[applause] Ian Jeffer ello. I'd first like to thank Lita for inviting me to talk, and I'd like to thank all the organizers for giving me this opportunity. So, diet microbiota interactions and the elderly. So, you've -- so, all about the microbiota -- so David Relman gave you these figures already, but these are good figures so I don't think there's any harm going through these again. So 10 times the number of cells in our body are bacteria to human, and these contain 150 times the number of bacterial genes to human genes. And so this might not be a one-to-one ratio because of splice variants and whatever but this is still a huge number. And if you think of genes as function, then these bacteria perform an awful lot of beneficial functions throughout the adult life, and these populations are relatively stable in the adults. And in stable form, they are involved in the absorption of minerals, the utilization of nutrients from our food. They interact with the amino mojetary [spelled phonetically] effects, the production of substrates, and even the regulation of insulin sensitivity and appetite control through the food fatty acid receptors 2 and 3. And so variations in these populations have been associated to a number of disorders including irritable bowel syndrome, IBD, ulcer colitis, and obesity. So why look at the gut microbiota in the elderly? Well, first of all, the elderly are an increasing portion of our society. This is a wonderful success of modern medicine but it comes with its own challenges. This cohort of our society have an increased susceptibility to infection, and so C. difficile infection is a big problem in residential care, and they also have this sort of systematic increased inflammatory status that we refer as inflamm-aging. And -- but the -- one of the best reasons for looking at the microbiota in the elderly is because there are changes in the microbiota composition and activity in this elderly population. It remains stable for adult life, and then as people get older, it begins to change. And so last but not least, there's the prospect of dietary modulation of these organisms to improve health of this cohort of people. So, to look at this, we need a dataset, and the dataset we use consists of 178 elderly individuals and 13 young controls. The largest group in this dataset are the community-dwelling individuals, but we also have 75 people in residential care. We divided these into two groups, individuals who'd been in residential care for longer than six weeks -- we refer to these as long-stay -- and less than six weeks, and we refer to these as rehabilitation. Within the people who live in community, we have 20 day hospital individuals who attend outpatient care, and for these individuals, we have a number of datasets. The primary dataset is a 16S ribosomal DNA amplicon dataset that measures the taxonomic composition for the individual subjects, but we also have a shotgun dataset based on Illumina sequencing, a metabolomics dataset, and we have dietary data for 168 of our 178 elderly individuals. So I'm a bioinformatician. When I get a dataset, I want to visualize it. So we have 5.4 million 16S reads. We clustered these into 47 and a half OTUs. My favorite definition of an OTU is an operational taxonomic unit. It's my favorite definition because it tells you absolutely nothing at all. The -- so OTUs are just reads clustered by similarity, in this case, 3 percent, and they generally correspond to either genera, species, or genus-level taxa. And so we can get this -- we can see the OTU composition for each individual person, and using this information, we can generate UniFrac distances and visualize the dataset. So this is a multivariate plot of the dataset. So the green samples here are our community samples, and the red up here are our long-stay individuals, and, as you can see, there's quite a good separation between the people living in the community and the people living in long-stay based on their microbial composition. The young controls here in purple cluster with the community, and we have individuals in day hospital and rehab who sort of form these intermediate groups. We can visualize this data a second way using one of these heat plots. Just generally, red is there, blue is absent. Along the top here, we have the samples, colored same as before, red being the long-stay individuals and green being the community, and, as you can see, based on this hierarchal tree, this is the strongest plate in the dataset. Along the side here, we have the OTUs. These are clustered. They are also color-coded at the family and fund level. The fund-level color-coding is blue for bacteroidetes, brown for firmicutes. And so what we can see is a couple of things. So, first of all, there's this -- a number of taxa are increased in the long-stay individuals, and we also see a number of taxa that are lost in the long-stay individuals compared to the community samples. We can see here that there are some clusters that have lost more microbial taxa than others, and so we can further divide this tree based on a particular height into eight groups; these are not distinct groups, they're just overlapping clusters. And we can take this information and overlay it back on the previous multivariate plot. And what we can see is that these eight clusters separate spatially along the plot, with these clusters over here being associated with long-stay, and these clusters over here being associated with the community samples. So we also have food data. This food data comes in the form of food frequency questionnaires. So these measure long-term dietary effects, and we have this information for the majority of our subjects. And it consists of 147 food types that were -- that are representative of the Irish diet, and these have previously been validated in previous studies. And we also generated what is known as the Healthy Food Diversity Index, and this pretty much does exactly what it says on the thing [spelled phonetically]. It measures how diverse and healthy your diet. Each individual food is given a health value, 0.268 for fruit and veg, and 0.0001 for lard. So, if you -- and this is multiplied up by, like, the diversity of the diet, and so if you have a diverse diet of fruit and veg, you get a very high value. If you have low diversity diet on mainly animal products, you have a very low value. And we can visualize this food frequency data using multivariate analysis. Similar to before, you've got your samples up here at the top. We use correspondence. This has the advantage, you can also visualize the variables, and so up here at the top, you have the samples. Our community samples are in green and our long-stay samples are in red. Very similar to the microbiota. Very good split. Down here, the foods are color-coded. Green is fruit and veg, and as you can see, it's skewed off towards the community. Brown is meat, skewed off towards the long-stay, as is blue, which is high sugar, high fat foods. So, generally, the community eat more fruit and veg; long-stay eat more meat and high fat, high sugar foods. Again, we can visualize this with a heat plot. The reason for doing this is, again, to look at the sort of diversity. We can see that there's a large number of foods here that are just not eaten in the long-term residential care, and a lot of these are fruit and veg.1 There are a number of foods that have reduced consumption in residential care, and there1 is -- and there are a number of foods that have increased consumption. So the people1 in residential care generally have a much less diverse diet than people in the community.1 And we can -- as we can see the same trends in both datasets, we can use a thing called1 Procrustes analyses to view both datasets together, and so all this really does is it1 takes the two datasets that are both sort of in multidimensional space, and it twists1 them in this multidimensional space to get the most covariance between the two datasets.1 And, in this way, we're able to visualize both the diet and the microbiota dataset on1 the same plot where the diet is at the end with the dot on it, and the microbiota is1 at the far end. And so what we can see -- so if we zoom in on the -- if we zoom in on the1 individuals that live in the community, what we can see is that individuals who have this1 type of diet have a corresponding type of microbiota, and individuals who have this1 type of diet have a corresponding type of microbiota. This is unweighted UniFrac. This1 is weighted UniFrac. And so up here, we have more of our Prevotella-associated microbiota.1 Down here, it's our bacteroides-associated microbiota.1 As people enter residential care, their diet starts to change, and after six weeks -- this1 is six weeks to a year -- their diet is recognized to be that of a person in residential care.1 And after about a year, the microbiota follow suit, and it is sort of recognizably the microbiota1 of individuals in residential care. So the diet changes first. The microbiota then follows1 over the course of the year.1 So the diversity of the microbiota in the diet. So, we -- down here, we have phylogenic1 diversity, which is a measure of the microbial diversity in the dataset, and these are our1 four diet groups -- did I forget to mention four diet groups? The four diet groups come1 from this hierarchical tree where we, again, split it just into four, and the Diet Group1 1 can be described low fat, high fiber. Diet Group 2 can be described as moderate fat and1 fiber. Diet Group 3 can be described as moderate fat and low fiber. And Diet Group 4 can be1 described as high fat, moderate fiber.1 And so across these diet groups, you can see there is a decrease in healthy food diversity1 from the community down into the residential care, and we can see an associated decrease1 in the microbial diversity from Diet Group 1 down to Diet Group 3. There's a small increase1 between Diet Group 4 and Diet Group 3 in the microbial diversity. This isn't significant,1 but I still think it's quite interesting because there is an increased consumption of -- a1 small increase of consumption of fiber in Diet Group 4, even though there's no increase1 in the actual healthy food diversity of this group, and so this is reflected in a small1 increase in microbial diversity. But overall, there's a very good correlation between microbial1 diversity and healthy food diversity within this dataset, so a more diverse diet leads1 to more diverse microbiota.1 We also have metabolomics dataset. So these are two multivariate plots where we separate1 -- where we see there's separation between community and long-stay, and we can -- and1 community and rehab. And we -- and so how does this relate to the microbiota? So there's1 a method that is similar to Procrustes analysis called co-inertia analysis, and so this allows1 us to visualize the samples and the variables at the same time.1 So if we just look at the top two panels -- so the first panel on the left -- and these are1 the samples, same as before. One side is -- the start of the arrow is metabolites. The end1 of the arrow is the microbiota. And you can see that the green community samples separate1 out from the red long-stay samples, and this separation is associated with, particularly,1 butyrate and acetate, but also propionate, valerate, and glutarate. So the short chain1 does a reduction in the production of short-chain fatty acids within the residential care individuals,1 and this is associated with a number of microbial changes represented here by the genus level,1 and we can see our bacteroidetes, ruminococcus, coprococcus, oxalobacter, among other genera1 that are more associated with the community than they are with the long-stay, and with1 the long-stay, you have the most associated genera is the parabacteroides.1 So to generate short-chain fatty acids, you need substrate, i.e. fiber, and you also need1 the microbial functionality, and so is the question that, just, the reduced fiber leads1 to reduced short-chain fatty acid production, but we also see reduction in marker genes1 for the production of butyrate, and acetate, and propionate. So for butyrate and acetate,1 this reduction is significant. For propionate, it's just a trend, but this makes sense based1 on the previous plot where the butyrate and acetate were much more strongly associated1 with the community than propionate.1 We also have number of measures of the health of these elderly individuals. We measure -- we1 use FIM and Barthel to measure frailty. We have information on cognitive decline in the1 Mini-Mental State Exam, and we have information -- geriatric depression and nutrition, and1 calf circumference, which is -- and mid-arm circumference, which is a measure of sarcopenia.1 And we removed possible confounders. We removed individuals who had used -- who had been given1 antibiotics within the previous month, and we also -- we also, in our statistical model,1 we adjusted for age, gender, location, and medication. So to correlate these clinical1 variables with the microbial populations, we correlated them with the two strongest1 trends in the dataset as defined by the multivariate analysis, and so we correlate the clinical1 variables with the trend going this way and the trend going this way. So this trend is1 from the community to the long-stay-type microbiota, and this trend here is more the high diversity1 microbiota to the lower diversity microbiota.1 And interestingly, we see IL-6 and IL-8 being associated with these two axes, and this is1 -- this was previously reported by Biagi et al [spelled phonetically] in 2010 as being1 associated with centenarians. And we can see the only centenarian in this dataset, our1 102-year-old up here, the healthiest person in the residential care. And also associated1 with the first axis, we see a decrease in calf circumference and weight, and along the1 second axis, we see a increase in the geriatric depression test.1 When we look at the community samples just on their own, we also see this decrease in1 the geriatric depression test, and when we look at the long-stay subjects only, what1 we see is as we go from more community-type -- as the individuals lose their community-type1 microbiota, there's an increase in -- there's an increase in frailty, and a decrease in1 calf circumference and weight and BMI. If we adjust for food, the association with weight1 and BMI disappears. The -- from the high diversity microbiota to the lower diversity microbiota,1 we also see that as you move from high diversity to low diversity, there's an increase in frailty,1 and as -- and also an increase in this inflammatory marker, the C-reactive protein, high levels1 of which is indicative of poor health.2 So just to give you an idea of what the side effects we're dealing with here, so the colors2 here are the same as the colors here. The community individuals in the middle, the long-stay2 at either side, high diversity, low diversity, and you can see that the blue here are the2 youngest of the long-stay, followed by the cyan, followed by the red, followed by the2 yellow, where the black is the -- the grey is the oldest. And so there's about a five-year2 difference on average between the red here and the yellow here, just to -- and when we2 look at the frailty, what we can see is that even though these people are younger, they've2 got a higher level of frailty than these cyan people over here. The level of frailty -- the2 difference -- the level of frailty between these is quite similar, considering that these2 individuals are five years older, and these people are the most frail. So when -- so there's2 an increase in frailty associated with the low diversity microbiota. We also see an increase2 in this C-reactive protein associated with the low diversity microbiota, particularly2 compared to the high diversity microbiota. So this -- these individuals are older than2 both the blue and the red here, and yet they have a much lower level of CRP.2 So, in summary, the microbiota in the elderly is different depending on community location.2 This is driven by habitual diet. Microbiota alterations correlate with health changes,2 especially in the long-stay, and so the hypothesis here is that diet shapes gut microbiota, which2 may impact on health in elderly people, and so we're hoping this leads to carefully-designed2 dietary interventions to promote healthier aging. And this is what we're doing as part2 of the New Age Consortium. So the New Age Consortium consists of about 30 different2 partners in -- across Europe and beyond, and we're going to look at 1,250 individuals in2 five different countrie .K., Netherlands, France, Italy, and Poland. Half of these individuals2 would be given a healthy Mediterranean-style diet, while half will be just given their2 regularly diet for 12 months. Over these, we'll have their microbiota information from2 before their diet and after, and we'll be measuring health indices throughout, as well2 as epigenetic and metabolomic datasets. So it's going to be -- it's going to be a very2 interesting long-term intervention.2 My challenges? My challenges are microbiota modulation and restoration in this cohort,2 and the challenges for -- in lots of different cohorts like -- and associated with this is2 the use of prospective longitudinal study and interventions. I would also like to echo2 Janet and the need for multiple omic datasets to really understand in a sort of integrative,2 comprehensive way what is happening in the gut microbiota, and the generation of dietary2 guidelines informed by the needs of the microbiota as we age.2 I'd like to thank you all for your attention. I'd be happy to take any questions. Thank2 you very much.2 [applause]2 The most important bi cknowledging the people I work with. I would like to acknowledge2 Paul O'Toole and Marcus Claesson. I'd also like to give a mention to Mr. Hugh Harris2 and Dr. Eibhlis O'Connor. Thank you very much. Okay. [laughs]2 Female Speake his talk is open for questions.2 Male Speake n your big European study, you're not going2 to take the Italians off their Mediterranean diets and put them on British stodgy diets,2 are you?2 [laughter]2 Ian Jeffer laughs] That would be an interesting study,2 but -- so the diet would be formed -- so it's -- the diet is going to be formulated specifically2 for elderly individuals, so it's good -- the needs of the elderly are going to be taken2 into account. So part of the project will be to sort of formulate an elderly-type food2 pyramid, and so, hopefully, we'll be able to improve somewhat on the diet that they2 are already on.2 Male Speake inaudible] in yogurt consumption and whether2 that was reflected in any changes?2 Ian Jeffer he short answer is there wasn't much difference2 in yogurt consumption, and we have not been able to identify compositional changes in2 the microbiota associated with consumption of yogurts in this dataset.2 Male Speake id you look at differences in antibiotic2 consumption?2 Ian Jeffer o antibiotic individuals were excluded if2 they had taken antibiotics a month prior. So although the -- although the effects of2 antibiotic treatment go beyond a month, this -- these effects generally reflect the sort2 of increase in antibiotic-resistant genes, and also maybe a loss of some species, such2 as bifidobacterium species are very prone to antibiotic treatment, and it's known that2 there is a loss of BIF species in elderly individuals -- well, a -- there's a lowering2 of the diversity of the BIF species in elderly individuals, with the exception of maybe bifidoadolesensus2 [spelled phonetically]. And so I would argue that after antibiotic treatment, this sort2 of new stable microbiota is then the microbiota associated with the individuals.2 So it would be impossible to exclude every individual that had antibiotic treatment for2 a long period of time, just due to the prevalence of antibiotic use in our society. By the time2 someone's treated, they have had multiple courses of antibiotic treatment, so it's no2 different for the elderly individuals. It would be a factor that would cause them -- that2 would destabilize their microbiome, I think, and induce some changes in the elderly.2 Female Speake ell, please join me in showing appreciation2 for this speaker.2 [applause]2 Female Speake hank you. Thanks, that was a very interesting2 talk. And now I'd like to introduce our next talk, which will be equally interesting and2 is in the pediatric realm, which is of interest of many of us here. This is Kathryn Dewey,2 and she is from the University of California, Davis, and will be speaking to us on Diet,2 Child Nutrition, and the Microbiome. Welcome, Dr. Dewey.