Interview with Gary Martin

After a PhD in Pharmaceutical Sciences from the University of Kentucky (in 1976), NMR spectroscopist Gary Martin spent the first 14 years of his career at the University of Houston before moving to Burroughs Wellcome, Co., in 1989, and then to Upjohn in 1996, which, through a series of mergers and acquisitions, left him working for Pfizer a few years ago. He has spent much of his career focused on the identification of natural product structures and subsequently synthetic compounds originating in drug discovery, and more recently the identification of impurity and degradant structures of drug molecules.

He is moving to Schering-Plough in their Summit, New Jersey, facility in Spring 2006, with responsibilities similar to those he has had over the last decade for the isolation and identification of impurities and degradation products of drug substances. Martin has also been a pioneer in the development of new NMR pulse sequences, and the development and evaluation of new probe technologies for the past fifteen years or so.

With more than 35 years experience in NMR spectroscopy, Martin revealed some of the insights he has gained in this field to Reactive Reports.

RR: With the diverse roles of chemists in a global pharmaceutical company, companies want to see a collaborative effort of sharing ideas and research. Is this currently happening?

GM: In my experience there has been good sharing of ideas between scientists across divisions but less effective sharing of data. This has led to the necessity for reacquiring fundamental data on a lead compound when it becomes necessary to characterize an impurity or degradation product. The structural data are undoubtedly available, but generally, they’ve not been in a repository that was accessible across the breadth of a given company. In many instances, you might want to acquire fresh data for the sample in hand, but it would be interesting to have the option to look at the data already available.

RR: What is the value of integration between MS and NMR for de novo elucidation of structures?

GM: It is critical if impurities and degradation products are to be successfully and quickly identified. Going a step further, infrared data should also be part of the data package acquired. What’s the point in having to run a 13C NMR spectrum to identify a potential nitrile group, which could take hours, an overnight, or longer, when a simple IR spectrum can fingerprint this functional group in minutes? Infrared data, unfortunately, has fallen out of favor in many structural laboratories.

RR: What are your feelings regarding the impact of NMR on impurities and degradants?

GM: I’ve generally preferred “tube type” probe formats for ease of usage. These have included first 3 mm probes developed in a collaboration between my labs at Burroughs Wellcome, Co., and Nalorac Corp. in the early 1990s, 1.7 mm submicro or SMIDG probes developed in collaboration between my labs at Pharmacia & Nalorac, and most recently cryogenic probes that were developed collaboratively first with Nalorac and continuing with Varian after Nalorac was acquired by Varian. You can think of small volume, high sensitivity NMR probes as an exercise in “scale.” You could always do the structure of an impurity or degradation product, it was just a question of the brute-force isolation of enough material to acquire the requisite NMR data. Smaller volume probes make the isolation step a much more facile process.

RR: How will NMR participate in biomarker ID?

GM: Biomarkers are analogous to drug impurities and degradation products. The primary difference is the matrix from which they’re being extracted. Once you have any type of trace sample in hand, the structural characterization of it becomes fundamentally the same.

RR: What contributions to NMR are you most proud of?

GM: This obviously goes in part back to small volume high sensitivity probes, but also to experimental methods contributions including long-range 1H-15N and accordion optimized long-range experiments.

RR: Are there any new industries that could benefit from NMR?

GM: There are a couple of factors associated with NMR spectroscopy that in my opinion tend to inhibit its use as an “analytical” technique. To begin with, most NMR hardware isn’t cheap and requires a skilled instrument operator to select the experiments to be performed to answer a specific structural question and then to acquire the necessary data. That operator, or a colleague, must also be capable of interpreting the data that results from the broad range of possible experiments that can be done with a modern NMR spectrometer. Moreover, interpreting the data can sometimes be much less straightforward than, for example, the interpretation of mass spectral data.

RR: Will NMR ever become a black box “detector”?

GM: For some applications, NMR is already essentially a black box “detector.” For instance, determining moisture content of foods and similar quantitative measurements or fluoride content of toothpaste batches. In this capacity, the black box NMR detector is alive and well.

Insofar as structure elucidation is concerned I doubt that NMR will ever become a black box detector like UV spectroscopy. Instrument vendors are working diligently to make spectrometers easier to operate but to get the most from an experiment, parameter choices will often need to be made, which moves NMR away from what might be called “push button” operation. Beyond the acquisition of the data, there remains the interpretation of those data.

RR: Who will win the MS vs. NMR battle?

GM: This question isn’t about winners and losers. I have a lot of friends who are mass spectroscopists. They are fond of teasing the NMR folk about having a hard time getting a molecular weight right. Conversely, the MS drivers are teased in return with equal vigor about never knowing which isomer it is. While neither of those quips is strictly true, they do reflect the fact that NMR and MS data are highly complimentary.

RR: What do you think of NMR in the current drug discovery process?

GM: When I have had direct interaction with drug discovery colleagues, it has usually been when they’ve been looking for new pharmacologically active leads by high-throughput screening (HTS) of vast compound libraries. Some of those libraries contain samples that have been stored for decades. Some fraction of these compound collections will have degraded over time; another fraction may have never been what the sample was labeled as years ago when instrumental techniques were not nearly as sophisticated as they are now. In many instances, the material in the sample vial is a mixture of compounds and the task is to identify the active component, sometimes with only vanishingly small quantities of material remaining that can’t be replaced.

RR: Will the arms race towards higher and higher NMR fields continue?

GM: The quest for higher field magnets will continue for some applications of NMR and was, in my opinion, at least, over some time ago for other applications. For the investigation of molecules such as carbohydrates and nucleic acids, there is considerable benefit to be derived from much higher field magnets. All of the resonances for all of the individual monomers in these molecules resonate in the same place, making spectra extremely congested and difficult to interpret. Any increase in the dispersion of these resonances is beneficial. Proteins certainly benefit from higher fields, but perhaps to a slightly lesser extent since they have inherently better dispersion than either carbohydrates or nucleic acids.

For small molecules, e.g., drugs or natural products, the focus shifts from the need for increased dispersion associated with higher magnetic fields to a need for higher sensitivity. While higher field magnets assuredly bring greater sensitivity, they are a very expensive way to get that sensitivity. The quest for higher sensitivity in my laboratories at Burroughs Wellcome, Co., in the early 1990s focused on the development of smaller diameter NMR probes to allow the successful interrogation of smaller samples by NMR methods. Around the same time, inverse or indirect detection NMR methods were becoming widely available, further increasing sensitivity and making it possible to fully characterize, including the acquisition of long-range proton-carbon chemical shift correlation data, samples of drug-sized molecules on less than 50 micrograms of material. Probe diameters have progressively shrunk over the intervening years and 1 mm diameter probes are now commercially available, as are highly mass sensitive capillary probes such as those developed by Protasis. While those developments were being pursued by many investigators, myself included, work was also on-going on the development of cryogenic NMR probes. Radio astronomers have immersed their receivers in liquid nitrogen for many years to improve sensitivity. The principal behind cryogenic NMR probes is essentially the same, albeit with helium gas as the cryogenic refrigerant and considerably greater technical difficulties than those facing radio astronomers. Many of the technical challenges associated with cryogenic NMR probes have now been overcome and they are becoming more widely available. Using cryogenic NMR probe capabilities, samples of steroids and similar-sized molecules of less than 10 micrograms can now be fully characterized using full heteronuclear two-dimensional NMR techniques.

RR: What is the most under-used NMR technique?

GM: My answer to this question may be in conflict with the answer that some or even many of my colleagues would give. A modern NMR spectrometer is capable of generating data far faster than investigators can digest it. Spectrometers with robotic sample changers are capable of grinding out data 24/7/365 if people keep the rack on the sample changer filled. Investigators are quite literally drowning in data, with the problem further exacerbated by the need to report those data to someone in intelligible form or to archive those data somewhere. In my opinion, an area where spectroscopists could benefit enormously is in the automated interpretation and assignment of NMR data. Rather than consuming highly trained NMR spectroscopists with the interpretation of routine data for many samples, it makes far better sense to focus their time and talents on problems that truly warrant their attention, leaving the solution of simpler problems to computer software.

RR: If you could steer a major NMR instrument company, what direction would you take them in?

GM: The NMR vendors have built wonderful instruments with a vast array of capabilities. Probe technologies, whether micro scale (1.7 and 1 mm) tubes, capillary, or superconducting probes have been developed to exploit what the NMR spectrometers are now capable of for exceedingly small samples of precious or rare natural products, metabolites, impurities or degradation products, and forensic samples, to name just a few. What hasn’t kept pace with the developments in instrumentation are the tools necessary for getting the data that the instruments can now generate into reports and databases that are meaningful to the people who have to consume those data! If I had the opportunity to steer a major NMR vendor, it would be in the direction of improving the interface between the spectrometer and the spectroscopist to facilitate the jobs that people have to perform using these instruments. All of the pieces seem to have come together on the instrumental side of the equation – now it’s time to revisit what people do with these wonderful tools.