Biogen, Inc. v. Amgen Inc., 115 F. Supp. 2d 139 (D. Mass. 2000)
September 25, 2000
United States District Court, D. Massachusetts.
*140 John Sylvia, Patrick T. Clendenen, Mintz, Levin, Cohn, Ferris, Glovsky & Popeo, P.C., Boston, MA, Eric R. Hubbard, Fish & Neave, New York, NY, James F. Haley, Kenneth B. Herman, Jane A. Massaro, Kathleen M. Walker, Fish & Neave, New York, NY, for plaintiff.
Thomas R. Murtagh, Mintz, Levin, Cohn, Ferris, Glovsky & Popeo, P.C., Boston, MA, Jan Van Gysegem, Cleary, Gottlieb, Steen & Hamilton, New York, NY, Paul F. Ware, Eileen M. Herlihy, Goodwin, Procter & Hoar, Boston, MA, Karen J. Kramer, Robert M. Galvin, Ricardo Rodriguez, Gary H. Ritchey, Darren B. Mitchell, Karen A. Gibbs, Cooley Godward LLP, Palo Alto, CA, Steven M. Odre, Karol M. Pessin, Thousand Oaks, CA, Lloyd R. Day, Jr., Vernon M. Winters, Day Casebeer Madrid Winters & Batchelder LLP, Cupertino, CA, for defendant.
*141 MEMORANDUM AND ORDER ON (1) AMGEN'S MOTION FOR SUMMARY JUDGMENT OF NON-INFRINGEMENT (DOCKET # 861) AND (2) AMGEN'S MOTION FOR SUMMARY JUDGMENT OF NON-EQUIVALENCE (DOCKET # 882)
STEARNS, District Judge.
In an August 6, 1998 Markman order (docket # 854), claim 1 of the '702 patent was construed as follows.
To be covered by the patent, a plasmid vector must contain the entire PLOL of bacteriophage λ as represented in Figure 6 of the patent and at least one endonuclease recognition site inserted at the converted Hae lll site at 73.1% of bacteriophage λ or at another site downstream of Hae lll, said endonuclease recognition site being within 300 base pairs of the Hinc ll site at -33, and prior to any sequences of λ DNA downstream of the Hae lll site.
Based on this construction, defendant Amgen, Inc., moves for summary judgment, contending that its accused vector neither literally infringes Biogen, Inc.'s '702 patent, nor infringes under the doctrine of equivalents.
A comparison of the material differences between the structure of Amgen's plasmid vector and that of the vectors claimed by Biogen in the '702 patent establishes literal non-infringement. As described in Figure 6 of the '702 patent the prototypical Biogen vector consists of 247 base pairs of bacteriophage λ DNA spanning the sequence beginning at the Sau 3A site at -133 and extending to the converted Eco RI (Hae lll) endonuclease recognition site at + 114. The sequence contains five structural elements: (1) the leftward promoter PL; (2) the leftward operator OL; (3) a transcription start site (+1); (4) the N utilization site (nut L) (+34 to +63); and (5) the Hae lll site. Figure 6 also shows a 114 base pair DNA fragment terminating at the Hae lll site, the earliest point at which translation occurs.
The Amgen vector by contrast, while initiating at Sau 3A, terminates at the Xba l site at +13. The Amgen vector thus does not contain nut L or the Hae lll site. As opposed to the 114 base pair DNA fragment depicted in Figure 6, the DNA fragment in Amgen's vector consists of 13 base pairs. The Hae lll recognition site depicted in Figure 6 is inserted outside of the PLOL sequence, while Amgen's endonuclease recognition site is inserted within PL OL. And finally, Amgen's vector inserts non-λ DNA in PLOL between base pair positions +4 and +13. Because there can be no dispute that Amgen's vector does not contain the entire sequence described in Figure 6 of the '702 patent, under the court's Markman construction Amgen is entitled to summary judgment on the claim of literal infringement. See Litton Systems, Inc. v. Honeywell, Inc., 140 F.3d 1449, 1454 (Fed.Cir.198) ("Literal infringement requires that the accused device contain each limitation of the claim exactly; any deviation from the claim precludes a finding of literal infringement.")
DOCTRINE OF EQUIVALENTS
A product or composition may not literally read on a patent claim and may yet infringe under the doctrine of *142 equivalents. Warner-Jenkinson Co., Inc. v. Hilton Davis Chemical Co., 520 U.S. 17, 25, 117 S. Ct. 1040, 137 L. Ed. 2d 146 (1997). The doctrine of equivalents is not above reproach, given its propensity to assume "a life of its own, unbounded by the patent claims." Id., at 28-29, 117 S. Ct. 1040. "In reconciling the uncertainty surrounding application of the doctrine of equivalents with the definitional and public-notice functions of the statutory claiming requirement, the Supreme Court [in Warner-Jenkinson] endorsed an element-by-element analytical framework for infringement." Litton Systems, 140 F.3d at 1454. "Each element contained in a patent claim is deemed material to defining the scope of the patented invention, and thus the doctrine of equivalents must be applied to individual elements of the claim, not to the invention as a whole." Warner-Jenkinson, 520 U.S. at 29, 117 S. Ct. 1040.
While Graver Tank & Mfg. Co. v. Linde Air Products Co., 339 U.S. 605, 70 S. Ct. 854, 94 L. Ed. 1097 (1950), set out what has become the familiar "triple identity" test of equivalence, comparing "the function served by a particular claim element, the way that element serves that function, and the result thus obtained by that element, ... the particular linguistic framework used is less important than whether the test is probative of the essential inquiry: Does the accused product or process contain elements identical or equivalent to each claimed element of the patented invention?" Warner-Jenkinson, 520 U.S. at 39-40, 117 S. Ct. 1040.
The first task in applying the "all-elements" rule is to identify the material limitations of the disputed claim. An illustration is provided by Litton Systems. There the district court construed the specification of a "Kaufman-type ion beam source" to include "[a]ny broad-beamed, multiapertured, gridded ion beam source, which includes any hollow cathode gun and any radio frequency gun." Id. at 1454. The Federal Circuit disagreed. Looking more closely to the specification, the Court identified the components of an ion beam gun of the "Kaufman-type" to include "a hot-wire cathode, an anode, grids, and magnets." Because the Litton Systems jury had been instructed to consider whether the accused product incorporated a generic ion beam source rather than an ion beam gun with the essential components of the Kaufman-type gun described in the patent, its finding of infringement by equivalents could not stand under the all-elements rule. Id. at 1455.
Infringement under the doctrine of equivalents is ordinarily a question of fact. Bayer AG v. Elan Pharmaceutical Research Corp., 212 F.3d 1241, 1251 (Fed.Cir. 2000). "Thus summary judgment may be granted [only] when no material fact is in dispute, or when no reasonable trier of fact could find facts whereby the nonmoving party could prevail, even when all justifiable factual inferences are drawn in favor of the nonmovant." Canton Bio-Medical v. Integrated Liner Technologies, Inc., 216 F.3d 1367, 1369 (Fed.Cir.2000). As will be noted, the material facts necessary for a resolution of this case are not in dispute, although many other facts are.
Turning to claim 1 of the '702 patent, it is not enough, as Warner-Jenkinson and Litton Systems teach, to simply say that it describes a bacteriophage λ *143 DNA vector with a leftward promoter and operator. More precisely, as depicted in Figure 6, the constituent structures of Biogen's vector are: (1) the leftward promoter (PL); (2) the three leftward operator sequences (OL1, OL2, and OL3); (3) a transcription start site; (4) the N utilization site (nut L); and (5) an endonuclease (Hae lll) recognition site. That there are substantial similarities between Biogen's and Amgen's vectors is self-evident. Both rely on PLOL to mediate gene expression. Both contain the components essential for PL to function, specifically the -35 region, the -10 region (the Pribnow box), the +1 transcription start site, and the three OL sequences, all identically spaced. Both use an artificially constructed endonuclease recognition site to insert the gene to be expressed. Both locate that site within 300 base pairs of Hinc ll to avoid transcription of contiguous downstream λ DNA. Where the vectors differ is that Biogen's contains the nut L sequence (from +34 to +63), while Amgen's does not. Whether this is a substantial difference requires an explanation of what the inventors (Fiers and Remaut) intended by including nut L in their claimed vectors.
As Biogen's Dr. Losick pointed out, nut L was the only DNA sequence downstream of +1 in Figure 6 "that was known [at the time the '702 patent application was filed] to have an effect, under some circumstances, on expression." Losick Decl. ¶ 67. The "circumstances" that Dr. Losick referred to are those in which either the host cell or the plasmid vector contains an active N gene. When either does, nut L interacts with the N gene to override downstream termination sequences that would otherwise abort transcription. Chamberlain Decl., Ex. B ¶ 63. Fiers explained that he and Remaut were concerned that unanticipated termination sites in cloned eukaryotic genes might preclude expression if the N gene were present. "It was not known, and this is precisely the reason that we anticipated possible transcriptional termination regions in the eukaryotic gene and therefore we developed the system which could be used plus or minus the N gene." Fiers Dep., June 11, 1997, at 1550-1551. Fiers and Remaut did so by placing the endonuclease recognition site at Hae lll, downstream of nut L. The ability of the '702 vector to express DNA despite the presence of an active N gene in the *144 host cell was specified in the patent as one of its attributes.
The preferred vectors of this invention are also those in which active N genes ... are absent. Therefore, by choice of an appropriate host, i.e., one containing or lacking an active chromosomal N gene, any of the vectors of the invention may be employed for expression of DNA sequences in the presence or in the absence of the N gene product.
'702 patent, col. 4:26-32.
It is undisputed that nut L is present in every vector claimed by Biogen. See Amgen's Statement of Facts [docket # 885] ¶ 41 and Biogen's Statement of Facts [docket # 895] ¶ 41. And it is equally undisputed that Amgen's vector does not contain nut L or use host cells that express the N gene. Id. at ¶ 46; Losick Decl. ¶ 67. In the Fiers and Remaut vectors the presence or absence of the N gene in the host cell is irrelevant because of nut L. The same is not true of Amgen's accused vector. Thus, Biogen's argument that "[a]ll of the proteins expressed in the patents-in-suit were expressed in N-minus systems (in which nut L is inactive), and two were expressed exclusively so," simply underlines the ability of the claimed vector to achieve expression when the N protein is present. Biogen's Statement of Additional Facts, [docket # 895] ¶¶ 17, 18. As Amgen accurately points out, "even Biogen admits that there are times when nut L is part of PLOL function. Either nut L is part of the invention or it is not. And if it is part of the invention, even sometimes, it is part of the invention for equivalents purposes." Amgen Reply [docket # 902], at 39. Biogen's refrain that because "the Amgen host/vector system neither contains the N gene nor produces N protein, ... the presence or absence of nut L in the Amgen system [is] irrelevant," simply proves Amgen's point, that its vector lacks one of the structural components of Figure 6.
Whether that difference is material depends on whether nut L in fact performs a function in Biogen's claimed vectors. (Merely eliminating a redundant or unnecessary element from a composition to invent around a patent is what the doctrine of equivalents is intended, at least in part, to prohibit). Here, there is no question but that nut L performs an essential function in the '702 vector as both Fiers and Remaut explained in the patent (as quoted above) and in their depositions.
Q. Now in order to achieve that overriding of the transcriptional termination that you are referred to, your PL containing plasmid vector encodes a site known as N utilization, correct?
A. That is correct, yes.
Q. And your PL containing plasmid vector encodes that site in the DNA such that that site is transcribed into a hairpin loop like structure in the RNA emanating from your plasmid structure, correct?
A. That's correct, yes.
Q. And by the existence of that N utilization structure, the N gene protein interacts with the structure on the mRNA so as to allow for an overriding *145 or an antitermination effect, correct?
A. Yes, that's correct.
Q. So, in that sense, that is the provision in your PL containing vector of a N utilization site, is it the case that your vector provides for a solution to unintended transcription stop signals occurring between the promoter and the structural gene sequence?
A. The vector has all of the necessary features to make use of the presence of an active N protein. And in that paper I was referring to, and also on other instances, we suggest to use a strain which applies the N protein entrance.
Remaut Dep., Sept. 12, 1997, at 1551-1552.
Q. Now, is it your understanding at the present time that the [nut L] site is needed for the antitermination action of the N gene product?
Q. Was it your understanding in September of 1980 that the [nut L] site was needed for the antitermination action of the N gene product?
Fiers Dep., January 23, 1997, at 887.
As Dr. Losick stated in his Declaration: before Fiers and Remaut did their work, scientists believed that it was useful to have a functional N gene co-transcribed by the PL promoter to achieve high levels of expression (i.e., the PL-initiated transcript contained both the N gene coding sequence and the coding sequence of the gene of interest). For example, Bernard et al., "Construction Of Plasmid Cloning Vehicles That Promote Gene Expression From The Bacteriophage Lambda PL Promoter," 5 Gene 59-76 (1979 ...), which was extensively discussed in the prosecution history, demonstrated greater PLOL-driven expression from vectors that included a co-transcriptional, functional N gene as compared to vectors which did not include a functional N gene. Hedgpeth et al., "Lambda Phage Promoter Used To Enhance Expression Of A Plasmid-Cloned Gene," 163 Molec. Gen. Genet. 197-203 (1978 ...), found that the "λ N product was necessary to overcome transcriptional termination sites between λ PL and the end of [the cloned gene]."
A skilled person who was familiar with Fiers and Remaut's invention would have understood that Fiers and Remaut took a different approach. They placed the desired gene to be expressed upstream of the N gene coding sequence. A skilled artisan would have recognized that this difference distinguished the Fiers and Remaut vectors from those of the prior art.
Id. ¶¶ 78, 79.
Morris (the inventor of Amgen's vector), on the other hand, deliberately eliminated nut L by placing the desired endonuclease recognition site prior to the nut L sequence.
And in that there are Nut sites that are very active and there is RNA's three sites that are very active. And those sites lent a great deal of structural uncertainty. And so what I set out to do when I was creating my systems was to totally eliminate them. And so that's what I indicated in this. They were completely removed because I did not want to I did not want those sequences present in my constructs. So I removed them.
Morris Dep., August 8, 1996, at 731. The result is a vector, that in the sense that Fiers and Remaut improved on the work of Horn and Wells and others by, among other things, placing the recognition site upstream of the N coding sequence, improved *146 on Fiers and Remaut by eliminating what was perceived in 1989 (when the infringement is alleged to have occurred) as another potential source of downstream instability in the λ phage vector. See August 6, 1998 Memorandum and Order, at 10-11.
There is a remaining issue. "Prosecution history estoppel provides a legal limitation on the application of the doctrine of equivalents by excluding from the range of equivalents subject matter surrendered during prosecution of the application for the patent." Cybor Corp. v. FAS Technologies, Inc., 138 F.3d 1448, 1460 (Fed.Cir.1998) (en banc). "Prosecution history estoppel can occur as a result of (i) amendments made to overcome patentability rejections or (ii) arguments made during prosecution that show `a clear and unmistakable surrender of subject matter.'" Bayer, 212 F.3d at 1251. The application of prosecution history estoppel (as its name implies) is a matter of law. Cybor, 138 F.3d at 1460.
Amgen's estoppel argument centers on Biogen's efforts, undertaken in 1988, to overcome the PTO's objection that the '702 patent was descriptively inadequate. Biogen offered a "compromise" to expedite allowance of the patent by amending its claims "to recite the specific structure of the plasmid vectors of the application." According to Biogen, the new Hae lll limitation meant that "each plasmid of the application contains the PL OL-Eco RI modified λ region of the pPLa23 as depicted in Figure 6." June 22, 1988 Amendment, at 8. The fact that Biogen's amendment was intended to cure a descriptive inadequacy is significant because prosecution history estoppel is usually understood to apply to amendments made to avoid prior art. As Justice Thomas noted in Warner-Jenkinson, the PTO often requests changes in claim language "without the intent to limit equivalents or, indeed, with the expectation that language it required would in many cases allow for a range of equivalents." Id., 520 U.S. at 32, 117 S. Ct. 1040. "Where the reason for the change was not related to avoiding the prior art, the change may introduce a new element, but it does not necessarily preclude infringement by equivalents of that element." Id. at 33, 117 S. Ct. 1040. (As Biogen argues, "[e]ach of the Federal Circuit cases cited by Amgen found estoppel only when an amendment or argument was made in connection with an outstanding prior art rejection." Biogen Memorandum [docket # 923], at 3).
Various Federal Circuit opinions do, however, more broadly imply that a patentee's "clear and unmistakable" surrender of subject matter during the prosecution of a patent application will work an estoppel even though the concessionary statement is not made to overcome prior art. See, e.g., Sextant Avionique S.A. v. Analog Devices, Inc., 172 F.3d 817, 828 (Fed.Cir. 1999) ("On the other hand, if the claims were amended for a reason that was not `related to patentability,' prosecution history estoppel does not apply absent a `clear and unmistakable surrender' of certain subject matter") (emphasis added); Bayer, 212 F.3d at 1252 ("Unmistakable assertions made by the applicant to the .. PTO ... in support of patentability, whether or not required to secure allowance of the claim, ... may operate to preclude the patentee from asserting equivalency.") The issue of just how broadly the doctrine of patent history estoppel sweeps was squarely framed by the Federal Circuit in granting a rehearing en banc in Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., Ltd., 187 F.3d 1381 (Fed.Cir.1999).
The following questions may be addressed in the briefs:
1. For the purposes of determining whether an amendment to a claim creates prosecution history estoppel, is "a substantial reason related to patentability," Warner-Jenkinson Co. v. Hilton Davis Chem. Co., 520 U.S. 17, 33, 117 S. Ct. 1040, 137 L. Ed. 2d 146 (1997), limited to those amendments made to overcome prior art under § 102 and § 103, *147 or does "patentability" mean any reason affecting the issuance of a patent?
If the Federal Circuit answers this question in the way I am inclined to think it will, estoppel would provide an additional or alternative ground on which summary judgment should enter for Amgen. To overcome the examiner's rejection Biogen told the PTO:
[a]s stated in the specification, the vectors of the prior art are disadvantaged by the great distance "between the sites available for insertion of cloned genes and the PL promoter." Applicants' vectors and methods overcome the problem confronted by the prior art by introducing "an Eco RI site at a short distance downstream from PL." Further, as disclosed in the specification as originally filed, applicants introduced the Eco RI site by converting the "single Hae lll site (73.1% λ, Figure 1) located about 150 nucleotides downstream from PL." Specification page 5, lines 19-25; page 6, lines 23-26; page 16, line 31 to page 17, line 11.
June 22, 1988 Amendment, at 7.
Based on this teaching, Biogen assured the PTO that "a person of skill in the art, employing the methods disclosed in the specification as filed, would not place a recognition site at any location other than in the region identified more precisely by the substitute claims.... Each of the plasmid constructions of this application utilizes the same region of phage λ and the same operative site, Hae lll (reconstructed to Eco RI)." Id., at 8-9. Thus, "each plasmid of the application contains the PL OL-Eco RI modified λ region of the pPLa23, as depicted in Figure 6." Id. at 8. From this language, a competitor would reasonably believe that Biogen had surrendered any claim to vectors lacking the elements of the Figure 6 sequence and, in particular, with recognition sites placed after or before the Hae lll region specified in claim 1. See Bayer, 212 F.3d at 1253.
For the foregoing reasons, Amgen's Motion for Summary Judgment on Literal Infringement [docket # 861] is ALLOWED. Amgen's Motion for Summary Judgment under the Doctrine of Equivalents [docket # 882] is also ALLOWED. Amgen's remaining pending dispositive motions [docket 639, 643 and 739] are consequently MOOT.
 A plasmid vector is a DNA molecule, which can be replicated when placed in a host cell, with or without an inserted gene.
 The structure of Amgen's vector is not in dispute.
 The promoter binds the RNA polymerase which transcribes the gene into messenger RNA (mRNA).
 An "operator" is a DNA sequence, which by binding (or unbinding) with a repressor protein, turns the promoter off or on. There are three such sequences associated with the λ PL promoter, designated as OL1, OL2, and OL3.
 An endonuclease recognition site is a naturally occurring or engineered portal in a DNA sequence into which another DNA fragment, including a gene whose expression is desired, can be inserted.
 The specification language construed by the Federal Circuit was as follows.
The ion beam gu[n] 4 is a commercially available ion [e]mitting ap[p]aratus generally known in the art as a Kaufman [sic] type ion beam gun. The gun's cathode 6 is a therm[i]onic emitter, i.e., it emits electrons by passing an electric current through it which heats the wire. The cathode 6 emits electrons which are accelerated towards the anode 8. The electrons being accelerated from the cathode to the anode strike argon atoms and in so doing dislodge electrons from the argon. The results are positively charged argon ions which are accelerated away from the anode and towards the grids 12 and 14. Permanent bar magnets 10 attached to the anode introduce a magnetic field into the area between the cathode and the anode....
Litton Systems, 140 F.3d at 1453-1454.
 By use of the word "structure," I am not suggesting that I endorse the "structure equals function" test advocated by Judge Lourie in his concurring opinion in Genentech, Inc. v. The Wellcome Foundation Ltd., 29 F.3d 1555 (Fed.Cir.1994). Thus, the fact that there is a "45 percent difference" in the structure of the two sequences (when a raw comparison of the numbers of base pairs is made), I do not deem material. I am referring rather to the structures of the composition depicted in Figure 6 of the '702 patent which comprise the elements of the claimed vector.
 It is not disputed that Amgen's vector is identical to the Figure 6 sequence from base pair position -130 through position +3.
 While both vectors use the same cl857 repressor to regulate expression, this is not a structural element, but rather an attribute of OL.
 While Biogen is correct that the law does not require "an equivalency of components," but only an equivalent for each limitation of the claim, see Biogen Surreply [docket # 909], at 38, nut L is clearly identified in Figure 6 of the '702 patent as an element of the claimed composition, and is present in every claimed vector. See Corning Glass Works v. Sumitomo Elec. U.S.A., Inc., 868 F.2d 1251, 1259 (Fed.Cir.1989) (Under the all elements rule, "[a]n equivalent must be found for every limitation of the claim somewhere in an accused device, but not necessarily in a corresponding component, although that is generally the case.")
 Dr. Remaut believed in 1981 that nut L might have an effect on transcription even when the N protein was absent. See Remaut et al., Plasmid vectors for high-efficiency expression controlled by the pL promoter of coliphage lambda, Gene 15:81 (1981), cited in Amgen Reply [docket # 902], at 39. Dr. Remaut's theory is set out in the patent at Table II and col. 20: 32-43.
 Eukaryotic refers to the large universe of life forms based on cellular structures containing nuclei.
 While Biogen points out that the patent specification does not mention nut L, but only describes "the N gene and its anti-termination effects," Biogen Surreply [docket # 909], at 39*, it is clear from their depositions that Fiers and Remaut understood that these effects were a consequence of the interaction between the N gene and nut L. See, e.g., Fiers Dep., January 22, 1997, at 887. The nut L site is, of course, referenced in Figure 6 of the patent.
 More precisely, Biogen states that "nut L may have a role in anti-termination of transcription if active N protein is present." Id. ¶ 18 (emphasis added). This is a bit disingenuous. Biogen's own expert, Dr. Losick, teaches that it does. See Losick Decl. ¶ 75 ("In bacteriophage λ, N protein coded for by the N gene, interacts at the nut L site to modify the RNA polymerase in such a way that the modified RNA polymerase will not recognize specific downstream transcription termination sites.")
 See, e.g., Losick Decl. ¶ 76.
 All prior art vectors had insertion sites in or downstream of the coding region of the N gene.
 Amgen's vector (pCFN536) was specifically constructed to express the protein G-CSF. There is no dispute that at least one of Biogen's claimed vectors is capable of similar expression.
 The importance of the way the question is framed lies in the fact that while Biogen has demonstrated, consistent with its burden under Warner-Jenkinson, 520 U.S. at 33-34, 117 S. Ct. 1040, that its amendment was made to overcome a descriptive inadequacy and not to avoid prior art, there can be no doubt that the amendment was also offered to overcome objections to the '702 vector's "patentability." See June 22, 1988 Amendment, at 6 ("[A]pplicants, in the spirit of trying to reach some compromise and expedite allowance of this application, have amended the claims to recite the specific structure of the plasmid vectors of the application and, thus, overcome this rejection.")
 As Amgen succinctly, and to my mind, accurately states, the Hae lll amendment not only specified "where to locate the endonuclease recognition site, but it also ... specifies where not to locate the claimed site." Amgen Memorandum [docket # 924], at 8. Were it not for the categorical statement in the June 22, 1988 Amendment, at 8-9, that "[t]he application does not refer to any site in the natural λ sequence `a short distance downstream from PL' other than the Hae lll site," I might agree with Biogen's argument that it intended to surrender only vectors with sites more than 300 base pairs downstream of Hinc ll and within the N gene coding sequences downstream of Hae lll. Biogen Memorandum [docket # 923], at 12.