Structure and Orientation of T4 Lysozyme Bound to the Small Heat Shock Protein α-Crystallin

Approach and general methodology

To determine the conformational state of T4L bound to α-crystallin, we analyzed proximities in pairs of spectroscopic probes introduced at selected sites that fingerprint the tertiary fold. The pairs consisted of either two nitroxide spin labels or a bimane and a tryptophan (trp). Trp quenches bimane fluorescence in the distance range below 15 Ų⁹. Spin labels undergo dipolar coupling in the 5-20 Å range which is manifested as broadening of the Electron Paramagnetic Resonance (EPR) lineshape ³⁰. Changes in bimane emission intensity reflect proximity changes between the trp and the fluorophore. This approach allows detection of local structural movements upon complex formation. It also overcomes the non-specificity of circular dichroism or FT-IR ²⁸ because the probes exclusively report on the structure of the substrate.

To promote binding of T4L to α-crystallin, the probes were introduced in the T4L-L99A background; a hydrophobic mutation in the C-terminal domain that significantly reduces the free energy of unfolding without perturbing the overall structure of T4L ³¹. Several criteria were employed to select appropriate pairs. First, we needed to sample structural motifs and domains across the molecule. Second, replacement of endogenous residues for cysteine and/or trp should not result in significant changes in stability beyond that due to the destabilizing mutation. Therefore, only surface sites were considered for cysteine replacement and subsequent probe attachment. Third, the degree of bimane quenching is influenced by the distance and orientation of the trp. The quenching of bimane by tyrosine residues is a complicating factor and we avoided introducing cysteines in their vicinities where possible. Using these criteria, two bimane/trp pairs were introduced into the C-terminal (N116Bi/N132W, G113Bi/K83W) and N-terminal (V57Bi/K16W, L39Bi/Y25W) domains. In addition, another pair was designed to bridge the active site cleft between the C- and N-terminal domains (Q141Bi/E22W). Each pair has a C_α-C_α distance less than 10 Å according to the crystal structure (Supplementary Table 1) although the actual distance between the probes depends on their relative projection. Figure 1 shows the location of the bimane/trp pairs in the crystal structure of T4L-L99A.

α-crystallin binds T4L in two discrete modes with vastly different affinities leading to distinct bimane emission characteristics ¹⁸. The manifestation of the two modes in the binding isotherm depends on the concentrations of the substrate and chaperone relative to the dissociation constants as well as the differences in the fluorescence emission in each mode. Therefore, the T4L concentrations were adjusted relative to the dissociation constants, K_D1 and K_D2 of the high and low affinity modes, to either include or eliminate contributions from low affinity binding. In addition, the majority of binding experiments were conducted with the triply phosphorylated analog of αB-crystallin, αB-D3 (S19D/S45D/S59D). αB-D3 has higher affinity for T4L than the WT which allows for quantitative analysis of two-mode binding isotherms ¹⁸; however, experiments with αA-crystallin yielded qualitatively similar characteristics.

Pattern of bimane quenching in native T4L-L99A

Under conditions that favor the folded state, all five T4L bimane/trp mutants demonstrated substantial quenching of bimane fluorescence relative to the corresponding single mutant. This observation is consistent with the mutants being properly folded and suggests that the substitutions do not significantly alter the tertiary structure. The thermodynamic stabilities of single and double mutant T4L were analyzed by denaturant unfolding. In most cases both the bimane (485 nm) and intrinsic trp (350/320 nm) unfolding curves report similar thermodynamic parameters as detailed in Table 1. ΔG_unf of the bimane-labeled mutants clusters in a narrow range around that of T4L-L99A. Except for N116Bi/N132W, the deviations are well within the error of the non-linear least-squares fit of the curves.

Global unfolding alters the intensity of bimane emission at all sites reflecting changes in its environment between the folded and unfolded states. In the bimane/trp pairs, bimane fluorescence increases considerably in the unfolded state indicating that the main determinant of emission intensity, quenching by the neighboring trp, is completely removed during unfolding (Figure 2a). For the single mutants, the sign of intensity changes does not follow a simple pattern except at loop sites where subtle changes are observed (Supplementary Figure 1). At G113Bi and L39Bi, we observe an increase in bimane emission upon unfolding, suggesting a native quenching phenomenon. L39Bi and G113Bi are 61% and 84% quenched, respectively, relative to N116Bi, a surface site with no nearby trp or tyr residues. The quenching observed at L39Bi is likely due to a native tyrosine residue at site 25. However, the magnitude of quenching is six-fold less than trp-induced quenching as evidenced by L39Bi/Y25W double mutant. The source of quenching at G113Bi is more enigmatic. The distance between residue 113 and the closest trp 138 is within the quenching radius, but the projection of 113 away from the core is expected to result in minimal interaction. In any case, quenching is increased in the G113Bi/K83W double mutant allowing the use of this pair to monitor unfolding.

Bimane quenching is eliminated upon binding of trp/bimane pairs to α-crystallin

Upon addition of a saturating amount (50-fold excess) of αB-D3 (or αA-crystallin), corresponding to high affinity binding, bimane fluorescence increases for the double mutants suggesting a reduction in the extent of bimane quenching by the neighboring trp. This effect is illustrated in Figure 2b for N116Bi/N132W where the bimane emission intensity of the double mutant is almost identical to the corresponding single mutant in the bound state but severely quenched in native, free T4L. Table 2 (and Supplementary Table 2) quantitatively reports the intensity change for all five double mutants. The results indicate that under high affinity binding conditions, native tertiary contacts are disrupted in both the C- and N-terminal domains as well as at their interface near the active site cleft. The marginal deviations in bimane intensities between the single and double mutants in the bound state can be attributed to small differences in the efficiency of labeling between the mutants as the protein concentration was the same in all samples.

Bound T4L is unfolded in both binding modes

Because low affinity binding occurs following saturation of high affinity sites, it is not possible to directly analyze the structure of low-affinity bound T4L. However, if the structure of bound T4L is distinct in the two modes, a characteristic binding isotherm is expected for the bimane/trp double mutants. A more native-like structure (i.e. compact) in the low affinity mode will lead to a closer proximity between the trp and bimane. Therefore, F₂, the low-affinity emission intensity, is expected to be significantly smaller than F₁, the high-affinity emission intensity. In contrast, if T4L is unfolded in both modes F₁ and F₂ are significantly larger than F_N, the normalized emission intensity in the native state. The simulated isotherms corresponding to these two cases are compared in Figure 3a. An unfolded conformation in the low affinity mode versus a compact conformation in the high-affinity mode leads to a biphasic isotherm as shown in Figure 3b. This case is excluded by the results of Table 2.

Experimentally, the binding isotherms of αB-D3 (Figure 3c) and αA-crystallin (Figure 3d) to the double mutants demonstrated a monophasic increase in intensity. The fractional emission intensities of T4L in both modes extracted from non-linear least-squares analysis is significantly higher than that of free, folded T4L (Table 3). Thus, low-affinity binding also increases the separation between the trp and bimane. This pattern of fluorescence is consistent with a loss of tertiary structure in both binding modes. The differences between F₁ and F₂ reflect the more hydrophobic environment of T4L in the high affinity mode (discussed below) which leads to an enhanced quantum yield.

Tertiary and secondary structure unfolding of bound T4L

The binding-induced unfolding of T4L was further confirmed via analysis of proximity between pairs of spin labels. Spin labels attached to the double mutant N116C/N132C have a broadened EPR lineshape in the folded state consistent with the short distance (8.7 Å, Supplementary Figure 2) in the crystal structure (Figure 4a). Binding to αB-D3 (or αA-crystallin) eliminates broadening due to spin-spin proximity as indicated by the almost superimposable lineshapes of the double mutant to that of the sum of single mutants in the bound state in Figure 4b. Furthermore, the lineshapes are superimposable at molar ratios of α-crystallin to T4L corresponding to exclusive high-affinity binding and combined two-mode binding (Supplementary Figure 3) which supports the conclusion that binding induces distance increases regardless of binding mode. Loss of broadening due to spin-spin interactions is also observed for the pairs E22C/Q141C and Y25C/L39C (data not shown).

Pairs of spin labels were introduced at i, i+4 in helix B in the N-terminal domain (S44C/K48C) and helix H in the C-terminal domain (V131C/K135C) to assess secondary structure unfolding in the bound conformation. In solution, the EPR spectra reveal substantial dipolar broadening consistent with the expected distance in a helix (Figure 4a). Upon binding, the spectral features arising from dipolar coupling disappear as demonstrated by a significant increase in intensity (Figure 4b). The increased distance between i, i+4 suggests that loss of tertiary structure is accompanied by helical unfolding. However, the lineshape of bound S44C/K48C shows residual homogenous broadening relative to the sum of singles in the bound state. The origin of spectral broadening is spin-spin interactions between residues i and i+4 separated by less than 20Å in an extended, unfolded peptide. Spin-spin interactions in this range can be detected by EPR as shown by the broadened lineshape of this double mutant in the presence of 4.5M guanidinium HCl (Supplementary Figure 4).

Bound T4L has a preferred orientation

In addition to reporting the increase in distance between spin labels in the bound state, the spectral lineshapes in Figure 4b describe the motional properties of spin labels ³². At sites 116 and 132, rigid limit EPR spectra are indicative of a highly restrictive steric environment. In contrast, the dominant component in the EPR spectrum at sites 44 and 48 is highly mobile reflecting considerable motional freedom of the spin labels.

To probe whether these motional differences represent an asymmetric pattern of contacts between the chaperone and T4L, we took advantage of the sensitivity of bimane emission to the polarity of its environment. Specifically, the wavelength of maximum emission (λ_max)has been shown to shift to larger values upon transition from a hydrophobic to a hydrophilic environment ³³. A set of single cysteine mutants sampling different regions of T4L-L99A was constructed to determine the wavelength of maximum emission under conditions corresponding to predominant high-affinity binding. Because the sites were solvent exposed in native, free T4L, the average emission maximum of bimane is 471 ± 3 nm. In the high-affinity complex with αB-D3 or αA-crystallin, induced shifts in λ_max are dependent upon the location of the probe. Overall, a bipolar distribution is observed with residues that map to the N-terminal domain (average bound λ_max 465 ± 2 nm) shifted less than those that map to the C-terminal domain (average bound λ_max 456 ± 3 nm). This suggests that bound T4L has a net orientation relative to the aqueous solvent with the unfolded N-terminal domain more exposed (Table 4 and Supplementary Table 3). Mapping λ_max of bound T4L onto a surface representation of T4L illustrates the polarized distribution in the bimane emission pattern between the C- and N-terminal domains (Figure 5). The shift in λ_max is almost identical at high and low T4L concentrations (Supplementary Figure 5) i.e. irrespective of contribution by the low affinity mode. Thus, low affinity binding does not significantly shift λ_max relative to native free T4L consistent with binding in a more solvent exposed region of the oligomer.

T4L orientation is not dependent on the location of the destabilizing mutation

Because L99 is located in the C-terminal domain, we investigated whether the net orientation reflects binding in the vicinity of the mutation due to local dynamic unfolding. To address this question, a mutation of similar nature, L46A, was introduced in the N-terminal domain. The crystal structure of L46A reveals that the mutation induces minor local repacking but the overall structure remains similar to the WT ³⁴. The mutation reduces ΔG_unf relative to wild type T4L, albeit to a lesser extent than L99A (Figure 6a). Denaturant unfolding of T4L-L46A indicates a 2-3 kcal/mol greater stability than L99A (Table 1), consistent with previously published data ³⁴. Reflecting its higher stability, the binding isotherm of L46A/T151Bi is characterized by an increase in K_D for both binding modes and an apparent reduction in the number of binding sites, n, relative to L99A/T151Bi (Figure 6b and Table 5). The reduction in n reflects the minor contribution of the low affinity mode which makes the binding isotherm under-determined for non-linear least squares analysis. Despite the differences in stability and binding, the shifts in bimane λ_max were insensitive to the location of the destabilizing mutation, resulting in identical λ_max patterns for both C- and N-terminal domain mutants (Supplementary Table 4).

Detailed analysis of the binding curves for the L46A (Supplementary Table 5) and L99A (Table 3) mutants conclusively shows that ΔG_unf is the main determinant of binding. Most L46A mutants bind only with high affinity at both 5 and 30 μM fixed concentration of T4L consistent with the greater stabilities of L46A relative to L99A. However, there are variations in the affinities of the mutants within a given background that do not correlate with ΔG_unf. These are particularly significant for N-terminal bimane sites in the L99A background (Table 3). It is likely that labeling T4L with bulky probes modulates binding in a site-specific manner.

EPR analysis confirms the unfolded structure of bound T4L and its asymmetric contacts in the complex

In agreement with the polarized distribution of bimane emission shifts, the EPR spectra of spin labeled V131C, N116C, V57C and K48C report distinct motional properties in the bound conformation (Figure 7). At site N116C, for instance, the spin label is highly immobilized and the λ_max of N116Bi is strongly blue shifted to 450 nm. Both results indicate that the attached probe is in a buried environment. V131Bi does not shift as strongly (final λ_max=459nm) and the corresponding EPR spectrum shows the titration of a mobile component (red arrow in Figure 7). Thus, there appears to be a direct correlation between the final bimane emission λ_max and the spin label mobility in the chaperone complex (Supplementary Figure 6).

A survey of the single T4L mutants labeled with iodoacetamide proxyl spin label revealed domain-specific motional properties in the bound conformation, regardless of binding mode. Spin labels introduced at sites in the C-terminal domain exhibit spectra characteristic of highly restricted motion (Figure 7a). In contrast, the spectra of sites in the N-terminal domain mutants are dominated by three sharp lines indicative of a highly mobile spin label, although a low field immobile component was also observed suggesting the presence of a second minor population of T4L where the spin label motion is restricted (Figure 7b).

The dominant mobile component indicates that this segment of the N-terminal domain does not directly contact α-crystallin. Furthermore, the lineshape is sharper than that of native, free T4L (inset, Figure 7b). The increased dynamics at the point of attachment of the spin label in the chaperone complex is consistent with the unfolding of this helix deduced from the observed proximity changes. Size exclusion chromatography before and after the EPR analysis did not detect significant dissociation of the complex (Supplementary Figure 7).

To confirm that the mobile lineshapes of residues in the N-terminal domain reflect a more solvent exposed location, we measured the accessibility of the spin labels to collision with NiEDDA. The EPR spectra of bound T4L mutants V57C and N116C are shown in the absence and presence of 50mM NiEDDA in Figure 8b. For V57C, NiEDDA induces spectral broadening by T₂ relaxation. In contrast, the spectrum of the bound C-terminal mutant N116C does not demonstrate any changes in the presence of NiEDDA. Power saturation experiments under high affinity binding conditions (30:1, αB-D3:T4L) reveal minimal accessibility to NiEDDA (Π=0.06) although an increase in accessibility (Π=0.12) was observed under low affinity conditions. These results clearly indicate that the C-terminal domain of T4L is buried in the αB-D3 complex, while the N-terminal domain remains partially exposed to solvent.

Concluding remarks

A detailed perspective on the mechanism of α-crystallin binding to its substrates emerges from the structural and thermodynamic analysis. Figure 9 presents a schematic of the thermodynamic coupling model that describes the chaperone activity of sHSP. Our results support an unfolded structure for stably bound T4L. However, a transient recognition step that precedes binding may involve more compact or native-like conformations as suggested by kinetically-determined aggregation assays. This implies that subsequent unfolding occurs on the sHSP oligomer. A detailed investigation of the rate constants of T4L binding is required to test this hypothesis.

The T4L accessibility profile in the bound state strongly suggests that the high-affinity binding sites are inside the sHSP oligomers while low affinity binding seems to involve a surface location of T4L. CryoEM analysis of complexes between sHSP and T4L are currently underway to verify this conclusion.

Materials

Monobromobimane was purchased from Molecular Probes. 3-(2-Iodoacetamide)-proxyl spin label was purchased from Sigma-Aldrich.

Cloning and Site-directed mutagenesis

T4L mutants were created using complimentary oligonucleotide primers containing the desired mutation and amplified via polymerase chain reaction as previously described ³². All mutants were cloned into pET20b+ expression vector. Plasmid DNA sequencing from transformed cells confirmed the presence of the desired mutations and the absence of unwanted changes. In this paper, mutants of T4L are named by specifying the original residue, the number of the residue, followed by the new residue. For bimane-labeled mutants, the suffix “Bi” is used.

Expression and Purification of T4L Mutants and α-Crystallin

T4L mutants were expressed in E. coli BL21(DE3) following standard protocols and using an induction temperature of 37°C for L46A mutants and 28-30°C for L99A mutants. T4L was purified using a two-step protocol and labeled as previously described, except that the 3-(2-Iodoacetamide)-proxyl was used as the spin label ¹⁷. T4L concentrations were determined using an extinction coefficient of 1.23 cm²/mg for the single bimane (or spin-labeled) mutants and 1.55 cm²/mg for the bimane/trp pair mutants after subtracting the bimane contribution to the absorbance. Four L99A mutants expressed into inclusion bodies (K83W/G113C, K16W/V57C, L39C, Y25C/L39C). These mutants were expressed at 37°C and the proteins were refolded as previously described ¹⁷.

Both αA- and the triply phosphorylated analog (S19D/S45D/S59D) of αB-crystallin, called αB-D3, were purified as previously described ^17,18. Protein concentrations for αA- and αB-D3 were determined using an extinction coefficient of 0.834 and 0.947 cm²/mg, respectively.

Thermodynamic Analysis of T4L mutants

Denaturant-unfolding curves were constructed by monitoring tryptophan and bimane fluorescence as a function of guanidinium HCl at 37°C in pH 7.2 buffer on a Photon Technology International (PTI) L-format spectrofluorometer. ΔG_unf of L99A/V57Bi/K16W was determined at 30°C due to aggregation and precipitation at 37°C. The curves were fit with a two-state unfolding model using nonlinear least squares methods.

Bimane Fluorescence Binding Analysis

Binding curves for each bimane-labeled mutant were obtained by incubating defined ratios of α-crystallin:T4L for 2hrs at 37°C in pH 7.2 buffer and collecting the bimane emission spectrum on a PTI L-format spectrophotometer with excitation and emission slit widths of 1 nm and 0.5 nm, respectively. The emission spectrum was collected from 410 to 500 nm after excitation of bimane at 380 nm. Binding of L99A-V57Bi/K16W was performed at 30°C. Binding isotherms were generated by plotting bimane emission at 485 nm as a function of the [α-crystallin/T4L] ratio. The curves were fit using the Levenberg-Marquart nonlinear least squares method in the program Origin (OriginLab Inc.) ¹⁸.

EPR Spectroscopy

Analysis of spin-labeled T4L was carried out on a Bruker EMX spectrometer at room temperature in pH 7.2 buffer. For binding, 30 μM T4L was mixed with αA or αB-D3 at an α-crystallin:T4L ratio of 3:1 (low and high affinity binding) or 30:1 (high affinity binding) at 37°C for 2hrs. The complex was purified by SEC on a Superose 6 column to ensure complex formation and to remove any residual free T4L. Following chromatography, the complex was concentrated to the original volume for binding. For quantitative solvent accessibility determination, power saturation experiments were performed with a Bruker Elexsys 500 spectrometer in the presence of 50mM NiEDDA.