Analysis
318 Against that complex background it is now necessary to ask whether the invention of claims 1 and 10 of the patent application are sufficiently enabled by the disclosure of P1. In particular, did P1, as at the priority date, provide sufficient information to the notional skilled addressee armed with the common general knowledge to perform the invention of the claim without undue burden over the whole scope of each claim? In deciding this question it is of course necessary to take account of my previous findings as to who is the notional skilled person or skilled team. It is also necessary to have regard to the nature of the invention of the claims, the field of technology, and the breadth of the relevant claims.
319 As to the nature of the invention, I do not think P1 discloses any principle of general application. It does not represent that all, or substantially all, system components (including a particular Cas9) derived from other bacterial species that have a Type II CRISPR/Cas system can be used to cleave DNA in eukaryotic cells, and does not state or imply that there is any reasonable scientific basis for concluding that most, or a substantial number of the many different bacterial species with a Type II CRISPR/Cas system, would be suitable for use in the compositions or the methods of the claims.
320 P1 shows that the inventors tested components derived from S. pyogenes only. There is nothing said in P1 which would indicate that S. pyogenes was likely to be representative of other bacterial species with a Type II CRISPR/Cas system or that the results of the experimentation with S. pyogenes derived components provided any reasonable scientific basis for inferring that Cas9 polypeptides derived from other bacterial species could also be expected to cleave DNA in eukaryotic cells.
321 Those observations are relevant to enablement because, as will be apparent from what I have previously said, P1 does not provide the skilled addressee with any encouragement, whether by the elucidation of a general principle or a series of working examples, that a system using Cas9 components derived from other bacterial species can reasonably be expected to cleave DNA in eukaryotic cells.
322 If, as I have found, P1 is not directed to a skilled team including a microbiologist, it follows that the claims are not sufficiently enabled. The molecular biologist is not told by P1 what other bacterial species have Type II CRISPR/Cas systems or how to determine the endogenous crRNA and tracrRNA sequences for such a species. The papers relied on by Professor Giffard when seeking to ascertain those sequences for S. thermophilus were not the common general knowledge of the molecular biologist and the CRISPRFinder, although a standard tool used by microbiologists, was not a standard tool used by molecular biologists. Nor was Professor Giffard's knowledge as to the proximity of the Cas1, Cas2, Cas9 and the CRISPR array to each other in the bacterial genome as a means of identifying bacteria with a Type II CRISPR/Cas system common general knowledge of molecular biologists. Further, for the reasons previously stated, I do not accept that deficiencies in the common general knowledge of molecular biologists could be filled by the use of literature searches that might generate results that included Bhaya (2011), Makarova (2011) or Deltcheva (2011). Accordingly, the following considerations are only relevant in circumstances where I am wrong about whether a microbiologist is a skilled addressee for P1 and the skilled team has the benefit of the skills and common general knowledge of a microbiologist such as Professor Giffard.
323 In his written evidence, Associate Professor Firestein appears to have accepted that not all CRISPR/Cas9 systems that work in vitro or in prokaryotic cells are necessarily able to cleave target DNA in eukaryotic cells. In Firestein 2, Associate Professor Firestein stated at para 129:
Furthermore, although AP Herold states that he is now aware that not all CRISPR/Cas9 systems that work in vitro or in prokaryotic cells are "necessarily able to target and cleave DNA in eukaryotic cells", in my opinion, this statement acknowledges that some systems that work in vitro also work in vivo. In this regard, my approach does not assume that in vitro activity correlates with in vivo activity, but rather tests and identifies systems that show activity in both contexts. Indeed, there are a number of reports in the literature that demonstrate multiple CRISPR/Cas9 systems, derived from different bacterial species, which are able to target and cleave genomic DNA in eukaryotic cells (as I discuss below in relation to Hou et al (N. meningitidis), Ran 2015 (S. aureus), Muller et al (S. thermophilus), Kim 2017 (C. jejuni) and Chatterjee et al (S. canis)).
324 It is clear from the context in which these observations were made that Associate Professor Firestein was referring to Type II CRISPR/Cas9 systems. I note that each of the articles to which he refers is concerned with a Type II CRISPR/Cas9 system derived from different bacterial species.
325 In my opinion, it would have been apparent to the skilled team as at the priority date that there was considerable uncertainty as to whether or not a CRISPR/Cas9 system derived from any particular bacterial species other than S. pyogenes would work in eukaryotic cells. The experts agreed that the ability of a CRISPR/Cas9 system to work in eukaryotic cells cannot be assumed by activity shown in an in vitro DNA cleavage assay and that additional work is required to test the CRISPR/Cas9 system in eukaryotic cells. They also agreed that the in vitro DNA cleavage assay is a relatively straightforward first approach that they would have undertaken to test a CRISPR/Cas9 system and, if it showed cleavage activity, would lead them to test the system in a eukaryotic cell.
326 Both Professor Thomas and Associate Professor Herold said that they would cease work on the candidate if it failed to show activity in vitro, although Associate Professor Firestein said that he would not assume that a candidate that did not work in vitro would not work in vivo. However, he did not refer to any examples in which a system that worked in vivo would not work in vitro. I regard Associate Professor Herold's and Professor Thomas' approach as more representative of the thinking of the notional skilled team on this topic.
327 Associate Professor Herold and Professor Thomas were also of the view that significant experimental work would need to be done to validate the use of the system in eukaryotic cells. Professor Thomas gave evidence that this would require conducting in vivo experiments at a wide range of different target sites and in a wide range of different cell lines. Associate Professor Firestein said in JER 1 that those steps would be straightforward and that the time taken to perform the work could depend on how much optimisation would be required for any particular system. The evidence of all three witnesses was somewhat vague as to how long such work would take but I am persuaded that it would involve a multi-step process requiring a significant amount of work with each step in the process dependant on the success of the previous step.
328 The skilled team would not know whether its work was likely to yield a product capable of achieving double stranded breaks at target locations in eukaryotic cells until the relevant candidate had been trialled and validated by in vitro and then in vivo experiments.
329 In relation to Associate Professor Firestein's statements in JER 1 and his affidavit evidence suggesting that experimental validation of a Type II CRISPR/Cas9 system derived from other species would be straightforward, I refer to the following exchanges in the concurrent evidence:
MR DIMITRIADIS: Would you agree, Professor Firestein, that your approaches that you've discussed in your affidavit would involve a very significant amount of work - a research project - firstly?
ASSOC PROF FIRESTEIN: It's difficult for me to - to speculate on that. It may involve a lot of work or it may be done quite quickly because, you know, if I bring up the case of streptococcal canis again, this is a system that's nearly 90 per cent homologous to strep pyogenes. It works with the same - well, not the same but let's say the tracrRNA is exactly the same, the crRNA differs by only one nucleotide.
Using that type of system, it could be repurposed quite quickly. If I had to go through hundreds of different CRISPR Cas9 systems, then, obviously that would be a significant amount of work. But it may also be that I - I hit upon a system much more quickly than that. And I think the important thing is that this is - that's why his approach and the same approach applies for every CRISPR Cas9 system that I'm utilising.
MR DIMITRIADIS: Professor Firestein, you used the language "I can't speculate". I suggest to you that that is because you have never engaged a process of this kind seeking to identify a component of a CRISPR Cas9 system from another bacterial species, correct?
ASSOC PROF FIRESTEIN: It is true that I have never sought to identify a CRISPR Cas9 from another species. And I refer back to my comments on that, that I've not found the NGG to be a practical limitation. The reason that I cannot speculate is, based on my broader experience in research, that sometimes we set out a logical approach to [sic] certain problem, we conduct experiments, sometimes those experiments work quite quickly, sometimes they take longer. And I also would, in that respect, disagree with Professor Herold's assertion that a Nature paper necessarily implies that there has been a lot of work involved.
330 Associate Professor Firestein therefore accepted that at least in some cases the amount of work involved in using a system derived from another bacterial species would be significant, but that in other cases the work could be done quite quickly. He cites as an example a system derived from S. canis that he suggested could be developed quite quickly, because it is nearly 90% homologous to S. pyogenes. His evidence was, in effect, that if he "hit upon" a system like that, it could be developed very quickly. There are several points to make about this evidence.
331 The evidence suggests that the system derived from S. canis was first characterised in a paper authored by Chatterjee et al published in 2018 ("Chatterjee 2018") which described a Cas9 protein derived from S. canis and the PAM sequences it recognised. I will say more about this paper shortly, but it constitutes post priority date information that would not have been available to the skilled team. In particular, Associate Professor Firestein's reference to the S. canis system being nearly 90% homologous to the S. pyogenes system seems to be derived from Chatterjee (2018) which reported S. canis Cas9 as having 89.2% sequence similarity to S. pyogenes Cas9. The sequence appears not to have been characterised or functionally validated until Chatterjee et al undertook that work.
332 However, I accept that some support for Associate Professor Firestein's reasoning is found in Jinek. This paper showed that of four S. pyogenes orthologs investigated in vitro, two Cas9 derived from L. innocua and S. thermophilus which are bacterial species that are homologous to S. pyogenes (54% and 58% homology respectively) could be complexed with the same S. pyogenes sgRNA to cleave DNA in prokaryotic cells. This was not so in the case of two Cas9 derived from C. jejuni and N. meningitidis which are bacterial species that have lower homology to S. pyogenes Cas 9 (16% homology for both). I also note that Associate Professor Herold and Professor Thomas agreed that homology between S. canis and S. pyogenes Cas9 and their crRNA and tracrRNA are nearly identical and that it is therefore very likely that the Cas9 and sgRNA components of these systems could be interchanged to mediate CRISPR/Cas9 cutting. Therefore, whilst I do accept the proposition that it may be easier to develop a functional system in bacterial species that are highly homologous to S. pyogenes, including because the same S. pyogenes sgRNA or a very similar sgRNA can be used, I think it must also follow that there is likely to be a very substantial proportion of Cas9 derived from other bacterial species that are not sufficiently homologous to S. pyogenes Cas9 to enable use of the same sgRNA or a very similar sgRNA to cleave DNA in eukaryotic cells.
333 Importantly, claims 1 and 10 are not confined to Streptococcus Cas9 polypeptides and extend to any bacterial species with a Type II CRISPR/Cas system whether or not within the Streptococcus genus and whether or not the Cas9 protein for that species is highly homologous to Cas9 derived from S. pyogenes. The disclosure required of P1 is that it be such as would enable the skilled team armed with the common general knowledge to make all, or substantially, all, embodiments within the scope of the claims without undue burden. Associate Professor Firestein's evidence concerning the ease with which the sgRNA used in Jinek could be re-purposed to Cas9 from other species would not hold true across the scope of the claims.
334 The respondents relied on the paper by Ran et al published in Nature in 2015 by a group of researchers associated with the Broad Institute which appears to be a collaboration involving MIT and Harvard. In their closing submission, ToolGen objected to the respondent's reliance on Ran (2015) and submitted that it and other post priority date papers should be given no weight on the issue of undue burden for the following reasons:
(a) no persons who did that work were called to give evidence;
(b) the articles are notable for their lack of reporting any difficulty at all, or any difficulty which was not easily overcome;
(c) the reviewers of the article were not called to give evidence to say precisely what aspect of the work they regarded as qualifying it for publication;
(d) the articles often involved many different tasks and experiments, and it is impossible to ascertain how much work was involved in any individual task relevant to the specific questions of enablement in this case;
(e) it is impossible to separate the work which may have gone into writing up and reporting upon the experiment in a form suitable for publication in a journal from the actual scientific work involved;
(f) the articles were not contemporaneous records of the amount of work done, but rather were highly polished accounts of work done, specifically drafted for the purpose of impressing reviewers and achieving publication; and
(g) the respondents have adduced no evidence that any of the work the subject of those papers was inventive or non-routine, and mere publication does not prove it to be so.
335 Each of the articles to which ToolGen's submission was directed was admitted into evidence without objection or limitation as to the use which might be made of it. That said, in deciding what weight to give Ran (2015) and other post priority date papers I have had regard to ToolGen's submission. However, I have also had regard to the fact that at least some of these papers were relied on by Associate Professor Firestein as demonstrating that multiple CRISPR/Cas9 systems derived from different bacterial species are able to target and cleave genomic DNA in eukaryotic cells. He accepted, speaking in the context of Ran (2015), that a paper published in Nature would be one that the publisher and reviewers considered to be a fairly substantial piece of original research or, in his words, "substantial in its concept and advancement of the field". He did not accept that you could draw any inference as to the amount of time the research work took. I agree with that, but I also accept Professor Thomas' evidence who, also speaking of Ran (2015), said that it must have reflected a very large amount of work.
336 Both Associate Professor Herold and Associate Professor Firestein referred to Chatterjee (2018) which was submitted in May 2018 and published in October 2018 in Science Advances as a research article. Chatterjee (2018) states that while numerous Cas9 homologs have been sequenced, only a handful of Streptococcus orthologs have been characterised or functionally validated. The authors describe how they characterised an orthologous Cas9 protein from S. canis which, as mentioned earlier, had a sequence homology of 89.2% similarity to that of S. pyogenes Cas9. They explain in some detail how they determined the PAM sequences recognised by S. canis Cas9 first in vitro and then in vivo in human cells.
337 Another paper by Kim (2017) submitted in October 2016 and published in February 2017 in Nature Communications (whose authors included Jin-Soo Kim and other researchers that are named inventors of P1 and the patent application) noted that several CRISPR/Cas9 orthologs had been used for genome editing. Kim (2017) describes a new Cas9 ortholog derived from C. jejuni an advantage of which is said to be its smaller size. The paper describes the steps taken by the authors to determine the PAM sequence for this Cas9 ortholog and to optimise the length of the sgRNA before delivering the system via an adeno-associated virus (AAV) to mammalian cells for in vivo genome editing.
338 With reference to Ran (2015), the respondents submitted:
179. The nature and standard of work required to develop another Type II CRISPR/Cas9 system for use in gene editing in eukaryotic cells are well illustrated by reference to the Ran paper. This was a publication in Nature, a prestigious journal, where the authors studied six CRISPR/Cas9 systems which showed cleavage in vitro but only identified two species which showed cleavage in vivo (including S. Aureus).
…
180. Ran was by members of the Broad Institute, which was not a typical academic laboratory, because of its breadth of expertise. The paper represents an enormous body of work, and is a combination of many peoples' efforts. Its publication in Nature reflects the cutting edge nature of the research involved. It was an original piece of research that is substantial in its concept and advancement of the field.
339 The research work the subject of Ran (2015) involved more than finding and validating another bacterial species that could be used in place of S. pyogenes. It appears that the researchers specifically focused on Cas9 derived from other bacterial species with a lower molecular weight than S. pyogenes Cas9 because these were thought to be better suited for delivery using AAV vectors. However, in my opinion, Ran (2015) does provide some insight into the work involved in identifying S. aureus derived Cas9, the crRNA and tracrRNA associated with it and, in particular, whether the work involved would constitute an undue burden.
340 Ran (2015) describes the steps taken by the researchers to determine the endogenous crRNA and tracrRNA sequences for Cas9 derived from the six species that were investigated, the PAM sequences for each of them (using a plasmid library) and the ability of each Cas9 paired with a single guide RNA to cleave target sites first, in vitro, and then in vivo in mammalian cells. This ultimately led the researchers to identify S. aureus Cas9 as one which produced "indels" (ie. insertions or deletions) with efficiencies comparable to S. pyogenes Cas9.
341 The authors of Ran (2015) state at page 186:
In search of smaller Cas9 enzymes for efficient in vivo delivery by AAV, we have previously described a short Cas9 from the CRISPRI locus of Streptococcus thermophiles LMD-9 (St1 Cas9, ~3.3 kb) as well as a rationally-designed truncated form of SpCas9 (ref. 18) for genome editing in human cells. However, both systems have important practical drawbacks: the former requires a complex protospacer-associated motif (PAM) sequence (NNAGAAW), which restricts the range of accessible targets, whereas the latter exhibits reduced activity. Given the substantial diversity of CRISPR-Cas systems present in sequenced microbial genomes, we therefore sought to interrogate and discover additional Cas9 enzymes that are small, efficient and broadly targeting.
(footnotes omitted)
In support of that statement the authors referenced Chylinski (2014) (co-authors including Makarova and Charpentier) published in 2014 and by Chylinski (2013) (co-authors including Charpentier) published in 2013 (both post priority date).
342 The authors of Ran (2015) then referred to their analysis of over 600 Cas9 orthologs with protein sizes approximately 1,350 and 1,000 amino acid residues in length. From the 600 Cas9 orthologs, the authors selected six candidates for profiling which involved ascertaining the crRNA and the tracrRNA for each Cas9, and designing a single guide RNA (sgRNA) for each of the six orthologs. The authors then identified the PAM sequence for each Cas9 by constructing a library of plasma DNA and performing an in vitro cleavage assay. The authors reported that the Cas9 orthologs, in combination with the sgRNA, successfully cleaved their targets in vitro. They then proceeded to investigate whether they would do the same in mammalian cells, after noting that "DNA cleavage activity in cell-free assays does not necessarily predict activity in mammalian cells". Of the six orthologs tested, only S. aureus produced indels with efficiencies comparable to those of S. pyogenes. The authors' investigation thereafter focused on S. aureus. It is apparent that by that stage of their work, they had successfully cleaved DNA in mammalian cells using components derived from S. aureus which they then sought to optimise in various ways.
343 I think Ran (2015) is useful in so far as it describes the work that its authors undertook in order to identify other Cas9 orthologs which could be used to cleave DNA in in vitro assays, and the work its authors took to confirm that S. aureus Cas9 could be used to cleave DNA in vivo in human cells. Ran (2015) and the various other post priority date papers also tend to show that for some years after the priority date there were various research groups attempting to identify, characterise and validate Cas9 orthologs derived from bacterial species other than S. pyogenes with the aim of identifying CRISPR/Cas9 systems that were better suited for use in AAV vectors, that were able to target a wider variety of PAM sites, and that also had reduced "off-target" effects (ie. the introduction of unintended breaks).
344 In another paper Wang (2013) received by the publisher in March 2013 and published in Cell in May 2013, the authors state:
There are several potential limitations of the CRISPR/Cas technology. First, the requirement for a NGG PAM sequence of S. pyogenes Cas9 limits the target space in the mouse genome. It has been shown that the Streptococcus thermophilus LMD-9 Cas9 using different PAM sequence can also induce targeted DNA cleavage in mammalian cells (Cong et al., 2013). Therefore, exploiting different Cas9 proteins may enable [sic] to target most of the mouse genome. Second, although the sgRNAs used here showed high targeting efficiency, much work is needed to elucidate the rules for designing sgRNAs with consistent high targeting efficiency, which is essential for multiplexed genome engineering. Third, although our off-target analysis for the seven most likely off targets of Tet1 and Tet2 sgRNAs failed to detect mutations in these loci, it is possible that other mutations were induced following as yet unidentified rules. A more thorough sequencing analysis for a large number of sgRNAs will provide more information about the potential off-target cleavage of the CRISPR/Cas system and lead to a better prediction of potential off-target sites.
345 The paper which was published not long after the priority date was by a research group associated with the Broad Institute. Associate Professor Firestein described the Broad Institute as "a very well-oiled machine … not like a typical academic lab". He was referring to the breadth of its expertise and the speed with which its research groups could generate data. The paper is consistent with what is in my view the effect of the evidence more generally that the development of new CRISPR/Cas9 systems for use in genome editing using bacterial species that recognised PAM sequences different from those recognised by S. pyogenes and S. thermophilus, would require considerable work in relation to the design of the sgRNA with high targeting efficiency and investigation of the different systems' off-target effects. I do not consider that the work involved in identifying and characterising systems derived from different bacterial species was likely to have been straightforward or routine. It appears to have been innovative and advanced research work undertaken by highly specialised groups of researchers which included leaders in the field. None of the experts who gave evidence had any experience either individually or as part of a team as at the priority date investigating the use of different CRISPR/Cas9 systems in genome editing or, in particular, the selection or characterisation of bacterial species that might be suitable for that purpose. Of course, that is hardly surprising given the state of the art as at the priority date.
346 On the question of undue burden, ToolGen placed considerable reliance on the Full Court's decision in Warner-Lambert Co LLC v Apotex Pty Ltd (No. 2) (2018) 129 IPR 205 ("Warner-Lambert FC"), the decision at first instance in the same case in Warner-Lambert, and also the decision of Heerey J in Eli Lilly & Co v Pfizer Overseas Pharmaceuticals (2005) 64 IPR 506 ("Eli Lilly"). Both cases were concerned with methods of treatment using known pharmaceutical compounds, and both were concerned with the application of s 40 of the Act prior to the its amendment by the RTB Act. It was also common ground in both cases that the patent specification in suit contained the information necessary to enable the skilled addressee to prepare some compositions within the claims. In Eli Lilly Heerey J said at [193]:
It would be necessary to test for oral bioavailabilty, toxicity and effectiveness, but the evidence shows that while these steps call for skill, they are essentially routine for those skilled in this area. The term routine here (and in other contexts in this case) is not used as a synonym for simple and easy. In the present case the hypothetical skilled workers at the hypothetical workbench are persons holding academic qualifications at the Ph D [sic] level together with practical experience. It would not be necessary to employ such persons unless the task they had to perform was a difficult one. Yet this does not of itself mean that the patent could not be worked without further invention.
His Honour's decision on this issue was upheld on appeal: Pfizer Overseas Pharmaceuticals v Eli Lilly & Co (2005) 225 ALR 416 ("Pfizer") at [342] per French and Lindgren JJ.
347 Pfizer was applied at first instance in Warner-Lambert FC at [262]-[263] which concerned a patent involving a method of treating pain using pregabalin. Dismissing the appeal, the Full Court in Warner-Lambert FC said at [129]:
In this connection, we accept that the appellants raise a valid point of distinction which is relevant to the respondent's criticism that the specification does not contain, for example, specific dosages or a safety and toxicity profile in respect of the use of the compounds for the treatment of pain in human subjects, and that a clinician would not use the compounds for this purpose without this information. Whilst the primary judge appears to have accepted that, from the clinician's perspective, this information would be necessary - with the consequence that further work would need to be carried out in this regard before the compound would be put to use in clinical practice - his Honour was not satisfied that this meant that the requirements of s 40(2)(a) of the Act had not been met. This was because, in the circumstances of the present case, the work required to put the invention into practice was of a routine (that is, non-inventive) nature for the person skilled in the art (now accepted on appeal), even though the work to be undertaken would require considerable skill, effort and resources or be complex, time-consuming and expensive.
348 However, it is clear that even if the work required of the skilled addressee is non-inventive and routine, it may still amount to undue burden. The skilled addressee is not expected to engage in an unreasonable amount of experimentation, research or study. If the work required involves "… prolonged study of matters presenting initial difficulty" the claim will not be properly enabled: see Gilead at [438] per Jagot J, upheld on appeal Idenix Pharmaceuticals LLC v Gilead Sciences Pty Ltd (2017) 134 IPR 1 at [144] (considering s 40(2)(a) of the Act before amendment by the RTB Act); see also Mentor Corp. v Hollister Inc [1993] RPC 7 at 13 per Lloyd LJ citing with approval the judgment of Buckley LJ in Valensi v British Radio Corporation [1973] RPC 337 at 377.
349 The routine work referred to in both Eli Lilly and Warner-Lambert FC included work necessary to support an application for regulatory approval. In Merck & Co Inc v Arrow Pharmaceuticals Ltd (2006) 154 FCR 31 at [108] the Full Court observed (although not in the context of sufficiency) that "… it is a matter of notoriety that prolonged testing for the purpose of regulatory approval must occur between the stage of patent application and commercial marketing". The same point was made by Jacob LJ who, when considering the concept of enabling disclosure, said of genetic engineering and pharmaceutical inventions, "[t]he work that goes into bringing them to market relates to testing efficacy and safety - not in actually making the invented product": Halliburton Energy Services Inc v Smith International (North Sea) Ltd [2006] EWCA Civ 1715 at [18]. His Lordship also observed that the test of "undue effort" and the words of the relevant statutory provision ("clearly enough and completely enough") emphasise that the question is one of degree. As to how one is to say when the work involved is too much, his Lordship said at [21]:
The answer is that the line is one to be drawn by an exercise of judgment, taking into account all of the relevant factors, one of which is of course the nature of the invention itself and its field of technology. But there are other factors too - for instance, the width of the patent claim or whether it has functional limitations which require too much work to explore.
350 Besides the width of the claim and any relevant functional limitations, it is also necessary to consider what information has been provided by P1. It is all very well for ToolGen to say that the necessary information forms part of the common general knowledge, but that does not mean that the deployment of the common general knowledge to solve a problem raised in P1 (how to avoid the 5'-NGG-3' PAM limitation) is routine work that does not amount to an undue burden. P1 provides no direction at all as to what options for approaching and solving the problem exist, how they should be prioritised or whether any of them can be expected to work.
351 The work that the notional skilled team would need to undertake at the priority date to perform the invention of claims 1 and 10 using a bacterial species other than S. pyogenes would in my opinion involve a significant research project that would not be straightforward or routine. My reasons are as follows.
352 First, to the extent that P1 might be understood as inviting the skilled addressee to attempt to perform the invention using Cas9 derived from another species, it offers no relevant encouragement or direction. Nor, as I have previously explained, does P1 disclose any principle of general application.
353 Second, the microbiologist would need to identify one or more suitable candidates. P1 provides no guidance on that topic. If the microbiologist were to rely on the Makarova (2011), the list of potential candidates would number around 120. If the microbiologist used Genebank for this purpose, the list of potential candidates would exceed 1,500. The fact that Professor Giffard chose S. thermophilus (which had been singled out for use in genome editing by Cong et al in early 2013) out of the 1,513 search results generated, seems to have involved a remarkable stroke of luck assuming his selection was not influenced by post priority date developments. Moreover, as previously noted, the S. thermophilus system did not relieve the 3'-NGG-5' PAM sequence limitation.
354 Third, assuming the microbiologist identified another candidate (eg. Professor Giffard's S. thermophilus) and determined its endogenous crRNA and tracrRNA sequences, it would be necessary to identify and characterise the mature crRNA and tracrRNA. P1 does not provide any guidance on to how that is to be done. I accept that this task may be within the skill of the microbiologist with expertise in CRISPR/Cas systems, but I do not consider that this would be routine work. There was no evidence to suggest that Professor Giffard had himself performed this task before the priority date.
355 Fourth, the RNA-seq data for most bacterial strains of interest was unlikely to be publicly available at the priority date. At that time, there were only around seven species of bacteria for which Type II repeat spacer expression data had been made available.
356 Fifth, while there were alternative methods of identifying and characterising the relevant sequences (eg. RNA-seq experiment, northern hybridisation/northern blot or RNase protection experiments) that could have been adopted by the skilled team, these depended on obtaining a bacterial isolate for the bacterial species of interest. None of the experts suggested that bacterial isolates for every known species of interest, or even a substantial proportion of them, were publically available at the priority date.
357 Sixth, the use of next generation sequencing needed to perform a RNA-seq experiment and to analyse the cut plasmids generated from the cleavage assays for PAM identification was not something that was routine at the priority date. This was a task that would need to be outsourced to a specialist laboratory.
358 Seventh, the requirement to identify the crRNA and tracrRNA molecules as the first step of a northern hybridisation (northern blot) experiment could be difficult for the reasons explained by Professor Thomas. In particular, it may be difficult to predict with certainty the mature crRNA from the spacer sequences of the CRISPR array because of the processing of pre-crRNA into mature crRNA. It may also be difficult to identify the tracrRNA because it would not be clear what part of the repeat sequence of the CRISPR array would be complementary to the tracrRNA.
359 Eighth, the compilation of a PAM variant library to identify and validate the PAM site was not routine work as at the priority date. None of the molecular biologists had used a PAM variant library at the priority date. Further, if using the in vivo approach proposed by Associate Professor Firestein, cellular repair mechanisms in eukaryotic cells may mutate the cut which would inhibit detection of the PAM. I accept Associate Professor Herold's evidence that using a barcode approach to overcome this issue would create considerable additional work which would need to be outsourced to a specialist laboratory.
360 Ninth, the alternative method of using an in silico approach for PAM site identification relied on identifying the protospacer of the phage where a particular spacer originated. This method depends on the DNA sequence for the phage being publicly available. Professor Thomas and Associate Professor Herold questioned whether that DNA sequences for all potential phages of interest were publically available at the priority date. Both accepted that, as at the priority date, this would be a matter for the microbiologist to investigate. Professor Giffard did not give evidence on the availability of phage sequences at the priority date. In the absence of evidence to the contrary, it cannot be assumed that the DNA sequence for any particular candidate would have been available. Nor would this method of identifying the PAM site be regarded as routine or straightforward.
361 Tenth, for any Cas9 derived from bacterial species that are not highly homologous to S. pyogenes, the S. pyogenes sgRNA could not be used, in which case Associate Professor Firestein's third approach of using the endogenous tracrRNA and crRNA to create a sgRNA would need to be adopted. Even if it was possible to use the S. pyogenes sgRNA in another system using Cas9 derived from a different bacterial species, this would not be possible for any Cas9 that was not highly homologous.
362 In my opinion the skilled team would be required to carry out prolonged research and experimentation and would most likely encounter significant difficulties along the way. Much of the work would be non-routine and would be carried out in circumstances where P1 provided no meaningful guidance or direction and no assurance of success.
363 I am persuaded that as at the priority date, P1 did not enable a skilled team including a molecular biologist specialising in genome editing in eukaryotic cells and a microbiologist with expertise in CRISPR/Cas systems in prokaryotes, to make the compositions of claim 1, or perform the methods of claim 10, using a bacterial species other than S. pyogenes, without undue burden.