Taking DRACOnian measures against viruses

When Alexander Fleming shamelessly took credit for a drug produced by the penicillium bread mould, our fight against bacteria completely changed. Even now, bacterial diseases are nowhere near the threat they were before that discovery. But today, we are seemingly having to put up with emerging viruses. Every year, we have to put up with entirely new strains of flu. There is nothing we can do about the viruses that cause the common cold. There are various antivirals out there, but they are specific for specific diseases, like flu, or HIV.
Generally, it’s been fairly difficult to make anti-viral drugs. Often, a lot of the problem comes from the way viruses work. When they are outside the cell, they don’t have any active metabolic processes that can be targeted by drugs. And when they are inside the host cells, the metabolic processes they’re mostly using are ours. So it’s very hard to target the viruses without attacking the host. Compared to bacteria, viruses present a very small target. 
Could a paper published in PLOS One called  “Broad Spectrum Anti-viral Therapeutics” represent a penicillin moment for viruses ? A trailblazer that can transform our relationship with viruses ? To work this out, let’s take a good long look at what these anti-viral therapeutics are, and what they do.  To do that, we have to talk about DRACO.

Are we talking about the DRACO from Harry Potter, or the dragon voiced by Sean Connery in Dragonheart ?

Whilst I compliment you on your knowledge of pop culture, you are wrong on both counts. DRACO is a handy acronym that stands for Double-stranded Ribonucleic acid Activated Caspase Oligomizer. DRACO is the drug that is being purported to extinguish all of those nasty viruses. To understand how it works, we need to take a look at how some of our most pernicious viruses work.
You may have heard about DNA, and how it’s genetic code is a blueprint for life. Your genome is encoded in your DNA. The nucleus of your cell holds all of the DNA, and acts like a library for your genetic code. When your cell needs a specific genetic code to make a certain protein, the nucleus makes an RNA copy of the appropriate gene and sends it out into the rest of the cell, where it can be used to construct a protein.
The goal of a virus is to enter a cell, and to hijack this process to make more viruses. Usually at some point, the virus will attempt to substitute it’s own genetic code for that of the host cell, tricking the host cell to make more viruses ^[1].
However, some viruses can store their genetic code using RNA only. Whilst RNA is less stable than the DNA, it means that these viruses can go straight to the cell machinery that translate RNA into protein, and get ahead with making virus based proteins.

 Influenza,  Hanta Virus , Dengue Fever, Rhinovirus, Lassa fever virus, and reoviruses all use RNA as the basis of their genomes, and at some point during their replication, form double stranded RNA. This last bit is important, because double stranded RNA is not usually found in human cells, and when they are found, they don’t last for very long^[3]. So essentially the creation of double stranded RNA has been the achilles heel for many viruses.
So naturally, humans (and other animals) have evolved ways to detect the presence of double stranded RNA. I’ll give one example, a protein known as TLR-3^[4].

TLR-3 detects dsRNA, and then it signals to other proteins within the cell, so that it can do a number of things that the virus won’t like. For instance, it leads the cell to produce interferon, which acts as a distress call to summon the immune system.  The most drastic reaction to the detection of viruses is the activation of caspases, which are the enzymatic equivalent of a self destruct button.

 However viruses can evolve very quickly, and have matched pace with us in this evolutionary arms race. Viruses have evolved a number of tricks to get past our natural defences against them. While they can’t evolve ways out of being detected by RNA binding proteins, they have evolved ways to short circuit the signalling cascade that can occur after their initial detection. Let’s say that TLR-3 binds dsRNA, it tells another protein, and then a virus protein get’s in the way, and the message is lost.

So the idea underlying DRACO is taking the dsRNA detecting ability of enzymes like TLR-3, and shortcutting all of the signalling pathways and go straight to pushing the self destruct button Calm down, it’s not as bad as it sounds.

Your body naturally kills of virus infected cells. a cell infected with virus is not your friend any more  it’s a factory for the enemy, churning out viruses to infect other cells. The normal immune response against viruses often involves killing off these cells, so don’t worry too much about them.

How does one make DRACOs ?

Now we’re getting into the actual paper, and away from the background. So how did this research group go about making DRACOs ?
They looked at a whole raft of different RNA binding proteins and looked at the parts of them which directly bind to the RNA. And then they looked at caspase proteins, and worked out which parts of those tend to cause human (or mouse cells) to self destruct. and they took those bits, and stuck them end to end.

Figure 1.A. of the paper shows the basic blueprint for these proteins. They also deliberately made versions of these DRACOs that were broken, to use as controls, just in case randomly giving cells proteins will protect them for no good reason. Figure 1.B. shows a western blot to prove that they created these proteins.

That’s great ! Do they work ?

So now that these DRACOs have been “designed” and created, the question is whether they work. So the first thing that we need to look at is whether these proteins actually get into cells. If they can’t do that, then there isn’t much hope that they can be effective.
The thing is, cells don’t just take up every protein that they get into contact with willy nilly. In order to gain passage into the cell, these proteins must have special tags on the end of them. In this case, they tried out DRACOs with PTD and TAT tags. They then added them to a culture with cells derived from humans (HeLa cells). They then extracted the cells from this,and tested whether the DRACOs had managed to get into the cells.

The western blot in Figure 2 A shows that DRACOs managed to get into the cells when they had the PTD or TAT tag added to them,  but not without it. Furthermore, in Figure 2 B, they added the DRACOs to a culture of cells, and took out samples at specific time points to test whether the DRACOs had entered. Judging by the image, the DRACOs were taken up as early as 10^[5]-15 minutes after administration. Figure 2C shows that the DRACOs were retained by these cells for about 6-7^[5] days after they were first applied.
So now we know it actually gets into the cells, the question is whether they can detect the presence of double stranded RNA (dsRNA), and furthermore whether they can cause cells to “self destruct” when they are present. So they took some human cells, and genetically modified them to produce dsRNA. If the DRACO’s worked , then they would immediately kill off these cells.

How do you work out whether these cells have caspases becoming active ?

To work out whether the cells were pushing their self destruct button by activating caspases, the scientists here did some clever cell manipulation. They added the gene for luciferase to the cells. Luciferases are enzymes that produce light when they grab onto specific chemicals called luciferins.
To these cells, they introduced a protein which mimics the kind of proteins that caspases usually act upon when they are activated, with one difference. Tied up within the structure of these proteins is Luciferin. So when the caspases are activated, they bind this substrate, the process causes luciferin to be released into the cell. This is then found by the luciferase enzymes which then cause light to be produced. So the researchers could work out how active the caspases are in these cells just by looking at how much light they are making.
So they gave some of these cells caspase inhibitors. If DRACO’s were naturally lethal to these cells, the presence of these inhibitors would make no difference to whether the cells would self destruct (a process known as apoptosis).

 In Figure 3, they added the DRACO’s to these cells, either with, or without the inhibitor. They also included a product which makes cells which have just self destructed glow in the dark.  The first four sections on the graph are simply controls, to pick up the background levels of cell death. Since the main function of caspases is to cause cell death to occur, you can guess what would happen if we were to add caspase inhibitors to a normal set of cells. The blue and red bars are both lower than the green bar , because they have the caspase inhibitors added. The next three sections show what happens when DRACO’s are added to the mix, and they show that they kill off a lot of cells. And importantly, you can tell that it’s performed using caspases, because in the presence of inhibitor, the cells do not die as much. In fact, the levels of death seen is more or less the same as the other controls with inhibitors.
Ah yes, but this doesn’t necessarily prove anything ! The DRACO’s were created using e.coli cells, and it could be possible that when extracting these cells, some other nasty substances were pulled out that could account for this effect. So in the next set of data, they add some of the e.coli extract, and show that actually it doesn’t have any effect, when compared with controls.
But you forget, cells naturally try to off themselves when confronted with dsRNA ! how do we know that the supposed effect of the DRACO isn’t caused by that ? Because they then tested the cells with added dsRNA. And while the cells did indeed die off more than controls, it was still much lower than when the DRACO’s were added.  They also added a compound known as camptothecin to deliberately trigger the self destruct in these cells.
So at the end of this last figure, we know that DRACO’s do what they say on the tin. When DsRNA is present, it activates Caspases and causes infected cells to Off themselves. But we haven’t yet tested them with real life actual viruses.
Whilst in the last experiment, we were looking at caspases only, in the next one we want to know whether it actually improves cell survival.
The theory behind DRACO’s is that the first cells to get infected should be the last cells to get infected, and die off before they can spread virus to the rest of the cells in the culture.

In Figure 4 A, they use rhinovirus, one of the viruses that cause the common cold, and add it to Normal Human Lung Fibroblasts, which are a cell type usually found in the lung. They are the normal target for rhinovirus. So they added the rhinovirus to the culture of lung cells, to see whether the presence of DRACO would save them. The first four sections show the controls, showing that no one part of the DRACO extract protects the cells on their own. Within 12 days post infection, all of the cells are dead.
The last two sections are the most interesting part of this graph. The last section shows that when the complete , fully functional DRACO is added to these cells, they are protected against rhinovirus infection.
Okay, that’s good. I should note that for this experiment, the DRACO’s were already in the cell culture when the viruses attacked. But what happens if you got rid of the DRACO? Could the virus bounce back, only being momentarily delayed ? How long do these cells need to be given DRACO to stop the infection?
Figure 4B goes some way to attempting to answer this question. A number of different types of DRACO were added to cell cultures, which were then exposed to the rhinovirus. After three days of marinating in DRACO filled medium^[6], the some cells were removed and left in plain old normal medium. Whether or not media were changed, the effects were still the same after seven days. This shows that whatever DRACO is doing, it’s happening within three days of infection.
In Figure 4C, we drive down further into this question. If you remember earlier, that DRACO’s can stick around in cells for around 6 days? So what happens if we treat cells with DRACO, and add in rhinovirus after 6 days ? it turns out that the cells survive. But the question is, can DRACO work if it is given after infection ? The last 3 sections of 4C show that it can work for up to 3 days after the initial infection.
So they next tested a whole raft of different types of DRACO’s (Figure 5 A) to see their efficacy against rhinovirus infection. In the previous studies, they showed that DRACOs can allow increased cell survival. But they haven’t yet shown a reduction in the levels of virus. Figure 5B solves this. Cultures of human lung cells were tested four days after infection to see detect the whether any rhinovirus was present. Turns out, the cultures with DRACO’s did not have any viruses, whereas the controls did.
The next experiment was a basic dose response, which asked the question ; What concentration of DRACO will save all of the cells in a culture ?
So they took a set of cell cultures, and added different concentrations of DRACO to them. They then infected each of these cultures with different species of virus. They used Rhinovirus, murine encephomyelitis and intriguingly murine adenovirus. At concentrations of 0.1 nM, no protective effects were found.  The DRACOs seemed to be effective against all the viruses at around 10nM. As with all experiments that involve a line graph and phenomena that could possibly be described by an equation , I wonder where the regression’s at ?  But I’ll talk more about that later.
But the interesting thing here is that for some reason the DRACO is effective against adenoviruses. This is interesting because adenoviruses do not have a genome of RNA- they have a genome of DNA, and are not noted for using dsRNA at any point during their life cycle.
In figure 6, they tested whether DRACOs were as effective against adenovirus as they were against rhinoviruses. DRACOs were effective against rhinovirus seemed to be even more effective against adenovirus, with it conferring 100% protection even if applied 3 days after infection. And a whole raft of DRACOs were effective against the adenoviruses. Similar tests were applied to the murine encephomyelitis, amapari, dengue and guama viruses, which are each viruses from different families, and all of them form dsRNA at some point in their replication cycle.  I could go into more detail about all of the viruses they’ve tested, but suffice to say that DRACOs look like they do what they say on the tin.
But a drug needs to do more than just stop viruses in a cell culture. A cell culture is basically just a culture of one cell type floating in fluid. The interior of an organism has connective tissue, a circulatory system and a whole variety of cell types all interacting with eachother in a complex structured mass which can’t be replicated in a cell culture. If this drug is to be effective, then it needs to be able to be absorbed into the body, and survive long enough to reach the same cells that the virus intends to infect. Since we’re talking about this drug fighting against different types of virus, then it’ll have to go into different places in the body. If it needs to fight against hepatitis C, then it’ll need to get to the liver. For rhinovirus, the lungs, and for something really horrible like hantavirus, it pretty much needs to get to the entire circulation.
So they need to know how fast this drug can pass through the circulatory system. Drugs have two routes out of the circulation system after they enter. Route 1 is via the kidneys, which act to filter out waste and toxic products from the blood. Route 2 is via the liver, which generally drains blood coming from the gut, and is the main place in the body where drugs accumulate and are detoxified. So when drugs end up in these areas, a fair bet is that they’ve gone through the circulation.

In figure 9 A, they inject the drug into the circulation of a mouse, and then tested out specific organs at specific times to see how long it took for the drug to appear in these organs. within 2 hours, the DRACOs appear in the lung and the kidneys. It starts to appear in the liver not long after, and it stays there for over a day.
So the question arises- can this drug actually work when given to a real live whole organism ? To assess this, they performed a survival experiment (Figure 9 B). They gave two groups of mice different types of DRACO, and one group no treatment at all.  The treatment was given the day before, and for three days after infection with H1N1 flu virus. 13 days after infection, most of the mice that didn’t receive the DRACO were dead, whilst most of the mice who received the drug survived. Some mice had their lungs extracted, and the numbers of viruses present in their lungs were assessed. The mice treated with the DRACOs had much fewer viruses within them than the untreated mice. In Figure 9C , a repeat of this experiment  was performed , only using different DRACOs, and smaller animal numbers.
But wait ! Let’s take another look at this infection model. The mice received intraperitoneal injections of DRACO of 200 microlitres. Now this may not be the most practical methods of application for use in people, it’s sort of (but not really) the equivalent of an iv drip. But perhaps if the DRACO was administered through a different method, such as via the lungs, it may be either more or less effective.

True to this, Figure 10 A shows the distribution of different types of DRACO to the lungs, after they were administered to the lung. The DRACO that remained in the lung the longest was also the one which was the most protective for mice infected with influenza.
So those were the experiments, but the question is, what can we take away from this research.

Summary

Cons

• Error Bars:  I know always tend to bang on about this when I review papers, but it’s really important that when you put error bars on a graph, that you explain what the hell they are. Now, to be fair, this isn’t a problem for all the figures, as some do note that the error bars show standard error of the mean. But when I see a graph with error bars, but doesn’t tell you what they mean, I often assume the worst, that they’ve let excel do it’s automatic error bar thing, or they are too embarrassed to show you the actual standard deviation.
• Statistics: This paper was relatively thin on the statistics. Often people tend to think that the only point of stats is to get a small p-value to “prove” that their pet theory is true. But it’s so much more than that ! It’s the best way to really get to grips with the shape of your data without your preconceptions getting in the way. The statistics they use are so basically to convert the numbers of dead/live cells into “percentage viabilities”. But they don’t seem to use any statistical analysis. I made a comment earlier about how they could have fitted their dose response data to an equation using regression, as it does look like a traditional hill slope. With that sort of information, you can actually get some interesting information about the similarities/ differences between the viruses. Without the stats, I have to make a guess that the variations in the cell viability are not large enough to account for these differences.
• dsCARE. In 2009, a group based in boston/ china also created anti-viral based dsRNA that had a similar design to the DRACOs described here , except there they were called dsRNA dependant caspase recruiters. There are only a few differences with this paper- they use microscopy to judge whether viruses are there or not, as well as cell viability, and use direct counts for virus.They even discovered adenoviruses vulnerability to dsRNA based treatment. But they don’t look at the drug distribution, a much broader set of viruses and cell types, show no animal work, and didn’t get the patents. Read it here.

Pros

• Breadth: They have shown quite effectively (and exhaustively) that DRACOs can prevent an infection propagating throughout a culture of cells. Now, I should note that without statistics, I am giving them the benefit of the doubt on this. But I think that the evidence they present is really compelling, and grounds for optimism. I mean the great thing about this paper is the sheer breadth of it. I may have name checked an earlier paper with a similar concept, but personally I like this one a fair bit more. This is the kind of paper that simply couldn’t have been published anywhere else, I mean there are 22 f***ing figures. That’s a huge amount of data. I mean, I’ve seen papers jump to far greater conclusions with less than a quarter of what this group is presenting.

So should we get excited ? Well a little bit. You can allow yourself a little chuckle perhaps. This is good news in the fight against viruses.  They may not have been the first to come up and test the concept for this drug, but they have done a lot to build a case for it being broad spectrum, and for the possibility of it being useful for treating viral diseases.  Even though the outlook is promising, this treatment is still in it’s early stages. The proteins may need to be altered, improved. They may even be ditched altogether and the genes encoding the drugs may be given. The treatment that will eventually result from these findings may look nothing like what was used here. It may look promising all the way to human tests, and then something could go wrong then.
This potentially could be a penicillin moment for viruses. But it’s too soon to tell. Such historic moments can only ever be judged with hindsight. I only hope that when I’m judging this moment in the future, it’ll be a future in which viruses have been knocked back as badly as bacteria once were.

Footnotes

[1] I should note that whilst this is the way that many viruses work, there are some exceptions. Mimiviruses, which generally are found in the sea infecting amoebas, may work in a completely different way that is mostly unknown at this moment.
[2] This footnote has no point to it. There is no reference in the text. It exists for its own sake, at the expense of the rest of the article.
[3] One exception is small interfering RNA. small interfering RNA’s do form double stranded RNA, which is recognised and rapidly broken down by the cell. The fact that double stranded RNA is broken down rapidly by the cell is why siRNA’s are so useful for a cell. For the first time, I can say that is the exception that proves the rule.
[4] TLR-3 stands for “Toll-Like receptor 3” because it was similar to a protein called Toll. and there were three more discovered before it that also looked like Toll. Yeah, it doesn’t really have anything to do with it’s function, but most proteins have names that occur by accident, and have nothing to do with their function.
[5] you may really need to squint your eyes to see the protein band here. This may be an instance of what my supervisor would refer to as “visible by photoshop”
[6] Feel free to use this phrase in your harry potter slash fic


Rider T.H., Zook C.E., Boettcher T.L., Wick S.T., Pancoast J.S., Zusman B.D. & Sambhara S. (2011). Broad-Spectrum Antiviral Therapeutics, PLoS ONE, 6 (7) e22572. DOI: 10.1371/journal.pone.0022572.t002