Nuclear Magnetic Resonance-Guided Structural Analysis of Moderate-Affinity Protein Complexes with Intrinsically Disordered Polypeptides

Abstract

Binding affinity of an individual binding site of an intrinsically disordered protein for its folded partner may be moderate. In such cases, a straightforward determination of the structure of the binding interface is difficult. We offer a hybrid protocol combining NMR chemical shift information, NMR spectral data on amino acid residue sequence substitution effects, residual dipolar coupling, and molecular dynamics simulation that allowed us to determine the structure of a complex between the intrinsically disordered tropomyosin-binding site of leiomodin and a coiled-coil peptide modeling the N-terminal fragment of tropomyosin. The protocol can be used for other moderate-affinity complexes composed of an intrinsically disordered peptide bound to a structured protein partner.

1. Introduction

The ability of a comparatively small and flexible intrinsically disordered protein (IDP) to extend and simultaneously bind to more than one distant binding site makes IDPs likely components of protein assemblies [1]. By wrapping around a large protein complex, a single IDP chain may tether several well-folded domains and ensure a high level of binding affinity and specificity. Even when each of the binding sites has only a moderate affinity for its binding partner, the multivalent mode of interaction results in a high cumulative strength of binding, an effect known in the field of antibody/antigen interactions as avidity [2]. Additionally, should regulatory processes lead to the need for assembly reorganization, binding domains of flexible IDPs can be displaced from their binding sites one by one. The kinetics of such multi-step mechanism of dissociation can be significantly faster than that of a dissociation following a two-state mechanism [3].

While moderate affinity of individual binding sites in a multivalent IDP provides a clear benefit for biological functions and regulation [4], it represents a considerable challenge for research if the goal is to determine the three-dimensional (3D) structure of the assembly. A common deconstruction approach which focuses on determining the structure of complexes between two components of the assembly is often problematic. Attempts to co-crystallize domains with low binding affinities may be unsuccessful [5, 6]. Fusing IDP domains with domains improving the binding helps co-crystallizing protein complexes [7], but may introduce misleading structural artifacts [8]. The usefulness of standard protein NMR is also limited, because kinetic processes of dissociation/association of a moderate-affinity protein complex lead to resonance line broadening [9]. Line broadening interferes with peak assignments and complicates collection of a sufficient number of structural restraints required for a high-definition structure.

When—due to their inherent limitations—classical methods of protein structure determination fail, hybrid methods of protein structure determination come to the forefront [10]. The general philosophy of hybrid methods is to generate a low-resolution structure from sparse data obtained by a range of structural methods and further refine it using computational techniques. In cases when NMR spectra are broadened non-uniformly, a partial spectral assignment is often possible, which can be used in combination with other methods to determine the structure [11]. The protocol described below uses a combination of NMR chemical shift information, effects of amino acid residue substitution on NMR spectra, residual dipolar coupling data, and molecular dynamics simulation to determine the structure of a complex between the tropomyosin-binding site of leiomodin and a peptide modeling the N-terminal fragment of tropomyosin [11]. Although the structure did not achieve the level of definition typical of an NMR structure of a well-behaved protein of equal size, it resolved a long-standing debate about the role of leiomodin in actin filament length regulation and maintenance.

Hybrid approaches tend to require case-by-case strategies of data collection and interpretation and therefore they are difficult to standardize. However, although the protocol presented here was developed specifically for the leiomodin/tropomyosin complex, many aspects of it can be generalized for other protein complexes. First, all the expertise required to follow this protocol—from cloning and site-directed mutagenesis to recombinant protein production, isotope labeling, NMR data collection/interpretation, and molecular modeling—is present in a typical protein NMR lab. Second, the expression protocol makes use of MFH, a reliable expression tag designed to produce mg quantities of disordered peptides and IDPs in Escherichia coli [12]. We have used it extensively to successfully produce a wide range of disordered peptides or peptides capable of spontaneous refolding [5, 13–15]. Although the MFH tag is directed to inclusion bodies in cells, it is very soluble upon extraction and does not precipitate in non-denaturing buffers, which makes it very useful [12]. Finally, mutagenesis or chemical modification of amino acid residues exposed on the surface of a well-folded component of the complex will produce a clear-cut localized response of an IDP NMR spectrum which helps map the interaction interface between an IDP and its partner. The sensitivity of chemical shifts of IDP NMR peaks allows one to make these substitutions very similar in chemical nature so as not to destroy the binding.

The following protocol details preparation and structure determination of the complex formed by two peptides, Lmod2s1 and αTM1a_1–14Zip. Lmod2s1 is an intrinsically disordered tropomyosin-binding region of leiomodin-2, a protein involved in regulation of the actin-thin filament length [16]. αTM1a_1–14Zip is a folded chimeric coiled-coil peptide modeling the N-terminus of tropomyosin, a protein bound to the actin filament and leiomodin-2 [13]. In this protocol we will report the preparation of Lmod2s1, the intrinsically disordered part of the complex. However, the preparation of the well-folded αTM1a_1–14Zip will not be covered, since approaches to express, isotopically label, and purify soluble folded proteins and their mutants at high yields are well known. Interested readers will find the details of recombinant production and purification of αTM1a_1–14Zip in our earlier work [11, 13].

2. Materials

2.1. Expression and Purification of the Lmod2s1 Peptide

2.1.1. BL21(DE3) Transformation

1. BL21(DE3) chemically competent E. coli cells.

2. Lysogeny Broth (LB) medium (sterile).

3. LB agar petri dish.

4. 100 mg/mL filter-sterilized ampicillin solution (1000×): sterilize a 100 mg/mL ampicillin solution in distilled water by filtrating it in sterile conditions through a filter with ≤0.2 μm pore size. Store at −20 °C.

5. LBA (LB plus ampicillin) agar plate: add 15 μL of 1000× ampicillin solution to 100 μL LB medium, spread the mix aseptically on the surface of a 100 × 15 mm LB agar petri dish, let it dry in a laminar flow cabinet for ~10 min.

6. Bacterial vector pET21-MFH-Lmod2s1 for inducible expression of fusion protein MFH-Lmod2s1 (see Fig. 1), 10 ng/μL (see Note 1).

7. Temperature controlled water bath.

8. Temperature controlled shaking incubator.

Fig. 1.

The DNA sequence of the recombinant insert in pET21-MFH-Lmod2s1. The bold sequence gaattc between MFH and the sequence encoding Lmod2s1 is a unique EcoRI cloning site. The atg sequence encoding Met used to release Lmod2s1 by cyanogen bromide is underlined

2.1.2. Expression of a ¹⁵N, ¹³C-Labeled Fusion Protein

1. LBA (LB plus ampicillin) medium (100 mL): add aseptically 100 μL of ampicillin solution (1000×, from Subheading 2.1.1) to 100 mL LB (from Subheading 2.1.1).

2. Sterile distilled water (1 L): autoclave.

3. 6 M HCl.

4. Filter-sterilized 100 mM (1000×) isopropyl β-D-1-thiogalactopyranoside.

5. Sterile 1 M MgSO₄ (20 mL): dissolve 4.93 g magnesium sulfate heptahydrate in distilled water, adjust volume to 20 mL, autoclave.

6. 100× BME vitamins.

7. Filter-sterilized 1000× trace elements: 20 mM FeCl₃, 4 mM CaCl₂, 2 mM ZnSO₄, 0.4 mM CoCl₂, 0.4 mM CuSO₄, 0.4 mM NiCl₂, and 0.4 mM H₃BO₃ in distilled water, filter-sterilize. Store at −20 °C.

8. Autoclaved 5× MIA salts (1 L): prepare 1 L of 250 mM Na₂HPO₄, 250 mM KH₂PO₄, 150 mM NaCl in distilled water, adjust pH to 7.0 with 6 M HCl, and autoclave.

9. Sterile 1× MIA salts (10 mL): mix aseptically 2 mL of 5× MIA salts with 8 mL of sterile water.

10. Source of ¹⁵N and ¹³C isotopes for minimal labeling expression medium (10 mL, per 100 mL of medium): dissolve 0.37 g ¹⁵N-ammonium sulfate and 0.4 g U- ¹³C₆-D-glucose in 5 mL of distilled water, filter-sterilize and collect into a 15 mL sterile conical centrifuge tube. Wash the filter by aseptically running an additional 5 mL of distilled water through it and collect the sterile wash into the same tube.

11. Minimal labeling expression medium MIA (100 mL) (see Note 2): to a sterile 2 L shake flask (see Note 3) add aseptically—in this order—20 mL of autoclaved 5× MIA salts, 68.6 mL autoclaved distilled water, ¹⁵N and ¹³C isotopes (10 mL, from step 10), 0.2 mL of autoclaved 1 M MgSO₄, 1 mL of 100× BME vitamins, 0.1 mL 1000× trace elements, and 0.1 mL 1000× ampicillin.

12. Large capacity, high speed centrifuge.

13. Desktop microcentrifuge.

14. UV-Vis spectrophotometer.

2.1.3. Ni-NTA Purification of the Fusion Protein

1. Ni-NTA resin (QIAGEN or similar), washed with water before use.

2. Dialysis tubing with a 12–14 kDa molecular weight cut-off.

3. Crystalline urea.

4. Autoclaved water.

5. 6 M HCl.

6. 1 M imidazole, pH 8.0: dissolve imidazole in autoclaved distilled water, add 6 M HCl until the pH is 8.0. Dilute with water to the molarity of 1 M.

7. 25/16× buffer A, pH 8.0: dissolve in 190 mL of autoclaved distilled water 2.02 g Na₂HPO₄, 0.11 g NaH₂PO₄•H₂O, and 5.26 g NaCl. Add 3 mL 1 M imidazole (pH 8.0). Do not adjust the pH. Store at 4 °C.

8. 1× buffer A, 8 M urea, pH 8.0, 100 mL: to 64 mL 25/16× buffer A, pH 8.0, add 48 g crystalline urea. Stir or shake at room temperature until urea is completely dissolved (see Note 4). Do not adjust volume or pH.

9. Elution Buffer: 0.6 × buffer A, 8 M urea, supplied with additional 250 mM imidazole, pH 8.0, 25 mL. Mix 6.25 mL 1 M imidazole, pH 8, with 9.75 mL 25/16× buffer A, pH 8.0, then add 12 g crystalline urea. Stir or shake at room temperature until urea is completely dissolved (see Note 4). Do not adjust volume or pH.

10. Large capacity, high speed centrifuge.

11. Ultrasonic sonicator.

12. Freeze-dryer.

2.1.4. Fusion Protein Cleavage with Cyanogen Bromide (see Note 5)

1. Cyanogen bromide.

2. Crystalline urea.

3. 6 M HCl.

4. 5 M NaOH.

2.1.5. Desalting of Cleaved Fusion Proteins

1. HPLC-grade distilled water.

2. HPLC-grade trifluoroacetic acid (TFA).

3. HPLC-grade acetonitrile.

4. 10% HPLC-grade trifluoroacetic acid (200 mL): using a graduated glass cylinder, measure 20 mL of TFA. Separately measure 180 mL of HPLC-grade water. In a glass beaker slowly add with stirring 20 mL of TFA into 180 mL water. Store in a glass bottle.

5. 0.1% HPLC-grade trifluoroacetic acid (100 mL): add 1 mL of 10% TFA to 99 mL water in a glass beaker, mix. Store in a glass bottle.

6. 0.1% HPLC-grade trifluoroacetic acid, 60% acetonitrile (100 mL): measure in a graduated glass cylinder 60 mL acetonitrile. In a glass beaker mix 1 mL of 10% TFA, 60 mL acetonitrile, and 39 mL water. Store in a glass bottle.

7. Sep-Pak C₁₈ Cartridge (see Note 6).

8. Freeze-dryer with a −80 °C or lower temperature cold trap.

2.1.6. The Removal of the Tag Protein

1. 6 M HCl.

2. Ni-NTA resin, washed with water before use.

3. 1× buffer A, 8 M urea, pH 8.0 (from Subheading 2.1.3).

2.1.7. Orthogonal Two-Step Reversed-Phase HPLC Purification of the Recombinant Peptides

1. 6 M HCl.

2. 0.1% HPLC-grade TFA.

3. HPLC-grade acetonitrile.

4. 0.1% TFA in HPLC-grade acetonitrile (100 mL): measure in a graduated glass cylinder 99 mL acetonitrile. In a glass beaker add to the acetonitrile 1 mL of 10% TFA, mix. Store in a glass bottle.

5. 1% ammonium bicarbonate.

6. HPLC reversed-phase Vydac 218TP C18 column, 250x × 4.6 mm 5 μm or comparable.

7. HPLC pH-resistant reversed-phase Waters XBridge BEH C18 Column, 130 Å, 5 μm, 250 mm × 4.6 mm or comparable.

8. HPLC machine equipped with a 2–5 mL injection loop, a UV detector (see Note 7), and capable of flow rates 1–2 mL.

2.2. NMR Data Collection and Analysis

2.2.1. NMR Sample Preparation

1. Autoclaved ultrapure water.

2. Purified ¹⁵N, ¹³C-labeled Lmod2s1.

3. Purified αTM1a_1–14Zip.

4. 2× NMR sample buffer: 100 mM disodium phosphate, 0.4 mM ethylenediaminetetraacetic acid (EDTA), 0.2% sodium azide, 0.4 mM sodium trimethylsilylpropanesulfonate (DSS), and 4× Pierce EDTA-free protease inhibitor cocktail in autoclaved ultrapure water. Do not adjust pH.

5. 0.1% TFA with 10% (v/v) acetonitrile (100 mL): measure in a graduated glass cylinder 10 mL acetonitrile. In a glass beaker add 1 mL of 10% TFA, 10 mL acetonitrile to 89 mL water, mix. Store in a glass bottle.

6. Deuterium oxide, >99.9 atom% D.

7. 1 M HCl.

8. 1 M NaOH.

9. A 5 mm Shigemi NMR microtube assembly.

10. A pH-meter with a microelectrode.

11. A tabletop microcentrifuge capable of creating the relative centrifugal force of 10,000 × g.

2.2.2. NMR Spectra Collection and Analysis

1. 1-Octanol.

2. Pentaethylene glycol monooctyl ether (C8E5).

2.2.3. Influence of Amino Acid Substitution in the Folded Component of the Complex on the IDP Spectra

1. Ac-αTM1a_1–14Zip [11].

2. αTM1a_1–28Zip [13].

2.3. NMR-Guided Molecular Dynamics Simulations (MDS) of the Lmod2s1 and αTM1a_1–14Zip Complex

1. A Linux OS computer for running MDS. Suggested hardware for running simulations with Amber [17–19] are available at the following link: https://ambermd.org/GPUHardware.php

   At least one GPU and one CPU are required to run the pmemd version of Amber, which offers a significant boost in speed over the sander version. The sander version runs on CPU only. With the pmemd version of Amber, a simulation of a ~25 k atom NPT system provides a trajectory of ~110 ns/day on a Titan-Xp GPU and ~40 ns/day on a Tesla K20c GPU at 1 fs time step. Alternatively, you may choose to utilize a cloud computing service such as Rescale or Penguin Computing accessed at the following links: Rescale: https://rescale.com; Penguin Computing: https://www.penguinsolutions.com/computing/

2. AMBER18 and AMBERTools18 program on the Linux computer or cloud computer. Directions for installing the latest version of Amber and AmberTools is available at the provided link: https://ambermd.org/GetAmber.php

3. UCSF Chimera, a molecular visualization program, on a Windows/Mac OS computer. Directions for downloading the latest version of UCSF Chimera is available at the provided link: https://www.cgl.ucsf.edu/chimera/download.html

4. Either WinSCP (Windows) or Cyberduck (Mac), an SSH file transfer program on a Windows/Mac OS computer. Directions for download of the programs are available at the provided links:

   WinSCP: https://winscp.net/eng/download.php

   CyberDuck: https://cyberduck.io/download/

5. Either PuTTY (Windows) or Cyberduck (Mac), an SSH terminal emulator on a Windows/Mac OS computer. Directions for download of the programs are available at the provided links:

   PuTTY: https://www.chiark.greenend.org.uk/~sgtatham/putty/latest.html

   CyberDuck: https://cyberduck.io/download/

3. Methods

3.1. Expression and Purification of Lmod2s1

3.1.1. BL21(DE3) Transformation by Heat Shock

1. Set the water bath temperature to 42 °C.

2. Thaw on ice 50 μL competent BL21(DE3) cells, transfer to an ice-cold 1.5 mL plastic microcentrifuge tube. Add 1 μL (10 ng) of the bacterial expression vector to the cells. Incubate on ice for 30 min.

3. Subject the cells to a heat shock in the water bath for 30 s.

4. Put the tube back on ice for 2 min.

5. Add 0.5 mL of LB to the tube and incubate it horizontally in the shaking incubator (275 rpm) for 45 min at 37 °C.

6. Plate 20 μL of the culture on a LBA agar plate and incubate the plate at room temperature for 1.5–2 days until the colonies reach the size of approximately 1 mm (see Note 8).

7. Seal the plates with Parafilm and store with the lid down at 4 °C.

3.1.2. Expression of a ¹⁵N, ¹³C-Labeled Fusion Protein (see Note 9)

1. Inoculate 5 mL of freshly prepared LBA in a 50 mL plastic centrifuge tube with a BL21(DE3) colony transformed with pET21-MFH-Lmod2s1 and incubate at 37 °C in a shaking incubator (250 rpm) overnight.

2. Centrifuge the overnight culture from step 1 at 2000 g for 10 min. Remove the supernatant with a sterile pipette, add 1 mL of sterile 1× MIA salts, resuspend the cells and centrifuge at 2000 g for 10 min again. Remove and discard the supernatant. Resuspend the cells in 1 mL of sterile 1× MIA salts one more time.

3. Add the resuspended cells to the minimal labeling expression medium MIA. Shake the flask at 220 rpm in a shaking incubator at 37 °C until the optical density at 600 nm reaches 0.5–0.7 (see Note 10). Add 0.2 mL 100 mM isopropyl β-D-1-thiogalactopyranoside and 0.05 mL of 100 mg/mL ampicillin (from Subheading 2.1.1) per 100 mL culture. Continue shaking the induced culture in the incubator for 10–12 additional hours which will normally end in the morning of the next day.

4. Withdraw 50 μL of the induced culture, add 450 μL of 1× MIA salts, measure the optical density of the cell suspension at 600 nm. For a good yield, it should be between 0.4 and 0.6.

5. Harvest cells by centrifuging the culture at 4000 g for 20 min, decant supernatant, freeze the cell pellet at −20 °C (see Note 11).

3.1.3. Ni-NTA Purification of the Fusion Protein (see Note 12)

1. Thaw the frozen cell pellet on ice, reconstitute in 20 mL of 1× buffer A, 8 M urea, pH 8.0 at room temperature.

2. Sonicate the resuspended cells on slush ice using a total of 300 pulses of 2 s interrupted by 8 s rest periods to avoid overheating.

3. Remove insoluble cell debris by centrifuging the disrupted cells at 15,000 × g, room temperature (see Note 13).

4. Equilibrate the Ni-NTA column with seven bed volumes of 1× buffer A, 8 M urea, pH 8.0.

5. Apply the cell lysate to the Ni-NTA column, wash with seven bed volumes of 1× buffer A, 8 M urea, pH 8.0.

6. Elute the fusion protein with the Elution Buffer (see Note14).

7. Dialyze the fusion protein against 10 mM HCl for 2–3 days with 2–3 solution changes (see Note 15).

3.1.4. Fusion Protein Cleavage with Cyanogen Bromide (see Note 5)

1. Add 6 M HCl to the dialyzed fusion protein to final 100 mM HCl. Place the solution in a 50 mL plastic conical centrifuge tube. If precipitate is observed, add crystalline urea up to 4–6 M until the solution is clear (see Note 15).

2. Add 400–1000× molar excess of crystalline cyanogen bromide, gently shake the tube until CNBr is dissolved, wrap the tube in foil and leave at room temperature overnight (see Note 16).

3.1.5. Desalting of Cleaved Fusion Proteins

1. Pre-wet the Sep-Pak cartridge with acetonitrile (see Note 6) according to the manufacturer’s recommendations.

2. Equilibrate the cartridge with 0.1% TFA (7 mL/per each gram of sorbent).

3. Under the chemical hood, apply the cleaved fusion protein to the cartridge.

4. Wash the cartridge with 0.1% TFA (7 mL/per each gram of sorbent).

5. Elute the cleaved fusion protein with 0.1% TFA/60% acetonitrile (see Note 17).

6. Dilute the eluted fraction 3 times by adding 2 volumes of 0.1% TFA, then freeze-dry.

3.1.6. The Removal of the Tag Protein

1. Equilibrate the Ni-NTA column with seven bed volumes of 1× buffer A, 8 M urea, pH 8.0.

2. Reconstitute the desalted cleaved fusion protein in 2 bed volumes of 1× buffer A, 8 M urea, pH 8.0, apply to the Ni-NTA column.

3. Collect the flow-through fraction, wash the column with 2–3 bed volumes of 1× buffer A, 8 M urea, pH 8.0, combine the wash and the flow-through fractions. Add 6 M HCl to final 0.2 M HCl, repeat desalting steps of Subheading 3.1.5.

3.1.7. Orthogonal Two-Step Reversed-Phase HPLC Purification of Lmod2s1

1. Purify Lmod2s1 on a Vydac C18 column with 1%/min acetonitrile gradient in the presence of 0.1% TFA. Freeze-dry.

2. Purify Lmod2s1 on an XBridge C18 column using a 1%/min ammonium bicarbonate/acetonitrile gradient (see Note 18).

3. Acidify the collected peptide fractions immediately after purification by adding 6 M HCl to final 0.1 M HCl.

4. Dilute the peptide fractions four times, apply to a reversed-phase column, desalt for 1 h with 0.1% TFA at the rate of 0.5 mL/min, then elute with a 1%/min acetonitrile gradient in the presence of 0.1% TFA (see Note 19).

5. Freeze-dry the peptide.

3.2. NMR Data Collection and Analysis

3.2.1. NMR Sample Preparation

1. Dissolve αTM1a_1–14Zip in 0.25–0.5 mL ultrapure water.

2. Reconstitute freeze-dried labeled Lmod2s1 in 0.25–0.5 mL ultrapure water (see Note 20).

3. Determine concentrations of both peptides using either the difference spectrum of the peptide dissolved in 6 M guanidine at pH 12.5 versus pH 7.1 or a microbiuret procedure (described in detail in Greenfield (2006)) [20].

4. Make 3–4 holes in the cap of a 1.5 mL microcentrifuge tube using a hypodermic needle.

5. Calculate the amount of labeled Lmod2s1 to give the final concentration of 0.5–0.6 mM if dissolved in 320 μL (see Note 21), withdraw the appropriate volume, transfer into the tube.

6. Transfer 2 to 3× molar excess of αTM1a_1–14Zip into the same tube. Close the cap, cover with foil, freeze, remove foil, and lyophilize.

7. Reconstitute the dry NMR sample in 160 μL 2× NMR buffer.

8. Adjust the pH to 6.5 or other desired value (typically <7, see Note 22) by titrating the sample with 0.5–5 μL aliquots of 1 M HCl or NaOH.

9. Add 32 μL D₂O.

10. Add water to 320 μL.

11. If the sample is cloudy, spin down the aggregates at 10,000 × g, for 10 min.

12. Transfer the cleared NMR sample into an NMR tube.

3.2.2. NMR Spectra Collection and Analysis

1. Record and assign 2D ¹⁵N-HSQC and ¹³C-HSQC and 3D HNCO, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, HBHA(CO)NH, HCCH-TOCSY, and ¹⁵N- and ¹³C-edited NOESY using NMRViewJ (One Moon Scientific) (see Note 23).

2. Freeze-dry the sample, transfer it into the 5% 1-octanol/C8E5 bicelles (see Note 24), record [21] and determine residual dipolar couplings.

3.2.3. Influence of Amino Acid Substitution in the Folded Component of the Complex on the IDP Spectra (see Note 25)

1. Repeat steps from Subheading 3.2.1 but use either Ac-αTM1a_1–14Zip or αTM1a_1–28Zip instead of αTM1a_1–14Zip.

2. Record a 2D ¹⁵N-HSQC spectrum for each complex.

3. Overlay the 2D ¹⁵N-HSQC spectra from Lmod2s1/αTM1a_1–14Zip and Lmod2s1/Ac-αTM1a_1–14Zip (or Lmod2s1/α TM1a_1–28Zip).

4. Map the residues for which peaks in the assigned 2D 15N-HSQC spectrum from Lmod2s1/αTM1a_1–14Zip do not overlap with those in the 2D ¹⁵N-HSQC spectrum for Lmod2s1/Ac-αTM1a_1–14Zip (or Lmod2s1/αTM1a_1–28Zip). These are the residues that are the most affected by the changes in the chemical structure of Ac-αTM1a_1–14Zip (or Lmod2s1/α TM1a_1–28Zip) compared to αTM1a_1–14Zip.

3.3. NMR-Guided MDS of the Lmod2s1/α TM1a_1–14Zip Complex

3.3.1. Generation of the Initial Structures of the Lmod2s1 and αTM1a_1–14Zip Complexes

1. In the NMRViewJ program, write the assigned Lmod2s1 chemical shifts (from Subheading 3.2.2) to an appropriately formatted file by clicking through the following sequence:

   >Assign

   >Atoms

   >File

   >Write Chemical Shift Files

   >Write TALOS PPM...

2. Save the TALOS PPM file for the Lmod2s1 chemical shifts.

3. Navigate to the TALOS+ web server at the link: https://spin.niddk.nih.gov/bax/nmrserver/talos/

4. Click the “Choose File” button and choose the Lmod2s1 TALOS PPM file that was saved in step 2.

5. Check the “Exclude Identical Sequences” box.

6. Check the “Apply Offset Correction” box.

7. Check the “Auto 1H Mode” option.

8. Enter and confirm your email and click the “Submit” button. Once the calculation is done, you will receive the results which will include the file pred.tab with predicted backbone dihedral angles (see Note 26).

9. In UCSF Chimera, initiate building a peptide structure by clicking through the following sequence.

   >Tools

   >Structure editing

   >Build structure

   >Start structure

   >Add peptide

10. Copy and paste the sequence of the Lmod2s1 peptide into the “peptide sequence” box, then press “Apply.”

11. In the dialog box that appears, set the dihedral angles to those that were predicted by TALOS+ in the pred.tab output file from step8, then press “Ok” or “Apply” to build the structure (see Note 27).

12. Open the αTM1a_1–14Zip structure (PDBID:1TMZ).

13. Activate the Chimera command line by clicking Favorites → Command Line. Isolate the first structure from the ensemble of 15 structures in 1TMZ by typing the following commands into the UCSF Chimera command line:

    select #1.2–15

    delete sel

14. Add a Gly residue to the N-terminus of both chains of the 1TMZ structure by typing the following commands into the UCSF Chimera command line:

    addaa gly,0 #1.1:1.A

    addaa gly,0 #1.1:1.B

15. Renumber residues in αTM1a_1–14Zip chains so all residues in the complex are numbered serially (Lmods1: res1–40, αTM1a_1–14Zip chain A: res41–73, αTM1a_1–14Zip chain B: res74–106) by typing the following commands into the UCSF Chimera command line:

    resrenumber 41 #1.1:0–32.A

    resrenumber 74 #1.1:0–32.B

16. Change chain IDs so all chains have a blank ID by typing the following commands into the UCSF Chimera command line:

    changechains A, ,

    changechains B, ,

17. Arrange the Lmod2s1 α-α hairpin over the top of the αTM1a₁14Zip coiled-coil dimer into a crisscross topology four helix bundle (see Fig. 2a) (see Note 28).

18. Combine the individual protein component models into one model by clicking through the following sequence in UCSF Chimera:

    >Favorites

    >Model panel

    Select all of the models

    >Copy/combine

    For the option “If original molecules have duplicate single-letter chain IDs, then”: choose “retain them”

    Check the “Close source models” box

    >Ok

19. Save the combined model as a PDB file called abc.pdb.

20. From your local computer, connect to your Linux computer with your SSH file transfer protocol program.

21. Move the abc.pdb to a working directory on your Linux computer.

Fig. 2.

Potential schematic topologies of a four α-helix bundle assembly in the Lmod2s1/αTM1a_1–14Zip complex. Lmod2s1 helices are shown in green, and αTM1a_1–14Zip helices are shown in black. 1YO7 (ID PDB) 4-helix bundle was used as the topology template. (a) Crisscross topology, and (b) side-by-side topology. (Modified from Tolkatchev et al. [11])

3.3.2. Running NMR-Guided NVT MDS with Dihedral Angle Restraints to Equilibrate the Initial Structure

1. Create a file called all_angles.txt specifying the lower and upper bounds of the phi and psi angles of all residues for which the “class” of prediction by TALOS+ was “good.” Use the pred. tab file from Subheading 3.3.1 (see Note 29). The format is:

   Res#

   3-letter_res_code

   angle_type

   lower_bound

   upper_bound

   Two sample lines are provided:

   17	GLU	PHI	−78.4	−38.4
   17	GLU	PSI	−58	−9.6

2. From your local computer, connect to your Linux computer with an SSH file transfer protocol.

3. Move the all_angles.txt file into the same working directory as abc.pdb (from Subheading 3.3.1) on your Linux computer.

4. From your local computer, connect to your Linux computer with an SSH terminal emulator.

5. Create a dihedral angle restraint file by typing the following command (see Note 30):

   makeANG_RST -pdb abc.pdb \
   -con all_angles.txt > angle_restraints.rst RST
   

6. Use the LEaP program in AmberTools to prepare input files for the simulations using the ff14SB forcefield [22] and TIP3P water [23] by typing the commands (see Note 31):

   tleap 
   source leaprc.protein.ff14SB 
   source leaprc.water.tip3p 
   xyz = loadpdb abc.pdb 
   addions xyz Na+ 0 
   solvatebox xyz TIP3PBOX 10.0 0.75 
   saveamberparm xyz abc_init.top abc_init.ncrst 
   quit
   

7. Prepare an input file for energy minimization by typing the commands (see Note 32):

   echo “ 
   &cntrl 
   imin=1 
   ntmin=1 
   ncyc=2500
   maxcyc=5000 
   nmropt=1 
   /
   &wt TYPE=END
   /
   DISANG=angle_restraints.rst
   “>min.in
   

8. Execute the energy minimization on the initial structure, using inputs generated in steps 6 and 7 by typing the commands:

   $AMBERHOME/bin/pmemd.cuda \
   -O \
   –i min.in \
   -p abc_init.top \
   -c abc_init.ncrst \
   -o abc_min.out \
   -r abc_min.ncrst \
   -inf abc_min.info &
   

9. Prepare an input file for initial NVT MDS by typing the following commands:

   echo “&cntrl 
   imin=0, 
   irest=0, 
   ntx=1, 
   ntt=3, 
   gamma_ln=1.0, 
   vlimit=10, 
   temp0=300.0, 
   ntf=2, 
   ntc=2,
   ntb=1, 
   dt=0.002, 
   nstlim=40000000, 
   ntwx=50000, 
   ntpr=50000, 
   ntwr=5000, 
   iwrap=1, 
   nmropt=1, 
   &end
   &ewald
   &end
   /
   &wt TYPE=END
   /
   DISANG=angle_restraints.rst
   “>md_NVT.in
   

10. Perform NVT MDS on the structure from step8 by typing the commands (see Note 33):

    $AMBERHOME/bin/pmemd.cuda \
    -O \
    –i md_NVT.in \
    -p abc_init.top \
    -c abc_min.ncrst \
    -o abc_md_NVT.out \
    -r abc_md_NVT.ncrst \
    -x abc_md_NVT.nc \
    -inf abc_md_NVT.info &
    

11. Perform an energy minimization on the structure from step 10 by typing the commands (see Note 34):

    $AMBERHOME/bin/pmemd.cuda \
    -O \
    –i min.in \
    -p abc_init.top \
    -c abc_md_NVT.ncrst \
    -o abc_md_NVT_min.out \
    -r abc_md_NVT_min.ncrst \
    -inf abc_md_NVT_min.info &
    

3.3.3. Running NMR-Guided Simulated Annealing MDS with Dihedral Angle Restraints to Eliminate Potential Metastable States

1. Prepare an input file for simulated annealing by typing the commands (see Note 35):

   echo “&cntrl 
   imin=0, 
   irest=0,
   echo “&cntrl 
   imin=0, 
   irest=0,
   ntx=1, 
   ntt=3, 
   gamma_ln=3.0, 
   vlimit=10, 
   ntf=2, 
   ntc=2, 
   ntb=1, 
   dt=0.001, 
   nstlim=130000000, 
   ntwx=50000, 
   ntpr=50000, 
   ntwr=5000, 
   iwrap=1, 
   nmropt=1, 
   &end
   &ewald
   &end
   /
   &wt TYPE = TEMP0, istep1 = 0, istep2 = 10000000, value1 = 300, value2 = 300,
   /
   &wt TYPE = TEMP0 , istep1 = 10000001, istep2 = 20000000, value1 = 310, value2 = 310,
   /
   &wt TYPE = TEMP0 , istep1 = 20000001, istep2 = 30000000, value1 = 320, value2 = 320,
   /
   &wt TYPE = TEMP0 , istep1 = 30000001, istep2 = 40000000, value1 = 330, value2 = 330,
   /
   &wt TYPE = TEMP0 , istep1 = 40000001, istep2 = 50000000, value1 = 340, value2 = 340,
   /
   &wt TYPE = TEMP0 , istep1 = 50000001, istep2 = 60000000, value1 = 350, value2 = 350,
   /
   &wt TYPE = TEMP0 , istep1 = 60000001, istep2 = 70000000, value1 = 360, value2 = 360,
   /
   &wt TYPE = TEMP0 , istep1 = 70000001, istep2 = 80000000, value1 = 350, value2 = 350,
   /
   &wt TYPE = TEMP0 , istep1 = 80000001, istep2 = 90000000, value1 = 340, value2 = 340,
   /
   &wt TYPE = TEMP0 , istep1 = 90000001, istep2 = 100000000, value1 = 330, value2 = 330,
   /
   &wt TYPE = TEMP0 , istep1 = 100000001, istep2 = 110000000, value1 = 320, value2 = 320,
   /
   &wt TYPE = TEMP0 , istep1 = 110000001, istep2 = 120000000, value1 = 310, value2 = 310,
   /
   &wt TYPE = TEMP0 , istep1 = 120000001, istep2 = 130000000, value1 = 300, value2 = 300,
   /
   &wt TYPE = END
   /
   DISANG=angle_restraints.rst
   “>annealing.in
   

2. Perform simulated annealing MDS on the structure from Subheading 3.3.2, by typing the commands (see Note 36):

   $AMBERHOME/bin/pmemd.cuda \
   -O \
   –i annealing.in \
   -p abc_init.top \
   -c abc_md_NVT_min.ncrst \
   -o abc_annealed.out \
   -r abc_annealed.ncrst \
   -x abc_annealed.nc \
   -inf abc_annealed.info &
   

3. Perform an energy minimization on the structure from Subheading 3.3.2, by typing the commands (see Note 34):

   $AMBERHOME/bin/pmemd.cuda \
   -O \
   –i min.in \
   -p abc_init.top \
   -c abc_annealed.ncrst \
   -o abc_annealed_min.out \
   -r abc_annealed_min.ncrst \
   -inf abc_annealed_min.info &
   

3.3.4. Running NMR-Guided NPT Production MDS with Dihedral Angle Restraints to Sample Many Conformations (see Note 37)

1. Prepare an input file for NMR-guided NPT production MDS by typing the commands:

   echo “&cntrl 
   imin=0, 
   irest=0, 
   ntx=1, 
   ntt=3, 
   temp0=298.0, 
   gamma_ln=3.0, 
   vlimit=10, 
   ntp=1, 
   taup=1.0, 
   ntf=2, 
   ntc=2, 
   ntb=2, 
   dt=0.001, 
   nstlim=400000000, 
   ntwx=50000, 
   ntpr=50000, 
   ntwr=5000, 
   iwrap=1, 
   nmropt=1, 
   &end
   &ewald
   &end
   /
   &wt TYPE=END 
   /
   DISANG=angle_restraints.rst
   “>md_NPT.in
   

2. Perform NMR-guided NPT production MDS on the structure from Subheading 3.3.2, by typing the commands (see Note 36):

   $AMBERHOME/bin/pmemd.cuda \
   -O \
   –i md_NPT.in \
   -p abc_init.top \
   -c abc_annealed_min.ncrst \
   -o abc_md_NPT.out \
   -r abc_md_NPT.ncrst \
   -x abc_md_NPT.nc \
   -inf abc_md_NPT.info &
   

3. Perform energy minimization on the structure from step 2 by typing the commands (see Note 34):

   $AMBERHOME/bin/pmemd.cuda \
   -O \
   –i min.in \
   -p abc_init.top \
   -c abc_md_NPT.ncrst \
   -o abc_md_NPT_min.out \
   -r abc_md_NPT_min.ncrst \
   -inf abc_md_NPT_min.info &
   

3.3.5. Refine the Structure by Running NMR-Guided MDS with Dihedral Angle and Residual Dipolar Coupling Restraints (see Note 38)

1. Create a file called all_RDCs.txt specifying the observed residual dipolar coupling values in Hz (from Subheading 3.2.2). The format is:

   Res#	observed_RDC

   Two sample lines are provided:

   15	3.19
   16	2.94

2. From your local computer, move the all_RDC.txt file into the same working directory as abc.pdb (from Subheading 3.3.1) on your Linux computer with your SSH file transfer protocol program.

3. From your local computer, connect to your Linux computer with an SSH terminal emulator.

4. Create a preliminary residual dipolar coupling restraint file by typing the following commands (see Note 31):

   makeDIP_RST.protein -pdb abc.pdb -type NH -file \ all_RDCs.txt
   

5. From your Linux computer, move the RST.dip output file to your local computer with your SSH file transfer protocol program.

6. Using a text editor, set the values of dobsl and dobsu to −/+ the uncertainty respectively in the observed RDC. Two example lines, corresponding to those in step 1 with a 1 Hz uncertainty, are provided:

   & align

   id(1) = 230, jd(1) = 231, dobsl(1) = 2.19, dobsu(1) = 4.19

   id(2) = 249, jd(2) = 250, dobsl(2) = 1.94, dobsu(2) = 3.94

7. Directly after the &align line, type the following (see Note39):

   ndip = 26, dcut =−1.0, dwt = 26*0.1,

   gigj = 26*−3.1631,

   s11 = 0, s22 = 0, s12 = 0, s13 = 0, s23 = 0,

   freezemol = .true.,

8. Save the RST.dip file and transfer it to your Linux computer with the SSH file transfer protocol.

9. From your local computer, connect to your Linux computer with an SSH terminal emulator.

10. Create a file to minimize the structure and to fit it to an alignment tensor by typing the commands (see Note 40).

    echo “ 
    &cntrl 
    imin=1 
    ntmin=1 
    ncyc=2500 
    maxcyc=5000 
    nmropt=2
    /
    &wt TYPE=END 
    /
    DISANG=angle_restraints.rst
    DIPOLE=RST.dip
    “>min_dip.in
    

11. Perform energy minimization and alignment tensor fitting on the structure from Subheading 3.3.4 by typing the commands (see Notes 36 and 41):

    mpiexec -np 8 $AMBERHOME/bin/sander.MPI \
    -O \
    –i min_dip.in \
    -p abc_init.top \
    -c abc_md_NPT_min.ncrst \
    -o abc_md_NPT_min fit.out \
    -r abc_md_NPT_min_fit.ncrst \
    -inf abc_md_NPT_min_fit.info &
    

12. In the abc_md_NPT_min_fit.info output file, locate the fitted alignment tensor components s11, s22, s12, s13, s23 and set those values in the RST.dip file created in steps 1–8 (see Note 42).

13. In the RST.dip file, set “freezemol = .false.”

14. Prepare an input file for NMR-guided MDS with dihedral angle and RDC restraints by typing the commands (see Note 40):

    echo “ 
    &cntrl 
    imin=0, 
    irest=0, 
    ntx=1, 
    ntt=3, 
    temp0=298.0, 
    gamma_ln=3.0, 
    vlimit=10, 
    ntp=1, 
    taup=1.0, 
    ntf=2, 
    ntc=2, 
    ntb=2, 
    dt=0.001, 
    nstlim=40000000, 
    ntwx=50000, 
    ntpr=50000, 
    ntwr=5000, 
    iwrap=1, 
    nmropt=2, 
    &end
    &ewald
    &end
    /
    &wt TYPE=END 
    /
    DISANG=angle_restraints.rst
    DIPOLE=RST.dip
    “>md_dip_NPT.in
    

15. Perform NMR-guided MDS with dihedral angle and RDC restraints by typing the commands (see Notes 36 and 41).

    mpiexec -np 8 $AMBERHOME/bin/sander.MPI \
    -O \
    –i md_dip_NPT.in \
    -p abc_init.top \
    -c abc_md_NPT_min.ncrst \ 
    -o abc_md_NPT_dip.out \
    -r abc_md_NPT_dip.ncrst \
    -x abc_md_NPT_dip.nc
    -inf abc_md_NPT_dip.info &
    

16. Perform an energy minimization on the structure from step15 by typing the commands (see Notes 36 and 41):

    mpiexec -np 8 $AMBERHOME/bin/sander.MPI \
    -O \
    –i min_dip.in \
    -p abc_init.top \
    -c abc_md_NPT_dip.ncrst \
    -o abc_md_NPT_dip_min.out \
    -r abc_md_NPT_dip_min.ncrst \
    -inf abc_md_NPT_dip_min.info &
    

17. Use the cpptraj program [24] in AmberTools to generate a . pdb file of the structure from step16 by typing the commands (see Note 43):

    cpptraj 
    parm abc_init.top 
    trajin abc_md_NPT_dip_min.ncrst 
    strip: WAT,Na+ 
    autoimage 
    trajout abc_md_NPT_dip_min.pdb
    

4. Notes

1. Expressed fusion protein MFH-Lmod2s1 consists of an MFH expression tag [12], followed by a purification His-tag, and the peptide of interest, Lmod2s1 (see Fig. 1). The gene can be synthesized and cloned into a commercial pET-21b vector between sites NdeI and XhoI at, e.g., GenScript Biotech, Piscataway, NJ. To allow separation of Lmod2s1 from the MFH tag by cyanogen bromide, which hydrolyzes peptide bonds at the C-terminus of methionine residues [25], an atg codon (Met) is inserted before the DNA sequence encoding Lmod2s1. Another available option to release a peptide of interest is to use enterokinase which is a serine protease cleaving peptide bonds at the C-terminus of lysine residues in the Asp-Asp-Asp-Asp-Lys sequence. If the presence of methionine in the peptide of interest precludes using cyanogen bromide for peptide release, a DNA sequence encoding the enterokinase recognition site should replace the atg codon as was done with αTM1a_1–14Zip, αTM1a_1–21Zip, and αTM1a_1–28Zip peptides [13] and human granulin peptides [26]. Otherwise, using cyanogen bromide is preferred to maximize yield and purity.

2. Assemble MIA immediately before inoculation. Alternatively, the medium can be frozen at −20 °C for up to 1 month until use.

3. For optimal culture aeration in MIA, the volume of the culture in a 2 L shake flask should be between 100 and 300 mL. Make sure that the flask is well-washed, thoroughly rinsed, and dried prior to use. Even a very small detergent impurity can considerably slow down the cell growth in a minimal medium.

4. To avoid carbamylation of primary amines in the protein, always prepare urea solutions immediately before use or freeze right after preparation and thaw immediately before use. Do not heat the solution to accelerate urea dissolution as this will accelerate conversion of urea into isocyanate.

5. If enterokinase is used to separate the peptide from the MFH tag, a protocol to produce large amounts of recombinant enterokinase (rEK) cost-effectively can be found in Skala et al. [27]. Other technical details on enterokinase use can be found in Tolkatchev [28].

6. A typical peptide binding capacity for a Sep-Pak C₁₈ cartridge is between 10 and 30 mg of peptide per gram of sorbent. To avoid peptide losses due to nonspecific binding, the amount of sorbent should not considerably exceed the required minimal sorbent amount.

7. Typical UV wavelengths for peptide detection are in the range of 214–280 nm.

8. Temperature conditions of the colony growth on a plate after the heat shock may be important for the level of expression of the recombinant fusion protein either in rich or minimal media. Growing colonies of BL21(DE3) transformed with pET21-MFH-Lmod2s1 at standard 37 °C instead of room temperature leads to a considerable drop in the expression level. Therefore, when transforming cells with a new construct it is advisable to test both temperature conditions for this step to optimize the expression level. It is also advisable to use freshly transformed cells for each protein production.

9. Collect small aliquots at each purification step for SDS-PAGE analysis. Use 15% SDS-PAGE gels for protein separation.

10. The doubling time of BL21(DE3) transformed with a pET-21b -based vector is typically 1–1.5 h in MIA at 37 °C.

11. The frozen cell pellet can be stored for up to several months at −20 °C.

12. Purification of His-tagged proteins in denaturing conditions (in the presence of 8 M urea) by Ni-NTA agarose is covered in detail in the QIAGEN “Handbook for high-level expression and purification of 6xHis-tagged proteins” which can be found on the company’s website www.qiagen.com. Therefore, the protocol provided in Subheading 3.1.3 is brief and focused on providing information specific to the Lmod2s1 peptide fused with MFH. Since the MFH tag protein is also unstructured [12], neither the fusion protein, nor Lmod2s1 need special care with respect to protein refolding after Ni-NTA purification in denaturing conditions.

13. The precipitate should be small compared to the initial size of the cell pellet and almost transparent. Large, non-transparent precipitate is a sign of poor cell lysis.

14. More wash steps with increasing imidazole concentration can be included to achieve a higher purity of the fusion protein. However, 90% purity is an acceptable purity level at this stage because at later stages of the protocol peptides undergo extensive reversed-phased HPLC purification where the impurities are effectively eliminated.

15. The most important purpose of dialysis is to remove imidazole, which interferes with cleavage procedures. The MFH tag is a highly soluble protein carrier and typically no precipitation of the fusion protein is observed in 1–20 mM HCl. However, if the fusion protein precipitates during the dialysis, after dialysis it can be brought back into solution by adding crystalline urea to 4–6 M urea. If enterokinase is to be used for cleavage, it remains active in urea concentrations of up to 1.0–1.2 M. Additionally, for enterokinase cleavage, the pH can be lowered to 6.5, which may also help with protein solubility depending on the fusion protein pI.

16. Cyanogen bromide (chemical formula CNBr) is highly toxic. Perform all operations under the well-maintained, unblocked operating fume hood. CNBr should be in 400 molar excess of the protein concentration if cleaved Met is not followed by a Ser, and in 1000 molar excess if Met is followed by a Ser. Do not exceed 0.25 g per 50 mL conical tube. If it does, use only a smaller portion of the recombinant protein to perform the reaction. Neutralize any CNBr residue on weighing boats or spatula with 5 M NaOH (leave to soak in the fume hood for 2 h).

17. The resolution of a Sep-Pak cartridge is comparatively poor, and it cannot be effectively used for separations as a typical chromatographic column having thousands of theoretical plates. However, it is sometimes possible to resolve the MFH tag from a peptide by using washes with increasing acetonitrile concentrations, e.g., in 10% increments. This allows skipping the step of removing MFH on Ni-NTA resin.

18. Depending on the pI of the peptide (which can be approximately predicted by Expasy ProtParam tool at https://web.expasy.org/protparam/), the second HPLC reversed-phase purification should use as a carrier solvent 1% ammonium bicarbonate if pI<7, or sodium phosphate, pH 11.2, if pI>7. Make sure to avoid direct switching between channels with pure acetonitrile and a buffer while priming/washing/equilibrating the column, as this could cause salt precipitation inside the HPLC system. Also, consider changing the organic mobile phase from pure acetonitrile to 70–85% acetonitrile in water.

19. Based on empirical observations, removing ammonium from the peptide sample to a degree that it does not interfere with NMR spectra takes time. One hour of washing with 0.1% TFA is sufficient if it is done on an HPLC column. It can also be done on a Sep-Pak column, if ~10× bed volumes of 0.1% TFA are used to desalt the peptide. Application of each 1× bed volume of 0.1% TFA should be followed by an approximately 5 min waiting time before the next 1× bed volume of 0.1% TFA is applied.

20. Peptides freeze-dried from 0.1% TFA/acetonitrile mixtures are acidic. A peptide’s solubility is typically minimal at pH ~ pI, where the peptide’s net charge is 0. Therefore, if the peptide is not well-soluble in acidic conditions created after adding to the freeze-dried peptide unbuffered water, it can be brought into solution by adding 0.5 mM disodium phosphate or careful titration with small volumes of 100 mM NaOH. Generally, it is best to first test the pH-dependent solubility of the isolated peptides and the complex they form using non-labeled peptide components.

21. 320 μL is a typical volume of an NMR sample before it is transferred into a 5 mm Shigemi NMR microtube. For different types of tubes this volume will vary.

22. Depending on the solubility and stability of the complex and solubility of its peptide components at different pH, other buffer alternatives may include deuterated acetate, deuterated HEPES, or deuterated Tris. Changing the NaCl concentration in the buffer is another variable that can help improve the solubility of the complex and the peptides.

23. Detailed description of standard NMR experiments and assignment of the NMR spectra used to determine 3D protein structure is not a part of this Chapter. A description of the 3D NMR experiments can be found in Sattler [29].

24. Detailed protocols on how to prepare oriented media can be found in Gebel [30].

25. For this part of the protocol, two peptides similar to αTM1a_1–14Zip were used, or more specifically Ac- αTM1a_1–14Zip and αTM1a_1–28Zip. The Ac-αTM1a_1–14Zip is acetylated at the N-terminus, and αTM1a_1–28Zip includes extra tropomyosin residues in its sequence [11, 13]. The choice of the modified structured components of the complex and methodology of their preparation is not discussed in this Chapter. Generally, modifications can be introduced chemically or by recombinant methods. They can include N- or C-terminal amino acid sequence extensions, site-directed substitutions via mutagenesis (such as Leu → Ile, Glu → Asp or Ser → Thr), modification of side chains via chemical methods, etc. The chemical nature of such a substitution should not destroy binding but simply alter 2D ¹⁵N-HSQC spectra of the IDP component therefore allowing mapping of the binding interface.

26. One may choose to repeat steps 1–8 of Subheading 3.3.1 for the assigned αTM1a_1–14Zip chemical shifts to confirm that the predicted dihedral angles agree with those in the solved NMR structure (PDB ID: 1TMZ). During the simulations, the backbone dihedral angles in αTM1a_1–14Zip may be restrained using predictions by TALOS+ [31]. An example of the dihedral angle predictions made for Lmod2s1 and αTM1a_1–14Zip by TALOS + are provided in supplementary S1 table in Tolkatchev et al. [11].

27. To change a dihedral angle, first, select the particular residue the conformation of which is to be changed. Then, type the desired values in the boxes. Do not press “Ok” or “Apply.” Instead, press “Set.” Proceed in the same manner until all of the dihedral angles have been changed, then press either “Ok” or “Apply.”

28. To do this in Chimera, both of the models, Lmod2s1 and αTM1a_1–14Zip, must be open. Deactivate one of them, then move the other into position. To deactivate a model, click through the following sequence:

    >Favorites

    >Model Panel

    In the row corresponding to the model you want to deactivate, uncheck the box in the column with the letter “A” at the top. As you arrange the peptides into the chosen topology, minimize the number of clashes. You can check for clashes by clicking through the following sequence:

    >Tools

    >Structure Analysis

    >Find Clashes/Contacts

    The less clashes your starting structure has, the easier it will be to energy minimize and equilibrate the system in subsequent steps.

    We chose the crisscross four helix bundle topology because, based on the NMR data in Subheading 3.2.3, the two-residue loop, Leu25 and Ser26, connecting the two α-helices of Lmod2s1, should be near the N-terminus of the αTM1a_1–14Zip dimer. However, this can also be achieved with a side-by-side four helix bundle topology (see Fig. 2b), where the Lmod2s1 α-α hairpin is on one side or the other of the αTM1a_1–14Zip dimer and the two-residue loop is oriented near the αTM1a_1–14Zip N-terminus. These topologies were found to be not stable nor do they agree with the NMR data [11].

    There is a wide variety of protein topologies [32, 33] and the correct topology for the system will depend on a number of factors. It is encouraged to try the topologies consistent with the data obtained in Subheading 3.2.3 and to choose one based on stability metrics.

29. The easiest way to do this is to open the pred.tab file in Excel and to do the operations of adding and subtracting the DPHI and DPSI from the PHI and PSI by writing a formula that can be dragged and filled. Follow the file format provided, placing the information in the appropriate columns, then save it as a plain text .txt file.

    The file should include only the amino acid residues for which the prediction is classified as “good.” You may choose to include angle restraints for only Lmod2s1 or for both Lmod2s1 and αTM1α_1–14Zip.

30. This command only uses the .pdb file to identify the residues and atoms in the system to match them up to the angles provided, which is why it is important to have the chains renumbered serially and the chain IDs consolidated in steps 15 and 16 of Subheading 3.3.1. The angle_restraints.rst file is what contributes to making the MDS NMR-guided. The angle restraints can be tightened or loosened by increasing or decreasing, respectively, the force constants rk2 and rk3, the default values of which are both 2.0 kcal/mol-rad². These values appear in the fifth line of the angle_restraints.rst. If you change them, be sure that rk2 = rk3 so that the harmonic restraint is symmetric.

31. Some of these commands are dependent on your system. For example, in the “addions” command, Na+ is used to neutralize the Lmod2s1/αTM1a_1–14Zip net negative charge. Cl− can be used instead to neutralize a net positive charge. If the peptides contain disulfide bonds, the names of the residues in the .pdb file must be changed from CYS to CYX and the bonds in LEaP must be explicitly specified using the “bond” command. If the peptides contain histidine residue(s), hydrogen atoms must be added/deleted to/from the imidazole ring(s) and the names of the histidine residues must be appropriately changed to achieve the desired histidine protonation state. In this particular case, there is a neutrally charged histidine residue in both chains of 1TMZ with the δ-nitrogen atom being protonated. To retain that protonation state, the residue name in the .pdb file should be changed from HIS to HID. Alternatively, the protons on the δ-nitrogens could be deleted. In this case, LEaP will automatically change residue names from HIS to HIE and add hydrogen atoms to the ε-nitrogen atoms. Hydrogens should be added beforehand if the .pdb file will be used to create NH residual dipolar coupling restraints. Therefore, every .pdb file may require a slightly different preparation/editing. It is encouraged to consult the Amber manual for help in steps that may be specific to your system.

32. The options “nmropt = 1,” and “DISANG = angle_restraints. rst,” are those that add restraints on the dihedral angles according to the NMR data, making this protocol an NMR-guided MDS.

33. It is important to start a simulation run from an energy minimized structure to ensure that there are no bad contacts/overlaps that may lead to large energies and instability in the simulation. Many systems may require more extensive equilibration protocols. The input .ncrst file to this simulation is the output .ncrst file from the energy minimization in step 8 of Subheading 3.3.2, not the output .ncrst from step 6 of Subheading 3.3.2.

34. The input files in the lines:

    –i min.in \
    -p abc_init.top \
    

    are the same as in step 8 of Subheading 3.3.2. However, the input in the line starting with “-c” changes depending on the structure being minimized. In step 8 it was the initial structure, represented by the abc_init.ncrst file. In step 11 of Subheading 3.3.2 it is instead the structure after being subjected to NVT MDS, represented by the abc_md_NVT.ncrst file, which was the output file specified in the line starting with “-r” in step 10 of Subheading 3.3.2. The file names can be changed provided they do not conflict with the flow of files through the protocol.

35. This input file is set up to start at 300 K, heat up to 360 K and cool back down to 300 K in 10 K increments, equilibrating for 10 ns at each temperature. These conditions were chosen after testing that showed the complex dissociated completely above 360 K and that 10 ns was sufficiently long to observe α_L-β and β-α_L transitions of Leu25 in Lmod2s1 [11]. Since the increase in temperature increases velocities, the sudden temperature increase can occasionally create a local instability. The vlimit option (e.g., “vlimit = 10”) limits atom velocities and can help to prevent this.

36. 36. The input file in the line:

    -p abc_init.top \
    

    is the same as in step10 of Subheading 3.3.2. The inputs in the lines starting with “-i” and “-c” depend on the type of simulation and the state of the structure at the beginning of the simulation. In step 10 the “ -i” input was the md_NVT.in file generated in step 9 of Subheading 3.3.2 and the “-c” input was the abc_min.ncrst file, which was the output specified in the line starting with “ -r” in step 8 of Subheading 3.3.2. In step 2 of Subheading 3.3.3, the “-i” input was the annealing.in file generated in step 1 of Subheading 3.3.3 and the “ -c” input was the abc_md_NVT_min.ncrst file, which was the output specified in the line starting with “ -r” in step 11 of Subheading 3.3.2. The file names can be changed provided they do not conflict with the flow of files through the protocol.

37. Before moving on to NPT production runs, repeat steps 2 and 3 of Subheading 3.3.3 several times to generate an ensemble of structures and to choose the one that agrees best with your NMR data to continue with. Name files for replicas uniquely to avoid overwriting previous files associated with the first replica.

38. Before moving on to RDC refinement, repeat steps 2 and 3 of Subheading 3.3.4 several times to generate an ensemble of structures and to choose the one that agrees best with your NMR data to continue with. Name files for replicas uniquely to avoid overwriting previous files associated with the first replica.

    Alternatively, you can use RDC values to validate your structure(s) from Subheading 3.3.4 by comparing experimental RDCs to the RDCs back-calculated from obtained structure (s) by a program such as REDCAT [34]. RDC values used for validation cannot be used for refinement and vice versa. NMRbox provides remote access to computers with REDCAT installed, available at the following link: https://nmrbox.org/

39. The “ndip” variable is the number of RDCs in your RST.dip file. “dwt” is the relative weight of each RDC. The current value assigns a weight of 0.1 to each RDC. You may choose to weight individual RDCs differently though. “gigj” is the product of the nuclear “g” factors, which is related to the gyromagnetic ratios of nuclei involved in the RDC. In this case, our RDC is for ¹⁵N-¹H, so the “gigj” is the product of their “g” factors. Typical values are ¹H = 5.5856, ¹³C = 1.4048, ¹⁵N =−0.5663, ³¹P = 2.2632. All 26 RDCs are associated with this value in our restraint file, but you may include other RDCs and assign them appropriate values. “s11, s22, s12, s13, s23” are components of the alignment tensor. Here they are set to zero so we can first have Amber fit the structure to an alignment tensor. To do so, it is also necessary that “freeze-mol = .true.” so that the structure is not allowed to change during the fitting.

40. The first line of the .in input files for the RDC refinement must either be blank or have an arbitrary title, because the simulation code (sander rather than pmemd) that is used for RDC refinement ignores the first line. If &cntrl is in the first line, as it is in other .in input files, you will get an error “cannot find cntrl namelist.”

41. Notice that in the first line, the command is mpiexec -np 8 $AMBERHOME/bin/sander.MPI rather than the usual $AMBERHOME/bin/pmemd.cuda. This is because RDC refinement is not supported by the pmemd code and it must be run on sander, which will use your CPU cores rather than your GPU. In this example, it is set to use 8 CPU cores with “-np 8”

42. At the end of the abc_md_NPT_min_fit.info file, you should see the fitted alignment tensor:

    Alignment tensor: −10.622 10.012 13.937
    10.012 11.014 13.737
    13.937 13.737 −0.392
    

    The values in the tensor are:

    Alignment tensor: s11 s12 s13 
    s21 s22 s23 
    s31 s32 s33
    

43. The file after the “parm” command is the.top file from step 6 of Subheading 3.3.2. The file after the “trajin” command is the -r output from step 16 of Subheading 3.3.5. If you keep the same file names as in the protocol, you can type the commands exactly as is. If you use your own file names, you will have to change them. The “strip :WAT, Na+” command will remove all solvent molecules from the system. If you used Cl− in step6 of Subheading 3.3.2 to neutralize a net positive charge in your system, replace “Na+” with “Cl−“ in the “strip” command. The “autoimage” command will write all of the peptide chains to the primary simulation box, which is important to visualize structures of multiple chains in a periodic simulation where coordinates of individual chains may have been written to an adjacent periodic box. You may use this list of commands, replacing the file after the “trajin” command with the appropriate.ncrst to generate a structure of the complex at any stage in the protocol if you would like to inspect it visually.