dissenting.
Today, the court overrides a jury verdict of infringement based on a tenuous obviousness analysis. It recognizes that the Stumpo paper only refers to the use of known substances and presumes that, to find infringement, the jury must have implicitly found that there was no motivation in Stumpo to utilize the disclosed cells with compounds not previously known to interact with the cell surface proteins. Based on the state of knowledge in the art that cells with heterologous cell surface proteins were ideal candidates for drug screening methods, the court then concludes that the jurys implicit finding is not supported by substantial evidence.
In reality, the court relies on the combination of Stumpo, Lester, and Sucov to establish that the use of heterologous cells with reporter gene constructs was known in the art to be useful in drug screening methods. Stumpo discloses cells identical to the 629 patent claims, but does not mention their use to test unknown compounds as possible drugs. Lester describes the utility of heterologous cell surface proteins for drug testing, but does not mention cells with reporter gene constructs, which are central to the method of testing of the 629 claims. Sucov was not even argued at trial, where Cadus argued that either Stumpo or Chen renders the 629 patent obvious. This analysis is inconsistent with the courts stated conclusion that the 629 patent is obvious over the Stumpo reference alone in view of the prior art as argued by Cadus. It fails to recognize that the 629 patent includes only method claims; ■ Sibia disclaimed all claims *1360to the cells themselves when Stumpo was brought to its attention.
The court is making an end-run around the requirement that there must be a motivation to modify the reference along the path taken by the 629 patent. See Kolmes v. World Fibers Corp., 107 F.3d 1534, 1541, 41 USPQ2d 1829, 1833 (Fed.Cir.1997) (Invention was not obvious where there was no suggestion or motivation to modify teaching of reference.). It combines a series of references not specifically argued to the jury to conclude that no reasonable jury could possibly find the absence of motivation in the prior art to modify the Stumpo paper to render the 629 patent obvious. Without citing any motivation to modify in any of the series of references, the court improperly concludes that it would have been unreasonable for the jury to find as a matter of fact that there was no such motivation. See Tec Air, Inc. v. Denso Mfg. Michigan, Inc., 192 F.3d 1353, 1359, 52 USPQ2d 1294, 1297-98 (Fed.Cir.1999) (Whether a reference provides a motivation to combine its teachings with other references is a question of fact underlying the legal determination of nonobviousness that we assume the jury resolved in favor of the verdict winner and leave undisturbed if it is supported by substantial evidence.).
The district court properly rejected Ca-dus motion for judgment as a matter of law, holding that there was substantial evidence to support a verdict of nonobviousness because the 629 patent was a combination of factors that was not apparent to a person of ordinary skill in the art. The trial court found additional support for the jurys verdict in evidence of secondary considerations of long-felt need and commercial success of the 629 patent. These are factual underpinnings of the legal conclusion of nonobviousness that the jury presumptively resolved in favor of Sibia because substantial evidence supported them. See id., 192 F.3d 1353, 52 USPQ2d at 1298.
This court improperly rejects this substantial evidence. It opens the door for accused infringers to string together a series of references, which collectively contain the elements of an apparatus (here, the cell with a heterologous cell surface protein and reporter gene construct) and various suggestions for the use of those separate references. It then would allow an inference of motivation to modify a single reference to render obvious a method claim for utilizing the' apparatus. All this is in violation of the well-settled mandate requiring a motivation to alter a single reference or to combine multiple references to render the claims of a patent obvious. See, e.g., id. at 1359, 192 F.3d 1353, 52 USPQ2d at 1298 (motivation to combine multiple references); B.F. Goodrich v. Aircraft Braking Sys. Corp., 72 F.3d 1577, 1582, 37 USPQ2d 1314, 1318 (Fed.Cir.1996) (motivation to modify a single reference); Grain Processing Corp. v. American Maize-Products Co., 840 F.2d 902, 907, 5 USPQ2d 1788, 1792 (Fed.Cir.1988) (“Care must be taken to> avoid hindsight reconstruction by using ‘the patent in suit as a guide through the maze of prior art references, combining the right references in the right way so as to achieve the result of the claims in suit.’”) (internal citation omitted); In re Fine, 837 F.2d 1071, 1075, 5 USPQ2d 1596, 1600 (Fed.Cir.1988) (“One cannot use hindsight reconstruction to pick and choose among isolated disclosures in the prior art to deprecate the claimed invention.”). The court has substituted itself for the jury, reweighed the evidence, and combined references that were not before the jury. I would sustain the jury’s verdict.