“Cherry Picking Guide” for Small Molecule Screening
BME 301
University of Wisconsin – Madison
March 12, 2004
Team:
Blake Hondl
Amit Mehta
Jon Millin
Ryan Pope
Client:
Noël R. Peters, M.S.
Keck-UWCCC Small Molecule Screening Facility
University of Wisconsin Hospital
Advisor:
Willis Tompkins
Department of Biomedical Engineering
University of Wisconsin – Madison
Abstract
The high throughput screening of chemicals in biological assays is a testing method used frequently in areas of cancer and RNA research. Often, investigators are searching for chemicals that interact specifically with a biological substance. To aid researchers in their studies, high throughput facilities can test a substance against a library of over 36,000 reagents. If certain reagents fulfill the desired characteristics of the investigator, secondary testing of those reagents is performed. This secondary testing requires the lab user to pipette an individual chemical from one well in a 384-well microtiter plate to a well in another plate of the same size. This is a time consuming, error-prone process because the small well sizes makes the identification of a single well difficult. This project aims to create a device that can increase the efficiency and decrease the error that exists with the manual procedures involved in retesting. The device will consist of a LCD screen interfaced with a computer and run by a software application. By taking input from either the user or a file listing the locations of wells of interest, the device will be able to illuminate the wells and effectively guide the user to them.
Problem statement
The goal of this project is to create a device to guide a micropipette user in the transfer of small volumes of compounds between 384-well microtiter plates. The device will increase the efficiency and ease of the transfers, and reduce user error. The input to the device should be from a MicrosoftÒ ExcelÒ file containing a list of wells to be retested, or from direct user input. Utilizing a dual plate design, the pipetting aid should effectively guide the user to both the current well, from which a reagent is being withdrawn, and the target well through the use of a visual signal such as illumination.
Background Information
The Small Molecule Screening Facility, located within the University of Wisconsin Comprehensive Cancer Center (UW-CCC), provides university investigators/researchers with cost-effective access to high throughput screening of structurally diverse, drug-like chemicals in biological assays. This is accomplished through the use of laboratory robotics. A Biomek® FX (Beckman-Coulter, Inc.) liquid handler is used to distribute assay reagents and to transfer compounds onto screening plates (Figure 1). The Biomek® FX is capable of transferring
0.5-250 uL of liquid into either 96-well or 384-well plates. The facility also uses an EnVision® plate reader (Perkin Elmer, Inc.) to detect absorbance, fluorescence, or luminescence in 96-well and 384-well plates (Figure 2). The primary focus of this project is of the 384-well plates.
Figure 1: Biomek® FX [10] Figure 2: EnVision® plate reader [10]
After reading the well-plate, typically one to four wells are identified as meeting the absorbance, fluorescence or luminescence requirements of the university investigator. These wells are generally referred to as “hits” by researchers and lab workers. The results of these wells must then be retested for further investigation. To perform the retesting, a lab user must combine the same reagents and biological materials used to create “hits” during the initial screening in a new 384-well plate. This combination is accomplished using a micropipette. The reagents are transferred from a standard 384-well reagent plate to an empty 384-well plate. This new plate is then run through the plate reader to verify the results of the initial screening, confirming or refuting that the reagent actually does meet the investigator’s requirements. The greatest source of testing error is introduced during the manual transfer procedure.
The 384-well plate is 90mm x 130mm and contains the 384 wells in an array containing 24 well columns and 16 well rows. The area containing the wells is 70mm x 107mm, making each well approximately 3.5mm x 3.5mm (Figure 3). Locating a specific well within the array and correctly withdrawing from the small target is a difficult task for a user, resulting in an unacceptably high rate of error.
Figure 3: Typical 384 well Plate
Patent Search
Research in the form of patent searches was conducted to identify any existing products similar or identical to the one the team was attempting to design. This was accomplished through the United States Patent Office (USPTO, 2004). The most similar device provides a method of identifying and arranging test tubes, which includes a rack containing an array of wells located at the intersection of mutually perpendicular columns and rows. The columns and rows are aligned on perpendicular edges of the rack. The test tubes are positioned in the wells and then marked with indices corresponding to the location of the respective wells. Although this device makes use of identifying test tubes and not microtiter plates, the team will consider the concept of alphanumeric identification as option for locating wells.
Design Constraints
The most important design constraint established by the client was that the device must work with her current computer and any similar computer that may be purchased in the future. She also specified that the device must require little or no maintenance and be serviced easily by a person with only a limited amount of programming knowledge. This is crucial because individuals working in her lab do not have extensive experience with computer programming and hiring persons with such experience can become quite costly.
Another important specification was that the device must interface with a MicrosoftÒ ExcelÒ file. The client proposed that a researcher who uses the lab’s services could submit such a file listing the locations of reagents that were found as “hits.” The client indicated that the device must take this file as input and effectively illuminate the appropriate well in the plate of reagents.
The client also specified that the device must be relatively small since her lab has limited bench space, particularly near the computer with which the device must interface. The device should be no larger than a standard notebook, or about 8.5 x 11 inches.
Furthermore, the device must be able to withstand exposure to any of the more than 36,000 chemicals in its operating environment. These chemicals must not damage or interfere with the operation of the device. Particularly important to consider is the solvent used in the screening process, dimethylsulfoxide (DMSO), which has the ability to dissolve many polymers. One notable exception to this is polypropylene, which does not dissolve. This is the material from which the well plates are manufactured.
Finally, the client specified that the device be lightweight, so that it can be easily moved, and less expensive than currently available commercial devices which cost approximately $1000.
See Appendix A for a complete and detailed listing of Product Design Specifications.
Current Competition
Commercial devices that accomplish the function of this project include the Matrix Memowell® 96-well pipetting aid and the Tomtec® Quadra Cherry Picker (Figures 4 and 5). The Memowell® pipetting aid is the device the client would like the design team to replace. This device is designed to operate with 96-well plates and does not fully function with 384-well plates. Memowell®, instead of illuminating a single desired well at a time, illuminates four wells. This does not significantly increase the efficiency of the user and does not satisfy the client’s specifications.
Figure 4: Matrix Memowell® 96 well Figure 5: Tomtec® Quadra Cherry Picker [9]
pipetting aid [7]
The Quadra Cherry Picker is an automated device that can be programmed to pipette to and from the desired wells in specified well plates. A user can supply a work list, in disk format, the source plate barcode identification (the barcodes are on the sides of the well plates), the desired well location, and the volume to be transferred. The Cherry Picker then performs the appropriate pipetting. It works with both 96-well plates and 384-well plates, making it quite suitable to the client’s lab. The Cherry Picker, however, costs approximately $150,000 and exceeds the client’s budget.
Design Options
After extensive research and consideration, the team generated three design alternatives that could be implemented to accomplish the goal of this project. These options were based on the use of fiber optics, light-emitting diodes, or a LCD screen. Each of these solutions could be used to easily identify specific wells on a 384-well plate. Although the designs differ in concept, a commonality exists in that they all must include a computer interface. Since each design must include such an interface, a discussion of computer-related issues precedes the description and analysis of the team’s design alternatives.
Computer Interface
As mentioned previously, the device must interface with the client’s computer (Figure 6). Currently, the client uses a Dell personal computer (PC) with the MicrosoftÒ WindowsÒ XP operating system (Figure7). In order for the device to interact with the PC, it must be connected via an external port. Available ports include a video port on a video card, a monitor port such as VGA, or serial and USB ports. Data will then be sent to the device through one of these external ports.
Figure 6: Both the user and the device must Figure 7: Dell PC, similar to the one used
interface with the client’s computer. by the client [1].
In addition to the device being connected to a computer, a program initiated by the user must also be able to control it. To create such a program, the design team must choose an appropriate language for programming. Common languages used include C++, Basic, and Java. All of these languages are high-level programming languages, making them mostly independent of a particular type of computer. Such languages are also easier to read, write, and maintain.
The team must also choose a programming environment in which it will create the application. MicrosoftÒ VisualÒ, Metrowerks™ CodeWarrior™, and MicrosoftÒ BasicÒ are all environments that allow a programmer to create an application. A compiler may also be necessary to create or run the application, but these are often included in the programming environment.
As previously discussed, the program will take input from a MicrosoftÒ ExcelÒ file. It should also take input from the user via a graphical display/keyboard interface, making it more flexible than an application based solely on file input.
Design Option 1: 384 Fiber Optic Array
The fiber optic design would utilize 384 individual optical fibers that would be fixed to form the 384-array. A single fiber optic fiber has three layers to it: the core, the cladding, and the buffer (Figure 8). The core is a thin glass or plastic center in which the light travels. The cladding is the outer optical material that reflects the light. Finally, the buffer coating is the outermost layer that protects the fiber from moisture and damage.
Figure 8: Cross-section of a single optical fiber [2].
The fiber is able to transmit light long distances because the cladding reflects the light and allows it to travel down the fiber at the speed of light. There are three different types of optical fibers: single-mode glass, multi-mode glass, and plastic. The single mode fibers have small cores (approximately 9 µm in diameter) and transmit infrared laser light with wavelengths between 1,300 and 1,550 nm. Multi-mode fibers have larger cores (approximately 62.5 µm in diameter) and are used to transmit infrared light, with wavelengths between 850 to 1300 nm, from light emitting diodes. These two options would not be applicable to this project since they require infrared light, which is not contained in the normal viewing (visual) spectrum [2].
The third fiber optic option is to use plastic fiber optics. These fibers are not used as widely as the other two. Plastic optical fibers have a diameter of about 1 mm and are able to transmit visible red light with a wavelength of 650 nm from LEDs [2]. This design, aside from requiring 384 fiber optics, would require a LED, mirror, and computer controlling software. The mirror would need the ability to tilt so that it could reflect the light from the LED down the appropriate fiber optic. Since some of the compounds used in chemical screening are dark in color, the client would like to posses the ability to illuminate a row and column simultaneously. Furthermore, 650 nm light, which is red in color, may not be the most appropriate wavelength.
The benefits of an optical design are that it would be very small in size, require very little power, even over long durations, and would result in precise illumination of a well. However, constructing such a device would be very time consuming. Every fiber optic would need to be correctly positioned so that it can be accurately controlled. Also, the cost of such a device would be considerably higher than that of other designs. The increased cost is due to the mirror that would need to be created, as well as the software necessary to control the mirror.
Design Option 2: 384 LED Array
Another design option similar to the fiber optic array is an array of light-emitting diodes. The 384 LED array would be controlled using a microcontroller. This would allow the user to light a row and a column simultaneously and determine the well he/she is attempting to extract a chemical from. This design would incorporate 384 surface mount LEDs which would be soldered onto a printed circuit board (Figure 9).
Figure 9: Three surface-mount LEDs [6].
The LEDs would be controlled so that both rows and columns would be illuminated using a microcontroller. A microcontroller is a programmable device that allows control over electronic circuits. The microcontroller would light up an individual LED at any particular time. By illuminating the appropriate LEDs for a short duration, but at a high frequency, the desired row and column would appear to remain lit. The design would require that each individual LED be soldered onto a printed circuit board (Figure 10). An example of a possible output, Figure 11 shows the pattern of illumination when column B and row 2 are lit up.