use floss in at
a toothpick?”.
least 10% of partial
Orthogonal the participants,
least square included in the
regression multivariateanalysis
discriminant analysis. wasAfter
usedadjustment
and basedfor on
multiple
the top 50testing,
functionsabundance
for each of 21 species
lifestyle wereanswer
trait and associated with
option basedthreeonor more lifestyleof
a combination
traits
VIP andp(corr)
and nine lifestyle
(VIP >1.9, traits were<associated
p(corr) with anasabundance
−0.65 or >0.65), of five or more common
described previously.
species (Figure 3). Results were concordant with the main analysis,
The model using the trait “do you think caries is a disease?” explained 29.2% where of “do you think
variance in the
original data and was able to predict 20.0% after cross-validation (PCV-ANOVA = 7.2 × 10−of
caries is a disease? ” was the strongest associated oral health trait. Sensitivity analysis 15 ).

uncommon
Unique species
enriched did not find
microbiota any association
functions (represented passing
by twocorrection
predictedfor multiple
genes or more testing.
(num-
Results
ber are presented
indicated in full in
in parenthesis Tables
after eachS3function))
and S4. linked to the answer “no” were Amino
sugar and nucleotide sugar metabolism (7), O-Antigen nucleotide sugar biosynthesis (4),
ABC transporters (4), Carbon metabolism (3), Starch and sucrose metabolism (3), Pentose
and glucuronate interconversions (2), Galactose metabolism (2), and Streptomycin biosyn-
thesis (2). Conversely, the answer “yes” was associated with to Propanoate metabolism (2),
Aminoacyl-tRNA biosynthesis (2), Arginine and proline metabolism (2), Phenylalanine,
tyrosine and tryptophan biosynthesis (2), Glycerolipid metabolism (2), Two-component
system (2), and Carbapenem biosynthesis (2).
Microorganisms 2021, 9, 1674 8 of 14
Microorganisms 2021, 9, x FOR PEER REVIEW 8 of 15

Figure 3. Association between lifestyle traits and relative abundance of common bacterial species.
Figure 3. Association between lifestyle traits and relative abundance of common bacterial species.
Z scores indicate the strength and direction of association and are obtained from linear regression
Z scores indicate the strength and direction of association and are obtained from linear regression
models. Associations with Benjamini–Hochberg-adjusted p values < 0.05 are indicated with (*). All
models.
modelsAssociations with Benjamini–Hochberg-adjusted
included adjustment p valueslevel.
for age, sex and highest educational < 0.05 are indicated with (*). All
models included adjustment for age, sex and highest educational level.
 cluded out of the total number of 10,543 predicted functions). The overall model indicated
that lifestyle traits associated with unhealthy oral health behaviours along with unhealthy
diet, physical work, lower educational level and not thinking that caries is a disease clus-
tered to the right (Figure 4A). Six lifestyle traits had significant association with predicted
Microorganisms 2021, 9, 1674 functions; “physical activity at work”, “tooth brushing < 1 per day”, “do you use floss or a9tooth-
of 14
pick?”, “do you use a fluoridated toothpaste?”, “do you use extra fluoride?” and “do you think
caries is a disease?” (Figure 4B) (Pcv-ANOVA < 0.05).

Figure4.4.Predicted
Figure Predictedmicrobiota
microbiotafunctions
functionsrelated
relatedtotolifestyle
lifestylecharacteristics
characteristicsand
andoral
oralhealth
healthbehaviours.
behaviours.Orthogonal
Orthogonalpartial
partial
least square regression analysis was used to evaluate the effect of participants’ (n = 401) oral health behaviours and lifestyle
least square regression analysis was used to evaluate the effect of participants’ (n = 401) oral health behaviours and lifestyle
characteristics (n = 17 lifestyle traits) on the composition of the predicted microbiota functions (n = 10,140). (A) Loading
characteristics (n = 17 lifestyle traits) on the composition of the predicted microbiota functions (n = 10,140). (A) Loading
score plot of oral health behaviours and lifestyle characteristics as dependent-y variables and predicted microbiota func-
score
tionsplot of oral health behaviours
as independent-x and lifestyle
variables. The loading characteristics as dependent-y
score plot illustrates variables
the correlation and predicted
between the y andmicrobiota
x variables.functions
(B) Bar
as independent-x variables. The loading score plot illustrates the correlation between the y and x
plot showing R -value (green bar, indicates the fraction of the original data explained by the model) and Q -value
2 variables. (B)
2 Bar (blue
plot
showing 2
R -value (green bar, indicates the fraction of thebyoriginal data explained by the model) 2
and* indicates
Q -value significant
(blue bar,
bar, indicates the fraction of the original data explained the seven-fold cross-validation model).
indicates
differencethe fraction
between ofrespondents’
the the original answers
data explained
and theby the seven-fold
indicated questioncross-validation model).
(PCV-ANOVA * < 0.05, * indicates
** < 0.01, significant
*** < 0.001).
difference between the respondents’ answers and the indicated question (PCV-ANOVA * < 0.05, ** < 0.01, *** < 0.001).
Having found a global shift in predicted functions based on lifestyle traits, we next
triedWith respectfunctions
to identify to the trait “do you use
important for floss or a toothpick?”,
the observed group the model explained
separation for the two26.6%
most
of the original data and predicted 18.6% after cross-validation (PCV-ANOVA = 2.1 × 10−16 ).
Answer “no” associated with unique enrichment in functions linked to e.g., Microbial
metabolism in diverse environments (9), Carbon metabolism (5), Fructose and mannose
metabolism (5), Carbon fixation pathways in prokaryotes (4), TCA cycle (3), Starch and
sucrose metabolism (3), Biosynthesis of cofactors (2), Peroxisome (2), Nicotinate and nicoti-
namide metabolism (2), Galactose metabolism (2), Glycolysis /Gluconeogenesis (2), Biosyn-
thesis of amino acids (2), Pyruvate metabolism (2) and Nitrogen metabolism (2). Functions
linked to “yes” were Biosynthesis of amino acids (10), ABC transporters (5), Phenylala-
nine, tyrosine and tryptophan biosynthesis (4), C5-Branched dibasic acid metabolism (3),
2-Oxocarboxylic acid metabolism (3), Aminoacyl-tRNA biosynthesis (3), Arginine and
proline metabolism (3), beta-Lactam resistance (3), Valine, leucine and isoleucine biosyn-
thesis (3), Glycine, serine and threonine metabolism (2), Carbapenem biosynthesis (2),
Cationic antimicrobial peptide (CAMP) resistance (2) and Glycerolipid metabolism (2).
This suggests that not considering caries to be a disease as well as not flossing their teeth
Microorganisms 2021, 9, 1674 10 of 14

was linked to a more carbohydrate focused functional profile or their microbiota. For more
specifics on enriched functions, see Table S5.

4. Discussion
Effective management of dental diseases requires maintenance or re-establishment of
a eubiotic oral microbiota [1]. It is therefore important to understand which lifestyle and
behavioural interventions are most effective in modulating the composition or function of
the oral microbiota. This study used a hypothesis-free method to identify lifestyle factors
which are associated with measures of oral microbiota and function. The main finding is
that favourable oral health behaviours, including interdental cleaning, were associated
with both the overall composition and the predicted functionality of the oral microbiota.
Results of this study could be used to understand lifestyle factors to test in clinical trials to
modulate the oral microbiota, as well as provide a model for prioritizing interventions for
other microbiotas.
The primary analysis was used to screen for oral health behaviours that are associated
with the overall microbiota composition and function, but did not provide inference
about whether those compositions and functions are relevant to health. Selected oral
health behaviours were therefore followed up with detailed analysis of individual species
abundance, and predicted functions taking potential confounders into account. This
identified that unfavourable oral health behaviours were associated with relatively higher
abundance of caries-associated species such as Campylobacter gracilis [24] and Scardovia
wiggsiae [25–27] and higher predicted levels of cariogenic functions, including starch and
sucrose metabolism [28–30], fructose and mannose metabolism, glycolysis, amino sugar
and nucleotide sugar metabolism [29,30], galactose metabolism, TCA cycle [30] and pentose
and glucuronate interconversions [29,31]. These species and functions are highly relevant
for dental caries, where dysbiosis occurs with selective proliferation of acid-producing and
acid tolerating species [1,32] and potential demineralization of tooth enamel. In our results,
we see a concerted pattern of association with proliferation of cariogenic species and a
carbohydrate-focussed predicted functional profile with increased sugar metabolism. Thus,
the results suggest that the microbiota in people with unfavourable oral health behaviours
is more cariogenic. This needs to be confirmed in future studies relating predicted oral
microbiota function to dental disease status, but is in keeping with the current clinical
paradigm which aims to create shift in the dynamic oral biofilm by targeted lifestyle
interventions [32].
One such intervention which has been advocated is interdental cleaning [33]. Previous
studies have shown flossing to result in altered beta diversity [5] and reduced relative
abundance of Streptococcus species [34]. Non-flossers have also been shown to have an over-
abundance of bacterial species implicated in caries and periodontitis [35]. In the present
study, “do you use floss or a toothpick?” was associated with several species thought to be
relevant to oral health. People who answered “no” to this question had higher relative
abundance of Actinomyces naeslundii, as reported previously [35]; however, the present
study identified novel associations including higher relative abundance of Fusobacterium
nucleatum subsp. animalis in people who replied “no” to “do you use floss or a toothpick?”.
Fusobacterium nucleatum, an anaerobe, was originally classed as part of the “orange” com-
plex in subgingival plaque by Socranksy et. al, which is related to disease progression in
periodontitis [36]. Flossing has shown to decrease the relative abundance of bacteria found
in periodontal disease [35]. A possible mechanism for this has previously been suggested
by Burcham et al. [5], who argued that interdental cleaning might disrupt small ecological
niches and induce inflammatory responses against oral bacteria on the one hand, while
allowing increased relative abundance of less specialized species on the other hand.
One of the most strongly associated questionnaire responses was whether participants
felt that caries was a disease or not. This simple question explained a surprisingly large
amount of variation in oral microbiota composition and function. This variable is therefore
likely to capture not only the degree of importance that participants place on dental caries
Microorganisms 2021, 9, 1674 11 of 14

prevention, but also a range of other attitudes and beliefs which are relevant to dental
diseases. Given that this question is simple but predictive of oral microbiota traits, it could
potentially be considered as an adjunct to caries risk assessment in clinical settings. Low
educational level and physical work (which is a proxy for socio-economic status in Northern
Sweden) were also associated with unfavourable microbiota composition, reflecting the
situation with clinical endpoints where socio-economic status is a well-known risk factor
for dental caries [37–39].
In this population, smoking and snus (Swedish moist snuff) use were not strongly
associated with oral microbiota composition in the main analysis. This finding differs
from previous studies which report that smoking causes proliferation of species including
Streptococcus sobrinus, Eubacterium brachy [40], Streptococcus mutans and Veillonella dispar [41].
At genus level, the genera of Atopobium [42], Streptococcus [42,43], Prevotella, and Veillonella
are all reportedly enriched in smokers [43]. The differences in results between these and
the present study may be related to statistical power, as tobacco use is becoming rarer over
time in Sweden and over 90% of the participants in this study group reported never having
smoked. It is also possible that smoking and snus use have somewhat opposing effects
on the microbiota which mask patterns of association, as smoking causes a decrease in
salivary pH [44] while snus is reported to increase salivary pH during use [45]. Finally,
there may be heterogeneity in the effects of different preparations of snus which could
further mask associations, depending on the snus pH which can either increase or decrease
pH in the mouth [46].
At present, there is little consensus about the required fasting time before donat-
ing saliva for oral microbiota analysis. Some authors have used prolonged fasting peri-
ods [47,48] and others have used and suggested shorter periods [49–52]. In this study,
time since last eating or drinking was not associated with microbiota composition in the
overall model. This may reflect that the measures used were based on relative abundance
measures standardized to the number of reads, which are likely to be relatively stable over
a short timescale (hours). This suggests that small differences in fasting time are unlikely
to introduce substantial bias into microbiota datasets, provided that relative abundance
measures are used. By contrast, measures of function based on RNA sequencing are likely
to change rapidly over short timescales since transcription occurs more rapidly than cell
proliferation. Studies using RNA-based methods may therefore be more sensitive to time
since fasting [47] than the DNA based method used in the present study.
The main strengths include the systematic approach to testing for association, the
inclusion of predicted functional analysis and the sample size which is relatively large for
studies of the oral microbiota. The limitations include the observational design which does
not allow for causal inference, and the limited statistical power for some analyses, given
that some species or behaviours were uncommon in this group and given the need for
strict correction for multiple testing. An additional limitation to consider when assessing
the results is the geographic limitations (Sweden) of the study population as the oral health
and lifestyle behaviours vary across the world and among various ethnic groups [53,54].
In summary, the study tested for association between lifestyle factors and oral mi-
crobiota composition and function, finding that favourable oral health behaviours are
associated with healthier composition and predicted function of the oral microbiota. The
results of the study could help prioritize the most important health behaviours and modifi-
able risk factors to target in prevention of dental diseases.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/

10.3390/microorganisms9081674/s1, Table S1: List of questions asked in the questionnaire with
response options, codings and short labels used in main text figures; Table S2: List of bacterial species
prevalence and relative abundance of species present in at least 10% of the participants; Table S3:
Linear regression models adjusted for age, sex and highest educational level for lifestyle traits and
common species.; Table S4: Lists of traits or species whose relative abundance was associated with
one or more species/lifestyle traits; Table S5: Top 50 PLS-DA predicted microbiota functions.
Microorganisms 2021, 9, 1674 12 of 14

Author Contributions: Conceptualization, I.J. and S.H. (Simon Haworth); formal analysis, S.H.
(Shirleen Hallang) and A.E.; resources, I.J.; writing original draft preparation, S.H. (Shirleen Hallang)
and S.H. (Simon Haworth); funding acquisition, I.J. All authors have read and agreed to the published
version of the manuscript.
Funding: This research was funded by the Swedish Patent Revenue Fund (grant 2017-019, I.J.).
S.H. (Simon Haworth) and S.H. (Shirleen Hallang) receive funding from the UK National Institute
for Health Research through the academic clinical fellowship scheme. The funders had no role in
the design of the study; in the collection, analyses, or interpretation of data; in the writing of the
manuscript, or in the decision to publish the results.
Institutional Review Board Statement: The study received ethical approval from the Swedish
Ethical Review Authority (dnr 2018/335-31 and dnr 09-134M).
Informed Consent Statement: The study adhered to the Helsinki declaration and all participants
provided signed informed consent to participate.
Data Availability Statement: Microbiota sequences are available at 10.6084/m9.figshare.14748276
and 10.6084/m9.figshare.14748126 and other data are available upon reasonable request and after
acquisition of mandatory ethical and other approvals.
Acknowledgments: The authors want to acknowledge Linda Eriksson and Pamela Hasslöv for
collecting clinical samples and Agnetha Rönnlund for excellent support during sampling and sam-
ple preparations.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Kilian, M.; Chapple, I.L.; Hannig, M.; Marsh, P.D.; Meuric, V.; Pedersen, A.M.; Tonetti, M.S.; Wade, W.G.; Zaura, E. The oral
microbiome—An update for oral healthcare professionals. Br. Dent. J. 2016, 221, 657–666. [CrossRef]
2. Kilian, M. The oral microbiome—Friend or foe? Eur. J. Oral Sci. 2018, 126, 5–12. [CrossRef]
3. Jepsen, S.; Blanco, J.; Buchalla, W.; Carvalho, J.C.; Dietrich, T.; Dörfer, C.; Eaton, K.A.; Figuero, E.; Frencken, J.E.; Graziani, F.; et al.
Prevention and control of dental caries and periodontal diseases at individual and population level: Consensus report of group
3 of joint EFP/ORCA workshop on the boundaries between caries and periodontal diseases. J. Clin. Periodontol. 2017, 44, S85–S93.
[CrossRef]
4. Chapple, I.L.; Bouchard, P.; Cagetti, M.G.; Campus, G.; Carra, M.C.; Cocco, F.; Nibali, L.; Hujoel, P.; Laine, M.L.; Lingstrom, P.;
et al. Interaction of lifestyle, behaviour or systemic diseases with dental caries and periodontal diseases: Consensus report of
group 2 of the joint EFP/ORCA workshop on the boundaries between caries and periodontal diseases. J. Clin. Periodontol. 2017,
44, S39–S51. [CrossRef]
5. Burcham, Z.M.; Garneau, N.L.; Comstock, S.S.; Tucker, R.M.; Knight, R.; Metcalf, J.L.; Miranda, A.; Reinhart, B.; Meyers, D.;
Woltkamp, D.; et al. Genetics of Taste Lab, Citizen Scientists Patterns of Oral Microbiota Diversity in Adults and Children:
A Crowdsourced Population Study. Sci. Rep. 2020, 10, 2133. [CrossRef]
6. Caselli, E.; Fabbri, C.; D’Accolti, M.; Soffritti, I.; Bassi, C.; Mazzacane, S.; Franchi, M. Defining the oral microbiome by whole-
genome sequencing and resistome analysis: The complexity of the healthy picture. BMC Microbiol. 2020, 20, 120. [CrossRef]
[PubMed]
7. Nearing, J.T.; DeClercq, V.; Van Limbergen, J.; Langille, M.G.I. Assessing the Variation within the Oral Microbiome of Healthy
Adults. mSphere 2020, 5, e00451. [CrossRef]
8. Cornejo Ulloa, P.; van der Veen, M.H.; Krom, B.P. Review: Modulation of the oral microbiome by the host to promote ecological
balance. Odontology 2019, 107, 437–448. [CrossRef] [PubMed]
9. Norberg, M.; Wall, S.; Boman, K.; Weinehall, L. The Västerbotten Intervention Programme: Background, design and implications.
Glob. Health Action 2010, 22. [CrossRef] [PubMed]
10. The Swedish Food Composition Database. Available online: https://www7.slv.se/SokNaringsinnehall/ (accessed on 2 Jan-
uary 2018).
11. Johansson, I.; Hallmans, G.; Wikman, A.; Biessy, C.; Riboli, E.; Kaaks, R. Validation and calibration of food-frequency questionnaire
measurements in the Northern Sweden Health and Disease cohort. Public Health Nutr. 2002, 5, 487–496. [CrossRef] [PubMed]
12. Nettleton, J.A.; Hivert, M.F.; Lemaitre, R.N.; McKeown, N.M.; Mozaffarian, D.; Tanaka, T.; Wojczynski, M.K.; Hruby, A.; Djoussé,
L.; Ngwa, J.S.; et al. Meta-analysis investigating associations between healthy diet and fasting glucose and insulin levels and
modification by loci associated with glucose homeostasis in data from 15 cohorts. Am. J. Epidemiol. 2013, 177, 103–115. [CrossRef]
13. Caporaso, J.G.; Lauber, C.L.; Walters, W.A.; Berg-Lyons, D.; Huntley, J.; Fierer, N.; Owens, S.M.; Betley, J.; Fraser, L.; Bauer, M.; et al.
Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012, 6, 1621–1624.
[CrossRef]
Microorganisms 2021, 9, 1674 13 of 14

14. Renaud, G.; Stenzel, U.; Maricic, T.; Wiebe, V.; Kelso, J. deML: Robust demultiplexing of Illumina sequences using a likelihood-
based approach. Bioinformatics 2015, 31, 770–772. [CrossRef]
15. Callahan, B.J.; McMurdie, P.J.; Rosen, M.J.; Han, A.W.; Johnson, A.J.; Holmes, S.P. DADA2: High-resolution sample inference
from Illumina amplicon data. Nat. Methods. 2016, 13, 581–583. [CrossRef]
16. Bolyen, E.; Rideout, J.R.; Dillon, M.R.; Bokulich, N.A.; Abnet, C.C.; Al-Ghalith, G.A.; Alexander, H.; Alm, E.J.; Arumugam, M.;
Asnicar, F.; et al. Author Correction: Reproducible, interactive, scalable and extensible microbiome data science using QIIME. Nat.
Biotechnol. 2019, 37, 852–857. [CrossRef]
17. QIIME2 Next-Generation Microbiome Bioinformatics Platform. Available online: https://qiime2.org (accessed on 20 Decem-
ber 2020).
18. Escapa, I.F.; Chen, T.; Huang, Y.; Gajare, P.; Dewhirst, F.E.; Lemon, K.P. New Insights into Human Nostril Microbiome from the
Expanded Human Oral Microbiome Database (eHOMD): A Resource for the Microbiome of the Human Aerodigestive Tract.
mSystems 2018, 3, e00187-18. [CrossRef] [PubMed]
19. Expanded Human Oral Microbiome Database (eHOMD). Available online: http://www.homd.org. (accessed on 20 Decem-
ber 2020).
20. Langille, M.G.; Zaneveld, J.; Caporaso, J.G.; McDonald, D.; Knights, D.; Reyes, J.A.; Clemente, J.C.; Burkepile, D.E.;
Vega Thurber, R.L.; Knight, R.; et al. Predictive functional profiling of microbial communities using 16S rRNA marker gene
sequences. Nat. Biotechnol. 2013, 31, 814–821. [CrossRef] [PubMed]
21. KO (KEGG ORTHOLOGY) Database of Molecular Functions. Available online: https://www.genome.jp/kegg/ko.html (accessed
on 15 December 2020).
22. Greengenes Database. Available online: http://greengenes.lbl.gov (accessed on 27 February 2021).
23. Hammer, O.; Harper, D.; Ryan, P. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontol.
Electron. 2001, 4, 1–9.
24. Aas, J.A.; Griffen, A.L.; Dardis, S.R.; Lee, A.M.; Olsen, I.; Dewhirst, F.E.; Leys, E.J.; Paster, B.J. Bacteria of dental caries in primary
and permanent teeth in children and young adults. J. Clin. Microbiol. 2008, 46, 1407–1417. [CrossRef] [PubMed]
25. Kameda, M.; Abiko, Y.; Washio, J.; Tanner, A.C.R.; Kressirer, C.A.; Mizoguchi, I.; Takahashi, N. Sugar Metabolism of Scardovia
wiggsiae, a Novel Caries-Associated Bacterium. Front. Microbiol. 2020, 11, 479. [CrossRef] [PubMed]
26. Kressirer, C.A.; Smith, D.J.; King, W.F.; Dobeck, J.M.; Starr, J.R.; Tanner, A.C.R. Scardovia wiggsiae and its potential role as a caries
pathogen. J. Oral Biosci. 2017, 59, 135–141. [CrossRef]
27. Eriksson, L.; Lif Holgerson, P.; Esberg, A.; Johansson, I. Microbial Complexes and Caries in 17-Year-Olds with and without
Streptococcus mutans. J. Dent. Res. 2018, 97, 275–282. [CrossRef] [PubMed]
28. Bradshaw, D.J.; Lynch, R.J. Diet and the microbial aetiology of dental caries: New paradigms. Int. Dent. J. 2013, 63, 64–72.
[CrossRef] [PubMed]
29. Xu, H.; Tian, J.; Hao, W.; Zhang, Q.; Zhou, Q.; Shi, W.; Qin, M.; He, X.; Chen, F. Oral Microbiome Shifts From Caries-Free to
Caries-Affected Status in 3-Year-Old Chinese Children: A Longitudinal Study. Front. Microbiol. 2018, 9, 2009. [CrossRef]
30. Kalpana, B.; Prabhu, P.; Bhat, A.H.; Senthilkumar, A.; Arun, R.P.; Asokan, S.; Gunthe, S.S.; Verma, R.S. Bacterial diversity and
functional analysis of severe early childhood caries and recurrence in India. Sci. Rep. 2020, 10. [CrossRef]
31. Shi, C.; Cai, L.; Xun, Z.; Zheng, S.; Shao, F.; Wang, B.; Zhu, R.; He, Y. Metagenomic analysis of the salivary microbiota in patients
with caries, periodontitis and comorbid diseases. J. Dent. Sci. 2021. [CrossRef]
32. Marsh, P.D.; Head, D.A.; Devine, D.A. Ecological approaches to oral biofilms: Control without killing. Caries Res. 2015, 49
(Suppl. 1), 46–54. [CrossRef]
33. Marchesan, J.T.; Byrd, K.M.; Moss, K.; Preisser, J.S.; Morelli, T.; Zandona, A.F.; Jiao, Y.; Beck, J. Flossing Is Associated with
Improved Oral Health in Older Adults. J. Dent. Res. 2020, 99, 1047–1053. [CrossRef]
34. David, L.A.; Materna, A.C.; Friedman, J.; Campos-Baptista, M.I.; Blackburn, M.C.; Perrotta, A.; Erdman, S.E.; Alm, E.J. Host
lifestyle affects human microbiota on daily timescales. Genome Biol. 2014, 15, R89. [CrossRef]
35. Corby, P.M.; Biesbrock, A.; Bartizek, R.; Corby, A.L.; Monteverde, R.; Ceschin, R.; Bretz, W.A. Treatment outcomes of dental
flossing in twins: Molecular analysis of the interproximal microflora. J. Periodontol. 2008, 79, 1426–1433. [CrossRef]
36. Socransky, S.S.; Haffajee, A.D.; Cugini, M.A.; Smith, C.; Kent, R.L., Jr. Microbial complexes in subgingival plaque. J. Clin.
Periodontol. 1998, 25, 134–144. [CrossRef] [PubMed]
37. André Kramer, A.C.; Pivodic, A.; Hakeberg, M.; Östberg, A.L. Multilevel Analysis of Dental Caries in Swedish Children and
Adolescents in Relation to Socioeconomic Status. Caries Res. 2019, 53, 96–106. [CrossRef]
38. Schwendicke, F.; Dörfer, C.E.; Schlattmann, P.; Foster Page, L.; Thomson, W.M.; Paris, S. Socioeconomic inequality and caries:
A systematic review and meta-analysis. J. Dent. Res. 2015, 94, 10–18. [CrossRef] [PubMed]
39. Costa, S.M.; Martins, C.C.; Pinto, M.Q.C.; Vasconcelos, M.; Abreu, M.H.N.G. Socioeconomic Factors and Caries in People between
19 and 60 Years of Age: An Update of a Systematic Review and Meta-Analysis of Observational Studies. Int. J. Environ. Res.
Public. Health 2018, 15, 1775. [CrossRef] [PubMed]
40. Belstrøm, D.; Holmstrup, P.; Nielsen, C.H.; Kirkby, N.; Twetman, S.; Heitmann, B.L.; Klepac-Ceraj, V.; Paster, B.J.; Fiehn, N.E.
Bacterial profiles of saliva in relation to diet, lifestyle factors, and socioeconomic status. J. Oral Microbiol. 2014, 6. [CrossRef]
41. Al Kawas, S.; Al-Marzooq, F.; Rahman, B.; Shearston, J.A.; Saad, H.; Benzina, D.; Weitzman, M. The impact of smoking different
tobacco types on the subgingival microbiome and periodontal health: A pilot study. Sci. Rep. 2021, 11, 1113. [CrossRef] [PubMed]
Microorganisms 2021, 9, 1674 14 of 14

42. Wu, J.; Peters, B.A.; Dominianni, C.; Zhang, Y.; Pei, Z.; Yang, L.; Ma, Y.; Purdue, M.P.; Jacobs, E.J.; Gapstur, S.M.; et al. Cigarette
smoking and the oral microbiome in a large study of American adults. ISME J. 2016, 10, 2435–2446. [CrossRef] [PubMed]
43. Al-Zyoud, W.; Hajjo, R.; Abu-Siniyeh, A.; Hajjaj, S. Salivary Microbiome and Cigarette Smoking: A First of Its Kind Investigation
in Jordan. Int. J. Environ. Res. Public Health 2020, 17, 256. [CrossRef] [PubMed]
44. Kanwar, A.; Sah, K.; Grover, N.; Chandra, S.; Singh, R. Long-term effect of tobacco on resting whole mouth salivary flow rate and
pH: An institutional based comparative study. European J. Gen. Dent. 2013, 2, 296–299. [CrossRef]
45. Andersson, G.; Warfvinge, G. The influence of pH and nicotine concentration in oral moist snuff on mucosal changes and salivary
pH in Swedish snuff users. Swed. Dent. J. 2003, 27, 67–75.
46. Hellqvist, L.; Boström, A.; Lingström, P.; Hugoson, A.; Rolandsson, M.; Birkhed, D. Effect of nicotine-free and nicotine-containing
snus on plaque pH in vivo. Swed. Dent. J. 2012, 36, 187–194. [PubMed]
47. Sullivan, R.; Heavey, S.; Graham, D.G.; Wellman, R.; Khan, S.; Thrumurthy, S.; Simpson, B.S.; Baker, T.; Jevons, S.; Ariza, J.; et al.
An optimised saliva collection method to produce high-yield, high-quality RNA for translational research. PLoS ONE 2020,
15, e0229791. [CrossRef] [PubMed]
48. Yano, Y.; Hua, X.; Wan, Y.; Suman, S.; Zhu, B.; Dagnall, C.L.; Hutchinson, A.; Jones, K.; Hicks, B.D.; Shi, J.; et al. Comparison
of Oral Microbiota Collected Using Multiple Methods and Recommendations for New Epidemiologic Studies. mSystems 2020,
5, e00156-20. [CrossRef]
49. Omori, M.; Kato-Kogoe, N.; Sakaguchi, S.; Fukui, N.; Yamamoto, K.; Nakajima, Y.; Inoue, K.; Nakano, H.; Motooka, D.; Nakano, T.;
et al. Comparative evaluation of microbial profiles of oral samples obtained at different collection time points and using different
methods. Clin. Oral Investig. 2021, 25, 2779–2789. [CrossRef]
50. Topkas, E.; Keith, P.; Dimeski, G.; Cooper-White, J.; Punyadeera, C. Evaluation of saliva collection devices for the analysis of
proteins. Clin. Chim. Acta 2012, 413, 1066–1070. [CrossRef] [PubMed]
51. Schulz, B.L.; Cooper-White, J.; Punyadeera, C.K. Saliva proteome research: Current status and future outlook. Crit. Rev. Biotechnol.
2013, 33, 246–259. [CrossRef] [PubMed]
52. Vogtmann, E.; Chen, J.; Kibriya, M.G.; Amir, A.; Shi, J.; Chen, Y.; Islam, T.; Eunes, M.; Ahmed, A.; Naher, J.; et al. Comparison of
Oral Collection Methods for Studies of Microbiota. Cancer Epidemiol. Biomarkers Prev. 2019, 28, 137–143. [CrossRef]
53. Peltzer, K.; Pengpid, S. Oral health behaviour and social and health factors in university students from 26 low, middle and high
income countries. Int. J. Environ. Res. Public Health 2014, 11, 12247–12260. [CrossRef]
54. Adair, P.M.; Pine, C.M.; Burnside, G.; Nicoll, A.D.; Gillett, A.; Anwar, S.; Broukal, Z.; Chestnutt, I.G.; Declerck, D.; Ping, F.X.;
et al. Familial and cultural perceptions and beliefs of oral hygiene and dietary practices among ethnically and socio-economicall
diverse groups. Community Dent. Health 2004, 21, 102–111.