P.I. David Brenner, Columbia University
Overview
This project calls for the design and construction of a new device that will use advanced, high-speed automated image analysis and robotics to examine tissue samples (e.g., a fingerstick of blood) quickly for quantitative indicators of radiation exposure (e.g., fragments of DNA; DNA repair complexes). Our goal is to develop a fully automated ultra-high throughput radiation biodosimetry workstation, using purpose-built robotics and advanced high-speed automated image acquisition. Maximum throughput will be 30,000 samples/day, compared with throughputs in current devices of a few hundred samples/day.
The basic system involves the well-characterized micronucleus assay in lymphocytes, with all the assays being carried out in-situ in multi-well plates.
By calling up pre-programmed options in timing, liquid handling, and image analysis, the device will also measure γ-H2AX foci yields, and micronucleus yields in reticulocytes, both providing “same-day answer” dose estimates.
By calling up pre-programmed options in liquid handling steps, the device will also measure micronuclei in other readily-accessible tissues, such as exfoliated cells from urine or buccal smears.
Automated Robotic System for High-Throughput Radiation Biodosimetry
System Layout
The robotic biodosimetry workstation consists of four main modules: centrifuge, cell harvesting system, liquid/plate handling robot and dedicated image acquisition/processing system.
The main tasks of this system are i) sample handling, ii) information logging and iii) imaging. The layout shown left includes robotic centrifugation, service robot, cell harvesting and liquid/plate handling.
The automated processing steps begins with loading blood samples, contained in PVC hematocrit capillaries, into a centrifuge which separates lymphocytes from red blood cells. After centrifugation the samples are transferred to a cell harvesting module, where lymphocytes are isolated by cutting the hematocrit tube. Plasma and lymphocytes are flushed into the appropriate well in a multi-well plate.
Product Development
Our product development strategy consists of three stages: breadboard, low-throughput prototype (6,000 samples per day) and high-throughput prototype (30,000 samples/day). In order to develop the breadboard a room was selected to impose dimensional constraints that would increase system portability. The automated biodosimetry room is located in the Mudd Engineering building at Columbia University Morningside Campus within the Department of Mechanical Engineering. The room has been equipped with an RS80 SCARA robot from Staubli, an O-Sprey UV laser system from Quantronix, a Sciclone ALH 3000 liquid handling robot from Caliper Life Sciences, a 5810RA Robotic Centrifuge from Eppendorf, and an industrial PC from iBASE technology running RTAI Linux for the low level control.
An implementation of the breadboard without the image acquisition/processing module is shown right.
Developing High-Throughput Imaging Systems for Biodosimetry
Introduction
Current automated imaging systems have limited throughput, mostly due to their non-specificity. We have therefore decided to build a dedicated high-throughput imaging system for performing the micronucleus assay exclusively, seeking creative solutions for rapid sample manipulation, automated focusing and image acquisition and analysis. The throughput of the imaging system currently under advanced stages of design and component testing is estimated at 5-6 minutes/96-well plate or 20,000-30,000 individual samples/day.Sample manipulation
The motion of the sample is separated into two components: a slower coarse motion and a rapid fine motion. The coarse motion is performed by a high speed stage (Parker motion) capable of few-g accelerations. This motion is used to move between adjacent samples (9 mm in 50 msec). The fine motion between fields of view within a single sample is performed, not by moving the sample but rather by steering light, using fast galvanometric mirrors as shown right. Typical transit times between adjacent fields of view of the microscope objective are 1-2 msec.
Focusing
A major rate limiting step in automated imaging system is focusing. In order to get good image quality, typically microscope objective lenses have rather small depth of field and are sensitive to the roughness of the sample being imaged. Our solution is to place a weak cylindrical lens in the optics path. Using an appropriately selected lens, a circular object will be imaged as circular when in focus and as elliptical when out of focus (see figure below), the aspect ratio being proportional to the distance from focus. The left image shows the resulting ellipse, while the image right shows the object as imaged by regular optics. The object-lens distance can then be corrected in one step.
Image acquisition and processing
For imaging we chose a CMOS camera, which has a much faster readout than the, lower noise, CCD cameras typically used. Analysis of the image is split between the camera and the frame grabber board to decrease the amount of data transferred to the controlling computer, the biggest bottleneck in current imaging systems. By using a dichroic mirror and two cameras, attached to the same frame grabber board we can simultaneously “see” the nucleus and cytoplasm and rapidly analyze their overlap obtaining the number and size of nuclei in each cell.
Lymphocyte-based biodosimetric assays for robotic handling
Introduction
The two assays used for automated biodosimetry are the micronuclei assay and the γ-H2AX assay. Both these assays require the separation of lymphocytes from whole blood as a first step. Whole blood is collected by minimally-invasive procedures such as a finger stick and transferred into glass capillaries coated with lithium heparin as an anticoagulant (QBC diagnostic Accutube).
Lymphocyte extraction
For lymphocyte separation, we have been successful in working with 50 µl of whole blood. The blood is transferred into the accutube by capillary action, followed by 50 µl of the lymphocyte separating medium, and centrifuged at a speed of 40 g for 20 mins. This yielded a good separation of the lymphocytes in the form of a clearly visible white band of lymphocytes with a count of 2100/µl of blood and about 80% purity.Although our current tests are with glass capillaries, we intend the final system to use plastic ones (such as Safe-T-Fill capillaries from RAM scientific), to increase safety and to facilitate capillary cutting. Initial tests indicate that the lymphocyte separation works as well with plastic capillaries.
Lymphocyte culture in the 96 microwell plates and micronuclei assay
In order to reduce handling time and to facilitate medium exchange and the addition/removal of reagents without needing to pellet out the lymphocytes by centrifugation each time, we have designed a system for culturing lymphocytes in 96 microwell plates (Multiscreen plates from Millipore). The underdrains of the microwell plates are easily detachable so that it allows easy removal of the processed, stained lymphocytes, embedded in the filters from the wells for imaging.The lymphocytes within each capillary after centrifugation were separated from the RBC pellet and dropped within the microwell. Cultures were set up in each well with complete medium containing 15% heat inactivated FBS, PHA (M-form), L-glutamine and antibiotics. After incubation at 37°C for 44 hrs cytochalasin B (in DMSO) was added in order to block cytokinesis. After 28 hrs with the cytochalasin B at 37°C, cells were treated with hypo and fixed in Carnoy’s fixative. Every time, the liquid already present within each well was drained out by the application of a positive pressure before the addition of any fresh reagents. Finally, the cells were allowed to dry and stained with Acridine Orange and DAPI and viewed under a fluorescent microscope.
Optimization of the γ-H2AX foci staining protocol
The need to detect DNA damage by radiation requires specific markers that can be easily seen and quantified, and γ-H2AX foci formation is one such event that can be used in this scenario. It is known that H2AX phosphorylation is specific to sites of DNA damage and is also indicative of amount of DNA damage. However, in order to use γ-H2AX as a quick screening tool, it must be optimized for sensitivity and rapidity which is what we are aiming to achieve.
The first aspect that we addressed is the image quality of foci in cells as it is important for uses in testing or as a diagnostic marker. Certain parameters need to be used in order to test the efficacy of the γ-H2AX, and the best procedure for producing the images. Light intensity ratios of foci, being one such parameter, can be optimized through antibody concentrations during chemiluminescence. The goal is to achieve the sharpest image possible and also to record the relationship between radiation level and foci counts. For the first experiments to determine the γ-H2AX induction, we worked with MEF cells in culture. It was seen that there was an increase of foci with increasing doses (left) of radiation in the fibroblasts. Experiments were also performed to measure which concentration of antibodies yielded the best image quality based on the contrast between the cell background and fluorescence signal given by the γ-H2AX foci. Cells that exhibited the largest intensity ratio were deemed the best for viewing for clarity and with the most distinction between foci and cellular background. Cells were treated with antibodies using various concentrations of 1-100, 1-500, 1-1000, and 1-2000 dilution of both the primary and secondary antibodies. Cell were then imaged and compared. It was found that the 1:100 concentrations for the primary antibody and the 1:500 for the secondary antibody yielded the best intensities ratios. A comparison was also done using different kinds of blocking agents, and it was found that even though Superblock (Pierce Biochemicals) yielded faster results, NFDM (Non fat dried milk) provided clearer pictures of the foci.
So far all the experiments regarding the γ-H2AX foci have been done in MEF cells, and we need to apply all these procedures to human lymphocytes and study the effects. We also need to find out the dose response for γ-H2AX foci formation in response to very low doses of radiation.
Collaborating Institutions
Center for Radiological Research, Columbia University
Department of Mechanical Engineering, Columbia University