Date Published: March 01, 2016
Publisher: International Union of Crystallography
Author(s): Daniele de Sanctis, Chloe Zubieta, Franck Felisaz, Hugo Caserotto, Max H. Nanao.
A case study of radiation-damage-induced phasing is discussed using ultraviolet light-emitting diodes to induce specific radiation damage.
Radiation damage during an X-ray diffraction experiment occurs owing to the absorption of X-rays by electron-rich sites in the macromolecule, in certain cases resulting in damage to these specific sites preferentially (Burmeister, 2000 ▸; Ravelli & McSweeney, 2000 ▸; Leiros et al., 2001 ▸). By tuning the dose of radiation, specific damage can be induced, creating the opportunity for de novo determination of the ‘substructure’ of radiation-damaged sites (Ravelli et al., 2003 ▸) and phasing. This method of phasing was named radiation-damage-induced phasing (RIP) and can be exploited in a manner analogous to the single isomorphous replacement (SIR) method. RIP has now been used to determine phases in macromolecular crystals for over a decade (Evans et al., 2003 ▸; Ravelli et al., 2003 ▸, 2005 ▸; Banumathi et al., 2004 ▸; Schiltz et al., 2004 ▸; Weiss et al., 2004 ▸; Zwart et al., 2004 ▸; Ramagopal et al., 2005 ▸). In RIP, the changes to structure factors either derive from the loss of electron density caused by radiation damage or the movement of existing atoms to new positions, for example changes in sulfur positions in disulfide-bond breakage. In contrast to the classical isomorphous replacement methods, which require a heavy-atom-derivatized and a native crystal, RIP can be performed using a single sample. One of the major limiting factors to the widespread use of RIP, however, is the degree of general radiation damage that is incurred during the course of the experiment. General radiation damage globally affects the crystal, often resulting in increased Wilson B factors, altered unit-cell parameters, increased mosaicity and decreased resolution (Burmeister, 2000 ▸; Ravelli & McSweeney, 2000 ▸). Since even small changes to unit-cell parameters have been known since the early days of isomorphous replacement to cause large changes in structure factors (Crick & Magdoff, 1956 ▸), this presents a serious problem for maximizing specific signal. Thus, for any radiation-damage-induced phasing experiment, one must maximize specific radiation damage and minimize general radiation damage by modulating the dose, which is often a challenging goal. To this end, various analytical and experimental techniques have been developed. The minimum acceptable general damage can be estimated from prior knowledge (Bourenkov & Popov, 2010 ▸) or determined empirically (Leal et al., 2011 ▸), but assessing specific damage is more complex. Ancillary tools such as UV–visible or Raman spectroscopy are available at many synchrotron light sources, but can require specialized expertise in their operation (Carpentier et al., 2010 ▸; Owen et al., 2011 ▸, 2012 ▸). A balance between maximum specific damage and minimum general damage is also obtainable post facto (de Sanctis & Nanao, 2012 ▸). Even with these tools, however, this can be a difficult balance to achieve, and various methods have been developed in order to maximize the signal (owing to specific damage) to noise (owing to global damage) ratio. Furthermore, although specific damage that maximizes the RIP signal typically occurs around ∼2 MGy (de Sanctis & Nanao, 2012 ▸), it can be difficult to properly calculate X-ray absorbed dose, even with calibrated diodes (Owen et al., 2009 ▸) and sophisticated new dose-calculation software (Zeldin et al., 2013 ▸), particularly in the now common situation where the crystal is much larger than the beam. Often, the exposure time required to induce sufficient damage for phasing is overestimated, resulting in poor phases and difficult substructure solution. RIP using UV light has previously been shown to cause much less general radiation damage and to produce effects with a different mechanism from X-ray-induced damage. For these reasons, UV light offers an attractive alternative to X-ray-induced radiation damage. This technique has been termed ultraviolet radiation-damage-induced phasing (UV-RIP; Nanao & Ravelli, 2006 ▸). UV-RIP is typically performed with a laser at a wavelength of 266 nm. The primary source of differences in structure factors in UV-RIP is the disruption of disulfide and thioester bonds, as well as a reduction of the occupancy of some heavy atoms, such as selenium (de Sanctis et al., 2011 ▸). While it has been shown that UV lasers are capable of inducing specific damage to macromolecular crystals, the technique can be limited by the costs associated with a dedicated UV laser setup, the technical requirements for precise laser-to-sample alignment, the attenuation of laser intensity owing to the use of fibre-optic cables and the specific safety requirements engendered by any laser experiment. Recently, however, high-power light-emitting diodes (UV-LEDs) at a variety of peak emissions in the UV range have become available at low cost. UV-RIP using UV lasers or diodes expands the repertoire of RIP techniques and provides an alternative to X-ray-induced radiation damage. Furthermore, UV-LEDs represent an excellent opportunity for the inexperienced RIP practitioner to experiment with the method because of their relatively low cost, high power and ease of alignment. Indeed, our results using these inexpensive and commercially available UV diodes gave similar results to previously published reports using more sophisticated UV laser setups. It should be noted that these results should also be achievable on home sources (Pereira et al., 2013 ▸). Here, we use UV damage to insulin crystals by high-power UV-LEDs as a case study for RIP and provide a detailed workflow of a typical UV-RIP experiment.
The workflow for solving the radiation-damage substructure from UV-induced damage is analogous to the solution of any RIP substructure. A UV-RIP experiment can be broken down into four distinct phases: (i) experimental setup, (ii) maximization of the difference signal before and after UV exposure, (iii) substructure solution and (iv) phasing, as discussed below.
Here, we demonstrate the steps necessary for RIP and show that UV-LEDs can be used to introduce sufficient specific radiation damage to determine phases. UV-LEDs are an inexpensive, easy-to-align and high-power alternative to UV lasers. They are versatile and simple enough to be incorporated into home-source systems and offer a method to induce radiation damage to the sample in a controlled and time-efficient manner compatible with virtually any diffraction setup. Indeed, the relative ease of inducing large differences in sulfur occupancies should be taken into consideration when using any technique that exposes crystals to UV light. Furthermore, UV-LEDs are available in a wide range of peak wavelengths, offering the exciting possibility of identifying different radiation-sensitive groups. The relatively wide spectral bandwidth results in a power output comparable to that of some UV lasers, and since our goal in using UV-LEDs is phasing, the increased power at the expense of bandpass is an acceptable compromise. UV-RIP and RIP in general proceed with the familiar steps used in other experimental phasing methods: analysis of the magnitude of the differences between reflections, substructure determination and finally phase calculation and improvement. While UV-RIP and RIP are closest in procedure to the single isomorphous replacement method, they differ in two key respects. Firstly, conventional scaling methods frequently overestimate the contribution of the damaged data set. This necessitates a slight downweighting of the damaged data set. Secondly, RIP substructures are generally comprised of many weak sites, and in many cases also contain negatively occupied sites. UV-RIP protocols reduce the number of weak sites somewhat; however, there are generally still many more weak sites than in an SIR experiment that has strong, but relatively few, sites in the substructure. These differences are not only of academic interest, but require the use of practical measures, specifically the rigorous iterative improvement of substructure by rounds of phase improvement and difference Fourier analysis. Despite these difficulties, RIP and UV-RIP also have some key advantages over SIR, including the potential for very high isomorphism, since the two data sets can be taken from the same crystal and indeed at the exact same position in the crystal, no requirement for direct chemical modification of the crystal and the ability to tune the amount of signal by changing the UV ‘burn’. Finally, UV-RIP can be combined with other methods such as long-wavelength sulfur SAD in the absence of heavy atoms (Rudiño-Piñera et al., 2007 ▸) and indeed could be performed on the same crystal. We hope that the ease of use of UV-LEDs makes the adoption of UV-RIP more widespread in general.