- SpectroLight 600/610
- Physical background
- Technical data
SpectroLight 600/610 a fully automated IN-PLATE DYNAMIC LIGHT SCATTERING INSTRUMENT, ADDITIONALLY EQUIPED WITH UV AND WHITE LIGHT IMAGING CAPABILITIES
SpectroLight 600/610 is designed to apply in situ dynamic light scattering (in situ DLS) to standard 96 well format (SBS-format) plates. In situ DLS is the ideal method for sample qualification, storage buffer identification, additive screening, long time stability analysis and many more applications. In membrane protein biochemistry, its applications are specific detergent identification for PDC formation, nanodisc/protein assembly and protein solubility. SpectroLight 600/610 provides the most material, time and man power efficient DLS approach currently available used for sample stability analysis and quality check, aging, oxidation, temperature denaturation, and chemical degradation. Furthermore it can be used to detect nucleation non-invasively long before crystals appear. The user can track each drop‘s progress towards crystallization.
The instrument provides:
- A fully automated non-invasive in-drop DLS and imaging system designed to operate on standard plates to perform high throughput DLS- measurements in sub-microliter volumes (min. vol. ~100 nl - max. vol. no limit).
- Assess homogeneity and assembly state e.g. in pre-crystallization sample analysis in sub-microliter sample volumes.
- Particle radius assignment in a range of 1 nm to ~5 µm.
- An automatic DLS-laser positioning algorithm.
- Single or serial measurements on a variety of conditions in parallel to yield a maximum of informations for biomolecule characterization.
- Monitoring of the ongoing crystallization process in SBS-format standard screening plates.
- Bright light imaging with 2 µm resolution enabled by a bulit in microscope.
- SpectroLight 610 combines in situ dynamic light scattering, white light imaging and intrinsic fluorescence imaging by connection of the external UV-light source XtalLight 100. Intrinsic fluorescence imaging ist mostly applied for rapidly recording of intrinsic fluorescence images mostly used to distinguish salt from protein crystals.
- No cleaning steps are involved in applying the instrument
- The software provides an integrated Laboratory Information Management System (LIMS) that stores all data in a SQL-data format. LIMS provides an easy access to all stored data. LIMS manages recording and cross referencing of all data.
- The software provides various DLS-results depiction and analyzation tools as well as comfortable data export functions.
- The instrument is capable to apply in situ DLS to non-SBS-standard crystallization equipment e.g. Terasaki-plates, capillaries and micro- channel chips. Depending on the dimensions of the equipment, those applications may require a special designed adapter.
- DLS-measurements are possible in capillaries of various diameters in a range of 0.7 to 0.1 mm even when they are embedded in a GCB- Domino for counter diffusion experiments.
Proteins solutions were loaded on a terasaki plate, sample volumes ~1µl, covered with paraffin oil.
|Figure 1.1||In situ DLS radius distribution plot of mistletoe lectin II (58 kDa) after elution from a affinity chromatography in its elution buffer.|
|Figure 1.2||In situ DLS radius distribution plot of mistletoe lectin II (58 kDa) after dialysis against the storage buffer.|
|Figure 1.3||In situ DLS radius distribution plot of mistletoe lectin II (58 kDa) in its storage buffer centrifugated for 40 minutes at 12000 g. A 9.1 nm (approx. hexameric) fraction and a ~150 nm fraction remained.|
|Figure 1.4||In situ DLS radius distribution plot of the pancreatic elastase (27 kDa) in its storage buffer shows a rh of 2.3 nm. The protein is ready for crystallization.|
|Figure 1.5||In situ DLS radius distribution plot of the elastase Inhibitor ShPI-1 a 5.1 kDa polypeptide, rh 1.1 nm.|
|Figure 1.6||In situ DLS radius distribution plot of a RNA hexamer under steady state conditions with a rh of 9.8 nm it appears to large to be in a monomeric state.|
|Figure 1.7||In situ DLS radius distribution plot of lysozyme 14 (kDa, 10 mg/ml) in 50 mM sodium acetate buffer, pH 4.5. The protein shows a rh of 1.2 nm.|
|Figure 1.8||In situ DLS radius distribution plot of Mistletoe lectin I (58 kDa, 1.2 mg/ml) in its storage buffer, rh 3.2 nm.|
Proteins solutions were loaded on a terasaki plate, sample volumes ~800 nl, covered with paraffin oil.
|Figure 2.1||In situ DLS radius distribution plot of mistletoe lectin III (56 kDa) in its storage buffer after 3 weeks storage. A ~6.8 nm (approx. hexameric) fraction and a 58 nm fraction was present.|
|Figure 2.2||In situ DLS radius distribution plot of mistletoe lectin III (56 kDa) in its storage buffer after 3 weeks storage. A ~6.8 nm (approx. hexameric) fraction and a 58 nm fraction was present.|
|Figure 2.3||In situ DLS radius distribution plot of a prion protein sample indicated fast growing of the aggregates.|
|Figure 2.4||In situ DLS radius distribution plot of Protein A after addition of salt. The protein had a rh of 3.8 in the beginning and was continiously measured over 6 days. During that time the protein formed aggregates of more than 100 nm.|
|Figure 2.5||In situ DLS radius distribution plot of glucose isomerase in its standard buffer measured over 7.5 h.|
|Figure 2.6||In situ DLS radius distribution plot of the elastase ShPI-1 compex (33.5 kDa). The measured rh was with 3.6 nm significantly larger to elastase alone (rh 2.3 nm, Figure 1.4). The 3D structure reveald that the inhibitor sticks out of the elastase and might be therefore influence the diffusion constant significantly.|
|Figure 2.7A||In situ DLS radius distribution plot of protein S after dialysis against an unoptimized storage buffer (rh ~10 nm).|
|Figure 2.7B||In situ DLS radius distribution plot of protein S after storage buffer optimization (rh ~7 nm) by means of dialysis against buffer with a different pH.|
|Figure 2.8A||In situ DLS radius distribution plot Protein N after elution from an affinity chromatography column in its elution buffer. In situ DLS revealed serveral aggregation states but also amounts of smaller molecules and oligomers between 1 to 10 nm radius.|
|Figure 2.8B||In situ DLS radius distribution plot Protein N after dialysis against the its storage buffer and centrifugation. Most of the aggregates could be removed however protein N is oligomerized. The oligomer shows a rh of 10 nm.|
|Figure 2.8C||In situ DLS radius distribution plot Protein N after storage buffer variation. The protein is now monodisperse with a rh of ~2.2 nm.|
Protein samples were loaded on terasaki plates covered with a 50:50 mixture of paraffin oil and silicon oil. The oil cover allowed access to the sample drop, therefore buffer addition was possible in order to crystallize the proteins. Sample solution volume ~800 nl + 800 nl.
|Figure 3.1A||In situ DLS radius distribution plot of mistletoe lectin I, a heterodimeric 58 kDa glycoprotein, rh 5.7 nm under steady state conditions.|
|Figure 3.1 B||In situ DLS radius distribution plot of mistletoe lectin I after precipitant addition. A strong nucleation signal could be detected. The drop was analysed with in situ DLS over a time period of 4 h.|
|Figure 3.1C||Summary of mean count rates of scattered light photons measured by in situ DLS of the same experiment. A characteristic boost from ~180 kHz to ~1.5 MHz could be observed after precipitant addition because of increased hydrodynamic radii but the drop remained clear for at least 5 days.|
|Figure 3.1D||Image of mistletoe lection I crystals of another batch crystallization experiment with similar conditions. Here crystals appeared after 5 days.|
|Figure 3.1E||In situ DLS radius distribution plot results of a mistletoe lectin I experiment with a too low precipitant concentration. The drop remained clear even after weeks probably due to the fact that the nuclei do not increased their size anymore.|
|Figure 3.2A||In situ DLS radius distribution plot of lysozyme under steady state conditions. The rh vary in dependence of the present NaCl concentration it is now measured to 2.1 nm.|
|Figure 3.2B||In situ DLS radius distribution plot of lysozyme after precipitant addition A relatively strong nucleation signal could be detected. The rh of the nuclei was detected at 100 nm at their first appearance.|
|Figure 3.2C||Image of the resulting lysozyme crystals. Needle shaped crystal of lysozyme crystals indicated to fast to intensive nucleation events.|
|Figure 3.2D||In situ DLS radius distribution plot of lysozyme after precipitant addition but in a vapor diffusion experiment not as batch crystallization for comparison.A relatively much weaker nucleation signal was be detected and not directly after precipitant addition.|
Some measurements have been done in mrc2 plates some in hanging drops plates.
|Figure 4.1A||In situ DLS radius distribution plot of a hanging drop crystallization experiment of crolectin from Crocus vernus after precitant addition.|
|Figure 4.1 B||Image of the drop 24 h after precipitant addition.|
|Figure 4.2A||In situ DLS radius distribution plot of a hanging drop crystallization experiment of the elastase ShPI-1 inhibitor after precipitant addition. No crystals have been.|
|Figure 4.3A||In situ DLS radius distribution plot of pancreatic elastase after precipitant addition.
|Figure4.3B||Image of the drop one week after precipitant addition.|
|Figure 4.4A||In situ DLS radius distribution plot of IRIP a type I RIP from I. hollandica after precipitant addition. A relatively weak second peak could be detected interpreted as nucleation signal.
|Figure 4.4B||Image of the drop 2 weeks after precipitant addition. Crystals of 0.2 mm in diameter have been discovered.|
|Figure 4.4C||In situ DLS radius distribution plot of the same IRIP crystallization drop. The laser was positioned between the crystals.
|Figure 4.5A||In situ DLS radius distribution plot of mistletoe lectin I after precipitant addition. A second peak of increasing intensity could be detected interpreted as nucleation signal.
|Figure 4.5B||Image of the drop thee days after precipitant addition. Crystals of 0.3-0.4 mm in diameter have been discovered.
|Figure 4.6A||In situ DLS radius distribution plot of plasmodium falciparum glutathion S-transferase after precipitant addition. A second peak of increasing intensity could be detected interpreted as nucleation signal.
|Figure 4.6B||Image of the drop five days after precipitant addition. Crystals of 0.4-0.7 mm in diameter have been discovered.|
Protein Detergent Complex Identification based on "empty" micelle and the same detergent solution in the presence of a membrane protein.
Liposomes are spherical vesicles with at least one lipid bilayer and have an aqueous solution core. Their sizes depend on buffer conditions but more importantly from their lipid composition and number of lipid bilayers. The liposomes investigated here belong to the unilamellar type that have just one lipid bilayer. If detergent is added to the buffer it removes lipids from the liposome. Depending on the detergent concentration, the liposomes may completely dissappear. This process can be monitored with in situ DLS even when the sample is covered with paraffin oil.
Lipid loaded Micelles are silghtly larger than the micelles of the pure detergent due to the larger required space of the integrated lipids. Size differences have been determined as ~ 3.2 nm.
Pure detergent Micelles here of micelles dodecyl-maltoside are highly uniform and compact spherical objects. Their precise size can be exploited to identify loading and protein detergent complexes.
In situ DLS analysis of detergent solutions as ~800 nl sitting drops covered with paraffin oil. Even detergent-micelles could be measured when coverd with oil .
|Figure 6.1||Examples for membrane proteins and detergents that made them water soluble for a successful crystallization.|
|Figure 6.2||In situ DLS radius distribution analysis of the non-ionic detergent CYMAL-4.|
|Figure 6.3||In situ DLS radius distribution analysis of the non-ionic detergent n-nonyl-β-D-glucopyranoside.|
|Figure 6.4||In situ DLS radius distribution analysis of the non-ionic detergent n-decyl-β-D-glucopyranoside.|
|Figure 6.5||Monitoring of long time aging effects on detergents in solution on n- Octyl-β-D-glycopyranoside.|
|Figure 6.6||Monitoring of long time aging effects on detergents in solution on Sucrose monododecanoate.|
|Figure 6.7||Scattered light intensity dependence on micelle concentration.|
Examples for membrane proteins and detergents that made them water soluble for a successful crystallization:
|Figure 8.1A||In situ DLS radius distribution plot of the pancreatic elastase (27 kDa) in its storage buffer (rh 2.3 nm).|
|Figure 8.1B||In situ DLS radius distribution plot of the elastase Inhibitor SHPI-1 a 5.1 kDa polypeptide (rh 1.1 nm).|
|Figure 8.1C||In situ DLS radius distribution plot of the elastase-ShPI-1-complex. The complex crystallized and the structure was solved to resolution of 2.0 Å (pdb ID: 3UOU).|
|Figure 8.2A||In situ DLS radius distribution plot from a long time measurement (25 h) of protein MK (native). DLS revealed an aging effect of the protein idicated by increasing aggregation. The effect could also be documented by a count rate plot of scattered light photons (Figure 8.2B).|
|Figure 8.2B||Summary of mean count rates from the in situ DLS measurements of protein MK (native). In consequence of the protein aging, the aggregation of protein MK increased and accordingly the count rate of scattered light photons increased too (Figure 8.2B).|
|Figure 8.2C||In situ DLS radius distribution plot from an even longer time measurement (60 h) of protein MK (native). The aging effect of protein MK proceeds and could be clearly detected as aggregation of the protein. Again, the effect could be documented as increasing count rate of scattered light photons (Figure 8.2D).|
|Figure 8.2D||Summary of mean count rates from the extended long time in situ DLS measurements of protein MK (native). Since the aging drives the aggregation, protein MK aging could be monitored by an increasing count rate of scattered light photons increased too (Figure 8.2D).|
|Figure 8.3A||In situ DLS radius distribution plot of protein MK with the chemical compound ξ in a 1:30 molar ratio in a long time experiment. Compound ξ is a organo-halogenic small molecule with the potential to unfold or denaturate protein MK. Unfolding or denaturation results in an increased aggregation that occurs faster dependent of the concentration ratios. This effect could be clearly detected by in situ DLS.|
|Figure 8.3B||Summary of mean count rates from the extended long time in situ DLS measurements of protein MK with compound ξ in a 1:30 molar ratio. The count rate increased significantly stronger than compared to Figure 8.2D because of the compound ξ induced unfolding or denaturation.|
|Figure 8.4A||In situ DLS radius distribution plot of protein MK with the chemical compound ξ in a 1:60 molar ratio. Compound ξ is a organo- halogenic small molecule with the potential to unfold or denaturate protein MK. Unfolding or denaturation results in an increased aggregation that occurs faster dependent of the concentration ratios. This effect could be clearly detected by in situ DLS.|
|Figure 8.4B||Summary of mean count rates from the extended long time in situ DLS measurements of protein MK with compound ξ in a 1:60 molar ratio. The count rate increased significantly stronger indicated by a steeper slope compared to Figure 8.3B.|
|Figure 8.5A||In situ DLS radius distribution plot of protein MK with the chemical compound ξ in a 1:90 molar ratio. Compound ξ is a organo- halogenic small molecule with the potential to unfold or denaturate protein MK. Unfolding or denaturation results in an increased aggregation that occurs faster dependent of the concentration ratios. This effect could be clearly detected by in situ DLS (see also Figure 8.3A and 8.5A).|
|Figure 8.5B||Summary of mean count rates from the extended long time in situ DLS measurements of protein MK with compound ξ in a 1:90 molar ratio. The count rate increased significantly stronger indicated by a steeper slope compared to Figure 8.4B.|
Bright light imaging is enabled by a built-in microscope down to a optical resolution of 2 µm. Visual inspection in combination with in situ DLS is a powerful combination for crystallization sample analysis since both methods complement each other and therefore the instrument covers a resolution range from 1 nm to ~ 4 mm.
|Figure 9.1A||Crystals of thermolysin iluminated whith bright light in a mrc2 plate at zoom step 0 that offers a field of view of 8.7 x 8.7 mm.|
|Figure 9.1B||The same well but at zoom step 1 that offers a field of view of 4.4 x 4.4 mm.|
|Figure 9.1C||The same drop at zoom step 2 with a field of view of 2.7 x 2.7 mm.|
|Figure 9.1D||Maximum maginfication at zoom step 4 with an optical resolution of about 2 µm and a field of view of 0.9 x 0.9 mm.|
• Fully automated plate imaging system
• Temperature controlled internal chamber
• Built-in laboratory microscope
• Modular set-up for optional fluorescence imaging
• Data management based on a SQL data base
• Fast imaging (a full plate in less than 80 s)
• Convenient export functions
The integrated Laboratory Information Management System, a relational data base for data aquisition and management.
|Figure 11.1A||All reccorded DLS-data are stored in a Laboratory Information Management System (LIMS) a SQL-database that is part of the operating software package "SpectroCrystal". An overview of all DLS-measurements or images of a plate (Figure 11.1B) is provided at a glance by the user interface design.|
|Figure 11.1B||Overview of all images of a plate at a glance).|
|Figure 11.2A||LIMS offers an easy access to all stored images or DLS-data (Figure 11.2B).|
|Figure 11.2B||LIMS offers an easy access to all stored DLS-data just by clicking of the icon of a measurement series.|
|Figure 11.3A||Comparison tools for measurement series are provided. Either plotted separately or summarized in one plot (Figure 11.3B).|
|Figure 11.3B||Summarized radius distribution comparison plot (Figure 11.3B).|
|Figure 11.3C||Screen composition information input files in *.xml and *.xls formats could be deposited and referenced to each individual plate.|
PHYSICAL BACKGROUND of SpectroLight 610 an IN-PLATE DYNAMIC LIGHT SCATTERING UNIT
SpectroLight 610 is equipped with a 100 mW laser diode (λ = 660 nm, red). The laser beam is focused into the sample solution (e.g. a crystallization drop in a well, (Figure 1A and B). Therefore , the system is equipped with a high precision X, Y, Z adjusting mechanics.
A detector optics is focused and aligned to laser beam as well (Figure 1A and B). All optical elements are integrated in an optical head (not shown). Particles that diffuse through the cross over point of both light paths could be detected. Even such a small volume is representative for the overall radius distribution of particles in the whole sample solution.
Scheme depiction of the in situ DLS optics
The laser beam (red) is focused into the sample drop. The detector optics is focused as well. That makes the optical path of the detector visible it is indicated by a green beam.
Image section of the in situ DLS optics scheme depiction.
In order to detect particle radii, the cross over point of the laser and detector optics has to be positioned in the sample drop volume.
Image of a Terasaki plate well with a laser beam passing through a sample drop of a protein solution. The sample solution is covered with paraffin oil. As long as the beam goes through air, the laser light is invisible. It becomes visible when the light goes through the sample drop because of the isotropic scattering properties of the protein particles. The detector beam is aligned to the laser beam, but invisible.
Scattering effect from particles e.g. proteins in solution
The detector optics guides the scattered light to the detector by light fibre. The detector reads out intensity fluctuations over different time intervals from a few hundred nanoseconds to several seconds. Every protein molecule forms a center of light scattering. If a particle is smaller than the wavelength of the incident light (<λ/4), the scattered light intensity distribution is isotropic and therefore angle-independent. Fluctuations in the scattered light intensity occur as the distance between moving particles in solution changes and these fluctuations allow conclusions about the velocities of the particles (Figure 2B).
Recorded scattered light intensity fluctuations
The random fluctuation of the count-rate is caused by interference fluctuations of scattered photons and represents the raw data, while the following formulas – ACF and Stokes-Einstein equation – use this information to calculate the hydrodynamic radius of the protein molecules in the sample.
Relative particle position dependency of the scattered light intensity by interference
A movement of particle 2 from A to B changes the phase shift from about 180° to 0°, i.e. from destructive to constructive interference. The required time for particle 2 to move from A to B can be calculated by the autocorrelation function (Figure 3A). The distance between A and B is given by the optical geometry described by the scattering vector (Figure 4C).
Formula of the measured autocorrelation function g(τ) of the scattered light intensity
Figure 3B: The measured autocorrelation function
The scattered light intensity is time dependent.The decay time is a measure of the intensity fluctuation velocity. For a given particle size, the ACF can be described as an exponential decay. For very small values of τ (~ 400 ns) g(τ) the scattered light intensity is almost equal to the intensity recorded at t. The reason is that the protein molecules in the detection volume had not enough time to move. However, if the time interval increases (τ ~ 10 µs) the protein molecules had time to move the distance 1/q (Equation, Figure 4C).
The autocorrelation fit function formula.
The ACF fit function could be converted to D, the diffusion constant. The mathematical expression is the autocorrelation fit function. By fitting the autocorrelation function on the measured values g(τ) the coefficients b and and D (Figure 5).
The autocorrelation fit function graph
The scattering vector (q) is variable of the autocorrelation fit function and pre-defined by the geometry θ of the optical system, the wavelength λ of the laser and the refractive index n of the surrounding medium.
Fit of the measured autocorrelation function by approximation
After fitting the ACF by least-squares methods, the coefficients b and D are calculated.
Stokes-Einstein equation: From the diffusion constant to the hydrodynamic radius rh
If the diffusion coeffcient D is known, the hydrodynamic radius rh of a particle in solution is caluclated by the Stokes-Einstein equation by solving the equation to rh. Crucial factors are the sample temperature T and the viscosity of the medium η. Especially the temperature is crucial since it also influences the viscosity.
The user interface shows all three mentioned curves at a glance. The lower curve is the number of counted photons (count-rate in MHz) at the vertical axis versus time at the horizontal axis. The measured autocorrelation function is shown as blue dots at the upper curve and approximated autocorrelation fit function is shown as well, as a red line. The main purpose of the autocorrelation function curves is to serve as a quality check criteria. The calculated hydrodynamic radii are shown in the curve in the middle in semi logarithimic scale.
An example of the influence of nucleation events on the autocorrelation function and calculated radii.
- The software comprises a laboratory Information Management System (LIMS) storing all data in a SQL-data base.
- The software provides various result depiction and analyzation tools.
- The software provides comfortable data export-functions.
- The instrument is capable to apply in situ DLS in non-SBS-standard crystallization equipment e.g. Terasaki-plates, capillaries and mircrochannels. Depending on the dimensions of the equippment, those applications require a special designed adapter.
- DLS-measurements in capillaries of various diameters in a range of 0.1 to 0.7 mm are possible even when they are embedded in a GCB-Domino for counter diffusion experiments.
Wavelength: 658 nm, optical power: 100 mW, adjustable
Photomultipliertube, dark count rate < 300 Hz quantum efficiency 5-7%, count sensitivity 1.5*105 Hz/pW
For single photon counting
Scattering angle 148°
Avalanche photodiode, higher sensitivity for wavelengths > 660 nm (optional)
Multi-tau architecture correlator to cover a wide sample time range
Sample time from 400 ns to 30 s
Total 208 channels, quasi logarithmic channel spacing
Sample concentration with standard laser (658 nm)
Minimum 0.1 mg/ml at 0.5 µl for ~30 kDa proteins and 0.3 mg/ml for ~14 kDa proteins (e.g. for lysozyme)
Maximum > >100 mg/ml
5 magnification steps: 0.63, 1.25, 2.0, 3.2, 6.4
Field of view: 10.5x7.6, 5.2x2.9, 3.3x2.5, 2.0x1.5, 1.0x0.75 mm
Resolution: 25 µm, 13 µm, 8 µm, 5 µm, 2.5 µm per pixel
CCD colour camera 1600 x 1200 pixels
other resolutions (optional)
Bright light integrated LED
UV by external light source (optional)
colour light source (optional)
Built-in temperature control
Range to 5 to 45°C (at ambient temperature 20°C)
Minimum droplet volume about 0.015 µL (manual dispensing), 0.015 µl (automated dispensing)
Particle sizes from 1nm to approx. 6 µm
Plates in SBS format
Sitting drop: e. g. MRC 96 well, Maxiplate 48 well,
Hanging drop: Cellstar
Others: Costar 3590, LCP plate
Douglas Instruments Vapor batch plates (with adapter)
customized sample holder (optional)
Table top system 650 mm x 270 mm x 450 mm (LxBxH)
Weight: approx. 24 kg
Power consumption: 115 to 230 V, 100 W
Mini PC attached to monitor (22 inch)
SpectroLight 600 software runs on Linux
Fully automated plate scanning with unique drop finding algorithm for DLS
Integrated LIMS database for storage and retrieval of images and DLS data
Control of light source parameters
Live display of camera image
Graphical histogramming software
Radius distribution 2D and 3D
Autopilot for scheduling of your individual measurement program
connection to external data base (optional)
connection to plate handling system (optional)
Karsten Dierks, Arne Meyer, Howard Einspahr and Christian Betzel
Dynamic Light Scattering in Protein Crystallization Droplets: Adaptations for Analysis and Optimization of Crystallization Processes,
Cryst. Growth Des., 2008, 8 (5), pp 1628–1634
Dominik Oberthuer, Emilio Melero-García, Karsten Dierks, Arne Meyer, Christian Betzel, Alfonso Garcia-Caballero and Jose A. Gavira
Monitoring and Scoring Counter-Diffusion Protein Crystallization Experiments in Capillaries by in situ Dynamic Light Scattering,
PLoS ONE 7(6): e33545. doi:10.1371/journal.pone.0033545
Arne Meyer, Karsten Dierks, Rana Hussein, Karl Brillet, Hevila Brognaro, andChristian Betzel
Systematic analysis of protein-detergent complexes applying dynamic light scattering to optimize solutions for crystallization trials,
Acta Crystallogr F Struct Biol Commun. 2015 Jan 1; 71(Pt 1): 75–81