Neutron Crystallography Contributes to Drug Design

Monday, January 7, 2013

Carbonic anhydrases are a large group of enzymes that are expressed in humans. Carbonic anhydrase (CA) catalyzes the reversible hydration of CO2 to make bicarbonate and a proton. As such, this enzyme has many physiological roles involved, but has also emerged as an important clinical target for diseases. Deregulation of CA has been implicated in glaucoma and cancer. Current CA inhibitors cross-react strongly with other CAs in the body, leading to unpleasant side effects due to high sequence conservation between isoforms. There is a strong need for isoform specific drugs, and conventional structure-based drug design has not yet yielded effective compounds. Neutron crystallography is the only technique that can show hydrogen (or its isotope deuterium) and hydrogen-bonds to the enzymes. This experimental approach promises to contribute to the drug design field in many profound ways.

Background

Up to one half of all atoms found in proteins are hydrogen (H) atoms. Even though these are the smallest atoms, they contribute in many important ways to protein folding, hydrogen bonding, substrate or inhibitor binding, protein solvation, and enzyme catalysis.1 Many enzymes use H2O, OH–, H+, and even H3O+ to mediate enzyme catalysis or inhibitor binding.2 The enzyme in the current study, human carbonic anhydrase II (HCA II), catalyzes the reversible carbon dioxide hydration reaction: CO2 + H2O → HCO3– + H+. The reaction has two parts: carbon dioxide hydration and a rate-limiting proton transfer event. HCA II uses a hydrogen-bonded network of water molecules to perform proton transfer, and we have been able to study this using neutrons in the past.3,4

CAs are also important medical targets, and CA inhibitors (CAIs) have been used for decades to treat glaucoma, altitude sickness, epilepsy, and certain forms of hypertension. There are 15 isoforms expressed in humans, and one of them, HCA IX, is only expressed in cancer cells when tumors become hypoxic. It is thought that HCA IX’s extracellular domain works to acidify the extracellular matrix, facilitating cancer cell escape resulting in metastasis.5 As a result, there is a huge effort underway to design isoform-specific CAIs, but due to the high level of sequence (and structure) conservation of these HCAs, there is significant cross-reactivity.

This off-target binding of drugs to other CAs leads to unwanted side effects. More atomic details on drug interactions with HCA II in terms of water involvement, hydrogen-bonding (H-bonding), ionization of bound drug, and interactions with amino acid residues will facilitate drug design that is tailored for specific isoforms.6

Neutron diffraction

The method of choice for obtaining atomic protein structures is X-ray crystallography. The magnitude of X-ray scattering from the electron cloud around an atomic nucleus is related to the Z number for that element, i.e., the more electrons an atom has, the better it will scatter X-rays. Due to this, the H atom hardly scatters X-rays, and ultrahigh-resolution data, >1.0 Å resolution, are needed to directly observe H atoms. However, a recent study by Gardberg et al. showed that even in high resolution X-ray data, only a small subset of all H atoms are seen.7

Neutron diffraction offers a complementary approach in that the neutrons are scattered from atomic nuclei of all elements to a similar extent. The atom types that are commonly found in proteins, such as C, N, O, and S, scatter neutrons with lengths of 6.6, 9.4, 5.8, and 2.8 fm, respectively (see Table 1). Neutrons also diffract differently from isotopes of the same element, e.g., H has a negative scattering length of –3.7 fm, while its isotope deuterium (D) scatters neutrons with a positive scattering length of 6.7 fm.

Table 1 – Neutron scattering lengths for the most common atoms found in proteins (adapted from NIST: http://www.ncnr.nist.gov/resources/n-lengths/)

A recent neutron structure report showed that close to 95% of all H atoms in a protein were seen at 1.1 Å resolution. In contrast, using 0.85 Å resolution X-ray data, only about 30% were observed.1 Less than 0.7% of all Protein Data Bank (PDB) entries fall in this elite category, making neutrons the method of choice for observing the highest number of H (or D) atoms.

The difference between positive and negative scattering lengths between neighboring atoms can cause signal cancellation. Also, H has a very large incoherent neutron scattering component (80 barns vs 2 barns for D). This interferes with signal-to-noise detection and high background. D atoms have the ideal scattering lengths for neutron experiments, and it is routine to either prepare perdeuterated or H/D exchange protein crystals prior to neutron data collection. This facilitates the observation of H-bonds, protonation state of amino acid side chains, water (or D2O) orientation, and H-bonding interactions.8,9  Figure 1 illustrates the utility of neutron crystallography.

Figure 1 – Nuclear density maps reveal the protonation state of amino acid side chains and can help distinguish between charged and neutral residues and also the orientation and H-bonding interactions of water molecules. a) Neutral histidine, b) charged histidine, c) correct orientation of glutamine, d) neutral lysine, e) charged lysine, f) characteristic shape of D2O molecule.

Drug or inhibitor binding to target proteins is mediated through a number of possible interactions that may include one or more of the following: energetic changes through water displacement, H-bonding to the protein directly or through intervening waters, electrostatic interactions with charged or polar amino acid side chains, metal coordination, aromatic stacking, or through other hydrophobic associations. After examining this list of potential interactions it is clear that X-ray crystallography is able to report on only a few of these binding interactions.

Neutron crystallography offers a novel approach to obtain most, if not all, of the drug binding interaction details that involve H/D atoms to a particular target.10–12 There are very few examples of neutron structures of drugs bound to their protein targets: Eschericia coli dihydrofolate reductase (DHFR) in complex with methotrexate, HIV-1 protease with inhibitor KNI-272, and porcine pancreatic elastase with FR130180.11,13,14 However, this list is for nonhuman enzymes with clinical drugs bound or disease-relevant enzymes with drugs bound that are not used clinically (i.e., they are experimental).

Recently the first neutron structure of a human target, carbonic anhydrase II (HCA II), in complex with its clinically used drug, acetazolamide (AZM), was reported.12 All of these studies have revealed specific H-bonds, waters, and other electrostatic interactions that mediate drug binding. With the advent of modern neutron protein crystallography instruments, we expect neutrons to contribute in a significant way to drug design in the future.

Experimental and instrumentation

The recent study on HCA II:AZM was conducted on a single, large H/D exchanged crystal of the enzyme:inhibitor complex. The room temperature neutron data to 2.0 Å resolution was collected at the Protein Crystallography Station (PCS) at Los Alamos National Laboratory (Los Alamos, NM) (Figure 2). The PCS is currently the only macromolecular diffraction instrument that was built at a spallation neutron source in North America. PCS is funded by the Department of Energy’s Office of Biological and Environmental Research (DOE-OBER) to operate a user program that serves many of the nation’s critical mission areas in medicine and human health, bioenergy, renewable biofuels, and genome science.

Figure 2 – Photograph of the 3He-filled position-sensitive PCS detector and schematic of the beam line. Left: The PCS detector covers 120° around the sample to maximize data collection efficiency. It comes equipped with κ-circle goniometer and has cryo capabilities to allow data collection at 293 or 100 Kelvin. Right: The PCS beam line has a 28-m flight pathlength and an in-line T-zero chopper that removes high-energy neutrons.

The PCS uses a position-sensitive 3He-filled neutron detector that detects neutrons with ~80% efficiency (Figure 2). The PCS receives spallation neutrons that are delivered in 20-Hz pulses in the thermal regime. Due to the pulsed nature of the incoming neutron beam, the instrument can resolve time-of-flight neutrons in the wavelength range between 0.6 and 7.0 Å, effectively increasing the number of neutrons (flux) in a shorter time frame compared to nuclear reactor-based instruments.15,16 A room-temperature X-ray diffraction data set to 1.6 Å resolution was collected on a similar crystal on an in-house Rigaku diffractometer (Tokyo, Japan). The structures were refined as described elsewhere.12

Results and conclusion

The current study on the neutron structure of the HCA II:AZM complex is the first report of a human target with a clinically used drug bound. It is known that AZM can have three possible charged forms that bind to HCA II when it is dissolved in water.17 All three species have pKa values in the physiological pH range (pKa from 7.2 to 8.7). Up to now there has never been crystallographic data supporting the preferential binding of any one of these three forms over the others.

From a pharmaceutical and drug development perspective, it is important to know which form binds best, since this information can be exploited to synthesize drug derivatives with new properties. A comparison of the nuclear and electron density maps shows the strong scattering from D atoms and highlights the complementarity of the two probes.

Figure 3 – Atomic details of acetazolamide (AZM) binding to the active site of HCA II determined by neutron and X-ray crystallography. Left: The 2Fo-Fc nuclear density map is shown in yellow and is contoured at 1.5σ. Right: The 2Fo-Fc electron density maps is shown in blue and is contoured at 2.0σ. Residues are as labeled; deuterium atoms are shown in cyan and hydrogen in white.

A detailed analysis of the neutron data (Figures 3 and 4) revealed atomic binding interactions between AZM and certain active site residues of HCA II. The data clearly show that the anionic form of AZM is bound, with the negative sulfonamide group bound to the active site zinc (Figure 3). The D atom on the acetoamido group H-bonds to the enzyme through W1120. Threonine 200 acts as a bifurcated H-bond to both W1120 and the backbone carbonyl of Proline 201 (Figure 4). This structural analysis provides important direct information on H-bonding and direct interactions between AZM and HCA II. Neutrons have tremendous application in this field because it is the only technique that can provide high-resolution, atomic details on H/D atom positions and interactions.

Figure 4 – Ball-and-stick diagram of binding interactions between AZM and the active site of HCA II determined by joint neutron and X-ray refinement. Residues are labeled and hydrogen-bonds are shown as dashed lines. Deuterium atoms are shown in cyan and hydrogen in white.

For future developments, the author’s laboratory is using neutrons to better understand the enzyme mechanism of CA to enable protein engineering. They have been able to improve the enzyme activity and stability by making strategic mutations based on neutron data. The laboratory is also looking at different drug targets such as antibiotic resistance enzymes, the Zn beta-lactamases, found in many pathogenic bacteria. These lactamase enzymes employ an activated (catalytic) water to hydrolyze lactam groups and the atomic details can only be visualized with neutron crystallography.

References

  1. Chen, J.C.; Fisher, Z. et al. Room temperature ultrahigh resolution time-of-flight neutron and X-ray diffraction studies of H/D exchanged crambin. Acta Cryst.  2012, F68, 119–23.
  2. Kovalevsky, A.Y.; Hanson, B.L. et al. Identification of the elusive hydronium ion exchanging roles with a proton in an enzyme at lower pH values. Angew. Chem. Int. Ed.  2011, 50, 7520–3.
  3. Fisher, S.Z.; Kovalevsky, A.Y. et al. Neutron structure of human carbonic anhydrase II: implications for proton transfer. Biochemistry  2010, 49, 415–21.
  4. Fisher, S.Z.; Kovalevsky, A.Y. et al. Neutron structure of human carbonic anhydrase II: a hydrogen bonded water network “switch” is observed between pH 7.8 and 10.0. Biochemistry  2011, 50, 9421–3.
  5. Ditte, P.; Dequiedt, F. et al. Phosphorylation of carbonic anhydrase IX controls its ability to mediate extracellular acidification in hypoxic tumors. Cancer Res.  2011, 71, 7558−67.
  6. Aggarwal, M.; Kondeti, B. et al. Insights towards sulfonamide drug specificity in alpha-carbonic anhydrases. Bioorg. Med. Chem. 2012; doi:10.1016/j.bmc.2012.08.019. Epub ahead of print.
  7. Gardberg, A.S.; Del Castillo, A.R. et al. Unambiguous determination of H-atom positions: comparing results from neutron and high-resolution X-ray crystallography. Acta Cryst.  2010, D66, 558–67.
  8. Glusker, J.P.; Carrell, H.L. et al. Using neutron protein crystallography to understand enzyme mechanisms. Acta Cryst.  2010, D66, 1257–61.
  9. Meilleur, F.; Myles, D.A.A. Neutron Laue macromolecular crystallography. Eur. Biophys. J. 2006, 35, 611–20.
  10. Kuroki, R.; Okazaki, N. et al. Towards investigation of the inhibitor-recognition mechanisms of drug target proteins by neutron crystallography. Acta Cryst. 2010, D66, 1126–30.
  11. Bennett, B.; Langan, P. et al. Neutron diffraction studies of Escherichia coli dihydrofolate reductase complexed with methotrexate. Proc. Natl. Acad. Sci. USA 2006, 103, 18493–8.
  12. Fisher, S.Z.; Aggarwal, M. et al. Neutron diffraction of acetazolamide-bound human carbonic anhydrase II reveals atomic details of drug binding. J. Am. Chem. Soc. 2012, 134, 14726–9.
  13. Adachi, M.; Ohhara, T. et al. Structure of HIV-1 protease in complex with potent inhibitor KNI-272 determined by high resolution X-ray and neutron crystallography. Proc. Natl. Acad. Sci. USA 2009, 106, 4641–6.
  14. Tamada, T.; Kinoshita, T. et al. Combined high-resolution neutron and X-ray analysis of inhibited elastase confirms the active site oxyanion hole but rules against a low-barrier hydrogen bond. J. Am. Chem. Soc. 2009, 131, 11033–40.
  15. Schoenborn, B.P.; Langan, P. Protein crystallography with spallation neutrons. J. Synchrotron Rad.  2004, 11, 80–2.
  16. Langan, P.; Greene, G. Protein crystallography with spallation neutrons: collecting and processing wavelength-resolved Laue protein data. J. Appl. Cryst.  2004, 37, 253–7.
  17. Supuran, C.T.; Winum, J.-Y. Drug Design of Zinc-Enzyme Inhibitors: Functional, Structural, and Disease Applications; John Wiley & Sons: Hoboken, NJ, 2009; pp. 439−46.

Dr. Zoë Fisher is Research Scientist II at Los Alamos National Laboratory, Bioscience Division B-11, P.O. Box 1663, Los Alamos, NM 87545, U.S.A.; tel.: 505-665-4105; e-mail: [email protected].