Genentech, Inc. v. Boehringer Mannheim GmbH

47 F. Supp. 2d 91 (1999)

GENENTECH, INC., Plaintiff,
v.
BOEHRINGER MANNHEIM GmbH
and
Boehringer Mannheim Corp., Defendants.

No. Civ.A. 96-11090-PBS.

United States District Court, D. Massachusetts.

April 14, 1999.

*92 Joseph Ferraro, Leora Ben-Ami, Philip E. Roux, Rogers & Wells, New York City, Gary W. Smith, Dennis D. Allegretti, Anthony J. Fitzpatrick, Renee Inomata, Burns & Levinson, Boston, MA, Sean Johnston, Genentch, Inc., Office of Patent Counsel, South San Francisco, CA, Gerard P. Norton, Rogers & Wells, New York City, for Genetech, Inc., a Delaware corporation, plaintiff.

William L. Patton, Steven A. Kaufman, Martin J. Newhouse, Crystal D. Talley, Ropes & Gray, Boston, MA, Peter F. Felfe, John E. Lynch, Felfe & Lynch, New York City, J. Barry Buzogany, Boehringer Mannheim Corporation—Therapeutics, Gaithersburg, MD, for Boehringer Mannheim, GmbH, a German corporation, defendant.

William L. Patton, Steven A. Kaufman, Martin J. Newhouse, Crystal D. Talley, Ropes & Gray, Boston, MA, Peter F. Felfe, John E. Lynch, John A. Bauer, Felfe & Lynch, New York City, J. Barry Buzogany, Boehringer Mannheim Corp., Gaithersburg, MD, for Boehringer Mannheim Corporation, an Indiana corporation, defendant.

William L. Patton, Steven A. Kaufman, Martin J. Newhouse, Crystal D. Talley, Ropes & Gray, Boston, MA, John G. Harkins, Jr., Neill C. Kling, Harkins Cunningham, Philadelphia, PA, for Centocor, Inc., defendant.

Dennis D. Allegretti, Anthony J. Fitzpatrick, Burns & Levinson, Boston, MA, Sean Johnston, Genentch, Inc., Office of Patent Counsel, South San Francisco, CA, for Genentech, Inc., counter-defendant.

*93 MEMORANDUM AND ORDER

SARIS, District Judge.

INTRODUCTION

Defendants Boehringer Mannheim GmbH and Boehringer Mannheim Corporation ("collectively BM") have moved for summary judgment on the claims of plaintiff Genentech, Inc. ("Genentech") that BM infringes three patents involving Genentech's tissue plasminogen activator ("t-PA") product. The three patents are United States Patent Nos. 4,342,832 ("the '832 patent"); 5,034,225 ("the '225 patent"); and 4,511,502 ("the '502 patent"). Genentech has filed a cross-motion for summary judgment with respect to the '502 patent. After hearings held on June 17-18, 1997 and July 23, 1997, which included the testimony and report of an impartial court-appointed expert, Dr. Connie Cepko,[1] this Court issued a memorandum and order, dated December 30, 1997, addressing issues of claim construction. See Genentech, Inc. v. Boehringer Mannheim, 989 F. Supp. 359 (D.Mass.1997); see generally Markman v. Westview Instruments, Inc., 517 U.S. 370, 391, 116 S. Ct. 1384, 134 L. Ed. 2d 577 (1996) (holding that issues of claim construction are matters of law to be determined by judge, not jury). Genentech has filed a cross-motions for summary judgment with respect to the '502 patent.

After extensive supplemental submissions on the pending motions for summary judgment to take into account the Markman decision, and after hearing, BM's motion for summary judgment is ALLOWED with respect to the '832 patent and the '502 patent, but DENIED with respect to the '225 patent.

UNDISPUTED FACTS

1. Tissue Plasminogen Activators

Serious heart attacks can be caused by the presence of a thrombus, which is a blood clot in the coronary blood vessels or coronary artery. The process that dissolves a thrombus is called thrombolysis, and the chemicals in the body that induce thrombolysis are called plasminogen activators. Two plasminogen activators, which are a natural part of a body's defenses, are urokinase-type plasminogen activator (u-PA) and tissue-type plasminogen activator (t-PA).

T-PA is a chemical that makes the body's natural plasminogen become plasmin, an enzyme. Plasmin cuts through fibrin, the substance which makes up blood clots. Although the body naturally produces small amounts of t-PA, this quantity does not activate sufficient plasminogen to cut through the large clots involved in heart attacks.

In the late 1970s and early 1980s, scientists, whose research was sponsored by Genentech, successfully reproduced t-PA through recombinant DNA technology by identifying the DNA encoding t-PA—527 amino acids ("aa") with glycosylation (the attachment of sugars) at three sites—and inserting the DNA into bacteria. The five domains (or regions) of the t-PA protein are:

—The Fibrin Finger ("F") domain (aa 4-49);
—The Epidermal Growth Factor ("E") domain (aa 50-87);
—The Kringle 1 ("K1") domain (aa 88-176) glycosylated;
—The Kringle 2 ("K2") domain (aa 177-262) glycosylated;
—The Serine Protease ("P") domain (aa 263-527) glycosylated.

In July, 1979, Genentech applied for the '832 patent for its particular method of reproducing t-PA and subsequently obtained approval from the Food and Drug Administration ("FDA") to sell t-PA under *94 the name Activase. Genentech also obtained a patent for purifying proteins ('502 patent) in 1985 and one for increasing the solubility of t-PA in a pharmaceutical composition by incorporating arginine ('225 patent) in 1991.

2. Basic Molecular Genetics

Analysis of the current dispute necessitates an understanding of basic molecular genetics and recombinant DNA technology. I borrow liberally from the excellent tutorials of Andrew C. Webb, a Professor of Biological Sciences at Wellesley College, BM's expert, and Dr. Joseph Oliver Falkinham III, a Professor of Microbiology at Virginia Polytechnic Institute, Genentech's expert. (See Dockets 224, 227.)

a. Proteins

The human body is comprised of tissues and organs constructed of microscopic cells which carry out their specific functions through the creation of molecules called proteins. Proteins are chains of repeating molecules called amino acids. There are twenty different amino acids. The order of amino acids in a protein is unique to the protein. Some proteins called enzymes serve as the engines of the cell by driving chemical reactions.

In the late 1970s recombinant DNA technology was developed to allow individual proteins to be readily mass produced. The basic principle behind this technology is to give bacteria the DNA for a particular protein and have the bacteria make large quantities of it. The recipe for a protein in the DNA database is known as a gene. The genetic language of DNA consists entirely of three letter words called codons that are spelled using a four letter alphabet. These four nucleotides are the building blocks of DNA: A (for adenine), C (for cytosine), G (for guanine) and T (for thymine). Each three letter combination is known as a "triplet codon." For example, the three nucleotides AAA code for the amino acid lysine. The order of nucleotides spells out the order of amino acids in the protein.

DNA is generally found as two strands wrapped around one another like a spiral staircase, in a form known as the double helix. The two strands stick to one another, through pairing of the bases on opposite strands. A always pairs with T and G always pairs with C. Any non-conventional base pairing, such as A with G, is said to be mismatched. The ends of DNA single strands are chemically distinct. The front is referred to as the 5' end, and the back as the 3' end. The two complementary DNA single strands are base paired such that the 5' end of one strand is positioned opposite to the 3' end of the other. This 5' to 3' orientation allows the base sequence of genes to be "read" in the correct direction.

In its normal state DNA is folded in bundles called chromosomes which are sequestered in the nucleus of the cell. The nucleus is surrounded by the cytoplasm which is where the proteins are made. Bacterial cells do not contain a nucleus and are called prokaryotic cells. The genetic material inside a bacterial cell consists of the large, circular chromosome and sometimes bacterial plasmids, which are small circles of DNA. Bacterial plasmids often contain coding information for antibiotic resistance together with noncoding sequences such as promoters and ribosome binding sites. These bacterial plasmids are essential to recombinant DNA technology.

b. DNA replication

When cells reproduce or divide, the DNA copies itself in a process called DNA replication. During replication, the original DNA serves as a template for the newly synthesized version. DNA replication or multiplication is carried out by the enzyme DNA polymerase which takes advantage of the specific chemical attraction between bases as a foundation for faithful duplication.

*95 Genes are organized in DNA as transcriptional units. Both DNA and RNA polymerase will only recognize the correct position in the DNA to start copying while the DNA molecule is still in its double stranded configuration. These positions are the origin of replication in the case of DNA polymerase and the promoter in the case of RNA polymerase. However, polymerase copying of the DNA template can only occur when the DNA is rendered single stranded by severing of the AT and GC base pairs holding the double helix together. This separation of complementary strands in a DNA double helix is called denaturation. In its double-stranded form the DNA does not provide a template since the bases are already paired. For polymerase enzymes to function, it is imperative that the bases be exposed over short distances, so that new base pairing can occur.

The order of events for replication and transcription of DNA is as follows: (i) the polymerase enzyme recognizes and binds to a specific region of double-stranded DNA; (ii) the double helix denatures in this region to become two antiparallel, single strands; (iii) the polymerase enzyme initiates copying of the template strand, adding the next complementary base to the growing chain only in a 5' to 3' direction; and (iv) as the polymerase moves down the template strand, it progressively opens the double helix to reveal more single-stranded template to be copied.

c. Expression

Expression refers to the process by which the machinery of the cell reads the code recorded in the DNA and assembles protein following that code. The expression of DNA to produce proteins is a process involving steps of transcription and translation.

Information stored in DNA is transferred from DNA in order to make proteins by a courier molecule called messenger RNA or mRNA. RNA is single stranded. The DNA sequence is copied into a mRNA by the enzyme RNA polymerase during a process known as transcription. The mRNA carries the DNA sequence to the ribosomes (located in the cytoplasm) that are responsible for assembling amino acids into proteins by the process of translation. Translation, then, results in protein synthesis; the protein recipe contained in the mRNA is read by the ribosomes, one codon at a time into a sequence of amino acids.

Once bound to the promoter, RNA polymerase separates the two complementary DNA strands at the initiation site, joins the first two ribonucleotide bases of the gene based on their complementarity to the template DNA sequence, and terminates transcription at another specific DNA base sequence. At this termination site, the polymerase falls off the template DNA and releases the completed mRNA copy of the gene. Control of gene expression is influenced by proteins which can increase, decrease or inhibit the binding of RNA polymerase to its promoter.

The DNA sequence encoding the protein carried in the newly synthesized mRNA bases is interpreted by another type of RNA called transfer RNA (tRNA) with the help of the cell's ribosomes. tRNA transfers amino acids from the cytoplasm to the site of the protein synthesis on the ribosome. There are different tRNAs for each of the twenty amino acids. The initiation of protein synthesis starts with a ribosome "scanning" down the mRNA. At the start codon (ATG), specifying methionine, the initiator tRNA attaches methionine and the elongation of the protein chain begins, by stepwise codon recognition. Once the ribosome reaches a stop codon, it falls off the mRNA and ejects the completed protein. The protein has an amino ("N")-Terminus end and a Carboxyl ("C")-Terminus end.

The order and composition of amino acids in a protein will determine the shape into which the protein spontaneously folds. Folding of the protein chain is crucial to *96 the production of a functional protein, and failure to fold correctly or subsequent unfolding (denaturation) is accompanied by a partial or total loss of functional activity. Discrete units within folded proteins are known as domains. In addition to the folding of the protein, there are other modifications that have to take place before the protein can function. One example is the attachment of sugars (glycosylation) to specific amino acids in the chain. It is also extremely common for proteins to be made initially as large, often inactive "precursor" proteins. Such precursors have leader sequences of amino acids that are cleaved by specific enzymes to yield the smaller "mature" proteins with full activity. Some precursor proteins include a leader or signal sequence which is a string of amino acids which are found at the amino ("N") terminus of the protein sequence. Tissue-type plasminogen activator (t-PA) is initially made with a leader sequence which is removed to yield mature t-PA.

d. Primers

Unlike RNA polymerase, DNA polymerase requires a "starter" or primer before it can add bases one to another. Whenever DNA is replicated, either in a cell (in vivo) or in a test tube (in vitro), primers consisting of small lengths of single-stranded DNA or RNA, called oligonucleotides, must first base pair with their complementary sequence on the DNA template before DNA polymerase activity can begin. Oligonucleotides are short (approximately 15-50 nucleotides long) stretches of single-stranded DNA or RNA that are chemically synthesized on programmable instruments. These primers precisely define the starting point of DNA replication and are then added to by the polymerase, one base at a time, in a reaction referred to as a "primer extension." The stretches of newly synthesized DNA are glued together by another enzyme called DNA ligase.

The synthesis of copy or complementary DNA (cDNA) involves the single-stranded template of mRNA and the DNA polymerase called reverse transcriptase. Reverse transcriptase also requires a primer and moves in a 5' to 3' direction along a single-stranded mRNA.

3. DNA Recombinant Technology or Genetic Cloning

In gene cloning, the DNA to be cloned is inserted into a cloning vector, such as a plasmid, to form hybrid or recombinant DNA molecules that are then introduced into host cells like bacteria. There they are replicated as the bacterial cells divide. Plasmids are circular DNA molecules which replicate themselves when inserted into bacteria.

The basic technology for gene cloning relies on the use of specific enzymes to "cut" and then "paste" together fragments of DNA to form recombinant DNA. "Cutting" is accomplished by use of enzymes called restriction enzymes which recognize and cut DNA at specific nucleotide sequences called restriction sites. The cloning site into which DNA is inserted contains recognition sites for multiple restriction enzymes. Restriction enzymes cut the foreign DNA at certain specific DNA nucleotide sequences. The DNA of the vector is also cut with the same enzyme so that the ends of the fragments match. Cleavage with one of these restriction enzymes linearizes the plasmid circle, allowing insertion of foreign DNA. The "pasting" or rejoining at DNA fragments (also called ligation) is accomplished by the use of an enzyme called DNA ligase. If the recombinant DNA (vector and foreign DNA) is fused to a promoter, the gene can be expressed by the host cell to make the protein.

An example[2] of the process of cloning genes using bacterial plasmid vectors is as follows:

*97 First, isolate the plasmid vector DNA from bacteria and the foreign DNA containing the gene of interest. The source of foreign DNA to be cloned can either be chromosomal or cDNA.

Second, digest both the vector and foreign DNA with compatible restriction enzymes which are the molecular scissors of recombinant DNA technology. They allow large stretches of double-stranded DNA to be cut into manageably sized pieces or fragments. Some cut the two strands of DNA opposite each other, creating a "blunt" end. Others cut asymmetrically to yield a staggered or "sticky" end, where a short two or four base sequence protrudes as a single stranded extension to the end of the DNA. Because blunt ends can join only inefficiently to other blunt ends, one solution is to add synthetic linkers or adaptors onto the blunt ends of DNA to create the sticky end of choice. They are chemically synthesized on a "gene machine" as short, complementary, single-stranded oligonucleotides that contain the recognition sequence for one or more restriction enzymes. Following synthesis, the complementary strands are allowed to base pair, creating the synthetic double stranded fragment.

Third, the vector and foreign DNA cut with restriction enzymes (possibly modified with adapters and linkers) are joined together permanently by another enzyme called DNA ligase.

Fourth, when a plasmid with the desired DNA and a properly positioned promoter is introduced into bacterial cells, like E. Coli, the cells will express the DNA carried by the plasmid to encode the protein. This process is called cloning.

4. Genentech's patented method

The direct testimony of Dr. Jeffrey V. Ravetch, a Professor at the Rockefeller University and Head of the Laboratory of Molecular Genetics and Immunology, provides a useful framework for understanding the pioneering nature of the '832 patent.[3]

In the early days of biotechnology, scientists used as their models various hormones to make recombinant human proteins. The DNA for the hormones was made chemically by hand. Because this process was time consuming, it was difficult to make DNA for large proteins. Genentech scientists David Goeddel and Herbert Heyneker, the named co-inventors of the '832 patent, resolved this problem. They isolated the desired gene from biological sources and then cloned it by using the "cDNA" technique.

However, making proteins from sources such as cDNA resulted in having cDNA which was either "too long or too short." This is because cDNA genes typically coded for not only the desired protein itself but also for the leader sequence attached at the beginning of the protein. The leader sequence in nature helps the protein emerge from the human cell and is clipped off automatically as the protein leaves the cell. But when a mammalian protein is produced in a bacterial (prokaryotic) cell, no mechanism may be present in the bacterial cell to clip off the mammalian leader sequence. Cutting off the leader sequence sometimes resulted in cDNA that was too short in that it did not code for the mature enzyme. In other situations, the cDNA obtained by reverse transcription from the mRNA was too short in that it coded for a protein which was less than the desired protein.

The '832 patent provides a method of making a cloning vehicle capable of expressing a gene for a particular polypeptide by combining cDNA and synthetic DNA so that the DNA (in Goldilocks' words) is neither too long, nor too short, but "just right." In Dr. Ravetch's words, "[i]t provides the method allowing a researcher *98 to work backwards, starting with the idea of a particular polypeptide sequence he desires to make and instructing him on how to create a DNA plasmid capable of expressing that polypeptide sequence." (Ravetch Dir. Test. ¶ II.A.5) (Docket 226.)

5. Reteplase

a. Background

Reteplase is a new thrombolytic for the treatment of heart attack victims developed by BM. On October 30, 1996 Reteplase was approved by the FDA for use in the United States. Reteplase is a member of a class of protein based drugs used to dissolve the clots which block coronary arteries during the onset of a heart attack. It is not a naturally occurring protein. The DNA encoding Reteplase was created by changing the DNA of human t-PA. BM's scientific director Dr. Stephan Fischer[4] refers to it as a "designer gene" because "it is not simply a recombinant version of a natural protein but is an entirely new molecule with new properties." (Fischer Decl. II ¶ 11.)

Reteplase has 355 amino acids and consists only of the Kringle 2 ("K2") and Serine Protease ("P") domains. Hence, it is called K2P. In contrast to t-PA, none of the amino acids is modified with glycosidic groups. (See id. at ¶ 13.) A side-by-side comparison (in Attachment A hereto) reveals the differences between the two molecules.

Reteplase was developed as a result of a research program to search for new plasminogen activators which had improved pharmacological activity as compared to t-PA. BM sought to make a non-glycosylated plasminogen activator which would have a better half-life, and therefore be a better therapeutic agent than t-PA which is a glycosylated molecule. Thus, BM chose to use E. Coli as a host cell to express recombinantly a new plasminogen activator. Bacterial cells, like E. Coli, differ from eukaryotic cells in that they do not have the cellular machinery to attach glycosidic (sugar) molecules to the recombinantly made polypeptide. The development of Reteplase involved the use and modification of many different plasmids, until reaching the production plasmid PA27-T2L which contains the DNA encoding Reteplase.

All of the subsequent cloning steps utilized site directed mutagenesis ("SDM") according to the method of Morinaga et al., Bio/Technology 636-639 (July 1984), or a technique called the polymerase chain reaction ("PCR"). Before describing the development of plasmid PA27-T2L in detail, I digress for a moment to describe these two Nobel Prize winning techniques generally.[5]

b. Site-Directed Mutagenesis (SDM)

Mutations are changes in the sequence of a segment of DNA, known as a "parent" sequence. The changes can be single nucleotide changes, additions of nucleotides or deletions of nucleotides. The new sequence is called the "progeny" sequence. A technique known as site-directed mutagenesis allows scientists to insert a mutation at a specific preselected place in a piece of parent DNA. This process is done in a test tube. The technique of SDM uses an organically synthesized single-stranded oligonucleotide that is perfectly base-paired to the template parent DNA sequence, except that it is "mismatched" at the site of the base to be changed or mutated. This organically synthesized oligonucleotide acts as a primer for the reaction. *99 DNA polymerase extends the primer during the enzymatic synthesis of new progeny DNA.

For example, the primer sequence might lack a specified region of the parent template DNA sequence. It is organically synthesized with a sequence that is complementary to the template DNA sequence on either side (flanking) of the DNA region to be deleted. On extension of this primer, the mismatched region would be absent from the progeny sequence, and would, in that sense, have been "deleted" from the parent DNA sequence. When the synthetic primer is hybridized to the template strand, the unwanted region forms a loop or is "looped out." The DNA polymerase is added to extend the primer enzymatically in a clockwise 5' to 3' direction to make the second strand that is complementary to the template strand. Thus, a heteroduplex DNA molecule (consisting of the mutated strand and the looped strand) is created. This heteroduplex is inserted into a bacterial cell. As the normal DNA replication process takes place, each strand is copied to obtain two populations of double stranded DNA: one population with the original sequence, and the other with the looped sequence. The population of progeny double stranded, homoduplex molecules is isolated and used to express a mutant protein (a "deletion mutant") with a shorter amino acid sequence than that coded by the parental DNA sequence.

A modified SDM method is known as the Morinaga or "gapped duplex" method. A key difference is that it uses a partially double-stranded DNA molecule having a single-stranded gap as a template region instead of a completely single-stranded template. The primer containing the desired change hybridizes to a template site in the single-stranded gap again yielding a loop-out of the unwanted region. DNA polymerase then fills any gap that remains between the ends of the primer and the double-stranded DNA via enzymatic synthesis, and DNA ligase joins the ends. Thus, only the gap is filled by enzymatic sequences, rather than the whole progeny strand.

c. Polymerase chain reaction (PCR)

The polymerase chain reaction ("PCR") is an in vitro reaction in which a specific region of DNA is amplified several million-fold by repeated synthesis using a DNA polymerase and specific oligonucleotide primers to define the ends of the DNA region to be amplified. To perform PCR, a scientist needs the sequence of DNA flanking the region to be amplified. A pair of primers is synthesized, usually 18-30 bases in length. One primer is complementary to the sequence at one end of the DNA region, and the other is complementary to the sequence on the opposite strand of the DNA at the other end of the region to be amplified. Primers initiate DNA synthesis by annealing or hybridizing to their complementary DNA base sequence and DNA polymerase then extends them only in a 5' to 3' direction, adding nucleotide bases according to the template sequence.

PCR reactions take place in the test tube where double-stranded DNA and oligonucleotide primers are mixed. The PCR cycle consists of repeated rounds of (i) separating template double DNA strand by heat denaturation; (ii) after cooling, annealing of primers to their complementary nucleotide sequence on the two separate, template DNA strands; and (iii) after reheating to the optimal temperature, synthesizing new DNA by polymerase extension of primers. By repeated heating to separate both original and newly synthesized strands, the net result of PCR after n cycles is 2n double stranded copies of the original DNA sequence between the primers.

What results are two double-stranded DNA molecules, one strand of which is the parental template and the other strand (the progeny) which incorporates the organically synthesized primer and enzymatically synthesized DNA. The amplification produces a piece of DNA that is a hybrid *100 piece containing (or marrying, if you will) the synthetic DNA primer and DNA copied from the target sequence.[6]

Fusion PCR is an adaptation of conventional PCR methodology. As in regular PCR, the purpose of fusion PCR is to join two separately amplified PCR fragments together by means of an overlap between nucleotides they have in common on the ends of the fragments.

An example will help to explain fusion PCR where two separate DNA fragments A and B are involved. Following PCR, the amplified, double-stranded DNA fragments are mixed, denatured by heating to yield single-strands, and then cooled to allow base pairs to reform ("renature"). Different combinations of renatured DNA strands can result. Complementary strands can hybridize over their entire length to recreate separate molecules of fragment A or B. Another possible combination is that one strand from fragment A can cross-hybridize with a strand from fragment B by base-pairing over their complementary, overlapping ends. In the presence of DNA polymerase, enzymatic DNA synthesis will take place in a 5' to 3' direction to create a fusion of fragment A to fragment B. Once the initial fusion product has been formed, conventional PCR using the original flanking primer can be used to generate large quantities of the fused DNA fragment.

d. BM plasmids[7]

1. Plasmid pBT95, encoding precursor t-PA

To make plasmid pBT95, BM isolated cDNA encoding precursor t-PA; that is, the DNA sequence for both (i) the mature t-PA coding region (i.e., the part of the gene coding for t-PA's five domains), and (ii) the leader sequence located upstream of the 5' end of the mature t-PA coding region.

The precursor t-PA had been made using the enzyme reverse transcriptase to copy t-PA messenger RNA ("mRNA") into cDNA. Plasmid vector DNA was combined with the cDNA by "cutting and pasting" the vector DNA and cDNA resulting in the recombinant plasmid designated pBT95.

pBT95 contained, in addition to the coding region for precursor t-PA, certain non-coding DNA sequences which affected the production of the t-PA protein, but are located outside the gene for t-PA. These included a promoter known as the tac promoter, a ribosome binding site (or Shine-Delgarno sequence), and an ATG start signal for translation, all located upstream of the 5' end. pB95 also included DNA known as the 3' untranslated region, located downstream of the 3' end.

2. Plasmid pePA98.1, encoding mature t-PA

In 1985, BM embarked on a research project to see if mammalian proteins could be expressed in E. Coli. It constructed a plasmid encoding mature t-PA by joining an organically synthesized linker designed to maximize expression to a cDNA fragment encoding t-PA and inserted that hybrid into a plasmid, designated pePA98.1. That hybrid was inserted into E. Coli for the purpose of expressing the cDNA encoding mature t-PA. This project was a failure because it produced improperly folded, biologically inactive protein. The improper folding problem was particularly acute with t-PA because the native protein possesses a very complex structure.

The t-PA leader amino acid sequence is not recognized, and thus, is not cleaved by E. Coli enzymes. It was therefore necessary, if active (i.e., mature) t-PA were to be expressed in E. Coli, to make plasmid *101 pePA98.1 in which the DNA coding for the leader sequence was removed from the precursor t-PA gene.

Plasmid pePA98.1 was made by cutting pBT95 at a restriction site close to the junction between the DNA coding for the leader sequence and the DNA coding for mature t-PA. A nuclease enzyme was used to "chew back" the sticky end left by the restriction enzyme, but this removed two nucleotides of the first codon coding for mature t-PA, which codes for the amino acid serine. Furthermore, in addition to the DNA coding for the leader sequence (35 codons) and the first two nucleotides of the codon for serine that were "chewed back," the cutting and "chewing back" operations also removed the start codon (ATG). These operations resulted in a mature t-PA coding sequence that was two nucleotides too short. BM replaced these two nucleotides. It organically synthesized partially complementary oligonucleotides to make a double stranded "linker" that provided, in addition to the first two nucleotides for serine, the obligatory start codon (ATG) for translation (which by necessity had been removed along with the leader sequence), additional nucleotides to create a sticky end to facilitate ligation to the plasmid, and additional nucleotides to optimize the distance between the ribosome binding site and the start codon.

The organically synthesized DNA fragment had the sequence:

5' AATTCTTATG TC 3'
3' GAATACAG 5'

As already noted, it was designed specifically to create the optimal distance between the ATG translation start signal (underlined above) and the ribosome binding site, and to facilitate insertion of the cDNA coding for mature t-PA into the expression vector. As shown, the last two nucleotides of the organically synthesized DNA fragment constituted two nucleotides removed by the chewing process, i.e., the TC of the TCT serine codon. The second T of the TCT serine codon was present on the cDNA.

In sum, plasmid pePA98.1 differed from pBT95 in that the DNA coding for the leader sequence of t-PA was removed, and the translation start signal ATG was moved from its previous location at the beginning of the leader to a new position next to the first codon (serine) of the DNA encoding mature t-PA.

3. Plasmid pePA126.1, encoding mature t-PA

Plasmid pePA126.1 was constructed in order to delete the 3' untranslated region of the gene for t-PA. The 3' untranslated region is the DNA that is immediately adjacent to, but downstream of, the 3' end of the DNA coding region (i.e., it is not in the coding region and therefore not translated).

The 3' untranslated region was removed by cutting pePA98.1 and pasting the cut DNA into a vector. The resulting plasmid was designated pePA126.1. All of the cutting and pasting occurred outside of the region encoding t-PA. Plasmid pePA126.1 contained the same t-PA coding sequence as pePA98.1 (i.e., all of the DNA coding for mature t-PA).

4. Plasmid pePA133, encoding mature t-PA

In 1985, BM undertook a limited project using a plasmid designated pePA133. This plasmid contained the same coding sequence as pePA98.1 but was a "low copy" plasmid. This project was successful and active t-PA was obtained by in vitro folding of t-PA isolated from inclusion bodies. A patent was originally filed in Germany in October 1985 and was issued in the U.S. in September 1995 as United States Patent No. 5,453,363 to Rudolph et al., assigned to BM.

Plasmid pePA133 was made in order to change the number of plasmid copies inside the bacterial host cell. Somewhat oversimplified, each plasmid contains a DNA sequence controlling the level of its *102 replication. Plasmid pePA126.1 contained a DNA sequence which resulted in a high plasmid copy number inside the bacterial cell.

Plasmid pePA133 was produced by excising the gene for mature t-PA, including its promoter, from plasmid pePA126.1, and combining this fragment, using cutting and pasting techniques, with a DNA fragment from another plasmid vector containing a DNA sequence yielding a low level of copies. It was thought that this change to low copy would help increase the amount of mature t-PA protein produced. Plasmid pePA133 contained the same t-PA coding sequence as pePA98.1 (i.e., all of the DNA coding for mature t-PA). In other words, pePA133 does not have any modification in the nucleotide sequence encoding the semisynthetic gene for t-PA. Only sequences outside the coding region were manipulated.

5. Plasmid pePA126fd, encoding mature t-PA

To construct another of its plasmids, plasmid pePA126fd, BM added a strong fd terminator of transcription to pePA126.1. This was done to increase the stability of the t-PA mRNA and hence increase the yield of t-PA protein in the host cells. The fd terminator was cut from the fd phage (a virus that infects bacteria) and ligated to the 3' end (but outside) of the t-PA coding sequence carried in plasmid pePA126.1. Thus, as with plasmid pePA133, the coding sequence of the resulting plasmid, pePA126fd, remained the coding sequence for mature t-PA; nothing was added to or subtracted from that coding sequence.

6. Plasmid pREM7685, encoding FK2 P

Using the DNA from plasmid pePA133 which codes for mature t-PA as a template, BM applied the Morinaga SDM technique to obtain a plasmid designated pREM7685 encoding only the F, K2, and P domains of t-PA. In this application of SDM, a cDNA fragment from the middle of the t-PA coding region is eliminated and a synthetic fragment which bonds to the coding information flanking the deleted region is incorporated into the plasmid. The sequences that encode the amino terminus end of the polypeptide are still synthetic DNA sequences that are provided from plasmid pePA98.1.

The following steps were performed in the construction of pREM7685:

(i) BM started with two identical pePA133 plasmids, each encoding mature t-PA (domains FEK1K2P). Plasmid 1(a) was cut with a restriction enzyme to make a single cut in the circle. Plasmid 1(b) was cut in two places with restriction enzymes outside the region to be deleted. In this case, plasmid 1(b) was cut just upstream of the F domain and also in the P domain.
(ii) One strand from plasmid 1(a) was then hybridized with one strand from plasmid 1(b) by heat denaturing each plasmid and allowing the strands to cross-hybridize producing a hybrid intermediate commonly referred to as a "gapped duplex."
(iii) A single-stranded oligonucleotide primer was synthesized having a sequence complementary to a short sequence on each side of the DNA to be deleted. In this case, one-half of the primer consisted of a short sequence (GCCTGTCAAA) complementary to the end of the F domain and the other half consisted of a short sequence (GGAAACAGTGA) complementary to DNA at the beginning of the K2 domain. In other words, the SDM elimination step was done using a synthetic oligonucleotide with a sequence of codons bonding to amino acid codons 46-49 and 176-179 of full length t-PA—coding already provided by the single stranded cDNA template—in order to effect a deletion of the cDNA sequences encoding the regions for amino acids 50-175.
(iv) The primer was then hybridized to the template DNA strand in the *103 "gap" resulting in the E and K1 domains forming a "loop."
(v) Next, the remaining gaps on either side of the primer were extended enzymatically by adding DNA polymerase and nucleotides, and the ends were then sealed with DNA ligase. The synthetic strand consisted of 21 nucleotides. The gap filled in enzymatically between the synthetic strand and the cDNA encoding portions of the P region constituted over 500 nucleotides. The resulting heteroduplex DNA was comprised of one DNA strand corresponding to the original parental sequence (FEK1K2P) and the other strand comprised of DNA corresponding to the mutant progeny sequence (FK2P).
(vi) This heteroduplex was then introduced into the bacterium, E. Coli. Once in E. Coli, each strand of the heteroduplex was copied independently, resulting in two plasmid populations—one a copy of the original parental plasmid (which coded for FEK1K2P) and the other a plasmid carrying a deletion mutant of the t-PA gene (coding for FK2P), the nucleotides coding for the E and K1 domains having been "deleted" by use of the Morinaga method.

The synthetically synthesized primer used in SDM was not incorporated at the N-terminus of the gene (i.e., the 5' end). On the contrary, it was incorporated into the middle part of the coding region for the mutant t-PA gene.

7. Plasmid pA27.3 encoding Reteplase (K2P)

Using the DNA from pREM7685 (encoding FK2P) as a template, BM again used the Morinaga SDM technique to remove the DNA region encoding the F domain of t-PA. The resulting plasmid, encoding only K2P, was designated pA27.3. Plasmid pA27.3 was the first plasmid made which contained DNA encoding the protein now known as Reteplase. Genentech contends that the production of this plasmid violates its patent. BM hotly disputes this contention. Understanding the production of this plasmid is key to understanding the infringement claim. See Attachment B.

The following steps were performed in the construction of pA27.3:

(i) BM started with two identical pREM7685 plasmids each encoding domains F, K2, and P of t-PA. Plasmid 1(a) was cut with a restriction enzyme to make a single cut in the circle. Plasmid 1(b) was cut in two places with restriction enzymes outside the region to be deleted. In this case, plasmid 1(b) was cut just upstream of the F domain and in the P domain.
(ii) One strand from one plasmid 1(a) was then hybridized with one strand from the other plasmid by heat denaturing each plasmid and allowing the strands to cross-hybridize to produce a hybrid "gapped duplex" intermediate.
(iii) A single-stranded oligonucleotide primer was synthesized having a sequence complementary to a short sequence on each side of the DNA to be deleted from the template strand. In this case, one-half of the primer consisted of a short sequence (TGTCTTACCAA) complementary to the codons for amino acids 1, 2 and 3 of mature t-PA and the other half consisted of a short sequence (GGAAACAGTGA) complementary to the beginning of the K2 domain.
(iv) The primer was then hybridized to the "template" strand such that it "straddled" the region between the DNA sequence that codes for amino acids 1-3 of mature t-PA and the beginning of the K2 domains causing the F domain to "loop out."
(v) Next, the remaining gaps on either side of the primer were filled in enzymatically by DNA polymerase and the ends then sealed with DNA ligase, resulting in the formation of a heteroduplex wherein the one strand comprised DNA corresponding to the pREM7685 parental plasmid (FK2P) and the other *104 strand comprised DNA corresponding to the mutant progeny sequence (K2P) The gap filled in enzymatically for portions of the K2 and P domains consisted of over 500 nucleotides.
(vi) This heteroduplex was then introduced into the bacterium, E. Coli. Once in E. Coli, each strand of the heteroduplex was copied independently, resulting in two populations—one a copy of the original parental plasmid, which coded for FK2P, and the other a plasmid carrying a deletion mutant gene for t-PA, which coded for K2P (i.e., Reteplase), the nucleotides coding for the F domain having been "deleted" by means of the Morinaga method.

Although plasmid pA27.3 expressed the Reteplase protein, the yield was very low and most of the host cells showed impaired viability. Thus, plasmid pA27.3 could not be used for commercial expression of Reteplase.

Significantly, synthetic primer used in the SDM occurs at the amino acid sequences at the N-terminus. During SDM, synthetic DNA was used to create the deletion of a specific sequence which BM wanted to eliminate. The synthetic primers (of 22 nucleotides) for the SDM annealed with the cDNA template at amino acids 1-3 and 176-179, and were incorporated in the plasmid to create the deletion of the F domain. The codons for amino acids 50 to 175 had been deleted in the predecessor plasmid pREM7685.

8. pA27fd encoding Reteplase

To increase the yield of Reteplase, the fd terminator was added downstream of the 3' end of the region coding for the protein. As noted above, in connection with BM's construction of the plasmid pePA126fd, the fd terminator is a naturally occurring DNA sequence found in a bacterial virus, known as the fd phage. The fd terminator does not code for any protein. It is a DNA sequence which regulates transcription. The expression of mammalian proteins in E. Coli. is significantly enhanced by adding the fd terminator downstream of the DNA coding for a protein.

This insertion of the fd terminator downstream of the K2P coding region to create plasmid pA27td was done by cutting and pasting DNA fragments from plasmids pA27.3 and pePA126fd (a plasmid containing DNA encoding mature t-PA and an fd terminator). These changes occur outside of the coding sequence for Reteplase.

9. The Final Production Plasmid for Reteplase pA27-T2L

To make the Reteplase production plasmid, pA27-T2L, the promoter in front of the DNA encoding Reteplase in plasmid pA27fd was switched. A promoter designated T2L was substituted for the existing "tac" promoter.

Plasmid pA27-T2L was obtained by combination of three fragments derived from plasmid pA27fd and another plasmid already containing the T2L promoter. One of the three fragments carrying the T2L promoter was made using a combination of conventional PCR and fusion PCR. This fusion PCR generated fragment contained the T2L promoter and part of the Reteplase coding region. The other two fragments used to construct pA27-T2L were isolated by cutting the DNA in plasmid pA27fd with restriction enzymes. These two fragments contained the other part of the Reteplase coding region, the fd terminator, and the remainder of the pA27-T2L plasmid vector outside the coding region. All three fragments were pasted together to make pA27-T2L.

None of the primers was incorporated into the DNA region encoding the N-terminus amino acid sequence of Reteplase.

DISCUSSION

1. Summary Judgment Standard

"Summary judgment is appropriate when `the pleadings, depositions, answers *105 to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.'" Barbour v. Dynamics Research Corp., 63 F.3d 32, 36 (1st Cir.1995) (quoting Fed. R.Civ.P. 56(c)), cert. denied, 516 U.S. 1113, 116 S. Ct. 914, 133 L. Ed. 2d 845 (1996). "To succeed [in a motion for summary judgment], the moving party must show that there is an absence of evidence to support the nonmoving party's position." Rogers v. Fair, 902 F.2d 140, 143 (1st Cir.1990); see also Celotex Corp. v. Catrett, 477 U.S. 317, 325, 106 S. Ct. 2548, 91 L. Ed. 2d 265 (1986).

"Once the moving party has properly supported its motion for summary judgment, the burden shifts to the non-moving party, who `may not rest on mere allegations or denials of his pleading, but must set forth specific facts showing there is a genuine issue for trial.'" Barbour, 63 F.3d at 37 (quoting Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 256, 106 S. Ct. 2505, 91 L. Ed. 2d 202 (1986)). "There must be `sufficient evidence favoring the nonmoving party for a jury to return a verdict for that party. If the evidence is merely colorable or is not significantly probative, summary judgment may be granted.'" Rogers, 902 F.2d at 143 (quoting Anderson, 477 U.S. at 249-50, 106 S. Ct. 2505) (citations in Anderson omitted). The Court must "view the facts in the light most favorable to the non-moving party, drawing all reasonable inferences in that party's favor." Barbour, 63 F.3d at 36.

2. The '832 Patent

Genentech asserts that the Reteplase production plasmids infringe the '832 patent. The determination of whether an accused process infringes a claim in a patent involves two steps. First, the court must construe the claim asserted to be infringed to determine its meaning and scope. Second, it must compare the properly construed claim to the accused process. See Tanabe Seiyaku Co. v. United States Int'l Trade Comm'n, 109 F.3d 726, 731 (Fed.Cir.1997) ("Tanabe"). To prove infringement, Genentech must prove by a preponderance of the evidence that the accused process embodies each and every claim limitation either literally or by equivalence. See Id.; Conroy v. Reebok Int'l, Ltd., 14 F.3d 1570, 1573 (Fed.Cir.1994); Warner-Jenkinson Co., Inc. v. Hilton Davis Chemical Co., 520 U.S. 17, 21, 117 S. Ct. 1040, 137 L. Ed. 2d 146 (1997). Whether a process infringes literally or under the doctrine of equivalents is a question of fact. Tanabe, 109 F.3d at 731.

a. The Claim

Entitled "Method of Constructing a Replicable Cloning Vehicle Having Quasi-Synthetic Genes," the '832 patent "provides a method of general applicability" for the production of useful proteins of known amino acid sequence, including antibodies and enzymes, and is "particularly suited to the expression of mammalian polypeptide hormones and other substances having medical applications." ('832 Patent, Col. 8, 11. 3-11.) The inventors claim that their application represents "the first occasion upon which a medically significant human polypeptide was directly expressed microbially, rather than in conjunction with extraneous protein." ('832 prosecution history, Markman Hearing, Pl.Ex. 2, at p. 103.)

To do this, the '832 patent teaches a method of making a plasmid using two types of DNA, synthetic DNA, and cDNA derived from messenger RNA. (See '832 patent, Col. 10, 11. 35 et seq.) Specifically, the '832 patent describes "a method of using organically-synthesized DNA to create an optimal 5' end (encoding the amino terminus of a protein) for the desired protein. The organically-synthesized DNA is joined to the cDNA, to create a chimeric [or semi-synthetic] gene." (Cepko Report at p. 1.)

Claim 1 of the '832 patent teaches the method of constructing the "replicable cloning vehicle" capable of expressing a *106 gene for a particular polypeptide of known amino acid sequence by combining cDNA and synthetic DNA. (See '832 patent, Col. 13, 11. 6-43.) Steps (a) and (b) involve "obtaining by reverse transcription from messenger RNA a first gene fragment" which encodes less than all of the amino acid sequence of the polypeptide. Critical to both step (a) and step (b) is that the whole reading sequence is not contained in the cDNA fragment. Step (c) involves providing by "organic synthesis one or more synthetic non-reverse transcriptgene fragments" for the "remainder" of the sequence. At least one of the synthetic fragments is described as coding for the amino terminus portion of the polypeptide. Step (d) involves joining the synthetic gene fragment(s) of Step (c) to the gene fragment(s) described in Steps (a) and (b) and inserting these gene fragments into a plasmid in proper reading phase. According to Ravetch, "[i]n order for the synthetic and cDNA fragments to combine so that a particular polypeptide can be expressed, they must be correctly positioned relative to each other so that gene code will be read to result in the particular polypeptide." (Ravetch Dir. Test. ¶ II.B.21.) See Attachment C.

Claim One of the '832 patent reads:

In the method of constructing a replicable cloning vehicle capable, in a microbial organism, of expressing a particular polypeptide of known amino acid sequence wherein a gene coding for the polypeptide is inserted into a cloning vehicle and placed under the control of an expression promoter, the improvement which comprises:
(a) obtaining by reverse transcription from messenger RNA a first gene fragment for an expression product other than said polypeptide, which fragment comprises at least a portion of the coding sequence for said polypeptide;
(b) where the first fragment comprises protein-encoding codons for amino acid sequences other than those contained in said polypeptide, eliminating the same while retaining at least a substantial portion of said coding sequence, the resulting fragment nevertheless coding for an expression product other than said polypeptide;
the product of step (a) or, where required, step (b) being a fragment encoding less than all of the amino acid sequence of said polypeptide;
(c) providing by organic synthesis one or more synthetic non-reverse transcript-gene fragments encoding the remainder of the amino acid sequence of said polypeptide, at least one of said fragments coding for the amino-terminus portion of the polypeptide; and
(d) deploying the synthetic gene fragment(s) of step (c) and that produced in step (a) or (b), as the case may be, in a replicable cloning vehicle in proper reading phase relative to one another and under the control of an expression promoter;

whereby a replicable cloning vehicle capable of expressing the amino acid sequence of said polypeptide is formed.

In the Markman hearing, the Court determined that the term "gene fragment" may refer to either single or double-stranded DNA, of any length, even two nucleotides. Genentech, 989 F.Supp. at 366. The Court construed the term "organically synthesized" to mean "the production of an oligonucleotide or fragment of a gene using organic chemistry without the use of an enzyme. It includes DNA made by replication of synthetic DNA." Id. at 364.

b. "Material Change" Under Section 271(g)

Under this Court's claim construction, BM concedes that the process for producing the intermediate plasmid pePA 98.1 which produces t-PA literally infringes the '832 patent because it combines synthetic DNA which codes for two of the *107 nucleotides for serine in the t-PA molecule with cDNA to direct expression of a known gene sequence in E. Coli.

However, BM argues that the Reteplase production plasmids do not infringe the patent. Stressing that these plasmids were produced outside the United States, in Germany, and have been materially changed from intermediate plasmid pePA98.1, BM contends that it is not liable for infringement under the Process Patent Amendments Act of 1988, Pub.L. No. 100-418, §§ 9001-9007, 102 Stat. 1107, 1563 (1988) (codified in various sections of 35 U.S.C.). This Act deems it an infringement to import, sell, offer to sell, or use in this country a product that was made abroad by a process protected by a United States patent. See 35 U.S.C. § 271(g). However, the Act does not apply if the product made by the patented process is "materially changed by subsequent processes" before it is imported. 35 U.S.C. § 271(g)(1).[10]

Section 271(g) requires two separate inquiries. First, it requires that a patentee establish that an accused infringer imported a product made by a process falling within the scope of one or more of the claims of the patent. See Novo Nordisk of North America, Inc. v. Genentech, Inc., 77 F.3d 1364, 1367-68 (Fed.Cir.1996) ("Novo Nordisk"). Second, once the patentee proves that the process falls within the literal scope of the patent, the court must determine whether the "materially changed" provision of Section 271(g) applies. "The `materially changed' exception of Section 271(g) requires, at a minimum, that there by a real difference between the product imported, offered for sale, sold, or used in the United States and the products produced by the patented process." Bio-Technology General Corp. v. Genentech, Inc., 80 F.3d 1553, 1560 (Fed.Cir.1996) ("BTG"). To determine whether the "materially changed" provision applies, the court must look to the substantiality of the change between the product of the patented process and the product that is being imported. See Eli Lilly & Co. v. American Cyanamid Co., 82 F.3d 1568, 1573 (Fed.Cir.1996).

C. Allocating the Burden of Proof under Section 271(g)

The statute is unclear as to whether the patentee or alleged infringer bears the burden of demonstrating "material change." The issue has been addressed, in passing, by two federal courts. In Ajinomoto Co., Inc. v. Archer-Daniels-Midland Co., 1996 WL 621837 (D.Del. Oct.21, 1996), the court discussed whether, in the circumstances of the case, the burden of proof for showing a "material change" under Section 271(g) was on the alleged infringer. See 35 U.S.C. § 295 (1994). The court's analysis hinged on Section 295's burden-shifting mechanism. Under Section 295, the accused infringer's product is presumed to have been made by the patented process if the trial court finds that (1) a substantial likelihood exists that the challenged product was made by the patented process, and (2) the patentee made a reasonable effort to determine the process actually used in the production of the product, but was unable to do so. "If the trial court makes these findings, the burden of establishing that the product was not made by the *108 patented process is on the accused infringer." Novo Nordisk, 77 F.3d at 1368 n. 6. Section 295 is inapplicable where the patentee is able to determine the process which the alleged infringer used. See Id. In Ajinomoto, the court concluded that, under section 295, if this "initial hurdle is surmounted, it is [the alleged infringer's] burden to demonstrate that its products have been materially changed." Ajinomoto, 1996 WL 621837, *12. However, because the initial burden was not met, the patentee was not afforded the benefit of the burden-shifting presumption. Id.

In Eli Lilly & Co. v. American Cyanamid Co., 896 F. Supp. 851 (S.D.Ind.1995), aff'd 82 F.3d 1568 (Fed.Cir.1996), involving a motion for a preliminary injunction, the Court held that, "when read as a whole, the two parts of section 271(g) require the plaintiff to demonstrate ... that the product made by the patented process is neither materially changed by subsequent processes nor a trivial and nonessential component of another product." Id. at 855-56. The Federal Circuit did not directly address the issue on appeal.

This case law suggests that a patentee must bear the burden of proof on the issue of "material change" unless it satisfies the burden-shifting requirements of Section 295. Here, because there has been ample opportunity for discovery and Genentech has in fact determined the process that BM used, Section 295 is inapplicable. Novo Nordisk, 77 F.3d at 1368 n. 6. Accordingly, Genentech bears the burden on the issue of material change.[11]

d. Analysis

As a preliminary matter, this Court must determine whether Reteplase is the "product" made by the patented process or whether the plasmid is the "product" for purposes of Section 271(g). BTG answers this question. Addressing a claim of the infringement of Genentech's '832 patent under Section 271(g) by a company that produced human growth hormone (hGH), the Federal Circuit determined that hGH was "a product which is made by a process patented in the United States," even though claim 1 of the '832 patent was literally directed to a method for producing a replicable cloning vehicle (e.g. a plasmid), not hGH. See BTG, 80 F.3d at 1560-61. Even though there was "little doubt that the plasmid product of the claimed process and hGH [were] entirely different materials, one being more than materially changed in relation to the other," the Federal Circuit held:

The legislative history precisely anticipated this fact situation and indicated Congress's intent that infringement of a process for making a plasmid is not to be avoided by using it to express its intended protein. Moreover, the '832 patent itself explicitly contemplates that the patented process will be used as part of an overall process for producing hGH; indeed the patent discloses in detail how to make hGH by carrying out the claimed process and other necessary steps. Thus, it cannot be said as a matter of law that the production of hGH is too remote from the claimed process of making a replicable cloning vehicle. We therefore find no error in the court's conclusion that hGH is a product that is "made by" the '832 patented process.

Id. at 1561.

BTG is also helpful in resolving the present dispute because of the Federal Circuit's reliance on legislative history, specifically the "two-phased test" that the Senate Judiciary Committee recommended for determining whether a product was "materially changed" prior to its importation. Id. at 1576. This report outlined the test as follows:

In order to give the courts Congressional guidance in what may be a difficult determination, the Committee notes that *109 the bill would establish the following two-phased test:
1. A product will be considered made by the patented process regardless of any subsequent changes if it would not be possible or commercially viable to make that product but for the use of the patented process. In judging commercial viability, the courts shall use a flexible standard which is appropriate to the competitive circumstances.
2. A product will be considered to have been made by a patented process if the additional processing steps which are not covered by the patent do not change the physical or chemical properties of the product in a manner which changes the basic utility of the product by the patented process. However, a change in the physical or chemical properties of a product, even though minor, may be "material" if the change relates to a physical or chemical property which is an important feature of the product produced by the patented process. Usually, a change in the physical form of a product (e.g., the granules to powder, solid to liquid) or minor chemical conversion (e.g., conversion to a salt, base, acid, hydrate, ester, or addition or removal of a protection group) would not be a "material" change.

S.Rep. No. 83, 100th Cong., 1st Sess. 50 (1987). As the touchstone for determining material change, the Committee admonished that the courts must "exercise careful judgement (sic) in distinguishing those products that are too far removed from the patented process, and those that have been changed only in insignificant ways." Id. at 49.

With respect to the first infringement inquiry under Section 271(g), Genentech argues that the process for constructing the BM production plasmids (pA27.3, pA27fd and pA27-T2L) violates claim one of the '832 patents because organically synthesized fragments are combined with the reverse transcriptase produced sequence of the Reteplase gene.

BM makes four arguments to refute Genentech's claim of literal infringement. Three are flawed; the last is not. First, BM argues that the first three steps of the method proscribe a process for constructing gene fragments for a "particular polypeptide of known amino acid sequence." Because BM did not perform step (a)— obtaining a gene fragment of cDNA by reverse transcription from mRNA—at a time when it knew the amino acid sequence of Reteplase, it insists, it did not infringe the patent. However, this "conflation" theory—as Genentech dubs it—is not persuasive. As Dr. Ravetch said: "[Claim 1] provides the method allowing a researcher to work backwards, starting with the idea of a particular polypeptide sequence he desires to make and instructing him on how to create a DNA plasmid capable of expressing that polypeptide sequence." (Ravetch Dir. Test. ¶ II.A.5.) At each of the stages of producing a plasmid, BM knew the amino acid sequence it desired.

Second, BM argues that in the construction of pA 27.3, there was no performance of step (a) of the '832 patent—obtaining a gene fragment by reverse transcription— because the cDNA was obtained by removing the cDNA from one plasmid and transferring it to another. Therefore, according to BM, there could be no performance of step (b) of the patent—shortening the product of step (a).

The patent refers to cDNA as a "[gene] fragment derived by reverse transcription from mRNA." ('832 patent, Col. 5, 11. 2-4.) Dr. Ravetch states: "cDNA is always obtained by reverse transcription by definition, whether it is immediately by reverse transcription or obtained by reverse transcription followed by other manipulations, such as replication in a plasmid and removal from the plasmid. Replicated cDNA is still cDNA obtained by reverse transcription, just as synthetic DNA which is replicated and then removed from a plasmid is *110 still `provided by organic synthesis ...'" (Ravetch Decl. ¶ 14-16.)[12]

BM does not dispute that cDNA is a gene fragment obtained by reverse transcription and that replicated cDNA is cDNA, but argues that the nomenclature does not absolve Genentech of the obligation to perform the step of obtaining cDNA by the process of reverse transcription during the construction of each plasmid. Dr. Ravetch points out that even in the '832 patent, the cDNA fragment used to create the designer gene for human growth hormone was replicated and transferred from plasmid to plasmid. (Id. at ¶ 16 and figures 4 and 5.) Therefore, the intrinsic evidence suggests that BM's argument that it must perform reverse transcription each time it makes an intermediate plasmid must be rejected. Even if BM were correct on this point, Genentech would be easily saved by the doctrine of equivalents. As Genentech persuasively points out, under the doctrine of equivalents, replicated cDNA is the substantial equivalent of reverse transcript cDNA. See Warner-Jenkinson, 520 U.S. at 39-40, 117 S. Ct. 1040. BM submits no evidence that they are not interchangeable, or that they are substantially different.

Third, BM argues that its process for making the production plasmids does not infringe claim one because step (c) requires that the missing coding information (the "remainder" of the gene), which includes information missing from the amino-terminus portion of the desired polypetide, be supplied by an organically synthesized gene fragment. At no point in the construction of the Reteplase plasmids, it argues, is there any "missing" coding information for the designer gene Reteplase because the template strand provides the coding information. Along a similar vein, BM contends that the SDM technique uses a synthetic primer to hybridize to the wanted regions of the template strand, causing unwanted coding regions (in this case the "domains") to "loop out." This is designed to delete or modify coding information, argues BM, not provide "missing" coding information.

The record does not support BM's "synthetic DNA does not add new information" theory. Rather, as Genentech points out, BM used synthetic primers complementary to a DNA template strand, which provided missing coding information at the N-terminus when the cDNA was too long or short. Dr. Cepko points out that in this plasmid, "an organically-synthesized gene fragment was joined to a cDNA" at the 5' end.[13] (Cepko Report at pp. 2, 4.) "Missing," as used in the patent, does not mean that the coding information is unknown; it means that the DNA sequences which reflect known coding information are missing.

BM presses on, pointing out that it does not infringe because the missing coding *111 information was only partially provided by a synthetically created primer. The key to this complex and difficult scientific dispute is understanding the process for the production plasmid pA27.3 which incorporates synthetic DNA at the N-terminus through the use of SDM. As Dr. Webb's illustrations to his supplemental declaration demonstrate,[14] the way in which the synthetically synthesized fragment (for amino acids 1-3 and 76-79) was joined to the DNA does not infringe claim one. That is so because the synthetic fragment and the cDNA were not deployed in proper reading phase in a plasmid in the manner set forth in steps (c) and (d). Rather, in the heteroduplex, there was a significant gap of over 500 nucleotides which was filed in enzymatically in order to provide the missing coding information for the remainder of the K and P regions. The missing coding information for the plasmid was only partially provided by organic synthesis at amino acids 1-3 and 76-79. The videotaped testimony of Dr. Ravetch does not dispute this point, and his chart depicts this enzymatic gap-filling with the same squiggly lines used by Dr. Webb. Dr. Ravetch states: "The synthetic DNA acts as a primer and then the target strand is copied based on the primer for the template that one is using." (Ravetch Decl. ¶ 50.) Compare Attachments B and C.

Accordingly, there is no evidence that BM deployed a synthetic gene fragment which provided the remaining coding information (i.e., "the remainder") and a cDNA derived by reverse transcriptase in a replicable cloning vehicle in "proper reading phase" relative to one another as required in step (d) in the production plasmid. Except for the construction of the initial plasmid, pePA98.1, none of the steps used by BM during this project involved the ligation of an organically synthesized gene fragment to a cDNA first gene fragment to provide missing coding information at the amino terminus. (See Fischer Decl. II ¶ 19.)

As a fallback, Genentech also argues that BM's production plasmids carry a synthetic fragment from pePA98.1, and therefore meet claim one of the '832 patent. Genentech disputes BM's theory of noninfringement that the use of additional steps like SDM "breaks the chain" of infringement. This "break the chain" argument requires a return to the language of Section 271(g) and its legislative history. The undisputed facts in the record demonstrate that Reteplase has not been made by a patented process because the additional processing steps to develop the production plasmids were not themselves covered by the patent. While BM does marry cDNA and synthetic DNA to obtain tailored DNA having exactly the desired DNA sequence, it does not perform the marriage ceremony in the same way prescribed by step (c) of claim one. It needs a marriage broker—the cDNA enzymatically created via DNA polymerase—to tie the bonds of matrimony between the synthetically synthesized strand and the cDNA derived from reverse transcriptase. For the reasons described above, constructing synthetic primers for use in SDM and PCR so as to create an enzymatic reaction is substantially different from providing missing coding information by deploying cDNA and the synthetic DNA "in proper reading phase."

Genentech cites the well-established patent doctrine that performance of additional steps not found in the patent does not absolve an infringer from liability. See Amstar Corp. v. Envirotech Corp., 730 F.2d 1476, 1484 (Fed.Cir.), cert. denied, 469 U.S. 924, 105 S. Ct. 306, 83 L. Ed. 2d 240 (1984) (noting that "non-infringement is *112 shown when an element or step in the claims is missing from the accused product or process, not vice-versa"). Although Amstar governs a determination of whether the process for producing the expression plasmids violates claim one, this reliance is misplaced in the context of an analysis of whether pePA98.1 was materially changed under Section 271(g). The additional steps, which were not covered by the patent, did change the physical and chemical properties of the plasmid 98.1, and its expression product t-PA, in material ways. In Eli Lilly, 82 F.3d at 1573, the Federal Circuit stated:

In the chemical context, a "material" change in a compound is most naturally viewed as a significant change in the compound's structure and properties.

Under this definition, the production plasmids pa27fd and pA27-T2L (and the protein they produce) have been materially changed by intervening steps. See Attachment A. First, they delete the coding information for the F, K1 and E domain. Second, in contrast to t-PA, none of the amino acids is modified with glycosidic groups. Third, while Genentech points to a study in the New England Journal of Medicine indicating that Reteplase did not provide any additional survival benefit in the treatment of acute myocardial infarction, (see Genentech Supp.Mem. in Opp., Ex. 41), Reteplase does have a longer half-life and is easier to administer. Additionally, BM has its own patent on the Reteplase molecule, U.S. Patent No. 5,223,256 issued to Stern, et al., and Genentech's t-PA product patents were cited by BM as prior art in the course of prosecuting the '256 patent. Accordingly, there has been a significant change in both its structure and properties. See Genentech, Inc. v. Wellcome Foundation Ltd., 29 F.3d 1555, 1558 (Fed. Cir.1994) ("Wellcome").

Finally, there is no evidence in the record concerning the commercial viability phase of the congressional test. Genentech makes this argument in passing but the exhibit it cites does not appear to discuss this point. In any event, the point is so inadequately addressed it is waived.

For the foregoing reasons, BM's motion for summary judgment is ALLOWED on the '832 patent.

3. The '225 patent

a. The Claim

The '225 patent claims a method for increasing the solubility and stability of t-PA by incorporating argininium. It provides:

1. A method of increasing the solubility of tissue plasminogen activator in a pharmaceutical composition containing same as active principle comprising incorporating argininium ion in said composition, wherein said argininium ion is present in an amount effective to increase the solubility of said t-PA.

In the Markman decision, the Court construed the terms "tissue plasminogen activator" as used in Claim 1 of the '225 patent, to mean native human t-PA, whether produced from natural source extraction or recombinant cell culture systems described in the patent, as well as certain described biologically active human tissue plasminogen activator equivalents which (i) are capable of catalyzing the conversion of plasminogen to plasmin; (ii) bind to fibrin; and (iii) share basic immunological properties of native t-PA. Genentech, 989 F.Supp. at 367. The Court construed the term "pharmaceutical compositions" to mean those that are stable for appropriate periods of time, acceptable in their own right for administration to humans, and readily manufacturable. A pharmaceutical composition is comprised of purified t-PA. Id. at 368. Finally, the term "incorporating" was construed to refer to adding argininium to a pharmaceutical composition of t-PA. Id.

BM argues that it does not literally infringe the '225 patent because (1) it does not add arginine to a pharmaceutical composition to increase solubility and (2) Reteplase *113 is not a "Tissue Plasminogen Activator".

b. Adding Arginine

As background,[15] BM initially adds arginine at high concentrations to the refolding solution in order to keep the protein in its properly folded state and thereby improve the protein yield during refolding. This is before Reteplase is in the pharmaceutical composition stage. From this point onward in the process for making Reteplase, the Reteplase is constantly bathed in high concentrations of arginine. It is in all of the solutions used in the subsequent purification steps: filtration, affinity chromatography, cationic exchange chromatography, anion exchange chromatography, and diafiltration. After the anion exchange chromatography, pure Reteplase is in a solution containing arginine at a certain concentration. This is called the bulk drug substance solution. In this bulk drug substance solution, arginine, added in earlier processing and purification steps, is present in an amount sufficient to maintain the solubility of the Reteplase. Because the bulk drug substance solution contains Reteplase in a higher concentration than is desired in the final drug product solution, it is blended with an excipient solution which results in the pharmaceutical composition containing Reteplase at the concentration level required by final product specification. This blending step does not change the arginine concentration and the molarity of arginine remains constant. Dr. Fischer states:

Accordingly, BM does not add arginine to a pharmaceutical composition containing reteplase to increase the solubility of the reteplase. Instead, arginine and reteplase are together through many steps of BM's process for making reteplase and never separated, well before the preparation of any pharmaceutical composition occurs. The concentration of arginine used in said prior processing steps is always high enough to maintain the reteplase in solution and there is no subsequent addition of arginine to increase reteplase solubility because this is not needed.

(Fischer Decl. I ¶ 25.) Based on this declaration,[16] BM argues that it does not add arginine to an already-purified pharmaceutical composition in order to increase solubility, and thus does not literally infringe.

Genentech submits the declaration of Dr. Alexander M. Klibanov, a Professor of Chemistry at M.I.T., who reviewed the documents submitted by BM to the Food and Drug Administration (FDA), to refute Dr. Fischer's declaration.[17] He concluded that the "bulk drug substance" is a "pharmaceutical composition" as the Court construed that claim because it is (1) stable for appropriate periods of time; (2) acceptable in its own right for administration to humans; and (3) readily manufacturable. He describes the production and purification process after the initial addition of arginine to the refolding solution:

The properly refolded reteplase is next subjected by a number of procedures to produce a purified and sterilized form of reteplase. At no time during these purification and sterilization steps is any arginine removed. Rather, arginine is repeatedly replenished. Moreover, none of these later steps require any further or additional refolding of reteplase (BM 06.022), because such refolding has been fully completed earlier, and once refolded, reteplase will not require arginine to remain refolded. The purified and sterilized reteplase, termed the "Bulk Drug Substance," is then frozen and transported *114 to BM's facilities in Mannheim, Germany, for storage for a length of time, up to 24 months.

(Klibanov Decl. ¶ 7) (internal citations omitted) The frozen bulk drug substance is then thawed and lyophilized for storage and transport to hospitals. Immediately prior to the lyophilisation (or freeze-drying step), arginine is again added to the bulk drug substance. (See id. at ¶ 9.) According to an explanation that BM gave the FDA, it chose arginine as an excipient to be added at this point for the following reason: "The amount of arginine (0.5 mol/1) used in the formulation was chosen to ensure stability and solubility of Reteplase. Reduction in arginine content reduces the stability of such formulations and induces the risk of preciptation of Reteplase." (Id. at ¶ 10) (citations omitted) (emphasis in original) Thus, Dr. Klibanov concludes that BM repeatedly adds arginine to a pharmaceutical composition of Reteplase called the "Bulk Drug Substance" to increase the solubility of Reteplase.

Because Dr. Fischer's declaration conflicts with Dr. Klibanov's on the key issue of whether arginine is added to a pharmaceutical composition to increase solubility, there is a disputed issue of material fact which precludes summary judgment. BM attempts to finesse the conflict by stating that at no point is the arginine concentration increased in order to increase the solubility of Reteplase. However, the claim refers to the addition of an amount of argininium to increase the solubility of t-PA. Accordingly, if any amount of argininium is added and that result achieved, it fulfills the '225 claim.

C. Equivalency

BM's second argument that Reteplase is not a "biologically active human tissue plasminogen activator equivalent" is more complex. The '225 patent provides:

The terms likewise cover biologically active human tissue plasminogen activator equivalents, differing in one or more amino acid(s) in the overall sequence, or in glycosylation patterns, which are though [sic] to be dependent on the specific culture conditions used and the nature of the host from which the tissue plasminogen activator is obtained.

('225 patent, Col. 4, 11. 10-16.) At the Markman hearing, Dr. Klibanov testified: "This explains where those changes in one or more amino acids or in glycosylation patterns may stem from. They may stem from the differences in culture conditions and also from the differences in the nature of the host." (Markman Tr. 1-144, 11. 9-13.) Based on this snippet of testimony by Genentech's expert on a clause not in dispute at the hearing, BM argues that in order to be a tissue plasminogen activator equivalent, the differences in amino acids in the overall sequence, as well as the glycosylation patterns, must be dependent on the specific culture conditions or host cells. The record is undisputed. Reteplase lacks glycosylation because it is produced in E. Coli., (Fischer Decl. I ¶ 12), which is unable to generate the carbohydrate side chains that are attached to many mammalian proteins. However, there is no evidence in the record that Reteplase's amino acid sequence is dependent on either culture conditions or host cell.[18] Rather, it was designed by the SDM and PCR techniques.

Genentech argues that the clause, "which are though[t] to be dependent on the specific culture conditions used and the nature of the host from which the tissue plasminogen activator is obtained," refers only to the term "glycosylation patterns" and not to the phrase "one or more amino acid(s) in the overall sequence." *115 While this sentence is not a paragon of clarity, Genentech has the better grammatical argument because the closest plural noun in the sentence which this clause could modify is "glycosylation patterns." This grammatical construction is also consistent with intrinsic evidence which shows that Genentech's European Patent Application Publication No. 93619, incorporated by reference into the '225 patent, specifically describes such genetically engineered deletion derivatives:

The potential exists, in the use of recombinant DNA technology, for the preparation of various human tissue plasminogen activator derivatives, variously modified by resultant single or multiple amino acid substitutions, deletions, additions or replacements, for example, by means of site directed mutagenesis of the underlying DNA. Included would be the preparation of derivatives retaining the essential kringle region and serine protease region characteristic generally of the human tissue plasminogen activator described specifically herein, but otherwise modified as described above.

(Declaration of Dr. Désiré Collen, dated February 4, 1998, Ex. 4, EPO93619 at p. 9, 11. 19-27) (Docket 342) ("Collen Decl.") This patent also recognized that "the location of and degree of glycosylation will depend on the nature of the host cellular environment." (Id. at 11 15-17.)[19] Accordingly, there is nothing intrinsic in the patent which would support the strained reading pressed by BM that the differences in amino acid sequences may not be due to changes caused by site directed mutagenesis.

Since Reteplase is not native t-PA, BM argues that the undisputed evidence demonstrates that Reteplase is not a biologically human tissue plasminogen activator equivalent because it does not meet one of the functional criteria of fibrin binding. BM relies on the declaration of Victor Gurewich, M.D., Director of the Vascular Research Laboratory and Co-Director of the Institute for the Prevention of Cardiovascular Disease at Beth Israel Deaconess Medical Center, and Professor of Medicine at the Harvard Medical School.[20] Dr. Gurewich explains that t-PA binds "strongly" to the fibrin in a blood clot via the finger and growth factor domains (the former being the most important domain responsible for fibrin-binding) while the Kringle 2 domain appears to be the least significant domain with respect to fibrin binding. (See Gurewich Decl. ¶ 10.) Because Reteplase is missing both the finger and growth factor domains, it is missing the structural determinants responsible for fibrin binding, and Reteplase's "weak affinity for fibrin occurs instead via a lysine binding site in the kringle 2 domain." (Id. at ¶ 12.) While Reteplase binds only weakly to fibrin through a different domain, it nonetheless binds to fibrin as required in the functional analysis.

BM's reliance on Wellcome, 29 F.3d at 1568, is partially misplaced because in that case the issue was whether there was sufficient evidence for a jury to find under the three prong doctrine of equivalents that FE1X (which also binds to fibrin through the second Kringle region) was equivalent to natural t-PA. The Federal Circuit concluded:

Even assuming that the K2 region does play a role in the binding function of both, that hardly establishes that the two bind to fibrin in substantially the same way with substantially the same results, particularly in view of the overwhelming and undisputed evidence that the two possess dramatically different properties and structure. *116 Id. Among the different properties cited were the weaker fibrin binding affinity and half-life of FE1X as compared to natural t-PA. In the instant case, the Court is not applying the doctrine of equivalents but a claim in the patent to equivalents as defined in a certain way. I "welcome" BM's Wellcome argument that this definition is "hopelessly overbroad" another day, but the issue before me is literal infringement.

Wellcome does flag the question whether t-PA and Reteplase share basic immunological properties. Dr. Gurewich points out that Reteplase has a different mechanism for binding to fibrin, and differences in half-life and mode of administration. For example, t-PA has a half-life of three to six minutes and Reteplase has a half-life of 13-16 minutes. (See Gurewich Decl. ¶ 21.) Dr. Klibanov submitted a declaration that stated: "Reteplase shares the basic immunological characteristics of human t-PA based on its reactivity with polyclonal antibodies made against naturally occurring t-PA." (Klibanov Decl. ¶ 25) (footnote omitted) He proceeds to cite various reports by BM to the FDA. Unfortunately, the language in these reports is highly technical and not self-explanatory. Moreover, there is no intrinsic or extrinsic evidence defining the word "immunologic" in this context.[21] Therefore, the record is inadequate to determine whether Reteplase shares the basic immunological characteristics of human t-PA.

Therefore, I deny BM's motion for summary judgment because there are disputed issues of fact concerning whether arginine is added to a pharmaceutical composition and whether Reteplase shares basic immunological characteristics of human t-PA.

4. The '502 Patent

a. The Claim

The '502 patent claims a three-step process of purifying proteins. It claims a method for extracting desired proteins from refractile bodies, dissolving them, and separating these proteins from high molecular weight contaminants by use of a molecular sieve or high speed centrifugation. Claim 1 reads as follows:

1. A process for purifying heterologous proteins which are expressed and deposited in insoluble, refractile form in a host cell culture, which process includes the steps of:
(a) isolating said insoluble refractile heterologous protein from said host cell culture;
(b) dissolving the isolated refractile heterologous protein in a strong denaturing solution; followed by
(c) removing high molecular weight impurities using a molecular sieve or high speed centrifugation techniques.

('502 patent, Col. 32, 11. 10-22.) During the Markman hearing, the key contested term was "molecular sieve" as used in paragraph (c). Genentech contended that the term was used to describe various techniques for separation based on size, including a filter, gel electrophoresis or gel permeation chromatography (also known as gel filtration). BM contended that the term "molecular sieve" referred only to a gel permeation chromatography or gel filtration, and did not include its use of "dead-end filtration." The court construed the term "molecular sieve" as used in the '502 patent to mean "gel permeation or gel filtration techniques which separate molecules based on size."

Under that claim construction, Genentech concedes that BM does not literally infringe the '502 patent because BM uses a depth filter, not a gel, to remove high molecular weight impurities. Generally speaking, filtration is a way to separate materials by size; the larger ones with diameters greater than the pores in the filter are trapped, while the small ones pass through the pores in the filter. Specifically, *117 BM removes insoluble impurities by pouring the mixture through a porous membrane, i.e. a filter, so that the insoluble material remains on top of the filter.[22]

b. The Purification Processes

Losing the Markman skirmish, Genentech insists it nonetheless wins the war because it argues that BM infringes step (c) of claim one under the doctrine of equivalents.[23] It urges the Court to conclude that BM's use of a filter is equivalent to high speed centrifugation since both serve the same role of removing high molecular weight insoluble impurities from the desired protein.

In his direct testimony, Genentech's expert, Dr. Charles L. Clooney, provided background on the '502 patent.[24] He is a Professor of Chemical and Biochemical Engineering at MIT.[25] The '502 patent concerns a process for isolating and purifying heterologous proteins. "Heterologous" means they are exogenous or foreign to the host cell. Prior to the invention of the '502 patent, scientists had problems isolating and purifying the heterologous proteins that aggregated inside the cells in insoluble protein refractile bodies. He distinguished "filtration" and "centrifugation" as follows. "Filtration membranes allow proteins of a certain size to pass through them, while retaining larger proteins and impurities." (Clooney Dir. Test. ¶ 9.) By comparison he describes centrifugation as follows:

Centrifugation is a separation process accomplished by spinning the components to be separated. The pellet is the solid material collected after a centrifugation. The supernatant is the liquid part of a centrifuged product, as opposed to the pellet part. For example, in perhaps the simplest type of centrifugation, a mixture of insoluble material dispersed in a liquid can be separated into a pellet which packs into the bottom of the spinning centrifuge tube and the remaining liquid (the supernatant), that can be poured off leaving the pellet separated.

(Id. at ¶ 12.) The way in which high speed centrifugation removes high molecular weight impurities from the protein of interest is based largely on size.[26] (See Clooney Decl. ¶ 9.) The result of high speed centrifugation is a solution containing primarily the protein of interest, substantially free of high molecular weight impurities. The rate of removal by centrifugation is related to the size of the particle squared and the difference between its density and the density of the surrounding medium. (See id. at ¶ 9 n. 1.) Dr. Clooney states: "I would consider, as would any individual of at least ordinary skill, that a depth filter to be (sic) interchangeable with the high speed centrifugation of step (c) of the '502 patent and would not consider such a substitution to be a substantial change." (Id. at ¶ 11.)

In high speed centrifugation, a mixture is placed into containers and spun at a high number of revolutions per minute. *118 According to the '502 patent specification ('502 patent, Col. 14, ll. 15-18), in a "high speed centrifugation" the speed, in revolutions per minute, must be sufficient to generate a force of between 25,000 and 40,000 times the force of gravity. In this approach it is carried out by spinning the protein and recovering the supernatant for further purification.

The suspension pellet resulting from the spin-down "which contains undissolved and precipitated host protein and debris is discarded." ('502 patent, Col. 20, ll. 49-51.) Genentech argues that the purpose of step (c) of claim 1 when high speed centrifugation is used is to free the desired protein from high molecular weight impurities which are insoluble.

Dr. Stephan Fischer, the scientific director at BM, describes the process for the production and purification of Reteplase as follows.[27] In the initial production steps, E. Coli bacterial cells containing the production plasmid for Reteplase are grown under conditions which favor the synthesis of Reteplase. Reteplase accumulates in the cell in the form of insoluble aggregates known as refractile or inclusion bodies. After harvesting, the bacterial cells are broken open and the inclusion bodies are isolated by low speed centrifugation at about 5,000 times the force of gravity and purified. These inclusion bodies are then solubilized, but not all of the inclusion body protein goes into solution. Therefore, the mixture is subjected to "dead end" filtration, whereby insoluble material is removed. In this procedure, the mixture is passed through a porous filter, so that all the insoluble material, which includes aggregates of molecules of both higher and lower molecular weight than Reteplase, remains on the top of the filter. All the material in solution (irrespective of its molecular weight) passes through the filter. After several steps, the solution is concentrated, and then diluted into a renaturation buffer where renaturation of Reteplase occurs. Following renaturation, the Reteplase is further purified by dead end filtration to remove any remaining insoluble material. After a series of additional purification steps—affinity chromatography, cation exchange chromatography, and anion exchange chromatography—pure Reteplase is obtained.

c. Equivalency

Under the doctrine of equivalents, a product that does not literally infringe the express terms of a patent claim nonetheless may be found to infringe if there is "equivalence" between the elements of the accused device and the claimed elements of the patented invention. See Warner-Jenkinson, 520 U.S. at 21, 117 S. Ct. 1040. The Supreme Court has stated that the inquiry in determining equivalence is as follows:

An analysis of the role played by each element in the context of the specific patent claim will thus inform the inquiry as to whether a substitute element matches the function, way, and result of the claimed element, or whether the substitute element plays a role substantially different from the claimed element.

Id. at 40, 117 S. Ct. 1040. Moreover, "[t]he known interchangeability of substitutes for an element of a patent is one of the express objective factors ... bearing upon whether the accused device is substantially the same as the patented invention." Id. at 36, 117 S. Ct. 1040. "[T]he proper time for evaluating equivalency—and thus knowledge of interchangeability between elements—is at the time of infringement, not at the time the patent was issued." Id. *119 at 37, 117 S. Ct. 1040. Where there is a disputed fact question on equivalency, it is "for the jury to decide whether the accused process was equivalent to the claimed process." Id. at 38, 117 S. Ct. 1040.

To determine equivalence, courts apply the "insubstantial differences" test. See Dawn Equip. Co. v. Kentucky Farms Inc., 140 F.3d 1009, 1015 (Fed.Cir.1998). Under that test, there is infringement by equivalents if the differences between the accused device and the asserted claim limitation are "insubstantial." See id. As the Supreme Court has acknowledged, however, the "insubstantial differences" test can be difficult to apply. See Warner-Jenkinson, 520 U.S. at 40, 117 S. Ct. 1040. Consequently, courts frequently return to the "function-way-result" test, which considers whether the elements of the accused device perform "substantially the same function, in substantially the same way, to achieve substantially the same result" as the limitation at issue in the asserted claim. Dawn Equip. Co., 140 F.3d at 1016; see also Eastman Kodak Co. v. Goodyear Tire & Rubber Co., 114 F.3d 1547, 1560 (Fed.Cir.1997).

Genentech argues that BM's use of a depth filter is equivalent to high speed centrifugation under the traditional function-way-result test. BM argues that use of depth filters does not meet this equivalency test because they do not operate in the same way as centrifugation. BM's argument is more persuasive. In the most obvious sense, centrifugation operates by spinning a solution and filtration operates by pouring a solution through a membrane. Moreover, they separate insoluble particles according to different principles. Dead-end filtration separates molecules primarily based on whether they are larger in dimension than the filter's pore size (not according to molecular weight) and solubility. To be sure, as BM's expert Dr. Webb stated: "For most proteins, there is a direct relationship between size and molecular weight (the sum of the weights of all the atoms in a molecule)." (Declaration of Dr. Andrew C. Webb, dated May 29, 1997, ¶ 6) (Docket 221) For this reason, the filtration removes high molecular weight impurities, although it accomplishes this result by separating impurities based on size. Centrifugation differentiates insoluble particles based on sedimentation rates of particles with different masses or densities. Thus, while they both play a substantially similar role—"remov[al of] high molecular weight components after the dissolution of the inclusion bodies"[28]—they do so in a substantially dissimilar way.

The "interchangeability" of the filter with high speed centrifugation is evidence of equivalence under the doctrine of equivalents but it is not dispositive. See Unidynamics Corp. v. Automatic Products Int'l, Ltd., 157 F.3d 1311, 1322 (Fed.Cir.1998) (holding that, at bottom, "[t]he doctrine of equivalents question involves whether the accused device is substantially different than the claimed device").

The Court's determination that the filtration and high speed centrifugation processes are not equivalent finds significant support in a recent Federal Circuit case. In Insituform Technologies, Inc. v. Cat Contracting, Inc., 161 F.3d 688, 693-95 (Fed.Cir.1998), cert. denied ___ U.S. ___, 119 S. Ct. 1254, 143 L. Ed. 2d 350 (1999), the Federal Circuit reversed a district court's ruling that two vacuum processes, one involving a cup and one involving a needle, were equivalent. The lower court held that the processes "do the same thing (suck a vacuum) by the same means (being connected to a vacuum source) to achieve the same result (produce a satisfactory vacuum ...)." Id. at 693. The Federal Circuit ruled, however, "that the district court's function-way-result analysis involves too much overlapping and is overly *120 broad." Id. By defining the "way" aspect of the test in such a broad manner ("being connected to a vacuum source"), the district court essentially read that requirement out of the test. The Federal Circuit held that such an analysis constitutes reversible error: "[o]nce the `way' is correctly defined and the structural differences [between the cup and needle processes] are properly considered, no reasonable trier of fact could have found the claimed single cup process and the accused multiple needle process to be equivalent." Id. at 694.

Thus, just as it was error to ignore the "structural differences" between the cup and needle processes, it would be error for this Court to ignore the manifest differences between the filtration and centrifugation processes. BM's motion for summary judgment is accordingly ALLOWED on the '502 patent.

ORDER

BM'S motion for summary judgment (Docket 95) is ALLOWED with respect to the claim of infringement on the '832 and '502 patent but DENIED with respect to the '225 patent. Genentech's motion for summary judgment of infringement of the '502 patent (Docket No. 192) is DENIED.

*121

*122

*123

*124

NOTES

[1] See Report of Connie Cepko, Professor of Genetics and Associate Investigator of the Howard Hughes Medical Institute at Harvard Medical School, dated July 14, 1997 (Docket 262) ("Cepko Report").

[2] See Tutorial Declaration of Dr. Andrew C. Webb, dated June 3, 1997, ¶ 32 (Docket 224) ("Webb Tutorial").

[3] See Direct Testimony of Dr. Jeffrey V. Ravetch, dated June 3, 1997 (Docket 226) ("Ravetch Dir. Test.").

[4] Dr. Fischer received his Ph.D. from the Technical University of Munich. He also has held post doctoral fellowships. See Declaration of Dr. Stephan Fischer, dated April 12, 1996 (Docket 53) ("Fischer Decl. II").

[5] See Webb Tutorial ¶¶ 36-42, Supplemental Declaration of Dr. Andrew C. Webb, dated March 6, 1998, ¶¶ 7-19 ("Supp.Webb.Decl."), and Declaration of Dr. Jeffrey V. Ravetch dated March 6, 1998, ¶¶ 33-43, 48-50 together with illustrative exhibits ("Ravetch Decl.").

[6] See Declaration of Dr. Martin Tracey, dated March 5, 1998, ¶ 19.

[7] The description of the development of the BM plasmids is drawn from Supp. Webb. Decl. and the illustrations thereto, Fischer Decl. II, Ravetch Decl., and the Videotaped Presentation of Dr. Jeffrey V. Ravetch, dated February 24, 1998.

[10] In its entirety, 35 U.S.C. § 271(g) provides: Whoever without authority imports into the United States or offers to sell, sells, or uses within the United States a product which is made by a process patented in the United States shall be liable as an infringer, if the importation, offer to sell, sale or use of the product occurs during the term of such process patent. In an action for infringement of a process patent, no remedy may be granted for infringement on account of the noncommercial use or retail sale of a product unless there is no adequate remedy under this title for infringement on account of the importation or other use, offer to sell or sale of that product. A product which is made by a patented process will, for purposes of this title, not be considered to be so made after (1) it is materially changed by subsequent processes; or (2) it becomes a trivial and nonessential component of another product.

[11] At any rate, the issue of burden allocation does not alter the outcome in the instant litigation. BM's motion for summary judgment succeeds even if it bears the burden.

[12] See also Tutorial Declaration of Dr. Joseph Oliver Falkinham III, dated June 16, 1997, p. 3 (Docket 227).

[13] Dr. Cepko volunteered an opinion regarding literal infringement although her task was limited to claim construction under Markman. Although a legal conclusion regarding infringement was beyond the scope of claim construction, she added these helpful observations about literal infringement:

The other aspect of the literal interpretation of the '832 patent given above that should be addressed concerns site directed mutagenesis (SDM) and the polymerase chain reaction (PCR). These methods were not in use in 1979 and most people skilled in the art were not familiar with them. Both methods employ organically-synthesized gene fragments. However, there is no evidence that Genentech foresaw or in any way covered these methods in the '832 patent. Other patents have been granted for these methods and the Nobel Prize was awarded for them (to Mullis and Smith). SDM and PCR were true innovations and have been very powerful for many applications, including the application debated here of making an altered gene. Although organically synthesized gene fragments used in PCR and SDM are joined to the 5' end of a cDNA, as covered literally in the '832 patent, there is no evidence that Genentech foresaw this application, nor was it in use in the art in 1979.

Cepko Report at pp. 3-4.

[14] Genentech does not press an argument that PCR or SDM is within the scope of the patent although they both involve the use of synthetic primers to create cDNA enzymatically. (See Supp. Webb Decl. ¶¶ 38-42 and accompanying illustrations 12 and 13.) Although Genentech refers to SDM in the patent ('832 patent, Col. 6, 11. 33-36), it does so in the context of creating restriction sites, not performing the steps of the patent.

[15] The background is taken from Declaration of Dr. Stephan Fischer, dated May 12, 1996 (Docket 54) ("Fischer Decl. I").

[16] Dr. Cepko also concluded that "arginine was not added at the point when the pharmaceutical composition was being formulated, but at an earlier step ..." (Cepko Report at p. 8.)

[17] See Declaration of Dr. Alexander M. Klibanov, dated January 20, 1998 (Docket 327) ("Klibanov Decl.").

[18] Dr. Ravetch states: "The definition of a t-PA equivalent in the '225 patent does not mean that culture conditions or particular hosts themselves cause changes in the amino acid structure of t-PA. Such a definition is not scientifically correct in a general sense." Declaration of Dr. Jeffrey V. Ravetch, dated February 5, 1998, ¶ 6 (Docket 339).

[19] Because I was able to construe the claim based on the intrinsic language of the patent, I need not explore the extrinsic evidence concerning skill of those in the art in December, 1985. (See Collen Decl. ¶ 20.)

[20] See Declaration of Dr. Victor Gurewich, dated May 12, 1997 (Docket 222) ("Gurewich Decl").

[21] Indeed, this was not a disputed term at the Markman hearing although BM pressed its argument that "Reteplase does not cross-react in the same fashion as t-PA." (Markman Tr. 3-63.)

[22] See Cepko Report at p. 5.

[23] There is no dispute that BM literally practices steps (a) and (b) of the '502 claim 1.

[24] See Direct Testimony of Charles L. Clooney, Ph.D., dated June 3, 1997 (Docket 226) ("Clooney Dir. Test.").

[25] See Declaration of Charles L. Clooney, Ph. D., dated February 5, 1998, ¶ 2 (Docket 341) ("Clooney Decl.").

[26] The following description is helpful: "The first step in a typical protein-purification scheme is centrifugation. The principle behind centrifugation is that two particles in suspension (cells, organelles, or molecules) having different masses or densities will settle to the bottom of a tube at different rates. Remember, mass is the weight of a sample (measured in grams), whereas density is the ratio of its weight to volume (grams/liter). Proteins vary greatly in mass but not in density." H. Lodish, et al, Molecular Cell Biology 88 (3d Ed.1995). In differential centrifugation which is used for the partial purification of proteins by separating soluble material from insoluble material, particles settle according to mass. See id. at 90-91.

[27] The Declaration of Dr. Stephan Fischer, dated May 2, 1997, ¶ 14 (Docket 99) states: "In dead end filtration molecules in solution are separated from those which are not in solution by passage through a filter. The insoluble material remains on top of the filter, while the material in solution passes through. Rather than separating on the basis of molecular weight, this purification procedure separates the desirable from the undesirable material on the basis of solubility."

[28] Deposition Transcript of Dr. Stephan Fischer, dated February 26, 1997, at p. 128, ll. 17-19 (Docket 340, Attachment D).