Request a Quote
BioNMR for Drug Discovery
BioNMR: An Essential Tool for Discovering Novel Drugs
David Fry, Ph.D.
NJ Bio, 350 Carter Rd, Princeton, NJ 08540, U.S.A.
Last updated on 12th January 2023
NMR is a powerful and versatile technique for obtaining structural information on molecules. BioNMR is the application of this methodology to larger biomolecules such as proteins, DNA, RNA, peptides, macrocycles, and a new class of drug called targeted protein degraders (TPDs). It can be utilized at every step in the drug discovery process, from hit finding, to lead selection, to optimizing a lead into a successful clinical candidate. There are other methods that can be helpful for these activities. However, BioNMR offers unique advantages that cannot be supplied by any other technique. There are cases of successful drug and clinical compound development where BioNMR played an essential role, and the drug could not have been discovered without it (1-3). This review will describe how BioNMR can help guide a drug discovery campaign and will highlight how BioNMR complements other methods, and when it is superior or uniquely capable versus other techniques.
I. Small Molecule Drugs
The Lead is Everything
Modern drug discovery focuses on individual protein targets and, when a small molecule is sought, starts with finding a lead – a compound that binds to the target and alters its function in a desired way. Through a medicinal chemistry campaign, the lead must then be developed into a clinical candidate by optimizing its activity and druglike properties. When one compares clinical candidates to their parent lead molecules, the basic attributes of the lead, with regard to how it interacts with the target protein, are usually retained. So, finding the right lead is the most critical step in a drug discovery program. BioNMR is uniquely powerful at identifying good leads.
The initial step in finding a lead is to evaluate many chemotypes, usually by some sort of screen involving a library of molecules. The result of the screen is a set of hits – molecules that register as positives in whatever screening technique is being utilized. It is critical to note that hits are not all of equal value. Only a rare few are true leads and have the potential to evolve into clinical candidates. BioNMR can sort out the quality of screening hits.
Evaluating hits by BioNMR is best done by observing the protein, adding the test molecule, and monitoring the response (Figure 1) (4). If there is no response, assuming the test molecule was soluble, then the hit is immediately classified as a false positive. The high-throughput methods used in modern compound screening are notorious for producing many false positives. These must be identified quickly to avoid wasting chemistry resources. BioNMR is an excellent method for this, and is the only technique that can definitively prove absence of binding. There are numerous examples of drug discovery programs where hit compounds thought to be legitimate binders were shown to be false positives using BioNMR (1, 5, 6).
Not only does BioNMR produce an immediate yes/no classification for the binding of screening hits, for the compounds that bind it also reveals the nature of the interaction (7). First, it identifies any undesirable effects of the compound on the target protein – such as unfolding the protein or causing it to aggregate. Also, by the nature of the initial response, and by performing wash-out experiments, BioNMR can identify covalent binding. Further, BioNMR can quickly identify the location on the protein where the compound is binding (Figure 2). Although this initial information is lower resolution, more of a “footprint” than a detailed structure, it is still of great value for ranking hits and initiating chemistry efforts.
How can BioNMR pick out the binders with the best potential as leads? Binding of a small molecule to a protein can be achieved in a variety of ways. In some cases, it involves embedding of the compound into a pocket of the protein, while in other cases binding is more superficial. BioNMR can differentiate between these two cases and help avoid further work on superficial binders with little potential for achieving druglike affinity.
For hits with high potential, the binding footprint obtained from BioNMR can be used in conjunction with computational modeling to dock the compound into the protein (Figure 3). The resulting model can suggest chemistry strategies for increasing affinity, for example, nearby subpockets that could be filled by adding appendages to the molecule, or potential interacting partners that could be engaged by derivatizing the molecule. With all of this information in hand, the most promising leads can be identified – ideally, compounds binding reversibly to the desired site, inserted into a protein pocket, with nearby additional interactions that could be exploited. BioNMR is the fastest way to arrive at this position in a drug discovery program, and as mentioned, it is the most critical step with regard to eventual success. There are examples in the literature of the use of BioNMR to focus a drug development program on a good lead and to efficiently kick-start the chemistry optimization campaign (1-3, 8-10).
Figure 1: BioNMR is a “gold standard” method for sorting out authentic hits from false positives.
Figure 2: Changes in the 2D NMR spectrum of a 15N-labeled protein upon addition of a compound can reveal its binding location.
Figure 3: Once a binding footprint has been identified by BioNMR (left; blue circles), molecular modeling can be used to dock the ligand into the structure of the protein (right).
Since BioNMR can detect binding, it can be used as a primary technique for screening (11). While this would appear to be a low-throughput approach, it has the advantage of providing simultaneous hit validation. In comparison, approaches commonly regarded as high-throughput must be followed up with a step that identifies the false positives. BioNMR compares quite favorably when considering total overall time for screening.
One unique capability of BioNMR is the interrogation of multiple protein targets in a single screen. The NMR spectra that are used to monitor binding are different for each protein. It has been shown that in a mixture of two or three proteins, the overlap is low enough that the individual protein spectra can still be reasonably visualized. Thus, it can be determined to which protein a particular hit is binding (Figure 4). This is a very efficient approach if one has a compound library to screen against multiple targets, or if one is trying to assess selectivity. BioNMR is the only technique that can perform this type of multiple-target screening.
Because it is a biophysical technique, BioNMR can be used to find hits for target proteins whose function is not yet known. For example, BioNMR is being used in drug discovery against SARS CoV-2 proteins (URL:https://covid19-nmr.com).
Figure 4: Simultaneous screening of multiple proteins using BioNMR. Multiple proteins can be distinguished in the 2D NMR spectrum (e.g. – MDMX vs. eIF4E; top). Selective perturbations then indicate which protein is binding the hit (e.g. – MDMX is binding the compound designated RO5506515; bottom).
BioNMR is one of the most sensitive methods for detecting binding. Therefore, it is one of the best methods for fragment screening, where hits are expected to have low affinities (12). Fragment screening employs small compounds in an attempt to discover leads by first identifying one sub-epitope. Neighboring sub-epitopes can be produced by chemically expanding that hit, or can be found by repeating the fragment screen with the first site saturated. This is an extremely efficient way to explore chemical space (13). In the conventional approach, using larger compounds, one hopes that two or three sub-epitopes are present on a single molecule. Therefore, fragment screening has a clear mathematical advantage – if one assumes three sub-epitopes were found by screening 100 fragments, to have these three sub-epitopes present on a single molecule would require screening 1003 (one million) conventional molecules.
There are now many examples of promising leads that have been discovered via fragment screening, and several examples of approved drugs which were developed starting with leads derived from fragment screening (14, 15, 16), including cases starting with BioNMR screening (2).
We have discussed proteins as targets for small molecule drug discovery because this has been the most traditional route, but all of the BioNMR methods just described can be used when the target is DNA or RNA (17).
In addition to the studies just described, which are rapid but of somewhat lower resolution, it is possible to determine full detailed 3D structures of proteins and protein-ligand complexes (Figure 5), and of DNA and RNA, using BioNMR (18). These efforts take longer and require more material, but yield structures with resolution that can be comparable to X-ray crystallography. BioNMR structures can be essential when the target of interest refuses to crystallize. This highlights the versatility of BioNMR, where one can fine-tune speed and resolution to fit one’s needs, as opposed to X-ray crystallography which is an all-or-nothing technique.
II. Medium-Sized Drugs
There is a growing interest in developing drugs in the molecular weight range of 1 to 5 kD. These are primarily peptides, macrocycles, and targeted protein degraders (TPDs). These molecules are large enough to have intricate 3D structures of their own. So, in addition to learning how they interact with their protein targets, it is very valuable to obtain structures of the molecules themselves. BioNMR can determine full, detailed structures for this class of molecule. In fact, it is usually impossible to obtain X-ray crystal structures of molecules in this size range.
Peptide, macrocycle, and TPD leads typically have deficiencies with regard to solubility and stability, which need to be improved along with potency during a drug optimization program. Having a structure of the lead allows one to intelligently design modifications, rather than make random changes and hope for improvement. For example, it has been shown that helical peptides can be stabilized via covalent linkage of side-chains (20), and by replacement of amino acid residues with organic spacers (Figure 6). These strategies require having the structure of the parent molecule to know where to locate these modifications, and how to correctly space them. TPDs have also been stabilized by replacing portions with more rigid spacers (21), and a structure of the TPD can be very useful for guiding this process.
Molecules in this size range are large enough to have what one may call “binding sites” where they interact with their target proteins. Accordingly, BioNMR can be used to map these sites, and provide binding footprints, as previously described for proteins. These studies may, in fact, also benefit by having the peptide, macrocycle, or TPD labeled with 15N and/or 13C. This can be accomplished synthetically.
III. Protein and Nucleic Acid Drugs
Drug classes whose members have the largest molecular weights include cytokines, antibodies, antibody-drug conjugates (ADCs), antisense oligonucleotides, and interfering RNAs. Most of these are close to, or exceed, the molecular weight limit for structural determination by BioNMR. However, BioNMR is very sensitive for performing a structural comparison of two versions of a protein. BioNMR has been used to evaluate biosimilars – protein therapeutics that are produced by a different source than the original manufacturer – to prove that they are sufficiently similar to the parent product (22). BioNMR can also be used for quality control, to check whether successively produced batches of a protein are identical.
For antisense oligonucleotides and other nucleotides that require artificial chemical modifications to impart stability, BioNMR can provide structural information to reveal how the modifications affect folding and base-pairing. This information can assist design strategies during the optimization process.
For antibody-drug conjugates, or any other protein drug that involves covalent attachment of a chemical group, such as PEG or a carbohydrate, BioNMR may be able to assess where these attachments have occurred, and any structural consequences. Due to the size of the resulting complex, this may require isotopic labeling of the protein drug, the attached polymer, or both.
We have described how BioNMR has unique strengths in providing structural information to a drug discovery program (Table I). As targets become more difficult, BioNMR becomes more essential to overcome the special hurdles presented. When hits cannot be obtained by screening conventional libraries, BioNMR is the method of choice for performing a fragment screen and efficiently exploring new regions of chemical space. When difficult targets produce an excessive number of false positives, BioNMR is a gold standard method for ruling these out, and for focusing the project on the very best leads. During lead optimization, BioNMR can allow downward adjustment of resolution to gain greater speed in obtaining crucial results. Conversely, when necessary, BioNMR can provide high resolution structures, particularly when X-ray crystallography is not successful.
BioNMR is typically the only method capable of delivering structures of medium-sized molecules, which are growing in popularity as novel classes of drugs. Modifying DNA and RNA to create drugs such as antisense oligonucleotides is also gaining in popularity, and BioNMR is able to examine these modified structures to see if they are folding and base-pairing as desired. Attachment of accessory molecules, such as toxins or PEG, to proteins has proven to be a successful strategy for producing novel drugs, and BioNMR has the potential to determine where these molecules are attached and if their presence causes any structural effects.
In conclusion, it is clear that in order to make a fully-armed effort toward discovering drugs, especially novel drugs and those against difficult targets, it is essential to include the unique capabilities provided by BioNMR.
|Drug Type||Deliverables from BioNMR|
|Small Molecule||► Screening
► Hit validation
► Hit ranking
► Binding footprint on target protein
► 3D structure of complex with target protein
(Peptide, macrocycle, TPD)
|► 3D structure of molecule
► 3D structure of modified analog
(Cytokine, antibody, ADC, antisense oligonucleotide, iRNA)
|► Site of attachment of appendage
► 3D structural checks and comparisons
(1) Fry, D. C., Graves, B. J., and Vassilev, L. T. Development of E3-substrate (MDM2-p53) Binding Inhibitors: Structural Aspects. Methods Enzymol. 2005, 399C, 622-633.
(2) Oltersdorf, T., Elmore, S. W., Shoemaker, A. R., Armstrong, R. C., et al. An Inhibitor of Bcl-2 Family Proteins Induces Regression of Solid Tumours. Nature 2005, 435, 677–81. doi:10.1038/nature03579
(3) Gesell J. J., McCoy M. A., Senior M. M., Wang Y-S., Wyss D. F. NMR-Based Screening Applied to Drug Discovery Targets. In: Webb G.A. (eds) Modern Magnetic Resonance. Springer, Dordrecht. 2008, 1419-1428.
(4) Williamson, M. P. Using Chemical Shift Perturbation to Characterise Ligand Binding. Prog. Nucl. Magn. Res. Spec. 2013, 73, 1–16.
(5) Zega, A. NMR Methods for Identification of False Positives in Biochemical Screens. J. Med. Chem. 2017, 60, 9437−9447.
(6) Musielak, B., Kocik, J., Skalniak, L., Magiera-Mularz, K., Sala, Czub, M., Stec, M., Siedlar, M., Holak, T.A., Plewka, J. CA-170 – A Potent Small-Molecule PD-L1 Inhibitor or Not? Molecules 2019, 24, 2804-2817.
(7) Fry, D. C., Emerson, S. D. Applications of Biomolecular NMR to Drug Discovery. Drug Des. Disc. 2000, 17, 13-33.
(8) Negi, A., Murphy, P. V. Development of Mcl-1 Inhibitors for Cancer Therapy. Eur. J. Med. Chem. 2021, 210, 113038.
(9) Wang, F., Jeon, K. O., Salovich, J. M., Macdonald, J. D., Alvarado, J., et al. Discovery of Potent 2‑Aryl-6,7-dihydro‑5H‑pyrrolo[1,2‑a]imidazoles as WDR5-WIN-Site Inhibitors Using Fragment-Based Methods and Structure-Based Design. J. Med. Chem. 2018, 61, 5623−5642.
(10) Schoepfer, J., Jahnke, W., Berellini, G., Buonamici, S., Cotesta, S., et al. Discovery of Asciminib (ABL001), an Allosteric Inhibitor of the Tyrosine Kinase Activity of BCR-ABL. J. Med. Chem. 2018, 61, 8120–8135.
(11) Wu, B., Barile, E., De, S. K., Wei, J., Purves, A., Pellecchia, M. High-Throughput Screening by Nuclear Magnetic Resonance (HTS by NMR) for the Identification of PPIs Antagonists. Curr. Top. Med. Chem. 2015, 15, 2032–2042.
(12) Ma, R., Wang, P., Wu, J., Ruan, K. Process of Fragment-Based Lead Discovery—A Perspective from NMR. Molecules 2016, 21, 854-868.
(13) Roughley, S. D., Hubbard, R. E. How Well Can Fragments Explore Accessed Chemical Space? A Case Study from Heat Shock Protein 90. J. Med. Chem. 2011, 54, 3989–4005.
(14) Tsai, J., Lee, J. T., Wang, W., Zhang, J., Cho, H., et al. Discovery of a Selective Inhibitor of Oncogenic B-Raf Kinase With Potent Antimelanoma Activity. Proc. Nat. Acad. Sci. 2008, 105, 3041–3046.
(15) Tap, W. D., Wainberg, Z. A., Ibrahim, P. N., Zhang, C., Healy, J. H., et al. Structure-Guided Blockade of CSF1R Kinase in Tenosynovial Giant-Cell Tumor. N. Engl. J. Med. 2015, 373, 428-37.
(16) Perera, T. P. S., Jovcheva, E., Mevellec, L., Vialard, J., De Lange, D.,et al. Discovery and Pharmacological Characterization of JNJ-42756493 (Erdafitinib), a Functionally Selective Small-Molecule FGFR Family Inhibitor. Mol. Cancer Ther. 2017, 16, 1010-1020.
(17) Yu, A.-M., Choi, Y. H., Tu, M.-J. RNA Drugs and RNA Targets for Small Molecules: Principles, Progress, and Challenges. Pharmacol. Rev. 2020, 72, 862-898.
(18) Fry, D. C., Emerson, S. D., Palme, S., Vu, B. T., Liu, C.-M., Podlaski, F. NMR Structure of a Complex Between MDM2 and a Small Molecule Inhibitor. J. Biomolec. NMR 2004, 30, 163-173.
(19) Alderson, T. R., Kay, L. E. NMR Spectroscopy Captures the Essential Role of Dynamics in Regulating Biomolecular Function. Cell 2021, 184, 577-595.
(20) Fry, D. C., Madison, V. S., Greeley, D. N., Felix, A. M., Heimer, E. P., et al. Solution Structures of Cyclic and Dicyclic Analogs of Growth Hormone Releasing Factor as Determined by Two-Dimensional NMR and CD Spectroscopies and Constrained Molecular Dynamics. Biopolymers 1992, 32, 649-666.
(21) Farnaby, W., Koegl, M., Roy, M., Whitworth, Diers, E., et al. BAF Complex Vulnerabilities in Cancer Demonstrated Via Structure-Based PROTAC Design. Nat. Chem. Biol. 2019, 672, 672–680.
(22) Japelj, B., Ilc, G., Marušič, J., Senčar, J., Kuzman, D., Plavec, J. Biosimilar Structural Comparability Assessment by NMR: From Small Proteins to Monoclonal Antibodies. Sci. Rep. 2016, 6, 32201.