EVATAR- Building the female reproductive tract in a dish

EVATAR – a box of reproductive tissues

Researchers have made a hand-sized device with all the tissues of the female reproductive system (1). They call it EVATAR (Eve + Avatar – a digital human).

EVATAR can be used to test if a new medicine is likely to cause hormonal problems or affect fertility in women.

Up until now, such tests could only be performed on human beings.

The device itself is like a box with many compartments. Each compartment contains tissues from the female reproductive organs. These include the ovaries, fallopian tube, uterus, and cervix.

The compartments are connected – allowing them to share nutrients and talk to each other using chemical signals (1).

Special electromagnetic valves control the flow of nutrients between compartments and a computer program controls the amount of nutrient in each compartment (1,2).

Photo credit : Northwestern University, Woodruff Lab

Why do we need this device?

Animals are not the best models for human disease or for studying how people will respond to a new medicine (3,4).

So what do we do when a study/ trial cannot ethically be performed in a human?

Traditionally, a scientist would take out some bits of a tissue from the human body and keep it alive in the lab (5).

They would then use these for further tests and studies.

Many human tissues, however, die or loose their function outside the body.

In the EVATAR device, researchers have succeeded in keeping all the tissues from the female reproductive system alive and functional for the 28-day menstrual cycle (1,6).


Testing for the behavior of tissues

During the menstrual cycle, the ovary in a woman’s body produces a mature egg. This mature egg is either fertilized or removed from the body(6)

The reproductive system, brain, and pituitary gland work together to make this happen(6).

The pituitary gland secretes gonadotrophins (namely Follicle Stimulating Hormone and Luteinizing Hormone).

The ovaries in the device responded to an external supply of these hormones by making their own chemical signals/ hormones – oestradiol, progesterone, Inhibin A and B.

The ovaries also produced a mature egg within the device.

Other tissues of the reproductive tract responded to the ovarian hormones.

The lining of the uterus – the endometrium, made more receptors to the hormones progesterone and estrogen at end of the 28-day cycle. The cells of the endometrium also multiplied, as expected during the menstrual cycle.

The ectocervix is the bit of the cervix that is externally exposed. It maintained it’s skin-like appearance and structure through the cycle.

This tissue also made receptors for the hormone progesterone only on day 0 and not on day 14 of the cycle, possibly controlled by the secretions of the ovary.

In the fallopian tube, the hair-like projections or cilia, that nudge the egg from the ovary into the uterus, were beating and functional even after 21 days in the device (7).

Researchers can now use this device to study infections and hormonal problems, as well as the reproductive system itself.

Inclusion of liver tissue in the EVATAR device

Tissue from the liver was included in the EVATAR model in one of the compartments.

The liver is not directly a part of the female reproductive system or cycle. However,  it breaks down a new medicine in the body( 8).

After 28 days, the liver tissue looked normal and made a healthy amount of the protein, albumin, within the EVATAR device.

Researchers can now test a new medicine on the reproductive system and the liver simultaneously.


Want to know more about EVATAR and things mentioned in this article? Here are some links:

1. The study describing the construction of the EVATAR device – “A microfluidic culture model of the human reproductive tract and 28-day menstrual cycle” (https://www.nature.com/articles/ncomms14584)

2. Draper Labs on the technology behind EVATAR – http://www.draper.com/news/fighting-cancer-boosting-fertility-promise-first-female-reproductive-system-chip

3. “Man or mouse? Why drug research has taken the wrong turning”, October 2016, NewScientist. (https://www.newscientist.com/article/mg23230973-700-man-or-mouse/)

4. Why do we need animal models? https://speakingofresearch.com/facts/the-animal-model/

5. Tissue culture – http://www.biotechnologynotes.com/animals/animal-cell-culture-history-types-and-applications/671

6. Dr. Woodruff explains the menstrual cycle https://www.coursera.org/learn/reproductive-health (Lecture 2.2)

7. Video of the cilia of the fallopian tube beating – https://www.nature.com/article-assets/npg/ncomms/2017/170328/ncomms14584/extref/ncomms14584-s2.mov

8. How does the liver work? (https://www.ncbi.nlm.nih.gov/pubmedhealth/PMH0072577/)

Thank you, Kelly McKinnon, Hunter Rogers (Dr. Woodruff’s lab) and Bernadette Sztojka for your feedback.


Can genetic variations define Ayurvedic Prakritis?

-A genome- wide study finds allelic differences between individuals that correlate with Ayurvedic body-types (Prakritis)

Snap-Shot of the study

Do the ayurvedic body types have genetic underpinnings? In a first step to answer this question, the authors evaluated differences between individuals whose body-type had been assigned by both Ayurvedic practitioners and a software. They found 52 variations across the genomes of 262 individuals which allowed them to be classified into ayurvedic Prakritis – Pitta, Kapha and Vata.

 Introduction to Ayurvedic prakritis

In Ayurveda, according to the ancient text Charaka Samhita –the body and mind must be brought together to lead a harmonious existence. People can be classified into Prakritis or types on the basis of relative contribution of the three constituents Pitta, Kapha and Vata (roughly translating to – arising from movement, digestion and accumulation – of toxic metabolites for instance) to their body. According to Ayurvedic philosophy- an understanding of this body type and the ability to maintain a diet and lifestyle suited to that body type translate to balance and health. Prakriti or Ayurvedic body-types which define a person’s intrinsic physical abilities, mental states and also have implications for their susceptibility to disease and response to drugs (1,2).

Background for this study

In a recent study (3), researchers have identified genetic variations associated with the traditional classification of people into Ayurvedic Prakritis – specifically if small differences (SNPs, Single nucleotide polymorphisms) throughout the human genome correlate with the Prakriti classification. All humans are genetically very similar to each other, differences between us (populations, races, ethnic groups etc.) are captured in variations of nucleotides which make up the DNA – these are called single nucleotide polymorphisms. Many studies have shown particular SNP or group of SNPs to be correlated with risk of disease, whether a person will develop resistance to therapy, etc. forming the basis for personalized medicine  (4,6).

 What did they do?

3416 individuals were classified for their prakritis by Ayurvedic practitioners as well as a software. Of these, DNA isolated from blood samples of 262 individuals (male, healthy, between 20-30 years), who were reliably classified as having a clear dominance of one of the three constituents (Vata, Pitta or Kapha) representing the “extreme” Prakritis were used for the genome-wide study. A microarray consisting of 1 million positions / SNPs was used to identify the genotype of these individuals.

 What did they find?

This study found 52 out of 1 million SNPs is sufficient to assign the Prakriti of individuals, irrespective of their ethnic background. One of the challenges in the study was that there was no control group – therefore each prakriti was compared to the other two. Subsequently, these 52 SNPs were able to cluster individuals into distinct groups by Principal Component Analysis. Interestingly, one of the SNPs in a PGM1 gene (Phosphoglucomutase 1), which codes for an important enzyme in sugar (glucose) metabolism, is significantly associated with Pitta dominant group that is known for efficient metabolism.

 Things to keep in mind about the study:

It is to be noted that the three categories compared and defined here represent extremes –and according to Ayurvedic principles most people are a composite of all three- Kapha, Vata and Pitta, with the dominant element defining their type. There are some previous studies, which have suggested that the Ayurvedic Prakriti classification may have a genetic or metabolic basis (2,5). They were conducted on fewer subjects and looked at fewer genes/ phenotypes compared to this study, which rigorously recruited a large number of subjects, used multiple methods of classifications (software and Ayurveda practitioners) and a genome wide approach. There is much work to be done in understanding the scientific basis of the Ayurvedic classification system and whether we can independently and reliably assign people (independent of race, gender and ancestry) to a type.

 What does this mean?

The 52 SNPs defined in this study can now be used independently in other populations and also provide a way of identifying new associations with metabolism and other phenotypes. For more than fifteen years now, we have been able to read our genome i.e – information in our genes but not been able to fully understand how it defines us as individuals. So on the one hand, this study by correlating phenotypes with genetic variations, helps us understand a little bit more about how genes make us who we are. On the other hand, by using modern genetic tools in the context of traditional knowledge, this study provides a rigorous way of assessing the framework of ayurvedic medicine.


1. Understanding personality from Ayurvedic perspective for psychological assessment: A caseS Shilpa and C. G. Venkatesha Murthy. Ayu. 2011 Jan-Mar; 32(1): 12–19.

2. Classification of human population based on HLA gene polymorphism and the conceptof Prakriti in Ayurveda. Bhushan P 1 , Kalpana J, Arvind C. J Altern Complement Med. 2005 Apr;11(2):349-53.

3. Genome-wide analysis correlates Ayurveda Prakriti. Govindaraj P, Nizamuddin S, Sharath A, Jyothi V, Rotti H, Raval R, Nayak J, Bhat BK, Prasanna BV, Shintre P, Sule M, Joshi KS, Dedge AP, Bharadwaj R, Gangadharan GG, Nair S, Gopinath PM, Patwardhan B, Kondaiah P, Satyamoorthy K, Valiathan MV, Thangaraj K. Sci Rep. 2015 Oct 29;5:15786. d

4. A database of humans SNPs and their recorded associations can be found here: http://www.snpedia.com/index.

5. Whole genome expression and biochemical correlates of extreme constitutional types defined in Ayurveda . Prasher B, Negi S, Aggarwal S, Mandal AK, Sethi TP, Deshmukh SR, Purohit SG, Sengupta S, Khanna S, Mohammad F, Garg G, Brahmachari SK; Indian Genome Variation Consortium, Mukerji M. J Transl Med. 2008 Sep 9;6:48.

6. http://www.scientificamerican.com/article/a-very-personal-problem/

Interview with Dr. K. Thangaraj

Q. What was the overlap between the prediction by the software, by different Ayurvedic practitioners? Have you tried to estimate if this classification is robust enough to suggest underlying genetic differences?

The first assessment was by Ayurvedic physicians and classified using their knowledge. To double check what they have done, we have used a computer software. The software is also based on various parameters specified by the Ayurveda physicians.There were many questions which the individual had to answer to be assigned a Prakriti. Across the three centres – Bangalore, Pune and Udupi, on average, 75% of the individuals were in concordance.

Q. Why not perform an enzyme profile or measure transcriptional differences for metabolic enzymes? Is there an advantage in taking an SNP approach?

The advantage is that the SNP does not changes, it is there from birth till the individual dies. Whereas transcription profiles may change at different times, depending on time of day, tissue type to tissue type etc. We have looked into that also. This is the very first step. We can extend this across countries, across ethnic groups and cluster them.

Q. Ayurvedic medicine believes in holistic changes including those of lifestyle, dietary along with medicines and does not rely overly on mechanistic explanations beyond the classification into types. There is less emphasis, if you will on dissection of cause-effect and more on restoring the overall balance. Do you agree with this? If yes, then what are the challenges with respect to the study design when you tried to apply the modern framework of science- which relies on a reductionist approach, finding a cause and targeting only that specifically, to the ayurvedic system?

I agree that there is a holistic approach, but the basic approach is to classify the individual. Based on the prakriti, they will make the changes to the diet, prescribe treatment etc., so this is a very important stage. The challenge is the following- I am a geneticist looking at diseases,looking at case-control studies is very easy – for every marker we can ask if the mutation is present or if the prevalence is higher in the patients versus the control. In this case all the individuals are normal and within the same age-group. The only difference is the prakiti. So we wanted to see how to differentiate these individuals- as there are three groups, not two. Then we decided to compare one prakriti against the other two types, and try to see if there is genetic variation between the groups. Then there was a lot of statistical analysis. We used 1 million markers, this has many advantages, the disadvantage is the robustness and having to come up with statistical analysis ourselves. (MT: So, by using 1 million regions, you may increase the chances of false associations, is that the worry?) Not, really false associations. For example we may not have information about a particular SNP in an individual. We need to use markers that are consistent between all the individuals. So, when data is not available, we need ways of retrieving the data. In that process we need to use genetic panel of markers which are Indian specific. We developed our own panel of markers – Dravidian, Indo-European and used as a reference and to extract what is the possible marker in a given position.

Q. Have you tried to validate your classification using the 52 SNP panel with an independent population? For instance, if you knew what a person’s SNP state is, how reliably can you assign their Prakriti? Would it be useful to perform a blind study in which both the SNP panel ayurvedic practitioners and the software performed the classification, with the aim of determining how often they match?

That is very interesting. What we did was, we has more than 300 samples analyzed for these 1 million genetic markers, from our initial studies on population genetics. We used some of those samples, as these are all populations specific – a very endogamous population. We tried to project some of those individuals into these three clusters, we did find that although the individuals have come from the same ethnic background (more homogeneous), they were falling in 2 or 3 different prakritis. The same ethinic background can be placed into different prakriti. This we tried to do with our own data, this is not as detailed as you suggest. Independent blind studies need to be done.

Q. You have excluded women from your study and restricted yourself to the Indian population, does this limit the applicability of your results?

Yes, of course. At this point we selected only males because we did not want any confounding effects, in the females there are many hormonal changes and so on.

Q. You have started a way of examining traditional medicine in the framework of modern science. What are the challenges and the future of this approach?

(The future of this approach) This has paved the way to do many more things. For example, the discovery of PGM1 has given the clue that you can take the phenotype of the particular prakriti and correlate it with the gene, this gene is involved in metabolism and individuals with the pitta prakirti have high metabolism. We can now use the characteristic feature of the prakriti and look into those genes in a detailed way. These are some of the futuristic aspects, one can take the study further with. We did try to look at the network of all the metabolic pathways genes, the problem is this will need transcription or metabolic profiles from tissues of these normal individuals.


That which could kill you can also cure you

An opensource database for therapeutic uses of venom


About this database:

Popular myth and folktales often involve a hero setting out in search of miraculous cures from deep within the unexplored and dangerous forest, slaying wild beasts and sampling rare herbs to save a doomed civilization. The use of toxic substances including toad and snake venom have been described in traditional medicine [1,5].

Venoms are substances produced by one animal to defend itself from predators and/or prey on other animals [2]. They are usually a complex mixture of organic compounds, the most potent of which have been found to be peptides, for example, the conotoxins from cone snails – are peptide neurotoxins made of small chains of 10-30 amino acids [3].

Even related species seem to have evolved their own cocktail or venom composition independently, giving rise to a huge diversity in venom types – over 10 million types [4,7]. Components of venoms have incredible specificity and speed of action, for example, a snake neurotoxic venom is so specific in its action that it only affects particular ion channels in neurons [5,6], a lizard neurotoxin must act fast enough to immobilize the prey so that it cannot disappear into a crevice. Both the specificity and speed have tremendous implications for therapeutics as delivering therapeutics to the right place in a timely manner is still quite a challenge. However, the diversity of venoms and their intrinsic toxicity pose a challenge in the systematic testing of these compounds for therapeutic use, further, this area of biology is not well studied and consequently there are huge gaps in our understanding of the mechanisms by which these compounds act. The creation of a new online database which allows easy retrieval of studies that have analyzed venoms from the viewpoint of their therapeutic use is a step in the right direction (https://venomkb.tatonettilab.org).


What did they do?

A striking feature of modern biology is the speed at which scientists gather information is surpassed by the ability of even specialists to assimilate them. In this project, the authors have made a database of venoms and their therapeutic uses called the Venom Knowledge Base or VenomKB. The database was constructed by scanning over 22 million studies whose titles and abstracts (research summaries) are cataloged in MEDLINE in a searchable format. Using the medical subheading “Venoms/therapeutic use” they retrieved 5117 pertinent articles. 275 of these articles were then sifted manually to form the first table of the database with ID, Venoms, Effects, PubMedID (link to the original research), and whether or not the entry is flagged for review.


271 |cobra venom cytotoxin| anticancer| 22888519|Not Flagged

The 2nd and 3rd tables of the database use automated searches to extract relevant information. The logic of both methods is described in the paper and the code is publicly available in a GitHub repository. The second method also results in the classification of venom and effect, where the effect is often a disease state.


26493| cobra neurotoxin proteins| paralysis| 7603413| Not Flagged

The third method is interesting because it takes a semantic approach and classifies data into three categories – subject, predicate and object

11567| cobra venom| causes| analgesia| 16539838| Not Flagged

The database right now grapples with control of false positives, reporting things as venom and effect when in fact it is not even a venom. However, both automated methods at least in the first pass appear to be quite sensitive in picking up venom-effect pairs. Both the automated search based tables have been manually scrutinized for obvious misclassification, however, the user will still have to follow the thread back to the original research paper to really get an idea of the nature of the link of the venom and the therapy and whether or not it is relevant. In spite of this, the ability to access data in this manner will promote more data-based hypothesis generation.This is relevant not only for venoms and their therapeutic effects but to all of biology which is rapidly becoming a data-driven exercise.

An interview with Joe Romano & Nick Tatonetti

Q. When making a database like the Venom Knowledge Base, what are the major challenges – in the design, representation, maintenance, storage, retrieval etc.? What are the resources required to maintain it and how is such work funded?

In our experience, the most challenging portion of designing VenomKB was defining semantics – essentially, the words we chose and what they mean intuitively to readers. It sounds like something that should be relatively straightforward, but since venom research is a pretty new field, it can be tough to achieve. For example, depending on the portion of the knowledge base you are looking at, some records can be “venom [x] treats condition/disease [y]”, or it could be “concept [x] does [some kind of action] to concept [y]” (where either [x] or [y] can be a venom compound). It sounds kind of nuanced, but in designing something that is meant to be read by the public, it creates a pretty important distinction. As for resources and funding, one of the great aspects of this study is that anyone with some web design experience can create a site like VenomKB for very minimal cost using resources that are totally available to the public. To give a rough figure, it costs us less than $50 per month to keep the site running and maintained, which is pretty well within the resources of many research groups. And overall, the research we do in our lab is funded by different branches of the National Institutes of Health (like much of the university-based biomedical research in the United States). Having a non-commercial funding source has its advantages, such as the fact that we can release everything related to VenomKB free of charge to anyone worldwide with an internet connection.

Q. At this moment, if I understand correctly -you have used search terms and string matches? Isn’t this a limited approach given the lack of standard ways of naming things in the field?

Actually, all of the methods we used to extract useful information from the literature were a bit more subtle than basic string search terms. In each case, we used medical and biological terminologies – essentially structured dictionaries that help with standardization in biomedical data – to find concepts and then transform those concepts into the data records you can look up in VenomKB. When we store the information in the database, we use the readable strings so people can understand what the data refer to, but the tools and algorithms that sit behind the scenes attach a semantic meaning to the text that the computer can understand and use to identify synonyms and duplicate meanings. Granted, these approaches aren’t perfect, for exactly the reason you mention (there isn’t great consistency presently in naming venoms and their components). Since there is no terminology that natively understands venoms, we use other ones that we feel come closest to fit our needs. Still, we acknowledge that there are many ways to improve our algorithms, and one of the projects we are currently working on is a resource that will standardize the names of venoms – hopefully solving this issue altogether!

Q. Traditional sources of knowledge (specific to communities, tribal healers, not part of allopathy for eg. ayurveda) are not on Pubmed in the form of papers, would it be possible to extend your database to Google books or libraries to mine the information there?

Yes, we have in fact considered this. There are really two challenges that prevented us from using sources like these at this stage: (1) In general, it’s harder to find data sets that are fully open access like PubMed is – for example, Google Books has faced legal issues due to claims of copyright infringement, and we have to be really cautious about things that could cause issues of legality. (2) Since PubMed almost always have abstracts available for view and download, much of the important information is condensed into a small amount of space that can be downloaded and searched through in its entirety on a modern computer. If we were to apply the same method to large collections of books, we might pick up a lot of irrelevant data (not to mention need far more computational power to perform the searches with). In short, yes, we do hope to incorporate data sets such as these for the very reason you mention, but we still need to define the best approach for doing so.

Q. You call this the Venom Knowledge Base, when do you think knowledge emerges from facts? For instance how do you account for studies that are badly designed or cannot be reproduced? There is general trend in science to create new information without knowing how to assimilate it, do you agree with this? What is the solution?

We totally agree with this statement. Bad science is an issue that remains regardless of how refined the peer-review process is. One of the features that VenomKB offers is the ability to connect purported knowledge (e.g., that a venom treats a certain disease, as claimed by a single study) to a number of biological databases that researchers can use to try and validate the claims made in the studies. This also touches on yet another study we’re in the process of designing right now – using information in VenomKB, we are planning to select 100 venoms and use modern genomics techniques to try and both validate information existing in VenomKB and to make new discoveries that may not have been picked up in existing literature. Either way, since VenomKB is a growing resource, we hope to add all of the validation and discovery study results as new modules to the knowledge base, separate but inherently linked to the existing database tables. Another functionality that helps transform raw data / information to actual knowledge is that VenomKB encourages community involvement. If you see something that looks suspicious, or maybe incorrectly included, there’s a button that lets us know that something probably shouldn’t be there, and we will take a look and make the decision whether to delete it or not. This helps bolster confidence in the data, and lets anyone at all get involved, even if they are not a scientist.

Q. If there is a student (high school, undergraduates) who wants to take on a science project -What kind of questions could they ask with this data?How accessible is the information to them? How can they contribute to this research?

One of the major draws of data science is that studies like this really have limitless potential for reuse. A few areas that I think would be particularly good for young biologists and computational biologists to explore include phylogenetic analysis of venomous animals, prediction of gene sequences that give rise to venom proteins, comparative analysis of venom proteins with drugs that are already approved by the FDA, and comprehensive literature reviews that aggregate known uses for a specific species’ venom. These are all examples of studies that we need in order to validate the reliability of VenomKB, and are straightforward to carry out using existing tools (usually open-source). With good advising, any of these projects could be within grasp of students who are relatively new to computational biology.

Q. What kind of access policies do you have for potential commercial use of the information? How can the industry contribute to sustaining this effort? What drove you to keep this in the public space, given the general trend of companies to make databases like this, at exorbitant prices, for private use?

VenomKB is fully open-access and open source, and we completely intend for it to remain that way. While we haven’t formally discussed what restrictions we may or may not place on reuse by corporate entities, we encourage any researchers – whether publicly or privately funded – to use the site to guide research that will benefit human health. Many pharmaceutical companies are devoting a lot of resources to studying ‘biologic’ drugs, a category that venom compounds definitely fall into. We like to hear from people who would like to make use of the data in their own research, so if any readers think they may be interested in doing so, drop us a line! To get to the final part of this question, open access is a rapidly growing trend in academic research. The journal we published in – Scientific Data – is entirely open access, as are a large number of others. The thought is that if research is meant to help humanity and benefit the largest number of people, we need to eliminate some of the barriers posed by cost and corporate ownership. Not all research can be done for free – the drug design process in particular requires major financial investments. However, things that can be done in settings like an academic research lab should try to make their data available and free if possible, for the purpose of lowering those same barriers.


1. “The Medicinal Use of Snakes in China”, Institute for Traditional Medicine, Portland, Oregon, 1997

2.  The toxicogenomic multiverse: convergent recruitment of proteins into animal venoms. Fry BG, Roelants K, et al., Annu Rev Genomics Hum Genet. 2009

3. Novel peptides of therapeutic promise from Indian Conidae. Gowd KH et al.,  Ann N Y Acad Sci. 2005

4. Complex cocktails: the evolutionary novelty of venoms. Casewell et al.,  Trends Ecol Evol. 2013

5. The Bite That Heals -Scientists are unlocking the medical potential of venom. By Jennifer S. Holland for National Geographic.

6. Alpha neurotoxins. Carmel M Barber et al., Toxicon. 2013

7. VenomKB, a new knowledge base for facilitating the validation of putative venom therapies. Romano JD and Tatonetti NP. Sci Data. 2015

Beauty is but skin deep?

-Mircroorganisms in our skin respond to vitamin B12



Many of the microorganisms that inhabit our skin surface i.e. the skin-microflora, are ‘Opportunistic pathogens’, usually harmless, perhaps beneficial, but capable of causing harm. We know very little about the conditions that  turn these microorganisms against us. The bacterium, Propionibacterium acnes for example, is present ubiquitously on human skin, but causes acne in a subset of individuals (1). This provides an interesting opportunity to study conditions that lead to acne formation (1).

Acne is a severe inflammation of the skin brought about in many ways and is associated with particular diets (dairy consumption for instance) (2-5). In a recent study, authors analyzed the skin-microflora of people presenting with and without acne (6). They found that a distinct set of genes were active in the microflora of people with acne when compared to that of people without acne.


What did they find?

What is the reason behind these differences? Are the microflora sensing something different in their environment (read, the skin of their human host)? To address this question, the authors focussed on P. acnes, notorious for getting deep inside the skin and causing inflammation (resulting in acne) (2).

The authors of this study then investigated the nature of the genes expressed by this bacteria uniquely in people with acne. They found diminished activity of genes that help in the synthesis of vitamin B12 and fatty acids. The production of Vitamin B12 is a multi-step process, requiring multiple different enzymes.The authors found that many of these enzymes were produced at much lower levels by the bacteria in individuals with acne. Clinicians have noted for a long time that some individuals develop acne (technically known as aceniform eruptions) as a side-effect of B12 supplementation (7). In this study B12 supplementation was given to a group of 10 people, 1 of whom developed acne within one week. Further investigations suggested that P. acnes responds to supplementation of B12 to the host by shutting down its own pathways for B12 synthesis. Such an auto-regulation for B12 production has been seen in other bacteria as well.

So, what is happening in patients who developed acne? Interestingly, in the patient who developed acne upon B12 supplementation, the authors were able to show that one of the intermediates of B12 synthesis was now being shunted into porphyrin synthesis. In lab grown cultures of P. acne, supplementation of B12 into the growth medium decreased its B12 synthesis and increased porphyrin synthesis. Porphyrins are known inflammatory agents and may underlie the development of acne. In fact, in other clinical studies patients who responded positively to acne treatment also showed decreased porphyrin levels (8).



Take-homes from this study

Not only does this study suggest an explanation for the why some people develop acne when they undergo B12 supplementation, it reveals an intricate coupling between humans and the microorganisms that inhabit them. This is the first study to our knowledge that shows metabolic effects in bacteria upon changes in the human nutrient condition. There are many open questions that emerge from this study, how do microorganisms sense the nutrient state of the host? Is metabolic shift a common theme in microorganisms that are usually friendly and then turn deadly?


What is cool about this study?

In a crowded microbial environment, like the human skin, picking up the changes in the activity of genes of one specific bacteria is non-trivial. It is also hard to get enough information to reconstruct pathways and links between these gene activity states to perform robust analysis across people (taking into account, the variations between individuals in the same group).


Pinch of Salt:

Don’t throw away your B12 supplements just yet!

It is important to understand, that although this study provides an explanation for why some people develop acne upon Vitamin B12 supplementation (7), nothing in this study suggests that people should stop taking Vitamin B12 supplements. This is particularly relevant as Vitamin B12 remains a problematic deficiency especially for vegetarians and vegans, with consequences for both physical and mental health (9-11).


1. About Propionibacterium acnes and its relationship to acne

2. “Epidemiology of acne vulgaris.” K.Bhate and H.C. Williams, Br J Dermatol. 2013

3. “The role of diet in acne: facts and controversies.” Batya B. Davidovici and Ronni Wolf, Clin Dermatol. 2010

4. Does diet really affect acne?” H. R. Ferdowsian and S. Levin, Skin Therapy Lett. 2010

5. “Evidence for acne-promoting effects of milk and other insulinotropic dairy products.” Melnik B.C., Nestle Nutr Workshop Ser Pediatr Program., 2011

6 .“Vitamin B12 modulates the transcriptome of the skin microbiota in acne pathogenesis.” Dezhi Kang et al., Sci Transl Med. 2015

7. “Vitamin B12-induced acneiform eruption.” Ilknur Balta and Pinar Ozuguz, Cutan Ocul Toxicol. 2014

8. “In vivo porphyrin production by P. acnes in untreated acne patients and its modulation by acne treatment.” Claudia Borelli et al., Acta Derm Venereol. 2006

9. “The neurology of folic acid deficiency.” E.H. Reynolds, Handb Clin Neurol. 2014

10.  “Vitamin B12 deficiency.” Alesia Hunt et al., BMJ. 2014

11. About Vitamin B12

Teixobactin: Can this new antibiotic help us sail through the doldrums of drug resistance?

iChip based discovery of a potent novel antibiotic


Why do we need a new antibiotic?

What can happen if you self-medicate on an antibiotic or do not finish a course of antibiotics that has been prescribed for you? When cattle are fed indiscriminately on antibiotics to keep them healthy? When sewage from hospitals is not completely treated and released into the community? The microbes that survive in these environments stop responding to antibiotics around them (1). Our world at present faces a daunting task of treating people infected with resistant forms of many bacteria. Often clinicians have to resort to potent broad spectrum antibiotics to treat infections that could be treated with the first line of drugs a few decades ago. The other aspect of the problem of antimicrobial resistance is the lack of new treatment options. Many of the current antibiotics are chemical modifications of ones that are known to work. Designing completely novel molecules, with antibiotic activity, synthetically, has not been very successful (2). In this rather bleak situation a new study brings a new ray of hope. In this study the authors have enriched hitherto uncultivated bacteria from the soil (3).


What is so special about this?

A very small percentage of bacteria in the soil can actually be grown in the laboratory (4). The development of tools/methods to grow more bacteria opens up a window for isolating new compounds with potential antimicrobial activities that these bacteria may be producing to their advantage in the complex niche of the soil. In this study, the authors diluted soil sample to contain single cells and grew them in special chambers embedded in the soil which allow for nutrient exchange with the soil. Earlier studies have shown this method to recover 50% of the bacteria from the soil. Previous studies have also demonstrated that once isolated, many of these bacteria can be grown in the lab.


How was this antibiotic found?

After screening over 10000 bacterial isolates in this way Ling et al., identified a new species Eleftheria terrae that produced a potent compound against Staphylococcus aureus (S. aureus can cause infections especially in hospitals, it also notorious for acquiring resistance to commonly used antibiotics). They examined this compound in greater detail, asking questions like- what does it look like (chemical structure)? How is it made in the bacteria (biosynthetic pathway)? To answer these questions, they undertook detailed chemical analysis by NMR and Marfeys’ structure analysis. The active compound which the authors named Teixobactin, was found to be a uricylated oligopeptide (the details of the structure are provided in the paper). They analyzed and were able to predict the pathway by which the E. terrae makes this antibiotic. The compound was found to be completely novel.


How does it work?

How does this antibiotic kill the target bacteria? The first clue was that it was more effective against gram positive that gram negative bacteria. These two classes  (distinguished on the basis of their appearance after staining them with dyes) of bacteria differ in the number and nature of protective coverings around them. Gram positive bacteria have a thick cell wall around them, whereas gram negative bacteria have a thin cell wall but also have an additional outer membrane. The cell wall is made up of repeating units of modified sugars known as peptidoglycan and is essential for structural integrity of bacteria. A breach in this structure would lead to the bacterium’s death. Interestingly, Teixobactin is not active against gram negative bacteria which have an outer membrane. However, a strain of E coli (a gram negative bacteria) with defects in the outer membrane is susceptible to this antibiotic. These data suggest that Teixobactin needs to have access to the bacterial cell wall for its activity. Consistent with this idea, they find that Teixobactin binds specifically to precursors of peptidoglycan and does not allow their incorporation into the cell wall. Instead of targeting the enzymes that carry out cell wall synthesis, Teixobactin, like the potent antibiotic, Vancomycin, interacts with structural components of the cell wall itself. It seems to target multiple precursors and bacteria die not only from lack of cell wall but also from accumulation of toxic intermediates of cell wall synthesis.


Does it work against pathogens?

They then asked, how effective is this antibiotic against common pathogens? Teixobactin was found to have potent activity against Staphylococcus aureus (which can cause disease under certain circumstances), Clostridium difficile (causes colitis) and Bacillus anthracis (anthrax). It also had good activity against hard to treat microorganisms like Mycobacterium tuberculosis (Tuberculosis) and enteroccocci (are intrinsically antibiotic resistant, causally associated with urinary tract infection among others). All of this is good news, however, how do we know that once in use, bacteria will not become resistant to Teixobactin? One way to answer this question is to subject bacteria to low levels of the antibiotic for prolonged period and then test if they still respond to it. Fortunately, no resistance to Teixobactin  emerged in either M tuberculosis (M Tuberculosis is notorious for acquiring resistance to multiple drugs) (6) or S aureus. Suggesting that resistance will probably be slow to evolve  against this antibiotic. The next step then was to assess if it was toxic to animal cells and/ or effective when used in animals.

The compound was found to be eminently suited for being used as a drug in animals. It was not toxic to mammalian cells, was active even in the presence of serum and was stable in  blood. Moreover, Teixobactin seems to have no carcinogenic properties. In mouse models of septicemia and pneumonia, mice treated with Teixobactin survived and responded well to  treatment. This makes Teixobactin a remarkable candidate for further studies with the  possibility of clinical trials in humans.


What does this finding mean?

The approach used to isolate and characterize Teixobactin is novel and paves way for the identification and characterization of many such compounds. We think that this may well be the beginning of a mining exercise where we explore more antimicrobials from the soil. It remains to be seen if Teixobactin can actually be used in humans, it is unclear how long that will take or last. At the very least, Teixobactin offers a tempting glimpse of what’s hidden in the soil and gives us a better appreciation for the microbial community we nonchalantly read upon.

Why are we more likely to get a cold in cold weather?


We don’t really know, what we do know is that some cold causing viruses grow better at lower temperatures. Rhinoviruses, one of the most common causes of the common cold, display a temperature dependent growth pattern. They were shown to grow better at cooler temperatures (33°C­ – 35°C)​ (1,2) like that in the upper respiratory tract than at the core body temperature of 37°C. Scientists hit a wall when they looked for a reason for this by analyzing the viruses themselves. Entry of the virus inside the cell, for instance, was not affected at cooler temperatures (33°C- 35°C). The failure to find a convincing mechanism such as a temperature dependent viral enzyme or gene product was puzzling until recently, when scientists turned the table and started looking at the virus infected cell rather than the virus itself.

More about the study

A study published in the Proceedings of the National Academy of Science (PNAS) looks at cellular defense mechanisms and anti­viral responses that come into play when Rhinoviruses infect cells (3). ​

​The authors of this study compared the response to infection at the warmer body temperatures (37°C) and at cooler temperatures as found in the nasal cavity (33°C).

In what we think is the first step to understanding the role of the host (human) in this process, this study used a mouse adapted strain of the virus. They generated this strain by growing the virus for many generations in mouse cells. Subsequently the virus acquired mutations, adapted and was able to infect mouse airway cells. Foxman et al., isolated mouse airway epithelial cells and used the adapted strain to infect these cells in the laboratory. In order to test the temperature sensitivity, they carried out infection experiments at 33°C (cooler) and 37°C (warmer) temperatures. They observed the expected decrease in number of viral particles (titers*) starting from 7 hours after infection at 37°C but not at 33°C, confirming that temperature does impact viral numbers. The changes observed were in fact in the infected cells which showed a lower antiviral response to the virus at cooler temperature.

When a cell gets infected by a virus, it puts out a signal saying “I am infected” by secreting molecules such as interferons and this is critical for mounting an antiviral response​ (4)​. The study by Foxman et al., shows that cells infected at cooler temperatures have lower expression of molecules critical to the anti­viral response. The authors artificially activated a defense pathway (The RLR pathway) which results in interferon production, and show that this pathway has a lower response at 33°C than at 37°C. In other words there is lower production of interferons, at cooler temperatures. Further, by genetically mutating either molecules of this pathway or a receptor of interferon in mouse airway cells, the authors find an increase in viral titers even at warmer temperature (37°C).

These data suggest the cells may be able to ward off a Rhinovirus infection at warmer temperatures (37°C) due to a robust anti­viral response resulting in the production of interferons. In the nasal epithelium which is in constant contact with the outside air, the temperature of the cells is likely to be low enough for Rhinovirus to get away with a successful infection. Rhinoviruses of course are one of many agents that cause cold and it is not known if other cold viruses are similarly checked at higher temperatures. It is likely that they use a repertoire of counter ­strategies to the host defense response, some aspects of which may be temperature dependent. The experiments in this study have been conducted on mouse cells grown in the lab. It remains to be seen if this holds true within living animals and whether it can be extended to human- ­cold virus interactions.

An interview with Dr. Ellen Foxman

Q. How would you place this work in context of the unanswered questions in the field?

Question #1: Why do Rhinoviruses grow better at nasal cavity temperature than at lung temperature? It has been known since the 1960s that most Rhinovirus strains replicate poorly at body temperature (37°C) and better at slightly cooler temperatures (33 – 35°C) such as the temperatures found in the nasal cavity. However, the reason for this was not known. In our study, we observed that Rhinovirus ­infected cells fight back against infection more at 37°C than at 33°C—in other words, the immune response triggered by the virus within infected is more robust at 37°C, and this is an important mechanism suppressing growth of the virus at 37°C. b. Question #2: Does temperature affect the immune response to diverse pathogens, or just the immune response to Rhinovirus? We found that two cardinal signaling pathways involved in immune defense were more active at body temperature than at nasal temperature: RIG­I like receptor signaling and Type I interferon receptor signaling. Since these pathways help defend us against in many viral different infections, our results raise the possibility that cool temperature also provides an advantage to viruses other than Rhinovirus. For example, many respiratory viruses cause colds more often than they cause lung infections; perhaps this is a reason why. That being said, it will be important to directly test other viruses, since viruses are tricky and many viruses have evolved ways to interfere with the immune responses we studied.

Q. How do you plan to take this study forward? What are the strengths and limitations of your model system?

The strength of our study was that we used a very well­ defined experimental system in which we could change one variable at a time to identify the immune system machinery needed to fight Rhinovirus within infected cells and to examine the effect of changing the temperature without changing anything else. Specifically, we used mouse primary airway cells grown in the laboratory. This way, we could compare cells from normal mice with cells from mice that differed by only one gene within the immune system. This allowed us to pinpoint which molecules within the immune system were important for defense against Rhinovirus, and which defenses were (or weren’t) affected by temperature. Also, by culturing cells in the lab, we were able to place them in incubators with controlled temperatures to clearly assess the effect of temperature without other confounding factors. b. Limitations/next steps: Although in general mice have been a good animal model for the human immune system, mice aren’t humans, and the next step in the study will be to examine in more detail how these mechanisms work within the human airway.

Q. Rhinoviruses are known to sometimes infect the lower respiratory tract (5) ​ what do you think is going on there?

One possibility is that the immune mechanisms required to block Rhinovirus infection don’t work as well in people who tend to have lung symptoms with Rhinovirus infection—for example, people with asthma. In our study, we found that if we used mouse cells lacking the necessary immune system machinery to block Rhinovirus infection, the virus could grow quite well at 37°C. There is some evidence that in airway cells from people with asthma, this machinery may not function properly; if this is the case, this might be what permits the virus to thrive at the warmer temperatures of the lung.

Q. Do you think alternating the temperatures (for example in a real world scenario inhaling steam or hot water gargling versus eating an ice cream) impact the success of Rhinovirus infection? In other words you perform the entire infection at one temperature, are there shorter time windows within which a temperature change would positively impact disease outcomes (for example gargling every morning or drinking hot water after eating an ice cream?)​ ?

We did do some temperature shift experiments (see Figure S3 in the paper.) We found that the level of the immune response tracked with the temperature of the cells during the time window when the virus was actively replicating; the temperature before the infection didn’t matter much. I would speculate that some exposure of infected cells to warm temperature at any point when the virus is actively replicating might be beneficial.

Q. What kind of experiments would you need to conduct to suggest to people that using different methods to increase the temperature of the upper respiratory tract – like drinking hot water may help fight cold? Have they been done?

The best way to prove that an intervention works is to directly test it, as you are suggesting. In this case, the best experiment would be to expose a group of volunteers to a fixed dose of Rhinovirus, and then place half of them on a well ­defined hot water drinking program (perhaps the other half could drink only cold water. If hot water program were effective, you would expect to see fewer colds develop in the hot water group than in the other group. This is a difficult study to perform: ideally, you would want to test a group of people who are identical in every way (genetics, behavior, environment, history of exposure to infections, etc.) except for the hot water drinking. In reality, this is quite hard to do, since every person is different! However, it might be possible to see an effect by studying a large group of people, especially if hot water drinking had a big impact (rather than a small effect) on whether or not colds developed after exposure to Rhinovirus. These types of studies can be very informative, but also can be complicated to interpret due to the inability to control all of the variables that may affect the outcome you are measuring (in this case, development of cold symptoms.) b. I do not know of any study considered to be definitive on this subject. However, I did a literature search and found a number of studies that have looked at the effect of hot liquids or steam inhalation on common cold symptoms, and I did find a number of these. You can read a few of these to get a feeling for the strengths and limitations. For example: i. Sanu and Eccles, 2008: This study tested the effect of hot liquid drinking on cold and flu symptoms in subjects who were recruited when they already had symptoms—the pathogen causing the symptoms is unknown. ii. Singh and Singh, 2013, meta-analysis of multiple studies looking at steam inhalation and common cold symptoms.

Q. You have emphasized on cell autonomous response to viral infections, what about the other aspects of the immune response? Do you think they could also contribute to temperature sensitivity?

We only looked at the cell autonomous immune responses in this study, and these were solely responsible for the temperature ­dependent blockade of Rhinovirus in our experiments. In the body, where many cell types are present, the responses we examined (RIG­I like receptor signaling and the Type I interferon response) can profoundly affect nearby and even distant cells through the action of secreted chemicals (cytokines). In this way, the phenomena we observed could also contribute to the temperature ­dependence of other immune responses; however, as yet we have no evidence for this.