RBX1 Target Analysis & Binder Design Strategy

🔬 GEM × Adaptyv Bio 2026 · Binder Design Competition

RBX1 Target Analysis
& Binder Design Strategy

Sequence · Structure · Function · Literature Review · Design Rationale

Target Protein

RBX1 / ROC1 (UniProt P62877)

Protein Size

108 amino acids · ~12 kDa

Competition Deadline

March 26, 2026

2

GEM × Adaptyv Bio 2026 — Competition Overview

Competition Brief

Competition Specifications

Parameter	Specification
Target protein	RBX1 (P62877, 108 aa)
Binder type	De novo only (no motif scaffolding)
Max binder length	250 aa
Novelty threshold	≥ 25% sequence edit distance from UniRef50
Max submissions	100 sequences per team
Experimental testing	300 total designs — expression + BLI binding affinity
Results announced	April 26, 2026 (Rio de Janeiro)
IP / licensing	Submissions made public on Proteinbase (ODC-ODbL)

🏆 Prizes

1st place: $1,000 USD · Runner-up: $100 USD · All participants receive experimental binding data

Core Challenges

RBX1 is only 108 amino acids — very small target, no large binding pocket
Contains 3 structural Zn²⁺ ions — chelators would be non-selective
Primary epitope (helix α2, E2-binding surface) competes with Glomulin
≥25% sequence novelty required — cannot directly repurpose GLMN fragments
Binder must be expressible and detectable by BLI

🔬 Experimental Readout

Adaptyv Bio will test designs for soluble protein expression (E. coli / yeast display) and binding affinity by Bio-Layer Interferometry (BLI).

⚡ Our Opportunity

No direct RBX1 RING domain binder has been reported in the literature. Glomulin (GLMN, PDB 4F52) provides a natural protein-based template with validated Kd ~36 nM — the ideal starting point.

3

RBX1 — Protein Identity & Sequence Features

Sequence & Identity

108

amino acids

~12

kDa MW

3

Zn²⁺ binding sites

74+

PDB structures

Domain Architecture (108 aa)

Full-Length Sequence — Color-Coded Key Residues

M A A A M D V D T P S G T N S G A G K K R F E V K K W N A V A L W A W D I V V D N C A IC R N H I M D LC I EC Q A N Q A S A T S E EC T V A W G VC N H A F H F HC I S RW L K T R Q VC P L D N R EW E F Q K Y G H
■ Zn²⁺-coordinating Cys/His residues ■ Key E2-binding residues (Trp87)

Post-Translational Modifications

• Met-1 cleavage → processed N-terminus is Ala-2
• N-terminal acetylation (Ala-2)
• Thr-9 phosphorylation (function under investigation)

Gene Aliases

RBX1 / ROC1 / RNF75 / HRT1
UniProt: P62877 · Gene ID: 9978
Chromosomal locus: 22q13.2

Subcellular Localization

Cytoplasm + Nucleus
Highly expressed in heart, skeletal muscle, testis;
constitutively expressed, no tissue restriction

4

RING-H2 Domain — Structural Architecture

Structural Detail

Three-Zinc Cross-Brace Coordination (Unique to RBX1)

Zn1 & Zn2 — Classic RING

Standard cross-brace coordination found in all RING E3 ligases; maintains RING fold stability

Zn3 — RBX1 Unique

Formed by a ~20 aa insertion loop (C75·C77·C94·C97); distinguishes RBX1 from all other RING proteins; caps the RING domain

Val38/Val39 Hinge — Conformational Gating

Dynamic RING Rotation

The RING domain undergoes 60°–170° rigid-body rotation via the Val38/Val39 hinge:

Closed state: un-neddylated cullin; RING domain auto-inhibited
Open state: cullin NEDD8-ylated; RING rotates to expose E2-binding surface
Only the open state productively recruits CDC34~Ub for substrate ubiquitination

Key Secondary Structure Elements

Element	Residues	Function
β-strand (N-term)	19–25	Inserts into CUL1, forms intermolecular β-sheet
Helix α1	~60–72	Zn2 coordination, structural support
Helix α2	~85–97	E2 recruitment surface — primary design target
Insertion loop	53–74	Unique Zn3 site formation; RBX1-specific

🔑 Key Design Constraints

• Any binder must not disturb the 3 Zn²⁺ coordination sites
• Conformational dynamics require design against an NMR ensemble (2LGV, 20 conformers)
• The E2-binding surface shifts between CRL states

5

Key PDB Structures — 74+ Total Entries

Structural Database

PDB ID	Complex	Resolution	Method	Significance / Design Value	Priority
2LGV	RBX1 RING domain — isolated (free)	NMR	Solution NMR	Only RBX1 structure without cullin; 20-conformer ensemble; reveals dynamics; best baseline for binder design	★★★★★
4F52	Glomulin – RBX1 – CUL1	3.0 Å	X-ray	Glomulin (natural protein inhibitor) masks E2-binding surface (~880 Å²); Kd ~36 nM; gold-standard design template	★★★★★
1LDD	CUL1–RBX1–SKP1–SKP2 (SCF complex)	2.0 Å	X-ray	Highest-resolution SCF structure; atomic-level CUL1–RBX1 interface; Zheng & Schulman (2002)	★★★★
4P5O	RBX1–UBC12~NEDD8–CUL1–DCN1	3.1 Å	X-ray	Trapped neddylation transition state; reveals RING open conformation; E2~UBL binding face detail	★★★★
5N4W	CUL2–RBX1–EloBC–VHL (CRL2 pentamer)	3.1 Å	X-ray	Closed→open conformational trajectory; VHL PROTAC platform relevance; novel RBX1 pose	★★★
8Q7H	CUL9–RBX1 (1.8 MDa hexamer)	3.4 Å	Cryo-EM	2024 structure; non-canonical CUL9 assembly; exposes novel RBX1 surfaces not seen in CRL1/2/4	★★★

Recommended Starting Structures for Binder Design

2LGV (NMR ensemble) — captures RING domain flexibility; use for pocket identification and conformational sampling.
4F52 (GLMN–RBX1, 3.0 Å) — provides atomic-resolution contacts for the E2-binding surface; use as the primary docking/design template.
Combining both maximizes coverage of accessible conformations.

AlphaFold2 Structure

AF2 predicted structure available as AF-P62877-F1 from UniProt. Useful as a complementary reference for computational design pipelines.

6

Biological Function — Catalytic Core of All Cullin-RING Ligases

Biological Function

Cullin-RING E3 Ubiquitin Ligase (CRL) Superfamily

~20%

of all cellular ubiquitination

350+

known protein substrates

8+

cullin family members

Key Substrates & Pathways

Substrate	F-box / Adaptor	Biological Effect
p27 (CDKN1B)	SKP2	Cell cycle G1→S entry
CDT1	CDT2	DNA replication licensing
β-catenin	β-TrCP	Wnt pathway off-switch
IκB	β-TrCP	NF-κB activation
HIF-1α	VHL (CRL2)	Hypoxia response regulation
Cyclin E	FBXW7	G1/S checkpoint control
RhoB	VHL (CRL2)	Tumor suppressor degradation

RBX1 Works with Three Distinct E2 Enzymes

• UBE2M/UBC12 — cullin neddylation (activation step)
• UBE2D/UBCH5 — ubiquitin chain initiation
• UBE2R/CDC34 — polyubiquitin chain elongation (degradation signal)

Embryonic Essentiality

RBX1 knockout mice die at E6.5–E7.5 (post-implantation lethality). Confirms indispensable role in early development. Partial rescue by p27 co-deletion. This also implies that inhibitors need to be precisely targeted to achieve a therapeutic window.

7

RBX1 in Cancer — Overexpression, Dependency & Therapeutic Window

Oncology Target

RBX1 Is Overexpressed Across Multiple Tumor Types

Cancer Type	Overexpression	Clinical Correlation
Anaplastic Thyroid (ATC)	62.6% of cases	Promotes PKM splicing, Warburg effect, metastasis
Lung Cancer (NSCLC)	High	Unfavorable prognosis marker
Gastric Cancer	High	Independent prognostic factor, poor survival
Bladder Cancer	Elevated	Disease progression correlate
Breast Cancer	Elevated	Validated in MDA-MB-231/461 cell lines
Hepatocellular Carcinoma	Elevated	Accelerated RhoB tumor suppressor degradation
Glioblastoma	High	siRNA reduces colony count 5–10-fold

Three-Phase Anti-Tumor Response to RBX1 Knockdown

Mechanism — Why Cancer Cells Are Addicted to RBX1

Over-Degradation of Tumor Suppressors

→ p27/p21 hyper-degradation (unchecked proliferation)
→ HIF-1α over-stabilization (hypoxia tolerance)
→ RhoB degradation (liver cancer metastasis)
→ Cyclin E/Myc accumulation (genomic instability)

Indirect Validation: MLN4924 Clinical Progress

MLN4924 (Pevonedistat) inhibits NAE (NEDD8-activating enzyme), indirectly blocking all CRLs including those containing RBX1. Active in Phase I/II/III for AML, MDS, and solid tumors. Validates the entire ubiquitination–neddylation axis as druggable.

Therapeutic Window

• Cancer cells are consistently more sensitive than normal cells
• Normal cells show relative tolerance to RBX1 loss
• Direct RBX1 inhibition may offer better selectivity than global CRL shutdown by MLN4924

Key References

Duan H et al. (2009) Cancer Res 69:4974 [PMID:19487272]
Song Y et al. (2023) Cell Biosci 13:30 [PMC9945352]

8

Key Binding Interfaces & Hotspot Residue Analysis

Binding Interfaces

① E2-Binding Surface (Helix α2) ⭐ Primary Target

Ile44 · Trp87 · Arg91 · Pro95 · Leu96 · Ala43 · Arg46

Core E2 recruitment interface, ~25×35 Å. Trp87 (W87A → completely abolishes di-Ub synthesis) and Arg91 (R91A → eliminates all activity) are the critical anchor residues. RBX1 binds CDC34~Ub with Kd = 18.6 µM (vs free E2 >1 mM) — ubiquitin itself participates in the recognition.
Reference structures: 2LGV (NMR) + 4P5O (transition state)

② Glomulin Inhibitory Interface — Natural Protein Template

RBX1: Ala43 · Ile44 · Ile54 · Ala58 · Glu55 · Gln57 · Trp87 · Pro95 · Leu96

GLMN uses a HEAT-like repeat fold to mask the RBX1 E2-binding surface: buried area ~880 Å², affinity Kd ~36–45 nM. Hydrophobic core (Met472/Leu567/Val571 → Ile44/Trp87) + electrostatic complement (Arg479 → Glu55). Gln57 is RBX1-specific, enabling selective design over RBX2.
Reference structure: 4F52 (3.0 Å)

③ CUL1-Binding Interface (N-terminal β-strand)

RBX1: Trp27 · Val30 · Leu32 · Trp33 (β-strand insertion)

RBX1 N-terminal β-strand (residues 19–25) inserts into the CUL1 α/β-subdomain forming an intermolecular antiparallel β-sheet. Universal interface for RBX1 binding to all cullins. Disrupting this would block CRL assembly entirely.
Reference structure: 1LDD (2.0 Å). Challenge: conserved across all 8 cullin paralogs — poor selectivity potential

④ DCN1-Assisted Neddylation Interface

RBX1–UBC12–NEDD8–CUL1–DCN1 pentameric transition complex

DCN1 (DCUN1D1) stabilizes the RBX1–UBC12~NEDD8 transition state, facilitating cullin neddylation. DCN1 small-molecule inhibitors (HZX-960, DI-1859) validated in preclinical studies. Locking this transition state is a potential allosteric strategy.
Reference structure: 4P5O (3.1 Å transition state)

      🎯 Core Conclusion: The E2-binding surface (interfaces ①②) is the optimal target for binder design. Glomulin provides natural protein-level validation (Kd ~36 nM), proving this surface is fully tractable for protein binders. RBX1 Gln57 offers a selectivity handle to discriminate RBX1 from RBX2.
    

9

Glomulin: Natural Protein Inhibitor & Design Template (PDB 4F52)

Design Template

GLMN–RBX1 Interface — Atomic-Level Contacts

GLMN Residues	RBX1 Residues	Interaction Type
Met472, Leu567, Val571	Ile44, Trp87	Hydrophobic core (primary anchors)
Tyr480, Ile483	Pro95, Leu96	Hydrophobic stacking
Arg479, Arg574	Glu55	Electrostatic (salt bridges)
Asn476, Lys425	Gln57, Asn47	Hydrogen bonds (selectivity key)
Asp429	Arg91	Salt bridge (catalytic residue lock)

Interface Parameters

880

Å² buried interface area

36

nM Kd (GLMN)

3.0

Å crystal resolution

Physiological Role of Glomulin

GLMN is a protein associated with glomuvenous malformation (GVM), a benign vascular skin tumor. Loss-of-function GLMN mutations prevent RBX1 binding, deregulating CRL activity and causing the disease phenotype. GLMN is a native competitive inhibitor of the RBX1 E2-binding surface, achieving Kd ~36 nM — ~30× tighter than free CDC34 E2.

Design Strategy Derived from GLMN

No need to copy the GLMN HEAT-repeat fold architecture — only the equivalent atomic contacts need to be recapitulated on a de novo scaffold
Functional requirements:
— Hydrophobic burial of Ile44 (core anchor)
— H-bonds to Glu55 / Gln57 (selectivity)
— Engage/lock Arg91 (block catalytic E2 binding)
Target interface size: ~880 Å² → optimal for a 60–120 aa de novo miniprotein
Kd ~36 nM is the benchmark affinity to match or exceed in competition designs
RBX1 vs RBX2 selectivity: Gln57 (RBX1) vs non-conserved equivalent (RBX2) — design handle for selectivity

⚠️ Design Constraints

• Competition requires ≥25% sequence edit distance from UniRef50 — cannot submit GLMN truncations directly
• Neddylation remodels the E2-binding surface slightly — verify binding against multiple conformational states
• GLMN is >600 aa — must be compressed to a minimal binding unit ≤250 aa

Key Reference

Duda DM, Olszewski JL, Schulman BA et al. (2012) "Structure of a glomulin–RBX1–CUL1 complex: inhibition of a RING E3 ligase through masking of its E2-binding surface." Mol Cell 47:371–382. PMID:22748924. PDB:4F52.

10

Existing Targeting Strategies & Literature White Space

Drug Landscape

Current Approaches to Targeting the CRL–RBX1 Axis

Approach	Agent	Direct Target	Status
NAE inhibitor (indirect)	MLN4924 (Pevonedistat)	NEDD8-activating enzyme	Phase I/II/III
DCN1 inhibitor	HZX-960, DI-1859	DCN1–UBC12 interface	Preclinical
PROTAC toolkit (use of CRL)	ARV-471 and others	VHL / CRBN recruit CRL for target degradation	Multiple clinical
siRNA / shRNA	Research tools	RBX1 mRNA	Preclinical only
Direct RBX1 RING domain binder	⚠️ None reported	E2-binding surface	White space!

📊 Literature Status (as of 2026)

PubMed searches for "RBX1 inhibitor binder design", "RBX1 RING domain small molecule", "ROC1 protein binder" return zero reports of a direct RBX1 RING domain binder or small-molecule inhibitor. All targeting approaches remain either indirect (upstream NAE) or use RBX1 machinery for PROTAC-mediated degradation of other targets.
→ A genuine white-space opportunity exists for direct de novo binder design.

Lessons from MLN4924

MLN4924 globally disables all CRL complexes by blocking cullin neddylation — broad efficacy but systemic toxicity from pan-CRL inhibition. A direct RBX1 E2-binding surface binder would be more precise, and could potentially be combined with specific F-box adaptor inhibitors to achieve substrate selectivity.

Why Design a Direct RBX1 Binder — Now?

Zero reported direct RING domain binders in the literature
Glomulin proves E2-binding surface achieves Kd ~36 nM — tractable epitope
74+ PDB structures provide atomic-level design foundation
AF2 + RFdiffusion + BindCraft toolchain now mature and validated
GLMN loss-of-function mutations cause human GVM — biological validation in vivo
Competition window (deadline 3/26/2026) aligns perfectly with tool availability

RBX1 vs RBX2 Selectivity Opportunity

RBX2 (SAG/RNF7) shares the RING fold with RBX1 but has divergent residues around the E2-binding groove (Gln57 region).
GLMN selectively binds RBX1 but not RBX2 — confirming that selectivity through this surface is achievable. RBX1 Gln57 is the primary selectivity determinant.

Affinity Target

• Goal: <100 nM Kd (outcompete free E2; approach GLMN level)
• BLI detection range: nM–µM
• Must express solubly in E. coli

11

De Novo Binder Design Strategy & Computational Pipeline

Design Strategy

Computational Design Workflow

Step 1

Target Prep

2LGV + 4F52

Step 2

Backbone Gen.

RFdiffusion

Step 3

Sequence Design

ProteinMPNN

Step 4

Structural Filter

AF2 / Boltz-2

Step 5

Submit & Test

BLI assay

Interface Targeting Priorities

Priority	Interface	Anchor Residues	Strategy
P1 Top	E2-binding surface (α2)	Ile44, Trp87, Arg91	Mimic GLMN contact geometry on de novo scaffold
P2	Val38/Val39 hinge	Val38, Val39	Lock RING conformation; prevent rotation
P3	Neddylated open state	Transition-state surface	Conformationally selective; trap inactive state

Tool References

RFdiffusion: Watson et al. Nature 2023 (PMID:37468640)
ProteinMPNN: Dauparas et al. Science 2022
BindCraft: Pacesa et al. Nature 2025 (s41586-025-09429-6)
Bennett et al. Nat Commun 2023 — AF2 filtering gives ~10× success rate improvement

Design Parameter Constraints

Target region	RING domain residues 40–108 (E2-binding face)
Primary anchors	Ile44, Trp87, Arg91 (essential contacts)
Binder length	60–150 aa (max 250 aa)
Prohibited	Zn²⁺ chelation; disruption of CUL1 β-sheet
Novelty	≥25% seq. edit dist. from UniRef50
Affinity goal	<100 nM Kd (BLI-detectable)
Template structures	2LGV (NMR 20-conformer) + 4F52

AF2 Filtering Metrics

pAE_interaction (complex PAE) < 10 Å
pLDDT of binder chain > 80
ipTM > 0.75
Multi-temperature ProteinMPNN sampling (T = 0.1–0.5)
Diverse backbone types to maximize coverage

Candidate Binder Scaffold Types

① Miniprotein (40–80 aa) — high stability, BLI-friendly, precedented
② α-helical bundle (80–130 aa) — precisely mimics GLMN contact geometry
③ β-hairpin + loop (60–100 aa) — fits concave E2-binding groove

12

Summary & Key References

Summary & References

Core Takeaways

RBX1 is a 108 aa RING-H2 protein — the catalytic engine responsible for ~20% of all cellular ubiquitination
Contains 3 structural Zn²⁺ ions; Val38/Val39 hinge allows 60°–170° RING rotation — critical for design considerations
Overexpressed in 10+ cancer types; RNAi knockdown causes sequential G2/M arrest → apoptosis → senescence
The E2-binding surface (helix α2: Ile44/Trp87/Arg91) is the primary design target; Glomulin (PDB 4F52) provides a natural protein template at Kd ~36 nM
No direct RBX1 RING domain binder exists in the literature — a genuine first-mover opportunity
Recommended pipeline: 2LGV + 4F52 → RFdiffusion → ProteinMPNN → AF2 filter (pAE <10) → submit top 100

        🚀 Action Plan: Use 4F52 E2-binding surface as the design epitope. Generate diverse backbones with RFdiffusion targeting Ile44/Trp87/Arg91. Design sequences with ProteinMPNN (multi-temp). Filter with AF2 (ipTM >0.75, pAE <10). Submit best 100 by March 26, 2026.
      

Key References

Zheng N et al. (2002) SCF complex structure. Nature 416:703. PDB: 1LDD

Duda DM, Schulman BA (2012) Glomulin–RBX1–CUL1. Mol Cell 47:371. PDB: 4F52 ⭐

Spratt DE, Shaw GS (2012) RBX1 NMR, E2~Ub recruitment. J Biol Chem 287:17374. PDB: 2LGV ⭐

Duan H, Sun Y (2009) RBX1 silencing — tri-phasic anti-tumor response. Cancer Res 69:4974

Cardote, Ciulli A (2017) Crystal structure of CUL2–RBX1–VHL. Structure 25:901. PDB: 5N4W

Watson JL et al. (2023) De novo protein design with RFdiffusion. Nature 620:1089

Bennett NR et al. (2023) Improving binder design with AF2 (~10× success). Nat Commun 14:2625

Song Y et al. (2023) RBX1 drives thyroid cancer metastasis. Cell Biosci 13:30