NEW YORK, May 9 - As a June deadline for commercializing its 10,000 gene human microarrays looms, Corning’s Microarray Technology division is finding out what Affymetrix already knows: It’s tougher than you think to make a highly reproducible high-density microarray.
“We are still working the bugs out of the process, and working through the internal validation,” said Thomas Hinman, division vice president and general manager of CMT Life Sciences. “It has taken us longer than we wanted to.”
The major task the company’s engineers are wrestling with is that of making an array that provides reproducible, consistent results.
Unforseen obstacles in the company’s manufacturing process mean the company, which entered into the microarray field last September, may not be able to meet a June launch date set earlier this year for the human arrays. “We are still working towards that,” said Hinman. These arrays will use gene content provided by Invitrogen, although other arrays the company develops later may use genes provided by Corning’s other content partner, Incyte.
When the human arrays do reach the market, they will not be significantly cheaper than Affymetrix arrays, as many analysts in the sector predicted. Corning may be able to mass produce many glass items cheaply, from catalytic converters to fiber optic cable, but CMT engineers have to solve the problem of quality before they tackle cost.
The good news for customers who are looking for viable alternatives to Affymetrix arrays is that CMT engineers have handpicked highly refined technologies from other divisions within the company to design an array that they believe will provide reliable results.
“Our focus is less on reducing the price and more on offering the best value,” Hinman said. “What scientists care most about is they want to know when they do experiment that they will get good data.”
Scott Greenstone, an analyst from Thomas Weisel Partners who covers Affymetrix, agreed with this observation about the scientific community, but said it would be difficult for Corning to convince scientists to switch from Affymetrix to Corning arrays.
“Affymetrix has several years of demonstrated product quality and consistency of results, and has demonstrated they can manage [to deliver] huge lots of consistent product to the scientific community,” said Greenstone. “Corning doesn’t have the reach into the scientific community nearly as much as Affymetrix so it will be hard for them to get their story out and capture the mind share of scientists.”
However, the Corning arrays could help researchers who already own some microarray technology to save costs, said Thomas Volkert, the microarray facility manager at MIT’s Whitehead Institute.
“They [Corning] have the advantage of being an open format array, and a facility that has a scanner and a printer wouldn’t have to buy any special equipment” to use them, said Volkert.
Volkert, who uses Affymetrix arrays, has also beta tested the Corning arrays as part of a $10 million collaboration between Corning and the Whitehead Institute.
“I think [the Corning] arrays can be as good as Affymetrix arrays, but theirs is a quite different approach,” Volkert said.
Given the different approach the company has developed using its own intellectual property, Corning believes it has a “path forward” into the market despite Affymetrix’s large intellectual property estate — no minor feat for any company entering the field as late as Corning did, with no IP in the microarray space.
From Catalytic Converters to Microarray Print Heads
Using scientific engineers from materials technology, biochemical and machine research areas, Corning has come up with an automated high-volume arraying process that bears little if any resemblance to Affymetrix's photolithographic method of making arrays. While Affymetrix synthesizes oligonucleotides directly on a silicon chip, Corning prints cDNA arrays on proprietary glass using technology borrowed from its various divisions.
Corning's array making process starts with a reservoir device that combines the company's ceramic catalytic converter technology with its Pyrex glass and its method for "drawing down" glass to make thin fiber optic cable. Company engineers take a short opaque Pyrex glass cylinder that has a Dense grid of 1,024 hollow square cells running through it from top to bottom (600 per square inch) — in essence, a glass version of a catalytic converter — then heat this glass, and "draw it down," pulling one end into a longer, narrow end the diameter of a penny. Tiny amounts of liquid can run through the narrowing cells of the reservoir, from the wide end to the narrow end.
The company then fills this glass grid funnel with genetic material using a robotic device that Corning engineers built specifically for that task. The robotic arm of the device dips a pipette into a well of a 96 well-plate filled with cDNA and reagents, sucks out the cDNA, then moves over to the wide end of the funnel and deposits the genetic material into one cell. The liquid sample travels down the funnel to the narrow far end. This process, which is visible both through the glass doors of the robotic machine and through special cameras set up at the narrow end of the funnel print head, is repeated until each cell of the funnel has been filled.
During the printing process the narrow end of the funnel makes contact with the pin plate, a circle with 1,024 100 micron-wide pins on it. These pins are coated with a proprietary Corning technology to be non-wetting, except on the tip, so they only pick up tiny amounts of cDNA on the tip.
On the automated array assembly line, 10 dipped pin plates make contact with a single slide, making the array, with over 10,000 cDNA data points. The substrate is a Corning silane coated glass slide using a glass borrowed from the Company's display glass division, which makes the glass for flat display computer monitors. Corning engineers chose this glass because it is ultra flat and has a low background fluorescence, according to Hinman, an 18-year Corning veteran.
"We've taken the technology from the flat panel display glass, from catalytic converters, from our photonics division, and melded it together to make something that's unique," said Hinman.
The slides take less than a minute to print each. During this printing process, the company has computer scanners set up along the assembly line that can check in real time whether all of the points on a particular array have been properly spotted. A tray of 50 slides is then removed from the assembly line, and undergoes a process of downstream quality control and enhancement to ensure array quality and performance. (The company was not willing to share details on this proprietary post-printing process.)
Currently, this whole process takes about a day, and occurs at a "pilot line" facility in Corning's Sullivan Park Process Research Center. The company produces its yeast arrays, which are now its sole commercial product, at another location within its plant. When human and other arrays move to full production, the company plans to move the array process out of the pilot line and into a separate production facility.
Eventually, the company plans to offer a full spectrum of gene arrays with content from human and other organisms' genes. The company is currently exploring new content partnerships, and is feeling out the customer base to find out which unmet needs its microarrays can best fill. The partnership with the Whitehead institute also helps Corning to get a better view of what is going on in the space, but more importantly, said Hinman, it gives Corning an idea where the whole field is going over the long term.
Through this partnership, Hinman said he expects Corning to be able to "bridge the molecular biology world to that of the physical sciences."