Promega Corp. v. Novagen, Inc.

6 F. Supp. 2d 1037, 1997 U.S. Dist. LEXIS 22355, 1997 WL 876025

Opinion

OPINION AND ORDER

CRABB, District Judge.

TABLE OF CONTENTS

I. STATEMENT OF CASE AND HOLDING.1039

II. EVIDENCE ADDUCED AT TRIAL.1041

A. Background.1041

B. Plaintiffs Claimed Inventions.1042

C. State of the Art of Protein Synthesis in October 1991 .1044

D. Person of Ordinary Skill in the Art.1045

E. Prior Art References.1045

1. Content.1046

2. Pelham .1046

3. Coen .1046

4. Stueber.1046

5. Perara and Lingappa.1047

6. Spirin and Baranov Papers.1047

7. Baranov Patent Application.1047

F. Plaintiffs Development of its Patents.1048

G. The Craig Paper.1048

H. Defendant’s Development of its Products.1049

I. Expert Witnesses.1049

J. Expert Opinions.1050

1. Bishop.1050

2. Anderson.1051

3. Mierendorf.1052

4. Dimond.1053

5. Jagus.1055

III. FACTS RELATING TO ALLEGATION OF INEQUITABLE CONDUCT.1056

IV. OPINION.1059

A. Inequitable Conduct.1059

B. Obviousness.1062

1. Coen reference.1064

a. Obviousness of independent claims.1064

b. Obviousness of dependent claims over Coen.1065

2. Stueber reference.1066

3. Perara and Lingappa reference.1066

4. Baranov patent application .1067

a. Obviousness of independent claims over Baranov.1067

b. Obviousness of dependent claims over Baranov.1069

C. Secondary Considerations.1071

D. Conclusion.1072

V. ORDER.1072

I. STATEMENT OF CASE AND HOLDING

Plaintiff Promega Corporation filed this civil action, alleging that defendant Novagen, Inc. was infringing plaintiffs rights under two patents for biotechnological inventions relating to protein synthesis, United States Patents Nos. 5,324,637 (’637) and 5,492,817 (’817), by manufacturing and selling a product called “Single Tube Protein System 2” or “STP2.” The STP product is designed for scientists engaged in research involving the synthesis of protein for the study of genes, gene expression and gene function. The product allows researchers to perform the two critical steps of translating DNA into messenger RNA and then transcribing the messenger RNA into protein in one single tube reaction. The ability to couple the two steps allows researchers to synthesize protein with much greater ease and speed, thus enabling them to undertake more studies and to test a greater number of hypotheses about gene function and expression.

In developing its product, defendant purchased a similar product known as TNT that had been manufactured by plaintiff, “deconstructed” the TNT to determine how it was produced and then essentially copied the TNT product. Such a procedure is not objectionable unless the TNT product is covered by a valid patent. Plaintiff believes it is: it has two patents that cover the product and that are presumed valid by virtue of their issuance by the United States Patent and Trademark Office. It has sued defendant for infringement. Defendant contends that neither of plaintiffs patents, ’637 or ’817, was valid for one or more of the following reasons. 1) Every element of every claim of plaintiffs patents could be found in public documents published before plaintiff applied for its ’637 patent in October 1991 (anticipation), 35 U.S.C. § 102; 2) prior art references taught or suggested to persons of ordinary skill in the art of developing kits such as STP and TNT the inventive concepts of the claims of the ’637 and ’817 patents (obviousness), 35 U.S.C. § 103; 3) plaintiff engaged in inequitable conduct in the prosecution of the ’637 patent. In addition, defendant denies that its STP2 product infringed the claims of plaintiffs patents.

After conducting discovery, the parties filed cross-motions for summary judgment on the validity, enforceability and infringement of the ’637 and ’817 patents. In an order entered February 28, 1997, I held, among other things, that defendant’s sale of STP2 and the protocol of defendant’s product infringed a number of the claims of both the ’637 and the ’817 patents, but, because there were disputed issues of fact, I denied summary judgment on whether plaintiff had engaged in inequitable conduct and whether certain prior art references anticipated or rendered plaintiffs patent claims obvious.

With the only issues left for resolution at trial being defendant’s affirmative defenses of anticipation, obviousness and inequitable conduct, defendant presented its evidence first. At the close of defendant’s case-in-chief, plaintiff moved pursuant to Fed. R.Civ.P. 50(a) for judgment as a matter of law. The motion was granted in part: defendant’s anticipation claims were dismissed in full on the ground that the evidence was insufficient to support a jury finding that any of the prior art references always and necessarily resulted in a coupled transcription-translation reaction so as to anticipate plaintiffs invention inherently. Certain of defendant’s obviousness claims were left for the jury to decide. The jury determined that four prior art references (known as “Bara-nov,” “Coen,” “Stueber” and “Perara and Lingappa”) each rendered obvious the independent claims of plaintiffs patents (claims 1, 34 and 68 of the ’637 patent and claims 1 and 13 of the ’817 patent), that the Baranov reference rendered obvious the dependent claims of the ’637 patent directed to specific concentrations of magnesium, potassium or ribonu-cleotide triphosphates, or NTPs, (claims 3, 4, 8,17, 37, 38, 48, 49 and 51) and that the Coen reference rendered obvious dependent claims 12, 13, 48 and 49 of the ’637 patent, but did not render obvious dependent claims 11 and 46 of that patent. Earlier in the trial, I determined as a matter of law that plaintiff had not engaged in inequitable conduct when it submitted the Second Declaration of Dr. Gregory Beckler to the patent office.

After trial, defendant filed a motion pursuant to Fed.R.Civ.P. 52, asking the court to reverse its decision on its inequitable conduct ruling with respect to the Second Declaration of Dr. Beckler, on the ground that Beckler’s trial testimony showed that plaintiff submitted his second declaration with the intent to deceive the patent examiner. For its part, plaintiff renewed its motion for judgment as a matter of law on defendant’s obviousness claims and filed another motion seeking a new trial or other appropriate relief under Fed.R.Civ.P. 59, contending that the evidence adduced at trial was insufficient to support the jury’s determination that the patents at issue were invalid for obviousness.

The ease is now before the court on the motions of both parties. I conclude that defendant has not shown any reason to reverse the determination that plaintiff did not engage in inequitable conduct but that plaintiff has shown that the jury’s verdict on obviousness cannot stand. No reasonable jury could have found by clear and convincing evidence that the claims of plaintiffs ’637 and ’817 patent are made obvious by any one of the Stueber, Coen, or Perara and Lingap-pa papers or the Baranov patent application.

To give a fair picture of the evidence that was before the jury, it is necessary to set it out in considerable detail. The essential question in deciding a motion for judgment as a matter of law is whether the evidence the jury could have believed in reaching its verdict was substantial enough to support its findings. Orthokinetics, Inc. v. Safety Travel Chairs, Inc., 806 F.2d 1565, 1573 (Fed.Cir. 1986). The question is not what the court might have believed, but what the jury could reasonably have determined. Id.

II. EVIDENCE ADDUCED AT TRIAL

A. Background

As all television watchers now know, every cell in every living organism contains DNA, the master copy of all the genetic information in the cell. Each gene contains information that can be used to synthesize a single, specific protein that has a particular purpose in the body. It might serve as a part of the structure of the cell, for instance, or as an enzyme to make the other chemical constituents of cells. This protein must be synthesized continuously in order to enable the organism to stay alive.

In the protein synthesis process, the DNA serves as the template for making messenger RNA, which is the intermediary between the DNA and the protein. The protein is actually a sequence of amino acids that join together and fall into a particular structure determined by the RNA, which is carrying the “message” of the sequence of nucleotides found in the DNA. The amino acids join to create a peptide bond; many peptide bonds make up a polypeptide; a polypeptide is a denatured form of protein made up of a strung out series of amino acids bound by peptides. For the purpose of this case, a protein has been defined as a full-length polypeptide that comes off the ribosome and folds correctly. (The ribosome is an intracellular particle consisting of several RNA molecules and proteins involved in protein synthesis; its function is to synthesize polypeptide chains having amino acid sequences specified by the genetic code of the messenger RNA.) The process of copying the message of the DNA is known as “transcription”; the process of making new protein from the transcript is known as “translation.”- Transcription must precede translation.

Broadly speaking, living organisms are either prokaryotic or eukaryotic. (A third kind has been identified but is irrelevant to this case.) Prokaryotic organisms such as bacteria have cells with no nucleus. Euka-ryotic organisms have cells that have a nucleus. In the cells of prokaryotic organisms, the two processes of transcription of the DNA into RNA and the translation of the RNA into protein occur in one place: within the cytoplasm. In eukaryotes, by contrast, transcription of RNA from DNA incurs within the nucleus where the DNA is contained and translation of the RNA into protein takes place outside the nucleus, within the cytoplasm.

Replicating the natural process of protein synthesis within the laboratory (in vitro, literally, in a glass) is a complex task requiring the reproduction of the conditions that are present in nature (in vivo). Plaintiff claims as its invention a method and a kit for enabling scientists to synthesize protein by performing in one test tube both the process of transcription (copying a particular DNA onto a messenger RNA) and that of translation (converting the information from the messenger RNA into a sequence of amino acids that form a protein). Laboratory transcription and translation is not a new concept; it has been practiced for many years in various forms. Plaintiffs patents claim methods of doing coupled transcription and translation in a eukaryotic cell-free extract and a kit for performing the coupled reaction.

Plaintiff does not claim to have invented the idea of transcribing and translating DNA to make protein, but to have made the process more reliable, faster and easier by enabling both transcription and translation to occur simultaneously in one reaction within a test tube. Its claimed inventive concept is the discovery that a researcher can do transcription and translation in a single test tube if the amount of magnesium in the reaction is kept within a certain range. Plaintiff argues that until this invention, no one had either identified the critical significance of magnesium to a coupled reaction or determined the optimal range of magnesium concentration.

For the first stage of transcription of DNA to messenger RNA to occur, it is necessary to have a template DNA, for the hard copy of the genetic information; an enzyme or polymerase, which is a protein of complicated structure that recognizes a particular sequence called the promoter on the DNA and begins copying there; precursor nucleotides (ribonucleotide triphosphates) that are the building blocks of the RNA; and an ionic environment that includes salt (a two-component compound) and more complicated salts, such as magnesium. Having sufficient magnesium is essential to the transcription reaction. The magnesium tends to be bound up by the nucleotides in the reaction, which are the precursors for the RNA. Without an excess of magnesium, the nucleotides will soak up the magnesium, leaving none for the enzymes that need magnesium as a cofactor for activation. The transcription process begins as the polymerase scans and recognizes a starting place on the DNA and then copies that DNA, putting nucleotides into transcript or messenger RNA until the polymerase comes to a DNA-sequence that tells it to stop. At that point, the transcript of the gene is terminated.

For translation to occur, it is necessary to have messenger RNA; the ribosome machinery, which is generally in the form of a lysate made from wheat germ or rabbits (a lysate is a mixture of substances formed by cells that have been destroyed or dissolved; another name for these lysates is “cell free extract”); amino acids delivered to the ribosome (transfer RNA); nucleotides for an energy source; and an ionic environment that includes magnesium. The lysates all contain some amount of magnesium but until plaintiff made what it claims to be its discovery, most people did not measure how much might be in any particular lysate; they were concerned primarily with the amount of added, magnesium (the exogenous magnesium rather than the endogenous magnesium).

A researcher who performs transcription and translation cannot just look at the final product and know whether RNA has been transcribed or protein has been produced. In order to evaluate the outcome, the researcher must start with a radioactive nucleotide. If this nucleotide is incorporated into the string of RNA, the researcher can confirm the synthesis of RNA by measuring the incorporation of the radioactive nucleotide into the product, using a machine that measures radioactivity. In the same way, by using a radioactive amino acid in the translation mixture, the researcher can check for the presence of that radioactive amino acid in the resulting product to determine whether any protein has been produced. To determine what protein has been produced, the researcher uses a procedure called electrophoresis, in which denatured proteins are loaded into a hole in a gel and the gel is put into an electric field. The gel is a sieve, made up of many different sized holes. Small proteins will make their way through the gel quickly; larger ones will move more slowly because their way will be blocked by some of the smaller holes, requiring them to move sideways to find a big hole. The researcher exposes the gel to an x-ray film to produce a picture for analysis. Knowing that each amino acid has a different molecular weight, the researcher can add the sizes of each amino acid and predict the size of the resulting protein. If the bands shown on the x-ray turn out to be the size predicted, that is an indication that the right kind of protein is being produced.

B. Plaintiffs Claimed Inventions

On October 11, 1991, plaintiff applied for a patent on coupling transcription and translation in one reaction; on- March 10, 1992, the patent examiner rejected nearly all the claims for the stated reason that they were anticipated by the Baranov et al. patent application and obvious over Baranov, in view of Suzuki. Plaintiff responded that Baranov did not teach a batch reaction, but the examiner was unconvinced and rejected the application a second time. Plaintiff appealed the rejection to the Patent Office Board of Appeals and Interferences, but withdrew its appeal in February 1994 and filed a continuation application, limiting its claims to batch reactions. Eventually, on June 28, 1994, the ’637 patent issued. The relevant claims read as follows:

I. A method for coupling transcription and translation in a cell-free extract derived from cells from the group consisting of plant and animal cells in a static reaction to produce protein, said method comprising

adding a DNA template to said extract, adding ribonucleotide triphosphates to said extract,

adding a RNA polymerase to said extract, and

adding a sufficient amount of a magnesium salt to said extract to raise the magnesium concentration to a level where RNA is transcribed from said template DNA and the RNA translates into said protein.

3. The method of claim 2 wherein said final magnesium concentration is about 2.5 mM [millimolar] to about 3.5 mM.

4. The method of claim 2 wherein said final magnesium concentration is about 2.6 mM to about 3.0 mM.

8. The method of claim 2 wherein the potassium concentration of said lysate is adjusted to about 40 mM to about 100 mM.

II. The method of claim 2 wherein said polymerase is selected from the group consisting of SP6, T7 and T3 RNA polymerases.

12. The method of claim 2 wherein said polymerase is selected from the group consisting of SP6, T7 and T3 RNA polymerases.

13. The method of claim 12 wherein a polymerase promoter sequence is located at one end of said multiple cloning region. 17. The method of claim 2 wherein 0.4 mM of each of said ribonucleotide triphosphates are added to said lysate.

34. A method for producing protein from a DNA template having a specific polymerase promoter sequence through coupled transcription and translation in a batch reaction, said method comprising the steps of preparing a solution of eukaryotic cell-free extract of cells selected from the group consisting of plant and animal cells, modifying said extract solution with sufficient concentrations of said template DNA having a specific polymerase promoter sequence, ribonucleotide triphosphates, amino acids and a polymerase corresponding to said promoter sequence of said template DNA, and adding a sufficient amount of a magnesium salt to said extract solution to raise the final magnesium concentration to a level where RNA is transcribed from said DNA template and said RNA translates into protein.

37. The method of claim 36 wherein said final magnesium concentration is about 2.5 mM to about 3.5 mM.

38. The method of claim 36 wherein said final magnesium concentration is about 2.6 mM to about 3.0 mM.

48. The method of claim 36 wherein said DNA template has a multiple cloning region.

49. The method of claim 48 wherein a polymerase promoter sequence is located at one end of said multiple cloning region.

51. The method of claim 36 wherein 0.4mM of each of said ribonucleotide triphosphates are added to said solution.

68. A kit for producing protein from a DNA template through coupled transcription and translation, said kit comprising the following components adapted to be used in a batch reaction: eukaryotic cell-free extract of cells selected from the group consisting of plant and animal cells, ribonucleotide triphosphates, RNA polymerase, and magnesium salt at a concentration whereby RNA is transcribed from DNA and RNA translates into protein.

In February 1996, plaintiff was granted its ’817 patent. The relevant claims of that patent read as follows:

1. A method for coupling transcription and translation in a eukaryotic cell-free extract to produce protein in a static reaction comprising the steps of:

adding a DNA template to the extract;

adding ribonucleotides triphosphates to the extract;

adding a RNA polymerase to the extract; and

adding a sufficient amount of magnesium salt to the extract to raise the magnesium concentration to a level where RNA is transcribed from the DNA template and RNA translates into protein.

3. The method of claim 1 wherein the RNA polymerase is selected from the group consisting of SP 6, T7 and T3 RNA polymerases.

13. A kit for producing protein from a DNA template through coupled transcription and translation, said kit comprising the following components adapted to be used in a batch reaction: eukaryotic cell-free extract, ribonucleotide triphosphates, RNA polymerase, and magnesium at a concentration whereby RNA is transcribed from the DNA template and RNA translates into protein.

The specifications of both the ’637 and the ’817 patents teach that the critical component in successful coupling of transcription and translation in eukaryotic systems is the concentration of magnesium ions in the reaction mixture. Magnesium occurs naturally in most rabbit reticulocyte lysates, typically in the range of 1.6 to 5 mM. Production of protein does not occur when magnesium concentrations present in the standard lysate are left unchanged. The inventive concept claimed in the ’637 patent is that the control of the magnesium concentration makes the difference between the success and failure of the coupled transcription and translation reaction. The meaning of “coupled transcription and translation” has been construed as covering any reaction in which “any” simultaneous transcription and translation occurs. Order of Feb. 28,1997 at 78.

C. State of the Art of Protein Synthesis in October 1991

Before plaintiff applied for a patent in October 1991, the state of the art for doing protein synthesis of eukaryotic DNA was to perform two separate sequential batch reactions, each optimized for its particular purpose of transcription or translation. Researchers tried various forms of optimization, such as using more magnesium in the first reaction than in the second or raising the temperature for the first reaction and lowering it for the second.

When researchers do separate transcription and translation reactions, they do a first reaction in which DNA is transcribed into RNA in a transcription reaction. They then add RNA from the transcription reaction to a second reaction for translation. Plaintiff marketed two-step transcription and translation kits for some time before filing the patent application that became the ’637 patent.

Before October 1991, researchers could use commercial products based on bacterial systems for synthesizing message from DNA, as well as procedures or protocols (the scientific version of a recipe) for doing such reactions. The researchers would make up a transcription reaction consisting of nucleotides; ions, including magnesium; a DNA template; and a RNA polymerase. The resulting transcription would continue for a matter of hours. When it was done, the researchers could extract the RNA, that is, purify it, before adding it to the translation reaction. Alternatively, they could add a small portion of the transcription reaction to the translation reaction, as described in the instructions published by Boehringer Mannheim, a company that, like plaintiff, manufactured a number of kits for performing transcription or translation. The instructions for one of the Boehringer Mannheim kits (exh. # 1065) advised researchers that a small amount of undiluted transcription reaction would stimulate translation efficiently when added to the translation reaction. The company advised researchers that they could run a transcription reaction, take some of the product and add it to Boehringer Mannheim’s rabbit reticulocyte lysate and produce protein. The company cautioned, however, that as the amounts of the transcription reaction increased, translational activity tended to decline, for two reasons:

firstly 1 [microliter] of the transcription reaction contains about the amount of RNA that is needed for efficient translation in the lysate. Large amounts of RNA have been shown to reduce translation. Secondly, the transcription reaction contains [magnesium chloride], 6 mmol/1, and m7G(5')G, 500 ¡xmol/1. Both reagents have also been shown to reduce translational activity when their concentrations are increased in the [translation reaction],

Boehringer Mannheim recommended adding EDTA to the transcription reaction before adding a small amount of the transcription reaction to the translation reaction. (EDTA binds up magnesium, leaving less free magnesium when the transfer of the transcription product is made to the translation system.)

Coupled transcription-translation systems were available or known in 1990 for proka-ryotic (no nucleus) systems and for vaccinia core (a DNA virus). Around 1990-91, persons skilled in the art thought that transcribing and translating required two separate batch operations and that if researchers failed to optimize either reaction, they would not get as much protein.

In 1990-91, plaintiff produced and sold a two-reaction kit (which is still available). If a researcher were to put both batches into the same test tube at the same time the reaction would not work because the ionic conditions are different for each step. The optimum magnesium concentration for the first reaction is six millimolar; the optimum for the second reaction is around two millimolar.

Researchers encountered certain problems in performing two-step reactions, particularly if they followed the procedure of extracting RNA before beginning the translation process. RNA is very susceptible to degradation by RNase, an enzyme that is ubiquitous in laboratories; the compounds needed for purifying the RNA away from the protein are caustic; and the linearizing of the DNA using a restriction enzyme is time-consuming.

Researchers have been interested in finding a way to combine the two steps of transcription and translation in a coupled (one-step) eukaryotic protein synthesis system since the early seventies. Their interest increased in the middle and late seventies, when the first cloning plasmids came into existence and researchers were able to clone copies of messenger RNA that existed within cells. (Plasmids are pieces of DNA found naturally in certain bacteria that can be cut to insert a protein, thereby enabling the researcher to study the expression of the gene as the plasmid replicates itself in the bacterial cells.) Around this same time, two-step reactions were developed for protein expression in eukaryotic extracts. The early versions used E. coli or RNA polymerase and eukaryotic translation extracts; researchers performed a separate transcription reaction and then transferred the messenger RNA into a translation reaction. In the early eighties, the development of phage polymerases had a strong effect on transcription reactions. Unlike E. coli RNA polymerase, which recognized not only the promoter in front of the gene to be studied but promoters in front of other genes, the phage polymerases were extremely specific for their promoters and had high transcriptional activity.

D. Person of Ordinary Skill in the Art

The “art” is the design of products such as the TNT kit that plaintiff sells to researchers to enable them to perform coupled, single batch transcription and translation. The person of ordinary skill in this art is one who has either a doctoral degree or a bachelor’s degree plus several years of experience in the field of molecular biology.

E. Prior Art References

Plaintiff cited a number of references in its patent application, including a patent application by Baranov et al. (Russian PCT patent application, W091/02076 (European Patent No. 0593757A1), an article by Spirin, Baranov and others in the journal Science, vol. 242, pp. 1162-1164 (1988); an article by Baranov et al. in the journal Gene, vol. 84, pp. 463-66 (1989)); and others. Plaintiff did not cite any articles by Coen, Stueber or Perara and Lingappa. (Defendant contends that these previously published articles and patent application would have taught or suggested to the person of ordinary skill in the art both that transcription and translation reactions could be performed in one step and the optimum magnesium concentration for making a one step reaction possible and effective, thus making plaintiffs invalid under 38 U.S.C. § 103, which prohibits obtaining a patent “if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains.” Defendant argued also that plaintiffs failure to cite Coen, Stueber and Perara and Lingappa in its patent applications rendered its patents invalid but this argument was rejected in the February 28 order.)

1. Content

As early as 1977, it was known that genes of the influenza virus RNA could be transcribed and translated from RNA using nuclease-pretreated rabbit reticulocyte lysate (that is, lysate pretreated to remove any messenger RNA so as to make the lysate highly sensitive to exogenous messenger RNA). A 1977 article by Content (exh. # 1021) describes the process as using a concentration of magnesium intermediate between the optimum for transcription and that for translation, or between five millimolar for translation and 1.5 to two millimolar for transcription. The Content article made no reference to transcribing DNA.

2. Pelham

Around 1978, an article by Pelham (exh. # 1048) discussed a coupled transcription and translation system for vaccinia cores. The reaction starts with the transcription of viral DNA and ends with protein. A scientist reading the article would understand when working with vaccinia cores that if a system for making messenger RNA is provided, messenger RNA can be translated in the same tube in which transcription takes place. According to the article, the transcription mix was incubated for 45 minutes and then diluted fivefold into a translation mix prepared as for coupled synthesis but lacking the magnesium-balanced nucleotide triphosp-hates contained in the transcription system. Pelham taught the addition of actinomycin D to inhibit transcription.

3. Coen

An article by Coen et al. (exh. # 1089), published in 1977, describes a linked transcription and translation procedure in which a transcription reaction using E. coli RNA polymerase is optimized and a small portion of the transcript product is taken out and diluted into a translation reaction optimized for translation. Coen shows a concentration of ten millimolar magnesium acetate for the transcription reaction and only two millimo-lar magnesium in the translation reaction. It is unclear from Coen whether the two milli-molar magnesium in the translation reaction is added magnesium or magnesium carried over from the transcription reaction into the translation reaction. Coen shows the four nucleotide triphosphates at a half millimolar in the transcription reaction and an unknown amount in the translation reaction equal to 20 percent of whatever was left over in the transcription reaction, with no nucleotides added. Coen adds EDTA (a magnesium binding chemical) in an apparent attempt to bind some of the magnesium coming from the transcription reaction. The transcription-translation system is described as linking. There is no reference to coupling and no reference to transcription taking place in the translation reaction.

4. Stueber

In a 1984 article, Stueber et al. (exh. # 1062) described an accurate and efficient two-step synthesis of single proteins from cloned DNA sequences in a test tube. The transcription process uses a template DNA with a promoter derived from bacterial viruses, an RNA polymerase derived from E. coli bacteria, with a buffer and in the presence of 10 millimolar magnesium acetate, other ions, four nucleotides, spermidine, and DTT, plus a ribonuelease inhibitor to prevent degradation of RNA. Stueber reported that after the transcription reaction had run for 20 minutes, he put the mixture on ice to cool it and then used some of it in a translation assay, using a translation system derived from wheat germ (wheat germ lysate) and 2.8 millimolar magnesium acetate in addition to whatever magnesium was in the wheat germ lysate. The translation process contin ued for 60 minutes. No transcriptional inhibitor was added. Stueber said nothing in the paper about transcription occurring during the translation reaction and included no data to show that there would have been transcription in the second reaction.

Stueber’s paper focused on the separate transcription reaction and the use of T5 RNA polymerase promoter with E. eoli RNA polymerase in an optimized transcription reaction. Stueber added the nucleotide ATP to the translation reaction after it started; the nucleotides GTP, CTP and UTP were carried over from the transcription reaction. He did the two reactions at different temperatures: 37 degrees C. for transcription; 25 degrees C. for translation.

5. Perara and Lingappa

In an 1985 article, Perara and Lingappa (exh. # 1050) described a two-step transcription and translation process using rabbit reti-culocyte lysate (made according to a procedure reported by Merrick) as the translation machine. The authors used DNA templates and a bacteriophage promoter. The transcription reaction was conducted for 60 minutes, after which a portion of the volume was transferred to a translation reaction without the addition of any transcriptional inhibitor. Perara and Lingappa reported .2 millimolar magnesium chloride in the reaction mixture. Id. at 2294. The reference does not inform a reader of the total magnesium concentration and it does not report any transcription occurring in the translation reaction.

Perara and Lingappa used two separate transcription and translation reactions, with both reactions optimized for the particular functions they were to perform. Like Stue-ber, Perara and Lingappa did the two reactions at different temperatures (40 degrees C. for transcription; 24 degrees C. for translation).

6. Spirin and Baranov Papers

In a 1988 paper in the journal Science, Spirin, Baranov and others wrote about a continuous flow translation system they had developed and which they said was a new method of doing in vitro translation that overcame problems that had existed for 20 years in the art of in vitro translation. The system involved adding messenger RNA to a prokaryotic extract, rather than using DNA templates. According to the authors, their continuous flow translation system enabled protein synthesis to continue for long periods of time.

In a 1989 article in Gene, Baranov and Ryabova wrote about the use of rabbit reticu-locyte eukaryotic extract for continuous translation. This was not a coupled system because it did not transcribe RNA. Instead, the RNA was added to a continuous translation system. Again, the system enabled protein synthesis to continue for a long time (as long as 90 hours) and produced about 5,000 times as much protein as other methods.

7.Baranov Patent Application

An international version of a Russian patent application by Baranov and others in the Spirin group was published in February 1991 (exh. # 1109). The application disclosed a continuous flow reaction for obtaining polypeptides in a cell-fi*ee system of coupled transcription and translation in both prokaryotic and eukaryotic extracts, conducted in a single reaction vessel. Baranov et al. took a cell-free extract such as rabbit reticulocyte lysate; a DNA template; nucleotides (the precursors to make RNA); amino acids (the precursors to make proteins); RNA polymerase; magnesium and other salts and placed all the ingredients into a reaction vessel, which was hooked up to a system that allowed new nucleotides, amino acids and magnesium to come into the system and subtracted spent, free agents, the derivatives, unused reagents and the protein product. Baranov described several different examples of the use of the system. Example 4 used rabbit reticulocyte lysate as the translation machinery, with RNA polymerase from bacteriophage SP6. The purpose of the continuous flow process is to remove products from the reactor through a semipermeable membrane while adding substrates simultaneously in order to maintain their original concentration. According to Example 4, 1.5 micromolar of magnesium was added at the outset. The example does not describe the preparation of the reticulocyte lysate used in the experi ment. The patent application does not describe any batch reactions.

F. Plaintiffs Development of its Patents

Gregory Beckler was the project leader of a research team at the plaintiff company that included Tom Van Oosbree, John Van Her-wynen and Dave Titus. The team worked unsuccessfully for over a year trying to develop Baranov’s continuous flow system as a commercial product. It undertook continuous flow experiments in an Amicon 8 MC stirred cell unit that has a hole for adding the feed solution (buffer A) into one chamber of the unit. The incubation mixture goes through another hole and the stir bar is activated to help force solutions through the membrane in order to remove spent ingredients and product. The unit is kept in a cold room because the product coming out into the tubes must be kept at 4 degrees C. The stirred cell unit keeps the actual incubation mixture at a much higher temperature (34 degrees C. for rabbit reticulocyte lysate).

The team experienced problems trying to replicate the Baranov continuous flow reaction. Because the reactions are fairly viscous, the membranes would clog, preventing the reaction from maintaining a constant volume. It took several hours until the proteins from the extract would stop coming out in the fractions coming off the continuous flow system. Plaintiff was never able to obtain the results reported in Baranov, although the team tried other devices and technologies and experimented with changing the variables from the amounts suggested. At best, the protein production never exceeded by very much what could be produced in a standard two-batch reaction; at worst, it was non-existent. Plaintiff abandoned work on the continuous flow project in March 1991.

In April 1991, after attending a meeting at which Spirin presented the continuous flow technology, Van Oosbree tried a coupled batch reaction, using the most sensitive reporter (luciferase) available. The experiment was a failure; no protein was produced. In a later experiment, however, Van Oosbree buried the concentration of the lysate and diluted it and made more protein. At that point, the team started focusing research efforts on inhibition. By chance, Dave Thompson tried a buffer known as TA that is used for restriction enzyme digestion and found that when he used it in a two-step batch reaction, he got rid of the inhibitory effects on translation. His initial thinking was that the inhibitor that was removed was the magnesium. On June 20, 1991, Thompson succeeded in making a coupled one-batch reaction, using the TA buffer.

Between April and June 1991, a group made up of Van Oosbree, a representative from plaintiffs research and development department and another from marketing voted down Beckler’s proposal to develop a coupled batch reaction, on the ground that it was not feasible and would take too long to develop.

Earlier, on May 13, 1990, Randall Dimond, plaintiffs Chief Technical Officer, had sent an e-mail message to Van Herwynen, stating:

Neither of the eukaryotic systems has transcriptional activity. However, I think we could possibly make them into a coupled assay by adding T7 or SP6 polymerase and appropriate buffers and nucleotides, etc. Dave Mead made the observation that when he added the whole Riboprobe transcription reaction (after transcription) to Lysate he got more protein synthesis than if he just added the RNA that had been produced. He felt this probably meant that he was continuing to get transcription after the Riboprobe mix had been added to the Lysate.

Plaintiff filed its initial patent application in October 1991 and produced its first TNT system in February 1992. In the first twelve months of production, the TNT product had sales of $600,000, and in the second year, $1 million. In 1993, plaintiff received an award for having one of the 100 most innovative products for that year.

G. The Craig Paper

In a 1992 article, Craig et al. described the authors’ successful attempts to combine coupled transcription and translation in a single reaction and noted that this had not been attempted previously, “possibly because it was thought that the optimum conditions for in vitro transcription and in vitro translation are so very different that no single set of conditions could be found under which both processes would function with reasonable efficiency.” (Exh. # 61 at 4987)

H. Defendant’s Development of its Products

Defendant never made any attempt to replicate the Baranov continuous flow process or to convert it to a batch process before plaintiff applied for its patent. When Robert Mierendorf, defendant’s vice-president, directed defendant’s scientists to conduct experiments to develop the product that became STP1, he did not give them the Bara-nov patent application, but the Craig paper that was written after plaintiffs patent application was filed.

Van Oosbree worked for plaintiff from 1984 to 1992 before going to work for defendant, where he was one of the scientists assigned to develop the STP product. His experiments in conjunction with the Craig paper did not work well, so he purchased a TNT kit produced by plaintiff, measured the endogenous magnesium in the kit, determined the nucleotide concentrations and adjusted defendant’s Red Nova lysate so that it would end up with a concentration similar to the TNT product.

When defendant produced its infringing product, STP (Single Tube Protein), it touted the advantages of the product as being more convenient, faster and capable of producing higher yields than the standard in vitro transcription and translation systems. In the accompanying manual, defendant recommended 2.5 mM magnesium concentration as the final magnesium concentration (the total magnesium in the system). In an advertisement, defendant described its system as “the next generation in coupled t ranscription/translation for rapid gene analysis.”

After plaintiffs ’637 patent issued, plaintiff wrote to defendant to advise it that defendant’s STP system was infringing the claims of the patent. Defendant withdrew its product from the market. In 1994, defendant introduced a new product, now referred to as STP2. Defendant tried to design the product to avoid infringement. The product is designed for in vitro synthesis of proteins from DNA templates containing a bacteriophage T7 or SP6 RNA polymerase promoter and can be used to express proteins from supereoiled plasmids. Both systems use a two-step reaction. The kit user combines a transcription mix including ribonucleotide triphosphates, RNA polymerase and magnesium salt with a DNA template and performs a transcription reaction for 15 minutes. In this step, no simultaneous transcription and translation occur. Following the completion of the transcription reaction, the user adds a sample of the reaction products from the transcription reaction, in unpurified form, to a translation mix consisting of rabbit reticu-locyte lysate, amino acids, and nucleotides. This translation step continues for 60 minutes. The translation reaction is a static or batch reaction. During the translation reaction, there is sometimes simultaneous transcription and translation, but not always. When the ingredients are combined in the second step, the magnesium concentration is 2.74 mM.

I. Expert Witnesses

(Much of the expert witness testimony focused on the question of anticipation under 35 U.S.C. § 102, that is, whether a single prior art reference anticipated plaintiffs invention by describing every element of the invention before plaintiff invented it and applied for a patent. I concluded at the end of defendant’s case that defendant had failed to show invalidity of the patents by anticipation; for this reason, I have not summarized the witnesses’ testimony on anticipation but have limited the description of their testimony to the issues of obviousness: whether earlier published articles or the Baranov patent application taught or suggested to one of ordinary skill in the art that both transcription and translation could occur in a coupled batch reaction in a single test tube, that the total magnesium concentration was crucial to the success of a coupled reaction and that the optimum magnesium concentration that would enable such a coupled reaction to occur would be between 2.5 mM to 3.5 mM total magnesium concentration (including the magnesium in the lysate and added magnesium).)

David Bishop is a doctor of philosophy employed by the University of Oxford, Ox ford, England. His life work has been the study of RNA viruses, including influenza. He has published about 800 scientific papers in the general field of molecular biology, is the editor of a monthly magazine and a reviewer of articles for a number of other journals. He is the named inventor on two patent applications involving the expression of protein synthesis in cells.

Carl W. Anderson is a senior scientist at the Brookhaven National Laboratory. He earned his Ph.D. in microbiology in 1970 and did postdoctoral work at the Cold Spring Laboratory on Long Island before joining Brookhaven in 1975.

Robert Mierendorf earned his Ph.D. in 1980, investigating the effects of hormones on transcription in cancer cells. He did postdoctoral work at the University of Wisconsin and joined plaintiff in 1984 as a senior scientist and director of research and development. While in plaintiffs employ, he led the group that did custom services for academic laboratories and others and he helped design certain products for in vitro transcription employing bacteriophage SP6 and T7 RNA polymerase. When plaintiff decided to terminate the activities of the custom service group in 1989, Mierendorf left and helped form the defendant company.

Gregory Beckler is employed as a senior research scientist in plaintiffs research and development department. He has a Ph.D. in cellular development biology and has published a number of articles in peer review journals.

Randall Dimond is plaintiffs vice president and chief technical officer. He earned a Ph.D. at the University of San Diego in 1975 and had additional training at the Massachusetts Institute of Technology in the area of in vitro protein synthesis. He joined the University of Wisconsin in 1977 as an assistant professor in the Department of Bacteriology. He left the university in 1984 to become vice president in charge of research at the plaintiff company.

Rosemary Jagus is a researcher at the Center of Marine Biotechnology, which is part of the University of Maryland. She has a Ph.D. in biochemistry and has worked at the National Institutes of Health and the University of Pittsburgh. Her research focuses on the regulation of protein synthesis.

J. Expert Opinions

1. Bishop

Bishop believes that it would have been obvious to anyone in the field to do a coupled batch reaction from reading the Baranov patent application for a continuous flow system. He would read the application as instructing the researcher to put together a volume of rabbit reticulocyte lysate and an equal volume of twofold concentration of buffer A, so that the final concentration of buffer A is onefold. In Bishop’s view, Baranov teaches that all of the ingredients listed in Baranov are present at the outset, making Baranov a batch reaction. It is only when the machine is turned on that the continuous flow reaction begins.

Bishop agrees that the way in which Bara-nov’s Example 4 is worded leaves it unclear whether the concentrations listed after the words “buffer A” are the total concentrations for the incubation mixture plus the buffer or just for the buffer. However, the application provides that the magnesium acetate concentration is maintained throughout the process and makes no such provision for any other form of magnesium, leading Bishop to think that the reference is to added magnesium and not just that in the lysate. Example 4 lists the constituents of the reaction as follows:

1 ml of the incubation mixture contains 600 (o-l of lysate from rabbit reticulocytes, o.l ml of plasmid with the gene of chloram-phenicol acetyl transferase under the promoter of SP6 polymerase obtained according to the method described (Molecular Cloning, 1989, Cold Spring Harbor Laboratory Press, ed. J. Sambrook, E.F. Fritsh, T. Maniatis, p. 1-21), 30,000 U of SP6 polymerase, 0.1 mg of pyruvate ki-nase, 50 U of the ribonuclease inhibitor from human placenta, 5 |xg of each of the protease inhibitors (leupeptin, chymotryp-sin) and « 2-macroglobulin in buffer A: 25 mM HEPES, pH 7.6, 1.5 mM MgAc2, 100 mM KAe, I mM ATP, 0.4 mM GTP, 0.4 mM CTP, 0.4 mM UTP, 0.25 mM spermi- dine, 4.0 mM dithiothreitol, 6 mM creatin phosphate, 20 p.M [35S] Met with specific radioactivity of 800 mCu/mmol, 20 pM of each of the other 19 amino acids.

Bishop reads Baranov as directing the researcher to make up one milliliter of a mixture, using 600 microliters of rabbit reticulo-cyte lysate plus enough buffer to render it the equivalent of the final concentration of buffer A.

Bishop testified that before October 1991, it was known in the art that transcription inhibitors can be added when a researcher wants to insure there is no transcription in the translation reaction and that there is no transcription inhibitor added in the second step of Perara and Lingappa or anywhere in the Stueber process. This leads him to believe that there is no difference between what these two references say and plaintiffs patent claims, with the exception of the word “coupled.” All of the ingredients for transcription are present in Stueber and in Per-ara and Lingappa and no inhibitors are present to stop transcription. On the basis of the experiments performed by defendant in preparation for trial, Bishop believes that the reactions disclosed in Coen and Perara and Lingappa would have made it obvious in 1990 to one of ordinary skill in the art to perform a two-step reaction that was coupled in the second step.

Bishop believes that the state of the art in 1990 and early 1991 was that researchers did not measure or account for the magnesium in the lysates they used.

2. Anderson

It is Anderson’s belief that in Baranov the initial mixture would include the magnesium added through buffer A (1.5 millimolar) and that this concentration would not change very much as the process continued. If the concentration within the vessel is the same as the concentration on the outside, the magnesium will not be driven out of the vessel. If the magnesium level did fall, it would not fall significantly for hours. He finds no reason to believe that the level would fall to 1.5 millimolar in 20 minutes; in 20 minutes only half of volume would have been exchanged. Thus, even if all magnesium were free to move, it would go down only to 2.5 millimolar (halfway between the three millimolar free magnesium in the reaction and the two milli-molar magnesium in the buffer). He believes Baranov would produce protein as a batch reaction.

Anderson’s opinion is that claim 1 of the ’637 patent would have been obvious to one of ordinary skill in the art at the time of the publication of Baranov patent application because it would have been obvious to convert Example 4 of the Baranov patent application from the continuous flow process to a static mode reaction to make protein. He would read Baranov as telling him to set up a reaction just as it is described for the continuous flow at the initial stages of the reaction, putting all of the components shown in Bara-nov into a test tube. Anderson believes that in light of Baranov, claim 34 of the ’637 patent would have been obvious at the time to one of ordinary skill in the art.

Anderson believes that in light of Stueber, the methods described in claims 1 and 34 of the ’637 patent would have been obvious to one of ordinary skill in the art at the time the patent application was filed. His scientific experience teaches him that transcription must occur in the Stueber translation because all the components required for transcription are present and no inhibitor is added. Also, he believes that the components of the kit claimed in claim 68 would have been obvious to one of ordinary skill in the art at the time in light of Baranov or in light of Stueber. His conclusions would be the same with respect to claims 1 and 13 of the ’817 patent. He knows of no one who tried to determine before this litigation began whether the second reaction in Stueber was coupled.

Anderson believes that the experiment Beekler performed was deficient because he did not include at least the conditions at the beginning of the reaction, specifically a reaction made up in buffer A containing 1.5 milli-molar added magnesium.

Anderson has never tried to run any continuous flow system for producing protein. He was unaware until trial that stirring went on inside the chamber in a continuous flow system.

3. Mierendorf

Mierendorf testified that at the time plaintiff applied for its patent, researchers generally defined the appropriate amount of magnesium to use in a translation reaction as the amount added to the reaction, usually .5 to 2.5 millimolar, and not the sum of the added magnesium and the magnesium within the lysate. At the time, standard ranges for magnesium in a transcription reaction were 6 to 10 millimolar magnesium chloride.

Mierendorf testified that in 1990, it was known in the industry that coupled transcription and translation could be done in a test tube, as shown from the virus core information, the Boehringer Mannheim instructions to add some transcription directly to the translation mix and the Baranov papers describing the continuous flow system of conducting a coupled system. He believes that under the conditions described by Perara and Lingappa, the transcription reaction does not stop before it is added to the translation reaction, that transcription would continue in the translation reaction and that protein would be produced inevitably. It is his opinion that in the transcription reaction described in both Stueber and Coen, transcription would continue during step two, the translation reaction, and that the transcript made would be translated into protein.

Mierendorf believes that in 1991, a person of ordinary skill in the art would have found the claims of the ’637 and the ’817 patents obvious in light of Stueber, in light of Coen and in light of Perara and Lingappa. He admits there is no data in either the Coen or the Perara and Lingappa paper that tell the reader whether the second reaction is coupled or not. He believes that a paper by Krieg and Melton showing that 40 to 70% of the nucleotides would still be present in a Perara and Lingappa type of reaction would lead a person to believe that full-length transcript would be produced during the translation reaction in a two-step process.

In September 1996, Mierendorf oversaw a number of experiments designed to prove that transcription would occur during the translation reaction under the conditions described in Coen and in Perara and Lingappa. He concluded that the experiments proved that transcription would occur and that protein would be produced.

It is Mierendorfs opinion that in 1991, a person of ordinary skill in the art would have found it obvious from the Baranov patent application to make a coupled transcription and translation reaction in a static mode. In September 1991, in preparation for trial, he conducted experiments that demonstrate to him that if 1.5 millimolar of magnesium acetate is added to the reaction described in Baranov’s Example 4, easily detectable amounts of protein will be produced. The only difference he finds between Baranov’s Example 4 as written and claim 1 of the ’637 patent is that the patent describes a static reaction; however, Baranov can be considered a static system before the pump is turned on. It is his opinion that claims 2 to 17, 34, 35-51, 68, 69, 70, and 73-75 of the ’637 patent would have been obvious to one of ordinary skill in 1991 in light of Baranov as would claims 1, 2, 3-12, 13 and 14-16 of the ’817 patent.

Mierendorf testified that the range of magnesium concentration was between 1.6 and 4.3 mM in lysates available in 1990-91; that he would say that as a general rule of thumb for bacteriophage enzymes, it is necessary to have about 4 mM magnesium above the level that would be bound by the nucleotides so that if you had more nucleotides in the reaction, more magnesium would be required; that it is generally known that it is necessary to provide enough magnesium in the translation reaction to saturate everything that binds the magnesium and leave a little extra for the translation to occur; that in 1990-91, most people thought of the appropriate amount of magnesium in the translation reaction as the amount added to the reaction, usually in the range of .5 to 2.5 mM; and that in 1990-91, it was well known in the art that one needed to optimize the various components of the transcription and translation reactions, that is, to try different concentrations of a given component while keeping the rest the same; that generally researchers focused on optimization of magnesium and potassium; and that it was known that magnesium was very important. Mierendorf calculated the range of the magnesium concentration used in Baranov Example 4, starting with the contribution of the lysate (.96 to 3 mM), adding 1.5 mM of exogenous magnesium acetate, resulting in a magnesium concentration of 2.46 mM to 4.5 mM.

In the fall of 1994, Mierendorf wrote an article on coupled in vitro translation in which he described the Perara and Lingappa article as having shown that SP6 RNA polymerase transcription reactions could be added directly to reticulocyte lysate for the production of protein, eliminating the need to purify the RNA prior to translation. Mier-endorf went on to say that Spirin and later Craig showed that transcription and translation could be coupled; he did not ascribe such a showing to Perara and Lingappa.

In performing experiments in May 1995 to determine whether the Perara and Lingappa reference was inevitably coupled, defendant used a different lysate, gene and cap. Defendant’s researcher was Tom Van Oosbree, who had helped invent the ’637 patent while working for plaintiff. Van Oosbree produced a chimpanzee alpha-globin fusion protein, which was not the protein he tried to produce. (He started with a beta-galactosidase gene.) Van Oosbree experimented with a number of different levels of magnesium. The experiments were unsuccessful.

In additional experiments performed during the same month, defendant attempted to replicate the Perara and Lingappa experiments, using a different rabbit reticulocyte lysate, a different protein, two different polymerases and polymerase promoters, a CITE translation enhancer, a slightly more concentrated reticulocyte lysate and more added magnesium than intended by Perara and Lingappa. The results did not show full length protein. In other experiments, similar substitutions were made and the results were generally inconclusive.

In attempting to determine whether coupling would occur under the Coen conditions, Mierendorf used a different reticulocyte ly-sate, although one manufactured by a similar procedure. He did not attempt to express the same protein. He used a stronger promoter and different amounts of DTT. He believes the substitutions are acceptable and would not change the outcome of the experiments. Some of the lanes in some of the gels resulting from the experiments did not show a protein band. In Mierendorf s opinion, the absence of a band does not prove that protein was not made in the reaction. He believes the gels demonstrate that Coen is necessarily and always coupled.

Mierendorf is aware of no attempt by anyone to determine whether the two step reactions used before 1991 were coupled until he did the experiments for this case.

4. Dimond ,

Dimond does not believe that either Stue-ber, Coen or Perara and Lingappa discloses a reaction as described in Claim 1 of plaintiffs patent and that none of the reference makes the invention in Claim 1 obvious. In reading the references, he saw no indication that the second reaction (translation) was coupled. Perara and Lingappa discuss how protein got into the membrane of the cell, specifically focusing on whether the amino acid sequence would guide the protein into the membrane if the sequence were in the middle of the protein or whether the sequence had to be at the beginning. In Dimond’s opinion, Perara and Lingappa do not add any information to what was already known in the art about transcription and translation.

Perara and Lingappa and Coen were unusual in transferring 1/5 of the transcription reaction to the translation reaction. Most researchers at the time would not have transferred that much. For example, the Boeh-ringer Mannheim protocol recommended adding only so much as would make up about four percent ,of the total volume of the translation reaction and also recommended adding EDTA to the reaction after doing a transcription reaction with the control RNA. Because the EDTA is a chelator (a chemical that will bind to divalent cat ions like magnesium), under the Boehringer Mannheim procedure, the transcription amount that is added to the translation reaction will not contain any free magnesium to affect the optimum conditions for the translation reaction.

In Dimond’s opinion, the Baranov patent application does not disclose batch reactions, but rather teaches away from them. In Dimond’s view, a batch reaction occurs when ingredients are put together in a test tube and incubated for a period of time to allow a reaction to take place. Baranov et al. do not do this in any of the examples in their patent application. He believes that the initial conditions of Baranov do not remain static throughout the reaction described in the application; in fact he cannot say that there is ever a static situation in the continuous flow system. He does not think that Baranov teaches the conditions one could use to get a coupled batch reaction. First, the patent application teaches that the continuous flow system is an improvement on batch systems, making it the “next generation” of protein synthesis. Second, Baranov teaches a method for overcoming the problems encountered in earlier attempts to do coupled batch reactions: removing the inhibitory by-products of the in vitro transcription reaction from the reaction by means of the flow of the buffer through the system. With this new method, proteins can be synthesized for a long period of time. Third, Example 4 would suggest to readers that if they wanted to convert the system to a batch reaction, they would need to do two separate batch reactions.

Dimond reaches this last conclusion for three reasons: 1) Baranov et al. are using a wide variety of different kinds of semipermeable membranes through which the fluid is flowing. In Example 4, they use very large amounts of SP6 RNA polymerase, which is a small protein that will move through the membrane. This leads Dimond to expect that as the continuous flow system runs, the E. coli or SP6 RNA polymerase will move out of the reaction vessel and will end up in the test tube with product. This indicates to him that most or all of the transcription is happening at the beginning.' 2) As the buffer flows through the system, the magnesium is going to end up at 1.5 millimolar. Persons of ordinary skill in the art would have realized that 1.5 millimolar is a very low magnesium concentration for transcription and would have interpreted that to mean either that transcription is not occurring at that stage of the process or that it is occurring only at a very low level. If the magnesium were higher earlier, the process would have more transcription at the beginning. 3) The data published with the European patent application support Dimond’s belief that transcription is occurring early in system, as does the Bara-nov and Ryabova article (exh. # 1057) that shows RNA introduced into the system at the beginning and never supplemented throughout the subsequent period of the reaction. When the RNA is introduced at the beginning, the reaction has a continuous and fairly linear production of the globin protein over that long period of time, indicating that the messenger RNA is stable in the system and not being degraded.

Example 4 of the Baranov application differs from the Baranov and Ryabova article because the patent application uses DNA and not RNA at the beginning of the reaction. Dimond would expect that if small amounts of RNA were being transcribed throughout the continuous flow system, the production of protein would be graphed in an upward curve, whereas if a lot of transcription were going on, it would soon saturate the ribosomes in the system, inhibiting the translation. The data on Example 4 (Exh. # 1104, Fig.5) show a linear production of protein, that is, a fairly constant rate of protein synthesis throughout the reaction, from which he concludes that all the RNA is generated early in Example 4. This would lead him to think that conversion of Example 4 to a batch reaction would require doing a first transcription batch reaction and then a translation batch reaction. However, he does not think anyone would try to do a batch reaction in reliance on Baranov because Baranov does not teach how to remove the inhibitors generated by the translation reaction.

Dimond testified that if 1.5 mM magnesium had been added to the initial mix in Baranov Example 4, protein would have been produced and the total magnesium level would have been within the range set out in claim 3 of the ’637 patent; that after the running conditions in Baranov are established the potassium concentration would be 100 mM, within the 40 mM to about 100 mM set out in claim 8 of the ’637 patent; that in buffer A, which is the running condition, Baranov shows .4 mM of three ribonucleotide triphosphates, and a different concentration of the fourth that is higher than .4.

5. Jagus

In Jagus’s opinion, Stueber does not make the claims of plaintiffs patents obvious. The Stueber paper added little to the state of the art of coupled transcription-translation other than the use of the T5 promoter with E. coli polymerase. It did not show a coupled reaction. Her opinion is that it is unlikely that the second reaction of Stueber is coupled because, as Roberts shows, the transcription reaction is essentially over in 15 to 20 minutes and the reaction product is not added to the translation reaction until after transcription is over. Both Roberts and Stueber teach two separate reactions and teach away from claim 1 of the ’637 patent.

Jagus notes that when Stueber was doing transcription, he was not using his enzyme with maximum efficiency; when he transferred the transcription reaction into the translation, he took it to a lower concentration so that in doing the transfer, even if the polymerase had still been working, the concentration of triphosphates would be such that it would be working much more slowly.

In Jagus’s opinion, the Perara and Lingap-pa paper does not render claim 1 obvious because the paper is teaching that a single sequence present in a protein molecule functions no matter where it is placed in the molecule. The authors discuss the method they were using only because they were producing a radiolabeled protein in vitro that they subjected to further experimentation and analysis. (Radiolabeling provides a means of tracking the protein to learn what happens to it in certain circumstances.) Per-ara and Lingappa do not disclose a coupled reaction or present any data to show that the second reaction is coupled.

Jagus agrees with Dimond that the widely different temperatures for the two reactions in Perara and Lingappa would mean that the polymerase could work in the second reaction at a maximum rate of one-twentieth of what it worked in the first reaction, even assuming that in the hour-long incubation before it went into the translation mix it was not inactivated by the by-products of the reaction and that the translation mix itself did not have any adverse effects on the ability of the polymerase to function. Perara and Lingap-pa’s system was inefficient, as shown by the very low level of protein produced.

It is Jagus’s opinion that Coen does not make claim 1 obvious. The paper describes a two-reaction system for producing protein and includes no data to indicate that the second reaction is coupled. Jagus believes that the reaction would not be coupled because the transcriptional activity would have ended during the incubation period used for the first reaction and, in addition, most of the nucleotide would be reduced to very low concentrations and EDTA added to bind magnesium. In her view, Coen teaches doing two reactions, each optimized. It teaches away from a coupled reaction.

It is Jagus’s belief that the Baranov patent application does not tell how to get coupled transcription and translation in batch reaction. She testified that the Craig paper that was published in 1992 (exh. # 61) is a paper from the laboratory of Dr. Richard Jackson, a laboratory that has made many important innovations in the use of cell-free systems. The paper describes the working of a coupled transcription-translation system and was considered sufficiently important to be published even though plaintiffs TNT kit was on the market already.

Jagus testified that one could determine the range of magnesium concentration in Baranov Example 4 if one assumed the concentration of magnesium in the lysate Bara-nov used as being in the range of 1.6 to 1.8 mM. With that assumption, one could calculate the total amount of magnesium concentration as being in the range of 2.46 to 2.58 mM, which is within the range of 2.5 to 3.5 given for coupled transcription and translation in the ’637 patent.

Jagus has found that the coupled system (with or without the kit) enables her to make much more protein than she was able to make before and to make it faster and more easily. It has opened up ways of doing things that she would lot have considered in the late 1980s, enabling her to screen for functions in a way she could not have done before. For instance, in studying the enzyme kinase, she has been able to use the translation factor made and radiolabeled in vitro to map the parts of the kinase that interact with a particular protein synthesis factor. The system has enabled her to ask different questions about her research.

III. FACTS RELATING TO ALLEGATION OF INEQUITABLE CONDUCT

(The essence of defendant’s inequitable conduct charge is defendant’s contention that plaintiff’s employee, Gregory Beckler, told the patent examiner that the protocol described in the Baranov patent application would not work when converted to a batch reaction and that Beckler misled the examiner intentionally when he failed to tell the examiner that he had not added any exogenous magnesium to the reaction he devised in an experimental effort to convert the continuous flow reaction to a batch reaction.)

The Baranov et al. patent application describes a coupled transcription and translation system that is intended to operate in a continuous flow mode in a commercial unit such as the Amicon 8 MC stirred cell unit. The lysate components necessary for sustained protein synthesis are retained by a semipermeable membrane of an ultrafiltration unit. Example 4 of the Baranov et al. patent application includes two sets of materials: an “incubation mixture” and “buffer A.” The incubation mixture is combined in an ultrafiltration unit with buffer A, which contains ingredients necessary for the coupled reaction. Buffer A is pumped continuously through the ultrafiltration unit into the incubation mixture and out through the semipermeable membrane, for as long as 34 hours. Protein synthesis begins approximately 2/£ hours after the reaction begins and continues throughout the 34 hours. The membrane is permeable to CAT specific protein, which has a molecular weight of about 30,000 Daltons, and to magnesium ions, which have a molecular weight of 58 Daltons.

Besides the rabbit reticulocyte lysate, buffer A contains magnesium acetate (MgAc2), which Baranov et al. report is included in buffer A at a concentration of 1.5 mM. It is known now that the concentration of magnesium for the initial reaction conditions of Baranov is the concentration of magnesium contained in the rabbit reticulocyte lysate of the incubation mixture plus the magnesium contained in buffer A. The concentration of magnesium in any given rabbit reticulocyte lysate may vary; standard rabbit reticulo-cyte lysate has a magnesium concentration of 4.2 to 5.0 mM. Baranov does not provide the concentration of magnesium in the rabbit reticulocyte lysate used, but Dr. Lyuba Rya-bova, a co-inventor on the Baranov patent application and the researcher who performed the experiment described in Example 4, has verified that 1.5 mM of magnesium was added to buffer A in Example 4. Ryabo-va arrived at the conclusion that 1.5 mM was the proper amount of magnesium to add to buffer A by performing a series of batch reactions using varying amounts of magnesium.

The Baranov patent application does not discuss conversion of its continuous flow system to a batch reaction and it does not disclose either the magnesium concentration of its initial reaction conditions or the concentration of magnesium that would be used for a batch reaction.

Plaintiff filed its original patent application (S.N.775, 136) on October 11, 1991. On March 10, 1992, the United States Patent and Trademark Office examiner rejected nearly all of the claims of plaintiffs original patent application, explaining that the invention as claimed in the patent application was anticipated by the Baranov patent application and obvious over the work of Baranov, in view of Suzuki. The examiner asserted that the Baranov patent application method would result in protein production even without the continuous flow dialysis described therein. Plaintiff responded that Baranov did not include any information useful for setting up a batch reaction. The examiner rejected plaintiffs application a second time on November 3,1992, stating:

Applicants [sic] remarks do not clearly indicate that Baranov et al. protocol described in the prior art does not work. A Declaration stating the results of the use of the Baranov et al. patent showing the Examiner that the protocol in the prior art cannot work as suggested in the rejection [i.e., as a batch mode] should be submitted if this is what Applicant means.

On February 3, 1993, plaintiff appealed the examiner’s decision to the United States Patent Office Board of Appeals and Interferences. On February 23, 1994, plaintiff withdrew the appeal and filed a continuation application to bring the patent application back to the examiner. The continuation application was assigned a new serial number (S.N. 149, 715). Legally and conceptually, the patent application remained the same but the examiner examined it anew. Plaintiff amended its claims to limit them to batch reactions and supplemented its earlier record and arguments with respect to all the rejections entered by the examiner, including those based in whole or in part upon the Baranov patent application. Plaintiff submitted a series of declarations, including one by Dr. Josephine Grosch explaining that plaintiffs product was a commercial success and two by Dr. Gregory Beckler describing his attempts to convert the continuous flow transcription translation reaction of Baranov to a batch reaction.

Dr. Grosch declares in part:

More than 75 scientific experts in a variety of disciplines and R & D Magazine editors selected the commercialized invention TNT Coupled Reticulocyte Lysate System as one of the 100 most technically significant new products of the year.

The commercialized invention TNT Coupled Reticulocyte Lysate System had total sales of over $600,000 in the product’s first year and over $1,000,000 in its second year.

In the second of the two declarations he submitted, Beckler states in part:

I designed and performed two experiments in an attempt to convert the continuous transcription / translation reaction of the Baranov reference to a batch reaction.

The first experiment (hereafter “Reaction 1”) utilized the incubation mixture alone as defined in Baranov Example 4. The second experiment (hereafter “Reaction 2”) utilized the incubation mixture and buffer A as defined in Baranov Example 4. See Exhibit 1 attached hereto for Table A which shows the components, concentrations, and amount added for a 0.5 ml batch Baranov incubation mixture. See Exhibit 2 attached hereto for Table B which shows the components, concentrations, and amount added for a 0.5 ml batch Baranov buffer A.

The text of the declaration does not mention that exogenous magnesium was not added to buffer A in Reactions 1 and 2, although that information is contained in table B attached to the declaration. The second line of table B shows that the only magnesium used in the reactions was 1.53 mM of magnesium, which was inherently present in the rabbit reticulo-cyte lysate. Beckler had read the ’637 patent application before submitting his declarations. That application stated, “Leaving magnesium concentrations at levels present in the standard lysate, protein production does not occur.”

In Reaction 1, Beckler utilized the incubation mixture alone as defined in Baranov Example 4 but changed some of the concentrations of the ingredients in attempts to make the mixture work in batch mode. For this first experiment, Beckler added the incubation mixture to water instead of buffer A. In Reaction 2, Beckler used the incubation mixture and buffer A as defined in Baranov Example 4 (again changing the ingredient concentrations to account for the conversion to batch mode), plus water to make 0.5 ml (13.15|xl). Beckler did not add any exogenous magnesium acetate to buffer A. He achieved a final magnesium concentration of 1.53 mM by using a 60% volume of a special Promega reticulocyte lysate (Flexi™) that has a magnesium concentration of 2.55 mM. Beckler completed and submitted a gel auto-diagram showing that no protein was produced using the conditions of Baranov.

On March 11, 1994, two of plaintiff’s attorneys, Beckler and another employee of plaintiff met with United States Patent and Trademark Office examiner Robert Wax and supervisory examiner David Schmickel and submitted the Second Declaration of Dr. Gregory Beekler in person. The Examiner’s Interview Summary Record reports that the second Beekler declaration demonstrates that “Baranov et al. as a static reaction does not work.” The examiner stated that he had erred in rejecting plaintiffs claims and withdrew that rejection. On March 22, 1994, the patent examiner issued a “Statement of Reasons for Allowance” stating:

Applicants have submitted declaration by Beekler that indicates that the Baranov et al. method of in vitro protein synthesis does not work as described. By performing the reaction as directed in the Baranov et al. reference, but in a batch mode, applicants have demonstrated that no protein is produced. Applicants had also presented evidence that the Baranov et al reference could not be used with some experimentation to produce proteins as is described in the reference in the continuous mode. As the reference did not enable those of ordinary skill in the art to make protein either in batch as claimed nor much protein in a continuous mode as detailed in the prior art the instant invention is not obvious over that reference. Applicants have presented additional data that indicates the invention is a commercial success.

On June 28, 1995, plaintiffs attorneys filed the same Beekler declaration, asking the patent examiner to allow its ’817 patent. On September 26, 1995, the patent office issued a “Notice of Allowability,” citing the same reasons for allowing the patent as mentioned with respect to the ’637 patent.

In preparation for trial, Dr. Robert C. Mierendorf, one of defendant’s scientists, reviewed Beckler’s reports of his attempt to replicate the experiments of Example 4 of the Baranov patent application and convert it to batch mode. Under Mierendorfs direction, scientists for defendant replicated Beekler’s “Reaction 2” experiment as Mier-endorf understood it. These same scientists performed the same experiment with the addition of exogenous magnesium in the amount of 1.5 mM in buffer A, as specified by Baranov. These experiments showed that if Beekler had added exogenous magnesium to the reaction mixture, protein would have been produced. Mierendorf reviewed the additional experiments performed by Robin Hurst at the plaintiff company on the same issue. Plaintiffs data revealed that its experiments yielded the same result as defendant’s experiments, that is, protein is produced when 1.5 mM of exogenous magnesium is added to buffer A.

Mierendorf understood from reading Beck-ler’s Second Declaration what experiments Beekler had conducted in response to the patent office’s suggestion. Mierendorf would have set up the conversion differently from the way Beekler did it. He believes that Beckler’s method was not a fair replication. He believes that if one skilled in the art is given the task of seeing whether something will work when converted to another mode, he or she will try various conditions. In his opinion the vast majority of those skilled in the art would have simply followed the Bara-nov protocol, added the ingredient's together in a tube and observed whether protein was produced. If that procedure had not worked, the researcher would have tried another way to interpret the reaction conditions fairly.

Randall Dimond, plaintiffs Chief Technical Officer, testified that Baranov explains what the magnesium concentration would be during the running phase of the experiments after the buffer had flowed through the system for a while. From his reading of Example 4, he can discern at least three different interpretations of what the initial magnesium concentration might be. 1) The magnesium listed refers only to added magnesium so the reaction should be adjusted to make the added magnesium 1.5 millimolar. 2) The 1.5 millimolar is the total magnesium, that is, the magnesium that was in whatever rabbit reti-culocyte lysate Baranov used plus the added magnesium. 3) The authors were saying that they were taking buffer A and mixing a certain volume of it together with the reticu-locyte lysate. (This approach is consistent with the approach taken in the Ryabova et al. paper (exh. # 1057), in which the authors describe a continuous flow reaction with reti-culocyte lysate, saying that they started with a certain volume of a micrococcal nuclease treated reticulocyte lysate into which they put the globin messenger RNA and added 250 microliters of feeding solution, producing the concentrations shown in the paper.)

If Dimond were going to try to convert Example 4 of the Baranov patent application to a single batch reaction (which he does not believe would work), he would start by running the continuous flow system for several hours until the components in buffer A had equilibrated so as to establish the running conditions, take the reaction out of the continuous flow reaction chamber, put it into a test tube and continue the incubation and try to characterize the properties of the reaction. In this way, he could determine what he thinks Baranov is trying to accomplish: the maintenance of certain concentrations in the continuous flow system. Dimond believes that the magnesium concentration would be 1.5 millimolar when equilibrium is achieved. In experiments he has done in the past he has observed that the endogenous magnesium in the reticulocyte lysate was lost during dialysis in a unit such' as Baranov used for the continuous flow system, making it necessary to add 1.5 millimolar of exogenous magnesium. In his experience, the magnesium concentration drops rapidly at first and then slows until it approaches whatever is being pumped through the system.

The difficulties in replicating Baranov made it impractical for researchers to determine running conditions in the manner Dimond would prefer. He thinks Beckler’s conversion was the most practical one that could be made. It is not reasonable simply to use the initial conditions to convert Example 4, because that is not what the patent application is teaching. Rather, the application’s whole thesis is that the best way of doing protein synthesis is by bringing buffer A to the running conditions Baranov et al. describe. This is supported by the authors’ failure to analyze the initial conditions or give any attention to what happens in the first half hour. In Dimond’s opinion it makes no sense to try to replicate the initial conditions in Baranov. They are not fully understandable and, in any event, are going to change rapidly in the system.

Dimond testified that whether using Bara-nov’s initial conditions would work depends on assumptions about what the initial conditions were. Example 4 does not disclose the nature of the initial lysate, whether the inventors added to the eukaryotic examples the enzymes necessary to generate ATP or what they used as energy to drive the translation. Converting Example 4 to a batch mode as written would work for no more than about four minutes because of the lack of an ATP regenerating system. (In the continuous flow system, ATP is being pumped into the reaction continuously, so it is not necessary to have an enzyme system for making ATP.) Beckler included an energy source in his conversion. Dimond believes Beckler’s declaration described a reasonable way to attempt to convert Baranov to batch reaction.

Dimond asked another researcher to repeat Beckler’s experiment. The replication showed that using the running conditions described in Example 4 of Baranov would not produce specific protein described in Example 4.

In Rosemary Jagus’s opinion, Baranov’s description of the initial reaction in Example 4 is ambiguous and there are at least three ways of interpreting it. She agrees with Dimond that transcription would occur early in the continuous flow system, followed later by translation, in part because she assesses the initial magnesium concentration as falling to the level of the running buffer (1.5 milli-molar) in one and a half to two hours, too little to support transcription under the running conditions.

The examiner listed on plaintiffs ’637 patent and on the Baranov U.S. patent is Robert Wax.

IV. OPINION'

A. Inequitable Conduct

In the February 28, 1997 order, I addressed three distinct arguments pertaining to plaintiffs alleged inequitable conduct and found two of them unsupported by the undisputed facts, but left the third argument for resolution at trial because there were material facts in dispute. Defendant’s third argument was that plaintiff had engaged in ineq uitable conduct when it submitted the Second Declaration of Dr. Gregory Beckler to the patent office because that document contained false or misleading statements about plaintiffs conversion of the continuous flow method in the Baranov reference to a static or batch reaction. Defendant contended that plaintiffs failure to point out explicitly to the patent examiner that Beckler did not add 1.5 mM of exogenous magnesium to the experiment was a “sly and subtle” attempt to deceive the patent examiner. After considering the evidence adduced on this issue at trial, I ruled that defendant had not proved by clear and convincing evidence that plaintiff had engaged in inequitable conduct.

In its Rule 52 motion, defendant argues that Beckler’s trial testimony proves that plaintiffs submission of Beckler’s second declaration amounts to inequitable conduct and points to Beckler’s admission at trial that his attempted conversion of Example 4 of the Baranov reference to a batch mode did not prove that Baranov would never produce protein if converted to a batch mode. According to defendant, this contradicts what Beckler told the patent examiner in his second declaration and is evidence that plaintiff deliberately misled the. examiner.

Defendant has framed its inequitable conduct argument in varying ways at various stages of this litigation. In its amended answer, it alleged that Beckler’s second declaration was untrue because he said he had used the Baranov buffer A when in fact he had not added 1.5 mM magnesium as specified by Baranov in Example 4. In its reply brief on its motion for summary judgment, defendant argued that Beckler had lied when he told the examiner that he had replicated Baranov in the continuous mode and that the experiments had failed. I did not address this allegation in the February 28 order and defendant has not pursued it. Now, in its motion for reconsideration, defendant contends that Beckler breached his duty of disclosure when he told the patent examiner that a proper conversion to a batch reaction of the Baranov continuous flow method could not work. Defendant argues that Beckler admitted the breach of duty when he testified at trial that the experiment he described in his second declaration does not prove that Baranov Example 4 cannot be converted successfully to a batch mode.

To a certain extent, defendant’s allegations of inequitable conduct could be characterized as a moving target and not the mere seman-tical difference that defendant would call them. Notwithstanding the fact that this last, newly framed allegation could be denied as untimely, I will address it along with the allegation that Beckler meant to mislead the examiner when he said that he used the incubation mix of Baranov and Baranov’s buffer A.

The standards governing inequitable conduct have not changed since the February 28, 1997 order and opinion. Section 1.56(a) of Title 37 of the Code of Federal Regulations provides that:

[e]ach individual associated with the filing and prosecution of a patent application has a duty of candor and good faith in dealing with the Office ....

Proof that an applicant engaged in inequitable conduct must be established by clear and convincing evidence and the conduct must be shown to be both material and intentional. Molins PLC v. Textron, Inc., 48 F.3d 1172, 1178, 1179 (Fed.Cir.1995). A statement or omission is material where there exists a substantial likelihood that a reasonable examiner would consider the information given or misstated important in deciding whether to allow the patent application to issue. 37 C.F.R. § 1.56; Halliburton Co. v. Schlumberger Technology Corp., 925 F.2d 1435, 1440 (Fed.Cir.1991). Materiality does not presume intent; intent must be proven on its own. Manville Sales Corp. v. Paramount Systems, Inc., 917 F.2d 544, 552 (Fed.Cir.1990). Intent need not be proven by direct evidence; most often it is proven by a showing of acts, the natural consequences of which are presumably intended by the actor, Molins PLC, 48 F.3d at 1180, and it may be inferred from the facts and circumstances surrounding the applicant’s conduct: Id. at 1180-81 (citing Paragon Podiatry Laboratory, Inc. v. KLM Laboratories, Inc., 984 F.2d 1182, 1189-90 (Fed.Cir. 1993)). The drawing of inferences relies heavily on the factfinder’s assessment of a witness’s credibility and character. Id. at 1181 (citation omitted). In making the overall determination whether inequitable conduct has occurred, a court must balance materiality and intent: the more material, the omission, the less culpable the intent required. Halliburton Co., 925 F.2d at 1439.

The materiality of Dr. Béckler’s second declaration is not in dispute. The patent examiner reversed his initial decision on obviousness after receiving the second declaration, changing plaintiff’s patent fortunes considerably. The real question is whether defendant has adduced clear and convincing evidence that plaintiff submitted the second declaration with the intent to deceive the patent examiner. Plaintiff filed Beckler’s second declaration with the intent to change the patent examiner’s mind about the initial decision that plaintiff’s patent claims were obvious over Baranov. Plaintiff understood the patent examiner’s concern that Beckler had simply taken all the ingredients from Baranov’s continuous flow Example 4, put them together in a test' tube and called the result an invention. In rejecting plaintiffs application on November 3, 1992, the patent examiner suggested that plaintiff submit a declaration explaining that Baranov would not work in a batch mode if plaintiff believed that to be true.

Plaintiff followed that advice by designing and performing two experiments in an attempt to convert Baranov to a batch reaction. He described these experiments in his second declaration, explaining that in the first experiment he had utilized the incubation mixture alone as defined by Baranov Example 4 and in the second experiment he had utilized the incubation mixture and buffer A as defined in Baranov Example 4. He included tables showing exactly what components he had used (and not used). He described the procedures he followed and the results (including those from a third experiment he had performed as a control, using the TNT product with the same gene product and the same reaction size as the first two experiments).

Beckler never stated in the declaration that the experiments showed that Baranov could not be converted to a batch reaction. He merely reported one effort that plaintiff argues represents a reasonable try at conversion. Defendant disagrees with the characterization of the effort, pointing out that plaintiff never added 1.5 mM of magnesium to the mix it used for the batch reaction. Defendant argues that Beckler should have read Example 4 as requiring added magnesium, citing its own experts’ trial testimony that it is- a common practice to state the amount of added magnesium, not the amount of total magnesium. Defendant argues that if Beckler had done that, the testimony of its experts (and of Beckler) shows that the procedure would have worked in a batch mode. Moreover, adds defendant, at trial, Beckler agreed with defendant that had he added the extra magnesium he would probably have been able to produce protein.

The fallacy in defendant’s argument is its reliance on after-acquired information to try to prove nefarious intent in 1992, when Beck-ler submitted the second declaration. It is easy to see now that the critical difference in making Baranov work is to add 1.5 mM magnesium at the outset. What defendant has failed to prove is that Beckler knew in 1992 that he should add magnesium and deliberately failed to do so, simply to mislead the examiner. Beckler testified at'trial that he believed from his analysis of the Baranov patent application and more than a year of working with the procedure that conversion to a batch reaction would require adjustments in the initial mix to approximate the running conditions of the continuous flow procedure. He tried to replicate the conditions that would exist within the Amicon unit after the initial components of buffer A had equilibrated. Beckler believed that because the concentration of magnesium in Baranov falls to 1.5 mM fairly quickly and before protein production occurs, a total of 1.5 mM magnesium would be the concentration to begin with in a batch reaction. He aimed at a final concentration of 1.5 mM by using a 60% concentration of plaintiffs rabbit reticu-locyte lysate, which has a magnesium concentration of 2.55 mM. Defendant has not shown either that this was an unreasonable hypothesis at the time or that Beckler knew that it was. Defendant cannot make this showing simply by adducing the opinions of experts that they would have read Example 4 differently from Beckler. Even if these opinions are untainted by hindsight, which is questionable, they do not prove that Beckler’s reading was unreasonable. The information contained in Example 4 is ambiguous at best. It could be read plausibly in several different ways and because of that, defendant may not rely on it to prove inferentially that Beckler tried to mislead the patent examiner.

Defendant’s contention that Beckler lied when he told the examiner that Baranov could not be converted to a batch reaction is undermined at the outset by defendant’s failure to prove that Beckler made any such assei'tion. What Beckler did do was submit a description of the experiment he had performed, indicating that Baranov did not make it obvious that the continuous flow procedure could be converted to a batch reaction that would produce protein. Beckler did not hide from the examiner the number of experiments he performed. He explained each one. The examiner knew when he read the declaration that Beckler had not proved the impossibility of any successful conversion; he had performed only one or two experiments. The examiner has a Ph.D., but it is not necessary to have an advanced degree to know that one cannot prove the impossibility of a proposition by doing only a few experiments. The examiner did not ask Beckler to prove that Baranov could never be converted to batch form; this is not what disproving obviousness requires. All that plaintiff (through Beckler) had to show was that Bar-anov did not make a successful conversion to batch obvious.

After submitting the second declaration, Beckler met with the patent examiner to discuss that declaration. Whatever plaintiff said in the declaration and during that meeting convinced the examiner to report in the Examiner’s Interview Summary Record that the second Beckler declaration demonstrates that “Baranov et al. as a static reaction does not work.” Relying on that information, the examiner decided that he had erred in rejecting plaintiffs claims and withdrew his rejection. On March 22, 1994, the patent examiner issued a “Statement of Reasons for Allowance,” in which he explained that plaintiff had demonstrated that no protein is produced by performing Baranov in a batch mode. Not surprisingly, defendant fastens on the examiner’s statements as revealing that some representative of plaintiff and probably Beckler had told the examiner that the conversion would never work under any conditions. In light of the prosecution history and the indisputable fact that plaintiff did not produce proof of more than one or two experiments, it is likely that the examiner used a shorthand way of saying that plaintiff had demonstrated that Baranov does not make it obvious to convert to a batch reaction that will produce protein. After all, the examiner was familiar with Baranov and what it disclosed from having been the examiner on the Baranov patent application.

Beckler’s testimony at trial corroborated his lack of deceit in filing his second declaration. Had he lied at the time he submitted the application, he would not have admitted at trial that he now believes the addition of extra magnesium would make the conversion possible. His admission at trial was that of a scientist who keeps an open mind about his work and is continuously reviewing the bases for his conclusions. His testimony showed not that he had lied in 1992, but that after listening to other scientists at the trial, he had come to see a different way of attacking a problem that had defeated him earlier. It was apparent from his testimony that he was not considering the legal or strategic consequences of his admission, but was thinking about the new perspectives he had gained from the experts.

I find no reason to believe that Beckler intended to deceive anyone when plaintiff filed his second declaration. It is evident that Beckler’s understanding of the Baranov reference has changed over the course of time; it does not follow from his present understanding of the reference that he did not believe what he said in 1992. Neither the evidence nor my evaluation of Beckler’s credibility supports such an argument.

B. Obviousness

Defendant proceeded to trial on three claims: invalidity of the patent claims for anticipation, invalidity for obviousness and invalidity for inequitable conduct. The last is an issue reserved for the court. The jury never considered the anticipation issues because they were dismissed before they went to the jury after defendant failed to show that any single prior art reference necessarily and always inherently anticipated the coupled transcription and translation process that plaintiff claimed as its invention. The jury found in favor of defendant on obviousness as to all but two of the disputed claims of both patents. Plaintiff argues now that no reasonable jury could have reached that conclusion on the evidence presented at trial.

Each claim of a patent carries a presumption of validity. 35 U.S.C. § 282. The presumption requires “the decisionmaker to employ a decisional approach that starts with acceptance of the patent claims as valid and that looks to the challenger for proof to the contrary.” Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1534 (Fed. Cir.1983). A challenger of a patent must establish facts that support a conclusion of invalidity by clear and convincing evidence. Texas Instruments v. United States Int’l Trade Comm’n, 988 F.2d 1165 (Fed.Cir. 1993). Pursuant to 35 U.S.C. § 103,

A patent may not be obtained ..., if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains.

Obviousness under 35 U.S.C. § 103 is a question of law. Para-Ordnance Mfg., Inc. v. SGS Importers Int’l, Inc., 73 F.3d 1085, 1088 (Fed.Cir.1995), cert. denied, — U.S. -, 117 S.Ct. 80, 136 L.Ed.2d 38 (1996). However, the analysis of obviousness rests on several factual inquiries: 1) the scope and content of the prior art; 2) the differences between the prior art and the claims at issue; 3) the level of ordinary skill in the art at the time of the invention; and 4) objective evidence of other considerations of non-obviousness, such as what the differences between the claimed invention and the prior art enabled the invention to achieve, that others in the field sought and failed to arrive at the claimed invention, that one of those others copied the claimed invention and that the invention met with outstanding commercial success. Panduit Corp. v. Dennison Mfg. Co., 810 F.2d 1561, 1566 (Fed.Cir.1987); see also ParaOrdnance Mfg. Inc., 73 F.3d at 1088; Miles Laboratories, Inc. v. Shandon, Inc., 997 F.2d 870, 877 (Fed.Cir.1993).

In determining whether a jury’s verdict of invalidity should be upheld, the court must decide whether there is “substantial evidence” in the record to support the verdict. Shatterproof Glass Corp. v. Libbey-Owens Ford Co., 758 F.2d 613, 621 (Fed.Cir. 1985). If the court is convinced that a reasonable jury could not have reached a verdict for the non-movant, it should grant the mov-ant’s motion for judgment as a matter of law. Perkin-Elmer Corp. v. Computervision Corp., 732 F.2d 888, 893 (Fed.Cir.1984). “[I]t may not be presumed that the jury found facts on which there is no evidence.” Newell Cos., Inc. v. Kenney Mfg. Co., 864 F.2d 757, 767 (Fed.Cir.1988) (citations omitted).

In evaluating the evidence to determine whether it is “substantial,” the court must employ the usual approach of viewing all evidence in the light most favorable to the non-moving party, drawing all reasonable inferences in favor of the non-movant, avoiding any determination of the credibility of the witnesses and refraining from substituting its view for the jury’s when the evidence is conflicting. A major portion of the evidence adduced at trial was devoted to the issue of anticipation. Defendant tried to show that each of four prior art references, Coen, Stue-ber, Perara and Lingappa and the Baranov patent application, anticipated plaintiffs patent claims inherently. Defendant’s experts testified that in the case of the two-step reactions described in Coen, Stueber and Perara and Lingappa, transcription must have occurred during at least a portion of the translation process because the components for transcription were present. Defendant performed a number of experiments to show that performance of the reactions described by Coen and by Perara and Lingappa would always result in coupled transcription and translation and protein synthesis. However, defendant did not replicate the prior art reactions precisely, making it questionable whether the results were reliable indicators of what had taken place in the original experiments. Moreover, the results defendant did obtain failed to show that protein would have been produced every time. Defendant made a similar argument with respect to the Bara-nov patent application, arguing that when all the components necessary for transcription and translation were present in the containment vessel before buffer A was added, it was inevitable that a transcription and translation process would have begun. However, defendant never established that transcription and translation did occur before the buffer was pumped through.

The legal standard for obviousness is whether “the prior art would have suggested to one of ordinary skill in the art that the [inventive] process should be carried out and would have a reasonable likelihood of success, viewed in light of the prior art.” In re Dow Chemical Co., 837 F.2d 469, 473 (Fed.Cir.1988). Because the parties gave the issue of anticipation so much attention during the trial, the jury may have believed mistakenly that it could consider what defendant argued was inherent in the prior art rather than only what the prior art would have suggested or taught to one of ordinary skill. It appears that the jury overlooked the absence in the prior art references of any suggestion that the reactions they were describing might involve coupling and the consequent production of protein.

The great difficulty for any factfin-der in deciding obviousness is always how to step back in time and evaluate what a prior art reference would have conveyed to those of ordinary skill before the inventor made his discovery. See, e.g., Interconnect Planning Corp. v. Feil, 774 F.2d 1132, 1138 (Fed.Cir. 1985) (invention must be viewed not with blueprint drawn by inventor, but in light of state of art at time); W.L. Gore & Assoc., Inc. v. Garlock, Inc., 721 F.2d 1540, 1553 (Fed.Cir.1983). See also Diamond Rubber Co. of New York v. Consolidated Rubber Tire Co., 220 U.S. 428, 434, 31 S.Ct. 444, 55 L.Ed. 527 (1911) (“Knowledge after the event is always easy ...”). Moreover, in considering a prior art reference, the reference must be considered as a whole, including portions that would lead away from the invention. Panduit Corp., 810 F.2d at 1566.

1. Coen reference

a. Obviousness of independent claims

When deciding whether the 1977 Coen reference would have made the invention of the ’637 and ’817 patents obvious to researchers of ordinary skill in the relevant art of producing protein synthesis kits such as plaintiffs TNT product, the jury had before it the following evidence.

® Bishop’s testimony that Coen is inevitably • coupled because all ingredients for transcription are present in the translation reaction and no inhibitors are added;

• Bishop’s testimony that the results of Mier-endorfs pretrial experiments showed that it would have been obvious in 1990 to one of ordinary skill in the art to perform a two-step reaction that was coupled in the second step;

• Mierendorf s unadorned testimony that plaintiffs patent claims would have been obvious in light of Coen and that it was his opinion from the experiments he conducted that transcription would have continued to take place after the thirty minutes of the first (transcription) reaction and that there would have been transcription occurring in the second (translation) reaction of Coen with a resulting translation into protein; and

®A communication from Dimond to one of the named inventors to the effect that another researcher had observed that he obtained more protein if he added the whole completed Riboprobe transcription reaction to the translation reaction than if he added just the RNA and that this might mean that he was continuing to get transcription after the addition.

The evidence supporting the jury’s verdict on the Coen reference cannot be characterized as substantial. It includes nothing from which a jury could have found that Coen would have suggested to a person of ordinary skill before October 1991 that the two-step reactions it described would be coupled. There was no evidence that anyone of any skill level had actually read Coen as suggesting or teaching that transcription might be occurring in a batch eukaryotic translation reaction before plaintiff applied for its patent. Mierendorfs opinion is not enough to support the finding because he gave no reasons for it. See In re Buchner, 929 F.2d 660, 661 (Fed.Cir.1991) (expert’s opinion on ultimate legal issue must be supported by something more than conclusory statement) (citing Application of Brandstadter, 484 F.2d 1395, 1405 (C.C.P.A.1973)); Ashland Oil, Inc. v. Delta Resins & Refractories, 776 F.2d 281, 294 (Fed.Cir.1985) (generally, opinion testimony of experts is entitled to some weight, but lack of factual support may render testimony of little probative value in validity determination). The omission of an explanation is not cured by Mierendorfs testimony that he knows now that coupling was occurring because he has done experiments in preparation for trial. What the experiments revealed and what he or anyone else learned after plaintiff filed its patent application do not illuminate the state of the art before the application was filed. Indeed, the fact that defendant had to undertake experiments to determine what happened in Coen and in Perara and Lingappa undermines to some extent its assertion that one of ordinary skill in the art would have known that coupling took place when the reactions described in those references were performed.

In placing so much reliance on the assertion that it proved that the two-step reactions were always coupled, defendant seems to forget that I held as a matter of law that the experiments defendant had done did not establish that coupling always took place in the two-step reactions. That was one of the grounds on which I dismissed defendant’s claims of anticipation, along with defendant’s failure to show that it had replicated the original conditions of Coen and Perara and Lingappa.

Bishop’s testimony about the inevitability of coupling is not relevant to the question the jury had to determine, which was whether Coen would have suggested to a person of ordinary skill that transcription would continue and that protein would be produced in the translation stage of the Coen experiment. His testimony is unpersuasive for other reasons as well. First, he testified that Coen is inevitably coupled because all the ingredients for transcription are present in the translation reaction. He did not explain how a person of ordinary skill would have known that “all the ingredients for transcription are present” before plaintiffs patent identified the critical ingredients for coupled transcription and translation. Without an explanation, Bishop’s testimony suggests reliance on after-acquired knowledge and impermissible hindsight. Second, his testimony rests on the experiments conducted by defendant well after the patent issued. Such post-invention knowledge is irrelevant to the determination of what would have been known in 1990 by one of ordinary skill.

The Coen paper itself made no reference to coupling or to transcription continuing in the translation reaction, used widely differing concentrations of magnesium for the two separate reactions, showed no added nucleotides in the translation reaction and used EDTA in an apparent attempt to bind some of the magnesium coming from the transcription reaction.

Defendant places weight on the e-mail message from Dimond, but the message adds nothing to the showing of obviousness. Dimond is not a person of ordinary skill in the art but rather one of extraordinary skill, as are defendant’s experts. The skill level of the scientist referred to in the message is not identified, making it impossible to conclude that he is one of ordinary skill. Finally, the message reveals nothing about what would have been suggested by one of the prior art references in dispute.

I cannot say that the jury had substantial evidence to support a conclusion that defendant had proven by clear and convincing evidence that the independent claims of plaintiffs ’637 and ’817 patents were obvious in light of Coen.

b. Obviousness of dependent claims over Coen

Defendant offered no evidence to show why dependent claims 12, 13, 48 and 49 of the ’637 patent might be obvious over Coen, other than Mierendorfs statement, which reads in full as follows:

Q. Lastly on the Coen paper that’s Exhibit 1087, do you have an opinion as to whether or not the claims of the ’637 patent that we discussed earlier would or would not have been obvious in view of the teachings of the Coen paper?

A. I do have an opinion.

Q. And what is your opinion?

A. That it would have been obvious.

Q. Would that opinion be any different if we were discussing the claims we discussed earlier from the ’817 patent?

A. No different.

Transcript, 3-B-36. 235 U.S.C. § 282 provides that each claim of a patent is presumed valid independently of the validity of the patent’s other claims. A challenger must establish the invalidity of each claim separately. Bausch & Lomb, Inc. v. Barnes-Hind/Hydrocurve, 796 F.2d 443, 446 (Fed. Cir.1986); Shelcore, Inc. v. Durham Industries, Inc., 745 F.2d 621 (Fed.Cir.1984). Defendant does not argue that Mierendorfs testimony was sufficient in and of itself to support the jury’s finding of obviousness; instead it argues that the jury was able to judge for itself from the patents whether the claims were obvious over Coen, as shown by its decision that claims 11 and 46 of the ’637 patent and claim 3 of the ’817 patent were not obvious.

Regardless of its verdict, a lay jury cannot determine the obviousness of claims involving complex issues of molecular biology without the testimony of an expert to explain why the claims might be obvious. Even if I had not determined that plaintiff is entitled to judgment as a matter of law on the obviousness of the independent claims of its two patents, I would have to find in favor of plaintiff on defendant’s allegations that the dependent claims of the patents are obvious in light of Coen.

2. Stueber reference

The evidence before the jury relating to the Stueber reference was essentially the same as that relating to the Coen reference, with a few exceptions.

• Bishop’s testimony, which was the same with respect to Stueber as to Coen (that the second reaction is inevitably coupled because all ingredients for transcription are present in the translation process and no inhibitor is added);

• Mierendorfs testimony that in his opinion transcription would continue in the translation reaction because no inhibitor is added to stop transcription;

• The Dimond communication concerning the possibility that transcription was continuing into translation when the entire Ribo-probe transcription was added;

• Anderson’s testimony that in his scientific opinion, if the Stueber reactions'were replicated, transcription would continue in translation and would produce proteins.

The jury did not have any experimental evidence because defendant did not perform any experiments using the Stueber reference.

Like the testimony on the Coen reference, the evidence of the obviousness of Stueber is far from substantial. It does not support the jury’s finding of obviousness because it does not establish what Stueber would have taught or suggested to a person of ordinary skill in the art at the time of the invention. Defendant argues that obviousness is shown by the admissions of Beckler and Jagus that coupled transcription and translation occur if all the components of the transcription reaction are present in the translation mix and within the right magnesium range, but this is evidence of what Beckler and Jagus know now. It is not evidence of what they knew before the invention or of what persons of ordinary rather than extraordinary skill would have known then. Defendant has failed to show that at the time of the invention, one of ordinary skill in the art would have expected successful coupled reactions in carrying out the process disclosed in Stueber or either of the two other two-step reactions.

3. Perara and Lingappa reference

The evidence before the jury on Perara and Lingappa was the following:

• Bishop’s testimony that transcription would occur inevitably in the second reaction be cause no transcription inhibitor was added to the translation mix. In Bishop’s view, the experiments conducted by Mierendorf show that it would have been obvious to one of ordinary skill in the art in 1990 that transcription was continuing in the translation reaction.

• Mierendorfs testimony that in light of Krieg and Melton’s paper showing that 40 to 70% of the nucleotides would remain at the end of a Perara and Lingappa transcription reaction, one would expect creation of full-length transcript even in translation.

This evidence is not substantial for the reasons outlined above in connection with Coen and with Stueber. In addition, defendant cannot rely on the disclosures of the Krieg and Melton paper because defendant failed to show at trial any suggestion that the prior art references at issue should be combined with any other prior art.

4. Baranov patent application

a. Obviousness of independent claims over Baranov

In arguing that there is substantial evidence to support the jury’s verdict that the independent claims of plaintiffs patents are obvious in light of the Baranov patent application, defendant points to the following evidence:

• Bishop’s testimony that when all of the ingredients listed in Baranov’s Example 4 were placed in the machine before turning it on, they were the equivalent of a batch reaction and would have produced protein;

• Anderson’s testimony that if he set up a batch reaction exactly as described for a continuous flow reaction in the Baranov Example 4, there would be protein produced and that it would have been obvious to him how to convert Baranov to a batch reaction;

• Mierendorf s testimony that one of ordinary skill in the art could have converted the experiment in Example 4 to static mode reaction;

• Articles from the Spirin group that reported comparative protein production data from noncoupled batch reactions;

• The patent examiner’s rejection of plaintiffs first patent application on the ground that one of ordinary skill would reasonably infer from the Baranov patent that dialysis was not needed to produce an analytical amount of protein and would proceed to perform the Baranov reaction as a batch and the “fact” that the examiner’s subsequent withdrawal of this rejection came about only because plaintiff misled the examiner about the nature of

• Beckler’s experiments at converting Bara-nov to a batch reaction;

• The Content paper showing that in a coupled prokaryotic reaction the optimum magnesium concentration is intermediate between the optimum for transcription and that for translation;

• The Pelham paper describing a eukaryotic cell-free system in which efficient coupled messenger RNA and protein synthesis are directed by a DNA template;

• Dimond’s agreement that if anyone had added 1.5 mM magnesium to the Baranov reaction some protein would have been produced;

• Beckler’s admission that Spirin’s continuous flow reactions could probably be converted to batch mode with enough experiments and that it was easier to do batch mode reactions than continuous flow; and

• Defendant’s experiments showing that Bar-anov Example 4 would work if 1.5 mM magnesium was added to the initial reaction.

Some of this testimony could not have been considered properly by the jury. For example, I determined at trial that defendant had not proved that Baranov Example 4 was a coupled batch reaction before the machine was turned on and therefore Baranov did not anticipate plaintiffs patents. 35 U.S.C. § 102. I ruled also that the jury could not consider any combinations of prior art references because defendant had not shown any suggestion in the art to combine the references with Baranov or any of the three two-step references at issue.

Even if the jury could have considered this evidence, it still would have lacked sufficient evidence to support its verdict of obvious ness. Nothing in Baranov teaches or suggests that the process be converted to a batch reaction. Two-step batch reactions were the standard in the art; Baranov et al. obtained a patent on the new idea of continuous flow that had the advantage of producing protein for long periods of time. Nothing in the patent application suggested that the process would also produce protein in batch or static mode; it was teaching away from the batch reaction, which was the old art. In this connection, it is worth noting that in previous articles, the Spirin group had included batch mode reactions as alternatives for continuous flow. The absence of any of those alternatives in the Baranov patent application is a strong indication that the continuous flow system was not intended to be an effective means of performing a batch reaction. The novel idea in the invention was the dialysis that removed spent agents and product from the reaction, enabling it to continue for more than a day or two at a time. Removal of spent agents and product would not be possible in a one-tube batch reaction. Moreover, graphs in the patent showed that transcription took place at the outset of the reaction, suggesting to persons of ordinary skill that if conversion were to be attempted, it should be by a two-step batch reaction so as to minimize the inhibitory effects of the transcription reaction on the translation reaction.

Defendant’s suggested combination of Content and Baranov is without foundation. Content concerned a prokaryotic (naturally coupled) reaction that had been in the public domain for fourteen years. No one had ever read it as suggesting that its disclosures could be utilized for the entirely different eukaryotic reaction. The same is true with respect to Pelham, which disclosed a coupled reaction for vaccinia core particles. Defendant has advanced no reason why anyone would apply its teachings to the Baranov continuous flow system, which does not use core particles.

The testimony of defendant’s experts to the effect that one of ordinary skill could have successfully converted Baranov to a static reaction does not respond to the key question, which is why Baranov Example 4 would have suggested to one of ordinary skill that such a conversion should be done in order to produce protein. Almost anything can be done, once the value of doing it is known or suspected. As I have noted, much in the Baranov patent application taught away from a static reaction. All the indications were that the patent had nothing to do with a batch mode. Defendant has not shown what would have suggested to a researcher of ordinary skill that the patent could be converted to batch mode and make protein, let alone how that could be accomplished. The admissions of Dimond and Beekler that Baranov Example 4 epuld be converted if 1.5 mM magnesium was added at the outset prove only that they know now that magnesium is the critical factor; they do not prove that a person of ordinary skill would have known that in 1991.

This leaves defendant with only the initial rejection of plaintiffs patent application by the patent examiner, a far too slender reed to support the jury’s verdict. The examiner withdrew his rejection after reviewing Beck-ler’s declarations to the effect that one reasonable effort to convert Baranov Example 4 did not produce protein, that the Baranov continuous flow system did not work as reported and that plaintiffs product had met with significant commercial success in the short time it had been on the market. Defendant maintains that the rejection was withdrawn only because plaintiff misled the examiner. I have explained in the discussion of inequitable conduct why that argument does not hold up.

Because the Baranov patent was before the patent examiner when he reviewed plaintiffs patent application, defendant has a particularly heavy burden of showing invalidity. Hewlett-Packard Co. v. Bausch & Lomb, Inc., 909 F.2d 1464, 1467 (Fed.Cir. 1990) (burden of proving invalidity is difficult “when the prior art was before the PTO examiner during prosecution of the application”); Central Soya Co. v. Geo. A. Hormel & Co., 723 F.2d 1573, 1577 (Fed.Cir.1983). Courts presume that qualified government agencies have done their jobs properly. American Hoist & Derrick Co. v. Sowa & Sons, Inc., 725 F.2d 1350, 1359-60 (Fed.Cir. 1984).

Although I am satisfied that defendant failed to adduce substantial evidence of the obviousness of the independent claims of plaintiffs ’637 and ’817 patents over Baranov, I am bolstered in that conclusion by the evidence that Thomas Van Oosbree, one of the inventors of the ’637 and ’817 patents and a person of more than ordinary skill in the art, tried in 1990 to convert Baranov to a batch reaction and failed. Van Oosbree had no incentive not to succeed. In fact, he had every reason to want to develop a commercially profitable adaptation of Baranov for defendant, his new employer. Compared to all of the after-the-fact interpretations by the experts, Van Oosbree’s unsuccessful efforts during the time in question are far more probative of the alleged ease of converting Baranov to a batch mode. S.C. Johnson & Son, Inc. v. Carter-Wallace, Inc., 614 F.Supp. 1278, 1300 (S.D.N.Y.1985) (what person skilled in the art actually did or failed to do at relevant time is best determinant of obviousness of invention) (citing U.S. Philips Corp. v. National Micronetics Inc. 550 F.2d 716, 722 (2d Cir.1977)). It is telling also that when Mierendorf wanted to develop a product for defendant, he did not suggest to his scientists that they consult Baranov for guidance on setting up a batch reaction.

b. Obviousness of dependent claims over Baranov

Although it follows from the finding that defendant failed to adduce substantial evidence from which a jury could find the independent claims of plaintiffs ’637 and ’817 patents invalid for obviousness that defendant cannot prove the dependent claims obvious, I will discuss them separately. Each claim defines a separate invention and is presumed valid on its own merits. Bausch & Lomb, 796 F.2d at 446. The dependent claims at issue (3, 4, 8, 17, 37, 38, 48, 49 and 51 of the ’637) patent specify particular concentrations of magnesium, potassium and nucleotides for the coupled batch reaction.

Plaintiff argues that it was érror for the court to admit the testimony of Robert Mierendorf on the subject of the dependent claims because he had not given any opinion on these claims in his expert reports. Plaintiff is correct; it was error to allow Mieren-dorf to testify on this subject. Even if the testimony is considered, however, it is far from the substantial evidence necessary to uphold the jury’s finding on these claims.

In order to prove that the dependent claims were obvious in light of Baranov, defendant had to show by clear and convincing evidence that Baranov taught or suggested to a person of ordinary skill in the art the importance of the particular concentration ranges of magnesium, potassium and nucleotides. The text of Baranov does not explain which of the various components are important in achieving the coupled reaction in the continuous flow system or what concentration ranges can be used for a coupled reaction. Baranov does not say how much magnesium is endogenous to the rabbit reticulocyte it uses or even specify the rabbit reticulocyte lysate it used or the method of preparation of the lysate. Baranov introduces more complications because the concentrations change when the continuous flow mechanism is turned on and Buffer a is allowed to flow through the system.

In testifying about the dependent claims of the two patents, Mierendorf did not explain how a person of ordinary skill in the art would have gathered from Baranov what ingredients were critical to the success of the reaction and the concentrations needed. • Instead, he merely noted his opinion that the dependent claims of the ’637 patent would have been obvious to one of ordinary skill in light of Baranov. His testimony about the amounts of magnesium used in Baranov and of the known advantages of optimizing the various components of the transcription and translation reactions does not inform the jury what the prior art would have taught or suggested to one of ordinary skill. It is easy to see now that controlling the amount of magnesium concentration is the crucial factor in coupling transcription and translation. That was not what defendant had to prove. Its task was to adduce evidence to show why this would have been understood in 1990-91. The testimony defendant elicited from Ja-gus and Dimond is of little value to defendant’s case. Jagus’s testimony is based upon an assumption of the concentration of magnesium in the lysate used by Baranov that defendant asked her to make. Her acknowledgment that one could calculate the total amount of magnesium on the basis of this assumption does not help the jury determine why persons of ordinary skill might have thought to make such a calculation in 1990-91 or might have thought that doing such a calculation would enable them to produce protein in a coupled batch reaction.

Defendant argues that the testimony of Mierendorf, Jagus and Dimond, together with the evidence that the examiner had concluded once that all the claims of the ’637 and ’817 patents were obvious, is substantial evidence to support the jury’s finding that the dependent claims in issue were invalid for obviousness. Even if the concentrations of magnesium, potassium and nucleotides did not fall within the disclosed ranges, defendant argues, it was well known in the art in 1990 that one needed to optimize the components of transcription and translation.

As with the independent claims, defendant failed to show what aspect of Bara-nov would suggest to a person of ordinary skill in the art that magnesium concentration was the critical factor in making a coupled reaction work and what the particular magnesium concentration might be. Baranov et al. do not disclose the initial magnesium present in Example 4 or even disclose the method of preparation of the lysate, suggesting strongly that they did not view the endogenous magnesium as a critical factor to be taken into consideration in determining the optimal magnesium concentration for a coupled reaction. (There is good reason to think that it was not critical for Baranov et al. to determine the total magnesium in the initial reaction; their concern was with the concentrations in the running conditions that would be established later.) Merely listing components without an explanation of their importance or the particular concentrations necessary does not make obvious the inventive concept of plaintiffs patents that the total concentration of endogenous and exogenous magnesium is critical to the success of a coupled batch reaction. At most, it might make it obvious to try varying the components, but that is not the test of obviousness. Application of Antonie, 559 F.2d 618, 620 (C.C.P.A.1977) (“Obvious to try is not the standard.”) An invention is obvious to try rather than obvious within the meaning of § 103 “ ‘where the prior art [gives] either no indication of which parameters [are] critical or no direction as to which of many possible choices is likely to be successful.’ ” Merck & Co., Inc. v. Biocraft Labs., Inc., 874 F.2d 804, 807 (Fed.Cir.1989) (quoting In re O’Farrell, 853 F.2d 894, 903 (Fed.Cir.1988)).

There is little probative value to defendant’s evidence of hypothetical conversions of Baranov Example 4 that might produce magnesium concentration ranges within those claimed in the patents. These backward looking, after-the-fact conversions shed no light on the suggestions and teaching of Bar-anov to a person of ordinary skill reading it in 1990-91. Mierendorf admitted that changing the magnesium concentration in a reaction was “one obvious thing to try” whenever a reaction is not working. Again, this is the “obvious to try” approach that the Federal Circuit has rejected.

In citing the examiner’s initial rejection of the magnesium range on the ground of obviousness, defendant neglects to state that the rejection was based on a reading of Baranov together with a reference by Suzuki and was withdrawn subsequently. Moreover, in the order granting summary judgment in part, I held that the Baranov and Suzuki combination could not render any of the patent claims invalid because defendant never presented any evidence of a suggestion to combine the teachings of Suzuki with those of Baranov. I conclude that the jury lacked substantial evidence to find the ■ dependent claims concerning magnesium concentrations (3, 4, 37 and 38) obvious over Baranov.

Interestingly, in arguing in support of the jury’s verdict on the obviousness of the dependent claim 8, defendant relies on Dimond’s testimony to the effect that the feed solution in Baranov Example 4 has 100 mM potassium, at least once the running conditions are established. In all its previous arguments, defendant has talked only about the hypothetical initial concentrations; indeed, it has accused Beckler of fraud and inequitable conduct for attempting to convert Baranov in a batch reaction by trying to replicate the running conditions of Baranov rather than simply putting into one test tube all the ingredients listed as the starting reaction. Defendant is less than consistent in arguing now that it would have been obvious to one of ordinary skill to understand from Baranov that the critical magnesium level was to be derived from the unknown lysate plus the added magnesium prescribed for the initial reaction but that the critical potassium level was that achieved when the running conditions had reached a state of equilibrium.

The Baranov initial reaction shows a potassium range of 100 mM, the high end of that set out in claim 8 of the ’637 patent, plus the potassium endogenous to the lysate. Following the logic of defendant’s argument that the magnesium concentration is to be derived from the initial reaction, Baranov teaches away from the invention in claim 8, by suggesting that more than 100 mM potassium concentration is critical to coupling.

Claims 17 and 51 of the ’637 patent require a concentration of 0.4 mM of each of the four ribonucleotide triphosphates. Baranov Example 4 discloses nucleotides in the buffer A feed solution of 1.4 mM of three of the nucleotides and 1.0 mM of the fourth. Defendant argues that because Baranov discloses concentrations of all four nucleotides of at least 0.4 mM, it renders obvious the claims that require exactly 0.4 mM. Defendant does not explain its reasoning or why it thinks the additional limitation of at least can be read into the claims. E.I. du Pont de Nemours & Co. v. Phillips Petroleum Co., 849 F.2d 1430, 1433 (Fed.Cir.1988) (improper to read a limitation from the specification into a claim); SSIH Equipment S.A. v. U.S. Int’l Trade Comm’n, 718 F.2d 365, 378 (Fed. Cir.1983) (courts “cannot alter what the pat-entee has chosen to claim as his invention”). Other claims of the ’637 patent claim different sets of specific nucleotide concentrations for wheat germ extracts, indicating that the specific concentrations are critical.

The only remaining dependent claims at issue are 48 and 49, which concern a DNA template with a multiple cloning region (claim 48) and a polymerase promoter sequence located at one end of the said multiple cloning region (claim 49). Defendant does not argue that it adduced any evidence of obviousness with respect to these two claims, other than Mierendorfs conelusory statement that these claims were obvious. This is not substantial evidence to prove clearly and convincingly that the claims are obvious.

C. Secondary Considerations

In considering obviousness, the factfinder must take into account such secondary considerations of nonobviousness as long felt but unsolved needs, failures of others, copying, and commercial success. Para-Ordnance Mfg., 73 F.3d at 1087-1088 (citing Graham v. John Deere Co., 383 U.S. 1, 17-18, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966)); Avia Group Int’l, Inc. v. L.A. Gear California, Inc., 853 F.2d 1557, 1564 (Fed.Cir.1988). This case presents unusually strong secondary considerations, beginning with the undisputed evidence that defendant developed its infringing product, not by studying the prior art references that allegedly made the invention obvious, but by buying plaintiffs kit and analyzing its components. This is potent evidence of nonobviousness. Specialty Composites v. Cabot Corp., 845 F.2d 981, 991 (Fed.Cir.1988) (“‘[CJopying the claimed invention, rather than one in the public domain, is indicative of unobviousness’ ”) (quoting Windsurfing Int’l Inc. v. AME, Inc., 782 F.2d 995, 1000 (Fed. Cir.1986)). It is difficult to give weight to defendant’s assertions of the obviousness of plaintiffs invention in light of the evidence that defendant tried unsuccessfully to make the same product and resorted to copying plaintiffs “obvious” invention. See Heidelberger Druckmaschinen AG v. Hantscho Commercial Prods., Inc., 21 F.3d 1068, 1072 (Fed.Cir.1994) (argument that “innovation is really quite ordinary carries diminished weight when offered by those who had tried and failed to solve the same problem, and then promptly adopted the solution that they are now denigrating”).

Additional evidence of nonobviousness is provided by the significant commercial success plaintiffs product has enjoyed. Defendant tries to play down the importance of commercial success, arguing that plaintiff cannot show a nexus between the inventive concept and the sales of its product. Defen- ■ dant cites Beckler’s testimony to the effect that some of the reactions covered by the claims would not result in increased protein production like that made possible by the TNT product. Plaintiff need not show that every claim of its patent would result in a commercially successful product; it need only show, as it has, that its product is the embodiment of claims in its patent. This provides the nexus between patent and product that is required. Demaco Corp. v. F. Von Langsdorff Licensing Ltd., 851 F.2d 1387, 1392 (Fed.Cir.1988) (nexus requirement is satisfied by showing that commercially successful product is invention disclosed and claimed in patent).

It is additionally probative of nonobviousness that no one developed a coupled euka-ryotic batch reaction between 1977, when Coen was published, and 1991, when plaintiff applied for its patent, although researchers were eager for such a process; that the Spirin group did not convert the Baranov continuous flow reaction to a batch when they realized the unlikelihood of ever commercializing the continuous flow system; that the Craig paper describes the coupled batch reaction process as a new effort that did “not seem to have been attempted previously” and one that resulted in “a considerable saving of time and effort”; that defendant advertised its STP product as the “next generation” in transcription and translation, promising speed, accuracy, convenience and high protein yields; that as late as April or May 1991, Van Oosbree and others thought that a coupled batch reaction was not feasible; and that Dr. Jagus has found that plaintiffs product enables her to do more experimental work than she could do previously.

D. Conclusion

From my review of the evidence defendant adduced at trial, I am persuaded that it was not substantial enough to support the verdict reached by the jury. Defendant’s experts offered little but after-the-fact, barebones conclusions about what the prior art references would have suggested or taught to persons of ordinary skill in the art. Much of defendant’s evidence rested on experiments intended to prove anticipation that were irrelevant to the determination of obviousness. The opinions of defendant’s experts and the experiments were insufficient to show by clear and convincing evidence that the contested claims of plaintiffs patent ’637 and ’817 patents would have been obvious in 1990-91 to a person of ordinary skill in the art. Plaintiffs showing of the strong secondary considerations of copying, long felt need, commercial success and peer recognition are additional reasons for finding defendant’s showing lacking.

V. ORDER

IT IS ORDERED that the motion of plaintiff Promega Corporation for judgment as a matter of law under Fed.R.Civ.P. 50 is GRANTED with respect to the jury’s verdict of obviousness as to the claims in issue of plaintiffs ’637 and ’817 patents; those claims have not been shown to be invalid; plaintiffs motion for a new trial or other appropriate relief under Fed.R.Civ.P. 59 is DENIED as moot; and the motion of defendant Novagen, Inc. for amendment of the judgment on defendant’s claim of unenforeeability of the patents on the ground of inequitable conduct is DENIED. The judgment entered herein on March 27,1997 is VACATED.

A scheduling hearing will be held at 8:30 a.m., October 17, 1997, to determine a date for trial of plaintiffs damages claims.