H has a neutron scattering length (−3.7 × 10⁻¹⁵ m or −3.7 fm) that is similar in magnitude but opposite in sign to those of other atoms found in proteins (O, 5.8 fm; N, 9.4 fm; C, 6.6 fm; S, 2.8 fm), while deuterium (D) has a positive scattering length (6.7 fm) and a significantly smaller incoherent neutron-scattering cross-section (~2.0 barns versus 80 barns for H). Consequently, the signal-to-noise ratio can be markedly improved in neutron scattering when D atoms can be substituted for H atoms. In contrast to neutrons, the diffraction of X-­rays depends on the number of electrons; as H and D are relatively electron-poor compared with the heavier atoms found in proteins, they diffract very weakly and appear practically invisible. All these factors together make it much easier to locate D or H in resulting nuclear density maps (Shu et al., 2000 ) and neutron crystallography can readily provide information on the protonation states of amino-acid residues, ligands and the nature of bonds involving hydrogen (Blakeley, Langan et al., 2008 ). Neutron crystallography can also be used to identify H atoms that are exchanged with D and the extent of this replacement, thus providing a tool for identifying isotopically labeled structural features, for studying solvent accessibility and macromolecular dynamics and for identifying minimal protein-folding domains (Bennett et al., 2008 ).

Neutron crystallography is also a powerful tool for studying the hydration of macromolecules, especially when it is combined with X-­­ray crystallography in joint (XN) refinement procedures (Blum et al., 2009 ). In electron-density maps, a water molecule is usually represented as a spherical density peak corresponding to the position of the O atom. However, in neutron scattering density maps, owing to the strong scattering contribution from D, the density associated with D₂O may no longer be spherical but rather extended. It can often be difficult to interpret these extended neutron scattering density peaks. However, we have found that using both X-ray and neutron data together can greatly help in this interpretation by allowing placement of the O atom and the subsequent interpretation of the extended neutron scattering density peak as either one or two D atoms. For this kind of analysis it is ideal to have neutron data to 2.0 Å resolution or better. Both the ability to accurately place and orient D₂O and to view the resulting hydrogen-bonded patterns they participate in can give enormous advantages in understanding how enzymes work.

Carbonic anhydrases (CAs) are ubiquitous enzymes that are found in all phyla of life and are intricately involved in many physiological processes (Tufts et al., 2003 ). Of all the isoforms, human carbonic anhydrase II (HCA II) is the most studied and probably the best understood. CAs catalyze the reversible hydration/dehydration of CO₂/HCO₃ ⁻. The reaction occurs in two distinct steps: the conversion of CO₂ to HCO₃ ⁻ and a subsequent proton-transfer (PT) step to regenerate the active site (see equations below; E = enzyme, B = H⁺ acceptor/donor, which can be buffer, His64 or bulk solvent; Christianson & Fierke, 1996 ; Silverman & Lindskog, 1988 ).The first step is fairly well understood and a crystal structure of the enzyme–substrate complex was recently determined showing the substrate-binding pocket for the first time (Domsic et al., 2008 ; PDB code 3d92). The second step, which is also the rate-limiting step of catalysis, is less well defined but is believed to proceed with the transfer of a H⁺ from the Zn-bound water to a proton-shuttling residue, His64, via a hydrogen-bonded network of water molecules that span the 8 Å distance to the bulk solvent (Lindskog & Silverman, 2000 ; Cui & Karplus, 2003 ; Fisher et al., 2005 ; Fisher, Maupin et al., 2007 ). His64 sits on the edge of the cone-shaped active site and is ~8 Å away from the ZnOH⁻/H₂O. In the crystal structures of HCA II determined at various pH values it has been observed that this proton shuttle can occupy two distinct conformations. The two conformations are the so-called ‘in’ and ‘out’ positions and it has been suggested that flexibility is a requirement for efficient PT from the solvent network to the bulk solvent (Nair & Christianson, 1991 ; Fisher et al., 2005 ; Maupin & Voth, 2007 ; Silverman & McKenna, 2007 ). There are several important residues (Tyr7, Asn62, Asn67, Thr199 and Thr200) that line the active site that participate in hydrogen bonds with these water molecules and it is thought that they are important for ordering the network (Fisher, Tu et al., 2007 ).

Time-of-flight wavelength-resolved Laue images were collected at room temperature on a Huber κ-circle goniometer at 27 usable settings, with approximately 32 h exposure time per diffraction image (Fig. 2 ). The crystal-to-detector distance was kept fixed at 730 mm, which corresponds to the cylindrical radius of the detector. Owing to the limited vertical span of the detector (120° horizontal span and 16° vertical span), the crystal was reoriented seven times using the κ and ω goniometer circles and ϕ scans were performed at each crystal orientation. Each image was processed using a version of d*TREK (Pflugrath, 1999 ) modified for wavelength-resolved Laue neutron protein crystallography (Langan & Greene, 2004 ). The integrated reflections were wavelength-normalized using LAUENORM (Helliwell et al., 1989 ) and then merged using SCALA incorporated into CCP4i (Evans, 2006 ; Diederichs & Karplus, 1997 ; Weiss & Hilgenfeld, 1997 ; Collaborative Computational Project, Number 4, 1994 ). The ‘tails’ of the wavelength range were cut off slightly, with a restricted range of 0.7–6.5 Å, from the original 0.6–7.0 Å wavelength distribution of the thermal neutrons so as to eliminate the least accurately measured reflections. The overall completeness was 82% to 2.4 Å resolution, with an R _merge of ~23% and a redundancy of 2.8. The multiplicity value was as expected for the low-symmetry space group P2₁ (Table 1 ). In fact, the diffraction pattern extended to about 2.2 Å resolution and 20 additional frames, requiring a further 30 d of neutron beam time, will be collected to complete the data set to this resolution as beam time becomes available on the PCS.