|Home | About | Journals | Submit | Contact Us | Français|
Metalloaminopeptidases (mAPs) are enzymes that are involved in HIV infectivity, tumor growth and metastasis, angiogenesis, and bacterial infection. Investigation of structure-function relationships in mAPs is a prerequisite to rational design of anti-mAP chemotherapeutics. The most intensively studied member of the biomedically important dinuclear mAPs is the prototypical secreted Vibrio proteolyticus di-zinc aminopeptidase (VpAP). The wild-type enzyme is readily purified from the supernatant of cultures of V. proteolyticus, but recombinant variants require expression in Escherichia coli. A greatly improved system for the purification of recombinant VpAP is described. A VpAP-(His)6 polypeptide, containing an N-terminal propeptide, and a C-terminal (His)6 adduct, was purified by metal ion affinity chromatography from the supernatant of cultures of E. coli. This single step replaced the sequence of (NH4)2SO4 fractionation, and anion exchange and hydrophobic interaction chromatographic separations of earlier methods. Traditionally, recombinant VpAP proenzyme has been treated with proteinase K and with heat (70 °C), to remove the N- and C-terminal regions, and yield the mature active enzyme. This method is unsuitable for VpAP variants that are unstable towards these treatments. In the new method, the hitherto noted, but not fully appreciated, ability of VpAP to autocatalyze the hydrolysis of the N-terminal propeptide and C-terminal regions was exploited; extensive dialysis of the highly purified VpAP-(His)6 full-length polypeptide yielded the mature active protein without recourse to proteinase K or heat treatment. Purification of variants that have previously defied isolation as mature forms of the protein was thus carried out.
Metalloaminopeptidases (mAPs) are key players in many biological processes, including cellular targeting of proteins [1, 2], protein degradation [1, 3, 4], aging [5-10], tissue repair [11-15], cataract formation [16-22], HIV infectivity [23-26], angiogenesis [27-31], and carcinogenesis [27, 32-34]. A biomedically important family of mAPs is that of the dinuclear mAPs, also known as “cocatalytic” mAPs . The dinuclear mAPs hydrolyze the N-terminal amino acid from peptides and proteins, with varying degrees of substrate specificity, and are characterized by an active site that contains two transition ions, usually Zn(II), separated by about 0.35 nm (3.5 Å) in a dinuclear active site. The most intensively studied and well characterized of these enzymes is the 32kD secreted di-zinc leucine aminopeptidase, VpAP, from Vibrio proteolyticus (formerly Aeromonas proteolytica) [36-40]. It appears that one of the metal ions in VpAP and related mAPs is involved in nucleophile activation, and is essential for catalysis, whereas the other is involved in substrate activation and stabilization of the transition state [36, 39, 41]; the non-essential metal ion may thus be key to substrate specificity, though conclusive evidence is lacking. Other dinuclear mAPs that have been studied include the angiogenesis target methionyl aminopeptidase, and the secreted aminopeptidases that are virulence factors in the pathogenicity of certain species of Clostridium, Streptomyces, Vibrio and Aeromonas [42-48]. VpAP itself exhibits a very high degree of structural homology with the protease domain of the prostate-specific membrane antigen (PSMA) . PSMA is highly expressed in prostate cancer cells , but its role in prostate cancer is currently unknown. Some potent inhibitors of VpAP and its homologs, including bestatin (ubenimex), ovalicin, and fumagillin, have also been shown to exhibit in vivo activity against tumor growth, angiogenesis, and HIV infectivity [8, 24, 51-59]. Wild-type VpAP is readily available in large quantities from V. proteolyticus, and it is an excellent model protein with which to explore the structure-function relationships of mAPs: studies on VpAP have provided information that may lead to the design of more effective and specific chemotherapeutic agents [35, 39-41, 60]. However, recent work on VpAP and VpAP homologs has shown that site-directed recombinant variants can provide important mechanistic information that is unavailable with the native proteins [61-64]. A need exists, therefore, for an efficient expression system and purification protocol for recombinant VpAP that can provide sufficient quantity of high-purity material for physicochemical, structural, and spectroscopic study of site-directed variants of VpAP.
VpAP has been isolated from cultures of V. proteolyticus since the mid-1960s. Investigation of the proteolytic activity associated with V. proteolyticus  led to the identification and isolation of VpAP by Prescott and Wilkes . Subsequent refinements of the purification procedure [67, 68] resulted in a protocol that yielded VpAP preparations that were sufficiently homogenous for crystallization. The purification of VpAP is complicated by the fact that the 32 kDa mature active protein is the product of substantial processing of a 54 kDa translated polypeptide, VpAP*, coded for by the native ORF [69, 70]. The 54 kDa VpAP* polypeptide consists of a 21 amino acid signal peptide, an 85 amino acid N-terminal propeptide, a 299 amino acid mature region, and a 99 amino acid C-terminal propeptide (Figure 1). The signal peptide is required for secretion, the N-terminal propeptide is required for folding but its presence inhibits catalytic activity, and the C-terminal propeptide is of undetermined function but appears to be species-specific [71, 72]. VpAP* is processed to the final 32 kDa mature protein only after secretion and during purification; roles in the processing of VpAP* have been proposed for VpAP* itself (and VpAP) in autoprocessing, for a co-expressed neutral protease, and for a heat-treatment step that has been traditionally employed [73-75]. An additional problem in the purification of VpAP for spectroscopic studies is the presence of two low Mr (< 2kD) brown pigments, one of which contains spectroscopically active Fe(III), that appear to be closely associated with VpAP; any successful protocol must separate these pigments from VpAP .
The commonly and successfully used protocol for the purification of VpAP from V. proteolyticus involves (i) rapid separation of the supernatant from V. proteolyticus cells; (ii) (NH4)2SO4 fractionation of the supernatant; (iii) heat treatment (70 °C for 5 – 7 h), followed by dialysis and clarification by centrifugation and filtration (0.22 μm); (iv) one or two hydrophobic interaction chromatographic fractionations (octyl-Sepharose); and (v) anion-exchange chromatography (Q-Sepharose). A highly detailed investigation into the requirements for the isolation of VpAP heterologously expressed in E. coli has been described by Bzymek, Holz and coworkers , and the resulting protocol has yielded both recombinant wild-type and site-directed variant forms of VpAP for subsequent crystallographic and physicochemical studies [38, 61-63, 71]. As suggested by earlier work [72, 77-79], it was found that VpAP was best expressed as a 52 kDa proenzyme (ΔspVpAP*; lacking the first 23 amino acids); the use of an appropriate leader sequence (e.g. the pelB sequence of pET26b(+)) results in secretion of the proenzyme, which can be purified from the supernatant. Essentially the same purification procedure as for the native enzyme from V. proteolyticus was adopted, except for an additional incubation with proteinase K (PK), which appears to mimic the effects of V. proteolyticus neutral protease and removes the N-terminal propeptide, prior to heat treatment.
In the course of our own studies, we have followed the earlier procedure  for the purification of recombinant wild-type and variant VpAP and have successfully isolated VpAP with amino acid substitutions in the active site pocket and in an adjacent hydrophobic pocket . However, we have also encountered problems, including (i) irreproducible, and generally low, yields of VpAP; (ii) unpredictability of the effects of PK on variant forms of VpAP*; and (iii) the inability to isolate some site-directed variants as the mature proteins, notably variants with substitutions of Asp99. In the present work, we describe a rapid purification scheme for heterologously expressed VpAP that reproducibly yields 10 mg of pure wild-type VpAP per liter of E. coli culture. In addition, we describe the expression and purification of the D99A, D99H, and D99M variant forms of VpAP, that had hitherto defied isolation as mature proteins. The advantages of the new method are discussed.
Restriction enzymes, T4 DNA ligase, and Antarctic phosphatase were purchased from New England Biolabs (Ipswich, MA). A dNTP cocktail and Pfu Turbo polymerase were purchased from Stratagene (La Jolla, CA). BL21 Star DE3 One Shot Chemically Competent E. coli cells were purchased from Invitrogen (Carlsbad, CA). LB medium and kanamycin were purchased from Fisher Scientific (Fair Lawn, NJ). IPTG was obtained from Anatrace Incorporated (Maumee, OH), aliquoted into 1 ml volumes of 1 M concentration, and stored at −20 °C until use. Pre-cast PAGE gels were purchased from BioRad Laboratories (Hercules, CA). Agarose bead-linked proteinase K from Tritirachium album was purchased from Sigma-Aldrich (St. Louis, MO). All other general chemicals (ACS grade) and buffer salts (Biotech or electrophoresis grade) were obtained from Sigma-Aldrich (St. Louis, MO). Deionized water was additionally filtered for organic and metal ion contaminants using a Milli-Q system (Millipore, Molsheim, France) and exhibited a resistance of > 18 MΩ.
cDNA was amplified by PCR from the plasmid pVSNMC , obtained from the Applied Enzymology Laboratory of the National Food Research Institute (Ibaraki, Japan), using the forward and reverse primers 5′-GGCTTCCATGGAAGACAAAGTGTGGATCTCA-3′ (forward; NcoI site underlined), and 5′-GCGCGCCTCGAGCTGATTGCC-3′ (reverse; XhoI site underlined). The PCR product was then inserted between the NcoI and XhoI sites of pET26b(+) (Invitrogen, Carlsbad, CA). The reverse primer lacked the VpAP native stop codon, thus transcription continued through the (His)6-coding sequence immediately downstream of the XhoI site in pET26b(+), and was stopped immediately after (His)6. The resulting construct, pAPMH, was used to express a wild-type polypeptide (ΔspΔCVpAP*(His)6) containing the N-terminal propeptide, the mature VpAP polypeptide, and a C-terminal (His)6 construct, but lacking the signal sequence and the C-terminal propeptide. pAPMH was also used as the template to generate VpAP site-directed variants.
Recombinant DNA for the D99A, D99M, and D99H variants of VpAP was generated using the following primers (mutations underlined):
A six-step PCR protocol was employed with the following parameters: step 1, 95 °C for 1 min; step 2, 95 °C for 30 s; step 3, primer Tm − 10 °C, 1 min; step 4, 68 °C for 8 min; step 5, 18 cycles of steps 2 – 4; step 6, 68 °C for 10 min.
All constructs were fully sequenced (Applied Biosystems ABI PRISM 3100 analyzer, Protein and Nucleic Acid Facility, Medical College of Wisconsin, Milwaukee, WI) and found to contain no mutations except for those intended. Sequence-verified constructs were used to transform BL21 Star (DE3) One Shot Chemically Competent E. coli cells (F- ompT hsdSB (rB-mB-) gal dcm rne131 (DE3)), using the transformation protocol provided by the manufacturer (Invitrogen, Carlsbad, CA). Frozen stocks were prepared by adding 1 ml bacterial culture (OD600 nm ≈ 0.6) to 1 ml 30% glycerol (in water), snap frozen in liquid nitrogen, and stored at −80 °C.
Starter cultures were prepared by inoculation of 20 ml LB, containing 30 μg/ml kanamycin, with a scraping from frozen stock cultures, followed by agitation at 250 rpm, 37 °C, in a 50 ml conical tube, until the OD600 nm reached 1.0. Cultures for VpAP expression were prepared by adding 2.5 ml of starter culture to a 1 L flask containing 250 ml LB (30 μg/ml kanamycin). Flasks were agitated in an orbital shaker at 250 rpm, 37 °C until OD600 nm = 0.6. IPTG was added (to a final concentration of 1 mM), and flasks were further agitated for 18 h, 250 rpm, 30 °C.
Steady state VpAP activity was estimated from spectrophotometric determination of the rate of formation of the product, p-nitroaniline (PNA), of hydrolysis of L-leucine-p-nitroaniline (LPNA; Sigma, St. Louis, MO), at 405 nm, 25 °C. The assay mixture contained 1.0 mM LPNA in 10 mM Tricine buffer, pH 8.0 [67, 80] and a differential extinction coefficient Δε405 nm = 10,800 M-1 cm-1 was employed . A “Unit” of activity was defined as a rate of hydrolysis of 1 μmol LPNA min-1 under these conditions. Apparent kinetic parameters K′m and k′cat were estimated by collecting the entire progress curve, with initial [LPNA] = 0.1 mM, and fitting the part of the curve that exhibited a linear relationship between t-1[ln([S0]/[St])] and t-1[S0 − St] [39, 82]. This approach indicated that the steady state assay routinely underestimated Vmax by a factor of 1.3, likely because of weak substrate inhibition, but steady state values were determined for wild-type VpAP for comparison to other studies. Quantities of ΔspΔCVpAP*(His)6 were estimated spectrophotometrically, taking ε280 nm = 58,110 M-1 cm-1, and purified mature VpAP protein was estimated taking ε280 nm = 43,950 M-1 cm-1 [76, 83-85]. SDS-PAGE analysis was carried out on precast 4 - 15 % polyacrylamide/Tris-HCl gels (BioRad “Ready Gels”). Samples of VpAP were generally loaded directly onto SDS-PAGE gels without pretreatment. Occasionally, samples were heated for 5 min at 95 °C in the presence of either 33 mM dithiothreitol, 66 mM β-mercaptoethanol, or both, prior to SDS-PAGE analysis. These pretreatments were found to diminish the intensities of bands showing anomalously high apparent Mr, but did not affect the number or the positions of the bands. SDS-PAGE analyses presented herein were carried out on samples that had not been pretreated, except where explicitly stated in the figure legend (i.e. only Figure 4).
Cultures were clarified by centrifugation (5200 × g, 10 min, 4 °C), and 1 L supernatant was applied (3.0 ml/min, 4 °C) to 15 ml TALON Superflow Metal Affinity Resin (Clontech, Mountain View, CA). The resin was washed with 75 ml of 50 mM HEPES buffer, pH 7.5, at 1.0 ml/min. HEPES buffer (50 mM, pH 7.5), containing imidazole (20 mM), was applied (3.0 ml/min) to effect protein elution. Fractions (6 ml) were assayed for activity and protein content. In order to speed future identification of fractions containing site-directed variants of VpAP that may show no detectable activity, the retention volumes for the fractions of wild-type VpAP with highest specific activity were characterized from the commencement of application of imidazole. Fractions containing protein and expressing VpAP activity were combined and dialyzed at 4 °C against 50 mM HEPES buffer, pH 7.5, to dilute imidazole to a final concentration of < 1 nM.
The dialysate from Step A was agitated (250 rpm, 37 °C) with ≈ 5 mg (≈ 0.025 units; ≈ 1 μg protein equivalent) PK agarose beads. Aliquots were periodically assayed for VpAP activity toward LPNA. When the activity toward LPNA ceased to increase upon further incubation with PK (generally ≈ 40 min), the VpAP-containing suspension was filtered to remove the PK beads (0.22 μm, Millipore mixed cellulose esters [MCE] syringe filter).
The solution from Step B was immediately incubated at 70 °C in a water bath, either until the activity began to decrease, or else for 1 h. The suspension was cooled on ice, filtered (0.22 μm), and dialyzed (50 mM HEPES, pH 7.5, 4 °C).
The solution from Step C was applied (2.0 ml/min, 4 °C) to a 15 ml TALON Co(II)-affinity resin column. HEPES buffer (50 mM, pH 7.5) was applied, and eluate fractions that contained the highest specific activity toward LPNA were pooled and concentrated by ultrafiltration (Amicon PM10). As in Step A, retention volumes were characterized.
Step A was carried out as for wild-type VpAP. Because the activities of the D99-substituted variants were extremely low at this stage, fractions were collected using the retention volumes characterized for wild-type. Protein content was estimated from A280 nm. Steps B and C, however, were replaced by Step E, which entailed extended (7 – 14 d) dialysis against HEPES buffer (50 mM, pH 7.5) at 4 °C. Self-processing of VpAP D99-substituted variants was monitored by SDS-PAGE on 4 – 20 % polyacrylamide gradient gels. When SDS-PAGE and N-terminal amino acid sequencing indicated complete processing, the dialysate was filtered (0.22 μm), and further processed as described for Step D for wild-type VpAP. The previously characterized retention volumes for wild-type VpAP were used as an aid to identifying fractions containing VpAP D99-substituted variants.
Proteins were separated on SDS-PAGE gels, transferred to polyvinylidene fluoride membranes, excised, and N-terminal sequenced by Edman degradation using a Beckman Porton LF3000G sequencer with in-line Applied Biosystems PTH analyzer, at the Protein and Nucleic Acid Core Facility at the Medical College of Wisconsin.
Circular dichroism spectra were recorded at 25 °C using a JASCO J-710 spectropolarimeter equipped with a 1 mm path-length quartz cell. Samples for spectropolarimetry were equilibrated with 5 mM phosphate buffer, pH 7.5, by dialysis.
VpAP was crystallized from a hanging drop. VpAP (0.5 mM) in 10 mM Tris buffer, pH 8.0, containing 10 mM KSCN and 400 mM NaCl was equilibrated with 100 mM Tris buffer, pH 8.0, containing 100 mM KSCN and 4.5 M NaCl, at 19 °C.
Diffraction data were collected on a single crystal of heterologously expressed wild-type VpAP, with Paratone-N (Hampton) as a cryoprotectant, at −160 °C on a Rigaku R-AXIS IV++ diffractometer using Crystal Clear software . An oscillation step size of 1.0° was used to collect 3 min exposures. Data were processed and unit cell parameters determined with HKL2000 . Phases were derived from a structure previously solved in this laboratory (accession code 2IQ6 ) and used to solve the structure by molecular replacement using Phaser . Model building was performed with the graphic program COOT . The structure was manually inspected with electron density maps and with 2Fobs − Fcalc and Fobs − Fcalc coefficients. Structures were refined using iterative cycles of energy minimization using REFMAC5  in the CCP4 program suite , alternating with manual map-fitting and model rebuilding. Progress of the refinement was confirmed by the steady decrease in both the Rcrystal and Rfree values. Procheck was used to check Ramachandran (, ψ) values .
10 L of E. coli culture, OD600 nm = 2.3, yielded 20 g wet weight of cells, which were discarded. Up to 10 mg of wild-type VpAP, with a steady state ([LPNA] ≈ 1 mM) specific activity of 120 Units/mg at 25 °C, was purified from the supernatant. The purification of wild-type VpAP is summarized in Table 1. Kinetic analysis indicated apparent Michaelis-Menten parameters k′cat = 82 ± 2 s-1, and K′m = 18 ± 1 μM.
VpAP-containing fractions that eluted during Co(II)-affinity chromatography upon the application of imidazole were characterized by three bands on SDS-PAGE, at apparent Mr ≈ 57 kDa, ≈ 42 kDa, and ≈ 32 kDa (bands A, B, and C of Figure 2). The relative intensities of the bands appeared to be the same for each fraction. The origin of the the “57 kDa” band, A, is unclear, as the cDNA coded for the 42kD ΔspΔCVpAP*(His)6 polypeptide, but that material evidently bound the Co(II) affinity resin with an affinity comparable to that for the material responsible for the other two bands. N-terminal amino acid sequencing of the other two bands, B and C, indicated the same N-terminal sequence of EDKVWI for each, corresponding to the N-terminal region of the 85 amino acid N-terminal propeptide, and these bands were only observed in fractions that eluted upon the application of imidazole, so contain the (His)6 adduct. Despite the apparent Mrs from SDS-PAGE, then, both bands B and C must be due to ΔspΔCVpAP*(His)6. The specific activity of the material following Step A was ≈ 6 Units/mg, consistent with the retention of the inhibitory N-terminal propeptide in > 95 % of the polypeptide molecules.
The anomalous behavior of VpAP on SDS-PAGE is worthy of comment. The apparent Mr of proteins from SDS-PAGE can be very sensitive to even very minor differences in physicochemical properties of polypeptides in some cases [94, 95]. Here, the origin of the anomalous behavior on SDS-PAGE is unknown, but may be related to the presence of cysteines with the potential to form disulfide bonds. Mature VpAP contains only two cysteines, Cys223 and Cys227, that form a disulfide bond. Larger recombinant constructs, however, contain up to three additional cysteines; one is in the N-terminal propeptide, and two reside in the C-terminal propeptide. The ΔspΔCVpAP*(His)6 construct contains three cysteines. It is highly likely, given the thermostability of VpAP, that the Cys223-Cys227 disulfide bond is very stable and that neither of these cysteines forms a disulfide bond with the third cysteine in any significant fraction of the sample. It is, however, possible that the cysteine that resides on the N-terminal propeptide engages in intermolecular disulfide bond formation. The presence of the N-terminal propeptide inhibits activity and so likely is adjacent to the active site in the folded polypeptide; given the extensive secondary structure and hydrophobic surface of VpAP, dimers thus formed would likely be of high stability and possibly partially resistant to SDS, leading to incomplete unfolding and the presence of multiple bands. In SDS-PAGE analysis of VpAP after Step A, pretreatment of samples by heating with β-mercaptoethanol and/or dithiothreitol was found to diminish the intensity of the “57 kDa” and “42 kDa” bands somewhat, but not completely. The thermostability of VpAP (see Step C) may be related to this phenomenon. An additional factor that may contribute to the anomalous behavior of unprocessed and partially processed VpAP on SDS-PAGE is the presence of regions of high hydrophobicity in both the mature VpAP sequence , and, particularly, in the N-terminal propeptide.
SDS-PAGE comparison of the material from Step A (Lane 2, Figure 3) with those from Steps B (Lane 3, Figure 3) and C (Lane 4, Figure 3) shows clear differences. Treatment with PK yielded a dominant species with an apparent Mr of 32 kDa, and some minor species of higher and lower apparent Mr. In contrast to results from Step A, N-terminal sequencing of the major band from Step B returned ASFVMP, indicating loss of the N-terminal propeptide ; the N-terminal region of mature natively-expressed VpAP begins MPPI, and is preceded in this recombinant protein by ASFV from the C-terminal sequence of the N-terminal propeptide. The specific activity of ≈ 110 Units/mg was consistent with removal of the N-terminal propeptide. More extensive incubation with PK (> 1 h) yielded a species with an apparent Mr of 28 kDa, with no change in catalytic activity (Figure 4). Two bands with apparent Mr of 28 and 32 kDa are routinely observed with preparations of natively-expressed VpAP, and mass spectrometry has indicated that the two bands correspond to species with almost indistinguishable Mrs of 32 ± 1 and 31 ± 1 kDa . Further processing of PK-treated VpAP by heat treatment (i.e. Step C) yielded a single species that was indistinguishable on SDS-PAGE (Lane 4, Figure 3) or by Co(II)-affinity resin binding behavior (Figure 5) from the major species observed prior to heat treatment; heat treatment served to denature fragments that had been cleaved upon PK treatment. In addition, following Step C, it was observed that the VpAP activity could be recovered from Co(II)-affinity resin with HEPES buffer, in the absence of imidazole, indicating that the C-terminal (His)6 region was no longer present following Steps B and C. The specific activity of 120 Units/mg for VpAP from Step C does not, in fact, indicate any enhancement of the activity per mole upon heat treatment, but reflects the removal of the cleaved N-terminal propeptide, about 60 % of which appeared to have denatured during PK-treatment and dialysis, and was removed by centrifugation, and the remainder of which was denatured by heat treatment.
The purpose of Step D was to separate any VpAP that had not had the C-terminal (His)6 removed during Step B and had yet remained undenatured during Steps B and C. SDS-PAGE analyses (Figure 5) indicated that no detectable amount of protein eluted from the Co(II)-affinity resin upon the application of imidazole.
10 L E. coli cultures, OD600 nm = 2.0 - 2.3, yielded 15 - 20 g wet weight of cells, which were discarded. From 1 to 2.5 mg of the D99-substituted variants of VpAP were purified from the supernatants. Steady state activities towards LPNA were significantly lower than for wild-type VpAP: the activities of VpAP:D99M, VpAP:D99H and VpAP:D99A were 0.030 ± 0.002, 0.004 ± 0.001, and 1.5 × 10-4 to 3.5 × 10-4 Units/mg, respectively. The purification of VpAP:D99M is summarized in Table 2. Apparent kinetic parameters of k′cat = 1.0 ± 0.1 min-1, and K′m = 31 ± 7 μM were estimated from kinetic analysis of VpAP:D99M; meaningful values for K′m could not be obtained for the other variants due to the very low activities.
Step A processing of supernatant containing ΔspΔCVpAP:D99X*(His)6, where X = A, H or M, proceeded as for wild-type VpAP, except that no activity towards LNPA was detectable at that stage. In addition, the apparent Mr of the bands that eluted from the Co(II)-affinity resin upon application of imidazole differed slightly from wild-type (SDS-PAGE of crude VpAP:D99A after Step A is shown as lane 4 of Figure 6 and crude VpAP:D99M is shown as Lane 1 of Figure 7; compare with wild-type VpAP, lane 5, Figure 2). The differences between the non-reducing SDS-PAGE profiles of wild type VpAP and of the VpAP:D99X variants may be related to the cysteine content, just as the cysteine content may be responsible for the presence of multiple bands, some with Mr higher than the molecular species present, in the first place. In wild-type VpAP, the Cys223-Cys227 disulfide bond is very stable, as attested to by the thermostability of VpAP. Asp99 is a key residue in a hydrogen bonding network that involves active site residues; indeed, it is of interest because of its possible structural and catalytic roles.
The Cys223-Cys227 disulfide bond is also a key structural feature of the active site. It is perhaps possible that some active site local structure is preserved, even in the presence of SDS, in wild-type VpAP during SDS-PAGE; it is noteworthy that mass spectrometry has verified that purified, mature VpAP does not run on SDS-PAGE corresponding to its correct Mr . The local structure may not be so preserved in VpAP:D99X variants, however; a hypersensitivity to PK and heat treatment are consistent with lower structural stability. In that case, at least six bands may be expected on SDS-PAGE, three dimeric species with intermolecular disulfide bonds between the N-terminal propeptide cysteine of one monomer and three monomeric species with a single disulfide; note that the suspected proximity of the N-terminal propeptide to the active site, suggested by the inhibitory effect on activity, would facilitate intramolecular disulfide bond formation between the N-terminal propeptide and either of Cys223 or Cys227 if the Cys223-Cys227 disulfide bond itself were not subject to additional stabilization. Further work that is outside the scope of the current study is clearly necessary to fully understand the behavior of VpAP polypeptide constructs on SDS-PAGE. However, there is available information on (i) the Cys content of mature and unprocessed VpAP constructs, (ii) the thermostability and stable disulfide bond in mature wild-type VpAP, (iii) the structural relationship between D99, the active site, and the disulfide bond, (iv) the much lower structural stability of D99 mutants and (v) the anomalous behavior on SDS-PAGE of purified, mature VpAP of mass-spectrometry-verified Mr. These data are consistent with, and suggest a mechanism for, multiple bands on SDS-PAGE for unprocessed VpAP constructs, a single band for purified VpAP, and a different and more complex band profile for VpAP:D99X variants unprocessed polypeptide constructs than for wild-type.
Upon treatment with PK (Step B), VpAP:D99A was rapidly degraded. After 10 min, bands at 28, 30 and 32 kDa were observable on SDS-PAGE, along with fragments of lower apparent Mr (lane 6, Figure 6). After 30 – 40 min, however, these bands had disappeared (lanes 8 and 9, Figure 6), and, after 50 min, no protein at all was observable on the gel (lane 10, Figure 6).
The lengthy methods of either SDS-PAGE or measurement of very low levels of activity are, obviously, impractical for real-time determination of when to terminate Step B, and this procedure could not be reliably or reproducibly employed to process ΔspΔCVpAP:D99X*(His)6. Instead, the alternative procedure of extended dialysis without PK, Step E, was employed. VpAP:D99M displayed the same SDS-PAGE profile as VpAP:D99A, following Step A (lane 1, Figure 7), and a similar sensitivity to PK treatment. However, upon extended dialysis at 4 °C against HEPES buffer, the bands in the range of apparent Mr = 30 − 40 kDa, and Mr ≈ 62 kDa, lost intensity and a band with apparent Mr ≈ 28 kDa emerged, corresponding to the 31 ± 1 kDa species characterized by mass spectrometry, until, after 13 d, only the band at Mr ≈ 28 kDa was observable (lane 2, Figure 7). The N-terminal amino acid sequence of this species was found to be TLASFVM, corresponding to the final six amino acids of the N-terminal propeptide, and the N-terminal methionine of mature VpAP; this confirmed that the N-terminus was autoprocessed during dialysis. Step D, a second application to Co(II)-affinity resin, was then carried out. SDS-PAGE analysis (Figure 8) clearly showed that VpAP:D99M isolated from Step E did not bind to the resin and that, therefore, the C-terminal region had also been removed during Step E. Circular dichroism spectropolarimetery of D99M yielded a spectrum indistinguishable from that of the “32 kDa” form of wild-type recombinant, fully active enzyme (Figure 9). If Mr for VpAP:D99M were really 28 kDa, this could only correspond to the loss of a large C-terminal α-helix, given the experimentally obtained N-terminal sequences, and the circular dichroism spectra would be expected to be significantly different.
Wild type VpAP protein readily formed hexagonal crystals (Figure 10) with a space group of P6122. The structure was determined to a resolution of 1.95 Å (1 Å = 0.1 nm; PDB code 3F4H) and the crystallographic data collection and refinement statistics are given in Table 3. The protein fold was essentially identical to those structures already published. For comparison, the present structure, 3F4H, and an earlier structure, 1LOK , were superimposable (LSQMAN) with a root mean square deviation of 0.120 Å in Cα position. The inter-Zn distances were 3.40 Å (3F4H) and 3.45 Å (1LOK). The crystal structure also provided information on the C-terminus of the fully processed protein. The residues comprising the sequence MGSATG were all clearly identifiable from the electron density and, as with natively expressed VpAP, only the short C-terminal sequence DTPTPGNQ could not be inferred from X-ray diffraction. In addition, the pre-N-terminal residues ASFV were not clear from the X-ray diffraction data. The possible range of Mr for the crystallized recombinant VpAP is, then, 31,388 to 32,640 D (not including Zn).
VpAP has been extensively studied, because of its high structural homology to many biomedically important metallohydrolases, and because of the actual and potential clinical utility of inhibitors of the enzyme [35, 36, 40]. In particular, the mechanism of action of the enzyme is seen as providing a prototypical template for the rational design of anti-cancer, anti-angiogenic and antibiotic chemotherapeutic agents. Despite much elegant work on the native wild-type enzyme, that has provided much insight into the catalytic mechanism, questions remain; even the roles of the distinct Zn(II) ions in the active site, once thought to be clear [60, 76, 97], are now somewhat controversial [39, 98]. The focus of recent structure-function research on VpAP, and on closely related enzymes, has been on site-directed variants [61-64, 98], as these may provide further insight into the structure-function relationships in VpAP.
A prerequisite to structure-function studies on site-directed variants of VpAP is a method by which these proteins can be expressed and purified. The challenges associated with heterologous expression and purification of VpAP are evident from earlier work, and have been particularly thoroughly addressed in a report by Bzymek, Holz and coworkers [71, 72]. Briefly, the problems include (i) the desirability of a mechanism by which VpAP is exported from the cell, to prevent inclusion body formation; (ii) the requirement for an N-terminal propeptide for correct folding; (iii) the inhibitory nature of the N-terminal propeptide, which makes identification and quantitation of unprocessed VpAP* by activity very difficult; (iv) anomalous behavior of VpAP* on SDS-PAGE gels; (v) the need for aggressive treatment with proteases and heat to process VpAP* to VpAP; (vi) multiple lengthy and inefficient chromatographic separations. Despite these challenges, Bzymek et al. were able to design a protocol for isolation of heterologously expressed VpAP that yielded wild-type and some site-directed variant forms of VpAP.
Despite the successes of the method of Bzymek, Holz and coworkers, two outstanding issues led us to revisit the expression and purification of VpAP from an E. coli expression system. The first was that we were unable to isolate some site-directed variants using this method. Where initial expression was low, the losses in terms of recoverable material during the traditional chromatographic steps were unacceptable. In addition, as our data clearly show, some variants were so sensitive to proteinase K treatment, and the procedure so difficult to control with variants exhibiting low specific activity, that incubation with proteinase K could not be reliably used to cleave the N-terminal propeptide. The second outstanding issue was that of final purity of the product. Many site-directed variants of interest in structure-function studies have very low activities compared to wild-type VpAP. Reported activities of e.g. 0.004 % of that of wild-type have been reported . Important mechanistic information is available from kinetic studies of variants with even very low activities; for instance, K′m informs on the ability of a kinetically compromised variant to form a Michealis complex with substrate. However, these studies rely on the level of contamination by other proteolytic enzymes being extremely low, much lower than can be determined by SDS-PAGE. While heat treatment may be expected to denature most proteins other than VpAP, there is no analytical way to determine that heat treatment does indeed destroy all contaminating proteases to a remaining level of 0.004 %. In addition, some VpAP site-directed variants are unstable toward heat treatment as well as toward proteinase K.
The expression of VpAP as a construct with a (His)6 adduct facilitated a very rapid and effective first purification step. Three bands were evident on SDS-PAGE after this first step; two of the bands were shown to be due to ΔspΔCVpAP*(His)6, indicating anomalous behavior on SDS. Given (i) the known propensity of both unprocessed and processed VpAP to behave anomalously on SDS-PAGE, (ii) the requirement of imidazole for elution of the unprocessed material from the Co(II)-affinity resin, (iii) a constant ratio of the intensities of the three bands among fractions from the Co(II)-affinity column, and (iv) that a single product resulted from processing of the unprocessed material, it is likely that all three bands were due to ΔspΔCVpAP*(His)6. Earlier attempts to employ a “ΔC” construct, i.e. one lacking 43 amino acids of the C-terminal propeptide (“pET27b(+)-43AAP”), resulted in a processed product with only 35 % of the activity of the native enzyme . In contrast, processing of ΔspΔCVpAP*(His)6, in which the entire C-terminal propeptide was deleted, provided VpAP with activity as high as any reported and with K′m in the range of typically reported values.
The use of a (His)6-containing construct allowed for rapid purification of VpAP but presented the concern that the (His)6 construct may not be removed during processing. As VpAP is a transition ion-dependent enzyme, structure-function studies addressing the active site would be compromised by binding of metal ions to the (His)6 sequence. In particular, Co(II) is commonly used as a spectroscopically active substitute for the native Zn(II) but (His)6 has a relatively high affinity for Co(II). An important final step, therefore, was the reapplication of fully processed VpAP to TALON Co(II)-affinity resin, and only fractions that did not bind, under the conditions that ΔspΔCVpAP*(His)6does bind, were retained. In practice, no SDS-PAGE-detectable material was actually observed in fractions collected upon subsequent application of imidazole, indicating highly efficient cleavage. The specific site of cleavage is unclear, but the crystal structure of wild-type VpAP indicated that it is somewhere in the short C-terminal sequence DTPTPGNQLE that immediately precedes (His)6. The ready crystallization, and sub-2 Å X-ray diffraction of the crystals, suggests a highly pure and homogeneous final product.
The use of the (His)6 construct was also applied to VpAP variants. One of the principal advantages was the rapid separation of fragile variants, such as the VpAP:D99X family, from other proteolytic enzymes. Studies with PK have shown these variants to be far less resistant than wild-type VpAP to proteolytic action, and the one-step Co(II)-affinity column isolation of ΔspΔCVpAP*:D99X(His)6 has replaced much lengthier and less effective hydrophobic interaction and anion exchange chromatographic steps that have been traditionally employed. Key to the use of ΔspΔCVpAP*:D99X(His)6, however, was the observation that solutions that had undergone exhaustive dialysis, followed by filtration, did not bind to Co(II)-affinity resin when reapplied. This observation suggested that ΔspΔCVpAP*:D99X(His)6 was capable of autoprocessing. Along with a role for a neutral protease expressed by V. proteolyticus in processing VpAP*, autoprocessing by VpAP has also long been suspected. However, as crude preparations have not been free of other proteolytic activity, it has been difficult to confirm such a role. In the case of the ΔspΔCVpAP*:D99X(His)6 variants, however, Co(II)-affinity resin chromatography provided a highly purified species, notwithstanding the anomalous behavior on SDS-PAGE. In addition to removal of the (His)6 adduct, exhaustive dialysis was shown by SDS-PAGE, and earlier mass spectrometry studies on wild-type VpAP, to have processed ΔspΔCVpAP*:D99X(His)6 to a single species with Mr = 31 kDa. This indicated removal of the N-terminal propeptide, and this was confirmed by N-terminal sequencing. It was confirmed by circular dichroism spectropolarimetry that extensive C-terminal processing did not occur, despite the apparent Mr of 28 kDa from SDS-PAGE. The 3 - 4 kDa C-terminal region that would have been lost from the 32 kDa protein to form a 28 kDa species would consist almost entirely of α-helix, and the circular dichroism spectrum would have differed significantly from that of wild-type in the 200 - 240 nm region.
In conclusion, we have developed protocols for purification of wild-type VpAP, and of protease-intolerant site-directed variants. The replacement of a C-terminal propeptide with a (His)6 adduct facilitates very rapid isolation of the expressed polypeptide from contaminating proteolytic activity, from the supernatant of cultures of E. coli. Post-translational processing of the purified polypeptide to the mature, functional enzyme can be effected by traditional exposure to heat or proteolytic activity or, for protease-intolerant variants, by an apparently autocatalytic mechanism. Isolation of a final product that is demonstrably devoid of the (His)6 construct is possible by a repeat application to Co(II)-affinity resin. This advance will facilitate the isolation and study of site-directed variants of VpAP with structure-function relationship interest.
The authors would like to thank Wei Yong and Prof. Jung-Ja Kim, Department of Biochemistry, Medical College of Wisconsin, for invaluable assistance with crystallography. This work was supported by the National Institutes of Health (NIAID R01 AI056231 to B.B.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.