15-PGDH Target Analysis & Binder Design Strategy

🔬 Adaptyv × Nucleate Berlin · Binder Design Competition 2025

15-PGDH Target Analysis
& Binder Design Strategy

Sequence · Structure · Conservation · Inhibitors · Design Rationale

Target Protein

15-PGDH / HPGD (UniProt P15428)

Protein Size

266 amino acids · ~29 kDa/subunit

Family

SDR (Short-Chain Dehydrogenase/Reductase)

Cofactor

NAD⁺ (obligate)

2

15-PGDH — Protein Overview & Biology

Protein Identity

266

amino acids / subunit

~29

kDa per subunit

2

subunits (homodimer)

1.65

Å best resolution (2GDZ)

NAD⁺

obligate cofactor

Domain Architecture (266 aa)

Key Biological Facts

Property	Value
Full name	15-hydroxyprostaglandin dehydrogenase
Gene	HPGD (SDR36C1) · Chr 4q34.1
Reaction	PGE₂ + NAD⁺ → 15-keto-PGE₂ + NADH
Quaternary structure	Obligate homodimer (α9 helix packing)
Expression	Lung, colon, placenta, stomach
Disease role	Tumor suppressor — inactivated in CRC, lung, gastric cancers
Therapeutic interest	Inhibition → anti-inflammatory, bone marrow, DMD

⚡ Why This Target

15-PGDH degrades PGE₂, a pro-inflammatory eicosanoid that suppresses immunity and promotes tumor growth. Inhibiting 15-PGDH raises tissue PGE₂ → accelerates hematopoietic recovery after transplantation and promotes muscle regeneration in DMD.

🔑 Binder Opportunity

The substrate binding pocket (adjacent to NAD⁺) is large, well-defined, and validated — SW033291 achieves Ki 0.1 nM. The dynamic lid domain (F185/Y217 hinges) is unique to 15-PGDH and offers a selectivity handle over other SDR family members.

⚠️ Design Constraint

NAD⁺ is obligate cofactor at near-saturating cellular concentrations (~0.5 mM). Avoid NAD⁺-site competition. Best approach: substrate-competitive pocket OR allosteric lid domain targeting.

3

Structural Architecture — SDR Fold & Catalytic Tetrad

Structural Detail

Active Site — Rossmann Fold + Catalytic Tetrad (PDB 2GDZ)

Reaction Mechanism

Y151 (pKa raised by K155) abstracts proton from C15–OH of PGE₂ → hydride transferred to NAD⁺ → 15-keto-PGE₂ released

GxGxxG Motif (pos 12–18)

G·A·A·Q·G·I·G — Gly16 (0.856) and Gly18 (0.903) coordinate NAD⁺ pyrophosphate via backbone NH

Catalytic Tetrad — Residue Detail

Residue	Role	Scorecon	Notes
Asn107	Substrate anchor	0.8448	H-bonds C15–OH of PGE₂; tetrad entry point
Ser138	Proton relay	0.9396	H-bond network bridging Tyr151 to substrate
Tyr151	Catalytic base ★	1.0000	Perfectly conserved — all SDR orthologs
Lys155	Tyr151 activator	0.9756	Ion-dipole lowers Tyr pKa ~4 units

Homodimer Interface

α9 Antiparallel Helix Packing

F161 · A146 · Y206 · L167 · L171

Primary dimerization element. Disrupting dimerization abolishes activity — both subunits required for correct active site geometry.

Interface Area ~1200 Å²

Hydrophobic core; conserved F264

Large buried surface. Stable under physiological conditions. Good target for protein binder approach via surface complementarity.

🏆 Best Structure: PDB 2GDZ (1.65 Å)

Highest-resolution human 15-PGDH structure with NAD⁺ bound. Complete substrate-binding cleft visible. Use as primary template for all docking and binder design calculations.

4

Key PDB Structures & Structural Resources

Structure Database

1.65

Å — best resolution (2GDZ)

8+

PDB entries available

2023

cryo-EM lid domain solved

AF2

AlphaFold model available

PDB ID	Complex / Ligand	Resolution	Method	Key Insight	Priority
2GDZ	Human 15-PGDH + NAD⁺ (homodimer)	1.65 Å	X-ray	Highest-resolution human structure; defines substrate cleft; Rossmann fold + catalytic tetrad in active conformation	★★★★★
2HHQ	NAD⁺ + substrate analog	2.1 Å	X-ray	Productive NAD⁺ + substrate co-binding; defines substrate vector and key pocket contacts	★★★★
2O8U	Indomethacin-bound	2.3 Å	X-ray	First small molecule co-crystal; substrate pocket accommodates larger scaffolds	★★★★
cryo-EM	SW033291-bound (Huang et al. 2023, Nat Commun)	3.2 Å	Cryo-EM	Reveals dynamic lid domain — F185/Y217 hinges; lid closes over inhibitor. First view of full conformational change	★★★★
AF-P15428	AlphaFold2 predicted model	in silico	ML	Full-length model including flexible N-term; useful for binder design vs. regions not in X-ray	★★★

Recommended Templates for Binder Design

2GDZ (1.65 Å, NAD⁺ complex) — primary template; highest-precision substrate pocket geometry.
Cryo-EM SW033291 structure (Huang 2023) — lid-closed conformation for allosteric strategies.
Combining open (2GDZ) + closed (cryo-EM) maximizes coverage of conformational states.

Lid Domain Uniqueness

Lid domain is disordered in apo structures, ordered only upon inhibitor/substrate binding. Binders that engage F185/Y217 achieve conformational selectivity not possible with rigid-site targeting.

5

Conservation Analysis — Valdar Scorecon (150 Orthologs)

Computational · MSA Analysis

Data Source

NCBI BLASTp vs SwissProt
E < 1×10⁻⁵ · ≥25% identity
150 homologs retained

MSA

MUSCLE v5.3
150 seqs × 732 columns
Henikoff sequence weighting

Scoring

Valdar (2002) Scorecon
BLOSUM62 normalized
Weighted pairwise scoring

Results

Score range: 0.0 – 1.0
13 positions > 0.80
Tyr151 = 1.0000 (perfect)

Per-Residue Scorecon Heatmap — All 266 Positions (hover for details)

Score: <0.35 0.35–0.50 0.50–0.65 0.65–0.75 0.75–0.85 0.85–0.95 >0.95 threshold 0.80

All 13 Highly Conserved Positions (Scorecon > 0.80)

Position	AA	Score	Module	Known Role
Tyr151	Y	1.0000	Catalytic Tetrad	General base; proton abstraction from C15–OH; perfectly conserved across all SDRs
Lys155	K	0.9756	Catalytic Tetrad	Activates Tyr151 by lowering pKa; essential for activity
Ser138	S	0.9396	Catalytic Tetrad	Proton relay; H-bond to Tyr151 and substrate
Pro183	P	0.9069	Lid Hinge Region	Structural constraint adjacent to F185 lid hinge
Ile133	I	0.9029	Substrate Pocket	Hydrophobic core; shapes substrate vector toward Y151
Gly18	G	0.9027	Rossmann GxGxxG	NAD⁺ pyrophosphate contact; critical for cofactor binding
Val176	V	0.8943	Lid Region	Hydrophobic core near lid base; positions lid for substrate entry
Gly131	G	0.8477	Substrate Pocket	Structural Gly enabling tight packing in catalytic domain
Asn107	N	0.8448	Catalytic Tetrad	Anchors substrate C15–OH; first member of tetrad
Gly16	G	0.8557	Rossmann GxGxxG	NAD⁺ phosphate binding via backbone NH
Ala92	A	0.8332	Cofactor Loop	Part of NAD⁺ binding loop; nicotinamide ring positioning
Asp64	D	0.8208	NAD⁺ Binding	Anchors adenine ribose 2'-OH; conserved SDR Asp
Met1	M	0.8487	N-terminus	High within SDR family; likely alignment boundary artifact

6

Conservation Insights — Key Findings for Binder Design

Interpretation

Critical Observations

★ Entire Catalytic Tetrad Is Highly Conserved

N107 (0.84), S138 (0.94), Y151 (1.00), K155 (0.98) — all 4 tetrad members rank in the top 13. Any binder contacting these positions will compete with the catalytic mechanism and show potent inhibition of enzymatic activity.

⚡ Lid Domain Hinges Are Low Conservation

F185 (0.163) and Y217 (0.288) — the cryo-EM-defined lid hinges — show low scores. The lid diverged across SDR family members. Lid-targeting binders will be species- and isoform-selective, a significant advantage for therapeutic development.

Pro183 — Underappreciated Hotspot (0.9069)

4th highest-scoring residue. Adjacent to F185 lid hinge. Acts as a rigid structural constraint before the lid. Targeting Pro183 may lock the lid in closed state — a novel allosteric mechanism not yet exploited by any known inhibitor.

GxGxxG: G18 > G16 (Both High)

Gly18 (0.903) and Gly16 (0.856) confirm NAD⁺ as obligate across all 150 orthologs. These glycines are structurally irreplaceable — define the cofactor selectivity filter of the Rossmann fold.

Score Distribution Summary

1

perfect score (1.00)

12

high (> 0.80)

~28

moderate (> 0.65)

4

catalytic tetrad in top 13

Conservation vs. Known Function

Residue	Score	Literature Role	Design Use
Tyr151	1.000	Catalytic base	Primary anchor — potency
Lys155	0.976	pKa modulator	Co-anchor with Y151
Ser138	0.940	Proton relay	Pharmacophore hydrogen bond
Phe185	0.163	Lid hinge (cryo-EM)	Species-selective contact
Tyr217	0.288	Lid hinge (cryo-EM)	Species-selective contact
Asp64	0.821	NAD⁺ adenine anchor	Avoid — cofactor binding site
Phe161	0.410	Dimer interface	Alternative allosteric site

        Design verdict: Lead with substrate pocket (Y151/K155/S138) for potency. Layer in lid domain (F185/Y217) contacts for selectivity. Avoid NAD⁺ site competition.
      

7

NAD⁺ Cofactor — Four Critical Roles in 15-PGDH Function

Cofactor Biology

NAD⁺ is not merely an electron acceptor — it structurally organizes the active site, sets reaction geometry, and gates substrate binding via ordered kinetics.

① Hydride Acceptor — Direct Chemistry

C4 of nicotinamide ← hydride from C15 of PGE₂

The B-face of the NAD⁺ nicotinamide ring accepts the pro-S hydride from PGE₂ C15–OH, producing NADH + 15-keto-PGE₂. This is the direct chemistry that inactivates PGE₂ inflammatory signaling.

② Structural Organizer — Rossmann Fold

G12·G16·G18 (GxGxxG) + D64 + N91 + A92

NAD⁺ binding via the Rossmann fold stabilizes the entire cofactor domain. GxGxxG contacts pyrophosphate backbone; Asp64 anchors adenine ribose 2'-OH. Cofactor binding is obligate for correct tertiary structure.

③ Active Site Organizer — Substrate Positioning

Nicotinamide ring · T188 · V186 · G184

The bound nicotinamide ring forms the floor of the substrate pocket, positioning PGE₂ C15 within 3.5 Å for hydride transfer. Without NAD⁺ the substrate pocket collapses — explaining the strict ordered binding mechanism.

④ Kinetic Gate — Ordered Bi-Bi Mechanism

Order: NAD⁺ first → PGE₂ second → NADH last

15-PGDH follows ordered bi-bi kinetics: NAD⁺ must bind first to open the substrate portal. NADH releases last after oxidation. This gating means binders to the substrate site must compete with substrate, not cofactor — a more tractable competition.

NAD⁺ Km & Cellular Context

Km(NAD⁺) ≈ 20–80 µM; cellular NAD⁺ ~0.5 mM → enzyme is normally NAD⁺-saturated in vivo. Binders competing with NAD⁺ face a ~10–25× concentration disadvantage.

NADH Product Inhibition

NADH is a competitive inhibitor (Ki ~15 µM). NAD⁺/NADH ratio partially controls enzyme activity in vivo. Oxidative stress → increased NAD⁺ → more PGE₂ degradation.

Design Implication

Avoid displacing NAD⁺ — extremely high competition barrier. Best targets: substrate pocket (opens only after NAD⁺ binds) or lid domain (gates substrate entry). Both are accessible from the enzyme exterior.

8

Known Small Molecule Inhibitors — Structural Templates

Ligand Chemistry

Compound	Affinity	Mode	Key Contacts	Clinical Status	Template Value
SW033291 Best-in-class	Ki ≈ 0.1 nM	Noncompetitive vs. PGE₂	S138 · Y151 · F185 · Y217	Preclinical (bone marrow, hematopoiesis)	★★★★★ — cryo-EM co-structure; spans pocket + lid
SW209415	Ki ≈ 2 nM	Noncompetitive	S138 · Y151 · L139 · I190	Preclinical	★★★★ — expanded pocket contact map
SW222746	Ki ≈ 5 nM	Competitive vs. NAD⁺	G16 · G18 · D64 · A92	Preclinical	★★ — NAD⁺ site; hard to outcompete in vivo
ML148	IC₅₀ ≈ 25 nM	Mixed	Q148 · V145 · T246	Probe compound (MLPCN)	★★★ — outer substrate pocket contacts
MF-300 (Epirium Bio) Phase 1 ✓	IC₅₀ ≈ 0.8 nM	Substrate-competitive	S138 · Y151 · K155 · L139 · I194	Phase 1 complete Sep 2025 (oral, DMD)	★★★★★ — first clinical; validates target in humans
Nimbus compound	IC₅₀ < 10 nM	Undisclosed	Not published	Early discovery	★★ — confirms active drugging competition

★ SW033291 — Pharmacophore Template for Protein Binder

SW033291 (Ki 0.1 nM) contacts S138, Y151, F185, Y217 — spanning both conserved catalytic tetrad and variable lid domain. The cryo-EM structure shows lid closing over the compound to form a ~900 Å² buried surface. A protein binder mimicking this pharmacophore but adding 2–3× more surface area should achieve sub-pM affinity. This is the primary template for Strategy A+B hybrid binder design.

Clinical Validation: MF-300

MF-300 (Epirium Bio) entered Phase 1 for DMD in 2024 and completed dosing September 2025. This is the strongest possible target validation — 15-PGDH inhibition is safe and mechanistically active in humans at drug doses.

9

Binder Design Strategy — Three Complementary Approaches

Design Strategy

Strategy A — Substrate-Competitive Pocket Binder

Target: Y151 · K155 · S138 · N107 · L139 · V145 · Q148 · I190 · I194

Design a protein that occupies the PGE₂ substrate binding site. Anchored by perfectly conserved Y151 (1.000) and K155 (0.976). Pocket opens only after NAD⁺ binds → design against 2GDZ (NAD⁺-bound form).

Rationale: SW033291 (Ki 0.1 nM) and MF-300 (IC₅₀ 0.8 nM) validate this pocket. A protein binder with 2–3× more buried surface area should reach sub-nM Kd by entropy-enthalpy compensation on a larger interface.

Strategy B — Lid Domain Conformational Capture

Target: F185 · Y217 · P183 · V176 · I190 · I194 · T188

Stabilize the "closed" lid conformation seen in cryo-EM. F185/Y217 are low conservation (0.16/0.29) → species-selective. Binder locks lid closed over substrate site — allosteric inhibition without touching NAD⁺.

Rationale: Higher selectivity vs. other SDR family members due to lid sequence divergence. Cryo-EM provides a precise structural model of the closed state for design template.

Strategy C — Dimer Interface Disruption

Target: F161 · Y206 · L167 · L171 · A146 · F264

Target the α9 antiparallel dimer interface (~1200 Å² buried). Disrupting dimerization abolishes activity — both subunits are required. Interface is hydrophobic; flat but sizeable.

Rationale: Isoform-selective — only homodimeric 15-PGDH is disrupted. Low risk of cross-reacting with SDR monomers. Lower conservation (0.4–0.5) means lower selectivity pressure but a novel epitope.

Prioritized Hotspot Residues

Residue	Score	Strategy	Contact Type	Priority
Tyr151	1.000	A	H-bond acceptor (OH), π-stacking	Must include — primary anchor
Lys155	0.976	A	Salt bridge / H-bond (ε-NH₃⁺)	Must include — co-anchor
Ser138	0.940	A	H-bond donor/acceptor (OH)	High priority pharmacophore
Pro183	0.907	B	Hydrophobic / structural constraint	High priority — lid anchor point
Phe185	0.163	B	Hydrophobic / π-stacking	Selectivity — use for isoform specificity
Tyr217	0.288	B	H-bond / aromatic stacking	Selectivity — use for isoform specificity
Asn107	0.845	A	H-bond donor/acceptor (amide)	High priority — tetrad entry
Phe161	0.410	C	Hydrophobic / π-stacking (dimer)	Alternative strategy

🏆 Recommended Primary Strategy: A + B Hybrid

Design a binder that anchors to the substrate pocket (Y151/K155/S138) AND reaches into the lid domain (F185/Y217) when lid is in the closed conformation. This mirrors SW033291's mechanism but with 2–3× greater buried surface area. Use 2GDZ + cryo-EM lid-closed model as dual templates for de novo backbone generation.

10

Computational Pipeline & Competition Summary

Pipeline · Summary

Recommended Computational Design Pipeline

Step 1

Structure Prep

2GDZ + cryo-EM

Step 2

Pocket Analysis

fpocket / SiteMap

Step 3

Hotspot Map

FTMap / Rosetta

Step 4

Backbone Design

RFdiffusion

Step 5

Sequence Design

ProteinMPNN

Step 6

Filter + Rank

AF2 · ESMFold · ΔG

Step 7

Submit 100 Best

Adaptyv BLI assay

Master Residue Reference

Category	Residues	Design Role
Catalytic Tetrad	N107 · S138 · Y151 · K155	Primary anchor — potency driver
Rossmann / NAD⁺	G12 · G16 · G18 · D64 · A92	Avoid — cofactor competition
Substrate Pocket	L139 · A140 · V145 · Q148 · I190 · I194	Shape complementarity
Lid Domain	P183 · F185 · V186 · T188 · Y217	Selectivity over other SDRs
Dimer Interface	F161 · L167 · L171 · Y206 · F264	Alternative allosteric site

Key References

• Cho et al. (2006) Proc Natl Acad Sci — PDB 2GDZ, 1.65 Å human structure
• Huang et al. (2023) Nat Commun — cryo-EM lid domain, SW033291 mechanism
• Seo et al. (2003) J Biol Chem — SDR catalytic mechanism
• Epirium Bio (2025) — MF-300 Phase 1 completion press release

Top 5 Insights for Competition Entry

Tyr151 is perfectly conserved (1.000) — computational and experimental data agree; use it as the primary anchor in every design
SW033291 (Ki 0.1 nM) proves the pocket — a protein binder with 2–3× more buried surface should reach pM range by design
Lid domain (F185/Y217) enables selectivity — low conservation makes these residues ideal for isoform-specific designs
MF-300 Phase 1 validates in-vivo safety — 15-PGDH inhibition is tolerated in humans at therapeutic doses
Ordered bi-bi kinetics: design against NAD⁺-bound form — use 2GDZ (NAD⁺ co-crystal) for maximal pocket fidelity

          Competition target: 100 designs, Strategy A+B hybrid. Pharmacophore: Y151·K155·S138·F185·Y217. Pipeline: RFdiffusion → ProteinMPNN → AF2 validation → BLI screening.
        

11

3D Structure — 15-PGDH Monomer (PDB 2GDZ, 1.65 Å)

Interactive · 3D Viewer

          Loading PDB 2GDZ...
Chain A · NAD+ complex
        

        Drag to rotate · Scroll to zoom · Right-drag to pan
      

Color Legend

White — Protein backbone

All other residues, chain A monomer

Red — Catalytic Tetrad

N107 · S138 · Y151 · K155

Scorecon: 0.84 to 1.00 (Y151 perfect)

Orange — Dimer Interface

A146 · F161 · L167 · L171 · Y206 · F264

alpha9 antiparallel packing (~1200 Ang2)

Purple — Lid Domain

Residues 181-220 · F185 · Y217 hinges

Disordered in apo; closes over substrate

Yellow — NAD+ cofactor

Obligate cofactor shown as sticks

SW033291 Binding (PDB 8CVN)

Cryo-EM (Huang 2023) shows SW033291 contacting both red catalytic residues AND the purple lid in closed conformation. This dual-site engagement explains Ki = 0.1 nM.

Dimer Interface

Orange residues contact the second monomer. Breaking dimerization abolishes activity, but the flat hydrophobic surface is harder to target than the substrate pocket.

12

Lead Binder — r2_cycle98, 88 aa

AF3 + Boltz2 validated

Loading complex...

● Catalytic triad ● Substrate site ● Lid (181-220) ● Binder

r2_cycle98

88-residue mini-binder designed against the open-lid conformation of 15-PGDH. Contacts the substrate pocket around S138, Y151, K155. Only design where both AF3 and Boltz2 agree on active-site binding.

STHERIHNVHEEVWKKMWPNSPEASEEAKELHRKWYEVWKKVHESRSWSEEEKKKMFPEIFFQIADIIIEWAYLIIWVQREAFEKYFP

Metric	AF3 (+MSA)	Boltz2
Ranking score	0.60	-
iPTM	0.55	-
iPSAE (active site)	0.322	0.77

Note on NAD+

When NAD+ is included in the AF3 prediction, overall confidence rises (iptm 0.55 to 0.72) but the binder gets displaced from the pocket. The current design competes with the cofactor for space.

13

Design Campaign — What Worked, What Didn't

Experimental

Closed state...

Closed (2GDZ) -- lid shut

Open state...

Open (MD 225ns) -- lid swings out

Lid dynamics

225 ns MD and BioEMU sampling both show the lid (res 181-220, purple) swinging 8-9 A open. Designing against the open frame was the first thing that gave any AF3 signal at all (closed-state designs scored near zero).

Strategies tested

Approach	AF3 iptm	On hotspot?
Active-site pocket (open)	0.55	Yes
Channel-entry cap	0.41	Marginal
Clamp across channel	0.29	No
Lid-lock staple (closed)	0.88*	No -- wrong site

*Lid-lock scored high but AF3 placed it on the N-terminal face (res 3-140), far from the lid. High iptm, zero hotspot contact.

Sequence optimization (4 rounds)

	Boltz2	AF3 rank	AF3 iPSAE_as
Seed	0.83	0.41	0.13
R1	0.76	0.56	0.18
R2 (r2_cycle98)	0.77	0.60	0.32
R3	0.79	0.61	0.12
R4	0.78	0.60	0.18

After R2 the Boltz2-guided optimizer started moving the binder away from the active site. The N52S mutation in R2 was the last productive change.

Oracle disagreement

Boltz2 gives iPSAE = 0.77 for the lead. AF3 gives 0.32. Protenix v2 gives 0. Optimizing against Boltz2 alone led to overfitting (iPSAE inflated 4x vs AF3). Cross-validation after every round caught this.

Point-mutation scan (190 variants, AF3)

Variant	AF3 rank	iptm
WT	0.60	0.55
N8L	0.63	0.59
Q79E	0.62	0.59
N8R	0.62	0.58

Single-seed AF3. Position 8 and 79 are the most mutable interface positions. Combinations not yet tested.