PROTEIN CRYSTALLIZATION STRATEGIES FOR STRUCTURAL GENOMICS
edited by
Naomi E. Chayen


Contents
Preface     xiii
List of Contributors     xv

1.    Structural Genomics in the United States: The NIGMS Protein Structure Initiative     1
Charles G. Edmonds and John C. Norvell    
2.    A Guide to Automation and Data Handling in Protein Crystallization     9
Bernhard Rupp
       2.1.    Introduction     11
       2.2.    High-Throughput Protein Structure Determination     13
                 2.2.1.    Why High Throughput?     13
                 2.2.2.    Definition of High Throughput     13
                 2.2.3.    High Throughput versus Low Throughput     14
                 2.2.4.    Estimation of Throughput     14
                 2.2.5.    The Data Explosion     17
                2.2.6.     Information Technology Infrastructure      18
                2.2.7.     Data Mining and Machine Learning      19 
       2.3.    Overview of Protein Crystallization     21
                 2.3.1.    Basics of Protein Crystallization     21
                 2.3.2.    Experimental Crystallization Setup Techniques     23
                 2.3.3.    Experimental Design Considerations     25

                 2.3.4.    Crystallization Screening as a Sampling Problem     26
                 2.3.5.    Robotics and Flexible Design Strategies     26
       2.4.    Implementation of Robotic Technology—General Considerations     28
                 2.4.1.    Efficiency versus Throughput     29
                 2.4.2.    Patent and Licensing Issues     30
                 2.4.3.    Service and Maintenance      31
       2.5.    Implementation of Robotics in the Academic Laboratory     31
                 2.5.1.    Objectives and Cost     32
                 2.5.2.    Bottlenecks     33
                 2.5.3.    Kits versus Custom Cocktails      34
                 2.5.4.    Limitations to Full Automation      35
                 2.5.5.    Two Basic Automation Scenarios      35
       2.6.    Instrumentation     36
                 2.6.1.    Liquid Handling Stations     36
                 2.6.2.    Plate-Setup Robotics     38
                 2.6.3.    Sealing      41
                 2.6.4.    Visualization      41
                 2.6.5.    Plate Handling and Storage      42
                 2.6.6.    Fully Integrated Systems      45
                 2.6.7.    Cryoprotection      45
                 2.6.8.    Robotic Sample Mounting      46
       2.7.    Conclusions     49
3.    Automated Liquid Handling Systems for Submicroliter Crystallization     57
Terese Bergfors
       3.1.    Introduction     59
                 3.1.1.    The Development Path of Crystallization Robotics      59
                 3.1.2.    The Shift Towards Nanoliter Crystallization      60
                 3.1.3.    Advantages of Crystallization on the Nanoliter Level      61
                 3.1.4.    Future Robot Technology for Crystallization      62 
       3.2.    Choosing a Robot     63
                 3.2.1.    Background of This Chapter     63
                 3.2.2.    What are the Automation Requirements?     63
                 3.2.3.    Multiuser or Service Laboratory?     63
                 3.2.4.    Reservoir Filling and Cocktail Production     64
                 3.2.5.    Types of Dispensing     65
                 3.2.6.    Accuracy and Precision     67
                 3.2.7.    Speed and Related Issues of Chilling and Evaporation Control     68
                 3.2.8.    Size     69
                 3.2.9.    Price     70
                 3.2.10.  Product Descriptions      70
       Appendix 3.1. Product Ddescriptions of the Crystallization Robots Listed in Table 3.1     73
4.    Soft-Polymer Microfluidic Applications to Protein Crystallization     87
James M. Berger
       4.1.    Introduction     89
       4.2.    Classic Crystal Growth Methods     91
       4.3.    Microfluidic Platforms for Crystallization     94
       4.4.    Capabilities Unique to Microfluidic Devices     97
       4.5.    Microfluidic Phase-Space Mapping     100
       4.6.    Horizon Methods     100
       4.7.    Conclusion     103
5.    Counter-Diffusion Capillary Crystallization for High-Throughput Applications     111
Juan M. Garcia-Ruiz and Joseph D. Ng
       5.1.    Why Counter-Diffusion Crystallization?     113
       5.2.    Principle of Counter-Diffusion Crystallization     114
       5.3.    Preparation and Setup     115
                 5.3.1.    Single Capillar     116
                 5.3.2.    Multiple Array Configurations     116
       5.4.    Getting the Protein Crystal to the X-ray Beam as Soon as Possible     118
                 5.4.1.    Taking a Walk Down Crystal Lane     120
                 5.4.2.    Mounting and Aligning Crystals in Capillary     121
                 5.4.3.    Preliminary X-Ray Analysis     121
                 5.4.4.    Data Collection and Electron Density Calculation Image Processing     122
       5.5. Important Points      124
6.    Scale-Down Approaches for Measuring Protein-Protein Interactions     127
Joseph C. Fanguy, Charles S. Henry, Steven C. Holman, Joseph J. Valente, and
W. William Wilson

       6.1.    Introduction     129
       6.2.    Scale Down of Static Light Scattering     133
                 6.2.1.    Configuration Description     133
                 6.2.2.    Flow Injection Components     136
                 6.2.3.    Data Acquisition System      137
                 6.2.4.    Samples for Demonstration Purposes      137
                 6.2.5.    Results for Demonstration Proteins      138
       6.3.    Scale Down of Self-Interaction Chromatography     143
                 6.3.1.    Evolution of Self-Interaction Chromatography (SIC)     143
                 6.3.2.    The Synergy of B and SIC     144
       6.4.    B and Solubility     146
       6.5.    Potential for Miniaturization     147
       6.6.    Conclusions     149
7.    Gearing Optimization Techniques Towards High-Throughput     153
Naomi E. Chayen
       7.1.    Introduction     155
       7.2.    Development of Optimization Procedures     156
       
         7.2.1.    Crystallization in Gels     157
                 7.2.2.    “Containerless” Crystallization       159
                 7.2.3.    Control of Evaporation Kinetics      160
                 7.2.4.    Optimization by Decoupling Nucleation and Growth      161
                 7.2.5.    Slowing Down Vapor Diffusion with an Oil Barrier      162 
       7.3.    Summary     163
       Appendix 7.1.  Experimental Procedures     166
                 A7.1.1.  Preparation of Gels     166
                 A7.1.2. Control of Evaporation     166
                 A7.1.3. Decoupling of Nucleation and Growth     167
8.    Miniaturization and Automation for High-Throughput Membrane Protein Crystallization
       in Lipidic Mesophases     169

Vadim Cherezov and Martin Caffrey
        8.1.    Introduction     171
        8.2.   
The In-Meso Crystallogenesis Robot Suite     174
                 
8.2.1.    Crystallization Robot     174
                  8.2.2.    Screen Solution Preparation     174
                  8.2.3.    Crystallization Plates     175
                  8.2.4.    Imaging Station     176
        8.3.    Crystallization Trials     176
                  8.3.1.    In-Meso Crystallization     176
                  8.3.2.    Screening     177
        8.4.    Robot-Performance Characterization     178
        8.5.    Crystallization Plates     179
        8.6.    Crystallization Robot     181
                  8.6.1.    Dispensing the Lipidic Cubic Phase     181
                  8.6.2.    Dispensing the Precipitant Solution     184
                  8.6.3.    Dehydration during Setup     185
        8.7.    Imaging Station     187
                  8.7.1.    Performance Characteristics     187
                  8.7.2.    Image Processing     188
        8.8.    Crystallization Trials      189
        8.9.    Quo Vadis?      189
        8.10.  Conclusions      192
9.    Automated Classification of Crystallization Experiment     359
Julie Wilson
       9.1.    The Need for Automated Classification     197
       9.2.    Edge Detection     198
       9.3.    Image Cropping     203
       9.4.    Artifacts not Related to the Crystallization Experiment     203
                 9.4.1.    Bubbles and Droplets     204
                 9.4.2.    Plastic Mold Effects     204
       9.5.    Masking the Drop     205
                 9.5.1.    Circular Masks     205
                 9.5.2.    Noncircular Masks     20
       9.6.    Analysis of the Crystallization Drop     208
                 9.6.1.    Feature Extraction     209
                 9.6.2.    Object-Based Features     209
                 9.6.3.   Drop-Based Features     212
       9.7.    Classification     214
                 9.7.1.    Learning Algorithms     214
                 9.7.2.    Decision Trees     215
                 9.7.3.    Linear Discriminant Analysis     216
       9.8.    Summary     217
Index     221




Abstract


Getting from gene to structure involves many steps: cloning expression, solubilization, purification, crystallization, and only then, the determination of the structure. Once protein is pure and soluble, the key to crystallography is the availability of high-quality crystals. Producing high-quality crystals has always been the bottleneck to structure determination and with the advent of proteomics this problem is becoming increasingly acute. As a result, crystallization is gathering a new momentum as evidenced by the increasing numbers of commercial companies selling crystallization kits and tools, the increased investment of pharmaceutical companies in crystallization equipment and expertise, a high demand for practical courses in crystal growth, and the launch by the International Union of crystallography of a new Journal, Acta Crystallographica F, formed in 2005 for publishing the recent explosion of data concerning crystallization.

The past five years have seen some of the greatest achievements in the field of protein crystallization. It is now feasible to screen thousands of potential crystallization conditions by dispensing trials consisting of nanoliter volumes in a high-throughput mode. This has cut the time of setting up experiments from weeks to minutes, a scenario that was unimaginable a few years ago. Even more incredible, is the revelation that diffracting crystals can be produced from protein samples in volumes as small as 5–20 nanoliter. The subsequent phase of image capture and analysis of the crystallization drops is also progressing in great strides.

Surprisingly, in spite of the impressive advances accomplished, the crystallization problem has not been solved. High throughput has not yet resulted in high output and the current challenge is to design new and improved techniques (of screening and optimization) for the production of useful crystals. Scientists worldwide have taken on the challenge by tackling the crystallization problem from a variety of different aspects.

Research advances in recent years have opened up the scope for the development of new methods and tools to overcome the bottleneck of protein crystallization. A variety of parameters that could previously not be explored are now accessible thanks to sophisticated apparatus and the development of new science-based techniques to monitor and control the process of crystallization. However, in order to become useful to the structural genomics effort, it is vital to miniaturize and automate these techniques and adapt them to cope with the vast numbers of “leads” resulting from the high-throughput screening procedures. Such efforts are those of the immediate future and the focus of this book.





Preface

We are currently living in an exciting age, where for the first time ever, human diseases are being understood at a molecular level. Protein crystallography plays a major role in this understanding because proteins, being the major machinery of living things, are often the targets for drugs. The fuction of these proteins is determined by their three-dimensional structures hence a detailed understanding of protein structure is essential for rational design of therapeutic treatments. Structural genomics, or more accurately, structural proteomics, which aims to determine the structures of thousands of proteins has emerged as a direct consequence of the genome project in which the human genes and other genes have been sequenced. Structural Genomics projects worldwide are detailed in http://www.isgo.org/.
  Getting from gene to structure involves many steps: cloning, expression, solubilization, purification, crystallization, and only then, the determination of the structure. Once protein is pure and soluble, the key to crystallography is the availability of high-quality crystals. Producing high-quality crystals has always been the bottleneck to structure determination and with the advent of proteomics this problem is becoming
increasingly acute. As a result, crystallization is gathering a new momentum as evidenced by the increasing numbers of commercial  companies selling crystallization kits and tools, the increased investment of pharmaceutical companies in crystallization equipment and expertise, a high demand for practical courses in crystal growth, the launch by the International Union of Crystallography of a new journal, Acta Crystallographica F, formed in 2005 for publishing the recent explosion of data-concerning crystallization, and the establishment of the International Organization for Biological Crystallizaion (http://www.iobcr.org).
     The past five years have seen some of the greatest achievements in the field of protein crystallization. It is now feasible to screen thousands of potential crystallization conditions by dispensing trials consisting of nanoliter volumes in a high-throughput mode. This has cut the time of setting up experiments from weeks to minutes, a scenario that wasunimaginable a few years ago. Even more incredible, is the revelation that diffracting crystals can be produced from protein samples in volumes as small as 5–20 nanoliter. The subsequent phase of image capture and analysis of the crystallization drops is also progressing in great strides.
    Surprisingly, in spite of the impressive advances accomplished, the crystallization problem has not been solved. High throughput has not yet resulted in high output and the current challenge is to design new and improved techniques (of screening and optimization) for the production of useful crystals. Scientists worldwide have taken on the challenge by tackling the crystallization problem from a variety of different aspects.
     Research advances in recent years have opened up the scope for the development of new methods and tools to overcome the bottleneck of protein crystallization. A variety of parameters that could previously not be explored are now accessible thanks to sophisticated apparatus and he development of new science-based techniques to monitor and control the process of crystallization. However, in order to become useful to the structural genomics effort, it is vital to miniaturize and automate these techniques and adapt them to cope with the vast numbers of “leads” resulting from the high-throughput screening procedures. Such efforts are those of the immediate future and the focus of this book.
     This book’s aim is to assemble a selection of reviews highlighting the state of the art in techniques and strategies developed over the last four years for crystallization in the context of structural genomics. It is dedicated to the steps from the time of having pure protein to the crystallization, detection, imaging, and production of a diffracting crystal.
       Most researchers involved in HTP consortia are engaged in solving structures and do not have the time or inclination to develop new methodology for crystallization. I have therefore chosen most of the Authors form the crystal growth community, for their expertise in development of methodology in their respective fields within crystallization.
       The first chapter provides a background to structural genomics describing how it was conceived, funded, and implemented in the USA. This is just one example of many excellent Structural Genomics projects
taking place worldwide (see http://www.isgo.org/) that could compile a book in their own right.
      Chapter 2 provides an overview of setting up high-throughput structural proteomics projects in different environments and guidance as to how to handle the data and know-how resulting from the numerous experiments.
       Chapters 3-5 describe different ways of conducting high-throughput initial screening for obtaining the first leads for crystallization. Chapter 3 reviews the current robotic systems for screening that are available on the market for conducting high-throughput crystallization trials. Until 2002, screening for initial crystallization conditions was performed almost exclusively using vapor diffusion and microbatch. Since then, screening by the free interface and counter-diffusion methods that were least used in the past due to handling difficulties and the requirement for large quantities of sample, have been miniaturized and automated thus gaining a new lease of life. Chapters 4 and 5 deal with miniaturization and adaptation to high throughput utilizing free interface and counter diffusion and gels, the former by microfluidics and the latter, using capillaries.
        Chapter 6 describes a procedure for screening with a built-in optimization. Screening is performed using the standard screening kits but instead of looking for leads such as crystals, precipitation etc, the aim is to find the conditions that are at the metastable zone, which contains the ideal conditions for crystal growth.
         Chapter 7 deals with strategies to optimize and improve the crystal quality once a lead has been obtained. These are methods that have been performed mainly manually and have recently been automated in order to adapt them to high-throughput experiments.
        Chapter 8 describes the miniaturization and automation of experiments with membrane proteins in liquid cubic phase.
         Dispensing trials with such amazing speed and ease has raised the issue of observation and monitoring of the vast numbers of trials. Major effort is currently being invested in designing image processing equipment for automated follow-up and analysis of the results as described in Chapter 9. Full automation of the visualization and monitoring of trials is expected to be the next major breakthrough in the field of crystallization.
        The success rate of obtaining high-quality crystals is improving and the coming years promise to bring further advances in the more complex techniques that will play a major role in crystallization and proteomics. This will raise the rate of producing high-quality crystals, and will equip the genome project to deal with its awesome task.


Naomi Chayen
London, England, May 2007