This paper describes a portable and readily aligned online microspectrophotometer that allows the spectral characterization of macromolecular crystals prior, during and after the X-ray diffraction experiment. The device, designed and commissioned in 2004, has been made publicly available to the general European macromolecular synchrotron community. It has proven to be a vital instrument for kinetic crystallographic studies as the X-ray beam can induce specific changes to particular sites that must be deconvoluted from the structural differences relevant to biological mechanism (Mees et al.
; Colletier et al.
; Burmeister, 2000
; Weik et al.
; Ravelli & McSweeney, 2000
Ideally, a user would have a detailed knowledge of all the likely spectral changes of their system prior to X-ray data collection at a synchrotron. A monochromatic device such as that of Sakai et al.
) would then allow fast absorption changes at a particular wavelength to be followed up to high optical densities. However, in our experience, it is also greatly beneficial to have fast monitoring of the full spectrum, so that not only changes in the background signal but also novel transient and spectrally different radiation-damage-induced intermediates can be observed simultaneously.
There are a few constraints to consider if good absorption spectra are to be recorded online. An orientation of the crystal has to be found where blockage (e.g.
by the loop material) or reflection (e.g.
from the crystal surface) of the light are minimized. As the quality of the absorption spectra changes upon rotation of the spindle axis, spectra have to be collected at a reference angle if quantitative comparisons are to be made. With modern goniometers the same sample position can be achieved with high accuracy, allowing interleaved X-ray diffraction and spectroscopic data collection where spectra are taken solely at a defined crystal position and the spindle is returned to the reference angle between the collection of successive frames (Wilmot et al.
). Alternatively, the appropriate data collection protocol could be determined from thin crystals held at a stationary orientation. For strong UV/VIS absorbers, such thin samples are essential. In terms of X-ray absorption, most crystals are highly transparent, making the damage induced within the crystal largely independent of crystal depth (Murray et al.
). A thin sample could thus serve to establish an X-ray data collection protocol using the online microspectrophotometer, whereas a thicker crystal could subsequently be used to collect the best possible X-ray diffraction data.
Another important factor that determines the quality of the spectra obtained is how well the crystals have been vitrified. A good cryoprotectant must be chosen (Garman, 1999
) and dry liquid and gaseous nitrogen should be used to prevent the deposition of ice particles on the sample. Badly vitrified and opaque crystals will show poor transmission of the monitoring UV/VIS light, in particular at short wavelengths. The measurement of either X-ray-induced optical luminescence or UV/VIS-light-induced fluorescence could overcome some of the problems caused by high sample optical density.
Two commonly observed X-ray-induced radicals have been characterized: the trapped electron and the disulphide radical. Their relative lifetimes at 100 K of 26 s and 300 s, respectively, are much longer than the typical exposure time on our brightest X-ray sources.
Trapped electrons have been identified in pure water (Hart & Boag, 1962
), with an absorption maximum of 580 nm at room temperature. Taking the extinction coefficient as 18700 M
), a sample thickness of 100 µm and an OD of 0.3 (a typical value as can be seen from Fig. 2), the concentration of trapped electrons during irradiation is estimated to be around 2 mM
. This absorption accounts for the bluish hue often observed marking the path of the X-ray beam in cryo-cooled crystals in loops. The absence of trapped electrons at acidic pH might be caused by the highly efficient scavenging by protons [hydrated electron + hydronium room-temperature rate constant is 2.3 × 1010
(Buxton et al.
)]. As described above, trapped electrons decay significantly faster than disulfide radical anions.
The absorption coefficient of the disulphide radical at between 410 nm and 425 nm, 410–425
, is estimated to be 4000–9000 M
(Favaudon et al.
). If the sample is assumed to be uniform, the thickness is taken to be between 50 and 100 µm and the absorbance of the irradiated crystal to be around 0.5, the concentration of this radical may be estimated at between 35 mM
(using 4000 M
and 50 µm) and 8 mM
(using 9000 M
and 100 µm). The concentration of disulfide bonds in an orthorhombic bovine trypsin crystal is 192 mM
, which implies that as a proportion these concentrations of the radical would only be observable crystallographically if highly accurate high-resolution data were available. In general, X-ray-induced species can be detected spectroscopically long before they can be crystallographically observed.
We observed little temperature dependence of the S radical and solvated electron lifetime in the temperature range from 100 to 160 K (Fig. 6). However, the maximum ODs of the trapped electrons do show a clear temperature dependence (Fig. 6). The saw-tooth patterns of Figs. 2 and 6 can only partly be explained by considering a negative decay rate constant k
[radical] and a positive build up rate k
when the beam is on. At least an additional third rate constant would be needed to describe the observation that the maximum OD was obtained after the second image was taken in the experiment shown in Fig. 2. Such a rate constant could be dose dependent, e.g.
reflecting a reduction of the number of available sites for trapped electrons as these sites become damaged themselves. A full characterization and understanding of the reaction kinetics behind Figs. 2 and 6 awaits further experiments and analysis: the results plotted in Fig. 6 show that temperature would be an interesting parameter to include in these studies. For cryo-cooled (not transferred to oil) orthorhombic bovine trypsin crystals the amorphous solvent within the crystal channels becomes an ultraviscous liquid at its crystallization temperature of 185 K (Weik et al.
). Figs. 6(a
) compare the radical concentrations and lifetimes at temperatures well below this value. Nevertheless, the temperature dependence of the kinetics of the disulfide radical, located within the protein region, and that of the trapped electrons, located in the solvent, are strikingly different.
Since the installation of high-magnification visualization cameras on MX beamlines, it has been observed that some crystals are seen to ‘glow’ during X-ray data collection. Visible luminescence has been observed both from individual amino acids and proteins in solution (Carter et al.
) but the signal has not been collected from cryo-cooled protein crystals. Using our online set-up, XEOL spectra were obtained from cryo-cooled bovine trypsin crystals at 100 K using an excitation energy of 13.3 keV. The complex spectra which are observed result from the interaction of X-rays at energies close to the absorption edge of elements in the protein crystal and surrounding solution. Multiple overlapping bands are a feature of the emission of optically excited states in a complex mixture of organic molecules and this signature is dependent on the excitation energy. In addition, diffusion of secondary electronic excitations and self-absorption of the optical emission by the sample can also contribute to the spectrum.
XEOL is distinct from ‘standard’ UV/VIS fluorescence and represents a complex multi-electron process that has been described as an ‘X-ray photon in and optical photon out event’ (Rogalev & Goulon, 2002
). Studies using soft X-rays as the excitation source are well documented. However the detailed processes involved at higher energies (i.e.
13 keV) have not been previously described for proteins and require further study. XEOL using tunable soft X-rays has been successfully utilized to study proteins labelled with fluorescein-isothiocyanate (FITC), commonly used in biological immunofluorescence (Kim et al.
). By exciting at the C, O and N K
-edges, these studies revealed that carbon, but not oxygen or nitrogen, is coupled to the luminescence from the FITC chromophore. The simplicity of the XEOL experiment with the microspectrophotometer opens up the possibility of performing a range of new studies.
The device described in this paper is available at the ESRF and has been utilized over the last few years to collect UV/VIS and fluorescence data by over 30 groups, trained and supervised by various authors of this paper. A recent example of its use was reported by Badyal et al.
), where our instrument was used to probe iron oxidation states within the active site of a heme peroxidase. Comparison of solution and crystal spectra allowed confirmation that no structural effects resulted from cryo-protection and cryo-cooling. Online UV/VIS measurements aided in the interpretation of the MX data during significant radiation-induced haem reduction. Subtle differences in the X-ray-reduced and chemical-reduced spectra could be rationalized in the context of the structural results and further demonstrate the application of the combined techniques. Another recent example of the use of the instrument can be found by Macedo et al.
) in which a combination of UV/VIS microspectrophotometry and X-ray crystallography has been used to test the efficacy of selected scavengers in reducing the undesirable photoreduction of metalloproteins.
The online microspectrophotometer was initially conceived as a tool that, by measuring the production of optically absorbing species known to be radiolytic products, could provide a radiation damage metric that would guide diffraction data collection. Unfortunately, it is now clear that the rate of radical production is much faster than the crystallographically observable damage to the crystals. Notwithstanding this realisation, the device has become a very welcome addition to the armoury of techniques that can yield complementary evidence to elucidate three-dimensional snapshots of macromolecules at work. Its full potential has yet to be realised.