Introduction

Peptide sequence design is the rational construction of amino acid sequences to achieve desired biological activity, pharmacokinetic properties, and physicochemical characteristics. This article covers the principles governing sequence-function relationships, design rules for major peptide classes, and computational tools for sequence optimization.

Fundamental Design Principles

Charge Distribution

The net charge of a peptide determines its solubility, membrane interaction, and receptor binding:

Charge Profile	Properties	Applications
Cationic (+2 to +8)	Membrane-active, antimicrobial	AMPs, CPPs
Anionic (−2 to −6)	Calcium binding, enzyme inhibition	Osteogenic peptides
Zwitterionic (0)	Neutral, receptor-specific	Hormone analogs
Amphipathic	Membrane insertion, pore formation	Channel-forming peptides

Hydrophobicity

The hydrophobic moment (μH) quantifies amphipathicity:

μH = √[(Σᵢ Hᵢ cos θᵢ)² + (Σᵢ Hᵢ sin θᵢ)²] / n

Where Hᵢ = hydrophobicity of residue i, θᵢ = angular position on helix wheel, n = number of residues.

Design targets:

• Antimicrobial peptides: μH > 0.5
• Membrane-active peptides: μH > 0.8
• Soluble peptides: μH < 0.3

Secondary Structure Propensity

Different amino acids have intrinsic tendencies to form specific secondary structures:

Residue	α-Helix	β-Sheet	Turn	Notes
Ala	High	Low	Low	Helix former
Val	Low	High	Low	Sheet former
Pro	Low	Low	High	Helix breaker
Gly	Variable	Low	High	Flexible
Leu	High	Low	Low	Helix former
Ile	Moderate	High	Low	Sheet former

Amino Acid Composition Rules

For antimicrobial peptides:

• Cationic residues (Lys, Arg): 20–30%
• Hydrophobic residues (Leu, Ile, Val, Phe): 40–60%
• Net charge: +2 to +8
• Amphipathic structure: α-helix or β-sheet

For hormone analogs:

• Conserved binding motifs: Preserve receptor contact residues
• Variable regions: Optimize pharmacokinetics
• D-amino acids: At protease-susceptible sites

For enzyme inhibitors:

• Active site mimetics: Match substrate stereochemistry
• Transition-state analogs: Tetrahedral geometry
• Binding determinants: Hydrophobic pockets, hydrogen bonds

Design Rules by Peptide Class

Antimicrobial Peptides (AMPs)

Design principles:

1. Net positive charge: +2 to +8 for membrane interaction
2. Amphipathic structure: Hydrophobic and hydrophilic faces
3. Helical content: >50% α-helix in membrane environment
4. Length: 12–50 residues (optimal 15–25)
5. Sequence patterns: Alternating hydrophobic and cationic residues

Sequence templates:

# α-helical AMP (16 residues)
X₃-KX₂-LX₂-KX₂-LX₂-KX₂-X₂
Where X = hydrophobic residue

# β-sheet AMP (defensin-like)
CX₃-CX₅-CX₉-CX₄-CX₆-C
Where C = cysteine (disulfide bonds)

Example sequences:

• Magainin 2: GIGKFLHSAKKFGKAFVGEIMNS
• LL-37: LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES
• Melittin: GIGAVLKVLTTGLPALISWIKRKRQQ

Hormone Analogs

Design principles:

1. Receptor binding domain: High-affinity interaction site
2. Effector domain: Activation or antagonism
3. Metabolic stability: Protect protease-susceptible sites
4. Pharmacokinetic optimization: Half-life, solubility

GLP-1 analog design:

# Native GLP-1 (7-36):
HAEGTFTSDVSSYLEGQAAKEFIAWLVKGR

# Modifications for stability:
1. Ala8 → Aib (α-aminoisobutyric acid)
2. Gly22 → Arg (resist DPP-IV)
3. Lys34 → Arg (reduce deamidation)
4. C-terminal amidation

Somatostatin analog design:

# Octreotide (Somatostatin analog):
dPhe-Cys-Phe-dTrp-Lys-Thr-Cys-Thr-ol

# Design features:
1. dPhe at position 1 (resist aminopeptidase)
2. dTrp at position 6 (helical constraint)
3. Cyclization via disulfide (Cys2-Cys7)
4. C-terminal threoninol (resist carboxypeptidase)

Enzyme Inhibitors

Design principles:

1. Transition-state mimicry: Tetrahedral geometry at scissile bond
2. Active site complementarity: Shape and charge matching
3. Binding affinity: Kd < 100 nM for therapeutic effect
4. Selectivity: >100-fold over related enzymes

Protease inhibitor templates:

# Serine protease inhibitor:
Ac-X₃-X₂-X₁-psi(CH₂NH)-X₁'-X₂'-X₃'-NH₂
Where psi = non-cleavable isostere

# Metalloprotease inhibitor:
X₃-X₂-X₁-PO(OH)-CH₂-X₁'-X₂'-X₃'
Where PO(OH) = phosphinic acid (zinc-binding group)

Cell-Penetrating Peptides (CPPs)

Design principles:

1. Cationic residues: Arg > Lys for uptake
2. Amphipathicity: Enhances membrane interaction
3. Length: 9–30 residues
4. Conjugation strategy: Cargo attachment point

Common CPP sequences:

• TAT (48-60): YGRKKRRQRRR
• Penetratin: RQIKIWFQNRRMKWKK
• Transportan: GWTLNSAGYLLGKINLKALAALAKKIL

Computational Design Tools

Sequence Analysis

Hydropathy plots:

• Kyte-Doolittle scale: Predict transmembrane domains
• Hopp-Woods scale: Predict antigenic sites
• Eisenberg scale: Predict membrane insertion

Charge analysis:

• Net charge calculation: Σ(+1 for Lys, Arg) + Σ(−1 for Asp, Glu)
• Charge distribution: Charge moment analysis

Structure Prediction

Secondary structure prediction:

• PSIPRED: Neural network-based prediction
• JPred: Multiple sequence alignment-based
• HNN: Hierarchical neural network

Tertiary structure prediction:

• AlphaFold2: Deep learning-based structure prediction
• RoseTTAFold: Three-track neural network
• Rosetta: Physics-based energy minimization

Molecular Dynamics

Simulation protocols:

1. System setup: Peptide in explicit solvent (TIP3P water)
2. Energy minimization: Steepest descent, 5000 steps
3. Equilibration: NVT (100 ps), NPT (100 ps)
4. Production: NPT, 100 ns – 1 μs
5. Analysis: RMSD, RMSF, secondary structure evolution

Key parameters to monitor:

• Root-mean-square deviation (RMSD): Structural stability
• Root-mean-square fluctuation (RMSF): Residue flexibility
• Radius of gyration (Rg): Compactness
• Solvent-accessible surface area (SASA): Exposure to solvent

Machine Learning Approaches

Sequence generation:

• Variational autoencoders (VAE): Learn latent space of peptide sequences
• Generative adversarial networks (GANs): Generate novel sequences
• Recurrent neural networks (RNNs): Sequence-to-sequence prediction

Property prediction:

• Random forests: Predict solubility, toxicity
• Support vector machines: Classify activity
• Deep neural networks: Predict binding affinity

Optimization Strategies

Directed Evolution

1. Library construction: Random mutagenesis, DNA shuffling
2. Selection: Phage display, yeast display, ribosome display
3. Screening: High-throughput assays
4. Iteration: Multiple rounds of selection

Rational Design

1. Alanine scanning: Identify critical residues
2. Charge scanning: Optimize net charge
3. Hydrophobicity scanning: Optimize amphipathicity
4. Length optimization: Determine minimal active sequence

Hybrid Approaches

1. Computational pre-screening: Predict promising candidates
2. Focused libraries: Limited randomization at key positions
3. Machine learning guidance: Train on experimental data

Design Rules Summary

General Rules

1. Start with known active sequences: Build on existing knowledge
2. Minimize length: Shorter peptides are easier to synthesize and optimize
3. Optimize charge: Match target (cationic for membranes, neutral for receptors)
4. Consider amphipathicity: Essential for membrane-active peptides
5. Protect against proteolysis: D-amino acids, cyclization, N-methylation

Class-Specific Rules

Peptide Class	Key Parameters	Optimization Targets
Antimicrobial	Charge, amphipathicity, length	MIC, hemolysis
Hormone analog	Receptor binding, stability	Kd, t½
Enzyme inhibitor	Binding affinity, selectivity	IC₅₀, Ki
Cell-penetrating	Charge, hydrophobicity	Uptake efficiency
Imaging agent	Binding, fluorescence	Kd, quantum yield

Practical Workflow

Step 1: Define Target Profile

1. Biological target: Receptor, enzyme, membrane
2. Desired activity: Agonist, antagonist, inhibitor
3. Pharmacokinetic requirements: Half-life, route of administration
4. Safety constraints: Toxicity, immunogenicity

Step 2: Initial Design

1. Select template: Based on known active peptides
2. Apply design rules: Charge, hydrophobicity, structure
3. Generate sequences: 10–100 candidates

Step 3: Computational Screening

1. Predict properties: Solubility, stability, binding
2. Model structures: AlphaFold2, molecular dynamics
3. Rank candidates: Score by predicted performance

Step 4: Experimental Validation

1. Synthesize top candidates: SPPS
2. Characterize: HPLC, MS, CD
3. Assay activity: Binding, functional assays
4. Iterate: Refine based on results

Summary

Peptide sequence design integrates principles from biochemistry, biophysics, and computational science. By understanding the relationships between sequence, structure, and function, and leveraging modern computational tools, researchers can design peptides with improved activity, selectivity, and pharmacokinetic properties. The field continues to advance with machine learning approaches that can explore larger sequence spaces and predict properties with increasing accuracy.

Deep dive: Explore Computational Design for advanced modeling techniques, or read about Peptide Modifications for post-design optimization.

Test yourself: Take the Peptide Design Quiz or study with Sequence Design Flashcards.