Can we turn biology into engineering?

Vijay Pande of Andreessen-Horowitz/Stanford has a thought-provoking piece in Scientific American in which he lays out a game plan for how we could potentially make biology more like engineering. He takes issue with what Derek (Lowe) once called the “Andy Grove fallacy”, which in a nutshell says that you can make fields like biotechnology and drug discovery as efficient as semiconductor or automobile engineering if you borrow principles from engineering.

There are parts of the piece that I resoundingly agree with; for instance, there’s little doubt that fields like automation and AI are going to have a significant impact on making biological experiments more reproducible, many of which are still more art than science and subject to the whims and sloppiness of their creators and their lab notebooks. Vijay is also optimistic about making biology more modular, so that one can string along parts of molecules, cells and organelles to enable better biological engineering of body parts, drugs and genetic systems. He also believes that bringing more quantitative measurements encoded into key performance indicators (KPI) will make the discipline more efficient and more mindful of its successes and failures. One point which I think is very important is that these kinds of approaches would allow us to gather more negative data, a data collection problem that still hobbles AI and machine learning approaches.

So far I am with him; I don’t believe that biology can’t ever benefit from such approaches, and it’s certainly true that the applications of AI, automation and other engineering-based approaches are only going to increase with time. But the article doesn’t mention some very fundamental differences between biology and engineering which I think demarcate the two substantially from each other and which make knowledge transfer between them highly problematic.

Foremost among these are non-linearity, redundancy and emergence.

Let’s take two examples which the piece talks about that illustrate all three concepts – building bridges and the Apollo project. Comparisons with the latter always make me wince a bit. Vijay is quite right that the right approach to the Apollo program was to break the problems into parts, then further break those parts up into individual steps such as building small models. The scientists and engineers working on the program gradually built up layers of complexity until the model that they tested was the moon landing itself.

Now, the fact is that we already do this in biology. For instance, when we want to understand or treat a disease, we try to break it down to simpler levels – organelles, cells, proteins, genes – and then try to understand and modulate each of these entities. We use animal models like genetically engineered mice and dogs as simpler representatives of complex human biology. But firstly - and as we keep on finding out - these models are pale shadows of true human biology; we use them because we can't do better. And secondly, even these ‘simple’ models are much more complex than we think. The reasons are non-linearity and emergence, both of which can thwart modular approaches. The sum of proteins in a cell is not the same as the cell phenotype itself, just like the sum of neurons in a human brain is not the brain itself. So modulating a protein for instance can cause complex downstream effects that depend on both the strength and nature of the modulating signal. In addition, biological pathways are redundant, so modulating one can cause another one to take over, or for the pathway to switch between complex networks. Many parts downstream, even ones that don’t seem to be directly connected, can interact with each other through complex, non-linear feedback through far-flung networks.

This is very unlike engineering. The equivalent of these unpredictable consequences in building a bridge, for example, would be for a second bridge to sprout out of nowhere when the first one is built, or the rock on the other side of the river suddenly turning from metamorphic to sedimentary, or the sum of weights of two parallel beams on the bridge being more than what simple addition would suggest. Or imagine the Apollo rocket suddenly accelerating to ten times its speed when the booster rockets fall off. Or the shape of the reentry vehicle suddenly changing through some weird feedback mechanisms as it reaches a certain temperature when it’s hurtling through the atmosphere. 

Whatever the complexities of challenging engineering projects like building rockets or bridges, they are still highly predictable compared to the effects of engineering biology. The fact of the matter is that the laws of aerodynamics and gravity were extremely well understood before the Apollo program (literally) took off the ground, so as amazing as the achievement was, it didn't involve discovering new basic scientific laws on the fly, something that we do a lot in biology. Aircraft design is decidedly not drug design. And all this is simply a product of ignorance, ignorance of the laws of biology and evolution – a clunky, suboptimal, haphazard, opportunistic process if there ever was one – relative to the laws of (largely predictable) Newtonian physics that underlie engineering problems.

The concept of modularity in biology therefore becomes very tricky compared to engineering. There is some modularity in biology for sure, but it’s not going to take you all the way. One of the reasons is that unlike modularity in engineering, biological modularity is flexible, both spatially and temporally. This is again a consequence of different levels of emergence. For instance, a long time ago we thought that the brain was modular and that the fundamental modules were neurons. This view has now changed and we think that it’s networks of neurons that are the basic modules. But we don’t even think that these modular networks are fixed through space and time; they likely form, dissolve and change members and locations in the brain according to need, much like political groups fleetingly forming and breaking apart for convenience. The problem is that we don’t know what level of modularity is relevant to addressing a particular problem. For instance, is the right ‘module’ for thinking about Alzheimer’s disease the beta-amyloid protein, or is it the mitochondria and its redox state, or is it the gut-brain axis and the microbiome? In addition, modules in biology are again non-linear, so the effects from combining two modules are not going to simply be twice the effects of one module – they can be twice or half or even zero.

Now, having noted all these problems, I certainly don’t think that biology cannot benefit at all from the principles of engineering. For one thing, I have always thought that biologists should really take the “move fast and break things” philosophy of software engineering to heart; we simply don’t spend enough time trying to break and falsify hypotheses, and this leads to a lot of attrition and time chasing ghosts down rabbit holes. More importantly though, as a big fan of tool-driven scientific revolutions, I do believe that inventing tools like CRISPR and sequencing will allow us to study biological systems at an increasingly fine-grained level. They will allow us to gather more measurements that would allow better AL/machine learning models, and I am all for this.

But all this will work as far as we realize that the real problem is not improving the measurements, it’s knowing what measurements to make in the first place. Otherwise we find ourselves in the classic position of the drunkard trying to find his keys below the lamp, because that’s where the light is. Inventing better lamps, or a metal detector for that matter, is not going to help if we are looking in the wrong place for the keys. Or looking for keys when we should really be looking for a muffin.

Intramolecular Hydrogen Bonds, We Hardly Knew Thee

Intramolecular hydrogen bonds (IHBs) are interesting beasts. They can be used to improve the potency of a drug by constraining it in a bioactive conformation, and they can be used to hide polarity and improve permeability; cyclosporin being the classic example of the latter.

But the exact amount of potency gain you get through formation of an intramolecular hydrogen bond is not clear. Conformationally you could constrain a molecule through an IHB, but you would still be introducing polarity and hydrogen bond donors and acceptors, and this will lead to some desolvation penalties that would have to be exactly compensated for by the IHB to lead to a net positive effect. There's a group from D. E. Shaw and AstraZeneca who have now looked at about 1200 cases of matched molecular pairs and their biological activities to figure out the contribution of IHBs to potency. The average gain they see? Close to zero. This means that on average, an IHB is as likely to blunt your potency as it is to improve it.

What's more interesting are the outliers. A small but distinct fraction of results display matched molecular pairs in which there is a gain of at least 2 to 3 log units. Inspection of these results sheds light on why these cases really benefit from the formation of IHBs, but it's also important to not overemphasize the positive role of the IHBs in every instance, especially in cases where the non-IHB matched pair displays particular severe repulsive interactions. 

For instance there's an N-Me vs N-H pair in which the latter can form an IHB while the former cannot. But not only can the former not form this bond, the N-Me probably causes a steric repulsion and causes a very different conformational profile, perhaps leading to a conformational penalty in binding to the target. Similarly in another case, the matched pair presents an N-H vs O difference. Here again, not only can the oxygen not form the bond, but it likely strongly repels the other oxygen participating in the IHB. Thus, these outliers are outliers not because the IHB is particularly stable, but because the complementary arrangement is particularly unstable.

Nonetheless, this is a nice study to keep in mind every time you want to use an IHB as a tactic for improving potency or permeability. It may well work, but then it may well not. As in most cases in drug discovery, the decision to incorporate an IHB-forming element will be dictated by many other factors including cost, resources and synthetic accessibility. As with many other tactics in the field, when it comes to IHBs, caveat emptor.

Macrocycles, flexibility and biological activity: A tortuous pairing

Here's an interesting paper from the Jacobson, Wells and Walsh labs at UCSF and Stanford that seeks to demonstrate how restricting the flexibility of macrocycles may lead to better inhibition of their targets from an entropic perspective. The authors are looking at a non-ribosomal peptide called thiocillin which inhibits the growth of Gram positive bacteria, especially MRSA.

What they wanted to determine was the effect of point mutations in the peptide on the inhibition. They performed saturation mutagenesis between positions 2 and 9 of the peptide and generated 152 mutants whose activities they tested in a minimum inhibitory concentration (MIC) assay. They found that 8 point mutants especially resulted in more potent analogs.

Now there can be several reasons why the potency went up, but one potential reason is entropy. Macrocycles, while often more rigid than their corresponding linear analogs, are still quite flexible. In fact, my own work with the macrocycle dictyostatin in graduate school showed how flexible even a supposedly constrained molecule can be. What this paper finds out is that in cases where the mutant lost activity, there was a corresponding increase in flexibility and entropy as measured by the number and distinctive nature of conformations from a conformational search technique which they have developed. Particularly striking changes in potency occurred when a single residue was modified from having a planar sp2 carbon to a non-planar sp3 carbon: in that case the saturated analog had many more conformations than the unsaturated one.

As someone who has always been partial to the impact of entropy and conformational flexibility on molecular activity, I like this kind of work. But I am not quite convinced yet that it is decreased flexibility that leads to more potent inhibition. For one thing, inhibition is not direct binding, and there are a variety of factors including changes in cell permeability and off target effects that could lead to the observed changes in inhibitory - not binding - affinity. Secondly, there were 152 mutants, and it's not clear to me how many were tested for flexibility: in other words, I am not sure there were enough controls to determine whether the flexibility-inhibition correlation really holds up. For instance, many of the mutants were inactive: were there instances in which some of these were actually less flexible and challenged the hypothesis? Another way to put it is to ask what the right null model for this dataset is. 

Thirdly, decreased or increased inhibition can be a result of both more conformations as well as conformational selection. For instance, two macrocycles can have similar conformations, but in one case a particular conformation more suitable for binding could be more stable (perhaps because of an intramolecular hydrogen bond) and represented to a higher degree in solution, making it easier for a protein target to pick it out. Lastly, it is not clear whether the improved affinity could simply have been a result of better interactions: although that seems unlikely for the sp2 vs sp3 pair above, it is nonetheless a factor that could be operating in other cases.

Entropy is an important consideration in drug design, but it's also trickier than it sounds to both understand its effects and implement its benefits. To their credit the authors acknowledge that rigidity is a necessary but not sufficient condition for increased affinity, and other studies seem to bear it out. Macrocyclization can also be counterintuitive: for instance in my own studies I found out that dictyostatin which is a macrocycle seems more flexible than its corresponding acyclic counterpart discodermolide. In that case it was fairly straightforward syn-pentane interactions which made the acyclic molecule rigid. In other cases it could be the opposite. In any case, this study serves as an interesting starting point for exploring the impact of flexibility on drug affinity, but it also serves to illustrate how thick the jungle of SAR really is.

Here's that prospective drug design study you asked for...

Improving ligand binding selectivity by
electrostatic optimization

Here's a rare beast: a set of ten prospective molecular design studies from Roche, mostly made possible by a combination of modeling and crystallography. It's a valuable contribution to the field in my opinion, especially since people are always asking about prospective successes of molecular modeling and because companies are reluctant to divulge that kind of data. I especially liked the authors' emphasis on qualitative rather than quantitative aspects of modeling approaches; this is a subtlety not always appreciated by critics and is in fact one that goes to the heart of chemistry as a predictive science.

The review deals with ten early discovery projects involving diverse targets where a variety of modeling techniques were used to improve affinity, selectivity, solubility, pharmacokinetic properties and a bunch of other desired druglike characteristics.

Some of the applications (filling hydrophobic pockets with small aromatic substituents or designing 'steric bumps' to get selectivity against other protein subtypes) are relatively straightforward while others (scaffold hopping, getting affinity by generalized electrostatic optimization, homology modeling) are more challenging and interesting. Here is a table displaying target type, approach and impact of the various protocols.

Most of the projects benefited from early crystallographic data and in fact make a case for getting this kind of data as early as possible, even when you have relatively weak hits (as I have found out through experience, you can get a perfectly reasonable co-crystal structure with a 10 µM hit). At the same time, crystallographic data can sometimes actually surprise and tell you where you went wrong; for instance there is an example of a tryptase inhibitor whose scaffold was redesigned and found to be favorable through modeling, only to realize from the crystal structure that the scaffold was in fact flipped through 180 degrees. That particular example illustrates that occasionally you can get the right answer through the wrong process, although knowing this fact as early as possible itself is quite useful.

Homology models pose a particular challenge for modeling; as I described in a previous post, a small change in the torsional angle of even a single residue can impact your ligand binding prediction. In this review the authors make a good case for using homology models even as crude aids. The crux of the matter is generating good hypotheses, and even crude models can help us do that. In this particular case, based on an initial lead, the model was used to predict a position to add polarity to the molecule. This led to a sulfonamide being replaced by an amide and a spiro ring system which retained potency and good properties.

The last few cases deal with ligand-based optimization in which the lack of 3D protein structural information required the use of 3D ligand overlays. The authors make another important point here: in one of their case studies they used a simple 'shape envelope' instead of detailed QSAR analysis to guide ligand design. As they point out, doing ligand overlaps for detailed QSAR analysis can be very tricky; the devil is in the molecular details and small differences in atom placement might throw you off. There have been several articles bemoaning the limitations of QSAR in recent years, so this sounds like a safe thing to do.

There are some obvious limitations to using such techniques which I am sure the authors are well aware of. Their list features hits, not misses, and many of the techniques which may have worked in these particular cases may not have worked in others. In addition, the review does not explore whether there is in fact a causal relationship between the technique used and the result obtained since other hypotheses aren't always explored. Nevertheless, it is unrealistic to expect researchers to try out every single hypotheses in a project, and what I find most useful in any case about this article is that it provides us with a checklist of things to try and conjectures to test. Science is about ideas, not answers, and as with anything else in drug discovery, if one thing fails you just hold your head high and try another.

The review concludes with a set of lessons which I think are valuable guidelines to think about in any molecular design project. The importance of the first lesson cannot be overemphasized: qualitative statements can often be more useful than quantitative analysis. I can't say this statement doesn't make me beam with pleasure. It is an antidote to those who think of chemistry as physics and expect quantitative predictions. As the case studies here demonstrate, not only are quantitative predictions often a fool's errand but in many cases they aren't even necessary. 

This is a point that I think is often lost on the critics of molecular modeling. The goal of modeling is not just to make detailed predictions; it is to cull unnecessary directions of inquiry, save labor, guide researchers into previously unexplored areas of thinking and generate hypotheses that can be quickly made and tested. It is to help think about molecular design in the broadest possible way. This is value that goes far beyond being able to rank order your latest deck of hits, and it's something to keep in mind the next time an experimentalist asks you whether you can do that.

The other lessons are also worth remembering: use molecular design to shape medicinal chemistry space, employ the principle of parsimony (and Occam's razor), annotate whenever possible (even simple visualization of molecular interactions can be eye-opening), realize the domain of applicability of your techniques (occasionally by stress-testing them) and perhaps most importantly, stay close to experiment. That last part is something all good modelers should know; don't use quantum chemical torsional calculations when you can look up features in the CSD, don't use homology models when you can convince the crystallographer to get you even a low-resolution crystal structure, don't use fancy scaffold morphing software when the medicinal chemist tells you that he or she can rapidly make alternative scaffolds. 

As the review concludes:

"Best practice in molecular design is best practice in all sciences: a relentless focus on clarity, simplicity and good experimental design. What is special about molecular design is the need to build solid hypotheses and to simultaneously foster creative thinking in medicinal chemistry. If we accept this, our focus may shift from the many semi-quantitative prediction tools we have to methods supporting this creative process. Further improvements in computational methods may then have less to do with science than with good software engineering and interface design. The tools are a just means to an end. Good science happens when they are appropriately employed."

A means to an end indeed. Modeling is a poor master but can be a very useful servant.

(Ir)rational drug design and the history of 20th century science

Here is an excellent overview of the hopes and foibles of "rational" drug design by Brooke Magnanti (Hat tip: Pete Kenny) which touches on several themes and names that would be familiar to those in the field: Ant Nicholls and OpenEye, Dave Weininger and Daylight fingerprints, Barry Werth's "The Billion Dollar Molecule" and Vertex, the inflated hopes of structure-based design, cheminformatics and screening etc. 

Those who are heroic survivors of that period would probably start with looking back with dewey eyes, followed by groans of disappointment. The bottom line in that article and several similar ones is that rational drug design and all that it entails (crystallography and molecular modeling in particular) has clearly not lived up to the hype. It's also clear that the swashbuckling scientists portrayed by Werth in his book for instance were more brilliant than successful. It's a tape of hope and woe that has played before, over and over again in fact.

It's clear that much of the faith in rational drug design until now has had a healthy component of irrational exuberance to it. Looking back at the inflated expectations of the 1980s and early 90s for designing drugs atom by atom, followed by the disappointing failures and massive attrition which rapidly succeeded these expectations, makes me wonder what it was exactly that got everyone into trouble. There was a constellation of factors of course, but the historian of science in me thinks that a major part of at least the psychological (and by extension, organizational) aspects of the issue have to deal with the stupendous successes of twentieth century science in generating a mountain of optimism which skeptics are still trying to chip away at.

It's quite clear that as far as scientific progress goes, the 20th century was the mother of all centuries. Very significant scientific advances (Newton, Maxwell, Darwin, Mendel) had undoubtedly occurred in earlier times, but the sheer rate at which science advanced in the last one hundred years far outstripped scientific progress in all previous centuries. Just consider the roster of both idea-based and tool-based scientific revolutions that we witnessed in the past century: x-rays, the atomic nucleus, relativity, quantum mechanics, nuclear fission, the laws of heredity, the structure of biomolecules, particle physics, lasers, computers, organic synthesis, gene editing...and we are just getting warmed up here.

By the 1980s this amazing collection of scientific gems had reached a crescendo, especially in the biomedical sciences. The rise of recombinant DNA technology, protein structure determination, and improved hardware, software and visualization virtually ensured that scientists started feeling very good indeed about designing drugs to block particular proteins at the molecular level. Philosophically too they were highly primed by the astounding reductionist successes of the past one hundred years. After all reductionism had uncovered the cosmic microwave background radiation from the Big Bang, given us the structure of elemental life proteins like hemoglobin and the photosynthetic complex, split the atom, doubled the number of transistors on a chip in eighteen months and taught us how to copy and paste genes. Designing drugs would be a natural extension, if not a job for graduate students, after all this success.

But what happened instead was that both scientifically and philosophically we ran into a wall. What we found out scientifically was that we still understand only a fraction of the complexity of biological systems that we need to for perturbing them with the fine scalpels of small organic molecules. Philosophically we found out that biological systems are emergent and contingent, so all the reductionist success of the past century is still not enough to understand them. In fact beyond a certain point reductionism would fundamentally put us on the wrong track. The past hundred years made us believers in Moore's Law, but what we got instead was Eroom's Law. Moore's Law is what reduces my running time from 12 mins/mile to 8:30 mins/mile in a year. Eroom's Law is what keeps it from reducing much further. Exponential technological success is not axiomatic and self-fulfilling.

I thus see a very strong influence of the success of twentieth century science in steering the wildly optimistic hopes of drug discovery scientists beginning in the 1980s. Hopefully we are wiser now, but institutional forces and biases still keep us from improving on our failures. As Pete Kenny says in his post for instance, obsession with specific technologies rather than a combined application of several technologies still biases scientists and managers in biotech and pharmaceutical organizations. The rise and ebb (did you just say "rise"?) of economic forces makes the job environment unstable and discourages scientists from pushing bold ideas that promise to break free from reductionist approaches. And much of our science is still based on sloppy theorizing without proper recourse to statistics and controls, not to mention an unbiased look at what the experiments truly are and are not telling us. 

Santayana told us that we are condemned to relive history if we forget it. But when it comes to the promises of rational drug design, what we should do perhaps is to purge our minds of the successes of the 20th century and remember Francis Bacon's exhortation from the 16th century instead: "All depends upon keeping the eye steadily fixed on the facts of nature. For God forbid that we should give out a dream of our own for a pattern of the world."

Image: "Cognition enhancer" (Source: Brooke Magnanti, Garrett Vreeland)

Identical ligands, unrelated proteins, similar energies - When language collides with the facts of nature

Recognition of aromatic rings by two very different
mechanisms but through similar binding energies

Over the years chemists have come up with many different ways to talk about the structure and energetics of molecules and especially to compare these parameters between various compounds. Doing this comparison is not just an academic exercise; for example, knowing which drug molecules are ‘similar’ or ‘different’ can be the deciding factor in picking one drug over another. It is also crucial for knowing the kinds of side effects that drugs can induce by interacting with off-target proteins.

Unfortunately the application of these simple descriptions to matters of molecular description is a very good example of what happens when language collides with fuzzy, ill-defined facts in nature. ‘Similarity’ is a classic example. When you are talking about two drugs being similar for instance, are you talking about their similarity purely in terms of molecular structure (which itself can be defined in many different ways), or their similarity in terms of their effects on cancer cells, or their similarity to engage a common protein target in the body, or through similar side effects? Clearly there are many different ways to define similarity and all these ways are subjective to a large extent.

But there is a problem with applying language to chemical concepts even at a very limited and basic level. A great example of this conundrum is hinted at by a paper from Brian Shoichet’s group at UCSF that just came out in the journal ACS Chemical Biology. The paper asks a very fundamental question: Do identical small molecules or ligands bind to very different proteins? The question in fact goes deeper: How do you define similarity and differences between various proteins to begin with?

To investigate this question, the authors consider 59 ligands bound to 119 different proteins in the PDB. Many bind with high affinity, ranging from low nanomolar to mid micromolar. What the study does is to classify these protein-ligand pairs into three groups. The first group consists of pairs in which the same atoms in identical ligands bind to similar or identical residues in different proteins. The second group consists of the same ligand atoms in identical ligands binding to similar kinds of residues (hydrophobic, positively charged etc.). The third group in a sense is the most interesting since it involves identical ligands binding to completely different proteins; in these cases the binding involves neither similar ligand atoms nor similar protein environments.

The authors find that a good two thirds of the set of protein-ligand pairs involve identical ligands binding to proteins with dissimilar residues. In addition, half of these involve ligands binding to proteins with completely different environments. There is thus no ‘pattern-matching code’ for the same ligand binding to different proteins.

Why do identical ligands bind in very different protein environments? The simple reason is because chemical binding is to a large extent a non-specific process, and there are many ways to skin the protein-ligand cat. Hydrophobic groups bump into hydrophobic groups, positively charged groups interact with negative charged ones and polar atoms snuggle up against other polar atoms. As the authors say:

"A reason why there is no simple code for ligand recognition among binding sites is that proteins have found multiple, at least superficially unrelated ways to recognize most common ligand groups. Thus, cationic amines can be recognized both by anionic residues such as aspartate or glutamate, but they can also be recognized by cation-Pi interactions. Nucleotide phosphates can be recognized by cationic residues such as arginines, but recognition by main chain amide nitrogens in a P-loop is also common. Ligand aromatic groups can stack with tyrosines, phenylalanines and tryptophans, but they can also form cation-Pi interactions many other variations might be mentioned."

But sometimes hydrophobic groups can also snuggle up against polar atoms or poke out into solvent and polar groups can nestle into hydrophobic pockets to various extents, simply because the other atoms in the ligand compensate for such uneasy alliances by forming favorable interactions. This can lead to the same ligands binding to very different protein atoms. As I mentioned in a previous post, atoms end up somewhere simply because they can. The differential placement of atoms in protein pockets is reflected in the different binding affinities that the authors see in their set.

From an evolutionary viewpoint this observation is very interesting. Protein-small molecule binding was constrained during evolution by the basic chemistry and physics of binding on one hand and by the damage incurred by too much non-specific binding on the other (as an extreme case, if every small molecule bound to every protein, there would be way too much noise and biological signaling networks would be effectively impossible). Thus there had to be a balance between promiscuity and specificity. Nature achieved this balance by tuning the affinity of small molecules for proteins over a wide range and by making sure that even weak affinity could translate to significant biological effects.

Unfortunately these are precisely the affinities that we ourselves want to finely tune in a drug discovery program and as the paper shows, this is always going to be an uphill battle because of the multitude of interactions and the lack of correlation between ligand and binding pocket structure (one conclusion from the paper is that you cannot always predict new targets for known ligands simply by computationally comparing binding sites).

But on another level I think this problem also speaks to the paucity of the language that we have for describing binding affinity and molecular interactions in general. Our metric for similarity in this case is the presence of similar ligand atoms binding to similar protein atoms. But nature can use another very simple measure of similarity – similarity in binding energy. It is not unreasonable to say that a ligand binds similarly to two proteins if it exhibits a similar binding affinity to both of them. And this binding affinity need not even be very different since even a few kcal/mol difference in binding energy can translate to a thousand fold difference in actual affinity (say from micromolar to nanomolar). Thus, what we call dissimilar binding may actually be judged as quite similar by nature. Consider the picture at the top of the post for instance: an aromatic ring can interact with a protein through either a stacking interaction with an aromatic amino acid or through a cation-pi interaction with a positively charged amino acid. The two interactions look very different, and yet they involve the same binding affinity. 

All this goes back to something we mentioned before: Similarity is in the eye of the beholder, and what our eye sees as squiggly lines of ligands and protein residues on a computer screen, nature sees simply as thermodynamics, kinetics and quantum mechanics, and all of it lying on a continuum. We might be dismayed to know that the same ligand is binding to very different proteins, but this is because nature may not be regarding them as very different to begin with. To figure out protein-ligand binding then, we may have to see things from the point of view of nature rather than that of our impoverished language.

What would be a "non-intuitive" prediction in medicinal chemistry?

Over the last two decades when computer-aided drug design was in development, one of the most common refrains you heard from medicinal chemists about its utility was that it did a poor job predicting  “non-intuitive” structural modifications to molecules. But the term is not always easy to define, and while the charge is often valid, it’s also sometimes unfair since what’s non-intuitive can be highly subjective and constitute a moving target. Also, the bar for non-intuitive ideas can be rather high based on the experience of particular medicinal chemists; it’s not always fair to hold a simple computational method up to the same standards as a chemist with thirty years experience (that’s not exactly what the software has been designed for…).

Nonetheless, it’s rather refreshing to hear modelers level the same charge against themselves, which is perhaps a sign that the entire field is now seeking higher standards than before. I was pleased to hear this sentiment noted several times in the ACS meeting in Boston which I just attended. But the question still stands: What prediction could a modeler make that would be deemed ‘non-intuitive’ or 'novel' by a fairly experienced medicinal chemist? There are at least a few cases that come to mind:

Binding and solution conformation prediction: Chemists are used to looking at 2D structures, and even a highly experienced chemist won’t be able to predict most of the time how a complicated-looking molecule will bind in a protein binding pocket. That is what docking is for, and it's one of the few areas of modeling which can claim a modest but solid degree of success. What is still a non-trivial problem is to predict the ensemble of conformations in solution which converge to a single conformation in the protein pocket. I worked on this problem myself in grad school, and it took up the majority of my half-decade or so spent there. The problem is in both determining the population of solution conformations and estimating the binding energy going from multiple to one conformation, and the general solution is still tedious and complicated.

The prediction of conformational changes in general is a problem that cannot be easily visualized by medicinal chemists without some kind of computational or experimental (especially NMR) support. Subtle structural additions like methyl groups or halogens can sometimes cause significant changes in solution conformational populations, which in turn may impact the binding conformation. Generally speaking it is impossible to understand these effects using intuition alone. Intramolecular hydrogen bonds which can stabilize conformations and improve membrane permeability are also hard to visualize or predict without some kind of computational analysis, especially for larger molecules like macrocycles.

Scaffold hopping: Another attractive idea which is not obvious to medicinal chemists. Scaffold hopping involves essentially locating the binding pharmacophore for a molecule and then finding (ideally) a completely different set of bonds and connectivities that would map on to the same pharmacophore. It is especially useful for transforming ring systems to one another or constraining an acyclic system in a ring. One utility of scaffold hopping is to locate bioisosteres. Computational techniques can be very useful here in principle, although pharmacophore detection can be spotty because of problems with false positives and negatives. Scaffold hopping is not just non-intuitive but is also a boon to getting around intellectual property which is usually every chemist’s nemesis.

Calculating desolvation penalties: An experienced medicinal chemist may be able to look at a compound and make a guess about its size or lipophilicity, but guessing desolvation penalties is intuitively quite hard except in obvious cases (as in the case of a positively or negatively charged group – even then, guessing the sum of all the interactions is challenging). One of the reasons is that desolvation being a point charge-dipole interaction, its energy goes up as the square of the charge (instead of just inversely as in Coulomb's law): small changes in heteroatom distributions can thus have significant and non-obvious effects on solvation/desolvation. 

Unfortunately calculating solvation energies is also still hard in a general sense for computational chemists, but progress continues to be made. One of the most successful predictions of modeling would be a case where a highly charged group compensates for all its desolvation by making perfectly formed hydrogen bonds with a protein - this is a very hard thing to predict as of now. Generally speaking though, desolvation penalties would, at least in principle, fall into the category of things that medicinal chemists wouldn’t often be able to guess.

Calculating strain energies: This is another case where even experienced medicinal chemists may not be able to make intuitive statements. Sometimes it’s obvious in an x-ray crystal structure that ligands are strained (manifested for instance in the form of bent amides, bent phenyl rings or any kind of non-planar conjugated systems). But other times the effects of strain can be invisible to the naked eye. The problem is that bond length changes of as little as tenths of an angstrom can translate into significant strain energies of several kcals/mol, and it is hard if not impossible for even seasoned medicinal chemists to actually see these strain-inducing elements without some kind of calculation. That’s where modeling can help.

Water molecules: We know well by now how complicated the behavior of water molecules in protein binding sites can be. Sometimes the kinds of predictions that modelers make about easily and productively displaced water molecules are rather obvious, such as when they are talking about a water molecule in a nice hydrophobic cavity. The subtle cases are harder to intuitively predict. For instance crystallographic waters may be firmly bound and therefore may have good enthalpy but may still be unhappy and displaceable because of an unfavorable entropy. Similarly as detailed in the link above, water molecules at ligand-solvent interfaces may have unexpected thermodynamic features. Unhappy water prediction methods like WaterMap and SZMAP are promising, but only when they can predict non-intuitive scenarios that are refractory to easy analysis by medicinal chemists.

Data analysis: Generally speaking the word ‘non-intuitive’ may also mask more mundane but useful goals like being able to analyze large amounts of data and suggest useful trends. In fact that’s a task that’s usually quite unsuited to the skills of a medicinal chemist because of its reliance on numbers and statistics and opacity to easy structural visualization.

Feel free to note others in the comments section which I might have left out. Even better are cases where modelers can make suggestions that aren't just non-intuitive but counterintuitive. For instance, if you can predict that a methyl group filling a pocket would lead to a drop in potency (steric reasons? trapped water?) that would be a good counterintuitive prediction. Or if you could predict that cyclization of a molecule would actually increase conformational flexibility because of alleviation of syn-pentane interactions (as I found out in my comparison of cyclic dictyostatin with acyclic discodermolide), that prediction would also fall into the same category. Counterintuitive predictions also provide acid tests of any model because of their emphasis on falsifiability.

I don’t claim that all the goals listed above are well within the purview of molecular modeling. What I am claiming is that there are several challenging tasks on which modeling has started to make inroads. And a good number of these could be called “non-intuitive” or even "counterintuitive". A decade or two down the line I don't think such predictions from modeling will be as rare as we currently think they are, and that's something that we all should look forward to.

The problem with molecular modeling is not just molecular modeling

I am attending the Gordon Conference on Computer-Aided Drug Design (CADD) in the verdant mountains of Vermont this week, and while conference rules prohibit me from divulging the details of the talks, even the first day of the meeting reinforces a feeling that I have had for a while about the field of molecular modeling: the problems that plague the field cannot be solved by modelers alone.

This realization is probably apparent to anyone who has been working the field for a while, but its ramifications have become really clear in the last decade or so. It should be obvious by now to many that while modeling has seen some real and solid progress in the last few years, the general gap between promise and deliverables is still quite big. The good news is that modeling has been integrated into the drug discovery process in many small and sundry ways, ranging from getting rid of duplicates and "rogue" molecules in chemical libraries to quick similarity searching of new proposed compounds against existing databases to refinement of x-ray crystal structures. These are all very useful and noteworthy advances, but they don't by themselves promise a game changing impact of modeling on the field of drug discovery and development.

The reasons why this won't happen have thankfully been reiterated several times in several publications over the last fifteen odd years, to the extent that most reasonable people in the field don't get defensive anymore when they are pointed out. There's the almost complete lack of statistics that plagued the literature, leading people to believe that specific algorithms were better than what they actually were and continuing to apply them (this aspect was well emphasized by the last GRC). There's the constant drumbeats about how badly we treat things like water molecules, entropy, protein flexibility and conformational flexibility of ligands. There are the organizational issues concerning the interactions between modelers and other kinds of scientists which in my opinion people don't formally talk about with anywhere near the level of seriousness and the frequency which they deserve (although we are perfectly happy to discuss them in person).

All these are eminently legitimate reasons whose ills must be exorcised if we are to turn modeling into not just a useful but consequential and even paradigm-shifting part of the drug discovery process. And yet there is one other aspect that we should be constantly talking about that really puts a ceiling on top of even the most expert modeler. And this is the crucial reliance on data obtained from other fields. Because this is a ceiling erected by other fields it's not just something that even the best modelers alone can punch through. And breaking this ceiling is really going to need both scientific and organizational changes in the ways that modelers do their daily work, interact with people from other disciplines and even organize conferences.

The problem is simply of not having the right kind of data. It's not a question of 'Big Data' but of data at the right level of relevance to a particular kind of modeling. One illusion that I have felt gradually creeping up the spines of people in modeling-related conferences is that of somehow being awash in data. Too often we are left with the feeling that the problem is not that of enough data, it's only of tools to interpret that sea of information.

The problem of tools is certainly an important one, but the data problem has certainly not been resolved. To understand this, let's divide the kind of data that is crucial for 'lower level' or basic modeling into three categories: structural, thermodynamic and kinetic. It should be obvious to anyone in the field that we have made amazing progress in the form of the PDB as far as structural information is concerned. It did take us some time to realize that PDB structures are not sacrosanct, but what I want to emphasize is that when serious structure-based modeling like docking, homology modeling and molecular dynamics really took off, the structural data was already there, either in the PDB or readily obtained in house. Today the PDB boasts more than a hundred thousand structures. Meticulous tabulation and analysis of these structures has resulted in high-quality datasets like Iridium. In addition there is no dearth of publications pointing out the care which must be exercised in using these structures for actual drug design. Finally, with the recent explosion of crystallographic advances in the field of membrane protein structure, data is now available for virtually every important family of pharmaceutically relevant proteins.

Now consider where the field might have been in a hypothetical universe where the PDB was just getting off of the ground in the year 2015. Docking, homology modeling, protein refinement and molecular dynamics would all have been in the inky backwaters of the modeling landscape. None of these methods could have been validated in the absence of good protein structure and we would have had scant understanding of water molecules, protein flexibility and protein-protein interactions. The Gordon Conference on CADD would likely still be the Gordon Conference on QSAR.

Apply the same kind of thinking to the other two categories of data - thermodynamic and kinetic - and I think we can see some of the crucial problems holding the field back. Unlike the PDB there is simply no comparable database of tens of thousands of reliable thermodynamic data points that would aid the validation of methods like Free Energy Perturbation (FEP). There is some data to be found in repositories like PDBbind, but this is still a pale shadow of the quantity and (curated) quality of structures in the PDB. No wonder that our understanding of energies - relative to structure - is so poor. When it comes to kinetics the situation is much, much worse. In the absence of kinetic data, how can we start to truly model the long residence times in protein-ligand interactions that so many people are talking about these days? The same situation also applies to what we can call 'higher order' data concerning toxicology, network effects on secondary targets in pathways and so on.

The situation is reminiscent of the history of the development of quantum mechanics. When quantum mechanics was formulated in the twenties, it was made possible only by the existence of a large body of spectroscopic data that had been gathered since the late 1870s. If that data had not existed in the 1920s, even wunderkinder like Werner Heisenberg and Paul Dirac would not have been able to revolutionize our understanding of the physical world. Atomic physics in the 1920s was thus data-rich and theory poor. Modeling in 2015 is not exactly theory-rich to begin with, but I would say it's distinctly data-poor. That's a pretty bad situation to be in.

The reality is very simple in my view: unless somebody else - not modelers - generates the thermodynamic, kinetic and higher-order data critical to advancing modeling techniques the field will not advance. This problem is not going to be solved by a bunch of even genius modelers brainstorming for days in a locked room. Just like the current status of modeling would have been impossible to imagine without the contributions of crystallographers, the future status of modeling would be impossible to imagine without the contribution of biophysical chemists and biologists. Modelers alone simply cannot punch through that ceiling.

One of the reasons I note this problem is because even now, I see very few (none?) meetings which serve as common platforms for biophysical chemists, biologists and modelers to come together and talk not just about problems in modeling but how people from these other fields can address the problem. But as long as modelers think of Big Data as some kind of ocean of truth simply waiting to spill out its secrets in the presence of the right tools, the field will not advance. They need to constantly realize the crucial interfacing with other disciplines that is an absolute must for progress in their own field. What would make their own field advance would be its practitioners knocking on the doors of their fellow kineticists, thermodynamicists and network biologists to get them the data that they need.

That last problem suddenly catapults the whole challenge to a new level of complexity and urgency, since convincing other kinds of scientists to do the experiments and procure the data that would allow your field to advance is a daunting cultural challenge, not a scientific one. Crystallographers were busy solving pharmaceutically relevant protein structures long before there were modelers, and most of them were doing it based on pure curiosity. But it took them fifty years to generate the kind of data that modelers could realistically use. We don't have the luxury of waiting for fifty years to get the same kind of data from biophysical chemists, so how do we incentivize them to speed up the process?

There are no easy ways to address this challenge, but a start would be to recognize its looming existence. And to invite more scientists from other fields to the next Gordon Conference in CADD. How to get people from other fields to contribute to your own in a mutually beneficial relationship is a research problem in its own right that deserves separate space at a conference like this. And there is every reason to fill that space if we want our field to rapidly progress.

Want to bind small molecules? Get a backbone

Here’s a paper from the Shoichet lab at UCSF that illustrates one of the major problems that drug designers encounter – predicting conformational changes (“entropy” to a physicist). What the study does is to plug a series of eight very simple congeneric ligands – benzene, methyl, ethyl and propyl benzene all the way to hexyl benzene - into a model protein cavity, in this case a lysozyme mutant, and observe the corresponding changes in protein conformation by solving the crystal structures. And the results aren’t exactly heartwarming for early phase drug discovery scientists.

Synthesizing congeneric series of ligands is a standard process in lead optimization and the elephant in the room which is often banished out of sight by drug designers is the possibility of large conformational changes in the protein caused by small changes in ligand structure (the other assumption is constancy in ligand binding orientation, and even that doesn’t always hold). The assumption is that any minor change in structure in the ligand would be accommodated by equivalent, small amino acid side chain movements in the protein.

This study shows that at best that assumption is a faith-based assumption which should always be considered provisional. What the authors observe is that instead of a smooth transition of amino acid side chain movements, you see a discrete and far more significant change in protein backbone movement, resulting in a subtle population of different states which bind the ligands. The difference in binding energy going from benzene to hexyl benzene is not too large – about 1.5 kcal/mol – but you are already seeing backbone movements. What is perhaps a bit more reassuring than this observation is that some of the discrete states are mirrored in lysozyme structures found in the PDB - but the authors looked at 121 structures to substantiate the result. Not the kind of numbers you would expect to find in the PDB for your typical novel drug discovery target.

The conclusions of the paper are a bit discomforting for at least two reasons. Firstly as mentioned above, drug designers often assume constancy or smooth and minor side chain changes in protein conformation when testing congeneric ligands in lead optimization. It’s quite clear that this is always a bit of a gamble: if something as simple as a change in molecular weight could lead to such divergent changes, what would small but important changes or reversals in polarity do? And then one also starts wondering how much weird or divergent SAR could potentially be explained by such unexpected backbone conformational changes.

Secondly, these kinds of changes pose a real problem for molecular modelers. As the paper says, you would need to go to pretty long MD (molecular dynamics) simulations or more radical protein modeling to look at backbone changes; even today, modeling backbone changes by either physics-based methods (like MD) or knowledge-based techniques (like Rosetta) is both less validated and more computationally expensive.

Lastly though, this study is another example of why drug discovery is hard even at a basic scientific level. Countless factors thwart the best intentions of drug designers at every stage, and uncertainty in predicting protein backbone conformational changes must rank pretty high on that list.

Thermodynamics in drug discovery: A Faustian bargain cooked in a devil's stew?

The binding of analogs of the anticoagulant melagatran
to thrombin demonstrates intricate water network and
protein conformation differences masked by similar binding modes

Driving drug design by studying the thermodynamics of ligands binding to proteins has always seemed like a good idea whose time has come. It all sounds very attractive: at its heart every molecular recognition event is driven by thermodynamic and kinetic principles, so in principle one would be able to figure out everything they wanted to know about the details of such interactions by accurately calculating or measuring the relevant parameters. Surely the main hurdles are experimental? The truth though is that we now have good experimental techniques to investigate thermodynamics, and yet nobody still seems to have figured out what’s the best way to apply the idea prospectively to new drug discovery projects. 

However there has been an increasing awareness of the breakdown of the free energy of binding into enthalpy and entropy: part of this awareness has been driven by analyses of drugs indicating that the best drugs have their enthalpies of binding rather than their entropies optimized. The conventional wisdom gleaned from these and other studies seems to be that optimizing enthalpy is tantamount to optimizing protein-ligand interactions while optimizing entropy is tantamount to displacing water molecules and building in hydrophobic modifications. This seems to indicate that one must try to optimize enthalpy through specific interactions early on in the drug development process, no matter that doing this is usually very challenging since you can often end up simply trading one hydrogen bond for another, leading to a net zero impact on binding affinity.

Now there’s a new, very comprehensive and readable review dealing with these matters out in J. Med. Chem. which demonstrates just what kind of a devil’s stew this matter actually is and which asks whether thermodynamics is still a 'hot tip' in drug discovery. The authors who are from Astra Zeneca in Sweden look at a variety of topics related to protein-ligand thermodynamics – the gory details of ITC which is the experimental workhorse used to determine the thermodynamic quantities, case studies showing that displacing water can sometimes help and sometimes hurt, the convoluted phenomenon of enthalpy-entropy compensation, the whole fundamental idea that water molecules are all about entropy and interactions with the protein are all about enthalpy.

They reach the conclusion that a lot of the conventional wisdom is, if not exactly incorrect, far too simplistic and often misleading. As always, reality is more complex and subtle than three round numbers. They find cases where displacing water improves the free energy of binding, not through entropy as one might expect but by strengthening existing interactions or effecting new ones, that is through enthalpy. There are also cases where displacing water makes things worse because the ligand is not able to pay the desolvation penalty imposed on its polar groups. There have already been reports looking at networks of water molecules not just in the protein active site but also around the ligand, and the subtle movements of these networks only serve to complicate any kind of water-based analysis. And as the authors demonstrate, simple observation of SAR can be very misleading when applied to conclusions regarding displacement of water molecules or formation of specific interactions: for instance, even ligands that have the same binding mode can showcase differing water networks and protein conformations.

The conclusion the paper reaches is not exactly heart-warming, although it points to some future directions.

“So can ligand binding thermodynamics still be regarded as a hot tip in drug discovery? No, not in a routine setting or with enthalpy and entropy regarded as isolated endpoints. Experimentally obtained thermodynamic data and crude derived parameters thereof are simply not well suited to be used for direct red/green decision-making. It is not unlikely, that some of the underlying parameters of the measured enthalpy and entropy might correlate with other interesting and relevant compound parameters. However, no such correlation has been convincingly shown thus far…Comparison of experimental ITC data with e.g. LLE (lipophilic ligand efficiency), simple solvent calculations or more rigorous free energy perturbations can enable the identification of compounds that do not behave as expected. Identifying and scrutinizing those outliers appears currently to be the most impactful use of thermodynamic profiling. The outliers could help to identify compound series that shift their binding mode, induce different motions in the target protein or distinguish intra-molecular hydrogen bonds from those between protein and ligand.”

The main question that the authors try to answer here is whether the measurement of free energy, enthalpy and entropy can prospectively help drug design, and their answer is largely negative. The fundamental reason is that all these quantities are composite effects so they mask individual contributions from protein, ligand and water. The contribution of a particular hydrogen bond to affinity is not an experimental observable, and trying to over-interpret thermodynamic data in order to divine such contributions may easily lead you down the rabbit hole. There is a multiplicity of such contributions that can result in the same number, so the problem is really underdetermined to a large extent. As the review indicates, all that prospective measurement of thermodynamic quantities can do is point to obvious outliers that might be causing very large protein conformational changes or leading to radically different ligand conformations. Although I would think that an eagle-eyed medicinal chemist armed with some structural expertise and robust SAR data might be able to reach the same conclusions.

Thermodynamics has always been one of those beloved children of drug discovery, one on whom the parents have pinned their high hopes but who still has to turn that potential into real achievement. As this review demonstrates, there is much complexity hidden in the heart of this prodigal child, and until one unravels this complexity his beatific smile will remain a cloudy crystal ball.

Reference: 

Ligand Binding Thermodynamics in Drug Discovery: still a hot tip?

Stefan Geschwindner , Johan Ulander , and Patrik Johansson

J. Med. Chem., Just Accepted Manuscript

DOI: 10.1021/jm501511f