Sir

I, as an inventor of a patent held by the University of California on cloned human growth hormone complementary DNA, was a witness in a trial involving a patent dispute between the university and Genentech (see page 289 of this issue and ref. 1).

My testimony concerned events that occurred 20 years ago, my work at the University of California, San Francisco (UCSF) and early work at Genentech, which collectively resulted in the expression of human growth hormone (hGH) in bacteria. This pioneering work with my colleagues at Genentech was the culmination of three years of previous efforts at UCSF by which I and my colleague John Shine had succeeded in cloning the main part of the coding sequence for human growth hormone. It had been a difficult personal time as this project had often involved working nights, owing to efforts by my lab head to stop my research on growth hormone, as documented in the 1987 book Invisible Frontiers: The Race to Synthesize a Human Gene by Stephen S. Hall (Tempus Books of Microsoft Press).

As I testified during the trial, the Nature paper2 reporting the landmark study by Genentech regrettably contains a technical inaccuracy. This inaccuracy concerns a plasmid, pHGH31, which represents one of the intermediate steps in the construction of the expression vector for hGH. In this plasmid, the coding region for amino acids 24-191 plus 3' noncoding sequence, all contained on a 551-base-pair Hae III complementary DNA fragment, is inserted by ‘GC tailing’ into the Pst I site of pBR322. Not this plasmid, but an equivalent one carrying the same 551-base-pair Hae III fragment inserted by linkers in the Hin dIII site of pBR322 and previously constructed by me and Shine at UCSF, was used as source of the natural coding region for amino acids 24-191 in the construction of the final hGH expression vector.

The existence of pHGH31 is questioned by the fact, acknowledged by Genentech, that there never was a sequence record showing hGH DNA sequence attached to a G or a C tail, even though such a record should have existed, according to the Nature paper. Several attempts at Genentech by a colleague and me to obtain pHGH31 were unsuccessful, primarily due to the poor quality of the RNA starting material available to us at the time. With increasing pressure to complete the expression work, my colleague and I agreed to use the University of California's cDNA clone for part of the work.

To be absolutely clear, I, like my coauthors, view it as mandatory that publications are correct in all aspects, including all technical and methodological details. Hence, I deeply regret that, contrary to my own principles and the principles of scientific endeavour, the Nature paper contains a technical inaccuracy.

As I emphasized during the trial, all scientific results and conclusions of the Nature paper are unambiguous and correct. The expression vector is exactly as described and the bacteria make the correct hormone at the levels described in the publication. The study reported in the paper forms the basis for the first human growth hormone preparation free of neurodegenerative agents3, and the first recombinant therapeutic to be marketed by Genentech, from which 100,000 children benefit worldwide.