So how does Respiratory Syncytial Virus infect your lungs?

This post is hopefully the beginning of a new series of stories on this blog covering the thinking behind the papers I cover from the point of view of those scientists doing the work. 

The first up is from Remi Villenave, a post-doc in my department at Queen's University, Belfast Centre for Infection & Immunity. Remi is currently working on understanding how Respiratory Syncytial Virus (RSV) infects and causes disease in humans. He studies this by using an in vitro culture of primary differentiated bronchial epithelial cells taken from children. This work is reported in a PNAS paper published a couple of weeks ago. 

In vitro modeling of respiratory syncytial virus infection of pediatric bronchial epithelium, the primary target of infection in vivo

Here is some answers he gave to a couple of questions I put to him last week:

Why study a virus like Respiratory Syncytial Virus, RSV? Why not something like HIV or ebola?


RSV is the primary cause of respiratory infections in infants and now it is more and more considered as an opportunistic pathogen, infecting the elderly and immunocompromised. [RSV is an extremely common viral infection and in older kids and immunocompetent adults causes mild cold-like symptoms.] Few numbers : 160 000 deaths and 60 millions infections worldwide annually according to WHO. Mainly in developing countries. However, some evidence suggest that RSV infections of infants might trigger wheezing and asthma later in childhood. So the impact of RSV might be underestimated. The last idea is subject to controversy and part of the literature failed to correlate RSV severe infection in infancy to asthma. [ There currently is no licensed vaccine to protect against RSV infection. There is however a monoclonal antibody to protect those in high-risk groups, although it is extremely expensive.]

RSV data in the UK from week 42 (2011) 

[For more information of RSV see the NCBI's piece here. And also the WHO page.]

Tell me a bit about RSV infection in general.

RSV is a respiratory pathogen that infects human and chimpanzees. [It was only isolated by Robert M. Chanock in the early 1960's.] An interesting feature of this virus is that even if you were infected as a kid, you are never fully protected, meaning you can be re-infected throughout life many times. This suggests that RSV is very good at hiding itself from the immune system or that RSV triggers a response that leads to inhibition or impairment of immune memory. Infection as a baby sometimes leads to hospitalization for severe bronchiolitis and even pneumonia. 

RSV is main cause of bronchiolitis in infants

At a cellular level, RSV infects ciliated cells of the lining of the upper and lower airways epithelia. Consequently, infected cells are dying, mucus is overproduced and as a result plugs of dead cells and mucus are obstructing airways. Additionally, RSV infection triggers inflammation of the airways, leading to oedema which in turn will contribute to the airway obstruction and make it even more difficult for the baby to breath. The cilia activity is also impaired which will make the mucociliary clearance infective.

Here's a 2002 paper initially detailing RSV infection of cultured epithelial cells. 

So describe this model you used and how you are using it to investigate RSV infection?

RSV proteins in green and cell cilia in red.

This model is based on an old clever idea from the 80's. The idea was to culture cells in the same physical conditions that you find in the epithelium, an air-liquid interface. The human epithelium (skin, gut, lung, nose, etc...), every organs at the interface between the body and the environment, is polarised, there is a basal side (the body) and an apical side (the external environment). In the late 80's, people started to recreate these conditions by seeding animal then human cells on semi-permeable membranes. These membranes are permeable for the medium but not for bigger elements such as cells, you can therefore put the cells on this membrane and they start growing. In this system, you have two compartments, the apical compartment, where the cells are, representing the external environment, and the basal compartment, representing the inside of the body, where the growth medium is. So in contrast with a classic culture system, the cells are not submerged by the growth medium, instead, the growth medium feed the cells from below, like in reality. 

For our study, we adapted this system in order to re-create a bronchial epithelium because RSV primarily infects the bronchial epithelium. In order to do this, we collected bronchial cells from children and cultured them at an air-liquid interface. When air-liquid interface was induced, something "magical" happened, the cells started to differentiate, meaning that all the cells that looked alike before started to differentiate in different cell types*. For instance, if you wait 3 weeks after air-liquid interface, you see ciliated cells, mucus-producing cells (aka goblet cells) and some other subtype that I am not aware of.

This model we created is a good system to study RSV infection as it mimics the bronchial epithelium physiology. Our hope was that by mimicking bronchial epithelium biology, our observation of RSV cytopathology would be close to what happen in-vivo, in individuals infected with RSV.  The results we showed in the paper seem to confirm that.

*I must say here that I still dont know if the cells we collected from children de-differentiated to stem cells during the expansion phase or if they retained some of their phenotype and therefore re-differentiated in their previous cell type. This would be interesting to know but would necessitate a bit of work.

Could you use this method to culture other primary cells? Could you even use mixed cultures?

This method is used to culture all kind of epithelia such as skin (used a lot in cosmetic research), gut epithelium, etc... You can absolutely mix different cells with this model, the limitation will be to find a growth medium in which the different cell types will be happy to grow. This is a difficult challenge.

You even used cultures from multiple donors. Is this important?

I think this is one of the crucial point, if not the most important, of our work. A major problem in scientific research is reproducibility. In other words, is what you see in your experiment real, or is it just an artefact that nobody else will see if they try to reproduce your experiment? Several studies have shown that a large proportion of what is actually published (even in top journals), is not reproducible, in other terms, is not real. This happen for many reasons. It can be because of fraud, fabrication of data (specially wit the high pressure to publish) or just because people do not care to reproduce their data with different methods to be sure that they are strong.

Bronchial airway brushing - how they got their primary cells.

The robustness of our data was something very important for us. What's the point to show something if you are not sure that its true? This is why we used cells from different donors. Although this creates big data variability, it allowed us to be sure that if we saw something, it was not an artefact. After all, everybody will respond differently to RSV infection but some characteristics are reproducible from individual to individual.

This model is clearly not a complete picture of what happens in vivo. There are no immune cells. Do you think this matters?

This model is just a first step towards modelling the human bronchial epithelium. And adding immune cells to it is the next step and it would be excellent to do it. However, having no immune cells allowed us to focus exclusively to the bronchial epithelium response to RSV infection. For example, a usual technique to quantify cytokines in children infected with RSV is to perform Broncho-alveolar lavages (BAL). However, titrating cytokines in BAL will give you a broad picture of what is happening during the infection. Here, we could decompose this picture and focus only on the epithelium. If you look at the IP-10 secretion, you can see enormous amount of IP-10 in BAL of sick children, but you have no idea what cells secrete this IP-10. Here our data suggest that it is the bronchial epithelium that secrete this high amount of IP-10 and not the immune cells around. Adding immune cells would definitely improve the model but would also complicate the data and the interpretation of the data.

What was wrong with using continuous cell lines or even mice? Could you not use primates?

Cells lines are not representative of the complexity of the human lung. Cell lines are cancer cells growing endlessly and passaged hundreds of time. Cell lines are good for high throughput experiments, were you need hundreds of different conditions quickly. Moreover, lots of tools are working with cell lines, for example transforming a cell line is easy, but transforming primary cells is more difficult. Mice is probably not the bestmodel to study RSV as it is semi permissive for RSV (the virus doesn't replicate very well) and infected mice do not reproduce the symptoms observed in human. However, mice have their advantages, such as tools available and knock down strains which are very useful. So every model has pros and cons, some have more cons than others.

The use of primates is difficult. Although they are relatively good models for lots of reasons (Murphy et al, 1992 -, Byrd et al, 1997), they present lots of ethical problems and are extremely expensive to use. Therefore, it is difficult to reach statistical significant numbers of individuals.

The chimpanzees is even a better model because it is extremely close to human (RSV has been first isolated in a chimpanzee). However, so many problems are associated with the use of chimpanzees in research that it is very difficult to use them for studies. On a personal level, I would prefer not to use them for RSV research.

Tell me about these two virus strains you are using. What is A2 and BT2a? How how are they different?

RSV particle schematic. From http://www.kuleuven.be/rega/mvr/images/RSV-1.jpg.

RSV A2 is a strain of RSV isolated in the 60's in Australia. It is the strain everybody is using around the world since many years to study RSV. The use of lab adapted strains such as RSV A2 is questioned more and more, because it gives results that are not relevant. For example, many studies identified heparin sulphate as RSV A2 receptor. It appears now that this is an artefact of this specific isolate being passaged countless times on inappropriate cell lines and heparin sulphate is not the 'real' receptor for RSV.

BT2a in contrast, is a clinical isolate, we collected during the winter 2006 inBelfast. We chose it because we felt that there was no logic in using a model of bronchial epithelium as authentic as we could and infect it with a non authentic strain of RSV, such as A2. This is why I think that it is important to use a clinical isolate.

It is difficult to say if A2 and BT2a are different because we did not sequence them fully. However, there are clear evidence in infectivity between both strains as the clinical isolate infects better than A2.

What happened when you infected your cultures [in vitro differentiated paediatric epithelial cells] with the two viruses? Were there any differences?

The two viruses infected the cultures. But BT2a infected much better than A2 and therefore induced more cytokines/chemokine secretion. However, A2 induced sometimes more cytokine than BT2a, suggesting that differences between the two viruses are not only due to the amount of infected cells.

These viruses seem to have the same growth kinetics although there seems to be differences in the numbers of virus infected cells. Why? Which is more "real"? Is A2 just churning out an 'incorrect' amount of virus?

There seems to be a difference in infectivity between both strains, but the growth kinetics are similar. One hypothesis is that production of A2 virions is more efficient than for BT2a. Its difficult to say if A2 is incorrect or if it is BT2a which is low. Another possibility could be that my method of titration is not accurate and “ favours” A2 compared to BT2a. I use HEp-2 cells and A2 is maybe more adapted to these cells, therefore it is growing better than BT2a. This could be a bias to the titration.

You keep comparing the results from this study to what happens in fatal cases. As in this paper here. Do you know the fatality rate of this virus? Do you not think you are focusing on a very rare event and maybe all infections are not like this?

In normal children the fatality rate is less than 1%. In high risk children (chronic lung disease, congenital heart disease, prematurity) it can reach around 40% (depending on studies). The definition of the bronchiolitis is very precise and we manage to observe few hallmark of it at a cellular level. The levels of cytokine we see, the amount of virus released on the apical side are consistent with what is seen in children hospitalized because of RSV. The presence of syncytia is also observed in autopsies. All the infections are certainly different and each individual will react differently, but from all the literature we read, some characteristics are reproduced in our model. Maybe we only modeled severe RSV infection, it is difficult to say.

Why weren't these goblet cells infected? Why wasn't every cell infected? Even other ciliated cells.

Goblet cells probably do not present RSV receptor at the surface. Same for all the cells I suppose. For other ciliated cells, there is maybe a refractory effect due to innate immunity. Or maybe, not all the ciliated cells are similar; there are maybe subtypes of ciliated cells with different characteristics. 

Why did you look at these particular cytokines? Are they important to RSV in particular? What about for other respiratory pathogens?

These cytokines are produced in human following RSV infection, this is why we chose them. They are important as some promote inflammation which is an important component of RSV disease.  These chemokines are usually found also in other respiratory infections where inflammation plays a big role.

If much of the disease is immune mediated why bother looking at the epithelial cells?

RSV-infected ciliated epithelial cells

There seems to be an immune driven component of the disease. But there is also a big epithelial involvement in the disease. The destruction of the epithelium, the impairment of the mucociliary clearance, the mucus hypersecretion also participate to the disease, in what proportion I would not know. But putting immune cells in this model would be interesting to study the immune component of the disease as well as the interaction between the epithelium and the immune cells. 

Can you comment on the differences that you see when you guys looked at non-differentiated airway epithelial cells?

This is interesting because the virus responses between A2 and BT2a are completely opposite in one model compared to another. In the non-differentiated cells A2 destroys everything and induces high cytokines levels, whereas BT2a infectivity is very limited. But I am not too surprised because I would expect that the phenotype of non-differentiated vs differentiated is extremely different. For example heparin sulphate is not express on the apical side of differentiated epithelium, I did not check on non-differentiated cells but I would suspect that there is some at the surface. This discrepancy between the 2 models could explain the differences observed.

Your conclusions state that the choice of RSV strain may be important in the results from these experiments but I do not really see such a big statistically significant difference when you look at A2 and BT2a. How come?

There are some in the chemokines titrations and overall BT2a is quantitatively different compared to A2. When you look at the non-differentiated model you can see lots of differences between the 2 viruses.

Could you have used a more 'wild-type' virus? Do you feel this would have made such a difference?

Its difficult to say but I passed BT2a 3 times. It is maybe too much to call it clinical isolate, but I think it's ok. 

Could your system be used to develop new antivirals or vaccine candidates?

New antivirals absolutely; vaccine candidates I don’t think so, unless you want to test the toxicity on the epithelium. Since there is no immune cells, it is not possible to study  adaptive immunity which is the base of vaccins.

Villenave, R., Thavagnanam, S., Sarlang, S., Parker, J., Douglas, I., Skibinski, G., Heaney, L., McKaigue, J., Coyle, P., Shields, M., & Power, U. (2012). In vitro modeling of respiratory syncytial virus infection of pediatric bronchial epithelium, the primary target of infection in vivo Proceedings of the National Academy of Sciences, 109 (13), 5040-5045 DOI: 10.1073/pnas.1110203109