Assembly landscape of the complete B-repeat superdomain from Staphylococcus epidermidis strain 1457

Abstract

The accumulation-associated protein (Aap) is the primary determinant of Staphylococcus epidermidis device-related infections. The B-repeat superdomain is responsible for intercellular adhesion that leads to the development of biofilms occurring in such infections. It was recently demonstrated that Zn-induced B-repeat assembly leads to formation of functional amyloid fibrils, which offer strength and stability to the biofilm. Rigorous biophysical studies of Aap B-repeats from S. epidermidis strain RP62A revealed Zn-induced assembly into stable, reversible dimers and tetramers, prior to aggregation into amyloid fibrils. Genetic manipulation is not tractable for many S. epidermidis strains, including RP62A; instead, many genetic studies have used strain 1457. Therefore, to better connect findings from biophysical and structural studies of B-repeats to in vivo studies, the B-repeat superdomain from strain 1457 was examined. Differences between the B-repeats from strains RP62A and 1457 include the number of B-repeats, which has been shown to play a critical role in assembly into amyloid fibrils, as well as the distribution of consensus and variant B-repeat subtypes, which differ in assembly competency and thermal stability. Detailed investigation of the Zn-induced assembly of the full B-repeat superdomain from strain 1457 was conducted using analytical ultracentrifugation. Whereas the previous construct from RP62A (Brpt5.5) formed a stable tetramer prior to aggregation, Brpt6.5 from 1457 forms extremely large stable species on the order of ≈28-mers, prior to aggregation into similar amyloid fibrils. Our data suggest that both assembly pathways may proceed through the same mechanism of dimerization and tetramerization, and both conclude with the formation of amyloid-like fibrils. Discussion of assembly behavior of B-repeats from different strains and of different length is provided with considerations of biological implications.

Significance

Staphylococcus epidermidis is a major pathogen responsible for device-related infections. The primary factor responsible for such infections is the accumulation-associated protein, Aap, through its ability to mediate formation of dense surface-adherent communities of bacteria known as biofilms. Our lab recently demonstrated that the B-repeat superdomain of Aap from strain RP62A undergoes Zn-dependent assembly to form functional amyloid fibrils that improve the strength and resilience of biofilms in vitro. These amyloid fibrils may be responsible for the difficulty in treating device-related infections. However, strain 1457 is commonly used for genetic manipulation. In this paper, the Zn-dependent assembly of B-repeats from strain 1457 is shown to lead to the same outcome of amyloidogenesis, although it occurs through different intermediate oligomeric states.

Introduction

Staphylococcus epidermidis has long been recognized as one of the most problematic organisms in medical device-related infections (1,2,3,4). S. epidermidis is a human commensal that is abundant on the skin but can also contaminate devices such as catheters, joint replacements, and even pacemakers after surgical implantation. Following bacterial adherence to a surface, the bacteria adhere to one another to form a resilient cluster known as a biofilm (3). While cell-to-cell accumulation can be mediated via production of the polysaccharide known as poly-N-acetylglucosamine (PNAG), data from a study of clinical isolates show the presence of the accumulation-associated protein (Aap) in 89% of prosthetic hip and knee joint infections (5). Furthermore, a recent study using a rat catheter model demonstrated that Aap is required for infection (6).

The cell-wall-anchored Aap, regardless of strain, has a well-conserved arrangement of domains (6,7). At the N-terminus, a series of short A-repeats with unknown function is present, followed by a lectin domain responsible for surface attachment (8,9,10) with SepA proteolytic cleavage sites on either side of the lectin domain (11). Downstream of the SepA site is the B-repeat superdomain containing 5–17 B-repeats, each composed of a G5 subdomain and spacer (also called E) subdomain. SepA cleavage occurs during biofilm formation, exposing the B-repeat superdomain and allowing for cell-to-cell adhesion via Zn²⁺-mediated B-repeat interactions (11,12,13). The B-repeat superdomain protrudes outward from the cell surface due to a highly extended stalk region that is rich in proline and glycine (14,15). At the C-terminus, an LPXTG Sortase A motif is present, which anchors Aap to the peptidoglycan layer of the bacterial cell wall (16,17).

Extensive biophysical and structural work has been performed on Aap B-repeats from S. epidermidis strain RP62A (7,13,18,19,20,21,22). However, much of the functional in vivo work has been performed using strain 1457, given that it is more amenable to genetic manipulation (6,23). The goal of this study is to compare B-repeat assembly and amyloidogenesis of Aap B-repeats from strain 1457 to previous analyses of B-repeats from strain RP62A, such that results from biophysical and structural studies can be translated more appropriately to B-repeats from other strains and to plan future genetic manipulations of S. epidermidis to rigorously test the impact of B-repeat assembly and amyloidogenesis.

Specifically, we begin by examining the assembly of wild-type (WT) Brpt6.5 from strain 1457 in the presence of Zn²⁺. Interestingly, an extremely large stable oligomer is observed. As with Brpt3.5 and Brpt5.5 from RP62A Aap, Brpt6.5 also assembles into functional amyloid-like fibrils (21). To examine dimerization, a Brpt6.5 mutant with the E203 position of each B-repeat mutated to alanine was used. Consistent with RP62A Aap Brpt1.5 crystallography and biophysical data (18), the Brpt6.5 7xE203A mutant was unable to dimerize in the presence of Zn, indicating similar mechanisms of dimerization. The RP62A Aap Brpt5.5 tetramer was recently identified as a critical step in assembly and aggregation into fibrils via the H85 position of each B-repeat (20). Similarly, distinct H85A mutations in Brpt6.5 either prevent putative tetramer formation or allow tetramer formation but preclude amyloidogenesis. In conclusion, the B-repeats from Aap from strain 1457 utilize similar mechanisms of assembly and aggregation, with significant differences in early-stage assembly equilibrium constants and the stability of intermediates on the way to the terminal amyloid-like fibrils.

Materials and methods

Protein design

The Brpt6.5 WT, 6xH85A, and 7xE203A constructs were synthesized by Life Technologies, GeneArt (Thermo Fisher Scientific, Waltham, MA) and subcloned into the pENTR vector. The 6xH85A mutant contained the following mutations: H85A, H213A, H341A, H469A, H597A, and H725A. The 7xE203A mutant contained the following mutations: H75A, H203A, H331A, H459A, H587A, H715A, and E843A. The VC (variant sequence mutated to consensus sequence) construct was also synthesized by Life Technologies, GeneArt and subcloned into the pENTR vector. The VC mutations included N789D, D791N, K793A, E796T, R798K, and K800V. Residues V842 and E843 are already present in the WT sequence, so no sequence modification of these two residues was necessary for the VC construct. Each construct contained a tobacco etch virus (TEV) protease cleavage site directly upstream of the gene. At the C-terminus, an added glycine was followed by another TEV protease cleavage site (ENLYFQ-G) and Strep-tag (WSHPQFEK) using the QuikChange II Site-Directed Mutagenesis Kit (Agilent, Santa Clara, CA). Single H85A mutants of Brpt6.5 were produced from the parent construct using QuikChange mutagenesis as was previously done for Brpt5.5 (20), with the exception that here the alanine residues from the 6xH85A mutant were being mutated back to histidine. The LR Clonase reaction was used to transfer the gene of interest and add Strep-tag to a destination vector containing an N-terminal His₆-MBP tag. Aap Brpt6.5 from strain 1457 contained amino acids 608–1454 from UniProt entry UniProt: A0A075IHN3. Aap Brpt5.5 from strain RP62A contained amino acids 1505–2223 from UniProt entry UniProt: Q5HKE8. B-repeat constructs used in this study are summarized in Fig. S1.

Protein expression and purification

Brpt6.5 constructs were expressed in BLR(DE3) cells grown to an OD₆₀₀ of ≈0.8 at 37°C, shaking, in 1 L of LB medium containing ampicillin and tetracycline antibiotics to maintain the cell line and plasmid. The cultures were then chilled to 10°C in an ice bath. Ethanol was added to a final concentration of 2%, and protein production was induced with 200 μM isopropyl β-D-1-thiogalactopyranoside (IPTG). Cultures were incubated at 20°C overnight with shaking. The cultures were centrifuged to pellet the bacteria, which were then resuspended in 20 mM Tris (pH 7.4) and 300 mM NaCl with Roche protease inhibitor cocktail tablet. Resuspended pellets were stored at −80°C.

Pellets were lysed by sonication. The lysate was then centrifuged, and the supernatant was filtered using 0.45-μm and 0.20-μm bottle-top filters. Protein was purified using a 20-mL nickel-nitrilotriacetate (Ni-NTA) column (Cytiva, Marlborough, MA) attached to an Akta M FPLC system (GE Healthcare, Marlborough, MA) with a 300 mM imidazole elution buffer. The fractions containing eluted protein were pooled and dialyzed into 20 mM Tris (pH 7.4) and 150 mM NaCl overnight at 4°C before being loaded onto a 5-mL StrepTrap HP column (Cytiva) and eluted with 2.5 mM desthiobiotin (Sigma-Aldrich, St. Louis, MO). The pooled elution fractions were then incubated at room temperature, rocking or stirring gently, with TEV protease for approximately 18–24 h, adding additional protease after the first 6–16 h. The protein solution was then filtered and run over a 20-mL Ni-NTA column to collect the flow-through, followed by the 5-mL StrepTrap HP column, again collecting the flow-through. When deemed necessary based on SDS-PAGE or analytical ultracentrifugation (AUC), an additional purification step using a 24-mL Superose 6 size-exclusion column (Cytiva) was implemented.

Circular dichroism

Circular dichroism (CD) experiments were performed on an Aviv 215 CD spectrophotometer with a Peltier junction temperature control system. Far-UV wavelength scans were performed using a 0.1-mm quartz cuvette at 20°C, with 5 scans being averaged and a 3-s averaging time at each wavelength step. The error in CD measurement is reported by the instrument as one standard deviation, determined during the 3-s averaging time at a given wavelength. The plotted error bars are the propagated errors across the five replicate scans. Protein was at 1.5 mg/mL in 50 mM 3-(N-morpholino)propanesulfonic acid (MOPS) (pH 7.2) and 50 mM NaCl. The conversion from machine units (millidegrees, θ) to mean residue ellipticity [θ], degrees cm² dmol⁻¹ residue⁻¹, was performed using Eq. 1:

$$\left[\theta \right]=\frac{\theta \times \text{MRW}}{10\times l\times c}\text{.}$$ (1)

MRW is the mean residue weight for WT (107.4 g mol⁻¹ residue⁻¹), VC (107.2 g mol⁻¹ residue⁻¹), 7xE203A (106.9 g mol⁻¹ residue⁻¹), or 6xH85A (106.9 g mol⁻¹ residue⁻¹). The pathlength, l, was in centimeters and concentration in mg/mL.

Thermal denaturation experiments were performed in the same cuvette and buffer and at the same protein concentration. The experiments followed the signal at 205 nm from 20°C to 80°C in 1°C steps. The averaging time at each step was 10 s. Where direct comparisons are made in the figures, each sample was dialyzed into the same buffer to ensure accurate comparisons were made, as pH and salt concentration differences may cause significant effects on thermal stability. Thermal denaturation experiments were analyzed in SigmaPlot 12.5 (Systat Software/Inpixon, Palo Alto, CA) using a two-state model correcting for pre- and post-transition linear changes (24). It was assumed that there was no change in heat capacity. Data are presented as fraction folded, according to Eq. 2:

$$\alpha =\frac{\left({\theta }_{\mathrm{t}}-{\theta }_{\mathrm{U}}\right)}{\left({\theta }_{\mathrm{F}}-{\theta }_{\mathrm{U}}\right)}\text{,}$$ (2)

where θ_t Is the MRE at a given temperature, θ_F is the MRE of the folded monomer, and θ_U is the MRE of the unfolded monomer (24).

Temperature-dependent wavelength experiments were carried out with 1.8 mg/mL protein in 50 mM MOPS (pH 7.2) and 50 mM NaCl with 5 mM ZnCl₂ added directly to the sample. Samples were examined in a 0.5-mm cuvette. A macro was used to run experiments sequentially without any timed manual input. Scans were single replicates with 3 s averaging time at each wavelength from 300 nm to 190 nm, at 10°C increments from 20°C to 80°C, then another scan after cooling to 20°C.

Temperature-dependent aggregation experiments were performed as previously described (20), but with 0.50 mg/mL Brpt6.5. Prior to the start of the experiment, 3.50 mM ZnCl₂ was added directly to the sample. A wavelength of 225 nm was used to monitor formation of amyloid-like aggregate formed by WT, and a wavelength of 218 nm was used to follow unfolding of the Brpt6.5 6xH85A mutant.

Sedimentation velocity analytical ultracentrifugation

AUC experiments were performed using a ProteomeLab XL-I (Beckman Coulter Life Sciences, Indianapolis, IN). Unless otherwise specified, sedimentation velocity experiments were run at 48,000 rpm using Spin Analytical (Berwick, ME) 1.2-cm meniscus-matching centerpieces and sapphire windows at 20°C. All velocity data were collected using interference optics unless otherwise noted. To match menisci, a short (≈5–10 min) spin at 3000–5000 rpm was used while monitoring the fringe display window to ensure successful transfer of buffer to the protein sector. The cells were removed from the rotor and gently rocked by hand to redistribute the protein gradient, then carefully realigned in the rotor. The AUC was allowed to equilibrate at least until the temperature display read 20.0°C for several minutes. The experiments were allowed to run overnight, usually at least 16 h, or until sedimentation of protein appeared complete. Sedimentation velocity data were analyzed by wide distribution analysis (WDA) within SEDANAL (25,26) or using the c(s) distribution model within SEDFIT (27). Sedimentation velocity experiments were conducted using protein samples at ≈0.5 mg/mL unless otherwise specified; some variation occurred due to dialysis of samples before loading into the centrifuge. The viscosity and density of buffer was estimated using SEDNTERP v.3.0.3 (28,29). In the case of MOPS buffers, the MOPS contribution was ignored due to the absence of this component in the database. The partial specific volume of the protein was estimated from the amino acid sequence using SEDNTERP v.3.0.3 (28,29).

The program SViMULATE v.1.0.2 (30) was used for hydrodynamic calculations between the sedimentation coefficient (S), frictional ratio (f/f₀), molar mass (M), and the diffusion coefficient (D). The appropriate partial specific volume, density, and viscosity were included as estimated by SEDNTERP v.3.0.3 (28,29).

SEDNTERP v.3.0.3 was also used to determine the hydrodynamic nonideality coefficient (k_s) according to Eq. 3:

$$\frac{1}{s\left(c\right)}=\frac{1}{s_0}\left(1+k_{\mathrm{s}}c\right)\text{,}$$ (3)

where s(c) is the concentration dependence of the sedimentation coefficient, s₀ is the sedimentation coefficient at zero concentration, c is the concentration, and k_s is the hydrodynamic nonideality coefficient. The values for s₀, k_s, and error estimates are listed in Table S1.

Sedimentation equilibrium analytical ultracentrifugation

Sedimentation equilibrium data were collected at 4000 rpm using absorbance optics at 233 nm, 240 nm, and 277 nm. Six-channel centerpieces were used. Loading concentrations were approximately 0.5, 0.3, and 0.1 mg/mL Brpt6.5 WT in 50 mM MOPS (pH 7.2), 50 mM NaCl, and 3.50 mM ZnCl₂. Equilibrium was verified by the MATCH utility within HeteroAnalysis (31). Data were analyzed within SEDANAL (32) using a model containing a monomer with a fixed molar mass, along with a second non-interacting ideal species.

Dynamic light scattering

Dynamic light scattering (DLS) experiments used a Malvern Zen 3600 Zetasizer Nano and a low-volume quartz cuvette. Brpt6.5 samples were adjusted to 0.50 mg/mL in 50 mM MOPS (pH 7.2) and 50 mM NaCl, filtered using a 0.2-μm syringe filter, and had 3.50 mM of filtered ZnCl₂ added just prior to the experiment. The DLS method took 2°C temperature steps with 120 s of equilibration time. Each measurement was made in triplicate (technical replicates) with automatic measurement duration. Changes in viscosity were considered at each temperature step in the Zetasizer program. The hydrodynamic radius (R_h) was directly reported by the Zetasizer program. The correlation data were then exported to SEDFIT v.16.36 using the “MakeDLSDAT-v1-0” macro written by Ulf Nobbmann (www.materials-talks.com). Correlation data were analyzed using the continuous I(D) distribution in SEDFIT to obtain the apparent diffusion coefficient (D^∗) of the primary peak. This peak contained each Brpt6.5 species present, meaning it is a weighted-average D^∗ of the species present. The D^∗ was then standardized to D_20,w using the following equation via SEDNTERP v.3.0.3 (28,29):

$$D_{20,\mathrm{w}}=D^{\ast }\frac{{\eta }_{\mathrm{T},\mathrm{b}}}{{\eta }_{20,\mathrm{w}}}\frac{293.15}{T}\text{,}$$ (4)

where η is the viscosity of the buffer at the experimental temperature (η_T,b) or of water at 20°C (η_20,w). The temperature of the measurement was converted to Kelvin for use in the standardization.

Turbidity assay

Brpt6.5 samples were dialyzed into the same buffer preparation of 50 mM MOPS (pH 7.2) and 50 mM NaCl and adjusted to 0.50 mg/mL. Triplicate samples of protein and dialysis buffer were prepared. A BioMate 3S UV-visible spectrophotometer (Thermo Fisher Scientific) was set to the multi-wavelength mode (280 nm, 400 nm, and 700 nm) and blanked with dialysis buffer. A 200-μL sample was transferred to a quartz low-volume cuvette, and a scan was taken to ensure expected baseline readings near 0.000 at 400 nm and 700 nm, with a 280 nm reading indicative of 0.5 mg/mL protein (or zero for buffer). Next, 30 1-μL aliquots of a 500-mM ZnCl₂ solution were added to the sample with gentle shaking between each addition and scan. The spectrophotometer measurements were taken at room temperature. This resulted in a final ZnCl₂ concentration of approximately 65 mM. The data collected at 400 nm are presented, as these values were within the linear range of the spectrophotometer (<1.0 OD) and showed greater signal than that observed at 700 nm, without any ZnCl₂ absorbance that was observed at 280 nm.

Thioflavin T assay

Samples of Brpt6.5 WT dialyzed into 50 mM MOPS (pH 7.2) and 50 mM NaCl were added to wells of a 96-well plate at a concentration of 4 mg/mL. Four replicate wells were set up for each condition. Data presented were collected after 24 h of shaking at 37°C at 200 rpm, with the plate sealed. Thioflavin T (ThT) was added to a final concentration of 20 μM at the start of the experiment. The total volume of each well was 100 μL. The plates were scanned using a BioTek Synergy plate reader (Agilent) with an excitation wavelength of 440 nm and emission wavelength of 482 nm, using the bottom optics position and a gain of 100. To test for significance, a Student’s t-test was performed in Microsoft Excel (two-tailed, two-sample unequal variance). Significance was identified when p < 0.05.

Transmission electron microscopy

A 5-μL sample was taken from the ThT assay at the 24-h time point. The sample was added to a 200-mesh Formvar carbon/copper grid. After 2 min, the grid was washed with 400 μL of Milli-Q water, and Whatman filter paper was used to wick the liquid off the grid. The grid was then stained for 30 s using 1% uranyl acetate that was added dropwise. The grid was then washed again with 400 μL of Milli-Q Water before being dried with filter paper and left to air-dry at least overnight. Images were captured using an AMT 2k CCD camera on a Hitachi 7600 microscope (Hitachi, Schaumberg, IL) at an accelerating voltage of 80 kV.

Results

A comparison of Aap from strain 1457 and RP62A

The overall domain arrangement of Aap is conserved across strain 1457 and RP62A, as shown in Figs. 1 A and S2. The primary differences between these and other strains are the number of B-repeats (6,7). In addition, the sequence of each B-repeat may be classified as a “consensus” or “variant” subtype. In the variant repeats, there are eight residues in the G5 subdomain of the B-repeat that replace those found in the consensus (Fig. 1 B (purple) and Fig. S1). Based on previous studies using Brpt1.5 constructs, it was found that consensus B-repeats were better able to assemble in the presence of Zn²⁺, while variant B-repeats conferred higher thermal stability of the folded monomer (7). Brpt6.5 from strain 1457 Aap is composed of six full consensus repeats that share 90%–100% sequence identity, with the C-terminal half-repeat (G5 subdomain) being the variant subtype. Interestingly, Shelton et al. observed a significant position-dependent effect of the variant sequence in the context of Brpt1.5, with the C-terminal G5 subdomain being the major determinant of Zn²⁺-dependent assembly and thermal stability (7). Thus, while Brpt6.5 (1457) and Brpt5.5 (RP62A) are both predominantly consensus, the presence of a variant B-repeat in the C-terminal half-repeat position in Aap from 1457 could have an impact on assembly or stability.

Figure 1.

A comparison of Aap from S. epidermidis strain RP62A and 1457. (A) Domain arrangements of Aap from each strain. B-repeats are colored according to whether the sequence cassette is consensus or variant, as described in the text (7). (B) Examples of the consensus and variant B-repeats are shown. Residues highlighted in purple represent the residues present in the variant cassette. Aap from strain 1457 only contains a variant cassette in the C-terminal half-repeat. The C-terminal half-repeat from Aap from strain RP62A is consensus. In both cases, the C-terminal half-repeat has lesser identity with the other G5 subdomains of Aap (highlighted in blue). Residue positions found to coordinate Zn²⁺ during dimerization are marked with an asterisk (^∗) and highlighted in green, while residue H85 is marked with a caret ( ˆ ) and highlighted in red to denote its involvement in tetramerization.

Brpt6.5 secondary structure and stability

B-repeats from strain RP62A Aap exhibit β-sheet and random coil content, regardless of the G5 sequence type (consensus or variant) or the number of B-repeats in the construct (7,13,18,20,21). The far-UV CD spectrum of Brpt6.5 WT (without Zn²⁺ present) was comparable to those observed previously for RP62A Aap B-repeats (Fig. 2 A). Thermal denaturation data were fitted to a two-state transition. Brpt6.5 WT showed an apparent T_m of 55.1°C (Fig. 2 B and Table 1), with a clear loss of β-sheet signal evident at 80°C relating to the fully unfolded state (Fig. 2 A). Additional constructs (Brpt6.5 7xE203A and Brpt6.5 6xH85A) were produced to analyze the mechanisms of Brpt6.5 oligomerization (described below). The Brpt6.5 7xE203A mutant contained similar secondary structure content and similar folding stability compared to WT (Fig. 2 and Table 1). The Brpt6.5 6xH85A also showed similar secondary structure content (Fig. 2 A) but showed a significant decrease in apparent T_m from 55.1°C to 50.5°C (Fig. 2 B). This is consistent with observations for Brpt5.5 5xH85A mutant from strain RP62A, as the H85 residue in each B-repeat is involved in a hydrogen-bonding network (20,22). Overall, the WT, 7xE203A, and 6xH85A constructs are properly folded, consistent with the expected secondary structure content, and are stable up to 50°C–55°C.

Figure 2.

Brpt6.5 WT and mutants are properly folded. (A) Far-UV CD wavelength scans showing similar secondary structure content across all constructs. Dark-colored lines were obtained at 20°C, while light-colored lines were obtained at 80°C. Error bars show ±1 standard deviation (SD) propagated over five replicate wavelength scans. (B) Thermal denaturation CD experiments monitoring the CD signal at 205 nm. Each dataset was fit to a two-state model. The markers indicate data points, while the solid lines show the best fit. Brpt6.5 6xH85A shows a significant decrease in thermal stability. Data are shown as fraction folded. CD data were collected in 50 mM MOPS (pH 7.2) and 50 mM NaCl at 1.5 mg/mL protein.

Table 1.

CD thermal denaturation results

Sample	Apparent Tm (°C)
WT	55.1 ± 0.2
7xE203A	54.8 ± 0.2
6xH85A	50.5 ± 0.5

Determined at 1.5 mg/mL Brpt6.5 WT in 50 mM MOPS (pH 7.2) and 50 mM NaCl in a 0.1-mm cuvette.

Zn²⁺-induced assembly of Brpt6.5 WT

In the context of Aap from RP62A, a major function of the B-repeat superdomain is to mediate intercellular interaction of adjacent cells in the developing biofilm. This occurs via a Zn²⁺-dependent mechanism that includes dimer and tetramer assembly states. The tetramer state can then undergo a conformational change that triggers aggregation into functional amyloid fibrils, which apparently function to further stabilize the biofilm. The Staphylococcus aureus ortholog of Aap, known as SasG, has also been shown to use Zn²⁺ in its role in intercellular adhesion (33). The relevance of these Zn²⁺-dependent interactions was confirmed in biofilms via the ability for DTPA (diethylenetriaminepentaacetate, a metal chelator) to inhibit or disrupt biofilm formation in S. epidermidis RP62A expressing Aap or S. aureus USA300 expressing SasG (13,21). Zn²⁺-dependent heterophilic interaction between Aap from strain 1457 and SasG from S. aureus strain SH1000 has been shown via single-cell force measurements (33). Based on this observation, Aap from strain 1457 is likely to mediate intercellular adhesion in biofilms via the same Zn²⁺-dependent mechanism as described for strain RP62A (13,21).

The ability of the WT Brpt6.5 construct from strain 1457 Aap to assemble in the presence of Zn²⁺ was evaluated using sedimentation velocity AUC. In the absence of Zn²⁺, Brpt6.5 WT sedimented relatively slowly at 48,000 rpm (Fig. 3), with a sedimentation coefficient (2.26 S) only slightly higher than that of Brpt5.5 (2.20 S (20)) and an estimated molar mass near that calculated from the sequence. Brpt6.5, like Brpt5.5, has an unusually large frictional ratio due to the extended conformation of the B-repeats (f/f₀ = 3.48; Fig. 3). In the presence of Zn²⁺, Brpt6.5 WT formed very large oligomers that had not been observed previously for other B-repeat constructs from strain RP62A Aap (Fig. 4). To determine the best experimental conditions with which to collect data on this system, a series of velocity experiments was run on Brpt6.5 + 3.50 mM ZnCl₂ at speeds ranging from 18,000 rpm to 60,000 rpm. At the two lower speeds, many scans can be collected for the faster-sedimenting species; however, the slower-sedimenting species (monomer) does not completely sediment (Fig. 4, A and B). At higher speeds the monomer fully sediments, but there is less time for data collection before the larger species sediment (Fig. 4, C and D). Analyses by both SEDANAL WDA and SEDFIT show strong signal relating to the ≈26 S species, with the SEDFIT c(s) distribution showing a major and a minor peak near ≈26 S at higher speeds.

Figure 3.

Brpt6.5 WT is monomeric in the absence of Zn²⁺. (A) Sedimentation velocity boundaries of Brpt6.5 WT at 48,000 rpm without ZnCl₂. The SEDFIT c(s) fit lines are colored from purple (early time scans) to red (late time scans). Data are shown as black markers behind the fit lines, with residuals shown in the bottom panel. The sedimentation velocity data were analyzed by SEDFIT c(s) (B). Data were collected in 50 mM MOPS (pH 7.2) and 50 mM NaCl at 0.5 mg/mL protein. Inset table shows parameters from the fit using the SEDFIT c(s) model; the expected molar mass based on sequence is 91.7 kDa.

Figure 4.

Brpt6.5 forms large oligomers in the presence of Zn²⁺. Sedimentation boundaries collected during sedimentation velocity AUC experiments of 0.5 mg/mL Brpt6.5 WT in the presence of 3.50 mM ZnCl₂ at (A) 18,000 rpm, (B) 32,000 rpm, (C) 48,000 rpm, and (D) 60,000 rpm. Data presented were collected using interference optics. The scan colors indicate time of data collection, from early time points (purple) to late time points (red). (A)–(D) were generated using GUSSI (34) after SEDFIT c(s) analysis. Both the TI and RI noise were removed. The data sets shown in (A)–(D) were analyzed by (E) SEDFIT c(s) analysis (27) or (F) SEDANAL WDA (25,26).

A series of experiments at varying protein and Zn²⁺ concentrations was conducted to better characterize the assembly of oligomers sedimenting near 26 S. First, Brpt6.5 WT was evaluated in the absence of Zn²⁺ across a range of protein concentrations. The lack of shift in the c(s) distribution toward higher s values (Fig. 5 A) and the decrease in s with increasing concentration (Fig. 5 C) indicates that Brpt6.5 WT does not assemble in the absence of Zn²⁺ but shows hydrodynamic nonideality that slows sedimentation (Table S1). This is consistent with previous reports for RP62A Aap Brpt1.5 and Brpt5.5 (7,13,20,35). A series of panels shows the c(s) distributions observed for Brpt6.5 WT with increasing amounts of ZnCl₂ (Fig. 5, D and E, black lines). The emergence of a boundary near 26 S was detectable at 1.75 mM ZnCl₂ and above. Experiments at 3.50 mM ZnCl₂ with Brpt6.5 loading concentrations ranging from 0.1 to 0.7 mg/mL revealed a stationary boundary near 26 S, indicating the presence of discrete species (Fig. S4).

Figure 5.

Assembly of the WT and VC mutant. A range of concentrations of Brpt6.5 WT (A) and Brpt6.5 VC (B) were examined in the absence of ZnCl₂. The resulting s_w values are plotted against protein concentration in (C). Brpt6.5 WT and VC mutant were then examined over a wide range of ZnCl₂ concentrations. (D) and (E) show the c(s) distributions with 0.5 mg/mL WT in black and VC in red. At 8.00 mM ZnCl₂, VC was too aggregated after dialysis into the buffer to perform the experiment. Dashed lines are shown at s = 2.2 S (monomer) and s = 26 S for reference. (F) The s_w obtained from c(s) distributions in (D) and (E) plotted against ZnCl₂ concentration. Conditions where significant loss of protein from aggregation was observed during dialysis are shown with red “+” symbols. (G) A bar chart shows the protein concentration in each experiment determined by integration of the area under the c(s) distribution from approximately 1 S–38 S. For (F) and (G), the WT data are shown in black, while the VC data are shown in red.

To better quantify the Zn²⁺-dependent sedimentation data, the weight-averaged sedimentation coefficient (s_w) was calculated over the entirety of each c(s) distribution (Fig. 5 F). A slight drop in s_w is observed at high Zn²⁺ concentrations due to aggregation of the larger oligomer(s) into insoluble species that were lost during dialysis (Fig. 5, F and G). Interestingly, in the lower S-value range, the monomer species is predominant until 6.50–8.00 mM ZnCl₂, where species up to ≈7 S are observed. This is a similar s range over which the Brpt5.5 dimer and tetramer were observed in previous reports (20,22). These observations suggest that the ≈26 S species appears to be thermodynamically stable intermediate(s) along the pathway to amyloidogenesis rather than a transient, heterogeneous set of oligomeric states (36,37). The paucity of species observed between monomer (2 S) and ≈26 S and the stationary boundary observed indicate cooperative assembly that favors formation of a terminal oligomer, as observed for other systems such as Mg²⁺-dependent assembly of tubulin (38,39).

Investigating the role of the C-terminal half-repeat

The previously characterized Brpt5.5 construct formed monomer (≈2 S), dimer (≈4 S), and tetramer (≈7 S) species in the presence of Zn²⁺ before aggregating into fibrils, which sedimented essentially immediately in AUC (14,21,22). We hypothesized that the formation of the ≈26 S intermediate for Brpt6.5 could be attributed to either the additional B-repeat or the presence of a variant cassette in the C-terminal half-repeat (7). To test the latter possibility, a mutant Brpt6.5 was designed where the C-terminal G5 subdomain variant sequence was swapped for the consensus sequence, resulting in a completely consensus Brpt6.5 construct (referred to as Brpt6.5 VC: Variant sequence mutated to Consensus sequence). The Brpt6.5 VC construct showed proper folding by CD with similar thermal stability as Brpt6.5 WT (Fig. S5 and Table S2). In the absence of Zn²⁺, Brpt6.5 VC remained monomeric and showed similar hydrodynamic nonideality (Fig. 5, B and C; Table S1). In the presence of Zn²⁺, Brpt6.5 VC was still capable of forming the ≈26 S species (Fig. 5, D–F). In fact, the VC mutation seemed to slightly encourage the formation of the ≈26 S species compared to WT, as can be seen in the higher s_w values at 1.75–3.50 mM Zn²⁺ (Fig. 5 F). However, the major difference observed between Brpt6.5 WT and VC is the propensity to form insoluble aggregates (or amyloid fibrils) at higher Zn²⁺ concentrations (Fig. 5 G). In conclusion, the variant cassette in the C-terminal half-repeat is not responsible for formation of the ≈26 S species. While it seems likely that the presence of the additional B-repeat is responsible for the formation of the stable intermediate(s) in the presence of Zn²⁺, there are other differences that remain between the Brpt5.5 and Brpt6.5 constructs, including the presence of a variant B-repeat in Brpt5.5 (Fig. 1 A; see B-repeat 8). Additional mutants will be needed to parse out the contribution(s) of each of the differences.

Characterization of the large Zn²⁺-induced species of Brpt6.5

To obtain more information about the identity of the ≈26 S species, sedimentation equilibrium AUC (EQ-AUC) experiments were conducted. While sedimentation velocity (SV-AUC) experiments can provide accurate molar mass estimates in many cases, there are possible complications when samples contain multiple species, especially when the species are interacting. An EQ-AUC experiment was performed at 4000 rpm, such that the ≈26 S species would be able to form a concentration gradient rather than sediment completely. At such a low speed, the monomer does not form an appreciable gradient (Fig. S6), so the analysis could be focused more on the larger species. More importantly, equilibrium experiments determine molar mass of species directly without requiring an estimate of the shape or diffusion coefficient of the species, as is the case for SV-AUC. EQ-AUC data for Brpt6.5 WT at three loading concentrations between 0.5 mg/mL and 0.1 mg/mL with 3.50 mM ZnCl₂ at 4000 rpm (Fig. S7) were collected. The three datasets were fitted locally to a model containing the monomer and an additional species. No significant trend in molar mass of the large species was observed. This may indicate that the larger oligomer is not interacting reversibly on the timescale of the experiment; alternatively, it is also consistent with isodesmic assembly terminated by an energetically favorable final step (see discussion). A global fit to the same model revealed a good fit, with molar mass of the oligomer estimated at ≈2.63 MDa (Table S3). This mass is consistent with a 28-mer.

While EQ-AUC does not provide hydrodynamic information, the shape of the ≈26 S species could be estimated with the input of the EQ-AUC molar mass estimate. This yielded a diffusion coefficient of 0.875 cm²/s and frictional ratio (f/f₀) of ≈2.67, indicating that the oligomer is highly elongated. Furthermore, an analysis of SV-AUC data using the c(s) model with bimodal f/f₀ (Fig. S8) provided a similar f/f₀ estimate of ≈2.54, with molar mass estimates close to the experimental value from EQ-AUC.

The diffusion coefficient and hydrodynamic radius (R_h) of the ≈26 S species was measured by DLS. A sample of Brpt6.5 in the presence of 3.50 mM ZnCl₂ was examined at 20°C (Fig. S9, A and B). While SV-AUC was able to separate the monomer from the larger oligomer, DLS was unable to resolve the species. This resulted in the DLS mass-weighted and intensity-weighted distributions showing a single peak, which was composed of all oligomers present. The resulting diffusion coefficient was 0.856 cm²/s (comparable to the diffusion coefficient of the ≈26 S species) and the Z-average R_h at 20°C was 27.3 nm. At 38°C the distribution shifted, while still showing up as a single peak with the polydispersity index increasing to above 0.1, a diffusion coefficient of 0.641 cm²/s, and an R_h of 34.9 nm (Fig. S9, C and D; Table S4). As the temperature increased, the diffusion coefficient decreased and the R_h increased, consistent with assembly into to larger sized species. At 40°C, however, a major increase in the polydispersity index was observed, with distributions showing high heterogeneity indicative of aggregation (Fig. S9, E and F; Table S4). Interestingly, Brpt5.5 also aggregates in the same temperature range when present with the same ZnCl₂ concentration (20).

Inhibition of Brpt6.5 assembly via E203A mutations

The dimerization of B-repeat constructs from strain RP62A Aap has been very well characterized (7,13,18,19,20,40). Biophysical data on Brpt1.5 constructs first determined the importance of a histidine in the Zn²⁺-binding site, then showed that 1–2 Zn²⁺ ions are bound per G5 subdomain during dimerization (13,40). X-ray crystallography structures then provided high-resolution detail of the Zn²⁺-binding site and also confirmed that B-repeats occupy a highly extended global conformation as monomers and form a dimer in a side-by-side orientation with essentially no change in the conformation of the protein (7,18). Brpt5.5 was found to have a dimerization mechanism similar to that of Brpt1.5 and Brpt2.5 constructs, with 1–2 Zn²⁺ ions binding per G5 subdomain during dimerization. Biophysical and small-angle X-ray scattering (SAXS) analyses of the Brpt5.5 monomer and dimer confirmed a highly extended monomer and a similar side-by-side dimer (19,20,22). In the Brpt1.5 crystal structure, H75 and E203 are critical G5 domain residues that ligate Zn²⁺ in trans across the dimer interface (13,40). Extensive mutagenesis revealed that these same positions in each G5 domain have conserved function in supporting dimerization in the longer Brpt5.5 construct (20,22).

To understand the dimerization of Brpt6.5 from strain 1457 Aap, a construct was designed that would be unable to dimerize if the mechanism of dimerization is consistent with RP62A Aap B-repeats. This construct, referred to as Brpt6.5 7xE203A, contains the equivalent of the E203A substitution that can completely inhibit Brpt1.5 dimerization (18) in each G5 domain of Brpt6.5. The Brp6.5 7xE203A mutant was monomeric in the absence of Zn²⁺, as expected (Fig. 6, A and D). Furthermore, its ability to dimerize or form the ≈26 S species in the presence of Zn²⁺ was compromised (Fig. 6, B and E), with only a minor increase in the sedimentation coefficient that has consistently been observed previously and is likely the result of Zn²⁺ binding to the monomer (18,19). There was no significant increase in s_w with protein concentration in the presence of Zn²⁺ (Fig. 6, C and D) or with increasing Zn²⁺ concentrations (Fig. 6, B and E). However, given the hydrodynamic nonideality observed for Brpt6.5 in the absence of Zn²⁺ (Fig. 5 D), very weak dimerization of the 7xE203A mutant in the presence of Zn²⁺ cannot be ruled out. Given the identical ability of the H75A/E203A mutations to inhibit Zn²⁺-dependent dimerization, it is concluded that B-repeats from strain 1457 Aap undergo dimerization via a mechanism similar to B-repeats from strain RP62A Aap.

Figure 6.

Assembly of Brpt6.5 7xE203A. (A) Brpt6.5 7xE203A analyzed in the absence of Zn²⁺ at increasing protein concentrations (0.22 mg/mL, black line; 0.45 mg/mL, gray line; 0.89 mg/mL, red line; 1.37 mg/mL, blue line). (B) The c(s) distributions from samples of 0.5 mg/mL Brpt6.5 7xE203A at increasing ZnCl₂ concentrations (−ZnCl₂, black line; 3.50 mM ZnCl₂, gray line; 5.00 mM ZnCl₂, red line; 8.00 mM ZnCl₂, blue line). (C) Distributions shown are 7xE203A at increasing protein concentrations in the presence of 8 mM ZnCl₂ (0.22 mg/mL, gray line; 0.45 mg/mL, red line; 0.91 mg/mL, blue line; 1.32 mg/mL, green line). 7xE203A (0.45 mg/mL) without ZnCl₂ (black line) is shown for reference. Distributions for samples in 8 mM ZnCl₂ overlay very closely, while the −Zn distribution is shifted to a lower s value. The s_w value from the distributions in (A)–(C) are shown in (D) and (E). (D) shows the s_w relationship with protein concentration in the absence of ZnCl₂ and in the presence of 8.00 mM ZnCl₂. (E) shows s_w versus ZnCl₂ concentration while maintaining a constant 7xE203A concentration.

Inhibition of Brpt6.5 assembly via H85A mutations

Considering that inhibition of dimer via the E203A mutations also prevented formation of the ≈26 S species, we were curious as to whether we could also prevent the ≈26 S oligomer via a set of mutations that inhibited tetramer formation in Brpt5.5. This could provide useful evidence of whether the ≈26 S species is formed via a tetramer-dependent pathway or via an independent mechanism.

Recent reports have shown the critical importance of the Brpt5.5 tetramer for the ability of Aap to form functional amyloid fibrils (20,21). The Brpt5.5 tetramer is formed by two dimers coming together in a side-by-side fashion, and tetramerization is mediated via Zn²⁺ binding to a histidine in position 85 of the spacer subdomain of each B-repeat. These conclusions are based on rigorous characterization of single- and multi-H85A mutants via biophysical and SAXS analyses (20,22). It was noteworthy that Brpt1.5 and Brpt2.5 are not observed in the tetramer state or as aggregate, indicating that an important factor in tetramer assembly is the number of B-repeats present in the construct.

To understand potential tetramer assembly of Brpt6.5, a mutant was constructed containing the H85A mutation in all six spacer subdomains, called 6xH85A. Notably, the C-terminal half-repeat consists of only a G5 subdomain, leaving only six spacer subdomains (and therefore six H85 positions) in Brpt6.5. This construct is analogous to the Brpt5.5 5xH85A construct, which was limited in assembly to the dimer state by preventing Zn²⁺ coordination in trans via the H85 residues (20,22). The Brpt6.5 6xH85A mutant is monomeric in the absence of Zn²⁺ (Fig. 7, A and B). The Zn²⁺-dependent assembly of Brpt6.5 6xH85A was directly compared with the previously characterized Brpt5.5 5xH85A (Fig. 7, C and D). There is a clear difference in the s_w values between the two proteins, with Brpt5.5 reaching nearly complete dimer at 3.5 mM ZnCl₂, while Brpt6.5 appears to exhibit weaker assembly. EQ-AUC experiments confirmed a monomer-dimer equilibrium assembly for Brpt6.5 6xH85A (Fig. S10 and Table S5). Overall, the 7xE203A and 6xH85A data suggest that the ≈26 S species formed by Brpt6.5 may be formed along the same pathway via dimer and tetramer species intermediates, similar to those observed for Brpt5.5. In the next experiment, single-H85A mutants were utilized to discern more details regarding the Brpt6.5 assembly.

Figure 7.

Assembly of Brpt6.5 6xH85A. (A) shows c(s) distributions of Brpt6.5 6xH85A in the absence of Zn²⁺ at increasing protein concentrations (0.19 mg/mL, black line; 0.41 mg/mL, gray line; 0.80 mg/mL, red line; 1.32 mg/mL, blue line). (B) Shows the s_w plotted against protein concentration. (C) The s_w of Brpt6.5 6xH85A or Brpt5.5 5xH85A (strain RP62A) at multiple ZnCl₂ concentrations determined from c(s) distributions shown in (D). (D) The c(s) distributions for Brpt6.5 6xH85A are shown in black, while those for Brpt5.5 5xH85A are shown in red. AUC experiments were performed in 50 mM MOPS (pH 7.2) and 50 mM NaCl. Where ZnCl₂ is present it was included in the dialysis buffer, and the protein concentration was approximately 0.5 mg/mL.

Periodicity of assembly in single-H85A mutants

Recent work on Brpt5.5 from strain RP62A revealed that single-H85A mutants in every other B-repeat spacer subdomain could completely inhibit tetramer formation (22). This was consistent with the proposed tetramer structure being formed by two dimers oriented side by side, each of which twists along its long axis (20,22). This particular tetramer orientation places alternating H85 positions along the length of the construct at the interface of the two dimers, where Zn²⁺ ions can be coordinated across dimers to form the tetramer. The remaining H85 positions protrude away from the tetramer and can mediate downstream aggregation into fibrils (22).

To determine whether Brpt6.5 might utilize a similar tetramer-dependent assembly process, single-H85A mutants were constructed and their assembly examined by SV-AUC at several ZnCl₂ concentrations (Fig. 8). We observe that certain H85A mutants of Brpt6.5 can form a ≈6 S oligomeric species such as Brpt5.5 only when the larger ≈26 S species is absent. Specifically, in the presence of 8 mM ZnCl₂, the H213A, H469A, and H725A mutants assemble into a ≈6 S species with apparent molar mass from local c(s) analyses suggesting the formation of a tetramer (masses = 369–381 kDa; f/f₀ = 2.97–3.08). Overall, the assembly data reveal a periodicity similar to that seen for Brpt5.5 (22), where, for example, H213A supports formation of the putative tetramer (≈6 S) but H341A is limited to dimerization (≈4 S; compare Fig. 8 B or D with Fig. 7 D). Interestingly, an H85A mutation in either the N-terminal or C-terminal spacer subdomain is insufficient to prevent formation of the ≈26 S species (Fig. 8, A and F). This could indicate that these H85 positions are not in the interface of the ≈26 S species or that, compared to Brpt5.5, the extra B-repeat of Brpt6.5 reduces the importance of any single H85A mutation on overall assembly.

Figure 8.

Investigation of Brpt6.5 single-H85A mutants. The c(s) distribution obtained from sedimentation velocity AUC experiments are shown for each of the single-H85A mutants that compose the 6xH85A mutant. Experiments were performed in the absence of Zn²⁺ (black line) or in the presence of 3.50 mM ZnCl₂ (gray line), 5.00 mM ZnCl₂ (blue line), and 8.00 mM ZnCl₂ (red line). Data were collected in 50 mM MOPS (pH 7.2) and 50 mM NaCl at approximately 0.5 mg/mL protein. ZnCl₂ was dialyzed into the sample when present. In each panel a schematic of Brpt6.5 is shown, where purple blocks are consensus repeats, the blue block is the variant C-terminal half-repeat, and the red block is the location of the specific H85A mutation in the spacer region (see Fig. 1 for additional details). Brpt6.5 H85A (A) and H725A (F) show the presence of the large ≈26 S species (marked with an asterisk). A single H85A mutation in the third (B, H341A) or fifth (C, H597A) spacer subdomain prevent formation of the tetramer (≈7 S) under these conditions. In contrast, a single H85A mutation in the second (D, H213A) or fourth (E, H469A) spacer subdomain allow for formation of the tetramer but prevent formation of the ≈26 S species. Vertical dashed lines indicate the predicted sedimentation coefficients for Brpt6.5 dimer and tetramer species from hydrodynamic modeling (see Fig. S11 and Table S6).

While the presence of the ≈26 S species complicates this analysis slightly, the same H85A periodicity is observed for Brpt6.5 as with Brpt5.5, suggesting that Brpt6.5 also assembles via dimer and tetramer intermediate species that cooperatively assemble until the ≈26 S species is formed, given that no intermediate oligomeric species are observed for Brpt6.5 WT. These data also highlight the importance of the middle four H85 positions in formation of the ≈26 S species whether they were involved in tetramer formation or not.

Brpt6.5 aggregation and amyloidogenesis

A primary function of Aap B-repeats is to assemble and aggregate into amyloid fibrils. Characterization of Brpt3.5 and Brpt5.5 aggregation has been reported using several approaches (20,21). Presented in Fig. 9 A is a turbidity assay wherein small aliquots of ZnCl₂ are titrated into a sample of Brpt6.5 WT, 7xE203A, 6xH85A, or buffer, and the turbidity at 400 nm is reported after each addition. As Zn²⁺-induced aggregation occurs, turbidity increases. Brpt6.5 WT readily aggregates beyond ≈7–10 mM Zn²⁺, whereas the Brpt6.5 6xH85A requires much higher Zn²⁺ concentrations. These results are consistent with Brpt5.5 results previously published (20,22). Interestingly, Brpt6.5 7xE203A exhibits aggregation with a sigmoidal trend at intermediate ZnCl₂ concentrations despite its inability to dimerize. This is a critical observation that could suggest that B-repeats may bypass the E203A-mediated dimer when at higher Zn²⁺ concentrations, allowing aggregation to occur via H85-mediated assembly. It could be the case that the presence of H85 in each B-repeat is still able to mediate downstream aggregation in a Zn²⁺-dependent manner.

Figure 9.

Brpt6.5 from strain 1457 aggregates into amyloid-like fibrils. (A) Light scattering of each Brpt6.5 construct in the presence of increasing ZnCl₂. Triplicate measurements were performed, and the error bars show ±1 SD. Temperature-dependent aggregation of Brpt6.5 WT (B) and Brpt6.5 6xH85A (C) in the presence of Zn²⁺ was measured by DLS and the data presented as radius of hydration (R_h). The DLS data are presented as black markers (average of three measurements) with error bars showing ±1 SD. (B) and (C) also show the CD signal in millidegrees as red markers with error bars showing ±1 SD, propagated over five replicate scans. The CD wavelength monitored in (B) was 225 nm, while 218 nm was monitored in (C). (D) shows far-UV CD measurements for each construct in the presence of 5 mM ZnCl₂ in 10°C increments from 20°C to 80°C. Asterisks mark the presence of a 225-nm minimum observed for Brpt6.5 WT (40°C) and 7xE203A (50°C). The black line shows a scan taken at 20°C after heating to 80°C and indicates the degree to which aggregation was reversible. (E) Fluorescence of thioflavin T (ThT) measured for samples after 24 h at 37°C with shaking. The bar chart indicates the average of four replicates, with error bars showing ±1 SD. The markers indicate the four individual data points for each sample. Asterisks denote significant differences between Brpt6.5 WT without Zn²⁺ and either in the presence of 5 mM Zn²⁺ or 10 mM Zn²⁺ according to a two-tailed, two-sample unequal variance Student’s t-test with p < 0.05. (F) TEM micrographs of Brpt6.5 + 10 mM ZnCl₂ from the ThT assay shown in (E).

A second approach to characterizing the aggregation propensity of Brpt6.5 is to start with Zn²⁺ present in the sample and then use DLS to follow aggregation in a temperature-dependent fashion. In parallel, the same sample can be monitored by CD to follow the formation of a minimum at 225 nm that has been associated with the formation of amyloid fibrils (20,21,41,42). For Brpt6.5 WT (Fig. 9 B), both DLS-measured aggregation and the appearance of a minimum in the CD signal at 225 nm occur concomitantly at ≈42°C. On the contrary, Brpt6.5 6xH85A shows only a decrease in R_h corresponding to unfolding of the highly extended folded species to random coil near 52°C (Fig. 9 C). Following a similar trend, the CD data show the development of a minimum measured at 218 nm (Fig. 9 C), consistent with unfolding observed in the absence of Zn²⁺ (Fig. 2). Fig. 9 D shows far-UV CD wavelength scans conducted on samples similar to those in the DLS/CD experiments. The temperature was increased in 10°C increments, revealing the appearance of the 225 nm minimum for Brpt6.5 WT and Brpt6.5 7xE203A, whereas Brpt6.5 6xH85A showed a strong minimum at ≈212 nm indicative of unfolding. The 225 nm minimum was previously characterized for Brpt5.5 (20,21), where the development of the 225 nm minimum corresponded to formation of amyloid fibrils. As the temperature is increased, all samples aggregate and the overall signal intensity is lost. In agreement with the Zn²⁺ turbidity assay that showed 7xE203A required slightly higher Zn²⁺ concentrations than WT, the 7xE203A required slightly higher temperatures to induce the fibril-related conformational change. Furthermore, after cooling back down to 20°C, the WT sample does not recover the original CD signal. This is further evidence of an irreversible aggregation that is consistent with previous observations of Brpt5.5 amyloid fibrils. Brpt6.5 6xH85A, on the contrary, recovered its original CD signal, indicating a degree of reversibility of the aggregation formed at high temperatures after unfolding. The DLS/CD observations are nearly identical to those observed for Brpt5.5, suggesting a very similar H85-mediated aggregation mechanism despite the different distributions of oligomeric states between the two B-repeat constructs (20,21,22).

Lastly, the aggregate formed by Brpt6.5 was examined in comparison to that previously published for Brpt5.5 (21). A fluorescent amyloid-detecting dye, ThT, was used in conjunction with transmission electron microscopy (TEM) to show the formation of amyloid-like fibrils. Fig. 9 E demonstrates that a significant increase in ThT fluorescence is observed for Brpt6.5 WT in the presence of Zn²⁺ after 24 h of incubation at 37°C. The Brpt6.5 7xE203A and 6xH85A mutants under similar conditions did not show comparable levels of ThT fluorescence (Fig. S12). Examination of the ThT-positive sample of Brpt6.5 WT with 10 mM ZnCl₂ by TEM (Fig. 9 F) showed aggregates similar in morphology to those observed for Brpt5.5 and in native S. epidermidis RP62A biofilms (21).

Discussion

This article describes an investigation of the mechanism of assembly and amyloidogenesis of the complete Aap B-repeat superdomain from S. epidermidis strain 1457. This is the first report to use a protein construct containing the complete B-repeat superdomain of Aap or SasG (the S. aureus ortholog) for biophysical analyses. While mechanisms of B-repeat assembly and amyloidogenesis have been published previously, all these studies were performed using protein constructs based on B-repeats from Aap from S. epidermidis strain RP62A (7,13,18,19,20,21,40). Most of the biological experiments designed to understand the role of Aap have used S. epidermidis strain 1457, which is amenable to genetic manipulation (6,8,9,11,23,43). Therefore, it is important to understand whether there are differences in B-repeat assembly and amyloidogenesis across strains so that the effects of Aap or B-repeat mutations can be evaluated reliably among biophysical, structural, and functional experiments. This work fills that gap in knowledge.

The evaluation presented in this study indicates that B-repeats of Aap from strains RP62A and 1457 behave similarly in terms of overall Zn²⁺-induced assembly into amyloid-like fibrils. Mutants of 1457 Brpt6.5 behaved largely as predicted, suggesting similar mechanisms of assembly and amyloidogenesis. There are some very intriguing differences, however. One such difference is the concentration of Zn²⁺ required for dimerization between Brpt6.5 6xH85A and Brpt5.5 5xH85A. Based on previous work, the variant sequence in the C-terminal G5 subdomain was shown to weaken Zn²⁺-dependent dimerization (7), but that was in the context of a Brpt1.5 construct. Clearly, the impact is not as significant in the context of Brpt6.5. Also factoring into dimerization is the number of B-repeats, whereby constructs containing a higher number of B-repeats are expected to dimerize in the presence of lower Zn²⁺ concentration due to the chelate effect (19,44). The most obvious difference is the ability for Brpt6.5 to form very large oligomers in the range of ≈28-mers that can then undergo irreversible assembly into amyloid fibrils. On the other hand, Brpt5.5 has a terminal stable species of tetramer, which then undergoes aggregation into fibrils without any detectable larger stable intermediates (20). If Brpt5.5 can in fact form similar higher-order oligomers such as Brpt6.5, these species must be very transient such that they are rapidly pushed toward amyloid fibril assembly.

Sedimentation of WT Brpt6.5 in the presence of Zn²⁺ results in a slow-moving boundary corresponding to monomer and a fast-moving stationary boundary for the large oligomer at ≈26 S without any detectable intermediate oligomeric species, indicating cooperative assembly. This behavior is reminiscent of Mg²⁺-dependent assembly of tubulin, which proceeds via an isodesmic (i.e., indefinite linear) assembly mechanism with a highly favorable final assembly step that terminates in a closed ring structure (38,39). In the case of Brpt6.5, our data with the E203- and H85A-containing mutant constructs suggest that assembly proceeds in a similar manner to Brpt5.5, which involves Zn²⁺ coordination in trans at the E203 site to form dimers and assembly of dimers into tetramers via a separate Zn²⁺ coordination step at the H85 site (20,22). Thus, we speculate that Brpt6.5 may assemble in an isodesmic-like fashion via dimer and tetramer intermediates, with a favorable final assembly step once the oligomer reaches a ≈28-mer. It is possible that, like tubulin, the ≈26 S oligomer species for Brpt6.5 may also form a ring- or tube-like structure that would allow additional interactions for the final assembly event, leading to favorable energetics and the observed cooperative behavior. Future work will be directed toward determining structural aspects of these larger oligomers, including SAXS and cryogenic electron microscopy approaches.

Despite these intriguing differences in the oligomeric intermediates of B-repeat Zn²⁺-dependent assembly, both the Brpt5.5 and Brpt6.5 constructs aggregate into amyloid fibrils under very similar Zn²⁺ concentrations and temperatures (≈40°C). This is a very important consideration for the biological implications of B-repeat assembly. Although different strains of S. epidermidis express Aap containing a wide variety in the number and cassette identity of B-repeats, there appear to be natural mechanisms in place to control the landscape of assembly and balance out the ability of Aap to mediate biofilm formation only under the desired environmental conditions. It is well known that enzymatic cleavage of the A-repeats and lectin domain switches Aap from a host adhesion function (via to the lectin domain) to an intercellular adhesion function (via self-assembly of the B-repeat domain of Aap molecules anchored to the staphylococcal surface) (11,12,13). However, there is clearly also the potential for variation in the extent and mechanism of self-assembly due to strain-dependent variations in the B-repeat superdomain itself.

Implications of this work are that B-repeats from Aap across S. epidermidis strains may differ in the distribution of oligomeric states formed in the presence of Zn²⁺ but that they share the common characteristic of nucleating functional amyloid fibers, which is not surprising based on their high sequence identity and the presumed biological importance of functional amyloid for biofilm mechanical strength. The results of this study will allow mutations in Aap from strain 1457 to be interpreted in a molecularly coherent manner in future genetic manipulation studies. For example, to test the impact of Aap amyloidogenesis on biofilm formation, one could produce H85A mutations in each B-repeat of Aap. Another important observation from this work is that while the E203A mutations prevented dimerization, aggregation was still observed at intermediate Zn²⁺ concentrations. This means that to test the impact of dimerization on biofilm formation, a genetically modified S. epidermidis 1457 strain should contain both the E203A and H85A mutations in the Aap B-repeats. While the S. aureus SasG B-repeats have only been biophysically and structurally analyzed in the monomer state, this work provides a guide to anticipating which variables may be important to consider for Zn²⁺-mediated assembly and biofilm formation in B-repeat surface proteins found in other staphylococcal species.

Acknowledgments

The authors thank Dr. Catherine Shelton and Dr. John Burgner for offering comments on the manuscript. We also thank Dr. Peter Sherwood and Dr. Walter Stafford for implementing updates within SEDANAL that allow for loading and analysis of greater than 999 scans. This is particularly useful when using interference optics to collect data using an equilibrium method without any time delay between scans to perform velocity experiments (99 steps of 99 scans).

This work was supported by National Institutes of Health funding from the National Institute of General Medical Sciences (R35-GM151986) awarded to A.B.H. and funding from the University of Cincinnati Graduate School Dean’s Fellowship awarded to A.E.Y. (2018-2019 AY).

Author contributions

A.E.Y. designed experiments, performed experiments, analyzed data, and prepared the manuscript. A.B.H. designed experiments, prepared the manuscript, and provided funding.

Declaration of interests

A.B.H. served as a Scientific Advisory Board member for Hoth Therapeutics, Inc., holds equity in Hoth Therapeutics and Chelexa BioSciences, LLC, and was a co-inventor on seven patents broadly related to the subject matter of this work.

Editor: John Correia.