Diversa Corporation v Maxygen, Inc - [2005] APO 43

ABSTRACTS OF DECISIONS

DECISION OF A DELEGATE OF THE COMMISSIONER OF PATENTS

Application : No. 703264 in the name of Maxygen, Inc.

Title: DNA Mutagenesis by Random Fragmentation and Reassembly

Action: Opposition under section 59 of the Patents Act 1990 by Diversa Corporation

Decision: Issued 05 October 2005.
Abstract

The invention relates to an in vitro method for obtaining mutated polynucleotides through a process of random fragmentation and reassembly of a double-stranded polynucleotide template or templates. In the method random fragments are denatured and allowed to reanneal, leading to the formation of new combinations and pairings of single-stranded fragments with overhanging ends. A polymerase is used to fill the overhangs and the process of denaturing, annealing and extension is repeated at least two further times. In a final step mutated polynucleotide products of the recombination and reassembly cycles are screened for polynucleotides having desired functional properties.

The opposition was successful on only one ground that claims 22 to 24 lacked clarity.

Costs were awarded against Maxygen up to the date on which the claims were amended. Post amendment, costs were awarded against Diversa Corporation.

PATENTS ACT 1990

DECISION OF A DELEGATE OF THE COMMISSIONER OF PATENTS

Re:Patent Application No. 703264 by Maxygen, Inc. and opposition under section 59 of the Patents Act 1990 by Diversa Corporation.

BACKGROUND

Patent application 703264 (29714/95) in the name of Maxygen, Inc (hereafter referred to as Maxygen) was filed as a convention application on 17 February 1995 claiming priority from US 08/198431, which was filed on 17 February 1994. The application was advertised as accepted on 25 March 1999. Diversa Corporation (hereafter referred to as Diversa) served a notice of opposition on 25 June 1999 and served their statement of grounds and particulars on 23 September 1999. Service of evidence in support was completed on 25 January 2001.
Amendments to the description and claims were filed on 18 May 2001 and 31 July 2001 and leave to amend was granted on 9 October 2001. Evidence in answer was deferred until leave to amend was granted and was completed on 20 September 2002. Further amendments to the specification and claims were then filed on 3 October 2002 and allowed on 30 April 2003.
Evidence in reply was completed on 16 November 2003.
Further evidence was then filed by Maxygen on 11 August 2004 and evidence in answer to this further evidence was filed by Diversa on 15 February 2005.
A second amendment to the statement of grounds and particulars was made on 12 November 2005 to introduce the ground of inutility and the new documents exhibited in Diversa’s evidence filed on 15 February 2005. Maxygen filed evidence in response to this evidence and the new ground on 13 May 2005.
The specification that was the subject of the hearing was the specification as accepted, including the amendments of 18 May 2001, 13 July 2001 and 3 October 2002.
The matter was heard in Canberra on Monday 1 August 2005 to Wednesday 3 August 2005. Diversa was represented by Mr James Cherry and Ms Sarah Couper of Freehills Patent Attorneys, Melbourne. Maxygen was represented by Ms Katrina Howard of Counsel, assisted by Dr Andrew Blattman and Dr Martin O’Brien of Spruson and Ferguson, Sydney and Ms Sharon Fujita and Dr Lori Giver of Maxygen.

THE EVIDENCE

Diversa’s evidence consists of three declarations made by Gerald Joyce, two by Anthony Weiss and one each by Mae Lin Gan, Ruth Bird and Ngaire Pettit-Young, together with various exhibits.
Maxygen’s evidence consists of three declarations made by David Kemp, three by Svante Pääbo, two by Charles Yanofsky, two by Steffan Ho, and one each by Jan Moller Mikkelsen, John Francis McCann and Andrew Nathaniel Blattman, together with various exhibits.
At the hearing Mr Cherry submitted that only part of Maxygen’s evidence of 11 and 13 May 2005 should be taken into account. Mr Cherry argued that Maxygen had only been granted permission to respond to Diversa’s evidence on utility and that Pääbo’s and Yanofsky’s final declarations and paragraphs 4 and 5 of Kemp’s final declaration were not strictly in response to utility.
After some discussion Maxygen conceded that their case might not suffer unduly if the evidence did not form part of the hearing proceedings because it did not raise new issues. Diversa also conceded that they had already considered the evidence during their preparation for the hearing and could provide submissions on these at the hearing if pressed.
Based on these admissions both parties agreed that the evidence would not be considered during the hearing. If it later became clear that the evidence was likely to have an impact on the outcome of the hearing Diversa would then have the opportunity to respond to the evidence in writing.
As will be apparent from the following decision, I have not needed to refer to this evidence when making my decision.

THE SPECIFICATION

The specification describes an in vitro method for obtaining mutated polynucleotides through a process of random fragmentation and reassembly of a double-stranded polynucleotide template or templates. The application of this technology to DNA has become known as DNA shuffling, or the “Stemmer technology”, in recognition of the contribution of Dr William Stemmer, one of the co-inventors of 703264.
The method described in the specification relies on the following basic steps.
a)A population of double-stranded polynucleotides is randomly cleaved to generate overlapping double-stranded fragments.
b)The fragments are denatured and annealed.
c)New combinations and pairings with overhanging ends are created as single-stranded random fragments anneal in areas of homology.
d)A polymerase is used to fill overhanging ends in paired fragments, with one strand acting as the template and the other as the primer.
e)Steps (b) to (d) are repeated multiple times, with an associated increase in the length of recombined fragments in each new cycle.
f)Products of (e) are screened or selected to identify recombined polynucleotides with desired functional properties.
When polynucleotides share regions of homology and heterology with respect to each other recombination will occur between related sequences as single-stranded fragments generated in step (a) anneal at regions of homology in step (c) and use heterologous sequence as a template during the polymerase extension in step (d).
The specification explains that at the priority date of the application the preferred methods of in vitro mutagenesis were oligonucleotide-directed mutagenesis and error prone polymerase chain reaction (PCR). However, neither of these methods is well suited to producing extensive or random mutations, or for readily generating large libraries of random mutants. Oligonucleotide-directed mutagenesis only introduces mutations into defined and predetermined areas of the template polynucleotide and error prone PCR only introduces mutations at a low rate.
At the priority date in vivo recombination was also used to introduce random mutations into a polynucleotide sequence. However, in vivo recombination systems are not random and introduce mutations at a relatively low rate because they rely on the presence of specific recombination sites within target sequences.
The specification also explains that at the priority date in vitro recombination during standard PCR was known, particularly where template DNA has been damaged by nicking or breaking. The specification does not deal with this aspect of the prior art in any detail, even though it represents some of the closest prior art and its similarity to the applicant’s technology forms much of the applicant’s case.
Although the specification does not provide an extensive comparison of standard PCR and the applicant’s methods it does highlight one specific difference between the invention and standard PCR. This is that there is no exponential amplification of the template in the Stemmer technology, as there is in standard PCR.
The specification also makes it clear that PCR primers are not required in the Stemmer technology. Example 1 explains that PCR primers were deliberately removed from the double-stranded polynucleotide template prior to the random cleavage step. Page 63 states:
“The removal of the fee primers was found to be important.”
In the absence of PCR primers, one member of a fragment pair acts as the primer in the polymerase extension reaction. At page 24 the specification explains that although the removal of PCR primers before fragmentation is not essential, removal of PCR primers results in a far greater degree and diversity of recombination.
When discussing this issue in his declaration, Maxygen’s expert Pääbo explained that the presence of an excess of PCR primers, as normally found in standard PCR reactions, shifts the equilibrium toward recombined fragments initiated by PCR primer extension. With a reduced likelihood of random-fragment primed extension there is an associated reduction in the range and number of mutant fragments produced by the steps outlined in the applicant’s method.
The specification contains a number of examples describing the use of DNA shuffling to mutagenise bacterial sequences and to generate chimeric and mutagenised cytokine and antibody sequences.
The specification has 92 claims. Claims 1, 9, 14, 22 and 25-29 are independent claims.
Claim 1 recites:

A method for forming a mutagenized double-stranded polynucleotide from a template double-stranded polynucleotide, comprising
(a)         treating a sample comprising a template double-stranded polynucleotide under conditions which provide for the cleavage of the template polynucleotide into random overlapping double-stranded fragments of the template double-stranded polynucleotide, and providing one or more single or double-stranded oligonucleotides, wherein said oligonucleotides comprise an area of homology and an area of heterology to the template double-stranded polynucleotide;
(b)         denaturing the resultant mixture of double-stranded overlapping fragments and oligonucleotide into single-stranded fragments;
(c)         incubating the resultant population of single-stranded fragments with a polymerase under conditions which result in the annealing of said single-stranded fragments at said areas of homology to form pairs of annealed fragments, said areas of homology being sufficient for one member of a pair to primer(sic) replication of the other thereby forming a mutagenized double-stranded polynucleotide;
(d)         repeating steps (b) and (c) for at least two further cycles, wherein the resultant mixture in step (b) of a further cycle includes the mutagenized double-stranded polynucleotides from step (c) of the previous cycle and the further cycle forms further mutagenized double-stranded polynucleotides whereby the average length of the mutagenized polynucleotides increases in each cycle; and
(e)         screening or selecting the further mutagenized double-stranded polynucleotides and thereby identifying a further mutagenized double-stranded polynucleotide having a desired functional property.
Claims 2-8 are dependent on claim 1. These claims further define the numbers, types, relative amounts and sizes of the double-stranded fragments, the number of repeats of steps (b) and (c) and a further method step comprising expression of the mutagenized polynucleotide.
Claim 9 recites:

A method of obtaining a chimeric polypeptide sequence having a desired functional property comprising:
(a)         treating a sample comprising different double-stranded template polynucleotides wherein said different template polynucleotides contain areas of homology and areas of heterology under conditions which provide for the cleavage of said template polynucleotides into random overlapping double-stranded fragments of a desired size;
(b)         denaturing the resultant overlapping double-stranded fragments of said different double-stranded template polynucleotides contained in the treated sample produced during step (a) into single-stranded fragments;
(c)         incubating the resultant single-stranded fragments with polymerase under conditions which provide for the annealing of the single-stranded fragments at the areas of homology to form pairs of annealed fragments, said areas of homology being sufficient for one member of a pair to primer (sic) replication of the other thereby forming a chimeric double-stranded polynucleotide sequence comprising template polynucleotide sequences;
(d)         repeating steps (b) and (c) for at least two further cycles, wherein the resultant mixture in step (b) of a cycle includes the chimeric double-stranded polynucleotides from step (c) of the previous cycle and the further cycle forms further chimeric polynucleotide sequences whereby the average length of the chimeric polynucleotide sequences increases in each cycle; and
(e)         screening or selecting the further chimeric polynucleotide sequence and thereby identifying a further mutagenized double-stranded polynucleotide having a desired functional property.
Claims 10 to 13 are dependent on claim 9. These claims further define the relative amounts and sizes of the template fragments.
Claim 14 recites:

A method for selecting or screening a library of displayed peptides or displayed antibodies for a desired functional property, the method comprising
(i)          obtaining a first plurality of selected library members comprising a displayed peptide or displayed antibody and an associated polynucleotide encoding said displayed peptide or displayed antibody and obtaining associated polynucleotides of said library members or copies thereof wherein said associated polynucleotides comprise a region of substantially identical sequence, and
(ii)         pooling and randomly fragmenting said associated polynucleotides or copies to form a population of double-stranded fragments;
(iii)        denaturing the population of double-stranded fragments into single-stranded fragments;
(iv)        incubating the resultant population of single-stranded fragments with a polymerase under conditions which result in the annealing of said single-stranded fragments at areas of overlap to form pairs of annealed fragments, whereby one member of a pair of primes replication of the other thereby forming recombinant polynucleotides;
(v)         repeating steps (iii) and (iv) for at least two further cycles;
wherein the average length of polynucleotides increases in each cycle to form a shuffled pool of recombined polynucleotides and
whereby a substantial fraction of the recombined polynucleotides of said shuffled pool are not present in the associated polynucleotides in step (i);
(vi)        expressing the recombinant polynucleotide to provide said library of displayed peptides or displayed antibodies; and
(vii)       selecting or screening said library of displayed peptides or displayed antibodies and thereby identifying a displayed peptide or displayed antibody having the desired functional property.
Claims 15 to 21 are dependent on claim 14. These claims recite using PCR and in vivo recombination to introduce additional mutations and screening the products of the method to identify mutagenized peptides with specific phenotypic characteristics.
Claim 22 recites:

A method for generating libraries of displayed antibodies suitable for affinity interaction screening,
the method comprising
(i)          obtaining a first plurality of selected library members comprising a displayed antibody and as(sic) associated polynucleotide encoding said displayed antibody, and obtaining said associated polynucleotides or copies thereof, wherein said associated polynucleotides comprise a region of substantially identical variable region framework sequence, and
(ii)         pooling and randomly fragmenting said associated polynucleotides or copies to from fragments thereof under conditions suitable for PCR amplification, performing PCR amplification, and
(iii)        thereby homologously recombining said fragments to form a shuffled pool of recombined polynucleotides comprising novel combinations of CDRs whereby a substantial fraction of the recombined polynucleotides of said shuffled pool comprise CRR combinations that are not present in the first plurality of selected library members
screening or selecting polynculeotides of said shuffled pool and thereby identifying a polynucleotide of said shuffled pool having a desired functional property.
Claims 23 and 24 are dependent on claim 22. These claims define an additional steps of affinity screening and of further cycles of shuffling following screening.
Claims 25 to 28 are omnibus claims with each claim reciting methods substantially as described in the examples. As discussed previously, each of the examples describes classical DNA shuffling involving random fragmentation and reassembly of double stranded DNA templates.
Claim 25 recites:

A method (sic) forming a mutagenized double-stranding polynucleotide from a template double-stranded polynucleotide from a template double-stranded polynucleotide substantially as herein described with reference to the Examples.
Claim 26 recites:

A method for obtaining chimeric polynucleotide sequence, substantially as herein described with reference to the Examples.
Claim 27 recites:

A method for generating a library of displayed peptides or displayed antibodies suitable for affinity interaction screening or phenotypic screening, substantially as herein described with reference to the Examples.
Claim 28 recites:

A method for generating libraries of displayed antibodies suitable for affinity interaction screening, substantially as herein described with reference to the Examples.
Claim 29 recites:

A method of recombining template polynucleotides to provide a chimeric polynucleotide having a desired function property, comprising
randomly cleaving template double-stranded polynucleotides to produce random double-stranded overlapping fragments of desired size:
subjecting the double-stranded overlapping fragments of a desired size to at least three cycles of denaturation, annealing and incubating in the presence of a polymerase, wherein
the denaturing steps denature double-stranded fragments in pairs; and
the incubation step forms chimeric double-stranded polynucleotides comprising template polynucleotide sequences by one member of an annealed pair of fragments priming replication of the other,
wherein the average length of chimeric polynucleotides increases in each cycle of the method; and
screening or selecting the chimeric double-stranded polynucleotides thereby identifying the chimeric polynucleotide having the desired functional property.
Claims 30 to 92 are dependent on claims 1, 9 and 29. They add features relating to the method of random fragmentation, the origin of the template polynucleotides, additional mutagenesis steps and screening for specific phenotypic properties.
The claims can be divided into three general applications of the Stemmer technology. The first application corresponds to the most widely recognised representation of the Stemmer technology in which shuffling is applied to mixed templates comprising regions of homology and regions of heterology with respect to each other. This application is defined in independent claims 9, 14 and 26-29 in full and 25 in part, and I will refer to it as “mixed shuffling”.
The second application corresponds to the earliest realisation of the Stemmer technology in which a single species of template is used. Mutations are introduced when oligonucleotides comprising an area of homology and an area of heterology with respect to the template are added. This application is defined in claim 1 in full and 25 in part, and I will refer to it as “spiked shuffling”.
The remaining application involves mixed antibody templates. This application is defined in claim 22 and I will refer to it as “PCR shuffling”.

PRIORITY

Diversa raised a number of issues to challenge the priority date of 703264. If the application is not entitled to its earliest priority date then a number of documents published between the priority date and the date of filing of the complete specification will also be relevant to the novelty and inventive step of the claims.
A first issue was that the priority document specifies annealing in regions of identity whereas the claims expand the requirement to one of homology.
However, in its discussion of homology at page 9 the priority document states:
“The term “homologous” or “homeologous” means that one single-stranded nucleic acid sequence may hybridize to a complementary single-stranded nucleic acid sequence. The degree of hybridization may depend on a number of factors including the amount of identity between the sequences and the hybridization conditions such as temperature and salt concentrations as discussed later. Preferably the region of identity is greater than about 5 bp, more preferably the region of identity is greater than 10 bp.”
In a later discussion of hybridization conditions and the minimum homology required for annealing and recombination the priority document also states:
“If a high frequency of crossovers is needed based on an average of only 4 consecutive bases of homology, recombination may be forced by using a low annealing temperature, although the process becomes more difficult.” (page 15)
As such, the priority document does not teach that fragments must share absolute identity. In contrast it teaches that fragments require a particular level of identity within homologous regions, and that the minimum level is 4 consecutive identical bases. This level is consistent with the definition of “homology” as understood in the art, particularly when the term is used to specify the level of homology required for annealing and polymerase extension. It is also consistent with “homology” as defined in the claims, where it is clearly stated that the level of homology must be sufficient for the fragments to anneal and then “for one member of a pair to primer (sic) replication of the other”.
Both parties’ experts also support this understanding of the boundaries of the term “homology”. Kemp explains in his second declaration that homology does not necessarily require absolute identity but rather is a matter of degree and Joyce in his third declaration acknowledges that “in some contexts, “homologous” does mean identical”.
Diversa also submitted that the claims defined a different invention to that described in the priority document because the priority document stated that the invention was a method of introducing mutations and the claims defined a method of identifying a mutated polynucleotide having a desired functional property. However, the priority document describes the invention as methods of shuffling and selection which allows for the creation of mutant polynucleotides and proteins having enhanced biological activity. As such, the priority document explicitly describes an invention that includes steps of identifying a mutated polynucleotioes having a desired functional property.
With all of the priority issues Diversa appears to be suggesting that the disclosure in the priority document should mirror the disclosure in the complete specification, even to the extent of using identical wording and providing identical claims. However, all that is required of the priority document is that it provides a real and reasonably clear disclosure of the features of the invention as claimed in the complete specification.
I am satisfied that the priority document meets this requirement and that, as such, each of the claims has as its priority date the date of filing of the basic application US 08/198431.

NOVELTY

The basic test for novelty is the “reverse infringement” test as stated in General Tire & Rubber Co v Firestone Tyre & Rubber Co Ltd, (1972) RPC 457 at pages 485, 486:
“If carrying out the directions contained in the prior inventor's publication will inevitably result in something being made or done which, if the patentee's patent were valid, would constitute an infringement of the patentee's claim”.
However, in applying this test regard must be given to whether the prior art publication has provided clear and unmistakable directions to do what patentee has claimed, as also stated in General Tire & Rubber Co v Firestone Tyre & Rubber Co Ltd at page 486:
"To anticipate the patentees claim, the prior publication must contain clear and unmistakable directions to do what the patentee claims to have invented. A signpost, however clear, upon the road to the patentee's invention will not suffice. The prior inventor must be clearly shown to have planted his flag at the precise destination before the patentee."
Diversa cited over 100 prior art documents against the novelty of the claims. As is to be expected with this number of citations many of the citations are repetitive and add little to the opposition proceedings. Although it might be argued that a larger number of the citations may have been relevant at the start of the opposition proceedings, prior to amendment of the claims, it is clear that most of the citations are no longer relevant.
The limited relevance of most of the cited prior art is also reflected in Diversa’s submissions on novelty and inventive step where they have restricted their comments to only a handful of the prior art documents cited in their original statement of grounds and particulars.
It is worth noting at this point that Diversa could have substantially reduced the complexity of these proceedings if they had identified those citations that they regarded as most relevant and that they intended to rely on at the hearing, prior to hearing. However, they did not do this. As a consequence, Maxygen was forced to prepare submissions on an extensive range of citations, most of which were not discussed at the hearing.
In my decision I have only referred to those documents that I consider most relevant to the invention, and I have divided these into four groups corresponding to the following subject matter: Jumping PCR, Grafting and Splice Overlap Extension, Random Pairing and Shuffling of Immunoglobulin Chains and In vivo Recombination.
Before discussing the citations it is worth first identifying the key principles underpinning the claimed methods. These features are:
a)Double-stranded polynucleotide templates are randomly cleaved prior to a first cycle of denaturation, extension and annealing.
b)Random template fragments prime polymerase extension.
c)The average length of extension products increases in each cycle.
d)The method is used to generate mutated polynucleotides having desired functional properties.

Jumping PCR Citations

The most relevant PCR citations are as follows:
Pääbo et al (1990) The Journal of Biological Chemistry 265(8), 4718-4721
Pääbo et al (1989) The Journal of Biological Chemistry 264(17), 9709-9712
Marton et al (1991) Nucleic Acids Research 19(9), 2423-2426
Meyerhans et al (1990) Nucleic Acids Research 18(7), 1687-1691
Horton and Pease in “Directed Mutagenesis: A Practical Approach” (1991), pages 217-247.
Saiki et al (1988) Science 239, 487-491
Shuldiner et al (1989) Nucleic Acids Research, 4409

Mixed shuffling claims

As explained previously, mixed shuffling represents the most well known application of the Stemmer technology, in which shuffling is applied to a starting population of mixed templates comprising regions of homology and heterology with respect to each other.
All of the jumping citations relate to a form of recombination that routinely occurs in the background of standard PCR reactions. This form of recombination has become known as “jumping PCR”, a name coined to describe the recombination that routinely occurs between damaged templates in PCR reactions. As such all of the citations share the feature that they are describing methods based on standard PCR in which an excess of flanking PCR primers are used to prime polymerase extension.
The mechanism of jumping PCR is explained in the following diagram taken from Pääbo (1989).

Figure 3 Concept of “jumping PCR.: Primers A and B use undamaged parts of five damaged templates to amplify a mosaic product. The primers are extended during the first PCR cycle up to points where ends of fragments (◊) cause the polymerase to stop. During subsequent cycles, these extended primers can anneal to other template fragments and be further extended. After a sufficient number of cycles the two primers have grown so long that their 3’ ends overlap and a full-length double-stranded molecule is formed.
The model illustrated above is supported by experts from both parties. Maxygen’s declarant Pääbo, the author of the model above and two of the prior art citations listed above, repeated his support for the model in his evidence. Diversa’s declarants Joyce and Weiss also both refer to Pääbo’s model of jumping PCR as a good depiction of the mechanism of jumping PCR.
According to this model, PCR primers initiate extension of the template during the first PCR cycle. The products of this first cycle then anneal and further extend in subsequent cycles to finally produce a recombined and reassembled double-stranded polynucleotide. This contrasts with the method as recited in the claims, where random template fragments prime extension in the first cycle and it is the products of these random-template initiated extensions that recombine and reassemble to produce the mutagenised or chimeric double-stranded polynucleotide product.
The jumping PCR citations can be divided into two groups:
a) a first group comprising Saiki et al, Shuldiner et al, Meyerhans et al and Pääbo et al (1989), that comments on jumping PCR as it occurs in the background of standard PCR reactions using uncleaved templates, and
b) a second group comprising Pääbo et al (1990), Marton et al and Horton and Pease, that assess jumping PCR in standard PCR reaction using randomly cleaved templates.
Saiki et al and Shuldiner et al represent two of the earliest discussions of recombination in standard PCR reactions. Both articles discuss the formation of recombinants during normal PCR and postulate that recombination occurs when there is incomplete extension of the annealed primer during a first cycle of PCR. In the next cycle the incomplete extension product can hybridise to allelic or homologous sequences and extend along the new template sequence to form a recombined extension product. If there is incomplete extension in this second cycle or in further cycles the same process can occur again leading to further recombination. Shuldiner et al also makes the observation that recombination rates are increased when DNA is nicked or degraded. Meyerhans et al and Pääbo et al (1989) disclose a similar finding and Meyerhans suggests that PCR mediated recombination may be exploited to produce chimaeric molecules.
As such all four citations disclose recombination in PCR reactions. However, based on the Pääbo model of jumping PCR, the PCR recombination disclosed in the citations does not mirror that defined in the claimed methods because PCR primers and not random fragments prime extension in the citations.
When addressing this issue at the hearing Mr Cherry submitted that the presence of an excess of flanking PCR primers did not preclude random fragments priming extension because random-fragment primed extension routinely occurs in the background of standard PCR reactions where jumping PCR occurs. However, the extent to which random fragment priming occurs in standard PCR reactions is a matter of speculation. Diversa has not provided any evidence to support the assertion that random fragment priming occurs in standard PCR reactions. Furthermore, this assertion also conflicts with the evidence provided by Diversa’s own experts, Joyce and Weiss, who both support a model of jumping PCR in which random fragments do not prime extension. In addition, random fragment priming is an undesired outcome in standard PCR reactions, where the presence of a large excess of short PCR primers that share identity with template sequences selects against priming by polynucleotides other than PCR primers.
Given this, I believe that in the current circumstances it is appropriate to rely on the evidence of the expert witnesses and accept that the citations do not disclose the feature of random-fragment primed extension.
A second distinguishing feature is that none of Saiki, Shuldiner, Meyerhans or Pääbo (1989) disclose deliberately cleaving DNA to generate random fragments prior to PCR. Although all four citations conclude that recombination is the result of incomplete extension, and Shuldiner et al observes that recombination rates are increased when DNA is nicked or degraded, there is still no deliberate random cleavage of polynucleotide templates.
A third feature that is also absent from at least three of the citations is that Saiki et al, Shuldiner et al and Pääbo (1989) all teach that jumping PCR is a nuisance, and should be avoided in PCR reactions, thereby teaching away from a method of using jumping PCR to generate mutants.
At the hearing Mr Cherry submitted that this distinction was of no consequence because the claims should be construed as defining methods suitable for producing mutagenised polynucleotides, rather than methods specifically undertaken for the purpose of generating mutagenised polynucleotides.
However, it is generally accepted that claims written in the form of a method “for” a specific purpose are properly construed as being restricted to that purpose, unless the specification expressly teaches that the broader construction should be applied. In the current situation I am satisfied that the usual construction is correct.
A method that is undertaken for the purpose of generating mutants is reflected in the broad description of the invention in title, “DNA Mutagenesis by Random Fragmentation and Reassembly”, and also in the abstract:
“In particular, a method for the production of nucleic acid fragments or polynucleotides encoding mutant proteins is described.”
This purpose is also reflected in the discussion of the prior art, where it is stated that “it would be advantageous to develop a method for the production of mutant proteins”, and in the examples, which each deal with methods of producing mutated and chimaeric polynucleotides, wherein the products of the methods provide improved properties that would not be present in non-mutated polynucleotides.
Given the tenor of the specification I am satisfied that the claims should be construed as relating to methods expressly undertaken for the purpose of producing chimaeric, recombined or mutated polynucleotides and that citations will only be relevant to the novelty of the claims if they disclose undertaking methods for the purpose of generating recombined polynucleotides. As such I am satisfied that Saiki et al, Shuldiner et al and Pääbo (1989) are all silent with respect to a further key feature of the claims.
With respect to Meyerhans et al, although the citation does suggest that jumping PCR might be used to generate recombinants, it does not provide further information about how recombinants might be identified. As such I do not believe that the citation provides clear and unmistakeable directions to do anything with random recombinants, least of all directions to screen or select recombinants for desired functional properties.
In summary, all four citations disclose processes in which at least three of the four key features of the invention are missing.
The second Pääbo paper, Pääbo et al (1990), and Marton et al represent the next step in the study of jumping PCR. Both citations disclose random cleavage of templates using sonication and DNAse treatment and demonstrate that this promotes the rate of PCR recombination. As such these citations provide one of the features that is lacking in the previous four citations.
However, both of these citations still disclose standard PCR conditions in which an excess of flanking PCR primers are added to prime extension and teach that jumping PCR is a nuisance and should be avoided. As such both of these citations are still silent with respect to two of the key features of the claims.
This leaves Horton and Pease. As explained by Mr Cherry at the hearing, Horton and Pease teaches both random cleavage and use of jumping PCR to generate random recombinants. However, as with all of the other citations it still discloses standard PCR, thereby still failing to provide a step in which random fragments prime extension. In addition, as with Meyerhans et al, it also provides scant detail with respect to identification of mutants. Horton and Pease’s sole reference to generating PCR recombinants is the following sentence:
“This implies that, if one wanted to cause random recombination in a PCR, it might be useful to damage the template DNA first, perhaps by treatment with an oxidising agent such as hydrogen peroxide”.
Although this provides a means of damaging DNA and the suggestion that jumping PCR may be used to cause random recombination it does not provide the further step of screening or selecting for mutagenised polynucleotides having desired properties.
As such, Horton and Pease still fails to provide two of the key features of the claims.
In summary, all of the citations fail to disclose a feature that is fundamental to the invention as claimed because all of the citations focus on standard PCR reactions in which PCR primers, rather than random fragments prime polymerase extension. In addition, of the four citations that disclose random cleavage of templates, none disclose the step of screening and selecting for mutants having a desired functional property.
As such, none of the jumping PCR citations provide clear and unmistakeable directions to the invention as claimed.

Spiked shuffling claims

In contrast to mixed shuffling, spiked shuffling starts with a single species of double-stranded polynucleotide template. Mutations are introduced by adding oligonucleotides comprising an area of homology and an area of heterology with respect to the template. Both oligonucleotides and random fragments can then prime extension.
In their submission Diversa argued that the oligonucleotides used in the claims could include PCR primers. As such the claimed methods cannot be distinguished from a citation such as Horton and Pease, which discloses a method of exploiting PCR recombination to produce mutated polynucleotides, the addition of oligonucleotides (PCR primers) and random cleavage of the template. However, this fails to appreciate that the claims specify that the oligonucleotides comprise regions of homology and heterology with the template sequences. This differs from the disclosure in Horton and Pease which simply discloses using standard PCR primer, which are generally identical to template sequences.
Also, as discussed previously, Horton and Pease also fails to provide clear directions to screen or select for polynucleotides having desired functional properties.
As such, none of the jumping PCR citations deprive the spiked shuffling claims of novelty.

PCR shuffling claims

The PCR shuffling claims are outliers among the claims because they do not define the specifics of denaturation, annealing and polymerase extension steps or identify which fragments prime extension. In claim 22 these details are replaced by a reference to PCR amplification.
Although the use of this term raises substantial issues with respect to the clarity and scope of the claim, it is still possible to reach a decision on the novelty of the claims by considering two further features of claim 22. These features are that the methods are used to generate recombined antibodies from templates comprising identical variable framework and heterologous CDR regions and that the mutagenised products of the method are screened or selecting for novel and desirable CDR combinations.

These features are absent from all of the citations. As discussed previously, none of the citations provide clear directions to use PCR recombination to generate mutants and then screen or select for mutants having a desirable functional property, let alone using PCR recombination to generate libraries of displayed mutant antibodies with recombined CDR regions and then screening or selecting antibodies with desirable functional properties. As such, none of the citations provides clear and unmistakeable directions to the PCR shuffling methods.

Citations relating to Grafting and Splice Overlap Extension (SOE) using PCR

The most relevant citations are:
Higuchi et al (1988) Nucleic Acids Research 16(15), 7351-7357
Prodromou and Pearl (1992) Protein Engineering 5(8), 827-9
Shi et al (1993) PCR Methods and Applications 3, 46-53
Daugherty et al (1991) Nucleic Acids Research 19(9), 2471-2476
Horton et al (1990) Biotechniques 8(5), 528-535
Horton et al (1989) Gene 77, 61-68
US 5023171
The processes of grafting and SOE recombine independent fragments to produce chimaeric or recombinant polynucleotides. The methods also include mechanisms for introducing targeted mutations at predetermined sites within polynucleotide products. Although a wide range of techniques fall within this category, all of the SOE methods involve generating fragments of specific and defined lengths and then arranging these fragments in a predetermined order. As such SOE does not involve random cleavage or the use of random fragments to prime extension.
Daugherty et al and Horton et al were the only two citations from this group that were considered by Diversa at the hearing. Daugherty et al teaches generation of template fragments as a consequence of internal PCR primers amplifying fragments within a longer DNA sequence. However, the fragments in Daugherty are not random and specific fragments of a defined length are arranged in a predetermined order in the final product. Also, the fragments are products of the PCR reaction rather than fragments generated by cleavage of the template prior to polymerase extension. As such the process taught in Daugherty is quite distinct from that defined in the claims.
With respect to Horton et al (1990), Weiss’ in his second declaration at paragraph 18 makes the point that Horton acknowledges that PCR random recombination may introduce further mutations into the SOE process where a partially elongated fragment of one gene acts as a primer on a different gene. Horton also states that such additional mutation or recombination may be viewed as either a hazard or a possible means of generating further diversity. Although this disclosure suggests some appreciation of the ability of jumping PCR to increase diversity in any standard PCR reaction, including PCR reactions during the SOE process, Horton et al still does not disclose the generation of random fragments and still discloses fragments whose extension is initiated using PCR primers.
In summary, all of the citations are silent with respect to random cleavage of templates, generation of random fragments and absence of PCR primers. As such, none are relevant to methods of the claims.

Citations relating to Random Pairing and Shuffling of Immunoglobulin Chains

The most relevant citations are:
WO 1993 006213
Winter et al (1994) Annual Review of Immunology 12, 433-55
WO 1991 016427
Random pairing and shuffling is similar to SOE in that discrete fragments are amplified using primers with complementary ends. When the fragments are combined the complementary ends anneal and act as primers for extension. However in contrast to SOE the fragments in shuffling are drawn from two pools or libraries of mixed sequences. Most frequently the two pools are pools of variable immunoglobulin sequences, for example light and heavy chain sequences. By applying shuffling a diverse array of antibodies can be generated and screened to select antibodies with desired affinities.
However, as with SOE, random pairing and shuffling does not involve an initial random cleavage step. In addition PCR primers, rather than random-fragments prime extension. As such, none of the citations provide clear and unmistakeable directions to the methods claimed in any of the group 1 to 3 claims.
Although citations disclosing random pairing and shuffling were discussed extensively by Diversa’s experts they were not mentioned in their submissions at the hearing.

Citations relating to In vivo recombination

The most relevant citations are:
Pompon and Nicolas (1989) Gene 83 ,15-24
EP 252 666
In vivo recombination was also extensively discussed in Diversa’s evidence in support, with Pompon and Nicolas and EP 252 666 detailed as two of the more relevant citations. However, as with many of the other citations, these documents were not discussed at the hearing or in Diversa’s later rounds of evidence.
In contrast to the claimed invention, in vivo recombination relies on cellular machinery to produce recombination between related sequences. As a consequence the process occurs via mechanisms that are quite distinct from those recited in the claims. In particular there is no treatment of a sample of template polynucleotides to produce random double-stranded fragments and no repeated cycles of denaturation, annealing and polymerase extension.
As such, none of the in vivo recombination citations provide clear and unmistakeable directions to the methods claimed in any of the group 1 to 3 claims.

INVENTIVE STEP

An appropriate test for inventive step is that given in Olin Mathieson v Biorex (1970) RPC 157 at page 187, and approved by the High Court in Aktiebolaget Hassle v Alphapharm Pty Ltd [2002] HCA 59 (12 December 2002):
"Would the notional research group at the relevant date in all the circumstances ... directly be led as a matter of course to try the invention claimed in the expectation that it might well produce a useful desired result.",
In respect of inventions relating to a specific combination of integers or features guidance is also provided in Minnesota Mining & Manufacturing Co v Beiersdorf (Australia) Ltd (1979-80) 144 CLR 253 at page 293:
"In the case of a combination patent the invention will lie in the selection of integers, a process which will necessarily involve rejection of other possible integers. The prior existence of publications revealing those integers, as separate items, and other possible integers does not of itself make an alleged invention obvious. It is the selection of the integers out of, perhaps many possibilities, which must be shown to be obvious."
Inventive step was not pursued by Diversa at the hearing. Mr Cherry explained that Diversa conceded that inventive step no longer played a major part in the opposition proceedings and that their major concerns were with the lack of novelty of the claims. In particular, both parties’ experts acknowledge the contribution made to the art by the Stemmer technology. This respect for the technology is reflected in Pääbo’s first declaration, where he describes his first impression of the technology.
“I first heard about Dr Stemmer’s work on molecular evolution of protein sequences (which is the subject of the opposed patent application) in the mid to late 1990’s. I will refer to this work as the “Stemmer technology”. My first reaction was that this was a brilliant new idea and I continue to hold this view today. It did not occur to me that our earlier work described in Pääbo et al. was related to the Stemmer technology. Subsequently, I have occasionally read papers described applications of the Stemmer technology in the major scientific journals, such Nature and Science. Still it did not occur to me that any of the work described in Pääbo et al. was related to the Stemmer technology.”
On the basis that Diversa has conceded that inventive step is no longer a major issue I will not go into detail in my discussion of the prior art, and I will simply focus on the key feature of the invention that is absent from all of the teachings of the prior art.
A common feature shared by all of the citations is the complete absence of any suggestion that recombination and reassembly could be achieved by randomly cleaving templates and conducting PCR-type denaturation, annealing and extension reactions in the absence of PCR primers. Having failed to appreciate this feature, none of the prior art teaches toward the key principles of randomly cleaving templates and relying on random-fragments to prime polymerase extension as claimed. As a consequence, none of the citations deprives the claims of an inventive step.

MANNER OF MANUFACTURE

Diversa submitted that the claimed invention was not a manner of manufacture because the use of PCR to generate mutants was well known in the art at the priority date of the application, as was the use of fragmented templates in PCR reactions. As such the claimed methods were nothing more than a mere new or analogous use of a known contrivance and/or a mere collocation of known integers.
This argument is relevant to the statement in the Commissioner of Patents v Microcell Ltd (1959) 102 CLR 232, that:
“a claim for the use of a known material in the manufacture of known articles for the purpose of which its known properties make that material suitable cannot be the subject matter of a patent.”
However, as suggested in the discussion on inventive step, at the priority date it was not known that randomly cleaved template fragments could recombine and reassemble in the absence of flanking PCR primers. It was also not known that this process could be exploited to generate mutated or chimaeric polynucleotides having desired properties. As such, the claimed methods are not exploiting known properties of PCR-type or polymerase extension reactions. In contrast, it is the previously undiscovered ability of random fragments to recombine and reassemble in the absence of PCR primers that is exploited in the claimed methods.
As such, I am satisfied that the claimed invention is a manner of manufacture.

SECTION 40

Diversa submitted that the scope of many of the words and phrases used in the claims was either unclear, or inconsistent with the invention as described in the specification. They summarised their problems in the following statement taken from their written submissions at the hearing:
“The opponent’s objections may be summarised as that the claims are too broad and unclear. Whether or not there is patentable material in the specification is a matter for the applicant to identify. The opponent acknowledges that at least one of the co-inventors of the opposed patent has made a contribution to the art. However, the claims travel well beyond that contribution.”
Diversa’s assertion that the claims lack clarity is particularly relevant to claim 22 and its dependent claims 23 and 24. Claim 22 provides the following description of the denaturation, annealing and extension steps of the method.
“(ii) pooling and randomly fragmenting said associated polynucleotides or copies to form fragments thereof under conditions suitable for PCR amplification, performing PCR amplification.”
This contrast with the specification at page 23, which states that the method of the invention differs from PCR because the method does not involve the amplification of template molecules seen in PCR.
“This method differs from PCR, in that it is an inverse chain reaction. In PCR, the number of polymerase start sites and the number of molecules grows exponentially. However, the sequence of the polymerase start sites and the sequence of the molecules remains essentially the same. In contrast, in nucleic acid reassembly or shuffling of random fragments the number of start sites and the number (not size) of the random fragments decreases over time.”
This view is also reflected by Maxygen’s experts who repeatedly state the invention is not PCR and does not involve the exponential amplification of template fragments that is seen in PCR.
At the hearing Ms Howard submitted that the clarity issue could be resolved by reading the claims in light of example 1, which refers to “PCR assembly” to describe the denaturation, annealing and extension steps of the invention.
However, this reference does not resolve the problem. It simply coins the phrase “PCR reassembly” for the invention and describes a process where there quite clearly is no PCR amplification. This leaves me with a specification that clearly states that the invention is not standard PCR amplification yet provides no guidance as to how this same term should be interpreted when it is used to define the steps of fragment recombination and reassembly in the claims.
As a consequence claims 22 to 24 lack clarity and I cannot determine their scope.
When I consider Diversa’s remaining submissions on the clarity of the claims, I believe that many of Diversa’s problems arise because they are looking at words and phrases in isolation, rather than in the context of the claims as a whole.
For example, Diversa submitted that it was not clear what the term “overlapping” meant when it was used to define “random overlapping double-stranded fragments of the template”. However, I believe that the meaning of this term is clear when it is considered in the context of the process described in the claim.
The term is used in the claim to define the products of random cleavage of multiple copies of the template. Random cleavage of a sample containing multiple copies of a template polynucleotide produces a population of overlapping fragments sharing regions of common sequence and derived from different copies of the same template sequence. When these fragments are denatured they can recombine with other single-stranded fragments derived from different overlapping double-stranded fragments. This produces double-stranded fragments with single-stranded overhangs that are then filled in the polymerase extension step. As such the term is quite clear once the reader considers what happens when multiple copies of the same template sequence are cleaved to form random fragments and how they will behave once denatured and reannealed.
Similarly Diversa also had problems with the statement that the “average length of mutagenised polynucleotide increases” in each cycle of denaturation, annealing and extension. Again this definition can be readily understood when the reader appreciates that each time one member of a pair of fragments primes replication of the other, the length of the primer increases. When this extended primer is subjected to a further cycle of denaturation, annealing and extension it will reanneal with a new partner, prime replication of this new partner and further increase its length. An inherent consequence of this process is that the average length of the extended polynucleotide fragments increases in each cycle.
Diversa also had further problems with other descriptive features found in the claims, for example the definition of percentage weights of a single species of random fragment in the cleaved template mixture (claim 2) and of the number of different fragment species (claim 3). However, as with the previous problems, these problems could all be resolved by considering the claim as a whole.
Diversa also submitted that the claims lacked fair basis because they define applications of the Stemmer technology that are not explicitly described in the specification. However, as with the priority issues Diversa appears to be suggesting that the claims should mirror the explicit disclosure in the body of the specification, even to the extent of defining nothing more than is described in the examples. However, the claims need not be limited to the examples so long as they do not stray beyond subject matter which is consistent with the invention as described in specification as a whole.
For example, Diversa submitted that the claims lacked fair basis because the term “double-stranded polynucleotide” was too broad. In support of this they pointed to the examples, which only describe recombination and reassembly using a double-stranded DNA template and did not provide any examples of the method applied to double-stranded RNA or RNA:DNA hybrids.
However, when elaborating on this in their written submissions Diversa acknowledged the disclosures at pages 24, 25 and 29 of the specification, which suggest that double-stranded RNA and DNA:RNA hybrids are suitable for use in the methods of the invention. These pages also provide methods for the preparation of RNA and DNA templates and for random cleavage of RNA and RNA:DNA hybrids.
Given this, and Diversa’s failure to provide any evidence that methods disclosed in the specification are not applicable to all species of double-stranded polynucleotides, there is nothing to suggest that the technology cannot be routinely applied to the full range of double-stranded polynucleotide templates.
Similarly, Diversa also submitted that the claims are not restricted to the preferred embodiments disclosed in the specification. For example, the claims do not restrict random fragment size to the 5 to 50 kb described as the preferred fragment size in the body of the specification. The claims also define the use of viral polynucleotides as templates yet this specific type of template is not explicitly disclosed or exemplified in the body of the specification. However, as discussed in relation to support for the full range of double-stranded polynucleotide templates, Diversa did not provide any evidence to suggest that the claimed method could not be routinely applied to template fragments of greater than 50 kb or to templates originating from a diverse range of organisms. As such, there is nothing to suggest that these issues result in a lack of fair basis for the claims.
In their submissions on priority Diversa also made the point that the priority issues that they had raised were also relevant to the fair basis of the claims. However, as discussed in the section on priority, there is sufficient support for these terms both in the priority document and the specification. As a consequence, I am satisfied that they do not lead to fair basis problems in the claims.
In conclusion, claims 22 to 24 lack clarity as a consequence of the use of the term PCR amplification to describe the denaturation, annealing and extension steps.
However, the remaining claims are both clear and fairly based on the specification.

UTILITY

Diversa submitted that the claimed invention lacked utility because the claimed methods included those where only three cycles of denaturation, annealing and extension are conducted and three cycles would not be sufficient to produce mutagenised polynucleotides.
In support of this Joyce made the following comment in his second declaration:
“If there is substantial diversity of nucleic acid fragments, resulting from random cleavage at many different points along the template polynucleotides, then more than a few cycles would be required to assemble the multiple fragments that would result. If no more than 2 or 3 fragments were to be joined then the claimed method would not differ from other methods used to generate libraries of recombinant polynucleotides”
However, these comments fail to appreciate that the claims do not require that a mutagenised or chimaeric product must be formed after only three cycles, the claims simply require that “at least” three cycles are conducted. As such, the only restrictions placed on the number of cycles are that there is a minimum of three, and as many as are required to produce a mutagenised or chimaeric polynucleotide. Although this requires that the practioner determine the appropriate number of cycles and extent of cleavage necessary to reassemble a mutagenised or chimeric polynucleotide, guidance for this is provided in the specification.
Furthermore, as Joyce appears to concede in his final sentence, a mutagenised product may still be produced even when only three cycles are conducted. As explained by Kemp in his second declaration, it is easy to envisage that a suitably mutagenised or chimaeric polynucleotide may be produced in as few as three cycles if the starting template is cleaved into only 2 or 3 fragments.

Given this, and the absence of any evidence to suggest that the claimed methods do not work, I have nothing before me to suggest that the claimed invention lacks utility.

CONCLUSION

The opposition was successful on only one ground, the ground of lack of clarity for claims 22 to 24.
Based on the evidence I have before me the priority date is valid and none of the claims lack novelty, an inventive step, utility or fair basis. The claims also meet the requirements for manner of manufacture.
As there is clearly patentable subject matter within the specification I allow the applicant 60 days from the date of this decision in which to file proposed amendments overcoming the problem noted above.

COSTS

It is general practice that in matters such as these costs follows the event. However, in the current circumstance the Opponent has only been successful on a minor clarity issue. In addition, the Opponent has significantly contributed to the complexity of the hearing by citing an unnecessarily large number of repetitive citations against the novelty of the claims and failing to acknowledge that the majority of these were no longer relevant following amendment of the claims.
Given this, I believe that it is appropriate to only award costs up to the 9 October 2001, the date on which the amendments to the claims were allowed, against the Applicant. Post amendment, I believe that it is appropriate to award costs against the Opponent.

Terry Moore
Delegate of the Commissioner of Patents

05 October 2005

Patent attorneys for the applicant : Spruson & Ferguson, Sydney
Patent attorneys for the opponent : Freehills Patent Attorneys, Melbourne