`Petitioner’s Reply
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`
`
`
`Celltrion, Inc.
`
`Petitioner,
`
`v.
`
`Genentech, Inc.
`
`Patent Owner
`
`Patent No. 6,407,213
`
`
`
`
`
`
`
`
`
`Inter Partes Review No. IPR2017-01373
`
`PETITIONER’S REPLY TO PATENT OWNER’S RESPONSE
`
`
`
`
`
`
`
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`
`2.
`3.
`
`4.
`
`5.
`
`6.
`
`B.
`
`C.
`
`Table of Contents
`
`All Challenged Claims Would Have Been Obvious over Queen-1989 or
`Queen-1990 and the PDB Database, Optionally in Light of Tramontano
`and/or Kabat-1987 and/or Hudziak ..................................................................... 4
`A.
`Claims 12, 42, 60, 65-67, and 71-79 are Obvious Over Queen-1989
`or Queen-1990, in Combination with the PDB ......................................... 4
`1.
`A POSA Following the Teachings of the Prior Art Would Have
`Identified the Claimed Residues for Substitution ........................... 4
`A POSA Would Have Used The PDB As Dr. Riechmann Did ...... 7
`A POSA Would Have Expected the Resulting Humanized mAb
`to Bind the Target Antigen ............................................................. 8
`Holding the Challenged Claims Invalid Would Not Have
`“Sweeping Consequences” ........................................................... 10
`Petitioner’s Arguments Regarding Queen-1989 Do Not Rely
`on Queen-1990 .............................................................................. 10
`Queen-1990 and Tramontano Further Support Petitioner’s
`Position .......................................................................................... 11
`The Consensus Sequence Limitation of Claims 4, 33, 62, 64, and 69
`Was Taught By the Prior Art ................................................................... 12
`The “Up to 3-Fold More Binding Affinity” Limitation of Claim 65
`Was Taught By the Prior Art ................................................................... 15
`The “Lack Immunogenicity Compared to a Non-Human Parent”
`Limitation of Claim 63 Was Taught By the Prior Art ............................ 16
`The Limitations Related to Binding p185HER2 of Claims 30-31, 33,
`42, and 60 Was Taught By the Prior Art ................................................. 17
`PO’s Attempts to Establish an Earlier Date of Invention for Claims 12, 42,
`60, 65, 71, 73-74, and 79 Are Insufficient ......................................................... 18
`Secondary Considerations of Non-Obviousness Do Not Render the
`Challenged Claims Nonobvious ........................................................................ 21
`A.
`The Use of a Consensus Sequence Would Not Have Been
`Unexpected .............................................................................................. 21
`PO Has Not Established that the Alleged Unexpected Results and
`Commercial Success Have a Nexus to the Challenged Claims .............. 22
`
`
`i
`
`D.
`
`E.
`
`B.
`
`I.
`
`II.
`
`III.
`
`
`
`
`
`
`TABLE OF AUTHORITIES
`
`IPR2017-01373
`Petitioner’s Reply
`
` Page(s)
`
`Cases
`In re Applied Materials, Inc.,
`692 F.3d 1289 (Fed. Cir. 2012) .............................................................................. 14
`
`Chiron Corp. v. Genentech Inc.,
`363 F.3d 1247 (Fed. Cir. 2004) .............................................................................. 20
`
`Coleman v. Dines,
`754 F.2d 353 (Fed. Cir. 1985) ................................................................................ 18
`
`Ecolochem, Inc. v. S. Cal. Edison Co.,
`227 F.3d 1361 (Fed. Cir. 2000) .............................................................................. 18
`
`Ecolochem, Inc. v. S. Cal. Edison Co.,
`91 F.3d 169, 1996 WL 297601 (Fed. Cir. 1996) .................................................. 3, 5
`
`KSR Int’l Co. v. Teleflex Inc.,
`550 U.S. 398 (2007) .................................................................................................. 6
`
`In re Kubin,
`561 F.3d 1351 (Fed. Cir. 2009) ................................................................................ 9
`
`Mahurkar v. C.R. Bard, Inc.,
`79 F.3d 1572 (Fed. Cir. 1996) ................................................................................ 18
`
`Medichem, S.A. v. Rolabo, S.L.,
`437 F.3d 1157 (Fed. Cir. 2006) .............................................................................. 18
`
`Ormco Corp. v. Align Tech., Inc.,
`463 F.3d 1299 (Fed. Cir. 2006) .............................................................................. 22
`
`In re Peterson,
`315 F.3d 1325 (Fed. Cir. 2003) ................................................................................ 7
`
`Santarus, Inc. v. Par Pharm., Inc.,
`694 F.3d 1344 (Fed. Cir. 2012) .............................................................................. 16
`
`
`
`ii
`
`
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`St. Jude Med., Cardiology Division, Inc. v. Board of Regents of the Univ.
`of Mich., IPR2013-00041, Paper 69 (PTAB May 1, 2014) .................................... 24
`
`Ex Parte Takeshi Shimono,
`Appeal 2013–003410 (PTAB Apr. 29, 2015) ......................................................... 22
`
`Statutes
`
`35 U.S.C. §102(b) ........................................................................................................ 20
`
`
`
`iii
`
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`After the Board considered the prior art cited by Petitioner and granted
`
`institution, PO abandoned its defense of claims 1, 2, 25, 29, 80, and 81 (POR, 18-
`
`22), essentially conceding that their limitations did not constitute patentable
`
`distinctions over the work of others.1
`
`Having made this concession, PO falls back to the position that certain
`
`elements distinguish a handful of remaining claims from the prior art: (1) the use
`
`of a “human consensus sequence” as the human framework (claims 4, 33, 62, 64
`
`and 69), (2) “lack of immunogenicity as compared to a non-human parent
`
`antibody” (claim 63), (3) “up to 3-fold more” binding affinity than the parent
`
`antibody (claim 65), and (4) binding to p185HER2 (claim 30). (POR, 1-4.) PO also
`
`alleges that even if some of the claimed substitutions were explicitly disclosed in
`
`the prior art, not all of them were. (POR, 3-4.) These elements do not make the
`
`otherwise-obvious claims patentable, however.
`
`Elements (1)-(4) above affect only a small number of claims. PO’s
`
`arguments do not change the fact that each element was explicitly disclosed in, or
`
`obvious from, the prior art.
`
`PO’s allegation that not all of the claimed residues are explicitly disclosed in
`
`the prior art ignores all that the prior art teaches, as well as the limitations of what
`
`1 Therefore, this Reply addresses the remaining challenged claims: 4, 12, 30, 31, 33,
`
`42, 60, 62-67, 69, and 71-79.
`
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`the patent teaches. The patent does not teach that all of the claimed substitutions
`
`and possible combinations thereof will work to improve the binding of all of the
`
`huge number of antibodies that could fall within the claims. Indeed, as the patent
`
`and prior art make clear, the substitutions needed to optimize antigen binding will
`
`vary from mAb to mAb. In fact, the patent does not teach that any specific
`
`substitution will work in any specific mAb falling within the claims, other than the
`
`handful of examples provided. Instead, the patent teaches that a POSA must
`
`conduct extensive modeling to determine which of the claimed substitutions might
`
`possibly improve binding in a given project, and then use serial mutagenesis—
`
`making mAbs with different substitutions and then testing them—to confirm which
`
`precise combination of possibilities optimizes that binding. But there is no
`
`material difference between this method and the humanization roadmap laid out in
`
`Queen-1989 and -1990. PO’s expert Dr. Wilson conceded that each of the steps
`
`required would have been routine to a POSA, and thus PO cannot credibly dispute
`
`that a POSA would have been able to identify the claimed substitutions when
`
`developing humanized mAbs that require them for optimal binding affinity. In
`
`fact, if this is not the case, the claims are not enabled.
`
`In addition, PO misconstrues the law when it comes to the laundry lists of
`
`substitutions recited in the Markush groups of claims 4, 12, 30, 31, 33, 42, 60, 62,
`
`63, 65-67, 69, and 71-78. As the Board found in its institution decision, a POSA
`2
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`applying the roadmap in Queen-1989 or -1990 would have identified at least some
`
`of the residues in each of claims 4, 30, 31, 33, 62, 63, 64 65, 66, 67, 69, and 78 as
`
`likely candidates for substitution during humanization. (Institution Decision, 16)
`
`(A POSA “would have identified nine positions in the light chain and eleven in the
`
`heavy chain as candidates for substitution, including those recited in the challenged
`
`claims.”). Since the claimed Markush groups merely require “one or more” of the
`
`recited substitutions, this is sufficient to render them obvious. See, e.g.,
`
`Ecolochem, Inc. v. S. Cal. Edison Co., 91 F.3d 169, 1996 WL 297601, *2 (Fed.
`
`Cir. 1996) (“[I]f utilizing one element of the [Markush] group is anticipated or
`
`obvious, the patentee is precluded from arguing that the claim is valid.”). In
`
`addition, as the Board found, a POSA would have identified all of the substitutions
`
`in claims 12, 42, 60 and 71-79 (Institution Decision, 16), which means they, too,
`
`are invalid.
`
`PO’s attempt to avoid invalidity through antedating also fails. The evidence
`
`of record is not corroborated and does not satisfy PO’s burden to prove an earlier
`
`invention date. Moreover, PO only attempts to establish an earlier invention date
`
`of claims 12, 42, 60, 65, 71, 73-74, and 79, and therefore concedes that it cannot
`
`antedate Petitioner’s prior art for claims 4, 30-31, 33, 62-64, 66-67, 69, 72, and 75-
`
`78.
`
`3
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`I.
`
`IPR2017-01373
`Petitioner’s Reply
`All Challenged Claims Would Have Been Obvious over Queen-1989 or
`Queen-1990 and the PDB Database, Optionally in Light of Tramontano
`and/or Kabat-1987 and/or Hudziak
`A. Claims 12, 42, 60, 65-67, and 71-79 are Obvious Over Queen-1989 or
`Queen-1990, in Combination with the PDB
`
`1.
`
`A POSA Following the Teachings of the Prior Art Would Have
`Identified the Claimed Residues for Substitution
`
`PO faults the Queen references as leading POSAs “to a broad genus of
`
`potential framework substitutions,” and alleges incorrectly that “Petitioner has
`
`provided no reason why a skilled artisan would have selected the specific
`
`framework substitutions recited in the challenged claims.” (POR, 46.) PO does
`
`not deny that a POSA following the prior art would identify the residues in claims
`
`12, 42, 60, 65-67, and 71-79 as candidates for substitution, only that a POSA
`
`would not limit the candidates to those specific residues. But this does not save the
`
`claims.
`
`The ’213 patent purports to identify a laundry list of murine framework
`
`residues that can be substituted for the corresponding human framework residues
`
`to improve binding affinity. However, the patent does not tell a POSA which of
`
`these residues to substitute in a specific humanized mAb project beyond
`
`huMAb4D5, anti-CD3, and anti-CD18. (Ex. 1143, ¶4.) And as Figures 5-6 and
`
`Table 3 of the patent make clear, each humanized mAb requires a different set of
`
`substitutions to optimize binding affinity. (Ex. 1001; Ex. 1143, ¶20.) While the
`
`
`4
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`patent claims a laundry list of possible substitutions to try, the only guidance in the
`
`’213 patent concerning how to select specific substitutions that will work in mAbs
`
`beyond the examples is, as PO’s expert Dr. Wilson acknowledged, remarkably
`
`similar to the guidance in Queen-1989 and Queen-1990. The patent requires a
`
`POSA to construct a computer model, identify the residues that, because of their
`
`positions within the VH and VL domains, could alter binding affinity, and then
`
`conduct trial-and-error mutagenesis to see which substitutions improve binding.
`
`(Ex. 1138, 116:1-122:1; compare Ex. 1050, 12:17-17:24 with Ex. 1001, 20:41-
`
`21:3; Ex. 1142, 76:19-80:13.) Dr. Wilson further conceded that it would have
`
`required nothing more than routine skill and experimentation to use these methods
`
`to identify specific residue(s) that would work in a given humanization project.
`
`(Ex. 1138, 116:1-122:1; see also Ex. 1142, 97:14-98:22.)
`
`As discussed in the Petition regarding Ground 1, a POSA following the
`
`Queen-1989 roadmap and using the PDB would have identified the substitutions at
`
`4L, 58L, 66L, 67L, 69L, 73L, 2H, 45H, and 69H using concededly routine skill.
`
`Since claims 4, 12, 30, 31, 33, 42, 60, 62, 63, 65-67, 69, 71-75, and 78 each merely
`
`require that “one or more” of the listed residues be selected for substitution, they
`
`are invalid. See Ecolochem, 1996 WL 297601, *2 (“[I]f utilizing one element of
`
`the [Markush] group is anticipated or obvious, the patentee is precluded from
`
`5
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`
`arguing that the claim is valid.”).2
`
`PO’s complaint that POSAs would have been faced with a large list of
`
`residues to choose from ignores that the patent likewise requires a POSA to choose
`
`the particular substitutions needed for a given humanization project from the large
`
`list of possibilities recited in the patent. PO has conceded that this kind of
`
`selection would have involved nothing but routine skill, thus it does not weigh
`
`against obviousness. See, e.g., KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 416
`
`(2007) (“The combination of familiar elements according to known methods is
`
`likely to be obvious when it does no more than yield predictable results.”).
`
`The fact that a POSA would have understood from the cited prior art
`
`references that additional residues were also candidates for substitution is also
`
`immaterial. A POSA would have understood that a given humanization project
`
`may require substitutions not described in the patent, and thus that the list of
`
`possible substitutions in the patent is incomplete. (Ex. 1143, ¶¶8-12.) Indeed,
`
`Perjeta® (pertuzumab)—which PO states was made using methods disclosed in the
`
`patent—contains substitutions not described in the patent. (Ex. 1138, 253:18-
`
`254:21.) PO offers no rationale why the selection process required by the patent is
`
`any less complex than the selection process disclosed in the prior art.
`
`2 Dr. Wilson’s opinions applied an improper legal standard, requiring every recited
`
`framework substitution to be obvious. (Ex. 1138, 91:3-92:14.)
`6
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`Additionally, PO does not point to any evidence that the particular
`
`substitutions listed in claims 12, 42, 60, 65-67, and 71-79 are critical to the claimed
`
`invention. The prior art identified a range of potential residues to be substituted in
`
`a humanization project, including the claimed residues. As a result, these residues
`
`are prima facie obvious, and the patentee must establish “that the [claimed residue]
`
`is critical, generally by showing that the claimed range achieves unexpected results
`
`relative to the prior art range.” In re Peterson, 315 F.3d 1325, 1330 (Fed. Cir.
`
`2003) (quoting In re Geisler, 116 F.3d 1465, 1469-70 (Fed. Cir. 1997)). PO has
`
`not made any such showing, and therefore, the selection of the residues listed in
`
`claims 12, 42, 60, 65-67, and 71-79 would have been obvious. (Ex. 1143, ¶¶26-27;
`
`Ex. 1001, Table 3; Ex. 1142, 131:10-132:5.)
`
`2.
`
`A POSA Would Have Used The PDB As Dr. Riechmann Did
`
`PO incorrectly argues that “the Queen references do not teach using the PDB
`
`database as Petitioner uses it.” (POR, 45.) PO does not dispute that the Queen
`
`references teach the use of computer modeling to identify residues that influence
`
`the conformations of the murine CDRs or interact directly with the antigen. (See
`
`Ex. 1003, ¶¶249, 260; Ex. 1034, 10031; Ex. 1050, 14:32-36.) The prior art teaches
`
`the use of the PDB data to do this. (See, e.g., Ex. 1050, 14:32-36; Ex. 1062, 902).
`
`Indeed, a POSA would have had to use data like that contained in the PDB to
`
`create an accurate molecular model. (Ex. 1143, ¶5.) Although PO points out that
`
`7
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`the Queen references describe modeling the murine antibody versus the humanized
`
`antibody, PO ignores that the modeling in Queen-1989 also considered “other
`
`antibody V domains with known crystal structure” when constructing the model
`
`(Ex. 1034, 10031), and Queen-1990 disclosed using “known antibody structures”
`
`in the PDB as rough models (Ex. 1050, 16). PO also does not allege that this
`
`distinction would alter the analysis in any respect, and a POSA would have
`
`recognized that the Queen method is a reliable means of identifying framework
`
`residue substitutions that would improve binding. (Ex. 1143, ¶¶6-7.)
`
`3.
`
`A POSA Would Have Expected the Resulting Humanized mAb
`to Bind the Target Antigen
`
`PO incorrectly argues that there is no evidence that a POSA would have
`
`expected a humanized mAb to bind an antigen.3 The claim language “bind an
`
`antigen” encompasses binding to any degree. PO’s expert admitted that in a
`
`humanization project “one approach to try to regain the binding affinity . . . was to
`
`make additional substitutions back to mouse in the human framework.” (Ex. 1138,
`
`28:2-8.)
`
`PO’s argument ignores all of the prior art discussing successful
`
`
`3 PO makes similar arguments with respect to claims 4, 33, 62, and 69. (POR, 56-
`
`57.) The arguments presented in this section apply with equal weight to those
`
`claims.
`
`8
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`humanization projects. For example, Queen-1989’s humanized anti-Tac mAb
`
`bound the target antigen with high affinity. (Ex. 1034, 10029, 10033 (“The
`
`resulting humanized antibody has a high affinity, 3 x 109 M-1, for its antigen.”).)
`
`Dr. Riechmann’s humanized CAMPATH mAb similarly had high binding affinity.
`
`(Ex. 1069, 3.) A POSA using the methodology set forth in Queen-1990 would
`
`have understood that the prior art methods for humanization had been successfully
`
`used to achieve mAbs with high binding affinity. (Ex. 1050, Abstract (disclosed
`
`antibodies “retain the same affinity as the donor immunoglobulin to the antigen”).)
`
`A POSA using prior art methods would have had a reasonable expectation of
`
`achieving results at least as good as those disclosed in the prior art. (Ex. 1138,
`
`104:12-105:5.) Additionally, the binding affinity of any antibody made according
`
`to the teachings of the prior art would be an inherent property of an obvious
`
`combination and cannot impart patentability. In re Kubin, 561 F.3d 1351, 1357
`
`(Fed. Cir. 2009).
`
`Furthermore, the ’213 patent provides no binding affinity data for most of
`
`the residue substitutions identified in the claims, which indicates either that a
`
`POSA would reasonably have expected antibodies with those substitutions to bind
`
`the target antigens, or that the patent lacks sufficient written description.
`
`
`
`
`
`9
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`Holding the Challenged Claims Invalid Would Not Have
`“Sweeping Consequences”
`
`4.
`
`PO’s hyperbole that finding the challenged claims obvious over the Queen
`
`references and the PDR would remove “patent protection for most if not all
`
`humanized antibodies” is baseless. (POR, 51.) The challenged claims are not
`
`directed to the primary sequences of the handful of specific humanized mAbs that
`
`the named inventors allegedly created. And PO already obtained a patent covering
`
`the variable domain sequences of its humanized 4D5 mAb, but that patent expired.
`
`(Ex. 1144, Claim 15; Ex. 1143, ¶29.) In this patent, PO is trying to reach far
`
`beyond the humanized mAbs the named inventors actually made, to grasp
`
`essentially huge numbers of mAbs made using basic prior-art humanization
`
`techniques pioneered by others. There is no evidence of record to support PO’s
`
`allegation that holding these over-reaching claims obvious would impact the
`
`patentability of different claims to novel humanized mAbs. Regardless, the only
`
`issue before the Board here is the invalidity of the challenged claims.
`
`5.
`
`Petitioner’s Arguments Regarding Queen-1989 Do Not Rely on
`Queen-1990
`
`PO is incorrect that Petitioner’s arguments in Grounds 1, 3, and 6 rely on
`
`Queen-1990 and not Queen-1989 because Dr. Riechmann used a 3.3 Å cutoff for
`
`identifying residues in contact with one another. (POR, 44-45.) Queen-1989 states
`
`that “a number of amino acid residues outside of the CDRs are in fact close enough
`
`10
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`to them to either influence their conformation or interact directly with antigen.”
`
`(Ex. 1034, 10031.) As explained by Dr. Riechmann,4 a POSA reading Queen-1989
`
`would have known that the operative distance would be approximately twice the
`
`interatomic distance—that is, approximately 3.3 Å for most protein atoms. (Ex.,
`
`1003, ¶255 n.17; Ex. 1143, ¶23; Ex. 1145, 261.) PO’s expert agreed,
`
`acknowledging that a POSA would have known that “Van der Waals and
`
`hydrophobic interactions can occur at distances of 3.5 to 4 Angstroms.” (Ex. 2014,
`
`¶184; Ex. 2045.)
`
`6.
`
`Queen-1990 and Tramontano Further Support Petitioner’s
`Position
`
`PO wrongly states that Tramontano “never suggested that substitutions at
`
`position 71H were desirable.” (POR, 15.) Tramontano explicitly disclosed that
`
`residue 71H is likely a “major determinant of the conformation of” the H2 CDR,
`
`thus it would have been obvious to a POSA to consider this residue for substitution
`
`where the murine and human framework residues differed at this position.
`
`(Petition, 7-8, 22, 50; Ex. 1003, ¶132; Ex. 1051, Abstract.)
`
`
`4 PO’s arguments that Dr. Riechmann “simply adopted the opinions” of Dr. Padlan
`
`are misplaced. (POR, 46-47.) Dr. Riechmann conducted his own thorough review
`
`of the relevant art and the ’213 patent and formed his own opinions regarding the
`
`’213 patent. (Ex. 1003, ¶10; Ex. 2039, 29:12-47:2; Ex. 1143, ¶36.)
`11
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`The Consensus Sequence Limitation of Claims 4, 33, 62, 64, and 69
`Was Taught By the Prior Art
`
`B.
`
`As discussed in the Petition, Queen-1990 explicitly discloses the use of a
`
`consensus sequence as the human framework in a humanization project. (Petition,
`
`36-37.) In arguing that Queen-1990’s disclosure of using “many human
`
`antibodies” to make a consensus is different than the claimed approach of using
`
`“all human antibodies,” PO offers no evidence that Queen’s approach was any
`
`different than
`
`
`
` (Ex. 1142,
`
`27:14-28:13, 32:17-20, 35:9-20.) Moreover, a POSA would have been motivated
`
`to use as many different known antibody structures as possible, to make the
`
`sequence as much of a consensus as possible. (Ex. 1143, ¶13-14.) A POSA would
`
`not have understood Criterion I as instructing a POSA to consider only a subset of
`
`human antibodies, since it does not explain what that subset is.
`
`PO’s argument that the Queen-1990 consensus sequence is different than
`
`that disclosed in the ’213 patent is also wrong. PO relies on a passage in Queen-
`
`1990 regarding “a representative collection” of human heavy chains, but this
`
`passage is taken out of context. This quote from Queen-1990 does not refer to the
`
`consensus sequence human framework, but relates to the other potential framework
`
`Queen-1990 describes. (Ex. 1050, 13:3-11; Ex. 1143, ¶15.)
`
`
`
`12
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`PO is also incorrect that a POSA would have understood the “rare” amino
`
`acids language in Criterion II of Queen-1990 as applying to the consensus
`
`sequence in Criterion I. (POR, 53.) Queen-1990 is clear that each of the criteria
`
`may be used separately. (Ex. 1050, 12:12-15 (“These criteria may be used singly,
`
`or when necessary in combination, to achieve the desired affinity or other
`
`characteristics.”); Ex. 1143, ¶16.)
`
`Queen-1989 also teaches the use of a consensus sequence as the human
`
`framework. Queen-1989 describes further humanizing a human antibody
`
`framework by replacing residues that are “unusual” in human antibodies with
`
`“residue[s] much more typical of human sequences . . . to make the antibody more
`
`generically human.” (Ex. 1034, 10032.) A POSA following that teaching, and
`
`informed by the sequence data in Kabat-1987 identifying the residues that are most
`
`“typical” in human immunoglobulins of particular subclasses and subunits, would
`
`have swapped out unusual residues with consensus residues until he or she ended
`
`up with a human consensus framework. (Petition, 51-52; Ex. 1003, ¶310; Ex.
`
`1143, ¶17-18.)
`
`PO’s argument that Queen-1989 does not teach a POSA to start with a
`
`consensus sequence is irrelevant. The ’213 patent claims do not require that a
`
`particular method be used to obtain the human consensus sequence.
`
`PO’s complaint that Queen-1989 replaces unusual human framework
`13
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`residues with murine residues is also misplaced. Queen-1989 describes how
`
`human antibodies exhibit “strong amino acid homology outside of the CDRs,” but
`
`that there are occasionally atypical residues. (Ex. 1034, 10031-32.) During
`
`humanization of anti-Tac, several atypical residues were identified, including 93H,
`
`95H, 98H, 106H, 107H, 108H, and 110H, as well as 47L and 62L, which were
`
`back-mutated to the murine residue. (Ex. 1034, 10032.) According to Queen-
`
`1989, back mutation to the murine residue only occurs when the murine antibody
`
`“has a residue much more typical of human sequences than does” the human
`
`framework. (Ex. 1034, 10032.). This was not done, as PO suggests, to maintain
`
`the confirmation of the murine CDRs, but instead “to make the antibody more
`
`generically human.” (Ex. 1034, 10032.)
`
`That Queen-1989 does not use the word “consensus” is irrelevant. Queen-
`
`1989 renders obvious all that it teaches a POSA, regardless of the terminology
`
`Queen et al. chose to use. See In re Applied Materials, Inc., 692 F.3d 1289, 1298
`
`(Fed. Cir. 2012) (“A reference must be considered for everything that it teaches”).
`
`Similarly, PO is incorrect that Kabat-1987 was only a reference to check the
`
`veracity of a potential sequence. A POSA would have used the information in
`
`Kabat-1987 to determine whether a particular residue was common. (Petition, 22-
`
`23, 51-52; Ex. 1003, ¶309; Ex. 1143, ¶19; Ex. 1142, 30:5-13, 57:23-58:6, 178:18-
`
`179:14; Ex. 1140, 60:3-12.)
`
`14
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`Finally, PO ignores that prior to the earliest possible priority date of the
`
`alleged invention, scientists had already created humanized antibodies using
`
`consensus sequences. (Ex. 1193, 106; Ex. 1138, 196:5-197:6 (discussing
`
`Genentech’s admission during EU patent proceedings that Dr. Riechmann used
`
`Foote’s consensus sequence in humanizing CAMPATH).) This is compelling
`
`evidence that it would have been obvious to use human consensus sequences in the
`
`human framework.
`
`C.
`
`The “Up to 3-Fold More Binding Affinity” Limitation of Claim 65
`Was Taught By the Prior Art
`
`The language “up to 3-fold more binding affinity” in claim 65 means just
`
`that: any improvement up to 3-fold more binding affinity, no matter how small.
`
`(Ex. 1138, 103:23-104:1; Ex. 1142, 118:8-17.) As explained in the Petition, a
`
`POSA optimizing binding affinity through mutagenesis would reasonably have
`
`expected to achieve a humanized antibody with an affinity around that of the
`
`parent antibody, i.e., an affinity that was slightly less or slight more than the
`
`parent. (E.g., Petition, 47-48.) A POSA also would have known that mutagenesis
`
`could lead to improved binding. Indeed, as Table 3 of the patent shows, one
`
`residue substitution can increase the binding affinity during humanization. (Ex.
`
`1001, Table 3; Ex. 1143, ¶26). While the specific binding properties of a
`
`humanized antibody can only be confirmed by testing, such mutagenesis would
`
`
`
`15
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`have been a routine part of antibody humanization. (Ex. 2039, 271:15-17; Ex.
`
`1142, 101:24-102:19; Ex. 1143, ¶20.)
`
`D.
`
`The “Lack Immunogenicity Compared to a Non-Human Parent”
`Limitation of Claim 63 Was Taught By the Prior Art
`
`PO incorrectly implies that claim 63 requires that the humanized antibody
`
`produce no immunogenic reaction, and uses an out-of-context quote from Dr.
`
`Riechmann as support. (POR, 60-61.) But PO does not dispute the Board’s
`
`construction that “lacks immunogenicity” refers to “a humanized antibody having
`
`reduced immunogenicity in a human patient as compared to its non-humanized
`
`parent antibody,” (POR, 18 (emphasis added).) As the Queen references state, the
`
`expectation that humanizing a murine mAb would reduce its immunogenicity was
`
`the very reason why humanization was undertaken in the first place. (Ex. 1034,
`
`10029; Ex. 1050, Abstract; see also Ex. 1138, 102:23-103:5 (“the goal of
`
`humanization is to retain binding affinity and reduce immunogenicity”); Ex. 1140,
`
`82:22-83:4; Ex. 1142, 111:1-6 (“Q: And in order to make a human therapeutic
`
`agent, the humanized antibody would need to lack immunogenicity compared to
`
`the nonhuman parent, right? Otherwise, you would just use the mouse? A:
`
`Agreed.”).) A POSA would thus have expected that making the antibody more
`
`human would make the antibody less immunogenic. (Ex. 1142, 112:5-9, 112:16-
`
`21 (“Q: So the fact that there were fewer mouse residues in the humanized variant
`
`
`
`16
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`versus the parent led to an expectation that is would lack immunogenicity
`
`compared to the parent, right? . . . A: Yes.”); Ex. 1143, ¶24.) Indeed, the patent
`
`does not contain any immunogenicity data, which underscores that the named
`
`inventors expected that the humanized mAbs would not be as immunogenic as the
`
`murine parent mAbs. Moreover, merely testing the immunogenicity of an
`
`otherwise obvious mAb does not render it patentable. See Santarus, Inc. v. Par
`
`Pharm., Inc., 694 F.3d 1344, 1354 (Fed. Cir. 2012).
`
`E.
`
`The Limitations Related to Binding p185HER2 of Claims 30-31, 33, 42,
`and 60 Was Taught By the Prior Art
`
`PO argues, in essence, that because there is no disclosure in the prior art of a
`
`humanized 4D5 mAb, such an antibody cannot be obvious. (POR, 62-63.) This
`
`argument misapplies the law of obviousness. PO’s expert admitted that, prior to
`
`the date of the alleged invention, it was known that overexpression of p185HER2 led
`
`to a poor prognosis in cancer, and work had been done to identify antibodies that
`
`would target p185HER2. (Ex. 1138, 19:7-20:1.) Also prior to the alleged invention,
`
`Hudziak identified that the murine antibody 4D5 downregulated p185HER2. (Ex.
`
`1021, 1169; Ex. 1004, ¶14; Ex. 1140, 22:1-24:7; Ex. 1138, 19:23-20:25, 22:8-12.)
`
`This would have motivated a POSA to use the humanization framework of Queen-
`
`1989 and apply that framework to 4D5 to make a therapeutic agent that would
`
`downregulate p185HER2. (Ex. 1004, ¶16; Ex. 1021; Ex. 1143, ¶28.) Dr. Leonard
`
`
`
`17
`PROTECTIVE ORDER MATERIAL
`
`
`
`
`IPR2017-01373
`Petitioner’s Reply
`confirmed that a POSA would have understood 4D5 to be a promising target for
`
`humanization, and the prior art taught a POSA how to humanize the antibody to
`
`generate a potential therapeutic agent that bound p185HER2. (Ex. 1004, ¶¶48-49.)
`
`As explained above, the prior art humanization process would have identified the
`
`potential residue substitutions identified in the ’213 patent, rending claims 30-31,
`
`33, 42, and 60 obvious.
`
`II.
`
`PO’s Attempts to Establish an Earlier Date of Invention for Claims 12,
`42, 60, 65, 71, 73-74, and 79 Are Insufficient
`
`
`
`
`
`
`
`Here, the default date of invent