Retroviral Vector-Mediated Expression of Hirudin by Human

Home , Hirudin

Gene Therapy (1999) 6, 385–392  1999 Stockton Press All rights reserved 0969-7128/99 $12.00 http://www.stockton-press.co.uk/gt Retroviral vector-mediated expression of hirudin by human vascular endothelial cells: implications for the design of retroviral vectors expressing biologically active proteins

JJ Rade1,2, M Cheung1, S Miyamoto1 and DA Dichek1,3 1Molecular Hematology Branch, National Heart, Lung, and Blood Institute, Bethesda, MD; 2Division of Cardiology, Johns Hopkins School of Medicine, Baltimore, MD; and 3Gladstone Institute of Cardiovascular Disease and Department of Medicine, University of California San Francisco, CA, USA

We constructed a hirudin cDNA cassette, HV-1.1, that and mass spectroscopic analysis revealed the presence of encodes mature hirudin variant-1 fused to the signal pep- an extra N-terminal serine residue, indicating aberrant tide of human tissue-type plasminogen activator (t-PA). cleavage of the t-PA signal peptide and likely accounting The cassette was subcloned into retroviral vectors and for the diminished activity. We therefore constructed a used to transduce human vascular endothelial cells in vitro. second cDNA cassette, HV-1.2, in which hirudin secretion Hirudin antigen and activity were measured by ELISA and was directed by the signal peptide of human growth hor- thrombin inhibition assays, respectively. Transduced cells mone. Hirudin expressed from the HV-1.2 cassette had a secreted up to 35 ± 2 ng/106 cells/24 h of biologically active specific activity of 13.5 ± 0.2 ATU/␮g. Protein sequencing hirudin; expression was stable for at least 7 weeks. and mass spectroscopic analysis demonstrated proper Recombinant hirudin, expressed from the HV-1.1 cassette, cleavage of the growth hormone signal peptide. Thus, we had a specific activity of 7.1 ± 0.2 antithrombin units per achieved high level retrovirus-mediated secretion of bio- microgram (ATU/␮g), compared with specific activities of logically active hirudin from endothelial cells in vitro. Use approximately 12 ATU/␮g for both native leech hirudin and of these vectors may permit sustained local antagonism of recombinant hirudin produced in yeast. Protein sequencing thrombin activity in vivo.

Keywords: gene therapy; endothelial cells; hirudin; signal peptide; thrombin

Introduction med in several animal model systems, have demonstrated that systemically administered hirudin can Autologous venous and prosthetic grafts are widely used prevent thrombosis and neointimal proliferation after as conduits for peripheral and coronary artery bypass arterial injury.7–10 Systemically administered hirudin has surgery. The utility of these conduits is limited, however, also been used in humans as an adjunct to thrombolytic by thrombosis and neointimal hyperplasia, both of which therapy and coronary angioplasty. Although some short- may result from generation of thrombin on the luminal term benefits of hirudin infusion have been reported, sys- graft surface. Thrombin activates platelets and cleaves temic hirudin administration has failed to prevent resten- 1 fibrinogen, leading to thrombosis and vessel occlusion. osis and has been associated in some settings with unac- Thrombin can also contribute to the development of graft ceptable bleeding complications.11–14 The hemorrhagic neointimal hyperplasia through its action as a smooth complications associated with systemic hirudin adminis- muscle cell mitogen, a mediator of inflammation, and tration may preclude the use of doses that are sufficient an agonist for the release of platelet-derived growth to achieve local thrombin inhibition at sites of vascular 2–5 factor. disease. Hirudin, a 65-amino acid protein secreted by the sali- We have previously demonstrated that adenovirus- vary glands of the medicinal leech, is a potent and spe- mediated gene transfer of hirudin in injured rat arteries 6 cific inhibitor of thrombin. By blocking both the catalytic results in hirudin secretion from the injured artery at a site and the anion-binding exosite of thrombin, hirudin level that is adequate to reduce neointimal formation.15 inhibits the interaction of thrombin both with fibrinogen Importantly, this effect was achieved without evidence of and with thrombin receptors that are present on the sur- systemic anticoagulation. Although adenovirus-mediated face of platelets and vascular cells. Recent studies, perfor- delivery of hirudin or other potentially therapeutic proteins to the vessel wall is relatively efficient, adenovirus- encoded transgene expression is transient and may be Correspondence: JJ Rade, Division of Cardiology, Johns Hopkins School of Medicine, Carnegie 565-C, 600 N Wolfe St, Baltimore, MD 21287- associated with a profound local inflammatory 16 6568, USA response. In addition, the high prevalence of pre-exist- Received 29 June 1998; accepted 24 September 1998 ing immunity to adenovirus in the human population Hirudin expression from transduced endothelium JJ Rade et al 386 may limit the ability of adenoviral vectors to deliver genetic material in a clinical setting.17 Retroviral vectors offer a promising alternative to adenoviral vectors for vascular gene therapy. Because retroviral vectors can insert genetic material into target cell genomic DNA, they have the potential to achieve stable long-term transgene expression. Moreover, many retroviral vectors are devoid of open reading frames encoding viral proteins. Elimination of these open reading frames should prevent destructive host inflammatory responses that are directed against foreign proteins expressed from transduced cells.18 Although achievement of efficient in vivo gene transfer to blood vessels has been a major problem with retroviral vectors,19,20 a recent report suggests that significant levels of in vivo retrovirus-mediated gene transfer may now be feasible.21 Alternatively, vascular endothelial cells transduced ex vivo with retroviral vectors might be introduced into the vasculature by seeding of these cells on to autologous vascular grafts or synthetic prostheses. We and others have demonstrated the feasibility of seeding retrovirally transduced endothelial cells on to prosthetic grafts, endo- vascular metal stents and native denuded arteries.22–24 As initial steps toward achievement of sustained local Figure 1 Retroviral vectors and cassettes expressing recombinant hirudin. (a) G HV-1.1 and G SHV-1.1 retroviral vectors containing the HV-1.1 thrombin inhibition via retroviral transduction of vascu- 1 1 cassette and G1HV-1.2 vector containing the HV-1.2 cassette. The vectors lar cells, we constructed retroviral vectors expressing consist of the Moloney murine leukemia virus 5Ј and 3Ј long terminal hirudin. We used vector backbones with and without repeats (LTR), non-coding regions (thin lines) containing the packaging internal splice donor and acceptor sites and constructed signal (⌿+), the HV-1.1 or HV-1.2 cassettes, the simian virus-40 early hirudin cDNA expression cassettes using two different promoter (SV40) and the neomycin phosphotransferase gene (NeoR) which heterologous signal peptide sequences to direct hirudin confers resistance to the neomycin analog G418. G1SHV-1.1 contains splice donor (SD) and splice acceptor (SA) sites, allowing synthesis of a secretion. Finally, the vectors were tested for their ability spliced mRNA encoding HV-1.1. (b) Hirudin cDNA cassettes, HV-1.1 to direct the secretion of intact, biologically active hirudin and HV-1.2, containing the coding sequence for mature hirudin variant- from cultured human endothelial cells. 1 (HV-1) downstream of the coding sequence for either the human t-PA signal peptide (t-PA Pre) or the human growth hormone signal peptide (hGH Pre). Also shown are the 5Ј and 3Ј untranslated sequences (thick Results black lines) and flanking sequences in the cloning vector (thin black lines) containing the restriction sites used for subcloning. Quantification of hirudin secretion from endothelial cells transduced with vectors expressing HV-1.1 Hirudin antigen and thrombin inhibitory activity were Table 1 Retrovirus-mediated hirudin expression by transduced measured in 72 h collections of conditioned medium har- endothelial cells

vested from untransduced human umbilical vein endo- 6 6 thelial cells (HUVEC) as well as from HUVEC transduced Vector ng/10 cells/24 h ATU/10 cells/24 h with the retroviral vectors G1HV-1.1, G1SHV-1.1 (Figure = ± ± G1HV-1.1 (n 6) 35 2 0.07 0.01* 1a), or the control vector G1XSVNa (not shown). = ± ± G1SHV-1.1 (n 6) 19 1 0.04 0.01 G1XSVNa lacks a hirudin expression cassette but is other- = ± ± G1HV-1.2 (n 4) 25 1 0.46 0.08* = ± wise identical to G1HV-1.1. Both the G1HV-1.1 and G1XSVNa (n 6) ND 0.03 0.01 = ± G1SHV-1.1 vectors contain the HV-1.1 hirudin expression None (n 6) ND 0.02 0.01 cassette, in which a synthetic cDNA for mature hirudin variant 1 is fused to the coding sequence for a human ATU indicates antithrombin units; ND, not detectable. Values tissue plasminogen activator (t-PA) signal peptide are expressed as mean ± s.e.m. Ͼ ± *P 0.02 compared with G1XSVNa. (Figure 1b). G1HV-1.1-transduced cells secreted 35 2 ng/106 cells/24 h of hirudin antigen, which was associated with a low level of thrombin inhibitory activity (Table 1). Secretion of hirudin antigen from G1SHV-1.1-trans- were analyzed by Western blotting (Figure 2). Native duced HUVEC was somewhat lower, demonstrating that hirudin can exist as oligomers at neutral pH;25 therefore, the existence of splice donor and acceptor sites in this we anticipated that several species would be identified construct did not increase protein expression levels. by the polyclonal anti-hirudin antibody. Several bands Thrombin inhibitory activity in the supernatant of were indeed present, with molecular masses correspond- G1SHV-1.1-transduced cells did not differ significantly ing to those of hirudin trimers, dimers and monomers. from controls. The pattern of bands detected in the lanes containing recHV-1.1 purified from the conditioned media of retro- Western blot analysis virus-transduced cells was similar to that present in lanes Recombinant hirudin (recHV-1.1) samples purified from containing either native leech hirudin or recombinant

the conditioned medium of G1SHV-1.1-transduced cells hirudin produced in yeast. Hirudin expression from transduced endothelium JJ Rade et al 387

Figure 3 Thrombin gel-shift Western blot analysis of purified hirudin species. Samples of purified hirudin were incubated with human ␣-thrombin for 5–10 min at room temperature. The thrombin (Ϯ hirudin) samples were then analyzed by 7.5% native polyacrylamide gel electrophoresis and Western blotting with antibodies to human ␣-thrombin. All lanes contain 2 NIH units of human ␣-thrombin with the indicated amounts of added hirudin: Lane 1, no hirudin. Lane 2, 1 ATU of native hirudin variant-1, from leeches (1 ATU of hirudin is defined by the ability to inhibit one NIH unit of thrombin). Lane 3, 1 ␮g recHV-1.1 purified from conditioned

medium of PA317 cells transduced with G1HV-1.1. Rec-hirudin shifts the mobility of all of the free thrombin, conﬁrming thrombin binding and suggesting a speciﬁc activity of у2 ATU/␮g.

Table 2 Biochemical characteristics of hirudin species

Figure 2 Western blot analysis of purified hirudin species. Samples of ␮ Hirudin species Specific activity Mass (AMU purified hirudin (1 g each) were subjected to 15% SDS-polyacrylamide Ϯ 0.1%) gel electrophoresis under reducing conditions, transferred to a nylon mem- ␮ brane, and detected by enhanced chemiluminescence using a polyclonal (ATU/ g) N-terminal rabbit anti-hirudin IgG antibody. Lane 1, native hirudin variant-1, from sequence leeches. Lane 2, recombinant hirudin (rec-hirudin), produced in yeast. Lane 3, recombinant hirudin (recHV-1.1) purified from the conditioned recHV-1.1 7.1 ± 0.2 SVVYT 7054.8 medium of HUVEC transduced with G1HV-1.1. Hirudin species are (retrovirus) identified as monomer, dimer and trimer based on their apparent molecu- recHV-1.2 13.5 ± 0.2 VVYTD 6969.9 lar mass. (retrovirus) native HV-1 (leech) 12.1 ± 0.1 VVYTD 6959.5 recHV-1 (yeast) 11.9 ± 0.2 VVYTD 6961.8 Thrombin gel shift Western blot analysis To characterize the ability of recHV-1.1 to bind thrombin, ATU indicates antithrombin units; AMU, atomic mass units; rec, purified samples from the conditioned media of G SHV- recombinant. N-terminal sequence is given in single letter amino 1 acid abbreviations. Specific activity values are expressed as the 1.1-transduced cells were incubated with purified human mean of quadruplicate measurements ± s.e.m. thrombin for 5–10 min at room temperature then subjected to native gel electrophoresis. During native gel electrophoresis proteins migrate according to both charge and mass, permitting differentiation of free thrombin parallel assays performed on commercially available from thrombin complexed to hirudin. Thrombin is a neu- native hirudin variant-1 obtained from leeches and trally charged molecule, with a major pI of 7.2–7.6.26 recombinant hirudin variant-1 produced in yeast. Pur- Hirudin has a strong negative charge (pI = 3.8–4.0); thus, ified recHV-1.1 had a specific activity of 7.1 ± 0.2 the thrombin–hirudin complex (which has an intermedi- ATU/␮g, approximately 60% of that measured for both ate pI of approximately 5)27 migrates faster than free native leech hirudin and recombinant hirudin from yeast. thrombin. RecHV-1.1 purified from the conditioned N-terminal amino acid sequence analysis of recombinant

media of retrovirus-transduced cells bound and shifted hirudin purified from conditioned medium of G1HV-1.1- thrombin in a fashion identical to that of native leech transduced cells revealed the unexpected presence of an hirudin (Figure 3). N-terminal serine residue. This additional residue appar- ently resulted from aberrant cleavage of the heterologous Biochemical and physical properties of recombinant t-PA signal peptide between Pro−2 and Ser−1 rather than hirudin the predicted cleavage site between Ser−1 and Val+1 To complete the characterization of recombinant hirudin, (Figure 4). Mass spectroscopy of the retrovirus-generated we measured the specific activity and performed N-ter- recHV-1.1 revealed a protein of 7054.8 atomic mass units minal amino acid sequence analysis and laser desorption (AMU), which is within the range of experimental error mass spectroscopy on hirudin samples purified from the for a hirudin variant-1 species lacking sulfation at Tyr+63 conditioned medium of retrovirus-transduced cells and containing an extra N-terminal serine residue (Table 2). Values were compared with those obtained in (average theoretical mass 7050.6 AMU). Hirudin expression from transduced endothelium JJ Rade et al 388

Figure 4 Protein sequence of the N-terminus of recHV-1.1 (a) and Figure 5 Duration of hirudin secretion from retrovirus-transduced endo- recHV-1.2 (b) as determined by translation of the synthetic cDNA thelial cells. HUVEC (passage 5) were transduced on day 0 with either sequence. The heterologous signal peptides (solid letters) are fused to hiru- ¼ ș G1HV-1.2 ( )orG1XSVNa ( ) and placed in selection with G418 on day din variant-1 (open letters). The predicted site of signal peptide cleavage 2. Before each determination, cells were passaged (up to a total of five −1 ϩ1 −1 ϩ1 is between Ser and Val for recHV-1.1 and between Ala and Val times), grown to confluence (or near-confluence), washed and placed in for recHV-1.2. The observed site of cleavage, as determined by N-terminal complete medium for 48 h. Hirudin antigen in the conditioned medium sequence analysis, is indicated for both proteins. Signal peptides are was determined by ELISA. Values are expressed as the mean of six experi- from tissue plasminogen activator (t-PA) and human growth hormone ments Ϯ s.e.m. (hGH). Amino acid sequences are indicated by single letter amino acid abbreviations. humans.29,30 To develop further the therapeutic potential of direct thrombin inhibition, we have sought to generate Because of the aberrant cleavage of the signal peptide viral vectors that can overexpress hirudin from vascular and the associated loss of antithrombin activity, we con- cells.15 In the present study we constructed and tested structed a second cDNA cassette (HV-1.2). This construct several retroviral vectors that express hirudin. The major contained a cDNA for mature hirudin variant-1 fused to findings of this study were: (1) retroviral vectors can the human growth hormone (hGH) signal peptide. The mediate the expression of biologically active hirudin G1HV-1.2 retroviral vector was generated as previously from human vascular endothelial cells; (2) retroviral described. Recombinant hirudin (recHV-1.2) purified transduction results in hirudin secretion that is stable for from the conditioned media of G1HV-1.2-transduced cells at least 7 weeks; and (3) heterologous signal peptides in had a specific activity of 13.5 ± 0.4 ATU/␮g, similar to retroviral vectors may be cleaved at unexpected sites. that of native leech hirudin and recombinant hirudin pro- The endogenous leech signal sequences that direct the duced in yeast. N-terminal amino acid analysis revealed secretion of native hirudins have never been reported. appropriate cleavage of the heterologous hGH signal Therefore, to achieve secretion of hirudin it was neces- peptide between Ala−1 and Val+1 (Figure 4). Mass spec- sary to select a heterologous signal peptide and fuse the troscopy of retrovirus-generated recHV-1.2 revealed a coding sequence for this peptide to the coding sequence protein of 6969.9 AMU, which is within the range of for mature hirudin. Recombinant hirudin has been experimental error for a hirudin variant-1 molecule lack- expressed previously in prokaryotes and lower eukary- ing sulfation at Tyr+63 (average theoretical mass 6963.5 otes, including E. coli, yeast and insect cells (using a bacu- AMU). lovirus vector).31–33 In each of these systems, hirudin ± HUVEC transduced with G1HV-1.2 secreted 25 1ng secretion was directed by heterologous signal peptides 6 of hirudin/10 cells/24h. Media from G1HV-1.2-trans- already known to function efficiently in these species. duced cells contained significant thrombin inhibitory Because our goal was to express hirudin in a mammalian activity (Table 1). Secretion of hirudin by transduced cell and because there is evidence that use of host cell HUVEC was stable for at least 7 weeks after transduction, species-specific signal peptides can result in enhanced then declined (Figure 5). Notably, at later time-points secretion and more reliable post-translational pro- (beyond 50 days), the general appearance of both trans- cessing,34,35 we initially used the human t-PA signal pep- duced and untransduced HUVEC was poor (ie large, tide to direct hirudin secretion from human endothelial vacuolated cells), and the cells no longer grew to conflu- cells. The t-PA signal peptide functions well in human ence. The decline in hirudin expression may thus be due endothelial cells36 and has also been used successfully as to cellular senescence. a heterologous signal peptide in other systems.37–39 Surprisingly, our efforts to express recombinant hiru- Discussion din using the t-PA signal peptide yielded a hirudin molecule (recHV-1.1)with an additional N-terminal serine Direct thrombin inhibitors have considerable promise as residue. Examination of the t-PA signal peptide sequence cardiovascular therapeutic agents. Direct thrombin inhi- suggested strongly that the presence of this serine was bition inhibits thrombosis and decreases vascular lesion due to aberrant signal peptide cleavage. This observed formation in animal models.8–10,28 Thrombin inhibition cleavage (between Pro−2 and Ser−1) was surprising potentiates thrombolysis and improves vessel patency in because cleavage at this site violated the ‘−3, −1 rule’ of Hirudin expression from transduced endothelium JJ Rade et al

40 389 von Heijne which predicted the site of signal peptide in vivo data generated with implanted, G1HV1.2-trans- cleavage to be between Ser−1 and Val+1. The aberrant duced baboon endothelial cells suggest that levels of cleavage may have been due to the absence of the 9- hirudin secretion that are attainable with current vectors amino acid t-PA propeptide, which is immediately down- may be adequate to inhibit both thrombosis and intimal stream of the signal peptide in the native t-PA precursor hyperplasia.49 Higher levels of hirudin production may molecule. At least two published studies have reported be even more effective. aberrant N-terminal cleavage consequent to deletion of We previously reported results of in vitro and in vivo propeptide sequences in mutant precursor proteins.41,42 investigations testing the hypothesis that seeding small- Consistent with this view, recombinant hirudin that we diameter prosthetic grafts with endothelial cells modified expressed using the hGH signal peptide (recHV-1.2) was by a retroviral vector to secrete t-PA would inhibit graft cleaved as predicted between Ala−1 and Val+1. The hGH thrombosis.50–54 Although the transduced cells secreted t- precursor protein does not contain a propeptide PA at physiologically meaningful levels, their adherence sequence;43 therefore, aberrant cleavage may be less to graft surfaces under flow conditions was significantly likely when the hGH signal sequence is used to direct decreased. Because adherence of the t-PA-transduced heterologous protein secretion. cells was improved in the presence of aprotinin, the poor Because the physical properties of hirudin can be dra- adherence of t-PA-transduced cells was likely due to matically affected by N-terminal amino acid additions, enhanced plasmin-mediated cleavage of extracellular the aberrant cleavage of the heterologous t-PA signal matrix and adhesion proteins. Expression of a therapeutic peptide is likely responsible for the decreased specific agent (such as hirudin) that lacks proteolytic activity activity of recHV-1.1. For example, experimental addition could represent a superior approach to graft seeding, as of a glycine or methionine residue to the N-terminus of hirudin-expressing cells have no apparent predisposition hirudin increases the thrombin inhibition constant (Ki)of to enhanced proteolysis; these cells could inhibit coagu- hirudin significantly,44 although thrombin binding is still lation without promoting degradation of their underly- = of high affinity (Ki 0.5–5.4 nm). The modestly decreased ing matrix. specific activity of recHV-1.1, approximately 60% that of Local thrombin inhibition, mediated by hirudin-trans- recHV-1.2, is therefore consistent with what would be duced endothelial cells, may eventually be useful not predicted to result from the addition of an N-terminal only in improving the performance of synthetic vascular serine residue to the hirudin molecule. An appreciation grafts, but also in preventing thrombosis and stenosis of of the biochemical complexities involved in heterologous native arteries and in preventing pulmonary thromboem- signal peptide processing is clearly essential in the design bolic disease. Further in vivo studies are now required to of hybrid expression cassettes for gene transfer and gene investigate the therapeutic potential of retrovirus-trans- therapy experiments. duced endothelial cells secreting hirudin. Retrovirus-transduced HUVEC expressed approximately 20–35 ng/106 cells/24 h of hirudin. This level of hirudin secretion is similar, on a molar basis, to levels of Materials and methods secretion of other proteins that have been expressed from endothelial cells by retroviral gene transfer.23,45 This com- Construction of hirudin cDNA cassettes parison suggests that our efforts to optimize hirudin A cDNA cassette, termed HV-1.1 (Figure 1b), coding for expression by using the viral LTR promoter and selecting mature hirudin variant-1 (HV-1) fused to the human t- transduced cells in G418 were reasonably successful. PA signal peptide and flanked by convenient restriction However, this rate of hirudin secretion is approximately sites, was constructed by a polymerase chain reaction one order of magnitude below the rate of hirudin (PCR)-based technique. The human t-PA signal sequence secretion we achieved using an adenoviral vector to was chosen as t-PA is a secreted product of human endo- express hirudin from HUVEC.15 Thus, endothelial cells thelial cells. Because of potential ambiguities regarding appear capable of producing and secreting a higher the cleavage site of the t-PA propeptide sequence,55 we amount of hirudin than was achieved in the present included only the signal or ‘pre’ peptide coding sequence study. Although differences in the promoters used in in the construct. Rules of signal peptide cleavage pre- these two vector systems preclude firm conclusions, the dicted that this ‘pre’ peptide would be cleaved immedi- superior performance of the adenoviral vector is likely ately N-terminal to the Val+1 residue of mature HV-1.40 due to the ability of this vector to transfer multiple trans- In the first PCR step, the two halves of the cassette, desig- gene copies to target cells,46 whereas our current proto- nated HV-1.1A (5Ј end) and HV-1.1B (3Ј end), were syn- cols for retroviral gene transfer achieve delivery of only thesized by seven cycles of PCR (denature, 94°C for 1 approximately one transgene copy per cell.47 This limi- min; anneal, 50°C for 2 min; extend, 72°C for 2 min) per- tation of retroviral gene transfer might be overcome by formed on two pairs of overlapping oligonucleotides. use of newer generation retroviral vectors, which can be HV-1.1A was synthesized from two 100-base oligonucleo- concentrated to higher titers and can effect delivery of tides: Sense: 5Ј CCGGAATTCC GGGCAGCGGC CGC multiple transgene copies per cell (up to 10–20) even in ATCATGG ATGCAATGAA GAGAGGGCTC TGCTGTG the absence of antibiotic selection, (Ref. 48, and Dr D Ory, TGC TGCTGCTGTG TGGAGCAGTC TTCGTTTCGC personal communication). Use of these vectors might per- CCAGCGTTGT 3Ј. Antisense: 5Ј GCGGATGCAT mit stable secretion of far higher levels of hirudin than TTGTTGCCCT GGCCGCAAAC GTTAGAGCCT TCGC are reported here. It is unclear what level of hirudin ACAGGC ACAGGTTCTG ACCGGATTCA GTGCAGT secretion will be required in order to achieve a thera- CAG TGTAAACAAC GCTGGGCGAA 3Ј. HV-1.1B was peutic effect in humans, and therefore we remain uncer- synthesized from two 75-base oligonucleotides: Sense: 5Ј tain as to whether retroviral vector-mediated expression TCCTGGGCTC TGACGGCGAA AAAAACCAGT GCGT of hirudin will be clinically useful. However, preliminary TACTGG CGAAGGTACC CCGAAACCGC AGTCTC 3Ј. Hirudin expression from transduced endothelium JJ Rade et al 390 Antisense: 5Ј CGCTCTAGAG CTATTACTGC AGGT medium containing 0.05 mg/ml G418 (Life Technologies, ATTCTT CCGGGATTTC TTCGAAGTCG CCGTCG Gaithersburg, MD, USA) for a period of at least 14 days TTGT GAGACTGCGG TTTCG 3Ј. In the second PCR before being used in experiments. This concentration of step, the HV-1.1A and HV-1.1B synthesis products were G418 killed 100% of untransduced cells within 10–14 amplified by 35 cycles of PCR (denature, 94°C for 1 min; days. anneal, 50°C for 2 min; extend, 72°C for 2 min) using 5Ј and 3Ј flanking oligonucleotides. The flanking primers Recombinant hirudin expression used for HV-1.1A amplification were as follows: Sense: Hirudin antigen was detected in cell-culture supernatants 5Ј CCGGAATTCCGGGCA 3Ј. Antisense: 5Ј GCGGAT by ELISA (Imubind Hirudin ELISA; American Diagnos- GCATTTGTT 3Ј. The flanking primers for the HV-1.1B tica, Greenwich, CT, USA). The limit of detection was amplification were: Sense: 5Ј TCCTGGGCTCTGACG 3Ј. routinely less than 0.2 ng/ml. Thrombin inhibitory Antisense: 5Ј CGCTCTAGAGCTATT 3Ј. The HV-1.1A activity in cell-culture supernatants was measured by a and HV-1.1B amplification products were gel purified, chromogenic substrate assay60 with the following modi- digested with NsiI and ligated. The crude ligation pro- fications: 100 ␮l of culture supernatant were added to ducts were used as templates to selectively amplify the 500 ␮l test reagent containing 0.0625–0.125 NIH U/ml full-length HV-1.1 cDNA cassette by 35 cycles of PCR human ␣-thrombin (Boehringer Mannheim, Indianapolis, (denature, 94°C for 1 min; anneal, 50°C for 2 min; extend, IN, USA), 200 mm Tris-HCl, pH 8.1, 840 mm urea, 25 mm 72°C for 2 min) using the following flanking primers: sodium chloride, 75 KIU/ml aprotinin (Boehringer) and Sense: 5Ј CCGGAATTCCGGGCA 3Ј. Antisense: 5Ј 100 ng/ml polybrene (Aldrich). The mixture was incu- CGCTCTAGAGCTATT 3Ј. Following gel purification the bated for 3–5 min at room temperature after which 200 ␮l amplified HV-1.1 cDNA was digested with EcoRI and were added to 25 ␮lof2mm D-phe-pip-arg-pNA (S-2238; XbaI and ligated into pBluescript IISK+ cloning vector KabiVitrum, Stockholm, Sweden). The absorbance at (Stratagene Cloning Systems, La Jolla, CA, USA) to gen- 405 nm was followed for 2 min using a Vmax Kinetic erate pBSHV-1.1. microplate reader (Molecular Devices, Menlo Park, CA, Due to experimentally determined aberrant cleavage of USA) and a Vmax (mOD/min) was measured. The the HV-1.1 heterologous t-PA signal peptide (see above), thrombin inhibitory activity, in antithrombin units per a second cDNA cassette termed HV-1.2 (Figure 1b) milliliter (ATU/ml), was calculated by comparison to was constructed by a PCR-based method described native hirudin variant-1 standards (native hirudin; Amer- previously.15 ican Diagnostica). All samples were within the linear range of the standard curve (0.031–0.5 ATU/ml). Because Construction of retroviral vectors the presence of heparin might confound interpretation of 56 The parent vector plasmid pG1XSVNa was generously the thrombin inhibition assay, heparin was omitted from supplied by Genetic Therapy (Gaithersburg, MD, USA). the culture medium that was used for these studies.

pG1HV-1.1 was constructed by ligating the HV-1.1 fragment, obtained by digesting pBSHV-1.1 with EcoRI and Purification of recombinant hirudin XbaI, into a similarly digested pG1XSVNa. The parent Recombinant hirudin was purified from media con- plasmid pG1samEN was a generous gift of Dr Richard A ditioned by PA317 retroviral producer cells transduced Morgan, Clinical Gene Therapy Branch, National Center with either pG1SHV-1.1 or pG1HV-1.2. After collection, for Human Genome Research, Bethesda, MD. The pG1sa- conditioned medium was centrifuged at 1000 g for 5 min mEN plasmid contains splice donor and acceptor sites to remove cellular debris and frozen at −20°C until use. that have been associated with increased expression of Hirudin was purified from conditioned medium in a two- 57 transgenes in retroviral vector constructs. pG1SHV-1.1 step procedure. In the first step, culture supernatants was constructed by ligating the HV-1.1 fragment, were applied to an A-25 DEAE Sephadex ion-exchange obtained by digesting pBSHV-1.1 with NotI, into a simi- column (Pharmacia LKB Biotechnology, Uppsala,

larly digested pG1samEN. pG1HV-1.2 was constructed by Sweden) equilibrated with 0.1 m ammonium acetate and ligating the HV-1.2 fragment, obtained by digesting 0.1 m sodium chloride, pH 6.0. Recombinant hirudin was pBSHV-1.2 with NotI, into a similarly digested eluted by a linear salt gradient 0.1–0.4 m sodium chloride, pG1XSVNa. Stocks of G1HV-1.1, G1SHV-1.1 and G1HV- 0.1 m ammonium acetate, pH 6.0. Fractions containing 1.2 (Figure 1a) amphotropic retroviral vectors were gen- hirudin antigen were pooled and further purified by erated by transfection of vector plasmids into a coculture FPLC (Pharmacia). Samples were loaded on to a ProRPC of GP+E8658 and PA31759 packaging cell lines. The titers HR 5/10 reversed phase column equilibrated with sol- of clonal retroviral stocks used in these experiments were vent A (0.1% triflouroacetic acid, water). Elution was per- typically greater than 1 × 105 G418-resistant colonies formed with a 0–50% gradient of solvent B (0.1% triflou- per milliliter. roacetic acid, acetonitrile) at 1 ml/min over 30 min. Fractions were assayed for hirudin antigen, pooled, and Cell culture and viral transduction concentrated by partial lyophilization before further Characterized human umbilical vein endothelial cells analysis. (HUVEC) from a single donor (Clonetics, San Diego, CA, USA; passages 2–10) were maintained at 37°C under 5% Western blot analysis

CO2 in complete EGM medium (Clonetics) containing 2% Samples were electrophoresed through a 15% SDS-poly- fetal calf serum (FCS). For retroviral transductions, sub- acrylamide gel using a Tricine SDS running buffer confluent HUVEC in 35-mm wells were incubated with (Novel, San Diego, CA, USA) essentially as described.61 1.5 ml of viral supernatant containing 8 ␮g/ml polybrene Samples were boiled for 5 min in loading buffer contain- (Aldrich Chemical, Milwaukee, WI, USA) for 6 h. Two ing 0.1 m dithiothreitol before electrophoresis. Proteins days later cells were placed into selection with complete were transferred to pre-wetted Immobilon-P membranes Hirudin expression from transduced endothelium JJ Rade et al 391 (Millipore, Bedford, MA, USA) using Towbin buffer62 sequence analysis was performed by William S Lane of without methanol in a Trans-Blot Electrophoretic Trans- the Harvard Microchemistry Facility (Cambridge, MA, fer Cell (Bio-Rad Laboratories, Hercules, CA, USA) USA). Samples were analyzed by automated Edman according to the manufacturer’s instructions. Pre-stained degradation using an Applied Biosystems Model 477A protein molecular weight standards (Life Technologies) Protein Sequencer equipped with an on-line 120 PTH were used to estimate molecular weights and to assess analyzer. transfer efficiency. After blocking overnight at 4°C with non-fat dry milk, membranes were incubated for 1 h Acknowledgements in blocking solution (Life Technologies) containing 1.0 ␮g/ml polyclonal rabbit anti-hirudin IgG antibody We thank Dr Henry Fales and Mr Edward Sokoloski for (5401; American Diagnostica) which recognizes the car- advice and technical assistance in obtaining the mass boxyl termini of several hirudin species. Membranes spectroscopy. We are also indebted to Dr Brian Safer for were washed in Tris-buffered saline with 0.5% Tween 20 his always thoughtful advice and assistance during pro- (TBST) and incubated for 1 h in blocking solution con- tein purification. Dr Dichek is an Established Investigator taining 1/4500 dilution of goat anti-rabbit biotinylated of the American Heart Association. This work was sup- antibody (Life Technologies). Following washing, mem- ported in part by grant HL53222 and by the Division of branes were incubated for 30 min in TBST containing Intramural Research of the National Heart, Lung and 1/2000 dilution of streptavidin–horseradish peroxidase Blood Institute. conjugate (Life Technologies). After extensive washing with TBST, hirudin species were detected by enhanced References chemiluminescence (ECL Western Blotting Detection Reagents; Amersham International, Amersham, UK) 1 Mann KG. Prothrombin and thrombin. In: Colman RW, Hirsh J, Marder VJ, Salzman EW (eds). Hemostasis and Thrombosis: Basic according to the manufacturer’s instruction. Principles and Clinical Practice, 3rd edn. JB Lippincott: Philadel- phia, 1994, pp 184–199. Thrombin gel shift Western blot analysis 2 Sugama Y et al. Thrombin-induced expression of endothelial P- Human ␣-thrombin (Boehringer) was incubated at room selectin and intercellular adhesion molecule-1: a mechanism for temperature for 5–10 min with either commercially avail- stabilizing neutrophil adhesion. J Cell Biol 1992; 119: 935–944. able native leech hirudin (American Diagnostica) or pur- 3 Harker LA, Hanson SR, Runge MS. Thrombin hypothesis of ified recombinant hirudin from the conditioned medium thrombus generation and vascular lesion formation. Am J Cardiol 1995; 75: 12B–17B. of G1SHV-1.1-transduced PA317 producer cells. Mixtures were electrophoresed through a 7.5% native polyacrylam- 4 McNamara CA et al. Thrombin stimulates proliferation of cultured rat aortic smooth muscle cells by a proteolytically acti- ide gel and proteins were transferred to an Immobilon- vated receptor. J Clin Invest 1993; 91: 94–98. P membrane as described above. After blocking, mem- 5 Okazaki H, Majesky MW, Harker LA, Schwartz SM. Regulation branes were incubated for 1 h in blocking solution con- of platelet-derived growth factor ligand and receptor gene taining 1.0 ␮g/ml polyclonal rabbit IgG directed against expression by ␣-thrombin in vascular smooth muscle cells. Circ human thrombin (4702; American Diagnostica). The Res 1992; 71: 1285–1293. bound rabbit IgG was detected with biotinylated goat 6 Markwardt F. Hirudin as an inhibitor of thrombin. Meth anti-rabbit IgG, as described above. Enzymol 1970; 19: 924–932. 7 Talbot MD et al. The effects of recombinant desulphatohirudin Specific activity determination of purified hirudin on arterial thrombosis in rats. Haemostasis 1991; 21 (Suppl. 1): The activity of purified hirudin was determined by a 73–79. 8 Just M, Tripier D, Seiffge D. Antithrombotic effects of recombi- thrombin inhibition assay using S-2238 (Chromogenix, nant hirudin in different animal models. Haemostasis 1991; 21 Mo¨lndal, Sweden). Purified hirudin (50 ␮l of an approxi- ␮ (Suppl. 1): 80–87. mately 200 ng/ml solution were added to 850 l of 150 9 Brill-Edwards P et al. Prevention of thrombus growth by mm Tris-HCl, pH 8.3, and 50 ␮l of 4 NIH U/ml human antithrombin III-dependent and two direct thrombin inhibitors ␣-thrombin (American Diagnostica) in 150 mm Tris-HCl, in rabbits: implications for antithrombotic therapy. Thromb pH 8.3, with 1 mg/ml protease-free bovine serum albu- Haemost 1992; 68: 424–427. min (Boehringer). After equilibration at 37°C for 3–5 min, 10 Sarembock IJ et al. Effectiveness of recombinant desulphatohiru- 50 ␮lof2mm S-2238 were added and incubated at 37°C din in reducing restenosis after balloon angioplasty of athero- for 2 min. The reaction was stopped by the addition of sclerotic femoral arteries in rabbits. Circulation 1991; 84: 232–243. 11 Topol EJ et al. Recombinant hirudin for unstable angina pectoris. 100 ␮l of 50% acetic acid, and the absorbance at 405 nm A multicenter, randomized angiographic trial. Circulation 1994; was measured in a 1-ml cuvette. Absorbance was sub- 89: 1557–1566. tracted from values obtained with thrombin-only controls 12 Serruys PW et al. A comparison of hirudin with heparin in the and this value was converted to ATU using 1.25 prevention of restenosis after coronary angioplasty. New Engl J 63 A405/min/ATU. The amount of purified hirudin was Med 1995; 333: 757–763. determined by ELISA. 13 van den Bos AA et al. Safety and efficacy of recombinant hirudin (CGP 39 393) versus heparin in patients with stable angina Molecular mass and N-terminal sequence undergoing coronary angioplasty. Circulation 1993; 88: 2058– determinations 2066. The molecular mass of hirudin species was determined 14 The Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) IIa Investigators. Randomized trial of intra- by matrix-assisted laser desorption mass spectroscopy venous heparin versus recombinant hirudin for acute coronary using a Kompact MALDI III mass spectrometer (Kratos syndromes. Circulation 1994; 90: 1631–1637. Analytical, Manchester, UK) and 3,5-dimethoxy-4- 15 Rade JJ, Schulick AH, Virmani R, Dichek DA. Local adenoviral- hydroxycinnamic acid (Aldrich) as matrix according to mediated expression of recombinant hirudin reduces neointimal the manufacturer’s instruction. N-terminal amino acid formation after arterial injury. Nature Med 1996; 2: 293–298. Hirudin expression from transduced endothelium JJ Rade et al 392 16 Newman KD et al. Adenovirus-mediated gene transfer into nor- peptide sequence of human preproparathyroid hormone for sig- mal rabbit arteries results in prolonged vascular cell activation, nal sequence function. J Biol Chem 1988; 263: 19771–19777. inflammation, and neointimal hyperplasia. J Clin Invest 1995; 96: 42 Matsuoka K, Nakamura K. Propeptide of a precursor to a plant 2955–2965. vacuolar protein required for vacuolar targeting. Proc Natl Acad 17 Schulick AH et al. Established immunity precludes adenovirus- Sci USA 1991; 88: 834–838. mediated gene transfer in rat carotid arteries. Potential for 43 Goeddel DV et al. Direct expression in Escherichia coli of a DNA immunosuppression and vector engineering to overcome bar- sequence coding for human growth hormone. Nature 1979; 281: riers of immunity. J Clin Invest 1997; 99: 209–219. 544–548. 18 Yang Y et al. Cellular immunity to viral antigens limits E1- 44 Wallace A, Dennis S, Hofsteenge J, Stone SR. Contribution of the deleted adenoviruses for gene therapy. Proc Natl Acad Sci USA N-terminal region of hirudin to its interaction with thrombin. 1994; 91: 4407–4411. Biochemistry 1989; 28: 10079–10084. 19 Flugelman MY et al. Low level in vivo gene transfer into the 45 Zwiebel JA et al. High-level recombinant gene expression in arterial wall through a perforated balloon catheter. Circulation rabbit endothelial cells transduced by retroviral vectors. Science 1992; 85: 1110–1117. 1989; 243: 220–222. 20 Nabel EG. Gene therapy for cardiovascular disease. Circulation 46 Connelly S et al. In vivo gene delivery and expression of physio- 1995; 91: 541–548. logical levels of functional human factor VIII in mice. Hum Gene 21 Zhu NL et al. Downregulation of cyclin G1 expression by retro- Ther 1995; 6: 185–193. virus-mediated antisense gene transfer inhibits vascular smooth 47 Kahn ML, Lee SW, Dichek DA. Optimization of retroviral vec- muscle cell proliferation and neointima formation. Circulation tor–mediated gene transfer into endothelial cells in vitro. Circ 1997; 96: 628–635. Res 1992; 71: 1508–1517. 22 Wilson JM et al. Implantation of vascular grafts lined with gen- 48 Ory DS, Neugeboren BA, Mulligan RC. A stable human-derived etically modified endothelial cells. Science 1989; 244: 1344–1346. packaging cell line for production of high titer 23 Dichek DA et al. Seeding of intravascular stents with genetically retrovirus/vesicular stomatitis virus G pseudotypes. Proc Natl engineered endothelial cells. Circulation 1989; 80: 1347–1353. Acad Sci USA 1996; 93: 11400–11406. 24 Nabel EG et al. Recombinant gene expression in vivo within 49 Harker L et al. Reduction in vascular lesion formation (VLF) by endothelial cells of the arterial wall. Science 1989; 244: 1342–1344. hirudin secreted from retroviral-transduced confluent endo- 25 Konno S, Fenton JW II, Villanueva GB. Analysis of the second- thelial cells on vascular grafts in baboons. Thromb Haemost 1997; ary structure of hirudin and the mechanism of its interaction 77 (Suppl.): 585. with thrombin. Arch Biochem Biophys 1988; 267: 158–166. 50 Dichek DA, Nussbaum O, Degen SJF, Anderson WF. Enhance- 26 Walsmann P. Isolation and characterization of hirudin from ment of the fibrinolytic activity of sheep endothelial cells by Hirudo medicinalis. Semin Thromb Hemost 1991; 17: 83–87. retroviral vector-mediated gene transfer. Blood 1991; 77: 533–541. 27 Braun PJ. Binding properties of hirudin determined by gel fil- 51 Dichek DA, Lee SW, Nguyen NH. Characterization of recombi- tration and gel electrophoresis. Thromb Res 1990; 59: 657–662. nant plasminogen activator production by primate endothelial 28 Meyer BJ et al. Dissolution of mural thrombus by specific throm- cells transduced with retroviral vectors. Blood 1994; 84: 504–516. bin inhibition with r-hirudin. Circulation 1998; 97: 681–685. 52 Dichek DA et al. Enhanced in vivo antithrombotic effects of et al 29 White HD . Randomized, double-blind comparison of hiru- endothelial cells expressing recombinant plasminogen activators log versus heparin in patients receiving streptokinase and transduced with retroviral vectors. Circulation 1996; 93: 301–309. aspirin for acute myocardial infarction (HERO). Circulation 1997; 53 Dunn PF et al. Seeding of vascular grafts with genetically modi- 96: 2155–2161. fied endothelial cells. Secretion of recombinant T-PA results in 30 Chesebro JH. Direct thrombin inhibition superior to heparin decreased seeded cell retention in vitro and in vivo. Circulation during and after thrombolysis. Circulation 1997; 96: 2118–2120. 1996; 93: 1439–1446. 31 Bergmann C et al. Chemical synthesis and expression of a gene 54 Shayani V, Newman KD, Dichek DA. Optimization of recombi- coding for hirudin, the thrombin-specific inhibitor from the nant t-PA secretion from seeded vascular grafts. J Surg Res 1994; leech Hirudo medicinalis. Biol Chem Hoppe Seyler 1986; 367: 731– 57 740. : 495–504. 32 Meyhack B et al. Desulfatohirudin, a specific thrombin inhibitor: 55 Berg DT, Grinnell BW. Signal and propeptide processing of expression and secretion in yeast. Thromb Res 1987; 48 (Suppl. human tissue plasminogen activator: activity of a pro-t-PA VII): 33. derivative. Biochem Biophys Res Commun 1991; 179: 1289–1296. 33 Benatti L, Scacheri E, Bishop DHL, Sarmientos P. Secretion of 56 McLachlin JR et al. Factors affecting retroviral vector function biologically active leech hirudin from baculovirus-infected and structural integrity. Virology 1993; 195: 1–5. insect eggs. Gene 1991; 101: 225–260. 57 Krall WJ et al. Increased levels of spliced RNA account for aug- 34 Bird P, Gething M-J, Sambrook J. Translocation in yeast and mented expression from the MFG retroviral vector in hematopo- mammalian cells: not all signal sequences are functionally equi- ietic cells. Gene Therapy 1996; 3: 37–48. valent. J Cell Biol 1987; 105: 2905–2914. 58 Markowitz D, Goff S, Bank A. A safe packaging line for gene 35 Tessier DC et al. Enhanced secretion from insect cells of a foreign transfer: separating viral genes on two different plasmids. J Virol protein fused to the honeybee melittin signal peptide. Gene 1991; 1988; 62: 1120–1124. 98: 177–183. 59 Miller AD, Buttimore C. Redesign of retrovirus packaging cell 36 Levin EG. Latent tissue plasminogen activator produced by lines to avoid recombination leading to helper virus production. human endothelial cells in culture: evidence for an enzyme– Mol Cell Biol 1986; 6: 2895–2902. inhibitor complex. Proc Natl Acad Sci USA 1983; 80: 6804–6808. 60 Spannagl M, Bichler H, Lill H, Schramm W. A fast photometric 37 van Zonneveld A-J, Veerman H, Pannekoek H. Autonomous assay for the determination of hirudin. Haemostasis 1991; 21 functions of structural domains on human tissue-type plasmin- (Suppl. 1): 36–40. ogen activator. Proc Natl Acad Sci USA 1986; 83: 4670–4674. 61 Laemmli UK. Cleavage of structural proteins during the 38 Devlin JJ et al. Novel expression of chimeric plasminogen acti- assembly of the head of bacteriophage T4. Nature 1970; 227: vators in insect cells. Biotechnology 1989; 7: 286–292. 680–685. 39 Kopfler WP et al. Adenovirus-mediated transfer of a gene enco- 62 Towbin H, Staehelin T, Gordon J. Electrophoretic transfer of ding human apolipoprotein A-I into normal mice increases cir- proteins from polyacrylamide gels to nitrocellulose sheets: pro- culating high-density lipoprotein cholesterol. Circulation 1994; cedure and some applications. Proc Natl Acad Sci USA 1979; 76: 90: 1319–1327. 4350–4354. 40 von Heijne G. Patterns of amino acids near signal-sequence 63 Johnson PH et al. Structure-function and refolding studies of the cleavage sites. Eur J Biochem 1983; 133: 17–21. thrombin-specific inhibitor hirudin. Haemostasis 1991; 21 (Suppl. 41 Wiren KW, Potts JT Jr, Kronenberg HM. Importance of the pro- 1): 41–48.