Hyenas & Microbes

By: Connie Rojas, PhD Candidate at Michigan State University

It has been a year of traveling! Earlier this year, I traveled to the Masai Mara National Reserve, Kenya (MMNR) to conduct my field work, and currently, I am in Mexico City doing a visiting scholarship at the Universidad Nacional Autonoma de Mexico (UNAM)! In between, I visited UC Berkeley for the 2018 Science and Technology Centers (STC) Director’s Meeting and San Antonio, TX for the Society for the Advancement of Chicanos/Hispanics and Native Americans in Science (SACNAS) annual conference, where I shared recent research findings as part of a symposium with other BEACONITES.  I am extremely grateful that my dissertation research on host-microbe interactions and the spotted hyena (Crocuta crocuta) microbiome has a strong field, laboratory, and computational component that allows me to travel and work with different collaborators!

 

I am a 4th year PhD Candidate in Dr. Kay Holekamp’s behavioral ecology laboratory at Michigan State University, in the Department of Integrative Biology and the Ecology, Evolutionary Biology, and Behavior program (EEBB). For my dissertation research, I am using next-generation sequencing technologies to assess how microbes and host-associated microbial communities (‘microbiome’) affect their host’s physiology, fitness, and behavior, and how host themselves, are influencing their microbiomes. I study these questions in a wild population of spotted hyenas. Hyenas are highly social carnivores and apex predators inhabiting much of Sub-Saharan Africa. Their societies are structured by linear dominance hierarchies, wherein an individual’s position in the hierarchy determines its priority of access to resources. Their social groups are also characterized by female dominance and male-biased dispersal.

 

For my field work, I traveled to my laboratory’s field camp at the MMNR and lived there for 4 months! I conducted 3 projects, all which investigated the role of microbes and microbiomes in shaping their host’s phenotype. One project involved me swabbing decomposing beef daily in order to survey microbial community succession across various stages of decomposition in the savanna environment. I wanted to emulate the environment and decomposition process of the carcasses hyenas eat and determine the types of beneficial and harmful microbes hyenas are acquiring this way. My second project was tons of fun; I conducted scent discrimination trials to ascertain if hyena scent gland secretions and their odors, which are hypothesized to be produced by microbes, contain information about the sender’s age, sex, and residency. I presented juvenile and adult female hyenas with the paste of two strangers (i.e. an immigrant male vs. adult female) and recorded the amount of time they spend sniffing each specimen. If my analyses show that hyenas spend a differential amount of time sniffing the paste samples, then this would indicate that the samples encode different information, and more importantly, that microbes are indeed contributing to their host’s chemical signaling! My last project was also very enjoyable and allowed me to interact with other animals in the reserve. I collected fecal samples from various species of antelope, elephants, and baboons to determine the role of host phylogeny in structuring the gut microbiomes of mammals in the savanna.

Right now, I am working with Dr. Valeria Souza (who gave an EEBB seminar in 2017; that is where I met her!) and Dr. Luis Eguiarte from the Institute of Ecology at the Universidad Nacional Autónoma de Mexico (UNAM) on the bioinformatics portion of my BEACON-funded gut microbiome project. This project investigates the socio-ecological drivers of gut microbiome structure and function in spotted hyenas, as well as its stability, its transmission across generations, and its potential to act as a reservoir for antibiotic resistance. We are using shotgun metagenomics (i.e. whole genome sequencing) to profile gut microbial community function and determine the metabolic pathways being provided by the community as a collective. Specifically, in their lab, I am being trained on the assembly, binning, annotation, and phylogenetic profiling of shotgun metagenomic data. From this data, we will be able to profile the taxonomic composition of the hyena’s gut bacterial communities, reconstruct the hyena’s diet, and survey the diversity of antibiotic resistance genes harbored by the community. We will also determine the relative importance of viruses in driving the evolution of these gut microbiomes and assay their heritability across generations and within an individual’s lifetime. The bioinformatic analyses are challenging and time-intensive, but I am making progress and it has all been very fun! Apart from work, I have been spending lots of time getting to know the city, eating as many tacos as I can, and making friends.

Although I am not looking forward to the bitter cold when I return to Michigan in January, I am looking forward to teaching my first class, taking a course on teaching college science, and co-organizing the 2019 EEBB Research Symposium, among other things. Until then, I am going to make the most of my time here in this great city!

Posted in Uncategorized | Comments Off on Hyenas & Microbes

Avida-ED in Action

BEACON scientists and educators featured in MSU Today reporting on a recent publication highlighting the use of Avida-ED in a newly developed undergraduate biology course. The course IBIO150 Integrative Biology: From DNA to Populations, was developed with non-Biology STEM majors in mind, students who need a rigorous major’s level course that covers core concepts emphasized in undergraduate biology reform. The unique component of the course is the incorporation of a digital evolution lab that uses Avida-ED, featuring series of exercises designed to address important concepts in evolutionary biology. In addition, students complete independent research projects. The incorporation of Avida-ED into the course is supported by current grants to the College of Natural Sciences through the Howard Hughes Medical Institute, in addition to an Improving Undergraduate Science Education (IUSE) grant Active LENS: Learning about Evolution and the Nature of Science. Avida-ED and other resources developed by BEACON have recently been featured in a video produced for the National Science Teachers Association (NSTA) TV.

Posted in BEACON in the News, Education, Evolution 101 | Comments Off on Avida-ED in Action

Ecology/Evolution Scientific Symposium at the SACNAS National Conference

The Society of Systematic Biologists (SSB) and BEACON have collaborated to organize a scientific symposium at the Society for the Advancement of Chicanos/Hispanics and Native Americans in Science (SACNAS) national conference in San Antonio, Texas, on October 11th, 2018 (10:30AM-12:00PM, Room 225B). The symposium title is “It’s Complicated: The Ecology and Evolution of Microbes and Their Hosts.”

Host-microbe interactions are ubiquitous and often drive evolution. Microbial parasites or pathogens harm hosts, whereas other host-associated microbes are beneficial or even necessary for host health. Diverse scientists at the forefront of the ecology and evolution of host-microbe interactions will provide a synthetic perspective of research on this important topic.

Speakers:

  • Connie Rojas, PhD Candidate – Michigan State University
  • Luis Zaman, PhD – LSA Collegiate Postdoctoral Fellow, University of Michigan
  • Lisa Barrow, PhD – NSF Postdoctoral Fellow, University of New Mexico
  • Kat Milligan-Myhre, PhD – Assistant Professor University of Alaska Anchorage

SACNAS is a national organization focused on increasing the proportions of underrepresented minorities in science, technology, engineering, and math (STEM) fields. It is the largest multicultural and multidisciplinary STEM diversity organization in the country. In 2017, the National Conference was attended by 3,845 scientists from diverse backgrounds, with 77% of the participants members of ethnic/racial groups that are significantly underrepresented in STEM fields.

Many people may not realize the complicated and often important relationships occurring all around (and inside!) us with microbes and their hosts. As a result of attending our scientific symposium session, attendees will learn about: (1) exciting, recent advances in science research that illuminate how microbes can drive the ecology and evolution of their hosts; (2) the questions and approaches for studying host-microbe and host- parasite interactions; (3) career options within ecology and evolution from presenters at diverse career stages;  and (4) the central role of ecology and evolution to the life sciences. Read the abstracts, below, for more details about each talk.

In addition to the scientific symposium, BEACON and SSB are also sponsoring a day-long Ecology/Evolution field trip on October 13th to Mitchell Lake Audubon Center and the San Antonio Zoo. The symposium and field trip were co-organized by Eve Humphrey, Maurine Neiman, and Alexa Warwick, with support from the Society for the Study of Evolution Diversity Committee and Education and Outreach Committee. The speakers, organizers, and other attendees will also participate in an Ecology/Evolution session of “Conversations with Scientists” to share scientific career options on October 11th (5:45-7:15PM, Room 221C). If you’ll be at SACNAS 2018, we hope to see you at one or all of these events!

Connie Rojas – Host and Ecological Traits Shape the Structure, Function, and Diversity of the Gut Microbiome in Wild Spotted Hyenas

Animal bodies harbor complex microbial communities, hereafter termed microbiota, that exert profound effects on their physiology, behavior, and evolution. In the mammalian gastrointestinal tract, resident microbes are known to synthesize essential vitamins, supply their host with energy released from the fermentation of indigestible carbohydrates, competitively exclude pathogens, and promote immune system and tissue development. In spotted hyenas (Crocuta crocuta) meerkats (Suricata suricatta), and ring-tailed lemurs (Lemur catta), microbiota inhabiting scent-gland secretions co-vary with the gland’s odorous metabolite profiles and contain well-documented odor producers, indicating they likely contribute to their host’s chemical signaling behaviors. Furthermore, in four species of insectivorous bats, bacteria isolated from the skin exhibit anti-fungal properties against the causal agent (Pseudogymnoascus destructans) for white-nose syndrome, suggesting a beneficial role of these microbes in pathogen defense. However, despite the importance of the microbiota, we know little about the forces shaping its structure and function, especially in wild animal populations. Here, I use 16S rRNA gene sequencing technologies to a) survey the gut microbiota of wild spotted hyenas and b) investigate the host social and ecological factors affecting the gut microbiota. Specifically, I assay if gut microbiota diversity and structure vary with hyena age, reproductive state, group size, and temperature and precipitation. Overall, this research will contribute to our understanding of the ways a host shapes its microbial communities, and how microbial communities, in turn, influence their host’s behavioral phenotype. In this talk I will also share a bit about my journey as a scientist from the perspective of a Latina first-generation college student and daughter of immigrant parents.

Dr. Luis Zaman –  Experimenting with Digital and Microbial Evolution

My path to evolutionary biology was unusual. I started as a computer scientist and ended up working in a wet lab with microbes and viruses. I’ll talk about how I ended up where I am, what I’m doing now, and why disciplined and undisciplined science are important in research.

 

 

 

 

Dr. Lisa Barrow – Variable Host Susceptibility and Enigmatic Parasite Distributions: Insights from Museum Collections and Genomics of Avian Haemosporidians

There are several outstanding questions in ecology and evolution of host-parasite interactions. Why do host species vary so drastically in their susceptibility to parasites? How localized or widespread are different parasites? What are the environmental and host range limits to parasite distributions? These questions are particularly important given the predicted influence of climate change on species distributions and the potential for emerging infectious diseases. Avian haemosporidians are intracellular parasites that infect birds across the globe, sometimes with devastating consequences. Together with multi-institution, student-driven teams, we have been tackling two complex avian haemosporidian systems in Peru and New Mexico, USA. Using molecular and microscopic screening of extensive museum collections, we found that ~35% of birds are infected. In Peru, we screened nearly 4,000 birds representing 40 families and 523 species. After accounting for several environmental, life history, and ecological predictors of infection, we found that host phylogeny explains substantial variation in infection rate. In other words, susceptibility is deeply conserved across the avian tree, and is likely related to conserved aspects of the immune system. In New Mexico, we sampled avian haemosporidian communities in three mountain ranges to better understand the limits to parasite distributions. Haemosporidian communities exhibit structure on fine spatial scales, with most lineages occurring in a single mountain range, but a few widespread generalists infecting multiple host species. Ongoing work incorporating new genomic methods is improving estimates of the host and environmental range limits of haemosporidians, providing important baselines for identifying potential host switches or range expansions.

Dr. Kat Milligan-Myhre – Use of an Evolutionary Model to Determine the Role of Host Genetic Background on Microbiota

Microbiota are the microbes that live in and on a host. Disruption of the microbiota can lead to painful inflammation in the host, which can become chronic, as in the case of inflammatory bowel disease. Our lab focuses on the role the host genes play on the relationship between the microbiota and their host. Thus, we adapted the evolution and biomedical model organism, threespine stickleback fish (Gasterosteus aculeatus), for host-microbe studies. Stickleback are ideal for these studies due to their large family sizes, genetic variation within and between populations that is similar to human genetic variation within and between populations, and the tools available to study these interactions. We compared the development and behavior in fish raised germ free, with conventional microbiota, with mock communities of up to eight microbiota members, or with microbiota disrupted by antibiotic or environmental contaminants. We found that the populations varied in their response to these manipulations, indicating that the genetic variation between the populations contributed greater to the relationship between microbes and the host than the variation within the populations. We will use these results as a basis for future studies to identify the critical windows in development in which disruptions to gut microbiota result in short- and long-term consequences to host health, and determine the extent to which the host genetic background contributes to the ability of healthy gut microbial communities influence to fitness.

Posted in BEACON Researchers at Work, BEACONites, Diversity in STEM, Member Announcements | Tagged | Comments Off on Ecology/Evolution Scientific Symposium at the SACNAS National Conference

Evolutionary Computation Experts Video Collection

This blog post is by Risto Miikkulainen [1,2], Paul Jarratt [2], and Andrew Turner [2] from (1) The University of Texas at Austin and (2) Sentient Technologies

Given recent advances in evolutionary computation technology, available computational power, and opportunities for AI in the real world, we believe evolution is on the verge of a breakthrough, i.e. becoming the next Deep Learning. In order to chart the possibilities as well as challenges, we sat down at Sentient and at GECCO 2018 with a number of EC experts in both academia and industry to share their ideas about where AI is heading, and the role evolutionary computation can play in its future. The result is a collection of video interviews; they are organized around a number of specific questions so that you can explore those of interest to you. You can check out the collection at https://www.sentient.ai/labs/experts.

Sentient plans to add more experts to this page in the future, so let us know if you’d like to contribute your point of view to this collection!

Posted in BEACON Researchers at Work, Education, Videos | Tagged , , , | Comments Off on Evolutionary Computation Experts Video Collection

A tale of scales and feathers… (ice and volcanoes)

This blog post is by MSU postdoc Murielle Ålund.

It all started with a-real life tetris game: trying to fit six gigantic coolers full of field gears, luggage, two kids and three adults into what was going to be our field car. And yet Greg Byford (Boughman Lab Research Technologist) had asked the rental agency for the biggest car they had!

Once finally on the road, as the amazing Icelandic landscapes were unfolding at each turn, I remember thinking that the trip – (and the postdoc position and the big move oversea with my family)- were already worth it, no matter how field work would go. And Iceland did not disappoint! It was such an amazing experience to spend almost six weeks in this beautiful country and to learn about Icelandic culture and fish.

Yes, the fish! I should probably have started with them, this is not a travel blog after all! The reason I got to visit such amazing places is that we are studying Threespine sticklebacks (Gasterosteus aculeatus) and how they adapt to their local environments in the Artic, a project led by Dr Janette Boughman. These tiny fish are impressive in their ability to colonize new habitats and adapt to very different environmental conditions all over the Northern hemisphere. They are very well known for their ability to repeatedly invade freshwater bodies from an original marine morph, which comes with drastic changes associated with the differences in salinity, food and predators. As Icelandic glaciers have been retreating for thousands of years and continue to do so at an accelerated rate (with a current estimated yearly surface loss of 0.2%)1, new lakes are continuously formed that can eventually be colonized by different populations of fish.

These glacial lakes are fed by constantly melting ice, and are thus extremely turbid, so much so that you can see the difference in color by looking at satellite images of the area! Once you have lost a trap in one of these lakes and realize that you cannot even see your own fingers in the water (making for an interesting time trying to find said trap by feeling it with your feet), you start wondering how sticklebacks can find food, avoid predators and choose a mating partner in these conditions!!!

This brings me to the main goals of our project: studying how sticklebacks adapt to widely different environments in these harsh and rapidly changing climates. We are specifically interested in the evolution of sensory systems, the idea being that other senses might compensate for the extreme low visibility of the turbid waters. To test that, we are collecting fish from glacial lakes, spring-fed lakes (these are very clear lakes) and from the sea and comparing their responses to visual and olfactory cues in a controlled behavioral experiment. We are then sampling and measuring their eyes, noses, lateral lines and comparing the different parts of their brains, to get a complete overview of their sensory systems and how developed their different senses are. In addition to being able to compare fish from lakes of different turbidity, we are also hoping to reconstruct a timeline of adaptation to these different habitats, as these lakes were formed anything in-between ten thousand and just hundred years ago, and have thus variable and known (max) colonization times. This will allow us to get an idea of the rate of evolution of sensory systems in sticklebacks.

This is a highly collaborative project. As I write, our amazing Icelandic team is still in the lab and in the field collecting fish and running trials. This includes our talented local technician Sven Wargenau from Hólar University, Julian Ohl (studying in Reykjavík for a Master’s degree in Environment and Natural Resources), the Boughman Lab’s brand new PhD student Brielle Dominguez and a visiting PhD student from Uppsala University: Javier Vargas Calle, specialist in gut microbiomes. Back here at MSU, Greg and I are starting to process some of the 1200 fish we already brought back and sending brain samples to the Hofmann Lab in Texas and eye samples to the Stenkamp Lab in Idaho, while the behavioral experiments are overseen by Dr. Jason Keagy at university of Illinois. To hear more about this project, come and listen to my talk at the BEACON congress next week!

That was for the tale of scales, ice and volcanoes. But the feathers you will ask? That would be enough for another full blog post… First let me say that the feathers were everywhere in Iceland, and bird watching there was amazingly easy: you can see rare species from your hot tub! But more seriously, my background is in studying speciation in birds, specifically two species of Eurasian passerines, collared and pied flycatchers (Ficedula albicollis and F. hypoleuca), that hybridize on a (much smaller) island in Sweden: Öland. During my PhD, I studied the consequences of hybridization for these two species and was particularly interested in their reproduction, and how sexual selection and fertility are affected by mating between different species.

My work involved quite a bit of fieldwork, catching, measuring and ringmarking thousands of young and adult birds, and collecting sperm samples and analysing them under the microscope directly in the forest. I first looked at what makes a male successful at siring as many chicks as possible in his nest and found that in collared flycatchers, sperm size matters, but differently depending on how “sexy” the males are: males with relatively small ornaments (white forehead patches) benefit from having long sperm, and vice-versa (coming out soon in Behavioural Ecology!2). I also found that hybridizing is really bad for these birds, as hybrids seem to be mostly sterile: the females lay empty eggs and the males do not manage to produce any functional sperm3. Since one of the species is pushed away from good territories by the other one, the females do not always really have a choice, and are sometimes constrained to mate with a male of the “wrong” species to secure a territory and food for their offspring. In a collaboration with a team at the Natural History Museum of Oslo, I found that females can still mostly avoid producing costly hybrids by seeking extra-pair copulations with males of their own species, and biasing against the sperm of the “unwanted” male inside their reproductive tract, so that it has fewer chances to fertilize their eggs4.

I find it fascinating that interactions between eggs, sperm, ovarian and seminal fluids can all influence the outcome of competition for fertilization and want to study this less understood phase of sexual selection in the future. Check my website for updates on this and more stories about birds, fish, cool behaviors and fascinating evolution!

 

Work cited:

1: https://notendur.hi.is/oi/icelandic_glaciers.htm

2: Ålund, M., Persson Schmiterlöw, S., McFarlane, S.E. and Qvarnström, A., 2018 Optimal sperm length for high siring success depends on forehead patch size in collared flycatchers, Behavioral Ecology, accepted

3: Ålund, M., Immler, S., Rice, A. M. & Qvarnstrom, A. 2013 Low fertility of wild hybrid male flycatchers despite recent divergence. Biol. Lett. 9:3, 20130169. (doi:10.1098/rsbl.2013.0169)

4: Cramer E.R.A. and Ålund†, M., McFarlane, S. E., Johnsen, A. and Qvarnström, A., 2016 Females discriminate against heterospecific sperm in a natural hybrid zone, Evolution 70 (8), 1844-1855. (doi: 10.1111/evo.12986)

Posted in BEACON Researchers at Work | Comments Off on A tale of scales and feathers… (ice and volcanoes)

Mapping Antibiotic Resistance in Pseudomonas aeruginosa Biofilms to Develop Better Therapies for Cystic Fibrosis

This blog post is by MSU graduate student Michael Maiden.

MSU researchers, Chris Waters, Michael Maiden and Alessandra Hunt, BPS, 04.30.18

Currently, I am a 7th year DO-PhD student in the physician scientist training program in Dr. Christopher Water’s drug development and biofilm laboratory in the department of Microbiology & Molecular Genetics. I was attracted to the Michigan State University and the College of Osteopathic Medicine and, specifically the DO-PhD program, because it offered the opportunity to work on clinically relevant projects that may lead to better therapies for patients in the future.

In the Waters’ lab my research is focused on developing new therapies for chronic infections caused by bacteria in the form of biofilms. Biofilms are a community of cells enmeshed in a self-made gel that renders the community up to 1,000x more resistant to antimicrobial therapies. For this reason, bacteria growing in biofilm communities are a major contributor to chronic infections and death.

One bacterial pathogen, that often infects and forms biofilms in patients, is Pseudomonas aeruginosa. In fact, P. aeruginosa is the leading cause of death in patients with cystic fibrosis (CF). CF is a debilitating genetic disease that results in dry and clogged airways, which trap bacteria and leads to life-long chronic infections, resulting in premature death between the ages of 30 and 40 YO.

A biofilm colony formed by P. aeruginosa surrounded by a secreted self-made mucus that makes the bacteria very difficult to treat.

By early adulthood, nearly 50% of CF patients are chronically infected with P. aeruginosa. To extend the lives of CF patients, it is essential to develop therapeutic interventions that eradicate P. aeruginosa before it is able to form a chronic infection in these patients.

We found that by treating with two specific antimicrobials, tobramycin and triclosan, we could kill up to 99% of P. aeruginosa cells growing in biofilm communities. Further, this combination was effective in as-little-as 2-hrs. These exciting results raised one very difficult question, how?

One way to determine how antimicrobials work is to go after well-known targets and pathways. By either turning them on or off, using various molecular techniques, you can test to see if your particular drug is working through that pathway. We tried this approach with little success. So, we turned to an un-biased evolutionary approach.

Using this method, we took advantage of the natural tendency for bacteria to evolve resistance to any antimicrobial given enough time and small enough doses so that some bacteria may survive and thus mutate their genome. We evolved P. aeruginosa cells growing in a biofilm and rendered them resistant to the combination, by slowing raising the dose with time. Next, we performed whole-genome sequencing to identify the genetic mutation(s) that could help to explain how they became resistant.

We found a novel mutation in P. aeruginosa renders the bacteria resistant to the combination. This mutation is located within an enzyme essential for protein synthesis. This gave us a valuable clue for how triclosan may be enhancing tobramycin activity, allowing us to formulate a model for how the two work synergistically. Subsequent experiments have supported this model.

Further, the mutation we identified in our evolution mutants has been identified independently in clinical CF isolates of P. aeruginosa, which renders them resistant to tobramycin. Thus, our artificial evolution work in the lab has been validated by the natural evolution taking place in the clinics, specifically in the lungs of CF patients.

We now have a possible lead for how our combination may be working synergistically against P. aeruginosa cells growing in biofilm communities. This new resistance mechanism could be targeted in the future to develop compounds that inhibit this resistance mechanism. Further, knowledge of this mechanism could pave the way for the future development of compounds that work in a similar fashion to our combination, thus, yielding much needed new antimicrobial therapies. Currently, we are exploring how this mechanism renders bacterial cells in a biofilm resistant to our combination.

As antimicrobial resistance continues to be a major threat to human health, it is important to develop better strategies that more effectively use our current antimicrobial arsenal. This combination may by an example of one such strategy. As a future clinician, I am grateful to be a part of a project with strong clinical implications. And as a scientist, I have always been interested in how organisms evolve. The opportunity to perform evolution studies in the lab is both exciting and rewarding, providing little hints into what sustains life and what great trials and tribulations all living organisms have gone through to maintain it.

Relevant MSU Today Articles:

https://msutoday.msu.edu/news/2018/ingredient-in-your-toothpaste-may-combat-severe-lung-disease

https://msutoday.msu.edu/news/2017/fighting-an-old-enemy-in-the-battle-against-cystic-fibrosis/

Posted in BEACON Researchers at Work | Comments Off on Mapping Antibiotic Resistance in Pseudomonas aeruginosa Biofilms to Develop Better Therapies for Cystic Fibrosis

Kalyanmoy Deb honored with IEEE Computational Intelligence Pioneer Award

BEACON’s own Professor Kalyanmoy Deb, the Koenig Endowed Chair in Electrical and Computer Engineering at Michigan State University, was honored today by the IEEE Computational Intelligence Society. At the World Congress of Computational Intelligence meeting in Rio de Janeiro, Brazil, he was given the IEEE Computational Intelligence Pioneer Award, which is given to at most one person each year who has made major contributions to the field. It recognizes contributions across one’s entire career. Prof. Deb was honored for his pioneering contributions to the field of evolutionary multi-objective optimization. Among those contributions was the algorithm NSGA-II, which has been more widely used than any other evolutionary multi-objective optimization tool. He has led the community in development of this new field that has spurred both widespread academic research and worldwide industrial application.

Posted in BEACON in the News | Comments Off on Kalyanmoy Deb honored with IEEE Computational Intelligence Pioneer Award

Learning an Evolvable Genotype-Phenotype Map

me!

This post is by MSU graduate student Matthew Andres Moreno

Hi! My name is Matthew Andres Moreno. I’m a graduate student finishing up my first year studying digital evolution with my advisor Dr. Charles Ofria.

Today, I’m going to talk to you about police detective work. Eventually, we’ll talk about evolvability and genotype-phenotype maps, but first let’s talk CSI.

Police Composite Sketches

Specifically, let’s think about how a police composite sketch works. First, someone sees a criminal and describes the face’s physical features with words. This description is the compact representation. Then, the police artist reconstructs the criminal’s face from the description.

Schematic of hypothetical police composite process. Mug shot and composite reconstruction were taken from the Crime Scene Training Blog.

Why does this work? It works because the witness has seen lots of faces and knows what the important bits to describe are. It works because the police artist understands the witness’ words and has also seen lots of faces — from experience, she knows that the mouth goes under the nose, the nose goes between the eyes, etc. and doesn’t need the witness to tell her absolutely everything about the face in order to draw it.

Well, autoencoders can also be used to reconstruct a corrupted input. This works something like a police sketch, too. Suppose that the criminal was wearing pantyhose that partially obscured his face. The witness can still describe the suspect’s face and the police artist can still draw it. Under the right conditions the missing part of the face can be reconstructed reasonably well.

Schematic of hypothetical police composite process with suspect in disguise (incomplete input). Mug shot and composite reconstruction were taken from the Crime Scene Training Blog.

Why does this work? It works because the witness can still see and describe part of the face. It works because the police artist understands the witness’ words and has also seen lots of faces — from experience, she can make a pretty good guess by cluing off the fact, for example, that faces have left-right symmetry or maybe that the criminal probably had a cheekbone and ear on the part of the face that was obscured. Again, because she’s seen lots of faces the police artist doesn’t need the witness to tell her absolutely everything about the face in order to draw it.

Deep-Learning and Autoencoders

The jig’s up.. it was all a set-up! A set-up, that is, to help you understand what autoencoders do. Unless you’re technically inclined, understanding exactly what autoencoders areisn’t particularly important for our discussion. Suffice it to say that what autoencoders are is a type of clever deep learning algorithm.

What autoencoders do is directly analogous to what the witness and police artist do. By looking at lots of examples of complex objects like faces, autoencoders learn to

  1. compactly describe the important features of a complex object (“encoding”, just like the witness) and
  2. reconstruct a complex object from that description (“decoding”, just like the police artist).

I’ll refer to these as the two powers of autoencoders.

The following graphic, a “latent space interpolation” between three faces, gives a neat glimpse of how autoencoders work and how powerful they are. The latent space refers to the set of all compact descriptions an autoencoder can read. To understand what’s going on here, let’s just look at the top row of images.

Autoencoder latent space interpolation with faces! Graphic from [White, 2016].

At the top-left, we see an image of a woman with curly hair. To get to the image immediately adjacent on the right, we use power 1 of autoencoders to generate a compact description and then use power 2 of autoencoders go reconstitute a face image.

Then, going left to right across the top row, things start to get interesting. We gradually change the compact description of the curly-haired woman until it matches the compact description of the red-haired woman on the far right. Each image shows an intermediate compact description that was reconstituted using power 2 of autoencoders. This visualization shows a very natural-looking transition between the two faces!

I won’t walk you through it, but the rest of the grid of images shown above was generated analogously.

Genotype-Phenotype Maps

What does any of this have to do with evolution? This year, I’ve been investigating how  autoencoders can be useful as genotype-phenotype maps in digital evolution. One idea of how this can work: use power 2 of autoencoders (the “decoder”) as the genotype-phenotype map. In this scheme, the genotype lives in the latent space.

In order to drive home the implications autoencoder genotype-phenotype maps on evolvability let’s talk through a little thought experiment. Think back the problem of police face reconstruction we’ve been thinking about. Suppose we’re trying to evolve a face that, as judged by the witness of a crime, maximally resembles the perpetrator. (Yes, this is a real thing people do [Frowd et al., 2004]). To accomplish this, we start out with a set of random genotypes that map to different phenotypes (images). The witness selects the images that most closely resemble the suspect’s face. Then, we mutate and recombine the best matches to make a new batch of images for the witness to consider As we iterate through this process, hopefully we generate images that more and more closely resemble the suspect’s face.

Consider trying to evolve a facial composite using the direct genotype-phenotype map. Under this map, the intensity of each pixel of the image is directly encoded in the genotype. First of all, the randomly generated images wouldn’t look very much like faces at all — they’d look more like static. Supposing that we were actually able to eventually get to an image that vaguely resembles a face at all, then what? Is there a path of pixel-by-pixel changes that leads to the suspect’s face where every pixel-by-pixel change more closely resembles the perpetrator’s face? I’d argue we’d be likely to sooner or later get stuck at a dead end where the image doesn’t resemble the perpetrator’s face but pixel-by-pixel changes to the image make it look less like the perpetrator’s face (or a face at all).

Evolving the composite using the direct genotype-phenotype map probably won’t work well.

What if instead of having the genotype directly represent the image at the pixel level, encode genotypes analogously to a verbal description then use a police artist who can draw a suspect from verbal descriptions to generate phenotypes. This is analogous to what the our “decoding” genotype-phenotype map, accomplishes.

(For those who are curious, software to evolve police composites use an indirect genotype-phenotype map based on eigenfaces [Frowd et al., 2004].)

Wrap-Up

This work — which we call AutoMap — was, in part, inspired by recent efforts efforts to understand evolvability in terms of learning theory [Kouvaris et al., 2017]. We hope that this work helps to strengthen an explicit connection between applied learning theory (i.e., machine learning) and evolvability. We’re also looking forward to expanding on the exploratory AutoMap experimental work that we’re taking to GECCO this summer.

If you’re interested in more detail, This blog is based on a more in-depth (but still non-technical and fun!) introduction to our work with AutoMap, which you can find here. If you want to check out our technical write-up on AutoMap, you can find the PDF here and the paper’s supporting materials here.

Finally, thanks also to my AutoMap coauthors Charles Ofria andWolfgang Banzhaf.

References

Frowd, Charlie D., et al. “EvoFIT: A holistic, evolutionary facial imaging technique for creating composites.” ACM Transactions on applied perception (TAP)1.1 (2004): 19-39.

Kouvaris, Kostas, et al. “How evolution learns to generalise: Using the principles of learning theory to understand the evolution of developmental organisation.” PLoS computational biology13.4 (2017): e1005358.

White, Tom. “Sampling generative networks: Notes on a few effective techniques.” arXiv preprintarXiv:1609.04468 (2016).

Posted in BEACON Researchers at Work, Member Announcements | Tagged , , , | Comments Off on Learning an Evolvable Genotype-Phenotype Map

On the Hunt: How Bacteria Find Food

This post is by MSU graduate student Joshua Franklin

Imagine you are half-starved, blindfolded, then placed into a large gymnasium with a plate full of freshly-baked cookies. How do you find the cookies? You could try to randomly walk around until you step on them. Interestingly, you are mathematically guaranteed to reach the cookies this way, even if the gym is infinitely large[1], but that could take a very long time, and you’re hungry now. Luckily, you can use your senses to speed things up.

It turns out that smell is the only useful sense here, as your vision is blocked and your other senses are either too short-ranged (taste, touch) or functionally unable (hearing) to help you find the cookies. But, generally, you cannot use smell to turn towards distant objects, because the change in odor intensity is too weak to detect when standing in place and turning around. So how can you find the cookies?

You could start by sniffing, walking for a while, then sniffing again. If the odor got stronger, then you must be moving towards the cookies. Great! You should keep moving in that direction. If the smell got weaker, that’s okay; just spin around and try again in a new direction. By following this procedure you will reach the cookies much more quickly than before (Figure 1).

Figure 1. A computer simulation of the paths taken by a person walking randomly (red) and a person using the cookie-finding procedure (blue). The black spot is the plate of cookies. Both strategies were allowed to walk for the same amount of time and both walkers move at the same speed. Even though the random walker starts closer to the cookies, they make far less progress than the cookie-finder.

It turns out that many bacteria use this cookie-finding procedure to help them move towards food, and in this context the procedure is called chemotaxis. Why do bacteria need to do this? Well, many kinds of bacteria can move using a flagella, which is a long filament that sticks off of the cell surface and rotates, working like a propeller to push the cell forward. But just like our noses are too weak to directly help us turn toward the cookies, bacteria are too small to directly sense the direction food is in. So bacteria use chemotaxis to move towards their food by swimming, sensing whether it has moved towards or away from the food, then deciding whether to keep swimming in the same direction or to change directions.

But there’s a hidden assumption here. Think back to the cookie-finding procedure. You smelled, walked, then smelled again and compared the current smell to the previous smell. That means that you need to remember what the smell was previously. Can bacteria actually do this? It turns out the answer is yes. A series of chemical reactions inside of the cell stores information about how strong the “smell” of the food was previously, so that the bacteria can tell if they have moved toward or away from the food source.

The reason chemotaxis works so well for bacteria is that, at the size of a bacterium, diffusion spreads chemicals into gradients very quickly. In fact, the bacterial world is dominated by diffusion-generated chemical gradients. This is hugely different from the world we normally see, where diffusion plays a only minor role[2]. From a bacterial point of view, the world is a series of chemical gradients that can lead them toward food or away from predators, and chemotaxis enables the cells to navigate these gradients effectively.

Figure 2. A diagram of the chemotaxis system in the E. coli. The network that controls the cell decision-making process is composed of only a handful of different proteins. Stars are molecules cells can “smell”, and the rectangular white bars are the sensory proteins. Letters in shapes are chemotaxis-associated proteins. Filaments coming off of the cell surface are flagella, which rotate to push the cell forward.

Our lab seeks to identify traits that affect how cells perform chemotaxis. Bacteria carry out chemotaxis using a set of proteins that detect food molecules outside of the cell and control the cell’s movement (Figure 2). Biochemical processes in this system control how well the cell performs chemotaxis in different environments. Cells which swim towards food more effectively will reproduce more often than cells which can’t, so natural selection will tend to optimize the chemotaxis system for a given environment.

While we have a pretty good understanding of how the chemotaxis system works, it is still difficult to predict how the strategy should be optimized to suit different environments. For example, how long should a cell swim before it is confident that it is going in the right or wrong direction? What if there are obstacles along the way? Which of the many chemicals that can be sensed should be followed? There are still many questions with unintuitive answers that need to be explored to understand why we see so much diversity in motile behavior and morphology in the microbial world.

Our lab has created a web app to explore the evolution of chemotaxis. The virtual environment consists of a rectangular world with a food gradient that increases from left to right. The simulation places cells into the world where they can perform chemotaxis to move towards the food. Each cell is given a number of traits, which control the chemotaxis system as described above and their fitness in the virtual world.

In this simulation cells that are better at chemotaxis reproduce more often than cells that are not as good. Eventually, cells with traits that are optimal for the specified environment will dominate the population. There are a number of environmental details you can change, such as the strength of the chemoattractant, the shape of the chemical gradient, and the presence of obstacles that the cell must navigate to reach the target. Different environments will select for different chemotaxis traits. For more details, see the guide on our website.

While the simulation is a simplified version of the E. coli chemotaxis system, it reproduces the behavior of real bacterial cells really well. Modeling and simulations allow us to explore the behavior of bacteria to generate hypotheses that can be tested in the laboratory. Part of my research involves using models and simulations to understand the performance tradeoffs that are imposed on bacterial motility.

The combination of biological and computational work in my research is a fantastic opportunity for me, as I have enjoyed programming since my early teens, and basically living in a forest as I was growing up nurtured my interests in biology. Being able to combine both of my main interests helps keep me engaged in my research despite the challenges it poses. When experiments are testing my patience I can generate news ideas by doing computer work, and when computer work wears thin I can refresh by returning to the bench.

Figure 3. Me in action at MSU Science Fest 2018.

One of my favorite features of my research is that, as this blog post shows, it provides an opportunity to design and write programs that allow students to study non-intuitive aspects of biology using interactive tools, without the constraints of setting up experiments. I think it can also help students appreciate the power of modeling and simulations in exploring the complexity of biological systems. Using educational programs for public outreach and education is something that I feel strongly about (Figure 3), and hope to expand upon as my graduate career continues.

[1]Technically, this assumes the floor is two-dimensional and space is discrete at some level. Perhaps even more interestingly, a bird flying around randomly in an infinite gymnasium would not be guaranteed to ever reach the cookies, even if given an infinite amount of time. For an approachable explanation of both phenomena, see: https://www.youtube.com/watch?v=stgYW6M5o4k

[2]When in school, I remember our teacher explaining diffusion by asking us what happens when when someone breaks a bottle of perfume in a store. We said “Everyone starts to smell it”. The teacher explained this as diffusion, but in reality it’s due to air currents. For example, in perfectly still air it would take about three days for oxygen (D = 0.176cm2/s, according to Wikipedia) to diffuse a distance of 10 feet. See: http://www.physiologyweb.com/calculators/diffusion_time_calculator.html

 

Posted in BEACON Researchers at Work | Tagged , , | Comments Off on On the Hunt: How Bacteria Find Food

Gallium cannot be used as a Trojan horse to fool Iron-selected bacteria

This post is by NCAT postdoc Akamu Jude Ewunkem, faculty Misty Thomas, grad student Sada Boyd, and faculty Joseph Graves Jr.

Antibiotics have heretofore been used as therapeutic agents (Butler et al., 2017). However, bacteria are increasingly developing resistance to these therapeutic agents. Due to this continuum of resistance evolution there has been a stagnation in the development of novel antimicrobial agents to treat multidrug-resistance organisms. However, alternative therapeutic options are currently being exploited for treatments. Iron acquisition is an alternate target of antimicrobial agents.

Iron is an important micronutrient for virtually all living organisms. Iron is involved in a wide variety of important metabolic processessuch as photosynthesis, respiration, the tricarboxylic acid cycle, oxygen transport, gene regulation and DNA biosynthesis (Weinberg, 2009). In bacteria, iron influences cell wall composition, intermediary metabolism, secondary metabolism, enzyme activity, and host cell interactions. Bacteria use surface proteins, heme group and siderophores for acquisition of iron. The competition for iron between host and bacteria is so important that many multicellular organisms have evolved defense mechanisms that sequester iron away from pathogenic microbes.

Gallium is a transition metal element, and has a similar ionic radius to that of iron. Thus, gallium can efficiently compete with iron for binding to iron-containing enzymes, transferrin, lactoferrin and siderophores. Gallium is used as a trojan horse to iron-seeking bacteria. Invading bacteria are tricked, in a way, into taking it up. However, while the binding of iron to a protein promotes protein function, the substitution of gallium for iron in a protein usually disrupts its function and may lead to adverse downstream effects in cells (Choi et al., 2017).

Our laboratory is evaluating the fitness of iron resistant bacteria in gallium. We utilized experimental evolution to create 12 iron (II) and 12 iron (III) selected replicates in Escherichia coli.  These cells had been selected in excess iron for 28 days. Within each selection regime, 5 had no history of silver resistance and 7 were derived from Agresistant replicates (Tajkarimi et al. 2017).  The control cells had no history of either iron or silver resistance. Fitness of these replicates was evaluated in the presence of increasing concentrations of Gallium (III) nitrate.  Our results indicated that bacteria selected in Iron (II) as well as Iron (III) showed a significantly superior 24 hour growth in Gallium nitrate compared to the controls (Fig 1 and 2).

Fig 1: Twenty-four-hour growth curve of Iron (II)-Selected E. coli in the presence of increasing concentrations of gallium (III) nitrate. Fe2+=Iron(II)-selected bacteria. Fe2+Ag= Iron (II)-selected bacteria with silver background.

Fig 2: Twenty-four-hour growth curve of Iron (III)-Selected E. coli in the presence of increasing concentrations of gallium (III) nitrate. Fe3+=Iron(III)-selected bacteria. Fe3+Ag= Iron (III)-selected bacteria with silver background.

For example, for the iron (III) resistant populations there was virtually no reduction in growth from 60—1000 mg/L of gallium, while for the controls growth was completely eliminated at 1000 mg/L. Interestingly, gallium was also very toxic to the ancestors of iron-selected cells with silver background (i.e. silver selected bacteria) (data not shown). Whole genome sequencing of our iron-selected bacterial cells demonstrated that mutations occurred in genes that confer anti-transition metal stress resistance. Examples of these genes include fecA (ferric citrate outer membrane transporer), rho (transcription termination factor), fur (ferric iron uptake regulon transcriptional repressor), murC (UDP-N-acetylmuramate: L-alanine ligase), dnaK (chaperone HSP70), tolC (transport channel), and nusA (transcription termination/antitermination factor).

In addition to whole genome sequencing, we utilized Nanostring technology to examine gene expression profiles of 50 genes we determined to be involved in iron metabolism or general metal resistance. We found striking patterns of expression difference in the presence of excess iron for genes: regulated by fur (ferric iron uptake regulon transcriptional repressor), involved in cell wall synthesis, general metabolism, transcription, transport, and transcription regulation.  Generally, genes in these categories in the iron resistant bacteria were significantly up-regulated, while these same genes were significantly down-regulated in the controls.

Thus, we hypothesize that the genomic profile and altered gene expression patterns of our excess iron resistant E. coli has also changed the way they interact with the iron analog gallium. Either these mutations reduce the rate that gallium enters the cell, increases the rate in which it is effluxed from the cell, or alter the targets of gallium toxicity once inside the cell.  These results suggest that gallium cannot be used as a Trojan horse to fool iron-selected bacteria, as there survivorship in the presence of increasing gallium suggests the capacity to rapidly evolve resistance to it. We will test this idea in subsequent experiments.

References

Butler, M. S., Blaskovich, M. A., & Cooper, M. A. (2017). Antibiotics in the clinical pipeline at the end of 2015. The Journal of antibiotics, 70(1), 3.

Choi, S. R., Britigan, B. E., Moran, D. M., & Narayanasamy, P. (2017). Gallium nanoparticles facilitate phagosome maturation and inhibit growth of virulent Mycobacterium tuberculosis in macrophages. PloS one, 12(5), e0177987.

Weinberg, E. D. (2009). Iron availability and infection. Biochimica et Biophysica Acta (BBA)- General Subjects, 1790(7), 600-605.

Tajkarimi, M., Rhinehardt, K., Thomas, M., Ewunkem, J. A., Campbell, A., Boyd, S., … & Graves, J. L. (2017). Selection for ionic-confers silver nanoparticle resistance in Escherichia coli. JSM Nanotechnol. Nanomed, 5, 1047.

Posted in BEACON Researchers at Work | Tagged , , , , | Comments Off on Gallium cannot be used as a Trojan horse to fool Iron-selected bacteria