What this chapter is about (big picture) 🧠

This section explains how solution NMR can give you both protein structure and protein dynamics—and why that combination is powerful. It opens with an example showing that changing a protein’s functional state (oxidized vs reduced cytochrome b5) changes not only structure, but also which backbone amides exchange slowly with solvent, revealing regions with restricted dynamics (shown as cyan spheres) 🧊💧

Core idea: what you must do first in protein NMR

Before you can get structure or dynamics, you need resonance assignment: figuring out which NMR signal belongs to which nucleus in the protein.

The chapter divides multidimensional experiments into two conceptual classes:

1. Coherence-transfer (through-bond) experiments → connect nuclei linked by covalent bonds using scalar (J) couplings.
2. Dipole–dipole (through-space) experiments → connect nuclei that are close in 3D space, even if not bonded.

This is exactly where COSY/TOCSY (through-bond) and NOESY (through-space) come in.

1) The classic 2D ¹H–¹H “starter kit”: COSY, TOCSY, NOESY 🔍

COSY (Correlation Spectroscopy) = “who is J-coupled to whom?”

• COSY shows cross-peaks between germinal or vicinal protons (typically ²J and ³J couplings).
• If protons are more than ~3 bonds apart, couplings like ⁴J are usually ~0 → no COSY cross-peak. Use: identifying direct neighbor relationships inside a residue’s proton network.

TOCSY (Total Correlation Spectroscopy) = “the whole spin system lights up”

• TOCSY spreads magnetization across an entire spin system via successive scalar couplings.
• So you get COSY-like peaks plus extra peaks connecting all protons in the same spin system.
• Patterns are interrupted by non-protonated heteroatoms. Because of backbone carbonyls, TOCSY patterns usually stay within one amino acid.
• Aromatics: a nonprotonated carbon breaks scalar pathways, so you often see two separate patterns (backbone HN/Hα/Hβ vs ring protons) with no scalar connectivity between them.

Use: “This spin pattern looks like this residue type.” Limitation: ambiguities—different residues can share very similar spin systems (examples given include overlap among Cys/Asp/Asn and aromatics for HN/Hα/Hβ patterns; and among Glu/Gln/Met when including Hγ).

NOESY (Nuclear Overhauser Effect Spectroscopy) = “who is close in space?”

• NOESY gives cross-peaks for proton pairs within about ~5 Å.
• Because it depends on through-space dipolar interactions, it can connect things TOCSY cannot:
  • Aromatic ring protons ↔ backbone protons (linking those “separate” aromatic patterns).
  • Crucially: backbone protons of sequential residues → enables sequence-specific assignment by linking spin systems along the chain.
• NOESY contains multiple levels of structural information:
  • Sequential contacts (neighbors in sequence) for assignment.
  • Short/medium-range NOEs (up to ~5 residues apart) that diagnose secondary structure (helices/turns).
  • Long-range NOEs (>5 residues apart in sequence) that define the tertiary fold and are especially important to assign unambiguously.

Turning NOESY peaks into distance restraints 📏

• You integrate NOESY cross-peak intensities to estimate proton–proton distances (used as restraints).
• But the text warns the relationship is not straightforward (mobility, spin diffusion, solvent exchange can distort intensity).
• Practical approach: acquire NOESY to avoid severe spin diffusion, then convert volumes into upper bounds on distances.
• You typically need many restraints (often ~15–20 per residue) to define the structure.

Why pure ¹H–¹H becomes hard as proteins get bigger 😵‍💫

• Proton chemical shifts span a relatively small range.
• Folded proteins have tons of proton pairs within 5 Å → huge number of peaks → overlap → ambiguous assignments and difficult integration.

2) Escaping overlap: go heteronuclear (¹⁵N, ¹³C) 🚀

¹⁵N-edited 3D NOESY/TOCSY: spreading peaks into a third axis

To reduce overlap, you add a third frequency dimension using ¹⁵N labeling:

• Peaks from the 2D ¹H–¹H spectrum get distributed across multiple ¹H–¹H planes, each corresponding to a ¹⁵N chemical shift (or narrow range).
• In a given plane, you mainly see correlations for protons linked to HN attached to N with that ¹⁵N shift. This increases the size range of proteins you can assign compared to unlabeled 2D.

Still, the chapter notes that beyond roughly 18–20 kDa, ¹⁵N editing alone is often insufficient.

Triple-resonance (¹H/¹⁵N/¹³C): the modern backbone assignment engine ⚙️

For larger proteins (and often even for smaller ones today), you use doubly labeled ¹³C,¹⁵N proteins and triple-resonance experiments.

Why triple resonance helps

• Uses heteronuclear J couplings (¹J or ²J) that are relatively large → efficient transfer, high sensitivity, shorter transfer times, and less loss from fast relaxation than ¹H–¹H methods.
• Spectra are simpler: few signals per frequency → less overlap.

How naming works (important detail)

Experiment names list nuclei in the order magnetization travels; nuclei in parentheses are relay nuclei whose frequencies are not detected.

• HNCO: HNᵢ → Nᵢ → COᵢ₋₁ (then back) and you record HN, N, CO.
• HN(CA)CO: uses Cα as a relay; still records HN, N, CO (Cα not directly recorded).

Backbone sequential assignment strategy (pairs)

The chapter emphasizes analyzing experiments in pairs:

• HNCO / HN(CA)CO: identifies COᵢ₋₁ vs COᵢ connections (one is more sensitive than the other).
• HNCA / HN(CO)CA: gives Cαᵢ and Cαᵢ₋₁ patterns; sequential links are made by matching Cα shifts. The figure described shows a “pair of strips” at a given ¹⁵N shift: HN(CO)CA gives the previous residue’s Cα, while HNCA gives both current and previous residue Cα peaks.
• CBCANH / CBCA(CO)NH: provides Cα/Cβ of residue i and i−1; lower sensitivity but gives Cβ, useful for residue type and secondary structure via chemical shift index (CSI).

Side-chain assignment (selected examples)

• HBHA(CO)NH and H(CC)(CO)NH correlate amide H/N of one residue with Hα/Hβ or side-chain protons of the preceding residue, but sensitivity is relatively low.
• HCCH-TOCSY is highlighted as very popular: uses strong ¹H–¹³C and ¹³C–¹³C one-bond couplings to get near-complete aliphatic ¹H/¹³C assignments (some overlap remains in long aliphatic chains).
• Aromatics are harder with HCCH-TOCSY (RF power issues due to aliphatic vs aromatic carbon shift differences). Alternatives include ¹H–¹H TOCSY for small proteins or aromatic-tailored ¹H–¹³C HSQC-NOESY.

3) Very large proteins: deuteration, TROSY, and beyond 🏋️

As molecular weight increases:

• Scalar-coupling coherence transfer suffers due to shorter transverse relaxation times.
• NOESY transfer happens with magnetization along z during mixing time; two processes occur:
  • Cross-relaxation (dipolar transfer)
  • Longitudinal relaxation (returns to equilibrium)
• For large proteins, longitudinal relaxation times are long, and cross-relaxation increases with molecular weight (proportional to rotational correlation time), so NOESY intensities can actually benefit.

TROSY and labeling strategies

• TROSY reduces line broadening using interference between relaxation mechanisms; works best for deuterated proteins at high fields and can push to very large systems (the text mentions up to ~100 kDa for good linewidths).
• With additional transfer elements (CRIPT/CRINEPT) combined with TROSY, detection in ²H,¹⁵N systems can extend to even larger complexes.
• There’s also a “protonless” approach where ¹³C is directly acquired (smaller magnetic moment → narrower lines), with experiment sets for sequential assignment below ~60–70 kDa.

¹³C–¹³C NOESY for large proteins

• Most intense peaks are between carbons separated by a single bond (strong and present regardless of MW).
• Longer-range C–C connectivities benefit from intraresidue spin diffusion (stronger at longer mixing times and larger size), making ¹³C–¹³C NOESY valuable for assignments in large proteins.

4) Practical “which experiments for what size?” (Table 4.1) 📌

The chapter summarizes minimal labeling and experiment choices by protein size:

• <10 kDa (unlabeled): ¹H–¹H COSY, TOCSY, NOESY
• <15 kDa (¹⁵N): ¹⁵N HSQC-TOCSY, ¹⁵N HSQC-NOESY
• <25 kDa (¹⁵N,¹³C): triple resonance
• <80 kDa (¹⁵N,¹³C + partial ²H): triple resonance
• >80 kDa (¹⁵N,¹³C + perdeuteration): TROSY triple resonance

It also notes a modern practical point: since ¹⁵N/¹³C labeling is now relatively affordable, many labs use triple-resonance approaches even for smaller proteins (up to ~25–30 kDa), but full deuteration/TROSY is mainly worth it only when size demands it.

COSY vs TOCSY vs NOESY vs HSQC (requested comparison) ✅

Here’s a clean “mental map”:

COSY 🧩 (through-bond, local)

• What it correlates: protons that are scalar (J) coupled (mostly ²J, ³J).
• What it tells you: “These two H’s are neighbors in the bonding network.”
• Best for: starting points inside a residue.

TOCSY 🕸️ (through-bond, whole spin system)

• What it correlates: all protons within a spin system (via multiple J-coupling steps).
• What it tells you: “These signals belong to the same residue’s spin system.”
• Best for: residue-type identification (spin pattern “fingerprints”), but not unique for all residues.

NOESY 📏 (through-space, structure)

• What it correlates: protons within about ~5 Å (dipolar/NOE).
• What it tells you: “These two H’s are close in 3D.”
• Best for: sequence-specific assignment (via backbone NOEs), secondary structure patterns, and especially long-range restraints for the fold.

HSQC 🎯 (heteronuclear “fingerprint map”)

HSQC (Heteronuclear Single Quantum Coherence) isn’t expanded in detail here, but the chapter uses it in context of size-dependent experiment sets (e.g., ¹⁵N HSQC-TOCSY and ¹⁵N HSQC-NOESY) and mentions ¹H–¹³C HSQC-NOESY for aromatics.

• What it correlates (conceptually): a proton with a directly attached heteronucleus (commonly ¹H–¹⁵N for backbone amides, or ¹H–¹³C for carbons).
• What it’s used for in practice: a clean, well-resolved “address system” (each amide gives one spot), and a launching point for 3D experiments like HSQC-TOCSY / HSQC-NOESY that reduce overlap by adding the heteronuclear dimension.

One-sentence memory trick 🧠✨

• COSY: neighbors by bonds
• TOCSY: family reunion of a whole residue’s spin system
• NOESY: neighbors in space
• HSQC: heteronuclear address book (clean fingerprint), often used as the hub for bigger 3D experiments