Using Optical Tweezers to Monitor Allosteric Signals Through Changes in Folding Energy Landscapes

Abstract

Signaling proteins are composed of conserved protein interaction domains that serve as allosteric regulatory elements of enzymatic or binding activities. The ubiquitous, structurally conserved cyclic nucleotide binding (CNB) domain is found covalently linked to protein with diverse folds that perform multiple biological functions. Given that the structures of cAMP-bound CNB domains in different proteins are very similar, it remains a challenge to determine how this domain allosterically regulates such diverse protein folds. Instead of a structural perspective, we focus our attention on the energy landscapes underlying the CNB domains and their responses to cAMP binding. We show that optical tweezers is an ideal tool to investigate how cAMP binding coupled to interdomain interactions remodel the energy landscape of the regulatory subunit of protein kinase A (PKA), which harbors two CNB domains. We mechanically manipulate and probe the unfolding and refolding behavior of the CNB domains as isolated structures or selectively as part of the PKA regulatory subunit, and in the presence and absence of cAMP. This approach allows us to dissect the changes in the energy landscape associated with cAMP binding, and to examine the allosteric interdomain interactions triggered by the cyclic nucleotide. This approach can be used to study other modular, multidomain signaling proteins found in nature.

1. Introduction

Signaling proteins are modular in nature. They are made up of conserved protein interaction domains, which serve as allosteric regulatory elements of many signaling networks (1). While the same interaction domain can be involved in a variety of different signaling pathways, the surface contacts that it mediates with ligands or other proteins are highly specific (2). The conserved residues of interaction domains are embedded in variable sequences, and these variable features increase the affinity of a protein for a signaling partner. Examples of protein interaction domains include the Src homology 2 (SH2) and the Src homology 3 (SH3) domains (3, 4). The SH2 domain recognizes phosphorylated tyrosine residues. The specificity of SH2 domains for a given signaling partner does not depend on the phosphorylated tyrosine alone but also on the amino acids surrounding it (1, 3). The SH3 domain recognizes proline residues on one of the three ‘spines’ of a polyproline type II helix. While proline residues are invariant and required for interaction, the rest of the residues that make up the helix are variable (1, 4). Binding selectivity is also enhanced by increasing the number of interaction sites between two peptides, by subcellular compartmentalization, and by cooperative assembly of large multiprotein complexes.

Cyclic nucleotide binding (CNB) domains are another example of structurally conserved and ubiquitous regulatory signaling modules in proteins (5, 6). CNB domains are not only found in all organisms, from bacteria, to archaea, to eukaryotes, but these domains are also found covalently linked to diverse protein folds that accomplish different biological functions, including gated channels, nucleotide exchange factors, transcription factors, and protein kinases (6) (Fig. 1a).

Figure 1.

(a) Cyclic nucleotide binding (CNB) domain in protein structures of diverse functionalities. Center: core structure of the CNB domain from the PKA regulatory subunit (isoform Riα) highlighting conserved structural motifs. Color-coding is consistent throughout the figure. The phosphate binding cassette (PBC) shown in red is bound to one cAMP (teal). The β-barrel is colored green and α-helices (N3A, α-B, and α-C) are colored violet. Top: PKA regulatory subunit RIα isoform (PDB: 1RGS), with two CNB domains: A (on the left) and B (on the right). Left: cyclic nucleotide-gated channel (HCN) (PDB: 1Q5O), with four subunits of 4-fold symmetry. Right: cAMP receptor protein (CRP) (PDB: 4A2U) with two subunits of 2-fold symmetry. Bottom: nucleotide exchange factor (EPACII) (PDB: 4MGI), with two subunits of 2-fold symmetry. For EPACII, there is an additional, lower affinity cAMP-binding site in the vicinity of the CNB domain. The second cAMP bound at such sites is shown in blue. All other secondary structures in this figure that are not CNB domains are colored in grey. Structures were generated using PyMOL. (b) Structure of cAMP.

The consensus structure of the CNB domain consists of a rigid 8-stranded β-barrel subdomain and a highly flexible α-helical subdomain which undergoes conformational changes upon ligand binding (Fig. 1a) (5). The phosphate binding cassette (PBC) motif, which contributes to the functional specificity of CNB domain protein families, is found at the bottom of the barrel. This motif consists of β-strand 6, a short turn of a helix, and β-strand 7. The canonical PBC motif is found in CNB domain protein families that bind cyclic nucleotides like cAMP, while the non-canonical PBC motif is found in CNB domain proteins that bind ligands other than cyclic nucleotides. The key residue for canonical PBC motifs is a buried arginine that coordinates with the exocyclic phosphate group of cAMP (Fig. 1b). When this residue is missing, the binding of a different small-molecule ligand is facilitated, such as chlorophenolacetic acid in the case of transcription factor CprK and heme in the case of the CO sensing transcription activator cooA (5). Sequence variation contributes to ligand specificity and gives rise to non-canonical PBC motifs (6). In addition to the small helix in the PBC, the helical subdomain also includes a helix N-terminal to the β-barrel called the N3A helix, and a helical motif C-terminal to the β-barrel called the B/C helix.

Upon signaling through cAMP, the canonical PBC motif binds the exocyclic phosphate and ribose oxygens of cAMP (Fig. 1b) through specific hydrogen bonds, protected by hydrophobic interactions. The adenine ring of cAMP interacts with the helical regions (B/C helix) of CNB domains through specific hydrophobic sandwich interactions. The conformation of the adenine lid plays a crucial role in the allosteric regulation of proteins with CNB domains, because the constraint of the B/C helix to the adenine ring of cAMP initiates a cascade of structural changes within CNB domain proteins that are consistent with their specific functions (6). Given that the structures of cAMP-bound CNB domains in different proteins are very similar, it remains a challenge to determine how this structurally conserved interaction domain is able to regulate such diverse protein folds.

Protein Kinase A (PKA) is a model system for investigating the coupling mechanism between cAMP binding and the allosteric activation of this enzyme. The PKA holoenzyme is a tetramer consisting of two catalytic subunits and two regulatory subunits, both with several different isoforms (7, 8). Each regulatory subunit harbors two CNB domains, termed CNB-A and CNB-B. When the regulatory subunits are bound to the catalytic subunits, PKA is inactive. However, upon binding cAMP, the regulatory subunits experience a large conformational change that releases active catalytic subunits (9–11).

The static structural snapshots of the regulatory subunit in its apo (without cAMP) and cAMP-bound forms are very revealing, but do not provide information about how the signals triggered by cAMP binding are allosterically transduced to the various domains in PKA. Proteins are dynamic in nature, and thus a more complete view of signaling relies on the plasticity of protein molecules: for a given protein there are alternative residue packing networks with coupled dynamic motions, and a protein populates and fluctuates between an ensemble of many conformational states (12–14). In order to determine the functional states of the PKA regulatory subunit, we need to interrogate the energy landscape of PKA and dissect how cAMP binding remodels this landscape and shifts the population distribution of conformational states, causing information to be transmitted downstream. Instead of protein-specific differences in domain structure, the underlying energy landscapes and their responses to cAMP binding may be unique for specific proteins. Optical tweezers provide a tool for studying the allostery of the PKA regulatory subunit in this dynamic view (Fig. 2).

Figure 2.

(a) Schematic representation of the optical tweezers setup. The target proteins are covalently attached to dsDNA handles at two pulling positions through disulfide bonds. The other 5’ end of the dsDNA handles has either biotin or digoxigenin (not shown), which interact with polystyrene beads coated with streptavidin (SA) or antibodies against digoxigenin (AD), respectively. (b) dsDNA handle attachment positions on PKA RIα are indicated by black arrows, where 110, 243, and 376 refer to the amino acid residue numbers in the full-length protein. The isolated CNB-A (light green) and CNB-B (dark green) domain structures were overlayed using PyMOL. (c) CNB domain constructs used in these experiments, where black arrows indicate the domain to unfold: Isolated CNB-A and CNB-B domains, and selective unfolding of CNB-A and CNB-B domains in the presence of the neighboring CNB domain.

In this chapter, we describe how to attach dsDNA handles to the CNB-A and CNB-B domains of PKA (15–17) using disulfide bonds (16, 18) (Fig. 2a). While the attachment of dsDNA handles via thiol groups has been widely used and is well-established (18–23), we follow this procedure with a selection step for functional and properly folded protein molecules (Fig. 3a). This step increases the reproducibility, homogeneity and throughput of data collection (16). By mechanically manipulating the protein with optical tweezers, we probe the unfolding and refolding behavior of the CNB domains as isolated structures, or selectively as part of the PKA regulatory subunit (Fig. 2b–c). These experiments are conducted in the presence and absence of cAMP (16, 17). Comparing the unfolding and refolding behavior of the CNB domains as isolated structures and as part of the PKA regulatory subunit allows us to dissect the changes in the energy landscape associated with cAMP binding, and to examine the allosteric interdomain interactions triggered by the cyclic nucleotide. Results from this integrated approach show that despite being almost structurally identical (9), each CNB domain carries unique but coordinated roles to allosterically activate PKA (17). This approach can be used for other modular, multidomain signaling proteins found in nature (1).

Figure 3.

(a) Covalent attachment of dsDNA handles to PKA-R constructs. In Step 1, purified proteins are covalently attached to dsOligos. In Step 2, the chimeric protein-dsOligos are functionally selected by using a cAMP-coupled agarose and eluted with increasing concentration of cAMP. In Step 3, functional chimeric protein-dsOligos are ligated with biotin- and digoxigenin-modified dsDNA. (b) Native polyacrylamide gel tracking attachment of dsOligos to the protein (Step 1) and elution after binding from the cAMP-coupled agarose (Step 2). (c) The ligation of chimeric protein-dsOligos with dsDNA handles with 5’ biotin (purple square) or digoxigenin (orange circle) (Step 3). DNA ladder for reference only since chimeric protein-dsOligos do not run as pure DNA. We observed the formation of chimeric protein-dsOligos with one dsDNA handle but those samples do not form tethers in the optical tweezers assay.

2. Materials

2.1. Transformation of E. coli and expression of PKA regulatory subunit (PKA-R) constructs

1. 25 g/L LB medium.

2. 50 mg/mL ampicillin.

3. LB-ampicillin agar plates.

4. BL21(DE3) pLysS competent cells for protein expression (New England Biolabs).

5. 1 M IPTG (Isopropyl β-D-1-thiogalactopyranoside).

6. Plasmid DNA: the PRSET plasmid harboring the Bos taurus regulatory subunit gene (ampicillin-resistant) is a gift from Susan S. Taylor’s lab at the University of California, San Diego (see Note 1).

2.2. cAMP-coupled agarose resin preparation

1. N-hydroxysuccinimide (NHS)-activated Sepharose 4 Fast Flow resin (GE healthcare Life Science).

2. 0.1 M HCl solution.

3. 100 μmol of 8-(2-aminoethyl)-amino-cAMP (8-AEA cAMP) (Biolog at https://www.biolog.de/8-aea-camp).

4. Incubation buffer: 0.2 M NaHCO₃, 0.5 M NaCl, pH 7.5.

5. 0.5 M ethanolamine, pH 8.3.

6. High pH buffer (pH 8.3): 0.1 M Tris-HCl.

7. Low pH buffer (pH 4.8): 0.1 M CH₃COONa, 0.5 M NaCl.

8. Storage buffer: 0.05 M H₃BO₃, 0.5 M NaCl, 10 μM IBMX, pH 8.5, with 20% ethanol.

2.3. cAMP-coupled agarose resin-based protein purification of PKA-R constructs

1. DTT (dithiothreitol).

2. Lysis buffer (pH 6.5): 20 mM MES, 100 mM NaCl, 2 mM EGTA, 2 mM EDTA, 5 mM DTT.

3. Protease inhibitor cocktail stock (stored at −20 °C): 0.313 g/mL Benzamidine in QH₂O, 48 mg/mL AEBSF in QH₂O, 1 mg/mL Pepstatin in methanol, 1 mg/mL Leupeptin in ethanol, TPCK & TLCK (3 mg each per mL in ethanol), 2.22 mg/mL IBMX in ethanol.

4. Ammonium sulfate (45% saturation).

5. 50 mL low salt wash buffer: same as lysis buffer.

6. 50 mL high salt wash buffer: lysis buffer with 0.5 M NaCl.

7. 12 mL elution buffer: lysis buffer with 18 mM cGMP.

8. Storage buffer: 50 mM MES, 200 mM NaCl, 2 mM EGTA, 2 mM EDTA, 5 mM DTT, pH 5.8.

9. Protein concentration: Pall centrifugal devices (10 kDa MWCO).

10. 60% glycerol (molecular biology grade), autoclaved.

2.4. Preparation of thiol-modified double-stranded oligonucleotide (dsOligo) for the covalent attachment of dsDNA handles to PKA-R constructs

1. Oligonucleotide sequences: (see Note 2).

   CS1/CS2 forward (FWD): 5’-Thiol-GTTACGCCTATTCCTATCATATGAAGACAC

   Reverse (REV) complimentary oligonucleotides with non-palindromic 5’ overhangs:

   CS1 (REV): 5’-phosphate- CGGAGTGTCTTCATATGATAGGAATAGGCGTAAC

   CS2 (REV): 5’-phosphate- CGACGTGTCTTCATATGATAGGAATAGGCGTAAC

2. Annealing buffer: 10 mM Tris, 10 mM EDTA, 50 mM NaCl, pH 7.4.

3. Deprotection buffer: 0.34 M sodium phosphate, pH 8.0.

4. 1 M DTT.

5. Purification column: Amersham PD-10 column.

6. Molecular biology grade ethanol.

7. 3 M sodium acetate, pH 5.2.

2.5. DTDP (2,2’-dithiodipyridine) activation of cysteine-modified PKA-R constructs and covalent attachment of dsOligos

1. DTDP (Millipore Sigma).

2. DNA binding buffer: 50 mM Tris, 100 mM NaCl, pH 7.6.

3. Micro Bio-Spin^™ P-6 gel columns (Tris buffer, Cat. No. 7326221, BioRad).

2.6. Functional selection of the chimeric protein-dsOligo constructs (see Note 3)

1. Low salt wash buffer: DNA binding buffer (50 mM Tris, 100 mM NaCl, pH 7.6).

2. High salt wash buffer: DNA binding buffer with 0.5 M NaCl.

3. Elution buffers: DNA binding buffer with varying concentrations of cAMP: 0.02 mM, 0.2 mM, 2 mM and 20 mM.

4. cAMP-coupled agarose resin, resin volume: 50 μL.

5. 60% glycerol (molecular biology grade), autoclaved.

2.7. Preparation of dsDNA handles (see Note 4)

1. pPROEX-Hta plasmid (see Note 5).

2. Biotin handle (332 bp): (see Note 6).

   Forward primer (FWR): 5’-TATTATTTTCTCCCATGAAGACGGTCCGCGACTG

   Reverse primer (REV): 5’-biotin-CGGTATCGTCGTATCCCACTACC

3. Digoxigenin (Dig) handle (315 bp): (see Note 7).

   FWR: 5’-GACGATACCGAAGACAGGTCGTGTTATATCC

   REV: 5’-digoxigenin-CCGTGCAGTCGATGATAAGCTGTC

4. 10 mM dNTP mixture (New England BioLabs).

5. 5 U/μL Taq polymerase.

6. Standard 10x Taq buffer.

7. 25 mM MgCl₂.

8. NucleoSpin Gel and PCR Clean-up kits (Clontech Laboratories).

9. BbsI restriction enzyme (Thermo Fisher Scientific) (see Note 8).

10. Buffer G (Thermo Fisher Scientific).

11. 0.8% w/v agarose gel.

2.8. Ligation of chimeric protein-dsOligo constructs with the dsDNA handles

1. 5 ng/μL biotin handle.

2. 5 ng/μL Dig handle.

3. Protein-dsOligo chimera.

4. T4 DNA ligase.

5. Ultra-pure water: Nuclease-free, molecular grade H₂O.

6. 10x ligation buffer: 400 mM Tris-HCl, 100 mM MgCl₂, 5 mM ATP, pH 7.8 (see Note 9).

2.9. Optical Tweezers

1. Our instrument is an opposed-beam single trap optical tweezers called miniTwezeers (24) (Fig. 4a). The instrument was assembled and calibrated by Steven Smith from Steven B. Smith Engineering. Assembly instructions and all parts of the optical tweezers can be purchased according to the list from the TweezersLab website (http://tweezerslab.unipr.it/cgi-bin/mt/home.pl).

2. Microfluidic chambers are purchased from Steven B. Smith Engineering. A chamber consists of three channels: top, center, and bottom. The top and bottom channels communicate with the center channel via dispenser tubes (Fig. 4b–c). A glass micropipette tip is added into the middle of the center channel. The chamber is held in a movable stage in X, Y and Z between two objective lenses (Fig. 4a).

3. Syringes:

   Becton Dickinson (BD) 10-mL syringe eccentric tip (Cat. No. 305482).

   Becton Dickinson (BD) medical 1-mL syringe slip tip (Cat. No. 309659).

4. Tubing:

   Becton Dickinson Intramedic polyethylene tubing PE10 (Cat. No. 427401).

   Becton Dickinson Intramedic polyethylene tubing PE50 (Cat. No. 427411).

5. Beads:

   Streptavidin coated 2.1-μm diameter polystyrene beads (SA bead) (Spherotech, catalog no. SVP-20-5, 0.5% w/v, 5 mL).

   Anti-digoxigenin coated 2.1-μm diameter polystyrene beads (AD bead) (Spherotech, catalog no. DIGP-20-2, 0.1% w/v, 2 mL) (see Note 10).

6. DNA binding buffer: 50 mM Tris, 100 mM NaCl, pH 7.6.

7. Storage buffer: 0.1% w/v Sodium Azide.

Figure 4.

(a) Photo of miniTweezers (24). Zoomed-in is the microfluidic chamber holder between the two objective lenses. (b) Microfluidic chamber. Top is a photo of the chamber with a ruler as size reference. Bottom is a cartoon representation of the chamber. The central channel (2) is connected to the top (1) and bottom (3) channels with dispenser tubes. The center channel contains a micropipette for bead suction. (c) bead coated with anti-digoxigenin has a ~3 μm diameter and it is trapped by the two counterpropagating lasers (not seen in photo). Streptavidin-coated bead with a diameter of ~2 μm is stabilized by suction on the tip of the micropipette.

3. Methods

3.1. Transformation of E. coli and expression of the PKA regulatory subunit (PKA-R) constructs (Fig. 2c)

1. LB medium: suspend LB Miller broth (Millipore Sigma) in 18 Milli-Q H₂O (QH₂O) to a concentration of 25 g/L. Autoclave and store at room temperature.

2. Ampicillin: dissolve ampicillin in QH₂O to a concentration of 50 mg/mL and store at −20 °C. The working concentration of ampicillin is 50 μg/mL.

3. LB-ampicillin agar plates: suspend LB broth with agar in QH₂O to a concentration of 40 g/L. Autoclave the solution and when the solution cools down to 50 °C, add ampicillin. Mix well and pour into the plates. After they solidify, store the plates at 4 °C for up to 2 weeks.

4. IPTG: dissolve IPTG in QH₂O to a concentration of 1 M, filter, then store at −20 °C. The working concentration of IPTG is 1 mM.

5. Two ampicillin agar plates are needed per transformation. Prepare plates, and label one with the construct name and the other one as negative control.

6. Thaw one tube (100 μL) of BL21(DE3) pLysS cells on ice for 5 minutes. Once cells are thawed, transfer 50 μL to a different tube and label the tube as negative control and the original tube with the construct name.

7. Add 20 ng of DNA plasmid harboring the protein of interest (different PKA-R constructs shown in Fig. 2c) to the tube with corresponding construct name and none to the negative control.

8. Incubate both tubes on ice for 30 minutes.

9. Heat shock both tubes at 42 °C in a water bath for 10 seconds.

10. Incubate both tubes on ice for 5 minutes.

11. Add 950 μL of pre-warmed LB medium to both tubes and incubate at 37 °C in a water bath for at least one hour.

12. After incubation, take both tubes out of the water bath, centrifuge them at 11363 rcf for 1 minute, then discard 900 μL of the supernatants.

13. Resuspend the cells in the remaining 100 μL of the supernatant left in the tubes after centrifugation. Plate the cells onto the corresponding construct name and negative control agar plates.

14. Incubate plates in the incubator at 37 °C overnight (see Note 11).

15. Pick a single colony from the construct plate and inoculate in 100 mL of LB-ampicillin medium (100 μL of ampicillin). Repeat for a total of three colonies. Incubate at 185 rpm at 37 °C overnight.

16. Measure the optical density of the cells at 600 nm (OD₆₀₀). Transfer each starter cell culture to 1200 mL LB medium (total of 3) with ampicillin (1.2 mL ampicillin) to an initial OD₆₀₀ of 0.2.

17. Incubate three Fernbach flasks at 125 rpm at 37 °C and grow cells until their OD₆₀₀ values reach 0.6.

18. After the OD₆₀₀ values reached around 0.6, induce cells with 1.2 mL of 1 M IPTG (IPTG final concentration 1 mM) and incubate overnight (for about 16-18 hours) at 125 rpm at 18 °C.

19. Harvest bacteria at 5180 rcf for 20 minutes at 4 °C.

20. Collect the bacterial pellets into 50-mL falcon tubes and note wet cell pellet masses. Store tubes at −80 °C until use. Expected wet cell mass per liter of culture to be around 1 g.

3.2. cAMP-coupled agarose resin preparation

1. Equilibrate 5 mL of NHS-activated agarose 4 Fast Flow resin with 0.1 M HCl.

2. Incubate 8-AEA-cAMP with the activated agarose resin in the incubation buffer at room temperature for 3 hours.

3. Block the reaction with 0.5 M ethanolamine (pH 8.3) for 2 hours.

4. Wash the resin three times by alternating between the high pH and low pH buffers.

5. Store the resin in the storage buffer at 4 °C.

3.3. cAMP-coupled agarose resin-based protein purification of PKA-R constructs

1. All solutions and steps during purification should be conducted at low temperatures or in a cold room at 4 °C.

2. Dissolve DTT in QH₂O to a concentration of 1 M, filter, then store at −20 °C.

3. Resuspend and homogenize the bacterial pellets in lysis buffer supplemented with the protease inhibitor cocktail (10 mM Benzamidine, 0.4 mM AEBSF, 1 μM Pepstatin, 1 μM Leupeptin, 28 μM TPCK & TLCK and 10 μM IBMX) and 5 mM DTT. Use 10 mL buffer per gram of pellet.

4. Break the homogenized cells at 10000 psi. Repeat three times to ensure all cells are broken and the solution is homogeneous (see Note 12).

5. Centrifuge the cells at 27216 rcf for one hour at 4 °C.

6. Precipitate the spun-down supernatant from the centrifugation with 45% ammonium sulfate for one hour at 4 °C with a stir bar.

7. Centrifuge the mixture at 12096 rcf for 10 minutes and discard the supernatant.

8. Resuspend the pellet in lysis buffer (maximum 30 mL) by gently mixing with a plastic pipette, and centrifuge at 12096 rcf for 10 minutes.

9. Batch bind the supernatant with 5 mL of the cAMP-coupled agarose resin on a rocker at minimum speed, overnight at 4 °C.

10. Wash the resin three times, with 50 mL of buffer for each wash. Use low salt buffer for the first wash, followed by high salt buffer, and lastly use the low salt buffer again.

11. Elute the protein from the resin three times. The first two times, 6 mL elution buffer is added to the resin and the mixture is incubated for 30 minutes with very gently stirring every 10 minutes. For the last elution, use 6 mL of the lysis buffer.

12. Concentrate the eluted protein to 5-10 mL using 10 kDa Pall Macrosep Advance Centrifugal device (see Note 13) and load onto a size exclusion chromatography column pre-equilibrated with PKA-R storage buffer. The proteins in storage buffer are ready for DNA handle attachment. The rest of the protein that is not used in the DNA handle attachment process can be stored with 30% glycerol at −80 °C.

3.4. Preparation of thiol-modified dsOligo (CS1 and CS2) for DNA handle attachment to PKA-R constructs

1. Dissolve the forward and reverse oligonucleotides in annealing buffer to ~500 μM each.

2. Set up 50-μL annealing reactions in a thermocycler using identical molar ratios of forward and reverse oligonucleotides. Annealing settings: 95 °C to 4 °C, in 1 °C/minute increments. CS1 7 reactions (350 μL in total) and CS2 7 reactions (350 μL in total).

3. Deprotect the thiol groups of the dsOligos:

   CS1: combine 350 μL of anneal product, 350 μL of the deprotection buffer and 28 μL of 1 M DTT (final concentration of DTT is 40 mM).

   CS2: combine 350 μL of anneal product, 350 μL of the deprotection buffer and 28 μL of 1 M DTT.

   Deprotect at 37 °C overnight.

4. Purify the deprotected dsOligos with the Amersham PD-10 desalting column using the gravity protocol. For each SH-dsOligo, 7 mL of final elution product will be obtained. Measure their concentrations using UV-absorbance (see Note 14).

5. Concentrate the purified dsOligos by ethanol precipitation at −20 °C overnight:

   for each dsOligo, divide the total 7 mL into six 5-mL tubes. Each 5-mL tube should contain: 1.16 mL of dsOligo, 116 μL sodium acetate and 3.83 mL of molecular biology grade ethanol.

6. Dissolve the concentrated dsOligos in 30-50 μL of annealing buffer (expected final concentration around 50000 ng/μL) and store them at −20 °C with 40 mM DTT.

3.5. DTDP activation of cysteine-modified PKA-R constructs and formation of the chimeric protein-dsOligo

1. Dilute both dsOligos (CS1 and CS2) to 2000 ng/μL in DNA binding buffer with 10 mM DTT and incubate overnight at 37 °C. The next day, take both CS1 and CS2 out of 37 °C and store them at 4 °C until PKA-R constructs are ready for covalent attachment of dsOligos.

2. Concentrate the purified PKA-R protein constructs using Pall 10 kDa MWCO centrifugal device to 5 mg/mL in 75 μL (see Note 13).

3. Remove excess DTT from the protein by three consecutives Micro Bio-Spin^™ P-6 gel columns pre-equilibrated with DNA binding buffer.

4. Dissolve DTDP in acetonitrile to 100 mM. The working concentration is 10 mM.

5. React protein with 10 μL of 100 mM DTDP and incubate at room temperature for 2 hours.

6. Remove excess DTDP by three consecutives Micro Bio-Spin^™ P-6 gel columns pre-equilibrated with DNA binding buffer.

7. Remove excess DTT from CS1 and CS2 by three consecutives Micro Bio-Spin^™ P-6 gel columns pre-equilibrated with DNA binding buffer.

8. Immediately after removing the excess DTT, react both CS1 and CS2 with the DTDP-activated protein in 1:1:1 molar ratio and incubate the reaction at 4 °C overnight. The mixture of the protein with the dsOligos, resulting in the chimeric protein-dsOligo, is called the handle-protein-handle (HPH) construct (Fig. 3a).

3.6. Functional selection of the HPH (chimeric protein-dsOligo) constructs

1. Incubate the HPH construct with the cAMP-coupled agarose resin overnight at 4 °C.

2. Remove the excess unreacted dsOligos by washing the resin with low salt wash buffer twice and high salt wash buffer once, using at least 20 times the volume of the resin in each round of washing.

3. Elute the HPH construct using 50 μL of increasing concentrations of cAMP elution buffers (0.02 mM to 20 mM) and incubating the elution buffers with the resin for 30 minutes before each elution (Fig. 3b).

4. Store the HPH constructs eluted with different concentrations of cAMP separately. Combine each 50 μL solution of HPH with 50 μL of 60% glycerol and distribute into 10 μL per tube. Store at −80 °C.

3.7. Preparation of dsDNA for ligation of handles to HPH

1. Resuspend all primers in ultra-pure water to 10 μM stocks.

2. Set up PCR reactions using the plasmid pPROEX-Hta as the template and corresponding forward and reverse primers for biotin and digoxigenin, 16 reactions each. Each reaction is 50 μL in total: ultra-pure water (36 μL), 10x buffer (5 μL), plasmid (1 μL), forward primer (1 μL), reverse primer (1 μL), dNTP (1.5 μL), Taq polymerase (0.5 μL), MgCl₂ (4 μL).

3. PCR cycle for biotin: 1. 95 °C (2 minutes), 2. 95 °C (30 s), 3. 61.5 °C (30 s), 4. 72 °C (1 minutes) 5. Repeat steps (2—4) 30 times, 6. 72 °C (10 minutes), 7. 4 °C (as long as needed).

4. PRC cycle for digoxigenin: 1. 95 °C (2 minutes), 2. 95 °C (30 s), 3. 60.3 °C (30 s), 4. 72 °C (1 minutes), 5. Repeat steps (2—4) 35 times, 6. 72 °C (10 minutes), 7. 4 °C (as long as needed).

5. Clean up PCR products using NucleoSpin Gel and PCR Clean-up kits. Combine all biotin PCR tubes into one column and elute the sample with 15 μL elution buffer. Combine all digoxigenin PCR tubes into a separate tube and elute the sample with 15 μL elution buffer. The concentrations for both should be around 500 ng/μL.

6. Digest PCR clean up products with BbsI at 37 °C overnight.

   Each BbsI reaction is 20 μL in total: BbsI enzyme (2 μL), commercial Buffer G (2 μL), DNA (1000 ng), and add ultra-pure water to make the total volume 20 μL.

7. After incubation, heat inactivate the enzyme at 65 °C for 20 minutes.

8. Load 40 μL of PCR product onto the agarose gel per well. Run at 100 V for 1 to 2 hours, then gel extract dsDNA handles using a PCR gel extraction clean-up kit.

3.8. Ligation of chimeric protein-dsOligo constructs with the dsDNA handles

1. For a 20 μL reaction: combine 100 ng of the biotin and 100 ng of the dig handle with 10 μL of the HPH construct, 2 μL of homemade ligation buffer, 1 μL of T4 ligase and ultra-pure water to a total volume of 20 μL.

2. Ligate overnight at 4 °C.

3. Run a ligation gel to check if the ligation is successful (Fig. 3c). Heat inactivate 10 μL of the sample at 55 °C for 10 minutes. Load onto 5% Acrylamide gel and run at 110 V for 40-60 minutes at 4 °C. If you observe a band around 1000 kb then the ligation is successful.

4. Store the rest of the ligation sample at 5 μL per tube at −80 °C.

3.9. Optical Tweezers Experiments

1. Wash the center channel of the microfluidic chamber with 10 mL of the DNA binding buffer and the top and bottom channels with 5 mL each of the DNA binding buffer. The center channel is connected to the top and bottom channels by two dispenser tubes (Fig. 4b).

2. Dilute 2 μL of the SA beads in 1 mL of the DNA binding buffer and inject into the bottom channel. Focus the lasers with the two objective lenses to form a laser trap (see Note 15). Trap a SA bead with the lasers from the bottom channel dispenser tube and move close to the micropipette. A recommended daily calibration step is performing a Stoke’s radius test with the trapped SA bead. The Stoke’s test result should match the known diameter of the SA bead, which is 2.1 μm (see Note 15). Stabilize the SA bead with suction from the micropipette tip connected to a syringe in the center channel.

3. Dilute the ligation product to 100-fold in the DNA binding buffer (see Note 16). Incubate 1 μL of the diluted product with 3 μL of the AD beads for 5 minutes at room temperature. Then dilute the sample in 1 mL of the DNA binding buffer (see Note 17) and inject into the top channel. Trap an AD bead with the lasers and bring close to the SA bead trapped by the micropipette tip to form a tether (Fig. 4c). This is a starting position from where the measurements take place (see Note 18).

4. Experimental conditions: we perform force ramp experiments, with a pulling speed of 75 nm/s, a sampling rate of 200 Hz, a refolding time of 10 s (see Note 19), and a refolding force of 2 pN.

5. Once data collection is complete (see Note 20), wash the center channel with 10 mL of the buffer and the top and bottom channels with 5 mL each of the storage buffer.

3.10. Analysis of Unfolding and Refolding Transitions Observed in Optical Tweezers Trajectories

Parameters underlying the folding energy landscape of a protein, such as the lifetime at zero force and the distance to the transition state, can be extracted from data collected in single molecule mechanical manipulation experiments with optical tweezers (15, 25–31). Using the method proposed by Dudko et al. (32–35), the force-dependent lifetimes of the folded and unfolded states of a protein are determined from a histogram of observed unfolding and refolding forces from force-ramp experiments (Fig. 5a). Here, we first illustrate the application of this method to mechanical unfolding and refolding corresponding to the CNB-A domain of the regulatory subunit of PKA in the presence of saturating amounts of cAMP (15, 16).

Figure 5.

(a) Force-extension plot of the truncated CNB-A domain bound to cAMP. Red curve shows unfolding of the protein (F_u) and blue curve indicates the refolding (F_f) of the protein. The unfolding and refolding trajectories are separated on the x-axis to better show the unfolding and refolding events. (b). Unfolding (red) and refolding (blue) force histograms of the truncated CNB-A domain bound to cAMP. The solid like corresponds to the reconstructed unfolding force probability distribution (red solid line) and refolding force probability distribution (blue solid line) using Equations 4 and 7. (c) Force-dependent folded state lifetimes (red squares) and unfolded state lifetimes (blue squares) of the truncated CNB-A domain bound to cAMP obtained from the histograms in panel (b) and using Equations 1 and 6. The dashed lines corresponds to fits using Bell’s model using Equation 3.

We begin analyzing the folded state of the cAMP-bound CNB-A domain by applying Equation 1 to an unfolding force histogram (Fig. 5b, red bars) with N bins of width ΔF that start at force F₀ and spans the force range (F₀, F_N = F₀ + NΔF). The i^th bin contains a number of counts C_i, and has a height h_i = C_i/(N_totΔF). N_tot is the total number of observed unfolding events and k = 1, 2, … is an index value that allows for evaluation of the following expression at the midpoint force of each bin:

$${\tau }_f\left(F_0+(k-1/2)\text{Δ}F\right)=\frac{\left(h_k/2+\sum _{i=k+1}^N{h}_i\right)\text{Δ}F}{h_k\dot{F}\left(F_0+(k-1/2)\text{Δ}F\right)}.$$ (1)

Here, $\dot{F}$ is the loading rate as a function of force and τ_f are the force-dependent lifetimes. Equation 2 (32, 34) can be applied to the force-dependent lifetime at each force bin:

$${\tau }_f(F)={\tau }_{0,f}{\left(1-\frac{vF\text{Δ}{x}_{f\to u}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{1-1/v}{e}^{-\text{Δ}{G}^{‡}/k_{B}T\left[1-{\left(1-vF\text{Δ}{x}_{f\to u}^{‡}/\text{Δ}{G}^{‡}\right)}^{1/v}\right]}$$ (2)

where τ_0,f is the lifetime of the folded state without applied force, F is the force, ${\mathrm{\Delta }x}_{f\to u}^{‡}$ is the distance to the transition state from the folded (f) to the unfolded (u) state, k_B is the Boltzmann constant, T is the temperature in Kelvin, ΔG^‡ is the free energy of activation at zero external force, and v is a scaling factor that reflects the underlying free-energy landscape (see Note 21). The lifetimes exhibit a linear behavior when v = 1, and Equation 1 simplifies to Bell’s model (36):

$${\tau }_f(F)={\tau }_{0,f^{\prime }}{e}^{-F\text{Δ}{x}_{f\to u}^{‡}/k_{B}T}.$$ (3)

The force-dependent lifetime plot for the cAMP-bound CNB-A domain is linear (Fig. 5c, red squares), and therefore was fitted to Equation 3, resulting in τ_0,f = 1.1·10³ s, and ${\mathrm{\Delta }x}_{f\to u}^{‡}=2.4\mathrm{nm}$. These values can be used to reconstruct the unfolding probability distribution, p_u(F), described by Equation 4 with force loading rate KV (product of spring constant of the tether and pulling velocity), β = 1/k_BT, and k_0,u as the unfolding rate (inverse of τ_0,f) at zero force (34):

$$p_u(F)=\frac{k(F)}{\kappa V}{e}^{\frac{k_{0,u}}{\beta \text{Δ}{x}_{f\to u}^{‡}\kappa V}}\times e^{-\left[\frac{k(F)}{\beta \text{Δ}{x}_{f\to u}^{‡}\kappa V}\right]{\left(1\frac{vF\text{Δ}{x}_{f\to u}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{1-\frac{1}{v}}}$$ (4)

with k_u(F), the force-dependent unfolding rate constants:

$$k_u(F)=k_{0,u}{\left(1-\frac{vF\text{Δ}{x}_{f\to u}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{\frac{1}{v}-1}\times e^{\beta \text{Δ}{G}^{‡}\left[1-{\left(1\frac{vF\text{Δ}{x}_{f\to u}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{\frac{1}{v}}\right]}.$$ (5)

The height of the experimentally observed unfolding force histogram can be normalized to match the reconstructed unfolding force probability distribution (Fig. 5b, solid red line). This comparison serves as another step to evaluate the fitted parameters obtained from the force-dependent lifetime plot. Lack of agreement between the observed histogram and the reconstructed unfolding probability distribution may indicate that the unfolding force histogram needs more events, especially at the extremes of the force histogram that occur with low probability (see Note 22). Moreover, as discussed below, the transformation of unfolding force histograms into force-dependent lifetimes requires input from the experimenter, which may lead to biases in the quantitative characterization of the folding energy landscape.

A similar process can also be used to analyze data measuring refolding events of the CNB-A domain bound to cAMP (Fig. 5a–b, blue trajectory and blue bars). Dudko and colleagues also describe this process, which reverses the direction of integration of the force histogram outlined in Equation 1 (34, 35):

$$\tau \left(F_0+(k-1/2)\text{Δ}F\right)=\frac{-\left(h_k/2+\sum _1^{i-1}{h}_i\right)\text{Δ}F}{h_k\dot{F}\left(F_0+(k-1/2)\text{Δ}F\right)}$$ (6)

The unfolded state lifetimes are also linear with force, and therefore were fitted to Bell’s equation (Eq. 3). This analysis yielded the unfolded state lifetime at zero force, τ_0,u = 2.0·10⁻⁴ s, and the distance to the transition state from the unfolded to the folded state, ${\mathrm{\Delta }x}_{u\to f}^{‡}$ = 8.2 nm (Fig. 5c, blue squares). Moreover, the distribution of folding forces, p_f(F), differs from the unfolding force histogram, and instead follows the distribution:

$$p_f(F)=\frac{k_f(F)}{\kappa V}\times e^{-\left[\frac{k_f^{(F)}}{\beta \text{Δ}{x}_{u\to f}^{‡}\kappa V}\right]\left[{\left(1+\frac{vF\text{Δ}{x}_{u\to f}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{1-\frac{1}{v}}\right]}.$$ (7)

With k_f(F), the force-dependent folding rate constants:

$$k_f(F)=k_{0,f}{\left(1+\frac{vF\text{Δ}{x}_{u\to f}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{\frac{1}{v}-1}\times e^{\beta \text{Δ}{G}^{‡}\left[1-{\left(1+\frac{vF\text{Δ}{x}_{u\to f}^{‡}}{\text{Δ}{G}^{‡}}\right)}^{\frac{1}{v}}\right]}.$$ (8)

Using Equations 6 and 7, we reconstructed the folding force probability distribution, which matches the experimentally observed histogram of refolding forces (Fig. 5b, solid blue line). The agreement between the two lends reliability to the fitted parameters.

Having obtained the lifetimes at zero force for the unfolded and folded states, we can use Equation 9 to estimate the thermodynamic stability of the CNB-A domain bound to cAMP:

$$\text{Δ}{G}^o=-RTln\frac{{\tau }_{0,u}}{{\tau }_{0,f}}.$$ (9)

In agreement with previous bulk studies (17, 37, 38), we find that the thermodynamic stability of the CNB-A domain bound to cAMP is 10.8 kcal/mol.

3.11. The Effects of Binning and the Number of Data Points on Fitted Energy Landscape Parameters

One challenge of carrying out the analyses proposed by Dudko and collaborators (32–34) is the appropriate construction of unfolding and folding force histograms. Defining the number or width of the force bins used to cover the observed unfolding or folding force distribution is a key part of the analysis that is determined by the experimenter (see Note 23). As shown by the results in Table 1, using bins of different sizes (0.5, 1 and 2 pN) can give quantitatively different fitted folded state lifetimes and distances to the transition state. For the CNB-A domain bound to cAMP, using a wider force bin (i.e., 2 pN vs. 1 pN bins) results in a ~3-fold lower τ_0,f. Conversely, a narrow force bin (i.e., 0.5 pN vs. 1 pN bins) results in a ~3-fold higher τ_0,f. Similar trends are observed for ${\mathrm{\Delta }x}_{f\to u}^{‡}$. Moreover, these binning effects can be more pronounced when fewer data points (i.e., a lower number of unfolding events) are available (Table 1, N = 392).

Table 1.

Kinetic parameters for the truncated cAMP-bound CNB-A domain

N=785
Bin size	τ0,F (s)	Δx‡F→U (nm)
1	1.1·103	2.4
2	3.5·103	2.0
0.5	2.8·104	2.6
N=392
1	5.8·103	2.2
2	1.6·103	1.9
0.5	1.0·103	2.4

3.12. Using Bootstrapping to Estimate the Errors in the Folded and Unfolded State Lifetimes and Distances to the Transition State

To reduce the effect of the chosen force bin size and produce more consistent results, a bootstrap procedure can be incorporated into the analysis of unfolding and refolding data with the Dudko method described above. The bootstrap procedure is carried out by repeating the Dudko analysis on a large number of resampled collections of the original unfolding or refolding force histogram. Each resampled collection is of size N_tot, with points sampled from the original data with replacement. Maintaining the same set of force bins in the force histogram, this analysis produces a collection of force-dependent lifetimes from the resampled data according to Equations 1 and 6. These lifetimes can be averaged into a single value at each force, with associated error from the standard deviation of the resampled datasets. The bootstrapped errors of the lifetimes as a function of force can be used as weights when fitting Equations 2 or 3.

An appropriate number of bootstrapped samples is determined by testing increasing numbers of iterations and ensuring that the standard deviation of the value of interest, in this case the lifetime at each force, has stabilized. The results of the analysis for the unfolding of CNB-A bound to cAMP are shown in Figure 6a. Based on the graph, 5000 bootstrapped samples are an appropriate choice for these analyses – at this point, the standard deviation values of the lifetimes have stabilized. Fitting Equation 3 to the bootstrapped folded-state lifetime values with weights from their standard error at each force provides fitting parameters that are less dependent on the force bin width and the number of unfolding events (Table 2). A similar analysis was conducted for refolding force histograms, requiring 5000 bootstrapped samples before the standard deviation of the unfolded state lifetime stabilized (Fig. 6b).

Figure 6.

Selection of an appropriate number of iterations for the bootstrapping process based on the mean of the standard deviations of the lifetime values at each force. (a) Shows this value for bootstrap analysis of unfolding data from the CNB-A domain with cAMP with a range of iterations. (b) Shows the same analysis for the refolding data. In both cases, 5000 iterations were determined to be appropriate because the mean standard deviation had become stable relative to the number of iterations.

Table 2.

Kinetic parameters for the truncated cAMP-bound CNB-A domain using bootstrapped data

N=785
Bin size	τ0,F (s)	Δx‡F→U (nm)
1	1.7·104	2.5
2	7.7·103	2.3
0.5	1.5·105	2.4
N=392
1	3.1·103	2.3
2	4.2·103	2.1
0.5	5.2·103	2.0

3.13. Dissecting the Effect of cAMP Binding and Interdomain Interactions in PKA-R

Having established an analytical and statistical strategy to extract the lifetimes at zero force and the distances to the transition state from unfolding and refolding force histograms, we quantitatively characterize the effect of cAMP binding, first to each CNB domain as an isolated protein domain, and second as part of PKA-R (Fig. 2b–c). The first experimental approach allows us to determine the effect of cAMP binding to the folding energy landscape of each CNB domain. From the second approach we can dissect how interdomain interactions, triggered by cAMP binding, further remodel the energy landscape of the protein. Altogether, this approach probes the underlying energetics of ligand-induced allosteric effects between CNB domains.

Figure 7a shows three unfolding trajectories corresponding to the CNB-A domain: as an isolated domain in the apo state (i.e., without cAMP), as an isolated domain bound to cAMP, and as part of PKA-R bound to cAMP. Relative to the apo state, cAMP binding increases the unfolding force by 8.6 pN, and interdomain interaction with CNB-B increases by another ~3 pN (Table 3). Applying Equation 1 in combination with the bootstrapping protocol, we obtained force-dependent folded state lifetimes and associated errors for each construct (Fig. 7b). We find that τ_0,f increases ~10-fold when the isolated CNB-A domain binds cAMP. Interactions with the CNB-B domain further increase τ_0,f of the CNB-A domain 100-fold (Table 3). Moreover, ${\mathrm{\Delta }x}_{f\to u}^{‡}$ for the CNB-A domain is reduced from 4.1 nm to 2.5-2.9 nm when it binds cAMP or interacts with the CNB-B domain. These results indicate that cAMP and interdomain interactions triggered by the cyclic nucleotide significantly increase the kinetic stability of the CNB-A domain and change its mechanical compliance.

Figure 7.

(a) Force-extension curves for the CNB-A domain in the apo state (1), bound to cAMP (2) and in PKA-R (3). (b) Force-dependent folded-state lifetimes for each of the constructs shown in panel (a).

Table 3.

Kinetic and thermodynamic parameters for the CNB-A and CNB-B domains in different constructs

CNB-A domain	Unfolding force (pN)	τ0,f (s)	Δx‡f→u (nm)	τ0,u (s)	Δx‡u→f (nm)	ΔG0 (kcal/mol)
Apo	8.8	2.1·103	4.1	1.4·10−4	9.7	9.8
cAMP-bound	17.4	1.7·104	2.5	2.0·10−4	8.2	10.8
cAMP bound in PKA-R	20.3	1.6·106	2.9	8.2·10−5	11.7	14.0
CNB-B domain	Unfolding Force (pN)	τ0,f (s)	Δx‡f→u (nm)	τ0,u (s)	Δx‡u→f (nm)	ΔG0 (kcal/mol)
Apo	7.3	2.8·102	3.9	3.7·10−3	6.3	6.7
cAMP-bound	12.0	7.2·103	3.5	1.6·10−3	7.6	9.1
cAMP bound in PKA-R	19.7	3.4·105	2.9	2.1·10−4	9.6	12.6

Figure 8a shows the unfolding trajectories of the CNB-B domain as an isolated domain in the apo state or bound to cAMP, and as part of PKA-R bound to cAMP. As in the case of the CNB-A domain, we observed that cAMP binding and interdomain interactions increase the mean unfolding force of the CNB-B domain (Table 3). However, we note that the magnitudes of force stabilization are different from those observed for the CNB-A domain (Fig. 7a). For instance, cAMP binding increases the mean unfolding force of the CNB-B domain in the apo state from 7.3 pN to 12 pN, and interdomain interactions with the neighboring CNB-A domain increase the mean unfolding force by an additional 7 pN. Fitting the force-dependent folded-state lifetimes plot (Fig. 8b) reveals that τ_0,f increases from 2.8·10² s in the apo state to 7.2·10³ s in the cAMP-bound state (a 25-fold increase) and then to 3.5·10⁵ s due to interdomain interactions with the CNB-A domain (an additional ~50-fold increase). Altogether, the single-molecule analysis of the two CNB domains indicate mutually stabilizing effects as they bind cAMP and interact with each other. However, we demonstrate that despite being structurally homologous, the stability contributions and responses to cAMP binding are unique to each CNB domain.

Figure 8.

(a) Force-extension curves for the CNB-B domain in the apo state (1), bound to cAMP (2) and in PKA-R (3). (b) Force-dependent folded-state lifetimes for each of the constructs shown in panel (a).

Figure 9a–b show force-dependent unfolded state lifetime plots for the CNB-A and CNB-B domains as isolated constructs in the apo state and bound to cAMP, or as part of the cAMP-bound PKA-R. We find that the presence of cAMP does not decrease τ_0,u (i.e., cAMP does not accelerate refolding) indicating that the CNB domains fold first before interacting with the cyclic nucleotide. In contrast, the presence of the neighboring CNB domain decreases τ_0,u in an asymmetric fashion, i.e., the CNB-A domain decreases τ_0,u of CNB-B by a factor of 10, whereas the CNB-B domain only marginally decreases τ_0,u for CNB-A.

Figure 9.

Force-dependent unfolded state lifetimes for the CNB-A (a) and CNB-B (b) domains in the apo state (1), bound to cAMP (2) and in PKA-R (3).

Integrating the analyses of unfolding and refolding transitions using multiple pulling geometries of a multidomain signaling protein like PKA-R allows us to estimate the thermodynamic stability of each CNB domain as well as ligand-induced coupling energies, which are the driving forces underlying the allosteric interactions triggered by cAMP. A folding energy landscape diagram (Fig. 10) shows the energetic contributions of cAMP binding and interdomain interactions to the stabilities of the CNB domains, which together lead to the cAMP-bound active state of the protein (9–11). Moreover, our approach further dissects the various modes of stabilization, such as stabilizing the folded state or destabilizing the unfolded state of each CNB domain. Lastly, we also observed how the mechanical compliance of these domains change when they bind cAMP and interact with the neighboring CNB domain. Such differences in compliance may contribute to the ability of these domains to change conformation depending on their binding partner, namely, cAMP (39), the PKA catalytic subunit (40–42), or phosphodiesterases (7, 43–45).

Figure 10.

Energy landscape and thermodynamic stability changes of the CNB domains (ΔG_apo) due to cAMP binding (ΔG_cAMP) and interdomain (ΔG_inter) contacts. The height of the energy barriers reflects the folded- and unfolded-state lifetimes of the CNB domains in the different states (Table 3). The energy landscapes have been normalized to the unfolded state.