Journal of

Structural Biology
Journal of Structural Biology 142 (2003) 180187 www.elsevier.com/locate/yjsbi

The promise of macromolecular crystallization in microuidic chips

Mark van der Woerd,a Darren Ferree,a and Marc Puseyb,*
b

Universities Space Research Association, National Aeronautics and Space Administration/Marshall Space Flight Center, Mail Code SD46, Huntsville, AL 35812, USA Biophysics SD46, National Aeronautics and Space Administration/Marshall Space Flight Center, Mail Code SD46, Huntsville, AL 35812, USA Received 5 February 2003

Abstract Microuidics, or lab-on-a-chip technology, is proving to be a powerful, rapid, and ecient approach to a wide variety of bioanalytical and microscale biopreparative needs. The low materials consumption, combined with the potential for packing a large number of experiments in a few cubic centimeters, makes it an attractive technique for both initial screening and subsequent optimization of macromolecular crystallization conditions. Screening operations, which require a macromolecule solution with a standard set of premixed solutions, are relatively straightforward and have been successfully demonstrated in a microuidics platform. Optimization methods, in which crystallization solutions are independently formulated from a range of stock solutions, are considerably more complex and have yet to be demonstrated. To be competitive with either approach, a microuidics system must oer ease of operation, be able to maintain a sealed environment over several weeks to months, and give ready access for the observation and harvesting of crystals as they are grown. 2003 Published by Elsevier Science (USA).
Keywords: Macromolecule crystallization; Microuidics; Lab on a chip

1. Introduction High-throughput methods in macromolecule crystallization are dependent upon the accurate formulation and dispensing of solutions having a well-dened composition. As the initial screening trials often use hundreds or even thousands of solutions of widely varying components one must either draw upon stock preparations of each solution or reproducibly formulate them on demand. The introduction of crystallization screening kits and methods (Cudney et al., 1994; Jancarik and Kim, 1991) has essentially formalized the initial procedures carried out by researchers when confronted with a previously uncrystallized macromolecule. This approach has also signicantly enabled the preliminary crystallization procedures for high-throughput methods. Robotic systems are able to draw from sets of standardized solutions and set up hundreds or even thousands of crystallization trials in a very short time. Crystallization
* Corresponding author. Fax: +256-544-6660. E-mail address: Marc.pusey@msfc.nasa.gov (M. Pusey).

optimization trials use a more narrowly dened range of components to form a series of solutions having systematically varied compositions, which are prepared as needed. High-throughput crystallization screening is now routinely done in nanoliter volumes. Precise and accurate solution preparation and dispensing processes that are relatively straightforward on the scale of liters or milliliters become very dicult with volumes ve or six orders of magnitude lower. NASA has also been interested in further developing its microgravity macromolecule crystallization hardware to embrace capabilities more closely approximating those found in a laboratory on Earth. Chief among its goals are a greater experimental density and the ability to conduct iterative experiments, in which subsequent crystallization experiments are planned and performed based upon the previous results. Empirical observations have suggested that crystallization conditions that are optimized on Earth are not necessarily optimized for use in microgravity as well, thus requiring either a large number of trials over a range of conditions or repetitive ight opportunities to rene the crystallization

1047-8477/03/$ - see front matter 2003 Published by Elsevier Science (USA). doi:10.1016/S1047-8477(03)00049-2

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187

181

conditions. Analysis of the ight data shows that for those proteins that are improved by crystallization in microgravity the success rate increases with the number of ights (Kundrot et al., 2000), i.e., an iterative process improves the nal crystal quality obtained. Currently microgravity experiments are sent into orbit with all parameters xed; iteration requires a second ight, which may be months or even years later. Under the present system the potential advantages of microgravity crystallization may be greatly oset by the long lead times prior to each experiment and the time between iterations. Optimization requires greater control over the crystallization solution composition than screening requires. In addition, free solutions are generally not acceptable practice in spaceight missions, and microuidics is seen as a means of implementing an iterative biocrystallization system with wholly contained solutions.

of concern. Rupp et al. (2002) point out in their recent work that the availability of protein is the true limitation given the current status of automation, and we therefore must at the very least focus on the use of the smallest possible quantities of protein to obtain crystals. This restriction on protein quantity does not only pertain to the volume of crystallization drops, but also to the equipment and processes used to set them up either through automation or by hand. The process of setup must therefore be ecient without appreciable loss of protein solution through dead volume and washing operations from start to nish. The use of microuidic technology potentially meets these requirements, as we will show below.

3. What are the characteristics of microuidics? The term microuidics correctly suggests that the technique involves liquids and small-scale devices, on the order of micrometers, in scale. Its synonymous term lab on a chip more appropriately conveys the true advantage, notably that multiple laboratory operations can be combined on the same chip. In a recent review of the functionality of microuidic systems in clinical and forensic applications, Verpoorte has shown that indeed a wide variety of analyses and other applications are not only feasible, but already in use in many cases (Verpoorte, 2002). The use of microuidic systems in enzyme assays was clearly proven to be advantageous in the demonstration by Hadd et al. of the analysis of b-galactosidase in a microuidic chip (Hadd et al., 1997). The authors demonstrate three very important points: rst, that microuidic devices are suitable for protein work; second, that very accurate delivery and measurement of enzymes can be made; and third, that although not perfect, the experiments were reproducible enough to determine MichaelisMenten constants and demonstrate inhibition by phenylethyl b-D -thiogalactoside. Later work, for example with protein kinase A (Cohen et al., 1999), shows that enzyme assays in a microuidic chip actually improve on the process compared to the macroscale laboratory operation because all operations take place within the same device, heterogeneous formats such as radiolabeling are not required, and there is potential for high-throughput automation. In addition, it was shown in enzyme assays that the handling of enzyme solutions and subsequent qualitative and quantitative enzyme activity measurements with dierent inhibitor compounds are highly reproducible (Kopf-Sill, 2002). Reproducibility to the extent shown in enzyme assays makes the technology very interesting for protein crystal growth. Another attractive property of microuidic devices, particularly desirable for macromolecular crystallization, is that the same results can be obtained using

2. Why do we need high throughput? In 2001 Venter et al. announced the completion of the rst draft of the human genome (Venter et al., 2001). Although this work is a draft and many of its implications are in some way only estimates, we have been forced into the realization that the amount of biological information available to us is, simply put, overwhelming. In their analysis of the human genome draft, Lander et al. (2001) estimate that it contains genes for over 30 000 proteins. We do not know the function of many of these proteins, and for the majority of those for which function is known we lack the detailed structural information that will enable further understanding of how that function can be explained. To further complicate the issue, the generic human genome is only the tip of the iceberg. The kind of personalized medicine suggested by Sander (2000) will require detailed mapping and analysis of an individuals genetic information. A very small part of this information has already become available by means of single-nucleotide polymorphism analysis (Sachidanandam et al., 2001). Mapping the genetic information from a large group of individuals is envisioned as the start of personalized medicine (Sander, 2000). We can therefore safely predict that the amount of genetic information will vastly increase in the near future and that the demand for structural information about the proteins encoded in it has only just begun. In many cases, understanding macromolecular function requires interpreting accurate structural information, which is reliably obtained by X-ray crystallography (Shapiro and Harris, 2000). The rate-limiting step in this technique is the production of high-quality crystals suitable for high-resolution X-ray analysis. Automated production of macromolecular crystals is therefore of great importance, but it is not necessarily the only issue

182

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187

microuidics while greatly reducing total reagent consumption. For example, the aforementioned b-galactosidase analysis consumed an average of only 120 pg enzyme and 7.5 ng substrate, resulting in a 10 000-fold reduction in reagent consumption compared to the same analysis in a macroscale laboratory (Hadd et al., 1997).

4. How does it work? Although it is not within the scope of this article to explain in exhaustive detail what the characteristics of microuidic devices are or how they are operated, it is important to explain the basic mechanics so that the reader may fully appreciate the advantages of the technology and how they apply to protein crystal growth. The volumes of solutions used inside a microuidic device must rst be considered. If we assume a microuidic channel with circular cross section, a radius of 10 lm, and a channel with a length of 1 cm, the total volume is 3.14 nl. Thus, if we were to empty the entire contents of a lled channel into a crystallization droplet, we would add 3.14 nl to it. Since it is possible to manipulate such a channel to deposit only part of its volume, perhaps onetenth of the volume, the delivery would be on the order of 300 pl. However, the volumes mentioned as an example are still greater than the lower limit volume dispensed with this technology. Intersections of channels allow injection of uid from one stream into another by rapid modulation of the uids respective driving forces, thereby allowing dispensing in amounts roughly two orders of magnitude smaller than stated above, i.e., in the range of a single picoliter (Fu et al., 2002). These volumes are very small compared to the regular laboratory crystallization setup in which we typically deliver liquids in microliter increments and are even small in contrast to existing high-throughput systems which use total dispensed volumes of 500 to 100 nl (Krupka et al., 2002; Luft et al., 2001; Rupp et al., 2002). To appreciate the next property of microuidic systems, we perform the following thought experiment: suppose we take a lled channel 1 cm in length and with 10 lm internal radius and then empty the channel contents into a droplet in 10 s. The linear ow speed is thus 1 mm/s (103 m/s). The dimensionless Reynolds number Re qvL=g 0:02, with q being the density of the medium (assume water, q 103 kg=m3 ); v the velocity (103 m/s); L the characteristic length, in this case L 2r 2 105 m; and g the viscosity of the medium (assume water, g 103 Pa s). The ow pattern of the solution in the channel is therefore strictly laminar, as the Reynolds number, which denotes the ratio of force due to inertia and the force due to viscosity, is far below the critical value of 2000 at which the ow changes from laminar to turbulent. Because the ow is strictly laminar, no turbulent mixing occurs, which is essentially the

method used for mixing bulk solutions. The only mixing of components that occurs in this system is by diusion. Intuition may incorrectly lead us to believe that therefore mixing of protein and precipitant is problematic in microuidic systems, but as the transverse mixing distance is never greater than the channel width (on the order of 10 lm), the time to completely mix protein and precipitantapproximated by Einsteins diusion equation t w2 =2D, where w is the width of the channel and D the diusion coecient of a protein, for example hemoglobin, 6.91011 m2 /s (Fu et al., 2002)is, in the example given here, approximately 0.7 s. The conclusion therefore is that mixtures containing protein, precipitants, and additives (both with considerably larger diffusion coecients and therefore much more rapid rates of diusion) can be made completely without mixing by delivering the components in sequence, or they can be thoroughly, completely, and reproducibly mixed upon delivery, to a degree that is unlikely to be achieved by manual methods or current robotic systems. One can also reproducibly obtain any degree of mixing between none and complete by control of the mixing time. Thus, microuidic systems add more value to the crystallization procedure than just small size: greater experimental control over the process is achievable than is possible in the standard laboratory. The last characteristic of microuidic systems of importance to macromolecular crystal growth is the surface-to-volume ratio. In the example above, with a volume of 3.14 nl and a channel 1 cm (104 lm) long with a radius of 10 lm, the surface-to-volume ratio is 0.2 lm1 . In contrast, for a small 0.5-ml Eppendorf tube, the ratio is 0.68 mm1 ; for comparison, 0:68 mm1 6:8 104 lm1 . Thus, the dierence is approximately three orders of magnitude. Since protein adsorbs to surfaces, the increased surface-to-volume ratio makes the technology more sensitive to proteinsurface interactions and these interactions will depend on the nature, for example hydrophobicity, of the protein. This increased sensitivity dictates a more rigorous approach in researching and modulating the nature of the surface. Heterogeneous nucleation is also an issue that needs further study when crystallization is performed in microuidic devices for the same reason. The preliminary results in our laboratory suggest that the high surface-tovolume ratio is a cause of altered crystallization kinetics, which may be an advantage if properly exploited.

5. What has been done using microuidics in protein crystal growth? One of the most recent publications of the use of the lab-on-a-chip approach for macromolecular crystallization was the description of successful batch crystallization by temperature variation. As a background to the

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187

183

method, Christopher et al. (1998) published a survey of protein solubility as a function of temperature, with an implicit application to crystallization (see also Table 3 in this reference), using a large-scale device. A multichambered thermal gradient device provides the temperature regime for capillaries with solutions containing protein and precipitant, which are inserted into a block. This block, which appears to cover a signicant part of a bench top, allows for accurate control of the local temperature and quick access to variable temperatures by simply moving samples to an alternate position. This same technique was established on a microchip by Juarez-Martinez et al. (2002) using a 1010 array of crystallization chambers, each having a volume of 5 ll, with a temperature gradient from 12 to 40 C. It is important to realize that these chips are entirely passive, that is, there are no channels present in chips of this nature and no ow occurs once the liquids are deposited into the wells (by pipette) and our discussion about the inherent advantages of microuidics is limited to the advantage of small volume in this particular application. The device was mounted on a pair of Peltier elements and could be operated with a temperature gradient in one axis and a precipitant concentration gradient in the other. Although the importance of temperature in macromolecule crystallization has been recognized, this is still a largely underutilized parameter. The crystallization chambers were manually loaded with premixed crystallization solutions and the wells sealed using clear tape. While not strictly a microuidics device, the overall design can potentially be readily modied to include apparatus for pumping and mixing solutions. Indeed, the use of a temperature gradient along one axis of a 2D crystallization chamber array is potentially readily implemented and may be a very valuable addition to both optimization and screening crystallization operations. Sanjoh and Tsukihara (1999) described a microuidics approach to macromolecule crystallization implemented on a silicon wafer. In this instance the goal was to microfabricate specic nucleation regions or anity sites within the crystal growth region of the device. The anity regions were placed on elevated stripes, posts, or trenches and consisted of both doped p- and n-type silicon layers. Tests with lysozyme showed that the crystals preferentially nucleated and grew on n-silicondoped regions. The approach did not include a means of actually moving solution about, but rather depended upon it being pipetted directly onto the silicon wafer placed within a sitting-drop pedestal, which was then equilibrated against a reservoir solution by the usual method. The ability to incorporate specically doped nucleation sites also indicates that one can incorporate active electronic or electromechanical functions within a crystallization device. These could be used to concentrate the macromolecule over a potential epitaxial

nucleation site to induce crystal formation. These devices show the potential of microfabrication for implementing active nucleation and crystal growth control, which approach has been previously demonstrated (Givargizov et al., 1991; McPherson and Shlichta, 1987; Rosenberger et al., 1993) but is still generally lacking in practical macromolecule crystallization applications. When we consider liquid ow and delivery and for a moment set aside the topic of macromolecular crystallization, the work of Thorsen et al. (2002) is quite interesting. The authors make a compelling comparison between electronic chips, with their associated miniaturization and automation in the electronic industry, and microuidic chips, with similar consequences forecast for the elds of chemistry and biology. In their work, Thorsen et al. present microuidic chips made from several layers of polydimethylsiloxane (PDMS) with hundreds of individual chambers, connected by channels that incorporate valves. The valves work on the garden hose principle-liquids are pumped through them by pressure and the channels can be closed by applying pressure to a channel perpendicular to the rst, eectively pressing the rst channel closed in a way similar to putting ones foot on a garden hose. In essence, by adjusting the channel geometry the imaginary wall thickness of the garden hose in our model is changed, so the pressure needed to deactivate it can be changed. This technique allows for access to individual picoliter-sized chambers. The authors show elegant examples of the use of this technology, for example in enzyme assay work or in individual (living) cell manipulation, but they also point out the preliminary status of this development by mentioning several disadvantages, all of which are applicable to macromolecular crystallization: PDMS is not compatible with organic solvents; the material is not impervious to small molecules (including water), which leads to some cross contamination; and some molecules, including proteins, are capable of adhering to the polymer. Another major problem of consequence to macromolecule crystallization is that the channels in the nonpressurized o state are open. For crystal growth, in which the important events occur long after the drop ingredients have been deposited, this will have to be addressed in order to prevent evaporative loss of crystallization solutions. Hansen et al. (2002) extended the technology described by Thorsen et al. (2002) to protein crystallization screening. The crystallization technique used is liquid liquid diusion. There are 144 parallel reactions executed in each chip, which require the growth solutions to be dispensed into 48 wells, by manual or robotic pipetting. The protein and precipitant solutions are pumped into small, individual chambers, which are subsequently connected to each other by removal of a barrier between them, which then initiates the mixing of protein and precipitant by diusion. The combined volume of the

184

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187

two chambers is 25 nl and the volumetric ratios of protein and precipitant solutions are preset by the nature of chips design, with three ratios of 1:4, 1:1, or 4:1 for each proteinprecipitant pair. The device is claimed to have picoliter accuracy and can be linearly scaled up to larger numbers of xed conditions. By using silicone elastomers they are able to fabricate micromechanical valves on the device. As the elastomer is gas-permeable, the solution priming problem is solved by simply placing the contents under a low hydrostatic pressure: the liquid displaces the gas, which diuses out through the elastomer. This approach also enables the lling of deadend channels that cannot be primed or lled using other methods. Both crystallization solutions and protein sample (3-ll volumes) were manually loaded into the device, which was then activated by placing under mild pressure (7 psi). The setup time for a single device was about 42 min. The authors demonstrated protein crystal growth in this hardware with 11 dierent macromolecules, at least one of which was harvested and exposed to X-rays. Some of these macromolecules, like the bacterial 70S ribosome, are dicult to crystallize and therefore clearly demonstrate that this technology can be successfully used for challenging crystallization systems. Some of the demonstrable advantages of this system include the robust solution metering, the lack of dependence on uid properties including viscosity, the ease of crystal harvesting, and the ability to perform liquidliquid diusion unhampered by convection in a standard gravity environment. Also, the authors noted that there was generally reduced precipitation during the initial mixing and in the equilibrated solutions than with the conventional methods, while equilibration and crystal growth times were faster. It is of particular interest to the crystal grower that the number of hits found by using such a chip exceeded the number achieved by standard microbatch or hangingdrop techniques. However, most of the generic disadvantages quoted above, such as the permeability of the elastomer and inability of sealing specic regions in the o state, remain unresolved. In addition it becomes clear that in this particular use of microuidics, optimization of crystallization conditions is dicult as one must start a new experiment with premixed stock (input) solutions. An obvious improvement would be to use a microuidics solution preparation chip, either separately or fabricated in situ, for the preparation of each precipitant solution. In this fashion more complex operations can be realized through the custom assembly of several simpler discrete devices, exactly analogous to the use of integrated circuits. Finally, the device described required manual harvesting of the crystal, accomplished by separation of the elastomeric portion of the chip from the substrate. This process results in opening of all crystallization wells to gain access to any

one. Thus all other crystals are lost in the harvesting of one. Manipulation and handling of a macromolecule crystal may aect the quality of the data obtained. Gavira et al. (2002) have recently described a procedure in which crystals are grown within X-ray capillaries and then subsequently harvested and the diraction data collected in situ without removing the crystals from their growth environment. It is not unreasonable to envision a system using microuidic chips to prepare solutions, mix them with protein solutions, and deposit the mixture into X-ray capillary tubes on the order of 100 lm in diameter in order to retain the advantages of microuidics while at the same time simplifying the crystal harvesting and data collection requirements.

6. Is the large-scale use of microuidics for macromolecular crystal growth feasible? Technically, is macromolecular crystallization possible in a microuidic system in the sense of scientic accomplishment, is it possible in the sense of engineering accomplishment, and does it give major advantages that are not otherwise achievable? Our discussion above illustrates that it is scientically achievable to create a device suitable for macromolecular crystallization and it is to be expected that some of the current disadvantages can be eliminated by improved design or use of alternate materials. Examples are found in the alternate design of pumps or valves (Hasselbrink et al., 2002; Yu et al., 2001) that can be implemented on microchips. The engineering feasibility explicitly refers to a common, yet not often considered problem: can we purchase identical crystallization devices, produced under adequate quality control? This issue seems somewhat trivial, but it actually is not if we think about how many crystallization trays any given laboratory uses for macromolecular crystallization and the complexity of microuidic devices. In numerous cases in which microuidic technology was mentioned in literature, the devices are actually not commercially available for this reason. The development of a microuidic device for crystallization condition screening and perhaps separately for crystallization condition optimization is a novel, incomplete process and it is therefore too early to comment on whether sucient unique advantages can be incorporated or disadvantages can be avoided. However, again drawing upon integrated circuits as an analogy, microuidics devices may be reducible to a standard set of discrete operations which can then be custom assembled to form more complex operations as needed. With this approach the success of a manufacturing investment does not have to rest upon a single application.

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187

185

7. What does it cost and what does the benet/cost ratio look like? A rough cost estimate indicates that the highest cost component in manually preparing a crystallization plate is of the labor. The cost of the plate is second, and those of the pipette tips and screening reagents third. The costs of the macromolecule are not considered, although reductions in volume will favor microor nanoscale approaches in their facilitating broader crystallization trials. Cost savings are primarily achieved through reduction in personnel or device costs and not in the reagents or disposables. Crystallization robots are justied by their ability to signicantly reduce crystallization trial setup times or at least allow the operator to perform other tasks while the robot is in operation. However, their purchase and maintenance costs keep them out of reach of many crystallography laboratories. Other costs are also not factored into the above, including those of subsequent storage, tracking, and observation once the crystallizations have been set up. Simple reductions in reagent volumes and costs are not justiable if they ultimately result in signicantly higher crystallization costs without a concomitant increase in successful crystallizations. Robotic nanoscale approaches can be justied as their costs per experiment are spread out over hundreds of thousands or even millions of crystallization trials. Reagent costs are low, and technician costs are now those of the robot operator. Using todays standard methods, plate costs are essentially xed, approximately equivalent whether setup is done manually or by robot. The purpose of the above is to place procedures for obtaining crystals in a more proper perspective. The crystallization approach taken by a laboratory or group must be matched to its needs and resources and not simply to the technical possibilities that are available. New approaches must oer some advantage such as being cost competitive or oering signicantly greater chances of success. The major costs associated with microuidics-based approaches will be for device fabrication and, potentially, associated equipment used to drive the operations. Crystal growth plates are injectionmolded plastic, a very mature and comparatively lowtechnology manufacturing process. Microuidic device manufacture will be more similar to that for integrated circuits. Micrometer-sized uid line dimensions require that the manufacturer put a premium on the cleanliness of the manufacturing process. While economies of scale may serve to bring the costs per device down, it is doubtful that they can approach the costs for injectionmolded crystallization plates. The decrease in per-chip costs will certainly not be comparable to those realized for the manufacture of integrated circuits. Whereas integrated circuits may see many years of use, microuidic

chips typically can be used only once and then must be discarded. This all becomes particularly relevant when contemplating microuidics-based macromolecule crystallization approaches. Low-throughput approaches are adequately covered by manual methods, while crystallization robots have heretofore covered high-throughput approaches, in which the complexity and expense are in a robot that typically uses the same crystallization plates as the low-volume approaches. With a microuidics approach both the crystallization device and the associated equipment for setup are likely to be complex and therefore expensive. Thus for microuidics the cost per device will likely always be signicantly higher than for current plate technology. In order to be successful, this approach must be competitive in the areas of experimental density, probability of success, and exibility.

8. Conclusion From the development work carried out to date microuidics may oer a powerful alternative approach to the growth of macromolecule crystals. Single-use devices, able to use an array of predetermined crystallization solutions, have been demonstrated and shown to be highly eective. Added complexity may be in the form of a linear expansion of the numbers of experiments or through enabling on-chip solution mixing and dispensing. The ability to set up crystallization arrays using premixed solutions ts into the current macromolecule crystallization screening practice. The considerably more complex ability to perform on-chip mixing of complex solutions suitable for crystallization optimization has yet to be demonstrated. Preliminary devices show considerable promise in terms of potential experimental density, minimal consumption of macromolecule solution, and success rate. Other technologies, particularly those involving the establishment of epitaxial nucleation sites and precise temperature control, may be readily implemented on microuidics devices and serve as powerful extensions to this approach. Regardless, several issues need to be addressed before a truly viable microuidics-based macromolecule crystallization system can be produced. Chief among these are the long-term sealing properties of the device, the compatibility of the materials used with all likely solution components, and the ability to harvest single crystallization wells while leaving the rest intact.

Acknowledgments The authors acknowledge discussions with Dr. Michael Spaid of Caliper Technologies Corporation.

186

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187 Wenning, S., Slezak, T., Doggett, N., Cheng, J.F., Olsen, A., Lucas, S., Elkin, C., Uberbacher, E., Frazier, M., Gibbs, R.A., Muzny, D.M., Scherer, S.E., Bouck, J.B., Sodergren, E.J., Worley, K.C., Rives, C.M., Gorrell, J.H., Metzker, M.L., Naylor, S.L., Kucherlapati, R.S., Nelson, D.L., Weinstock, G.M., Sakaki, Y., Fujiyama, A., Hattori, M., Yada, T., Toyoda, A., Itoh, T., Kawagoe, C., Watanabe, H., Totoki, Y., Taylor, T., Weissenbach, J., Heilig, R., Saurin, W., Artiguenave, F., Brottier, P., Bruls, T., Pelletier, E., Robert, C., Wincker, P., Smith, D.R., DoucetteStamm, L., Rubeneld, M., Weinstock, K., Lee, H.M., Dubois, J., Rosenthal, A., Platzer, M., Nyakatura, G., Taudien, S., Rump, A., Yang, H., Yu, J., Wang, J., Huang, G., Gu, J., Hood, L., Rowen, L., Madan, A., Qin, S., Davis, R.W., Federspiel, N.A., Abola, A.P., Proctor, M.J., Myers, R.M., Schmutz, J., Dickson, M., Grimwood, J., Cox, D.R., Olson, M.V., Kaul, R., Shimizu, N., Kawasaki, K., Minoshima, S., Evans, G.A., Athanasiou, M., Schultz, R., Roe, B.A., Chen, F., Pan, H., Ramser, J., Lehrach, H., Reinhardt, R., McCombie, W.R., de la Bastide, M., Dedhia, N., Blocker, H., Hornischer, K., Nordsiek, G., Agarwala, R., Aravind, L., Bailey, J.A., Bateman, A., Batzoglou, S., Birney, E., Bork, P., Brown, D.G., Burge, C.B., Cerutti, L., Chen, H.C., Church, D., Clamp, M., Copley, R.R., Doerks, T., Eddy, S.R., Eichler, E.E., Furey, T.S., Galagan, J., Gilbert, J.G., Harmon, C., Hayashizaki, Y., Haussler, D., Hermjakob, H., Hokamp, K., Jang, W., Johnson, L.S., Jones, T.A., Kasif, S., Kaspryzk, A., Kennedy, S., Kent, W.J., Kitts, P., Koonin, E.V., Korf, I., Kulp, D., Lancet, D., Lowe, T.M., McLysaght, A., Mikkelsen, T., Moran, J.V., Mulder, N., Pollara, V.J., Ponting, C.P., Schuler, G., Schultz, J., Slater, G., Smit, A.F., Stupka, E., Szustakowski, J., Thierry-Mieg, D., Thierry-Mieg, J., Wagner, L., Wallis, J., Wheeler, R., Williams, A., Wolf, Y.I., Wolfe, K.H., Yang, S.P., Yeh, R.F., Collins, F., Guyer, M.S., Peterson, J., Felsenfeld, A., Wetterstrand, K.A., Patrinos, A., Morgan, M.J., Szustakowki, J., de Jong, P., Catanese, J.J., Osoegawa, K., Shizuya, H., Choi, S., Chen, Y.J., 2001. Initial sequencing and analysis of the human genome. Nature 409, 860 921. Luft, J.R., Woley, J., Jurisica, I., Glasgow, J., Fortier, S., DeTitta, G.T., 2001. Macromolecular crystallization in a high throughput laboratorythe search phase. J. Cryst. Growth 232, 591595. McPherson, A., Shlichta, P.J., 1987. Facilitation of the growth of protein crystals by heterogeneous/epitaxial nucleation. J. Cryst. Growth 85, 206214. Rosenberger, F., Howard, S.B., Sowers, J.W., Nyce, T.A., 1993. Temperature dependence of protein solubilitydetermination and application to crystallization in X-ray capillaries. J. Cryst. Growth 129, 112. Rupp, B., Segelke, B.W., Krupka, H.I., Lekin, T.P., Schafer, J., Zemla, A., Toppani, D., Snell, G., Earnest, T., 2002. The TB Structural Genomics Consortium crystallization facility: towards automation from protein to electron density. Acta Crystallogr. D 58, 1514 1518. Sachidanandam, R., Weissman, D., Schmidt, S.C., Kakol, J.M., Stein, L.D., Marth, G., Sherry, S., Mullikin, J.C., Mortimore, B.J., Willey, D.L., Hunt, S.E., Cole, C.G., Coggill, P.C., Rice, C.M., Ning, Z., Rogers, J., Bentley, D.R., Kwok, P.Y., Mardis, E.R., Yeh, R.T., Schultz, B., Cook, L., Davenport, R., Dante, M., Fulton, L., Hillier, L., Waterston, R.H., McPherson, J.D., Gilman, B., Schaner, S., Van Etten, W.J., Reich, D., Higgins, J., Daly, M.J., Blumenstiel, B., Baldwin, J., Stange-Thomann, N., Zody, M.C., Linton, L., Lander, E.S., Altshuler, D., The International SNP Map Working Group, 2001. A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms. Nature 409, 928 933. Sander, C., 2000. Genomic medicine and the future of health care. Science 287, 19771978.

References
Christopher, G.K., Phipps, A.G., Gray, R.J., 1998. Temperaturedependent solubility of selected proteins. J. Cryst. Growth 191, 820826. Cohen, C.B., Chin-Dixon, E., Jeong, S., Nikiforov, T.T., 1999. A microchip-based enzyme assay for protein kinase A. Anal. Biochem. 273, 8997. Cudney, R., Patel, S., Weisgraber, K., Newhouse, Y., McPherson, A., 1994. Screening and optimization strategies for macromolecular crystal growth. Acta Crystallogr. D 50, 414423. Fu, L.M., Yang, R.J., Lee, G.B., Liu, H.H., 2002. Electrokinetic injection techniques in microuidic chips. Anal. Chem. 74, 5084 5091. Gavira, J.A., Toh, D., Lopez-Jaramillo, J., Garcia-Ruiz, J.M., Ng, J.D., 2002. Ab initio crystallographic structure determination of insulin from protein to electron density without crystal handling. Acta Crystallogr. D 58, 11471154. Givargizov, E.I., Kliya, M.O., Melik-Adamyan, V.R., Grebenko, A.I., DeMattei, R.C., Feigelson, R.S., 1991. Articial epitaxy (graphepitaxy) of proteins. J. Cryst. Growth 112, 758772. Hadd, A.G., Raymond, D.E., Halliwell, J.W., Jacobson, S.C., Ramsey, J.M., 1997. Microchip device for performing enzyme assays. Anal. Chem. 69, 34073412. Hansen, C.L., Skordalakes, E., Berger, J.M., Quake, S.R., 2002. A robust and scalable microuidic metering method that allows protein crystal growth by free interface diusion. Proc. Natl. Acad. Sci. USA 99, 1653116536. Hasselbrink Jr., E.F., Shepodd, T.J., Rehm, J.E., 2002. High-pressure microuidic control in lab-on-a-chip devices using mobile polymer monoliths. Anal. Chem. 74, 49134918. Jancarik, J., Kim, S.-H., 1991. Sparse matrix sampling: a screening method for crystallization of proteins. J. Appl. Crystallogr. 24, 409411. Juarez-Martinez, G., Steinmann, P., Roszak, A.W., Isaacs, N.W., Cooper, J.M., 2002. High-throughput screens for postgenomics: Studies of protein crystallization using microsystems technology. Anal. Chem. 74, 35053510. Kopf-Sill, A.R., 2002. Successes and challenges of lab-on-a-chip. Lab Chip 2, 42N47N. Krupka, H.I., Rupp, B., Segelke, B.W., Lekin, T.P., Wright, D., Wu, H.-C., Todd, P., Azarani, A., 2002. The high-speed Hydra-PlusOne system for automated high-throughput protein crystallography. Acta Crystallogr. D 58, 15231526. Kundrot, C.E., Judge, R.A., Pusey, M.L., Snell, E.S., 2000. Microgravity and macromolecular crystallography. Cryst. Growth Des. 1, 8799. Lander, E.S., Linton, L.M., Birren, B., Nusbaum, C., Zody, M.C., Baldwin, J., Devon, K., Dewar, K., Doyle, M., FitzHugh, W., Funke, R., Gage, D., Harris, K., Heaford, A., Howland, J., Kann, L., Lehoczky, J., LeVine, R., McEwan, P., McKernan, K., Meldrim, J., Mesirov, J.P., Miranda, C., Morris, W., Naylor, J., Raymond, C., Rosetti, M., Santos, R., Sheridan, A., Sougnez, C., Stange-Thomann, N., Stojanovic, N., Subramanian, A., Wyman, D., Rogers, J., Sulston, J., Ainscough, R., Beck, S., Bentley, D., Burton, J., Clee, C., Carter, N., Coulson, A., Deadman, R., Deloukas, P., Dunham, A., Dunham, I., Durbin, R., French, L., Grafham, D., Gregory, S., Hubbard, T., Humphray, S., Hunt, A., Jones, M., Lloyd, C., McMurray, A., Matthews, L., Mercer, S., Milne, S., Mullikin, J.C., Mungall, A., Plumb, R., Ross, M., Shownkeen, R., Sims, S., Waterston, R.H., Wilson, R.K., Hillier, L.W., McPherson, J.D., Marra, M.A., Mardis, E.R., Fulton, L.A., Chinwalla, A.T., Pepin, K.H., Gish, W.R., Chissoe, S.L., Wendl, M.C., Delehaunty, K.D., Miner, T.L., Delehaunty, A., Kramer, J.B., Cook, L.L., Fulton, R.S., Johnson, D.L., Minx, P.J., Clifton, S.W., Hawkins, T., Branscomb, E., Predki, P., Richardson, P.,

M. van der Woerd et al. / Journal of Structural Biology 142 (2003) 180187 Sanjoh, A., Tsukihara, T., 1999. Spatiotemporal protein crystal growth studies using microuidic silicon devices. J. Cryst. Growth 196, 691702. Shapiro, L., Harris, T., 2000. Finding function through structural genomics. Curr. Opin. Biotechnol. 11, 3135. Thorsen, T., Maerkl, S.J., Quake, S.R., 2002. Microuidic large-scale integration. Science 298, 580584. Venter, J.C., Adams, M.D., Myers, E.W., Li, P.W., Mural, R.J., Sutton, G.G., Smith, H.O., Yandell, M., Evans, C.A., Holt, R.A., Gocayne, J.D., Amanatides, P., Ballew, R.M., Huson, D.H., Wortman, J.R., Zhang, Q., Kodira, C.D., Zheng, X.H., Chen, L., Skupski, M., Subramanian, G., Thomas, P.D., Zhang, J., Gabor Miklos, G.L., Nelson, C., Broder, S., Clark, A.G., Nadeau, J., McKusick, V.A., Zinder, N., Levine, A.J., Roberts, R.J., Simon, M., Slayman, C., Hunkapiller, M., Bolanos, R., Delcher, A., Dew, I., Fasulo, D., Flanigan, M., Florea, L., Halpern, A., Hannenhalli, S., Kravitz, S., Levy, S., Mobarry, C., Reinert, K., Remington, K., Abu-Threideh, J., Beasley, E., Biddick, K., Bonazzi, V., Brandon, R., Cargill, M., Chandramouliswaran, I., Charlab, R., Chaturvedi, K., Deng, Z., Di Francesco, V., Dunn, P., Eilbeck, K., Evangelista, C., Gabrielian, A.E., Gan, W., Ge, W., Gong, F., Gu, Z., Guan, P., Heiman, T.J., Higgins, M.E., Ji, R.R., Ke, Z., Ketchum, K.A., Lai, Z., Lei, Y., Li, Z., Li, J., Liang, Y., Lin, X., Lu, F., Merkulov, G.V., Milshina, N., Moore, H.M., Naik, A.K., Narayan, V.A., Neelam, B., Nusskern, D., Rusch, D.B., Salzberg, S., Shao, W., Shue, B., Sun, J., Wang, Z., Wang, A., Wang, X., Wang, J., Wei, M., Wides, R., Xiao, C., Yan, C., Yao, A., Ye, J., Zhan, M., Zhang, W., Zhang, H., Zhao, Q., Zheng, L., Zhong, F., Zhong, W., Zhu, S., Zhao, S., Gilbert, D., Baumhueter, S., Spier, G., Carter, C., Cravchik, A., Woodage, T., Ali, F., An, H., Awe, A., Baldwin, D., Baden, H., Barnstead, M., Barrow, I., Beeson, K., Busam, D., Carver, A., Center, A., Cheng, M.L., Curry, L., Danaher, S., Davenport, L., Desilets, R., Dietz, S., Dodson, K., Doup, L.,

187

Ferriera, S., Garg, N., Gluecksmann, A., Hart, B., Haynes, J., Haynes, C., Heiner, C., Hladun, S., Hostin, D., Houck, J., Howland, T., Ibegwam, C., Johnson, J., Kalush, F., Kline, L., Koduru, S., Love, A., Mann, F., May, D., McCawley, S., McIntosh, T., McMullen, I., Moy, M., Moy, L., Murphy, B., Nelson, K., Pfannkoch, C., Pratts, E., Puri, V., Qureshi, H., Reardon, M., Rodriguez, R., Rogers, Y.H., Romblad, D., Ruhfel, B., Scott, R., Sitter, C., Smallwood, M., Stewart, E., Strong, R., Suh, E., Thomas, R., Tint, N.N., Tse, S., Vech, C., Wang, G., Wetter, J., Williams, S., Williams, M., Windsor, S., Winn-Deen, E., Wolfe, K., Zaveri, J., Zaveri, K., Abril, J.F., Guigo, R., Campbell, M.J., Sjolander, K.V., Karlak, B., Kejariwal, A., Mi, H., Lazareva, B., Hatton, T., Narechania, A., Diemer, K., Muruganujan, A., Guo, N., Sato, S., Bafna, V., Istrail, S., Lippert, R., Schwartz, R., Walenz, B., Yooseph, S., Allen, D., Basu, A., Baxendale, J., Blick, L., Caminha, M., Carnes-Stine, J., Caulk, P., Chiang, Y.H., Coyne, M., Dahlke, C., Mays, A., Dombroski, M., Donnelly, M., Ely, D., Esparham, S., Fosler, C., Gire, H., Glanowski, S., Glasser, K., Glodek, A., Gorokhov, M., Graham, K., Gropman, B., Harris, M., Heil, J., Henderson, S., Hoover, J., Jennings, D., Jordan, C., Jordan, J., Kasha, J., Kagan, L., Kraft, C., Levitsky, A., Lewis, M., Liu, X., Lopez, J., Ma, D., Majoros, W., McDaniel, J., Murphy, S., Newman, M., Nguyen, T., Nguyen, N., Nodell, M., Pan, S., Peck, J., Peterson, M., Rowe, W., Sanders, R., Scott, J., Simpson, M., Smith, T., Sprague, A., Stockwell, T., Turner, R., Venter, E., Wang, M., Wen, M., Wu, D., Wu, M., Xia, A., Zandieh, A., Zhu, X., 2001. The sequence of the human genome. Science 291, 13041351. Verpoorte, E., 2002. Microuidic chips for clinical and forensic analysis. Electrophoresis 23, 677712. Yu, Q., Bauer, J.M., Moore, J.S., Beebe, D.J., 2001. Responsive biomimetic hydrogel valve for microuidics. Appl. Phys. Lett. 78, 25892591.