High-throughput method for detecting intramolecular hydrogen bonds (IHBs)

We have talked about the importance of intramolecular hydrogen bonds (IHBs) as motifs to hide polar surface area and improve the permeability of molecules a couple of times here. The problem is that there's no high-throughput method for detecting these in, say, a library of a thousand molecules. The best method I know is temperature and solvent dependent NMR spectroscopy but that's pretty time consuming. Computational methods can certainly be useful but they can sometimes lead to false positives by overemphasizing hydrogen bonding between groups in proximity.

Now here's a promising new method from Pfizer which could lend itself to high-throughput screening of IHBs. The authors essentially use supercritical fluid chromatography (SFC) to study retention times of molecules with and without IHBs. The solvent system consists of supercritical CO2 spiked with some methanol. They pick a judicious set of matched molecular pairs, each one of which contains a hydrogen bonded and non-hydrogen bonded version. They then look at retention times and find - not surprisingly - that the hydrogen-bonded versions which can hide their polar surface area have lower retention times on the column. 

They corroborate these findings with some detailed NMR studies looking at solvent and temperature dependent chemical shifts. At the beginning, when they plot the retention times vs the total polar surface area (TPSA) they get a nice correlation for non IHB compounds. For compounds with IHBs however, the TPSA is an imperfect predictor; what they find in that case is that a new parameter called EPSA based on the retention time is a better predictor of IHBs compared to TPSA.

This seems to me to be a potentially valuable method for quickly looking at IHBs in relatively large numbers of compounds, especially since we have now started worrying about how to get "large" molecules like macrocycles and peptides across membranes. And I assume every medium or large company will probably have access to SFC, so the technology itself should not pose a measurable barrier. One more handy tool to look at "beyond rule of five" compounds.

Reference: 

A High Throughput Method for the Indirect Detection of Intramolecular Hydrogen Bonding

Gilles H Goetz, William Farrell, Marina Shalaeva, Simone Sciabola, Dennis P Anderson, Jiangli Yan, Laurence Philippe, and Michael J. Shapiro

J. Med. Chem., Just Accepted

Publication Date (Web): March 18, 2014 (Article)

DOI: 10.1021/jm401859b

Ivano Bertini

Bertini with Harry Gray at Caltech (Image: NSMB)

I thought I should take note of the unfortunate fact that Ivano Bertini has passed away. I first heard of him when I came across the famous textbook on bioinorganic chemistry which he co-authored with Gray, Stiefel and Valentine. I think it's still the best introduction to the subject. After laying out the basic properties and abundances of inorganic ions in biology and the environment, it goes on to describe in careful detail the role of major metalloproteins in key biological processes. Each of these proteins is depicted as an elegant molecular machine performing metal-catalyzed room-temperature reactions with an efficiency that we chemists can only dream of accomplishing.

Bertini made very important contributions to the NMR structure elucidation of metalloproteins and solved more than 150 structures during a distinguished career that was cut short, an unprecedented record. Before his work, NMR of paramagnetic proteins was thought to be exceedingly difficult because of paramagnetic relaxation; this is the same reason why oxygen in an NMR sample precludes the acquisition of good data for NOE experiments, requiring the sample to be purged with nitrogen. 

An obituary in Nature Chemical Biology (paywall) gives a good sense of both the man and his characteristically bold approach to science:

One could not avoid knowing Ivano: he was 'loud' in all senses. In a room full of people, his booming voice would always tell if he was around. He was also tall and large and would speak loudly to the heart of any new acquaintance, making himself unforgettable. He also had a loud love for science. In his office, he had a banner that said, “La scienza è come l'amore: non puoi non pensarci sempre” (Science is like love: you can't help thinking about it all the time)...

The story of the first solution structure of a paramagnetic protein is a typical example of Ivano's response to scientific challenges. In the early nineties, about ten years after the first protein NMR structure, it was implicit that paramagnetic relaxation prevented NMR analysis of paramagnetic proteins. 

On a midsummer Sunday at Ivano's country house, however, we were reading a recent review article by a well-known NMR spectroscopist that explicitly stated it would never be possible to solve NMR structures of paramagnetic proteins. Ivano said, “Do you believe it?” We said, “No.” He then said, “This is a project that will need the whole lab.” The next day, we were all at work; the paper was published 14 months later. With that work, a taboo had been broken, and many structures of paramagnetic proteins have been solved since then.

Metalloproteins continue to be of intense interest in chemistry, biology and medicine and we will all continue to benefit from Bertini's legacy.

'SAR by C13 NMR'

The biggest utility of NMR spectroscopy in drug discovery is in assessing three things; whether a particular ligand binds to a protein, what site on the protein it binds, and what parts of the ligand interact with the protein. Over the last few years a powerful technique named ‘SAR by NMR’ has emerged which is now widely used in ligand screening. In this technique, changes in the resonances of ligand and protein protons are observed to pinpoint the ligand binding site and corresponding residues. Generally when a ligand binds to a protein, both its and the protein’s rotational correlation time decreases; the result is a broadening of signals in the spectrum which can be used to detect ligand binding. One of the most effective methods in this general area is Saturation Transfer Difference (STD) spectroscopy. As the name indicates, it hinges on the transfer of magnetization between protein and ligand; the resulting decrease in intensity of ligand signals can provide valuable information about proximity of ligand protons with specific protein residues.

But these kinds of techniques suffer from some drawbacks. One straightforward drawback is that signals from protein and ligand may simply overlap. Secondly, the broadening may be so much as to virtually make the signals disappear. Thirdly from a practical perspective, it is hard to get sufficient amounts of N15-labeled protein (usually obtained by growing bacteria on a N15-rich source and then purifying the proteins of interest).

To circumvent some of these problems, a team at Abbott Laboratories has come up with a neat and relatively simple method which they call ‘labeled ligand displacement’. The method involves synthesizing a protein-binding probe that has been selectively labeled with C13. Protein binding broadens and diminishes the signals of this probe. However, when a high-affinity ligand is then added, it displaces the probe and we get recovery of the C13 signals. The authors illustrate this paradigm with several proteins of pharmaceutical interest, including heat-shock protein and carbonic anhydrase.

The method is relatively simple. For one thing, using a commercially available C13-labeled building block for synthesizing a ligand is easier than obtaining a N15-labeled protein. The biggest merit of the method though is the fact that it hinges on C13 signals very specific to the probe; thus there is no complicating overlap of signals. And finally, the ligand seems to be general enough to be applied to any protein. Only time will tell how much it is utilized, but for now it seems like a neat addition to the arsenal of NMR methods for studying protein-ligand interactions.

Swann, S., Song, D., Sun, C., Hajduk, P., & Petros, A. (2010). Labeled Ligand Displacement: Extending NMR-Based Screening of Protein Targets ACS Medicinal Chemistry Letters, 1 (6), 295-299 DOI: 10.1021/ml1000849

Conformations of the stevastelins: A reassessment

Shameless self-promotion: my paper on the conformational analysis of cyclic antiviral peptides called stevastelins is now online on the Biopolymers site. Here's a brief overview.

The stevastelins are cyclic peptides that show promising antiviral activity against the vaccinia viral VHR phosphatase. These peptides are phosphorylated in vivo before they can inhibit their target protein. A group in Germany previously did a meticulous analysis of four diastereometric analogs of these peptides which included their synthesis, biological characterization and conformational analysis. However, the conformational analysis was done using force field conformational searches from a single force field, constrained by variables from the NMR data (coupling constant derived dihedral angles and NOESY derived distances). Using such a protocol, the group concluded that each of the four diastereomers exists as a single conformational family in solution.

The problem with constrained conformational searches (or constrained molecular dynamics for that matter) is that they constitute a rather self-fulfilling exercise, with the assumption that there is in fact a single conformation of the molecule under question. However, as I have often discussed on this blog, any molecule with a couple of rotatable bonds is going to exist as multiple conformers in solution, so an assumption of a single conformation would be fuzzy unless supported by more data. NMR by itself is of scant value in determining these conformations for thermodynamic and kinetic reasons. Plus, analyzing conformations using a single force field can be fraught with ambiguity, since every force field comes with its own set of parameters and convergence criteria. Especially trusting energies from force fields can be dangerous. In this case, the stevastelin peptides have 9 rotatable bonds each, so I thought it worthwhile to apply our previously developed and applied NAMFIS (NMR Analysis of Molecular Flexibility In Solution) methodology combining NMR variables with structures from extensive conformational searches to the enumeration of the conformational behavior of these interesting molecules.

The paper essentially describes the conformational variability obtained for each of the diastereomers. Many of the conformations are very similar to the previously postulated families, but some are quite different. There are also some striking observations that are corroborated; for instance, the use of a d-serine truly seems to 'lock' the peptides in a single conformation. Such a lock could be effected to counter the entropic penalty that a multiconformational ensemble of molecules might have to pay. The instructive general observation is that subtle changes in sterechemistry at one or two chiral centers can dramatically affect conformational behavior, a fact that continues to surprise and confound medicinal chemists. I also note that if the NMR data for the phosphorylated peptides were available, an interesting comparison of the conformational pool for the phosphorylated and unphosphorylated counterparts could be attempted. This would shed light on whether phosphorylation leads to less conformational variability or simply increases the proportion of a chosen subset of conformations of the peptides.

Comments, criticism and general feelings of chagrin are welcomed.

Jogalekar, A. (2010). Conformations of stevastelin C3 analogs: Computational deconvolution of NMR data reveals conformational heterogeneity and novel motifs Biopolymers DOI: 10.1002/bip.21504

Book review: The Dale Carnegie of Silicon-29 NMR

Chemical Shifts and Coupling Constants for Silicon-29 (Landolt-Börnstein: Numerical Data and Functional Relationships in Science and Technology - New Series / Condensed Matter)

By Radha Raman Gupta (Contributor), Manfred Dieter Lechner (Contributor), Heinrich Marsmann (Contributor), Bozhana Mikhova (Contributor), Frank Uhlig (Contributor)

If you read this book, you cannot help but be instantly transformed into a silicon-29 NMR lover and specialist, purveyor of a wonderful, deep and satisfying field of inquiry. This book whose publication was a watershed did for silicon-29 NMR what Dale Carnegie's book did for the average shy person and what Ayn Rand did for capitalism. It finally brought silicon-29 NMR lovers out of the shadows of their self-imposed silence into the mainstream community. As for the price, I can say only one thing; this is one of the very very few products, whether in the shoe store, car store or pet store, whose price is completely justified by its quality. It's a clarion call to all silicon-29 NMR specialists who can finally reclaim the dream that the original topaz and mica lovers only dared to contemplate. I am pretty sure that even after several decades it will remain the most referenced and readily available book on the market, and will continue to be a godsend for those hesitant men and women who are uncertain about their future in the silicon-29 NMR field. Three words; buy it now.

Much ado about protein dynamics

Let me alert you, in case you haven't noticed, to the latest issue of Science which is a special issue on protein dynamics.

There is much of merit here, but this article is especially relevant to drug discovery. It talks about the interaction of small molecules and how it reshapes the energy landscape of protein conformational motion. One of the most useful ways of thinking about small molecule-protein interactions is to visualize a protein that fluctuates between several conformational states which are in equilibrium. A small molecule can inhibit the protein by preferentially stabilizing one of these states.

The article illustrates this concept with several examples, most notably inhibition of kinases. Many kinases exist in an inactive and active state and kinase inhibitors stabilize and block one of these states. Such conformational trapping can also reduce mobility of the protein. The article also describes how certain kinase inhibitors such as imatinib and nilotinib trap the kinase in the inactive state while others such as dasatinib trap it in an active state. Although all three of these are classified as ATP-competitive inhibitors, dasatinib blocks the Abl-Bcr kinase by effecting an allosteric movement of a particular loop. Allosteric inhibitors of kinases are of value since they won't target the highly conserved ATP-binding site, thus reducing problems with selectivity. But allosteric targeting is difficult since many times it involves targeting shallow, poorly defined sites including those involved in protein-protein interactions. HTS campaigns aimed at disrupting P-P interactions usually give very poor results. However, recent tools and especially NMR with labeled residues may improve the detection of weakly binding molecules that may be missed in assays (where the limit is usually 30 µM). HSQC spectra are generally taken of the protein, with and without the inhibitor, and changes in residue resonances can give an indication of conformational changes.

In any case, this article and the others are worth reading. Basically it seems that the remodeling of energy landscapes of proteins by either small molecules or other signals is a concept acquiring central traction. Such a concept could essentially tie together the dual problems of protein folding and inhibiting proteins with small molecules.

Reference:
Lee, G., & Craik, C. (2009). Trapping Moving Targets with Small Molecules Science, 324 (5924), 213-215 DOI: 10.1126/science.1169378

So salt bridges are not stable in water? Shocking

Three salt bridges seen in this protein in the xtal structure were not observed by detailed NMR experiments in water. Here's the abstract:

NMR investigations have been carried out on the B1 domain of protein G. This protein has six lysine residues, of which three are consistently found to form surface-exposed salt bridges in crystal structures, while the other three are not. The Nζ and Hζ chemical shifts of all six lysines are similar and are not affected significantly by pH titration of the carboxylate groups in the protein, except for a relatively small titration of K39 Nζ. Deuterium isotope effects on nitrogen and proton are of the size expected for a simple hydrated amine (a result supported by density functional theory calculations), and also do not titrate with the carboxylates. The line shapes of the J-coupled 15N signals suggest rapid internal reorientation of all NH3+ groups. pKa values have been measured for all charged side chains except Glu50 and do not show the perturbations expected for salt bridge formation, except that E35 has a Hill coefficient of 0.84. The main differential effect seen is that the lysines that are involved in salt bridges in the crystal display faster exchange of the amine protons with the solvent, an effect attributed to general base catalysis by the carboxylates. This explanation is supported by varying buffer composition, which demonstrates reduced electrostatic shielding at low concentration. In conclusion, the study demonstrates that the six surface-exposed lysines in protein G are not involved in significant salt bridge interactions, even though such interactions are found consistently in crystal structures. However, the intrahelical E35−K39 (i,i+4) interaction is partially present.

The title was meant in half-jest of course and I don't mean to disparage such studies. But I think it just goes to show the kind of difficult, tedious and careful work that has to be often carried out in science even to reach "obvious" conclusions.

An an aside though, this conclusion was not at all obvious for a fair amount of time. There was a vigorous debate in the 90s kicked off by Bruce Tidor's paper arguing that salt bridges are not really that energetically important in protein stabilization, especially on surfaces. People who believed in the intense power of the holy electrostatic attraction did not really believe this. While the debate still continues, to my knowledge the general consensus is now on the side of the original Tidor proposition; salt bridges mostly provide only a marginal energetic gain (1-2 kcal/mol) to protein stability. This has been shown to be so primarily because of the loss in solvation and especially long-range solvation that formation of a salt-bridge incurs. Well, let the "obvious" research continue.

References:
1. Tomlinson, J., Ullah, S., Hansen, P., & Williamson, M. (2009). Characterization of Salt Bridges to Lysines in the Protein G B1 Domain Journal of the American Chemical Society DOI: 10.1021/ja808223p

2. Z.S. Hendsch and B. Tidor. Do salt bridges stabilize proteins? A continuum electrostatic analysis. Protein Sci. 3: 211-226 (1994)

Hexacyclinol: Case Closed

Seems that hexacyclinol is finally dead and buried and the tomb has been sealed. If anyone had any doubt that there was something wrong with the original structure, this paper should resolve it. For those who have had the good fortune of not hearing about hexacyclinol, see the links at the end.

In any case, the present authors from Italy simply calculate H1 chemical shifts and coupling constants for the original disputed structure (1) and the newly proposed and calculated structure (2). Recall that one of the things James La Clair claimed when John Porco synthesized what Scott Rychnovsky felt (Yes, it does sound like one of those convoluted Shakespearean chains of people) was the correct structure was that the two different structures could possibly give rise to the same or extremely similar NMR spectra.

Yes, it's probably true that in the infinitely large universe of organic compounds one could possibly locate two compounds which could give the same NMR spectra. In real life, such a possibility for two complex compounds that look so different is very unlikely. However, the assertion that this could be so is a testable and falsifiable hypothesis and therefore a good one.

In the present paper, the Italian group finally tackles the problem using a set of high-level calculated NMR spectra that include H1 chemical shifts, coupling constants, and even COSY plots at the DFT level. The answers are unambiguous and clear; Porco and Rychnovsky's hexacyclinol gives much smaller errors between measured and calculated parameters compared to La Clair's "hexacyclinol". At the same time, there is some striking similarity between many of the NMR parameters. In the end though, a competent organic chemist would be able to tell that the spectra belong to different structures. As the authors put it rather clearly,

The structure of hexacyclinol is confirmed to be (2). Furthermore, if (1) had been synthesized or was formed from an unforeseen reaction, its NMR spectra are sufficiently different from those of (2) as to guarantee their distinction.

Short of a miracle that the famed Bionic Brothers could spring now, I don't see what Dr. La Clair can do to validate his original proposal and synthesis. This should also again be a question for Angewandte Chemie who published the paper; I am sure all of us are curious to know what happened there.

I have talked on this blog several times before (see below) about computationally calculated NMR parameters coming of age, and this paper should reinforce their tremendous power in resolving confusion between organic structures. H1 chemical shifts can now be predicted to within a remarkable 0.1 ppm accuracy. Hopefully these studies should now encourage the widespread use of the relevant tools in organic chemistry.

Previous posts on hexacyclinol and computational prediction of NMR data: here, here, here, here and here

Update: Also discussed at In The Pipeline and The Chemistry Blog. As mentioned in the comments, I too find it remarkable that Angew Chem has not published some kind of explanatory note by now, after all the blog and news publicity that the issue got.


Reference:
Giacomo Saielli, Alessandro Bagno (2009). Can Two Molecules Have the Same NMR Spectrum? Hexacyclinol Revisited Organic Letters DOI: 10.1021/ol900164a

Samoquasine is perlolidine after all

Here is another nice example of the application of C13 chemical shifts calculated using quantum chemistry to solve a dispute concerning the structure of a natural product. This one is from Peter Wipf's group at Pittsburgh who use C13 shifts calculated at the B3LYP/6-31+G(2d,p) level to ascertain that an alkaloid named samoquasine is identical to another one named perlolidine.

Samoquasine was isolated in 2000. At the time the original authors thought that they had isolated a new alkaloid and gave it a new name. Then in 2002 they proclaimed that the "new" substance was actually the same as an earlier identified substance called perlolidine. Then along came some other chemists in 2003 who again said that in fact samoquasine is different from perlolidine. Time to put an end to the squabble!

To resolve the dispute, Wipf's group constructed 48 isomers that could correspond to all combinations of samoquasine/perlolidine. They calculated chemical shifts for all 48 isomers and found out that only the chemical shifts of one particular isomer (perlolidine!) showed the least deviation with measured C13 chemical shift for samoquasine/perlolidine. This indicated that the original authors' assertion, that samoquasine was identical to perlolidine was in fact correct. Not only were the latter authors wrong, but Wipf also castigates their melting point and their lack of measurement of UV spectra that could have helped settle the dispute back then. In short, the previous study was not exactly a nice piece of experimental work. Ouch.

Carl Djerassi once said that the great value of computers in structure determination is to generate several straw men and then identify which ones are in fact straw men. This study serves to drive that point home and again shows the value of computational prediction of NMR chemical shifts in organic chemistry. However, I also wonder how much the results would have changed had the authors used the MPW1PW91 functional which has been found to be optimum for such calculations. Also recall that this technique was demonstrated in a famous incident a few years ago.

Reference:
Cody Timmons, Peter Wipf (2008). Density Functional Theory Calculation of
C13 NMR Shifts of Diazaphenanthrene Alkaloids: Reinvestigation of the Structure of Samoquasine A
The Journal of Organic Chemistry, 73 (22), 9168-9170 DOI: 10.1021/jo801735e

∆G, ∆G†† and All That: Implications for NMR

Since we were on the subject of NMR and determining conformations, I think it would be pertinent to briefly discuss one of the more slippery basic concepts that I have seen a lot of chemistry students (naturally including myself) get plagued with; the difference between thermodynamics and kinetics. I find myself often besieged by a distinction between these two important ideas that encompass all of chemistry. Simply saying that thermodynamics is "where you go" and kinetics is "how you get there" is not enough of a light to always assuredly guide students through the sometimes dark corridors of structure and conformation.

Going beyond the fact that thermodynamics is defined by the equilibrium free energy difference (∆G) between reactants and products and that kinetics relates to the activation barrier (∆G^††) for getting from one to the other, I want to particularly discuss the importance of both these concepts for determining conformation by NMR spectroscopy.

There are two reasons why determining conformations in solution can become a particularly challenging endeavor. The first reason is thermodynamics. Again consider the all-important relation ∆G = -RTlnK which makes the equilibrium constant exquisitely sensitive to small changes in free energy (∆G). As mentioned before, an energy difference of only 1.8 kcal/mol between two conformations means that the more stable one exists to the extent of 96% while the minor one exists to the extent of only 4%. In practice such energy differences between conformers are seen all the time. A typical scenario for a flexible molecule in solution will posit a complex distribution of conformers being separated from each other by tiny energy differences ranging from say 0.5-3 kcal/mol. Again, the above exponential dependence of equilibrium constant K on ∆G means that the concentration of minor conformers which are higher in energy than the more stable ones by only 3 kcal/mol will be so tiny (~0.04%) as to be virtually non-existent. NMR typically cannot detect conformers which are less than 2-3% percent in solution (and it's too much to ask of NMR to do this), but such populations exist all the time.

Thus, thermodynamics is often the bane of NMR; in this case the technique is plagued by its low sensitivity

If thermodynamics is the bane, kinetics may be the nemesis. Rotational barriers between conformations (∆G^††) can be even tinier compared to thermal energy available to jostle molecules around at room temperature. For example, the classic rotational barrier for interconversion in ethane (whose origins are still debated by the way) is only 3 kcal/mol. Energy available at room temperature is about 20 kcal/mol which will make the ethane conformations interconvert like crazy. So even for energy barriers that are several kcal/mol, conformational interconversion is usually more than adequate to observe averaging of conformations and consequently all associated parameters- most importantly chemicals shifts and coupling constants- in NMR. The resolution time of NMR is on the order of tens of milliseconds, while conformational interconversion is on the order of tens of microseconds or less. Now in theory one can go to lower temperatures and 'freeze out' such motions. In many such experiments, line broadening at lower temperatures is observed, followed by separation of peaks at the relevant temperature. But consider that even for a barrier as high as 8-10 kcal/mol, NMR usually gives distinct, separate signals for the different conformers only at -100 degrees celsius. For barriers like those in ethane, the situation would be hopelessly challenging. As an aside, that means that sharp, well-defined resonances at room temperature do not indicate lack of conformational interconversion but can simply mean that conformational interconversion is fast compared to the NMR time scale.

Thus, kinetics is also often the bane of NMR; in this case the technique is plagued by low resolution time

Now there may be situations in which either thermodynamics or kinetics is favourable for carrying out an NMR conformational study. But for the typical flexible organic molecule, both these factors are usually pitted against the technique; rapid interconversion because of low rotational barriers, and low thermodynamic energy differences between conformers. Given this fact, it probably should not sound surprising to say that NMR is not that great a technique. However, as is well known to every chemist, its advantages far outweigh its drawbacks. Conformational studies comprise but one important aspect of countless NMR applications.

Nonetheless, when conformational studies are attempted, it should always be kept in mind that thermodynamics and kinetics have both conspired to make NMR an unattractive method for our purposes. Thermodynamics leads to low populations. Kinetics leads to averaging of populations. And yet the average information gained from NMR is invaluable and can shed light on individual solution conformations when combined with a deconvolution technique like NAMFIS or molecular dynamics. On the other hand, fitting the average data to a single conformation for a flexible molecule is inherently flawed and unrealistic. No one who has tried to take pictures of a horse race with a low-shutter speed camera should believe that NMR by itself is capable of teasing apart individual conformations in solution.

For determining conformations then, NMR alone does provide a wealth of data locked inside a safe. Peepholes in the door may illuminate some aspects of the system. But you need a key, best obtained from other sources, that will allow you to open the door and savor the treasures unearthed by NMR in their full glory.

Does a protein-bound ligand exist in only one conformation?

I have been thinking a lot recently about studies in which people have determined the bound conformation of a ligand by transfer-NOESY experiments, essentially by transferring magnetization off another ligand to the protein and then back to the ligand of interest. With the known bound conformation of the first ligand, one can apparently locate the conformation of the second one. Many such unknown protein-bound conformations have been worked out. In my field of research, the ones which are relevant are of agents that bind to tubulin, especially discodermolide. In this case, the conformation of discodermolide was deduced via competition transfer-NOESY experiments with epothilone. These experiments are non-trivial to carry out and, as is the case for other biomolecular NMR studies, should be interpreted carefully. But in the end they look like nifty techniques that can shed light on unknown bioactive conformations, something that's very valuable for drug design.

Essentially it's again a problem of fitting the bound conformation NMR data to a single conformation. In solution we know for sure that this is a fallacious step. The (not so) obvious assumption in doing this for bound conformations is that there's got to be only one conformation in the active site too. But I have always wondered if a ligand in a protein active site could also have multiple conformations. MJ's comments on a past post and the discussion there makes me think that even in a protein active site, there could possibly be multiple conformations of a ligand, something that runs counter to what we conventionally think. How diverse those conformations might be is a different question; one would probably not expect large conformational changes. But even 'small' conformational changes could be significant enough to distinguish between different conformations in the active site. It's a problem worth thinking about.

Article on NAMFIS in IIT-D magazine

A short holiday break and a rather protracted bout of the flu have kept me from blogging. So I will link to an article of mine that just got published in the magazine of the Chemical Society of the Indian Institute of Technology (IIT), Delhi. The article is written for the layman and talks about the importance of realizing that flexible molecules have multiple conformations in solution. Such conformations cannot be determined by NMR alone due to their rapid interconversion.

In the article, I describe NAMFIS (NMR Analysis of Molecular Flexibility In Solution), a joint computational-NMR approach which can derive a Boltzmann population for flexible molecules in solution. This information can be very useful for deducing, for example, the protein-bound conformation of a drug. But it can also be useful under other circumstances where determining conformation is important, such as for organic molecules assembling on a surface. Comments, criticism and questions are of course always welcome.

Hexacyclinol as a test case: ab initio C13 chemical shift prediction

Anybody heard of this natural product called hexacyclinol and how doubts were raised about its synthesis and structure? Kidding obviously. I am going to assume that any organic or related chemist who has not heard of hexacyclinol has not heard of Robert Burns Woodward by default.

Well, in any case, recall that the high point of that deb(acle)ate was Scott Rychnovsky's demonstration by using quantum chemical prediction of C13 chemical shifts that a structure quite different from hexacyclinol fit the C13 NMR data much better compared to JJLC's structure. To do this Rychnovsky used DFT methods and the mpw1pw91 functional which was tried, tested and proven to be a reliable tool for C13 chemical shift prediction by Bifulco and others. (excellent general review here which deals with calculation of both shifts and 2 and 3 bond homo and heteronuclear coupling constants)

The point of value for the organic chemist from the whole exercise was the fact that C13 chemical shift prediction could not just be used to distinguish regioisomers whose identity might be ambiguous but, based on Rychnovsky's analysis, also can be used to correctly assign misassigned C13 peaks. To me this is the greatest benefit of the analysis for the practicing organic chemist.

Henry Rzepa and Christopher Braddock at Imperial College in London have now demonstrated the application of this increasingly valuable method to the correct assignment of some interesting halogenated natural products called obtusallenes. In this case there was ambiguity about the positions of a chlorine and a bromine. The proton chemical shifts were very similar and could not be used to assign the positions. Rzepa and Braddock used the mpw1pw91 functional not just for the chemical shift calculation but also for the optimization. Fast computational power has made this possible now. The bottom line is that average C13 shift deviations are much more for the incorrect regioisomer. Using the method, the authors also re-assigned two ambiguous peaks. In addition, they determine that the 6-31G (d,p) basis set gives some errors for certain functional groups while using the aug-cc-pVDZ basis set (all that's left to say is "warp speed" now) basis set eliminates these errors.

A short, neat demonstration of the increasing value of quantum chemical NMR prediction methods for the practical organic chemist.

Braddock, D.C., Rzepa, H.S. (2008). Structural Reassignment of Obtusallenes V, VI, and VII by GIAO-Based Density Functional Prediction. Journal of Natural Products, 71(4), 728-730. DOI: 10.1021/np0705918

The negative IQ people at the State Department

This is really ridiculous. We are writing a paper with a friend and collaborator of mine who is a NMR specialist at a prominent university in the US. He just came back from India after a stay of more than two months. He said that his original trip was planned for only a month. So why did it take so long?

Apparently, his visa was delayed. The fine folks at the US State Department saw the dreaded word "nuclear" in his job description. Alarm bells went off in their experienced minds. Nuclear Magnetic Resonance? Surely this is suspicious. Off they went doing s background check for more than a month. In the end of course they found nothing. But my friend had to stay for an extra month, delaying his work here, not to mention our own work.

This is outrageous. NMR is one of the most important techniques ever in chemistry, biology, materials science and drug discovery. For crying out loud, life-saving MRI is based on it. Every single day, hundreds, if not thousands of papers are published in journals worldwide that involve the use of NMR in one way or the other. Four Nobel Prizes have been awarded to NMR scientists. My own PhD. thesis is mostly based on the interpretation of results obtained using NMR (I have mentioned about it here) NMR has nothing remotely to do with atomic bombs.

But the bull-headed rocks at the State Department cannot even distinguish between the "nuclear" in NMR and that in "nuclear weapons". Why can't they hire specialists who actually know something basic about science (and common sense) instead of randomly spouting gut reactions and going ballistic every time they see the word "nuclear"? In some ways, it would give people like me sadistic pleasure to think of all those floor scrubbers in the department running around trying to find out if I have a Jihadist background. But as everyone knows, unfortunately in the end the person who will lose the most will be me.

Despicable, and it reminds me of Goverdhan Mehta's shoddy treatment at the American consulate. But considering the ultimate authority they answer to, we can trust them not to look at trivial things like facts and details.

At least now I know what word to not include in my job description when I file for a Visa. "Magnetic Resonance" will have to do. Sigh.

Three lessons

When you have been in graduate school for two or three years, it's a good question to ask which are the important lessons you have learnt from your time there. These are not the "lessons of life" I am talking about- they are perhaps even more important- but technical lessons which spring to your mind. In my case, three lessons strikingly come to my mind. All of them are related to work that our group has done, but also to other people's work and general thought. All of them deal with material published in the literature that is unfortunately "fiction", and articles in journals don't seem to acknowledge this fact. Here they are:

1. Oxidation states for transition metals greater than 1 (eg. Au3+, Cu2+) are fictional and non-existent (related post and references)
2. Single "average" structures derived from NMR for flexible molecules are fictional and "virtual" (related post and references)
3. 3-fluoropiperidines exist in solution almost exclusively as the axial-F conformer. (related post and references)

The beastly behemoth

The damn thing took me more time in Chemdraw than I expected. If anyone wants the cdx file, I will be happy to mail it to them. That way it could save you some slave labour. Please be merciful and don't point out any mistakes in the structure...at least for now.



Going back to the original structure determination papers for maitotoxin (Yasumoto, Kishi; 1993, 1994, 1996), I find that some of the NOE assignments are kind of fuzzy, because the spectrum was naturally incredibly crowded in the disputed region. Regular 2D NOESY produced a smatter of fizz.


(Reference DOI: 10.1039/a901979k)

So they resorted to a fancier technique, PFG 3D NOESY, to pick out the cross peaks. The third axis in this NOESY is the C13 shifts, so that the protons are also separated on the basis of C13 shifts which differ more than proton chemical shifts (although still not by much, 3-4 ppm in some cases).

If I were asked if I believe in the assignments and the NOE correlations, I would probably say yes, but mainly on the basis of a process of elimination, and not on the basis of unambiguous confidence. But then I guess such is the nature of painstaking NMR structure determination of such complex natural products.

In any case, Spencer, Gallimore, and Nicolaou are definitely on to something, as described in the post before the last one, and we may learn something quite interesting about the NMR data or the biosynthesis.

MD? Check...at least in this case

I have often talked on this blog about the fallacy of deriving single average conformations for molecules from average NMR data. How can a molecule with ten rotatable bonds (and frequently more) have any single or dominant conformation in solution? And yet, one repeatedly comes across publications dealing with single average conformations derived for organic molecules.

Peptides, being favourite targets for solution conformational analysis, are often subjected to such analysis. One comes across many instances where people want an alpha helix, or a 3/10 helix, or a gamma turn. They then design a peptide which they think is constrained to form that motif. They get their average NMR data (coupling constants and distances from NOE data), and then do a constrained molecular dynamics run, and see their favourite motif on the screen. But of course, what you put in is what you get out. If you are constraining the molecule to fit the NMR data in the first place, it's not going to stray away from it, and not surprisingly you see what you want to see (Isn't this a general theme and trap for so many of us...to be tempted to see what we want to see?).

In general, molecular dynamics is not a very efficient method of searching conformational space. Plus, every MD program uses some force field, and force fields are well-known to give fairly accurate geometries if well parametrized, but not good energies. So there is no guarantee that all the "low-energy" conformations of any molecule will be sampled. But in theory, if one could actually search all of conformational space with a good force field, then one would explore the whole ensemble of structure that a molecule adopts in solution. Even with faster computers though, that goal is going to unrealizable for a general situation.

However, MD can be a good way of exploring conformational space if the molecule is highly constrained in the first place. In such a case, it can serve as a good way of deconvoluting the average NMR data. And this is prcisely what Nikiforovich and Marshall, two researchers who do seem aware of the general problem of assigning multiple conformations, have done (J. Med. Chem, DOI: 10.1021/jm070084n). They have taken a cyclic pentapeptide, Pro-Ala-Ala-Ala-Ala, and done long 100 nanosecond (which is a huge amount of time for a MD run) on the molecule. Most importantly, their MD runs are unconstrained. After running the MD with two different force fields and different starting velocities, they consistently see the same results; a major conformer without a gamma turn, and a minor one (about 6%) with a gamma turn, a usually forbidden motif. They also gratifyingly note that the average distances from their ensemble of MD structures match well with the NMR data.

This analysis was feasible and realistic because of a couple of reasons; firstly, the molecule is small (although it still has 8-10 rotatable bonds) and one can actually do a 100 ns MD run on this. Secondly, they used many different conditions that converged to the same two conformational profiles, and thirdly they used unconstrained MD to remove any bias. I am not completely satisfied with it though, because the molecule still does have 10 rotatable bonds and is quite flexible. But it is a ring, and some of the bonds are not as flexible, so in general it is constrained and I think the conclusions are quite believable.

For once, it gave me a satisfied feeling to see an analysis that gave results that are quite likely to be true, complete, and sweet.

P.S. Garland Marshall also has one of the more memorable quotes in science that I have heard:

"Nature does not shave with Ockham's Razor!"

Sad and totally unexpected...

Two years ago, we were working on determining the stereochemistry of some sphingolipid analogs. To this end, we corresponded several times with Prof. Luigi-Gomez Paloma from Salerno, Italy to seek his help in implementing a pulse sequence to determine 2 and 3 bond HMBC coupling constants. We were already aware of his pioneering efforts to compute stereochemistry by comparing experimental and calculated coupling constants using the mPW1PW91 density functional. Later, he also worked on calculating and comparing C13 NMR shifts for finding stereochemistry.

While the project fell through, he was quite helpful. I became acquainted with his work, and was quite happy to see Scott Rychnovsky use the same technique and functional in his work on hexacyclinol. While K C Nicolaou does not explicity say it, I suspect he used the same functional in his recent revision of the structure of maitotoxin. For reference, I still have a mail or two from Prof. Gomez-Paloma.

So it can be imagined what I must have felt when, completely unexpectedly, I saw a simple sentence on top of Nicolaou's recent paper.

"In memory of Luigi-Gomez Paloma"

Wow. That was unexpected and an unexpected way of getting to know it. Sad news.

Here are two of his pioneering papers. 1 (coupling constants) and 2 (C13 shifts). Prof. Gomez-Paloma was a talented organic chemist and an expert in NMR. He was definitely most instrumental in giving synthetic chemists the confidence to use quantum chemical NMR calculations to determine stereochemistry of natural products.

Least favourite NMR solvent, but a swell liquid nonetheless

Coronene has a poll on what everyone's least favourite NMR solvent is. I have not used all the solvents on the list, but among the ones I have used, DMSO was probably my least favourite, as it also seems to be for others taking the poll.

In my case, the reason was specific. I needed to do NOESY spectroscopy on a degassed, sealed sample. And even ice kept freezing up the DMSO when we were connecting it to the vacuum. For CDCl3, we could happily use lig. N2 and be happy, but not so for DMSO; we had to patiently do the degassing for almost half an hour. In addition, DMSO is viscous and radically changes the tumbling of molecules, causes aggregation and hydrogen bonding etc. Needless to say, one has to re-experiment with his/her NOESY parameters.

On the other hand, it is instructive to see what a great molecule DMSO is for biological applications. It can be used as a drug-delivery vehicle, although this use is debated. DMSO is widely used as a cryo-protectant, and the serine protease inhibitor PMSF is dissolved in DMSO as it will naturally get inactivated in water.

Ronald Breslow and the late Charlotte Friend also discovered that this simple molecule can induce cell differentiation and later, it was found to be a HDAC (histone deacetylase) inhibitor, a class of molecules that's heavily researched these days. Interaction with HDACs intimately interferes with transcription. For example, the recent celebrity 'wine molecule' resveratrol is purported to increase expression of the HDAC protein SIRT1, possibly affecting longevity.
Incidentally, Breslow's latest paper is an article in Nature Biotechnology, in which DMSO is used as the starting lead to develop SAHA, a HDAC inhibitor which is now in Phase 2 clinical trials as an anticancer agent.

DMSO also "enters the cell and kills the herpes virus". Apparently, this is an 'indisputable fact', and seems to be the basis of scented DMSO creams as the picture above shows. More interesting talk for the cocktail party; guys, beware of the DMSO girl.

A remarkably simple molecule with some remarkable (and occasionally annoying) properties, this DMSO.