Abstract

   Periodontitis is an extremely prevalent disease worldwide and is driven
   by complex dysbiotic microbiota. Here we analyzed the transcriptional
   activity of the periodontal pocket microbiota from all domains of life
   as well as the human host in health and chronic periodontitis. Bacteria
   showed strong enrichment of 18 KEGG functional modules in chronic
   periodontitis, including bacterial chemotaxis, flagellar assembly, type
   III secretion system, type III CRISPR-Cas system, and two component
   system proteins. Upregulation of these functions was driven by the
   red-complex pathogens and candidate pathogens, e.g. Filifactor alocis,
   Prevotella intermedia, Fretibacterium fastidiosum and Selenomonas
   sputigena. Nine virulence factors were strongly up-regulated, among
   them the arginine deiminase arcA from Porphyromonas gingivalis and
   Mycoplasma arginini. Viruses and archaea accounted for about 0.1% and
   0.22% of total putative mRNA reads, respectively, and a protozoan,
   Entamoeba gingivalis, was highly enriched in periodontitis. Fourteen
   human transcripts were enriched in periodontitis, including a gene for
   a ferric iron binding protein, indicating competition with the
   microbiota for iron, and genes associated with cancer, namely nucleolar
   phosphoprotein B23, ankyrin-repeat domain 30B-like protein and
   beta-enolase. The data provide evidence on the level of gene expression
   in vivo for the potentially severe impact of the dysbiotic microbiota
   on human health.

Introduction

   Dysbiosis of the human microbiome may play a crucial role in a variety
   of complex diseases, e.g. type II diabetes, rheumatoid arthritis,
   allergy, inflammatory bowel disease, periodontitis and even extreme
   obesity^[32]1–[33]5. The transition from the healthy state to dysbiosis
   is thought to be driven by key microbial players in the community.
   These species can influence the host immune system and worsen the
   habitat conditions for the commensals which dominate the healthy
   symbiotic community^[34]6. Chronic periodontal disease is a prevalent
   oral disease worldwide. According to a report from CDC (Centers for
   Disease Control and Prevention) there are 47.2% of adults over 30 years
   old that have some form of periodontal disease in the USA^[35]7, and
   the situation is also serious in other countries.

   The oral microbiome is one of the most complex and dynamic microbial
   communities in the human body, comprising several hundreds of different
   species of bacteria. Moreover, archaea, protozoa and viruses are
   present, and millions of different genes are expressed. Dysbiosis of
   the oral microbiota can interfere with the normal function of the host
   immune system resulting in enhanced development of periodontitis^[36]8.
   Moreover, periodontal disease is associated with many other complex
   diseases. For instance, it can increase the risk for cardiovascular
   disease^[37]9, rheumatoid arthritis^[38]10 and cancer^[39]11. The oral
   cavity is easily accessible and thus an ideal system to study the
   interaction between host and microbiome.

   The periodontal metatranscriptome contains the transcripts of genes
   from all members of the microbiota, including bacteria, archaea,
   viruses and phages, protozoa, and fungi, as well as from the human
   host. It reflects the activities of all of those groups simultaneously
   without culturing bias. Moreover, the transcriptional activities
   reflect the interactions between all members of the community and with
   the human host in vivo. Less than a handful of studies on the
   metatranscriptome of periodontitis are currently available, and each of
   them shed light on a different aspect of the complex microbiota.
   Duran-Pinedo and co-workers demonstrated the expression of putative
   virulence factors in commensal oral microorganisms^[40]12 and
   identified GO terms associated with disease progression^[41]13. Jorth
   et al. compared microbial communities in healthy and diseased
   periodontal pockets in the same individual^[42]14 and found that
   dysbiotic communities were less diverse but more similar among each
   other than healthy periodontal pocket communities. They identified key
   players and the metabolic enzymes that they transcribed, and suggested
   that although the species composition in periodontal pockets varies
   widely, the metabolic networks activated in disease are
   conserved^[43]14.

   Archaea comprise a minor component of the oral community^[44]15. The
   dominant Archaeon is Methanobrevibacter oralis ^[45]16 which has been
   categorized as a periodontal pathogen due to its strong association to
   disease^[46]17 and pocket depth^[47]18. Methanogens have been shown to
   be co-occurring with Prevotella intermedia, a fermentative species,
   possibly due to interspecies hydrogen transfer^[48]18.

   Viruses are the most abundant living entity on the planet. A number of
   human viruses can be detected in the oral cavity^[49]15. The salivary
   virome is dominated by bacteriophages which may act as reservoirs of
   virulence factors^[50]19. Interestingly, CRISPRs from healthy
   individuals cover a wider phylogenetic host spectrum than those of
   periodontitis patients^[51]20. Accordingly, a metatranscriptome study
   showed that more diverse phages were present in periodontally healthy
   individuals^[52]21. Subgingival biofilms contain clearly different
   viromes in health and periodontal disease^[53]22. Thus, bacteriophages
   might be important drivers of the community composition of bacteria in
   the oral cavity, yet their precise role for disease progression is
   currently not understood^[54]23.

   Very little indeed is known about protozoa in the oral cavity. Two
   species are commonly observed, Entamoeba gingivalis described for the
   first time^[55]24 in 1849 and Trichomonas tenax ^[56]15. Although they
   are strongly correlated to periodontitis^[57]25 it is now thought that
   they are not pathogens, but merely commensals feeding on abundant
   bacteria and debris associated with poor oral hygiene^[58]15.

   We have previously analyzed the shifts in the taxonomic composition of
   the microbial communities in subjects with chronic periodontal
   disease^[59]26. We then analyzed the metatranscriptomes of those
   samples^[60]27 and were able to show significant changes in the
   composition of the active community as well as enrichment of certain
   COG categories in periodontitis. Gene expression of the commensal
   Prevotella nigrescens shifted towards pathogenicity in samples from
   individuals with chronic periodontitis, confirming the concept of
   accessory pathogens. A detailed comparison of butyrate synthesis
   related transcripts showed that, contrary to previous assumptions,
   Fusobacterium nucleatum transcribed butyrate synthesis genes to a
   similar extent both in health and disease; however, in disease both the
   functional and the taxonomic diversity of expressed butyrate synthesis
   pathways increased, i.e. butyrate synthesis transcripts from additional
   taxa and additional enzymatic routes were detected. We then discovered
   3 potential functional biomarkers based on the metatranscriptomes which
   are highly predictive and can discriminate the dysbiotic communities
   from the healthy ones^[61]27.

   Here we analyzed the data further. We used a newly developed in-house
   pipeline to gain a deeper understanding of the taxa that drive
   functional shifts in dysbiosis. Previously, we had used MG-RAST^[62]28
   with the COG database^[63]29 to characterize the functional alterations
   of the communities. Here, we used the KEGG database^[64]30 to assign
   genes to pathways and functional modules. It provides information not
   only about the possible general functional category of a gene, but also
   assigns given genes to specific pathways. By mapping transcripts to
   KEGG reference genes and using gene set enrichment analysis (GSEA) for
   pathways or modules, we were able to identify altered pathways or
   modules as well as those microbial taxa causing them. Additionally, we
   analyzed archaeal, protozoan and viral transcripts thus unraveling the
   activity of all three domains of life in periodontitis. Periodontal
   pocket samples always contain varying amounts of human “contamination”
   derived from epithelial cells. When we analyzed those transcripts we
   found hints for possible interactions between the microbiota and the
   human host.

Results and Discussion

   In this study, we used a dataset from our previous work^[65]27 which is
   a part of the “German National Cohort”. These samples were taken from
   different individuals at several sites (two paper points per site), and
   the sampling paper points (one paper point per site was chosen)
   originating from the same individual were pooled together for RNA
   isolation. Fourteen periodontal pocket metatranscriptome datasets were
   analyzed, of which 4 were derived from individuals with chronic
   periodontitis and 10 from periodontally healthy individuals. Study
   population and methods for sampling, extraction of total RNA, depletion
   of rRNA, and sequencing have been described^[66]27. Our previous
   analyses clearly showed that samples AU_01 and AU_11 are
   outliers^[67]26, [68]27. Those microbial communities were distinct from
   all other samples for reasons that we do not know. We excluded them
   from the present analysis because we wanted to identify functional
   changes in chronic periodontitis occurring consistently in the majority
   of patients. The short reads aligner BWA with the BWA-MEM^[69]31
   algorithm was utilized to map metatranscriptomic sequences onto DNA
   sequence references. The following alignment references were utilized: