Casting a wide investigative net to define microbial behaviors that lead to periodontitis
NIDCR grantee Dr. Jorge Frias-Lopez discusses his results in harnessing genomic technology to uncover the molecular profiles of periodontitis
Jump to Q&A »
In the early 1980s, Americans turned on their television sets at night and settled in for sit-coms with predictable plots and regrettable fashions. Like now, every 15 minutes the networks cut to commercials to laud products and build a better you. Slotted among the better yous was Madison Avenue’s first crack at introducing American consumers to a previously unpublicized threat to their oral health: Gingivitis.
Authoritative male voice: “When you ignore plaque, you’re gambling with gingivitis, an early form of gum disease.” The winning bet: A dental product that specifically “kills germs that cause plaque and gingivitis.”
While the commercials made sense – practicing good oral hygiene is always a good idea – they raised a few questions. Among them was which germs really needed to be killed to prevent gingivitis and its possible advance to tooth-threatening periodontitis? At the time, researchers were divided into two schools of thought. The specific-plaque camp placed blame squarely on a handful of bad germs (pathogens) that live in the sticky plaque that forms along and under the gumline, known as the supragingival and subgingival spaces. Their data suggested that the pathogens expanded their numbers and geographical footprint over time to infect nearby cells and trigger gingivitis. The non-specific-plaque camp took the opposite view. They hypothesized that some as-yet undetermined dark factor specifically within the subgingival space caused the resident microbes, collectively known as the oral microbiota, to overgrow their normal boundaries and infect tissues of the mouth.
That’s where the science and Madison Avenue stood until in the early 1990s when the British microbiologist Philip Marsh split the difference between the two theories and proposed a new hypothesis. He wrote, “Some of the arguments surrounding these hypotheses may be, in part, about semantics (e.g. the definition of ‘specific’ or ‘non-specific’), since plaque-mediated diseases, while not necessarily having a totally specific etiology, do show evidence of specificity.”
The Dynamic Oral Microbial Community
Fluorescence microscopy shows multiple species of bacteria in dental plaque
Marsh then unified the best of both theories to propose his ecological plaque hypothesis. He argued that dominant microbes might cause periodontal disease, but they do so only if a specific environmental catastrophe within the plaque unleashes their ill behavior. “It is a basic tenet of microbial ecology that a major change to an ecosystem produces a corresponding disturbance to the resident community,” he explained. “Examples of this have been reported in ecosystems as diverse as soil, skin, and water.”
As Marsh’s hypothesis gained adherents in the 1990s, periodontal research adapted a number of existing ecological terms. Most notably, dental plaque became a biofilm, a dynamic, polymicrobial community that, like lichens growing along a craggy coast, colonizes various and sundry pits, fissures, and other oral surfaces. From this ecological perspective, the way forward in periodontal research involved gaining a more detailed understanding of these microbial ecosystems, the environmental disturbances that destabilize them, and the opportunistic pathogens that exploit the imbalance to cause disease.
DNA Technology and the Oral Biofilm
By the late 1990s and into the early 2000s, sequencing DNA had become technically feasible and more affordable. The NIDCR and other research funding agencies hurried to support multi-lab efforts that assembled the first full-sequence DNA maps of several key oral pathogens. These highly anticipated new resources meant that researchers for the first time could pore over the genetic blueprints (genomes) of these bugs and perform comparative DNA analyses across species (genomics) to generate more strategic, evolution-based hypotheses about what made a bad bug bad. Because genomics brought its own unique investigative perspective and language to biology, the periodontal lexicon underwent another predictable shift. Most notably, the freshly minted genomic term microbiome gave microbiota a sudden run for its investigative money. Initially, the new M word referred to the collection of microbial genomes in a given ecosystem (tooth, rock, plastic), shifting the investigative focus from the bugs themselves to the A, C, G, Ts in their genetic codes. The focus eventually dialed back to the current more-general connotation: all of the microbes in a biofilm, with an emphasis on their genomic information.
While the microbiome meme entered through one window, a new generation of genomic technologies arrived through another. These powerful laboratory tools allowed scientists to pursue an emerging scientific discipline called metagenomics. It builds on the premise that a biofilm is not a hodgepodge of microbial tenants bottled up in a sticky matrix but a single, collective, living organism. To study this community, scientists could extract the genomes from the various microbes living in a given biofilm and then sequence them to compile a working inventory of all the microbial genes present in the organism.
“The oral biofilm is not a hodgepodge of microbial tenants bottled up in a sticky matrix but a single, collective, living organism.”
For a typical oral biofilm, which houses up to hundreds of distinct microbial species, the inventory yields a catalogue of genes that easily can stretch into the hundreds of thousands. This opened two investigative doors. One, this body of biological information could be entered into a searchable online database as a bioinformatic tool to generate better working hypotheses and enable more targeted future studies. Two, these inventories were generated from biofilms living in their natural environments, not cultured under artificial conditions in a laboratory. It was the difference between learning French on the streets of Paris and in a college-language lab.
This new ability to inventory biofilms brought another shift in terminology. Instead of putting a bolder, genome-friendly coat of paint on the 1970s term biofilm, scientists opted for the modern metagenome as a synonym for microbiome to emphasize that the data emanate from DNA taken directly from a natural environment in the course of performing metagenomics.
But metagenomes came to the investigative fore with inherent limitations. Because polymicrobial communities are complex, DNA collection can be inefficient, some DNA fragments are easier to sequence than others and can possibly introduce bias, and the required sequencing capacity to churn out the data can be enormous. In addition, although the genetic code contained in metagenomes can provide broad hints about the possible activities performed in a microbiome in its natural environment, it offers no direct evidence that these activities actually take place in time (when) and space (where).
Uncovering Gene Activity in the Oral Biofilm
To tap more directly into the life of the microbiome, researchers learned to harvest messenger RNA (mRNA) from microbiomes, convert it back to DNA, and sequence it. What’s mRNA? When the information encoded in a gene is expressed, the DNA is copied chemically into an mRNA molecule, an information-bearing intermediary that will be translated elsewhere in the cell as the directions to make a specific protein. By sequencing the information, scientists are in a position to intercept and unencrypt what amounts to a biological order form to make a protein. When viewed across the myriad bugs layered into a microbiome, mRNA (metatranscriptome) serves as a measurable record of which of the hundreds of thousands of genes present in this multi-layered microbial community are actually expressed at any one time. Or, more tellingly, metatransciptomes can serve as a before-and-after record of how gene expression shifts in response to environmental changes, such as an increased acidity or mechanical stress.
The metatranscriptome brought an important new investigative wrinkle to the study of periodontal disease. Scientists could develop methods to harvest mRNAs and as comprehensively as technically possible from the subgingival spaces of people in good oral health and those with periodontitis. After sequencing the most abundant transcripts, they would have genomic snapshots that provide for the first time composite pictures to compare functionally what stays right for some and goes wrong for others. Or, they could take their analyses to the next conceptual level. Instead of trying to identify who is there by name, scientists can focus on the collective tasks, or functions, that oral microbiomes tend to perform in periodontal sickness and health. It’s much like walking down a busy city street. While nodding and greeting everyone by name would be great, it’s more important to your safety to be aware of your surroundings and monitor the behaviors, good and bad, of those congregating on the street.
This brings us back to theory. Because the study of the oral microbiome has improved by technological leaps and bounds in recent years, scientists have made headway parsing the specificity of the earlier plaque hypotheses and modelling their discoveries within Marsh’s ecological plaque theory.
Big Data and Periodontitis
The modeling suggests microbes that were labelled decades ago as periodontitis-causing pathogens might be more disruptive in the subgingival space rather than overtly pathogenic. For example, the oral bacterium P. gingivalis, long suspected of causing periodontitis, appears in reality to have mastered how to paralyze a key component of our natural immunity. Less efficient community policing in the subgingival space creates an opportunity for the usually law-abiding (commensal) residents to take advantage of the anything-goes dynamic to grow, turn opportunistically pathogenic, and cause periodontitis. This pathogenic tipping point takes us to one last term, which reflects the coalescence of microbiome research throughout the body. The term, coined by German gastroenterologists in the late 1950s, is dysbiosis.
“[Dysbiosis] can be likened, perhaps, to that of a crew team,” wrote Drs. George Hajishengallis and Richard Lamont recently of this dysbiotic microbial synergism that seems to underlie periodontitis. “All the oars need to be manned but the identities of individual crew members, provided they are capable of rowing, are not important for forward progression. Some functions, such as cox, are important for coordinating characteristics such as direction and speed.”
Now compelling metagenomic and metatranscriptomic data have begun to enter the periodontitis research literature. Among the most fascinating is a paper published in the August 2014 issue of the ISME Journal by a team of researchers at the Forsyth Institute in Cambridge, MA. The NIDCR-funded team compared the metagenomes and metatranscriptomes of several people with chronic periodontitis to those of healthy volunteers. The results, though preliminary, provide a fascinating glimpse into progression. In a follow-up article, published in the August issue of the journal Infection and Immunity, the scientists found a needle in their many haystacks of data to generate a new lead in understanding how P. gingivalis can disrupt a healthy biofilm.
A Q&A with Dr. Jorge Frias-Lopez
Dr. Jorge Frias-Lopez
The Forsyth Institute, Cambridge, Massachusetts
The Science Spotlight recently spoke with Jorge Frias-Lopez, Ph.D., a microbial ecologist at Forsyth and a senior author on both papers. He offered some perspective on his findings and the larger picture of genomics as an engine of discovery for periodontal disease.
Let’s start with the end in mind. If I were a periodontist, why should I keep my eye on metagenomes and metatranscriptomes?
Virtually every periodontist has wondered why some sites in the mouth remain stable and others progress to advanced disease? Same patient, same oral cavity, different biofilms. Ideally, periodontists could probe a suspicious site during an examination, characterize its microbial content, and determine whether another episode of periodontal disease will occur. Right now, that type of diagnostic specificity isn’t possible. Even if your dentist had a point-of-care technology handy to attempt it, the test results would be meaningless. The science simply isn’t there to program a point-of-care device on what to look for, when to look for it, and why.
Metagenomics and metatranscriptomics are the first steps in determining the what, the when, and the why that will program that future diagnostic device. The work builds on biology’s phenomenal progress in genomics over the past few decades and now allows research groups like the one here at Forsyth to cast a wide investigative net and haul in vast amounts of the messenger RNA produced by oral biofilms at all stages of the disease process. By hauling in these raw data sets, it is possible to profile the underlying biology of progression and ascertain the functions that seem to drive the process.
By functions, you mean collective behaviors, not pinning the blame on individual micro-organisms?
Yes. Rather than saying, these two organisms are usually prevalent and must be up to no good, I want to know which set of communal activities defines the potential for progression at a specific site. I think function ultimately is going to be more relevant to explain progression than who lives there. Sure, knowing who lives there and the genes encoded in their genomes offers a glimpse into their potential to do certain things. But potential doesn’t tell you anything about what’s really happening in the biofilm.
Let’s go to your paper and see what’s happening. Your group collected supra- and subgingival plaque samples from six healthy volunteers and seven people with severe periodontitis. Each sample was deep-sequenced. By deep sequenced, I’m referring to a next-generation sequencing strategy that gives a greater depth of coverage to identify the sequences contained in the RNA molecules suspended in the biofilm and their frequency of expression. Just to give readers an idea of the scale – no, avalanche - of data generated, what were the yields?
Our yields from participants ranged from 400,000 to 20 million sequences for the metagenome alone. For the metatranscriptome, the yields ranged from 900,000 to 39 million sequences per participant. Now, of the 1,278,494 genes identified via the deep sequencing and which already were entered into a genomic reference database, 702,106 had at least one hit in more than one sample.
That’s roughly equivalent to the population of Detroit?
That’s right. You treat your sample as though it were a single organism, in much the same way that Detroit is the sum of its many neighborhoods. From this perspective, we wanted to know what all of those 702,106 genes were doing. The deep sequencing gave us the genetic code and frequency of expression. That’s one, vital side of the coin. The flip side of the coin is bioinformatics. I created a huge database with all of the genomes that are available to the oral community. Once you have the transcription profiles from the deep sequencing, you try to map those transcripts to your database. The database allows you to figure out which gene is expressed and its phylogenetic origin. In other words, our results are only as good as our database.
And at this point, it’s pretty good?
I think so. As of now, the database has 312 species and more than 500 genomes. There may be some organisms that are important, and we don’t have their genomes yet. In the paper’s supplementary section, we found that 95 percent of the sequences from an “in silico” community match the right gene in our database. What’s even better, only 0.1 percent went to the wrong place. So, that’s the error rate that you could have in your analysis.
But I noticed in the paper that your group reports on behavior and the micro-organisms themselves. If the group behavior is what will matter most in defining progression of periodontitis, why straddle both sides of the investigative fence?
Good point. This is our first publication to present our community-wide analyses. We thought it was important to weave in as a common frame of reference the likely sources of a pathogenic behavior within the group dynamic.
What’s interesting is this common frame of reference allows your group to present in the paper something old and something new. Let’s start with the old, and I’m referring here to the Red Complex. What is the Red Complex?
In the late 1980s, the late Sig Socransky and late Anne Haffajee here at the Forsyth Institute performed seminal work that allowed them to catalogue into a color-coded system the consortia of dominant bugs associated with periodontal health, gingivitis, and periodontitis. The Red Complex describes the latter and consists of three oral pathogens - Porphyromonas gingivalis, Treponema denticola, and Tannerella forsythia. The key word is associated. At the time, the field lacked the needed tools to drill down and look at function and progression.
One way to measure pathogenicity – and which Socransky and Haffajee couldn’t measure at the time - is the production of virulence factors. In Figure 5 of your paper, 64 species are ranked by the number of putative virulence factors that each contributes to the metatranscriptome. How does the Red Complex fare?
Quite well actually. P. gingivalis and T. denticola ranked in the top five, and T. forsythia comes in fairly high on the list. This was very reassuring for us. You would expect to find the pathogens that everybody talks about to be near the top in producing virulence factors. If they didn’t make the cut, I would have been suspicious that something was wrong with our analysis. But the Red Complex was there, and so were a number of new organisms. So, all in all, we saw some really interesting things. I think we have a solid base now to proceed. We can be fairly sure that our results reflect what’s really going on during severe periodontitis.
What was most interesting about the new organisms?
There were a lot of organisms that are not considered at this point to be periodontal pathogens that were producing large numbers of putative virulence factors. I say, “putative” because this is based on homology.
Meaning the gene sequences, to various degrees, resemble known virulence factors?
Right. It may turn out that not all of them are virulence factors. But the huge number of them being produced suggests to me that the whole community is becoming more virulent, not just the Red Complex organisms. Somehow the new environment induces the production of these virulence factors in the whole community. It’s just a competition to make them. To me, that was the most surprising and interesting. It indicated that the community becomes more dysbiotic and more virulent under those environmental conditions.
Bacteria, though, aren’t the only occupants of oral biofilms. What about archaea, the single-celled organisms that are one of the proposed three domains on the tree of life?
We saw some archaea, but their numbers and activities were at a really low level. I’ve referenced other papers that mention this. Archaea are there, but they are never in large numbers. I should note that we didn’t look for fungi, although it’s certainly possible to do so.
What about viruses?
No, and that surprised me. But my colleague and co-author on the paper Ricardo Teles mentioned that to see expression, you must catch viruses at the exact moment that they are replicating. Viruses are not like bacteria. Either they are there doing nothing. Or, they are actually replicating in high numbers. If you don’t take your sample right at the moment that they are active, you won’t see them.
Your paper lists several other community functions in addition to virulence factors. Let’s run through some of them. You saw a lot of genes involved in iron acquisition. What is it about iron?
Iron is extremely limited in nature. A large number of pathogens need iron to grow. For many, it is a driving force in their life cycle. So, it wasn’t surprising that we saw so many genes expressed that are involved in iron acquisition. Actually, the data were reassuring. What I mean is although our data represent a snapshot in time, they show that what we expected was happening. That’s a real advantage of our metagenomic approach. You don’t see this type of activity in the lab under artificial conditions. The one surprise here was the proportion of genes, for example, in the Red Complex, that were involved in this function. These bacteria spend a LOT of energy at the point of disease to take in iron.
During periodontitis, you found that a lot of antibiotic-resistance genes were expressed. How interesting was that?
Very interesting. At this point, I don’t have a clear explanation. It’s something that we’ll look into in the future. One of the nice things about metatransciptomic studies is you can always go back and ask which organisms express these genes? What is their role in this resistance? But I haven’t had the time to look in detail at those questions.
You spent a lot of time and effort characterizing the role of a bacterium named TM7 in disease. Why TM7?
TM7 is associated with periodontal disease. But these associations are based solely on phylogenetic analysis and the high numbers of the bacterium in disease. The reason is TM7 presently is a non-cultivatable organism.
You can’t grow it in the lab?
Exactly. So, I thought it would be interesting to look at the functions of TM7 in the real microbial world using a metagenomic approach, and it worked. We started making some sense of the potential role of TM7 in disease or health.
What did you find?
We found that during severe periodontitis, TM7 expresses certain genes that it doesn’t during health. Some of these genes are involved in iron acquisition and others produce putative virulence factors. But, again, we are still in the early stages of our research. I wouldn’t make a big deal about these findings yet.
The oral microbiofilm contains many species of bacteria.
But the larger point is you could at least begin to study TM7?
Yes. These genomic techniques allow you to peek into the functions of an organism without any preconceived ideas about them. I’m a microbiologist by training, and, for me, it’s so interesting to gain a first glimpse of what these uncultivable organisms might be doing in the real world.
Let’s jump to your group’s second paper. Based on some earlier work and data from the first paper, you decided to look at a curious effect that P. gingivalis seems to have on the bacterium Streptococcus mitis. Why?
S. mitis is an early colonizer of a developing oral biofilm and abundant in healthy microbiomes that don’t cause disease. In previous work in culture, we introduced P. gingivalis and another periodontal pathogen called Aggregatibacter actinomycetemicomitans to see how our healthy biofilm model reacted to the new arrivals. We were surprised by what we saw.
What was that?
First of all, we saw expression profiles of healthy organisms changed completely. We observed that when those pathogens were added, S. mitis in particular started to upregulate the expression of enzymes called transposases like crazy. That caught my attention. At first, I thought the transposases helped S. mitis mobilize, or shuffle, its DNA to rise to the challenge of living with its new neighbors. But the high number of transposates was suspicious. I mean, if you want to mobilize DNA, why overdose on tranposases? A colleague suggested that it may have nothing to do with mobilization. It might have everything to do with cell death.
Did you look into that?
Yes. What we found is when we added P. gingivalis alone, it caused S. mitis cells to die. There is something that this bug does that induces S. mitis cell death. That’s interesting because, it is well known that when you have a single-species biofilm that grows old, part of the population kills itself in order to release DNA and other nutrients that allow the others to survive. What’s interesting here is it’s not that S. mitis senses that the biofilm is old. The biofilm isn’t old at all. But P. gingivalis does something – and we still don’t know what that something is – that induces the death of S. mitis cells. Ecologically, that has really big consequences. If we can prove in the future that P. gingivalis makes other organisms die—particularly health-associated bacteria, it may have important implications for understanding how biofilms become diseased.
A live-dead bacteria viability assay using confocal microscopy shows addition of P gingivalis to a healthy biofilm of S. mitis cells (A) causes them to die (B).
It also adds another piece to the “keystone pathogen” hypothesis. Briefly, Hajishengalis and colleagues proposed that P. gingivalis populates biofilms in low abundance but hovers in the shadows to disrupt immune function to its dysbiotic benefit. So these data suggest that P. gingivalis might wreak havoc not only on the immune system but at least on some of the good bugs in the biofilm?
That’s right. Hajishengalis observed that the community changed. We don’t know in which ways the community changed or which set of signals the keystone pathogen uses to change the community to its advantage. The other interesting thing is we still don’t know what the signal is to cause S. mitis cell death. If we can somehow learn the language that these organisms use to tell S. mitis to die, you can think of scenarios in which you can use those signals to kill unwanted organisms instead of using antibiotics. That’s an idea that I have, but first we must identify this signal.
But what’s unique about the oral biofilm is the high density of cells and the species that have specific cell-to-cell interactions. How is that unique as opposed to biofilms in other parts of the body? Is our understanding of the biofilm more developed in the mouth?
Well, I think the density per square inch is even greater in the gut. But one of the advantages that we have in the oral biofilm is that we have a wealth of historical knowledge that other sites in the body don’t. In the mouth, we have a pretty good idea of its spatial, or at least its sequential, colonization. We know which organisms may be important in periodontal disease. In the gut, they’re still trying to piece together which bugs are important and what the core community is in a normal gut. So, we are way ahead of most of the other sites in the body.
We haven’t talked at all about the role of inflammation in driving chronic periodontitis. Is immunity in your long-term research picture?
It is. My ultimate goal is to understand immune function, too. What happens to the host and its defenses when the disease is active? I want to try to make the link between what we see in the bacterial community and the specific response of the host under different conditions. I'd also like to do a dual analysis and glean a more comprehensive picture of what’s going on.
In other words, the bugs are the initiators, and inflammation is the promoter of chronic disease?
Yes. The final goal may be to subclassify diseases according to the type of virulence that is active at a site and its effects on immune function. The idea being that periodontal diseases are truly plural, not singular. But, above all, it would be to characterize these functions. If you see early expression of these genes at a site in your mouth, you run the risk that it will advance to periodontitis in weeks or months. The other thing that we did, based on our knowledge of these functions, is try to devise new strategies to reduce disease. That will happen when we know more about metagenomic function.
And you’re getting there.
Yes we are. The tools are available now to begin performing these types of analyses that generate meaningful data. There are just a plethora of questions to ask in the laboratory. I hope more people get involved in this research area to generate more information to tell us what’s going on there.
Thanks so much for your time.