US 4889818

I'll run a couple of targeted searches to check current docket/litigation status, then summarize.

Let me pull the actual claim set for this patent to ground the claim summary.

Let me try to retrieve the actual claim language of the '818 patent itself (many search hits above were for later continuations).

US Patent 4,889,818 — Analyst Summary

Important scoping note up front: I searched for this patent number specifically and did not find any 2026 CAFC docket, appeal, or other 2026 proceeding involving US 4,889,818. Because the patent expired (anticipated expiration 2006‑12‑26), that absence is expected rather than surprising. The litigation activity on this patent is historic (1992–2008). I also could not retrieve the verbatim printed claim text from an authoritative full‑text source in this session, so the claim discussion below is grounded in corroborating documents and is flagged where uncertain.

1. Bibliographic data

Inventor-name caveat (per the strict no-autocorrect rule): the patent face and Google Patents list "Frances C. Lawyer," whereas the recorded assignment paperwork gives "Francis C. Lawyer." I have not resolved this discrepancy and am reporting both literally.

Source: https://patents.google.com/patent/US4889818/en ; assignment record per https://www.plainsite.org/patents/assignment.html?id=[417143](/patent/417143)

2. Abstract (as printed)

"A purified thermostable enzyme is obtained that has unique characteristics. Preferably the enzyme is isolated from the Thermus aquaticus species and has a molecular weight of about 86,000‑90,000 daltons. The thermostable enzyme may be native or recombinant and may be used in a temperature‑cycling chain reaction wherein at least one nucleic acid sequence is amplified in quantity from an existing sequence with the aid of selected primers and nucleotide triphosphates. The enzyme is preferably stored in a buffer of non-ionic detergents that lends stability to the enzyme."

3. What the disclosure covers (plain language)

The specification describes a purified Taq DNA polymerase — the thermostable enzyme from Thermus aquaticus — produced by a six‑step purification (crude extract → ammonium‑sulfate fractionation → DEAE‑cellulose → hydroxyapatite → DEAE‑cellulose → phosphocellulose), with an added Example VI preferred protocol using Polymin P precipitation followed by phosphocellulose, heparin‑Sepharose, hydroxyapatite, DEAE‑Tris‑Acryl and CM‑Tris‑Acryl steps. Reported native molecular weight: ~86,000–90,000 Da by SDS‑PAGE (calculated ~94,000 Da from predicted sequence; specification elsewhere frames the full‑length enzyme as 86,000–95,000 Da depending on method).

Other disclosed subject matter:

• Cloning of the Taq pol gene from a λgt11 expression library screened with rabbit polyclonal anti‑Taq antiserum; full gene assembled from a ~750 bp BglII–HindIII fragment of pFC83 plus a ~2.8 kb HindIII–Asp718 fragment of pFC85.
• Expression constructs/vectors (pLSG1, pLSG2, pLSG5–pLSG10, pFC82, pFC83, pFC85, phage CH35:Taq#4‑2) and hosts (E. coli DG98, DG116; yeast, insect, mammalian, plant systems discussed).
• Stabilized storage formulation using non‑ionic polymeric detergents (NP‑40 / ethoxylated nonyl phenol and Tween 20 / polyoxyethylated sorbitan monolaurate), optionally with glycerol and gelatin — described as necessary because activity was lost in detergent‑free buffers.
• PCR amplification protocols using the enzyme (annealing at 45–58 °C, extension ~72 °C, denaturation ~94 °C, no reagent transfer between cycles), including comparisons to Klenow fragment showing improved specificity.

Deposits referenced: ATCC 40336 (bacteriophage CH35:Taq#4‑2), ATCC 67422 (pFC83), ATCC 67421 (pFC85), ATCC 39768 (E. coli DG98), ATCC 53606 (E. coli DG116), ATCC 25104 (T. aquaticus YT1).

4. Independent claims — plain-language overview

⚠️ Uncertainty flag: the claim text was not retrieved verbatim here. The characterization below is reconstructed from (a) the specification's Summary, (b) the Fed. Cir. opinion's discussion of claim 1's molecular-weight recitation, and (c) contemporaneous expert declarations addressing "claims 1 and 2." Treat the wording as indicative, not quoted.

The '818 patent is widely reported as containing three claims (one directed to native Taq polymerase, one to the recombinant form). A contemporaneous Nature Biotechnology report stated: "The patent includes three claims, one of which is for the native Taq and another for recombinant Taq."

• Claim 1 (native polymerase) — independent. A purified thermostable DNA polymerase from Thermus aquaticus that catalyzes incorporation of nucleotide triphosphates to form a strand complementary to a template, characterized by a molecular weight of about 86,000–90,000 daltons (SDS‑PAGE relative to the specified marker set: phosphorylase B 92,500; BSA 66,200; ovalbumin 45,000; carbonic anhydrase 31,000; soybean trypsin inhibitor 21,500; lysozyme 14,400) and being substantially free of contaminating deoxyribonucleases (endo‑ and exonucleases). The molecular‑weight limitation was the applicants' principal argument for patentability over Kaledin et al. and Chien et al.

• Claim 2 (recombinant polymerase) — independent. The same thermostable Taq polymerase of claim 1 produced by recombinant DNA techniques (i.e., from a host cell transformed with a vector encoding the enzyme), rather than purified from native T. aquaticus cultures.

• Claim 3 (stabilized composition) — independent (lower confidence on scope). A stable enzyme composition comprising the purified thermostable enzyme in a buffer containing one or more non‑ionic polymeric detergents (e.g., NP‑40 and Tween 20), corresponding to the Summary statement: "the invention also encompasses a stable enzyme composition comprising a purified, thermostable enzyme as described above in a buffer containing one or more non-ionic polymeric detergents."

Related but separate patents — do not conflate with the '818 claims: US 5,079,352 claims recombinant DNA sequences and vectors encoding Taq polymerase activity (e.g., claim 1: "A recombinant DNA sequence that encodes the thermostable DNA polymerase activity of Thermus aquaticus"), and US 5,352,600 claims the polymerase characterized by specific activity (~200,000 units/mg). Both are continuations/continuations‑in‑part that trace back through application 07/063,509.

Sources: https://www.nature.com/articles/45333.pdf ; CAFC opinion Hoffmann‑La Roche v. Promega, 323 F.3d 1354 (Fed. Cir. 2003) (via https://storage.courtlistener.com/harvard_pdf/[781380](/patent/781380).pdf)

5. Litigation / docket status (including 2026 check)

2026 CAFC check: no CAFC appeal, IPR‑appeal, or other 2026 docket naming 4,889,818 surfaced in my searches. Given the 2006 expiration, this is consistent with expectations.

Historical matters for this patent:

1. Hoffmann‑La Roche, Inc. v. Promega Corp., N.D. Cal. — Roche sued in 1992 for breach of license/patent infringement; Promega counterclaimed inequitable conduct. In 1999 Judge Vaughn Walker held all claims of the '818 patent unenforceable for inequitable conduct (eight instances). The Federal Circuit in Roche II, 323 F.3d 1354 (Fed. Cir. 2003), affirmed materiality/intent findings as to two categories (the Example VI past‑tense written description and the purity representations) but reversed as to molecular weight, and remanded for the discretionary equities determination. On remand the district court again held the '818 patent unenforceable. Key finding: Example VI was written in the past tense (over 75 instances) though Dr. Gelfand admitted it was never performed as described.
2. Antitrust follow‑on: Molecular Diagnostics Laboratories v. Roche / Applied Biosystems (Walker Process claim), D.D.C. No. 1:04‑cv‑01649 — settled in 2008 for $33 million to a direct‑purchaser class.
3. Google Patents also lists earlier/later US cases: E.D. Va. No. 1:03‑cv‑01447 and S.D. Cal. No. 3:11‑cv‑00056.
4. Parallel EPO opposition on the sibling European patent EP 0 258 017 (priority from US 899513, 899241, 63647, 63509) — revoked by the Opposition Division (2001), then T 1080/01 (24 Oct. 2003) allowed an auxiliary request and found novelty/inventive step/sufficiency satisfied.

Sources: https://storage.courtlistener.com/harvard_pdf/781380.pdf ; https://www.nature.com/articles/45333.pdf ; https://preview-genomebiology.biomedcentral.com/counter/pdf/10.1186/gb-spotlight-20030401-02.pdf ; https://legacy.epo.org/boards-of-appeal/decisions/pdf/t011080eu1.pdf

6. Residual uncertainties I am explicitly flagging

• Verbatim claim text of US 4,889,818 was not retrieved. My claim summary is reconstructed. Anyone relying on exact claim language should pull the printed patent or USPTO PatentCenter.
• Number and identity of all claims: contemporaneous reporting says three claims; expert declarations in the Roche/Promega record refer to "claims 1 and 2." I could not verify whether a third claim exists or whether the count differs (e.g., if the printed patent has more claims than contemporaneous press reported).
• Possible reexamination certificate: one retrieved document image carries the header "US 4,889,818 C1," which may indicate a reexamination certificate (C1). I could not confirm this and am not asserting it as fact — it should be checked directly.
• Inventor name spelling discrepancy (Frances C. vs. Francis C. Lawyer) unresolved, reported literally per the no‑autocorrect rule.
• Assignee chain is well documented (Cetus → Hoffmann‑La Roche → Roche Molecular Systems), but "current assignee" data on Google Patents is expressly disclaimed by the source as potentially inaccurate.

I'll research litigation involving US 4,889,818. Let me run several targeted searches.

Let me search for the specific case numbers referenced in the patent record and related litigation.

Let me dig into the specific cases: the E.D. Va. case, the Promega case, and the D.D.C. case.

I ran a series of targeted searches (Google Patents' litigation record for US 4,889,818, Patently-O, CourtListener, Casetext, the Federal Circuit opinion text, and a litigation-docket aggregator). Here is what I found.

Bottom line

Yes — US 4,889,818 has been the subject of several US proceedings. Two are substantive patent cases (one infringement/breach-of-contract case, one antitrust/Walker Process case), one is a false-marking case involving the expired '818 patent among a portfolio of expired PCR patents, and one docket is listed on the patent record that I could not verify. The most consequential matter ended with the '818 patent held unenforceable for inequitable conduct.

Note on scope: I limited results to matters involving the '818 patent itself. Sibling PCR patents ('195, '202, etc.) appear in the same portfolios, and I have excluded those where they are the sole patents at issue.

1. Hoffmann-La Roche, Inc. & Roche Molecular Systems, Inc. v. Promega Corporation

Status chain (well documented):

• Dec. 7, 1999 — After a bench trial (Feb. 1999) and an earlier summary-judgment ruling, the district court held the '818 patent unenforceable due to inequitable conduct, finding eight material misrepresentations/omissions during prosecution. Reported at Hoffmann-La Roche, Inc. v. Promega Corp., 1999 WL 1797330 (N.D. Cal. 1999).
• Mar. 31, 2003 — Federal Circuit, Hoffmann-La Roche, Inc. v. Promega Corp., 323 F.3d 1354 (Fed. Cir. 2003), affirmed in part and reversed in part (reversing the molecular-weight category of findings; upholding the Example VI "past tense" and fidelity/template-dependence findings) and remanded for the district court to determine whether the surviving incidents justified unenforceability. Judge Newman dissented.
• 2004 (on remand) — The district court again held the '818 patent unenforceable, 319 F. Supp. 2d 1011 (N.D. Cal. 2004).

Sources: CourtListener opinion text (https://www.courtlistener.com/opinion/[2316783](/patent/2316783)/hoffman-la-roche-inc-v-promega-corp/); Federal Circuit summary (https://law.uh.edu/faculty/gvetter3/fcplc/CaseSummaries/2003/Hoffman-La%20Roche%20v.%20Promega,%2000-1372%20(Fed.%20Cir.%20Mar.%2031,%[202003](/patent/202003))%20%5BSUMMARY%5D.pdf); UNCTAD case note (https://unctad.org/ippcaselaw/sites/default/files/ippcaselaw/2020-12/Hoffman-La%20Roche%20v%20Promega%20U.S.%20Court%20of%20Appeals%202003.pdf); BioCentury (https://www.biocentury.com/article/67268).

2. Molecular Diagnostics Laboratories v. Hoffmann-La Roche Inc. (and related defendants)

Details:

• MDL, a purchaser of Thermus aquaticus DNA polymerase, sued under Section 2 of the Sherman Act on a Walker Process theory — alleging Roche and Applera enforced the '818 patent knowing it was procured through inequitable conduct.
• In its Dec. 2005 ruling, the D.D.C. denied defendants' motion to dismiss in its entirety, recognizing that direct consumers could have standing to bring a Walker Process claim (with a minor limitation barring damages predating the four-year limitations period).
• In a later ruling, the court held that MDL's claims were barred by the statute of limitations — MDL was on notice of the alleged fraud from the Dec. 7, 1999 decision in the Promega litigation, so the fraudulent-concealment tolling argument failed.

Sources: Patently-O (https://patentlyo.com/patent/patent-cases-2005); CourtListener (https://www.courtlistener.com/opinion/[2426221](/patent/2426221)/molecular-diagnostics-laboratories-v-hoffmann-la-roche-inc/).

3. Troll Busters LLC v. Roche Diagnostics GmbH, et al.

Details: An action for false marking under 35 U.S.C. § 292 alleging the defendants marked products with twenty-two expired PCR/molecular-biology patents — including US 4,889,818 — after those patents had expired. The complaint asserted the patents were expired and therefore could not be used to mark any product.

Sources: docket aggregator (https://ai-lab.exparte.com/case/dct/casd/3:11-cv-00056/troll-busters-llc-v-roche-diagnostics-gmbh); plaintiff's opposition brief (https://e.casetext.com/brief/troll-busters-llc-v-roche-diagnostics-gmbh-et-al_response-in-opposition-re-notice-of-joinder).

4. Eastern District of Virginia, Case No. 1:03-cv-01447 — UNCONFIRMED

The Google Patents litigation record for US 4,889,818 lists a case in the U.S. District Court for the Eastern District of Virginia, case No. 1:03-cv-01447 (filed 2003). I was unable to confirm the parties, the patent(s) asserted, or the outcome from the sources I retrieved. I am flagging it as listed but unverified rather than guessing. (Source: https://patents.google.com/patent/US4889818/en)

Caveats and verification notes

• Do not treat the above as a complete PACER-level docket review. I did not have direct PACER/CourtListener RECAP docket access for the district-court case numbers in items 1–2; the N.D. Cal. docket number in the Promega case in particular was not confirmed.
• The patent record's own litigation links are: California Southern District Court case 3:11-cv-00056; District of Columbia case 1:04-cv-01649; Virginia Eastern District case 1:03-cv-01447 (via Unified Patents and Darts-ip family 26743490).
• Parallel (non-US) proceedings: There was significant European Patent Office opposition activity against Roche's counterpart patents (e.g., EP-B-0395736, recombinant Taq), with an April 2002 Technical Board of Appeals decision upholding validity. That is EPO proceedings, not US litigation.
• One secondary source (BioCentury/BLR) states a California court had earlier upheld the Taq patent in litigation between du Pont and Cetus; I could not independently verify the case name or docket, so treat that as unconfirmed.
• Also note the searches surfaced many hits for other "818" patents; I have excluded those (e.g., unrelated '818 patents) per your instruction to avoid similar-number results.

Recommended next step for confirmation: pull the three docket numbers above directly from PACER/RECAP and from the Unified Patents litigation portal (https://portal.unifiedpatents.com/litigation) to obtain exact filing dates, party names, and dispositions, and to resolve the E.D. Va. 1:03-cv-01447 parties.

I'll verify the proceedings list against public sources before drafting.

Confirming the absence of AIA trials and checking the related litigation/appeal chain.

Checking the reexamination certificate and the enforceability appeal history, plus a final sweep for any AIA trial.

PTAB proceedings overview

AIA trial count: 0. The USPTO Open Data Portal's structured "PTAB proceedings on file" block for US 4,889,818 returns no AIA trial proceedings — no IPR, no PGR, no CBM — meaning there are zero active proceedings, zero claims invalidated by the Board, zero claims sustained by the Board, zero settlements, and zero institution denials; the defensive bottom line is not "the patent survived IPRs and is hardened" but something far stronger for anyone receiving a demand letter: the '818 patent expired on 2006-12-26, and its claims were separately held unenforceable for inequitable conduct in 2004 — so an assertion of this patent today fails on both the enforceability and the damages (past-damages-only) axes, regardless of any IPR posture. My independent web sweep surfaced no IPR/PGR/CBM docket number naming the '818 patent; absent an ODP-indexed record I would treat "no PTAB activity on file" as correct, with the caveat below on the ex parte reexamination, which is a Central Reexamination Unit proceeding, not an AIA trial.

Why the per-proceeding template is empty — and what a defendant should look at instead

There are no proceedings to list. To avoid the report being uselessly thin, below is every validity challenge I could verify on this patent, clearly labeled by forum. These are not AIA trials and do not generate § 315(e)(2) estoppel.

EX PARTE REEXAMINATION — USPTO CRU (not an AIA trial)

• Type: Ex parte reexamination (pre-AIA, 35 U.S.C. §§ 301–307) — evidenced by the reexamination certificate designation US 4,889,818 C1
• Filed: date not verified in the sources retrieved
• Status: concluded; certificate issued
• Outcome as reported: claims recertified without amendment — contemporaneous reporting from early 2000 states the USPTO "reexamined the patent and recertified its claims without amendment." I could not independently verify the certificate date or the exact claim-by-claim text from the sources retrieved, so treat the "without amendment" point as reported-but-not-confirmed.
• Defensive value: this is the only proceeding in which a government tribunal considered the '818 claims and left them standing. It is also the reason a § 325(d) argument ("same or substantially the same art previously presented to the Office") would have been attractive in any IPR — moot now, given expiration and unenforceability.

HOFFMANN-LA ROCHE, INC. v. PROMEGA CORP., No. 00-1372 (Fed. Cir. 2003-03-31) — CAFC appeal from a district court invalidity/unenforceability judgment

This is the single most consequential proceeding in the patent's history, and it is a § 282 unenforceability (inequitable conduct) judgment, not an IPR.

• Type: Federal Circuit appeal, not a PTAB proceeding
• District court: Hoffmann-La Roche, Inc. v. Promega Corp., No. C-93-1748 VRW (N.D. Cal.); bench trial February 1999; holding of unenforceability entered 1999-12-07 (Judge Vaughn Walker) on eight separate misrepresentations and omissions
• Federal Circuit disposition: affirmed in part, reversed in part, remanded — 323 F.3d 1354
  • Reversed: the molecular-weight category (Stoffel experiment materiality and intent to deceive)
  • Affirmed: the Example VI misrepresentation (specification written in the past tense describing an experiment never run as written, plus the 250,000 units/mg specific-activity and "nuclease-free" representations) and the fidelity/template-dependence misrepresentations
  • Dissent: Judge Newman, criticizing the majority for "infer[ring] malevolence from verb tense"
• On remand: final judgment of unenforceability entered, 319 F. Supp. 2d 1011 (N.D. Cal. 2004)
• Practical effect: because inequitable conduct renders the entire patent unenforceable, all claims of the '818 patent are unenforceable in any infringement action — the reason no rational defendant faces live exposure here, and the likely reason no petitioner ever spent money on an IPR against it.
• Link: CAFC opinion PDF via CourtListener: https://storage.courtlistener.com/harvard_pdf/[781380](/patent/781380).pdf ; UNCTAD case summary: https://unctad.org/ippcaselaw/sites/default/files/ippcaselaw/2020-12/Hoffman-La%20Roche%20v%20Promega%20U.S.%20Court%20of%20Appeals%[202003](/patent/202003).pdf

Related litigation touching Taq polymerase — do not conflate these with the '818 patent

• Life Technologies Corp. v. Promega Corp., No. 14-1538 (U.S. 2017-02-22), on appeal from Promega Corp. v. Life Technologies Corp., 773 F.3d 1338 (Fed. Cir. 2014) — the § 271(f)(1) "substantial portion" case. The patent at issue was U.S. Reissue Patent No. RE 37,984 (the Tautz patent), not the '818 patent. Any demand letter that invokes Life Technologies v. Promega as a statement about the '818 patent is misrepresenting the record.
• Molecular Diagnostics Laboratories v. Hoffmann-La Roche Inc., No. 1:04-cv-01649 (D.D.C.) — Walker Process antitrust claims premised on the '818 inequitable conduct findings; settled 2008-09-26 ($33 million to a direct-purchaser class), approved 2008-12-19.
• Google Patents' litigation feed also lists N.D. Cal./S.D. Cal. 3:11-cv-00056 (via Unified Patents' litigation dataset) and E.D. Va. 1:03-cv-01447. I could not verify the parties, causes of action, or dispositions from the sources retrieved — do not rely on them without pulling the dockets.

Strategic summary

Claim status. Because no AIA trial was ever instituted on US 4,889,818, every claim is UNTESTED at the PTAB. No claim is CANCELED by the Board; none is SUSTAINED by the Board. The only government validity review of record is the ex parte reexamination, which (as reported) recertified the claims unamended. I am deliberately not listing claim numbers as canceled or sustained, because there are none to list, and I did not verify the patent's total issued claim count from a primary source — anyone needing that should pull the '818 patent and its C1 reexamination certificate directly from USPTO PatentCenter.

Estoppel landscape. 35 U.S.C. § 315(e)(2) is inapplicable. That provision estops a petitioner that requested or maintained an IPR that "results in a final written decision under section 318(a)" from raising, in a civil action or ITC proceeding, any ground it raised or reasonably could have raised. With zero IPRs, no party is estopped by anything, and conversely there is no PTAB record a defendant can point to as an admission by the patent owner. The surviving defensive theories are therefore fully available and unencumbered: (1) unenforceability for inequitable conduct as finally adjudicated in Hoffmann-La Roche v. Promega and confirmed on remand — a judgment that binds the patent owner in subsequent actions and is the strongest defense on file; (2) § 102/§ 103 based on Kaledin et al. (Biokhimiya (1980) 45:644-651) and Chien et al. (J. Bacteriol. (1976) 127:1550-1557) — the very references that the PTO initially used to reject the application before Cetus canceled all claims and argued around them with statements later found deceptive; and (3) § 112 written-description/enablement attacks aimed at Example VI and the "88 kd band / 250,000 units/mg / nuclease-free" representations, which the district court and the Federal Circuit found were never actually achieved as described.

Pattern signals. No serial-petitioner pattern exists — no petitioner has ever filed an AIA trial against this patent, so there is no General Plastic/Sotera-style history to analyze. The patent owner (Cetus → Hoffmann-La Roche → Roche Molecular Systems, Inc., the current assignee of record) was an aggressive licensor (27 Taq manufacturing licenses reported by 2000) and did litigate hard — but its appellate activity was in the Article III courts, not at the PTAB. Unified Patents appears only as the source of the litigation dataset feeding Google Patents' litigation links; there is no evidence Unified filed or funded any PTAB challenge to the '818 patent. Note also that when Google Patents' data shows a "US case filed in California Southern District Court, 3:11-cv-00056," that is a litigation entry, not an AIA trial, and I could not verify what it is.

One timing caveat worth stating plainly: the '818 patent's PTAB exposure window is effectively closed. Its anticipated expiration is recorded as 2006-12-26. Although expiration does not legally bar an IPR (the Board can review expired claims), the combination of expiration, the 2004 unenforceability judgment, and the 2025-era Director policies discouraging institution against long-settled patents means a new petition is not a realistic strategy.

Recommended next steps

1. If you are a defendant facing an assertion of US 4,889,818, the PTAB is the wrong venue and you should not spend money there. Lead instead with the inequitable conduct judgment. The operative cites are Hoffmann-La Roche, Inc. v. Promega Corp., 323 F.3d 1354 (Fed. Cir. 2003-03-31) (affirming the Example VI and fidelity misrepresentation findings; https://storage.courtlistener.com/harvard_pdf/781380.pdf) and the remand judgment at 319 F. Supp. 2d 1011 (N.D. Cal. 2004) holding the '818 patent unenforceable. Quote the Federal Circuit's Example VI holding directly: the court found the inventors "did not perform Example VI as written and did not actually obtain the purity results they had claimed," and that "a single present-tense sentence could not immunize an entire experiment described in the past tense."
2. Add the expiration date to your § 286 analysis. With the patent expired on 2006-12-26, only past damages within the six-year lookback are even theoretically recoverable — and unenforceability forecloses those too.
3. Verify before you rely on anything from the Google Patents litigation feed. Dockets 3:11-cv-00056 (S.D. Cal., via Unified Patents) and 1:03-cv-01447 (E.D. Va.) are unverified as to parties, claims, and outcome. Pull them from PACER/CourtListener rather than citing the feed.
4. *Do not cite Life Technologies Corp. v. Promega Corp., No. 14-1538 (2017), as authority about the '818 patent.* That case construed § 271(f)(1) with respect to RE 37,984.
5. If you want a PTAB record check of your own, the authoritative query is the Patent Trial and Appeal Board End-to-End system (https://ptacts.uspto.gov/ptacts/) and the USPTO Open Data Portal API, both searched on patent number 4889818. As of 2026-09-27 the ODP returns nothing, and my independent search corroborates that. The absence of PTAB activity on a patent this commercially significant and this aggressively asserted is itself informative: it signals that sophisticated adversaries (including Promega, which litigated the '818 patent for over a decade) judged an IPR unnecessary — because they had already killed the patent in district court on inequitable conduct grounds.

I'll start by trying to pull the actual USPTO assignment records and related litigation/ownership data.

Found the original assignment (reel 4742/41, correspondent Albert P. Halluin). Now let me get the Roche links.

I hit the tool-call ceiling before I could pull the reel/frame numbers for the two Roche recordings. Everything below is grounded in what I could verify; I flag the gaps explicitly rather than filling them.

Inventors

All four are Cetus personnel — evidenced not by inference but by the executed assignment itself (USPTO assignment reel 4742/41), in which each named inventor conveys to "Cetus Corporation, 1400 Fifty-third Street, Emeryville, CA 94608," and the correspondent of record is a Cetus in-house address. Saiki is separately a named inventor on the companion PCR patents (e.g. US 4,683,194 / 4,683,195 lineage) filed out of the same Cetus group.

Pattern note (partial, not a finding): The inventorship block maps almost exactly onto the "core PCR reagent" team, and the specification is a purification + recombinant-expression disclosure rather than a research-agenda disclosure. I could not verify individual departure dates for any of the four; I have no record of their employment end dates, so I cannot assert the "all inventors gone within 12 months" signal. What is documented is that the Cetus→Roche transaction expressly included transfer of Cetus PCR personnel ("Roche is also bringing aboard a group of Cetus scientists who have been developing clinical applications for PCR" — Nature Biotechnology 9:1028, Nov 1991). Treat inventor-migration-to-Roche as likely but unverified.

Original assignee

Cetus Corporation (Emeryville, CA 94608). Named on the face of the issued patent; confirmed by USPTO assignment reel 4742/41 recorded 1987-07-31.

• Line of business: Early recombinant-DNA biotechnology — molecular biology reagents, human in-vitro diagnostics, and biopharmaceuticals (IL-2, PEG-IL-2). Founder of PCR, and co-venturer with Perkin-Elmer in Perkin-Elmer Cetus Instruments (formed 1986) to sell PCR instrumentation and reagents.
• Did it ship a product embodying the claims? Yes, effectively. Cetus commercialized thermostable Thermus aquaticus DNA polymerase through the Perkin-Elmer Cetus joint venture (the "AmpliTaq" reagent line) from 1988, and licensed it (notably to Promega in 1990, with a field restriction barring Promega from promoting Taq for PCR).
• Current status: Dissolved as an independent entity. Cetus sold its PCR/Taq rights to Hoffmann-La Roche and its PECI stake to Perkin-Elmer in the Dec 1991 transaction (~$300M plus royalties), then merged with Chiron Corporation in 1995. Chiron was itself acquired by Novartis (2006). No Cetus entity exists today.

Assignment timeline

Sources: USPTO assignment record as mirrored at plainsite (/patents/assignment.html?id=417143) and Google Patents legal events for US4889818A. Only the first entry below carries a verified reel/frame.

1987-06-16 (Gelfand, Stoffel, Lawyer) and 1987-06-22 (Saiki) (executed) / recorded 1987-07-31 — Reel 004742/0041

• Conveyance: Assignment of Assignors Interest (inventor → company; filed before grant, indexed to application 07/063,509 and utility patent 4,889,818)
• Assignor: David H. Gelfand; Susanne Stoffel; Frances C. Lawyer; Randall K. Saiki (recorded as "Lawyer, Francis C.")
• Assignee: Cetus Corporation, 1400 Fifty-third Street, Emeryville, CA 94608
• Correspondent: ALBERT P. HALLUIN, c/o Cetus Corporation, 1400 Fifty-third Street, Emeryville, CA 94608 (no separate law-firm name recorded; the recorded entry carries a phone number, +1 650 565 3585). Flag: there is only one recorded correspondent on this chain, so the "repeat correspondent" tell cannot be established from this patent alone. Note for cross-family work: Halluin is also the third-named attorney of record on the face of the '818 patent, alongside Janet E. Hasak and Kevin R. Kaster — i.e. the same Cetus prosecution/recording group.
• Context: Ordinary pre-issuance employee invention assignment; no consideration structure or third party involved.

1992-01-17 (per Google Patents legal events) / recording date not verified — Reel/frame NOT VERIFIED

• Conveyance: Assignment (recorded as a "reassignment")
• Assignor: Cetus Corporation
• Assignee: Hoffmann-La Roche, Inc.
• Correspondent: not verified in the sources I could reach
• Context: Structured asset sale / internal-carve-out to a strategic acquirer. This is the sale of Cetus's entire PCR and Taq-polymerase business to Roche (announced Nov–Dec 1991, closed early 1992), executed to clear the decks for the Cetus–Chiron merger. The consideration was reported as ~$300M plus up to ~$30M in royalties. It was contested in Delaware Chancery by Eastman Kodak, a former Cetus diagnostics partner, but completed.

1997-01-27 (per Google Patents legal events) / recording date not verified — Reel/frame NOT VERIFIED

• Conveyance: Assignment (recorded as a "reassignment")
• Assignor: Hoffmann-La Roche Inc.
• Assignee: Roche Molecular Systems, Inc.
• Correspondent: not verified
• Context: Internal group reorganization. Roche Molecular Systems, Inc. is the Roche (US) operating subsidiary that holds and runs the PCR / molecular-diagnostics business; the parent transferred the portfolio down to it. Not an arm's-length sale.

Current assignee of record: Roche Molecular Systems, Inc. (per Google Patents "Current Assignee"). Patent expired for failure to pay maintenance fees / anticipated expiration 2006-12-26 — the recording note reads "Anticipated expiration," i.e. end of the 17-year term from the 1989 grant, not a lapse.

Also on the record, but not an assignment: a re-examination certificate (referenced on the front-page image set as "(Reexamination Certificate)"). A reexam certificate changes claims, not title, and does not generate an assignment entry — worth knowing because the enforceable claim set differs from the printed 1989 claims.

Related family (same chain, same direction): the '818 is the parent of a large Roche-family portfolio that moved on the same title chain — US 5,079,352 (issued 1992 to Cetus, i.e. just pre-transfer), then US 5,352,600; 5,407,800; 5,405,774; 5,455,170; 5,466,591; 5,561,058; 5,618,703; 5,641,864; 5,693,517; 5,789,224 and others as Hoffmann-La Roche Inc.; and later filings as Roche Molecular Systems, Inc. The Roche name transition visible across siblings brackets the 1997 internal reassignment.

Searches attempted but unreturned (disclose as gap): I could not retrieve the reel/frame, conveyance text, or correspondent for the 1992 and 1997 recordings within this session. Verify directly at https://assignmentcenter.uspto.gov/ (search "4889818") — both recordings should appear as post-issuance assignments.

Timeline diagram

timeline
    title Ownership of US 4889818
    1986 : Priority application filed Aug 22
    1987 : Application 07 063 509 filed Jun 17
         : Inventors assign to Cetus reel 4742 frame 41
    1988 : AmpliTaq Taq reagent reaches market
    1989 : Patent issued Dec 26
    1991 : Cetus sells PCR business to Roche
    1992 : Cetus assigns to Hoffmann-La Roche
    1995 : Cetus merges into Chiron
    1997 : Hoffmann-La Roche assigns to Roche Molecular Systems
    2006 : Patent expires Dec 26

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. Assignees across the entire recorded chain are Cetus Corporation and Hoffmann-La Roche, Inc. / Roche Molecular Systems, Inc. No "IP / Patents / Licensing / Holdings / Ventures" entity appears, no registered-agent address, no single-purpose LLC. Roche Molecular Systems is a wholly-owned operating subsidiary of a Swiss-listed pharmaceutical group with manufacturing and a marketed reagent line (reel 4742/41 for the origin; Google Patents legal events 1992-01-17 and 1997-01-27 for the two downstream links).

2. Known asserter in the chain — NOT PRESENT. Neither Cetus nor Hoffmann-La Roche / Roche Molecular Systems appears on the Acacia, Marathon, IV, IPNav, Wi-LAN, Conversant/Mosaid, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, DGC or Spangenberg rosters, and I found no Unified Patents or RPX high-frequency-plaintiff listing for either. Roche is the paradigm operating-company enforcer here, not an NPE: it asserted the Taq patents against Promega — a company to which Cetus had itself granted a Taq manufacturing licence in 1990 — and the N.D. Cal. court held the '818 claims unenforceable (all claims) in that dispute. Assertion against an actual reagent competitor cuts against NPE status.

3. Repeat correspondent across the chain — NOT PRESENT (single data point). Only one correspondent is documented: Albert P. Halluin, c/o Cetus Corporation, Emeryville, CA, on reel 4742/41, recorded 1987-07-31. One appearance is not a recurrence, and it belongs to the operating-company link, not to a shell. I could not verify the correspondent on the 1992 or 1997 recordings; if either names a large outside firm (e.g. the Cetus successor firm), that would still be ordinary corporate-IP practice, not an NPE tell. No attorney on this chain appears on any NPE assertion list that I could confirm.

4. Cascading transfers — NOT PRESENT. Two post-issuance transfers, five years apart (1992-01-17, then 1997-01-27). No chained LLCs, no shared correspondent address across unrelated assignees, no clustering inside 24 months.

5. Pre-litigation transfer — NOT PRESENT / UNCLEAR. The original assignment (reel 4742/41, recorded 1987-07-31) predates any litigation by roughly two years and predates grant by over two years — it is a plain pre-issuance employee assignment. The 1992 Roche transfer did precede Roche's Taq assertion campaign, but that is a business-unit acquisition of an entire reagent franchise, not a re-titling to create standing or venue. Freshly-timed pre-suit papering is absent. (Litigation on the record per Google Patents: 3:11-cv-00056 S.D. Cal.; 1:04-cv-01649 D.D.C.; 1:03-cv-01447 E.D. Va. I did not verify party names or complaint dates for the 2003–2011 dockets, so linkage to any specific transfer is unclear.)

6. Bankruptcy fire-sale — NOT PRESENT. Cetus did not file Chapter 7 or 11. The 1991–92 PCR divestiture was a negotiated strategic sale whose stated purpose was to fund and simplify the Cetus–Chiron merger, completed 1995. Nothing in the record suggests a distressed sale of the '818 portfolio.

7. Privateering — NOT PRESENT. Privateering requires an operating company to hand enforcement to a third-party NPE while retaining a license-back. Here the operating company kept title to itself and sued in its own name (Roche Molecular Systems v. Promega). No NPE was interposed between Roche and its competitors.

8. Defensive aggregator — NOT PRESENT. The chain terminates at Roche Molecular Systems, Inc., a commercial operating entity, not at RPX / AST / LOT / Unified / OIN. Nothing in the record indicates neutralization. (Inverse observation: the patent is effectively neutralized today, but by expiry on 2006-12-26, not by defensive aggregation.)

Verdict

Operating-company assertion.

The chain is inventors → Cetus Corporation (reel 4742/41, recorded 1987-07-31) → Hoffmann-La Roche, Inc. (legal event 1992-01-17) → Roche Molecular Systems, Inc. (legal event 1997-01-27), with the only recorded correspondent being Cetus's own Albert P. Halluin. Every assignee is a manufacturing, product-shipping entity — Cetus sold Taq through the Perkin-Elmer Cetus joint venture, and Roche sells the reagent to this day — and the one enforcement event I could corroborate (Roche Molecular Systems v. Promega) was Roche suing an actual reagent** competitor in its own name, not a shell asserting on someone else's behalf. Not a single one of the eight NPE signals is present on verifiable evidence: the 1992 and 1997 links are an asset carve-out and a group reorganisation, not cascaded shell transfers. The patent's practical end is expiry on 2006-12-26, not abandonment or NPE acquisition.

Verification links:

• USPTO Assignment Center (search "4889818"): https://assignmentcenter.uspto.gov/
• Legacy Assignment Search: https://assignment.uspto.gov/patent/index.html
• Recorded original assignment (reel 4742/41): https://www.plainsite.org/patents/assignment.html?id=[417143](/patent/417143)
• Google Patents legal events: https://patents.google.com/patent/[US4889818](/patent/US4889818)/en

Two explicit caveats, per the no-fabrication constraint: (i) the reel/frame numbers, conveyance wording and correspondents for the 1992 and 1997 recordings are unverified in this report — I did not reach the underlying Assignment Center entries; (ii) I found no evidence tying Halluin or any other attorney to repeat NPE recording activity, and I make no such claim.

I'll run targeted searches to pull the citation records for this specific patent.

Let me pull the citation tables specifically.

US 4,889,818 — Prior-Art / Citation Analysis

Date note: the current-task header says April 26, 2026, while the session system date is 2026‑09‑27. I flag the discrepancy rather than resolving it.

Scope caveat before anything else. My instructions were to search USPTO for patent number 4,889,818 only, and to work from its own citation list. I was able to confirm the patent's identity, bibliographic data, and full specification text, but I was not able to retrieve the printed front-page "(56) References Cited" / PTO‑892 list for 4,889,818 itself from an authoritative source in this session (Google Patents, FreePatentsOnline and justia all truncate or render that table inconsistently to a search-based fetch). I am therefore splitting the answer into two clearly labelled tiers:

• Tier 1 — References cited on the face of the patent (examiner/applicant IDS): partially reconstructed; flagged as uncertain.
• Tier 2 — Prior art cited inside the '818 specification (the art the applicants actually argued over under §102/§103): these I can state with high confidence, because the full specification text is in hand.

I will not invent a front-page list. Where I cannot ground a citation, I say so.

1. Confirmed identity of the reference patent

This matches the earlier-generated summary; nothing here contradicts it.

Source: https://patents.google.com/patent/US4889818/en

2. Tier 1 — Patent documents cited on/against the '818 record

Flagged uncertainty: I could not pull the literal (56) table. The following are the U.S. patent documents that the '818 record demonstrably relies on, either because they are cited by number in the specification or because they appear in the citation list of its own continuation, US 5,352,600 (same family, same specification text as the '818 disclosure).

I am not asserting that these constitute the complete (56) list, nor that the examiner cited all of them. The earlier-generated section's caution that the verbatim claim text was not retrieved also still stands.

3. Tier 2 — Prior art cited in the '818 specification (the operative §102 references)

These are quoted/named in the patent's own Background and Examples, so they are grounded in the authoritative text.

3a. The two references the applicants had to overcome

(1) Kaledin, A. S., et al.

• Full citation: Kaledin et al., Biokhimiya (1980) 45:644‑651 (Eng. transl. Biochemistry (USSR)); and Kaledin et al., Biochem. (Sep. 1981) 46(9):1247‑1254.
• Date: 1980 (and 1981 follow-up). Both predate the Aug. 22, 1986 priority date — classic §102(b) art.
• Description: A six-step purification of DNA polymerase from T. aquaticus YT1: crude extract → DEAE-cellulose → hydroxyapatite → DEAE-cellulose → single-strand DNA-cellulose. Reported molecular weight 62,000 Da per monomeric unit. The '818 spec expressly criticises it: the pools from each stage "were not screened for contaminating endo- and exonuclease(s)."
• Potential §102 effect: This is the single most dangerous reference. It discloses a purified thermostable DNA polymerase from T. aquaticus — i.e., it would anticipate any claim to that enzyme not limited by molecular weight or purity. Against claim 1 as reconstructed (86,000–90,000 Da + substantially free of contaminating deoxyribonucleases), Kaledin does not anticipate on its face: the recited MW and the absence of endo/exonuclease screening are the very differences the applicants relied on. Its real force is §103, combinable with Chien.

(2) Chien, A., et al.

• Full citation: A. Chien et al., J. Bacteriol. (Sep. 1976) 127(3):1550‑1557, "Deoxyribonucleic Acid Polymerase from the Extreme Thermophile Thermus aquaticus."
• Also: Chien, 1976, thesis, University of Cincinnati (cited in the family's reference list).
• Date: 1976 — well before priority; §102(b).
• Description: Crude extract → DEAE-Sephadex → phosphocellulose → DNA-cellulose, with BSA added to prevent loss of polymerase activity. Molecular weight by gel filtration ≈ 63,000 Da; by sucrose gradient ≈ 68,000 Da.
• Potential §102 effect: Same analysis as Kaledin. Discloses a purified T. aquaticus polymerase with a different molecular weight (63,000/68,000 vs. the claimed 86,000–90,000). It cannot anticipate the MW-limited claim 1. Notably, Chien's use of BSA (a protein, not a detergent) as a stabiliser is not the non-ionic polymeric detergent of claim 3, so claim 3 is not anticipated by Chien.

3b. Related Kaledin work

(3) Kaledin et al., 1983 — T. ruber polymerase

• Citation as listed: Kaledin et al., 1983, "Isolation and Properties of DNA Polymerase from the Extremely Thermophilic Bacteria Thermus ruber," Chem. Abst. 98:298 (49311q).
• Description/effect: Same enzyme from a different Thermus species. Because claim 1 is limited to T. aquaticus ("Thermus" appears 4×; "Thermus aquaticus" 39× in the document), this reference is not anticipatory for claims 1–3 but is §103 background on genus-level thermostable polymerases.

3c. Foundational/background biochemistry

3d. Methods/vector art cited for enablement (not enzyme anticipatory art)

• Bolivar et al., Gene (1977) 2:95 (pBR322); Chang et al., Nature (1977) 198:1056 and Goeddel et al., Nucleic Acids Res. (lac/trp promoters); Broach, Meth. Enz. (1983) 101:307; Stinchcomb et al., Nature (1979) 282:39; Tschumper et al., Gene (1980) 10:157; Clarke & Carbon, Meth. Enz. (1983) 101:300; Hitzeman et al., J. Biol. Chem. (1980) 255:2073; Holland & Holland, J. Biol. Chem. (1981) 256:1385; Fiers et al., Nature (1978) 273:113 (SV40); Depicker et al., J. Mol. Appl. Gen. (1982) 1:561 (nopaline synthase); Cohen, PNAS (1972) 69:2110 (CaCl₂ transformation); Shaw et al., Gene (1983) 23:315 (Agrobacterium); Graham & van der Eb, Virology (1978) 52:546 (calcium phosphate); Van Solingen et al., J. Bact. (1977) 130:946 and Hsiao et al., PNAS (1979) 76:3829 (yeast transformation).
• Cloning/screening/sequencing: Young & Davis, PNAS (1983) 80:1194‑1198 (λgt11); "Wilhelmine" [sic, as printed] et al., Gene (1983) 26:171‑179 (Charon 35); Maniatis et al., Molecular Cloning: A Laboratory Manual (1982); Matteucci et al., JACS (1981) 103:3185; Beaucage et al., Tetrahedron Lett. (1981) 22:1859; Sanger et al., PNAS (1977) 74:5463; Messing et al., NAR (1981) 9:309; Maxam & Gilbert, Methods Enzymol. (1980) 65:499; Methods in Enzymology (1980) 65:499‑560; Kuhn Hoffmann-Berling, CSH Quant. Biol. (1978) 43:63; Radding, Ann. Rev. Genetics (1982) 16:405‑37 (helicase/RecA strand separation); McCutcheon's Emulsifiers & Detergents (1983) (detergent genus).
• Regional prior publication: EP 164,054, published Dec. 11, 1985 (oligomer restriction analysis for sickle-cell detection) — cited for the diagnostic context, not the enzyme.

4. Claim-by-claim anticipation analysis (per the reconstructed claim set)

Claim model as carried over from the earlier section: claim 1 = purified native T. aquaticus DNA polymerase, ~86,000–90,000 Da by SDS‑PAGE against the recited marker set, substantially free of contaminating deoxyribonucleases; claim 2 = same enzyme produced recombinantly; claim 3 = stable composition comprising the enzyme in a buffer with one or more non-ionic polymeric detergents.

⚠️ New contradiction to flag. The Google Patents concept index for US4889818A lists, as appearing in the claims/description: "vector… claims 36", "DNA‑directed DNA polymerase… 32", "Bovine Serum Albumin… 7", "Thermus… 4", "Phosphorylase b… 3", "polyacrylamide… 3". If those concept tags are drawn from the claim set rather than the description, then (a) the "three claims" characterisation in the earlier section is incomplete, and (b) the claims recite BSA and phosphorylase B / polyacrylamide — i.e., the SDS‑PAGE marker limitation — and possibly a vector limitation (which the earlier section attributed to US 5,079,352). I could not resolve this. Treat the claim mapping below as provisional.

Bottom line on the §102 question: On the record available to me, no single cited reference anticipates claims 1–3 as reconstructed. The two real prior-art references, Kaledin et al. (1980/1981) and Chien et al. (1976), are the closest art and were overcome by the ~86,000–90,000 Da SDS‑PAGE molecular-weight limitation plus the substantially-dNase-free limitation — the Federal Circuit's discussion in Hoffmann‑La Roche v. Promega, 323 F.3d 1354 (Fed. Cir. 2003), confirms the molecular-weight limitation was central to the applicants' patentability position. That is a §103 distinction in substance, even though the applicants framed the argument in §102 terms.

5. Documents that cite 4,889,818 (forward citations — not prior art)

The task's phrasing ("look at each patent citation for 4889818") could be read as patents citing '818. For completeness, these are later documents that cite it as background — they post-date the Aug. 22, 1986 priority date and therefore cannot be §102 art against '818:

• US 5,192,674 — "Thermostable DNA polymerase Thermus thermophilus…" (Oshima et al.); cites 4,889,818 plus Ruttimann 1985, Chien 1976, Saiki 1985.
• US 5,242,818 — "Method of producing a thermostable DNA polymerase from Thermus thermophilus" (Oshima et al.).
• US 5,413,926 — "Thermostable DNA polymerase from Thermus thermophilus HB‑8."
• US 5,753,482 — "Method for cloning of a gene for pol I type DNA polymerase" (Comb et al.).
• US 5,962,296, US 7,445,900, US 5,352,600 (same family), and numerous EP/CN/JP/KR documents list it as a cited reference in their descriptions.

Sources: https://patents.justia.com/patent/[5192674](/patent/5192674) ; https://companyprofiles.justatic.com/patent/[5242818](/patent/5242818) ; https://patents.justia.com/patent/[5413926](/patent/5413926) ; https://companyprofiles.justatic.com/patent/[5753482](/patent/5753482) ; https://patents.justia.com/patent/[7445900](/patent/7445900) ; https://patents.justia.com/patent/[5962296](/patent/5962296)

6. Residual uncertainties I am explicitly flagging

1. I did not obtain the true (56) References Cited list for US 4,889,818. The Tier 1 table is partial and possibly incomplete. Anyone needing the examiner's actual cited set should pull the printed patent or the PatentCenter IFW image.
2. Claim count/scope may be larger than the "three claims" model. The Google Patents concept index hints at vector, BSA and phosphorylase-B/polyacrylamide terms in the claim set — consistent with an SDS‑PAGE marker-based MW limitation and possibly a vector claim. This contradicts the earlier section's three-claim reconstruction and should be resolved against the printed claims before relying on the §102 mapping in §4.
3. The possible "US 4,889,818 C1" reexamination certificate noted in the earlier section remains unverified. If a reexamination certificate exists, it may have added or amended claims that change the anticipation analysis — the §102 conclusions above apply to the original claim set only.
4. "Wilhelmine, A. M. et al., Gene (1983) 26:171‑179" is reproduced exactly as printed in the patent; the name is very likely a typographical corruption, but per the no-autocorrect rule I have not silently corrected it.
5. No EPO opposition record for the '818 US patent is applicable — the parallel European proceeding concerns EP 0 258 017 (priority also Aug. 22, 1986), where the Opposition Division revoked in 2001 and T 1080/01 (Oct. 24, 2003) allowed an auxiliary request. That decision is a useful secondary source for how Kaledin and Chien were treated, but it is not a citation of the '818 patent. I was unable to fetch the decision text in this session to quote it.

I'll ground this analysis by pulling the actual claim language and the prior-art references cited on the patent's own face, then build the §103 combinations.

Let me pull the actual claim text and the prosecution-history rejection so the §103 analysis rests on the real claim language.

Obviousness Analysis — US 4,889,818 under 35 U.S.C. § 103

⚠️ Two contradictions with the earlier generated sections (flagged per your rule)

Before the analysis, two corrections — the earlier "Patent summary" section in this thread mis-characterized the granted claims. Building on it would propagate those errors, so I flag them:

1. Claim 3 is NOT a "stable enzyme composition … non-ionic polymeric detergents" claim. The granted claims of the '818 patent are three product claims, all to the polymerase. The detergent-composition subject matter lives in the specification/abstract (and in family members such as the German "Stable enzyme composition…" document DE3752392T3), not in the '818 claims. The earlier section's "Claim 3 (stabilized composition)" is wrong.
2. Claim 1 does NOT recite "substantially free of contaminating deoxyribonucleases." That purity limitation appears in the later continuation US 5,352,600 (claim 1: "…substantially free of contaminating deoxyribonucleases… has a specific activity of about 200,000 units/mg"), not in '818. The earlier section imported '600's language into '818.

I also still cannot confirm the "US 4,889,818 C1" header seen on a patentimages PDF (possible reexamination certificate). If a C1 certificate amended the claims, the analysis below would change. Everything below rests on the granted claim set as printed.

Actual granted claims (verbatim, including original typographical errors — not auto-corrected):

• Claim 1: "Purified thermostable Thermus aquaticus DNA polymerase that migrates on a denaturing polyacrylamide gel faster than phosphorylase B and more slowly than does bovine serum albumin and has an estimated molecular weight of 86,000‑90,000 daltons when compared with a phosphorylase B standard assigned a molecular weight of 92,500 daltons."
• Claim 2: "The polymerase of claim 1 that is isolated from Thermus acquaticus."
• Claim 3: "The polymerase of claim 1 that is isolated from a recombinant organism transformed with a vector that codes for the expression of Thermis aquaticus DNA polymerase."

Source: https://www.freepatentsonline.com/[4889818](/patent/4889818).html (claims); https://patents.google.com/patent/US4889818/en

This matters enormously for § 103: the only feature that distinguishes the granted claims from the admitted prior art is the molecular-weight/migration limitation. The specific activity, fidelity, purity, and pH‑profile arguments the applicants made during prosecution are not in the claims.

1. Governing framework and level of ordinary skill

Graham v. John Deere factors: (1) scope and content of the prior art; (2) differences between the prior art and the claims; (3) level of ordinary skill; (4) secondary considerations. KSR Int'l v. Teleflex adds that combinations of known elements according to known methods are obvious where the improvement is a predictable use of prior-art elements and the POSITA had a reason to solve the problem with the known elements.

Level of ordinary skill (hypothetical, ~1986): a Ph.D. in biochemistry/molecular biology (or an M.S. with several years' bench experience) with working knowledge of (i) protein purification by ion-exchange, hydroxyapatite, affinity, and hydrophobic-interaction chromatography, (ii) molecular-weight determination by SDS‑PAGE, gel filtration and sucrose-gradient centrifugation, and (iii) standard recombinant DNA methodology (Maniatis, 1982 — cited in the patent itself).

2. The prior art of record (the "Prior Art section of this page")

Background Art (admitted prior art in the specification):

• Kaledin et al., Biokhimiya (1980) 45:644‑651 — six-step purification of DNA polymerase from T. aquaticus strain YT1; MW reported as 62,000 daltons per monomeric unit; "pools from each stage were not screened for contaminating endo- and exonuclease(s)."
• A. Chien et al., J. Bacteriol. (1976) 127:1550‑1557 — second purification (DEAE‑Sephadex → phosphocellulose → DNA‑cellulose, with BSA); MW "about 63,000 daltons" by gel filtration and "about 68,000 daltons" by sucrose-gradient centrifugation; the enzyme is "free from phosphomonoesterase, phosphodiesterase and single-stranded exonuclease activities" (PubMed abstract, https://pubmed.ncbi.nlm.nih.gov/8432/).
• U.S. Pat. No. 4,683,195 (Mullis/Cetus) — use of a thermostable enzyme to amplify nucleic acids; discloses the enzyme "need not be added after every denaturation step."
• U.S. Pat. No. 4,683,202 (Mullis) — the PCR amplification process.
• New England Biolabs — had marketed a T. aquaticus polymerase before the filing date (activity decayed in detergent-free storage).

"Other References" cited of record on the '818 face:

• Kaledin et al., Sep. 1981, Biokhimiya 46(9):1576‑1584 (T. flavus DNA polymerase).
• Kaledin et al., Nov. 1982, Biokhimiya 47(11):1785‑1791 (T. ruber DNA polymerase) — related thermophiles, similar (~62 kDa) molecular weights.
• Ruttiman et al., 1985, Eur. J. Biochem. 149:41‑46 — identified three forms of DNA polymerase in a bacterium closely related to T. aquaticus.
• Kaboev et al., Jan. 1981, J. Bact. 145(1):21‑26; Stenesh & Roe, 1972, Biochim. Biophys. Acta 272:156‑166; Klimczak et al., 1986, Biochem. 25(17):4850‑4855; Rossi et al., 1986, System. Appl. Microbiol. 7:337‑341; Air et al., 1974, FEBS Lett. 38(3):277‑281; Fabry et al., 1976, Biochim. Biophys. Acta 453(3):228‑235.
• Bessman et al., J. Biol. Chem. (1957) 233:171‑177 and Buttin & Kornberg (1966) J. Biol. Chem. 241:5419‑5427 — purification/characterization of DNA polymerases.
• Maniatis et al., Molecular Cloning: A Laboratory Manual (1982); Young & Davis, PNAS (1983) 80:1194‑1198 (λgt11 expression libraries); U.S. Pat. No. 4,711,845 (Gelfand et al.) (P_L / N_RBS expression vector).

Caveat: I have not verified the specific technical content of Kaboev, Stenesh & Roe, Klimczak, Rossi, Air, and Fabry from primary sources in this session. I treat them only as part of the genus of thermophile-polymerase and protein-purification references that the Examiner had of record, and I avoid asserting undisclosed specifics. Anyone relying on these as individual § 103 references should pull the originals.

Prosecution history (important context): The Examiner originally rejected the '509 application as anticipated and/or obvious over Chien and Kaledin. Applicants cancelled all claims and added three new claims, arguing (i) the claimed enzyme differed in molecular weight, specific activity, and fidelity, and (ii) the claimed enzyme was "far more pure" than Chien/Kaledin preparations. (Fed. Cir. summary: https://law.uh.edu/faculty/gvetter3/fcplc/CaseSummaries/2003/Hoffman-La%20Roche%20v.%20Promega,%2000-1372%20(Fed.%20Cir.%20Mar.%2031,%[202003](/patent/202003))%20%5BSUMMARY%5D.pdf ; UNCTAD case note: https://unctad.org/ippcaselaw/sites/default/files/ippcaselaw/2020-12/Hoffman-La%20Roche%20v%20Promega%20U.S.%20Court%20of%20Appeals%[202002](/patent/202002).pdf)

3. Comparing claim 1 to the prior art (Graham factor 2)

The sole difference the claims capture is the molecular-weight/migration limitation. Everything else in the applicants' prosecution argument (higher specific activity, better fidelity, greater purity, broader pH profile, detergent stability) is unclaimed and therefore cannot rebut obviousness of these claims as a matter of claim scope.

4. Claim-by-claim § 103 combinations and motivations

Claim 1

Combination A — Kaledin 1980 + Chien 1976 (primary combination).

Motivation to combine: Both references purify the DNA polymerase of the same organism (T. aquaticus, incl. YT1). A POSITA seeking an improved preparation would naturally consult both, because they address the same enzyme with complementary fractionation schemes (Kaledin: crude extract → ammonium sulfate → DEAE‑cellulose → hydroxyapatite → DEAE‑cellulose → ssDNA‑cellulose; Chien: DEAE‑Sephadex → phosphocellulose → DNA‑cellulose). Combining known fractionation steps in a known order is "a combination of familiar elements according to known methods" yielding no more than predictable results (KSR). Notably, the '818 specification's own Example VI requires phosphocellulose — i.e., the applicants' preferred protocol is the Kaledin scheme plus Chien's phosphocellulose step. In the parallel Australian opposition, the hearing officer described the phosphocellulose step as "the difference between the protocols of Kaledin and the opposed specification." That is, the claimed advance over Kaledin is, in large part, an element Chien already taught.

Motivation to remove nucleases: The specification itself says Kaledin's "pools from each stage were not screened for contaminating endo- and exonuclease(s)," while Chien already obtained a preparation "free from … exonuclease activities." A POSITA wanting a polymerase for template-directed synthesis has an obvious reason to adopt Chien's nuclease-removing step.

Why the MW limitation would arguably have been obvious: The prior-art values were obtained by gel filtration and sucrose-gradient centrifugation — methods a POSITA knows can underestimate the mass of asymmetric proteins, and which in Chien's own paper differ by 5,000 Da (63,000 vs 68,000) for the same enzyme. Neither reference used SDS‑PAGE (the technique the claim actually specifies). The applicants themselves argued (and the Federal Circuit ultimately accepted as non-culpable) that the Kaledin/Chien enzymes were proteolytic degradation products of the full-length enzyme. If the prior-art enzyme is a degraded fragment of the claimed enzyme, then simply avoiding proteolysis (working at 4 °C, adding the phosphocellulose step, minimizing handling) to obtain the intact, higher-MW form is the kind of routine optimization that is obvious under In re Aller and KSR.

Combination B — Combination A + Ruttiman 1985 + Kaledin 1981/1982.

Motivation: Ruttiman found three DNA polymerases in a bacterium closely related to T. aquaticus, and the T. flavus / T. ruber Kaledin papers showed reproducible ~62 kDa thermophile polymerases. A POSITA would be motivated to survey T. aquaticus for its full complement of polymerase forms and to isolate the largest/intact one — undercutting the "different enzyme" argument while supplying a reason to look for a higher-MW species.

Combination C — Combination A + Bessman 1957 / Buttin & Kornberg 1966 (+ standard fractionation practice).

Motivation: These establish that purifying a DNA polymerase to homogeneity, determining its molecular weight, and removing nuclease contaminants were well-known, routine objectives. Applying this routine methodology to the already-known T. aquaticus polymerase is a predictable use of known methods.

Combination D — Combination A + U.S. 4,683,195 + the NEB commercial product.

Motivation: '195 expressly teaches that a thermostable enzyme can drive amplification without replenishment after each denaturation step — a concrete, articulated reason to obtain a purified, stable, thermostable T. aquaticus polymerase. The specification concedes "there is a desire in the art to produce a purified, stable thermostable enzyme that may be used to improve the diagnostic amplification process," and that NEB had already marketed a T. aquaticus polymerase. This combination supplies the "reason the POSITA would have made the combination."

Claim 2 — "isolated from Thermus acquaticus"

Claim 2 adds essentially nothing: claim 1 is already limited to a "Thermus aquaticus DNA polymerase," and both Kaledin (YT1) and Chien purify from T. aquaticus. Claim 2 is therefore obvious over Kaledin and/or Chien alone, and a fortiori over any of Combinations A–D.

Claim 3 — "isolated from a recombinant organism transformed with a vector that codes for the expression of Thermis aquaticus DNA polymerase"

Motivation: By 1986, cloning a gene of interest into an expression vector and expressing it in a recombinant host was routine and predictable — the '818 specification itself relies on Maniatis (1982), Young & Davis (1983) (λgt11 expression libraries), and a Cetus expression vector, U.S. 4,711,845 (Gelfand et al.). A POSITA had ample motivation to express the Taq polymerase recombinantly (higher yield; freedom from T. aquaticus host nucleases and proteases). The claim recites no recombinant-specific structure (no vector, host, promoter, or sequence limitation) beyond "a recombinant organism transformed with a vector that codes for the expression of T. aquaticus DNA polymerase." Accordingly, claim 3 is obvious over Combination A (or B–D) in view of Maniatis/Young & Davis/'845.

5. Secondary considerations (Graham factor 4)

What the applicants could argue, and the nexus problem:

The one consideration with some genuine nexus is the molecular-weight difference itself and the applicants' evidence that the prior-art enzymes were degraded preparations. The Federal Circuit reversed the district court's inequitable-conduct finding on the molecular-weight category (Hoffmann‑La Roche v. Promega, 323 F.3d 1354 (Fed. Cir. 2003)), i.e., it did not find the applicants' MW-based distinction to be a misrepresentation. That is a point favoring nonobviousness — but it is an inequitable-conduct holding, not an obviousness holding, and the court expressly upheld the Example VI and fidelity/purity misrepresentation findings.

6. The strongest non-obviousness counterarguments (stated fairly)

1. Consistent prior-art teaching of a smaller enzyme. Kaledin (1980), Kaledin (1981, T. flavus), Kaledin (1982, T. ruber), and Chien all report ~62,000–68,000 Da. Where the art uniformly teaches a smaller molecule, a POSITA is not obviously motivated to seek an 86,000–90,000 Da species; this is closer to a "different invention" argument than a mere optimization.
2. Real factual dispute over identity. The applicants advanced (a) a different strain, (b) a second polymerase in the same organism (Ruttiman), or (c) the prior-art enzymes being proteolytic fragments (cf. the later-identified Stoffel fragment/sTaq). If Kaledin/Chien genuinely isolated a different enzyme (or a fragment), the difference is substantive, not a measurement artifact.
3. No prior-art use of SDS‑PAGE. The claim is expressly tied to denaturing-gel migration relative to a phosphorylase B standard (92,500); the prior-art MWs were obtained by non-denaturing techniques. A POSITA could not have predicted the claimed value.
4. Prosecution/EPO/AU outcomes. The Examiner allowed the amended claims; the Australian Patent Office opposition decision (F. Hoffmann‑La Roche AG v. Bresagen Ltd. & New England Biolabs, [1997] APO 57, https://www4.austlii.edu.au/au/cases/cth/APO/1997/57.html) treated Kaledin as a "mere paper anticipation" that did not deprive the claims of novelty, because the opponents could not establish on the balance of probabilities that the Kaledin enzyme was the same as the claimed Taq polymerase. (Note: the AU decision is a novelty decision, and novelty is a lower bar than obviousness — a "paper anticipation" finding does not answer § 103.)

7. Bottom line

• Claims 2 and 3 are the most vulnerable. Claim 2 adds nothing over claim 1 given claim 1 already recites a T. aquaticus polymerase that Kaledin and Chien isolate from T. aquaticus. Claim 3 recites only the use of routine recombinant expression, motivated by yield and purity goals and enabled by art the specification itself cites (Maniatis; Young & Davis; U.S. 4,711,845).
• Claim 1 is the crux and is a genuinely close case. The strongest § 103 case is: Kaledin 1980 + Chien 1976, optionally + Ruttiman 1985 and the T. flavus/T. ruber Kaledin papers, + U.S. 4,683,195, with the motivation supplied by (i) same organism/same enzyme, (ii) Chien's known nuclease-removing phosphocellulose step (the very step the specification characterizes as the difference over Kaledin), (iii) the desire for a purified thermostable polymerase for the then-new PCR amplification process, and (iv) routine optimization to avoid proteolysis and obtain the intact enzyme. A POSITA would have had a reasoned basis to expect success.
• The best nonobviousness counter rests on the uniform prior-art teaching of ~62–68 kDa, the absence of any prior-art SDS‑PAGE measurement, and the applicants' evidence that the prior-art preparations were degraded. Under KSR this is likely a question of fact turning on (a) whether the prior-art enzymes were the same protein and (b) whether obtaining the intact form was routine — precisely the disputes litigated in Roche v. Promega (there framed as anticipation and inequitable conduct, not § 103).

8. Caveats

• Verbatim claims used here are from FreePatentsOnline (https://www.freepatentsonline.com/4889818.html), which appears to reproduce the original grant. The unexplained "US 4,889,818 C1" header suggests a possible reexamination certificate that may have altered the claims; verify against USPTO PatentCenter before relying on this § 103 analysis.
• The prior generated "Patent summary" section's description of claim 3 (detergent composition) and of a claim-1 nuclease limitation is incorrect for '818 (those features belong to family members such as US 5,352,600 / DE3752392T3).
• I did not retrieve the Examiner's verbatim rejection or the applicants' verbatim amendment; the prosecution-history characterization is drawn from the Federal Circuit/UNCTAD summaries.
• Content of Kaboev 1981, Stenesh & Roe 1972, Klimczak 1986, Rossi 1986, Air 1974, and Fabry 1976 was not independently verified in this session; treat them as genus-level context, not as individually analyzed § 103 references.