cDNA Cloning and Functional Expression of Jerdostatin, a Novel RTS-disintegrin from Trimeresurus jerdonii and a Specific Antagonist of the α1β1 Integrin*

Jerdostatin represents a novel RTS-containing short disintegrin cloned by reverse transcriptase-PCR from the venom gland mRNA of the Chinese Jerdons pit viper Trimeresurus jerdonii. The jerdostatins precursor cDNA contained a 333-bp open reading frame encoding a signal peptide, a pre-peptide, and a 43-amino acid disintegrin domain, whose amino acid sequence displayed 80% identity with that of the KTS-disintegrins obtustatin and viperistatin. The jerdostatin cDNA structure represents the first complete open reading frame of a short disintegrin and points to the emergence of jerdostatin from a short-coding gene. The different residues between jerdostatin and obtustatin/viperistatin are segregated within the integrin-recognition loop and the C-terminal tail. Native jerdostatin (r-jerdostatin-R21) and a R21K mutant (r-jerdostatin-K21) were produced in Escherichia coli. In each case, two conformers were isolated. One-dimensional 1H NMR showed that conformers 1 and 2 of r-jerdostatin-R21 represent, respectively, well folded and unfolded proteins. The two conformers of the wild-type and the R21K mutant inhibited the adhesion of α1-K562 cells to collagen IV with IC50 values of 180 and 703 nm, respectively. The IC50 values of conformers 2 of r-jerdostatin-R21 and r-jerdostatin-K21 were, respectively, 5.95 and 12.5 μm. Neither r-jerdostatin-R21 nor r-jerdostatin-K21 showed inhibitory activity toward other integrins, including αIIbβ3, αvβ3, α2β1, α5β1, α4β1, α6β1, and α9β1 up to a concentration of 24 μm. Although the RTS motif appears to be more potent than KTS inhibiting the α1β1 integrin, r-jerdostatin-R21 is less active than the KTS-disintegrins, strongly suggesting that substitutions outside the integrin-binding motif and/or C-terminal proteolytic processing are responsible for the decreased inhibitory activity.

The integrin family of cell adhesion proteins promotes the attachment and migration of cells on the surrounding extracellular matrix (1,2). Through signals transduced upon integrin ligation by extracellular matrix proteins, several integrins play key roles in promoting angiogen-esis and tumor metastasis (3). However, although antagonists of several integrins (e.g. ␣ 5 ␤ 1 , ␣ v ␤ 3 , and ␣ v ␤ 5 , the primary targets of endostatin, an endogenous negative regulator of angiogenesis (4)) are now under evaluation in clinical trials to determine their potential as therapeutics for cancer and other diseases (5,6), the precise regulation and exact action of integrins is still unclear (7,8). Thus, the integrins ␣ 1 ␤ 1 and ␣ 2 ␤ 1 are highly up-regulated by vascular endothelial growth factor in cultured endothelial cells, resulting in an enhanced ␣ 1 ␤ 1 -and ␣ 2 ␤ 1 -dependent cell spreading on collagen and it has been reported that these integrins provide critical support for vascular endothelial growth factor signaling, endothelial cell migration, and tumor angiogenesis (9). The ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins are highly expressed on the microvascular endothelial cells, and blocking of their adhesive properties by monoclonal antibodies (9,10) or by the snake venom disintegrin obtustatin (11) significantly reduced the vascular endothelial growth factor-driven neovascularization ratio and tumor growth in animal models. Moreover, null-mice lacking integrin ␣ 1 ␤ 1 develop normally, but exhibit reduced vascularity of the skin (9) and have reduced number and size of intratumoral capillaries (12). Ongoing studies with ␣ 2 knock-out mice also suggest a critical role in angiogenesis for the ␣ 2 ␤ 1 integrins (13,14). Thus, inhibitors of the ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins alone or in combination with antagonists of other integrins involved in angiogenesis may prove beneficial in the control of tumor neovascularization. ␣ 1 ␤ 1 and ␣ 2 ␤ 1 belong to the I-domain bearing subfamily of integrins, and specifically interact with collagen (15). However, despite sharing large structural homology, these two integrins have distinct collagen binding preferences: ␣ 1 ␤ 1 integrin is a very selective receptor of basement membrane type IV collagen, whereas ␣ 2 ␤ 1 is highly specific for fibrillar collagen types I-III (16,17). Substitution of the cytoplasmic domains of the ␣ 1 and ␣ 2 subunits in transfected human mammary epithelial cells revealed that the two integrins participate in different signal transduction pathways (18). Noteworthy, the ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins are the targets of snake venom toxins belonging to different protein families. C-type lectin-like proteins include selective and potent (i.e. EMS16 from Echis multisquamatus; IC 50 ϭ 6 nM) inhibitors of ␣ 2 ␤ 1 (19,20), whereas the only to date known snake venom proteins that specifically antagonize the function of the ␣ 1 ␤ 1 integrin are the disintegrins obtustatin (IC 50 ϭ 2 nM) from the venom of Vipera lebetina obtusa (11,21), viperistatin (IC 50 ϭ 0.08 nM) from Vipera palestinae (22) and lebestatin (IC 50 ϭ 0.4 nM) from Macrovipera lebetina. 3 The crystal structure of EMS16 in complex with the integrin ␣ 2 I-do-main has provided insight into the structural basis of the integrin inhibitory specificity of this C-type lectin protein (23). On the other hand, the primary ␣ 1 ␤ 1 binding motif of obtustatin and viperistatin is a KTS tripeptide located in a lateral position of the mobile disintegrins active loop (24), which displays concerted motions with the C-terminal region (25). Now, we report the molecular cloning, primary structure, recombinant expression, and integrin inhibitory characteristics of two conformers of jerdostatin, an RTS-containing disintegrin from the Chinese Jerdon pit viper Trimeresurus jerdonii, and of a R21K mutant. In cell adhesion assays, the recombinant (r-) jerdostatin 4 conformers of both the wildtype (r-jerdostatin-R21) and the R21K mutant (r-jerdostatin-K21) selectively blocked, albeit with different potency, the adhesion of K562 cells expressing the integrin ␣ 1 ␤ 1 to collagen IV. . ␣9and mock-transfected SW480 cells were generated as described (26). K562 and Jurkat cell lines, which express ␣ 5 ␤ 1 and ␣ 4 ␤ 1 integrins, respectively, were purchased from ATCC (Manassas, VA).

EXPERIMENTAL PROCEDURES
PCR Amplification of Jerdostatin cDNA-The T. jerdonii venom glands were collected from Yiliang, Yunnan, China. Isolation of mRNA and reverse transcription was conducted using the PolyATtract System 1000 kit and Reverse Transcription System kit, respectively, according to the manufacturer's protocols. DNA was amplified by PCR using total reverse transcriptase-PCR products as template. The forward primer, 5Ј-CCAAATCCAG(C/T)CTCCAAAATG-3Ј, and the reverse primer, 5Ј-TTCCA(G/T)CTCCATTGTTG(G/T)TTA-3Ј, were designed according to the highly conserved 5Ј-and 3Ј-noncoding regions of the cDNAs encoding for elegantin-2a from Trimeresurus elegans (GenBank accession number AB059572), elegantin-1a from T. elegans (GenBank accession number AB059571), and HR2a from Trimeresurus flavoviridis (27). The PCR amplification protocol included 35 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 2 min. The recovered PCR products were cloned into PMD18-T vector (TaKaRa), and then transformed into Escherichia coli strain JM109. The white transformants were screened by PCR and the positive clones were subjected to sequencing on an Applied Biosystems model 377 DNA sequencing system.
Generation of r-Jerdostatin R21K Mutant-Site-directed mutagenesis was performed essentially as described in the QuikChange sitedirected mutagenesis kit of Stratagene (La Jolla, CA). To this end, plasmid pET-32a containing the wild-type jerdostatin sequence flanked by NcoI and XhoI restriction sites was used as the template in the PCR (denaturation at 94°C for 2 min, followed by 12 cycles of denaturation (30 s at 94°C), annealing (60 s at 55°C), and extension (12 min at 68°C), and a final extension for 10 min at 68°C) using the forward primer 5Ј-GGAACAACATGCTGGAAAACCAGTGTATCAAGTCATTAC-TGC-3Ј and the reverse primer 5Ј-ACTTGATACACTGGTTTTC-CAGCATGTTGTTCCTGCCGGC-3Ј in which the Arg codon AGA has been substituted AAA (Lys) (in boldface). The mutant DNA was sequenced to confirm the absence of undesired mutations.
Purification of KTS-disintegrins-Obtustatin and viperistatin were purified from the venoms of V. lebetina obtusa and V. palestinae, respectively, using the previously described two-step reversed-phase HPLC (11,21,22). The purity of the disintegrins was assessed by SDS-PAGE. The monoisotopic masses of the purified disintegrins were determined either by electrospray ionization mass spectrometry with a triple quadrupole-ion trap hybrid instrument (QTrap from Applied Biosystems) equipped with a nanospray source (Protana, Denmark) or by MALDI-TOF mass spectrometry (MS) using an Applied Biosystems DE-Pro spectrometer, operated in delayed extraction and reflector modes, and ␣-hydroxycinnamic acid saturated in 0.1% trifluoroacetic acid in 70% acetonitrile as the matrix. A tryptic peptide mixture of Cratylia floribunda seed lectin (SwissProt accession code P81517) prepared and previously characterized in our laboratory was used as a mass calibration standard (mass range 450 -3300 Da). For determination of isotope-averaged molecular masses, the instrument was operated in the linear mode using 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid) saturated in 70% acetonitrile and 0.1% trifluoroacetic acid as the matrix. The mass calibration standard consisted of a mixture of the following proteins, whose isotope-averaged molecular mass in daltons are given in parentheses: bovine insulin (5,734.6), E. coli thioredoxin (11,674.5), and horse apomyoglobin (16,952.6).
Protein concentration was determined with the bicinchoninic acid (BCA) protein quantification kit (Pierce) with bovine serum albumin as a standard, or by amino acid analysis (after hydrolysis in 6 N HCl for 24 h at 110°C in air-evacuated and sealed ampoules) using a Biochrom (Amersham Biosciences) amino acid analyzer.
Cloning and Production of Recombinant r-Jerdostatin-Thioredoxin-His 6 Fusion Proteins-The jerdostatin cDNA coding for wild-type and R21K fragments were amplified by PCR using primers synthesized by Sigma-Genosys (Haverhill, UK). The forward primer was 5Ј-CGTGC-CATGGATTGTACAACTGGACCATG-3Ј, which contained a NcoI restriction site (underlined) and the sequence coding for the first six residues of the protein. The reverse primer was 5Ј-GCCTCGAGTAT-TAGCCATTCCCGGGATAAC-3Ј, which includes a restriction site for XhoI (underlined), a stop codon (in italics and bold), and the last six C-terminal residues of jerdostatin. The PCR protocol included denaturation at 94°C for 2 min, followed by 40 cycles of denaturation (10 s at 94°C), annealing (15 s at 55°C), and extension (20 s at 72°C), and a final extension for 7 min at 72°C. The amplified fragments were purified using the Perfect Pre Gel Clean Up kit (Eppendorf, Hamburg, Germany) and cloned in a pGEM-T vector (Promega, Madison, WI). E. coli DH5␣ cells (Novagen, Madison, WI) were transformed by electroporation using an Eppendorf 2510 electroporator following the manufacturer's instructions. Positive clones, selected by growing the transformed cells in Luria broth (LB) medium containing 100 g/ml ampicilin, were confirmed by PCR amplification using the above primers, and the PCRamplified fragments were sequenced (using an Applied Biosystems model 377 DNA sequencer) to check the correctness of the sequences of the wild-type and the R21K jerdostatins open reading frame.
To construct an expression vector of jerdostatin-thioredoxin-His 6 wild-type and mutated fusion proteins the pGEM-T-jerdostatin plasmid and a pET32a vector (Novagen) were digested with NcoI and XhoI for 12 h at 37°C and the 132-bp jerdostatin fragments and the pET32a vector were purified after agarose gel electrophoresis with the Eppendorf Perfect Pre Gel Clean Up kit. The jerdostatin fragments and the open pET32a vector were ligated with T4 DNA ligase (Invitrogen) overnight at 13°C. These constructs were used to transform electrocompetent E. coli DH5␣ cells. The plasmidic DNAs from positive clones were used to transform (by electroporation) E. coli Origami B cells (Novagen). Another pool of cells were transformed with mock pET32a plasmid and used as negative control for the recombinant expression of jerdostatin-thioredoxin fusion protein.
Recombinant Expression of Jerdostatin-Thioredoxin-His 6 Fusion Proteins-Positive E. coli Origami B clones, shown by PCR to contain the jerdostatin-thioredoxin fusion constructs, wild-type or R21K mutant, were grown overnight at 37°C in LB medium containing 100 g/ml of ampicillin, 33 g/ml of kanamycin, and 12 g/ml of tetracyclin, followed by a 1:10 (v/v) dilution in the same medium. For the induction of the expression of the recombinant fusion proteins, isopropyl ␤-D-thiogalactosidase was added to a final concentration of 1 mM, and the cell suspensions were incubation for another 7 h at 37°C. Thereafter, the cells were pelleted by centrifugation, resuspended in the same volume of 20 mM sodium phosphate, 150 mM NaCl, pH 7.4, washed three times with the same buffer, and resuspended in 100 ml/liter of initial cell culture of 20 mM sodium phosphate, 250 mM NaCl, 10 mM imidazole, pH 7.4. The cells were lyzed by sonication (15 cycles of 15 s sonication followed by 1 min resting) in an ice bath. The lysates were centrifuged at 10,000 ϫ g for 30 min at 4°C, and the soluble and the insoluble fractions were analyzed by SDS-15% polyacrylamide gel electrophoresis.
Purification of Recombinant Jerdostatin Molecules-The jerdostatinthioredoxin-His 6 fusion proteins, wild-type and R21K mutant, were purified from the soluble fraction of positive E. coli Origami clone, the lysate was purified by affinity chromatography using an Ä KTA Basic chromatograph equipped with a 5-ml HisTrap HP column (Amersham Biosciences) equilibrated in 20 mM sodium phosphate, 250 mM NaCl, 10 mM imidazole, pH 7.4, buffer. After absorbance at 280 nm of the flowthrough fraction reached baseline, the bound material was eluted at a flow rate of 1.5 ml/min with a linear gradient of 10 -500 mM imidazole for 60 min. The purified protein fractions (checked by SDS-PAGE) were pooled, dialyzed against 50 mM Tris/HCl, pH 7.4, and digested with 0.25 units of enterokinase (Invitrogen) per mg of recombinant protein. The reaction mixture was freed from enterokinase by chromatography on a 0.5-ml column of agarose-trypsin inhibitor (Sigma) equilibrated and eluted with 50 mM Tris-HCl, pH 7.4. Jerdostatin was separated from thioredoxin-His 6 by chromatography of the agarose-trypsin-inhibitor non-bound fraction on a HisTrap column (as above), and the nonbound and retarded fractions, both containing jerdostatin, were further purified by reverse-phase HPLC followed by size-exclusion chromatography using an Ä KTA Basic chromatograph equipped with a Superdex Peptide column (Amersham Biosciences) eluted with phosphate-buffered saline buffer at a flow rate of 0.3 ml/min. The purity of the isolated proteins was assessed by SDS-PAGE, reverse-phase HPLC, N-terminal sequence analysis (using an Applied Biosystems Procise instrument), and MALDI-TOF mass spectrometry as described above for the KTSdisintegrins, and nanoelectrospray ionization mass spectrometry using a QTrap instrument (Applied Biosystems) equipped with a nanoelectrospray source (Proxeon, Denmark). Protein concentration of purified recombinant jerdostatin was determined spectrophotometrically using an ⑀ at 280 nm of 10,677 M Ϫ1 cm Ϫ1 calculated by amino acid analysis as above.
In-gel Tryptic Digestion and Mass Fingerprinting-The recombinant expression of the jerdostatin-thioredoxin-His 6 fusion proteins and the purification of r-jerdostatin molecules were monitored by SDS-PAGE and mass fingerprinting. To this end, SDS-PAGE separated polypeptides were subjected to automated digestion with sequencing grade bovine pancreatic trypsin (Roche) at a final concentration of 20 ng/l of 50 mM ammonium bicarbonate, pH 8.3, using a ProGest digestor (Genomic Solutions) following the manufacturer's instructions. Digestions were done with prior reduction with dithiothreitol (10 mM for 15 min at 65°C) and carbamidomethylation with iodoacetamide (50 mM for 60 min at room temperature). The tryptic peptide mixtures were freed from reagents using a C18 Zip-Tip pipette tip (Millipore) activated with 70% acetonitrile and equilibrated in 0.1% trifluoroacetic acid. Following protein adsorption and washing with 0.1% trifluoroacetic acid, the proteins were eluted with 3 l of 70% acetonitrile and 0.1% trifluoroacetic acid. For mass fingerprinting analysis, 0.85 l of the digests were spotted onto a MALDI-TOF sample holder, mixed with an equal volume of a saturated solution of ␣-cyano-4-hydroxycinnamic acid (Sigma) in 50% acetonitrile containing 0.1% trifluoroacetic acid, dried, and analyzed with an Applied Biosystems Voyager-DE Pro MALDI-TOF mass spectrometer, operated in delayed extraction and reflector modes, as above. The peptide mass fingerprint obtained from each electrophoretic band was compared with the expected proteolytic digest of the fusion protein using the program PAWS. 5 Collision-induced Dissociation by Tandem Mass Spectrometry-For peptide sequencing, the protein digest mixture was subjected to electrospray ionization tandem mass spectrometric (MS/MS) analysis using a QTrap mass spectrometer (Applied Biosystems) equipped with a nanoelectrospray source (Protana, Denmark). Doubly charged ions selected after Enhanced Resolution MS analysis were fragmented using the Enhanced Product Ion with the Q 0 trapping option at 250 atomic mass units/s across the entire mass range. For MS/MS experiments, Q 1 was operated at unit resolution, the Q 1 to Q 2 collision energy was set to 35 eV, the Q 3 entry barrier was 8 V, the linear ion trap Q 3 fill time was 250 ms, and the scan rate in Q 3 was 1000 atomic mass units/s. Collisioninduced dissociation spectra were interpreted manually or using the on-line form of the MASCOT program. 6 5 Proteometrics, available at prowl.rockefeller.edu. 6 www.matrixscience.com.

Quantitation of Free Cysteine Residues and Disulfide Bonds-For
quantitation of free cysteine residues and disulfide bonds (30), the purified proteins dissolved in 10 l of 50 mM HEPES, pH 9.0, 5 M guanidine hydrochloride containing 1 mM EDTA) were heat denatured at 85°C for 15 min, allowed to cool at room temperature, and incubated with either 10 mM iodoacetamide for 1 h at room temperature, or with 10 mM 1,4-dithioerythritol (Sigma) for 15 min at 80°C, followed by addition of iodoacetamide at 25 mM final concentration and incubation for 1 h at room temperature. Carbamidomethylated proteins were freed from reagents using a C18 Zip-Tip pipette tip (Millipore) after activation with 70% acetonitrile and equilibration in 0.1% trifluoroacetic acid. Following protein adsorption and washing with 0.1% trifluoroacetic acid, the PE-proteins were eluted onto the MALDI-TOF plate with 1 l of 70% acetonitrile and 0.1% trifluoroacetic acid and subjected to mass spectrometric analysis as above.
The number of free cysteine residues (N SH ) was determined using Equation 1, The regions of acostatin-␣ containing the three cysteine residues absent in jerdostatin, and the jerdostatin containing the short disintegrin-specific cysteine residue are boxed. C, cartoon depicting the proposed common ancestry of the messenger precursors coding for the short disintegrin jerdostatin and dimeric disintegrins. The proposed evolutionary pathway includes the removal of the metalloproteinase domain from a PII-metalloproteinase precursor gene. Key events in the emergence of jerdostatin appear to be the substitutions of the first three cysteine residues (Cys 6 , Cys 7 , and Cys 12 in the dimeric disintegrin subunit precursor) by His, Glu, and Asn, respectively, impairing thereby dimerization through either homologous Cys A7 -Cys B12 and Cys A12 -Cys B7 linkages, as reported for Schistatin (44), or Cys 7 -Cys 7 and Cys 12 -Cys 12 , as determined for EMF-10 (40); the appearance of a novel cysteine residue at position 101 (short-coding precursor numbering) between the 9 th and 10 th cysteine of the dimeric disintegrin subunit precursor (C) enabling the short disintegrin-specific disulfide bond is depicted by a broken line, and the proteolytic processing of the N-and C-terminal regions (scissors). The proposed disulfide bond pattern for jerdostatin is as determined for obtustatin (24). The two conserved disulfide bonds in the structures of dimeric disintegrin subunits and the short disintegrins are represented by thick lines. DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 where M IA is the mass of the denatured but nonreduced protein incubated in the presence of iodoacetamide; M NAT is the mass of the native, HPLC-isolated protein; and 57.05 is the mass increment because of the carbamidomethylation of one thiol group.

Cloning and Expression of the RTS-disintegrin Jerdostatin
The number of total cysteine residues (N Cys ) can be calculated from Equation 2, where M CM is the mass (in Da) of the reduced and carbamidomethylated protein.
Finally, the number of disulfide bonds N S-S can be calculated from Equation 3.
Cell Adhesion Studies-Adhesion studies of cultured cells labeled with 5-chloromethyl fluorescein diacetate were performed essentially as described (31,32). For inhibition studies, increasing concentrations of disintegrins were incubated for 30 min at 37°C with 1 ϫ 10 5 5-chloromethyl fluorescein diacetate-labeled cells in the wells of a 96-well enzyme-linked immunosorbent assay plate, previously covered with collagen IV (2 g/ml in 100 l of Hank's balanced salt solution containing 3 mM Mg 2ϩ ). After washing with the same buffer, the adhered cells were lysed with Triton X-100 and the plate was read using a FLx800 fluorescence plate reader. The percentage inhibition was calculated by comparison with the fluorescence values obtained from control samples without integrin inhibitors.
NMR Spectroscopy-For one-dimensional 1 H NMR analyses, each of the two HPLC fractions of wild-type r-jerdostatin was dissolved in 5% D 2 O, 95% H 2 O and placed in 5-mm Shigemi H 2 O/D 2 O susceptibility matched NMR tubes. Final concentrations of the samples were determined by UV-visible spectroscopy and were 2.4 and 1.4 mM for conformers HPLC-1 and HPLC-2, respectively. NMR spectra were recorded on a Bruker Avance NMR spectrometer operating at a 1 H frequency of 500.13 MHz and equipped with conventional BBI dual 1 H-Broadband probe. The spectra were processed on a SGI O 2 work station running the XWIN-NMR version 3.1 software. Processing included stages of apodization with a Gaussian function and zero filling to double number of points. Solvent signal suppression was achieved either by using the double WATERGATE pulse sequence (33,34) or by low-power irradiation at the water resonance frequency. A recycling relaxation delay of 1.5 segments between transients was employed in all experiments. Spectra recorded for both samples included one-dimensional double WATERGATE 1 H and one-dimensional 1 H NOE.

RESULTS AND DISCUSSION
The Structure of the Jerdostatin Open Reading Frame Provides Clues for Its Evolutionary Emergence-Jerdostatin represents a novel non-RGD short disintegrin encoded by a cDNA amplified from the venom gland mRNA of T. jerdonii by reverse transcriptase-PCR using primers complementary of the highly conserved 5Ј-and 3Ј-noncoding regions of other Trimeresurus disintegrins genes. The cDNA of jerdostatin comprised 369 bp (GenBank accession code AY262730) coding for an open reading frame of 333 bp including a signal sequence (1-20), a pre-peptide (21-68), and an obtustatin-like short disintegrin domain (residues 69 -110) (Fig. 1A).
Although the vast majority of disintegrins, including all known monomeric PIII, long, and medium-sized disintegrins, are derived by proteolysis of a large mosaic metalloproteinase precursor (35), the ␣-subunit of the dimeric disintegrin acostatin from Agkistrodon contortrix contortrix venom has been reported to be coded for by a shortcoding mRNA (36) similar to the jerdostatins messenger (Fig. 1B). Noteworthy, the jerdostatin pre-peptide sequence encompasses a region that is a homolog of the N-terminal sequence of acostatin-␣ harboring the first 3 cysteine residues of the mature molecule. Current biochemical and genetic data support the view that the different groups of the disintegrin family evolved from a common ancestor and that structural diversification occurred through disulfide bond engineering (35). In line with this view, the jerdostatin cDNA structure reported here, which represents the first complete open reading frame of a short disintegrin, points to a mechanism for the emergence of jerdostatin from a shortcoding gene similar to that of acostatin-␣. Fig. 1C depicts a scheme of the proposed evolutionary pathway, which involves substitutions of the first three N-terminal cysteine residues, the appearance of the short disintegrin-specific cysteine at the C-terminal region (underlined), and proteolytic processing of the precursor molecule at the N-and C-terminal regions. It is worth noting that the known native -fold of short disintegrins adopt a slightly different disulfide bond pattern than that of the dimeric disintegrin chains (Fig. 1C), providing further possibilities for the evolution of the structure and function of this family of integrin antagonists.

Recombinant Expression of Two HPLC Conformers of Jerdostatin-
The deduced primary structure of jerdostatin exhibits 80 -85% amino acid sequence identity with the KTS-disintegrins lebestatin from M. lebetina, obtustatin from V. lebetina obtusa venom (11,21), and viperistatin isolated from the venom of V. palestinae (22) (Fig. 2). Noteworthy, the 7-9 different residues between jerdostatin and the KTS-disintegrins are segregated within the C-terminal half of the molecule, including the integrin-recognition loop and the C-terminal tail, two structural elements that form a continuous functional epitope in the three-dimensional structure of obtustatin (24,25). In particular, jerdostatin contains a novel RTS motif instead of the KTS tripeptide found in lebestatin, obtustatin, and viperistatin. The KTS motif has been shown to endow disintegrins with selective inhibitory activity of the in vitro adhesion of integrin ␣ 1 ␤ 1 to immobilized collagen IV (21) and of angiogenesis in vivo (11). To investigate the biological activity of jerdostatin, the wildtype disintegrin was expressed in E. coli Origami B cells as a jerdostatinthioredoxin-His 6 fusion protein. The Origami B cells are derived from a lacZY mutant of E. coli BL21 and provide mutations in both the thioredoxin reductase (trxB) and glutathione reductase (gor) genes, greatly enhancing disulfide bond formation in the bacterial cytoplasm (37). Induction of the expression of the recombinant fusion protein construct was independent of the addition of the (0.1 mM) Lac operon inducer isopropyl ␤-D-thiogalactosidase, and the recombinant fusion protein was produced in approximately equal amounts in the soluble and insoluble cell lysate fractions (Fig. 3, lanes c and d). The expression of the fusion protein in these two subcellular fractions was assessed by MALDI-TOF mass fingerprinting of in-gel tryptic digests followed by collision-induced dissociation of selected monoisotopic ions. In particular, the simultaneous presence of ions from thioredoxin (903.3 (2 ϩ ), MIAPILDEIADEYQGK, and 634.3 (2 ϩ ), LNIDQNPGTAPK) and jerdostatin (595.4 (2 ϩ ) LKPAGTTCWR, and 627.9 (2 ϩ ), TSVSSHYCTGR) demonstrated that the protein band corresponded to the expected recombinant fusion protein product.
Affinity chromatography on a HisTrap column of the enterokinase degradation mixture of the jerdostatin-thioredoxin-His 6 fusion protein yielded major (80%) non-bound and minor (20%) retarded fractions. Both protein fractions eluted at the same position from the Superdex Peptide size-exclusion column used to complete the purification protocol, exhibited distinct reverse-phase HPLC elution profile, and had the same amino acid sequence (AMDCTTGPCCRQCKLKP . . . ) (Fig. 2), MALDI-TOF native isotope-averaged molecular mass (4898.6 Da) (Fig.  3B), and tryptic peptide mass fingerprinting expected for reduced and carbamidomethylated r-jerdostatin. Sequence analysis of the tryptic peptides was done by collision-induced fragmentation tandem mass spectrometry and confirmed the MALDI-TOF mass fingerprint assignments. The final purification yields of the two jerdostatin isoforms, designated according to their elution order from the reverse-phase HPLC column as conformers-1 and -2 of wild-type r-jerdostatin (r-jerdostatin-R21), were about 2 and 0.5 mg, respectively, per liter of Origami B cell culture. The monoisotopic molecular masses of r-jerdostatin-R21 conformers 1 and 2, measured by nanoelectrospray ionization mass spectrometry, were both 4894.6 Ϯ 0.3 Da (Fig. 3B, inset), which matched accurately the calculated value for the cDNA-derived amino acid sequence of the disintegrin (Fig. 2) with fully oxidized cysteine residues (calculated mass 4894.8 Da). Furthermore, mass spectrometric analysis of the reduced and carbamidomethylated conformers 1 and 2 yielded the same isotope-averaged molecular mass of 5363.2 Da. Incubation of r-jerdostatin-R21 isoforms 1 and 2 with iodoacetamide under denaturing but nonreducing conditions did not change their molecular masses. Hence, the mass difference of 464.6 Da between the native and reduced and carbamidomethylated proteins clearly indicated that each r-jerdostatin isoform contained eight cysteine residues engaged in the formation of 4 disulfide bonds. Taken together, these data, along with the different behaviors of r-jerdostatin-R21 conformers 1 and 2 on reversephase HPLC suggested that the two isoforms may represent structural conformers of r-jerdostatin. The stronger binding of conformer 2 to the C18 matrix indicates that r-jerdostatin conformer 2 exposes more hydrophobic surface than conformer 1. We sought to investigate the possibility that alternative disulfide bond connectivities could account for the different chromatographic behaviors of the two r-jerdostatin-R21 conformers. Determination of their disulfide bond pattern(s) was not possible, however, because both proteins were resistant to enzy-  5 kDa). B, MALDI-TOF (linear mode) mass spectrum of purified HPLC conformer 1 of r-jerdostatin-R21. The inset shows the triply charged isotope cluster of r-jerdostatin-R21 determined by nanoelectrospray mass spectrometry. The same results were obtained with the HPLC conformer 2 of r-jerdostatin-R21. C, isotope-averaged molecular mass of purified HPLC conformer 1 of r-jerdostatin-K21 determined by MALDI-TOF mass spectrometry. The same result was obtained with the HPLC conformer 2 of r-jerdostatin-K21. DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 matic proteolysis and chemical degradation with oxalic acid. Degradation with oxalic acid has been previously used to assign disulfide bonds in echistatin (38) (short, RGD-containing disintegrin), bitistatin (39) (long RGD-disintegrin), and EMF10 (40) (dimeric disintegrin), but did not work with obtustatin, indicating that the KTS/RTS-disintegrins are unique among disintegrins regarding their unusual stable conformation.

Cloning and Expression of the RTS-disintegrin Jerdostatin
Conformers 1 and 2 Represent, Respectively, Native and Unfolded r-Jerdostatin-R21 Molecular Species-The folding status of conformers 1 and 2 of r-jerdostatin-R21 was assessed by one-dimensional 1 H NMR (41,42). NMR spectra of the two samples showed a good signal-to-noise ratio, and water signal line width in the spectra of both samples, even without water suppression, was very narrow, indicating a good field homogeneity. Spectra recorded using the double WATERGATE solvent signal suppression method are shown in Fig. 4. The narrow proton peaks along with good resonance dispersion of at least 50 different peaks in the amide and aromatic proton regions (Fig. 4A) clearly indicate that conformer 1 possesses a well folded structure. On the other hand, the NMR spectrum of conformer 2 shown in Fig. 4B displays low resonance dispersion and broad peaks, which are indications of an unfolded flexible protein. Spectral differences between both r-jerdostatin-R21 conformers are particularly dramatic in the methyl proton region, from 0.5 to 1.5 ppm, where the dispersed and narrow set of peaks of conformer 1 contrasts with the presence of a wide band at the 1 ppm position in conformer 2. It is also worth noting that the peak at 10.02 ppm, which belongs to the HN side chain of the single r-jerdostatin tryptophan residue at position 20 (Fig. 2), remains at the same position in the spectra of both r-jerdostatin conformers, suggesting that this proton is exposed to the solvent and does not participate in intra-protein interactions. However, the significant broadening experimented by this proton peak in the spectrum of conformer 2 (Fig. 4B) indicates a larger degree of flexibility of Trp 20 in the unfolded versus the folded (conformer 1) r-jerdostatin-R21 species.
Conformers 1 and 2 of r-Jerdostatin-R21 Exhibit the Same Integrin Inhibitory Specificity Although Distinct Potency-Conformers 1 and 2 of r-jerdostatin-R21 were screened against a panel of integrins using the same cell adhesion inhibition assays described for the KTS-disintegrins obtustatin (11,21) and viperistatin (22). Both r-jerdostatin conformers proved to be selective inhibitors of the binding of the ␣ 1 ␤ 1 integrin to collagen IV (TABLE ONE, Fig. 5), and none of them showed inhibitory activity toward other integrins such as (TABLE ONE). The restricted integrin specificity of conformers 1 and 2 was highlighted by the fact that neither r-jerdostatin-R21 species blocked the adhesion of ␣ 2 -K562 cells to collagen ligands (TABLE ONE), a functional featured shared by the KTS-disintegrins. However, the IC 50 of conformer 1 (180 nM) was, respectively, about 90, 900, and 2250 times less potent than obtustatin (IC 50 2 nM), lebestatin (IC 50 0.2 nM), and viperistatin (IC 50 0.08 nM), inhibiting the binding of cells expressing integrin ␣ 1 ␤ 1 to immobilized collagen IV. On the other hand, r-jerdostatin-R21 conformer 2 (IC 50 5950 nM) was 33 times less active than conformer 1 (Fig. 5, TABLE ONE). These functional data in conjunction with the one-dimensional 1 H NMR analysis of the two r-jerdostatin-R21 conformers discussed above and shown in Fig. 4 supports the view that conformer 1 has a native, fully active dis-

Cell
Integrin Ligand  integrin-fold, whereas conformer 2 may represent a non-native, activity compromised disintegrin molecule. The lower inhibitory activity of the RTS versus KTS-disintegrins suggests that the amino acid residues that differentiate jerdostatin from obtustatin/viperistatin may create a distinct chemical environment responsible for its decreased potency, although these differences do not affect the restricted inhibitory selectivity of jerdostatin toward integrin ␣ 1 ␤ 1 . Among them, Arg 21 , Val 24 , and Ser 25 belong to the integrin binding loop, Arg 32 and Glu 35 lay at the face opposite to the integrin binding loop, Ser 38 forms part of the hydrophobic core of the protein, and Pro 40 , Asn 42 , and Gly 43 reside in the C-terminal region of the molecule (Fig. 6). Mutations at positions 32, 35, and 38 may not significantly alter the conformation of the disintegrin, and may therefore represent neutral mutations. In line with this assumption, pairwise comparison of the amino acid sequences and inhibitory potency of the KTS-disintegrins shown in Fig. 2 indicate that substitution at positions 38 and 40 impair the potency of lebestatin versus viperistatin. An extra mutation R24L in obtustatin further decreases 1 order of magnitude the ␣ 1 ␤ 1 blocking activity of this disintegrin when compared with lebestatin. In agreement with this reasoning, comparison of the ␣ 1 ␤ 1 inhibitory activities of viperistatin and obtustatin, using synthetic peptides representing their integrin binding loops (viperistatin, 19 CWKTSRTSHYC 29 ; obtustatin, 19 CWKTSLTSHYC 29 ), showed that the 25-fold increased inhibitory activity of viperistatin over obtustatin was because of an Arg/Leu mutation at position 24 of the integrin binding loop and a Gln/Pro substitution at position 40 of the C-terminal region (22).
The possible contribution of integrin binding loop residues Arg 21 , Val 24 , and Ser 25 to the decreased ␣ 1 ␤ 1 inhibitory ability of r-jerdostatin-R21 versus its homologue KTS-disintegrins was assessed using 9 sets of positional-restricted combinatorial synthetic peptides. Each set contained 19 peptides representing the entire integrin-recognition loop of obtustatin but differing in the residue at a single position ( 19 CX 1 KTSLTSHYC 29 ; 19 CWX 2 TSLTSHYC 29 ; etc., where X n is an equimolar mixture of all amino acids except cysteine). Compared with an obtustatin control loop peptide, sets X 2 , X 3 , X 4 , and X 7 , exhibited about 5-fold enhanced activity, whereas X 1 , X 5 , X 6 , X 8 , and X 9 showed 2-5-fold decreased activity. Although the neat differences in activity were modest, probably because of compensatory effects, the results were recurrent and converged to indicate that most of the integrin binding loop positions may play a functional role either through direct interactions with the receptor, or indirectly by maintaining the active conformation of the loop, as shown for Thr 22 of obtustatin (25). According to the NMR structure of obtustatin, residues at positions X 1 , X 2 , X 4 , and X 5 , are surface-exposed amino acids of the integrin binding loop. Noteworthy, X 2 corresponds to Lys 21 , and thus the results indicating that substitutions at this position enhanced the ␣ 1 ␤ 1 inhibitory ability of the peptides suggests that KTS may be a suboptimal ␣ 1 ␤ 1 inhibitory motif. On the other hand, this result provided circumstantial evidence against a more potent inhibitory activity of KTS versus RTS. To further check this hypothesis, a single R21K mutant (r-jerdostatin-K21) was generated by site-directed mutagenesis, as described under "Experimental Procedures." The r-Jerdostatin-K21 Mutant Is a Weaker Inhibitor of Integrin ␣ 1 ␤ 1 Than Wild-type r-Jerdostatin-R21-Recombinant jerdostatin exhibiting the integrin-binding motif 21 KTS 23 instead of 21 RTS 23 , r-jerdostatin-K21, was expressed and purified as described for wild-type r-jerdostatin-R21. Similar to wild-type disintegrin, two protein fractions differing in their HisTrap and reverse-phase HPLC elution times, but displaying the same amino acid sequence and the expected molecular mass of the fully disulfide-bonded mutant protein (4870 Da) (Fig. 3C), were purified. Furthermore, both conformers of the Lys 21 mutant selectively impaired the adhesion of ␣ 1 -K562 cells to collagen IV (TABLE ONE), although conformer 1 was significantly more potent than conformer 2. Nonetheless, the r-jerdostatin-K21 conformers are weaker inhibitors than the homologous wild-type proteins (TABLE ONE), indicating that the KTS motif is a less potent antagonist of the integrin ␣ 1 ␤ 1 than RTS.
Concluding Remarks-Based on its structural and functional characteristics, we propose that jerdostatin belongs, together with obtustatin, viperistatin, and lebestatin to the novel class of short-sized ␣ 1 ␤ 1 -specific disintegrins. A distinct feature of jerdostatin is its novel 21 RTS 23 motif, which appears to represent a more potent inhibitor of integrin ␣ 1 ␤ 1   DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 than KTS. The fact that recombinant wild-type jerdostatin is less active than KTS-disintegrins obtustatin, viperistatin, and lebestatin isolated from their natural sources, suggests that amino acid residues of jerdostatin outside of the integrin binding motif and departing from the primary structures of the KTS-disintegrins may create a distinct chemical environment responsible for the lower inhibitory activity of jerdostatin, although these substitutions do not affect the restricted inhibitory selectivity of jerdostatin toward integrin ␣ 1 ␤ 1 . On the other hand, NMR studies have revealed that the integrin binding loop and the C-terminal tail of obtustatin (24,25) and echistatin (43) are structurally linked and display concerted motions in the 100 -300-ps time scale, strongly indicating that these two functional regions may form a conformational epitope engaged in extensive interactions with the target integrin receptor. Jerdostatin, like lebestatin, possesses serine and proline at positions 38 and 40, respectively, making it unlikely that these residues account for the decreased functional activity of the recombinant disintegrin. However, a distinct feature of r-jerdostatin is the presence of two C-terminal residues ( 42 NG 43 ), which in all venom-isolated members of the ␣ 1 ␤ 1 -specific short disintegrins are post-translationally removed. The structural and functional consequences of the lack of C-terminal processing deserve further detailed investigation.