Spectral properties of commonly used media. The spectra of various commercial cover slips, films and substrates for hanging-drop or sitting-drop crystallization experiments are shown. The top-to-bottom order of the legend coincides with the line colors of the graph. We can see that the 0.1 mm cover slip from Hampton Research, the HP260 tape, the substrate from Grace Bio-Labs and the ClearSeal film attenuate UV light minimally and are thus the most effective for high-throughput screening.

Comparison of prototype UV microscopes. Each model selected demonstrates an important feature when evaluating the efficacy of a UV microscope. The star ratings indicate the featural strengths of a microscope. List prices for each microscope are as follows: microscope 1, $80 000; microscope 2, $55 000–$80 000; microscope 3, $35 000; microscope 4, $200 000. * denotes work in progress. The word turnkey means that the microscope is ready to operate without extensive training. Other device types for UV–Vis monitoring of protein crystals (not part of this study) include the DUVI, which is part of the SpetroImager-501 system that measures dynamic light scattering in situ (Dierks et al., 2008 ▶; Meyer et al., 2009 ▶), and the Minstrel HT UV (Rigaku, Carlsbad, California, USA). Other device types for X-ray diffraction screening of protein crystals in situ include the recently developed Oxford Diffraction PX scanner (Varian, Palo Alto, California, USA) and the robotic arms that hold the crystallization plate vertical to a synchrotron source (Jacquamet et al., 2004 ▶) as implemented, for example, on the FIP beamline at the European Synchrotron Radiation Facility (Roth et al., 2002 ▶).

Analyses of tryptophan residues. The environmental distribution of 2665 tryptophans from a sample set of crystal structures show that (a) tryptophans tend to be amphipathic in nature, being found in 42.9% ± 16.8% polar environments, and (b) have an average of 84.2% ± 19.2% of their surface area buried within proteins. (c) The percentage of open reading frames that are without a tryptophan is on average 18.5% ± 6.4% in the shown genomes.

A difficult test to compare the abilities of UV microscopes to detect a protein crystal. Two examples of a protein crystal are shown in hanging-drop experiments. (a) A crystal of the cytoplasmic domain of the sodium bicarbonate cotransporter (NtNBCe1) is shown. (b) Microcrystals of inositol 1,4,5-trisphosphate receptor-binding protein (IRBIT) are shown. Fluorescence (FL): the partial attenuation of UV light by a 0.96 mm glass cover slip distinguished the microscopes based on the quality of their optical hardware to detect emitted light by epi-illumination with 1 s exposures. The ability of microscope 1 to detect emitted light from the crystal is lower than the others, if the camera is indeed detecting any light. Microscope 2 is the most sensitive as is apparent from the bright and sharp images for both crystal examples. Interestingly, there seemingly appears to be a hint of external or reflected light, i.e. light that does not originate strictly from the crystal(s) and/or perhaps an enhanced sensitivity of the camera. The FL images for microscope 3 (zero gain) and microscope 4 are similar in intensity but not in clarity. Brightfield (BF): BF images are useful for identifying individual protein and salt crystals when compared with FL images. However, a few of the BF images shown have anomalies: the images from microscope 1 were actually inverted and reverted; here they have been correctly oriented using third-party software. Also, note that the BF image from microscope 2 is significantly dark or black around the rim, masking crystals (see white arrow) and prohibiting usable side-by-side images of BF and UV. Furthermore, note that both microscopes 1 and 4 do not give a one-to-one positional correspondence or register between the BF and FL images owing to the fact that there are two light sources on the camera that are mounted at different locations. Absorbance (AB): microscope 4 can also detect protein crystals by transilluminating UV light from the bottom of the tray and detecting from the top of the microscope. Note that microscope 4 clearly identifies the NtNBCe1 crystal as protein by AB, thereby suggesting that AB is a more robust method than FL. However, microscope 4 is not able to clearly identify IRBIT as protein by AB, perhaps owing to the location of microcrystals at the edge of the drop, thereby showing some limitation under these conditions (see Fig. 4 ▶).

Comparing tryptophan fluorescence with tryptophan absorbance. The cover slip holding the IRBIT protein crystal in Fig. 3 ▶(b) is now flipped, removed from the tray and directly exposed to BF (a) and UV (b, c) light microscopy. Here, the crystals are confirmed to be protein (i.e. knowing that there is no other aromatic component in the crystallization condition) by tryptophan fluorescence (bright crystals in b) and by tryptophan absorbance (black-colored crystals in c). If the crystals were salt they would appear dark in (b) and white in (c).