- Email: [email protected]
Functional expression of bitistatin, a disintegrin with potential use in molecular imaging of thromboembolic disease
Protein Expression and Purification 39 (2005) 307–319 www.elsevier.com/locate/yprep
Functional expression of bitistatin, a disintegrin with potential use in molecular imaging of thromboembolic disease Linda C. Knight*, Jan E. Romano Nuclear Medicine Division, Radiology Department, Temple University School of Medicine, Philadelphia, PA 19140, USA Received 9 November 2004 Available online 8 December 2004
Abstract Bitistatin is a single-chain disintegrin which contains 83 amino acids and is internally crosslinked with seven disulfide bonds. This platelet aggregation inhibitor, which binds with high affinity to the aIIbb3 integrin, has potential use as the basis for a radiotracer to locate thrombi and emboli by scintigraphic imaging. A method amenable to large-scale, consistent production of bitistatin was sought. A synthetic gene coding for bitistatin was inserted into two different Escherichia coli expression vectors. One vector expressed recombinant bitistatin (rBitistatin) as a cleavable fusion protein and the other expressed rBitistatin as an isolated protein. In both cases, rBitistatin contained an additional amino acid (Gly) at the N-terminus compared with the native protein. The fusion protein was purified by affinity chromatography, then cleaved enzymatically to release rBitistatin, which was purified by reversedphase high performance liquid chromatography (HPLC) to a single active form. The rBitistatin produced as an isolated protein was purified from cell lysate by HPLC in a reduced form, then refolded, and purified again by HPLC. Yields of active rBitistatin averaged 12 mg/L for expression as an isolated protein, 10 times as high as when the fusion protein was employed. Structural assays confirmed the expected mass and sequence of the product. Functional assays (inhibition of platelet aggregation in vitro, equilibrium binding to platelets in vitro, and binding of labeled protein to experimental thrombi and emboli in vivo) confirmed that rBitistatin retained the functional characteristics of native bitistatin. 2004 Elsevier Inc. All rights reserved. Keywords: Bitistatin; Platelet aggregation inhibitor; Thromboembolic disease; Radionuclide imaging
Disintegrins from the venom of snakes are the most potent known antagonists of integrin function [1]. They were first described as inhibitors of platelet aggregation, by their binding to integrin aIIbb3 (the fibrinogen receptor, also known as glycoprotein IIb/IIIa) on the surface of platelets. All contain multiple cysteines at highly conserved positions, which result in a high degree of intramolecular crosslinking [1]. Disintegrins are also characterized by an Arg-Gly-Asp (RGD), Lys-Gly-Asp (KGD) or other receptor recognition sequence, presented at the apex of a flexible loop. Their affinity for integrin aIIbb3 is much higher than linear RGD con-
*
Corresponding author. Fax: +1 215 707 8110. E-mail address: [email protected] (L.C. Knight).
1046-5928/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.pep.2004.11.005
taining peptides or short cyclized peptides containing RGD. Bitistatin is a 83-amino acid monomeric disintegrin [1]. It is a potent inhibitor of platelet aggregation and has been shown to have beneficial effects in vivo such as inhibiting acute restenosis of experimental coronary artery interventions [2,3] and preventing platelet loss during extracorporeal circulation [4]. It is being investigated as the basis for a radiotracer imaging test for locating thrombi and emboli [5,6]. Bitistatin was originally isolated from the venom of the puff adder, Bitis arietans [2]. For eventual clinical use, however, isolation from snake venom was not considered sufficiently reliable or consistent, so we sought a recombinant product. Numerous reports indicate that snake venom disintegrins are synthesized in the venom gland as larger
308
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
proteins with multiple domains, and are cleaved in the gland to release the disintegrin domain [7,8]. For thrombus imaging, we wanted only the disintegrin domain, and aimed to express only the disintegrin domain to mimic the native bitistatin found in collected venom. One difficulty with expressing this molecule in Escherichia coli is that expressed protein in the cytoplasm of the cells is kept in a reduced form. Bitistatin contains 14 cysteines, which in the native molecule are fully crosslinked to form seven internal disulfide bridges. Correct folding of disintegrins is important for their biologic activity, to maintain the correct conformation of the RGD receptor-binding site. Several investigators have reported that reduction of native disintegrins abolishes their activity for inhibiting platelet aggregation [9– 11]. A challenge of this project was to refold and oxidize the expressed rBitistatin to form its disulfide bridges correctly. In this report, we have successfully developed a reproducible procedure for producing recombinant bitistatin from E. coli. The refolded product has biologic activity which is equivalent to native bitistatin. Note. The molecule discussed here is one of four isoforms of bitistatin previously reported. This isoform was previously reported as bitistatin 4 [1].
Methods General All chemicals were obtained from either Fisher Scientific (Pittsburgh, PA) or Sigma Chemical (St. Louis, MO) unless otherwise specified. Solvents and chemicals were the highest purity available and were used as received without further purification. HPLC separations Reversed-phase HPLC was used for purification and analysis. In all cases, a widepore C18 column packing ˚ pores, 10 lm particle size, Varian, Walnut Creek, (300 A CA) was used. The elutions employed 0.1% TFA (aq) as mobile phase A and 0.1% TFA in acetonitrile as mobile phase B. Semipreparative HPLC method 1 used a 21.4 · 250 mm column connected to a gradient HPLC system (two Rainin HPXL pumps with 50 mL heads and Knauer UV/Vis detector; Varian, Walnut Creek, CA). The column was eluted with a linear gradient of 0–60% mobile phase B over 30 min. The flowrate was 16 mL/min and elution was monitored by UV absorbance at 280 nm. Semipreparative HPLC method 2 used a 10 · 250 mm column with a gradient of 0–40% mobile phase B over 27.1 min, using the pumps and detector described above. The flowrate was 3.5 mL/min and elution was monitored by UV absorbance at 280 nm. Analytical HPLC used a 4.6 · 250 mm column, connected to an
analytical system (two model 510 pumps, gradient controller, model 486 UV detector; Waters, Milford, MA). The detector data were collected and analyzed with an AllChrom data analyzer (Alltech, Deerfield, IL) running in a personal computer. The column was eluted with a linear gradient of 0–60% mobile phase B over 30 min. The flowrate was 1.0 mL/min and elution was monitored by UV absorbance at 280 nm. Purification of native bitistatin Bitistatin was purified from freeze-dried B. arietans venom (Miami Serpentarium Laboratories, Punta Gorda, FL) using ion-exchange chromatography followed by reversed-phase HPLC as previously described [6]. The product was purified to a single peak and evaluated by SDS–PAGE, by MALDI mass spectrometry and by peptide mapping. Expression of bitistatin as a fusion protein The oligonucleotide sequence needed to code for bitistatin was deduced from the amino acid sequence of bitistatin (Fig. 1). The full-length double-stranded DNA was constructed using splicing by overlap extension [12] from four synthetic oligonucleotides of 85–88 nucleotides each (Paragon Biotech, Baltimore, MD). The final construct included a BamHI site at the 5 0 end, and the 3 0 end included a stop codon followed by an EcoRI site. The purified double-stranded synthetic DNA was ligated into the BamHI/EcoRI sites of pGEX-KT expression vector [13]. The resulting vector, referred to as pGEX-GSTrBit, predicts a protein containing glutathione S-transferase (GST) followed by a polyglycine spacer (‘‘kinker’’) region, a thrombin-cleavable sequence (LVPRR), a glycine, and the bitistatin sequence (Fig. 2). Thrombin cleavage of the fusion protein would result in recombinant bitistatin with an additional Gly at its N-terminus. Recombinant pGEX-GSTrBit vectors were introduced by transformation into E. coli BL21(DE3) (Novagen, Madison, WI), and bacteria were grown at 37 C in Luria–Bertani medium containing ampicillin (100 lg/mL) until the culture reached OD595 = 0.6. Expression of the fusion protein was induced with 1 mM isopropyl-1-thio-galactopyranoside (IPTG) for 3 h. The cells were harvested by centrifugation (1800g
Fig. 1. Amino acid sequence of native bitistatin (isoform bitistatin 4 after Gould et al. [1]).
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
309
Fig. 2. Expression construct in pGEX-GSTrBit vector. The expression product consists of glutathione-S-transferase (GST), followed by a polyglycine kinker (a spacer to enable thrombin access), a thrombin recognition/cleavage site, and bitistatin. The sequence of bitistatin is shown in bold. Cleavage by thrombin results in release of bitistatin which differs from native bitistatin in that it has an additional Gly at its amino terminus. Only one strand is shown, and the gene for GST is not shown in its entirety. The location of the BamHI restriction site is underlined.
for 25 min at 4 C), resuspended in 1/10 original volume of PBS, and spun down again. Cells were lysed by resuspension in 1/10 original volume of B-PER cell lysis reagent (Pierce, Rockford, IL) in the presence of 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 25 lg/mL aprotinin, and 25 lg/mL leupeptin. After 10 min of lysis, cellular debris was removed by centrifugation (27,000g for 15 min at 4 C). The lysate was incubated for 1 h at 4 C with glutathione–agarose beads (Pierce, Rockford, IL) to selectively bind the expressed fusion protein. After six washes to remove unbound impurities, the beads were suspended in TBS (50 mM Tris and 150 mM NaCl), pH 8.0, and placed on a tube rocker at 4 C to permit bitistatin to refold and oxidize to form disulfide crosslinks. Thrombin cleavage of the fusion protein was performed while the GST was still attached to glutathione–agarose [14]. Bovine thrombin (Amersham Biosciences, Piscataway, NJ; 2 U/mg fusion protein) and CaCl2 (final concentration 2.5 mM) were added to cleave rBitistatin from the fusion protein. Rocking of the tube containing the beads was continued at 4 C
for 24–48 h. Refolding of cleaved protein was monitored by sampling the bead supernatant and analyzing it on analytical reversed-phase HPLC. When there was no further change in the HPLC profile, the supernatant was collected from the beads and fractionated by semipreparative reversed-phase HPLC, method 2. Peaks were collected manually as they eluted. The collected fractions were tested for inhibition of ADP-induced human platelet aggregation. The peak with the highest potency in this assay was characterized further. Expression of bitistatin without a fusion partner Because of concerns about introducing viral contaminants from the use of human or bovine thrombin for cleavage, we wanted to express bitistatin without a fusion partner. The gene for bitistatin was cleaved from the BamHI/EcoRI sites of the pGEX-GSTrBit vector and separated on an agarose gel. The band containing the bitistatin gene was excised from the gel and purified with a QIAquick gel extraction kit (Qiagen, Valencia, CA) (Fig. 3A).
310
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 3. Construction of plasmid for expression of bitistatin without a fusion partner. (A) Cleavage of bitistatin gene from pGEX-GSTrBit plasmid. (B) Modification of bitistatin gene with synthetic duplex extension. (C) Ligation of extended bitistatin gene into pET-5a plasmid. (D) Removal of unwanted bases to form final expression vector, pET-rBit.
To adapt the gene to a new expression vector, a doublestranded extension was prepared. Two oligonucleotides were synthesized: 5 0 TACATATGG and 5 0 GATCCCAT
ATG. Both oligonucleotides were phosphorylated on the 5 0 end. The synthetic oligos (dissolved in water) were mixed together in equimolar quantities and annealed by
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
311
Fig 3. (continued)
boiling 20 min followed by cooling at room temperature. The resulting double-stranded DNA segment (Fig. 3B) has a sticky end for ligation to the BamHI site on the rBitistatin gene cleaved from pGEX-GSTrBitis, and a sticky end for ligation into an NdeI site on a new vector. The yield of duplex was assumed to be 50%, and the mixture was carried on to the next step without further purification. The double-stranded extension was combined in 100-fold molar excess with the rBitistatin gene cleaved from pGEX-GSTrBitis. Ligation was carried out for 20 h at 4 C in the presence of T4 DNA ligase. The reaction was purified using QIAquick (Qiagen). This purified product was ligated into the NdeI/ EcoRI sites of pET-5a vector (Promega, Madison, WI). This construct, referred to as pET5-E3, should express bitistatin with a single Gly added at the N-terminus (Fig. 3C), and should theoretically provide exactly the same protein as the thrombin cleavage product of the fusion protein. This assumption is based on the removal of the N-terminal methionine in E. coli [15]. The vector was transformed into BL21trxB (DE3)pLysS cells (Novagen, Madison, WI). DNA sequencing was performed to confirm the correct construct. Because plasmid pET5-E3 contained an additional 4 bp which were not needed but which resulted in unwanted separation of the initiation codon from the Shine-Dalgarno sequence, plasmid pET5-E3 (purified by Wizard Plus SV Miniprep, Promega) was digested with NdeI and EcoRI. The 268 bp fragment containing the rBitis gene was purified on an agarose gel, purified
by QIAquick gel extraction, and ligated into the NdeI and EcoRI sites of freshly digested pET-5a vector (Fig. 3D). This construct, referred to as pET-rBit, was used to transform BL21trxB (DE3)pLysS cells (Novagen, Madison, WI). DNA sequencing was performed to confirm the correct construct. Bacteria harboring the pET-rBit plasmid were grown at 37 C in Luria–Bertani medium containing ampicillin (100 lg/mL), chloramphenicol (34 lg/mL), and kanamycin (15 lg/mL) until the culture reached OD595 = 0.6. Expression of rBitistatin was induced with 1 mM IPTG for 3 h. The cells were harvested by centrifugation (3000g for 10 min at 4 C), resuspended in 1/10 volume of PBS, and spun down again. Cells were lysed by three cycles of freeze–thawing in TBS, pH 7.5, containing 25 mg/ml aprotinin, 25 mg/ml leupeptin, 1 mM PMSF, and 0.1% Triton X-100. After adding MgCl2 to 10 mM and DNAse I to 10 lg/mL, cell debris was removed by centrifugation. Dithiothreitol (DTT) was added to a final concentration of 20 mM to reduce the expression products and prevent formation of disulfide crosslinks during purification. The lysate was diluted with 0.1% trifluoroacetic acid (TFA). After centrifuging to remove any precipitate, the supernatant was filtered through a 0.22 lm syringe filter (Corning Life Sciences, Acton, MA). A portion of the filtrate was loaded onto a semipreparative reversed-phase HPLC system and processed using method 1. Major peaks were collected manually. The peak eluting from 19 to 21 min was collected and freeze-dried (Centrivap, LabConCo, Kansas City,
312
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
MO), then stored at 70 C until refolding could be carried out. Refolding to allow reduced bitistatin to oxidize and form disulfide crosslinks was accomplished by dissolving the reduced bitistatin to a final concentration of 0.1 mg/mL in TBS, pH 7.5, containing 2.5 mM CaCl2. Glutathione and oxidized glutathione (USB, Cleveland, OH) were then added to final concentrations of 5.0 and 2.5 mM, respectively, to facilitate disulfide exchange [16]. The solution (100–150 mL) was stirred in a sterile 250 mL capacity polycarbonate conical vent-cap flask (Corning) at 4 C for 3 days. It was then placed in a 20 C freezer for 3–4 weeks. The product was then stored at 70 C until final purification. Final purification of refolded rBitistatin was accomplished using semipreparative HPLC method 2. The refolded rBitistatin solution was filtered (0.22 lm) before loading it into the system. Elution was monitored by UV absorbance at 280 nm. The first peak to elute (at 20–21 min) was manually collected and freeze-dried. Structural evaluations A sample of the final product was injected into the analytical HPLC system to determine its profile (retention time and peak shape). Elution was monitored by UV absorbance at 280 nm. Mass spectrometry (MALDI-MS) and amino acid analysis were performed on the final product. Mass mapping was performed on a reduced and alkylated product. Briefly, rBitistatin (1 mg/mL in 0.2 M Tris, pH 8.3, containing 2 mM EDTA) was reduced with dithiothreitol (DTT; 180:1 molar ratio DTT:rBitistatin) for 2 h at 37 C under an argon atmosphere. This was followed by addition of iodoacetamide (2900:1 molar ratio iodoacetamide:rBitistatin) and incubation at 37 C was continued for a further 30 min. The product was purified by reversed-phase HPLC and freeze-dried. The reduced, alkylated material was subjected to trypsin digestion for various times from 5 min to 8 h, and the fragments were analyzed by MALDI-MS. The fragments found were compared to the theoretical tryptic fragmentation pattern of the amino acid sequence. The expression products were analyzed by SDS–polyacrylamide gel electrophoresis (SDS–PAGE). For monitoring of expression and initial purification, precast 10–20% gels (Owl Separations, Portsmouth, NH) were run with Tris–Tricine SDS buffer, pH 8.5, and stained with Coomassie blue [17]. For testing the final product, precast 4–20% gradient gels, Tris–Hepes–SDS, pH 7.0 (Precise Protein Gels, Pierce, Rockford, IL), were run with 100 mM Tris, 100 mM Hepes, and 3 mM SDS, pH 8.0. The gels were stained with GelCode blue (Pierce, Rockford, IL). Molecular weight standards were run in separate lanes.
Functional assays Inhibition of platelet aggregation in vitro Under a protocol approved by the Institutional Review Board, blood was obtained from healthy human donors who had not taken aspirin or related medications for two weeks. The blood was anticoagulated with 1/10 volume of 3.8% sodium citrate and was centrifuged at 160g for 12 min to obtain platelet-rich plasma (PRP). The concentration of platelets in human PRP was determined using a hemacytometer, then diluted with the donorÕs own plasma to obtain a platelet concentration of 300,000/lL. The study was performed using an aggregometer (Payton Scientific, Buffalo NY). For each measurement, 400 lL of adjusted PRP was placed in a cuvette and stirred at 37 C. Vehicle (saline) or various concentrations of peptide were added in a volume of 10 lL and the mixture was stirred at 37 C for 1 min while recording light transmission. ADP was then added (final concentration 10 lM) to induce platelet aggregation. Platelet aggregation was determined by change in light transmission through the PRP, which was recorded for 5 min after ADP addition. The IC50 was determined (by least-squares fitting) as the concentration of peptide required to produce 50% inhibition of the response to ADP in the presence of the vehicle. Results were normalized to the results for the tetrapeptide Arg-GlyAsp-Ser (RGDS). Test of binding of radiolabeled rbitistatin to platelets Bitistatin was radiolabeled with 125I (Perkin–Elmer Life Sciences, N Billerica, MA) using IodoGen (Pierce, Rockville, IL) as previously described [6], and purified by gel chromatography. Citrated human blood and platelet-rich plasma (PRP) were obtained as described above. Gel-filtered platelets (GFP; plasma-free) were prepared by filtering 4 mL of citrated platelet-rich plasma through Sepharose 2B (2.5 · 8 cm column, equilibrated with an elution buffer containing 0.4 mM NaH2PO4, 12 mM NaHCO3, and 10 mM Hepes, pH 7.45, with 137 mM NaCl, 2.7 mM KCl, 0.1% glucose, and 0.2% bovine serum albumin). Fractions containing the highest platelet concentrations were pooled, adjusted to a final platelet concentration of 108 platelets/mL, and used within 1 h of preparation. Equilibrium binding assays were done as previously described [18–20]. Aliquots of the platelets were stimulated by addition of 10 lM ADP, combined with 125I-labeled rBitistatin or native bitistatin over a range of concentrations, and incubated for 1 h at 37 C. Platelets and bound radiotracer were separated from free radiotracer by centrifugation of triplicate samples through 0.1 ml of 30% (W/V) sucrose at 13,000g for 3 min. Nonspecific binding was determined in the presence of 9 mM EDTA. The Kd and Bmax were determined by Scatchard analysis [21].
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Test of binding to thrombotic lesions in vivo To confirm that recombinant bitistatin is capable of binding to thrombi and emboli in vivo, rBitistatin was radiolabeled with 123I (MDS Nordion, Kanata, Ontario) and tested in a standardized canine model of experimental thrombi and emboli [5]. The method of labeling rBitistatin with 123I was as previously described [6]. Thrombi (in a femoral vein) and pulmonary emboli were induced 24 h before radiotracer administration by placement of embolization coils (Cook, Bloomington, IN) in three animals. Approximately 10 mCi of [123I]rBitistatin and 50 lCi of 125I-native bitistatin were injected into a peripheral vein of each anesthetized animal. Scintigraphic lateral images of the chest and anterior views of the hind legs were acquired during the next 4 h, using a large field of view gamma camera (MaxiCamera, General Electric Medical Systems, Milwaukee, WI). The camera was interfaced to an image acquisition system (NucLear Mac, Scientific Imaging, Denver, CO) running in a Macintosh computer (Apple, Cupertino, CA) for digitization of the image data. At 4 h after radiotracer administration, thrombi and emboli were recovered for weighing and counting for radioactive content (Wizard 1480, Perkin–Elmer, Downers Grove, IL) for comparison with blood, muscle, and lung. A standard of the administered dose was also counted, to convert the counts/min/g of each tissue to percent of injected dose per gram of tissue. Statistical evaluation methods Comparison between two groups was performed using a two-tailed t test. Comparison among three groups was performed using ANOVA. A p value <0.05 was used to reject the null hypothesis. The statistical calculations were performed using Microsoft Excel (Microsoft, Redmond, WA).
Results Expression of rbitistatin as a fusion protein GST::rBitistatin fusion protein was purified from cell lysate by binding to glutathione–agarose. Fig. 4 shows SDS–PAGE of cell growth and affinity purification of the fusion protein, resulting in a band with an apparent molecular weight of 37 kDa, in agreement with the theoretical sum of GST (28 kDa), and rBitistatin (9 kDa). The amount of GST::rBitistatin fusion protein purified from cell lysate averaged 26.3 mg/L of culture. Thrombin cleavage of GST::rBitistatin fusion could theoretically be done either in solution (after release of fusion protein from the agarose support) or while the GST::rBitistatin is still bound to the support. The first method was attempted, but resulted in considerable precipitate formation which lowered yields of the rBitistatin
313
Fig. 4. SDS–PAGE of pGEX-GSTrBit expression of GST::rBitistatin fusion protein and thrombin cleavage. The 10–20% gel was run with Tris–Tricine buffer, pH 8.5, under reducing conditions. Lane 1, molecular weight standards; lane 2, cell culture before IPTG; lane 3, cell culture 3 h after IPTG; lane 4, cell lysate; lane 5, supernatant after incubating cell lysate with glutathione–agarose; lane 6, immobilized GST::rBitistatin; lane 7, thrombin digest of GST::rBitistatin (unfractionated); lane 8, thrombin digest HPLC fraction 1; lane 9, thrombin digest HPLC fraction 2; lane 10, thrombin digest HPLC fraction 3; lane 11, final product rBitistatin; and lane 12, native bitistatin.
(data not shown). The second method, suggested by Guan and Dixon [14], was more successful in our hands. As seen on SDS–PAGE, thrombin treatment cleaved the fusion protein, releasing material which migrated as a single band and had approximately the same mobility as native bitistatin under reducing conditions (Fig. 4, lanes 7 and 12). Analysis of the product of thrombin digestion on reversed-phase HPLC revealed three prominent peaks followed by a tail (Fig. 5). Separation of the different forms and analysis of their bioactivity (inhibition of platelet aggregation) revealed that peak 2 had platelet activity comparable to native bitistatin, whereas other peaks were less active. The separated peaks had approximately the same mobilities on SDS–PAGE (Fig. 4). Peak 1, however, stained poorly and migrated slightly farther into the gel, suggesting that it may have been degraded. Peak 2 was purified, but because it was difficult to resolve from neighboring peaks, yields were low. The amount of thrombin-cleaved, purified refolded rBitistatin averaged 1–2 mg per original liter of culture. Expression of rbitistatin without a fusion partner In the direct production method using the pET-rBit vector, separation of reduced cell lysate on reversed-phase HPLC resulted in the elution profile shown in Fig. 6. The peak which eluted from 19 to 21 min had the same mobility as native bitistatin on SDS–PAGE run under reducing
314
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 5. Reversed-phase HPLC profile of supernatant from glutathione–agarose beads treated with thrombin for 24 h. Peaks 1, 2, and 3 were collected separately. Peak 2 contained functionally active rBitistatin.
Fig. 6. Semipreparative reversed-phase HPLC profile of DTT-reduced lysate from bacteria containing pET-rBit. The clarified lysate was loaded onto a C18 widepore column, and eluted by HPLC method 1 (see text). The elution was monitored by UV at 280 nm. The indicated peak (*) contained reduced rBitistatin. The black bar indicates the eluate collected for refolding.
conditions. Fortunately, this peak was well separated from other cell components and was therefore easy to purify by collecting the entire peak. In the direct production method, an average of 15.6 ± 3.1 mg of purified, reduced rBitis was obtained per liter of culture.
Fig. 7. Analytical reversed-phase HPLC of rBitistatin from pET-rBit plasmid. (A) Reduced rBitistatin prior to refolding. (B) Refolded mixture prior to purification. The large peak at 9 min is glutathione. (C) Purified final product.
Reduced rBitistatin exhibited a retention time of about 21 min on analytical HPLC (Fig. 7A). As refolding was carried out, the peaks observed on reversedphase HPLC shifted gradually to earlier retention times. After initial aeration by stirring at 4 C, the period of storage at 20 C was found to be important for achieving the highest yields of correctly folded rBitistatin. Fig. 7B shows an elution profile of refolded rBitistatin after it had nearly completed refolding. The profile consists of an initial spike followed by a broad secondary peak. In some cases, the secondary peak was merely a tail following the spike. The initial spike was found to contain the material which had the best bioactivity for inhibiting platelet aggregation (see below). The final product after purification eluted as a single sharp peak on reversedphase HPLC (Fig. 7C). The average yield was 12 ± 3 mg of purified, refolded rBitis per liter of starting culture. To determine which part of the refolded rBitistatin had the highest functional activity, a batch of (unpurified) refolded material was separated on HPLC into narrow fractions. The fractions were freeze-dried and individually analyzed on analytical HPLC to confirm that each had a discrete retention time. Fraction A was a pure peak corresponding to the initial spike (retention time 19.2 min on analytical reversed-phase HPLC). Fractions C, D, and E were pure peaks eluting
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
315
at 19.85, 20.05, and 20.41 min, respectively. It was not possible to obtain a pure fraction between A and C, so fraction B contained two overlapping peaks of retention times (areas) 19.2 min (44%) and 19.6 min (56%). Samples of each fraction were tested for their potency in inhibiting platelet aggregation. Only fraction A had potency similar to that of native bitistatin; all other fractions were significantly less potent (Table 2). Additional samples of each fraction were treated with DTT to fully reduce disulfides, and then analyzed on HPLC. The reduction caused all peaks to shift to the same retention time, 20.7 min (not shown). Structural analysis On analytical HPLC, there was no difference in retention time between rBitistatin produced directly in the pET-rBit vector and rBitistatin cleaved from the fusion protein (Fig. 8). The retention time of rBitistatin was similar to that of native bitistatin (not shown). SDS–PAGE analysis of rBitistatin is shown in Fig. 9. Under reducing conditions, both rBitistatins had the same mobility as native bitistatin. This 9 kDa protein typically migrates on SDS–PAGE at a higher apparent molecular weight of about 13 kDa. On SDS–PAGE under nonreducing conditions, the bitistatin and rBitistatin bands are more diffuse and tend to elute from the gel during staining; however, they migrate with approximately
Fig. 9. SDS–PAGE of native and recombinant bitistatins. Gel (4–20% gradient) run under reducing conditions. Lane 1, molecular weight standards; lane 2, native bitistatin; lane 3, rBitis from pGEX-GSTrBit; and lane 4, rBitis from pET-rBit.
the same mobility as under the reducing conditions (not shown). rBitistatin had an observed mass of 9076.8 on MALDI-MS. This is consistent with the expected value of 9079, within the error of the instrument. This is also in agreement with the expected removal of the N-terminal methionine in E. coli [15], leaving a Gly at the N-terminus. Native bitistatin (without the N-terminal Gly) had an observed mass of 9023.8 (theoretical 9022). The MALDI-MS masses did not change after treatment with vinylpyridine under denaturing (6 M guanidinium hydrochloride, 50 mM Hepes, pH 9.0) but nonreducing conditions, indicating that neither native Bitistatin nor rBitistatins contained free sulfhydryl groups. The results of amino acid analysis (Table 1) agreed well with the expected result. Peptide mapping was done by tryptic digestion of reduced and alkylated rBitistatin. Expected fragments were found, covering 95% of the sequence of bitistatin. The section of the sequence which was not confirmed was between Thr59 and Arg61, in a section of the protein where trypsin digestion results in very small fragments. Functional analysis
Fig. 8. Reversed-phase HPLC of purified rBitistatin: (A) from pGEXGSTrBit and (B) from pET-rBit.
Ability to inhibit platelet aggregation Fig. 10 shows examples of inhibition curves for rBitistatin, native bitistatin, and RGDS for single determinations. There was no significant difference between native bitistatin (IC50 125 ± 21 nmol/L, mean ± SD) (n = 5) and rBitistatin produced by cleavage of fusion protein (124 ± 43 nmol/L, p = 0.97) (n = 6) or rBitistatin produced by direct expression (134 ± 24 nmol/L, p = 0.50) (n = 9).
316
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Table 1 Amino acid analysis of rBitistatin AA
Calc
Found
AA
Calc
Found
AA
Calc
Found
Asx Ser Glx Gly His Thr Arg
14 5 9 8 2 4 3
13.46 4.49* 8.15 7.81 2.15 4.02 3.01
Ala Cys Pro Tyr Val Met Lys
5 14 5 1 2
5.16 8.50* 4.99 0.67 1.94
Ile Leu Phe Trp NorLeu AHX
2 2 1 2
2.05 2.21 1.14
5
4.94
Notes. *Cys and *Ser partially and
**
**
Trp completely destroyed during hydrolysis.
Table 2 Test of fractions isolated from refolded recombinant bitistatin for their ability to inhibit platelet aggregation Fraction
Retention time
IC50 (nmol/L)b
Aa B C D E
19.2 min 19.2 (44%) + 19.6 (56%) 19.85 min 20.05 min 20.41 min
145 266 957 1178 709
a
Corresponds to final purified rBitistatin. Concentration needed to inhibit ADP-induced human platelet aggregation by 50%. b
Fig. 10. Inhibition of human platelet aggregation by native bitistatin (solid diamonds), recombinant bitistatin (open triangles), and the tetrapeptide RGDS (solid squares).
Binding to ADP-stimulated platelets Both 125I-labeled native and recombinant bitistatin bound to a single class of receptors on ADP-stimulated platelets in a saturable manner (Fig. 11). The equilibrium binding constants were similar: 8.1 nmol/L for native bitistatin and 7.7 nmol/L for recombinant bitistatin. Binding to thrombotic lesions in vivo The ability of rBitistatin to bind to thrombi and emboli in vivo was tested in a standardized canine model. Fig. 12A shows the images obtained with 123I-labeled rBitistatin clearly delineating an experimental pulmonary embolus which was only 5 mm in size. Fig. 12B shows an image of [123I]rBitistatin accumulating in a deep vein thrombus in the leg of the same animal. The target-to-background contrast obtained with [123I]rBitistatin compares favorably with the image contrast
Fig. 11. Equilibrium binding of 125I-labeled native and recombinant bitistatin to ADP-stimulated platelets. (A) 125I-native bitistatin. Kd = 8.1 nmol/L, Bmax = 55,246 sites/platelet. (B) 125I-recombinant bitistatin. Kd = 7.7 nmol/L, Bmax = 79,587 sites/platelet. Each data point represents the average of three determinations.
obtained previously with 123I-native bitistatin [6]. In this study, the percentage of injected dose which accumulated in each tissue (thrombus, embolus, blood, muscle, and lung) was not significantly different between the two radiotracers ([123I]rBitistatin and 125I-native bitistatin) (p > 0.05).
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 12. Planar scintigraphic images of experimental thromboembolic disease with [123I]rBitistatin. (A) Left lateral view of chest of dog showing uptake in pulmonary embolus (arrow) at 1 h 50 min after radiotracer administration. Radiotracer accumulation in the embolus is more intense than in surrounding background tissues. Radiotracer is also seen in the blood pool in the chambers of the heart, the major blood vessels, and in the spleen (only the top of the spleen is in the field of view). (B) Anterior view of legs showing uptake in femoral vein thrombus (arrow) at 1 h 40 min after radiotracer administration. Radiotracer accumulation in the thrombus is more intense than in surrounding blood in the major vessels or surrounding muscle tissue.
Discussion Many groups have had an interest in recombinant expression of disintegrins, because this method enables the production of larger quantities of reproducible material than are available from natural sources. In addition, it provides the framework for site-directed mutagenesis to study the function and specificity of disintegrins. Because of their integrin-inhibitory activity, disintegrins
317
are being explored as the basis for design of drugs to inhibit cell adhesion and associated processes in diseases such as thrombosis, atherosclerosis, and cancer. Bitistatin, when labeled with an imaging radionuclide, has been shown to have potential as a diagnostic imaging agent for locating thrombi and emboli [5,6], and also has promise for imaging tumor angiogenesis [22]. Recombinant bitistatin was initially expressed as a fusion protein to simplify the purification process from crude cell lysate. However, refolding of rBitistatin into a correct orientation with proper disulfide pairing did not occur efficiently in the fusion with GST. Several distinct forms were found on reversed-phase HPLC. Only one peak had activity for inhibiting platelet aggregation but the three main peaks had similar mobility on reduced SDS–PAGE. This suggests that rBitistatin assumed several different folding configurations in the fusion protein, or that disulfide crosslinking was incomplete in some. Peak 1 may represent partially degraded rBitistatin. This might be a consequence of adding thrombin protease (which might contain other proteases) or the glutathione–agarose support (which might retain some proteases from the cell lysate). The different forms of rBitistatin eluted close together on HPLC, resulting in low yields of the desired component. Expression of rBitistatin as an isolated protein was achieved by transferring the gene into another expression vector. Purification of expressed rBitistatin in a reduced form was simplified by prior knowledge of its retention time on reversed-phase HPLC. Fortunately, reduced rBitistatin elutes well apart from other eluting cellular proteins in this system. The refolding and disulfide-pairing process in this case could be carried out with highly purified reduced rBitistatin in the absence of contaminating proteins and resulted in high yields of the correctly folded form, with no apparent degradation. It was found that addition of glutathione and oxidized glutathione was helpful in optimizing the folding process. In addition, refolding to the desired form appeared to accelerate when the 4 C solution was chilled further to 20 C. Other snake venom disintegrins have been produced by recombinant expression in E. coli. Various approaches have been used to achieve correct refolding and disulfide bond formation. Echistatin, which has 49 amino acids and four disulfide crosslinks, was produced from a synthetic gene as a fusion protein [23]. The expressed protein was found in inclusion bodies. After cleavage from its fusion protein, echistatin was completely reduced and refolded, enabling more than 30% of the reduced protein to be recovered as correctly folded echistatin. Kistrin, which contains 68 amino acids and six disulfide crosslinks, was produced as a fusion protein in an expression vector which directed the product to the periplasm [24]. It was cleaved from its fusion partner and purified by reversed-phase HPLC. Correctly folded recombinant kistrin was found, but the yields
318
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
were only 0.15–0.30 mg of correctly folded kistrin per liter of culture. The disintegrin domain of jararhagin was cloned as a fusion with thioredoxin, and expressed in a vector which should allow disulfide bond formation in the cytoplasm in E. coli [25]. Upon purification, however, it was found that only a fraction of the total disintegrin was correctly folded and crosslinked. These studies suggest that it is unlikely that the isolated disintegrin domain, coupled to a foreign protein, will refold efficiently in the same configuration as it does in the venom gland when coupled to its natural neighboring domains. Our experience with rBitistatin as a fusion protein agrees with the kistrin and jararhagin disintegrin reports, in that we found a mixture of refolded forms and obtained relatively low yields of the correctly folded form. In this study, when we purified the rBitistatin in a reduced form and refolded it in isolation, we were able to recover an average of 75% of reduced rBitistatin as correctly folded protein, for an overall yield of 12 mg/L of original culture. During the refolding of rBitistatin which had been expressed as an isolated protein, analysis was done to see which fractions had bioactivity and whether they represented differently folded forms of rBitistatin. The finding that all peaks shifted to the same retention time upon DTT reduction suggests that the fractions all contained the same peptide with different or incomplete disulfide bridging patterns. The fraction with the earliest retention time on HPLC had the lowest IC50 for inhibition of human platelet aggregation. The fully reduced form had the latest retention time. This is in agreement with the observations of others who tracked the refolding of recombinant proteins with reversed-phase HPLC [26–28]. Reversed-phase HPLC separates molecules in part by hydrophilicity, with the most hydrophilic forms eluting first. It may be that during optimal folding of disintegrins, hydrophobic residues fold toward the interior, and hydrophilic residues remain on the outside of the molecule. Bitistatin binds to the aIIbb3 integrin on platelets, which is also known as the glycoprotein IIb/IIIa receptor. Binding to this receptor blocks fibrinogen binding and, in adequate concentrations, inhibits aggregation of platelets in response to an agonist such as adenosine diphosphate (ADP). Recombinant bitistatin retains the ability to inhibit platelet aggregation in platelet-rich plasma prepared from the citrated blood of normal human donors. Radiolabeled rBitistatin binds to platelets with an affinity which is no different from native bitistatin, and has a similar number of apparent binding sites. The number of binding sites found in this study (55,000– 80,000) agrees with the reported number of aIIbb3 sites (40,000–80,000 sites per platelet) [29]. Most importantly, radiolabeled rBitistatin is equivalent to native bitistatin in its ability to bind to platelet deposits (thrombi and emboli) in vivo and to produce high-contrast images
of the thrombus over background within a short time after radiotracer administration. In this project, rBitistatin was produced by expression in E. coli as a fusion protein and also as an isolated protein. The final products of these two methods appear to be the same, however the yield is much higher for production as an isolated protein. The findings of this study suggest that recombinant bitistatin is suitable for further development as a radiopharmaceutical for imaging thromboembolic disease in vivo. Acknowledgments The authors are grateful to Dr. Alan Maurer for assistance with the animal model, to Drs. Kwamena Baidoo, Virginia Heatwole, and Bill Wong for assistance with the initial expression vector, to Drs. William Moore, Tapan Ganguly, Juan Calvete, and Kaye Speicher, and to Tom Beer and Nicole DiFlorio, for help with molecular analyses. DNA synthesis and sequencing was performed by Paragon Bioscience and by the University of Pennsylvania DNA Sequencing Facility (funded by the National Cancer Institute). Protein analyses were performed by the University of Pennsylvania Protein Chemistry Laboratory and the Wistar Institute Proteomics Facility. This work was supported by NIH R01 HL 54578.
References [1] R.J. Gould, M.A. Polokoff, P.A. Friedman, T.-F. Huang, J.C. Holt, J.J. Cook, S. Niewiarowski, Disintegrins: a family of integrin inhibitory proteins from viper venoms, Proc. Soc. Exp. Biol. Med. 195 (1990) 168–171. [2] R.J. Shebuski, D.R. Ramjit, G.H. Bencen, M.A. Polokoff, Characterization and platelet inhibitory activity of bitistatin, a potent arginine-glycine-aspartic acid-containing peptide from the venom of the viper Bitis arietans, J. Biol. Chem. 264 (1989) 21550–21556. [3] R.J. Shebuski, I.J. Stabilito, G.R. Sitko, M.H. Polokoff, Acceleration of recombinant tissue-type plasminogen activator-induced thrombolysis and prevention of reocclusion by the combination of heparin and the Arg-Gly-Asp-containing peptide bitistatin in a canine model of coronary thrombosis, Circulation 82 (1990) 169– 177. [4] O. Shigeta, P. Gluszko, S.W. Downing, W. Lu, S. Niewiarowski, L.H. Edmunds, Protection of platelets during long-term extracorporeal membrane oxygenation in sheep with a single dose of a disintegrin, Circulation 86 (Suppl. II) (1992) II398–II404. [5] L.C. Knight, K.E. Baidoo, J.E. Romano, J.L. Gabriel, A.H. Maurer, Imaging pulmonary emboli and deep venous thrombi with 99mTc-labeled bitistatin, a platelet-binding polypeptide from viper venom, J. Nucl. Med. 41 (2000) 1056–1064. [6] L.C. Knight, A.H. Maurer, J.E. Romano, Comparison of Iodine123-disintegrins for imaging thrombi and emboli in a canine model, J. Nucl. Med. 37 (1996) 476–482. [7] C.P. Blobel, J.M. White, Structure, function and evolutionary relationship of proteins containing a disintegrin domain, Curr. Opin. Cell Biol. 4 (1992) 760–765.
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319 [8] D. Yamada, Y. Shin, T. Morita, Nucleotide sequence of a cDNA encoding a common precursor of disintegrin flavostatin and hemorrhagic factor HR2a from the venom of Trimeresurus flavoviridis, FEBS Lett. 451 (1999) 299–302. [9] T.-F. Huang, J.C. Holt, H. Lukasiewicz, S. Niewiarowski, Trigramin. A low molecular weight peptide inhibiting fibrinogen interaction with platelet receptors expressed on glycoprotein IIb– IIIa complex, J. Biol. Chem. 262 (1987) 16157–16163. [10] M.S. Dennis, W.J. Henzel, R.M. Pitti, M.T. Lipari, M.A. Napier, T.A. Deisher, S. Bunting, R.A. Lazarus, Platelet glycoprotein IIb–IIIa protein antagonists from snake venoms: evidence for a family of platelet-aggregation inhibitors, Proc. Natl. Acad. Sci. USA 87 (1990) 2471–2475. [11] Z.R. Gan, R.J. Gould, J.W. Jacobs, P.A. Friedman, M.A. Polokoff, Echistatin: a potent platelet aggregation inhibitor from the venom of the viper, Echis carinatus, J. Biol. Chem. 263 (1988) 19827–19832. [12] R.M. Horton, H.D. Hunt, S.N. Ho, J.K. Pullen, L.R. Pease, Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension, Gene 77 (1989) 61–68. [13] D.J. Hakes, J.E. Dixon, New vectors for high level expression of recombinant proteins in bacteria, Anal. Biochem. 202 (1992) 293– 298. [14] K.L. Guan, J.E. Dixon, Eukaryotic proteins expressed in Escherichia coli: an improved thrombin cleavage and purification procedure of fusion proteins with glutathione S-transferase, Anal. Biochem. 192 (1991) 262–267. [15] P.H. Hirel, J.-M. Schmitter, P. Dessen, G. Fayat, S. Blanquet, Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid, Proc. Natl. Acad. Sci. USA 86 (1989) 8247–8251. [16] D.B. Wetlaufer, V.P. Saxena, A.K. Ahmed, S.W. Schaffer, P.W. Pick, K.J. Oh, J.D. Peterson, Protein thiol-disulfide interchange and interfacing with biological systems, Adv. Exp. Med. Biol. A 86 (1977) 43–50. [17] H. Scha¨gger, G. von Jagow, Tricine–sodium dodecyl sulfate– polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa, Anal. Biochem. 166 (1987) 368–379. [18] B.S. Coller, A new murine monoclonal antibody reports an activation-dependent change in the conformation of and/or
[19] [20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
319
microenvironment of platelet GP IIb/IIIa complex, J. Clin. Invest. 76 (1985) 101–108. J.S. Bennett, G. Vilaire, Exposure of platelet fibrinogen receptors by ADP and epinephrine, J. Clin. Invest. 64 (1979) 1393–1401. L.C. Knight, J.E. Romano, A.H. Maurer, In vitro platelet binding compared with in vivo thrombus imaging using aIIbb3-targeted radioligands, Thromb. Haemost. 80 (1998) 845–851. D.R. Burt, Receptor binding methodology and analysis, in: R.A. OÕBrien (Ed.), Receptor Binding in Drug Research, Marcel Dekker, New York, 1986, pp. 4–29. P. McQuade, L.C. Knight, M.J. Welch, Evaluation of 64Cu and 125 I-radiolabeled bitistatin as potential agents for targeting avb3 integrins in tumor angiogenesis, Bioconjug. Chem. 15 (2004) 1068–1075. Z.-R. Gan, J.H. Condra, R.J. Gould, R.A. Zivin, C.D. Bennett, J.W. Jacobs, P.A. Friedman, M.A. Polokoff, High-level expression in Escherichia coli of a chemically synthesized gene for [Leu28]echistatin, Gene (1989). M.S. Dennis, P. Carter, R.A. Lazarus, Binding interactions of kistrin with platelet glycoprotein IIb–IIIa: analysis by site-directed mutagenesis, Proteins: Struct. Funct. Genet. 15 (1993) 312–321. A.M. Moura-da-Silva, A. Lı´nica, M.S. Della-Casa, A.S. Kamiguti, P.L. Ho, J.M. Crampton, R.D.G. Theakston, Jararhagin ECD-containing disintegrin domain: expression in Escherichia coli and inhibition of the platelet–collagen interaction, Arch. Biochem. Biophys. 369 (1999) 295–301. M.D. Jones, L.O. Narhi, W.C. Chang, H.S. Lu, Refolding and oxidation of recombinant human stem cell factor produced in Escherichia coli, J. Biol. Chem. 271 (1996) 11301–11308. Y.C. Liou, M.E. Daley, L.A. Graham, C.M. Kay, V.K. Walker, B.D. Sykes, P.L. Davies, Folding and structural characterization of highly disulfide-bonded beetle antifreeze protein produced in bacteria, Protein Expr. Purif. 19 (2000) 148–157. J.X. Zhang, D.P. Goldenberg, Mutational analysis of the BPTI folding pathway: I. Effects of aromatic fi leucine substitutions on the distribution of folding intermediates, Protein Sci. 6 (1997) 1549–1562. S.J. Shattil, Function and regulation of the b3 integrins in hemostasis and vascular biology, Thromb. Haemost. 74 (1995) 149–155.
Functional expression of bitistatin, a disintegrin with potential use in molecular imaging of thromboembolic disease Linda C. Knight*, Jan E. Romano Nuclear Medicine Division, Radiology Department, Temple University School of Medicine, Philadelphia, PA 19140, USA Received 9 November 2004 Available online 8 December 2004
Abstract Bitistatin is a single-chain disintegrin which contains 83 amino acids and is internally crosslinked with seven disulfide bonds. This platelet aggregation inhibitor, which binds with high affinity to the aIIbb3 integrin, has potential use as the basis for a radiotracer to locate thrombi and emboli by scintigraphic imaging. A method amenable to large-scale, consistent production of bitistatin was sought. A synthetic gene coding for bitistatin was inserted into two different Escherichia coli expression vectors. One vector expressed recombinant bitistatin (rBitistatin) as a cleavable fusion protein and the other expressed rBitistatin as an isolated protein. In both cases, rBitistatin contained an additional amino acid (Gly) at the N-terminus compared with the native protein. The fusion protein was purified by affinity chromatography, then cleaved enzymatically to release rBitistatin, which was purified by reversedphase high performance liquid chromatography (HPLC) to a single active form. The rBitistatin produced as an isolated protein was purified from cell lysate by HPLC in a reduced form, then refolded, and purified again by HPLC. Yields of active rBitistatin averaged 12 mg/L for expression as an isolated protein, 10 times as high as when the fusion protein was employed. Structural assays confirmed the expected mass and sequence of the product. Functional assays (inhibition of platelet aggregation in vitro, equilibrium binding to platelets in vitro, and binding of labeled protein to experimental thrombi and emboli in vivo) confirmed that rBitistatin retained the functional characteristics of native bitistatin. 2004 Elsevier Inc. All rights reserved. Keywords: Bitistatin; Platelet aggregation inhibitor; Thromboembolic disease; Radionuclide imaging
Disintegrins from the venom of snakes are the most potent known antagonists of integrin function [1]. They were first described as inhibitors of platelet aggregation, by their binding to integrin aIIbb3 (the fibrinogen receptor, also known as glycoprotein IIb/IIIa) on the surface of platelets. All contain multiple cysteines at highly conserved positions, which result in a high degree of intramolecular crosslinking [1]. Disintegrins are also characterized by an Arg-Gly-Asp (RGD), Lys-Gly-Asp (KGD) or other receptor recognition sequence, presented at the apex of a flexible loop. Their affinity for integrin aIIbb3 is much higher than linear RGD con-
*
Corresponding author. Fax: +1 215 707 8110. E-mail address: [email protected] (L.C. Knight).
1046-5928/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.pep.2004.11.005
taining peptides or short cyclized peptides containing RGD. Bitistatin is a 83-amino acid monomeric disintegrin [1]. It is a potent inhibitor of platelet aggregation and has been shown to have beneficial effects in vivo such as inhibiting acute restenosis of experimental coronary artery interventions [2,3] and preventing platelet loss during extracorporeal circulation [4]. It is being investigated as the basis for a radiotracer imaging test for locating thrombi and emboli [5,6]. Bitistatin was originally isolated from the venom of the puff adder, Bitis arietans [2]. For eventual clinical use, however, isolation from snake venom was not considered sufficiently reliable or consistent, so we sought a recombinant product. Numerous reports indicate that snake venom disintegrins are synthesized in the venom gland as larger
308
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
proteins with multiple domains, and are cleaved in the gland to release the disintegrin domain [7,8]. For thrombus imaging, we wanted only the disintegrin domain, and aimed to express only the disintegrin domain to mimic the native bitistatin found in collected venom. One difficulty with expressing this molecule in Escherichia coli is that expressed protein in the cytoplasm of the cells is kept in a reduced form. Bitistatin contains 14 cysteines, which in the native molecule are fully crosslinked to form seven internal disulfide bridges. Correct folding of disintegrins is important for their biologic activity, to maintain the correct conformation of the RGD receptor-binding site. Several investigators have reported that reduction of native disintegrins abolishes their activity for inhibiting platelet aggregation [9– 11]. A challenge of this project was to refold and oxidize the expressed rBitistatin to form its disulfide bridges correctly. In this report, we have successfully developed a reproducible procedure for producing recombinant bitistatin from E. coli. The refolded product has biologic activity which is equivalent to native bitistatin. Note. The molecule discussed here is one of four isoforms of bitistatin previously reported. This isoform was previously reported as bitistatin 4 [1].
Methods General All chemicals were obtained from either Fisher Scientific (Pittsburgh, PA) or Sigma Chemical (St. Louis, MO) unless otherwise specified. Solvents and chemicals were the highest purity available and were used as received without further purification. HPLC separations Reversed-phase HPLC was used for purification and analysis. In all cases, a widepore C18 column packing ˚ pores, 10 lm particle size, Varian, Walnut Creek, (300 A CA) was used. The elutions employed 0.1% TFA (aq) as mobile phase A and 0.1% TFA in acetonitrile as mobile phase B. Semipreparative HPLC method 1 used a 21.4 · 250 mm column connected to a gradient HPLC system (two Rainin HPXL pumps with 50 mL heads and Knauer UV/Vis detector; Varian, Walnut Creek, CA). The column was eluted with a linear gradient of 0–60% mobile phase B over 30 min. The flowrate was 16 mL/min and elution was monitored by UV absorbance at 280 nm. Semipreparative HPLC method 2 used a 10 · 250 mm column with a gradient of 0–40% mobile phase B over 27.1 min, using the pumps and detector described above. The flowrate was 3.5 mL/min and elution was monitored by UV absorbance at 280 nm. Analytical HPLC used a 4.6 · 250 mm column, connected to an
analytical system (two model 510 pumps, gradient controller, model 486 UV detector; Waters, Milford, MA). The detector data were collected and analyzed with an AllChrom data analyzer (Alltech, Deerfield, IL) running in a personal computer. The column was eluted with a linear gradient of 0–60% mobile phase B over 30 min. The flowrate was 1.0 mL/min and elution was monitored by UV absorbance at 280 nm. Purification of native bitistatin Bitistatin was purified from freeze-dried B. arietans venom (Miami Serpentarium Laboratories, Punta Gorda, FL) using ion-exchange chromatography followed by reversed-phase HPLC as previously described [6]. The product was purified to a single peak and evaluated by SDS–PAGE, by MALDI mass spectrometry and by peptide mapping. Expression of bitistatin as a fusion protein The oligonucleotide sequence needed to code for bitistatin was deduced from the amino acid sequence of bitistatin (Fig. 1). The full-length double-stranded DNA was constructed using splicing by overlap extension [12] from four synthetic oligonucleotides of 85–88 nucleotides each (Paragon Biotech, Baltimore, MD). The final construct included a BamHI site at the 5 0 end, and the 3 0 end included a stop codon followed by an EcoRI site. The purified double-stranded synthetic DNA was ligated into the BamHI/EcoRI sites of pGEX-KT expression vector [13]. The resulting vector, referred to as pGEX-GSTrBit, predicts a protein containing glutathione S-transferase (GST) followed by a polyglycine spacer (‘‘kinker’’) region, a thrombin-cleavable sequence (LVPRR), a glycine, and the bitistatin sequence (Fig. 2). Thrombin cleavage of the fusion protein would result in recombinant bitistatin with an additional Gly at its N-terminus. Recombinant pGEX-GSTrBit vectors were introduced by transformation into E. coli BL21(DE3) (Novagen, Madison, WI), and bacteria were grown at 37 C in Luria–Bertani medium containing ampicillin (100 lg/mL) until the culture reached OD595 = 0.6. Expression of the fusion protein was induced with 1 mM isopropyl-1-thio-galactopyranoside (IPTG) for 3 h. The cells were harvested by centrifugation (1800g
Fig. 1. Amino acid sequence of native bitistatin (isoform bitistatin 4 after Gould et al. [1]).
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
309
Fig. 2. Expression construct in pGEX-GSTrBit vector. The expression product consists of glutathione-S-transferase (GST), followed by a polyglycine kinker (a spacer to enable thrombin access), a thrombin recognition/cleavage site, and bitistatin. The sequence of bitistatin is shown in bold. Cleavage by thrombin results in release of bitistatin which differs from native bitistatin in that it has an additional Gly at its amino terminus. Only one strand is shown, and the gene for GST is not shown in its entirety. The location of the BamHI restriction site is underlined.
for 25 min at 4 C), resuspended in 1/10 original volume of PBS, and spun down again. Cells were lysed by resuspension in 1/10 original volume of B-PER cell lysis reagent (Pierce, Rockford, IL) in the presence of 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 25 lg/mL aprotinin, and 25 lg/mL leupeptin. After 10 min of lysis, cellular debris was removed by centrifugation (27,000g for 15 min at 4 C). The lysate was incubated for 1 h at 4 C with glutathione–agarose beads (Pierce, Rockford, IL) to selectively bind the expressed fusion protein. After six washes to remove unbound impurities, the beads were suspended in TBS (50 mM Tris and 150 mM NaCl), pH 8.0, and placed on a tube rocker at 4 C to permit bitistatin to refold and oxidize to form disulfide crosslinks. Thrombin cleavage of the fusion protein was performed while the GST was still attached to glutathione–agarose [14]. Bovine thrombin (Amersham Biosciences, Piscataway, NJ; 2 U/mg fusion protein) and CaCl2 (final concentration 2.5 mM) were added to cleave rBitistatin from the fusion protein. Rocking of the tube containing the beads was continued at 4 C
for 24–48 h. Refolding of cleaved protein was monitored by sampling the bead supernatant and analyzing it on analytical reversed-phase HPLC. When there was no further change in the HPLC profile, the supernatant was collected from the beads and fractionated by semipreparative reversed-phase HPLC, method 2. Peaks were collected manually as they eluted. The collected fractions were tested for inhibition of ADP-induced human platelet aggregation. The peak with the highest potency in this assay was characterized further. Expression of bitistatin without a fusion partner Because of concerns about introducing viral contaminants from the use of human or bovine thrombin for cleavage, we wanted to express bitistatin without a fusion partner. The gene for bitistatin was cleaved from the BamHI/EcoRI sites of the pGEX-GSTrBit vector and separated on an agarose gel. The band containing the bitistatin gene was excised from the gel and purified with a QIAquick gel extraction kit (Qiagen, Valencia, CA) (Fig. 3A).
310
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 3. Construction of plasmid for expression of bitistatin without a fusion partner. (A) Cleavage of bitistatin gene from pGEX-GSTrBit plasmid. (B) Modification of bitistatin gene with synthetic duplex extension. (C) Ligation of extended bitistatin gene into pET-5a plasmid. (D) Removal of unwanted bases to form final expression vector, pET-rBit.
To adapt the gene to a new expression vector, a doublestranded extension was prepared. Two oligonucleotides were synthesized: 5 0 TACATATGG and 5 0 GATCCCAT
ATG. Both oligonucleotides were phosphorylated on the 5 0 end. The synthetic oligos (dissolved in water) were mixed together in equimolar quantities and annealed by
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
311
Fig 3. (continued)
boiling 20 min followed by cooling at room temperature. The resulting double-stranded DNA segment (Fig. 3B) has a sticky end for ligation to the BamHI site on the rBitistatin gene cleaved from pGEX-GSTrBitis, and a sticky end for ligation into an NdeI site on a new vector. The yield of duplex was assumed to be 50%, and the mixture was carried on to the next step without further purification. The double-stranded extension was combined in 100-fold molar excess with the rBitistatin gene cleaved from pGEX-GSTrBitis. Ligation was carried out for 20 h at 4 C in the presence of T4 DNA ligase. The reaction was purified using QIAquick (Qiagen). This purified product was ligated into the NdeI/ EcoRI sites of pET-5a vector (Promega, Madison, WI). This construct, referred to as pET5-E3, should express bitistatin with a single Gly added at the N-terminus (Fig. 3C), and should theoretically provide exactly the same protein as the thrombin cleavage product of the fusion protein. This assumption is based on the removal of the N-terminal methionine in E. coli [15]. The vector was transformed into BL21trxB (DE3)pLysS cells (Novagen, Madison, WI). DNA sequencing was performed to confirm the correct construct. Because plasmid pET5-E3 contained an additional 4 bp which were not needed but which resulted in unwanted separation of the initiation codon from the Shine-Dalgarno sequence, plasmid pET5-E3 (purified by Wizard Plus SV Miniprep, Promega) was digested with NdeI and EcoRI. The 268 bp fragment containing the rBitis gene was purified on an agarose gel, purified
by QIAquick gel extraction, and ligated into the NdeI and EcoRI sites of freshly digested pET-5a vector (Fig. 3D). This construct, referred to as pET-rBit, was used to transform BL21trxB (DE3)pLysS cells (Novagen, Madison, WI). DNA sequencing was performed to confirm the correct construct. Bacteria harboring the pET-rBit plasmid were grown at 37 C in Luria–Bertani medium containing ampicillin (100 lg/mL), chloramphenicol (34 lg/mL), and kanamycin (15 lg/mL) until the culture reached OD595 = 0.6. Expression of rBitistatin was induced with 1 mM IPTG for 3 h. The cells were harvested by centrifugation (3000g for 10 min at 4 C), resuspended in 1/10 volume of PBS, and spun down again. Cells were lysed by three cycles of freeze–thawing in TBS, pH 7.5, containing 25 mg/ml aprotinin, 25 mg/ml leupeptin, 1 mM PMSF, and 0.1% Triton X-100. After adding MgCl2 to 10 mM and DNAse I to 10 lg/mL, cell debris was removed by centrifugation. Dithiothreitol (DTT) was added to a final concentration of 20 mM to reduce the expression products and prevent formation of disulfide crosslinks during purification. The lysate was diluted with 0.1% trifluoroacetic acid (TFA). After centrifuging to remove any precipitate, the supernatant was filtered through a 0.22 lm syringe filter (Corning Life Sciences, Acton, MA). A portion of the filtrate was loaded onto a semipreparative reversed-phase HPLC system and processed using method 1. Major peaks were collected manually. The peak eluting from 19 to 21 min was collected and freeze-dried (Centrivap, LabConCo, Kansas City,
312
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
MO), then stored at 70 C until refolding could be carried out. Refolding to allow reduced bitistatin to oxidize and form disulfide crosslinks was accomplished by dissolving the reduced bitistatin to a final concentration of 0.1 mg/mL in TBS, pH 7.5, containing 2.5 mM CaCl2. Glutathione and oxidized glutathione (USB, Cleveland, OH) were then added to final concentrations of 5.0 and 2.5 mM, respectively, to facilitate disulfide exchange [16]. The solution (100–150 mL) was stirred in a sterile 250 mL capacity polycarbonate conical vent-cap flask (Corning) at 4 C for 3 days. It was then placed in a 20 C freezer for 3–4 weeks. The product was then stored at 70 C until final purification. Final purification of refolded rBitistatin was accomplished using semipreparative HPLC method 2. The refolded rBitistatin solution was filtered (0.22 lm) before loading it into the system. Elution was monitored by UV absorbance at 280 nm. The first peak to elute (at 20–21 min) was manually collected and freeze-dried. Structural evaluations A sample of the final product was injected into the analytical HPLC system to determine its profile (retention time and peak shape). Elution was monitored by UV absorbance at 280 nm. Mass spectrometry (MALDI-MS) and amino acid analysis were performed on the final product. Mass mapping was performed on a reduced and alkylated product. Briefly, rBitistatin (1 mg/mL in 0.2 M Tris, pH 8.3, containing 2 mM EDTA) was reduced with dithiothreitol (DTT; 180:1 molar ratio DTT:rBitistatin) for 2 h at 37 C under an argon atmosphere. This was followed by addition of iodoacetamide (2900:1 molar ratio iodoacetamide:rBitistatin) and incubation at 37 C was continued for a further 30 min. The product was purified by reversed-phase HPLC and freeze-dried. The reduced, alkylated material was subjected to trypsin digestion for various times from 5 min to 8 h, and the fragments were analyzed by MALDI-MS. The fragments found were compared to the theoretical tryptic fragmentation pattern of the amino acid sequence. The expression products were analyzed by SDS–polyacrylamide gel electrophoresis (SDS–PAGE). For monitoring of expression and initial purification, precast 10–20% gels (Owl Separations, Portsmouth, NH) were run with Tris–Tricine SDS buffer, pH 8.5, and stained with Coomassie blue [17]. For testing the final product, precast 4–20% gradient gels, Tris–Hepes–SDS, pH 7.0 (Precise Protein Gels, Pierce, Rockford, IL), were run with 100 mM Tris, 100 mM Hepes, and 3 mM SDS, pH 8.0. The gels were stained with GelCode blue (Pierce, Rockford, IL). Molecular weight standards were run in separate lanes.
Functional assays Inhibition of platelet aggregation in vitro Under a protocol approved by the Institutional Review Board, blood was obtained from healthy human donors who had not taken aspirin or related medications for two weeks. The blood was anticoagulated with 1/10 volume of 3.8% sodium citrate and was centrifuged at 160g for 12 min to obtain platelet-rich plasma (PRP). The concentration of platelets in human PRP was determined using a hemacytometer, then diluted with the donorÕs own plasma to obtain a platelet concentration of 300,000/lL. The study was performed using an aggregometer (Payton Scientific, Buffalo NY). For each measurement, 400 lL of adjusted PRP was placed in a cuvette and stirred at 37 C. Vehicle (saline) or various concentrations of peptide were added in a volume of 10 lL and the mixture was stirred at 37 C for 1 min while recording light transmission. ADP was then added (final concentration 10 lM) to induce platelet aggregation. Platelet aggregation was determined by change in light transmission through the PRP, which was recorded for 5 min after ADP addition. The IC50 was determined (by least-squares fitting) as the concentration of peptide required to produce 50% inhibition of the response to ADP in the presence of the vehicle. Results were normalized to the results for the tetrapeptide Arg-GlyAsp-Ser (RGDS). Test of binding of radiolabeled rbitistatin to platelets Bitistatin was radiolabeled with 125I (Perkin–Elmer Life Sciences, N Billerica, MA) using IodoGen (Pierce, Rockville, IL) as previously described [6], and purified by gel chromatography. Citrated human blood and platelet-rich plasma (PRP) were obtained as described above. Gel-filtered platelets (GFP; plasma-free) were prepared by filtering 4 mL of citrated platelet-rich plasma through Sepharose 2B (2.5 · 8 cm column, equilibrated with an elution buffer containing 0.4 mM NaH2PO4, 12 mM NaHCO3, and 10 mM Hepes, pH 7.45, with 137 mM NaCl, 2.7 mM KCl, 0.1% glucose, and 0.2% bovine serum albumin). Fractions containing the highest platelet concentrations were pooled, adjusted to a final platelet concentration of 108 platelets/mL, and used within 1 h of preparation. Equilibrium binding assays were done as previously described [18–20]. Aliquots of the platelets were stimulated by addition of 10 lM ADP, combined with 125I-labeled rBitistatin or native bitistatin over a range of concentrations, and incubated for 1 h at 37 C. Platelets and bound radiotracer were separated from free radiotracer by centrifugation of triplicate samples through 0.1 ml of 30% (W/V) sucrose at 13,000g for 3 min. Nonspecific binding was determined in the presence of 9 mM EDTA. The Kd and Bmax were determined by Scatchard analysis [21].
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Test of binding to thrombotic lesions in vivo To confirm that recombinant bitistatin is capable of binding to thrombi and emboli in vivo, rBitistatin was radiolabeled with 123I (MDS Nordion, Kanata, Ontario) and tested in a standardized canine model of experimental thrombi and emboli [5]. The method of labeling rBitistatin with 123I was as previously described [6]. Thrombi (in a femoral vein) and pulmonary emboli were induced 24 h before radiotracer administration by placement of embolization coils (Cook, Bloomington, IN) in three animals. Approximately 10 mCi of [123I]rBitistatin and 50 lCi of 125I-native bitistatin were injected into a peripheral vein of each anesthetized animal. Scintigraphic lateral images of the chest and anterior views of the hind legs were acquired during the next 4 h, using a large field of view gamma camera (MaxiCamera, General Electric Medical Systems, Milwaukee, WI). The camera was interfaced to an image acquisition system (NucLear Mac, Scientific Imaging, Denver, CO) running in a Macintosh computer (Apple, Cupertino, CA) for digitization of the image data. At 4 h after radiotracer administration, thrombi and emboli were recovered for weighing and counting for radioactive content (Wizard 1480, Perkin–Elmer, Downers Grove, IL) for comparison with blood, muscle, and lung. A standard of the administered dose was also counted, to convert the counts/min/g of each tissue to percent of injected dose per gram of tissue. Statistical evaluation methods Comparison between two groups was performed using a two-tailed t test. Comparison among three groups was performed using ANOVA. A p value <0.05 was used to reject the null hypothesis. The statistical calculations were performed using Microsoft Excel (Microsoft, Redmond, WA).
Results Expression of rbitistatin as a fusion protein GST::rBitistatin fusion protein was purified from cell lysate by binding to glutathione–agarose. Fig. 4 shows SDS–PAGE of cell growth and affinity purification of the fusion protein, resulting in a band with an apparent molecular weight of 37 kDa, in agreement with the theoretical sum of GST (28 kDa), and rBitistatin (9 kDa). The amount of GST::rBitistatin fusion protein purified from cell lysate averaged 26.3 mg/L of culture. Thrombin cleavage of GST::rBitistatin fusion could theoretically be done either in solution (after release of fusion protein from the agarose support) or while the GST::rBitistatin is still bound to the support. The first method was attempted, but resulted in considerable precipitate formation which lowered yields of the rBitistatin
313
Fig. 4. SDS–PAGE of pGEX-GSTrBit expression of GST::rBitistatin fusion protein and thrombin cleavage. The 10–20% gel was run with Tris–Tricine buffer, pH 8.5, under reducing conditions. Lane 1, molecular weight standards; lane 2, cell culture before IPTG; lane 3, cell culture 3 h after IPTG; lane 4, cell lysate; lane 5, supernatant after incubating cell lysate with glutathione–agarose; lane 6, immobilized GST::rBitistatin; lane 7, thrombin digest of GST::rBitistatin (unfractionated); lane 8, thrombin digest HPLC fraction 1; lane 9, thrombin digest HPLC fraction 2; lane 10, thrombin digest HPLC fraction 3; lane 11, final product rBitistatin; and lane 12, native bitistatin.
(data not shown). The second method, suggested by Guan and Dixon [14], was more successful in our hands. As seen on SDS–PAGE, thrombin treatment cleaved the fusion protein, releasing material which migrated as a single band and had approximately the same mobility as native bitistatin under reducing conditions (Fig. 4, lanes 7 and 12). Analysis of the product of thrombin digestion on reversed-phase HPLC revealed three prominent peaks followed by a tail (Fig. 5). Separation of the different forms and analysis of their bioactivity (inhibition of platelet aggregation) revealed that peak 2 had platelet activity comparable to native bitistatin, whereas other peaks were less active. The separated peaks had approximately the same mobilities on SDS–PAGE (Fig. 4). Peak 1, however, stained poorly and migrated slightly farther into the gel, suggesting that it may have been degraded. Peak 2 was purified, but because it was difficult to resolve from neighboring peaks, yields were low. The amount of thrombin-cleaved, purified refolded rBitistatin averaged 1–2 mg per original liter of culture. Expression of rbitistatin without a fusion partner In the direct production method using the pET-rBit vector, separation of reduced cell lysate on reversed-phase HPLC resulted in the elution profile shown in Fig. 6. The peak which eluted from 19 to 21 min had the same mobility as native bitistatin on SDS–PAGE run under reducing
314
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 5. Reversed-phase HPLC profile of supernatant from glutathione–agarose beads treated with thrombin for 24 h. Peaks 1, 2, and 3 were collected separately. Peak 2 contained functionally active rBitistatin.
Fig. 6. Semipreparative reversed-phase HPLC profile of DTT-reduced lysate from bacteria containing pET-rBit. The clarified lysate was loaded onto a C18 widepore column, and eluted by HPLC method 1 (see text). The elution was monitored by UV at 280 nm. The indicated peak (*) contained reduced rBitistatin. The black bar indicates the eluate collected for refolding.
conditions. Fortunately, this peak was well separated from other cell components and was therefore easy to purify by collecting the entire peak. In the direct production method, an average of 15.6 ± 3.1 mg of purified, reduced rBitis was obtained per liter of culture.
Fig. 7. Analytical reversed-phase HPLC of rBitistatin from pET-rBit plasmid. (A) Reduced rBitistatin prior to refolding. (B) Refolded mixture prior to purification. The large peak at 9 min is glutathione. (C) Purified final product.
Reduced rBitistatin exhibited a retention time of about 21 min on analytical HPLC (Fig. 7A). As refolding was carried out, the peaks observed on reversedphase HPLC shifted gradually to earlier retention times. After initial aeration by stirring at 4 C, the period of storage at 20 C was found to be important for achieving the highest yields of correctly folded rBitistatin. Fig. 7B shows an elution profile of refolded rBitistatin after it had nearly completed refolding. The profile consists of an initial spike followed by a broad secondary peak. In some cases, the secondary peak was merely a tail following the spike. The initial spike was found to contain the material which had the best bioactivity for inhibiting platelet aggregation (see below). The final product after purification eluted as a single sharp peak on reversedphase HPLC (Fig. 7C). The average yield was 12 ± 3 mg of purified, refolded rBitis per liter of starting culture. To determine which part of the refolded rBitistatin had the highest functional activity, a batch of (unpurified) refolded material was separated on HPLC into narrow fractions. The fractions were freeze-dried and individually analyzed on analytical HPLC to confirm that each had a discrete retention time. Fraction A was a pure peak corresponding to the initial spike (retention time 19.2 min on analytical reversed-phase HPLC). Fractions C, D, and E were pure peaks eluting
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
315
at 19.85, 20.05, and 20.41 min, respectively. It was not possible to obtain a pure fraction between A and C, so fraction B contained two overlapping peaks of retention times (areas) 19.2 min (44%) and 19.6 min (56%). Samples of each fraction were tested for their potency in inhibiting platelet aggregation. Only fraction A had potency similar to that of native bitistatin; all other fractions were significantly less potent (Table 2). Additional samples of each fraction were treated with DTT to fully reduce disulfides, and then analyzed on HPLC. The reduction caused all peaks to shift to the same retention time, 20.7 min (not shown). Structural analysis On analytical HPLC, there was no difference in retention time between rBitistatin produced directly in the pET-rBit vector and rBitistatin cleaved from the fusion protein (Fig. 8). The retention time of rBitistatin was similar to that of native bitistatin (not shown). SDS–PAGE analysis of rBitistatin is shown in Fig. 9. Under reducing conditions, both rBitistatins had the same mobility as native bitistatin. This 9 kDa protein typically migrates on SDS–PAGE at a higher apparent molecular weight of about 13 kDa. On SDS–PAGE under nonreducing conditions, the bitistatin and rBitistatin bands are more diffuse and tend to elute from the gel during staining; however, they migrate with approximately
Fig. 9. SDS–PAGE of native and recombinant bitistatins. Gel (4–20% gradient) run under reducing conditions. Lane 1, molecular weight standards; lane 2, native bitistatin; lane 3, rBitis from pGEX-GSTrBit; and lane 4, rBitis from pET-rBit.
the same mobility as under the reducing conditions (not shown). rBitistatin had an observed mass of 9076.8 on MALDI-MS. This is consistent with the expected value of 9079, within the error of the instrument. This is also in agreement with the expected removal of the N-terminal methionine in E. coli [15], leaving a Gly at the N-terminus. Native bitistatin (without the N-terminal Gly) had an observed mass of 9023.8 (theoretical 9022). The MALDI-MS masses did not change after treatment with vinylpyridine under denaturing (6 M guanidinium hydrochloride, 50 mM Hepes, pH 9.0) but nonreducing conditions, indicating that neither native Bitistatin nor rBitistatins contained free sulfhydryl groups. The results of amino acid analysis (Table 1) agreed well with the expected result. Peptide mapping was done by tryptic digestion of reduced and alkylated rBitistatin. Expected fragments were found, covering 95% of the sequence of bitistatin. The section of the sequence which was not confirmed was between Thr59 and Arg61, in a section of the protein where trypsin digestion results in very small fragments. Functional analysis
Fig. 8. Reversed-phase HPLC of purified rBitistatin: (A) from pGEXGSTrBit and (B) from pET-rBit.
Ability to inhibit platelet aggregation Fig. 10 shows examples of inhibition curves for rBitistatin, native bitistatin, and RGDS for single determinations. There was no significant difference between native bitistatin (IC50 125 ± 21 nmol/L, mean ± SD) (n = 5) and rBitistatin produced by cleavage of fusion protein (124 ± 43 nmol/L, p = 0.97) (n = 6) or rBitistatin produced by direct expression (134 ± 24 nmol/L, p = 0.50) (n = 9).
316
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Table 1 Amino acid analysis of rBitistatin AA
Calc
Found
AA
Calc
Found
AA
Calc
Found
Asx Ser Glx Gly His Thr Arg
14 5 9 8 2 4 3
13.46 4.49* 8.15 7.81 2.15 4.02 3.01
Ala Cys Pro Tyr Val Met Lys
5 14 5 1 2
5.16 8.50* 4.99 0.67 1.94
Ile Leu Phe Trp NorLeu AHX
2 2 1 2
2.05 2.21 1.14
5
4.94
Notes. *Cys and *Ser partially and
**
**
Trp completely destroyed during hydrolysis.
Table 2 Test of fractions isolated from refolded recombinant bitistatin for their ability to inhibit platelet aggregation Fraction
Retention time
IC50 (nmol/L)b
Aa B C D E
19.2 min 19.2 (44%) + 19.6 (56%) 19.85 min 20.05 min 20.41 min
145 266 957 1178 709
a
Corresponds to final purified rBitistatin. Concentration needed to inhibit ADP-induced human platelet aggregation by 50%. b
Fig. 10. Inhibition of human platelet aggregation by native bitistatin (solid diamonds), recombinant bitistatin (open triangles), and the tetrapeptide RGDS (solid squares).
Binding to ADP-stimulated platelets Both 125I-labeled native and recombinant bitistatin bound to a single class of receptors on ADP-stimulated platelets in a saturable manner (Fig. 11). The equilibrium binding constants were similar: 8.1 nmol/L for native bitistatin and 7.7 nmol/L for recombinant bitistatin. Binding to thrombotic lesions in vivo The ability of rBitistatin to bind to thrombi and emboli in vivo was tested in a standardized canine model. Fig. 12A shows the images obtained with 123I-labeled rBitistatin clearly delineating an experimental pulmonary embolus which was only 5 mm in size. Fig. 12B shows an image of [123I]rBitistatin accumulating in a deep vein thrombus in the leg of the same animal. The target-to-background contrast obtained with [123I]rBitistatin compares favorably with the image contrast
Fig. 11. Equilibrium binding of 125I-labeled native and recombinant bitistatin to ADP-stimulated platelets. (A) 125I-native bitistatin. Kd = 8.1 nmol/L, Bmax = 55,246 sites/platelet. (B) 125I-recombinant bitistatin. Kd = 7.7 nmol/L, Bmax = 79,587 sites/platelet. Each data point represents the average of three determinations.
obtained previously with 123I-native bitistatin [6]. In this study, the percentage of injected dose which accumulated in each tissue (thrombus, embolus, blood, muscle, and lung) was not significantly different between the two radiotracers ([123I]rBitistatin and 125I-native bitistatin) (p > 0.05).
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
Fig. 12. Planar scintigraphic images of experimental thromboembolic disease with [123I]rBitistatin. (A) Left lateral view of chest of dog showing uptake in pulmonary embolus (arrow) at 1 h 50 min after radiotracer administration. Radiotracer accumulation in the embolus is more intense than in surrounding background tissues. Radiotracer is also seen in the blood pool in the chambers of the heart, the major blood vessels, and in the spleen (only the top of the spleen is in the field of view). (B) Anterior view of legs showing uptake in femoral vein thrombus (arrow) at 1 h 40 min after radiotracer administration. Radiotracer accumulation in the thrombus is more intense than in surrounding blood in the major vessels or surrounding muscle tissue.
Discussion Many groups have had an interest in recombinant expression of disintegrins, because this method enables the production of larger quantities of reproducible material than are available from natural sources. In addition, it provides the framework for site-directed mutagenesis to study the function and specificity of disintegrins. Because of their integrin-inhibitory activity, disintegrins
317
are being explored as the basis for design of drugs to inhibit cell adhesion and associated processes in diseases such as thrombosis, atherosclerosis, and cancer. Bitistatin, when labeled with an imaging radionuclide, has been shown to have potential as a diagnostic imaging agent for locating thrombi and emboli [5,6], and also has promise for imaging tumor angiogenesis [22]. Recombinant bitistatin was initially expressed as a fusion protein to simplify the purification process from crude cell lysate. However, refolding of rBitistatin into a correct orientation with proper disulfide pairing did not occur efficiently in the fusion with GST. Several distinct forms were found on reversed-phase HPLC. Only one peak had activity for inhibiting platelet aggregation but the three main peaks had similar mobility on reduced SDS–PAGE. This suggests that rBitistatin assumed several different folding configurations in the fusion protein, or that disulfide crosslinking was incomplete in some. Peak 1 may represent partially degraded rBitistatin. This might be a consequence of adding thrombin protease (which might contain other proteases) or the glutathione–agarose support (which might retain some proteases from the cell lysate). The different forms of rBitistatin eluted close together on HPLC, resulting in low yields of the desired component. Expression of rBitistatin as an isolated protein was achieved by transferring the gene into another expression vector. Purification of expressed rBitistatin in a reduced form was simplified by prior knowledge of its retention time on reversed-phase HPLC. Fortunately, reduced rBitistatin elutes well apart from other eluting cellular proteins in this system. The refolding and disulfide-pairing process in this case could be carried out with highly purified reduced rBitistatin in the absence of contaminating proteins and resulted in high yields of the correctly folded form, with no apparent degradation. It was found that addition of glutathione and oxidized glutathione was helpful in optimizing the folding process. In addition, refolding to the desired form appeared to accelerate when the 4 C solution was chilled further to 20 C. Other snake venom disintegrins have been produced by recombinant expression in E. coli. Various approaches have been used to achieve correct refolding and disulfide bond formation. Echistatin, which has 49 amino acids and four disulfide crosslinks, was produced from a synthetic gene as a fusion protein [23]. The expressed protein was found in inclusion bodies. After cleavage from its fusion protein, echistatin was completely reduced and refolded, enabling more than 30% of the reduced protein to be recovered as correctly folded echistatin. Kistrin, which contains 68 amino acids and six disulfide crosslinks, was produced as a fusion protein in an expression vector which directed the product to the periplasm [24]. It was cleaved from its fusion partner and purified by reversed-phase HPLC. Correctly folded recombinant kistrin was found, but the yields
318
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319
were only 0.15–0.30 mg of correctly folded kistrin per liter of culture. The disintegrin domain of jararhagin was cloned as a fusion with thioredoxin, and expressed in a vector which should allow disulfide bond formation in the cytoplasm in E. coli [25]. Upon purification, however, it was found that only a fraction of the total disintegrin was correctly folded and crosslinked. These studies suggest that it is unlikely that the isolated disintegrin domain, coupled to a foreign protein, will refold efficiently in the same configuration as it does in the venom gland when coupled to its natural neighboring domains. Our experience with rBitistatin as a fusion protein agrees with the kistrin and jararhagin disintegrin reports, in that we found a mixture of refolded forms and obtained relatively low yields of the correctly folded form. In this study, when we purified the rBitistatin in a reduced form and refolded it in isolation, we were able to recover an average of 75% of reduced rBitistatin as correctly folded protein, for an overall yield of 12 mg/L of original culture. During the refolding of rBitistatin which had been expressed as an isolated protein, analysis was done to see which fractions had bioactivity and whether they represented differently folded forms of rBitistatin. The finding that all peaks shifted to the same retention time upon DTT reduction suggests that the fractions all contained the same peptide with different or incomplete disulfide bridging patterns. The fraction with the earliest retention time on HPLC had the lowest IC50 for inhibition of human platelet aggregation. The fully reduced form had the latest retention time. This is in agreement with the observations of others who tracked the refolding of recombinant proteins with reversed-phase HPLC [26–28]. Reversed-phase HPLC separates molecules in part by hydrophilicity, with the most hydrophilic forms eluting first. It may be that during optimal folding of disintegrins, hydrophobic residues fold toward the interior, and hydrophilic residues remain on the outside of the molecule. Bitistatin binds to the aIIbb3 integrin on platelets, which is also known as the glycoprotein IIb/IIIa receptor. Binding to this receptor blocks fibrinogen binding and, in adequate concentrations, inhibits aggregation of platelets in response to an agonist such as adenosine diphosphate (ADP). Recombinant bitistatin retains the ability to inhibit platelet aggregation in platelet-rich plasma prepared from the citrated blood of normal human donors. Radiolabeled rBitistatin binds to platelets with an affinity which is no different from native bitistatin, and has a similar number of apparent binding sites. The number of binding sites found in this study (55,000– 80,000) agrees with the reported number of aIIbb3 sites (40,000–80,000 sites per platelet) [29]. Most importantly, radiolabeled rBitistatin is equivalent to native bitistatin in its ability to bind to platelet deposits (thrombi and emboli) in vivo and to produce high-contrast images
of the thrombus over background within a short time after radiotracer administration. In this project, rBitistatin was produced by expression in E. coli as a fusion protein and also as an isolated protein. The final products of these two methods appear to be the same, however the yield is much higher for production as an isolated protein. The findings of this study suggest that recombinant bitistatin is suitable for further development as a radiopharmaceutical for imaging thromboembolic disease in vivo. Acknowledgments The authors are grateful to Dr. Alan Maurer for assistance with the animal model, to Drs. Kwamena Baidoo, Virginia Heatwole, and Bill Wong for assistance with the initial expression vector, to Drs. William Moore, Tapan Ganguly, Juan Calvete, and Kaye Speicher, and to Tom Beer and Nicole DiFlorio, for help with molecular analyses. DNA synthesis and sequencing was performed by Paragon Bioscience and by the University of Pennsylvania DNA Sequencing Facility (funded by the National Cancer Institute). Protein analyses were performed by the University of Pennsylvania Protein Chemistry Laboratory and the Wistar Institute Proteomics Facility. This work was supported by NIH R01 HL 54578.
References [1] R.J. Gould, M.A. Polokoff, P.A. Friedman, T.-F. Huang, J.C. Holt, J.J. Cook, S. Niewiarowski, Disintegrins: a family of integrin inhibitory proteins from viper venoms, Proc. Soc. Exp. Biol. Med. 195 (1990) 168–171. [2] R.J. Shebuski, D.R. Ramjit, G.H. Bencen, M.A. Polokoff, Characterization and platelet inhibitory activity of bitistatin, a potent arginine-glycine-aspartic acid-containing peptide from the venom of the viper Bitis arietans, J. Biol. Chem. 264 (1989) 21550–21556. [3] R.J. Shebuski, I.J. Stabilito, G.R. Sitko, M.H. Polokoff, Acceleration of recombinant tissue-type plasminogen activator-induced thrombolysis and prevention of reocclusion by the combination of heparin and the Arg-Gly-Asp-containing peptide bitistatin in a canine model of coronary thrombosis, Circulation 82 (1990) 169– 177. [4] O. Shigeta, P. Gluszko, S.W. Downing, W. Lu, S. Niewiarowski, L.H. Edmunds, Protection of platelets during long-term extracorporeal membrane oxygenation in sheep with a single dose of a disintegrin, Circulation 86 (Suppl. II) (1992) II398–II404. [5] L.C. Knight, K.E. Baidoo, J.E. Romano, J.L. Gabriel, A.H. Maurer, Imaging pulmonary emboli and deep venous thrombi with 99mTc-labeled bitistatin, a platelet-binding polypeptide from viper venom, J. Nucl. Med. 41 (2000) 1056–1064. [6] L.C. Knight, A.H. Maurer, J.E. Romano, Comparison of Iodine123-disintegrins for imaging thrombi and emboli in a canine model, J. Nucl. Med. 37 (1996) 476–482. [7] C.P. Blobel, J.M. White, Structure, function and evolutionary relationship of proteins containing a disintegrin domain, Curr. Opin. Cell Biol. 4 (1992) 760–765.
L.C. Knight, J.E. Romano / Protein Expression and Purification 39 (2005) 307–319 [8] D. Yamada, Y. Shin, T. Morita, Nucleotide sequence of a cDNA encoding a common precursor of disintegrin flavostatin and hemorrhagic factor HR2a from the venom of Trimeresurus flavoviridis, FEBS Lett. 451 (1999) 299–302. [9] T.-F. Huang, J.C. Holt, H. Lukasiewicz, S. Niewiarowski, Trigramin. A low molecular weight peptide inhibiting fibrinogen interaction with platelet receptors expressed on glycoprotein IIb– IIIa complex, J. Biol. Chem. 262 (1987) 16157–16163. [10] M.S. Dennis, W.J. Henzel, R.M. Pitti, M.T. Lipari, M.A. Napier, T.A. Deisher, S. Bunting, R.A. Lazarus, Platelet glycoprotein IIb–IIIa protein antagonists from snake venoms: evidence for a family of platelet-aggregation inhibitors, Proc. Natl. Acad. Sci. USA 87 (1990) 2471–2475. [11] Z.R. Gan, R.J. Gould, J.W. Jacobs, P.A. Friedman, M.A. Polokoff, Echistatin: a potent platelet aggregation inhibitor from the venom of the viper, Echis carinatus, J. Biol. Chem. 263 (1988) 19827–19832. [12] R.M. Horton, H.D. Hunt, S.N. Ho, J.K. Pullen, L.R. Pease, Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension, Gene 77 (1989) 61–68. [13] D.J. Hakes, J.E. Dixon, New vectors for high level expression of recombinant proteins in bacteria, Anal. Biochem. 202 (1992) 293– 298. [14] K.L. Guan, J.E. Dixon, Eukaryotic proteins expressed in Escherichia coli: an improved thrombin cleavage and purification procedure of fusion proteins with glutathione S-transferase, Anal. Biochem. 192 (1991) 262–267. [15] P.H. Hirel, J.-M. Schmitter, P. Dessen, G. Fayat, S. Blanquet, Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid, Proc. Natl. Acad. Sci. USA 86 (1989) 8247–8251. [16] D.B. Wetlaufer, V.P. Saxena, A.K. Ahmed, S.W. Schaffer, P.W. Pick, K.J. Oh, J.D. Peterson, Protein thiol-disulfide interchange and interfacing with biological systems, Adv. Exp. Med. Biol. A 86 (1977) 43–50. [17] H. Scha¨gger, G. von Jagow, Tricine–sodium dodecyl sulfate– polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa, Anal. Biochem. 166 (1987) 368–379. [18] B.S. Coller, A new murine monoclonal antibody reports an activation-dependent change in the conformation of and/or
[19] [20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
319
microenvironment of platelet GP IIb/IIIa complex, J. Clin. Invest. 76 (1985) 101–108. J.S. Bennett, G. Vilaire, Exposure of platelet fibrinogen receptors by ADP and epinephrine, J. Clin. Invest. 64 (1979) 1393–1401. L.C. Knight, J.E. Romano, A.H. Maurer, In vitro platelet binding compared with in vivo thrombus imaging using aIIbb3-targeted radioligands, Thromb. Haemost. 80 (1998) 845–851. D.R. Burt, Receptor binding methodology and analysis, in: R.A. OÕBrien (Ed.), Receptor Binding in Drug Research, Marcel Dekker, New York, 1986, pp. 4–29. P. McQuade, L.C. Knight, M.J. Welch, Evaluation of 64Cu and 125 I-radiolabeled bitistatin as potential agents for targeting avb3 integrins in tumor angiogenesis, Bioconjug. Chem. 15 (2004) 1068–1075. Z.-R. Gan, J.H. Condra, R.J. Gould, R.A. Zivin, C.D. Bennett, J.W. Jacobs, P.A. Friedman, M.A. Polokoff, High-level expression in Escherichia coli of a chemically synthesized gene for [Leu28]echistatin, Gene (1989). M.S. Dennis, P. Carter, R.A. Lazarus, Binding interactions of kistrin with platelet glycoprotein IIb–IIIa: analysis by site-directed mutagenesis, Proteins: Struct. Funct. Genet. 15 (1993) 312–321. A.M. Moura-da-Silva, A. Lı´nica, M.S. Della-Casa, A.S. Kamiguti, P.L. Ho, J.M. Crampton, R.D.G. Theakston, Jararhagin ECD-containing disintegrin domain: expression in Escherichia coli and inhibition of the platelet–collagen interaction, Arch. Biochem. Biophys. 369 (1999) 295–301. M.D. Jones, L.O. Narhi, W.C. Chang, H.S. Lu, Refolding and oxidation of recombinant human stem cell factor produced in Escherichia coli, J. Biol. Chem. 271 (1996) 11301–11308. Y.C. Liou, M.E. Daley, L.A. Graham, C.M. Kay, V.K. Walker, B.D. Sykes, P.L. Davies, Folding and structural characterization of highly disulfide-bonded beetle antifreeze protein produced in bacteria, Protein Expr. Purif. 19 (2000) 148–157. J.X. Zhang, D.P. Goldenberg, Mutational analysis of the BPTI folding pathway: I. Effects of aromatic fi leucine substitutions on the distribution of folding intermediates, Protein Sci. 6 (1997) 1549–1562. S.J. Shattil, Function and regulation of the b3 integrins in hemostasis and vascular biology, Thromb. Haemost. 74 (1995) 149–155.