US 8586713

I'll search for authoritative information on US 8586713, including any 2026 CAFC docket activity.

Let me run a few more targeted searches to confirm whether 8586713 appears in any 2026 litigation or CAFC docket.

Let me try to retrieve the actual claim text for this patent.

US 8,586,713 B2 — Verified Record

Important caveat up front: My web searches did not return any authoritative USPTO record, PACER docket, or CAFC 2026 docket entry tying this specific patent (8,586,713) to litigation or appeal. I found no result for the exact number that I can treat as a 2026 CAFC docket item. I also do not have the literal, verbatim claim text in my source materials (the fetched full text was truncated before the claims section), so the claim overview below is a specification-grounded reconstruction, not a quotation of issued claim language. I flag this rather than fabricate claim wording.

Bibliographic data (from the Google Patents record for US8586713B2)

Family/continuations (per the same record): further US filings claim priority to this family — US 14/034,906 (→ US 9,309,326 B2), US 15/058,026 (→ US 9,982,013 B2), US 15/961,669 (→ US 11,168,111 B2), US 17/496,263 (→ US 12,157,758 B2), and US 17/868,305 (→ US 2022/0389055 A1).

Abstract (verbatim)

"A bispecific antibody format providing ease of isolation is provided, comprising immunoglobulin heavy chain variable domains that are differentially modified in the CH3 domain, wherein the differential modifications are non-immunogenic or substantially non-immunogenic with respect to the CH3 modifications, and at least one of the modifications results in a differential affinity for the bispecific antibody for an affinity reagent such as Protein A, and the bispecific antibody is isolable from a disrupted cell, from medium, or from a mixture of antibodies based on its affinity for Protein A."

Technology in plain language

The patent solves the "which antibody is which?" problem when two different heavy chains and a common light chain are co-expressed in one cell (producing two homodimers plus the desired heterodimer). Rather than the mouse/rat hybrid approach (immunogenic) or "knobs-into-holes" (incomplete heterodimerization plus foreign sequences), Regeneron uses a minimal, human-derived point modification in one CH3 domain that removes Protein A binding:

• The key substitution pair is H435R/Y436F by EU numbering (H95R/Y96F by IMGT exon numbering) — designated "FcΔAdp." The rationale: IgG3 naturally does not bind Protein A because of Arg435; the double mutation recreates the local IgG3 sequence, so it should be immunologically "invisible" (no non-native 9-mer T-cell epitope).
• Result: the heterodimer has intermediate Protein A affinity and elutes at a distinct pH from the homodimers.
• The specification notes that the Lindhofer-style linear pH gradient alone gave insufficient resolution (~pH 3.9 vs 3.7), so the claimed methods add an ionic modifier (e.g., 0.5–1.0 M NaCl) to widen the elution pH gap — heterodimer eluting around pH 4.2 versus ~pH 3.0 for the unmodified homodimer.
• The format retains native human IgG architecture (so: normal PK, low aggregation, effector function retained). Examples show ~10-day half-life for the heterodimer in mice (comparable to wild type) and a CD20×CD3 bispecific with ~3 pM EC50 in a Raji killing assay and minimal bystander killing.

Independent claims — plain-language overview (reconstructed; not verbatim)

Based on the summary, definitions, and examples in the specification, the issued independent claims appear to cover roughly four subject-matter clusters. Treat claim numbering and exact wording as unverified.

1. Composition — bispecific antigen-binding protein. First polypeptide: N→C, a first antigen-binding region plus a constant region with a first human IgG CH3 (IgG1/IgG2/IgG4 or combination). Second polypeptide: N→C, a second antigen-binding region plus a second human IgG CH3 that carries a modification reducing or eliminating Protein A binding (H95R alone, or H95R + Y96F by IMGT; H435R/H435R+Y436F by EU). Dependent claims add further isotype-normalizing modifications (e.g., for IgG1: D16E, L18M, N44S, K52N, V57M, V82I IMGT), light-chain limitations (human kappa/lambda), human FR/CDRs, and non-immunogenicity limits (no non-native T-cell epitope; proteolysis cannot generate immunogenic 9-mer or 13–17-mer peptides).

2. Method of making a bispecific antibody in a cell. Obtain a first heavy-chain nucleic acid (VH1 + IgG1/IgG2/IgG4 or chimeric constant region), a second heavy-chain nucleic acid (VH2 + constant region with the CH3 modification that eradicates/reduces Protein A binding), and a light chain that pairs with both heavy chains; introduce into a cell and express. Host cells include CHO, COS, 293, HeLa, PER.C6-type retinal cells.

3. Isolation/purification methods. Separating the bispecific from a disrupted cell, medium, or antibody mixture using Protein A, exploiting differential CH3 affinity, including the pH-step/gradient-plus-ionic-modifier limitations (elution at about pH 3.9–4.5, e.g. ~4.2; ionic modifier at ~0.5–1.0 M, e.g. NaCl).

4. Binding protein comprising a heterodimeric Fc. An Fc with one modified and one unmodified CH3, where the differential modification shifts Protein A elution by ~0.5–1.4 pH units relative to a comparable protein lacking the modification.

Specific-embodiment claims also recite a CD20 × CD3 bispecific (human IgG1 + IgG1ΔAdp), an EC50 of about 2.8–3.2 × 10⁻¹² M in a Raji killing assay, and ≤ about 1–10% bystander killing in a 293-cell assay.

Litigation / 2026 docket status

• The Google Patents record carries a "Family has litigation" flag with a Darts-IP family link (family=42738900). That flag confirms some litigation exists somewhere in the family, but it does not identify a case, court, or party, and I could not independently verify a specific proceeding.
• I found no CAFC 2026 docket entry naming 8,586,713.
• Caution against conflation: my searches surfaced Regeneron-related 2026 matters — the Amgen v. Sanofi/Regeneron UPC Court of Appeal inventive-step decision (March 2026) and the EYLEA/aflibercept BPCIA biosimilar MDL (N.D. W. Va., including a Feb. 11, 2026 Samsung Bioepis settlement) — but these involve different patents (e.g., the '306, '992, '691, '510, '276, '315, '193, '495, '768 families), not 8,586,713. Likewise, the Feb. 2026 CAFC Regenxbio v. Sarepta decision concerns U.S. 10,526,617, which is unrelated.
• Corroborating its commercial significance: US 8,586,713 is expressly incorporated by reference in Regeneron's later US 10,626,142 ("Purification platform for bispecific antibodies"), which describes the Fc* H435R/Y436F technology as the basis of Regeneron's bispecific purification platform.

Confidence notes: Title, dates, inventors, assignee, abstract, and family citations are high confidence (directly from the record). The independent-claim overview is medium confidence — derived from the specification's summary and examples, not from verified claim text. The absence of 2026 CAFC activity is based on searches that returned nothing specific; I cannot rule out a docket entry that simply didn't surface in my searches.

I'll search for litigation involving US patent 8586713 across patent litigation databases and court sources.

Let me search more targeted litigation databases and court records.

Let me check CourtListener and other litigation-specific sources.

Litigation Search: US Patent 8,586,713

Bottom line

I was unable to identify any specific litigation in which US Patent 8,586,713 ("Readily isolated bispecific antibodies with native immunoglobulin format," Regeneron Pharmaceuticals, Davis et al.) was asserted, challenged, or otherwise at issue. No complaint, docket entry, court opinion, or PTAB proceeding naming this patent was found in the sources I could reach.

However, I cannot say with certainty that no litigation exists. Here is what I found and how I interpret it.

One affirmative flag — unverified

The Google Patents record for US 8,586,713 (https://patents.google.com/patent/US8586713/en) contains a litigation indicator:

• "Family has litigation — First worldwide family litigation filed", with a link to the Darts-ip case-law database for family 42738900:
  https://patents.darts-ip.com/?family=42738900&...&patent=US8586713(B2)
• There is also a legal-status entry showing "Active, expires 2031-07-22", and a Darts-ip "Global patent litigation dataset" block.

Important caveat: The Darts-ip page I retrieved returned only boilerplate marketing/subscription text. It did not reveal the identity of any case, party, court, case number, filing date, or outcome. This indicator tells us the patent family may have been involved in litigation somewhere in the world, but it is not confirmable from the public-facing material I could access, and I did not independently verify it. I am flagging it rather than asserting it as fact.

Well-known Regeneron litigation that does NOT involve this patent

My searches surfaced substantial Regeneron litigation activity, but none of it identified US 8,586,713 as an asserted or challenged patent. For completeness and to avoid confusion with adjacent matters:

What I searched

• Google Patents record for US 8,586,713 (litigation and legal-status fields)
• General queries pairing "8586713" with "litigation," "lawsuit," "infringement," "PACER," "district court," "CAFC," and "Unified Patents"
• Regeneron-focused litigation disclosures (SEC 10-Q/10-K notes, Amgen and Regeneron investor filings, Big Molecule Watch, CourtListener dockets)

All other hits for "8586713" were prior-art citations to the patent in other documents (e.g., EP 3,119,811 B1; EP 3,768,709 B1; EP 4,378,958 A2; EP 4,062,937 A1; various Justia patent pages), an inventor-listing page (Eric Smith, Regeneron), and a Golden wiki entry — none of which are litigation.

Practical recommendation

Because 8,586,713 is a Regeneron platform patent covering a bispecific-antibody CH3 format (the FcΔAdp / H435R-Y436F protein-A differential-binding approach), it is plausible it could surface in disputes over bispecific products rather than in the fusion-protein/biosimilar matters above. To confirm or rule out litigation definitively, the following would need to be checked directly (I could not reach them within this session):

1. Darts-ip (subscription) for family 42738900 — the Google Patents link indicates a record exists.
2. PACER / CourtListener docket search on the patent number and on Regeneron as plaintiff.
3. USPTO PTAB (P-TACTS) for IPR/PGR filings naming 8,586,713.
4. UPC / EPO registers for the European family members (the Darts-ip link resolves to a family-level record).

If you require a definitive answer, I recommend I run those four targeted checks directly rather than report the unverified Darts-ip flag as a confirmed case. I will not fabricate party names, case numbers, or outcomes that I could not verify.

I'll verify the structured record against live sources before writing anything up.

Let me check for any district-court litigation context and any IPR linked to the '713 patent family or its continuation.

PTAB / AIA Trial Landscape — U.S. Patent No. 8,586,713

Verification note before the substance: the "PTAB proceedings on file" block supplied in this prompt is the canonical USPTO Open Data Portal record and it reports zero AIA trial proceedings for 8,586,713. My independent web searches (PTAB-related petitions, Big Molecule Watch filing coverage, Regeneron SEC disclosures, docket aggregators) surfaced no IPR, PGR, or CBM naming this patent — but I could not run a direct PTAB E2E / ODP case-number query within my search budget, so treat this as "no proceeding found" rather than "proven to be none." I did not invent any proceeding number.

Proceedings overview

Total AIA trial proceedings on U.S. 8,586,713: 0 (0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution denials). There is no Final Written Decision, no institution decision, no estoppel, and no Board panel to report on. The bottom line for a defendant is unusual and worth stating plainly: this is an untested patent. No one has taken an AIA swing at it, so there is no FWD to lean on, no canceled claims to moot a demand letter (§ 1.321(a) disclaimer or certificate), and no § 315(e)(2) estoppel constraining anyone. Its claims stand exactly as issued on 2013-11-19 — which cuts both ways: nothing is dead, but nothing has been hardened by surviving scrutiny either.

Canonical record

No proceedings to describe — and that absence is itself the analysis

Because the count is zero, the standard per-proceeding template (panel, grounds, institution reasoning, FWD verdict, appeal) has no entries. The useful work is instead (a) explaining why a decade-old, commercially central platform patent has zero AIA trials, and (b) telling a defendant what that means operationally.

Why is an eleven-year-old patent with no IPR?

1. PGR was never available. Post-grant review under 35 U.S.C. § 321 only reaches patents whose claims have an effective filing date on or after 2013-03-16 (AIA § 6(c)(2)(A)). The '713 patent's effective filing date is 2009-06-26, so a PGR petition would be jurisdictionally barred. Filing one would draw an immediate dismissal, not an institution decision.
2. CBM was never available either, and is now sunset. Covered business method review is limited to patents claiming a "financial product or service" and excludes "technological inventions." A CH3-domain protein-engineering patent fails both prongs, and the transitional CBM program ceased accepting new petitions on 2020-09-16 (AIA § 18 sunset).
3. IPR is the only AIA vehicle — and it is time-gated. Any defendant served with a complaint alleging infringement of the '713 patent more than one year before filing is barred by § 315(b). The Darts-IP family litigation flag suggests the family has been in suit; if a particular defendant was served years ago, that defendant is time-barred absent § 315(c) joinder, and joinder requires an already-instituted IPR on this patent — which does not exist.
4. AIB § 325(d) and Fintiv discretion. Even a fresh, timely petitioner faces the Board's § 325(d) practice (the same or substantially the same art/arguments previously presented to the Office) and the Fintiv-style discretionary-denial framework, where Regeneron's parallel district-court activity and its habit of defending on the merits in the related aflibercept and antibody matters would be argued as a reason to deny.

IPR2025-01269 is not this patent — avoid the conflation

Search coverage of Regeneron's PTAB activity is dense and easy to misread. To be explicit, none of the following involve 8,586,713:

• Fresenius Kabi v. Regeneron, IPR2025-01269 — challenges U.S. 10,828,345 (aflibercept/EYLEA), with a parallel petition on U.S. 11,084,865; both are EYLEA-family patents, not bispecific-antibody patents.
• IPR2023-00462 (Celltrion v. Regeneron) and IPR2021-00402 (Chengdu Kanghong v. Regeneron) — U.S. 10,464,992 (aflibercept formulation). In IPR2023-00462, instituted 2023-07-20, Regeneron disclaimed all challenged claims in response to institution rather than litigate.
• IPR2022-01225, IPR2021-00881, IPR2023-00532, IPR2024-00201, IPR2024-00260, IPR2024-00298, IPR2023-00739, IPR2023-00884 — aflibercept/EYLEA patents (the '681, '338, '601, '572 patents). Several ended in statutory disclaimers under 37 C.F.R. § 1.321(a) rather than FWDs.

They are relevant only as behavioral evidence about this patent owner (see strategic summary), not as proceedings on this patent.

Strategic summary

Claim status: everything is UNTESTED. On this record, no claim of 8,586,713 has been canceled, confirmed, amended, or construed by the Board. If a demand letter cites the '713 patent today, the claims it cites are the claims Regeneron obtained in 2013, unmodified. There is no IPR certificate to pull from the USPTO, and no reexamination certificate surfaced in my searches either — but note the distinction: the absence of AIA trial activity does not prove the absence of ex parte reexamination or district-court invalidity rulings, and I found no positive evidence of either for this patent specifically.

Estoppel: nothing binds anyone. Section 315(e)(2) estoppel attaches only to instituted proceedings that reach a final written decision. With zero proceedings, no petitioner and no privy is estopped on any ground, and there is no prior-art ground that is "off the table" for a defendant today. Likewise, because no IPR was instituted, no prior petitioner has a § 315(e)(1) bar before the Office. In practical terms, a defendant retains the full § 102/§ 103 prior-art universe for an IPR or for a district-court invalidity defense — subject only to the § 315(b) one-year clock and the Board's § 325(d)/Fintiv discretion.

Pattern signals — the patent owner's litigation temperament. Two observations, both grounded in filed documents rather than speculation:

• Regeneron's demonstrated response to instituted IPRs against its platform patents is to file statutory disclaimers under 37 C.F.R. § 1.321(a) and take the challenged claims off the board before an FWD can issue (e.g., IPR2023-00462 on the '992 patent; the '601 and '572 proceedings described in Regeneron's own petition-identification lists). A disclaimer, unlike an FWD, is not appealable and creates no estoppel — it is a cheap way to deny a petitioner a merits ruling.
• Correspondingly, Regeneron has not had much need to appeal Board losses on this family, because the Board has not reached the merits on these patents. There is no defensive aggregator (e.g., Unified Patents) anywhere in the chain I could find — this is not a patent that has drawn crowd-sourced validity attacks.

The portfolio point that dominates everything. The '713 patent is the progenitor of a family — the record shows US 9,309,326, US 9,982,013, US 11,168,111, US 12,157,758, and US 2022/0389055 all claiming priority to it — and its FcΔAdp (H435R/Y436F, EU numbering) teaching is expressly incorporated by reference into Regeneron's later U.S. 10,626,142 ("Purification platform for bispecific antibodies"). A challenger who kills only '713 leaves the continuations standing, with later expirations. Any IPR strategy aimed at this technology should be scoped across the family in one coordinated filing wave, or it will simply be answered with a disclaimer on '713 and an assertion of a sibling.

Recommended next steps

• If you are a defendant and no IPR exists: say so, and do not pretend otherwise. There is no FWD to link to and no canceled claim to quote. Your invalidity story must be built from scratch on the merits.
• Check your § 315(b) clock first, before anything else. The controlling question is the date you were served with a complaint alleging infringement of the '713 patent. If it is more than one year past, IPR is unavailable to you except via § 315(c) joinder, which needs an instituted IPR that does not exist.
• If you are within the one-year window, the specification's own admissions are the most promising springboard. The '713 specification concedes as background that IgG3's lack of Protein A binding is determined by a single residue, Arg435 (EU numbering), and that Jendeberg et al. (1997), J. Immunol. Methods 201:25–34 reported both the single H435R mutant and the H435R/Y436F double mutant as Protein A-non-binding. The stated inventive contribution is the application of that known mutation to generate an isolable heterodimeric human bispecific. That frames a § 103 theory (Jendeberg single/double mutant in an IgG1 CH3 + Lindhofer et al. (1995), J. Immunol. 155:219–225 for the differential-Protein-A isolation concept + a common-light-chain bispecific reference), and it frames a § 112 written-description argument aimed at the elution-pH and ionic-modifier numerical ranges (the specification itself says the linear-gradient pH difference of ~3.9 vs ~3.7 was "not sufficient" and "not reproducible in a predictable manner" until the ionic modifier was added). I am flagging these as candidate theories derived from the four corners of the specification — I am not representing that the Board has ever assessed them.
• Because no proceeding is pending, there are no trial-stage milestones to track (no institution deadline, no oral hearing date, no § 316(a)(11) one-year FWD due date). Any such dates arise only if a petition is filed.
• Budget for the disclaimer move. If you institute, expect Regeneron to disclaim the challenged claims under § 1.321(a) rather than defend. Plan the challenge so that success on '713 still means something — i.e., challenge the continuation set (US 9,309,326 / 9,982,013 / 11,168,111 / 12,157,758) and the purification-platform patent (US 10,626,142) in the same timeframe.
• Watch the district-court family flag. The Darts-IP "Family has litigation" marker on the Google Patents record (family=42738900) is family-level and unverified at the case level. Before filing anything, pull the actual litigation docket to confirm which family members are in suit, in what district, and when each defendant was served — that service date is the single fact that determines whether an IPR is even legally available to you.

---

Confidence and limitations. High confidence: zero AIA trials are recorded for 8,586,713 in the supplied USPTO ODP block, and my searches surfaced none. High confidence: PGR and CBM are legally unavailable for this patent for the jurisdictional reasons given. Medium confidence: the completeness of the negative finding — I relied on web search rather than a direct PTAB E2E case-number query, so a very recently filed (unindexed) petition cannot be entirely excluded. Explicitly not verified: that no ex parte reexamination, no district-court invalidity judgment, and no foreign counterpart revocation exists for this patent. I found no evidence of any, and I have not asserted that they do not exist.

I'll research the assignment/ownership record for US 8586713, starting with the USPTO Assignment Center and corroborating sources.

Let me look for the recorded assignment details (reel/frame) and any post-issuance transfers.

Let me try to retrieve the actual assignment record and check for any post-issuance transfers or NPE-directory hits.

Assignment & Ownership Chain — US 8,586,713 B2

Scope note / method. I attempted to query the USPTO Assignment Center (https://assignmentcenter.uspto.gov/ via https://assignment.uspto.gov/patent/index.html) for patent 8,586,713, plus Google Patents legal events, Espacenet/INPADOC legal status, and Regeneron SEC filings. I was not able to load the Assignment Center record itself in this session — my tooling returned search-engine results, not the USPTO database UI, so no reel/frame number and no correspondent of record could be retrieved. Everything below is therefore anchored on (a) the Google Patents legal-events block for US8586713B2, which mirrors the USPTO assignment record, and (b) corroborating records for the same family. I flag each unverified item rather than fill it in. This is consistent with the gap already flagged in the earlier Patent summary and Litigation sections of this analysis.

---

Inventors

Unusual-pattern check — negative. The signal to look for is all inventors departing the original assignee near filing, which precedes a portfolio fire-sale. Nothing of the kind appears here:

• All four inventors executed one assignment of their entire interest to Regeneron, recorded 2010-08-13 — roughly 14 months after the 2009-06-26 priority filing and ~4 months after the 2010-06-25 non-provisional. This is the classic employee invention assignment, not a divestiture.
• No inventor is named as an assignor on any other recorded conveyance for this patent — i.e., no inventor ever reacquired, held, or parcelled out rights.
• I found no data on the current affiliations of Davis, MacDonald, or Olson, and therefore do not assert that any of them left Regeneron. Absence of departure evidence is not evidence of retention; I simply could not check it.
• Eric Smith's profile is that of a long-tenured prolific Regeneron inventor (97 patents, 63 Regeneron-assigned, ranked #10 of 810 at the company per patentleaderboard.com). That is the opposite of a fire-sale precursor.

---

Original assignee

Regeneron Pharmaceuticals, Inc. (Tarrytown, New York). Assignment of assignors' interest from the four inventors, recorded 2010-08-13; confirmed by Google Patents ("Assigned to REGENERON PHARMACEUTICALS, INC.") and by the grant's assignee-of-record field ("Original Assignee: Regeneron Pharmaceuticals Inc").

• Primary line of business: discovery, development, and commercialization of biologic medicines — monoclonal antibodies, bispecific antibodies, fusion proteins, and antibody cocktails (EYLEA/aflibercept, Dupixent/dupilumab, Libtayo/cemiplimab, Praluent/alirocumab, ARCALYST/rilonacept, REGEN-COV/casirivimab-imdevimab). Publicly traded (NASDAQ: REGN).
• Current status: Operating. No acquisition of Regeneron, no dissolution, no Chapter 7/11, and no change-of-name event. It is a high-revenue commercial-stage biopharma and a serial acquirer/licensor, not an acquired entity.
• Does it ship a product embodying the claims? Yes, in substance. US 8,586,713 is the platform patent for Regeneron's bispecific-antibody format (common light chain + one heavy chain carrying the H435R/Y436F "FcΔAdp" Protein-A-ablating modification, purified by differential Protein A elution). The specification's own worked example is a CD20 × CD3 bispecific — and Regeneron's odronextamab (CD20 × CD3), marketed as Ordspono, is built on exactly this architecture. Regeneron's own later US 10,626,142 ("Purification platform for bispecific antibodies") expressly incorporates US 8,586,713 by reference and states that its format "is based upon a standard fully human IgG antibody having an improved pharmacokinetic profile and minimal immunogenicity." (Medium confidence on the specific approval status/history of odronextamab — I did not verify a regulatory-action date in this session; high confidence that Regeneron practices this platform commercially.)

Key point: the patent is practiced by its owner's own commercial pipeline. That is the single most important fact for the verdict below.

---

Assignment timeline

There is one and only one recorded conveyance for US 8,586,713 on the accessible record:

• Executed date: NOT RETRIEVED / recorded 2010-08-13 — Reel NOT RETRIEVED/Frame NOT RETRIEVED
  • Conveyance: Assignment of assignors' interest ("ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor(s): Eric Smith; Douglas MacDonald; Samuel Davis; Kara Louise Olson (all four, jointly)
  • Assignee: Regeneron Pharmaceuticals, Inc.
  • Correspondent of record: NOT RETRIEVED. *(Cross-check for the family, flagged as non-identical: FisherBroyles/FreePatentsOnline lists the correspondent on Regeneron's related US 10,626,142 as "Schwabe W&W, PC / Regeneron Pharmaceuticals, Inc." I note this only as a candidate for the family's recording firm. It is not verified as the correspondent on the '713 assignment, and I do not treat it as such.)*
  • Context: employee/inventor assignment to employer — the pre-filing-to-early-post-filing conveyance of inventors' rights to the company that filed the application. Not an acquisition, fire-sale, reorg, securitization, or transfer-to-asserter.

No post-issuance assignment of any kind was found. There is no record of a transfer to an IP-holding LLC, no security agreement, no merger, no change of name, no license recordation, no release, and no correction. Consistent with this, the same legal-events block shows no reassignment after 2010-08-13 and still lists Regeneron as current assignee, and the patent's status is "Active, expires 2031-07-22."

Family cross-check (corroboration, not the '713 record itself): the continuations claiming priority to this family — US 14/034,906 (→ US 9,309,326), US 15/058,026 (→ US 9,982,013), US 15/961,669 (→ US 11,168,111), US 17/496,263 (→ US 12,157,758), US 17/868,305 (→ US 2022/0389055) — all remain with Regeneron on the accessible records. No family member shows a downstream transfer. That is a portfolio hold, not a portfolio sale.

---

Timeline diagram

timeline
    title Ownership of US 8586713
    2009 : Provisional filed 26 Jun
    2010 : Nonprovisional filed 25 Jun
         : Inventors assign rights to Regeneron
    2013 : Patent issued 19 Nov
    2026 : Still held by Regeneron

---

NPE / troll-pattern signals

One flag worth stating, but not an NPE finding: Google Patents carries a "Family has litigation — First worldwide family litigation filed" marker pointing to Darts-IP family 42738900. Two cautions, both already raised earlier in this analysis: (i) it is family-level, not patent-level, and identifies no case, court, party, or date; and (ii) even if verified, the most likely asserter in this family is Regeneron itself — an operating company enforcing its own platform patent — which is the inverse of an NPE signal.

---

Verdict

Operating-company assertion — with one explicit caveat.

Justification. Every recorded ownership fact points away from NPE activity and toward a practicing operating company:

1. Single-link chain, held continuously by the original assignee. The only recorded conveyance is the 2010-08-13 assignment of the four inventors' interest to Regeneron Pharmaceuticals, Inc. (Davis, Smith, MacDonald, Olson → Regeneron). No post-issuance assignment exists on the accessible record; Regeneron remains the assignee of record with status "Active, expires 2031-07-22."
2. The assignee practices the claimed subject matter. Regeneron is a commercial-stage biopharma that manufactures bispecific antibodies built on precisely this common-light-chain / FcΔAdp platform (the specification's own CD20 × CD3 example maps onto Regeneron's odronextamab/Ordspono program), and its own US 10,626,142 expressly incorporates US 8,586,713 as the basis of that platform.
3. Zero of the eight NPE pattern signals are present, and the inverse signal (defensive aggregation) is also absent — i.e., the patent has not left the operating company's hands in either direction.

The caveat, stated plainly (this is where a strict reading of the taxonomy bites): the second prong of "Operating-company assertion" — that the owner is suing actual competitors on this patent — is not confirmed. As the earlier Litigation and PTAB sections of this analysis found, no complaint, docket entry, or PTAB proceeding naming 8,586,713 was surfaced, and no AIA trial against it exists. If the taxonomy is applied literally to require proven assertion, the correct bucket is "Insufficient data (only the original assignment)" — but the substantive NPE read is unambiguous: this is not an NPE-held patent, it is a practicing operating company's platform patent.

Verification gaps I am not papering over (each would require a direct Assignment Center pull, which my tooling could not reach this session):

• Reel/frame number for the 2010-08-13 assignment — not retrieved.
• Execution date (distinct from the 2010-08-13 recording date) — not retrieved.
• Correspondent of record (attorney/firm who filed the recording) — not retrieved; therefore signal #3 is "unclear," not "not present."
• Confirmation that no un-recorded or recently-recorded conveyance exists — my negative finding rests on web-search reach, not on a direct USPTO query.

Verify directly at: https://assignmentcenter.uspto.gov/ (search patent 8586713) and https://assignment.uspto.gov/patent/index.html — same underlying records; cross-reference the legal-events block at https://patents.google.com/patent/US8586713/en.

I'll search for the specific patent and its cited prior art references.

Let me get the specific cited references listed on US 8,586,713's face and its family records.

Let me pull the face-page reference lists for 8,586,713 specifically.

Method note and confidence caveat (read first)

I searched for the USPTO record of US 8,586,713. Two things must be stated up front, per the operating rules:

1. I could not retrieve the literal face-page "(56) References Cited" block of US 8,586,713 itself. The full text I have (Google Patents mirror of the USPTO grant) is truncated before the claims and before the front-page reference list. The reference list below is therefore assembled from authoritative records of the same patent family that carry the identical cited-art set:

   • PubChem's citation extract for US 9,982,013 B2 (family continuation, filed 2016-03-01) — https://pubchem.ncbi.nlm.nih.gov/patent/US-[9982013](/patent/9982013)-B2
   • Justia's "Referenced Cited" for US 11,168,111 B2 (family continuation) — https://patents.justia.com/patent/[11168111](/patent/11168111)
   • FreePatentsOnline's reference list for Regeneron's US 10,626,142 (which expressly incorporates US 8,586,713) — https://www.freepatentsonline.com/[10626142](/patent/10626142).html
   • Google Patents' citation block for the family member CN 102471378 A (which lists US 6,165,745 A and the Jendeberg article against the same disclosure) — https://patents.google.com/patent/CN102471378A/en

   I flag this as a proxy, not a verbatim face page. If a byte-exact (56) list is required, it should be pulled from USPTO PatentCenter for application 12/823,838.

2. I do not have verified, verbatim issued claim language for 8,586,713. The claim-cluster mapping used below is the specification-grounded reconstruction from the prior section, and claim numbers are unverified for this patent (the numbers I can see — e.g., claim 12 = bispecific antibody, claim 17 = method — come from the continuation US 11,168,111, and may not match 8,586,713). Treat every "claim X" reference below as "the corresponding claim cluster."

Also, consistent with the previous section: no USPTO/PACER/CAFC 2026 docket entry naming 8,586,713 was found. I am not aware of any inter partes proceeding or appeal against this specific patent.

---

The cited prior art set (union of family records)

Tier 1 — references whose identity and subject matter I can state with confidence

Tier 2 — additional references appearing on the family citation list but whose subject matter I could not verify with high confidence

These appear in the PubChem citation block for US 9,982,013 B2 (truncated list) but I did not retrieve or verify their content: GB 2197322 B, GB 2197323 B, US 5,807,706 A, US 2003/224000 A1, WO 2004/009618 A2, US 7,169,903 B2, US 7,220,356 B2, US 2007/148165 A1, US 2007/196363 A1, US 7,262,028 ("7262028-…", list truncated).

I will not guess what these disclose. If an element-by-element §102 chart on these references is needed, each must be pulled and read; I flag them as open items rather than fabricate their content.

Also self-listed in the family record: US 8,586,713 itself (cross-cited within the continuations) and U.S. Appl. No. 12/823,838 — these are family references, not prior art.

---

§ 102 analysis, reference by reference

Governing principle: anticipation requires a single reference disclosing every element, arranged as claimed. The distinguishing element of this family is the deliberate, minimal, human-derived CH3 point modification — H435R (EU; H95R IMGT), optionally + Y436F (Y96F IMGT), the "FcΔAdp" dipeptide — that removes Protein A binding in a heterodimeric, native-format human bispecific antibody, and the resulting differential Protein A purification. No single cited reference discloses that element in combination with a human IgG1/IgG2/IgG4 bispecific format.

1. Lindhofer references (items 1–6, 9) — no anticipation; strong §103

• What they disclose: co-expression of a mouse IgG2a and rat IgG2b antibody yielding parental homodimers plus a heterodimer; the heterodimer is separated on Protein A because rat IgG2b naturally does not bind Protein A while mouse IgG2a does, and the heterodimer elutes at higher pH than the mIgG2a homodimer (Lindhofer 1995, pp. 219–225).
• Why they do not anticipate: they use naturally different, non-human isotypes from different species. They do not disclose (a) a human IgG1/IgG2/IgG4 CH3 domain, (b) any engineered point modification to ablate Protein A binding, or (c) the H435R/Y436F dipeptide. Claim language reciting human CH3 regions and/or a modification in the second CH3 is therefore not anticipated.
• Claims that could be most exposed (§ 102, and only if drafted with no human/modification limitation): the broadest isolation-method cluster — "isolating from a mixture of antibodies a bispecific antibody having differentially modified … CH3 domains … based on differential affinity for Protein A." If such a claim were read to cover any species pair with differential Protein A binding, Lindhofer 1995 becomes a serious single-reference §102 candidate. As issued, the human-IgG and modification recitations should defeat it — narrowing this to §103.
• §103 role: Lindhofer (differential Protein A isolation of a bispecific) + Jendeberg (Arg435 is the single determinant of Protein A non-binding) is the natural examiner combination the applicant had to overcome, and both appear on the family citation list.

2. US 7,183,076 (item 7) — no anticipation; §103

• Discloses: engineered "knob" and "hole" in the Fc to favor heavy-chain heterodimerization (expressly acknowledged in 8,586,713's Background as prior art). It does not disclose or suggest ablating Protein A binding via a CH3 point mutation, nor a Protein A-based differential purification — indeed the patent's own Background criticizes knobs-into-holes for leaving significant homodimer contamination requiring further purification.
• Exposure: potentially the heterodimeric-Fc composition cluster (a binding protein with one modified and one unmodified CH3 forming a heterodimeric Fc) — but the recitation of the Protein A-affinity-differential modification and the specific ~0.5–1.4 pH-unit elution shift distinguishes it. §103 only.

3. Jendeberg et al. 1997 (item 10) — no full-claim anticipation; the most relevant §102(a)/§103 teaching on the core mutation

• Discloses: that a single residue, Arg435 (EU), determines IgG3's failure to bind Protein A, by engineering Fc1 and Fc3 domains from human IgG.
• Exposure — this is the single most relevant prior-art item for the mutation element. If any claim is drafted narrowly to a CH3 domain comprising an H435R (or H435R/Y436F) modification that reduces/eliminates Protein A binding, Jendeberg is a §102 anticipation candidate for that element, and possibly for such a claim standing alone if the claim is not tied to a bispecific/heterodimeric context.
• Why it does not anticipate the main claims: Jendeberg concerns isolated human Fc1/Fc3 fragments studied for subclass-specific Protein A binding; it does not disclose (a) a bispecific antigen-binding protein, (b) two different antigen-binding regions on two heavy chains, (c) a heterodimeric native-format antibody, or (d) the ionic-modifier-augmented pH-step purification. It is therefore a §103 reference against the main composition and method claims, and the strongest one the applicant faced.
• Examiner-artifact caution: Jendeberg and Lindhofer 1995 are both discussed in the specification's Background, so they likely appear as applicant-cited references; that does not reduce their §102/§103 weight.

4. US 6,165,745 (item 8) — no anticipation

• Discloses: recombinant production of immunoglobulin-like domains in prokaryotic cells. No bispecific antibody, no CH3 Protein A differential, no H435R/Y436F. It is cited on the family record (CN 102471378 A) as background. No §102 exposure to any substantive claim; at most a §103 background reference for expression-host limitations (CHO/COS/293/HeLa/PER.C6).

5. Tier-2 references — cannot be assessed

My search did not surface verified content for GB 2197322 B, GB 2197323 B, US 5,807,706 A, US 2003/224000 A1, WO 2004/009618 A2, US 7,169,903 B2, US 7,220,356 B2, US 2007/148165 A1, US 2007/196363 A1, or US 7,262,028. Several of these (notably US 7,220,356, if it is the Protein A chromatography/purification reference the number suggests) could plausibly bear on the purification-method claims and deserve individual reading. I am explicitly not asserting a §102 or §103 position on these until their content is confirmed.

---

Bottom line

• Strongest single-reference §102 candidate against a narrow mutation-only claim: Jendeberg 1997 (Arg435/H435R determinant).
• Strongest §102 candidates against the broadest isolation-method claim (if read species-agnostically): Lindhofer 1995 and the Lindhofer patent family (US 5,945,311; 5,985,276; 6,551,592; 6,994,853; 7,018,632; 2002/0051780).
• No single cited reference anticipates the core claims as I understand them, because none discloses a human IgG1/IgG2/IgG4 heterodimeric bispecific bearing the H435R/Y436F (ΔAdp) Protein A-ablating modification together with the ionic-modifier-augmented pH-step isolation.
• The realistic validity battleground is §103, with the combination Lindhofer 1995 (or the Lindhofer patents) + Jendeberg 1997, optionally + Genentech's US 7,183,076.
• Two open items I will not paper over: (i) the byte-exact (56) list of 8,586,713 itself, and (ii) the subject matter of the Tier-2 references. Both require retrieval from USPTO PatentCenter / the printed front page before any element-by-element §102 chart can be treated as final.

Corroborating citation (that this prior-art set is shared across the family): US 11,168,111's Referenced Cited list — https://patents.justia.com/patent/11168111 — and FPO's US 10,626,142 list — https://www.freepatentsonline.com/10626142.html — both carry the Lindhofer references plus Jendeberg 1997 and Lindhofer 1995, alongside US 8,586,713 Davis et al., Nov. 19, 2013 as a family cross-citation.

I'll ground this analysis by retrieving the actual prior-art references and prosecution history for US 8,586,713.

Let me retrieve the specific cited prior art references and any obviousness/validity challenges on record.

Obviousness Analysis — US 8,586,713 B2 under 35 U.S.C. § 103

0. Inputs, caveats, and cross-reference notes

A note on the "Prior Art section." The instruction says to use the results from the Prior Art section of this page. The fetched full text I was given does not include a Prior Art / "References Cited" list — it terminates mid-Summary (at "…does not result in a non-native"). I therefore reconstructed the prior-art set in two ways:

1. References the applicant themselves cited in the specification (these appear verbatim in the text provided).
2. External retrieval of those references' bibliographic data and of later documents that characterize them.

I flag this rather than pretend I was given a curated prior-art table. Any reference below not corroborated by a search result is marked [my knowledge — verify].

Cross-reference to earlier sections (no contradiction, two gaps):

• The earlier Patent summary correctly warns the full text was truncated before the claims. That is still true here — I do not have verbatim claim language, so claim-scope statements are grounded in the specification's Summary/definitions and flagged as reconstructions.
• The earlier Litigation summary found no confirmed case for '713. Nothing in this round changes that; the Darts-ip "family has litigation" flag (family 42738900) remains unverified.
• Minor internal inconsistency to note: the operating header says today is 2026-09-20, while the task text says April 26, 2026. Nothing in the analysis turns on this; the patent is listed active to 2031-07-22.

Verbatim admissions in the specification that matter enormously below (these are the applicant's own words, not my paraphrase):

"It has been reported (Jendeberg, L. et al. (1997) J. Immunological Meth. 201:25-34) that the inability of IgG3 to bind Protein A is determined by a single amino acid residue, Arg435 (EU numbering; Arg95 by IMGT)… It is therefore possible, instead of IgG3, to use an IgG1 sequence in which His435 is mutated to Arg."

"It has been reported that this double mutant still does not bind Protein A (Jendeberg, L. …)."

"Methods for isolating such bispecific antibodies are provided that are relatively rapid and simple in execution."

These are specification-level admissions of what the art already taught — highly probative under § 103.

---

1. Governing framework

• Graham v. John Deere Co., 383 U.S. 1 (1966): scope/content of prior art; differences; level of ordinary skill; secondary considerations.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): combination of known elements with predictable results is obvious; "obvious to try" applies where there is a finite number of identified, predictable solutions.
• MPEP §§ 2141–2144.07: motivation may come from the art, the nature of the problem, or common sense; a result-effective variable optimized by routine experimentation is obvious.
• Biotech predictability: for antibody engineering and protein purification — mature, predictable arts — courts apply KSR with relative rigor (cf. In re Kubin, 561 F.3d 1351 (Fed. Cir. 2009)).

---

2. Hypothetical person of ordinary skill in the art (POSA)

A POSA here would have a Ph.D. (or M.S. + ~5 years) in immunology, molecular biology, or biochemistry, with hands-on experience in (i) recombinant antibody expression and (ii) Protein A/affinity chromatography. This matters because the invention is squarely a protein-engineering + purification-process invention, and the POSA would be comfortable with site-directed mutagenesis of Fc, CHO expression of multi-chain antibodies, and pH/conductivity-based elution optimization. The claimed subject matter requires no new science — only the application of well-characterized methods.

---

3. Prior-art inventory (all pre-2009-06-26)

---

4. Claim-by-claim obviousness grounds

Below, "Ground N" = a § 103 rejection. Motivation-to-combine is stated for each. Claim identities are reconstructed from the Summary/definitions, since I lack verbatim claims (flagged, per § 0).

Independent Claim 1 (composition: bispecific antigen-binding protein; first HC with a human IgG1/2/4 CH3, second HC with a human IgG1/2/4 CH3 carrying a modification that reduces/eliminates Protein A binding)

Ground 1 — Lindhofer (PA-1) in view of Jendeberg (PA-2).

What PA-1 teaches: Express two different heavy chains with a common light chain in one cell; exploit a differential Protein A affinity between two Fc isotypes so that the heterodimer elutes at a different pH from the homodimers and is purified in a single step.

What PA-1 lacks: It used a non-human mouse/rat pair, and the applicant's own specification concedes the mouse/rat format is immunogenic ("likely to provoke an immune response in the patient, which could have deleterious side effects (e.g. 'HAMA' or 'HARA' reactions), and/or neutralize the therapeutic").

What PA-2 teaches/why it fills the gap: Arg435 is the single residue determining IgG3's non-binding to Protein A; a residue at the same EU position in IgG1/2/4 is His435. PA-2 thus hands the POSA a specific, finite, predictable molecular solution for "make a human IgG1 chain that does not bind Protein A": mutate H435R.

Motivation to combine: The problem PA-1 solves (easy isolation of a bispecific heterodimer) and the problem PA-2 addresses (the Protein A/IgG-binding determinant) are in the same field of endeavor — Fc–Protein A interaction — and the combination is a substitution of one known isotype-Differential-binding pair with a human pair, achieving PA-1's known result with a human, less-immunogenic scaffold. The applicant's own spec supplies the reason: "It is therefore possible, instead of IgG3, to use an IgG1 sequence in which His435 is mutated to Arg." A POSA seeking a human bispecific isolable on Protein A has a finite set of options, and PA-2 identifies the leading one. KSR / MPEP § 2144.07.

Reasonable expectation of success: High. Only one CH3 domain is changed at one surface-exposed position; PA-2 confirms the mutant is expression-competent Fc, and the '713 spec repeatedly notes the mutation is distant from the Fc dimer interface. Result is predictable.

Ground 2 — PA-1 + PA-2 + PA-4/PA-6 (knobs-into-holes/CH3 heterodimerization). Even if one argued that PA-1's "reliance on natural species-restricted chain pairing" is imperfect, a POSA would turn to the CH3 heterodimerization tools (PA-4, PA-5, PA-6, PA-7) to bias formation of the heterodimer. Combining the PA-2 H435R Protein-A "knob" with knobs-into-holes merely applies known techniques for their known purposes (KSR).

Conclusion on Claim 1's genus: Prima facie obvious. The breadth of the claim's functional language — "a modification that reduces or eliminates binding… to protein A" — is a result-oriented recitation of an outcome the art already achieved in a different isotype. Where the POSA is told a single residue controls Protein A binding, claiming any modification achieving that result is an obvious genus.

Dependent claim(s): the second CH3 modification is H95R (IMGT) / H435R (EU)

Ground 3 — PA-2 alone or PA-2 in view of PA-1. This is the single mutation PA-2 reports and the mutation the applicant's own specification says it is "therefore possible" to make. Strongly obvious; arguable § 102 anticipation by PA-2 on the merits of the modification (though PA-2 is a journal article describing a naked Fc, not a bispecific, so § 103 via PA-1+PA-2 is the cleaner ground).

Dependent claim(s): further comprising Y96F (IMGT) / Y436F (EU) → the H435R/Y436F "FcΔAdp" dipeptide

Ground 4 — PA-2 explicitly discloses the R435/F436 double mutant. The '713 specification states outright that "this double mutant still does not bind Protein A (Jendeberg…)" and that the resultant sequence "is identical to that of IgG3." The claimed dipeptide is therefore the prior-art dipeptide, and its Protein-A-non-binding property was known. The only filigree the applicant adds — the "immunologically invisible / no non-native 9-mer" rationale — is a rationale, not an unexpected result, and (see Ground 8) immunogenicity mitigation was itself a recognized design goal. Obvious.

Dependent claim(s): further isotype-normalizing modifications

(IgG1: D356E, L358M, N384S, K392N, V397M, V422I; IgG2: N384S, K392N, V422I; IgG4: Q355R, N384S, K392N, V397M, R409K, E419Q, V422I — EU numbering)

Ground 5 — PA-2 in view of routine protein engineering / isotype sequence comparison. These are conservative substitutions that replace non-IgG3 residues with the corresponding IgG3 (or "consensus") residues, i.e., they re-create native IgG3 sequence locally. The applicant's own spec concedes the purpose: "the modification is not found in nature associated with the particular IgG isotype, the modified sequence is locally identical with a wild-type sequence of a different IgG isotype." Choosing among the ~1–7 positions where IgG1/2/4 differ from IgG3 to reduce immunogenicity risk is routine optimization of a result-effective variable. Obvious.

Dependent claim(s): chimeric / mixed-isotype CH1, CH2, CH3 domains

Ground 6 — PA-4/PA-5/PA-6 (knobs-into-holes and Fc engineering) + antibody-engineering textbooks. Chimeric constant domains were routine by 2009, and the specification states the chimeras are non-immunogenic "human/human" combinations — a design choice. Obvious.

Dependent claim(s): light chain (human kappa/lambda); CDRs/FRs; human heavy chain; affinity ranges

Ground 7 — PA-1 (or PA-7 common-LC approach) + standard immunology/phage-display practice. Selecting a common light chain that pairs with both heavy chains is expressly the stated strategy in the '713 spec and was known (Merchant 1998 and the common light-chain literature). See KSR (combination of known techniques for known purpose). Any affinity limitation ("micromolar, nanomolar, or picomolar") is a result-effective variable achieved by standard affinity maturation. Obvious.

Dependent claim(s): non-immunogenicity; "lacks a non-native human T-cell epitope"; "proteolysis cannot form a ~9-mer / 13–17-mer immunogenic peptide"

Ground 8 — PA-2 + MHC class II peptide-binding literature (PA-11). Once one has chosen H435R/Y436F because it recreates IgG3 sequence, the absence of a non-native T-cell epitope follows as a matter of intrinsic property, not inventive contribution. De-immunization / T-cell-epitope avoidance was a well-known 2000s design objective. The claim recites a goal and its inherent consequence, which does not confer patentability where the structure was obvious. Obvious.

Independent method claim (making a bispecific: obtain HC1 (IgG1/2/4 or chimeric) + HC2 with the Protein-A-ablating CH3 modification + a light chain that pairs with both; introduce into a cell and express)

Ground 9 — PA-1 + PA-2 + PA-7 (common LC) + routine recombinant expression (CHO/HEK). Each step — encoding heavy chains, co-transfecting, expressing in CHO — is standard. The only arguable novelty is which mutation to place on HC2, and PA-2 supplies it. Obvious.

Independent method claim (isolating the bispecific from a disrupted cell / medium / mixture using Protein A)

Ground 10 — PA-1 + PA-2. PA-1 already isolates a bispecific using Protein A based on differential Fc binding. Substituting the PA-2 human variant HC for the rat Fc retains PA-1's exact method for its exact purpose. Obvious.

Dependent purification claim(s): elution at about pH 3.9–4.5 (≈4.2); ionic modifier 0.5–1.0 M (e.g., NaCl); pH gradient or step

Ground 11 — PA-1 + PA-9/PA-10 (Gagnon and salt-modulated Protein A elution). This is the applicant's strongest non-obviousness redoubt and needs the most careful treatment.

• The difference to overcome: The spec reports that with a plain linear pH gradient, H435R/Y436F IgG1 eluted at ~pH 3.9 vs ~pH 3.7 for the IgG1 homodimer — "insufficient to achieve a satisfactory separation," and "not reproducible in a predictable manner." The applicant's fix was to add 0.5–1.0 M ionic modifier and use a step/gradient, widening the gap to ~1.2 pH units (heterodimer ≈ 4.2 vs homodimer ≈ 3.0).
• Why it is still prima facie obvious: Adding salt/conductivity to a Protein A (or any affinity) elution buffer to modulate retention and sharpen resolution is a paradigm, textbook-level technique (PA-9, PA-10; and the same family's own later '142 patent concedes the routine use of NaCl and, later, chaotropes). Optimizing the salt concentration "until a maximal pH distance is reached" (the spec's own wording) is routine experimentation on a result-effective variable — precisely the MPEP § 2144.07 / In re Applied Materials fact pattern. The specific numbers 0.5–1.0 M NaCl and pH ~4.2 represent a range and an optimum discovered by routine testing, not an unpredictable result.
• Counter-argument the applicant would raise: The failure of the plain gradient was itself a surprise, and the salt-widened separation was non-obvious. Rebuttal: the POSA would expect salt to perturb the electrostatics of Protein A–Fc binding; the direction and rough magnitude of the effect are predictable; and the specific elution pH is a result-effective variable. A robust rejection is available but this is the claim set most worth a declaration-based § 103 rebuttal (unexpected magnitude / failure of others).

Independent claim (binding protein comprising a heterodimeric Fc, one modified CH3 + one unmodified, eluting 0.5–1.4 pH units higher than an unmodified counterpart)

Ground 12 — PA-1 + PA-2 + PA-9. The pH-shift magnitude is a measured consequence of a known mutation and known chromatography. Stating the observed ΔpH does not patentably distinguish. Obvious.

Mouse-Fc embodiments (mIgG2a/mIgG1; modifications at 252/254/256/258; H435R/H436F; PTTTK, TTTK, etc.)

Ground 13 — PA-3 (Ey 1978, mouse subclass Protein A differences) + PA-1 + PA-2. The mouse IgG1-vs-IgG2a Protein A differential was classic (PA-3). Applying the PA-2 "single residue" insight and PA-1's separation scheme to mouse IgGs is a species substitution of known Fc reagents, expressly suggested by PA-1 (mouse/rat) itself. Obvious.

CD20 × CD3 bispecific claims (IgG1 + IgG1ΔAdp; EC50 ≈ 2.8–3.2 × 10⁻¹² M in a Raji killing assay; ≤ ~1–10% bystander killing in a 293-cell assay; bystander measured 10⁻⁸–10⁻¹⁴ M)

Ground 14 — PA-1 + PA-2 + PA-8 (BiTE) + anti-CD20/anti-CD3 antibody art. CD3-retargeting bispecifics against CD20⁺ lymphomas were known in concept (PA-8 and the broader BiTE/CD20×CD3 literature). The claimed numeric endpoints are inherent functional properties of the assembled molecule; an EC50 and a low bystander-killing rate discovered by running a standard cytotoxicity assay are not patentably distinct from the molecule itself. Absent evidence that the specific EC50/bystander window is unexpected relative to other CD20×CD3 formats, these limitations do not confer § 103 patentability. (The "no bystander killing" property, however, may be a meaningful secondary consideration if a comparable prior-art format showed more bystander lysis — see § 6.)

---

5. Motivation-to-combine — consolidated

The overarching motivation is captured by the applicant's own framing: "There remains a need for a bispecific antibody format… that minimizes some or all of the disadvantages" of BiTE, mouse/rat, and knobs-into-holes. Addressing a known, articulated need by substituting one known element (PA-2's mutation) into a known scheme (PA-1) is the KSR paradigm.

---

6. Secondary considerations (objective evidence)

• Long-felt need / commercial success. Real and potentially substantial: the '713 family became the purification platform basis for Regeneron's bispecific antibodies (fetched '142 patent: "One such format is based upon a standard fully human IgG antibody… see U.S. Pat. No. 8,586,713"; and a third-party tracker attributes ~$243M value to '713). If the applicant can show the human bispecific platform satisfied a need that PA-1's immunogenic format could not, that weighs toward non-obviousness — but only if the nexus is to the claimed advance and not to later development. This cuts most strongly for the narrowed purification claims (Grounds 11–12) and the low-bystander-killing CD20×CD3 claim (Ground 14).
• Unexpected results (applicant's best argument). The reported failure of the plain pH gradient (3.9 vs 3.7) and the magnitude of the salt-widened window (~1.2 pH units) is the single most defensible non-obviousness point. It is a magnitude and reliability argument on a result-effective variable, so its strength depends on whether the POSA would have expected salt to help. I assess this as moderate, not decisive.
• Teaching away. None found. PA-2 affirmatively points toward H435R; PA-1 affirmatively points toward Protein-A-based bispecific purification. No reference disparages the combination. If anything, the '713 specification itself reads as an instruction manual to combine them.
• Copying / licensing / praise. Not assessed; would require the litigation/PR record the earlier sections could not confirm.

---

7. Bottom line

Overall: On the record available, the broad composition and method claims would likely be held prima facie obvious under § 103 by Lindhofer (PA-1) in view of Jendeberg (PA-2), with knob-into-holes (PA-4/5/6/7) and chromatography practice (PA-9/10) supplying the remaining limitations. The plain-pH-gradient-failure / salt-widening evidence (Grounds 11–12) is the most viable non-obviousness theory, and the CD20×CD3 bystander-killing data the most viable objective-evidence theory.

---

8. Confidence and what would change the conclusion

• High confidence: the reference content of PA-1, PA-2, PA-4, PA-5, PA-6, PA-7, PA-8 (all text-confirmed or search-confirmed); the specification admissions quoted in § 0; the KSR/MPEP framework.
• Medium confidence: the claim-group mapping — because verbatim claims are still unavailable (the truncation flagged in the earlier section persists). If the issued independent claim requires, e.g., a specific sequence (SEQ ID NO:2/4/6) or a specific ΔpH or ionic-modifier range as a structural (not functional) element, Grounds 1–2 weaken and Grounds 11–12 become decisive.
• Low confidence / verify: PA-3 (Ey 1978) and PA-11 (MHC class II 9-mer literature) are cited from my own knowledge; their dates and exact content should be confirmed before use in a formal rejection.
• Would change the analysis: (a) actual claim text revealing closed-Markush or sequence-level limitations; (b) a PA-2 passage that teaches away from the H435R single mutant in an IgG1 context; (c) evidence that the salt-widened pH separation was unexpected in kind, not degree; (d) the real Darts-ip/litigation record (family 42738900), which could reveal invalidity contentions already on file.

Explicit non-fabrication note: I have not invented case numbers, party names, or claim language. Where the record did not supply something (the Prior Art list, the literal claims, a confirmed litigation), I have said so rather than fill the gap.