Ultrahigh and High Resolution Structures and Mutational Analysis of Monomeric Streptococcus pyogenes SpeB Reveal a Functional Role for the Glycine-rich C-terminal Loop

Background: SpeB is a protease secreted from the pathogenic bacterium Streptococcus pyogenes.

Results: Apo and inhibitor complex structures of SpeB reveal conformational changes associated with activation and binding.

Conclusion: The glycine-rich C-terminal active site loop is required for substrate recognition and product release.

Significance: Crystal structures and mutagenesis afford insights into the specificity of SpeB and the functional role of the active site loops in binding biological substrates.

Abstract

Cysteine protease SpeB is secreted from Streptococcus pyogenes and has been studied as a potential virulence factor since its identification almost 70 years ago. Here, we report the crystal structures of apo mature SpeB to 1.06 Å resolution as well as complexes with the general cysteine protease inhibitor trans-epoxysuccinyl-l-leucylamido(4-guanidino)butane and a novel substrate mimetic peptide inhibitor. These structures uncover conformational changes associated with maturation of SpeB from the inactive zymogen to its active form and identify the residues required for substrate binding. With the use of a newly developed fluorogenic tripeptide substrate to measure SpeB activity, we determined IC₅₀ values for trans-epoxysuccinyl-l-leucylamido(4-guanidino)butane and our new peptide inhibitor and the effects of mutations within the C-terminal active site loop. The structures and mutational analysis suggest that the conformational movements of the glycine-rich C-terminal loop are important for the recognition and recruitment of biological substrates and release of hydrolyzed products.

Introduction

The Gram-positive bacterium Streptococcus pyogenes is a highly disseminated human-specific pathogen that causes a wide range of diseases, including pharyngitis, tonsillitis, common skin rashes, rheumatic and scarlet fever, and meningitis (1). The majority of these bacterial infections remain localized to skin and soft tissue; however, many S. pyogenes strains can penetrate deeper into tissues and induce severely invasive diseases, such as sepsis, necrotizing fasciitis, and toxic shock syndrome, with high associated mortality (2). Despite our current arsenal of antibiotic treatments, S. pyogenes remains at the forefront of bacterial pathogens that cause significant morbidity and mortality throughout the world. As such, it is imperative to identify new approaches and methods to control these common, but often fatal, infections. S. pyogenes secretes a variety of documented virulence factors into the extracellular space that promote infectivity and down-regulate host immune responses. The cysteine protease SpeB is one such factor that has been extensively studied for its ability to promote S. pyogenes infection.

SpeB, or streptopain, is a highly conserved secreted cysteine protease that is found in the majority of S. pyogenes strains (3). Production, secretion, and activation of SpeB is highly regulated, and its proteolytic activity has been linked to immunomodulating activities during infection, including 1) degradation of host immunoglobulins to promote immune system evasion (4); 2) cleavage of the cytokine precursor interleukin-1β (IL-1β) to its mature form, resulting in inflammation and septic shock (5); and 3) release of the peptide hormone bradykinin from the precursor H-kininogen to produce hypotension during septic shock (6). Additionally, SpeB has been implicated in degradation of fibronectin and vitronectin, two host extracellular matrix proteins involved in tissue integrity (7), and in the liberation of proteins tethered to the streptococcal cell surface, including M proteins and C5a peptidase that are proposed to enhance S. pyogenes virulence (8). Consideration of the potential host targets and the careful regulation of SpeB proteolytic activity suggest an important role for the protease (9). However, due to conflicting results from human tissue samples and animal models, the relative importance of SpeB to S. pyogenes pathogenicity has been highly debated (10).

S. pyogenes exclusively infects humans, and active SpeB is secreted from the bacteria in individuals with invasive disease (11). Notwithstanding this protease secretion in human S. pyogenes infections and establishment of host and bacterial substrates, some data suggest an inverse relationship between SpeB production and disease severity (12–14). Some recent results have suggested that down-regulation of SpeB expression and inhibition of protease activity promote the accumulation and activation of host protease plasmin on the GAS bacterial surface, thereby promoting infectivity (13). Similarly, a clinical correlation has been reported between S. pyogenes invasive disease severity and diminished SpeB expression (12). Regardless of its potential role as a virulence factor, understanding of the structure and function of the secreted protease SpeB will help in elucidating the biology and chemistry at the interface between human epithelial cells and S. pyogenes and potentially provide insight into the role of proteases secreted from other bacteria.

SpeB, like most extracellular proteases, is synthesized as an inactive zymogen to protect the intracellular components of the bacteria from proteolytic activity during protein production. SpeB is composed of an N-terminal signal sequence (residues 1–27), a prodomain (residues 28–145), and a catalytic C-terminal region (residues 146–398). Once secreted into the extracellular milieu, SpeB is susceptible for maturation to the active protease by autocatalysis, mature SpeB, trypsin, and subtilisin (15–18). Mutational analysis and structural studies have already revealed residues essential for catalytic activity (Cys¹⁹² and His³⁴⁰) (19) and substrate binding (Trp³⁵⁷ and Trp³⁵⁹) (20–22). These studies, in combination with kinetic interrogation, have identified the preferred peptide sequences targeted by SpeB for cleavage (17). Based on this information and the sequence of the SpeB residues hydrolyzed in its self-activation, we have designed and synthesized a fluorogenic tripeptide substrate, acetyl-Ala-Ile-Lys-7-aminomethylcoumarin (Ac-AIK-AMC),² which demonstrates robust catalytic turnover by SpeB, and a specific peptide inhibitor acetyl-Ala-Gln-Ile-(S)-2,6-diaminohexanal (Ac-AEIK-CHO). We then determined the crystal structure of mature SpeB with and without our peptide inhibitor. This structural information would help to elucidate any conformational changes associated with active site binding and afford an understanding of the interactions between the protease and potential substrates as well as how to target the protease for specific small molecule drug discovery.

Structures of the SpeB zymogen and mature SpeB have been previously determined by both x-ray crystallography and NMR. The protease has a canonical papain-like fold (20–22). Here, we report the highest resolution crystal structure to date of mature apo-SpeB at 1.06 Å resolution from a crystal obtained in neutral pH conditions and high resolution crystal structures of SpeB in complex with the general proteinase inhibitor trans-epoxysuccinyl-l-leucylamido(4-guanidino)butane (E64) and in complex with the SpeB-specific N-terminal acetylated peptide aldehyde inhibitor Ac-AEIK-CHO. These structures elucidate the residues essential for substrate binding on the nonprime side of the SpeB active site and the conformational changes of the protein associated with binding. Unexpectedly, these crystal structures provide the first view of a “floppy” glycine-rich C-terminal loop, adjacent to the active site, in both the apo and complex isoforms. With this structural information, we performed alanine-scanning mutagenesis on all of the loop glycine residues to understand the functional role of this large and unique loop that is not found in other papain-like proteases. Our structural and biochemical results provide insight into the structure, function, and broad substrate specificity of SpeB and new information that could aid in the development of small molecules capable of the specific regulation of SpeB activity.

EXPERIMENTAL PROCEDURES

Protein Preparation and Purification

The S. pyogenes 10782 zymogen SpeB clone (residues 28–398) was generated without the N-terminal secretion leader sequence (residues 1–27) using standard PCR-based cloning and verified via double-stranded plasmid sequencing. SpeB is overexpressed as a C-terminal His₆ tag fusion (additional amino acids for the affinity tag include LEH₆, residues 399–406) from Escherichia coli BL21DE3pLysS (Stratagene) in a pET23b vector (Novagen). Cells were grown in 2× YT medium supplemented with 200 μg/ml ampicillin and 50 μg/ml chloramphenicol at 37 °C to an A_{600 nm} of 0.8–1.0. Flasks were then transferred to 25 °C, and protein expression was induced with 0.5 mm isopropyl 1-thio-β-d-galactopyranoside for 5 h. SpeB self-maturates during expression and results in the fully active protease (residues 146–398). Cells were immediately harvested and resuspended in ice-cold 100 mm Tris, pH 8.0, 100 mm NaCl (buffer A) and subjected to three cycles of lysis by microfluidization (Microfluidics). The cell lysate was clarified by centrifugation at 45,000 × g for 30 min at 4 °C, and soluble fractions were loaded onto a 1-ml HisTrap HP nickel-nitrilotriacetic acid affinity column (GE Healthcare) pre-equilibrated with buffer A and eluted with buffer A containing 250 mm imidazole. The eluted protein was immediately diluted 5-fold with buffer B (50 mm NaOAc, pH 4.5) and purified by cation exchange chromatography (HiTrap SP HP, GE Healthcare) with a 20-column volume gradient to 50% of buffer B containing 1 m NaCl. Fractions corresponding to SpeB were pooled and dialyzed overnight at 4 °C in phosphate-buffered saline (PBS), pH 7.4, and 1 mm DTT. If required, pure SpeB protein was further subjected to gel filtration chromatography (Superdex 75, GE Healthcare) in PBS and 1 mm DTT to remove any remaining contaminants. Mature SpeB was concentrated to ∼10 mg/ml using Millipore Ultrafree-15 devices with a molecular weight cut-off of 3500 and immediately stored at −70 °C. Pure SpeB yields are ∼10–20 mg/liter of culture with >95% purity, as assessed by SDS-PAGE (supplemental Fig. 1A).

Mutations of Glycine Residues within C-terminal Loop

SpeB point mutations were generated using QuikChange (Qiagen) mutagenesis. SpeB C-terminal loop mutants were constructed using forward primer and complement 5′-CAG CTA CAG GAG CCG TTG CTA CTG C-3′ for C192A, 5′-CTT CAG CTC TTG CTA CTG GTG-3′ for G378A, 5′-CTT CAG CTC TTGGTG TTG GTG-3′ for T379V, 5′-CTT CAG CTC TTG GTG CTG GTG-3′ for T379A, 5′-CTT CAG CTC TTG GTA CTG CTG-3′ for G380A, 5′-CTT CAG CTC TTG GTA CTG GTG-3′ for G381A, 5′-CTT CAG CTC TTG GTA CTG GTG-3′ for G382A, 5′-GGT ACT GGT GGC GGC GCA GC-3′ for G384A, 5′-GGT ACT GGT GGC GGC GCA GA-3′ for G384D, and 5′-GGT ACT GGT GGC GGC GCA GG-3′ for G385A. Purification was performed exactly as for wild type SpeB. The point mutations did not affect structural stability of any of the recombinant expressed SpeB proteins as determined by differential scanning fluorimetry (data not shown) (23).

Crystallization and X-ray Data Collection

Crystals of mature apo-SpeB were grown by sitting drop vapor diffusion by mixing equal volumes (2 μl) of SpeB (10 mg/ml) and reservoir solution consisting of 0.15 m sodium nitrate and 15% PEG 3500 at 4 °C. Inhibitors E64 (G-Biosciences) and Ac-AEIK-CHO were added in a 3-fold molar excess to SpeB and incubated for 2 h at 25 °C and immediately used for co-crystallization experiments. Crystals of SpeB in complex with E64 and Ac-AEIK-CHO were grown from solutions consisting of 0.2 m magnesium nitrate with 20% PEG 3500 at 4 °C and 0.15 m sodium nitrate with 26% PEG 3500 at 25 °C, respectively. The His₆ tag was not removed because the protein crystallized readily.

Data for all three SpeB x-ray structures were collected on single, flash-cooled crystals at 100 K in a cryoprotectant consisting of mother liquor and 20% PEG 400 and were processed with HKL2000 (24) in orthorhombic space group P2₁2₁2 (Table 1). The calculated Matthews coefficient (V_M = 2.1 ų Da⁻¹) suggested one monomer per asymmetric unit with a solvent content of 42%. X-ray data for the apo and E64 complex SpeB structures were collected to 1.06 and 1.50 Å resolution, respectively, on beamline 11.1 at the Stanford Synchrotron Radiation Lightsource (Menlo Park, CA). For the SpeB·Ac-AEIK-CHO complex, data were collected to 1.37 Å resolution on the Advanced Photon Source beamline 23-ID-B at the Argonne National Laboratory (Argonne, IL). Data collection and processing statistics are summarized in Table 1.

TABLE 1.

SpeB x-ray data processing and structure refinement statistics

	Apo	E64	Ac-AEIK-CHO
Protein Data Bank ID	4D8B	4D8E	4D8I
Space group	P21212	P21212	P21212
Unit cell parameters (a, b, c) (Å)	114.99, 50.11, 41.89	114.61, 50.68, 37.38	114.91, 50.14, 37.46
Data processing
Resolution range (Å) (outer shell)	20–1.06 (1.08–1.06)	50–1.50 (1.53–1.50)	50–1.37 (1.39–1.37)
Unique reflections	107,464 (4612)	35,699 (1662)	45,004 (2221)
Completeness (%)	96.8 (83.6)	99.6 (95.5)	98.8 (100.0)
Redundancy	2.2 (4.8)	2.4 (5.0)	2.5 (4.5)
Rmerge (%)a	7.0 (59.2)	10.0 (57.3)	9.6 (72.9)
Average I/σ(I)	33.6 (2.1)	23.7 (1.7)	17.8 (2.0)
Refinement
Resolution range (Å)	19.7–1.06 (1.07–1.06)	35.6–1.5 (1.54 -1.50)	35.6–1.38 (1.41–1.38)
No. reflectionsb (test set)	107,138 (5358)	35,508 (1773)	44,934 (2270)
Rcryst (%)c	14.4 (42.4)	15.8 (30.5)	16.4 (27.7)
Rfree (%)c	16.0 (40.2)	18.8 (34.0)	19.1 (30.2)
Protein atoms/Waters	1960 (1865 H)/449	1967/383	2006/377
CV coordinate errord (Å)	0.15	0.21	0.17
r.m.s. deviation, bonds (Å)/angles (degrees)	0.008/1.31	0.020/1.85	0.017/1.63
B-values, protein/waters/ligands (Å2)	11.3/23.7/16.4	12.6/28.9/24.1	10.4/24.2/16.2
Ramachandran statistics (%)
Most favored	97.6	97.6	97.3
Additional allowed	2.4	2.4	2.7
Generously allowed	0.0	0.0	0.0

^a R_sym = 100·Σ_hΣ_i|I(h)_i − 〈I(h)〉|/Σ_hI(h), where I(h)_i is the ith measurement of reflection h, and 〈I(h)〉 is the average measurement value.

^b Reflections with I > 0 were used for refinement.

^c R_cryst = Σ_h‖F_o| − |F_c‖/Σ_h|F_o|, where F_o and F_c are the structure factor amplitudes from the data and the model, respectively. R_free is R_cryst with 5.0% test set structure factors.

^d Cross-validated Luzzati coordinate errors.

Structure Solution and Refinement

The apo-SpeB structure was determined by molecular replacement with Phaser (25, 26), using the previously published mature SpeB structure (2UZJ) (21) as the initial search model. The apo-SpeB structure then served as the search model for the E64 and Ac-AEIK-CHO SpeB complex structures. All structures were manually built with Coot (27) and iteratively refined using Phenix (28) with cycles of conventional positional refinement. For the ultrahigh resolution apo-SpeB structure, anisotropic B-factor refinement was included, and hydrogen atoms were included in the model coordinates. For the SpeB complex structures, TLS B-factor refinement was carried out in the last round of refinement with the SpeB monomer split into nine TLS groups as determined by TLS Motion Determination in Phenix (28). TLS refinement resulted in improved electron density maps with a minimal change in R_cryst and R_free. The electron density maps clearly identified that inhibitors E64 and Ac-AEIK-CHO were covalently attached to Cys¹⁹² within the active site. Water molecules were automatically positioned by Phenix using a 2.5σ cut-off in F_o − F_c maps and manually inspected. All three structures revealed a nitrate molecule (from the buffer) bound to residues located within the C-terminal loop. For the apo-SpeB structure, the final R_cryst and R_free are 14.4 and 16.0%, respectively; for the E64 complex structure, the final R_cryst and R_free are 15.8 and 18.8%, respectively; and for the Ac-AEIK-CHO complex structure, the final R_cryst and R_free are 16.4 and 19.1%, respectively (Table 1).

All models were analyzed and validated with PROCHECK (26, 29), WHATCHECK (30), and Molprobity (31) on the Joint Center for Structural Genomics (JCSG) Web server. Analysis of backbone dihedral angles with the program PROCHECK (29) indicated that all residues for the three structures are located in the most favorable and additionally allowed regions in the Ramachandran plot with no outliers. Coordinates and structure factors have been deposited in the Protein Data Bank (32), with accession entries 4D8B (apo), 4D8E (E64 complex), and 4D8I (Ac-AEIK-CHO complex). Structure refinement statistics are shown in Table 1.

SpeB Fluorogenic Substrate Design and Kinetic Assays

We synthesized an N-terminal acetylated, fluorogenic SpeB tripeptide substrate Ac-AIK-AMC, based on previously determined optimal cleavage sequences (17). This substrate was used to determine the catalytic activity of native SpeB and effects of point mutations on catalytic efficiency and to establish IC₅₀ values for E64 and Ac-AEIK-CHO. Unless otherwise stated, SpeB was incubated at 50 nm in 50 μl of PBS, pH 7.4, 0.1 mm EDTA, 10 mm DTT, 0.1% CHAPS at 25 °C with the activity measurement initiated by introduction of 50 μm Ac-AIK-AMC. The C-terminal His₆ tag was not removed from the SpeB protein preparations because the presence of the tag would not interfere with in vitro kinetics. The His₆ tag is positioned on the opposite side of the protein surface more than 50 Å from the active site (supplemental Fig. 1B). The increase in fluorescence due to substrate hydrolysis was measured every 30 s for a 10-min duration in 96-well plates on a PerkinElmer EnVision plate reader (excitation 355 nm, emission 460 nm). All components of the assay are stored as frozen aliquots and thawed immediately prior to the assay. Michaelis-Menten (K_m and k_cat) and IC₅₀ values were determined using GraphPad Prism software (GraphPad, Inc.).

SpeB IC₅₀ Determination

SpeB was incubated at 50 nm in the presence of increasing amounts of E64 or Ac-AEIK-CHO (10 nm to 10 μm) in a reaction buffer consisting of PBS, pH 7.4, 0.1 mm EDTA, 10 mm DTT, and 0.1% CHAPS and incubated for 20 min at 25 °C. 50 μm Ac-AIK-AMC was subsequently added, and the rate of substrate hydrolysis was measured for 10 min by the increase in fluorescence.

Cleavage of Inactive C192A Zymogen SpeB

C192A SpeB was incubated at 10 μm with 200 nm active wild type or SpeB mutants in a reaction buffer consisting of PBS, pH 7.4, 0.1 mm EDTA, 10 mm DTT, and 0.1% CHAPS. The mixtures were agitated at 37 °C, and aliquots were removed every 10–30 min. Reactions were immediately quenched with the addition of 1% SDS-reducing gel-loading buffer and boiled. Samples were subjected to SDS-PAGE to determine the extent of substrate cleavage in the inactive C192A SpeB.

Ac-AIK-AMC Substrate Synthesis

The SpeB Ac-AIK-AMC peptide substrate was synthesized using standard Fmoc solid phase synthesis chemistry starting with Fmoc-Lys(carbamate)-AMC Wang resin (EMD Biosciences). After completion of the peptide synthesis and its N-terminal acetylation with acetic anhydride and diisopropylethylamine in dichloromethane, the substrate was released from the resin with a mixture of trifluoroacetic acid (TFA), triisopropylsilane, and water (95%, 2.5%, 2.5%). Crude Ac-AIK-AMC was purified by reverse-phase HPLC using a C18 Xterra column (Waters), and all fractions containing the desired product were lyophilized. The final purity of Ac-AIK-AMC exceeded 95% and was verified by mass spectrometry: expected m/z 530.3, LC/MS (electrospray ionization) m/z 530.4 (MH⁺). Michaelis-Menten kinetic measurements for the hydrolysis of Ac-AIK-AMC by a 50 nm concentration of the wild type mature SpeB yielded k_cat = 0.69 s⁻¹, K_m = 246.6 μm, and k_cat/K_m = 2.8 × 10³ s⁻¹ m⁻¹ (Table 2 and supplemental Fig. 2).

TABLE 2.

Michaelis-Menten kinetic values of SpeB glycine-rich loop mutants

Mutation	Km	kcat	kcat/Km
	μm	s−1	s−1 m−1
Wild type	246.6 ± 16.9	0.69 ± 0.02	2.8 × 103
G378A	216.3 ± 8.7	0.40 ± 0.01	1.9 × 103
G380A	138.7 ± 10.8	0.39 ± 0.01	2.8 × 103
G381A	208.9 ± 8.8	0.60 ± 0.01	2.9 × 103
G382A	223.9 ± 9.0	0.57 ± 0.01	2.5 × 103
G384A	117.8 ± 8.8	0.52 ± 0.01	4.4 × 103
G385A	247.7 ± 15.2	2.14 ± 0.05	8.6 × 103
T379A	213.8 ± 7.1	0.63 ± 0.01	3.0 × 103
T379V	202.3 ± 8.3	0.62 ± 0.01	3.0 × 103
G384D	183.2 ± 13.1	0.35 ± 0.01	1.9 × 103

Synthesis of Ac-AEIK-CHO Inhibitor

An N-terminal acetylated, SpeB peptide inhibitor Ac-AEIK-CHO was designed based on the nonprime side cleavage sequence required for self-removal of the zymogen prodomain (residues 28–145). The Ac-AEIK-CHO peptide inhibitor was synthesized using standard Fmoc solid phase synthesis chemistry starting with N-ϵ-t-butoxycarbonyl-l-lysine-2-chlorotrityl resin (AnaSpec). Release from the resin with a mixture of acetic acid, trifluoroethanol, and dichloromethane (10%, 10%, 80%) afforded a C-terminal carboxylic acid peptide with tert-butyl-protected glutamate and butoxycarbonyl-protected lysine side chains (33) (1; supplemental Fig. 3). The resulting peptide was dried under vacuum and purified by reverse-phase HPLC and lyophilized. Compound 1 was dissolved in acetonitrile with equal molar concentrations of N-methylmorpholine hydrochloride and isobutyl chloroformate added in rapid succession at room temperature for 30 min to yield the mixed anhydride (2; supplemental Fig. 3), which was dried under vacuum, dissolved in water, and placed on ice. Sodium borohydride (NaBH₄) was added to 2 at a 1.1 molar ratio, brought to room temperature, and reacted for 30 min to produce the C-terminal alcohol (34) (3; supplemental Fig. 3), which was extracted with ethyl acetate and dried under vacuum. Compound 3 was dissolved in acetonitrile and a 1.1 molar ratio of Dess-Martin periodinane, dissolved in an equal volume of acetonitrile, and slowly added dropwise at room temperature and reacted overnight to yield the resulting C-terminal aldehyde (4; supplemental Fig. 3), which was purified by standard aqueous workup and dried under vacuum. Deprotection of the peptide aldehyde 4 was performed with a mixture of TFA, triisopropylsilane, and water (95%, 2.5%, 2.5%). Crude Ac-AEIK-CHO was purified by reverse-phase HPLC using a C18 Xterra column (Waters), and the desired product fractions were lyophilized to produce the final product (5; supplemental Fig. 3). The purity of Ac-AEIK-CHO exceeded 90% and was verified by mass spectrometry: expected m/z 487.6, LC/MS (electrospray ionization) m/z 488.2 (MH²⁺). All intermediates and reactions were monitored by LC/MS.

RESULTS

Apo-SpeB Structure

The SpeB zymogen (residues 28–398), without its N-terminal secretion sequence (residues 1–27), was cloned from S. pyogenes strain 10782 and overexpressed in E. coli. The protease fully matures during expression with self-removal of the N-terminal prodomain (residues 28–145) and purifies as a monomer as evidenced by gel filtration (supplemental Fig. 1A). The apo-SpeB structure was determined by molecular replacement using the mature SpeB structure (Protein Data Bank entry 2UZJ) (21) as the initial search model. The final R_cryst and R_free values for the apo structure were 14.4 and 16.0%, respectively, with 97.6% of the residues in the most favored region of the Ramachandran plot (Table 1). The structure includes residues 146–398, an ordered leucine residue from the C-terminal His₆ tag (LEH₆), a nitrate ion, and 548 water molecules. The nitrate was introduced via the crystallization conditions and is not required for biological activity. The bound nitrate has a significantly higher B-value in comparison with the protein, most likely due to incomplete occupancy (Table 1). Three regions of unaccounted for positive electron density within solvent channels were not modeled in the final F_o − F_c maps and are most likely cryoprotectant PEG 400 at low occupancy.

The crystals of apo-SpeB were obtained in conditions with a biologically relevant pH of 7.0, which resulted in a different crystal lattice and space group in comparison with the mature SpeB x-ray structure 2UZJ crystallized at pH 4.0 (21). Due to crystal packing interactions in the 2UZJ structure, SpeB was proposed to exist in a homodimeric arrangement with the interface situated along a crystallographic axis. This arrangement in the crystal lattice resulted in residue Ser¹⁸² (our numbering system) adopting a cis-peptide bond and contributing to the oxyanion hole of the opposite subunit as well as a key hydrogen bond to an unidentified ligand that was bound within the active site at the dimer interface (21). This interface does not exist in the crystals grown at neutral pH. Our gel filtration purification profile and x-ray structure strongly support a monomer as the biologically relevant form of SpeB. Our results are consistent with the x-ray structure of the zymogen form of SpeB (Protein Data Bank entry 1DKI) at pH 4.2 (20) and the NMR structure of mature SpeB at pH 6.0 (Protein Data Bank entry 2JTC) (22).

The apo-SpeB structure was determined to 1.06 Å resolution with excellent electron density throughout the entire protein structure. SpeB belongs to the C10 family of bacterial cysteine endopeptidases and adopts the canonical papain fold that is conserved across the CA clan of proteases. Briefly, SpeB is composed of a C-terminal subdomain consisting of a four-stranded antiparallel β-sheet that is sandwiched between four α-helices (three above, one below) from the N-terminal subdomain (Fig. 1) (20–22). Importantly, the initial F_o − F_c electron density maps after molecular replacement exhibited unambiguous electron density for the glycine-rich C-terminal loop (residues 368–390) that is unique to SpeB (Fig. 1). This loop has been shown to be highly flexible by NMR (2JTC) and was designated as a disordered region in both the zymogen (1DKI) and mature (2UZJ) SpeB structures (20, 21). In the absence of ligand, the C-terminal loop is positioned over the active site and forms a type II β-turn (Gly³⁷⁸-Thr³⁷⁹-Gly³⁸⁰-Gly³⁸¹) (Fig. 1). Gly residues within the β-turn and the aliphatic side chain of Leu³⁷⁷ secure the C-terminal loop to the protein surface via hydrophobic interactions with the indole side chain of Trp³⁵⁹. Our structure is consistent with the NMR study of mature SpeB that observed NOEs among the same loop residues and Trp³⁵⁹ (22). In the crystal structure, the neighboring symmetry-related molecules do not provide any interactions that would account for the observed stabilization of the C-terminal loop.

FIGURE 1.

Crystal structures of mature SpeB reveal an ordered C-terminal Gly-rich loop. A, schematic representations of apo (4D8B; cyan), E64-bound (4D8E; green), and Ac-AEIK-CHO-bound (4D8I; rose) mature SpeB show the expected canonical papain-like fold. The N-terminal α-helical domain supplies the four α-helices that sandwich the C-terminal four-stranded antiparallel β-sheet. Residues 376–385 of the C-terminal loop, E64, and Ac-AEIK-CHO are shown in stick representations (yellow, carbon; red, oxygen; blue, nitrogen). A blue F_o − F_c simulated annealing omit map contoured at 2.5σ provides compelling evidence for an ordered structure of the C-terminal loop that was not observed in previous x-ray structures of SpeB (20, 21). B, chemical structure and atom numbering of the general cysteine protease inhibitor E64 and of the covalent complex formed upon nucleophilic attack by the active site residue Cys¹⁹². C, chemical structure of a new Ac-AEIK-CHO inhibitor based on the recognition sequence for zymogen prodomain removal during self-activation. Nucleophilic attack by the active site residue Cys¹⁹² results in covalent attachment of the peptide and protonation of the aldehyde oxygen to form a hydroxyl.

Structures of SpeB in Complex with E64 and Ac-AEIK-CHO

Mature SpeB was incubated with a 3-fold excess of either the general cysteine proteinase inhibitor E64 or our substrate mimetic peptide inhibitor Ac-AEIK-CHO for 2 h at room temperature prior to crystallization (Fig. 1, B and C). The apo-SpeB structure served as the search model for both molecular replacement solutions. The final R_cryst and R_free values were 15.8 and 18.8%, respectively for the E64 complex and 16.4 and 19.1%, respectively, for the Ac-AEIK-CHO complex. Both structures have greater than 97.2% of the residues in the most favored region of the Ramachandran plot (Table 1). As with the apo structure, both complex structures crystallized with identical crystal packing and had continuous electron density for residues 146–398, an ordered leucine residue from the C-terminal His₆ tag, a bound nitrate, and water molecules. After the initial refinement, clear residual density was present in F_o − F_c maps corresponding to E64 or Ac-AEIK-CHO covalently attached to the active site residue Cys¹⁹² (Fig. 2A). For both structures, the compounds were refined with 100% occupancy and have B-values similar to adjacent active site residues. The overall conformation of SpeB is highly conserved between the E64-bound and Ac-AEIK-CHO-bound structures, as evidenced by an r.m.s. deviation of 0.13 Å for all Cα atoms with a maximum deviation of 0.34 Å (r.m.s. deviation of 0.26 Å with maximum deviation of 2.69 Å for all atoms).

FIGURE 2.

SpeB active site interactions and electrostatic surface potential. A, E64 and Ac-AEIK-CHO bound to SpeB are shown in stick representations and colored according to the legend of Fig. 1. Specific main-chain and side-chain interactions provided by the SpeB active site to the inhibitors are labeled (carbon, green for E64 complex and rose for Ac-AEIK-CHO complex) with ordered water molecules as red spheres. Simulated annealing electron density for the bound inhibitors is shown as a blue F_o − F_c omit map contoured at 2.0σ. Although not biologically significant, the side-chain Nϵ2 of Asn²⁰⁸ from a symmetry-related crystal contact provides a potential hydrogen bond with an oxygen from the peptide inhibitor P3 glutamate and further stabilizes the peptide, as indicated by improved F_o − F_c electron density, when compared with the similar region of E64. B, schematic representation of the hydrogen bonding network and corresponding distances within the SpeB active site to E64 or the bound Ac-AEIK-CHO peptide. The bound inhibitors E64 and Ac-AEIK-CHO are labeled in blue, and a potential hydrogen bond provided by side-chain Nϵ2 of Asn²⁰⁸ from a symmetry-related crystal contact (non-physiological) to the peptide inhibitor P3 glutamate side chain is labeled in purple. C, the electrostatic potential of the surface surrounding the SpeB active site was calculated with Adaptive Poisson-Boltzmann Solver (52). E64 and Ac-AEIK-CHO are colored as in Fig. 1. Blue, positive potential (≥5 mV); white, neutral potential (0 mV); red, negative potential (≤−5 mV). E64 and Ac-AEIK-CHO are primarily exposed to bulk solvent with only 30 and 43% of the inhibitors buried by SpeB active site residues, respectively. The solvent-exposed regions of the bound inhibitors were determined in CCP4 (26) with a solvent radius of 1.4 Å.

E64-bound SpeB Interactions

The SpeB active site, as with the majority of conserved papain fold proteases, consists of a catalytic dyad composed of residues Cys¹⁹² and His³⁴⁰, which exist as a zwitterionic pair (35). The nucleophilicity of the negatively charged Cys¹⁹² is greatly stabilized by both its location at the N terminus of the central α-helix (residues 192–204), to exploit the dipole moment, and the interaction with His³⁴⁰ (35, 36) (Fig. 2A). His³⁴⁰ also serves as the acid/base catalyst during substrate cleavage. The majority of specific interactions provided by the SpeB active site residues are concentrated near the carboxyl end of E64 (Fig. 2A). Opening of the E64 epoxide ring occurs upon nucleophilic attack by SpeB Cys¹⁹² and results in covalent attachment of the C2 atom of E64 to the Cys¹⁹² thiol with a bond distance of 1.9 Å (Figs. 1B and 2B). The Gln¹⁶² side-chain Nϵ2 and main-chain amide of Cys¹⁹² comprise the oxyanion hole that stabilizes the interaction to one of the E64 carboxyl oxygens. The other E64 carboxyl oxygen forms potential hydrogen bonds with the Nδ1 of the catalytic His³⁴⁰ imidazole and a bound water molecule that is also bridged to the side-chain Nϵ1 of Trp³⁵⁹ (Fig. 2B). The E64 carbonyl (C4 and O4) is stabilized by interactions with the main-chain amide of Ser²⁸² and a highly ordered water molecule, which is positioned via hydrogen bonds to the main-chain amides of both Cys¹⁹² and Val¹⁹³. The E64 leucine side chain is inserted into the hydrophobic S2 binding pocket formed by Val³³⁴, Gly³³⁹, and Ala³⁴¹ (Fig. 2C). The interactions between SpeB surface and the E64 amino-4-guanidinobutane moiety are mediated through ordered water molecules. This region of E64 has less well defined electron density and higher B-values, indicative of increased flexibility (Fig. 2B). E64 is largely exposed to bulk solvent with only 30% of its surface buried by the SpeB active site (Fig. 2C). Overall, the interactions between E64 and SpeB are highly conserved with previously determined E64-bound papain-like proteases, including Ervatamia coronaria ervatamin-A (Protein Data Bank entry 3BCN) (r.m.s. deviation of 2.04 Å for 68 conserved core Cα atoms) and ervatamin-C (Protein Data Bank entry 2PRE) (r.m.s. deviation of 2.03 Å for 68 conserved core Cα atoms) (37), and Staphylococcus aureus staphopain (1CV8) (r.m.s. deviation of 1.57 Å for 62 conserved core Cα atoms) (38).

Ac-AEIK-CHO-bound SpeB Structure

We synthesized a specific peptide C-terminal aldehyde inhibitor Ac-AEIK-CHO based on the nonprime side recognition sequence for zymogen prodomain removal during self-activation (Ala¹⁴²-Glu¹⁴³-Ile¹⁴⁴-Lys¹⁴⁵/Gln¹⁴⁶). Standard Michaelis-Menten kinetics revealed that Ac-AEIK-CHO inhibits mature SpeB activity with an IC₅₀ of 205 nm (supplemental Fig. 4). The C1 atom of the peptide C-terminal aldehyde is susceptible to nucleophilic attack by SpeB and results in formation of a covalent adduct between Ac-AEIK-CHO C1 and the Cys¹⁹² thiol with a bond length of 1.9 Å (Figs. 1C and 2B). As a consequence, the aldehyde oxygen is reduced to a hydroxyl. An ordered water molecule that is positioned by the side-chain Oϵ1 of residue Gln¹⁶² is a likely candidate to supply the proton to the aldehyde oxygen (Fig. 2B). This reduction occurs concomitant with sp³ hybridization of the C1 atom (Fig. 2A). The reacted peptide inhibitor hydroxyl group is positioned in close proximity to the oxyanion hole, created by residues Gln¹⁶² and Cys¹⁹², and the electron-withdrawing properties of these residues stabilize the oxyanion intermediate during catalysis. With iterative rounds of model building and refinement of the Ac-AEIK-CHO-bound structure, additional electron density indicated that the reacted peptide hydroxyl had an alternate conformation positioned toward the side-chain Nδ1 of His³⁴⁰ (Fig. 2A). Notwithstanding, the peptide inhibitor is not entirely representative of a biological substrate that would include residues to bind on the prime side of the SpeB active site. Under these circumstances, protein substrates with trans-peptide bonds would position a carbonyl oxygen toward the oxyanion hole. Additional ordered water molecules are coordinated near the peptide inhibitor C-terminal hydroxyl moiety that do not contribute to peptide binding or electrostatic perturbations of SpeB active site residues (Fig. 2B). These waters would most likely be displaced or absent after binding of a biologically relevant protein substrate, which would provide interactions with the prime side region of the SpeB active site.

In comparison with the E64 complex structure, many of the same residues provide key interactions for binding and positioning of the Ac-AEIK-CHO (Fig. 2B). The peptide P1 Lys side chain is positioned via interaction of the Nζ with the side-chain hydroxyl from Ser²⁷⁹ and has elevated B-values and reduced electron density in comparison with the rest of the bound peptide, suggesting flexibility within this region (Fig. 2A). SpeB has been shown to preferentially cleave biological protein substrates with a variety of residues at P1, including Arg, Leu, Asn, and Phe (18, 39, 40). Based on the Ac-AEIK-CHO structure, the S1 pocket can accommodate a variety of substrate P1 residues and prefers basic side chains as evidenced by the strong electronegative potential of the SpeB S1 pocket (Fig. 2C). Similar to E64, the main-chain carbonyl of the peptide inhibitor P2 Ile is anchored into a highly electropositive patch provided by the main-chain amide of Ser²⁸² and a water molecule ordered by the main-chain amide of Val¹⁹³ (Fig. 2B). The main-chain carbonyl of Ser²⁸² provides a key interaction to the peptide inhibitor main-chain amide of P2 Ile (Fig. 2B) that is not observed in the E64-bound SpeB structure where the E64 carbonyl (C11 and O5) superimposes on the peptide P2 Ile amide. The peptide inhibitor P2 Ile side chain is anchored into the S2 hydrophobic pocket similarly to the E64 Leu moiety. Despite the difference in Ile and Leu side-chain branching of Ac-AEIK-CHO and E64, respectively, the S2 pocket residues are conformationally conserved. This is surprising because the preference of Ile over Leu at P2, in otherwise identical substrates, has been reported to result in an 86-fold increase in catalytic efficiency (41). As with other papain-like proteases, the specificity of the S2 pocket of SpeB is a primary determinant in substrate recognition (17, 42). Our SpeB·Ac-AEIK-CHO structure illustrates an ability to accommodate a number of branched hydrophobic side chains in the S2 pocket, including Ile, Tyr, Met, and Val, which have been identified in a number of biological SpeB substrates, and accounts for the relatively broad specificity of SpeB (18, 39, 40). As with E64, there is a lack of specific interactions from the SpeB active site to the P3 Glu and P4 Ala residues of Ac-AEIK-CHO with 57% of the peptide inhibitor surface exposed to bulk solvent (Fig. 2C). The Ac-AEIK-CHO P3 glutamate is positioned by water molecules that are ordered by SpeB surface residues (Fig. 2B). For most papain-like proteases, the S3 pocket is not well defined and is located on the surface of the protease (37). Our structure supports the substrate specificity profiles of SpeB that have suggested that three amino acids on the nonprime side of the substrate are sufficient for recognition and that alteration of residues at the P3 or P1 position does not drastically influence catalytic efficiency by the protease (17). Additional prime side specificities have been established (18, 39, 40), but cannot be validated, with the current x-ray structures.

Conformational Movements Associated with Activation and Active Site Binding

The SpeB prodomain (residues 28–145) preserves the protease in a catalytically inactive state by the insertion of an α-helix into the active site that displaces both the C-terminal (residues 368–390) and catalytic (residues 332–341) loops (Fig. 3A) (20–22). Activation of the protease via removal of the prodomain allows the C-terminal loop to flip ∼150–170° to adopt the conformation observed in the apo structure (Fig. 3A). Superposition of the zymogen (1DKI) and active apo-SpeB structures, not including the C-terminal loop, indicate a highly conserved structure as evidenced by an r.m.s. deviation of 1.61 Å for all Cα atoms with a maximum deviation of 8.5 Å (compared with 1.41 Å with a maximum deviation of 10.7 Å for all atoms). The additional large conformational movement is found in the catalytic loop (residues 332–341), which contains His³⁴⁰ (Fig. 3A). Upon removal of the prodomain, the catalytic loop is free to move and results in conformational alignment of His³⁴⁰ with a catalytically active orientation and promotion of Cys¹⁹² nucleophilicity (Fig. 3A) (21, 22). Importantly, movement of the catalytic loop also exposes and forms the hydrophobic patch critical to substrate recognition.

FIGURE 3.

Conformational changes associated with activation and active site binding. A, zymogen SpeB (1DKI) is shown as a schematic representation with the static regions of the protein colored gray, zymogen prodomain in light blue, and the RGD motif in rose with active site residues Cys¹⁹² and His³⁴⁰ represented as sticks and colored purple. The highly flexible catalytic (residues 332–341) and C-terminal (residues 368–390) loops are colored purple and orange, respectively, and labeled accordingly. The two ends of the C-terminal loop with missing residues in the zymogen x-ray structure are connected with an orange line. B, apo-SpeB structure is colored as in A. The catalytic loop moves ∼8.5 Å, relative to the zymogen structure, and orients the catalytic His³⁴⁰ for catalysis. Removal of the prodomain allows the C-terminal loop to flip 150–170° to form a type II β-turn over the active site. C, structure of SpeB in complex with E64 reveals the additional loop movements that occur upon the transition from apo to bound forms. The catalytic loop and C-terminal loop move ∼4.5 and 17.5 Å, respectively upon binding E64 or Ac-AEIK-CHO. The E64 complex is colored as in A. D, superposition of the apo-SpeB (carbon, cyan) and E64-bound (carbon, green) structures reveals the large conformational movement of the C-terminal loop to accommodate substrate binding. Hinge residues Ala³⁷⁶ and Gly³⁸⁵ have a 23.6 and 51.0° rotation, respectively, from the active site occluded C-terminal loop conformation in the apo structure to the open orientation depicted in the E64- and Ac-AEIK-CHO-bound structures. Residues with conserved orientations between the apo and bound states are labeled in black.

Comparison of the apo-SpeB with the E64- and Ac-AEIK-CHO-bound structures reveals several large conformational changes in loops surrounding the active site upon ligand binding. As with the transition from zymogen to active apo, the most significant movements occur within both the C-terminal loop (residues 368–390) and the catalytic loop (residues 332–341) with a maximum Cα displacement of 17.5 and 4.5 Å, respectively (Figs. 1 and 3A). Ala³⁷⁶ and Gly³⁸⁵ represent the C-terminal loop hinge points where the loop swings open and away from the active site by 23.6 and 51.0°, respectively, upon binding (Fig. 3B). This movement abolishes the type II β-turn (Gly³⁷⁸-Thr³⁷⁹-Gly³⁸⁰-Gly³⁸¹) in the apo structure and increases the solvent accessibility of the entire protein surface by 1063 Ų and the active site cleft by ∼450 Ų. Despite the proper alignment of His³⁴⁰ for catalysis in the apo active site, the catalytic loop shifts an additional 4.5 Å to accommodate substrate binding. This conformational transition would not be permitted without the concerted relocation of the C-terminal loop upon substrate binding because clashes would occur between the two loops (Fig. 3A). For the rest of the protein, minimal conformational rearrangements occur upon substrate interaction as evidenced by an r.m.s. deviation of 0.42 Å with a maximum deviation of 1.90 for all protein Cα atoms (compared with 0.58 Å with a maximum deviation of 7.49 Å for all atoms). These loop movements associated with active site binding would not be possible if SpeB existed in the proposed homodimeric arrangement and do not favor the formation of a transitory homodimeric SpeB as previously suggested (21). Furthermore, gel filtration analysis of SpeB in the presence of 5 μm E64, 50-fold above the IC₅₀ of 98 nm (supplemental Fig. 4), shows that the presence of E64 does not affect the monomeric state (supplemental Fig. 1).

Mutational Analysis of Glycine Residues within the C-terminal Loop

The C-terminal loop (residues 368–390) contains six Gly residues, all of which adopt different φ/ψ angles between the apo and bound structures of SpeB (Fig. 4A). These six Gly residues were mutated to Ala to explore the effects on SpeB substrate binding and catalysis with respect to restricted glycine φ/ψ angles that promote either the apo-SpeB loop (occluded state) or the bound SpeB loop (open state) conformations (Fig. 3B). Activity of the SpeB Ala mutations was measured by hydrolysis of the small tripeptide fluorogenic substrate Ac-AIK-AMC. G378 is located at n + 2 from the hinge residue Ala³⁷⁶, and mutation to Ala would restrict the loop to an occluded state because the φ/ψ angles in the apo and bound structures are −99/−163 and 100/123, respectively. Forcing the loop to remain positioned over the active site would block access of substrates and result in a deleterious effect on catalysis (Fig. 4B). As predicted, the G378A mutation caused an inhibition of catalytic turnover (k_cat) that results in a modest decrease in catalytic efficiency (k_cat/K_m) of ∼34% relative to wild type SpeB (Table 2).

FIGURE 4.

Effects of Gly-rich C-terminal loop alanine-scanning mutagenesis. A, Ramachandran plot of changes in the φ/ψ angles of the six Gly residues found in the C-terminal Gly-rich loop in the apo and E64-bound SpeB structures reveals conformational isomerism between two well defined conformers. B, table of changes in the φ/ψ angles in the apo versus bound state for residues in the C-terminal loop and effects of mutations on catalytic efficiency upon the hydrolysis of the tripeptide fluorogenic substrate Ac-AIK-AMC. Mutation of Gly³⁷⁸ to Ala results in locking the loop into the active site occluded orientation and blocks accessibility to substrates. Mutation of the hinge residue Gly³⁸⁵ to Ala favors the open conformation and results in an approximately 3.1-fold increase in SpeB catalysis. C, prodomain removal in a catalytically inactive zymogen C192A SpeB and effect of mutations in the C-terminal Gly-rich loop. Mutation of C-terminal Gly residues to Ala restricted the C-terminal loop and reduced zymogen prodomain removal considerably. These results suggest that a flexible C-terminal loop is critical to aid in the removal of hydrolyzed products of natural protein targets.

As opposed to the negative effects mutation to Gly³⁷⁸ has on SpeB catalysis, mutation of Gly³⁸⁴ and the hinge residue Gly³⁸⁵ to Ala would promote the open state observed in the SpeB complex structures as these residues adopt φ/ψ angles unique to glycine only in the apo (occluded) structure (Fig. 4B). Such mutations should stimulate activity of SpeB by affording substrates unrestricted access to the active site. As expected, locking the loop in the open state via mutation of both Gly³⁸⁴ and Gly³⁸⁵ to Ala resulted in an increase of ∼1.6- and ∼3.1-fold, respectively, in the k_cat/K_m for Ac-AIK-AMC hydrolysis (Fig. 4B). For Gly³⁸⁴, the increase in efficiency is a result of improved substrate binding because the K_m increases almost 2.1-fold over the wild type (Table 2). The G385A mutation had no influence on substrate binding but did result in an ∼3.1-fold improvement in k_cat, suggesting that the loop is not required for catalysis.

Gly³⁸⁰, Gly³⁸¹, and Gly³⁸² form the type II β-turn in the apo structure (Fig. 3B). Mutations at these positions are not predicted to result in significant consequences on SpeB activity because these residues are located near the tip of the C-terminal loop away from the hinge residues and would have limited effects on its overall flexibility and positioning. Mutation of Gly³⁸¹ and Gly³⁸² resulted in anticipated negligible changes in both substrate binding and rates of Ac-AIK-AMC hydrolysis (Table 2). However, despite a catalytic efficiency similar to that of wild type, for G380A, k_cat was reduced by 44%, and K_m improved by 1.8-fold (Table 2). The addition of a methyl group to Gly³⁸⁰ would provide a potential van der Waals interaction with a bound substrate only if the C-terminal loop assumed the occluded apo conformation. This result suggests that the C-terminal loop could sample this occluded position upon substrate binding to the active site. The G380A mutation, however, does negatively affect the catalytic rate, suggesting that reduced flexibility of the C-terminal loop hinders product release.

Additional mutations included Thr³⁷⁹ to Ala and Val to explore the structural and/or functional role of the Thr³⁷⁹ side-chain hydroxyl and Gly³⁸⁴ to Asp because this mutation has been observed in allelic variants of SpeB and significantly reduces protease activity (22). The Thr³⁷⁹ hydroxyl does not contribute significantly to substrate binding or catalysis as observed by the insignificant consequences that mutation to either Ala or Val has on SpeB activity (Fig. 4B). The allelic variant G384D, which was reported to have a 12-fold decrease in proteolytic activity of SpeB as measured by azocasein hydrolysis (22), results in a minimal decrease of ∼33% in Ac-AIK-AMC catalysis.

Overall, the results of the alanine-scanning mutagenesis of Gly residues within the C-terminal loop illustrate a conserved net effect on k_cat of Ac-AIK-AMC turnover notwithstanding the residue location within the loop. Two alanine mutations (Gly³⁸⁰ and Gly³⁸⁴) did improve the K_m of substrate binding, which was countered by a decrease in k_cat (Table 2). However, substrate binding is assumed to be much slower than hydrolysis under these steady state in vitro conditions, and the observed changes in k_cat can alter the rate constants that comprise K_m without altering the affinity of SpeB for Ac-AIK-AMC. Taken together, our mutational analysis results suggest that the C-terminal loop does not significantly contribute to substrate affinity, but its structural flexibility is the limiting factor in the catalytic efficiency with respect to the recognition of the small tripeptide substrate and release of the hydrolyzed product.

The C-terminal loop Ala mutants were also assessed for their ability to remove the prodomain of the inactive C192A SpeB zymogen and to determine if the mutations also altered specificity profiles for a biologically relevant substrate. Due to limitations in detection, the zymogen was added to a final concentration of 10 μm, well above the predicted K_m, and introduced product release as a potential rate-limiting step. Overall, the individual C-terminal loop Gly mutations to sterically hindered Ala residues slowed hydrolysis of the zymogen considerably. Therefore, a flexible C-terminal loop seems critical to aid in the removal of hydrolyzed products (Table 2). Moreover, residues not important to Ac-AIK-AMC hydrolysis had a significant impact on protein substrate catalysis. For instance, mutation of Thr³⁷⁹ to Ala, which was not significant for Ac-AIK-AMC turnover, resulted in a drastically reduced hydrolysis of prodomain, whereas mutation to Val did not substantially affect the rate of zymogen activation (Fig. 4C). From these results, we surmise that Thr³⁷⁹, particularly its branched side chain, is most likely involved in substrate recognition and/or product release. G380A, G384A, and G384D mutants had an improved K_m of Ac-AIK-AMC binding with a concomitant decrease in catalytic turnover. With respect to a biological substrate, these mutations have similar zymogen maturation intermediate profiles; however, the accumulated amounts of intermediates at various time points differ among the SpeB mutants (Fig. 4C). The individual cleavage sites are therefore cleaved at different rates upon mutation of residues in the glycine-rich active site loop, which suggests that the mutations affect the propensity to recognize and hydrolyze certain cleavage sequences (39). Mutation of Gly³⁸¹ and Gly³⁸² to Ala had negligible affects on k_cat and K_m for Ac-AIK-AMC hydrolysis but had a deleterious effect on prodomain removal, most likely due to decreased loop flexibility. The G385A mutant, with a 3.1-fold increase in k_cat, has a similar rate of zymogen prodomain removal as wild type protein (Fig. 4C). Because these measurements were carried out in saturating conditions, we cannot assess increased efficiency of this mutation under these experimental conditions because product release is the rate-limiting step.

DISCUSSION

The crystal structures of the S. pyogenes secreted cysteine protease SpeB in the mature apo form and in complexes with E64 and with Ac-AEIK-CHO provide a detailed view of the conformational changes that occur upon activation and substrate binding. These structural results, in combination with the development and application of a newly synthesized robust fluorogenic substrate to probe SpeB activity, the design of a new specific peptide inhibitor, and alanine-scanning mutagenesis, offer significant detailed insights into the broad substrate specificity of SpeB and the functional role of the active site loops.

Comparison with Previous SpeB Structures

Interactions between the SpeB active site and E64 have been previously characterized by NMR and mutational analysis (22). Likewise, an E64-like molecule was co-purified and crystallized in the mature SpeB x-ray structure 2UZJ; however, the identity of the compound was not confirmed (21). Specific interactions between E64 and SpeB residues, as indicated by chemical shift perturbations in the NMR studies, are well conserved in our x-ray structure. In the NMR study, residues that caused the most significant chemical shifts, including Gly²⁸¹ and Val³³⁴, were mutated to alanine, and the effects on azocasein hydrolysis and prodomain removal of a catalytically inactive C192S mutant of the SpeB zymogen were examined (22). Mutation of Gly²⁸¹ to Ala resulted in a 15.3–46-fold decrease in catalytic efficiency, and the authors concluded that the region of the protein containing Gly²⁸¹ (residues 279–281) directly interacted with E64. Based on our E64 complex structure, the addition of a methyl group via the G281A mutation would introduce steric van der Waals clashes and occlude a portion of the active site from binding E64. Mutation of the hydrophobic S2 pocket residue Val³³⁴ to alanine precipitously reduced SpeB activity by 131–420-fold (22). The V334A mutation would not drastically change the volume of the S2 pocket; however, because this residue is the primary source of hydrophobicity within the pocket, reduction to an alanine side chain would greatly alter the electrostatic surface potential and modify the substrate specificity of SpeB (Fig. 2C). Both G281A and V334A mutants proteolyze the inactive zymogen SpeB with altered digestion intermediates, supporting the likelihood of changes in substrate specificity due to both mutations (22).

Role of C-terminal Loop in SpeB Function

In the NMR study of mature SpeB, the C-terminal loop had increased flexibility and dynamic motion upon binding E64 (22). The conformational exchange was measured in the picosecond to nanosecond time scale range and not on the slower microsecond to millisecond time scale typical for standard active site loops associated with enzyme catalysis (43, 44). In this study, increased hydrolysis of the fluorogenic substrate for the G384A and G385A mutants, which lock the C-terminal loop in the open conformation, provide evidence that the C-terminal loop is not required for substrate hydrolysis by SpeB. Consistent with the NMR findings, the C-terminal loop does not conform to the induced fit theory, whereby active site loops are proposed to be essential for the protection of bound substrates and reactive intermediates and/or help to stabilize transition state intermediates during the course of a reaction. Many kinases, including protein kinase A (PKA) and MEK1, contain glycine-rich active site loops that position ATP for γ-phosphate transfer and are similar in length to the C-terminal loop of SpeB (45). However, unlike SpeB, these kinase loops exhibit induced fit and partially close over the active site upon ATP binding (45–48). It is therefore possible that the primary biological function of the C-terminal loop is not associated with the regulation of SpeB hydrolysis but is involved in substrate recognition and product release.

Targeting the SpeB Active Site for Inhibitor Design

SpeB has been studied as a virulence factor for decades, although its importance for S. pyogenes pathogenesis, as assessed by genetic knock-out models, has been highly debated. The development of specific small molecules directed toward SpeB would promote the freedom to apply reversible, rapid, and dosable chemical methodologies to the functional studies of SpeB under native bacterial conditions and infection models. The broad substrate specificity that SpeB shares with other papain fold cysteine proteases presents a challenge in the development of specific small molecule active site-directed inhibitors. Furthermore, high affinity compounds typically contain an irreversible reactive warhead, such as the epoxide moiety of E64, that exploits the nucleophilicity of the protease active site residues and, in turn, tends to diminish specificity toward individual target proteins (49). Highly ordered waters identified in the apo-SpeB active site provide insight into the potential design of novel active site-directed inhibitors because these atoms can serve as a foundation on which a compound scaffold can be constructed (Fig. 5). For example, superposition of E64 in complex with SpeB shows that several of the apo-SpeB active site-bound waters are conserved with E64 atoms. Redesign of E64 to accommodate hydrogen bond acceptors and/or donors at sites where water is found would help to promote specificity and affinity toward SpeB. Surprisingly, two highly ordered water molecules within the apo-SpeB structure superimpose with the E64 Ile side-chain methyl groups located within the hydrophobic S2 pocket (Fig. 5). These ordered waters suggest that hydrogen bond donors and acceptor moieties should also be explored in the design of any small molecules targeting the SpeB S2 binding pocket.

FIGURE 5.

Apo-SpeB active site for inhibitor design. Ordered water molecules within the apo-SpeB structure can help guide the design of specific active site-targeted small molecules. Waters are represented as red spheres with electron density shown as a blue 2F_o − F_c map contoured at 1.5σ. Superposition of E64 and the electrostatic surface potential, colored as in Fig. 2, shows conservation in water molecule positions with atoms in E64.

Comparison of SpeB with Other Proteins with Glycine-rich Active Site Loops

Along with SpeB, the C10 peptidase family includes Porphyromonas gingivalis PrtT peptidase, P. gingivalis periodontain, and Prevotella intermedia interpain A (40). Sequence alignments show that PrtT peptidase and periodontain share a conserved C-terminal loop similar in length and glycine content with SpeB. Interpain A, the only other member of the family with structural information, consists of a truncated C-terminal loop but has an extended catalytic loop that may serve a similar function to the SpeB C-terminal loop (Fig. 6). Unfortunately, interpain A and the other C10 family members have not been extensively studied, and there is little information on substrate specificity and kinetics.

FIGURE 6.

Superposition of papain-like proteases. Schematic representations of x-ray structures of papain-like proteases are shown superimposed on SpeB in complex with E64, including P. intermedia interpain A (3BBA) (53), S. aureus staphopain in complex with E64 (1CV8) (38), T. cruzi cruzain bound to a nonpeptidic tetrafluorophenoxymethyl ketone (3IUT) (54), C. papaya papain in complex with remnants of a cysteine protease inhibitor (2CIO) (55), and human cathepsin X (1EF7) (56) with α-helices in red, β-strands in yellow, and smoothed loops in green. Catalytic cysteines and small molecule inhibitors are depicted in a stick representation and colored as in Fig. 1.

Superposition of a number of papain-like proteases shows that the structural core of the proteases is conserved but with very little structural similarity in the loop regions (Fig. 6). Similar to SpeB, S. aureus staphopain hydrolyzes bradykinin (50) and fibronectin (51) and prefers branched hydrophobic residues at P2 and Arg residues at P1. Despite these similarities in both substrate preferences and structure, staphopain belongs to the C47 peptidase family and shares only 11% sequence homology with SpeB. As with interpain, staphopain has a shortened C-terminal loop and an elongated catalytic loop (Fig. 6). Members of the large C1 peptidase family, including Carica papaya papain, Trypanosoma cruzi cruzain, and human cathepsin X, all share sequence target preferences with SpeB because they also prefer hydrophobic P2 and basic P1 residues within their substrates. Nevertheless, these proteases lack active site loop conservation with SpeB (Fig. 6). These observations further support our proposal that the primary biological function of the SpeB C-terminal loop is for substrate recognition.

These x-ray structures, the use of a new fluorogenic peptide substrate and substrate mimetic peptide inhibitor, and alanine-scanning mutagenesis have shed light on conformational changes that take place during the life cycle of activation and substrate binding to the active site of SpeB. Our findings demonstrate that the C-terminal loop has a significant role in the recognition and recruitment of substrates and release of products but is not necessary or sufficient for substrate hydrolysis. These insights and the new peptides will provide excellent opportunities and valuable tools that will aid in further probing the function and biological role of SpeB during S. pyogenes infections.