Circular Permutation Prediction Reveals a Viable Backbone Disconnection for Split Proteins: An Approach in Identifying a New Functional Split Intein

30-second read

CP-site prediction acts as a prior for safe protein disconnection, followed by experimental validation of a new functional split intein.

Central question

Can a CPred site predicted to tolerate new termini be repurposed as a split-protein breakpoint whose fragments still assemble, fold, and perform protein trans-splicing?

Intuition

If a backbone position tolerates new termini in a circular permutation, it may also tolerate being split into two chains. The authors use CPred to choose sites and validate progressively: does the single-chain CP retain its fold, do the two fragments still fold, and can they still ligate two exteins?

Why it matters

This is a complete predictor-to-experiment loop. Rather than stopping at AUC, the study constructs SP36 and measures structure, thermal stability, binding, and PTS kinetics, showing how computational ranking narrows protein-engineering search.

Prerequisites

• Understand inteins, exteins, and protein trans-splicing (PTS).
• Know the difference between a circular-permutation site and a true split site.
• Know the basic meaning of NMR HSQC, circular dichroism, melting temperature, and ITC.
• Understand rate constant, half-time, yield, and binding affinity as distinct endpoints.

paper-specific guide · plain → technical → input → output → source

Method walkthrough

1. 01 · Select NpuInt breakpoints with CPred

   Choose local score peaks, avoid the active site, and deliberately favor the less-explored N-terminal half.

   Technical reading: CPred is run on the 137-residue Npu DnaE intein (PDB 2KEQ); E12–Y13 and N36–G37 (scores >0.8) are selected, with the natural N102–I103 site as a reference.

   Input: The NpuInt solution structure and known active/split-site context.

   Output: A prioritized experimental list at sites 12, 36, and 102.

   Boundary: Among nine Ssp DnaB sites scoring above 0.7, only five are functional; a high score narrows the search but does not guarantee success.

   PDF pp. 2–3, Assessment of the Intein CP Site Prediction and Figure 1

2. 02 · Validate structure and thermal stability from CP to SP

   First construct single-chain CP12/36/102 to verify that new termini preserve the fold, then split them into SP12/36/102 and test two-fragment assembly.

   Technical reading: 13Cα/13Cβ secondary shifts, 1H–15N HSQC, and CD are compared with native C1G; 224-nm CD thermal denaturation estimates Tm and van't Hoff thermodynamics.

   Input: Purified CP and SP constructs.

   Output: Secondary/tertiary fold correspondence, Tm, ΔH, ΔS, and ΔG.

   Boundary: Spectral similarity and high Tm support folding/stability but do not alone establish PTS catalytic function.

   PDF pp. 2–8, Figures 2–5 and Table 1

3. 03 · Measure PTS kinetics, yield, and fragment affinity

   Attach GB1 to both intein fragments, test whether fragment assembly produces GB1–GB1, and measure fragment binding by ITC.

   Technical reading: SDS-PAGE time courses and mass spectrometry confirm ligated product and support PTS rate constant, t1/2, and yield estimates; ITC compares Kd, ΔH, and TΔS for SP102N/C, SP36N/C, and cross-pairs.

   Input: GB1–intein fragment substrates and complementary/cross fragment pairs.

   Output: Functional PTS kinetics, ligation yield, and binding specificity/affinity.

   Boundary: Stronger affinity correlates with faster PTS in this system, but the data do not establish a universal causal law.

   PDF pp. 6–9, Figures 6–7 and PTS/association sections

Key result

Computational screening narrowed experimental search and found a workable split, translating CP knowledge into a synthetic-biology tool.

Evidence-guided deep reading

Paper facts, project readings, and teaching models are labelled separately.

paper-fact

CPred is a negative filter, not a success certificate

Functional split sites across three inteins mostly lie near CPred local maxima. All five functional Ssp DnaB sites exceed 0.7, but four nonfunctional sites also exceed 0.7; low scores below 0.5 correspond strongly to invalid splitting.

The study therefore does not simply choose the highest score; it combines a score peak, distance from the active site, and application-oriented fragment length.

Source locator: PDF pp. 2–3, Assessment of the Intein CP Site Prediction and Figure 1

paper-fact

An evidence chain from fold to function

CP36/CP102 secondary shifts and HSQC spectra largely match native C1G, with major shifts concentrated near old/new termini. SP constructs also show well-structured spectra. CP Tm values all exceed 85°C, SPs remain above 65°C, and SP102 is about 91.2°C.

Functionally, SP102 has a rate constant of 9.0×10^-3 s^-1, t1/2 about 1.3 min, and about 80% splicing by 5 minutes. New SP36 reaches 6.2×10^-4 s^-1, t1/2 about 18 minutes, and about 65% yield by one hour. SP36's 36-aa N-terminal fragment is amenable to chemical synthesis.

Source locator: PDF pp. 2–8, Figures 2–6 and Table 1

project-reading

Each endpoint answers only its own question

HSQC/CD address fold similarity, Tm addresses thermal stability, Kd addresses fragment association, and rate constant/yield address PTS function. No single endpoint substitutes for the other three.

Retrospective validation using only successful constructs overestimates utility. A complete next round should preregister candidates, include low-score controls, and report expression/purification failures.

Source locator: PDF pp. 2–9, Results and Discussion; PDF p. 9, Summary

Study design and evaluation

Data and samples

Known functional/nonfunctional split sites in Npu DnaE, Ssp DnaE, and Ssp DnaB are first compared. NpuInt CP12/36/102 and SP12/36/102 are then constructed, with native C1G and natural SP102 as references.

Baselines

• Native single-chain NpuInt C1G, natural split SP102, slower SP12, and literature Ssp DnaB split sites.

Metrics

Spectral/CD correspondence and Tm

Compare secondary/tertiary fold and heat-induced unfolding transition.
Boundary: Does not directly measure enzyme function.

PTS rate constant / t1/2 / yield

Quantify trans-splicing speed and completion fraction.
Boundary: Depends on assay substrate, temperature, and concentration; values from different conditions cannot be ranked directly.

ITC Kd

The equilibrium dissociation constant of complementary intein fragments; smaller values indicate stronger affinity.
Boundary: Affinity is not catalytic rate, and mutations/fusion partners affect measurement.

Reported result

CPred-guided site 36 yields a folded, PTS-active SP36 with Tm 67.9°C, rate constant 6.2×10^-4 s^-1, t1/2 about 18 minutes, and about 65% yield at one hour. SP102 binds at 1.7 nM versus 1.2 μM for SP36, roughly 700-fold different and consistent with their PTS-speed difference without proving universal causality.

PDF pp. 7–9, Figures 6–7 and Table 1

teaching-model · not a reported experiment

Teaching example (not a reported experiment)

Turn a predictor hit into a validation funnel

Project teaching model: a 150-aa enzyme has three CPred peaks: A near the active site, B near the N-terminus, and C in a buried helix.

1. Rank by score, active-site distance, fragment length, and secondary structure; prioritize B while retaining A/C as controls.
2. First construct single-chain CP-B and test whether new termini preserve the fold by spectroscopy/stability.
3. Then construct split-B and measure fragment association, enzyme activity, and product identity in sequence, recording failures at every layer.

Takeaway: A staged funnel locates failure in prediction, folding, assembly, or catalysis instead of preserving only a success story.

outside-scope

Evidence boundary versus FAST

The question is breakpoint viability, not geometric alignment error.

Lawful source and access

12 pages · SHA-256 f76bf53dc1b14c1680674f323e99d02b24c4cd71db7508459022246e85c50871

Lawful open full text.

Europe PMC open-access PDF

Limits and misreadings

• Validation on limited intein systems does not guarantee universal generalization.

Source locator map

1. PDF pp. 2–3, Assessment of the Intein CP Site Prediction and Figure 1
2. PDF pp. 2–6, Figures 2–5
3. PDF pp. 7–8, Figure 6 and Table 1
4. PDF pp. 7–9, Figure 7 and fragment-association analysis
5. PDF p. 9, Table 2 and Summary

Check understanding

1. Why construct a CP variant before a split variant?

   Answer: It first isolates whether new termini disrupt the fold, then tests whether separated fragments can reassemble.

   The two steps separate topology tolerance from fragment association.

2. Is CPred >0.7 sufficient to call a functional split?

   Answer: No. Ssp DnaB contains high-scoring but nonfunctional sites.

   The score provides enrichment, not a deterministic rule.

3. What does SP36's larger Kd than SP102 mean?

   Answer: SP36 fragments bind more weakly.

   This is consistent with slower PTS but cannot establish universal causality from two constructs.

Completion task: Write a preregistered validation matrix for a new split enzyme with at least three high-, two medium-, and one low-score site, specifying pass/fail criteria for expression, fold, Tm, Kd, activity, and product identity.

Paper-specific glossary

Intein

A protein internal segment that excises itself and ligates flanking exteins.

Protein trans-splicing

Covalent ligation of exteins from separate polypeptides after split-intein fragments assemble.

HSQC

An NMR experiment whose internuclear correlation peaks provide a residue-level fold fingerprint.

Kd

Equilibrium dissociation constant; under the same conditions, smaller generally means stronger binding.