Alterations of the nasal and oral microbiota in multiple sclerosis

Summary

Background

While changes in the gut microbiota have been reported in multiple sclerosis (MS), little is known about the nasal and oral microbiota, which are important modulators of the mucosal immune system. The gingival microbiota can drive systemic inflammation, the pharyngeal microbiota can lead to autoimmune-mediated neurologic diseases, and altered nasal bacterial toxigenic genes are reported in active MS.

Methods

We investigated nasal, gingival, and oropharyngeal microbiota from 69 MS to 40 healthy control subjects by 16S rRNA sequencing and identified site-specific microbiota differences in both the nasal and oral microbiota related to relapsing and progressive MS. We identified microbes associated with MS diagnosis, disease progression, disease modifying therapy, smoking, and anatomical site using linear regression analysis while controlling for confounding factors and subject demographics.

Findings

We identified site-specific microbiota differences in both the nasal and oral microbiota related to relapsing and progressive MS. Using baseline microbiota samples and clinical outcomes over time, we identified potential beneficial microbes, including Dolosigranulum pigrum in the nasal microbiota and Prevotella sp. in the oral microbiota, as well as potential detrimental Streptococcus sp. in the oral microbiota.

Interpretation

Taken together, our data suggest that nasal and oral microbiota are altered in MS, are linked to the disease course and provide an avenue to better understand MS pathogenesis and treatment.

Funding

This work was supported by the Nancy Davis Race to Erase MS Young Investigator Award (LMC), NIH/NINDS R01NS087226 (HLW), Water Cove Charitable Foundation (HLW), and the Clara E. and John H. Ware Jr. Foundation (HLW, LMC).

Research in context.

Evidence before this study

It is well established that the gut microbiome shapes systemic immune responses and may play a role in multiple sclerosis (MS). Microbiota at other mucosal surfaces, including the gingival microbiota, nasal microbiota, and oropharyngeal microbiota also modulate immunity and may play a role in neurologic diseases. Three studies have reported difference in the salivary microbiota by 16S sequencing and one study has investigated bacterial toxin genes from nasal samples by qPCR. However, differences in the nasal, gingival, and oropharyngeal microbiota have not yet been described in MS.

Added value of this study

We investigated the alterations in the nasal and oral microbiota in MS and associated with changes in 2-year clinical outcomes. We rigorously tested and controlled for potential confounders, including smoking, oral hygiene, respiratory infection, allergies, disease modifying therapy, and subject demographics. We found distinct bacterial changes in the nasal and oral microbiota in relapsing and progressive MS vs. healthy controls, including an enrichment of potential pathogens and a depletion of bacteria that have previously been shown to have a protective role in models of MS. We identified potentially beneficial microbes including Dolosigranulum pigrum in the nasal microbiota and Prevotella sp. in the oral microbiota and potentially detrimental Streptococcus sp. in the oral microbiota that was linked to worsening clinical disability score.

Implications of all the available evidence

Our study suggests that MS microbiota residing in three unique biologic niches in the gingiva and the upper respiratory tract are associated with the disease course in MS. Several of our putative protective or detrimental bacteria are consistent with studies in experimental animal models. Taken together, this provides evidence that the oral and nasal microbiota represent potential environmental contributors to MS pathogenesis.

Introduction

Multiple sclerosis (MS) is an autoimmune mediated demyelinating disease in which the immune system attacks myelin and damages nerve cell bodies in the central nervous system and is the most prevalent neurological disability affecting young adults in the US.^1,2 Both environmental and genetic factors contribute to MS pathogenesis.³ The microbiome is a complex ecological community that inhabits multiple anatomical locations of the human body and has a fundamental role in shaping systemic immune and metabolic homoeostasis.⁴ It is well-recognised that the gut microbiome modulates MS by affecting mucosal, systemic, and CNS autoimmunity,5, 6, 7 and transferring gut microbiota from MS subjects to mice worsens experimental autoimmune encephalomyelitis.^8,9 However, microbiota at other mucosal sites, including the nasal and oral microbiome, also modulate mucosal immune responses¹⁰ but have not been extensively characterised in MS and in relation to disease progression over time.11, 12, 13, 14, 15 We investigated alterations in the nasal and oral microbiome in MS (NOMMS) and how the nasal and oral microbiome relate to clinical outcomes, while controlling for factors such as smoking and oral hygiene known to affect the nasal and oral microbiota.

The nasal and oral microbiome reside in the proximity of mucosa-associated lymphoid tissue (MALT) located in the most cranial pharyngeal mucosa and interact closely with mucosal lymphocytes in the epithelium, mucosa, and lamina propria and aid in defence against invading pathogens.^16,17 The nasal microbiota, which contains Staphylococcus and Corynebacterium as predominant species, is highly variable including both normal microbiota and opportunistic pathogens that may worsen inflammation.¹⁸ For example, an increase of nasal superantigen enterotoxin A positive Staphylococcus aureus was associated with relapses in MS.¹⁹ The gingival microbiota includes microorganisms residing in the supragingival and subgingival areas and some gingival microbes drive Th17 cell expansion in an IL-6 and IL-23 dependent manner.^20,21 Porphyromonas gingivalis, the major etiological agent of chronic periodontitis, worsened EAE and induced proinflammatory responses of CNS glial cells.²² The oropharyngeal microbiota has more diverse bacterial communities compared to the nasal microbiota; key members include Streptococcus, Rothia, Veillonella, Prevotella, and Leptotrichia species.¹⁶ Streptococcus pyogenes oral infection has been associated with a robust Th17 cell-mediated immune responses and development of autoimmune diseases, including scarlet fever, glomerular nephritis, and paediatric autoimmune neuropsychiatric disorders associated with streptococcal infections (PANDAS).²³ While Epstein–Barr virus (EBV) infection is strongly associated with the development of MS,24, 25, 26 only a small proportion of EBV infected individuals develop MS, suggesting other environmental factors may be important. Streptococcus sanguinis, a resident of the oral microbiota, has been reported to induce EBV reactivation via H₂O₂ secretion thus exacerbating inflammatory symptoms in the CNS,²⁷ suggesting a link between bacteria in the oral microbiota and EBV infection. Taken together, these studies suggest that alterations in the nasal and oral microbiome shape the mucosal immunity and have the potential to influence neurologic and autoimmune diseases including MS.

To date, there have been limited studies of the nasal and oral microbiota in MS. There are four microbiome studies which investigate either the saliva samples or total oral swab samples.11, 12, 13^,15,28 However, important site-specific relationships between the microbiota and inflammation in the gingiva, oropharynx, and nasal mucosa, which may play unique roles in MS onset and progression has not been extensively studied. Here, we investigated the nasal and oral microbiome in relapsing and progressive MS and identified microbes associated with disease worsening. We assessed covariates that may affect the oral and nasal microbiota composition, including age, sex, body mass index (BMI), disease modifying therapy (DMT), smoking, and adjusted our linear regression models to identify altered nasal and oral microbes in MS. We found site-specific opportunistic pathogens linked to disease worsening as well as normal microbiota species linked to clinical improvement, consistent with experimental findings in independent studies. Our results suggests that the nasal and oral microbiota may be relevant for shaping the disease course in MS.

Methods

Ethics

Ethical approval was obtained from our institutional research board (IRB) at the Brigham and Women's Hospital under protocol number 2017P000748. All subjects provided written informed consent.

Sample collection

Study subjects were enrolled and given the choice to self-collect or have a study coordinator collect samples. DNA-free cotton swabs with wooden handles (Puritan, catalogue 25-806 1WC FNDA) were used throughout all sample collection process. All the swab samples were collected at our clinical centre either by a trained study coordinator or under the direct supervision of the study coordinator, immediately following instructions, and utilising a mirror to facilitate the collection. In addition, the sample collector (subject or coordinator) wore gloves to minimise any contamination from skin microbiota. Any difference in the sampling methods were minimised by the in-person one-on-one instruction and supervision. Nasal samples were collected by swabbing the inside of the nostril for 10 s on both the right and left side. Gingival samples were collected by swabbing, moving back and forth across the upper gingival surface for 10 s on both right and left sides while the upper lip was grasped and lifted away from touching the swab. Oropharyngeal samples were collected by swabbing the back of the throat for 1–5 s. Each swab was placed in a labelled collection tube and transported to the research lab within 2 h after collection. All collected samples were stored at −80 °C.

Recruitment, and demographic and clinical data collection

A total of 69 MS and 40 HC subjects were included in this study (Table 1). Participants were recruited through the MS clinic at Brigham and Women's Hospital (Boston, MA). Inclusion criteria for MS subjects followed the 2017 McDonald criteria²⁹ and aged 18–65. Inclusion criteria for HC subjects are (1) self-declared overall good health with no history of MS (2) age 18–65. Exclusion criteria for MS and HC subjects are (1) pregnancy (2) antibiotic use within the past 6 months (3) inhaled corticosteroid use within the past month (4) receipt of nasally delivered live, attenuated, cold-adapted influenza vaccine within the previous 28 days. Microbiota samples of MS and HC subjects were collected on the day of enrolment. Clinical outcomes of MS subjects were followed for up to two years after enrolment. Samples were collected from June 2017 to March 2019.

Table 1.

Subject demographics.

Characteristics	HC	RRMS	Prog MS
Subject (n)	40	53	16
Female subject (n)	22	38	10
Female subject (%)	55.0	71.7	62.5
Age (years, mean ± SD)	39.8 ± 12.2a	42.1 ± 9.8a	56.7 ± 5.2b
BMI (mean ± SD)	25.9 ± 5.8a	29.9 ± 9.9a	26.9 ± 7.6a
Race
White	36	50	13
Asian	1	0	0
South Asian	1	0	1
Black or African American	0	1	2
More than one race	1	1	0
Unknown or not reported	1	1	0
Ethnicity
Hispanic or latino	3	2	1
Not hispanic or latino	37	51	15
DMT
HC	40	0	0
Dimethyl Fumarate	0	18	1
Fingolimod	0	7	0
Glatiramer acetate	0	5	0
Interferon	0	5	1
Natalizumab	0	4	1
Teriflunomide	0	4	0
Anti-CD20	0	3	9
Anti-CD52	0	1	0
Methotrexate	0	0	2
Mycophenolate mofetil	0	0	1
>1 DMT (Mycophenolate mofetil & glatiramer acetate)	0	1	0
Untreated	0	5	1
Vitamin D supplement (%)	23.1	92.3	80
Smoking (%)	25.6	46.2	40.0
Smoking habits (n)
Cigarettes	8	20	6
Cigar	1	0	0
Tobacco	0	0	0
Marijuana	1	1	0
>1 Habits (Cigarettes & Marijuana)	0	2	0
>1 Habits (Cigarettes & Tobacco)	0	1	0
Oral health
Oral Disease (%)	35.9	26.9	31.3
Dental procedure visit (%)	87.2	92.3	75
Oral Hygiene Cumulative Score (mean ± SD)	9.46 ± 2.85a	10.29 ± 2.89a	10.50 ± 3.18a
Allergy symptoms (%)	12.8	11.5	0
Autoimmune disease other than MS (%)	2.6	13.5	12.5
Chronic sinusitis (%)	7.7	5.8	6.3
MS duration (years, mean ± SD)	–	12.5 ± 8.3a	21.9 ± 10.6b
EDSS (mean ± SD)	–	1.78 ± 1.31a	5.59 ± 1.87b

Abbreviations: HC, healthy control; MS, multiple sclerosis; RRMS, relapsing remitting MS; Prog MS, progressive MS; BMI, body mass index; DMT, disease modifying therapy; EDSS, expanded disability status scale.

Superscripts a and b were assigned for significant differences between groups, following the basic principles such that groups sharing at least one letter are not statistically different.

The following clinical and study participant data was collected. MS diagnosis, including relapsing remitting vs. progressive MS. Disability, as measured by the expanded disability status scale (EDSS), was determined by an MS physician at the time of sample collection and annually for up to two years following sample collection. Study subjects were asked to complete a demographic, health status and diet questionnaire. Questionnaires were administered using REDCap³⁰ and included i) demographic information including year of birth, race, ethnicity, sex, height, and weight, ii) supplements including vitamin D intake and probiotics, and iii) other factors that may contribute to inflammation including environmental and food-borne allergies, asthma, autoimmune diseases in addition to an MS diagnosis, recent respiratory tract infection, and diagnosed chronic sinusitis history. To address oral health, the questionnaire further included questions about dental history (bridges, crowns/caps, dentures, extractions, fillings, gum surgery, oral surgery, root canals, sealants, teeth whitening, and veneers), oral disease diagnoses (dental caries, gingivitis, periodontal disease oral cancer, dental abscess, reoccurring ulcers), oral hygiene practices regarding the frequency of brushing, flossing, using mouthwash, and times of visiting the dentist in a year, and smoking history of cigarettes, cigars, tobacco, or marijuana.

DNA extraction and 16S rDNA library preparation

Samples were stored at −80 °C. DNA was extracted using a QIAGEN DNeasy Powerlyzer Power Soil kit (QIAGEN). Amplicons spanning variable region V4–V5 of the bacterial 16S rRNA gene were generated with primers containing barcodes (515 F, 926 R) from the Earth Microbiome project^31,32 using HotMaster Mix (Quantabio 10847-708) and paired-end sequenced on an Illumina MiSeq platform at the Harvard Medical School Biopolymer Facility.

16S sequencing and statistical analysis

Paired-end reads were sequenced on the MiSeq and used 310 bp X 300 bp reads at a Harvard Biopolymers facility, and the sequence analysis was performed in QIIME2 v2023.09.³³ Demultiplexed forward and reverse reads were trimmed at base position 280 and 200 respectively and further denoised using DADA2 algorithm.³⁴ Minimum overlap for the forward and reverse reads to merge was set to 50 bp. Other parameters followed the default settings in the DADA2 plug-in.

Reads taxonomy were classified through a pre-trained Naïve Bayes QIIME2 classifier using the oral microbiome weighted SILVA138.1³⁵ as the reference database. Sequence quality control steps were performed as follows: (1) mitochondria, chloroplasts were removed as contaminants; (2) features with a minimum frequency below 10 were removed; (3) features occurred in less than 2 samples were removed; (4) samples contained less than 1000 features were removed. The feature table was cleaned down to 314 samples and 916 features (ASVs) with a total frequency of 2,712,131. For select ASVs of interest with incomplete identification, additional identification was performed using BLASTn, National Center for Biotechnology Information.³⁶

Mean relative abundance of taxa among groups were regenerated at the phylum (L2), genus level (L6), and species level (L7) respectively using R tidyverse (v1.2.1). 16S V4–V5 phylogeny was reconstructed based on the align-to-tree-mafft-fasttree pipeline. To compare all co-variates across the three anatomical sites, the ASV table was initially rarefied at 1000 sampling depth to calculate the β diversity dissimilarities across anatomical locations. Based on striking differences by anatomical site, the ASV table was next split into 3 sub-tables for nasal, gingival, and oropharyngeal samples and rarefied at sampling depths of 1000 (nasal), 2400 (gingival), and 3000 (oropharyngeal) respectively. Alpha and beta diversity indices of each anatomical site were calculated at the respective sampling depths afterwards.

Statistics

Alpha diversity metrics were calculated in QIIME2 and differences were tested in GraphPad Prism 10 (GraphPad, San Diego, USA). As the data was not normally distributed, we performed the nonparametric Kruskal–Wallis test and corrected for multiple comparisons using Benjamini–Hochberg (BH) method to control the False Discovery Rate (FDR).

Beta diversity differences were determined in QIIME2 using the Permutational Multivariate Analysis of Variance (PERMANOVA), Permutational Multivariate Analysis of Dispersion (permdisp) with multiple comparison correction using the Benjamini–Hochberg (BH) method. Results were plotted in GraphPad Prism 10. The Analysis of Dissimilarity (ADONIS) test, which supports multifactor (PERMANOVA) tests was used to identify the contribution of demographic, clinical, or oral hygiene factors to microbiome variation in QIIME2.³⁷ This included MS diagnosis, MS treatment, MS duration, EDSS, age, sex, BMI, race, ethnicity, smoking preference, oral health indicators (oral disease, oral hygiene, and dental procedure), vitamin D supplementation, autoimmune disease (not MS), most recent asthma attack, upper respiratory infection, allergy symptoms on the day of visit, and chronic sinusitis as the potential contributors to the microbiome variation. We individually tested the 20 covariates using the ADONIS test to evaluate their effect on site-specific microbiota beta-diversity, then further included a covariate in a multi-variate analysis if the covariate had a significant effect in the mono-variate analysis. We further integrated the results of nasal, gingival, and oropharyngeal ADONIS tests and plotted in RStudio (2023.3.1.446 with R version 4.4.1) using R packages tidyverse v1.2.1 and ggplot2 v3.5.1.

We tested the differential abundances of taxonomic features using the Microbiome Multivariable Association with Linear Models (MaAsLin2 v1.18.0) at the phylum, genus, species, and ASV levels to determine the potential multivariable associations between our clinical metadata and microbial meta-taxonomic features.³⁸ We applied the log-transformed general linear model (GLM), the logit-transformed GLM, and the Compound Poisson Linear Model (CPLM) and decided on the CPLM based on residual plots of representative features, current literature, and expectations of microbiota compositional data. Prior to testing, we filtered out the features that had <10% prevalence in either healthy control (HC), relapsing remitting MS (RRMS) or progressive MS (Prog MS). Several changes in bacterial relative abundance were identified using MaAsLin2, while controlling for statistically relevant confounders. i) Anatomical compositional differences in the nasal, gingival and oropharyngeal were identified while controlling for demographic, smoking, and oral health confounders. Further analyses were split by anatomical site. Compositional changes linked to ii) MS diagnosis and iii) DMT were identified while controlling for demographic, smoking, and oral health confounders. Similarly, bacterial iv) Compositional differences linked to disease progression as measured by EDSS change in two years and v) linked to smoking to were adjusted for age, sex, and BMI. Significant MaAsLin2 findings are reported as p value < 0.05 denoted with asterisks (∗) and MaAsLin q value (a false discovery rate approach using Benjamini–Hochberg (BH) correction for multiple comparisons) denoted with a plus sign (+) in the figures.

Role of funders

This work was supported by the Nancy Davis Race to Erase MS Young Investigator Award (LMC), NIH/NINDS R01NS087226 (HLW), Water Cove Charitable Foundation (HLW), and the Clara E. and John H. Ware Jr. Foundation (HLW, LMC). The funding sources did not have any role in the writing of the manuscript or the decision to submit.

Results

Demographics and clinical characteristics of study subjects

We investigated nasal, gingival, and oropharyngeal microbiome samples from 40 healthy controls (HC), 53 relapsing MS (RRMS) and 16 progressive MS subjects (Table 1). Progressive MS subjects had increased disease duration and increased Expanded Disability Status Scale (EDSS) scores compared to RRMS subjects, and increased age compared to RRMS and HC subjects (Figure S1), consistent with the finding that age is a risk factor for progressive MS.³⁹ Other characteristics regarding demographics and lifestyle were assessed by standardised questionnaires. We found no differences between groups in terms of BMI, sex, vitamin D supplementation, smoking, oral disease, dental visits, oral hygiene, allergy symptoms, other autoimmune diseases or chronic sinusitis (Figure S1 and Table 1).

Smoking has adverse effect on MS and impacts the microbiome composition,^40,41 therefore we assessed the smoking habits of our study participants (Tables 1 and 2). Of our healthy control group, 8 participants smoked cigarettes, 1 participant smoked cigars, and 1 participant smoked marijuana. In the RRMS group, 2 participants smoked cigarettes and marijuana, 1 participant smoked cigarettes and tobacco, 20 participants smoked cigarettes, and 1 participant smoked marijuana. In the progressive MS group, 6 participants smoked cigarettes. Because the data of smoking types in our work is uneven and sparse, we further classified participants into tobacco exposure group (cigarettes, cigars, or tobacco) or non-tobacco exposure group (marijuana or none) and examined the effects of tobacco exposure on microbiome variation.

Table 2.

Characterisation of subject smoker types.

Characteristics	HC	RRMS	Prog MS
Cigarettes	20%	40%	38%
Cigarettes & Tobacco	0%	2%	0%
Cigarettes & Marijuana	0%	4%	0%
Cigars	3%	0%	0%
Marijuana	3%	2%	0%
Not a smoker	73%	53%	56%
no data	0%	0%	6%

Dental health affects human microbiome in the upper respiratory tract, and inflammation in the gingiva can affect systemic inflammation.⁴² We assessed the oral health of our participants in three ways, including oral disease, dental visit for procedures, and the customised oral hygiene cumulative score based on daily practice (Tables 1 and 3). We surveyed the personal daily oral hygiene of our participants, and assessed the frequency of brushing, flossing, using mouthwash and visiting dentists. The four indices were scored as such: for visiting a dentist, 0—never, 1—less than once a year, 2—once a year, 3—twice a year, 4—more than twice a year; and for the frequency of brushing, flossing, and using mouthwash, 0—never, 1—once a month, 2—once a week, 3—once a day, 4—twice a day, 5—more than twice a day. The scores were combined to become our oral hygiene cumulative score. The maximum oral hygiene cumulative score is 19 and occurs if the participant visited the dentist more than twice a year (scored 4), brushed more than twice a day (scored 5), flossed more than twice a day (scored 5), and used mouthwash more than twice a day (scored 5). A score greater than 12 is consistent with CDC oral hygiene practice guidance.⁴³ Separate from the oral hygiene score, we also examined microbiome changes linked with oral disease including dental caries, gingivitis, periodontal disease, dental abscess, reoccurring ulcers, and oral cancer. Since dental procedures can increase the chance of infection, the procedures we assessed included fillings, crowns, root canals, extractions, bridges, oral surgery, sealants, gum surgery, teeth whitening, dentures, veneers, apicoectomy, and implant. Of note, more than 60% of our participants had more than one dental procedure.

Table 3.

Oral Health related measurements.

	HC (total n = 40)		RRMS (total n = 53)		Prog MS (total n = 16)
	n	%	n	%	n	%
Characteristics
Dental procedures
Bridges	1	3%	1	2%	3	19%
Crowns	12	30%	21	40%	7	44%
Dentures	0	0%	1	2%	1	6%
Extractions	15	38%	7	13%	6	38%
Fillings	25	63%	35	66%	10	63%
Gum surgery	3	8%	4	8%	2	13%
Oral surgery	18	45%	19	36%	5	31%
Root canals	12	30%	19	36%	7	44%
Sealants	4	10%	10	19%	0	0%
Teeth whitening	2	5%	6	11%	1	6%
Veneers	0	0%	3	6%	0	0%
Apicoectomy	0	0%	0	0%	0	0%
Implant	0	0%	1	2%	0	0%
Oral diseases
Dental caries	11	28%	4	8%	1	6%
Gingivitis	2	5%	4	8%	2	13%
Periodontal disease	2	5%	3	6%	3	19%
Oral cancer	0	0%	0	0%	0	0%
Dental abscess	0	0%	4	8%	0	0%
Reoccurring ulcers	0	0%	1	2%	0	0%
Oral hygiene (Score)
Brushing frequency
More than Twice a day (5)	4	10%	5	9%	1	6%
Twice a day (4)	27	68%	36	68%	9	56%
Once a day (3)	8	20%	11	21%	4	25%
Once a week (2)	1	3%	1	2%	0	0%
Once a month (1)	0	0%	0	0%	0	0%
Never (0)	0	0%	0	0%	1	6%
No data	0	0%	0	0%	1	6%
Flossing frequency
More than twice a day (5)	0	0%	3	6%	0	0%
Twice a day (4)	3	8%	4	8%	1	6%
Once a day (3)	13	33%	22	42%	8	50%
Once a week (2)	15	38%	12	23%	3	19%
Once a month (1)	2	5%	6	11%	1	6%
Never (0)	7	18%	6	11%	2	13%
No data	0	0%	0	0%	1	6%
Mouthwash frequency
More than twice a day (5)	0	0%	1	2%	1	6%
Twice a day (4)	0	0%	7	13%	2	13%
Once a day (3)	11	28%	4	8%	3	19%
Once a week (2)	6	15%	11	21%	3	19%
Once a month (1)	5	13%	3	6%	3	19%
Never (0)	18	45%	27	51%	3	19%
No data	0	0%	0	0%	1	6%
Dentist visiting frequency
More than twice per year (4)	3	8%	7	13%	1	6%
Twice per year (3)	21	53%	31	58%	9	56%
Once per year (2)	7	18%	10	19%	2	13%
Less than once per year (1)	7	18%	4	8%	3	19%
Never (0)	2	5%	1	2%	0	0%
No data	0	0%	0	0%	1	6%

Because vitamin D supplementation can modulate immunity and the microbiome,⁴⁴ and its level are associated with the risk of developing MS⁴⁵ and MS disease activity and progression,⁴⁶ we assessed vitamin D supplementation of our participants. Vitamin D supplementation was highly prevalent in RRMS (92.3%) and progressive MS (80.0%), while the HC group had a lower rate of vitamin D supplementation (23.1%). Unsurprisingly vitamin D supplementation was more frequent in MS compared to HC, as vitamin D is commonly prescribed in MS.⁴⁷ We also assessed recent allergy symptoms, other types of autoimmune diseases, and chronic sinusitis as these potentially shift the oral and nasal microbiome compositions (Tables 1 and 4).

Table 4.

Autoimmune Disease not MS, allergy, asthma, and upper respiratory infection information.

	HC	RRMS	Prog MS
Autoimmune diseases not MS
Autoimmune hepatitis	0%	2%	0%
CNS vasculitis	0%	0%	0%
Lupus	0%	0%	0%
Psoriatic arthritis	0%	2%	0%
Retinal vasculitis	3%	0%	0%
Rheumatoid arthritis	0%	2%	0%
Thyroid disease	0%	2%	6%
Type I diabetes	0%	0%	0%
Allergy Symptoms on day of Sample Donation	13%	11%	0%
Last attack of asthma
Unsure or never	98%	89%	81%
Over a year ago	0%	9%	19%
Within the last year	0%	0%	0%
Within the last month	0%	2%	0%
Within the last week	0%	0%	0%
Last attack of upper respiratory infection
Unsure or never	18%	15%	19%
Over a year ago	25%	17%	38%
Within the last year	45%	57%	38%
Within the last month	8%	4%	6%
Within the last week	3%	8%	0%

Microbiota community differences in nasal and oral microbiota

We determined that anatomical location is the biggest contributor for the microbiome variance among the host and environmental variables by the analysis of distances, ordinating and non-parametric inference by similarity (ADONIS) test and based on weighted UniFrac distance (Fig. 1a). We visualised microbiota community differences by disease status and anatomical site using principal coordinate analysis of weighted UniFrac distances (Fig. 1b, Figure S2). Because we found no differences in β-diversity between MS and healthy controls (Figs. 1a and 2a–b, Figure S2), we pooled all microbiome samples together for the analysis of anatomical sites. We found the greatest separation between the nasal and oral samples, and a significant but lesser separation between gingival and oropharyngeal samples (Fig. 1b, q < 0.05). The significant difference of β-diversity resulted from both between group distance (Fig. 1c–e, PEMANOVA, q < 0.05) and within group sample dispersion (Fig. 1f, permdisp q < 0.05), suggesting both a difference in composition and ecological variance in our participants (Fig. 1c–f).

Fig. 1.

Nasal, gingival, and oropharyngeal microbiota differences. (a) Effect of host factors on total microbial β-diversity including nasal, gingival, and oropharyngeal microbiome was measured using ADONIS test based on weighted UniFrac distance matrix. (b) Principal Coordinate analysis (PCoA) of microbiome samples based on weighted UniFrac distance matrix showing significantly different clustering between nasal, gingival, and oropharyngeal swabs. (c–e) Box-whisker plots of pairwise PERMANOVA (n = 999) based on weighted UniFrac distance indicating difference of between group distances. (f) Box-whisker plot of pairwise permdisp (n = 999) based on weighted UniFrac distance indicating difference of data dispersion. For box-whisker plots, q values (p adjusted for false discovery rate using Benjamini & Hochberg method) was matched with asterisks as follows: q < 0.001, ∗∗∗; q < 0.01, ∗∗, q < 0.05, ∗.(g) Effect of host factors on site-specific microbial β-diversity was measured using ADONIS test based on weighted UniFrac distance matrix. Factors has a p value < 0.05 were indicated by asterisks (∗). URI, upper respiratory infection. Nasal sample size: n = 107, gingival: n = 106, oropharyngeal: n = 107.

Fig. 2.

Microbiome differences examined across disease diagnoses in nasal, gingival and oropharyngeal samples. (a) Principal Coordinate analysis (PCoA) of microbiome samples based on weighted UniFrac distance matrix showed no significant difference across RRMS, Prog MS, and HCs. (b) Box-whisker plots of pairwise PERMANOVA (n = 999) based on weighted UniFrac distance indicated no difference of between group distances. (c) alpha diversity metrics for observed features (richness), Shannon's diversity, Pielou's evenness, and Faith's phylogenetic diversity. Difference of alpha diversities between MS diagnosis were tested by Kruskal–Wallis test followed by Benjamini–Hochberg test for multiple comparison corrections. ∗ denotes q < 0.05.(d–f) Microbial species altered in RRMS vs. HC, Prog MS vs. HC and Prog MS vs. RRMS described at ASV (amplicon sequence variant) level. Bacterial relative abundance data were regressed to MS diagnoses (3-level categorical data: RRMS, Prog MS, HC) using CPLM model without further normalisation and transformation, while adjusted for age, sex, BMI, smoking, and oral health indices. ∗ denotes MaAsLin p < 0.05, + denotes MaAsLin q < 0.05. Abbreviations: N, nasal, G, gingival, O, oropharyngeal. HC sample size n = 40, RRMS n = 53, Prog MS n = 16.

We then separately assessed the nasal, gingival, and oropharyngeal microbiome and examined the degree to which twenty host and environmental variables (see Methods) contributed to the microbiome variance using ADONIS test (Fig. 1g). β-diversity used in ADONIS test were calculated based on weighted UniFrac distances to account for taxonomic relative abundance as well as phylogenetic relatedness. For the nasal microbiota, sex and allergy symptoms significantly altered microbiota composition. For the gingiva, autoimmune diseases other than MS and ethnicity had a small but significant effect on the microbiota. For the oropharynx, autoimmune diseases other than MS and a history of recent upper respiratory infections (URI, p = 0.07) contributed to microbiota alterations. Taken together, these data indicate that both host factors and sources of inflammation shape microbiota community structure in a site-specific manner.

We next examined changes in bacterial composition at the phylum, genus, and species level and found clear differences across the nasal, gingival, and oropharyngeal samples (Figure S3). We identified microbiota differences using MaAsLin model and revealed major different phyla in each anatomical site and the corresponding driving species (Figures S4 and S5). The nasal microbiota had increased Actinobacteriota vs. the gingival and oropharyngeal microbiota while the gingival microbiota had increased Firmicutes vs. the nasal and oropharyngeal microbiota, and the oropharyngeal microbiota had increased Bacteroidota vs. the nasal and gingival microbiota (Figure S4a–c, MaAsLin q < 0.05, denoted with a “+”). At the genus and species level (Figures S4d–f and S5), we found expected nasal microbiota, including Corynebacterium species and Staphylococcus warneri, gingival bacteria Actinomyces viscosus, Streptococcus infantis, and Haemophilus parainfluenzae, and oropharyngeal bacteria Streptococcus salivarius, Prevotella histicola, and Veillonella dispar. These data identified niche-specific bacteria while adjusting for age, sex, BMI, smoking, and oral health indices, which represent unique communities of mucosal associated microbes.

MS microbiota exhibits decreased oropharyngeal diversity

Because of the large impact of anatomical site on microbiota composition, we next examined MS microbiota alterations separately in the nasal, gingival, and oropharyngeal microbiota. We examined differences in the bacterial overall variance (β-diversity) between HC, RRMS and progressive MS within each anatomical site using the weighted UniFrac distance index (Figure S6). The UniFrac distances within and between HC, RRMS, and progressive MS (Prog MS) groups showed no differences in terms of dispersion (within group analysis by permdisp) as well as between group separation (between group analysis by PERMANOVA) in the three anatomical sites indicating that microbial communities of MS do not differ at the broad community level from HC, consistent with our study of the gut microbiota.⁴⁸

We further assessed four α-diversity indices within each anatomical site. We found significant changes in oropharyngeal α-diversity by diagnosis (Fig. 2c). RRMS has a reduced oropharyngeal α-diversity compared to HC as revealed by three indices including the observed feature, Shannon diversity, and Faith phylogenetic diversity. Prog MS also showed decreased Shannon diversity compared to HC in oropharyngeal microbiota. There were no differences in nasal and gingival α-diversity metrics. Taken together, these data suggest that fewer types of bacteria inhabit the oropharyngeal MS microbiome vs. oropharyngeal HC microbiome.

Nasal and oral microbiota altered in MS

We examined the nasal and oral microbiota changes in MS at the genus level (Figure S7a–c) and at the ASV level (Fig. 2d) while adjusting for age, sex, BMI, smoking, and oral health as covariates. We found the greatest number of changes in the oropharyngeal microbiome between MS and healthy controls, followed by the gingival and nasal microbiomes. In the oropharyngeal microbiome, Veillonella ASV_03, Alloprevotella ASV_24, and Rothia ASV_28 were increased in RRMS compared to HC, consistent with other reports of elevated Veillonella.^12,15 The elevated Rothia ASV_28 in the oropharynx of RRMS vs. HC (Fig. 2d) was most closely related to Rothia mucilaginosa (Identity 99.2% similar), and next most closely related to Rothia terrae (98.4% similar) and Rothia aeria strains (98.1% similar) (Table S1). Of note, Rothia species have divergent impacts in different diseases49, 50, 51 and we found changes only at the ASV but not genus level (Figure S7a). S. infantis ASV_09, and Haemophilus ASV_14 were increased in Prog MS compared with RRMS and HC subjects (Fig. 2e–f), and Haemophilus has been reported to be increased in MS oral microbiota.^11,13 Prevotella species (Prevotella nigrescens ASV_30, Prevotella ASV_13, and Prevotella pallens ASV_11) and Selenomonas sputigena ASV_10 were decreased in RRMS vs. in HC (Fig. 2d). In addition, Prevotella ASV_87 was decreased in Prog MS vs. HC (Fig. 2e). When comparing progressive MS vs. RRMS (Fig. 2f), we found increased S. infantis ASV_09, Prevotella ASV_13, and Hemophilus ASV_14, and decreased Haemophilus ASV_41 in progressive MS. Haemophilus ASV_14 had 100% similar sequence identity to Haemophilus haemolyticus (Table S1), while Haemophilus ASV_41 had a 100% sequence identity to H. parainfluenzae (Table S1).

In the gingiva, Tannerella ASV_18, Lautropia ASV_26, and P. pallens ASV_27 were increased in RRMS compared to HC (Fig. 2d); Tannerella and Lautropia have been associated with other neurologic diseases.^52,53 Surprisingly, we found decreased abundance of H. parainfluenzae ASV_12 in both RRMS and Prog MS compared to healthy controls (Fig. 2d–e). In addition, Alloprevotella ASV_37, Campylobacter showae ASV_29, and Fusobacterium ASV_08 were decreased in RRMS vs. HC (Fig. 2d), while Clostridia UCG-014 was decreased in Prog MS vs. HC (Fig. 2e). Alloprevotella was shown alleviating EAE,⁵⁴ and Fusobacterium depletion was associated with MS relapse risk.⁵⁵ When comparing progressive MS vs. RRMS, we found increased Fusobacterium ASV_08 and decreased Paludibacteraceae ASV_90 in progressive MS (Fig. 2f). In the nasal microbiota, we found only one significant change. Enterococcus durans ASV_25 decreased in RRMS vs. HC (Fig. 2d), which was shown to reduce EAE severity.^56,57

Oral and nasal microbiota associated with MS disease progression

We next examined the association of microbiome with MS disease progression as defined by 2-year change in the EDSS from the microbiome collection baseline (Fig. 3a). Using multivariate linear regression, we identified microbiota at the genus level (Fig. 3b) and at the ASV level (Fig. 3c) that are linked with 2-year EDSS change, adjusted for age, sex, and BMI as covariates for all three anatomical sites. As MS is three times common in women vs. men,^58,59 we further performed sex-stratified analyses and identified bacterial genera and ASVs associated with 2-year-change in EDSS in males (Figure S8a, c; n = 12) and females with MS (Figure S8b, d; n = 38).

Fig. 3.

Microbiota associated with disease progression. (a) Distribution of EDSS change data binned by baseline EDSS. (b) Microbial alteration related with MS disease progression in 2 years described at the genus level. (c) Microbial alteration related with MS disease progression in 2 years described at ASV (amplicon sequence variant) level. Bacterial relative abundance data were regressed to EDSS change in 2 years (numeric data) using CPLM model without further normalisation and transformation, while adjusted for age, sex, BMI. ∗ denotes MaAsLin p < 0.05, + denotes MaAsLin q < 0.05.

In the oropharynx of all MS subjects (males and females), we found that Rothia ASV_66, Actinomyces ASV_61, Prevotella ASV_59, and Capnocytophaga sputigena ASV_35 were associated with disease worsening (Fig. 3c). Recent studies have reported elevated levels of Actinomyces in MS compared to healthy controls in both the oral¹² and gut microbiota.⁶⁰ We found that Veillonella ASV_78 and Butyrivibrio ASV_73 were associated with clinical improvement (Fig. 3c). In our sub-analysis, we found that the Butyrivibrio genus and Butyrivibrio ASV_73 was associated with clinical improvement in both males and females (Figure S8a and d). Butyrivibrio gut species are butyrate producers which are correlated with lower MS severity in Prog MS.⁶¹

In the gingiva (Fig. 3c), we found two groups of known invasive oral microbes linked with disease worsening, including the sanguinis-group streptococci⁶² (S. sanguinis ASV_58, Streptococcus parasanguinis ASV_64, and S. gordonii ASV_17, see Table S1) and HACEK bacteria known to cause infective endocarditis (Cardiobacterium hominis ASV_51 and Abiotrophia defectiva ASV_63, see Table S1).⁶³ We found S. infantis ASV_62 is linked with disease worsening in males (Figure S8c) and S. sanguinis ASV_58 is linked with disease worsening in females (Figure S8d). Consistent with our findings in the oropharynx, gingiva-residing A. viscosus ASV_42 was associated with disease worsening. Additional bacteria linked with disease worsening included opportunistic pathogens Leptotrichia hongkongensis ASV_56,⁶⁴ and periodontal pathogen Lautropia ASV_26, and members of the normal oral microbiota Neisseria (ASV_65 & ASV_60) and Veillonella (ASV_67). We found Prevotella (ASV_79, ASV_30, ASV_70, ASV_68) associated with disease improvement, a genus that reduces CNS autoimmunity^65,66 and has been reported to increase with MS disease modifying therapy.⁶⁷ We further identified Prevotella ASVs using BLASTn and found that ASV_30 is most closely related to P. nigrescens (100% identity), ASV_68 is most closely related to P. denticola (100% identity), ASV_70 is most closely related to P. jejuni (100% identity), and ASV_79 is most closely related to Prevotella oris (namely Segatella oris, 99.73% identity) (Table S1). Interestingly, Prevotella ASV_70 and ASV_79 association with clinical improvement were prominent in females not males (Figure S8d). Leptotrichia buccalis ASV_74 and Leptotrichia ASV_72 were associated with disease improvement, which have previously been found increased in the MS oral microbiota,¹⁵ and decreased in Sjogren's syndrome.⁶⁸ Unexpectedly, Treponema lecithinolyticum ASV_83 and Treponema socranskii ASV_76 were associated with disease improvement (Fig. 3c) and only prominent in females (Figure S8d), and these two species are related with aggressive inflammation in periodontitis.⁶⁹ We identified Fusobacterium animalis ASV_69 (Table S1) associated with disease improvement. Furthermore, normal oral microbiota Streptococcus anginosus ASV_86, and Parvimonas micra ASV_71, Eubacterium brachy group ASV_75 was associated with disease improvement (Fig. 3b–c and Table S1), especially in the females (Figure S8b and d).

In the nasal microbiota, D. pigrum ASV_84 was associated with clinical improvement (Fig. 3c and Figure S8d), which is a human nasal microbial resident associated with nasal health and homoeostasis.⁷⁰ We also found nasal Anaerococcus octavius ASV_53 associated with disease worsening (Fig. 3c and Figure S8d), which has recently been reported as a cause of bacteraemia.⁷¹

Effect of disease modifying therapy on oral and nasal microbiota

We further examined nasal and oral microbiota changes affected by MS disease modifying therapy to determine whether the MS microbiota changes we observed were related to DMT. We first examined changes in β-diversity in DMT groups that had more than 5 patients/treatment and found no difference in β-diversity between anti-CD20, dimethyl fumarate, interferon, glatiramer acetate, fingolimod, and natalizumab compared with untreated patients and HCs (Fig. 4a). This suggested that DMTs had minimal effects on the nasal and oral microbiota structure. We then examined whether specific microbiota was altered based on DMT, while controlling age, sex, BMI, smoking, and oral hygiene score as covariates. We found no changes in the oropharyngeal and gingival microbiota at the ASV level. In the nasal microbiota at the ASV level, Corynebacterium propinquum was increased in anti-CD20 treated subjects vs. untreated MS, and Finegoldia was increased in interferon and natalizumab treated subjects (Fig. 4b). At the genus level, we found increased nasal Finegoldia in natalizumab treated subjects vs. untreated (Figure S9a), which is consistent with the ASV level results. Natalizumab also increased nasal Prevotella and Anaerococcus genus and oral Lactobacillus and Atopobium (Figure S9a, c), while decreased oral Gemella, Treponema, Alloprevotella, and Neisseria (Figure S9b–c). Furthermore, natalizumab, dimethyl fumarate, and fingolimod were associated with the most microbial genera changes across nasal, gingival, and oropharyngeal samples (Figure S9a–c), while the changes were treatment dependent. Overall, while changes were detected between DMT-treated and untreated subjects, these were minimal compared to alterations in MS.

Fig. 4.

Microbiota alterations subject to disease modifying therapies (DMTs). DMT groups were included into analysis only when having 5 participants at least. (a) Principal coordinate analyses based on beta diversities indicated DMT not related to microbial alterations in nasal and oral microbiota of MS participants. (b) Microbial alteration related with DMT described at ASV (amplicon sequence variant) level. Bacterial relative abundance data were regressed to DMTs (categorical data: anti-CD20, dimethyl fumarate, interferon, glatiramer acetate, fingolimod, natalizumab, untreated and HC) using CPLM model without further normalisation and transformation, while adjusted for age, sex, BMI, smoking, and oral health indices ∗ denotes MaAsLin p < 0.05, + denotes MaAsLin q < 0.05. Sample size HC n = 39, anti-CD20 n = 12, DMF n = 18, Interferon n = 6, natalizumab n = 5.

Effect of tobacco use on oral and nasal microbiota

Because tobacco use can alter the microbiota and increase MS risk, we then analysed the effect of tobacco exposure on the microbiome in individuals who smoked cigarettes, cigars, and/or other tobacco products vs. non-tobacco users. We identified altered microbes while adjusting for age, BMI, and sex covariates (Figure S10). We found at the ASV level, Actinomyces of the oropharyngeal and A. octavius of the nasal were positively associated with tobacco use (Figure S10b), which were also associated with disease progression as measured by worsening EDSS (Fig. 3c). This suggests that smoking may select for microbes that contribute to MS progression. Gingival Parvimonas was significantly elevated with tobacco exposure at both the genus and the ASV level (Figure S10), however, was not associated with disease progression or severity, suggesting that the effect of tobacco exposure on this particular microbe is independent of clinical outcomes. Furthermore, we identified decreased Neisseria in gingival (Figure S10b, ASV_60) and oropharyngeal samples (Figure S10b, ASV_65) in tobacco users, consistent with a previous report.⁷²

Discussion

In this study, we investigated the structural and compositional changes in the nasal, gingival, and oropharyngeal microbiota to extend our work on the gut microbiota in MS.^67,73,74 The nasal and oral microbiota closely interact with the mucosal surfaces of the upper respiratory tract. In health, the nasal and oral microbiota shape mucosal immune homoeostasis, educate the immune response, and help defend against pathogens. Conversely, in disease, the nasal and oral microbiota may drive systemic inflammation. Therefore, we asked whether the nasal and oral microbiota were altered in MS and associated with disease worsening over time. Using well-controlled co-variates (e.g., smoking, oral hygiene, DMT, and demographics), and a longitudinally tracked MS population, we identified nasal and oral microbes that are altered in MS and are linked to clinical outcomes. Several of these beneficial or detrimental associations are consistent with independent animal studies, discussed below.

The contribution of the microbiota to MS pathogenesis was demonstrated in a germ-free experimental autoimmune encephalomyelitis (EAE) SJL/J mice, which did not develop paralysis.⁷⁵ Early studies focused on the gut microbiota and showed that transfer of gut microbiota from individuals with MS to animal models of EAE worsened disease, demonstrating a causal role.^8,9 We and others have identified changes in the gut microbiota of individuals with MS,^{8,9,61,67,73,74,}76, 77, 78, 79, 80, 81, 82, 83, 84, 85 including increased Methanobrevibacter⁶⁷ and Akkermansia^61,67,73 and reduced butyrate-producers such as Butyricimonas⁶⁷ and Blautia.^61,73 Furthermore, the gut microbiota is altered in progressive MS, and is linked to differential clinical outcomes.^67,73,74,86

While there have been many studies of the gut microbiome in MS, there are fewer studies on the nasal and oral microbiota. A case-controlled study of the salivary microbiota alterations in MS reported depletion of Aggregatibacter and Streptococcus and enrichment of Leptotrichia and Fusobacterium which related to inflammation.¹⁵ The depletion of Streptococcus was also reported in MS patients and their CIS monozygotic twin compared to the healthy control group.¹¹ The saliva microbiome obtained from another female MS cohort also revealed increased Fusobacterium and eight other genera including Bacteroides, Porphyromonas, Prevotella, Veillonella, Actinomyces, Propionibacterium and Bifidobacterium.¹² Notably, Prevotella administered intratracheally ameliorated EAE severity in rats,⁸⁷ suggesting that alterations detected in the MS salivary (oral) microbiota have functional roles in modulating MS pathogenesis. In addition, the three studies on the salivary oral microbiota^11,12,15 in MS and another on the oral swabs of MS¹³ had few consistent microbiota changes at the genus level, which could be due to variations in sample type and sequencing techniques across studies and regional differences. Our study provides novel findings related to site specific oral and nasal microbiota communities, linked to disease subtype and progression.

Since more than 80% of individuals with MS have a relapsing-remitting course, we first defined the nasal and oral microbiota alterations in RRMS vs. HC. We found reduced E. durans in RRMS nasal samples, which can be found in fermented food products (cheese).⁵⁶ Given that E. durans has mucosal anti-inflammatory properties,⁵⁶ we postulate nasal E. durans may be beneficial for MS. E. durans has been recognised as a potential probiotic species and has been shown to decrease disease severity of models of experimental autoimmune encephalomyelitis (EAE)⁵⁷ and experimental colitis.⁸⁸ This may be linked to a decrease in inflammation as administering E. durans increases Treg cells and IL-10 production and downregulates pro-inflammatory cytokines expression (IL-6, IL-1beta, and TNF-alpha) in colitis.^88,89 E. durans produces butyrate that induces anti-inflammatory effects in an in vitro model of a compartmentalised Caco-2/leukocyte coculture model.⁹⁰ E. durans has an inhibitory effect on NF-kB activation in intestinal epithelial cells.⁹¹ While most studies of E. durans focus on the gut microbiome, one study reported that oral administration of E. durans ameliorated EAE,⁵⁷ and another study systematically evaluated the safety of E. durans oral supplementation as a probiotic for human and animal use.⁹²

In the gingiva, we found increased Tannerella, Lautropia, and Prevotella pallens in RRMS. Tannerella sp. triggers inflammatory responses and is related to oral diseases,⁹³ which related to cognitive decline in two AD case-cohort studies,^53,94 and Lautropia was reported to be inversely associated with stroke.⁵² In addition, we found decreased gingival Alloprevotella, C. showae, Fusobacterium, and H. parainfluenzae in RRMS. Administration of an oral strain of Alloprevotella rava was shown to alleviate EAE symptoms,⁵⁴ indicating a potential beneficial role of Alloprevotella for MS. A study of the gut microbiota in paediatric MS found that Fusobacteria depletion was associated with increased risk of relapse,⁵⁵ also suggesting a beneficial role for Fusobacterium across anatomical sites. Little is known about H. parainfluenzae in MS, but Haemophilus influenzae, a close related species of H. parainfluenzae, is related to MS pathology.⁹⁵

In RRMS vs. HC oropharyngeal samples, we observed reduced microbial diversity and evenness, increased Veillonella, Alloprevotella and Rothia and decreased S. sputigena, P. nigrescens, Prevotella pallens, and Prevotella ASV_13. Although oral Prevotella species are less studied in MS, the Prevotella gut species, P. histicola and P. copri increased with DMT including Copaxone and beta-interferon in MS.^65,67,80 Furthermore, P. histicola suppressed EAE by reducing CNS inflammation and demyelination.⁹⁶ P. pallens is a core member of the oral microbiota with no connection to oral disease⁹⁷ and was reported to be decreased with hyposalivation in a cohort of female subjects with Sjogren's syndrome.⁹⁸ P. nigrescens was reported to be linked with TLR2 activation mechanism and the induction of Th17 responses, chronic oral inflammation, and periodontitis.⁹⁹ P. melaninogenica was also reported in related to TLR2 stimulation in the upper airway,¹⁰⁰ which was identified as increased in male oral microbiota in our study regardless of MS diagnosis and other demographics (Figure S7e, ASV_119). In addition, S. sputigena is associated with tooth cavities.¹⁰¹

Rothia oral species can have multi-faceted roles and may depend on the species and the anatomical locations, and the specific disease in discussion. Rothia dentocariosa, associated with periodontal inflammatory disease, was reported to induce TNFα production by a TLR2-dependent mechanism,⁵¹ while R. mucilaginosa isolated from cystic fibrosis negatively correlated with inflammatory cytokines in chronic obstructive pulmonary disease, and the administration of R. mucilaginosa lowered LPS-mediated lung inflammation and histopathologic damage.⁵⁰ Rothia also negatively correlated with pro-inflammatory markers and associated with pulmonary rehabilitation in chronic obstructive pulmonary disease,⁴⁹ indicating a beneficial role. We found detrimental associations of different Rothia ASVs, including an elevation in Rothia ASV_28 in the oropharynx of RRMS vs. healthy controls (Fig. 2d) and Rothia ASV_66, associated with disease worsening (Fig. 3c), both of which were most closely related to R. mucilaginosa (ASV average nucleotide identity 99.2% and 99.73% similar, respectively).

In progressive MS, subjects have a gradual deterioration of function over time¹ with 65% of RRMS patients developing secondary progressive MS.¹⁰² We investigated which oral and nasal microbiota were unique or shared in relapsing vs. progressive and found a reduced microbial diversity in the oropharyngeal samples in progressive MS vs. HC, which is similar to α-diversity changes in relapsing MS vs. HC. Examining changes in oropharyngeal composition in progressive MS vs. HC, we found increased S. infantis, Haemophilus, and Lactobacillales P5D1-392 and decreased Prevotella. Consistent with our findings, increased Streptococcus and Haemophilus has been observed in the oral microbiota of untreated MS subjects vs. HC.¹³ Intriguingly, several Streptococcus and Haemophilus species were depleted following B-cell treatment, raising the possibility that anti-B cell therapy may select for beneficial oral microbiota.¹³ Our finding of increases in both S. infantis and Haemophilus in MS could be indicative of proinflammatory conditions in the oropharynx, given that they both cause respiratory and systemic infections.^103,104 In contrast to the oropharynx, we found decreased H. parainfluenzae and Clostridia UCG-014 in the gingiva Progressive MS vs. HC.

Next, we asked which microbiota differed between RRMS and progressive MS. In gingival samples, we found increased Fusobacterium and decreased Paludibacteraceae F0058, and in the oropharynx, we found increased S. infantis, Prevotella ASV_13, and Haemophilus ASV_14 and decreased Haemophilus ASV_41 in the oropharynx in Progressive MS vs. RRMS. Examining which findings were shared in both RR and progressive MS, we found Prevotella species was consistently decreased vs. HC in the oropharynx and H. parainfluenzae ASV_12 consistently decreased in the gingiva regardless of MS subtypes, whereas all other nasal and oral microbial changes were unique to RRMS vs. HC and progressive MS vs. HC. This is consistent with our previous work on the gut microbiome,⁷³ in which we found both shared and some unique microbiota in RR and progressive MS.

Examining sex-specific differences, we found Haemophilus species are more abundant in female vs. male nasal and oral microbiota (Figure S7d), while related to disease worsening in males but not females (Figure S8c). Eight Haemophilus ASVs were associated with females in oral samples regardless of MS diagnoses and other demographics (Figure S7e). Haemophilus has been shown increased in MS gut microbiota,⁷⁶ associated with MS pathogenesis⁹⁵ and with Guillain-Barre syndrome.¹⁰⁵ Haemophilus was reported to increase inflammation¹⁰⁶ and worsen EAE.¹⁰⁷ However, no existing study mentioned Haemophilus has larger impact in females MS or in male MS.

We next identified microbes associated with disease worsening or improvement. We found that nasal D. pigrum was associated with improvement, which has been suggested to be a beneficial species in the human upper respiratory tract and a nasal probiotic candidate for acute respiratory infection prevention and treatment.¹⁰⁸ Furthermore, D. pigrum has been shown to inhibit S. aureus.^109,110 Of note, S. aureus strains that produce the superantigens Enterotoxin A and B are linked with MS exacerbations and worsen EAE in mice,¹¹¹ suggesting that beneficial associations with D. pigrum may be linked to intra-species microbiota interactions. In addition, we identified nasal A. octavius as being associated with MS disease worsening, which has been described in bacteraemia, but have not been explored in the context of MS or other neurologic diseases. Increased D. pigrum and decreased A. octavious were further associated with disease worsening in our female cohort but not male cohort (Figure S8d).

In gingiva samples, we found that P. oris ASV_79, P. nigrescens ASV_30, P. jejuni ASV_70, and P. denticola ASV_68 were associated with disease improvement, further supporting a potential beneficial role of oral Prevotella species. We found that gingival S. sanguinis, S. parasanguinis, S. gordonii, and A. defectiva were associated with disease worsening. Oral S. sanguinis was reported to increase EBV reactivation potential, which could contribute to MS.²⁷ S. parasanguinis is suggested as a major coloniser in the oral microbiota related with dental plaque formation¹¹² and is reported to be elevated in the RRMS gut microbiome compared to healthy controls.⁸⁵ Kageyama et al. analysed 144 pairs of oral and gut microbiota of human adults using saliva and stool samples and found that S. parasanguinis likely translocated from the oral cavity to the gut.¹¹³ We also found that one Streptococcus ASV_86, closely related to the S. anginosus group, was associated with disease improvement (especially in females), which was unexpected as this group of bacteria are the most common etiologic agents of brain abscesses.¹¹⁴

We identified gingival Lautropia ASV_26 associated with disease worsening, both in males and in females. Lautropia was also increased in RR vs. HC and increased in untreated MS vs. HC. Lautropia species was also found reduced in subjects with tobacco exposure which is consistent with works in oral microbiota and brain health.^115,116 Lautropia is a Gram-negative commensal bacteria in the upper respiratory tract. While sequence BLAST results showed that Lautropia ASV_26 had 100% sequence identity to Lautropia mirabilis and Lautropia ASV_52 had 100% sequence identity to Lautropia dentalis, interventional studies with oral administration is needed to examine the role of Lautropia on MS clinical outcomes.

We also identified gingival Fusobacterium ASV_69 associated with disease improvement (especially in females), and found decreased gingival levels of Fusobacterium ASV_08 (F. animalis via BLASTn) in RRMS vs. HC and increased in Prog MS vs. HC. Fusobacterium can be either commensal microbiota and potential pathogens involved in periodontitis or peritonsillar abscesses,¹¹⁷ depending on the species and sub-species and host factors. Two independent studies found oral Fusobacterium increased in MS vs. HC.^12,15 The differential associations of oral Fusobacterium in MS may be linked to decreased pathogenicity of F. animalis vs. Fusobacterium nucleatum and Fusobacterium necrophorum.^117,118 There is a case report of a female MS patient on immunosuppressive therapy (dimethyl fumarate) that developed spondylodiscitis caused by oral F. nucleatum translocation through the blood circulation.¹¹⁹ Furthermore, F. necrophorum showed elevated IgG sero-reactivity when tested against CSF from patients with demyelinating disease (including MS) vs. control and other neurologic diseases.¹²⁰ In addition to the oral cavity, Fusobacterium can be detected in the gut microbiota, but are less frequent, and are associated with colorectal cancer.¹²¹ In the gut microbiota, one study found decreased Fusobacterium in MS vs. HC,¹⁵ and another study in paediatric MS study found that the absence of Fusobacterium in gut microbiota was associated with relapse risk.⁵⁵ Thus, while our study links oral F. animalis with clinical improvements in MS, the relationship of Fusobacterium in MS may be both species- and site-specific.

Because disease modifying therapy may either directly or indirectly affect the microbiome, we investigated the extent to which DMTs affected the oral and nasal microbiota. Furthermore, it has been reported that MS DMTs can increase susceptibility upper respiratory tract inflammation, including nasopharyngitis, upper respiratory tract infection, and flu-like symptoms.¹²² In our study, we found few DMT-related changes in the nasal, gingival and oropharyngeal microbiota (Fig. 4 & Figure S9). Anti-CD20 treated subjects had increased nasal C. propinquum and gingival Actinomyces, while MS subjects treated with interferon and natalizumab had increased nasal Finegoldia sp. compared to untreated MS subjects. These bacterial strains belong to the mentioned genera which have been rarely studied in MS or neurological diseases. Corynebacterium was not associated with MS disease diagnosis or progression in our data, and Finegoldia was found increased in RRMS and HC compared to progressive MS only at the genus level. DMT-modulated microbes have little effect on disease outcomes, which may be due to limited effects of MS DMT shaping nasal and oral mucosal immune responses or to a limited sample size, though it is worth mentioning that MS treatments may reduce immune responses and increase susceptibility to bacteria and viruses. For example, natalizumab and fingolimod treatments for MS leads to susceptibility to the John Cunningham (JC) virus, which otherwise causes no infection in immunocompetent individuals.^123,124 Others have reported that anti-CD20 can increase the risk of COVID infection.¹²⁵ Thus, DMTs may affect upper respiratory pathogens. This highlights the need for future studies of the nasal and oral microbiota in MS using metagenomics to address strain level changes. Another factor for determining potential pathogenicity is host immunologic function. It is possible that disease modifying therapies which suppress immune function could result in a higher potential for commensal microbiota to become invasive or pathogenic. A classic example in which pathogens such as S. aureus can be frequently detected in the normal nasal microbiota, but do not always cause disease. The differential function may be determined by altered host physiology or by pathogenicity factors, as is suggested by data associating S. aureus toxigenic strains with recent MS relapses.^19,111

An important question is whether the oral microbiota affects the gut microbiota. Few bacteria are shared between the oral and gut microbiota due to a low pH in the stomach, the different physiology of the intestine vs. the mouth, and the large biomass of intestinal microbiota. Nevertheless, there are reports that adherent oral microbes, such as F. nucleatum and oral Streptococcus species, are found along the GI mucosa, including both oral and faecal S. parasanguinis as mentioned above.¹¹³ Because we used a different MS cohort to collect nasal and oral microbiome samples than our gut microbiota cohort, a direct comparison was not possible. We did find, however, that gingival levels of E. brachy were associated with improvement in 2-year EDSS, and in our study of the gut microbiota, we found that faecal levels of E. brachy were associated with increased deep grey matter brain volume (less atrophy).⁷⁴ Thus, there appears to be beneficial associations with both oral and gut levels of E. brachy.

Limitations of our study included the use of descriptive data that relied on patient recall and reporting, including tobacco exposure (cigar, cigarettes, and tobacco), which could have been subjected to response error and recall bias. We applied 16S amplicon sequencing rather than whole genome sequencing to profile the bacterial and archaeal compositions, thus additional studies may help resolve species levels associations. Our in-house 16S library preparation protocol using the standardised Earth Microbiome Program primers allowed to prepare the sequence library from the V4-5 region, which allows comparison against one of the most commonly used microbiome sequencing approaches, although other V regions may provide higher taxonomic resolution. While studies suggest V1–V2 regions may be optimal for respiratory samples, the V3–V4 and V5–V7 were tested as giving a similar resolution of the alpha diversity as well as the compositional differences on the genus level.¹²⁶

Another limitation is the number of subjects vs. number of microbial comparisons. This study had 69 MS subjects and 40 HC subjects, which greater than current studies of the saliva microbiota in MS (Boussamet et al., Scientific Reports, 2024, n = 14 MS, 21 HC; Troci et al., Scientific Report, 2022, n = 36 MS, 38 HC; Zangeneh et al., PLoS One, 2021, n = 30 MS, n = 30 HC; Boullerne et al., Journal of Neuroimmunology, 2020, n = 1 MS, n = 1 CIS, n = 80 HC).11, 12, 13^,15 Because of the large number of microbiota variables vs. the number of subjects, it was not always possible to identify changes that overcame the FDR correction, which we address by reporting by FDR adjusted q values and non-adjusted p-values. Several of the targets we identify at p < 0.05 have been shown to affect CNS autoimmunity in animal models and are found across several analyses, which strengthens their potential findings and can be interpreted as hypothesis generating associations to be experimentally validated in future studies.

In summary, we found distinct bacterial changes in the nasal and oral microbiota in subjects with both relapsing and progressive MS. Using baseline microbiota samples and clinical outcomes over time, we identified potential beneficial microbes, including D. pigrum in the nasal microbiota and Prevotella sp. in the oral microbiota, and potential detrimental Streptococcus sp. in the oral microbiota. Although the oral and nasal microbiota is less studied than the gut microbiota in MS, our findings are consistent with studies from other groups11, 12, 13, 14, 15 and in vivo studies in which orally administered microbes modulate EAE.^54,57,96 Taken together, our study suggests that MS microbiota residing in three unique biologic niches in the gingiva and the upper respiratory tract are associated with the disease course in MS and provides an avenue to better understand MS pathogenesis and treatment.

Contributors

L.M.C., and H.L.W. contributed to conceptualisation, investigation, funding acquisition, project administration, resources, and supervision; E.C., R.S., B.I.G., and M.W. contributed to subject recruitment and sample acquisition; A.S. and V.W. contributed to extracting and sequencing the microbiota; S.L., L.M.C. contributed to methodology, data curation, formal analysis, visualisation and validated the data. S.L. contributed to software and writing the original draft; S.L., L.M.C., H.L.W., B.I.G., and F.M. contributed to review & editing the final manuscript. All authors have read and approved the manuscript.

Data sharing statement

The raw data that support the findings of the study have been deposited in the NCBI in the Sequence Read Archive (SRA) under BioProject ID PRJNA1202517. The codes were deposited and made publicly available at https://github.com/shuqili/NOMMS.

Declaration of interests

The authors report there are no competing interests to declare.