Insights into the potential function and membrane organization of the TP0435 (Tp17) lipoprotein from Treponema pallidum derived from structural and biophysical analyses

Abstract

The sexually transmitted disease syphilis is caused by the bacterial spirochete Treponema pallidum. This microorganism is genetically intractable, accounting for the large number of putative and undercharacterized members of the pathogen's proteome. In an effort to ascribe a function(s) to the TP0435 (Tp17) lipoprotein, we engineered a soluble variant of the protein (rTP0435) and determined its crystal structure at a resolution of 2.42 Å. The structure is characterized by an eight-stranded β-barrel protein with a shallow “basin” at one end of the barrel and an α-helix stacked on the opposite end. Furthermore, there is a disulfide-linked dimer of the protein in the asymmetric unit of the crystals. Solution hydrodynamic experiments established that purified rTP0435 is monomeric, but specifically forms the disulfide-stabilized dimer observed in the crystal structure. The data herein, when considered with previous work on TP0435, imply plausible roles for the protein in either ligand binding, treponemal membrane architecture, and/or pathogenesis.

Introduction

Syphilis remains a serious global health problem, both in industrialized¹ and in third-world countries.² The persistence of syphilis, the emergence of macrolide resistance in its etiologic agent, Treponema pallidum,³ and the current lack of a vaccine have prompted ongoing studies on the disease and its causative agent.

T. pallidum is an obligate human parasite with a minimal genetic makeup of a single circular chromosome that encodes just over 1000 genes.⁴ Gene products needed for the synthesis of vital metabolites, such as fatty acids, nucleotides, and most amino acids, are absent from T. pallidum. Therefore, the organism must acquire these nutrients from its obligate human host. The epicenter of T. pallidum's nutrient acquisition is its cell envelope, comprising two membranes separated by a periplasmic space.^5,6 Remarkably, the outer membrane has a dearth of detectable proteins,^4,7,8 and many of the periplasmic proteins are membrane-anchored there via lipoyl groups covalently attached to their respective amino termini.⁹ These lipoproteins can be major immunogens,^10,11 and ostensibly carry out vital functions for the bacterium. Because T. pallidum cannot be continuously cultivated in vitro, standard genetic approaches cannot be used to study the functions of its gene products. Accordingly, our efforts have focused on identifying undercharacterized periplasmic lipoproteins and attempting to garner functional information from their crystal structures.^12–19

This report focuses on TP0435 (also referred to as “Tp17” or the “tpp17 gene product”), a 14-kDa lipoprotein that is a known syphilis antigen.^20–22 TP0435 is a potent immunogen despite comprising only about 2% of the detergent-soluble protein complement of T. pallidum.^11,21,22 It also induces proliferation in syphilitic rabbit splenocytes²³ and elicits the production of TNF-α, an inflammation signaling factor, by murine macrophages.^24,25 Additionally, the tp0435 gene is one of the most highly expressed genes in T. pallidum during rabbit infection.²⁶ Previous studies²⁵ had shown that the protein can oligomerize under oxidizing conditions, but is resolved to monomers when treated with reducing agents. Despite this body of information, the function of TP0435 has remained obscure.

As an approach to further clarify potential structure-function relationships for TP0435, we engineered a recombinant form of the protein (rTP0435) lacking lipidation. Subsequently, we crystallized this protein²⁷ and determined its three-dimensional structure at a resolution of 2.42 Å. The structure has an eight-stranded antiparallel β-barrel with a “basin” at one end. Two molecules of the protein are present in the asymmetric unit of the crystals, and they are linked by a disulfide bond. Hydrodynamic studies showed that this link can be specifically formed in solution under oxidizing conditions. The hydrodynamic behavior of rTP0435 and its structural similarity to other proteins suggest possible roles in ligand binding and/or treponemal membrane architecture.

Results

The crystal structure of rTP0435

In the crystal structure of rTP0435, the protein is dominated by an eight-stranded antiparallel β-barrel (Fig. 1). Packed against one end of this barrel is the amino-terminal α-helix (“helix A”). Unlike transmembrane β-barrels (e.g. the barrel domain of FecA²⁸), there is no pore running through the center of rTP0435; instead, the center of the barrel is packed with amino acid side chains. However, one end of rTP0435 has a shallow basin [ Fig. 1(C)]. For illustrative purposes, we term this the “basin” end of the barrel. Because the other end of the barrel houses both termini, we call it the “terminal” end of the barrel. The basin end also has a small 3₁₀ helix (helix B). The topology and numbering of secondary-structure elements is shown in Figure 1(A,B).

Figure 1.

Structural aspects of rTP0435. (A) Ribbons representation of the structure. The β-strands are shown in purple, α-helices are shown in green, and regions with no regular secondary structure depicted are in light blue. The amino- (N) and carboxyl- (C) termini are marked, and the designations of the secondary-structural elements are shown (strands are numbered, helices are lettered). (B) A topology diagram of the structure. The coloration is the same as in part (A). (C) The basin of rTP0435. A cut-away view of the protein's surface is shown. The basin is marked, and basin-lining residues are shown in green. The orientation of the protein is the same as that depicted in part (A). (D) The electrostatic surface potential of rTP0435. A legend to the coloration is shown. The upper part of this panel shows the basin of the protein, rotated 90° toward the viewer compared with panel C. In the lower part, there are two copies of the protein; the two A helices of the proteins meet in the center, and the two basins are oriented to the left and right.

The basin end of the barrel is remarkable for the number of disordered residues occurring there. Of the eight basin-end interstrand loops observed in the two copies of rTP0435 in the asymmetric unit, only one is fully ordered (residues 112–121). This loop is stabilized by contacts with a crystallographic symmetry mate; whereas we term it a “loop,” it actually contains a short span of β-strand (“7a”). Of the 59 disordered residues in the asymmetric unit, 53 are located at this end of the protein. Examination of the crystal packing shows that all the basin ends in the unit cell are oriented toward one another (not shown), but clearly they do not interact with one another in a way that would give rise to a static, interpretable structure.

The surface-electrostatic features of the protein are unremarkable except for a patch of negative potential near to helix A [ Fig. 1(D)]. In the asymmetric unit, two copies of rTP0435 come together to form a continuous patch of negative electrostatic potential. Given this feature's proximity to the proteins' respective amino-termini and the dimeric nature of the protein (see below), it is likely that, in vivo, this feature would be oriented toward the membrane to which the proteins are tethered.

Comparisons with other protein structures

We queried the structure of rTP0435 against the Protein Data Bank (PDB) for similar structures using DALI²⁹ and SSM.³⁰ Unsurprisingly, the top “hits” from such searches were those that had already been identified as structurally homologous using sequence queries alone.^27,31 The top match was a “putative lipoprotein” from Shewanella oneidensis (accession code: 3LHN; no attendant publication). This protein has only 25% amino acid identity with TP0435. SSM matching demonstrated that rTP0435 has a root mean square deviation (r.m.s.d.) of 1.6 Å over 82 comparable C_α atoms. Other notable hits included a putative lipoprotein from Streptococcus pneumoniae (3GE2; r.m.s.d. of 1.6 Å over 66 comparable C_α's; 15% amino acid identity) and the amino-terminal domain of NlpE from Escherichia coli³² (2Z4I; r.m.s.d. of 1.7 Å over 69 comparable C_α's; 25% amino acid identity). Only NlpE has a known function; it is involved in the CpxA/R stress-response pathway.³³ It is also thought to be involved in copper homeostasis, a function putatively mediated by a CXXC motif in the protein.^32,34 Whereas the homology to NlpE is significant, other data weaken the case for an NlpE-like function for TP0435. For example, no homologues to the E. coli CpxA or CpxR proteins are known in the T. pallidum proteome. It therefore appears that this signaling system is not present in the spirochete. Further, although a CXXC motif is present in the immature TP0435 primary sequence, the most likely cleavage point (between S24 and C25 in the numbering of the immature sequence) of the protein following lipidation eliminates one of these cysteines. The processed C1 is followed by another cysteine residue four residues away (CTTVC), but engagement of copper by this membrane-proximal motif seems improbable. Finally, NlpE contains a carboxyl-terminal, five-stranded β-barrel domain not present in TP0435. Given these differences, TP0435 probably does not function similarly to NlpE.

Secondary-structure matching reveals another intriguing structural homology: avidin from Gallus gallus. The best match to rTP0435 comes from the structure 1LEL,³⁵ avidin complexed to biotinyl p-aminocaproic acid (BCAP). In avidin's equivalent of the “basin end,” it binds this biotin mimic. Other avidin structures^35,36 also demonstrate that biotin and other ligands bind here. Given this structural homology, it seems possible that the function of TP0435 is to bind a small molecule or other partner at its basin end; binding of the partner could be accompanied by a disorder-to-order transition in the basin-end interstrand loops. Notwithstanding this attractive hypothesis, biotin itself is an improbable target of the molecule. The biotin-avidin dissociation constant is extremely low. Thus, if rTP0435 bound similarly to biotin, which is available in the cytoplasm of E. coli, it likely would have co-purified with the protein. There is, however, no evidence for a small-molecule ligand bound to rTP0435 in our electron-density maps. Additionally, amino acids known to bind to biotin and biotin analogues in avidin are not conserved in TP0435 (Table I), and the analogous residues in the treponemal protein often have different chemical characteristics than their avidin counterparts (not shown). Despite the low likelihood of TP0435 binding biotin, eight-stranded β-barrel motifs have been shown to bind to diverse ligands, such as retinol³⁷ and pheromone 2-(sec-butyl)thiazole.³⁸ Given this flexibility, the possibility that TP0435 binds a ligand in the periplasm cannot be discounted.

Table I.

Biotin-Contacting Avidin Residues and their Analogs in rTP0435

Avidin residue	rTP0435 analog
L214a	L29
S216	T39
Y233	K51
T277	E94
W297	G110
N318	Y125

^aOnly residues that both contact biotin and have clear structural analogs in rTP0435 are listed.

The quaternary structure of rTP0435

As noted above, there are two copies of rTP0435 in the asymmetric unit of the crystals. These two molecules, which contact each other at their respective terminal ends, are related by a pseudo-twofold rotational axis (Fig. 2). These two copies are very similar structurally, exhibiting only a 0.4 Å r.m.s.d. between 90 comparable C_α atoms. At this axis, there is a disulfide bond that links the S_γ atoms of C18 of the respective molecules [ Fig. 2(B)]. This raised the possibility that TP0435 exists as a covalently stabilized dimer in solution. We submitted the model to the PISA server³⁹ to examine putative oligomeric assemblies of the protein. The only potentially stable oligomer that was found was this homodimer, which has 1300 Ų of surface area buried at the monomer-monomer interface. Although intermolecular disulfide bonds occurring at two-fold axes are rare, there are several examples in the database. Among them are signaling molecules (bone morphogenic protein-2, tissue growth factor β3, myostatin^40–42) and a CAP (cysteine-rich secretory/antigen 5/pathogenesis-related 1^43,44) protein from Ostertagia ostertagi (ASP-155); the role of this latter protein is poorly defined, but it is secreted⁴⁴ and therefore is likely to be involved in host/pathogen interactions.

Figure 2.

The disulfide-linked dimer of rTP0435. (A) The contents of the asymmetric unit. Coloration is as described in Figure 1(A). The respective C18s of the two molecules are shown as ball-and-stick models, with the carbon atoms drawn in green and the sulfur atoms in yellow. The black symbol in the middle of the figure represents the pseudo-twofold noncrystallographic axis present in the asymmetric unit. (B) Evidence for the disulfide bond. The orientation of the molecules is identical to that in part (A). A kicked mF_o–DF_c omit map is shown, contoured at the 3-σ level (both C18 residues had been omitted from the calculation). The map is superposed on the final refined coordinates of the structure. Secondary-structural elements are shown faded for clarity.

To investigate whether this dimer existed in solution, we used analytical ultracentrifugation in the sedimentation velocity (SV) mode. The c(s) distribution shown in Figure 3 demonstrates that, as purified, rTP0435 is a monomer. However, when allowed to incubate at room temperature for 3 days at pH 8.5, a significant peak for dimeric rTP0435 was evident. The formation of the dimer could be prevented by the inclusion of 1 mM tris(2-carboxyethyl)phosphine (TCEP) in the incubation buffer. Dimer formation is therefore dependent on an oxidizing environment. There are two free cysteines in the rTP0435 construct that could cause a disulfide linkage in solution: C18 [ Fig. 2(B)] and C34, which is in a disordered loop region between β-strands 1 and 2. To ascertain which of these is responsible for the disulfide-linked solution dimer, we generated the rTP0435 mutant C18S. After performing the same high-pH incubation strategy as with the wild-type protein, we observed almost no dimer formation (Fig. 3). This experiment thus confirmed that a large majority of the observed disulfide-stabilized dimer is specifically formed between the C18's of two copies of rTP0435.

Figure 3.

Hydrodynamic properties of rTP0435. The c(s) distributions for SV experiments under differing conditions are shown. The peaks corresponding to monomer (M) and dimer (D) are marked. A legend to the coloration is shown in the figure; the abbreviation “l.i.” means “long incubation” (72 h).

Whenever considering the oligomeric status of a lipoprotein, it is important to examine whether the proposed state is compatible with the known topological constraint of membrane tethering at the amino-termini of the proteins. In the structure of rTP0435, the first visible amino acid residue in both copies is K10. There are thus nine residues forming the presumably flexible tethers between the membrane anchor and the observed structure. If fully extended, the tethers would reach a length of about 35 Å. Because both amino-termini of the dimer components are oriented in the same direction, it is plausible that this assembly exists in vivo. This proposed arrangement orients the basin ends of the dimer components away from each other and parallel to the membrane (Fig. 4, right).

Figure 4.

The modeled in vivo topology of the TP0435 dimer. One monomer is shown in blue, the other in salmon. Carbon atoms are colored according to their respective protomer, and the sulfur atoms are colored yellow. The “reduced” state of the monomers is shown on the left, but, under oxidizing conditions, the disulfide-linked dimer may be formed (right). The positions of the C18 residues and the approximate positions of the C34's are indicated on the right side of the figure.

Discussion

The crystal structure of rTP0435 reveals a disulfide-stabilized dimeric β-barrel structure (Figs. 1, 2, and 4). Hydrodynamic studies (Fig. 3) demonstrated that this structure also exists in solution. These observations raise an important question: does this oligomer exist in vivo (i.e. in the natural periplasmic/membrane proximal milieu of T. pallidum)? Previous studies on this protein²⁵ examined the mobility of Tp17 isolated from whole T. pallidum cells on detergent-permeated polyacrylamide gels. The predominant oligomeric form of the protein observed in those experiments was monomer, with a minority of the protein migrating as a dimer that was resolvable with the addition of the reducing agent 2-mercaptoethanol. Almost as abundant as the dimer was a polymeric form of the protein that was too large to cross the barrier between the stacking and separating gels. These data suggest that disulfide-linked forms of the protein can exist in vivo. Thus, joint consideration of the current and previous results implies that the disulfide-linked form that we observed in the crystal structure is present in vivo.

Another means of exploring whether the disulfide-linked dimer is functionally relevant is to examine whether C18 is conserved in TP0435-like proteins in other organisms. However, besides matches of 100% identity in various subspecies of T. pallidum (including T. pallidum spp. pertenue, the etiologic agent of yaws⁴⁶) and in T. paraluiscuniculi, only low-homology matches to TP0435 are known at this time. These latter homologues do not have conserved cysteines homologous to C18. It is striking that homologues of this protein do not even occur in many other treponemes (e.g. T. denticola and T. phagedenis). It appears that TP0435 and its disulfide-stabilized dimer perform very specific functions that are relevant only to two pathogenic treponemes: T. pallidum and its subspecies (human pathogens) and T. paraluiscuniculi (a rabbit pathogen). Given that all of these organisms cause treponematoses,^45,47 TP0435 may be important for the specific mode of pathogenesis employed by these bacteria.

Interestingly, the mutation C18S drastically diminished but did not eliminate disulfide formation (Fig. 3). A small amount of dimer was observed in C18S, and its formation could be prevented by the addition of TCEP. These observations demonstrate that disulfide links between C34's are also possible, because C34 is the only cysteine available to form disulfide bonds in the C18S protein. In the proposed topological model of TP0435 (Fig. 4), the C34's of the dimer are located facing away from one another on the basin ends of the dimer. This fact opens the possibility that the C18-linked dimers may polymerize via their exposed C34's, forming a large, linear aggregate. This might explain the existence of the large aggregates of TP0435 observed in SDS-PAGE experiments described above.²⁵ The functional consequences of the existence of such aggregates in the periplasm are unclear. However, a long, linear array of membrane-tethered proteins could have significant architectural effects on the membrane to which they are attached.

That the oligomerization status of TP0435 depends on the solution reduction potential (see Fig. 3 and Akins et al.²⁵) underscores its potential to serve as a redox sensor in the periplasm of T. pallidum. In many Gram-negative organisms, the DsbA/B system is responsible for facilitating the formation of disulfide bonds in the periplasm.^48,49 However, the limited genome of T. pallidum does not have any apparent homologs to this redox machinery. The redox (and thus oligomerization) state of TP0435 molecules are therefore probably subject to local variations in the reduction potential of their in vivo environment(s). Under this scenario, two TP0435 monomers could come together under oxidizing conditions and become covalently linked (Fig. 4), or vice versa under reducing conditions. There is a precedent for the disulfide-mediated oligomerization state of a protein acting as a redox sensor in bacteria: the RegB protein of Rhodobacter capsulatus.^50,51 In this system, sufficiently oxidizing conditions in the R. capsulatus cytoplasm induce the formation of an intermolecular disulfide bond, transforming the dimeric RegB to a tetrameric form, which results in the diminution of its native histidine-kinase activity.⁵¹ If TP0435 does serve as a redox sensor, presumably it must communicate its redox/oligomerization state to downstream effectors so that a response may be mounted. The identity and nature of this hypothetical signal-transduction pathway is unknown.

Another possible function of TP0435 inferred by the structure is ligand binding. The resemblance of the protein to avidin is marked. However, a structural alignment of rTP0435 and avidin showed that key biotin-contacting residues in the former protein are not conserved in the latter (Table I). While these dissimilarities seem to rule out biotin binding by TP0435, we cannot rule out the idea that another small molecule binds in the basin of the protein. Another possibility is that the basin-end binds to a proteinaceous partner. A recent exhaustive yeast two-hybrid analysis of the T. pallidum proteome⁵² shows several potential binding partners; the three most probable were TP0467 (an 82-amino acid hypothetical protein), TP0629 (another hypothetical protein), and TP0726 (a 65-amino acid putative flagellar protein). The former two targets do not seem likely, as they apparently do not exist in the periplasm. It is tempting to ascribe meaning to the TP0726 interaction because the T. pallidum flagella exist in the periplasm.^5,6 However, the function and location of TP0726 is unknown. In the past, we have identified treponemal periplasmic binding partners by examination of the relevant operons.^17,18 However, the tp0435 gene is apparently monocistronic, making this avenue of discovery unfruitful.

Our structural and hydrodynamic studies offer several substantial clues regarding the function of TP0435. However, they do not definitively elucidate the protein's function. Additional studies to probe the potential ligand- or protein-binding proclivities of the protein are warranted. Also, an expanded exploration of the in vivo state of the protein could offer further insights into the relevance of the potential covalent linear arrays. Finally, an examination of the role of TP0435 in pathogenesis could reveal why it is present only in three pathogenic treponemes.

Materials and Methods

Protein preparation, crystal growth, and X-ray diffraction data collection

Details are presented in a previous publication.²⁷ To summarize, rTP0435 (lacking acylation) was hyperexpressed in E. coli cells with an amino-terminal SUMO tag. Following affinity purification, SUMO-specific protease (LifeSensors, Malvern, PA) was used to remove the tag, and the protein was subjected to size-exclusion chromatography in Buffer A (20 mM HEPES pH 7.5, 100 mM NaCl, 2 mM n-octyl β-d-glucopyranoside (β-OG)) and concentrated to 11 mg/mL. Crystallization was carried out using the hanging-drop vapor-diffusion method with a solution of 40% (v/v) PEG 400, 0.2M lithium sulfate, and 0.1M Tris pH 8.5 comprising the reservoir. The crystals were cryoprotected in the reservoir solution supplemented to 60% PEG 400, then plunged into liquid nitrogen. Data were collected at the Structural Biology Center of the Advanced Photon Source of Argonne National Laboratory. The diffraction data had a d_min spacing of 2.4 Å and the symmetry of space group R3 (unit-cell dimensions a = b = 85.7 Å, c = 85.4 Å in the hexagonal setting). The R_merge for all data was 0.048, and the Wilson B-factor was 68.1 Ų.

Site-directed mutagenesis and protein concentration determination

The C18S mutation was introduced into the plasmid carrying the wild-type rtp0435 sequence using the QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). The mutation was confirmed by DNA sequencing. The mutant protein was expressed and purified as described earlier.²⁷ Protein concentrations were determined in Buffer A using spectrophotometry. Extinction coefficients were calculated using the Protparam tool⁵³ of ExPASy (http://www.expasy.org).

Structure determination

The crystal structure of rTP0435 was determined using molecular replacement in the program Phaser.⁵⁴ The search model was based on that of a “putative lipoprotein” from Shewanella oneidensis (PDB accession code 3LHN). Based on a structure-aided sequence alignment,⁵⁵ all atoms of identical residues in the model were retained; nonidentical residues were truncated at the last common side chain atom. Using this model, a promising solution was obtained; two copies of the model were placed with a log-likelihood gain of 151. However, difference electron-density maps generated using these model phases were poor, as evidenced by main chain breaks in the density and low correlations between model side chains and electron-density features. These phases were improved using the NCSREF script available in the CCP4⁵⁶ package. Briefly, the script calculated phases from the molecular-replacement model, determined the noncrystallographic (NCS) operators from this model, and improved the calculated phases with the solvent-flattening, histogram-matching, and NCS-averaging algorithms available in dm.⁵⁷ The model was then refined against the data and these improved phases in Refmac5,⁵⁸ then again for one cycle without the phases. The procedure was iterated several times, resulting in an averaged map from dm. This map was significantly improved compared with the initial difference electron-density map, and it was sufficient to allow for correction of errors in the model using Coot.⁵⁹ The corrected model was initially refined in PHENIX⁶⁰ using the simulated-annealing, positional, and group B-factor refinement protocols using NCS restraints. In later stages of refinement, the NCS restraints were released, and the positional, TLS, and individual B-factor refinement protocols were used. Riding hydrogen atoms were used throughout refinement. Stereochemistry and B-factor restraints were determined using an automated procedure in phenix_refine. The final model has good geometry (Table I), with a final R-value of 0.231 and an R_free-value of 0.275. Numerous residues in loop regions could not be reliably located in the electron-density maps and were thus not modeled (see Table II). Structural figures were rendered in PyMol (Schrödinger LLC, Portland, OR). The structure of rTP0435 has been deposited in the Protein Data Bank with accession number 4U3Q.

Table II.

Refinement Statistics

Resolution (Å)a	37.0–2.4 (2.55–2.4)
No. reflections	9,024 (1294)
Completeness (%)	99.1 (95.0)
No. protein atoms	1,473
No. waters	1
Rwork	0.231
Rfree	0.275
Avg. B-factor, protein (Å2)	75.3
Avg. B-factor, solvent (Å2)	44.7
r.m.s. Deviations
Bond lengths (Å)	0.003
Bond angles (°)	0.7
Maximum-likelihood coordinate error (Å)	0.4
Ramachandran distributionb
Favored (%)	97.7
Allowed (%)	100.0
Missing residues	A: 9, 30–36, 55–62, 85–93, 115–119, 131–132; B: 9, 32–35, 55–63, 86–94, 131–132
Accession number	4U3Q

^aValues in parentheses represent those for the highest resolution shell.

^bAs defined by Molprobity.⁶²

Analytical ultracentrifugation

Sedimentation velocity (SV) analytical ultracentrifugation experiments were performed in a model XL-I (Beckman-Coulter, Indianapolis, IN) centrifuge at 20°C. Charcoal-filled Epon centerpieces were sandwiched between sapphire windows in an aluminum housing. The sectors were filled with either buffer in the reference sector or protein in the sample sector. The proteins were present in either Buffer A (see above) or in Buffer B (20 mM Tris pH 8.5, 100 mM NaCl, 2 mM β-OG). In some experiments, the buffers were supplemented to 0.5 mM in TCEP. After dilution to the experimental concentrations from stock solutions, some proteins were dispensed into the cells immediate, while others were incubated at room temperature for 72 h before dispensation. After insertion into an An50-Ti rotor, the assembled centrifugation cells were equilibrated at the experimental temperature under vacuum in the centrifuge for at least 2.5 h. Centrifugation at 50,000 rpm was then commenced, and absorbance optics tuned to 280 nm were used to acquire the concentration-profile data. The data were analyzed using the c(s) distribution model in SEDFIT.⁶¹ All figures derived from these analyses were generated in the program GUSSI (http://biophysics.swmed.edu/MBR/software.html).

Glossary

ASP

Ancylostoma-secreted proteins

BCAP

biotinyl p-aminocaproic acid

β-OG

n-octyl β-d-glucopyranoside

CAP

cysteine-rich secretory/antigen 5/pathogenesis-related 1

r.m.s.d.

root-mean-square deviation

rTP0435

the recombinant version of TP0435

TCEP

tris(2-carboxyethyl)phosphine