Capping motifs stabilize the leucine-rich repeat protein PP32 and rigidify adjacent repeats

Abstract

Capping motifs are found to flank most β-strand-containing repeat proteins. To better understand the roles of these capping motifs in organizing structure and stability, we carried out folding and solution NMR studies on the leucine-rich repeat (LRR) domain of PP32, which is composed of five tandem LRR, capped by α-helical and β-hairpin motifs on the N- and C-termini. We were able to purify PP32 constructs lacking either cap and containing destabilizing substitutions. Removing the C-cap results in complete unfolding of PP32. Removing the N-cap has a much less severe effect, decreasing stability but retaining much of its secondary structure. In contrast, the dynamics and tertiary structure of the first two repeats are significantly perturbed, based on ¹H-¹⁵N relaxation studies, chemical shift perturbations, and residual dipolar couplings. However, more distal repeats (3 to C-cap) retain their native tertiary structure. In this regard, the N-cap drives the folding of adjacent repeats from what appears to be a molten-globule-like state. This interpretation is supported by extensive analysis using core packing substitutions in the full-length and N-cap-truncated PP32. This work highlights the importance of caps to the stability and structural integrity of β-strand-containing LRR proteins, and emphasizes the different contributions of the N- and C-terminal caps.

Introduction

Repeat proteins are composed of tandem repeated structural motifs that stack together to form a folded domain. These proteins are stabilized by strong interactions between adjacent repeats, but lack sequence-distant contacts commonly observed in globular proteins.¹ Despite this absence of sequence-distant contacts, repeat proteins are highly cooperative in their unfolding. This modular architecture and high cooperativity permit dissection of energetic contributions of discrete structural units, making them ideal candidates for folding studies.²

Among the most common types of repeat motifs are β-strand-containing leucine-rich repeats (LRRs). LRRs have conserved LxxLxLxxN/CxL amino acid sequence motifs (x represents any amino acid), and fold into an extended β-sheet structure such that the conserved hydrophobic leucines point towards the protein core (Supporting Information Fig. S1).³ Most LRR proteins also have capping motifs on either or both termini.⁴ These motifs are hypothesized to serve several purposes. They can shield the hydrophobic core from solvent, preventing aggregation.⁵ They can become structured in the transition state ensemble, and thus, may guide folding, as was observed for the N-terminal α-helical cap of the LRR domain of InlB.⁶ Terminal caps can also be essential in maintaining structural integrity of the adjacent LRR repeats, as seen for the C-terminal β-strand cap from the 15-repeat LRR protein YopM.⁷ Finally, the terminal caps of some LRR domains have been implicated in binding directly to partner proteins.⁸

One LRR domain that shows recognizable capping structures is the N-terminal domain of PP32, a member of the evolutionarily conserved acidic nuclear phosphoprotein family. The LRR domain of PP32 is composed of five LRRs flanked by N-terminal α-helical and C-terminal β−hairpin capping motifs (Fig. 1). Both caps are highly conserved in primary sequence. The N-terminal cap has the same structures in three closely related constructs.^9–12 The C-terminal cap also shows similar structural features, although the constructs differ in their C-terminal boundaries, complicating detailed structural comparison [Fig. 1(B,C)]. Alignment of C-terminal caps in a subfamily of LRR proteins including PP32 reveals a consensus sequence YRxxϕxxxϕPxϕxxLD (ϕ represents a hydrophobic residue, x represents any residue).¹³ In PP32, the flanking residues Y and D (Y131 and D146) form a structural hydrogen bond [Fig. 1(B,C)]. This conserved C-terminal sequence has been proposed to be necessary for nuclear targeting,¹⁴ and to shield the hydrophobic core from solvent, as well as contribute to the overall stability of the LRR domain.

Figure 1.

Structure and sequence of PP32. A: Schematic representation of the hAnp32A. The folded N-terminal LRR domain is shown as colored shapes (N-terminal capping motif in purple, repeats 1–5 in red, green, blue, magenta, and orange, respectively, the C-terminal capping motif in cyan), and the C-terminal acidic region is shown as a black line. Ribbon representation of (B) the crystal structure of the LRR domain of human Anp32A (PP32)⁹ and (C) the NMR structure of the LRR domain of mouse Anp32A.¹⁰ The conserved hydrogen-bonded Y131 and D146 are shown in black. D: Sequence alignment of the five LRRs of hAnp32 1-154. Conserved hydrophobic residues and asparagines are highlighted in cyan. The conserved hydrogen bond donor and acceptor Y131 and D146 are bolded. The five C-terminal residues included in this study (residues 150–154) are highlighted in yellow. Construct hAnp32A 1-149 is missing the residues highlighted in yellow. Construct hAnp32A 1-145 is missing residues in the red box.

To test whether N- and C-terminal caps are essential for folding and for maintaining short-range structural integrity in the LRR protein PP32, we have created several PP32 variants with modified capping structures and examined the effects on equilibrium stability and overall structural integrity. Deleting the C-cap, which removes the acceptor D146 of the conserved hydrogen bond, completely unfolds PP32. However, the disruption of the hydrogen bond alone is not responsible for this unfolding, since construct PP32 Y131F/D146L is folded and has highly similar structure to wild-type PP32. A construct lacking the N-cap (Δ_NCap PP32) is destabilized across the LRR domain, but retains significant secondary structure. Solution NMR and core packing mutational studies of Δ_NCap and full-length PP32 show that, in addition to providing long-range stability, the N-terminal cap is required for structural integrity of the two adjacent LRRs (one and two).

Results

Defining the C-terminus of the PP32 LRR domain

Members of the Anp32 family contain a highly conserved N-terminal LRR domain, followed by a variable C-terminal acidic region [Fig. 1(A)]. The boundaries of the LRR domain of PP32 (human Anp32A) were first suggested by x-ray crystallography of a construct spanning residues 1–149 [Fig. 1(B,D)].⁹ In this structure, the C-terminal residues form a single β-strand β7 (residues 144–145). Subsequent NMR studies of two PP32 homologs, mouse Anp32A¹⁰ and human Anp32B,¹² included C-terminal residues up to 164, where the acidic region begins. Some of these additional residues form a well-ordered structural motif that appears to stabilize a C-terminal β-hairpin [residues 144–145 {β-strands β7} pairing with residues 148–149 {β-strands β8}; cyan, Fig. 1(C)]. NMR ¹⁵N spin relaxation studies showed both proteins to be well structured from residues 1–154.^10,12 Therefore, the added residues may stabilize the C-terminal capping motif for the LRR region of PP32.

To determine whether these additional C-terminal sequence elements are important for structure and stability, we compared secondary structures and folding properties of constructs 1–145, 1–149, and 1–154 [Fig. 1(D)] from human Anp32A (hAnp32A). Far-UV CD spectra for both hAnp32A 1–149 and 1–154 show minima at 217 nm [Fig. 2(A)], characteristic of proteins with β-sheet structure. The two spectra also have nearly identical shapes, indicating similar secondary structures. However, hAnp32A 1–154 displays a stronger negative ellipticity, consistent with ordering of the β-hairpin structure (β7 and β8) in the 154-residue construct, as observed by NMR.^10,12 Surprisingly, the CD spectrum of hAnp32A 1–145 shows little signal at 217 nm but instead a minimum around 200 nm, characteristics of an unfolded protein, suggesting that residues 146–149 (which includes the conserved hydrogen bond) are critical for maintaining the folded structure.

Figure 2.

CD spectra and equilibrium unfolding of hAnp32A with different C-termini. A: Far-UV CD shows characteristic β-strand signal with a minimum at 217 nm for both hAnp32A 1-149 (dashed line) and 1–154 (solid line) but disordered polypeptide signal for hAnp32A 1-145 (dotted line). B: Normalized urea-induced unfolding transitions monitored by far-UV CD for hAnp32A 1-149 (squares, dashed line) and 1–154 (circles, solid line). Lines result from fitting a two-state unfolding model to the data.

To determine the effect of residues 150–154 on stability, urea-induced unfolding transitions for hAnp32A 1–149 and 1–154 were monitored by far-UV CD. Both denaturation curves are sigmoidal and can be well-fitted by an equilibrium two-state model in which the unfolding free energy is linearly dependent on urea concentration [Fig. 2(B)]. The fitted m-values for both constructs are the same within error (Table I). As m-values have been shown to be correlated with the size of cooperative unit,¹⁵ hAnp32A 1–149 and 1–154 appear to be undergoing structural transitions of the same size. The magnitudes of the fitted m-values (2.86 and 2.88 kcal mol⁻¹ M⁻¹) suggest both proteins unfold in a single concerted reaction.^† The C_m and values for hAnp32 1–154 are substantially higher than those for hAnp32A 1–149 [Fig. 2(B), Table I], indicating that the C-terminal residues 150–154 contribute significantly to the stability of the LRR domain of hAnp32A (by more than 2 kcal mol⁻¹). C_m and values for hAnp32A 1–161 are the same as for 1–154 (data not shown), consistent with observation from the NMR studies that residues following 154 are highly dynamic and unstructured.^10,12 These observations provide a thermodynamic boundary for the LRR domain of the hAnp32A (residues 1–154). For the remainder of this work, we will consider construct 1–154 as the hAnp32A LRR domain, which we will refer to as “PP32” for brevity.

Table I.

Thermodynamic Unfolding Parameters for hAnp32A 1-149, hAnp32A 1-154, and Cap Variants

		m-Value	CM
hAnp32A 1-149	−5.72 ± 0.05	2.88 ± 0.09	1.99 ± 0.05
hAnp32A 1-154 (PP32)	−7.93 ± 0.18	2.86 ± 0.02	2.77 ± 0.04
PP32 Y131F	−6.97 ± 0.09	3.05 ± 0.04	2.28 ± 0.01
PP32 D146L	−3.49 ± 0.26	3.05 ± 0.22	1.15 ± 0.02
PP32 Y131F/D146L	−4.72 ± 0.14	2.79 ± 0.14	1.69 ± 0.06
ΔNCap PP32	−4.13 ± 0.09	2.37 ± 0.04	1.74 ± 0.02

Values determined from urea-induced denaturation. in kcal·mol⁻¹; m values in kcal·mol⁻¹·M⁻¹, C_M in M (urea).

The role of the conserved hydrogen bond between Y131 and D146

Removing only four C-terminal residues from construct hAnp32A 1–149 leads to global unfolding. The resulting C-terminal deletion construct, hAnp32A 1–145, cannot form the conserved hydrogen bond in the C-terminal cap since it is missing the acceptor D146. To determine if this hydrogen bond is essential to the folding and stability of PP32, we replaced Y131 with phenylalanine and D146 with leucine (constructs Y131F, D146L, and Y131F/D146L) in the hAnp32A 1–154 construct. Unlike the hAnp32 1–145 construct, these variants are folded by far-UV CD [Fig. 3(A)]. To determine if there are significant changes in tertiary structure upon removal of the conserved hydrogen bond, we compared ¹⁵N-¹H-HSQC spectra of wild-type PP32 and Y131F/D146L. The HSQC spectra reveal well-dispersed resonances, indicating that both proteins are well-folded [Fig. 3(B)]. Overall, the spectrum of Y131F/D146L looks similar to that of wild-type PP32, with many of the peaks overlaying. However, some peaks are perturbed in the double variant.

Figure 3.

Solution spectroscopy and equilibrium unfolding of PP32 and PP32 Y131F/D146L. A: Far–UV CD shows characteristic β-strand signal with a minimum 217 nm for PP32 (black), and variants Y131F (red), D146L (magenta), and Y131F/D146L (blue). B: ¹H-¹⁵N-HSQC spectra of PP32 (black) and PP32 Y131F/D146L (blue) in 20 mM sodium phosphate, 50 mM NaCl, 0.2 mM TCEP, pH 6.8, recorded at 600 mHz, and 20°C. C: CSPs of the PP32 Y131F/D146L variant compared with wild-type. Y131 and D146 are represented in sticks. Spheres represent Cα's. CSPs of residues for which assignments can be transferred from PP32 to PP32 Y131F/D146L are displayed on a blue to white scale. Residues for which amide peaks disappear are displayed in black, and those that move in a crowded region of the HSQC and therefore cannot be assigned with certainty are displayed in red. D: Normalized urea-induced unfolding transitions monitored by far-UV CD for PP32, Y131F, D146L, and Y131F/D146L (colors as noted above). Lines result from fitting a two-state unfolding model to the data.

To map these chemical shift changes, we assigned the resonances for wild-type PP32 using standard 3D NMR experiments and for Y131F/D146L by comparison to wild-type PP32. The chemical shift perturbations (CSPs) are mostly small and localized to the sites of substitution [residues 131 and 146, Fig. 3(C)]. The N-terminal cap and first four LRRs show little to no chemical shift changes. The high chemical shift dispersion across the entire domain, along with the restriction of CSPs to the site of the substitutions, suggests that the overall LRR fold is maintained, despite the removal of this conserved hydrogen bond.

To determine the energetic consequences of removing the conserved hydrogen bond on folding, urea-induced unfolding transitions for the variants were monitored by far–UV CD [Fig. 3(D)]. Based on the decreased and C_m values, these substitutions are all destabilizing, indicating that the folded state of PP32 is indeed stabilized by this capping hydrogen bond (Table I). The differences in the extent of destabilization for Y131F and D146L single site variants indicate additional interactions between these two residues and their intermediate surroundings. Moreover, there is significant nonadditive free energy between these substitutions, with the larger destabilization in the wild-type background, compared with the singly substituted backgrounds. This nonaddictive effect suggests that the true contribution of the hydrogen bond maybe overestimated from analysis of the single mutants, which likely pay a penalty for desolvation and burial of a hydrogen bonding group. All three variants have nearly identical m-values to wild-type PP32 [Fig. 3(D), Table I], suggesting fully cooperative unfolding transitions, despite the highly destabilizing effects.

The role of the conserved N-terminal α-helical capping motif

To investigate the structural and energetic contributions of the N-terminal capping motif, we removed the 18 residues prior to the first repeat (Δ_Ncap PP32). Far-UV CD spectra for full-length and Δ_Ncap PP32 have nearly identical shapes, suggesting that the two proteins have similar secondary structures (data not shown). Urea-induced unfolding reveals that Δ_NCap PP32 has lower and C_m values than full-length PP32 [Fig. 4(A); Table I], indicating that PP32 is less stable without the N-terminal capping motif. Unlike the deletion of residues 150–154, the m-value for Δ_NCap PP32 decreases, consistent with a smaller cooperative unfolding unit.

Figure 4.

Equilibrium unfolding and NMR spectroscopy of PP32 and Δ_NCap PP32. A: Normalized urea-induced unfolding transitions monitored by far-UV CD for PP32 (black) and Δ_NCap PP32 (red). Lines result from fitting a two-state unfolding model to the data. B: Superposition of the HSQC spectra of PP32 (black) and Δ_NCap PP32 (red) in 20 mM sodium phosphate, 50 mM NaCl, 0.2 mM TCEP, pH 6.8, recorded at 600 mHz, and 30°C. C: CSPs resulting from removal of the N-terminal cap. Residue numbering is based on the PP32 construct. D: Mapping of CSPs displayed in (C) onto the structure of PP32. The 18-residue α-helical-capping motif is in black. Spheres represent the Cα's of the assigned residues. CSPs are displayed on a white to blue scale.

To obtain a more detailed structural picture of the consequences of N-terminal cap deletion, we studied Δ_Ncap PP32 by NMR. The HSQC spectrum shows a large number of well-dispersed resonances, indicative of a folded protein [Fig. 4(B)]. However, comparison of full-length and Δ_NCap PP32 HSQC spectra shows that many peaks have significantly shifted. To map these chemical shift changes, we assigned the resonances for Δ_NCap PP32. Large CSPs were observed for N-terminal repeats one and two, whereas the last three repeats and C-terminal capping motif remain largely unperturbed [Fig. 4(C,D)]. Using the program TALOS,¹⁶ which predicts secondary structures based on chemical shifts and sequence information, we observe highly similar secondary structure distributions in full-length and Δ_NCap PP32 despite significant chemical shift changes in repeats one and two. In particular, the β-strands of the LRRs are retained along the entire Δ_NCap PP32 construct.

To determine whether the large CSPs simply resulted from a local change in chemical environment upon removal of the N-terminal capping motif, or from a larger conformational rearrangement of repeats one and two, we measured residual dipolar couplings (RDCs) for both PP32 (data not shown) and Δ_NCap PP32 [Fig. 5(A)]. RDCs provide information on orientations between directly bonded atoms and the overall molecular frame, and thus, present a sensitive means to compare the higher-order structural details of related proteins. From repeat two to the C–terminus, the strong correlation between the RDCs of full-length and Δ_NCap PP32 indicates high structural similarity [Fig. 5(B)]. In contrast, for the first repeat, there is relatively poor correlation between the RDCs of full-length and Δ_NCap PP32, suggesting that the structure of the first repeat has undergone a significant conformational rearrangement.

Figure 5.

RDC of Δ_Ncap PP32. A: RDC values along the sequence of Δ_Ncap PP32 in liquid crystalline media of 5% C12E5/Hexanol. B: Correlation between RDCs of Δ_Ncap PP32 and of corresponding PP32 residues. The data are colored by repeat identity: red, 1; green, 2; blue, 3; magenta, 4; orange, 5; and cyan, C-terminal capping motif. The black line represents perfect correlation of RDCs between full-length and Δ_Ncap PP32.

To directly probe whether the conformational changes increase dynamics upon removing the N-terminal cap, we carried out relaxation studies on the Δ_NCap PP32 (Fig. 6). R₁ values are nearly constant along the sequence. However, R₂ values are elevated for the first 55 residues, but converge to a constant value towards the C-terminus. These data are consistent with μs-ms-timescale fluctuation in the N-terminal two repeats. Similar to the R₂ profile, heteronuclear NOE intensities are high and uniform (around the theoretical value of 0.8) for most of the protein, except for the first 35 residues on the N-terminus, where the intensities are varied and slightly depressed. These observations indicate that the C-terminal region of Δ_NCap PP32 is folded and rigid whereas the first repeat and half of the second repeat experience fast backbone fluctuation on the ps-ns timescale. For mouse Anp32A¹⁰ and human Anp32B,¹² the R₁, R₂, and heteronuclear NOE profiles have been shown to be relatively constant along the entire domain (including the N-terminal α-helical capping motif and the first LRR). Consistent with these changes in dynamics, we observed an increase in the intensities of many N-terminal HSQC resonances of Δ_NCap PP32 with increasing temperature (data not shown). The enhanced R₂ and depressed heteronuclear NOE values seen here indicate that the α-helical cap rigidifies LRRs one and two.

Figure 6.

The ¹⁵N spin relaxation parameters of Δ_NCap PP32. Residue specific R₁ (top), R₂ (middle) and heteronuclear NOE (bottom) profiles are shown with error bars. The relaxation experiments were carried out at pH 6.8, 30°C.

Removal of the N-terminal α-helical capping motif disrupts side chain packing of the N-terminal LRRs

To determine the extent to which deletion of the N-terminal cap disrupts side-chain packing, we made conservative single-site Leu→Ala and Val→Gly substitutions at core positions of both full-length and Δ_Ncap PP32. Far-UV CD spectra of full-length PP32 packing variants are nearly identical in shape to those of wild-type PP32, indicating that the β-sheet structure of PP32 is retained [Fig. 7(A)]. However, spectra of many of the same variants of Δ_NCap PP32 differ from that of the parent construct. These differences depend on the location of the substitutions. Packing substitutions in the C-terminus shift spectra towards random coil, whereas substitutions in the N-terminus show less pronounced spectral broadening [Fig. 7(A)].

Figure 7.

CD spectroscopy and equilibrium unfolding of core packing substitutions in PP32 and Δ_NCap PP32. A: Far-UV CD shows disruption of β-sheet structure upon C-terminal packing substitutions in Δ_NCap, but not in full-length PP32. For direct comparison of the shape of far-UV CD spectra, the traces are normalized to −1 at 216 nm. Wild-type constructs are in black. Packing variants are as depicted in the ribbon diagram: red, L22A; magenta, L37A; blue, L47A; light green, L60A; dark green, L83A; orange, L109A; pink, V135G. Full-length constructs are shown with solid lines. Δ_NCap constructs are shown with dashed lines. B: Normalized urea-induced unfolding transitions monitored by far-UV CD for PP32 and variants (circles, solid line) and Δ_NCap PP32 and variants (squares, dashed line). Lines result from fitting a two-state unfolding model to the data. Lines for Δ_NCap PP32 L83A, L109A, and V135G are only for guide and do not result from fitting.

To quantitatively assess of the effects of these packing substitutions on structural stability, urea-induced unfolding transitions were monitored by far-UV CD [Fig. 7(B)]. Based on decreases in C_m, full-length PP32 variants are significantly destabilized across the entire domain (Table II). C-terminal substitutions (L83A, L109A, and V135G) are similarly destabilized in the Δ_NCap background.^‡ However, substitutions in the first two repeats of Δ_NCap (L22A, L37A, L47A, and L60A) have about the same stabilities as the parent (Δ_NCap) construct. These four sites are in the region of increased dynamics (based on R₂ and heteronuclear NOE) and structural change (based on CSPs and RDCs). These data are consistent with a structural ensemble in which packing is disrupted in the first two LRRs of Δ_NCap PP32, but retained in LRR3 through the C-terminus.

Table II.

Thermodynamic Unfolding Parameters for Core Packing Substitutes in Full-Length and Δ_NCap PP32

		m-Value	CM
PP32	−7.93 ± 0.18	2.86 ± 0.02	2.77 ± 0.04
PP32 L22A	−3.88 ± 0.05	2.36 ± 0.03	1.64 ± 0.01
PP32 L37A	−3.68 ± 0.15	2.28 ± 0.07	1.62 ± 0.01
PP32 L47A	−2.98 ± 0.17	2.22 ± 0.06	1.34 ± 0.05
PP32 L60A	−3.98 ± 0.10	2.62 ± 0.01	1.52 ± 0.04
PP32 L83A	−4.46 ± 0.02	3.25 ± 0.02	1.37 ± 0.01
PP32 L109A	−4.56 ± 0.08	3.37 ± 0.17	1.35 ± 0.06
PP32 V135G	−4.07 ± 0.19	3.61 ± 0.16	1.13 ± 0.01
ΔNCap PP32	−4.13 ± 0.09	2.37 ± 0.04	1.74 ± 0.02
ΔNCap PP32 L22A	−4.07 ± 0.25	2.34 ± 0.11	1.64 ± 0.03
ΔNCap PP32 L37A	−3.58 ± 0.16	2.23 ± 0.08	1.60 ± 0.03
ΔNCap PP32 L47A	−2.76 ± 0.12	2.01 ± 0.07	1.37 ± 0.04
ΔNCap PP32 L60A	−3.03 ± 0.37	2.14 ± 0.28	1.42 ± 0.04

Values determined from urea-induced denaturation. in kcal·mol⁻¹; m-values in kcal·mol⁻¹·M⁻¹, C_M in M (urea). Parameters for PP32 Δ_NCap L83A, L109A, and V135G could not be determined.

Discussion

Several studies have examined the contributions of individual units to the structure and stability of linear repeat proteins through addition and removal of repeats.^2,7,17–22 Most studies have focused on α-helical repeat proteins, which appear to be tolerant to manipulations of whole repeats.^17,18,23 In contrast, deletion of β-strand-containing repeat motifs could result in large-scale disruption of adjacent (and nonadjacent) repeats. Removing the C-terminal motifs from YopM caused unfolding of several preceding repeats.⁷ The prevalence of terminal capping structures in β-strand containing repeat proteins (and LRR proteins in particular) may serve as a means to protect otherwise labile β-strand repeats. Removing the N-terminal caps from β-strand-containing LRR proteins InlB and YopM resulted in disordered polypeptides (N. Courtemanche, E. Kloss and DB, unpublished). Deletion of the C-terminal cap from pertactin β-helix resulted in mixture of aggregates and soluble oligomers.²⁴ In this study, we seek to further understand the roles of capping motifs on structures and stabilities of LRR proteins.

C-terminal capping structures are essential for stability and the structural integrity of the entire PP32 LRR domain

Despite making little contact with the rest of the protein, the β-hairpin cap on the C-terminus of PP32 is critical for stability and folding (Fig. 2). Removing the last five residues (the second strand of the β-hairpin, hAnp32A 1–149) decreases the stability of PP32 by more than 2 kcal/mol (Table I). Removing the next four residues (the first strand of the β-hairpin, hAnp32A 1–145) causes the entire protein (N-terminal capping motif and all five LRRs) to unfold. This is a more global structural disruption than that produced by a similar C-terminal deletion in YopM, where removing the β-cap causes only the three preceding LRRs to unfold.⁷ This complete loss of structure explains the loss of function in yeast SDS22, where deletion of the C-terminal cap disrupts nuclear localization.¹⁴ It is important to note that the increased tolerance of YopM to β-cap removal might be a result of the larger size of YopM (15 repeats). Unlike the pertactin β-helix, PP32 remains monomeric upon C-terminal cap deletion.²⁴ For PP32, the cap protects the structural integrity of the entire molecule. Global unfolding in hAnp32A 1–145 is not solely the result of loss of the conserved hydrogen bond between Y131 and D146 (Fig. 3), highlighting the importance of other interactions within the C-terminal cap.

The N-terminal cap stabilizes and constrains the first two repeats

In contrast to LRR proteins InlB and YopM, PP32 is folded and well-behaved without its N-terminal cap. In addition to being less stable [Fig. 4(A)], Δ_Ncap PP32 has lower than predicted m-value based on the number of residues removed,^§ indicating that the cooperative unit for Δ_Ncap PP32 is smaller than the construct itself. However, Δ_Ncap PP32 has similar secondary structure to the full-length PP32, based on far-UV CD and TALOS¹⁶ (data not shown). Interestingly, the RDC values of the two constructs correlate well, except for the first repeat, suggesting that the first repeat of Δ_NCap PP32, as a unit, might align differently than in full-length PP32, relative to the rest of the LRR domain [Fig. 5(B)]. Relaxation studies show that the first two LRRs experience increased dynamics both in fast and slow timescales upon N-terminal cap deletion (Fig. 6). Together, the data point to a less ordered, or packed, N-terminus, but well-folded native-like C-terminus in Δ_NCap PP32, as confirmed by mutational analysis of the two constructs (Fig. 7). Upon removal of N-terminal cap, PP32 seems to have molten globule-like structure, a partially folded intermediate where secondary structures are formed but rigid tertiary packing is disrupted.^25,26

A possible mechanism for the folding of PP32

We have been able to obtain folded PP32 that is missing a significant segment of its N-terminus [18 residues, Δ_Ncap PP32, and up to 65 residues (data not shown)]. However, removing only four C-terminal residues (146–149) unfolds the protein, suggesting that the C-terminus might be the most stable region of the molecule. Studies of many globular proteins, and α-helical repeat proteins (myotrophin and Notch ankyrin) have shown that folding pathways are determined by local stabilities.^18,27–33 Based on these observations, our results suggest that for the LRR domain of PP32, the C-terminus folds first, and may nucleate the rest of the fold. This proposed mechanism differs from the LRR domain of InlB, which folds through a polarized N-terminal pathway involving the N-cap and first two repeats. The C-terminal folding mechanism for PP32 seems at odds with a cotranslational folding model, as folding of the nascent PP32 chain will be prevented until the C-terminus is released from the exit tunnel.

Materials and Methods

Subcloning, protein expression, and purification

The gene encoding the PP32 was a kind gift from the laboratory of Dr. Cynthia Wolberger. LRR encoding segments were subcloned into the Ndei and XhoI sites of pET24b (Novagen, Madison, WI) using PCR. A tryptophan codon was added to facilitate protein concentration determination. Point substitutions were made using Quikchange (Stratagene, La Jolla, CA).

Constructs were expressed in Escherichia coli BL21 cells in autoinduction media³⁴ at 20°C overnight. Bacteria were pelleted, lysed in 20 mM NaPO₄, 500 mM NaCl, 25 mM imidazole, 0.1 mM TCEP (pH 7.4), and purified via Ni²⁺ chromatography. Purified proteins were dialyzed into 20 mM Na PO₄, 150 mM NaCl, 0.1 mM TCEP (pH 7.8), and were frozen at −80°C. Protein concentrations were determined as described.³⁵

Circular dichroism spectroscopy

All CD measurements were carried out using Aviv Model 400 CD spectrometer (Lakewood, NJ). Far-UV CD spectra were collected in an 0.1 cm path-length quartz cuvette with protein concentrations ranging from 15 to 30 μM. Spectra were averages of three wavelength scans, each with 1 nm step size and 10 s signal average per step.

Equilibrium unfolding

Urea-induced unfolding was monitored by CD at 220 nm. Urea (Amresco, Solon, OH) was deionized by chromatography over mixed-bed resin (Bio-Rad, Hercules, CA). Urea concentration was determined by refractometry.³⁶ Urea titrations were carried out using a computer-controlled Microlab syringe titrator (Hamilton, Reno, NV). Samples contained 2–4 μM protein, 20 mM NaPO₄, 150 mM NaCl, and 0.1 mM TCEP (pH 7.8). At each urea concentration, samples were equilibrated for 5–10 min at 20°C and CD signal was averaged for 30 s. Two-state analysis of equilibrium unfolding transitions were carried out as described by Street et al.³⁷

NMR spectroscopy

¹⁵N- and ¹⁵N,¹³C-labeled PP32 and variants were expressed in M9 minimal medium supplemented with ¹⁵NH₄Cl and ¹³C-labeled glucose (Cambridge Isotope Laboratories, Andover, MA) at 20°C overnight and purified as described above. NMR samples contained 0.6–1.2 mM proteins, 20 mM NaPhos, 50 mM NaCl, 0.1 mM TCEP, 0.1 mM EDTA, and 5% D₂O (pH 6.8). For RDC experiments, 0.6 mM full-length and Δ_NCap ¹⁵N-labeled PP32 constructs were aligned using a liquid crystalline medium containing 5% (w/w) pentaethylene glycol monododecyl ether (C12E5), with 1-hexanol (Sigma-Aldrich, St. Louis, MO) at mole ratio (to C12E5) of r = 0.85.³⁸ NMR experiments comparing PP32 and PP32 Δ_NCap were performed at 30°C and those comparing PP32 and PP32 Y131F/D146L were performed at 20°C. Spectra were collected on a Bruker Avance II 600 MHz spectrometer equipped with a cryoprobe, processed using NMRPipe,³⁹ and displayed and analyzed with Sparky.⁴⁰

Resonance assignments for PP32 and PP32 Δ_NCap were carried out using triple-resonance spectra, including HNCACB, CBCA(CO)NH, HNN, HBHA(CO)NH, HNCA, and ¹⁵N-edited HSQC-NOESY. Assignments were determined using CARA.⁴¹ CSPs between constructs were determined from differences in resonance positions in ¹H–¹⁵N HSQC spectra using the equation Δδ = [(Δδ_H)² + (Δδ_N/5)²]^1/2, where Δδ_H and Δδ_N are differences in the ¹H and the ¹⁵N chemical shifts, respectively, for a given residue. RDC values for full-length and Δ_NCap PP32 were each measured with two sets of IPAP experiments in alignment media and in isotropic media.⁴² Peak positions in the IPAP-HSQC spectra were determined using the program PATI.⁴³ R₁, R₂, and heteronuclear NOE measurements for Δ_NCap PP32 were collected using standard pulse sequences. ¹⁵N T₁ experiments used relaxation delays of 97.5, 195, 299 (duplicated), 494, 696, 897, and 1099 ms. ¹⁵N T₂ experiments used relaxation delays of 7.6, 15.2, 22.8, 30.4 (duplicated), 38, 45.6, and 53.2 ms. Relaxation rates were determined using the program RELAXFIT, with 500 Monte Carlo trials for error analyses.⁴⁴

Glossary

CSPs

chemical shift perturbations

hnNOE

heteronuclear NOE

LRR

leucine-rich repeat

NOE

nuclear overhauser effect

RDC

residual dipolar coupling

Supporting Information

Additional Supporting Information may be found in the online version of this article.

Supporting Information Figure 1.