Generation of Anti-complement “Prodrugs”

Expression of biologically active molecules as fusion proteins with antibody Fc can substantially extend the plasma half-life of the active agent but may also influence function. We have previously generated a number of fusion proteins comprising a complement regulator coupled to Fc and shown that the hybrid molecule has a long plasma half-life and retains biological activity. However, several of the fusion proteins generated had substantially reduced biological activity when compared with the native regulator or regulator released from the Fc following papain cleavage. We have taken advantage of this finding to engineer a prodrug with low complement regulatory activity that is cleaved at sites of inflammation to release active regulator. Two model prodrugs, comprising, respectively, the four short consensus repeats of human decay accelerating factor (CD55) linked to IgG4 Fc and the three NH2-terminal short consensus repeats of human decay accelerating factor linked to IgG2 Fc have been developed. In each, specific cleavage sites for matrix metalloproteinases and/or aggrecanases have been incorporated between the complement regulator and the Fc. These prodrugs have markedly decreased complement inhibitory activity when compared with the parent regulator in vitro. Exposure of the prodrugs to the relevant enzymes, either purified, or in supernatants of cytokine-stimulated chondrocytes or in synovial fluid, efficiently cleaved the prodrug, releasing active regulator. Such agents, having negligible systemic effects but active at sites of inflammation, represent a paradigm for the next generation of anti-C therapeutics.

results in production of proinflammatory mediators (such as C3a and C5a) and formation of the cytolytic membrane attack complex (MAC). Host cells are protected from damage by proteins present on cell membranes, complement regulatory proteins (CReg). These proteins include complement receptor 1 (CR1; CD35), membrane cofactor protein (CD46), and decay accelerating factor (DAF; CD55), which act early in the C cascade to inactivate the amplification enzymes, and CD59, which acts late in the cascade and prevents MAC formation (1,2). The genes encoding the regulators of the activation cascades are linked on chromosome 1 and the proteins are structurally related, comprising a number of protein domains known as short consensus repeats (SCRs). These domains consist of ϳ60 amino acids, many of which are highly conserved, and are linked end-to-end in the CReg to give flexible, elongated structures (3,4). DAF and membrane cofactor protein comprise four SCR domains, whereas the most common isoform of CR1 has 30. Under normal circumstances the CReg are sufficient to protect cells from damage by homologous C. However, the active products that enable complement to perform its physiological roles can inappropriately target self-tissues; C-mediated inflammation and tissue destruction is an important drive to pathology in diseases such as rheumatoid arthritis (RA), systemic lupus erythematosus, glomerulonephritis, multiple sclerosis, ischemia/reperfusion injury, and transplantation, where it acts to sustain the "vicious cycle" of inflammation and tissue damage. In RA, soluble products of C activation are present in the synovial fluid of affected joints and complement deposits are evident on synovial tissue (5)(6)(7). In addition, affected joints are full of leukocytes (neutrophils and T cells) attracted to the site by a gradient of C5a and other chemoattractants. Although C itself is not always the primary cause of disease in these inflammatory diseases, it acts to sustain the proinflammatory cycle and perpetuate tissue damage.
The involvement of C in the perpetuation and exacerbation of these disorders has driven the search for therapeutic reagents capable of inhibiting dysregulated C activation. Drugs mimicking the action of CReg have been effective at controlling pathological activation of complement in vivo in many animal models of disease (8,9). These reagents are frequently recombinant forms of membrane CReg that have been engineered without their membrane-anchoring domains as soluble proteins (10). Some of these reagents, such as soluble CR1 (sCR1), have progressed to trials in humans and have been beneficial in the treatment of acute inflammation, such as in adult respiratory distress syndrome (11). However, there are drawbacks associated with anti-C therapy, including rapid clearance of reagents in vivo, which can be within minutes of administration, and the side effects of long term systemic inhibition of the C system. Individuals genetically deficient in C components are prone to recurrent bacterial infections and immune complex disease, provoking caution in the use of anti-C agents. Such agents might be suitable for short term therapy, for example, in myocardial infarction or adult respiratory distress syndrome, but cannot be used for chronic treatment.
To circumvent the short half-life of CReg-based therapeutic reagents, we and others have previously fused CReg to antibody Fc domains, generating "antibody-like" molecules that have a much enhanced half-life in vivo and are effective in animal models of C-mediated disease (12,13). We demonstrated that fusion of DAF to antibody domains increased the half-life in rats from 20 min to 33 h. However, functional analysis of two reagents, rat DAF-Ig and rat CD59-Ig, indicated that the CReg function was substantially impaired in the fusion protein. Comparison of activities (moles of CReg) indicated that DAF-Ig had a 10-fold reduction in activity, whereas CD59-Ig had a 35-fold reduction in activity (13). Insertion of "spacing" domains between the CReg and the antibody hinge partially restored function, which was totally restored if the Fc domains were removed with papain. These data implicated steric constraints in the reduction in function.
We have taken advantage of this phenomenon to generate anti-C prodrugs by deliberately fusing functional domains of human CReg to inflexible antibody hinge regions. Insertion of a short enzyme site between the CReg, illustrated here with DAF, and the Ig hinge generates a reagent that can be targeted by specific enzymes to release active DAF from the fusion protein in the relevant tissue. Our model prodrugs contain short regions from aggrecan, the major proteoglycan in cartilage that is degraded by aggrecanases and MMPs in arthritis (14). Aggrecan comprises two globular domains at the amino terminus (G 1 , G 2 ) separated by a polypeptide stretch of ϳ150 amino acids, termed the interglobular domain (IGD), followed by a long glycosaminoglycan attachment region that separates G 2 from a third globular domain at the COOH terminus, G 3 . Extensive proteolytic degradation of aggrecan is associated with cartilage destruction in arthritic diseases; two major sites of catabolism have been identified within the IGD of aggrecan. The major site of cleavage appears to be hydrolysis of the peptide bond Glu 373 -Ala 374 (human sequence enumeration (15)) generated by the activity of the aggrecanases (members of a disintegrin-like and metalloproteinase with thrombospondin motifs (ADAMTS) family, namely ADAMTS-4, -5, and -1) (16,17), whereas a secondary cleavage is generated by hydrolysis of the Asn 341 -Phe 342 bond by a number of MMPs and recently reported by ADAMTS-4. Various other minor sites of cleavage within the IGD have also been characterized (18,19). Cleavage of aggrecan is enhanced in inflammatory joint disease resulting in cartilage damage, indeed soluble products of aggrecan catabolism generated by aggrecanases and MMPs can be found in synovial fluid and cartilage extracts of patients with arthritis (14, 20 -22). In vitro culture systems have also demonstrated that a variety of inflammatory cytokines, such as IL-1␣ and TNF-␣ stimulate cleavage at these same sites (23)(24)(25)(26). The enzyme cleavage sites, and the minimal sequence lengths for recognition by target enzymes, have been previously documented (27)(28)(29)(30).
The prodrugs described here combine three important characteristics. First, the reagent has enhanced circulating half-life because of the antibody Fc domains; second, it is designed to have little or no systemic activity; and third, it is a targeted reagent, having potential to unleash powerful anti-C activity at sites that express an abundance of the target enzyme.

Materials
Chemicals and reagents were from Fisher (Loughborough, United Kingdom) or Sigma unless otherwise stated. All tissue culture reagents and plastics were from Invitrogen. pDR2⌬EF1␣ was a gift from Dr. I. Anegon (INSERM U437, Nantes, France) and has been described previously (31). Restriction enzymes were from Amersham Biosciences, T4 DNA ligase from Promega (Southampton, UK), dNTPs from Bioline (London, UK), and Vent DNA polymerase from New England Biolabs Ltd. (Hitchin, UK). Primers and other molecular biology reagents were from Invitrogen. Chinese hamster ovary cells were obtained from the European Collection of Animal Cell Cultures (ECACC, Salisbury, UK). All tissue culture reagents and plastics were from Invitrogen (Paisley, UK). Chinese hamster ovary cells were cultured in Ham's F-12, 50 units of penicillin/streptomycin, 1 g/ml amphotericin B, 2 mM glutamine, 1 mM sodium pyruvate, 4% fetal calf serum. Goat anti-human Fc-HRP was from Sigma, goat anti-mouse Ig-HRP was from Bio-Rad. Neoepitope antibodies BC-3 recognizing the new NH 2 terminus ARGSV . . . generated by aggrecanases, BC-4 recognizing the new COOH terminus . . . DIPEN, and BC-14 recognizing the new NH 2 terminus FFGVG . . . generated by MMPs and aggrecanases, and anti-human CD55 (HD1A) were produced in-house (24,26,33,34

Generation of Fusion Proteins
Construction of Ig Fusion Protein Expression Vectors-Total RNA was prepared from human peripheral blood mononuclear cells using Ultraspec total RNA isolation reagent (Biotecx Laboratories Inc., Houston, TX) and was reverse transcribed according to standard protocols using Superscript reverse transcriptase (Invitrogen) and oligo(dT) (CC-AGTGAGCAGAGTGACGAGGACTGGAGCTCAAGCT 17 ). DNA encoding the hinge, CH2, and CH3 regions of human IgG 2 or IgG 4 was amplified by PCR with primers (see below) that incorporated restriction sites (indicated in bold) enabling direct ligation into the expression vector pDR2⌬EF1␣. DNA encoding either three or four SCR of human DAF was amplified using plasmid template and primers that incorporated restriction sites (indicated in bold) enabling direct ligation into pDR2⌬EF1␣ immediately upstream of and in-frame with DNA encoding the Ig domains. Sequencing confirmed that no errors had been introduced by PCR. Primers for IgG and DAF were as follows: IgG 2 sense, 5Ј-CACGGATCCGAGCGCAAATGTTGTGTCG; IgG 4 sense, 5Ј-GTTGGATCCAAATATGGTCCCCCATG for DAF-IgG 4 and AACGGA-TCCGTGGACAAGAGAGTT for DAF4-IGD75-IgG 4 prodrug; IgG 2 and IgG 4 antisense, 5Ј-CGAGATATCGGGGAGCGGGGGCTTGC; DAF sense, GCTTCTAGACTAACCCGGCGCGCCATGACC; DAF4-IgG 4 antisense, TAGGGATCCTCCTCTGCATTCAGGTGGTGG; DAF4-IGD-75-IgG 4 prodrug antisense, TGGGGATCCCTTGGAAGTTAGAGATTT-TC; three SCR forms of DAF antisense TGCGGATCCATAAATTT-CTCTGCACTCTGG.
Incorporation of Enzyme Sites-To incorporate the short enzyme site derived from the aggrecan IGD ( Fig. 2) into the DAF (3 SCRs)-IgG 2 construct, complementary DNA oligomers were used that encoded the polypeptide flanked on both sides by DNA forming the restriction site for BamHI (sense primer, 5Ј-GCAGGATCCGGAGAAGACTTTGTAG-ATATTCCAGAAAATTTCTTCGGCGTCGGTGGAGAGGAGGACGGA-TCCACG, encoding amino acids GEDFVDIPENFFGVGGEED). Oligomers were annealed, digested with BamHI, and ligated into the vector encoding DAF3-IgG 2 at the BamHI site (located at 3Ј end of DAF, 5Ј end of IgG 2 ). To generate the larger construct, DAF4-IGD75-IgG4, DNA encoding 75 amino acids of aggrecan IGD sequence (sequence, GYTGEDFVDIPENF 342 FGVGGEEDITVQTVTWPDMELPLPRNITEG EA 374 RGSVILTVKPIFEVSPSPLEPEEPFTFAP) was amplified by PCR from plasmid template, primers encoded the BamHI restriction site at either end of the construct enabling ligation into the expression vector between DAF and the Ig hinge. The major aggrecanase and MMP cleavage sites are underlined. In all cases sequencing confirmed orientation and fidelity of DNA sequence.
Cell Transfection and Purification of Fusion Proteins-Chinese hamster ovary cells were transfected with expression vectors using Lipo-fectAMINE (Invitrogen) according to the manufacturer's instructions and stable lines were established by selection with 400 g/ml hygromycin B (Invitrogen). Tissue culture supernatant was collected and passed over a Prosep A column (Bioprocessing Ltd., Consett, UK) to purify the fusion protein. The column was washed with PBS and with 0.1 M citrate buffer, pH 5.0, to remove most contaminating bovine Ig and the fusion protein was eluted with 0.1 M glycine/HCl, pH 2.5. Eluted protein was neutralized with Tris, concentrated by ultrafiltration, and dialyzed into PBS. In some cases reagents were applied to a Superose 12 size exclusion column to exchange buffer to MMP digest buffer and to remove minor contaminants such as bovine Ig. Protein concentration was determined using BCA protein assay reagent (Pierce & Warriner, Chester, UK) according to manufacturer's instructions and using bovine serum albumin as a standard. Purity was assessed by SDS-PAGE and either Coomassie or silver stain (35).
Soluble DAF without Fc-Soluble forms of DAF comprising either the amino-terminal three or all four SCR domains from human DAF have been described previously (36 -38). In brief, DAF was expressed as a soluble protein in Pichia pastoris with a carboxyl-terminal oligohistidine tag to enable purification on nickel-nitrilotriacetic acid columns. The soluble forms of DAF produced had NMR spectra typical of SCRcontaining proteins and bound a large range of monoclonal CD55 antibodies (37).

Hemolysis Assay
Sheep E (TCS Microbiology, Claydon, UK) were sensitized by incubating 1 volume of 4% E (v/v) with 1 volume of 1/250 rabbit anti-sheep E (Amboceptor; Behring Diagnostics) for 20 min at 37°C. Sensitized cells were washed twice in GVB and resuspended to 2%. A 50-l aliquot was incubated with 50 l of normal human serum and 50 l of a dilution of DAF-Ig or control protein. All dilutions were made in GVB and serum was previously titered to yield between 50 and 70% lysis in the absence of inhibitor. Cells were incubated at 37°C for 30 min, pelletted, and 75 l of supernatant was removed for measurement of absorbance at 415 nm (hemoglobin released). Control incubations included cells incubated in buffer only (0%) or in 0.03% Nonidet P-40 (100%). Percent lysis was calculated as follows: percent lysis ϭ 100 ϫ (A 415 test sample Ϫ A 415 0% control)/(A 415 100% control Ϫ A 415 0% control). A non-regulatory fusion protein was used as a negative control protein and inhibitory activity of the test protein was calculated: percent inhibition ϭ 100 ϫ (% lysis with negative control protein Ϫ % lysis test sample)/% lysis with negative control protein (13). The IH 50 represents the concentration of test protein at which 50% inhibition was obtained. Where reagents had been incubated in MMP digest buffer and activity subsequently assessed by hemolysis assay, an equivalent volume of this buffer was added to all other incubations.

SDS-PAGE and Western Blot
Proteins were separated by SDS-PAGE using the Hoeffer Mighty Small Gel system (Hoeffer Scientific Instruments, Newcastle-under-Lyme, UK). Gels were electroblotted onto nitrocellulose that was then blocked for 30 min at room temperature in PBS, 5% nonfat dried milk and incubated overnight with primary antibody (BC-4, BC-3, and BC-14, all at 1/100 dilution tissue culture supernatant; HD1A anti-DAF at 10 g/ml) or polyclonal goat anti-human Fc-HRP (1/1000). Blots incubated with mouse monoclonal antibodies were then washed three times in PBS, 0.1% Tween 20 for 5 min each, followed by incubation with HRP-conjugated goat anti-mouse Ig at a dilution of 1:1000 in PBS, 5% milk for 1 h. Blots were then washed three times in PBS, 0.1% Tween 20 for 5 min each, followed by PBS, and were developed using a chemiluminescent substrate (SuperSignal, Pierce & Warriner) and Kodak x-ray film. Where samples were analyzed following culture with chondrocytes, an alkaline phosphatase-conjugated secondary antibody was used to detect primary antibody and blots were developed as described previously (39,40).

Cleavage of the Prodrug by MMPs and Aggrecanase
Reagents that were subjected to enzymatic cleavage by MMPs or aggrecanase were first buffer-exchanged into digest buffer on a Super-ose 12 column. Purified enzymes (MMP3, MMP8, or ADAMTS-4) were added to the final concentrations stated in the text and reagents were incubated at 37°C for the specified time. To analyze cleavage by native enzyme released from cytokine-stimulated chondrocytes, porcine chondrocytes were isolated from metacarpophangeal joints as described previously (39,40), embedded in agarose, and cultured with prodrug and various cytokines as stated under "Results." After 4 days, medium was removed, dialyzed against water, and lyophilized. Samples were reconstituted in SDS-PAGE sample loading buffer and proteins were separated on a 10% reducing SDS-PAGE gel. Proteins were transferred to nitrocellulose and immunoblotted with neoepitope antibodies as described above.
To analyze ex vivo cleavage by native enzyme, synovial fluid (SF) was collected aseptically from the knees of patients with rheumatoid arthritis. All patients gave written consent and had symptomatic knee effusions needing aspiration for clinical relief. These human SFs were preadsorbed on Prosep A prior to addition of the prodrug to deplete human IgG already present in the joint fluid. SF was also collected from knee joints of rats with antigen-induced arthritis (AIA), the disease was induced as described elsewhere (8) and fluid was aspirated from the knee joint 9 days following disease induction. The prodrug was added to the fluid at 0.33 (human SF) or 0.2 mg/ml (rat SF) and incubated for 16 h at 37°C. Fluid was incubated with 50 l of Prosep A (solid phase) to bind human Fc fragments, the pellet was washed once with PBS, and bound protein was eluted with HCl, pH 1.5. The eluate was lyophilized, reconstituted in SDS-PAGE loading buffer, and analyzed by Western blotting as described above, blots were probed with neoepitope antibodies reacting with the Fc portion of the cleaved molecule.
In vivo cleavage was analyzed in the joints of rats with AIA 9 days following disease induction, 200 g of the DAF4-IGD75-IgG 4 prodrug or an equivalent molar concentration of a reagent lacking an enzyme site (DAF3-IgG 2 ) was injected into the knee joint. Four hours later, fluid was aspirated, human Fc-containing fragments were harvested and analyzed as described above.

In Vitro Functional Analysis of Human DAF-Ig-
We have previously demonstrated that fusion of rat DAF or CD59 to human IgG 1 Fc domains reduces their C-inhibitory function (13). In that study, comparison of rat DAF-Ig to soluble DAF (no Fc) that had been generated in the same cell line demonstrated an approximate 10-fold increase in function when compared on a moles of DAF basis and that of CD59-Ig to soluble CD59 a 35-fold increase in regulatory function. To generate the most effective prodrug based on human CReg, we first determined which fusion protein best abrogated function of human DAF. We assessed the effect on function of both antibody isotype and the length of the DAF molecule incorporated into the fusion protein. The two most "inflexible" human antibody isotypes are documented as IgG 4 and IgG 2 (42), this is likely because of their short upper hinge regions, only three amino acids for IgG 2 (43,44). Fusion of DAF to these isotypes therefore had the best potential for abrogation of DAF function. Fusion proteins consisting of the amino-terminal three or four SCRs from DAF fused to human IgG 2 or IgG 4 Fc domains were engineered, expressed, and purified from tissue culture supernatant. SDS-PAGE and Coomassie stain analysis of these fusion proteins is shown in Fig. 1A. Both recombinant and native human IgG can exist as non-covalent and covalently associated forms (43,44), the half-form of DAF-IgG 4 is particularly prominent and can be seen here in lanes 3 and 4 (indicated by large arrows). Gel filtration studies indicate that these forms are indeed non-covalently associated dimeric forms of DAF-IgG (data not shown). These reagents were tested in hemolysis assays for their ability to inhibit human C (Fig. 1b). Their activities were compared with those of soluble forms of DAF (sDAF) comprised solely of the equivalent SCR domains, the amount of each protein required for 50% inhibition was calculated and relative activities were calculated based on the number of DAF moieties in the reagent (Table I). To obtain accurate concentrations, the mass of each protein was determined by MALDI-TOF mass spectrometry (Table I). Fusion proteins comprising all four SCR domains of DAF demonstrated an approximate 10-fold reduction in function. Fusion of three SCR to the human IgG 4 hinge also reduced activity by 10-fold, however, fusion of three SCR to the most inflexible human IgG 2 hinge generated a protein in which the functional activity was dramatically reduced by ϳ60-fold.
Generation of Prototype Prodrugs-We have designed and engineered various different prodrugs, differing in the number of DAF SCRs incorporated into the construct, the length of antibody upper hinge, and the length of target enzyme site(s) taken from within the IGD of cartilage aggrecan. All reagents incorporated polypeptide stretches that contained enzyme sites for aggrecanases and/or MMPs that are expressed at high levels at inflammatory sites, including inflamed joints of arthritic patients. We demonstrate here two prodrugs from a large pool of reagents. The first reagent, termed DAF4-IGD75-IgG 4 , comprised four SCRs from DAF joined to human IgG 4 with 75 amino acids of IGD sequence inserted. Whereas the long enzyme site restored some function to the fusion protein (data not shown), it had potential for cleavage by both MMPs and aggrecanases. The second reagent, termed DAF3-DIPEN-IgG 2 , comprised three SCRs of DAF fused to human IgG 2 through a 19-amino acid MMP cleavage site (Fig. 2). DAF3-IgG 2 had markedly restricted function (Table I; Fig. 1) and was therefore a good candidate for a prodrug; three SCRs of DAF regulated the classical pathway of C as efficiently as all four SCRs (Fig. 1). Western blot of the supernatant from cells transfected with either prodrug expression vector demonstrated the presence of both DAF and Fc in each fusion protein (Fig. 3). Both prodrugs were purified from culture supernatant by protein A affinity chromatography.
In Vitro Cleavage of Prodrugs-Incubation of the appropriate prodrug with MMP3, -8, or with ADAMTS-4 (aggrecanase-1) resulted in efficient release of DAF from the Fc (Fig. 4), product sizes were as expected for cleavage at the correct location in the linker. DAF4-IGD75-IgG 4 was cleaved at the major MMP site (Asn 341 -Phe 342 ) by MMP3 and MMP8, although cleavage at other sites is evident, because of the presence of a multiple MMP cleavage sites in the aggrecan IGD (18). ADAMTS-4 also efficiently cleaved the prodrug at the aggrecanase site (Glu 373 -Ala 374 ). The prodrug DAF3-DIPEN-IgG 2 was efficiently cleaved at the MMP site by MMP3, the aggrecanase site was not included in this reagent. Cleavage at these different sites resulted in differing lengths of IGD remaining attached to the DAF moiety of the Fc, resulting in the altered mobility of this protein by SDS-PAGE. Prodrugs incubated at 37°C for 24 h in the absence of enzyme, or prodrugs without enzyme sites but incubated with enzymes were not degraded or digested (not shown).
To confirm the specificity of cleavage of the prodrugs, antibodies that recognize neoepitopes of the newly formed amino and carboxyl termini following cleavage of aggrecan by MMPs or aggrecanases were used to detect cleavage following incubation of any of the illustrated prodrugs with target enzymes (Fig.  5) (24,33). Prodrugs were incubated with enzymes, Western blotted to nitrocellulose, and probed with antibodies BC-3, BC-4, or BC-14. BC-3 recognized the new amino terminus formed following cleavage at the major aggrecanase site, whereas BC-4 and BC-14 recognized neoepitopes formed following cleavage at the major MMP site (BC-4, new carboxyl terminus; BC-14, new amino terminus). Cleavage of DAF4-IGD75-IgG 4 at both sites by MMPs or at the aggrecanase site by ADAMTS-4 was demonstrated (panel a). MMP8 also cleaves the aggrecan IGD at the aggrecanase site, as illustrated in Fig.  5a as has been previously described (18,32). The DAF3-DIPEN-IgG 2 prodrug was cleaved by MMP3 at Asn 341 -Phe 342 as demonstrated by Western blot with BC-14 (Fig. 5, panel b). This prodrug was not cleaved in the absence of MMP3 and the parent molecule, without the 19-amino acid insert, was not cleaved by MMP3 (Fig. 5b).
Activation of Prodrug-Insertion of the enzyme site may act as a "spacing domain" and restore some function to the parent DAF-Ig fusion protein (13). When the 19-amino acid "DIPEN" site was inserted into DAF3-IgG 2 , a 3-fold gain of function compared with the parent molecule was evident (Fig. 6). De- FIG. 1. Purification of DAF-Ig fusion proteins. a, all four SCR of human DAF or the amino-terminal three SCR were fused to human IgG 2 Fc or IgG 4 Fc. Proteins were purified by affinity chromatography on protein A and subjected to SDS-PAGE on a non-reducing 7.5% gel (loading indicated above lanes), proteins were stained with Coomassie Blue. Small arrows indicate the intact molecule, large arrows indicate the half-forms of IgG 4 . b, inhibitory activities of the purified proteins and sDAF were assessed by hemolysis assay and the concentration required for 50% inhibition was determined sDAF (SCRs 1-4, q), sDAF (SCRs 1-3, OE), DAF4-IgG 2 (E), DAF4-IgG 4 (Ⅺ), DAF3-IgG 4 (‚), DAF3-IgG 2 (ࡗ).

TABLE I Activities of DAF-Ig
Activities of DAF-Ig fusion proteins were compared with that of sDAF (3 or 4 SCRs depending on the fusion protein). The molarity of each reagent was calculated from the mass, and correction was made for two DAF moieties in the fusion proteins. Each point is the mean Ϯ S.D. of triplicate determinations. spite this, DAF3-DIPEN-IgG 2 was still Ͼ20-fold less active than soluble DAF comprising three SCR only. Incubation of this prodrug with MMP3 released DAF from the prodrug and restored function to the DAF moiety in the reagent (Fig. 6). Incubation of the parent DAF3-IgG 2 with MMP3 or incubation of the prodrug at 37°C without MMP3 did not alter the function in the assay. Cleavage of Prodrugs by Enzyme Released from Cytokinestimulated Chondrocytes-To assess the cleavage of these two prodrugs by aggrecanases and MMPs synthesized and released by chondrocytes, agarose-chondrocyte cultures were established and treated with or without the cytokines IL-1␣, TNF-␣, or the chemical agent retinoic acid. All of these reagents are known stimulators of cartilage catabolism. IL-1␣ and TNF-␣ are known to up-regulate the production of the aggrecanases and some MMPs, whereas retinoic acid is known to up-regulate the production of the aggrecanases but not the MMPs (23)(24)(25)(26)40). Following incubation of chondrocytes with these reagents and the DAF4-IGD75-IgG 4 prodrug, the neoepitopes generated by the aggrecanases and MMPs were detected using BC-3 and FIG. 2. Schematic illustration of the anti-C prodrug. Aggrecan IGD is cleaved at various sites by MMPs and aggrecanases (A). A portion of the IGD containing the major MMP site was inserted between the three amino-terminal SCR of DAF and the IgG 2 antibody hinge (B). This generated a reagent with little Cregulatory activity until cleaved by MMPs at the site of inflammation, resulting in release of the active DAF moiety (C).

FIG. 3. Generation of DAF-Ig prodrug.
Supernatants from cells transfected with prodrug were analyzed by non-reducing SDS-PAGE on a 7.5% gel. Proteins were blotted to nitrocellulose and probed with anti-DAF (HD1A) and HRP-linked secondary antibody or HRPlinked anti-human Fc. Blots were developed using enhanced chemiluminescence and x-ray film.
BC-14, respectively, demonstrating that cleavage of the major aggrecanase site (BC-3; Fig. 7, a and b) and the major MMP site (BC-14; Fig. 7, c and d) had occurred in the prodrug. Chondrocytes cultured ex vivo can release some MMPs constitutively, this is evident in Fig. 7, c and d. To confirm the specific cleavage of the prodrug, two doses of IL-1␣ were used to stimulate the chondrocytes (Fig. 7, b and d), the data showed a dose-dependent increase in cleavage at both the aggrecanase and MMP site in the presence of IL-1␣. MMP-mediated cleavage of the DAF3-DIPEN-IgG 2 reagent, which incorporates the MMP site but not the aggrecanase site, was also demonstrated using BC-14 (Fig. 7e). In contrast, the DAF3-DIPEN-IgG 2 reagent was not cleaved when incubated with unstimulated chondrocytes (Fig. 7e), this may reflect biological variation of MMP expression and activation in the chondrocytes isolated from different animals. In addition, lack of cleavage of the smaller prodrug by MMPs could be because of increased steric hindrance at the hinge resulting in decreased accessibility for the enzyme.
Cleavage of Prodrug in SF ex Vivo and in Vivo-To assess cleavage and activation of the reagents by enzymes in inflammatory SF, fluid was aspirated from the knee joints of patients with RA and from rats with antigen-induced arthritis (9 days post-disease induction). The DAF4-IGD75-IgG 4 prodrug was added to the fluid, incubated at 37°C, and human Fc-containing fragments were purified. Analysis by Western blot using the anti-IGD neoepitope antibodies BC-14 (detecting cleavage at the major MMP site) and BC-3 (detecting cleavage at the aggrecanase site) demonstrated that enzymes in the human SF cleaved the prodrug at the MMP site with little or no evidence FIG. 4. Digestion of prodrug with MMPs and ADAMTS-4. DAF4-IGD75-IgG 4 was incubated at 60 g/ml with 4.5 g/ml either MMP8 or MMP3 (a), at 120 g/ml with 4.5 g/ml ADAMTS-4 (b), or DAF3-DIPEN-IgG 2 was incubated at 200 g/ml with 2 g/ml MMP3 (c). Samples were removed at specified times and analyzed on 11% reducing SDS-PAGE gels. Bands were visualized by silver staining. Identity of the DAF and Fc moieties was confirmed by Western blot with specific antibodies (not shown).
FIG. 5. Specificity of prodrug digestion. DAF4-IGD75-IgG 4 (50 g/ml) (a) or DAF3-DIPEN-IgG 2 (100 g/ml) (b) were digested with MMP3 (2 g/ml), MMP8 (3.7 g/ml), or ADAMTS-4 (5.8 g/ml), subjected to reducing SDS-PAGE, and Western blotted to nitrocellulose. Blots were probed with antibodies specific for neoepitopes formed following cleavage at the major MMP site Asn 341 -Phe 342 (BC-4, anti-IPEN; BC-14, anti-FFGV) or the aggrecanase site Glu 373 -Ala 374 (BC-3, anti-ARGS). Blots were developed using enhanced chemiluminescence and captured on x-ray film. of aggrecanase catabolism ( Fig. 8a; results are representative of four experiments in two individuals with RA). In contrast, enzymes in the rat SF cleaved efficiently at the aggrecanase site (Fig. 8b). Cleavage in situ was also assessed by injecting reagent directly into the knee joint. Cleavage of two different reagents was assessed: DAF4-IGD75-IgG 4 , which contained sites for both groups of enzymes, and DAF3-IgG 2 , which had no enzyme site. Four hours following injection of the reagent, fluid was harvested and human Fc-containing fragments were purified. Western blot analysis demonstrated efficient cleavage of the DAF4-IGD75-IgG 4 prodrug at the aggrecanase site correlating with the ex vivo data (Fig. 8b), no cleavage was detected at the MMP site (data not shown). DISCUSSION We describe here the generation of anti-C prodrugs that can be targeted to sites of tissue damage and inflammation. The concept derives from the observation that fusion of CReg to Ig domains can abrogate the function of the C-regulatory moiety, probably because of the inaccessibility of the active sites of the CReg to their large multimolecular ligands (13). Deliberate generation of greater steric constraints, by removing the nonessential COOH-terminal SCR domain from our chosen CReg, DAF, and by using the most inflexible antibody isotype, IgG 2 , as source of Fc, generated reagents with substantially reduced activity. Fusion of the three SCR from human DAF to Fc from IgG 2 , the most inflexible antibody isotype, resulted in a molecule with a 60-fold impairment of function ( Figs. 1 and 6). To target the reagent to a specific disease site and to enable its activation, we inserted short enzyme cleavage sites between DAF and the antibody hinge. These sites were derived from the IGD of aggrecan, the major proteoglycan in cartilage that is extensively proteolysed and damaged in arthritis by MMPs and aggrecanases, enzymes that are expressed at high levels in the inflamed joint. The prodrugs are designed to be cleaved by these enzymes at sites of inflammation and to release active DAF. We show here using two prodrugs, one designed for testing enzymatic cleavage efficiencies and the other for analyzing function, that exposure of the prodrugs to MMPs or aggrecanase, purified (Fig. 5), secreted from cytokine-stimulated chondrocytes (Fig. 7), or in the complex milieu of SF itself ( Fig. 8) results in specific cleavage. Specificity of cleavage is demonstrated by using anti-neoepitope antibodies that detect the new amino and carboxyl termini in the cleaved substrate. Cleavage by enzymes in SF is demonstrated ex vivo in both humans and rats and in vivo in rats. The predominant cleavage resulting from enzymes in human rheumatoid SF was at the MMP site and that in rat SF was at the aggrecanase site. This probably reflects the stage of disease, aggrecanase is reported to be present at high levels at the early stages of arthritis during proteoglycan depletion, whereas MMPs predominate at later stages when collagen becomes degraded and the cartilage becomes fibrillated exposing bone. The AIA rodent model of inflammatory arthritis mimics the early stage of cartilage degradation while the samples of human RA SF would reflect late stage disease (14,41). We also demonstrate that cleavage of the prodrug by the target enzymes restores C-regulatory function (Fig. 6). FIG. 6. Restoration of DAF function following cleavage of prodrug. The prodrug DAF3-DIPEN-IgG 2 was digested with MMP3 for 6 h at 37°C and its anti-complement function was analyzed by hemolysis assay (f). Control incubations were as follows: prodrug incubated without enzyme (OE), prodrug taken fresh from frozen storage (ϩ), the "parent" reagent (DAF3-IgG2) incubated with (q) or without (ϫ) MMP3, sDAF (SCRs 1-3) (ࡗ). Hemoglobin released from erythrocytes was taken as a measure of hemolysis, percent inhibition compared with a negative control protein was calculated and plotted against protein concentration (expressed as moles of DAF, logarithmic scale). Each point is the mean Ϯ S.D. of triplicate determinations. The DAF4-IGD75-IgG 4 prodrug contains a relatively long target polypeptide comprising several different enzyme sites, and retains substantial C regulatory ability. The DAF3-DIPEN-IgG 2 prodrug was engineered specifically as an anti-C prodrug and had very little anti-C activity, being at least 60fold less active than sDAF without the insertion of 19 amino acids of aggrecan IGD, and 20 -25-fold less active after inclusion of this site. These findings illustrate the importance of insertion of substrate domains that are as short as possible. We have demonstrated a similar effect previously, in that introduction of serine-glycine "linking" regions between other CReg, such as CD59, and antibody hinge regions can restore some C-regulatory activity (13). We chose a 19-amino acid stretch as a target for MMPs in our prodrug, however, others have demonstrated that MMPs can recognize even shorter polypeptide sequences (28), it may therefore be possible to truncate this MMP target site further still and to fully abrogate DAF function. The minimal polypeptide sequence required for substrate recognition by aggrecanse has also been analyzed. Carboxyland amino-terminal truncation of recombinant aggrecan IGD indicates that aggrecanases can cleave the substrate if 24 amino acids NH 2 -terminal to the primary cleavage site are present (29). Significantly, COOH-terminal truncation of the IGD actually increased enzyme activity in this study and only 13 amino acids were required for substrate recognition. Further truncation and "fine-tuning" of the aggrecanase enzyme site in our prodrugs will permit generation of reagents with a greater still reduction in systemic anti-C activity. It may also be possible to generate an even more efficient anti-C prodrug by utilizing a CReg other than DAF. Fusion of CD59, the ligand of which is the macromolecular MAC, to human IgG 2 may generate such steric hindrance to MAC binding that the reagent is totally systemically inactive, we have previously shown that rat CD59 fused to human IgG 1 Fc has very little Cregulatory activity (13). A CD59-containing prodrug will enable generation of a reagent that specifically targets the terminal stages of the C pathway.
The DAF3-DIPEN-IgG 2 prodrug fulfills many of the requirements for the ideal anti-C therapeutic. Although the half-life of this agent has not been tested in vivo, comparable work with other agents suggests that it will have a long circulating halflife because of the Ig domains (13); the prodrug has little or no systemic anti-C activity, leaving the C system intact to deal with infection and immune complex solubilization; activity is "targeted" for release at the inflammatory site. Our model prodrug is designed for an inflamed joint where MMPs and aggrecanases are highly expressed, however, these reagents may have therapeutic potential in other chronic diseases in which C is implicated in pathology. These may include arteriosclerosis, where levels of MMPs are also elevated. Indeed, the technology could equally be applied to other non-C reagents, such as anti-tumor toxins, as long as the reagent is inactive, or substantially inactivated when fused to antibody hinge regions. Similarly other enzymes expressed at specific sites may be utilized, providing they meet the following criteria: the inserted enzyme site should be short to avoid restoration of function; the enzyme must be generated specifically at the disease site; the enzyme must be capable of recognizing and cleaving a short "linear" polypeptide sequence. The principles of logical design of anti-C therapeutics illustrated here solve many of the problems associated with "first generation" agents and provide the real prospect of treating C-driven pathology even in chronic diseases.