The protein makes the poison: Dancing fruit flies and terfenadine

"Chemophobia" is the name of the exasperating phenomenon in which every material substance is branded as a "chemical" and made to look dangerous irrespective of context. Since everything in the universe is supposedly material, by definition chemophobia extends to everything. The media in particular has eagerly latched on to this idea, forgetting that almost everything (not just chemicals but life, liberty and the pursuit of happiness) is dangerous in the wrong quantities and context and harmless in the right ones. 

Matt Hartings at Sciencegeist had the excellent idea for us bloggers to do our part in dispelling chemophobia. He wants us to write about our favorite toxic chemical compounds. This will not only give us an opportunity to explore the many incarnations of toxicity but will also help inform the public about the highly context-specific safety and toxicity of chemicals.

My fellow bloggers have done a great job so far in documenting the various facts and myths about toxic molecules (you can find summaries on Matt's blog). A resounding theme in their posts is that "the dose makes the poison". It's an idea which goes back to Paracelsus in the 15th century and sounds intuitively true (consider the widespread injunction against gluttony), but which seems surprisingly recalcitrant to being universally accepted. This dose-specific toxicity especially makes its appearance in medicine, with unfortunate reports of celebrities fatally overdosing on prescription drugs regularly appearing in the news media. Strangely, the same media which readily accepts the fact that prescription drugs are safe as long as they are not abused in large quantities abandons its critical attitude when talking about "chemicals" in our food and clothing.

Dose-specific toxicity is indeed of paramount importance in medicine, but if you delve deeper, the common mechanism underlying the toxicity of many drugs often has less to do with the specific drugs themselves and more to do with the other major player in the interaction of drugs with the human body - proteins. Unwarranted dosages of drugs are certainly dangerous, but even in these cases the effect is often mediated by specific proteins. Thus in this post, I want to take a slightly different tack and want to reinforce the idea that when it comes to drugs it's often wise to remember that "the protein makes the poison". I want to reinforce the fact that toxicity is often a function of multiple entities and not just one. In fact this concept underlies most of the side-effects of drugs, manifested in all those ominous sounding warnings delivered in rapid fire intonations in otherwise soothing drug commercials.

What do I mean by "the protein makes the poison"? Almost every drug demonstrates its effects by binding to specific proteins which may be involved in particular diseases, and the goal of pharmaceutical research is to find molecules that target and inhibit or activate these proteins. There is of course much more to a drug than just inhibition of a protein, but that's the fundamental challenge. This goal was delineated during the turn of the twentieth century in Paul Ehrlich's notion of a "magic bullet", a compound that would hit only the rogue protein and nothing else. We are still trying to implement Ehrlich's program and in the process have discovered how hideously complicated the process is.

The thing is, in spite of much progress we still understand woefully little about the human body. When we design a drug to inhibit one protein, it has to contend with the thousands of other proteins in the body which perform crucial functions. Making a drug that binds to a protein is essentially like designing a key to fit a lock. Even if you think you have a perfect key that fits only one lock, the number of locks with similar structures is so large that it's very likely for parts of the key to fit other locks. And if these other locks or proteins play fundamental roles in normal physiological processes, you may be in trouble. In fact there's a name for this group of unwanted proteins - antitargets - and there are entire books written on how to avoid them.

Ideally you have to contend with every other protein when your goal is to target only one, but somewhat fortunately, the history of drug research has found out a handful of key proteins which seem to be often hit, leading to side-effects. In this post I will focus on two, and I will illustrate both through the example of the drug terfenadine (illustrated on top). Interestingly, the story of terfenadine reinforces the idea about both dosage and protein-specific toxicity.

Terfenadine was introduced in 1985 as an anti-allergy drug. Things seemed to be going well for its maker Hoechst Marion Roussel until 1990 when troubling reports emerged of a serious and potentially lethal side effect. This side effect was a perturbation of the heart's rhythm. It can be of several types, all of which are usually lumped under the title of "arrhythmias". In particular, terfenadine caused two phenomena with the impressive names of QT prolongation and torsades de pointes.

The heart is a pump, but it's also a kind of electrical motor with its own electrical cycle. This cycle is governed by the influx and outflux of various ions into heart cells; most commonly, sodium, potassium, calcium and chloride. The cycle shows up as peaks and troughs in electrocardiograms (ECGs). Each peak is alphabetically labeled, and the interval between the trough Q and the peak T is particularly important. It turns out that several drugs including terfenadine prolong this interval, essentially throwing the heart's rhythm out of sync. It is not hard to see that the consequences of disturbing this very fundamental rhythm of life can be catastrophic; the heart can stall, go into cardiac arrest and kill the unfortunate victim. QT prolongation can also be part of a larger indication called torsades de pointes, characterized by a specific shape of the ECG.

But what's responsible for this effect at the molecular level is a unique protein called human ether a-go-ho (hERG). The protein is an ion channel conducting potassium ions through heart cells, thus its crucial significance in maintaining heart rhythm should not be surprising. Terfenadine and several other drugs (most notably some antidepressants and antipsychotics) bind to this protein with high affinity and can even block it. The amusing name of the protein points to an amusing origin. The protein was a product of the human analog of genes discovered in fruit flies by researchers at the University of Wisconsin who were studying mutations in these genes. They found that the mutant flies' legs started to shake when they were anesthetized, making the insects look like entomological versions of Elvis. Another scientist at the City of Hope remembered where he had seen humans doing a similar dance; at the Whisky a Go Go nightclub in West Hollywood. It was the ultimate in anthropomorphization. Here's what part of the protein looks like.

When the FDA found out about the dangerous side effects that terfenadine mediated through the hERG ion channel, they sent a letter to doctors who were prescribing the drug and issued a black box warning. In 1997 the FDA finally withdrew terfenadine; after all there were several anti-allergy medications out there and there was no need to market an especially dangerous one. Since then, testing potential drugs against hERG is a mandatory part of seeking FDA approval and there is much research dedicated to finding out specific molecular features of drugs which may turn out to be hERG blockers; one common determinant seems to be the presence of a positively charged basic nitrogen atom. There are entire lists of drugs including marketed ones that can cause QT problems to varying extents under different conditions.

So there it is, toxicity mediated not just by a particular "chemical" but by its interaction with a particular protein. But the story does not end there. Toxicity problems with terfenadine seemed to occur - you guessed it - only at high dosages. The dose indeed made the poison. But there was an added twist. Some patients experienced hERG blockage only when they were taking other drugs, most notably the antibiotic erythromycin. Surprisingly this also happened when they were drinking quantities of, of all things, grapefruit juice. Grapefruit juice has also turned out to be important in the effects of other popular drugs like statins for heart disease.

What was going on? When terfenadine is administered, like any foreign molecule it has to first get through the gut wall and the liver to enter the bloodstream. And it's in the liver that it encounters a protein called cytochrome P450. This crucial protein is the great gatekeeper of the human body, denying entry to thousands of molecules which it deems to be poisonous. It is responsible for the metabolism of about 75% of all drugs. It served a necessary function during evolution when organisms had to keep potentially poisonous chemicals out, but it haunts drug discovery scientists in their dreams because of its ability to affect drug structures in unexpected ways. The centerpiece of P450 is an iron atom that oxidizes electron-rich bonds in molecules. Most of the times the protein induces an oxidation reaction in a drug that changes it to something else. As a further testament to the complexities of drug development, that "something else" itself can be toxic, beneficial or neutral. In case of terfenadine there was a stroke of good luck; cytochrome P450 was transforming the compound into another drug called fexofenadine. Chemists will recognize the small difference in the structures - a single carboxylate group at the terminal end. 

But as is often the case in the wonderful world of pharmacology, this tiny difference had momentous consequences; fexofenadine no longer bound to hERG with high affinity to cause QT-prolongation. What happened at high doses was that terfenadine saturated cytochrome P450 and some of it made its way into the bloodstream without being transformed into fexofenadine. Similarly the compounds in grapefruit juice preferentially bound to cytochrome P450, again allowing terfenadine to get past the protein. And this terfenadine which escaped the clutches of cytochrome P450 blocked hERG. One thing is clear here; it is chilling to contemplate the effects of terfenadine had it not been metabolized to fexofenadine by P450 in the first place.

This fascinating (at least for me) story of terfenadine drives home many important points regarding toxicity. Firstly, it takes two to tango, and toxicity is always a function of a drug and its target and not just of the drug alone. Secondly, we again had a case where "the dose made the poison". And thirdly, the reason this was true was because of a guardian angel, a protein which changed terfenadine into something else that was not toxic; a corollary of this point is that it takes a tiny change to turn a toxic compound into a non-toxic one.

There should be little more evidence needed to prove that toxicity is a many splendored, context-specific thing.

All images are from Wikipedia