Concentrating on concentrators: Students design and test novel microfluidic ultrafiltration system for biological samples at the beamline
Biological solution scattering experiments are often the culmination of months, or even years of preparation. The seemingly mundane little droplets of liquid we put into the X-ray beam are rare bits and pieces of the machinery of life, painstakingly separated and purified from Nature’s unimaginably complex brew. Suspended delicately in solution, biological molecules are fussy, sensitive, and sometimes barely present at all. Researchers play a game of roulette when they try to concentrate samples enough to get useful X-ray scattering signals: not enough concentration and the signal is too weak, too much concentration and the molecules may crash out of solution becoming irretrievably lost.
Recognizing the real need for a means of concentrating samples during X-ray scattering measurements, master of chemical engineering students Manjie Huang and Melanie MacMullan teamed up to design and test a novel microfluidic sample concentrator chip based on the principle of ultrafiltration. Putting samples under gentle controlled pressure, porous membranes allow water and salt to pass through while retaining proteins and other large biomolecules. The process is similar to reverse osmosis used in many household water purification systems. But could this process work fast enough to be useful? Can wet filtration membranes be embedded in plastic chips without leaking? Would the fragile transparent X-ray windows necessary for scattering measurements survive the pressure? Is the concentration in front of the window reasonably uniform?
Students Melanie MacMullan (shown) and Manjie Huang connecting computer-controlled pressure regulation system for testing their microfluidic sample concentrator chip on G1 beamline.
Huang and MacMullan set out to answer these questions using a combination of simulation and experiment. They used computational fluid mechanics to model the transport of dilute protein samples over a porous membrane and past an X-ray window. Based on parameters obtained from test measurements of membrane patches, the results looked promising. Huang busily spent the summer at the Cornell Nanoscale Facility working out the grueling details of how to assemble and bond a leak free chip with a filtration membrane and fragile X-ray windows. If a concentration factor of 10 could be achieved in an hour or less, the technology could potentially be valuable to beamline users.
Simulation and experiment show how protein samples concentrate on prototype ultrafiltration X-ray scattering chip. As concentrated protein sweeps past the X-ray window (red color in simulated chip inset) a peak appears in the intensity of the actual scattering data (green curve).
As a synchrotron facility, CHESS is unique in being closely integrated with a major university campus. X-ray beamtime that opens up unexpectedly on short notice is rapidly filled with enthusiastic students willing to work at any hour for the chance to collect data at a world-class facility. Just such a time opened up for Huang and MacMullan during the final late evening hours of beamline setup at G1 station in the fall of 2016. After several hours assembling equipment, the test was ready to go. As the first protein signal flowed off the concentrator chip and into the beam, X-ray scattering signal started to rise as hoped for: a factor of 5 … 10 … 15 … nearly 20! With time only for a few trial runs in the early morning hours, the results were reproducible and unmistakable: the chip outperformed our wildest expectations.
Timelapse video of microfluidic ultrafiltration concentrator chip in action. Over the course of approximately 20 min, a colored macromolecule called blue dextran starts to visibly concentrate in the lower righthand part of the image.
As always with new technical advances, much work remains to convert a successful feasibility study into routine, easy-to-use technique that can be adopted by some of our many visiting biology researchers. As CHESS is a nexus of scientific activity where exchange of information is nearly as valuable as the data collected, students can and do play a vital role transforming ideas into practical solutions to important problems.
Submitted by: Richard Gillilan, MacCHESS, Cornell University 12/13/2016
Eddy Arnold of Rutgers University has been studying the HIV virus for a long time. A frequent user of CHESS, Arnold has been using the technique of X-ray crystallography to investigate the structure of HIV proteins and learn more about inhibitors of those proteins which might lead to drugs to fight AIDS. His efforts have been reported here several times (1, 2).
Jeffrey DeStefano, an HIV researcher at the University of Maryland, College Park and post-doc Gauri Nair published a paper in 2008 about their development of a new inhibitor of HIV RT (reverse transcriptase), the viral enzyme which copies viral RNA into DNA in order that it might be incorporated into the host cell's chromosomes by another viral enzyme, integrase (3). The RT inhibitor consisted of a 38 base pair piece of DNA with a particular sequence which binds tightly to the enzyme. Such a specific piece of DNA is called an aptamer.
Arnold and DeStefano then collaborated to study the aptamer using structural biology. Not only did they manage to solve the structure of the RT-aptamer complex, they found that the new inhibitor bound so tightly to RT that it stabilizes the complex and helps it to form better crystals than were previously available (4). Before this, researchers trying to crystallize RT in order to study it with X-ray crystallography had to either chemically cross-link the enzyme to its substrate, or co-crystallize it with an antibody fragment. The new system is not only easier, it provides better results. The X-ray structure was solved at the CHESS F1 beamline.
With the RT-aptamer complex in hand, Arnold and DeStefano had a new tool to study the HIV RT enzyme, and they put it to use in solving structures of the enzyme-aptamer complex in various states (5). By adding calcium, a metal which discourages the reaction catalyzed by RT, they solved a structure with an unreacted inhibitor AZTTP bound. By adding magnesium instead, they solved a structure of RT with the AZTMP inhibitor (the reaction product of AZTTP) enzymatically linked to the aptamer; but the aptamer binds so tightly to the enzyme that the aptamer does not translocate to the next position in the way it usually does. Another structure includes the inhibitor alpha-CNP, and yet another the inhibitor foscarnet (phosphonoformic acid). The structures produced by this set of experiments provides a treasure trove of information about the enzyme and the way it works, information that can be put to use in designing better inhibitors and hopefully better drugs to fight the HIV virus. Some of the data for these structures was collected at CHESS beamline F1, and some at NSLS beamline X25.
The foscarnet structure is especially intriguing. HIV evolves rapidly, and tends to becomes resistant to drugs over time. Foscarnet is used as an anti-viral therapy for cytomegalovirus and herpes simplex virus infections in some immunocompromised patients. Foscarnet is a weak inhibitor of HIV RT, but it has the curious property that when HIV develops resistance to foscarnet, it regains susceptibility to AZT, a commonly-used drug. Likewise, when HIV develops resistance to AZT, it regains susceptibility to foscarnet. So if foscarnet could be used as the starting point for designing a better inhibitor, that inhibitor would likely make a good complement to the AZT class of drugs. The hope is that the crystallographic structures provided by this research might eventually lead to such advances in AIDS medication.
 Novel Aptamer Inhibitors of Human Immunodeficiency Virus Reverse Transcriptase. DeStefano, J.J. and Nair, G.R. Oligonucleotides (2008) 18(2) 133-144, (2008)
 Structure of HIV-1 reverse transcriptase bound to a novel 38-mer hairpin template-primer DNA aptamer. Miller, M. T., Tuske, S., Das, K., DeStefano, J. J. and Arnold, E. Protein Science, 25: 46-55, (2016)
5D3G Structure of HIV-1 Reverse Transcriptase Bound to a Novel 38-mer Hairpin Template-Primer DNA Aptamer
 Conformational States of HIV-1 Reverse Transcriptase for Nucleotide Incorporation vs Pyrophosphorolysis-Binding of Foscarnet. Das, K., Balzarini, J., Miller, M.T., Maguire, A.R., DeStefano, J.J., and Arnold, E. (2016), ACS Chem. Biol., 11 (8), pp 2158-2164, (2016)
5I42 Structure of HIV-1 Reverse Transcriptase in complex with a DNA aptamer, AZTTP, and CA(2+) ion;
5I3U Structure of HIV-1 reverse transcriptase N-site complex; catalytic incorporation OF AZTMP to a DNA aptamer in crystal;
5HP1 Structure of HIV-1 reverse transcriptase in complex with a DNA aptamer and foscarnet, a pyrophosphate analog;
5HRO Structure of HIV-1 reverse transcriptase in complex with a DNA aptamer and an Alpha-carboxy nucleoside phosphonate inhibitor (alpha-CNP).
Submitted by: David J. Schuller, MacCHESS, Cornell University, 09/06/2016
Cornell Chronicle. By Tom Fleischman
The In-Situ-1 crystallization plate, developed by Mi-Te-Gen and its founder, Robert Thorne, is shown along with a patch from the SpaceX CRS-8 mission in April, on which the plate flew for experiments by drugmaker Eli Lilly.
Long-range electron transfer in the cytochrome c peroxidase and cytochrome c complex requires stringent conditions
Mitochondrial cytochrome c peroxidase (CcP) binds to cytochrome c (Cc) to break down hydrogen peroxide to water. This reaction is a series of steps that involves heme-oxygen chemistry and long-range electron transfer (ET) (Figure 1). Firstly, hydrogen peroxide reacts with Fe(III) heme of CcP to form an Fe(IV) iron oxo species [Fe(IV)=O], and oxidizes its nearby tryptophan 191 (W191) to a radical cation (W•?). Secondly, the presence of W•? facilitates the transfer of electrons from Cc proteins when Cc Fe(II) is oxidized, causing the reduction of CcP W•? to W191. The CcP Fe(IV)=O then re-oxidizes W191 back to its radical cation state, resulting in the eventual formation of Fe(III) and water.
Figure 1: W191 mediates long-range ET between the two heme centers (~16Å), resulting in two short hops instead of one long-range leap.
Since aromatic residues play a crucial role in mediating the long-range ET of many biological redox systems (Payne et al.), the authors investigated the effect of substituting W191 for other residues in the CcP:Cc complex. CcP:Cc was chosen because it is a standard model system for studying long-range ET between proteins (Volkov et al.).
Since W191 is a residue that is critical in the reaction in the CcP:Cc system, three CcP W191X variants were created (where X = Y, F or G; tyrosine, phenylalanine or glycine respectively) to test for the effect of amino acid substitution on ET. Crystal structure complexes of CcP W191X variants bound to Cc showed similar 1:1 association modes and heme pocket conformations (Payne et al.). Unlike the wild-type (WT) W191 that formed a hydrogen bond with aspartate 235 (D235), none of the aromatic substitutions (Y191 and F191) had any polar contacts (Figure 2). The absence of the indole group in the W191G variant produced a water-filled cavity, where redox-active ligands were added. Ligands that bound in the W191G pocket were structurally disordered and did not form hydrogen bonds with Asp235 either. Furthermore, electron paramagnetic resonance (EPR) spectroscopy showed that only the WT and W191Y variants could form a stable W/Y radical formation (Figure 3).
Figure 2: Superposition of the W191X variants. WT in white; W191Y pink; W191F blue; W191G yellow.
Figure 3: EPR spectra of the CcP variants. Peroxide induced the formation of radicals that was captured by EPR. WT and W191Y signals were significantly larger than the other variants (W191F not shown).
To study how the CcP W191 variants affected the long range ET to Cc, the heme in the CcP variants was substituted with zinc-porphyrin (ZnP), which ET could now be photoinduced. These variants had low ET rate compared to the WT, with the W191G being weaker than the W191F variant. As for the W191Y variant, low net reactivity towards Cc(II) could be due to the inability of: 1) the ZnP•? to produce enough W191Y radical, or 2) the resulting neutral W191Y radical to oxidize Cc(II) effectively.
In conclusion, this study showed that there must be both structure stability and effective redox potential in order to carry out significant long-range ET between the CcP and Cc proteins. At this point it seems that only WT W191 meet these requirements.
 Payne TM, Yee EF, Dzikovski B, Crane BR, Constraints on the Radical Cation Center of Cytochrome c Peroxidase for Electron Transfer from Cytochrome c. Biochemistry. 2016 Aug 30;55(34):4807-22.
 Volkov AN, Nicholls P, Worrall JA. The complex of cytochrome c and cytochrome c peroxidase: the end of the road? Biochim Biophys Acta. 2011 Nov;1807(11):1482-503.
Submitted by: Teck Khiang Chua, MacCHESS, Cornell University 09/06/2016
Six summer science workshops at CHESS followed a most memorable Users’ Meeting on June 7th when our user community first heard the big news about plans to reconfigure the accelerator and beamlines to use only a single source of particles. This will involve relocating the five experimental stations on the A, B, C, and D beamlines, and upgrading the replacement stations with independently tunable high-flux undulator sources. The goal of the workshops was to identify pressing and important scientific needs for a future high-energy x-ray source utilizing unique capabilities of the Cornell accelerator and special types of organization and user support.
This “biomolecules” workshop was held on June 8th and involved 11 invited speakers and 36 attendees. 19 remote viewers asked questions and participated in discussions through the on-line YouTube stream. The local organizing committee consisted of Richard Gillilan and Marian Szebenyi (MacCHESS) and Nozomi Ando (Princeton). This workshop focuses on motion of biomolecules (proteins, nucleic acids, complexes) which occurs as they perform their functions. “Motion” includes conformational changes, ranging from large domain “hinging” motions to relatively small loop motions (e.g. to allow and deny access to an active site); oligomerization changes; interaction with partner molecules.
The full agenda for the workshop with speakers’ names and presentation titles and abstracts are available online here: http://meetings.chess.cornell.edu/sciworkshops2016/Workshop2.html
Lee Makowski (Northeastern University) introduced first the topic that protein function depends on motion and that motion is a lot more complex than static structures. Measurements on many time and length scales will be needed, along with a strong computational framework, to understand this difficult topic. Solution small- and wide-angle scattering, time-resolved crystallography, microfluidic mixers, laser and thermal pumping systems synchronized to the storage-ring x-ray pulses, are all parts of the experimental arsenal needed to make striking new discoveries in biomolecular systems in fields as diverse as plants, animals and human disease and health.
During the workshop it grew common to discuss biomolecules as “molecular machines.” In order to understand how molecular machines work, it is necessary to look at their dynamics, i.e. how the parts move and interact when the "machine" is working. Static structures, from crystallography or cryoEM, provide snapshots which are useful in deducing how biomolecules might function, but more dynamic measurements are necessary to determine what actually happens during their operation. Workshop attendees discussed how the most promising methods for obtaining dynamic information include time-resolved SAXS/WAXS (small-angle/wide-angle X-ray scattering from solution), time-resolved serial crystallography using a large number of small crystals, and diffuse scattering from crystals. The use of anomalous scattering, obtained through measurements at multiple wavelengths, can be applied to SAXS/WAXS and diffuse scattering experiments to provide more complete information than we can get from single-wavelength experiments.
An upgraded CHESS will be a great place to pursue study of dynamics because (1) it will provide the high-flux, tightly focused beams necessary to work with small samples utilizing short exposure times and (2) the CHESS source could be configured to provide x-ray pulses with large numbers of photons and separated in time and (3) the variety of different experiments proposed fits well with the demonstrated flexibility of CHESS stations and staff.
After the meeting, CHESS scientists, organizers and participants joined forces to summarize notes and compose “white papers,” capturing the scientific need and opportunities for innovative work using an upgraded CHESS source. CHESS scientists are now working with members of the CHESS External Advisory Committee and members of the CHESS Users’ Executive Committee to refine, combine, reduce and/or sharpen the ideas captured from the workshops. The CHESS staff is enormously grateful to members of the user community and beyond who’ve helped shape the future of CHESS.
Stay tuned to the CHESS eNewsletter to hear exciting updates on the CHESS-U upgrade.
Some of the participants at Workshop 2 in the Physical Sciences Building.
Submitted by: Ernest Fontes and Rick Ryan, CHESS, Cornell University 07/13/2016
The Macromolecular Diffraction Facility at the Cornell High Energy Synchrotron Source (MacCHESS) held its sixth highly successfully BioSAXS Essentials workshop from May 13th to 16th, 2016. Thirty students from fourteen different institutions (including ones as far away as UC Irvine and the University of Puerto Rico) attended the workshop in person, and fifteen students from eleven institutions and companies attended the course remotely via WebEx and YouTube Live. Five different expert instructors gave students a day and a half of lectures and hands-on tutorials in SAXS fundamentals, data collection, and data processing. On-site students then had two and a half days of hands-on data collection at two different MacCHESS beamlines (G1 and F1), allowing them to obtain valuable practical experience and collect data from research samples.
Dr. Richard Gillilan giving the opening lecture of the BioSAXS Essentials 6 workshop. There were thirty students on site for the workshop.
The workshop kicked off on the 13th with an overview of SAXS and fundamental scattering principles by Dr. Richard Gillilan (MacCHESS), followed by an extremely thorough presentation on best practices for preparing samples for SAXS experiments by Dr. Kushol Gupta (UPENN Medical School). Dr. Gillilan closed out the morning with a two-part lecture covering the basics of SAXS data collection and analysis, with a focus on assessing/ensuring data quality at the beamline.
After lunch (and, perhaps more importantly, coffee) on the 13th, Dr. Thomas Grant (Hauptman-Woodward Institute) gave the first of his two talks on advanced processing methods. He first provided a summary of the processing Dr. Gillilan discussed in the morning session, and then focused on inverse Fourier transform methods and bead model reconstructions from SAXS data. This was followed by a discussion on best publication practices by Dr. Gillilan, which provided crucial information on how to write up and publish SAXS data and models.
Dr. Ileana González-González (Universidad del Turabo, Gurabo, Puerto Rico) prepares her sample for a SEC-SAXS experiment. Students, in groups of 2-4, got 6 hours of beamtime at stations G1 or F1 to obtain hands-on experience and training with Dr. Gillilan and Dr. Hopkins. They were able to collect data on standards and research samples.
The final lectures of the day were also on advanced measurement and processing techniques. The first of these was given by Dr. Steve Meisburger (Princeton), and covered the use of Size Exclusion Chromatography coupled SAXS (SEC-SAXS), including advanced techniques for deconvolving poorly-separated SEC-SAXS data developed at CHESS by the Ando group (Princeton). Dr. Gupta gave his second talk of the day, discussing the analysis of mixtures, flexibility, atomistic modeling, and contrast matching/SANS experiments. Finally, Dr. Grant closed out the day with an overview of rigid body modeling, hybrid techniques, and time resolved SAXS at synchrotrons and XFELs.
The morning of the 14th was devoted to hands-on data processing tutorials, run by Dr. Jesse Hopkins (MacCHESS) and Dr. Meisburger. The first tutorial covered the basics of data processing and verification at the beamline using RAW (currently supported/developed by Dr. Hopkins). The second tutorial walked students through using advanced modeling techniques, including finding the P(r) function with GNOM and using DAMMIF to reconstruct bead models from measured scattering profiles.
Dr. Arnab Modak (University of Connecticut) loads his sample by hand into the sample cell at G1.
During the night of the 13th, and around the clock from noon on the 14th to 5 pm on the 16th students collected SAXS data at the G1 and F1 MacCHESS beamlines with guidance and training from Dr. Gillilan and Dr. Hopkins (the F1 beamline was converted to SAXS for the workshop). Students collected data in groups of 2-4, with at least 6 hours of data collection per group. They had the option of carrying out either standard robotically assisted SAXS or SEC-SAXS during their beamtime. While standard proteins were available for training, most students collected data on research samples they had brought from their labs.
The F1 station is usually a macromolecular crystallography station. In preparation for the course, it was temporarily converted to SAXS use, shown here. Much of the conversion was done by Master’s students Melanie MacMullan (front left) and Manjie Huang (front right) working with Dr. Gillilan (back). Additional work was done by Dr. Hopkins and research support specialists Bill Miller and Scott Smith (not pictured).
Jesse Hopkins, MacCHESS, Cornell University 06/09/2016
Many cellular processes, particularly intercellular signaling, require enzyme-catalyzed reactions to occur inside a cell membrane. There are four known families of membrane-immersed proteases (enzymes which break protein chains); all four carry out important functions and damage to them is implicated in pathologies including cancer, Parkinson's disease, impaired resistance to parasites, and more. When a mutation results in overactivity of a membrane protease, an inhibitor of the protease can be effective treatment for a disease. Designing such inhibitors has proven difficult, largely because of incomplete understanding of the catalytic process in the intramembrane environment.
The Urban group (Johns Hopkins) is studying E. coli GlpG, which is a member of the membrane-embedded rhomboid protease family, using measurement of enzyme kinetics coupled with crystallographic structure determinations. In previous experiments, the group found that the overall rate of the reaction (clipping off a transmembrane helix in order to release a signaling protein) is determined by the time needed for the substrate to get into the active site, rather than the reaction rate in the site (, see CHESS eNews #15). More recently, the focus has shifted to examining the interaction of inhibitors with GlpG. A number of complexes of rhomboid proteases with inhibitors have been examined previously, but these have all been inhibitors that bind irreversibly to the enzyme's active site and distort it in the process. The Urban group's approach is to use short peptides with an aldehyde moiety at the C-terminus; such an inhibitor binds in the active site and forms an intermediate just like the natural substrate, but is then trapped in that state.
Figure 1: Cartoon showing stages in the operation of a rhomboid protease (blue), and its inhibition by a substrate-mimicking peptide aldehyde (red star with tail). A target protein (green) binds first to the interrogation (I) site and then transfers to the scission (S) site for cleavage.
As reported in a recent publication , kinetic studies of the inhibition of GlpG by various peptide aldehydes showed that the inhibition is non-competitive, meaning that inhibitor and substrate can both be bound to the enzyme at once. This strongly supports the model shown in Figure 1; binding of the substrate to the “interrogation” site (which is the step that determines the reaction rate) can occur whether or not there is an inhibitor molecule bound in the “scission” site. In addition to the kinetic work, several crystal structures of protease-inhibitor complexes were determined, using data collected at CHESS. For most of these, the protease molecules were embedded in bicelles, providing an environment similar to natural membranes. A mutation (conversion of tyrosine 205 to phenylalanine) improved crystallizability with minimal effect on activity. The crystal structures revealed:
- the conformation of the tetrahedral intermediate produced when substrate (or inhibitor) binds to the catalytic serine;
- specific interactions of the substrate/inhibitor with the protein. These largely involve backbone rather than sidechain atoms, consistent with the broad specificity of rhomboid proteases.
Several of the specific interactions involve regions outside the active site proper, as seen in Figure 2. This was unexpected, since there is no requirement for any particular residues in the P2 and P3 positions of substrates, and is helpful in designing inhibitors. The peptide aldehyde Ac-VRMA-CHO used in the structure study is in fact a potent inhibitor of GlpG in vivo, better than irreversible inhibitors used previously.
Figure 2: Binding of inhibitor (cyan) to GlpG (yellow). Left, GlpG Y205F mutant in bicelle membrane with Ac-VRMA-CHO bound; this is a “gate-open” form, in which substrate access to the active site is possible. Center, GlpG in detergent with Ac-VRMA-CHO bound; this is a “gate-closed” form. Right, close-up of Ac-VRMA-CHO binding.
The Urban group's work suggests a strategy for improving inhibitors of rhomboid proteases: focus on reversible inhibitors such as peptide aldehydes and systematically investigate modifications to residues previously considered “nonesssential”, i.e. those which do not contact the active site.
 S.W. Dickey, R.P. Baker, S. Cho, S. Urban, “Proteolysis inside the membrane is a rate-governed reaction not driven by substrate affinity”, Cell 155, 1270-1281 (2013).
 S. Cho, S.W. Dickey, S. Urban, “Crystal structures and inhibition kinetics reveal a two-stage catalytic mechanism with drug design implications for rhomboid proteolysis”, Molecular Cell 61, 329-340 (2016).
Submitted by: Marian Szebenyi, MacCHESS, Cornell University 03/09/2016
Macromolecules typically produce only small crystals; to observe diffraction from them (and determine the molecular structure) we need the intense, highly collimated beam from a synchrotron source. And, the tiny beam has to hit the tiny crystal. The location of the beam can be well determined and doesn't change, but how do we know where the crystal is?
The traditional method for imaging protein crystals at most macromolecular crystallography (MX) beamlines employs an on- or off-axis digital bright-field microscope. Accurate visualization of the crystal is necessary to keep it centered in the X-ray beam as it rotates during an oscillation crystallography experiment. Precise centering becomes more difficult when the visibility of the sample is obstructed by ice or excess cryoprotectant. Several methods have been developed in the past decade to address the issues related to sample visualization on the MX beamline. Among others, raster scanning the sample with an attenuated X-ray beam and UV-excited fluorescence are growing in popularity in beamline use for the visualization of microcrystals (>10 µm). However, both UV and X-ray based methods can damage the sample during imaging due to the ionizing nature of radiation at UV and X-ray wavelengths.
Figure 1: Comparison of bright field (BFM) and visible light excited fluorescence (FM) micrographs of a thermolysin crystal at the A1 station. The crystal is mounted in a nylon loop utilizing NVH immersion oil as the cryoprotectant. Crystals were illuminated with a 5 mW 405 nm CW laser and imaged through an FGL435 long-pass Schott glass filter.
A new method for imaging protein crystals was discovered at MacCHESS and is now implemented on both the A1 and F1 stations here (Figure 1). The method utilizes visible light (405 nm) to excite conjugated double bond and delocalized electron systems natively present within a protein crystal. The magnitude of fluorescence is heavily temperature dependent with about a 10-fold stronger fluorescence at cryogenic temperatures. This novel imaging method can also be used for screening crystallization trays using fluorescence confocal microscopy (Figure 2). Unlike methods relying on non-linear optics, visible light excited fluorescence does not occur with salt crystals and thus provides a mild method for discriminating between salt and protein crystals in screening trays .
Figure 2: Fluorescence confocal imaging of protein crystals in a screening tray. Left to right: lysozyme, thaumatin, thermolysin and trypsin. Crystals were illuminated with a 405nm CW laser.
 Lukk T, Gillilan RE, Szebenyi DME and Zipfel WR. “A visible-light-excited fluorescence method for imaging protein crystals without added dyes” J. Appl. Cryst. (2016). 49, 234–240
Submitted by: Tiit Lukk and Marian Szebenyi, MacCHESS, Cornell University 03/03/2016