A Milk - Oriented microbiota - Mom - In infants - How babies find their moms - David mills

David Mills: I want to thank the organizers. This has been a wonderful meeting, and a chance to learn -- see so much impressive work on the microbiota. I always like to also say I am in the wine department at UC Davis. I am going to be talking about milk. I'm also in the food science department at UC Davis. So I work on two great beverages. We've learned a lot about probiotics in the last 100 or so years. And there's been a lot of nice talks today where people have mentioned bifidobacteria, lactobacilli, in one way or another as a probiotic. And I think it helps us to go backwards a little bit to see where this came from. I mean, the probiotic concept, of course, emanated from Mechnikov a long time ago, and although, while some people say it, he didn't get his Nobel Prize for this, he got the Nobel Prize for, I think, phagocytosis. And he also was not the person to first use the word dysbiosis. I guess it was a word that was in play at the time. But he used it a lot in his "Prolongation of Life," written in 1906, and talked about how one could use fermented milks as a way of reestablishing a dysbiotic microbiota. And his rationale was rather simple, that milks that were fermented didn't putrefy, and his vision was it was putrification in the GI tract that was causing problems. And so if you drink a lot of fermented milks, you should be able to wipe out that problem. Needless to say, it wasn't that -- universally accepted at the time. And if you read Paul de Kruif's "Microbe Hunters" in 1926, he made a rather wry comment about the Bulgarian bacillus became the rage. Companies were formed, directors grew rich off these silly bacilli. And I look at this and I say, "Well, you know, maybe we haven't gone that far in 100 years." You know, people are still selling probiotics, and they're getting rich. I don't know if I'd call them silly bacilli anymore. But, of course, we have made a lot of progress. We know a lot more. I am going to talk about some of the challenges related to that a little bit later. But what I wanted to focus on is what might be the next generation of probiotics and prebiotics? So before I go to that, I need to get into a little bit of definition of what a probiotic is, and what a prebiotic is. So -- and there's some misperceptions on how these are used. So a probiotic is a live organism that, when administered in adequate amounts, confers a health benefit on the host. When administered: that means it's outside of your body, and you're administering it either on or in your body. A prebiotic is a selectively fermented ingredient that allows specific changes in the composition or activity -- this is the modified, new description -- of a gastrointestinal microflora -- microbiota, of course -- that confers a health benefit as well. Prebiotics do not enrich probiotics unless you've added the probiotic into the body. So, often, people will talk about prebiotics enriching the probiotics in the intestinal microbiota. Well, probiotic can only be there if you actually put it there. It might enrich other microorganisms that are in the gut that are in the same genera and class as probiotics, but a probiotic, by definition, is something you're adding in from the outside. That gets us to the concept of symbiotics, which, of course, is combinations of pre and probiotics so that you can actually enrich what you are also adding in. We, at UC Davis, have been very interested in the concept of symbiotics. And we're looking for natural models for symbiotic effect, and landed on one that is rather obvious, and that is human milk, because human milk really serves two biological entities, and one is the human host, of course, but also the microbiota of that host. And it actually is involved in establishing that microbiota of the host. And that has also been studied for many, many years. Again, about 100 years ago, Tissier identified, with the high-tech instrument of a microscope, that there was a lot of what he called Bacterium bifidus in breastfed infant feces. Many, many years later, others started to identify, well, what was the factor that might enrich what turned out to be bifida bacteria in the feces of breastfed infants, and identified it as some sort of complex glycan. And you can see the quote there from the work of Gyorgy. So this concept has been out there. And we've been very interested in what are the factors in milk that shape that microbiota. Well, of course, milk is mostly water. But there's a variety of macro and micro nutrients. It has a fair amount of lactose, of course, and that's food for the infant. Has a fair amount of lipids, is food for the infant; fair amount of protein that's broken down, is also food for the infant. Also has a huge amount of glycans, human milk oligosaccharides. And I'll talk about those in a second. Of the agents that are in human milk, there's a variety that shape the microbiota. So there's lysosome, there's lactoferrin. There's a variety of free fatty acids that can have some either passive or active inhibitory effects. So we are suppressing, or potentially suppressing, that microbiota and shaping it with some of the constituents in milk. But it's the glycans that are also present in human milk that are really intriguing and thought to shape that microbiota as well. And what is really amazing about the glycans that are in milk is how complex they are, and that the various linkages that -- by how they're put together, the enzymes needed to break those linkages are really not expressed in the human. So you're delivering a huge amount of glycans that the infant can't eat. And so it's clearly aimed at the microbiota and shaping the microbiota in some way, at least in part. And this is a representation from one of the reviews we've written. It just sort of shows all the kinds of linkages. And this is not all of the glycans in milk, by any means. I happen to have the great joy of working with an amazing glycochemist at UC Davis by the name of Carlito Lebrilla. And he has worked for the last 10 years trying to understand the complexity of human milk glycans, and free human milk oligosaccharides in particular. And I'm showing you sort of a generic structure here, just to show you the different kinds of sial-fucosyl [sp] linkages, et cetera, that compose human milk oligosaccharides. And so they're very complex structures. This is a composite. There's many different types of structures. But when you look at human milk, you'll -- he notices that most of the human milk oligosaccharides are in the 4 to 10 degree of polymerization range, and many of those are fucosylated. And that's really the bolus of moms are delivering into infants. But there's still a large number of smaller-quantity larger oligosaccharides present. In humans, there's a higher proportion of fucosylated oligosaccharides that's sialyated. And so far, there's somewhere between 150 to 200 different structures. What is this bolus of glycans doing for the infant? Well, the first and most obvious idea was that those glycans look like the glycans on epithelial surfaces. And so maybe they're just acting as decoys for an epithelial surface. So a pathogen comes along, binds them, and flushes out of the infant. And that has been demonstrated in a variety of realms. And it's clearly one of the roles that human milk glycans play. There's also some evidence for direct immune stimulation, and also, particularly with regard to the sialic acid in neural development. But we've been focused on the enrichment of specific microbes, in a sense, a prebiotic enrichment coming from human milk. And one of the things I want to point out -- and this is a nice slide I got from Lars Bode in San Diego -- is the difference in structural diversity that human milk delivers in terms of free oligosaccharides compared to current prebiotics, which are up here on the right. And often current prebiotics are put forth as very similar to the glycans in human milk. And I think you can see from this slide that that's really just not true. Human milk has a range of structures that are present. And one could imagine that it has a -- that range of structures have a range of activities that evolve to be. What does breast milk enrich in infants? Well, I talked to you before about what Tissier identified over 100 years ago. Well, that has been validated a bunch of different times with non-culture-based studies since. And, in general, breast milk enriches bifidobacteria populations. And people have done PCR surveys, and they've done fish surveys. And the most recent one, of course, is the slide that a lot of people in this talk have been using, from Jeff Gordon's lab, which looked at infants -- excuse me -- people across time. If focused -- if you focus in on the infants, particularly during probably the lactation stage or the breastfeeding stage, up to 75 percent of all babies had amplicons that matched to bifidobacterium. So, in a sense, this is the one common diet we have, and it enriches a very unique and common clade of bacteria in us all, apparently, at least a good chunk of us. Other folks have found the same phenotype, and there's a variety of high level of bifs -- or instance, this is a Japanese study -- that are enriched in babies, and, in this case, one month of age. But they also noticed that there's a clade of infants that had a low level of bifs. And it was sort of a split with sort of a no-man's land in between. And we've noticed the same phenotype in a lactation study at UC Davis. And this is some work presented by Zack Lewis here -- that we noticed that babies sort of fall into either a low bif or a high bif clade. And we were curious about that, and I'll come back to this in a second. But why bifidobacteria? Why would you enrich bifidobacteria, or why would nature enrich bifidobacteria in an infant? Well, we can get some clues by the folks who do work on probiotics. And so this is a wonderful work published in Nature a couple of years ago by a Japanese group that was screening different bifidobacteria for its ability to prevent infection of an E. coli 0157 strain. And so they were just rotating different bifidobacteria through to see which one was protective. And they found a couple of strains. And this one -- it was a B. longum -- that was protective against the 0157 challenge. And they went and they sequenced the strain, and tried to figure out, well, what's different about this strain than the strains that weren't protective? And they noticed that, basically, it was the ability, in this case, to consume fructose to ferment the particular sugar source that was present in the mouse chow. So in other -- and they showed, of course, that because it could ferment fructose better, it had the genes for the transport and metabolism of fructose, it produced more acetate. And they went on to show, really nicely, how acetate was able to modulate tight junction function in a protective way. And so this really leaves one with a rather simple explanation of why bifidobacteria might be effective inside of breastfed infant, is that there's a whole bunch of human oligosaccharides coming down, and they are wonderfully able to consume them. And, of course, one of their major end products is acetate. So maybe that's one of the simple rationales for why this is protective. We have some other work going on with Chuck Stevenson and some folks in Bangladesh who are doing a vaccine trial. And they are trying to understand the impacts of -- this is actually a vitamin A study, in fact, a vitamin A on their various vaccine responses that they're looking at in the first year of life. And we've been following the microbiota of these kids, and they're all breastfed kids. And if you look at the data here, these are a bunch of different kids, and they're each measured at three time points: six, 11, and 15 weeks. And they all have a bunch of acinetobacter bacteria, mostly that's all bifidobacteria in them, and whereas some of the kids at the far end of the scale here don't. And one of the things they noticed when they started segregating out this data is that the vaccine responses that they were testing were much better in the kids that were more colonized by bifidobacteria. So maybe there's another rationale here for why bifidobacteria are protective in stimulating a healthy immune response. Okay, different moms might secrete different types of milk glycans. And I know at a previous talk it was mentioned how the FUT-2 allele expresses one 2' fucosyl linkage -- this one right here -- on various glycans that are on your epithelium. Well, of course, mother's milk has those same kind of glycans. And so a mom might be a secretor mom or a non-secretor mom. And really, it's whether you produce that linkage or not. So we were kind of curious: Is this FUT-2 allele linked, and the delivery of secretor milk or non-secretor milk, linked to any of the differences in bifidobacterial populations that we see? Previous studies have shown that secretor milk is differentially protective than non-secretor milk. And this is a work some time ago by Ardeth [spelled phonetically] Morrow and David Newburg, that showed different levels of the two fucosyl linkage seemed to be protective, at least in an epidemiology-type analysis. So we were curious if it -- maybe the secretor/non-secretor separation is the reason we see this separation here. And so we did some looking. And it does appear that those moms that were secretors, delivering secretor milk into their babies, enriched more bifidobacteria. Those moms that weren't secretors enriched less bifidobacteria, at least over time. And you saw larger populations in case of streptococci or enterococci -- or Enterobacteriales. And you can kind of see it better here in terms of the percent babies with bifidobacteria established from either six to 120 days of life. Basically, there's more bifs and they established faster in a secretor -- a baby getting secretor mom milk than one that's getting non-secretor mom milk. And it even separates down to species types, too. You see more B. longums in the secretor babies, as it were, and more B. brevis in the non-secretor. Well, can you see these glycans being consumed in the feces? And that's one of the things we wanted to try to understand. And so working with Carlito Lebrilla again, we were able to simply glycol profiles -- not simple, but glycol profile the feces. And I would then do the microbiota of the feces at the same time. And so you get a profile that looks like this. This is a day -- week one, week two, week four, and week 12, and these are a bunch of different master charge ratios that represent, in a sense, different degrees of polymerization, different glycans, not at the isomer levels, but just different composition. And you normalize it to the first week, and then you watch it over time. And you can see, in this baby, we see glycans go up a little bit by week two, but then they dramatically drop. And if you go look at the microbiota shifts, we see a large increase in bifidobacteria by week 12. And so we start to see correlations between bifidobacteria populations and glycans disappearing from the feces. And this happens to be a B. longum/infantis clade. But Carlito Lebrilla's lab's able to go down to the isomer level. And so they can take any one of those master charge ratios and split it up by the various isomers that are actually present, and figure out, is it a one-two linkage that's going away, or a one-four linkage that's going away. And so we get wonderfully detailed data on which specific glycans are going away in these poops. And we've been working on this on a variety of infants now, and trying to map the bifidobacteria populations that are present and the glycans that are missing. But you got to be careful about associations. There's been a lot of talk about associations at this meeting. [laughter] Even though things look really obvious, you don't necessarily know the order. I think you could imagine a lot of different orders for those. [laughter] So you got to be careful on associations. And so we need mechanisms. So which bifidos actually grow on human milk oligosaccharides? We've done a lot of screening. Often, most of the infant-borne bifidobacteria do grow on some of the base human milk oligosaccharides that are present -- lacto-N-tetraose, lacto-N-neotetraose. But once you start adding sialyl oligos or fucosyl oligos, that's when it starts to differentiate. B. infantis, which is actually B. longum subspecies infantis, grows pretty much on everything you throw at it. B. bifidum also grows on most things you throw at it. B. longum -- and that's B. longum, subspecies longum, and B. breve, on the other hand, really just do well mostly on lacto-N-tetraose and lacto-N-neotetraose. Although we have isolated, and sequenced, and have characterized a variety of B. brevi and longum strains that grow really well on HMO and they -- that'll be for a different time. Justin Sonnenberg and Angela Marcobal did a wonderful job of examining competition in an gnotobiotic mouse model between B. infantis, which we know grows well on this particular sugar -- lacto-N-neotetraose, and bacteroides B. theta, which also grows well on human milk oligosaccharides. And what they did is they just put the sugar into the water at a specific moment, when they had both organisms in an gnotobiotic mouse. And when they did that, B. infantis suddenly dominated that population wonderfully. As soon as they took it out of the water, went right down to parity with the bacteroides, and then when they put the sugar back in the water, it goes right back up. And so we see two organisms that actually are able to grow on the same sugar, but in competition in situ in an gnotobiotic mouse, for whatever reason, bifidobacteria are wonderfully competitive. And we're trying to understand that. We've been profiling the different glycans that bifidos consume. This is work done quite a few years ago. And this is, again, master charge ratios across here. And we noticed that that bottom bolus of oligosaccharides that moms deliver are wonderfully consumed by B. infantis, but not so much consumed by some of these other strains. The other strains, again, just consume lacto-N-tetraose. But Carlito can look at individual isomers. And so we've now been creating heat maps from a variety of strains like this, where the size of the bubble here, over this particular sugar, represents the amount of consumption. And so we're starting to now pattern a range of strains, and there's lots of strains differences. And we can try to relate this back to, perhaps, what we see in the feces. What about whole genome analysis? Because not all these strains grow on human milk oligosaccharides. So we've been comparing the strains that do and that don't grow, and sequencing lots of genomes. This is a couple representative genomes that we started off with. And it's rather easy to find the genes associated with deconstruction of these glycans. There are a variety of glycosyl hydrolases and transporters. And in B. infantis, they stick out pretty nicely in a single cluster, allows us to make a model. We've done RNA-Seq and proteomics, and clearly showed that genes unique to milk-associated bifidobacteria are upregulated during growth on milk sugars. And that give us a model for how this catabolism actually happens. This is a slide; that's one slide for a two-page D thesis. [laughter] And so we characterized all of the sialidases, fucosidases, hexosaminidases, galactosidases, and the surface-binding proteins in B. infantis, and showed which ones are induced and which ones are actually involved with human milk oligosaccharides or not. And please forgive me, Dave and Dan, for doing that in one slide. If you grow bifidos on human milk oligosaccharides, they also bind to Caco-2 cells better in and in vitro model. And they -- and protectively modulate. And they induce tight junction proteins in a protective way and anti-inflammatory cytokines. And this is really not something people pay attention to. So when people study probiotics, they don't really pay attention to what sugar am I growing the probiotic on when I'm doing the test? And this comparison is with the same strain grown on lactose. And so it's something that I would like the probiotic folks to think about. We have to think about what the probiotics are growing on in situ, and hopefully design our in vitro tests that we're trying to match that. So we end up with a model where B. infantis is able to transport and deconstruct these very complex oligos inside the cell. There's another model for B. bifidum. It does most of it outside the cell. And you can imagine how that might create different competition strategies inside the intestine. We think B. infantis can, of course, compete well, eat all the different glycans, and protectively modulate. And that allows us to make an initial proposal that complex milk glycans enhance this particular probiotic effect. But can we translate this? We've been working with Mark Underwood who works in the neonatal intensive care unit at UC Davis, and he's been doing a study where we put a HMO-positive B. infantis into premature infants. Of course, premature infants have dramatic dysbiosis. Often they're colonized by proteobacteria. And so a common approach to try to help them is to use probiotics of a variety of sort, bifidobacteria being one of them. And so we were looking at an HMO-minus B. lactus strain versus an HMO-plus B. infantis strain. And this was just published in the Journal of Pediatrics. When we -- when Mark added the B. infantis strain in, when he was feeding formula to these infants, it never really colonized at a high level in these infants. But when he fed it to infants that were getting mom's milk, it dramatically rose and persisted, persisted through the washout, and actually persisted when we added the B. lactus in. You never saw the B. lactus. And so I think this is -- this gets us to a stage where we start to understand that milk can really provide us a model of how to modulate the microbiota, and that the specificity of that modulation is driven in part by the glycan complexity that's present in milk, and -- and this is most important -- the cognate bacterial catabolism, the ability of the bacteria to actually consume it. B. animalis lactis that we used in as the control in Mark's study is what we use as our negative control for HMO growth. So intelligent understanding of the strains is really critical to be able to partner, perhaps, with specific prebiotics to make more effective therapies. This is a lot of detailed mechanistic research. And, of course, mechanistic research is in the basic realm, but it's also wonderful for translational purposes because it leads to diagnostics. We have all sorts of ways of understanding if a bifidobacterium that we put into somebody is actually acting the way we would predict it should. And that gets to my gaps, needs, and challenges, with several minutes left to go. I like to show this slide, and it's a pretty common one when people are talking about metagenomics and the influence between the host and the microbiome, in that there's so much we don't know. And we have such wonderful tools now. We're just starting to chip away at all of these subjects. And so when we talk about gaps, needs, and challenges, I look at this, and I start -- there's so many challenges when I look at it. But I'm enthused by the amount of new tools that we have. Within the last five years, we suddenly can sequence in a normal research lab that hasn't been doing microbiome work. I mean, the translatable tools become much more accessible. One of the points I want to make with this slide -- and it was made by Johanna Lampe as well -- is that we don't take a lot of dietary effects into our microbiota work. And I think this is also a challenge for the food science world and the food industry. If the work I show you I hope has convinced you that if we have a detailed understanding of the structures of our food, we might have a better clue on how they're modulating the microbiota that are enriched inside of us. That can feed back to help us design better foods, not just drugs. We can hopefully design better diets and foods in this way. And so this would be one of the challenges. So I would list a couple of things. There's certainly more mechanistic research needed. And we not only need to do our systems biology, but we need to get down to the strain level and do examinations at the strain level. We do need to encourage interdisciplinary research. And Vince is somewhere around here; I want to thank him for saying that before. We also need to do it in a way where assistant professors don't get lost in that process. So we need better ways to encourage folks to enter this field that can be part of large grants. Certainly we need better animal models and continued tool developments. And I'm going to throw glycomics into there, that wasn't talked about so much at this meeting. But we need a lot more glycomics if we're going to study the various glycan structures. And we need to be able to stratify clinical populations. I have a bunch of different examples, and I won't go through them all [laughs] because I don't have any time. But I will leave that with -- I'd like to thank, of course, the team that I work with at UC Davis. I couldn't get anything done without the other professors I work with, and the amazing students and post docs that I work with. It's really a joy. Also, of course, acknowledge my funding agencies, and my conflict of interest statement. Thank you very much. [applause] Female Speaker:
Okay, our last speaker of the afternoon is Dr. Brett Finlay. And Dr. Finlay is professor of the Michael Smith Laboratories in the Department of Biochemistry, Molecular Biology, and Immunology at the University of British Columbia in Canada. Dr. Finlay.