On Halloween night, while space warriors pranced outside, researchers began work, without fanfare, in the Harvard special containment laboratory for recombinant DNA research. The new facility, tucked away on the fourth floor of the Biological Laboratories, is subject to some of the most elaborate safety requirements specified by government DNA research guidelines. The lab, known as a P-3 facility because it requires a high level of physical containment, now houses two Harvard research teams. Behind the reinforced glass doors, entered with a special magnetic identification card, the scientists are experimenting with the genetic code. At the same time, debate continues over another kind of code--safety regulations Harvard scientists must follow as they conduct their research [see sidebar].

For the past year, the Holy Grail of genetic engineering has been the production of human insulin by E. coli bacteria cells. This work, if successful, would provide an almost unlimited source of natural human insulin for diabetics at a low price, a very tangible medical benefit which would go far towards convincing the public of the benefits of recombinant DNA research.

Exaggerated and contradictory headlines have chronicled progress in this research in Cambridge and California over the past year. Some observers, however, have felt the the scientists quest for insulin-production is more of a race. Researchers, they point out, are not ignorant of the potential prize of being first--lucrative contracts with drug companies and a likely Nobel Prize.

Critics point to one of the research groups working to induce bacteria to manufacture insulin which already has a contract with a large pharmaceutical company, Eli Lilly & Co. The speed at which the work has advanced--and possible shortcuts scientists may have taken--have raised doubts.

Walter Gilbert '53, American Cancer Society Professor of Molecular Biology, and his Harvard research team are working on the insulin problem. Gilbert has been using the MIT P-3 lab for the past year pending completion of the Harvard lab. This past July his group developed a bacterial strain which synthesized rat proinsulin, a modified form of insulin.

The central problem is finding a way to induce the bacterial cells to "express" the inserted foreign insulin genes. Researchers have to trick the cell into "reading" the added DNA along with the rest of its genes in order to translate its coded instructions for making insulin. And once the insulin is synthesized, the stage at which Gilbert's team is near, the researchers must smuggle the insulin safely out of the cell and isolate it in high quantity.

"We use a special trick to make the experiment work," Gilbert says. The insulin gene is inserted into a plasmid at a site before the end of another gene, which codes for a protein known as penicillinase. The insulin and penicillinase proteins are synthesized in a fused form when the genes are read together.

The penicillinase, a natural product of the bacteria, serves as a carrier and marker, transporting the insulin to the cell surface where it normally resides. At the surface, the penicillinase and insulin can be exposed to certain radioactively labeled substances which attach specifically to these two proteins. Bacterial cells which are successfully manufacturing the protein are thus identified.

One of the California groups, led by Howard Goodman and Bill Ruggers, inserted the insulin gene already in bacteria last year but they have been unsuccessful in getting the E. coli to read it, according to Gilbert. The other West Coast project, run by Genentech Inc. and an organic chemist, Dr. Keiichi Itakura, announced in September that it had successfully produced human insulin using E. coli bacteria.

Itakura and his associates reported that, "Artificial genes that 'command' laboratory bacteria to manufacture human insulin have been synthesized." Rather than using natural animal genes for insulin, this group built an artificial copy of the human insulin gene in two short segments and inserted these separately into E. coli plasmids.

The short insulin segments were hidden in large bacterial cell proteins during synthesis and safely escorted from the cells but the group has had problems putting the two insulin segments together in the proper three-dimensional conformation.

"They have managed to (produce human insulin), but it is inefficient," Gilbert said of the recent announcement. He said that the California artificial gene method was only applicable for a small gene such as insulin.

"One point of this work," Gilbert said, "is to make large amounts of proteins and to use insulin as a model system for proteins of any size." He added, "We're not limited by the size of the proteins."

Certainly, Gilbert's group is in competition with the California researchers but Gilbert smiles, hesitating to say that he is involved in a "race." The thrill of a good race appeals to him. Rather than being harmful to the research, such pressure probably stimulates better work, Gilbert says, adding, "It can be detrimental, or one can simply view it as part of the effervescence of the field of science. It's not a question of added pressure. I like to rush," he insists.

Yet, Gilbert says that there is something less than total cooperation between the groups attacking the insulin problem. While techniques are traded freely and he has communicated quite a bit with the artificial DNA group, he has been out of touch with the other California group which is employing an approach like his.