Skip to main content
NCBI home page
Acta Crystallographica Section F: Structural Biology Communications logoLink to Acta Crystallographica Section F: Structural Biology Communications
. 2015 Jun 27;71(Pt 7):823–830. doi: 10.1107/S2053230X15009061

Towards time-resolved serial crystallography in a microfluidic device

Ashtamurthy S Pawate a, Vukica Šrajer b, Jeremy Schieferstein a, Sudipto Guha a, Robert Henning b, Irina Kosheleva b, Marius Schmidt c, Zhong Ren b,d, Paul J A Kenis a, Sarah L Perry a,e,*
PMCID: PMC4498702  PMID: 26144226

An X-ray compatible microfluidic crystallization platform enables in situ time-resolved serial Laue crystallography.

Keywords: serial crystallography, Laue diffraction, time-resolved protein crystallography, protein crystallization, microfluidics, photoactive yellow protein

Abstract

Serial methods for crystallography have the potential to enable dynamic structural studies of protein targets that have been resistant to single-crystal strategies. The use of serial data-collection strategies can circumvent challenges associated with radiation damage and repeated reaction initiation. This work utilizes a microfluidic crystallization platform for the serial time-resolved Laue diffraction analysis of macroscopic crystals of photoactive yellow protein (PYP). Reaction initiation was achieved via pulsed laser illumination, and the resultant electron-density difference maps clearly depict the expected pR1/pRE46Q and pR2/pRCW states at 10 µs and the pB1 intermediate at 1 ms. The strategies presented here have tremendous potential for extension to chemical triggering methods for reaction initiation and for extension to dynamic, multivariable analyses.

1. Introduction  

Developments in the field of protein crystallography continue to enable structural analysis of ever-smaller crystals and more challenging targets. However, the ultimate goal remains the direct observation of functional motions within a protein. The challenge of obtaining dynamic structural information stems from (i) the necessary X-ray exposure times for obtaining a diffraction pattern, which are typically on the order of milliseconds to seconds for monochromatic, synchrotron radiation, compared with femtoseconds to milliseconds for functional motions in proteins, (ii) the large X-ray dose required for the collection of data at multiple time points during a reaction and (iii) difficulties associated with synchronizing macromolecular dynamics within protein crystals.

Traditionally, time-resolved crystallography has been enabled through the use of polychromatic, Laue diffraction (Ren et al., 1999 ▸). At synchrotron sources, Laue diffraction enables data collection with a time resolution of ∼100 ps, the duration of a synchrotron X-ray pulse (Moffat, 1997 ▸, 2014 ▸; Graber et al., 2011 ▸). Polychromatic radiation allows a significant fraction of the reflections (reciprocal space) to be in a diffracting condition during exposure, resulting in significantly shorter exposure times compared with monochromatic diffraction and allowing the measurement of integrated diffraction peaks without sample oscillation (Hedman et al., 1985 ▸; Cornaby et al., 2010 ▸). More recently, ultrabrilliant X-ray free-electron lasers (XFELs) have enabled data collection using single femtosecond X-ray pulses (Neutze & Moffat, 2012 ▸; Kupitz et al., 2014 ▸; Tenboer et al., 2014 ▸).

However, for both the Laue method and XFELs, the high X-ray flux results in significant radiation damage, a challenge that is exacerbated by the need to perform data collection at biologically relevant temperatures and map out structural changes in the time domain. The effects of radiation damage can be circumvented by taking advantage of an approach termed ‘serial crystallography’ (Chapman et al., 2011 ▸; Hunter & Fromme, 2011 ▸), which extends the concept of combining data from multiple crystals (Yonath et al., 1998 ▸; Cherezov et al., 2007 ▸; Cornaby et al., 2010 ▸; Liu, Chen et al., 2011 ▸; Liu, Zhang et al., 2011 ▸; Liu et al., 2012 ▸; Liu, Liu et al., 2013 ▸) to the limit of a single frame of data per crystal. This approach has been critical for structural biology efforts at XFEL sources, where radiation damage only allows the collection of a single diffraction image from each crystal (Hedman et al., 1985 ▸; Chapman et al., 2011 ▸; Boutet et al., 2012 ▸; Johansson et al., 2012 ▸, 2013 ▸; Smith et al., 2012 ▸; Barends et al., 2013 ▸; Demirci et al., 2013 ▸; Kern et al., 2013 ▸, 2014 ▸; Redecke et al., 2013 ▸; Hirata et al., 2014 ▸; Kupitz et al., 2014 ▸; Tenboer et al., 2014 ▸). Serial methods have been extended to analysis at synchrotron sources (Wang et al., 2012 ▸; Gati et al., 2014 ▸; Heymann et al., 2014 ▸; Stellato et al., 2014 ▸), including for Laue methods (Dejoie et al., 2013 ▸; Perry et al., 2014 ▸). However, these large-scale serial crystallography methods suffer from the need to manipulate crystals and/or from inefficient sample utilization (Chapman et al., 2011 ▸; Zarrine-Afsar et al., 2012 ▸; Liu, Wacker et al., 2013 ▸; Hunter et al., 2014 ▸; Lyubimov et al., 2015 ▸).

For time-resolved studies, detection of transient structural intermediates requires a rapid triggering event, or pump (for example, light or a temperature jump), to synchronize structural changes within a crystal that are then probed by short X-ray pulses. Traditionally, one crystal or a small number of crystals are used for data collection, which requires multiple pump–probe cycles on the same crystal volume for the collection of complete data at different time points and crystal orientations. This is feasible if the reaction is reversible and the protein undergoes a complete reaction cycle and naturally resets to the initial state before the next pump–probe cycle (Neutze & Moffat, 2012 ▸; Ren et al., 2012 ▸; Schotte et al., 2012 ▸; Jung et al., 2013 ▸; Schmidt et al., 2013 ▸; Kupitz et al., 2014 ▸; Moffat, 2014 ▸). Unfortunately, many reactions in crystals are either irreversible or are effectively irreversible owing to radiation damage, slow timescales for reversion to the initial state or limitations of the crystal lattice (Moffat, 1997 ▸, 2014 ▸; Graber et al., 2011 ▸; Miller, 2014 ▸). To facilitate studies of irreversible processes, a large number of crystals or crystal volumes is necessary since only one trigger event is possible for each fresh crystal (volume). This also requires methods for rapid sample exchange, and standard methods for room-temperature crystal mounting are not appropriate for such data collection. As a result, even when a means for rapid triggering is available, many important biological systems have been resistant to classical time-resolved crystallography. However, serial time-resolved approaches have recently been demonstrated using laser photoinitiation and XFEL sources to analyze photosystem II and photoactive yellow protein (PYP) (Ren et al., 1999 ▸; Kupitz et al., 2014 ▸; Tenboer et al., 2014 ▸). Instead of repeated photoinitiation on the same crystal, these experiments took advantage of a liquid jet for rapid crystal delivery into the X-ray beam and averaging of data from a very large number of crystals.

Recently, we reported a low-profile microfluidic platform for protein crystallization that eliminates manual handling of protein crystals and was used effectively for the serial collection of Laue diffraction data at a synchrotron source (Perry et al., 2014 ▸) and for anomalous diffraction analysis (Perry et al., 2013 ▸). Our approach enables the reproducible growth of isomorphous crystals with low mosaicity (Guha et al., 2012 ▸; Perry et al., 2013 ▸) and provides an alternative platform for the implementation of large-scale serial crystallography with higher sample utilization than liquid (Chapman et al., 2011 ▸) or paste jets (Liu, Wacker et al., 2013 ▸) and without the need for sample manipulation associated with fixed target-based strategies (Zarrine-Afsar et al., 2012 ▸; Hunter et al., 2014 ▸; Lyubimov et al., 2015 ▸) reported previously. Furthermore, our platform is compatible with optical methods for reaction initiation, and integrated fluidic control could be used to enable chemical triggering and/or the formulation of a wide range of sample conditions for dynamic crystallography experiments (Ren et al., 2013 ▸).

Here, we present, for the first time, the application of our microfluidic crystallization platform for time-resolved data collection using synchrotron Laue radiation. We validate our approach by merging data collected from multiple crystals of PYP grown on-chip, first using repeated photoinitiation to drive the accumulation of the pB1 intermediate in a steady-state-like experiment, followed by the collection of time-resolved data to probe transient structural changes on the microsecond and millisecond timescales.

2. Methods  

2.1. Expression and purification of PYP  

Polyhistidine-tagged apo PYP from Halorhodospira halophila BN9626 was heterologously expressed and purified from Escherichia coli, reconstituted to holo PYP with the chromophore p-coumaric acid and treated with enterokinase to cleave the polyhistidine tag using procedures reported previously (Kort et al., 1996 ▸).

2.2. Fabrication and operation of microfluidic crystallization devices  

Crystallization chips were fabricated and operated as described previously (Guha et al., 2012 ▸; Khvostichenko et al., 2013 ▸; Perry et al., 2013 ▸, 2014 ▸). The microfluidic chip design consists of separate half-wells for protein/precipitant and micro-seed solutions (Fig. 1 ▸). Dedicated valve lines for each set of half-wells allow independent filling with the protein/precipitant solution and the seeding solution.

Figure 1.

Figure 1

Open in a new tab

Optical micrograph of PYP crystals grown in a microfluidic crystallization platform showing a view of four wells from a larger 96-well chip. Micro-seed solution was filled into the middle sets of half-wells and a mixture of protein and precipitant was introduced into the outside sets of half-wells. Crystals formed after mixing and overnight incubation.

2.3. Crystallization of PYP  

PYP was crystallized in microfluidic chips using previously published crystallization conditions (Yamaguchi et al., 2007 ▸). The PYP crystals were crushed in 3 M ammonium sulfate to prepare a microseed stock solution from which serial dilution was carried out by successively diluting 50% of the microseed solution into 3 M ammonium sulfate. PYP was then crystallized in 96-well microfluidic chips by microseeding. A 1:1 mixture of 20 mg ml−1 PYP (in 10 mM HEPES pH 7.4, 50 mM NaCl) and 2.6 M ammonium sulfate was loaded into one set of half-wells, while a PYP micro-seed solution in 3 M ammonium sulfate was loaded into the corresponding adjacent chamber. A microseed serial dilution of 1:12.5 from the original stock solution produced 1–2 crystals per chamber and was used in all subsequent crystallization experiments. One complete chip required 6 µl protein/precipitant solution and 6 µl microseed solution. Ammonium sulfate at pH 5.0 was used for the initial set of experiments. The pH was adjusted to 7.0 for time-resolved experiments. The isolation valve between the two half-wells was actuated for 20 min to allow mixing of the two solutions by free-interface diffusion. Crystals grew to variable dimensions after incubation overnight at room temperature (Fig. 1 ▸).

2.4. On-chip X-ray diffraction  

Data were collected in polychromatic mode at 12 keV (1.03 Å, 5% bandwidth) on the 14-ID-B beamline (Graber et al., 2011 ▸) at the Advanced Photon Source at Argonne National Laboratory. The microfluidic chips were mounted directly on the ϕ spindle of the goniometer (Fig. 2 ▸). Positioning and alignment of the chips was performed using a high-resolution camera oriented at 30° with respect to the X-ray beam and a medium-resolution camera oriented at 60°, as described previously (Perry et al., 2014 ▸). Sample visualization and positioning were performed using a graphical user interface with the in-house uControl software at the 14-ID-B beamline. Sample positioning was achieved using a click-and-translate routine coupled to the high- and medium-resolution cameras. Sample centering along the path of the X-ray beam was achieved by visually focusing the sample, taking advantage of the very small depth of field of the high-resolution camera. The interface enabled identification and alignment of all crystals on a chip followed by automated data collection. uControl interacts with the BioCARS timing module and the LaueCollect software for control of the laser pulse for photocycle initiation followed by X-ray exposure after an appropriate time delay.

Figure 2.

Figure 2

Open in a new tab

A 96-well chip mounted on the 14-ID-B beamline at BioCARS. The triggering laser and high-resolution (30°) and medium-resolution (60°) cameras are indicated. The chip is facing the high-resolution camera. The red arrow indicates the incident X-ray beam; the cyan arrow indicates the incident laser beam.

Initiation of the PYP photocycle was achieved using nanosecond pulses from a Nd:YAG/OPO laser (OPOTEK Opolette 355 II HE; 7 ns pulse duration; 20 Hz). The laser beam was focused to a spot size of 100 × 100 µm at the sample, delivering 4.8 mJ mm−2 at 485 nm. The laser beam path was perpendicular with respect to the X-ray beam (Fig. 2 ▸). Data were collected from microfluidic chips initially oriented at 30° with respect to the X-ray beam (i.e. parallel to the high-resolution camera). The robust PYP crystals permitted the potential collection of multiple frames of data from each individual crystal at room temperature (Fig. 3 ▸). For strongly diffracting crystals, light and dark data were collected at 5° intervals over the range 20–40° from stationary crystals. For complete data sets, data were merged from multiple crystals (see Tables 1 ▸ and 2 ▸). The X-ray beam was 90 × 70 µm. A MAR CCD 165 detector was used with a sample-to-detector distance of 110 mm.

Figure 3.

Figure 3

Open in a new tab

In situ Laue diffraction from a PYP crystal on-chip. Diffraction spots can clearly be observed, even when coincident with the diffuse background scattering from the polymer materials of the chip. The inset images provide more detail of the quality and intensity at high Bragg angles. The sample-to-detector distance was 110 mm.

Table 1. Crystallographic statistics for data with an accumulated pB1 state obtained using on-chip serial Laue analysis.

Values in parentheses are for the highest resolution shell.

Data set Dark Accumulated pB1
No. of crystals 5 5
No. of images 30 29
Resolution () 1001.80 (1.881.80) 1001.80 (1.881.80)
Space group P63 P63
Unit-cell parameters () a = 66.9, b = 66.9, c = 41.0 a = 66.9, b = 66.9, c = 41.0
Single reflections
R merge(F 2) 0.075 0.087
R merge(F) 0.047 0.054
F/(F) 37.0 (26.8) 32.6 (20.2)
Total observations 27306 (1766) 23432 (1333)
Unique observations 8414 (778) 8061 (710)
Multiplicity 3.2 (2.2) 2.9 (1.9)
Single and harmonic reflections
Unique observations 8500 (785) 8121 (713)
Completeness (%) 87.1 (67.0) 83.2 (60.7)
Open in a new tab

Table 2. Crystallographic statistics for time-resolved data obtained using on-chip serial Laue analysis.

Values in parentheses are for the highest resolution shell.

Data set Dark 10s 10 ms
No. of crystals 10 10 10
No. of images 22 21 22
Resolution () 1001.80 (1.881.80) 1001.80 (1.881.80) 1001.80 (1.881.80)
Space group P63 P63 P63
Unit-cell parameters () a = 66.9, b = 66.9, c = 41.0 a = 66.9, b = 66.9, c = 41.0 a = 66.9, b = 66.9, c = 41.0
Single reflections
R merge(F 2) 0.074 0.080 0.081
R merge(F) 0.043 0.046 0.047
F/(F) 37.5 (34.4) 35.2 (31.6) 34.4 (28.5)
Total observations 18517 (980) 17044 (842) 17389 (944)
Unique observations 7393 (573) 7134 (527) 7247 (522)
Multiplicity 2.5 (1.7) 2.4 (1.6) 2.4 (1.8)
Single and harmonic reflections
Unique observations 7456 (578) 7189 (529) 7301 (559)
Completeness (%) 76.8 (49.8) 74.0 (45.4) 75.3 (48.2)
Open in a new tab

Two strategies for data collection were pursued. Initial experiments were performed in a steady-state-like measurement (Genick et al., 1997 ▸) where the crystal was exposed to a repeated pump–probe sequence (pump, 5 ns laser pulse; probe, an X-ray pulse train of 18 individual X-ray pulses generated by the storage ring operating in 324 singlet mode, delayed by 1 ms from the laser pulse, with a 2 s wait time between each photocycle initiation). Here, the accumulation of 80 pump–probe sequences was used for each diffraction frame prior to detector readout. The wait time used in this experiment was insufficient for the protein photocycle to reset to the dark state at pH 5.0 (Tripathi et al., 2012 ▸), thus driving the accumulation of high concentrations of the pB1 intermediate by the end of 80 pump–probe sequences.

A second set of experiments was designed to collect actual time-resolved structural information. For these experiments, crystals were grown at pH 7.0 in order to accelerate the recovery time of the photocycle (Tripathi et al., 2012 ▸). Here, the accumulation of ten pump–probe sequences was used for each diffraction frame prior to detector readout: pump, 5 ns laser pulse; probe, an X-ray pulse train of 11 individual X-ray pulses generated by the storage ring operating in 24-bunch mode, delayed by either 10 µs or 1 ms from the laser pulse, with a 3 s wait time between photocycle initiations to enable full recovery to the ground state. From each crystal, a dark, laser-off diffraction frame was collected followed by a sequence of two laser-on diffraction frames at 10 µs and 1 ms after photocycle initiation.

2.5. Data analysis  

Laue diffraction images were processed with the Precognition/Epinorm software (Renz Research; Tables 1 ▸ and 2 ▸). Each crystal was indexed separately and data from multiple crystals were merged. PYP crystallizes in space group P63, which allows two alternative indexing conventions, which were selected randomly by the software during indexing. Prior to merging the data from different crystals, this indexing ambiguity had to be resolved. For the steady-state-like accumulation of the pB1 intermediate, this indexing ambiguity was handled manually. Data were merged for each new crystal with the previously processed crystals using both indexing alternatives: (hkl) versus (khl). The values of R merge were then compared to determine the consistent indexing alternative. For the time-resolved data we took advantage of new software developed by M. Schmidt that makes use of the Pearson (cross) correlation coefficient (Brehm & Diederichs, 2014 ▸) between intensities of symmetry-related reflections found in two diffraction patterns taking a previously known lambda curve into account. The resulting cross-correlation matrix is then diagonalized and the indexing convention is recovered from the second eigenvector. Negative entries of this eigenvector mean flip the indices from hkl to khl, while positive entries mean keep the indexing convention. The data were integrated to 1.80 Å resolution (Tables 1 ▸ and 2 ▸). This resolution was chosen both to maintain a completeness in the highest resolution shell, on the order of 50%, and based on the quality of the subsequent electron-density difference maps.

Weighted difference electron-density maps for various time points (Figs. 4 ▸ and 5 ▸) were calculated using weighted difference structure-factor (SF) amplitudes ΔF w = w[|F light(t)| − |F dark|] and phases derived from the reference PYP structure (PDB entry 2phy; Borgstahl et al., 1995 ▸). |F dark| is the laser-off SF amplitude and |F light(t)| is the corresponding laser-on SF amplitude at time t (Ren et al., 2001 ▸). The R cryst between F dark and F calc (2phy) is 19.6% without further structural refinement. The weight w was calculated as (Ren et al., 2001 ▸)

2.5.

where ΔF = |F light(t)| − |F dark| is the difference in the SF amplitudes between time t and the dark state. σ2 ΔF = σ2(F dark) + σ2[F light(t)] is the variance of ΔF determined as the sum of the variances of F dark and F light(t). The symbol 〈x〉 denotes the mean value of x in a data set. Difference electron-density maps (Figs. 4 ▸ and 5 ▸) were calculated and displayed in XtalView (McRee, 1993 ▸). Maps were contoured at ±3σ and ±5σ, where σ is the root-mean-square value of the difference electron density across the asymmetric unit. Time-resolved maps were compared with data collected from a single, large, capillary-mounted PYP crystal on 14-ID-B using the methods described in Schmidt et al. (2013 ▸). Models of the pR1/pRE46Q and pR2/pRCW intermediates were obtained from PDB entry 1ts7 and that of pB1 from PDB entry 1ts0 (Ihee et al., 2005 ▸).

Figure 4.

Figure 4

Open in a new tab

Difference electron-density map of PYP depicting the structural changes in the chromophore binding pocket, including conformational changes of the p-coumaric acid, Arg52 and Cys69 that occur owing to accumulation of the pB1 intermediate, compared with the dark state. Maps are contoured at ±3σ and ±5σ and are superimposed on a stick representation of the protein structure in the dark state (yellow, PDB entry 2phy) and the pB1 intermediate (green, PDB entry 1ts0).

Figure 5.

Figure 5

Open in a new tab

Difference electron-density maps of PYP depicting the structural changes in the chromophore binding pocket, including conformational changes of the p-coumaric acid that occur (a) 4–10 µs after photoinitiation and (b) 1–4 ms after photoinitiation. The left-hand maps (a1 and b1) show the data obtained using our microfluidic approach at room temperature. We compare this with those (right; a2 and b2) for data collected from a large single crystal mounted in a traditional glass capillary at 0°C (Schmidt et al., 2013 ▸). Although the temperature was different for the maps shown, in both cases data were collected in the time domain where the concentrations of the intermediates plateaued (Schmidt et al., 2013 ▸) and similar signal is expected for both maps. Maps are contoured at ±3σ and ±5σ and are superimposed on a stick representation of the protein structure in the dark state (yellow, PDB entry 2phy), the pR1/pRE46Q state (cyan, PDB entry 1ts7), the pR2/pRCW state (red, PDB entry 1ts7) and the pB1 intermediate (green, PDB entry 1ts0).

3. Results and discussion  

The goal of this work was to demonstrate the feasibility of time-resolved protein crystallography via serial Laue crystallo­graphy in a microfluidic device. The bacterial blue light photoreceptor photoactive yellow protein (PYP) was used as a model system since its photocycle has been very well characterized by a number of previous standard time-resolved crystallographic studies. The on-chip crystallization and the collection of serial Laue diffraction data have been reported previously (Perry et al., 2014 ▸), as has the successful collection of single-wavelength anomalous diffraction for de novo structure determination (Perry et al., 2013 ▸). These results suggested the strong potential for observing the type of small structural changes anticipated in a time-resolved diffraction experiment.

PYP crystals form in a needle-like crystal habit, with typical dimensions of 10–60 × 10–60 × 100–300 µm as crystallized on-chip (Fig. 1 ▸). For this work, multiple orientations as well as both dark and light data were collected from a limited number of crystals and were used to demonstrate the proof-of-concept feasibility of our approach. This data-collection strategy was facilitated by the stability of PYP crystals, which successfully resist significant radiation damage (Schmidt et al., 2012 ▸).

For initial experiments, a repeated pump–probe sequence was used to drive the accumulation of the pB1 intermediate. Crystallization was performed at pH 5.0, where recovery to the ground state is slow (Tripathi et al., 2012 ▸). 30 frames of laser-off and laser-on data were collected from five crystals to a resolution of 1.80 Å and merged into data sets that were 87.1 and 83.2% complete for the dark state and the accumulated pB1 state, with R merge(F) values of 0.047 and 0.054, respectively (Table 1 ▸). The resulting electron-density difference map clearly depicts the formation of pB1 (Fig. 4 ▸). The transition between the dark state and pB1 is characterized by trans-to-cis isomerization of the p-coumaric acid chromophore as well as opening of the Arg52 gate. The areas of positive and negative electron density observed in Fig. 4 ▸ match well similar results reported previously (Genick et al., 1997 ▸; Schmidt et al., 2004 ▸; Ihee et al., 2005 ▸).

Time-resolved experiments were designed to examine the regimes where pR states and the pB state are dominant. At 10 µs, the pR1/pRE46Q and pR2/pRCW states are expected, while at 1 ms pB1 is anticipated (Fig. 5 ▸). Data from 22 frames and ten crystals were merged to a resolution of 1.80 Å with completenesses of 76.8, 74.0 and 75.3% and R merge(F) values of 0.043, 0.046 and 0.047 for the dark, 10 µs and 1 ms data, respectively (Table 2 ▸). Our crystals diffracted to a similar resolution as traditional, single-crystal reports for PYP: 1.80 Å compared with 1.60–1.80 Å (Tripathi et al., 2012 ▸; Schmidt et al., 2013 ▸). Our results also match well with reports from traditional single-crystal Laue reports in terms of R merge(F) and completeness (Tripathi et al., 2012 ▸; Schmidt et al., 2013 ▸).

We compared the electron-density difference maps obtained from on-chip data with maps obtained using traditional single-crystal methods (Fig. 5 ▸; Schmidt et al., 2013 ▸). The data of Schmidt and coworkers were obtained at slightly different delay times compared with our data (4 µs instead of 10 µs and 4 ms instead of 1 ms) and at 0°C instead of room temperature. However, analysis of the time-dependent populations of intermediates from the reference data (Schmidt et al., 2013 ▸) confirms the equivalence of the two data sets, as data in both cases were collected at time delays in the plateau region at intermediate concentrations. On the microsecond timescale we observe the presence of a mixture of the pR1/pRE46Q and pR2/pRCW states. Although stronger electron-density difference signals were observed for the single-crystal data compared with our microfluidic results, we clearly observed the displacement of the S atom from Cys69 and the carbonyl group from the linked thioester group (Fig. 5 ▸ a). On the millisecond timescale we observe changes in the electron density corresponding to the change in position of the phenolic ring in the pB1 state (Fig. 5 ▸ b). As expected, the electron-density difference maps comparing the dark and pB1 state for these time-resolved data were less defined than for the experiment in which repeated pulse–probe cycles were used to accumulate high concentrations of pB1 (Fig. 4 ▸).

In comparing the electron density obtained using traditional single-crystal methods with our serial microfluidic approach, we believe that the weaker electron density observed in our experiments is the result of parameters related to the signal-to-noise ratio of the X-ray diffraction data rather than a difference in reaction initiation efficiency. In traditional methods, which utilize large single crystals (100 × 100 × 700 µm), the optical density of PYP crystals is such that incident laser light is unable to effectively initiate the photocycle through the entire depth of the sample. In these experiments, the sample was aligned such that the X-rays only probed the laser-illuminated volume near the surface of the crystal, where photoinitiation is more efficient (60–80 µm). On-chip crystallization resulted in crystals with a typical thickness on the order of 10–60 µm, which is in the range where efficient photoinitiation can be achieved throughout the entire crystal volume. Furthermore, a higher incident laser power was used in our experiments (4.8 versus 4.0 mJ mm−2) and our microfluidic chips are optically transparent and do not interfere with transmission of the laser signal. Thus, photoinitiation of the reaction should not be limiting in our experiments.

However, signal-to-noise considerations are particularly significant in the context of time-resolved experiments, where structural changes produce only small variations in the overall diffraction signal (Helliwell et al., 1998 ▸; Ren et al., 2013 ▸). Neglecting variations in crystal quality, signal to noise can be adversely affected by either attenuation of the diffraction signal itself or a mismatch in the size of the X-ray beam relative to the sample. Unfortunately, the overall size of our crystals (10–60 × 10–60 × 100–300 µm) was smaller than the X-ray beam size (90 × 70 µm), resulting in a higher background level. Matching the size of the X-ray beam to the crystal, either by growing larger crystals or by using a microfocused X-ray beam, would improve the quality of the observed diffraction signal by eliminating this source of noise. Additionally, the materials used in our microfluidic device can attenuate the observed diffraction signal. Our current microfluidic chip design has a total material path length of 127 µm compared with a maximum crystal thickness of 60 µm (Perry et al., 2014 ▸). This achieves a path-length ratio of device materials to crystal of approximately 2:1. Further optimization of the device architecture and manufacturing procedures would enable a significantly higher signal from the crystal and/or the analysis of microcrystals. Such efforts are ongoing.

4. Conclusions  

The work reported here demonstrates the efficacy of a serial, microfluidic approach for time-resolved Laue crystallography. Laser initiation was an effective trigger of the PYP photocycle in the context of our optically transparent microfluidic devices. Small slices of diffraction data were merged to generate high-quality electron-density difference maps depicting the expected pR1/pRE46Q and pR2/pRCW states at 10 µs and the pB1 state at 1 ms. This approach takes advantage of X-ray compatible microfluidic platforms to both grow a large number of high-quality, isomorphous crystals and to facilitate efficient, high-throughput serial time-resolved Laue diffraction analysis without the need for sample handling and mounting. This method has the potential to enable the study of structural dynamics for irreversible biochemical reactions. Short Laue exposures could also help to outrun secondary radiation damage (Warkentin et al., 2013 ▸). Looking forward, the integrated fluid-handling capabilities of our microfluidic platform could be harnessed to enable chemical triggering (for example, substrate addition or a pH jump) to investigate a wide range of biologically and medically relevant protein targets that have thus far been resistant to time-resolved studies. Building on this idea, it would also be possible to extend these types of studies to multivariable analyses to expand the scope and context of the resultant dynamic structural information (Ren et al., 2013 ▸).

Acknowledgments

This work was funded by National Institutes of Health grant GM086727. Use of the Advanced Photon Source was supported by the US Department of Energy Basic Energy Sciences Office of Science under Contract No. DE-AC02-06CH11357. Use of BioCARS was supported by the National Institutes of Health National Institute of General Medical Sciences grant 1R24GM111072. MS was supported by NSF-0952643 (CAREER) and NFS-STC 1231306.

References

  1. Barends, T. R. M., Foucar, L., Botha, S., Doak, R. B., Shoeman, R. L., Nass, K., Koglin, J. E., Williams, G. J., Boutet, S., Messerschmidt, M. & Schlichting, I. (2013). Nature (London), 505, 244–247. [DOI] [PubMed]
  2. Borgstahl, G. E. O., Williams, D. R. & Getzoff, E. D. (1995). Biochemistry, 34, 6278–6287. [DOI] [PubMed]
  3. Boutet, S. et al. (2012). Science, 337, 362–364.
  4. Brehm, W. & Diederichs, K. (2014). Acta Cryst. D70, 101–109. [DOI] [PubMed]
  5. Chapman, H. N. et al. (2011). Nature (London), 469, 73–77.
  6. Cherezov, V., Rosenbaum, D. M., Hanson, M. A., Rasmussen, S. G., Thian, F. S., Kobilka, T. S., Choi, H.-J., Kuhn, P., Weis, W. I., Kobilka, B. K. & Stevens, R. C. (2007). Science, 318, 1258–1265. [DOI] [PMC free article] [PubMed]
  7. Cornaby, S., Szebenyi, D. M. E., Smilgies, D.-M., Schuller, D. J., Gillilan, R., Hao, Q. & Bilderback, D. H. (2010). Acta Cryst. D66, 2–11. [DOI] [PMC free article] [PubMed]
  8. Dejoie, C., McCusker, L. B., Baerlocher, C., Kunz, M. & Tamura, N. (2013). J. Appl. Cryst. 46, 1805–1816.
  9. Demirci, H. et al. (2013). Acta Cryst. F69, 1066–1069.
  10. Gati, C., Bourenkov, G., Klinge, M., Rehders, D., Stellato, F., Oberthür, D., Yefanov, O., Sommer, B. P., Mogk, S., Duszenko, M., Betzel, C., Schneider, T. R., Chapman, H. N. & Redecke, L. (2014). IUCrJ, 1, 87–94. [DOI] [PMC free article] [PubMed]
  11. Genick, U. K., Borgstahl, G. E. O., Ng, K., Ren, Z., Pradervand, C., Burke, P. M., Šrajer, V., Teng, T.-Y., Schildkamp, W., McRee, D. E., Moffat, K. & Getzoff, E. D. (1997). Science, 275, 1471–1475. [DOI] [PubMed]
  12. Graber, T. et al. (2011). J. Synchrotron Rad. 18, 658–670. [DOI] [PMC free article] [PubMed]
  13. Guha, S., Perry, S. L., Pawate, A. S. & Kenis, P. J. A. (2012). Sens. Actuators B Chem. 174, 1–9. [DOI] [PMC free article] [PubMed]
  14. Hedman, B., Hodgson, K. O., Helliwell, J. R., Liddington, R. & Papiz, M. Z. (1985). Proc. Natl Acad. Sci. USA, 82, 7604–7607. [DOI] [PMC free article] [PubMed]
  15. Helliwell, J. R., Nieh, Y.-P., Raftery, J., Cassetta, A., Habash, J., Carr, P. D., Ursby, T., Wulff, M., Thompson, A. W., Niemann, A. C. & Hädener, M. A. (1998). Faraday Trans. 94, 2615–2622.
  16. Heymann, M., Opthalage, A., Wierman, J. L., Akella, S., Szebenyi, D. M. E., Gruner, S. M. & Fraden, S. (2014). IUCrJ, 1, 349–360. [DOI] [PMC free article] [PubMed]
  17. Hirata, K. et al. (2014). Nature Methods, 11, 734–736. [DOI] [PubMed]
  18. Hunter, M. S. & Fromme, P. (2011). Methods, 55, 387–404. [DOI] [PMC free article] [PubMed]
  19. Hunter, M. S. et al. (2014). Sci. Rep. 4, 6026. [DOI] [PMC free article] [PubMed]
  20. Ihee, H., Rajagopal, S., Šrajer, V., Pahl, R., Anderson, S., Schmidt, M., Schotte, F., Anfinrud, P. A., Wulff, M. & Moffat, K. (2005). Proc. Natl Acad. Sci. USA, 102, 7145–7150. [DOI] [PMC free article] [PubMed]
  21. Johansson, L. C. et al. (2012). Nature Methods, 9, 263–265. [DOI] [PMC free article] [PubMed]
  22. Johansson, L. C. et al. (2013). Nature Commun. 4, 2911. [DOI] [PMC free article] [PubMed]
  23. Jung, Y. O., Lee, J. H., Kim, J., Schmidt, M., Moffat, K., Šrajer, V. & Ihee, H. (2013). Nature Chem. 5, 212–220. [DOI] [PMC free article] [PubMed]
  24. Kern, J. et al. (2013). Science, 340, 491–495.
  25. Kern, J. et al. (2014). Nature Commun. 5, 4371. [DOI] [PMC free article] [PubMed]
  26. Khvostichenko, D. S., Kondrashkina, E., Perry, S. L., Pawate, A. S., Brister, K. & Kenis, P. J. A. (2013). Analyst, 138, 5384–5395. [DOI] [PMC free article] [PubMed]
  27. Kort, R., Hoff, W. D., Van West, M., Kroon, A. R., Hoffer, S. M., Vlieg, K. H., Crielaand, W., Van Beeumen, J. J. & Hellingwerf, K. J. (1996). EMBO J. 15, 3209–3218. [PMC free article] [PubMed]
  28. Kupitz, C. et al. (2014). Nature (London), 513, 261–265.
  29. Liu, Q., Dahmane, T., Zhang, Z.-N., Assur, Z., Brasch, J., Shapiro, L., Mancia, F. & Hendrickson, W. A. (2012). Science, 336, 1033–1037. [DOI] [PMC free article] [PubMed]
  30. Liu, Q., Liu, Q. & Hendrickson, W. A. (2013). Acta Cryst. D69, 1314–1332. [DOI] [PMC free article] [PubMed]
  31. Liu, Q., Zhang, Z. & Hendrickson, W. A. (2011). Acta Cryst. D67, 45–59. [DOI] [PMC free article] [PubMed]
  32. Liu, W., Wacker, D. et al. (2013). Science, 342, 1521–1524. [DOI] [PMC free article] [PubMed]
  33. Liu, Z.-J., Chen, L., Wu, D., Ding, W., Zhang, H., Zhou, W., Fu, Z.-Q. & Wang, B.-C. (2011). Acta Cryst. A67, 544–549. [DOI] [PMC free article] [PubMed]
  34. Lyubimov, A. Y., Murray, T. D., Koehl, A., Araci, I. E., Uervirojnangkoorn, M., Zeldin, O. B., Cohen, A. E., Soltis, S. M., Baxter, E. L., Brewster, A. S., Sauter, N. K., Brunger, A. T. & Berger, J. M. (2015). Acta Cryst. D71, 928–940. [DOI] [PMC free article] [PubMed]
  35. McRee, D. E. (1993). Practical Protein Crystallography. San Diego: Academic Press.
  36. Miller, R. J. D. (2014). Science, 343, 1108–1116. [DOI] [PubMed]
  37. Moffat, K. (1997). Methods Enzymol. 277, 433–447. [DOI] [PubMed]
  38. Moffat, K. (2014). Philos. Trans. R. Soc. B Biol. Sci. 369, 20130568. [DOI] [PMC free article] [PubMed]
  39. Neutze, R. & Moffat, K. (2012). Curr. Opin. Struct. Biol. 22, 651–659. [DOI] [PMC free article] [PubMed]
  40. Perry, S. L., Guha, S., Pawate, A. S., Bhaskarla, A., Agarwal, V., Nair, S. K. & Kenis, P. J. A. (2013). Lab Chip, 13, 3183–3187. [DOI] [PMC free article] [PubMed]
  41. Perry, S. L., Guha, S., Pawate, A. S., Henning, R., Kosheleva, I., Srajer, V., Kenis, P. J. A. & Ren, Z. (2014). J. Appl. Cryst. 47, 1975–1982. [DOI] [PMC free article] [PubMed]
  42. Redecke, L. et al. (2013). Science, 339, 227–230.
  43. Ren, Z., Bourgeois, D., Helliwell, J. R., Moffat, K., Šrajer, V. & Stoddard, B. L. (1999). J. Synchrotron Rad. 6, 891–917.
  44. Ren, Z., Chan, P. W. Y., Moffat, K., Pai, E. F., Royer, W. E., Šrajer, V. & Yang, X. (2013). Acta Cryst. D69, 946–959. [DOI] [PMC free article] [PubMed]
  45. Ren, Z., Perman, B., Šrajer, V., Teng, T.-Y., Pradervand, C., Bourgeois, D., Schotte, F., Ursby, T., Kort, R., Wulff, M. & Moffat, K. (2001). Biochemistry, 40, 13788–13801. [DOI] [PubMed]
  46. Ren, Z., Šrajer, V., Knapp, J. E. & Royer, W. E. (2012). Proc. Natl Acad. Sci. USA, 109, 107–112. [DOI] [PMC free article] [PubMed]
  47. Schmidt, M., Pahl, R., Srajer, V., Anderson, S., Ren, Z., Ihee, H., Rajagopal, S. & Moffat, K. (2004). Proc. Natl Acad. Sci. USA, 101, 4799–4804. [DOI] [PMC free article] [PubMed]
  48. Schmidt, M., Srajer, V., Henning, R., Ihee, H., Purwar, N., Tenboer, J. & Tripathi, S. (2013). Acta Cryst. D69, 2534–2542. [DOI] [PMC free article] [PubMed]
  49. Schmidt, M., Šrajer, V., Purwar, N. & Tripathi, S. (2012). J. Synchrotron Rad. 19, 264–273. [DOI] [PMC free article] [PubMed]
  50. Schotte, F., Cho, H. S., Kaila, V. R. I., Kamikubo, H., Dashdorj, N., Henry, E. R., Graber, T. J., Henning, R., Wulff, M., Hummer, G., Kataoka, M. & Anfinrud, P. A. (2012). Proc. Natl Acad. Sci. USA, 109, 19256–19261. [DOI] [PMC free article] [PubMed]
  51. Smith, J. L., Fischetti, R. F. & Yamamoto, M. (2012). Curr. Opin. Struct. Biol. 22, 602–612. [DOI] [PMC free article] [PubMed]
  52. Stellato, F. et al. (2014). IUCrJ, 1, 204–212. [DOI] [PMC free article] [PubMed]
  53. Tenboer, J. et al. (2014). Science, 346, 1242–1246. [DOI] [PMC free article] [PubMed]
  54. Tripathi, S., Šrajer, V., Purwar, N., Henning, R. & Schmidt, M. (2012). Biophys. J. 102, 325–332. [DOI] [PMC free article] [PubMed]
  55. Wang, X. et al. (2012). Nature Struct. Mol. Biol. 19, 424–429. [DOI] [PMC free article] [PubMed]
  56. Warkentin, M., Hopkins, J. B., Badeau, R., Mulichak, A. M., Keefe, L. J. & Thorne, R. E. (2013). J. Synchrotron Rad. 20, 7–13. [DOI] [PMC free article] [PubMed]
  57. Yamaguchi, S., Kamikubo, H., Shimizu, N., Yamazaki, Y., Imamoto, Y. & Kataoka, M. (2007). Photochem. Photobiol. 83, 336–338. [DOI] [PubMed]
  58. Yonath, A. et al. (1998). Acta Cryst. A54, 945–955. [DOI] [PubMed]
  59. Zarrine-Afsar, A., Barends, T. R. M., Müller, C., Fuchs, M. R., Lomb, L., Schlichting, I. & Miller, R. J. D. (2012). Acta Cryst. D68, 321–323. [DOI] [PubMed]

Articles from Acta Crystallographica. Section F, Structural Biology Communications are provided here courtesy of International Union of Crystallography

RESOURCES