Siga Technologies said last week that it will use molecular modeling technology developed by Molsoft in a $3.2 million contract to computationally design small-molecule compounds to inhibit biowarfare agents for the United States Air Force.
The value of the contract nearly doubles the company's 2004 annual revenue of $1.8 million, and adds to the firm's growing pool of federal biodefense funding, which includes two Small Business Innovation Research grants that Siga received from the NIH in late 2004: a $5.8-million award to develop a drug against smallpox, and a $6.3-million award to combat arenaviruses.
Bernard Kasten, Siga's CEO, told BioInform that the 10-year-old company has been building its computational biology capabilities for the last three years, and now considers molecular modeling a "core strength." One key to Siga's computational drug-design strategy is a very close collaboration with La Jolla, Calif.-based Molsoft, founded in 1995 by Ruben Abagyan of the Scripps Research Institute.
"We work hand-in-glove with Molsoft on projects of this nature, so it's a very strong partnership for Siga," Kasten said. Several members of Siga's 35-person research staff are housed at Molsoft's headquarters, and the firms are co-recipients of a number of research grants, albeit not the recent Air Force contract, which was awarded solely to Siga.
"Because of the rather unique situation with biodefense, we hope to move [the compounds] through the regulatory process more quickly — perhaps two-plus years."
Under the terms of the contract — awarded through the US Army Medical Research and Materiel Command, but funded through the Air Force — the company will work with Molsoft to design a drug to inhibit a smallpox receptor called G1-L. The companies will use Molsoft's software for a number of steps in the process, including homology modeling to reconstruct the receptor pockets on the target and computational docking to screen up to 70,000 compounds against those pockets.
"The first phase is building the model of your target," Abagyan explained. "The second is to pick where the druggable pocket is, then characterize that pocket, deal with issues of flexibility in that pocket, and finally, of course, dock small molecules to screen a large library."
Molsoft uses its suite of ICM (internal coordinate mechanics) tools for these projects. Abagyan said that he originally designed ICM as a global-optimization algorithm for peptide prediction, homology modeling, loop simulations, and docking. The 10-person company has more recently added new capabilities for graphics, molecular animations, chemistry, and sequence analysis, he said.
Kasten said that this process will generate around 100 compounds that will be tested in the wet lab, "and the most promising of those will undergo further development through animal testing, and then become drug candidates."
He said that the G1-L project "is a new application of a very proven pathway for Siga." The company currently has two lead antiviral compounds — one against smallpox called SIGA-246 and one against arenavirus called SIGA-294 — in its pipeline that were developed in a similar manner. Kasten said that the company is wrapping up preclinical work on these candidates in order to file investigational new drug applications.
A Quicker Path to Validation?
Kasten acknowledged that molecular modeling, virtual screening, and other computational methods lumped under the category of "rational drug design" have had their "ups and downs" in the drug-development sector, but added that the successful application of these approaches is "more in the practice and the practitioner than in the theory. I think there are bits of art that are important. Certainly Molsoft has those, and certainly we've developed some of our own."
Computational drug design is a staple in the pharmaceutical industry, but has not yet been fully validated due to the long timelines associated with the development and approval process. Even though computational approaches promise to reduce the time and costs required to identify a set of lead candidates, it has not yet been determined whether compounds that have been computationally designed to inhibit human proteins exhibit better efficacy or ADMET properties than those developed using traditional approaches like high-throughput screening.
Kasten said that the drive to ramp up countermeasures for biowarfare agents may hasten validation of the technology, however. "Traditionally, one looks at a rather protracted process of drug development spanning maybe four to five years in the laboratory and another eight years in clinical development. We're able to compress the time in the laboratory to a year," he said.
Furthermore, he said, "because of the rather unique situation with biodefense, we hope to move [the compounds] through the regulatory process more quickly — perhaps two-plus years."
One aspect of this "unique" situation, he said, is a decision by the US Food and Drug Administration in 2002 to loosen its regulations in order to speed the approval process for biothreat inhibitors. Under this move, the FDA now accepts efficacy data solely from animal models as part of the approval process, and has eliminated the requirement for human efficacy data.
Kasten said Siga believes the Department of Defense "will carry these things through to first use — to IND and ultimately NDA approval — because of their need for the protection of the warfighters is so compelling."
While Siga's computational approach has yet to be validated by IND or NDA approval, Kasten is confident that the company will see fewer failed trials "because we've been so meticulous in our design."
He said that Siga and Molsoft have designed the compounds "to have good oral availability. We've designed them to have absence of noxious induction of metabolites or enzymes that would alter metabolism. All those things are planned ahead, so we're sparing a lot of iterations on the chemistry side in determining the final compound to be advanced to a drug."
In the case of SIGA-246, he said, "We modeled it very successfully, and we've compared it with known successful drugs against about 15 different parameters, and those parameters would be ones that would predict noxious side effects — toxicity, metabolic changes that would cause trouble, those kinds of things — and it happens that Siga-246 falls right in the mean predicted for a successful drug. We find that it has a very clean toxicology profile."
Abagyan said that there are a number of "subtle" differences between the work Molsoft does for Siga and the work it does for its biopharma clients, such as Bristol-Myers Squibb, AstraZeneca, Vertex, and Biogen, which tend to concentrate on inhibitors for human proteins.
"When you target bacterial or viral proteins," he said, "you have to design with the large variability of your target in mind. Because it will not be just one viral protein — it will be rapidly mutating, and you need to study which things mutate at what rate, and try to design the targeting strategy that is least vulnerable. That's usually not a problem for human proteins, unless it's cancer."
In addition, he said, viral and bacterial proteins differ from human proteins in permeability and other properties that may have an impact on side effects.
Abagyan added that modeling may offer promise in the antimicrobial market because "some of the targets are so challenging that the only hope is to use these new computational approaches."
With modeling and virtual screening, he said, "even in the existing chemical space, I can screen a couple million credible candidates, while with the best high-throughput screening approach, it's almost an order of magnitude less. And if you add the virtual space — where you specify reactions and building blocks and generate molecules — it can be even larger numbers."
Going forward, he said, "the ability to go through an order or two orders of magnitude larger pools of molecules in the computer — of course, provided your method is good enough to distinguish right from wrong — is a great opportunity for the most difficult cases, and many viral and bacterial targets are those cases."
— Bernadette Toner ([email protected])
Will Federal Funds Help Siga Retain Nasdaq Compliance?
A day after Siga announced its $3.2-million contract with the USAF, the company disclosed that it had received two letters from the Nasdaq exchange regarding its compliance with listing requirements.
One letter was welcome news — a follow-up to an August letter in which Nasdaq notified the firm that it had violated a listing requirement to have a minimum of $2.5 million in stockholder equity, $35 million of listed securities, or $500,000 of net income for two of the three most recently completed fiscal years.
In early September, Siga responded to the notice, claiming that its stockholder equity as of Aug. 31 was "in excess of" $2.5 million and that the firm had not recognized $1 million in revenues in the quarter ended June 30 because it was awaiting approval for the second half of an NIH grant that began Aug. 1.
In the most recent letter, dated Sept. 22, Nasdaq notified Siga that the company "provided a definitive plan evidencing its ability to achieve and sustain compliance." The firm must prove that it is compliant in its filing for the quarter ended Sept. 30, however, "and if at the time of Siga's next periodic report Siga does not evidence compliance, it may be subject to delisting."
Siga received a second letter dated Sept. 22, which notified the firm that its share price had closed below the minimum $1.00 per-share requirement for 30 consecutive business days. The company has 180 calendar days, or until March 21, 2006, to regain compliance. To do so, its stock must close at or above $1.00 for a minimum of 10 consecutive trading days.
While the announcement of the USAF contract on Sept. 27 sent Siga's shares as high as $1.41 that day, to close at $1.07, word of the Nasdaq notices swiftly brought the price back down below the $1.00 threshold. The company's shares closed at $.96 on Sept. 28 and were at $. In mid-afternoon trading Sept. 29.