US 5789650

I'll research this patent across the requested sources.

Let me search for claim text and any litigation/docket activity.

Let me try to retrieve the abstract and claim text directly.

US 5,789,650 — Patent Profile

Sources consulted: Google Patents full-text record for US5789650A (https://patents.google.com/patent/US5789650/en), corroborating citation records in later EPO/US/CN/JP/TW/RU patent documents, and targeted searches of Federal Circuit 2026 docket material.

1. Bibliographic Data

Note on dates: the "priority date 1990-08-29" shown by Google Patents is characterized there as an assumption, not a legal conclusion. The 1992-03-18 filing date is the actual application filing date of the '650 patent itself.

This patent is one of the core "HuMAb-Mouse" / minilocus patents — the Lonberg & Kay series later cited in essentially every human-antibody-platform specification (e.g., EP 2 740 743, EP 3 698 113, EP 4 008 348 all cite "US5789650, Lonberg").

2. Abstract

Uncertainty flag: the fetched Google Patents full text did not include the verbatim abstract block for US5789650A. What follows is a faithful reconstruction from the patent's own specification, not a quotation. A close analogue of the abstract language appears in the family's Japanese counterpart (JP2009045066A):

The invention provides heterologous immunoglobulin transgenes and transgenic non-human animals harboring such transgenes, wherein the transgenic animal is capable of producing heterologous antibodies of multiple isotypes by undergoing isotype switching. Classical isotype switching occurs by recombination events involving at least one switch sequence region in the transgene; non-classical isotype switching may occur, e.g., by homologous recombination between the human δ-associated (σμ/Σμ) sequences. The transgenes and animals produce a first isotype required for antigen-stimulated B-cell maturation (IgM) and can switch to encode and produce subsequent heterologous isotypes of therapeutic/diagnostic utility (IgG, IgA, IgE). Also provided are B-cells and hybridomas from such animals, methods of inducing heterologous antibody production, methods of determining whether isotype switching has occurred, and vectors/methods for functionally disrupting or suppressing endogenous immunoglobulin loci.

Verbatim-quote caveat: should you need the exact abstract text for a filing or chart, pull it from USPTO PatentCenter / the printed front page rather than relying on the above.

3. Plain-Language Overview of the Disclosure

The patent's central problem statement: transgenic animals had been made carrying heterologous Ig genes, but "there have been no reports of heterologous transgenes that undergo successful isotype switching." Without switching, such animals are locked into making only IgM (the isotype needed for B-cell maturation) and cannot produce therapeutically useful IgG/IgA/IgE from human sequences.

The invention addresses this by building unrearranged ("minilocus") heterologous heavy-chain transgenes that contain:

• V, D, and J gene segments in germline configuration with functional recombination signal sequences (RSSs) — so functional V-D-J joining occurs, including incorporation of recognizable D-region sequence into CDR3;
• multiple C_H genes (e.g., human μ plus γ) with switch regions and the upstream 5′ flanking transcriptional regulatory/sterile-transcript elements needed for class switching;
• optionally regulatory elements borrowed from the host species (e.g., rodent enhancers in a mouse) to boost expression and allelic exclusion.

Supporting subject matter includes: rearranged and unrearranged heavy/light chain transgenes; light-chain miniloci (κ); synthetic V-segment repertoires; in vivo transgene construction by homologous recombination of co-injected overlapping fragments; positive–negative selection (neo/HSV-tk) targeting vectors for knocking out endogenous mouse J_H and J_κ; and antisense-RNA/antisense-oligonucleotide/anti-Ig-antiserum approaches for suppressing endogenous Ig without genomic disruption.

4. Independent Claims — Plain-Language Overview

Significant uncertainty here. The search results returned the patent's disclosed subject matter and its indexed claim concepts (transgenic mouse; DNA; recombination; isotype switching; upstream; nucleotide; germ cell; rearrangement; V gene; D gene; J gene; response; coding region; chromosomal DNA), but not the verbatim claim set. The following is derived from the specification and indexed concepts and should be verified against the issued claims before being relied on:

Claim family A — Transgenic animal composition claims. A transgenic non-human animal (in the preferred embodiment a transgenic mouse) whose germ cells and somatic cells contain a heterologous immunoglobulin heavy-chain transgene. The transgene comprises, in unrearranged form, at least one V_H segment, at least one D segment, at least one J_H segment, and a constant region; the animal functionally rearranges the transgene and produces a heterologous antibody in response to antigen.

Claim family B — Isotype-switching / multi-C_H claims. A transgenic animal (or transgene) in which the heavy chain transgene contains at least two C_H genes with operably linked switch sequences, such that the animal produces heterologous antibodies of multiple isotypes by isotype switching — i.e., a non-switched isotype (μ) plus a switched isotype (γ, α, or ε).

Claim family C — Improved V-D-J rearrangement / D-region incorporation. A heavy chain transgene in which recognition of incorporated D-region sequence in the rearranged V region is achieved, reflecting the "at least about 1%, preferably at least about 10%, more preferably >30–50% of transgene-encoded V regions include recognizable D region sequences" limitation described in the specification.

Claim family D — Combination animal claims. A transgenic animal containing both a heterologous heavy chain transgene (unrearranged) and a heterologous light chain transgene (rearranged or unrearranged), wherein endogenous Ig loci are disrupted (J_H knockout) or suppressed.

Claim family E — Method claims. Methods of (i) producing a heterologous antibody — e.g., immunizing the transgenic animal with an antigen and recovering the antibody or B-cells/hybridomas; and (ii) determining whether isotype switching has occurred in a transgenic animal containing an Ig transgene.

Claim family F — Endogenous-locus suppression/disruption. Vectors (positive–negative selection vectors) and methods for functionally disrupting endogenous heavy/light chain loci by homologous recombination into embryonic stem cells, and antisense/antiserum-based suppression methods.

Bottom line on claims: the claim set is broad, genus-level, and composition-oriented. The likely true independent-claim count and exact wording cannot be confirmed from the sources retrieved — treat the above as a content map, not a claim chart.

5. Litigation / CAFC 2026 Docket Check

No results found. Searches of Federal Circuit 2026 docket and opinion material, and of general litigation databases, returned nothing tying US 5,789,650 to any pending or decided CAFC appeal in 2026.

This is unsurprising and internally consistent: the patent's legal status is Expired – Lifetime, with anticipated expiration on August 4, 2015 — over a decade before the 2026 date you specified. An expired patent can still be the subject of an appeal only in narrow circumstances (e.g., pending infringement damages, PTAB appeal of a pre-expiration challenge, or an FDA/patent-term matter). I found no such record.

Explicit uncertainty statement: my search coverage of the CAFC docket was via web search, not a direct authenticated query of the Federal Circuit's CM/ECF or PACER docket system. I did not find a docket entry, and I cannot rule out a low-visibility matter that is not web-indexed. If a docket confirmation is required for your purposes, query appellate CM/ECF for the Federal Circuit (cafc.uscourts.gov / PACER) by patent number and by the parties Genpharm International, Inc., Janssen Alzheimer Immunotherapy, and their successors (the Genpharm estate passed through acquisition to Janssen/Johnson & Johnson).

6. Historical Context (training data, not search-verified)

For orientation only — this is background I hold from training, not confirmed by the searches above, and it should be independently verified:

• The Lonberg/Kay Genpharm minilocus patents were central to the HuMAb-Mouse platform, later commercialized through Medarex and subsequently Bristol-Myers Squibb.
• The Genpharm patent estate (including the '650) was involved in long-running third-party citation activity rather than being a frequently-asserted litigation patent in recent years.
• I do not have high-confidence, verifiable details of any specific litigation judgment invalidating or upholding US 5,789,650, and I will not invent one.

7. Confidence Summary

I'll research litigation involving US patent 5789650 across patent litigation databases. Let me run several targeted searches.

The first pass surfaced prior-art citations rather than litigation. Let me search the litigation-specific databases.

Still mostly prior-art citations. Let me query litigation-specific databases and docket sources directly.

Litigation search report — U.S. Patent No. 5,789,650

Bottom line: I found no evidence of any district court litigation, ITC investigation, or PTAB proceeding involving U.S. Patent No. 5,789,650.

I want to be explicit about my confidence level and the limits of what I could verify, because "no results" is only meaningful if you know what was searched.

What I searched

• General web queries for the patent number in both formats (5789650 and "5,789,650") paired with litigation, infringement, lawsuit, and party names.
• Queries aimed at the requested sources (Unified Patents litigation portal) and at RPX Insight, Docket Alarm, and CAFC opinion/order repositories.
• Queries pairing the patent with its assignee/inventor chain (Genpharm International, Lonberg, Kay) and with the Abgenix/Xenomouse family.

Important limitation: I was working through a general web-search interface, not a direct, authenticated query into PACER, the Unified Patents litigation caselist, or RPX Insight. I could not run a docket-level search. So my negative finding is a "nothing surfaced" result, not a certified clearance that no suit was ever filed. For a definitive answer you should run the number directly in PACER's case-locator, the Unified Patents litigation caselist, and RPX/InSight, and check the patent's file history for any § 290 notices.

What the searches did return (and why it isn't litigation)

Every hit for 5789650 was a patent-prosecution citation, not a litigation record. The number appears in:

• Prior-art/background citation lists in unrelated later patents — e.g., EP 2,740,743 A2 (US 5789650 A, Lonberg [0090]), EP 2,330,129 B1 ([0043]), EP 1,685,152 B1 ([0084] [0254]), EP 2,865,687 A1 ([0125]).
• "Incorporated by reference" boilerplate in later antibody patent families, where '650 is listed alongside its Lonberg/Kay siblings — U.S. 5,545,806; 5,569,825; 5,625,126; 5,633,425; 5,661,016; 5,770,429; 5,814,318; 5,877,397; 6,255,458 (see e.g. RU 2,830,232 C2; RU 2,811,912 C2; EA 040561 B1).

My separate searches for litigation turned up patent numbers that merely look similar — the '650 and '789 patents in Patriot Universal Holdings, LLC v. Formax, Inc. (E.D. Wis.) are unrelated meat-forming-machine patents; likewise the '649/'789 hits in the Zebra and Packet Intelligence/Cisco matters. None of these involve U.S. 5,789,650. I flag them only so you can disregard them.

What I can say about the patent itself (from the patent record)

• Title: "Transgenic non-human animals for producing heterologous antibodies"
• Inventors: Nils Lonberg and Robert M. Kay
• Assignee: Genpharm International, Inc.
• Application: US 07/853,408, filed March 18, 1992
• Priority: August 29, 1990
• Granted: August 4, 1998
• Legal status: Expired – Lifetime; anticipated expiration August 4, 2015

The 2015 expiration is itself relevant to your question: any damages claim would have had to be filed within the six-year lookback of 35 U.S.C. § 286, i.e., by roughly mid-2021, and only for pre-expiration conduct. A patent this old, in a field (transgenic antibody-producing mice) where rights were consolidated rather than litigated, having no trace of enforcement in the databases I could reach is a plausible, if not conclusive, result.

One context note, flagged as uncertain: this patent belongs to the Lonberg/Kay "minilocus" family that underlies the HuMAb-Mouse® technology (Genpharm, later Medarex, later Bristol-Myers Squibb), which was a contemporary competitor to Abgenix's Xenomouse/Kucherlapati family. I have a recollection that this technology area was the subject of patent interference practice before the USPTO rather than district court litigation, but I could not confirm from my searches that U.S. 5,789,650 specifically was a subject of any interference, and I will not assert it as fact. Interferences are not litigation and would appear on the patent's face/file history rather than in a litigation database — that is the first place I'd check next.

Recommended next steps to close the gap

1. USPTO Patent Center / Patent Public Search — check the '650 file wrapper for interference documents, terminal disclaimers, and any § 290 litigation notices.
2. PACER Case Locator — search "5,789,650" across all districts (filed 1998–2021).
3. Unified Patents litigation caselist and RPX Insight — direct docket queries rather than keyword search.
4. CourtListener / CAFC — search for the patent number plus "Genpharm," "Medarex," "Abgenix," "Bristol-Myers Squibb."

If you want, I can draft the exact query strings for those databases, or, if you can pull a docket/§ 290 result, I'll analyze any case you find (claim scope, asserted claims, outcomes).

I'll verify the structured data with targeted searches before finalizing.

PTAB Proceedings Report — U.S. Patent No. 5,789,650

Patent: "Transgenic non-human animals for producing heterologous antibodies" — Lonberg & Kay, Genpharm International, Inc. (later Medarex / Bristol-Myers Squibb lineage). Application 07/853,408, filed 1992-03-18; priority 1990-08-29; granted 1998-08-04; expired 2015-08-04.

Proceedings overview

Total AIA trial proceedings on file: 0. There are no inter partes reviews, no post-grant reviews, no covered business method reviews, no derivation proceedings, and no institution decisions — denied or otherwise — involving U.S. Patent No. 5,789,650 at the PTAB, per the USPTO Open Data Portal structured data supplied in this prompt and confirmed by web search (I found no IPR/PGR/CBM petition, institution decision, FWD, or Board notice naming this patent). The defensive posture this creates for a defendant is not "hardened patent" but "moot patent" — nothing has been canceled (there is nothing to point to as a claim-level win), but the patent expired on 2015-08-04 and, as explained below, every infringement claim on it is now time-barred under 35 U.S.C. § 286.

⚠️ Cross-reference note / no contradiction. The previously generated litigation section reported no district court, ITC, or PTAB activity for this patent, with the caveat that it could not run authenticated docket queries. The PTAB finding here is consistent with that: zero proceedings in both venues. The one search hit that looks like PTAB activity for this family (an IPR exhibit numbered to a Lonberg/Kay sibling patent) is a false positive and is addressed below.

Proceedings on U.S. 5,789,650

None. No proceeding number exists to report, and I will not construct one. Per the constraints on this task, I am not filling in the per-proceeding template (type / filed / status / panel / grounds / institution / FWD / settlement / appeal / defensive value) because there is no proceeding to describe — doing so would require fabricating a docket number, a petitioner, and claim-level outcomes.

Near-misses I checked and rejected (flagged so you can disregard them)

Strategic summary

Claim status: everything is UNTESTED. No claim of 5,789,650 has been canceled, confirmed, or construed by the Board, because no AIA trial was ever filed on it. If your demand letter or complaint cites claims of the '650, you are not facing a "hardened" patent that has survived an IPR — you are facing a patent with no claim-level adjudication history at all, and correspondingly no FWD to lean on and no estoppel record to neutralize. The trade-off is that you also have no free kill shot on the record; you'd have to build the invalidity case from scratch.

Estoppel landscape: completely clean. Because no IPR, PGR, or CBM was ever instituted against this patent, § 315(e)(2) estoppel does not attach to anybody. No non-party, no privy, and no prior petitioner has been estopped from raising any § 102/§ 103 ground in a civil action or ITC proceeding on this patent. There are no § 325(d) "same or substantially the same art" findings, no Fintiv-type discretionary-denial records, and no claim-construction precedent (BRI or Phillips) from the Board to work around. Practically, a defendant today inherits an empty board: no ground is pre-consumed. (The corollary also holds — no prior petitioner's failure tells you which art the patent owner has beaten, so you get no free intelligence.)

Why the Board was never engaged — and why that is not an accident of enforcement strategy. Three structural reasons, in order of importance:

1. The patent expired before the AIA matured as a tool. The IPR regime opened 2012-09-16; the '650 expired 2015-08-04. That is a roughly 34-month window in which the patent was both alive and IPR-eligible, and the patent's economic value was concentrated in its earlier term, during which AIA trials did not exist. Any pre-AIA validity fight on this family would have been run as an ex parte reexamination or a patent interference in the examining corps, not an AIA trial — that is the right place to look next (see below).
2. PGR and CBM were never available. PGR applies only to patents with effective filing dates on or after 2013-03-16; the '650 claims carry a 1990 priority date, so PGR was categorically unavailable. CBM required a "covered business method" claim (financial-product/service) and excluded "technological inventions"; a transgenic-immunoglobulin patent would not qualify — and the CBM program sunset on 2020-09-16 regardless.
3. The family is a platform-licensing portfolio, not a troll portfolio. The Lonberg/Kay minilocus family ('806, '825, '126, '425, '016, '429, '318, '397, '650, '129, '458) underlies the HuMAb-Mouse® technology. Rights in this area consolidated through corporate transactions (Genpharm → Medarex → Bristol-Myers Squibb) rather than through serial assertion of individual patents, which is exactly the profile that produces zero IPRs. I flag as unverified (and explicitly do not assert) that this family may have been the subject of pre-AIA interference practice with the competing Kucherlapati/Abgenix (XenoMouse) family — if so, that would appear on the face of the patent and in the file wrapper, not in any PTAB or litigation database, and it is not an AIA trial proceeding.

Pattern signals: no repeat petitioner (there are no petitioners); no PTAB appeals by the patent owner (there are no FWDs to appeal); no defensive aggregator such as Unified Patents in the chain (Unified's caselist contains no challenge to this patent). The absence of PTAB activity here is not the usual "the troll's patent just hasn't been hit yet" signal — it is a matured, expired, platform patent whose enforcement window closed quietly.

Recommended next steps

1. The threshold point for any defendant: the '650 is dead as an assertion vehicle, independent of any PTAB analysis. The patent expired 2015-08-04, and under 35 U.S.C. § 286 no recovery is available for infringement occurring more than six years before the filing of a complaint. The six-year tail on the last possible infringing act ran out on 2021-08-04. A demand letter citing 5,789,650 in 2026 is therefore not merely weak on validity — it cannot support recoverable damages at all, and post-expiration "infringement" is a legal impossibility. That should be the lead paragraph of any response, and it is a stronger and cheaper point than any IPR.

2. If you nonetheless want the invalidity record built (e.g., for indemnity posture, or because a counterparty is asserting the family rather than the '650 itself): there is no FWD to link, so the work is a fresh prior-art search against the 1990-08-29 priority date. Note that the family's own earlier-filed siblings and the Abgenix/Kucherlapati patents (U.S. 5,939,598; 6,075,181) are the natural starting points, along with the Surani U.S. 5,545,807.

3. Close the file-wrapper gap on the interference question — this is the one genuine unknown. Pull the USPTO Patent Center / Patent Public Search file wrapper for 5,789,650 and its parent 07/574,748 (filed 1990-08-29) and check for: (a) any interference declarations or settlements, (b) any § 290 litigation notices, (c) terminal disclaimers or common-ownership statements tying the family together, and (d) any pre-AIA reexamination of this or a sibling patent. I could not verify interference activity and will not assert it.

4. If any proceeding is found on a later pull (I did not find one, and the ODP data agrees), re-run this analysis with the actual docket number — but be alert that in this family the recurring false positive is IPR2017-01129, which is Immunex/Amgen U.S. 8,679,487 and used U.S. 5,545,806 as an exhibit, not 5,789,650.

Confidence statement: I am confident in the negative finding, with the same limitation flagged in the litigation section — I reached PTAB sources through a general web index and the supplied ODP block, not a docket-level query, and PTAB E2E / PTAB Public Search is the authoritative check. Note the asymmetry: for a patent with active proceedings, a negative search result is a weak signal; for a patent that expired in 2015 with a 34-month IPR window and zero hits anywhere, a negative is a strong and corroborated signal.

I'll research the assignment record for US 5,789,650 across assignment databases and corporate-ownership sources.

Let me dig into the specific assignment records and the post-1992 ownership chain.

Assignment & Ownership Analysis — U.S. Patent No. 5,789,650

Methodology and a hard limitation, stated up front. I could not open the USPTO Assignment Center (assignmentcenter.uspto.gov) or the Assignment Search index (assignment.uspto.gov) as an authenticated query interface, and I could not retrieve reel/frame numbers for this patent. Everything below is drawn from the patent record itself, the Google Patents legal-events/assignee fields, and public corporate-transaction reporting. No reel/frame is invented anywhere in this report. Where a value was not retrievable, I say so. For a reel/frame-level verification you must run the patent number directly at the Assignment Center link at the end of this report.

Inventors

Unusual-pattern check: not present. There is no evidence of a coordinated inventor exit preceding a portfolio sale. The inventors assigned to their employer five weeks after the application was filed (see below), which is ordinary employee-assignment timing. The "all inventors depart within 12 months" tell does not fire here — and note the direction is the opposite of the fire-sale pattern: the technology, not the people, was the thing that moved in 1997, and Lonberg moved with it.

Original assignee

Genpharm International, Inc. (also styled "GenPharm International, Inc."), a privately held California biotechnology company.

• Primary line of business: transgenic animal technology — specifically the HuMAb-Mouse® platform for generating fully human monoclonal antibodies from immunized transgenic mice. This is an operating R&D company, not a holding vehicle. SBIR/NIH award records list its address as 2350 Qume Drive, San Jose, CA 95131, with 250–500 employees and an "IP Holdings" band of 25–49 patents.
• Did it ship a product embodying the claims? It practiced the claims — the HuMAb-Mouse strain itself is the embodiment, and GenPharm out-licensed it (Centocor, Eisai, Eli Lilly, LeukoSite, and others). It did not sell the mouse as a consumer product; the commercial model was collaborative research/licensing. This is the key distinction from an NPE: the claims were worked internally in a real lab, and the transactions surrounding them were corporate M&A, not shell transfers.
• Current status: not operating. GenPharm was acquired by Medarex, Inc. — deal announced May 1997 at up to $65 million in MEDX common stock, and reported as completed in October 1997. Medarex was itself acquired by Bristol-Myers Squibb in 2009. GenPharm International, Inc. survives as a Medarex/BMS subsidiary entity (it appears as the contracting "Medarex" party in a 2007 Celldex amendment, at 521 Cottonwood Drive, Milpitas, CA), and it founded Genmab A/S in 1999 as a 100%-owned subsidiary. So: acquired twice, no bankruptcy, no dissolution, no fire-sale.

Assignment timeline

Finding: only one assignment is of record for this patent — the original inventor-to-employer assignment. There is no post-issuance assignment chain, no security agreement, no change-of-name record, and no transfer to any LLC on the face of the record I could reach.

• 1992-04-27 (recorded) — execution date not separately retrievable; assignors' interest assigned "of interest"
  • Conveyance: Assignment (recorded on Google Patents as a "reassignment" event; USPTO text: "ASSIGNMENT OF ASSIGNORS INTEREST.")
  • Assignor: KAY, ROBERT M. and LONBERG, NILS
  • Assignee: GENPHARM INTERNATIONAL, INC.
  • Reel/Frame: not retrievable from the sources I could access — I am not supplying a placeholder or a fabricated number.
  • Correspondent: not retrievable. I have no evidence of who recorded this instrument, and I will not guess. There is accordingly no correspondent-recurrence signal to report for this chain — which is itself relevant, because the "one lawyer behind many shells" tell requires at least two recorded links, and there is only one.
  • Context: Ordinary employee invention assignment to the operating employer, executed in the same window as the March 18, 1992 filing. Not a fire-sale, not a securitization, not a transfer to an asserter.

Subsequent ownership (corporate succession — NOT recorded assignments I could verify against this patent):

• 1997-05 announced / 1997-10 completed — GenPharm International, Inc. acquired by Medarex, Inc. in a stock-for-stock transaction valued at up to $65 million. Present in the chain as M&A, not as a patent-recording.
• 2009 — Medarex, Inc. acquired by Bristol-Myers Squibb; Medarex entities continue as Medarex, L.L.C. / Medarex, Inc., and BMS's Princeton address (Route 206 & Province Line Road) appears on later co-ownership listings alongside E. R. Squibb & Sons, L.L.C. — consistent with successor-in-interest treatment. I did not confirm a recorded assignment of the '650 to any of these entities.

Important interpretive caveat — flagged as a contradiction to watch: the Google Patents "Current Assignee" field still reads "Genpharm International Inc." That is almost certainly a records artifact, not a statement that a dissolved-into-Medarex 1997 acquisition target still independently owns the patent. Successor-in-interest transfers of large biotech portfolios are frequently effected by merger operation of law and recorded, if at all, on a blanket basis — or not recorded as to each individual patent. Do not read the stale assignee field as establishing that GenPharm is an independent owner today.

Timeline diagram

Events at and after 1997 are the corporate-succession context for the asset; only the 1992 entry is a recorded assignment of this patent.

timeline
    title Ownership of US 5789650
    1990 : Earliest priority
    1992 : Inventors assign to Genpharm International
    1997 : Genpharm acquired by Medarex
    1998 : Patent granted Aug 4
    2009 : Medarex acquired by Bristol-Myers Squibb
    2015 : Patent expires Aug 4

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. There is no transfer to any entity bearing an IP / Patents / Licensing / Holdings / Ventures suffix in the record I could reach. The only recorded transfer is inventors → GenPharm International, Inc., an identified operating biotech with a physical R&D address (2350 Qume Drive, San Jose) and NIH award history. No registered-agent address, no single-member LLC, no Delaware/Texas incorporation appears anywhere in the chain.

2. Known asserter in the chain — NOT PRESENT. The chain does not touch Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. The named parties are GenPharm, Medarex, and Bristol-Myers Squibb — all operating life-sciences companies. Notably, the nearest thing to an assertion in this technology space was a trade-secret suit against GenPharm (Cell Genesys/Abgenix, filed February 1994), in which GenPharm was the defendant, not the plaintiff — the inverse of the NPE posture. That dispute is also outside this patent.

3. Repeat correspondent across the chain — NOT PRESENT / UNVERIFIABLE. With only one recorded instrument and no retrievable correspondent of record, there is no recurrence to measure. If a later blanket Medarex/BMS recording exists, the correspondent on that instrument is the one worth pulling — it would show whether the same outside counsel handled the whole Lonberg/Kay family. Recommend pulling it; I could not.

4. Cascading transfers — NOT PRESENT. There are no consecutive LLC-to-LLC hops in under 24 months. Two genuine corporate acquisitions occurred, but ten years apart (1997 and 2009), each by a public operating company, each for stock, each publicly disclosed — the opposite of a rapid shell cascade.

5. Pre-litigation transfer — NOT PRESENT. No infringement suit naming this patent was found in the prior research phase, so there is no suit to date a transfer against. The single 1992 assignment predates issuance by six years.

6. Bankruptcy fire-sale — NOT PRESENT. Neither GenPharm nor Medarex filed for bankruptcy. GenPharm's trajectory was: cancelled a planned 1994 IPO after the Cell Genesys trade-secret suit → settled and cross-licensed in March 1997 (Abgenix/Cell Genesys/Xenotech/Japan Tobacco paid GenPharm and Medarex ~$38.6M in 1997–98) → sold into Medarex from a strengthened, not distressed, IP position. The 1994–96 contraction was a litigation-driven downsizing, not a Chapter 7/11.

7. Privateering — NOT PRESENT. No operating company transferred these patents to an NPE to assert against competitors. Medarex asserted the HuMAb-Mouse portfolio through licensing and cross-licenses (Kirin, Genmab, Pfizer, Centocor) — direct, disclosed, first-party licensing by an operating company, which is ordinary biopharma practice.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. It terminates at a large operating pharmaceutical company.

Verdict

Insufficient data — with a strong non-NPE finding.

I am choosing this label honestly rather than flatteringly: the verdict options key "Insufficient data" to "no records, or only the original assignment," which is exactly what is of record here — a single 1992-04-27 inventor-to-employer assignment to Genpharm International, Inc., with no reel/frame retrievable and no post-issuance chain. On the evidence actually available, not one of the eight NPE signals is present, and the asset sits in the hands of operating life-sciences companies by way of two ordinary stock-funded acquisitions (GenPharm → Medarex, 1997; Medarex → Bristol-Myers Squibb, 2009). The "Operating-company assertion" verdict would overstate what I can prove, because the assertion half is missing — no suit naming this patent was found.

Why not a shell-entity concern despite the stale "Genpharm International Inc." current-assignee field: the entity named is a real, addressable, NIH-funded operating company that was acquired and absorbed, not a purpose-built licensing vehicle. The correct next action is records-based, not inference-based — pull the file wrapper and any blanket Medarex/BMS recordings.

Verify at: USPTO Assignment Center · USPTO Patent Assignment Search · Google Patents — US 5789650

Query strings for the operator: search the patent number 5789650; then search assignor/assignee names GenPharm International, Medarex, Medarex, L.L.C., Bristol-Myers Squibb, and E. R. Squibb & Sons and cross-filter by the Lonberg/Kay sibling patent numbers (US 5,545,806; 5,562,126; 5,633,425; 5,661,016; 5,770,429; 5,814,318; 5,877,397; 6,255,458) to expose any blanket recording that lists the '650 among many properties — a blanket instrument is the most probable place the successor transfers are hiding.

I'll research the prior-art citations of US 5,789,650. Let me query multiple sources.

Let me pull the front-page "References Cited" for the '650 patent specifically, and identify the cited documents.

Prior-Art Analysis — US 5,789,650 (Lonberg & Kay, Genpharm International)

0. Scope, sources, and a mandatory verification caveat

I searched for the record of US 5,789,650 specifically (application 07/853,408, filed March 18, 1992, granted August 4, 1998). Two things must be stated up front, because they bound everything below:

(a) I could not retrieve the verbatim front-page "References Cited" block for US 5,789,650. The Google Patents full-text record that was fetched begins at the abstract/description and was truncated before the citation tables; it rendered the classified subject-matter terms and the very long specification but not the U.S. Patent Documents / Foreign Patent Documents / Other References tables. I was working through web search and public mirrors (patents.google.com, uspto.report, Espacenet), not an authenticated USPTO Patent Public Search / PatentCenter session. A document-level query against Patent Public Search (or the '650 file wrapper in Patent Center) is still required to certify the exact list.

(b) I explicitly did not substitute a similar or related patent number. The instructions require literal treatment of patent numbers. In the course of searching I did encounter the "References Cited" table on the US 6,300,129 record (a different patent number — a later continuation in the same 07/853,408 family, which lists 07/853,408 as a parent). I am not presenting that table as the '650's citation list. I flag its contents separately in §4 solely as corroborating family provenance, clearly labelled, precisely so that it is not mistaken for the '650's own citations.

Accordingly, the prior art I can attribute to US 5,789,650 with confidence is the corpus the '650 specification itself identifies as prior art ("The references discussed herein are provided solely for their disclosure prior to the filing date of the present application…"). That is a legitimate and substantial "cited references" set for the '650. Anything else is flagged as unverified.

Critical date framework for § 102 (pre-AIA, this patent is pre-AIA):

Consequence: a reference published after March 18, 1991 cannot be a § 102(b) bar. A reference published after ~Aug. 1990 can generally only be § 102(a)/(e)/(g) art if the invention date is later than the reference. A § 102(b) rejection of the '650's independent claims can only rest on art published on or before March 18, 1991.

1. Most relevant prior art — ranked (the references that actually matter)

2. Each citation, with § 102 mapping

The prior-art references below are grouped by the claim family they potentially touch. Claim families (A)–(F) are as mapped in the previously generated section — verbatim claim text of the '650 was again not retrievable in this pass, so the mapping is to claim scope, not to numbered claims, and is flagged accordingly.

2.1 References aimed at Claim Family A — "transgenic non-human animal containing a heterologous Ig heavy-chain transgene"

2.2 References aimed at Claim Family B — "isotype switching / at least two C_H genes with operably linked switch sequences"

2.3 References aimed at Claim Family C — "functional V-D-J rearrangement with recognizable D-region incorporation"

2.4 References aimed at Claim Family D/E — knock-out hosts, hybridomas, methods

2.5 Documents that are not prior art to the '650 (important to exclude)

3. The § 102 bottom line (honest assessment)

No single cited reference appears to be a clean § 102 anticipation of the '650's principal independent claims. The reason is structural: every pre-bar reference either

(i) discloses a rearranged single-chain transgene (Yamamura 1986; Rusconi 1985),
(ii) discloses a μ-only transgene whose switching is extra-transgenic (Durdik 1989; Gerstein 1990),
(iii) discloses human D-segment / switch-region sequence as genomic material without a transgenic animal (Ichihara 1988; Nikaido 1981; Mowatt 1986; Sideras 1989), or
(iv) discloses knock-out or suppression methodology (Zijlstra 1989; Mansour 1988; Weiss 1984).

The '650's asserted contribution — an unrearranged heterologous minilocus containing V + D + J + ≥2 C_H genes with operably linked cis switch regions and upstream sterile-transcript regulatory elements, that functionally V-D-J rearranges and isotype-switches in vivo — is not disclosed by any one of them.

Therefore the cited art is materially stronger under § 103 than under § 102. A § 102 rejection would require a reference disclosing all the elements; the closest candidates (WO 90/04036 and Bruggemann 1989) fail on the D-segment and two-C_H/switch limitations, and the switching references (Durdik, Gerstein) fail on the heterologous two-C_H cis-construct limitation. Combinations such as WO 90/04036 + Durdik 1989 + Yasui 1989, or Bruggemann 1989 + Sideras 1989 + Ichihara 1988, are the natural § 103 attack.

Caveat: I could not test this conclusion against the '650's literal claim language, which was not retrievable in this pass. If a claim is drafted narrowly to, e.g., a specific construct (pHC1/pHC2/pIGM1) or a specific D-segment (DHQ52), the anticipation analysis changes. Retrieve the claims before finalizing.

4. Corroborating family provenance — clearly labelled, NOT the '650's citation list

For completeness and to prevent conflation, here is what the US 6,300,129 record (a different patent number — the continuation in the same 07/853,408 family) shows on its face. I am not attributing these to the '650:

• U.S. Patent Documents: US 5,175,384 (Krimpenfort, Dec. 1992); US 5,204,244 (Fell et al., Apr. 1993); US 5,434,340 (Krimpenfort, Jul. 1995); US 5,698,196 (Matsushima, Dec. 1997); US 5,702,946 (Doerschuk, Dec. 1997); US 5,939,598 (Kucherlapati et al., Aug. 1999).
• Foreign Patent Documents: EP 0 315 062 (May 1989); WO 90/04036 (Apr. 1990); WO 90/12878 (Nov. 1990); WO 91/00906 (Jan. 1991); WO 91/10741 (Jul. 1991); WO 92/03918 (Mar. 1992); WO 96/02576 (Feb. 1996).

Two of these I verified independently:

• WO 90/04036 A1 — "Production of antibodies from transgenic animals," Bruggemann / Surani, Medical Research Council + AFRC, priority Oct. 12, 1988, published Apr. 19, 1990 (verified via Espacenet).
• WO 91/00906 A1 — "Chimeric and transgenic animals capable of producing human antibodies," Wood & Kaufman, Genetics Institute, priority Jul. 12, 1989, published Jan. 24, 1991 (verified).

Not verified: EP 0 315 062, WO 90/12878, WO 91/10741. I will not guess their titles. (I have a low-confidence recollection that WO 91/10741 is a Cell Genesys / Kucherlapati "generation of xenogeneic antibodies" filing, consistent with the later US 5,939,598 / 6,075,181 line — but I did not confirm it and it should be checked before use.)

Note the timing consequence: WO 91/00906 (Jan. 24, 1991) and WO 91/10741 (Jul. 1991) fall on opposite sides of the March 18, 1991 bar date. WO 91/00906 is a potential § 102(b) reference; WO 91/10741 could only be § 102(a) art, and only against claims not entitled to the Aug. 1990 priority. That distinction is decisive and should be argued explicitly.

5. Contradictions / flags against the previously generated sections

1. No contradiction found with the earlier sections' bibliographic data: application 07/853,408, filed Mar. 18, 1992, granted Aug. 4, 1998, priority Aug. 29, 1990, app 07/853,408 confirmed as a parent of US 6,300,129 (Appl. 08/758,417). Consistent.
2. Refinement, not contradiction: the earlier sections described the claim set as "not retrieved verbatim." That remains true; this pass did not close that gap, and I have therefore mapped art to claim families, not numbered claims.
3. New flag: the "priority date 1990-08-29" is characterized by Google Patents as an assumption. This matters for § 102, because if the '650's claims are not entitled to Aug. 1990, several 1990–1991 references (Gerstein, Oettinger, Pettersson, Yamada, Sanz, Mills 1991) move from "possibly pre-invention" to squarely § 102(a) art. Priority entitlement should be tested before any validity opinion is issued.

6. What still must be pulled to certify this

1. USPTO Patent Public Search / PatentCenter, Application 07/853,408 — the '650's actual "References Cited" tables (PTO-892 / PTO-1449) and any § 290 notices. This is the one action that converts §1–§2 above from "specification-attributed" to "record-certified."
2. The '650's literal claim set — required before any § 102 element-by-element mapping can be stated as a rejection rather than a scope correspondence.
3. Publication dates: confirm the exact issue/publication dates of Mills et al. 1991, Yamada et al. 1991, and Sanz 1991 against the March 18, 1991 bar date.
4. Identity verification for EP 0 315 062, WO 90/12878, and WO 91/10741.
5. File-wrapper check for interference — the Abgenix/Cell Genesys "xenogeneic antibodies" line (US 5,939,598 / 6,075,181) and the Genpharm line were contemporaneous; interferences, if any, appear in the file history rather than in a litigation database, and would be relevant context though not prior art per se.

If you can supply the '650's front-page citation table or its claim set, I will produce the element-by-element § 102 chart and the corresponding § 103 combination analysis.

I'll ground the obviousness analysis in the actual prior-art references of record for this patent. Let me pull the citation records.

Excellent — that surfaced the actual cited-reference corpus for the family. Let me pull the patent-specific reference list and verify the two most important switching references.

Obviousness Analysis — U.S. Patent No. 5,789,650 under 35 U.S.C. § 103

Date note / contradiction flag: the earlier sections of this analysis were prepared against a stated date of 2026-09-28; this task states 2026-04-26. This does not affect anything substantive — the patent expired 2015-08-04 either way — but flagging per the operating rules.

Threshold caveat (carried forward, not repeated): the verbatim claim set was never retrieved. The analysis below is run against the claim families (A–F) mapped in the earlier Independent Claims section and against the patent's own formulaic transgene definition
(V_H)x—(D)y—(J_H)z—(S_D)m—(C1)n—[(T)—(S_A)p—(C2)]q
which the specification states is the descriptive basis for the heavy-chain transgenes. Where a conclusion depends on precise claim language, I say so.

Governing law: this application was filed 18 March 1992, so pre-AIA § 103 applies. The controlling Graham/KSR framework applies with the pre-AIA critical dates discussed in § II below.

I. The prior art corpus actually of record

The page's "prior art" material consists almost entirely of applicant-cited non-patent literature and patent family members. I verified the two most consequential references through independent sources; I verified the existence and titles of a broader set through the reference lists of co-family and related patents (e.g., the USPTO-printed reference list reproduced for the sibling US 6,300,129, and a third-party pre-issuance submission reference list reproduced for US 9,253,965).

Important qualification: the absence from this corpus of examiner-cited US patent references means a real § 103 challenge would need the front-page "References Cited" list and the file wrapper, not just the specification's bibliography.

II. Critical-date analysis (this is where a real challenge is won or lost)

Consequence: The § 102(b) core that is safe on any priority theory is the 1988-and-earlier switching/biochemistry/locus-structure art (Nikaido, Siebenlist, Mowatt, Mills, Szurek, Banerji, Sado, Berman, Ichihara) plus Thomas & Capecchi/Smithies. Durdik and Brüggemann sit in the contested band — they are § 102(a) art absent a Rule 131-style swearing behind, and § 102(b) art if the claims are ultimately limited to the 1992-03-18 date (which is likely for subject matter first added in 07/853,408, notably some of the suppression and specific vector disclosures).

III. Element-by-element obviousness mapping

IV. The combinations that render the claims obvious

Combination 1 — the core animal claims (families A, B, C, D)

Brüggemann 1989 + Nikaido 1981/1982 + Mowatt 1986 + Szurek 1985 + Mills 1983 + Berman 1988 + Banerji 1983, optionally + Durdik 1989

Brüggemann 1989 supplies a complete, working transgenic-mouse minilocus: unrearranged human V, D and J linked to human μ. It also states the objective verbatim — a repertoire of human antibodies and human mAbs against human antigens. Its disclosed deficiency is equally explicit: only human IgM, ~4 % of B cells, ~50 µg/ml.

The specification's own problem statement then supplies the motivation in the applicant's words: "isotypes other than IgM are desired for therapeutic uses" and a transgenic animal that cannot switch is "limited to producing a heterologous antibody of a single isotype."

The solution is a one-variable modification of Brüggemann's construct: append a second C_H gene plus its switch region. Every piece needed to do so was known and published: Sμ, Sγ1, Sγ3 sequences and their repeat structure (Nikaido, Mowatt, Szurek, Mills); the μ switch region and J-Cμ enhancer are physically on the very 18 kb J-containing fragment used to build the minilocus (Mills 1983, as the specification concedes); and the human heavy-chain NotI fragment carrying μ, δ, γ3, γ1 and α1 on a single fragment had been mapped (Berman 1988 — a finding the patent itself relies on and cites for exactly this point).

Motivation to combine is therefore not merely "suggestion in the art" but the stated goal of the closest reference plus the stated medical need. Reasonable expectation of success is supported by: (i) S-region sequence conservation between mouse and human — the specification concedes "similarity between pairs of human and mouse clones 5′ to the C_H genes has been found to be confined to the S regions, a fact that confirms the biological significance of these regions"; (ii) Ott 1987, showing switch recombination operates on an introduced construct in murine pre-B cells; and (iii) Durdik 1989, showing isotype switching of a microinjected μ heavy-chain gene in a transgenic mouse — i.e., switching was already observed in the exact host system.

If Durdik's disclosure is as its title states, the isotype-switching limitation (family B) is squarely anticipated-in-substance or at minimum obvious over Durdik + any reference teaching a second C_H with its switch region.

Combination 2 — the "T—S_A—C2" regulatory limitation (E5)

Stavnezer 1988 + Sideras 1989 + Berton 1989 + Rothman 1990 + Banerji 1983/Petterson 1990

These references collectively establish the then-current mechanistic model: sterile/germ-line transcripts precede and direct switch recombination, and lymphokines (IL-4, IFN-γ) regulate switching by regulating those transcripts. A POSITA designing a switchable transgene would therefore include the promoter/enhancer elements immediately 5′ of each switch region — precisely the claimed T element. Banerji 1983 (mouse J-Cμ enhancer) and Petterson 1990 (rat IGH 3′ enhancer) taught that Ig enhancers drive high-level, B-cell-specific expression and are portable between rodent constructs. Berton 1989 (IL-4 induces, IFN-γ inhibits the γ1 germline transcript) supplies motivation to place an IFN-inducible element upstream — which the specification itself adopts.

The specification's own statement — "increasing the transcription of a particular sterile transcript typically can be expected to enhance the frequency of isotype switch recombination involving adjacent switch sequences" — is an admission that the design rule was known and predictable.

Combination 3 — the D-region-incorporation limitations (family C)

Ichihara 1988 + Siebenlist 1981 + Berman 1988 + Brüggemann 1989

Brüggemann 1989 included D elements, but the field's problem was limited/poor rearrangement. Ichihara 1988 provided a fully sequenced 15 kb fragment containing 12 individual D segments; Siebenlist 1981 mapped the D subregions; Berman 1988 mapped the V_H families and their proximity to C_H. Increasing the number of V and D segments to increase combinatorial and junctional diversity is a predictable, purely quantitative variation of a known construct — a classic KSR "finite number of identified, predictable solutions" case. The claim's numerical thresholds ("at least about 1 %", "10 %", ">50 %" of V regions containing recognizable D sequence) read as results-effective amounts of an obvious design choice, and would need to be shown to be unexpectedly superior over Brüggemann's construct.

Combination 4 — the J_H-knockout host (families D, F)

Thomas & Capecchi 1987 + Zijlstra 1989 + Johnson 1989 (+ Weaver 1985 for rationale)

Zijlstra 1989 is the closest: it used neo as a positive marker and HSV-tk as a negative marker to disrupt, by homologous recombination in ES cells, a locus of the same general class (β2-microglobulin, in the MHC/immune system). The patent itself concedes it followed "the design employed by Jaenisch and co-workers." Johnson 1989 extended targeting to non-expressed genes, obviating any concern that J_H sequences are silent in ES cells.

Motivation: Weaver 1985 showed a transgenic μ gene suppresses endogenous rearrangement; Ritchie 1984 and Storb 1986 addressed allelic exclusion — but the specification itself warns that allelic exclusion would not reliably prevent the endogenous mouse locus from dominating ("the expression of successfully rearranged… transgenes is expected to have a dominant effect … [but] another way to generate a nonhuman that is devoid of endogenous antibodies is by mutating the endogenous immunoglobulin loci"). The mouse J_H region spans only 1.3 kb and the targeting fragment (15 kb KpnI) was known — so the knockout is a routine application of a known method to a known gene. This combination is very strong.

Combination 5 — suppression variants (family F)

Uhlmann & Peyman 1990 + Weiss 1984 + Pepin 1991 (antisense RNA transgenes)

Applying antisense technology — expressly described as a general, target-agnostic knockdown principle — to endogenous Ig chains, as an alternative to time-consuming breeding of knockout homozygotes, is a textbook obvious combination. The specification's own rationale ("does not require the time-consuming breeding") is the motivation expressed in the patent. The limitation that the antisense be selected to hybridize to endogenous but not transgene Ig sequences is a routine design constraint met by sequence comparison.

Combination 6 — in vivo transgene construction

Murray & Szostak 1983 + Burke 1987 + Anand 1989 + Cook 1988 + Smithies 1985 + Song 1987 (and the copending USSN 07/574,747, which is the applicant's own work and is not prior art against itself but evidences obviousness of the technique generally)

Homologous recombination between overlapping DNA fragments in mammalian cells was known (Smithies 1985, Song 1987); YAC/PFGE methodology for isolating very large Ig fragments was known. Using overlapping fragments with ~1–2 kb of shared homology to assemble a large transgene in a zygote is an application of a known mechanism to a known problem (size limits of microinjection).

V. Why a POSITA would have combined them (the legal argument)

1. Same field, same problem, overlapping disclosures. All the recombination and targeting references are in transgenic-animal/molecular-immunology; the switch references are in Ig class-switching. Courts treat combination of references "from the same field of endeavor" as requiring little motivation (In re Keller; In re Merck).
2. The closest reference states the goal the patent claims to have achieved. Brüggemann 1989 frames the objective as a human antibody repertoire and human mAbs — the exact utility the '650 asserts.
3. The patent's own specification concedes the design rules and the need. The statements that IgM-only is "of limited therapeutic utility," that switch sequences may be grafted ("Switch regions may be grafted from various C_H genes and ligated to other C_H genes"), that mouse switch sequences may function better in mice, that sterile-transcript transcription enhances switching — these are admissions of what the art knew and expected.
4. The variation is a small, identified, predictable set. KSR: where there is a design need and a finite number of identified solutions, the combination is obvious. Adding one γ C_H + Sγ to a μ-only minilocus is not an unbounded search.
5. Reasonable expectation of success. Durdik 1989 (switching in a transgenic mouse) + Ott 1987 (switching on an introduced construct) + S-region conservation + Berman 1988 (μ, δ, γ3, γ1, α1 physically linked) together supply it.

VI. Where an obviousness challenge would be contested (and could fail)

I would not present this as a clean invalidity case. Real counterweights:

1. Brüggemann 1989 was an express invitation to a "repertoire of human antibodies," yet the field did not produce switchable human-transgene mice for years. The '650's own admission that unrearranged-transgene experiments were "variable… producing incomplete or minimal rearrangement" is evidence of unpredictability, and Brüggemann's own numbers (4 % of B cells, 50 µg/ml) are poor — arguably a teaching that the minilocus route was marginal, which cuts against "predictable variation."
2. Durdik 1989 involved a mouse (self-species) gene, not a heterologous one. If Durdik switched a murine μ gene in a mouse background, it does not directly teach switching of a human transgene in a mouse, and the patent can argue the human-switch/mouse-recombinase question was genuinely uncertain. (Note the specification appears to argue the opposite for a different purpose — that mouse switch sequences would work better — which is an internal tension worth exploiting.)
3. Objective indicia. Strong long-felt-need and failure-of-others evidence exists in this field, and commercial success/reconstitution of a functional human humoral response in mice carrying human miniloci is documented in the applicant's own later publications (Taylor et al. 1992; Lonberg et al. 1994) and is the subject of the sibling patents. That evidence could rebut even a prima facie case — subject to the usual nexus inquiry, since the commercial success attaches to the whole platform (heavy + light + knockout), not necessarily to the specific claimed subject matter.
4. Priority/§ 102(b) fragility. Several of the most useful references (Durdik, Brüggemann 1989) are § 102(a) art only if priority holds at 1990-08-29, and cannot be § 102(b) art (published within one year). A challenge built on them must overcome a possible Rule 131 showing of earlier invention or establish that the claims are not entitled to the 1990-08-29 benefit.
5. Claim-scope uncertainty. My claim mapping is reconstructed. If the issued claims are narrower than I have assumed — e.g., if an independent claim requires a specific copy number, a specific ratio (IgG:IgM ≈ 10:1), a specific serum titre, or the specific ΔH chain-3′ enhancer — the obviousness case weakens considerably because those are results-effective parameters needing separate art.

VII. Bottom line

The single most dangerous reference is Durdik et al., PNAS 86(7):2346–2350 (April 1989) — "Isotype Switching by a Microinjected μ Immunoglobulin Heavy Chain Gene in Transgenic Mice." If its content matches its title, it places isotype switching in a transgenic mouse squarely in the prior art, and it would do so before even the 1990-08-29 priority date.

Do not rely on my characterisation of Durdik, Brüggemann 1991 (Eur. J. Immunol. 21:1323–1326), Yamamura, Ott 1987, or Yasui 1989 for a filing — I verified their bibliographic existence and titles via reference lists, but did not retrieve their full texts. Before any § 103 paper is advanced, pull: (a) the issued claim set of the '650; (b) the front-page "References Cited"; (c) full texts of Durdik 1989, Brüggemann 1989 and Brüggemann 1991, Ott 1987, and Szurek 1985; and (d) the file wrapper for any Rule 131/132 earlier-invention showing that would move the critical date to 1990-08-29.