New machines make prep work, protein analysis, and chip spotting faster, better, and cheaper — and humans redundant
Clontech Checks Its Chips
Dmitry Bochkariov, research scientist at BD Biosciences Clontech in Palo Alto, says a new manufacturing and quality control robot has put the company’s glass oligonucleotide microarray business in a new league. “Customers who have had access to Affymetrix [chips] for broad screening of data are now prepared to request significantly higher quality control measurements. It’s to our really big advantage to be able to provide it, and because of this we might get customers that we wouldn’t get otherwise,” says Bochkariov, who started building Clontech’s glass microarray manufacturing business in 1997.
Clontech, which introduced its first microarrays to the market in 2000, began shipping glass chips fabricated on its new RoboArrayer five months ago. The robot, from RoboDesign International in Carlsbad, Calif., incorporates a “vision system” that lets Clontech ensure the quality of every single spot on every array it produces.
Before now, the company assured chip quality by checking hybridization on four or five from each batch of 100 arrays. Based upon the results, the entire lot passed or failed. Discarding 100 arrays due to one or two bad ones wasted time and money. But passing the entire lot based on four positive results was a gamble too. “You say, OK, these are fine so the rest ought to be fine as well, but that [doesn’t] really work well because you cannot control quality of the surface treatment on the initial glass slides that you use, so you have to assume that that’s fine as well,” says Bochkariov. “It’s one unknown on top of another unknown, which may result in inferior performance of the final product.”
Instead, the $75,000 RoboArrayer can be programmed to check quality on dozens of parameters for each individual chip. For instance, Bochkariov says that just by checking the diameter of each dot on a slide, he can predict with high certainty areas of good quality. That way he’s making an informed decision on every array that gets passed onto the manufacturing lot. “It’s like being blind and all of a sudden getting sight,” he says.
The base of the RoboArrayer is a table with mechanical motion controls with linear slide motors. Another table on top moves side to side, and a fixed column moves front to back. A second column moves up and down on the first one, permitting control of three axes of movement. The instrument can fabricate 140 of Clontech’s one-by-three inch, 3,800-spot slides in six hours.
Once it fabricates the slide, which is done without any human intervention and a class 100 clean room, a CCD camera scans the whole lot to pick and analyze images of each array.
Bochkariov says the instrument has improved productivity three or four-fold, but that the most important difference is in quality, which he guesses has improved hundreds of times over. “We calculate how many spots on each slide are in the marginal area of diameters. Let’s say we allow one percent of spots to be in the marginal area of quality. Anything above that would be rejected from the lot.” According to its criteria now, even one bad spot out of 3,800 on an array will result in that slide being rejected. “You cannot do this if you don’t have the vision system,” says Bochkariov.
The vision system also enables Clontech to store all of these images. “This way, a customer who did experiments on those arrays and found some potentially valuable information can … [ask us to] go back to our database and check the information on that particular spot on that particular slide.”
Have these new measures gained Clontech any new customers? Says Bochkariov, “I’m working on it right now!”
— Adrienne Burke
Futuristic Febit Fabrication
What do you get by clicking your mouse exactly nine times? If you’re using Febit’s automated bench-top microarray technology and software program, you get a customized oligonucleotide microarray, according to Cord Stähler and his brother Peer, CEO and CSO of the Mannheim, Germany-based company. The tenth click begins its synthesis, in situ.
The machine, Geniom One, due on the European market in January 2003, is comparable to an all-in-one fax, printer, scanner, and copier. It has the ability to produce microarrays based on any known sequence, hybridize the sample to the probe, and fluorescently detect the data, all within a day. Among its applications are genotyping and expression profiling, and in the future, the Stählers say, diagnostics and personalized medicine. The machine is currently being beta-tested at the European Molecular Biology Lab, the German Cancer Research Center, and the German Resource Center for Genome Research.
The Stählers formed the 75-person Febit four years ago along with their father, Fritz, Manfred Muller, and Bernhard Jurisch. The Geniom instrument is Febit’s first product and has been in development since the company’s founding. “It is a hurdle you have to go over to get the level of [automation], and you have to invest in that,” says Cord.
He and his brother say huge demand led to the creation of the technology, which eliminates the “headache of handling oligonucleotides,” according to Peer, who likens it to a CD writer because of its reproducibility and quality.
The technology can process in a day eight to 16 arrays in parallel, containing 6,000 to 8,000 features. “Our features are very small, so we are in the league of Affymetrix,” Cord claims. The Stählers say there is a “clear roadmap” to increase the number of arrays that can be handled in parallel.
The length of the probe can be varied between eight and 60 mers, and the Stählers cite their microfluidics system as key to reducing the sample volume required to a tenth of the industry standard for other kits.
The entire package will cost Ý350,000, with a price tag of Ý130 per array. “That is far below the actual market price,” says Cord. Febit anticipates expanding to the US in 2004, after it ramps up its production capacities.
— Dana Frisch
Better Beamlines in Berkeley
Not content with having the world’s brightest source of x-ray light to get at a protein’s 3D image, Lawrence Berkeley National Laboratory in San Francisco has been adding robots and advanced software to make the effort truly high throughput.
“Now that the genome sequence has reached a number of milestones, the knowledge of the 3D structure [of proteins] is important to determine how molecules work,” says Thomas Earnest, head of LBNL’s Berkeley Center for Structural Biology. “If you can pick your best crystal and optimize without human intervention, you gain time savings by three or four fold.”
In March 2001, the center converted one of its beamlines, an x-ray accelerated along the curve of a synchrotron, into an automated operation. In this process, which the lab predicts will eventually save weeks in protein analysis while freeing researchers from repetitive tasks, a robot mounts and centers onto an x-ray field a protein crystal lifted from a container housing 112 cryogenically cooled crystals. Last winter the center retrofitted a second beamline and recently added “smart software” that chooses the best crystal to study, tracks the data, and builds a 3D image from the results, according to Earnest.
It will take about a year to integrate the software with the hardware before researchers can reduce from days to hours the time it takes to collect data, and from weeks to days the time required to process the data. Currently, two beamlines have robots and two additional lines are under construction and slated for operation by the fall.
Time savings have already been realized, and organizations funding the effort are already betting with dollars that greater efficiency will be realized: Amgen, Roche Biosciences, Howard Hughes Medical Institute, Syrrx, and the Genomics Institute of the Novartis Research Foundation have all made sizeable grants to Berkeley Labs in exchange for time on the beam, Earnest says.
Additionally, a number of companies has kicked in millions of dollars at various stages of the tool’s development to guarantee their own sizable chunk of x-ray time: Howard Hughes contributed an initial $8 million for beamline construction and continues to pay $600,000 annually for operational costs for two beams.
The funds allow Howard Hughes investigators to use a beamline for 75 percent of its operating time. The other 25 percent goes to researchers submitting proposals, who get their time for free unless the research is proprietary. In that case, they can expect to fork over approximately $1,500 for an eight-hour spin on the machine.
The Genomics Institute of the Novartis Research Foundation and Syrrx, meanwhile, share 75 percent of the time reserved on another beamline after kicking in an initial $2 million and promising to shell out $450,000 in annual operating costs.
Earnest says that part of the center’s approximately $4 million annual budget also comes from the US Department of Energy and the National Institutes of Health.
Other players are also getting into the automated-synchrotron protein-analysis game: The structural molecular biology group at Stanford and the European Synchrotron Radiation Facility in Grenoble, France, are slated to have automated facilities in operation within a year.
— Ken Howard
The Shaking Roe Bot
It pays to sell a robot to Bruce Roe. Zymark won a bid to build an automated liquid handling and microplate management station for Roe’s Advanced Center for Genome Technology at the University of Oklahoma earlier this year and installed a custom made instrument in May. With only word-of-mouth promoting, the company has since sold two more Mini-Staccatos to high-throughput sequencing labs and is in discussions with at least nine others. All of those customers were once postdocs in Bruce Roe’s lab, which is now known for contributing the sequence of chromosome 22 to the Human Genome Project.
Says Zymark system designer Ed Alderman, “We’ve learned in the past that if you find a thought leader in an arena, they’ll do most of your selling for you.” Alderman figures that Roe has “somewhere on the order of 60 postdocs, about 20 of whom are fairly high up the food chain in labs around the country” — Rick Wilson, Elaine Mardis, Sandy Clifton, and Stephanie Chissoe are a few of the names Roe drops. Not a bad early market for a $200,000 instrument.
Roe is nothing if not a thought leader. On the Sanger Centre team 25 years ago that developed DNA prep and sequencing methods that became Human Genome Project protocols, he has gone on to improve and automate each step and bottleneck in the process. “Over the years, because I’m not an engineer but I am slightly innovative, I’ve said, ‘Gee, we ought to automate these things,’” Roe says.
Zymark’s Mini-Staccato, which integrates the company’s Sciclone liquid handling station with its Twister II microplate handler, addresses an inefficiency Roe saw in his DNA prep procedures — the transfer of solutions to shakers. Unwilling to adopt the magnetic bead prep techniques that labs such as Baylor and Whitehead are using, Roe wanted to automate the procedure he already had in place. That method he says is “simple, it’s very cheap — under four cents a prep — and it’s in a 384-well format.” Furthermore, it gets 3,000 clean reads off an ABI Prism 3700 capillary, whereas others get about 600.
Roe describes his new Zymark instrument as “a rack with 30 384-well microtiter plates piled up like towers on a table the size of desk.” A robotic arm takes the top plate, hits a barcoder to record which one it is, and puts it down on a shaker. Four shakers are arranged in a square, so the arm places four plates down at a time. A 384-well pipetting head moved by a gantry from above dips into solution A and adds tris buffer to each of the four plates that are then shaken for several minutes. It adds solution B, a lysis solution, next and the plates are shaken for another 10 minutes. Finally, the 384-well-head adds solution C, an acetate high-salt solution to make the proteins in the genomic DNA precipitate. The plates are shaken for another 10 minutes before the robot moves them onto another rack where the tower is rebuilt.
In the 1980s, Roe says, one good lab tech could isolate 24 templates in a day. And through the chromosome 22 project DNA prep ran 22 hours a day. Now, two people monitor the Zymark instrument for three hours a day to prep 20 plates, or 7,680 templates. The instrument sits idle otherwise. “That’s as many plates as we need in a day,” says Roe, whose lab is now sequencing the syntenic regions of chimp, baboon, and zebrafish and doing comparative genomics of chromosome 22 with other species.
Roe says that next he’ll tackle a bottleneck around isolating BACs. And Zymark says it’s ready when he is to design changes that would give the lab a start-to-finish automated sequencing platform that requires no human intervention. Alderman acknowledges that Zymark hadn’t done much business in genomics before now, but the company is eager to build customized genomics lab systems out of its standard components. It’s no wonder. Zymark’s VP of sales was a postdoc in Roe’s lab.
— Adrienne Burke