Targeting TNF-α with a tetravalent mini-antibody TNF-TeAb

Abstract

Anti-TNF-α [anti-(tumour necrosis factor-α)] therapy is widely considered to be among the most efficient treatments available for rheumatoid arthritis, psoriatic arthritis and inflammatory bowel disease. In the present study a tetravalent mini-antibody, named ‘TNF-TeAb’, was constructed by fusing the tetramerization domain of human p53 to the C-terminus of an anti-TNF-scFv [anti-(TNF-α–single-chain variable fragment)] via a long and flexible linking peptide derived from human serum albumin. TNF-TeAb was overexpressed as inclusion bodies in the cytoplasm of Escherichia coli, purified to homogeneity by immobilized- metal affinity chromtaography under denaturing conditions and produced in functional form by using an in vitro refolding system. In vitro bioactivity assays suggested that tetramerization of TNF-scFv resulted in an enormous gain in avidity, which endowed TNF-TeAb with a stronger ability to inhibit both receptor binding and cytolytic activity of TNF-α. TNF-α targeting therapy in rats with collagen-induced arthritis demonstrated that TNF-TeAb provided a much more significant therapeutic effect than did TNF-scFv in suppressing arthritis progression, attenuating inflammation and destruction in joints, and down-regulating pro-inflammatory cytokines and anti-(type II collagen) antibody. The conclusions are therefore (i) that multimerization of the antibody fragment by a self-association peptide is an efficient way to increase its avidity and (ii) that TNF-TeAb has potential applicability for anti-TNF-α therapy.

INTRODUCTION

TNF-α (tumour necrosis factor-α), originally identified by its ability to cause haemorrhagic necrosis of specific tumours [1], is now known to play an important role in promoting immunological, inflammatory and pathobiological reactions. Enhanced TNF-α synthesis is associated with the development of RA (rheumatoid arthritis), psoriatic arthritis and IBD (inflammatory bowel disease). Anti-TNF-α therapy is widely considered to be among the most efficient treatments available for such disease [2,3].

ScFvs (single-chain variable fragments) are the smallest functional units of the antibody molecules that still retain the complete antigen-binding site. Their small size and ability to be produced in Escherichia coli make them an interesting starting point for protein engineering, with potential for use in immunodiagnostics and therapy. Unfortunately, recombinant antibody fragments have exhibited poor in vivo targeting efficiency, probably due to their low binding avidity and fast clearance from the blood circulation, resulting in low total dose accumulation. The targeting efficacy of antibody fragments can be improved by the use of multimeric formats of antibody fragments with higher avidity and a molecular mass slightly above the renal-filtration threshold [4–6]. This has been confirmed in the studies in which Fab and scFvs were multimerized by chemical linkage [7–11]. A more efficient way to multimerize an antibody fragment is the modification of its polypeptide sequence by recombinant DNA technology and the subsequent purification of the multimeric protein from the bacterial host. One strategy is to shorten the flexible peptide linker of the scFv fragment so as to make the formation of a monomer impossible. These so-called ‘diabodies’ (bivalent diabodies) have been shown to be stable under in vivo conditions and to enrich efficiently at xenografts [12,13]. Another strategy is the fusion of heterologous or homologous self-assembling peptides or protein domains, such as leucine zippers, helix–loop–helix motif, streptavidin and the tetramerization domain of human p53, to the C-terminus of the scFv fragment [6,14–17], leading to the spontaneous assembly of the scFv fragment to the multimeric molecule directly in the cytoplasm of E. coli or in vitro refolding system. By using this approach for an anti-p185^HER−2 scFv fragment to form so-called mini-antibodies, very promising tumour-targeting data were reported [18]. However, it has not been documented so far whether such multimeric scFv fragments are applicable for inflammation targeting therapy.

In the present paper we report the engineering and production of a tetravalent mini-antibody formed by using the tetramerization domain of p53 as a multimerization module and its in vivo application for targeting TNF-α. The comparison of the uni- and tetra-valent formats is expected to provide insight into the feasibility of employment of such multivalent molecules for anti-TNF-α therapy and serves as a proof of this new therapeutic concept under in vivo conditions. Our results demonstrate that the tetravelant mini-antibody can be readily produced in functional form from E. coli inclusion bodies and that its avidity obtained by tetramerization is high enough to result in a satisfactory therapeutic effect for rat CIA (collagen-induced arthritis). The present study, therefore, provides a stable foundation for the development of a novel agent for anti-TNF-α therapy.

MATERIALS AND METHODS

Plasmid, strains and cells

Plasmid pTCM was constructed in our laboratory by cloning a DNA fragment encoding c-Myc tag into plasmid pTMF between the EcoRI and BamHI sites [19] and was used for protein expression. E. coli strain Top10 was used for cloning and maintaining plasmids throughout the experiments. E. coli strain BL21 (DE3) star was used as a host cell for protein expression. The murine fibrosarcoma cell line L929 was stored in our laboratory and, when required, cultured in RPMI-1640 medium containing 10% (v/v) FBS (foetal bovine serum; Hyclone) at 37°C in a 5%-CO₂ incubator.

Regents and materials

Anti-c-Myc tag mAb (monoclonal antibody) (9E10) and HRP (horseradish peroxidase)-conjugated polyclonal goat anti-mouse IgG were from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Anti-TNF-α mAb was from Peprotech (Rocky Hill, NJ, U.S.A.). CD3-scFv was constructed and stored in our laboratory. rhTNF-α [recombinant human TNF-α; 2×10⁷ units/mg (a unit of TNF-α is defined as the concentration (ng/ml) at which the activity is 50% of the maximum response in a standard TNF-α-dependent L929 cell killing)] and recombinant rat TNF-α were from PeproTech. Restriction endonuclease, T₄ DNA ligase, DNase, and RNase were from TaKaRa Biotechnology (Dalian, China). Pfu (Pyrococcus furiosus) DNA polymerase was a product of BioAsia (Shanghai, China). 2-Mercaptoethanol, Triton X-100, reduced and oxidized glutathione, bovine CII, and complete Freund's adjuvant were from Sigma Chemical Co. (St Louis, MO, U.S.A.). Actinomycin D was from Fluka Chemical Company (Buchs, Switzerland). Ni-NTA (Ni²⁺-nitrilotriacetate)–agarose was from Pharmacia Biotechnology Company (Piscataway, NJ, U.S.A.). The ELISA commercial kits for rat TNF-α and IL-β (interleukin-β) were from Biosource International (Camarillo, CA, U.S.A.). The anti-CII IgG commercial kit was from Chondrex (Redmond, WA, U.S.A.).

Animals

Male Lewis rats, 6–7 weeks old, with a mean bodyweight each of 175–200 g, were obtained from the Chinese Academy of Medical Sciences (Beijing, China). The animals were fed with standard rodent chow and water. The health status of the animals was monitored in accordance with the guidelines from the Chinese Veterinary Board. Rats were subdivided into the following groups: (1) normal control (n=5); (2) CIA plus saline (n=10); (3) CIA plus dexamethasone (0.1 mg/kg) (n=6); (4) CIA plus TNF-scFv (15 mg/kg) (n=6); (5) CIA plus TNF-TeAb (15 mg/kg) (n=6).

Cloning of the constructs

Plasmid pTS for the production of TNF-scFv was constructed previously by insertion of the TNF-scFv gene into plasmid pTCM via XhoI/EcoRI sites and maintained in E. coli strain Top10 [20]. The HSA (human serum albumin)–p53–c-Myc gene (Figure 1A below), comprising residues 490–513 of HSA [21,22], residues 319–360 of the human p53 [17,23] and a C-terminal c-Myc tag, was assembled by an overlapping PCR with nine synthetic oligonucleotides (Figure 1A and Table 1) according to a published method [24]. Plasmid pTTA for production of TNF-TeAb was constructed by insertion of the HSA–p53–c-Myc gene into pTS via EcoRI/BamHI sites (Figure 1B) and transformed into E. coli strain Top10.

Figure 1. Construction of the tetravalent mini-antibody TNF-TeAb.

(A) DNA and amino acid sequences of the HSA-p53-c-Myc gene. The HSA linker is shown in bold, the tetramerization domain of p53 is italicized and the c-Myc tag is underlined; P1–P9, primers for construction of HAS–p53–c-Myc gene. (B) Construction scheme of the tetravalent mini-antibody TNF-TeAb. Shown are the XhoI/BamHI cassettes of the expression vector pTCM for production of TNF-scFv and TNF-TeAb. Light-grey shading indicates the nucleotide sequence of a primer and the dark-grey shading is the overlapping region of two primers.

Table 1. Oligonucleotides used for the construction of the HSA–p53–c-Myc gene.

The underlined nucleotides in primers P1 and P9 indicate the sites of action of the endonucleases mentioned.

Primer	Sequence (from the 5′-end to the 3′-end)
P1	TTCGAATCCGCGCTGGAAGTGGA EcoRI
P2	GAATTCGCGCTGGAAGTGGATGAAACCTATGTGCCGAAAGAATTTAACGCGGAA
P3	TTTTTTGAGCTCAATATCCGCATGAAAGGTAAAGGTTTCCGCGTTAAATTCTTT
P4	GATATTGAGCTCAAAAAAAAACCGCTGGATGGCGAATATTTTACCCTGCAGATT
P5	CAGTTCGCGAAACATTTCAAAGCGTTCGCGGCCGCGAATCTGCAGGGTAAAATA
P6	GAAATGTTTCGCGAACTGAACGAAGCGCTGGAACTGAAAGATGCGCAGGCGGGC
P7	TTCTTCGCTAATCAGTTTCTGTTCGCCCGGTTCTTTGCCCGCCTGCGCATCTTT
P8	AAACTGATTAGCGAAGAAGATCTGAACGGATCC
P9	GGCGGATCCGTTCAGATCTTCTT BamHI

Expression and Western-blotting analysis of the constructs

The plasmids pTS and pTTA were transformed into E. coli strain BL21 (DE3) star for expression. The transformant cells harbouring pTS or pTTA were grown overnight at 37°C in 250 ml of Luria–Bertani broth (LB) medium supplemented with kanamycin (50 μg/ml). For the expression of either of the constructs on a large scale, 25 ml of this culture was then used for incubation of 1 litre of LB medium containing 1% glucose and kanamycin (50 μg/ml) in a 5-litre baffled shake flask. The culture was grown at 37°C, and expression was induced with 0.4 mM IPTG (isopropyl β-D-thiogalactopyranoside) [for the construct production when it reached an attenuance (D₆₀₀) of 0.6–0.8]. Expression was continued for 4–5 h at 37°C until the culture reached a final D₆₀₀ of 4–5. The cells from a 1-litre expression culture were harvested by centrifugation, resuspended in 250 ml of buffer containing 50 mM Tris/HCl (pH 8.0), 100 mM NaCl, 2 mM EDTA, 100 μg/ml lysozyme, 10 μg/ml DNase and 10 μg/ml RNase, and then thoroughly lysed by sonication (Ultrasonic Homogenizer 4710; Cole Parmer Instrument Company, Vernon Hills, IL, U.S.A.; 800 W, four bursts/min, one burst lasting 10 s) at 4°C for 30 min. The sonicated material was centrifuged subsequently at 14000 g at 4°C for 20 min, and the supernatant and pellet were analysed by SDS/PAGE to determine the expression characteristics of the construct. Immunodetection by Western-blotting analysis was performed according to the standard protocol [25]. Briefly, proteins produced in E. coli cells were subjected to SDS/PAGE, electrotransferred to a nitrocellulose membrane (Pall Gelman Sciences, Port Washington, NY, U.S.A.; 0.45 μm pore size), and incubated with blocking buffer at 25°C for 1 h. The membrane was incubated with mouse anti-c-Myc mAb (9E10) and then HRP-conjugated goat anti-mouse IgG. Finally, the proteins were identified by visualization with diaminobenzidine as the substrate.

Purification and refolding of the constructs

For purification of the constructs, the sonicated pellet obtained from a 1-litre expression culture was washed twice with a buffer containing 20 mM Tris/HCl, pH 8.0, 3% (w/v) Triton X-100 and 1 M urea, resuspended in 200 ml of denaturing buffer containing 20 mM Tris/HCl, pH 8.0, 8 M urea, 150 mM NaCl, 10 mM imidazole and 10 mM 2-mercaptoethanol, and stored at 25°C with gentle shaking to ensure that the inclusion bodies were entirely dissolved. The protein was then purified by IMAC (immobilized-metal affinity chromatography) under denaturing conditions. Briefly, the denatured mixture was centrifuged at 14000 g at 4°C for 20 min and the supernatant was then loaded on to a 10 ml Ni-NTA–agarose column that had been pre-equilibrated with the above denaturing buffer at a flow rate of 0.5 ml/min. After loading of the mixture, the column was washed with a buffer containing 20 mM Tris/HCl, pH 8.0, 8 M urea, 500 mM NaCl, 20 mM imidazole and 10 mM 2-mercaptoethanol at a flow rate of 1 ml/min until the absorption reached the baseline. The target protein was then eluted with a buffer containing 20 mM Tris/HCl, pH 8.0, 8 M urea, 150 mM NaCl, 250 mM imidazole and 10 mM 2-mercaptoethanol and checked for its purity by SDS/PAGE. For refolding of the constructs, the purified protein was diluted to a concentration of 250 μg/ml with the denaturing buffer in a 50 ml dialysis bag and subjected to a stepwise urea-removal dialysis (protein solution/buffer solution 1:50, v/v) against borate buffers (0.9% boracic acid/0.3% NaOH, pH 9.5) containing 4, 2, 1 or 0 M urea. Each dialysis step took 4 h and utilized a 0.5 M L-arginine/4 mM reduced glutathione/4 mM oxidized glutathione system as refolding additives at urea concentrations of 2, 1 or 0 M. After the stepwise dialysis, the protein solution was dialysed against a 100-fold volume of Tris/HCl, pH 8.0, buffer three times to remove the residual impurities, followed by centrifugation at 14000 g at 4°C for 30 min. The protein solution was then passed through a 0.22-μm-pore-size filter to remove potential aggregates and micro-organisms. Finally, the protein was concentrated by ultracentrifugation using Centricon micro-concentrators (Amersham Pharmacia Biotech, now Amersham Biosciences, Little Chalfont, Bucks., U.K.), quantified by the Bradford method using BSA as standard, and freeze-dried.

Analytical gel filtration

Analytical gel filtration of the constructs was performed on an Amersham Pharmacia Biotech HPLC system using a TSK-G3000-SW_XL column (7.8 mm diameter×300 mm long; Tosoh Bioscience, Tokyo, Japan) equilibrated with degassed PBS. Portions (20 μl) of the constructs were injected at a concentration of 200 μg/ml. For calibration of the TSK-G3000-SW_XL column, the standard proteins IgG (molecular mass 150 kDa), BSA (66 kDa) and carbonic anhydrase (29 kDa) were used. The molecular mass of the constructs was determined from their retention times.

Antigen-binding ELISA

The antigen-binding activity of the antibody constructs was detected by using a direct ELISA system similar to that described previously [26]. Briefly, 100 ng of rhTNF-α/well was bound to the surface of a 96-well flat-bottomed plate by incubation for 18 h at 4°C in 100 μl of 0.1 M carbonate buffer, pH 9.6. After washing and blocking, the plate was incubated with 100 μl of construct at 37°C for 2 h. CD3-scFv was used as negative control. The specifically bound construct was detected by using mouse anti-c-Myc mAb as the prime antibody and HRP-conjugated polyclonal goat anti-mouse IgG as the secondary antibody. HRP-conjugated IgG was then detected by addition of 100 μl of 1 mM OPD (o-phenylenediamine) solution containing 0.016%(v/v) H₂O₂. Colour development was assessed at 490 nm with an ELISA reader.

Receptor-binding inhibition

The ELISA plate was coated overnight at 4°C with 50 μl/well of 200 μg/ml TNF receptor, which was extracted from the L929 cells in accordance with the membrane antigen preparation method described previously [27]. The antibody construct was pre-incubated overnight with 100 ng of rhTNF-α in 100 μl of PBS buffer containing 2% BSA at 4°C. TNF-receptor-coated wells were washed with PBS containing 0.05% Tween 20, and 50 μl of construct/rhTNF-α mixed solution per well was added and the mixture incubated at 25°C for 2 h. The wells were washed and 50 μl of 1 μg/ml mouse anti-TNF-α mAb was added per well and the mixture incubated at 25°C for 1 h. The wells were washed and 50 μl of HRP-conjugated goat anti-mouse polyclonal IgG, at a dilution of 1:5000, was added per well and the mixture was incubated at 25°C for 1 h. After the final wash, 50 μl of OPD solution was added per well and the mixture was incubated in the dark at 25°C for 15–30 min. A 50 μl portion of 2 M H₂SO₄ was then added to each well to stop the reaction. The A₄₉₀ was read and used to calculate the percentage inhibition of rhTNF-α binding using the following equation:

CD₃-scFv was used as negative control in this assay.

Neutralization of TNF-dependent cytolytic activity

TNF-dependent L929 killing activity was quantified as previously described [26]. Neutralization of TNF-dependent cytolytic activity by the antibody constructs was quantified by use of a modification of the TNF bioassay. Briefly, L929 cell cultures were harvested in EDTA, washed, and resuspended to 2.5×10⁵ cells/ml in RPMI-1640 medium containing 10% FBS. A 100 μl portion of cell suspension was added per well to a 96-well assay plate and the plate was incubated overnight at 37°C under 5% CO₂ in a humidified incubator. A 100 μl portion of assay medium (RPMI-1640 medium containing 2% FBS) was added per well to another plate, and the construct was serially diluted in the 100 μl assay medium by 2-fold from row 3 to row 12, followed by addition of 50 μl of rhTNF-α per well at a final concentration of 1 ng/ml from row 2 to row 12. Row 1 was supplemented with 50 μl of assay medium to an identical volume. A 50 μl portion of actinomycin D solution per well was then added at a final concentration of 1 μg/ml. The plate was incubated at 37°C for 2 h, and 100 μl of this construct/rhTNF-α mixed solution was transferred to the corresponding wells of the assay plate with cell culture. The plate was then incubated for 24 h at 37°C under 5% CO₂ in a humidified incubator. After incubation, 10 μl/well of 5 mg/ml MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] solution was added and the mixture incubated at 37°C under 5% CO₂ for 4 h. This was followed by addition of 50 μl of MTT lysing solution, containing DMSO, per well and incubation at 25°C for 30 min. A₅₇₀ was then read and the percentage inhibition was calculated using the following equation:

where A_570,Ab/Ag is the absorbance of the wells with the addition of construct and TNF-α; A₅₇₀,_Ag is the absorbance of the wells with addition only of TNF-α; A₅₇₀,_N is the absorbance of the wells without addition of either the construct or TNF-α. CD3-scFv was used as negative control in this assay. To choose a species of animal for an in vivo bioactivity study, the recombinant rat TNF-α was used in this assay and the experimental protocol was the same as that of rhTNF-α.

Induction of arthritis

CIA was induced in rats as described previously [28], by multiple intradermal injections at the base of the tail and into three or five other sites on the back, of 250 μg of bovine CII (type II collagen) in 125 μl of 0.1 M acetic acid emulsified in an equal volume of Complete Freund's adjuvant containing 2 mg dry weight of Mycobacterium tuberculosis/ml. Rats were challenged again with the same dose of preparation 7 days later by an intraperitoneal injection. Disease developed about 11 days after the second immunization.

Treatment of rat CIA with the constructs

On the eleventh day after the second immunization, animals were randomized to receive the treatment listed in the ‘Animals’ subsection above, timed to coincide approximately with the onset of arthritis pathology. Dexamethasone, TNF-scFv and TNF-TeAb were each dissolved in physiological saline (0.9% NaCl) to a concentration of 0.1, 1 and 1 mg/ml. TNF-scFv and TNF-TeAb were administered by intraperitoneal injection of 15 mg/kg body weight at 3-day intervals and dexamethasone was orally administered daily with a dose of 0.1 mg/kg. The CIA control group received intraperitoneal injections of physiological saline, coinciding with the injection time of TNF-scFv and TNF-TeAb. The treatment period continued for 3 weeks (21 days).

Arthritis assessment

The severity of the arthritis was quantified at 3-day intervals by a clinical score measurement from 0 to 4 as follows: 0=normal; 1=erythema; 2=erythema plus slight swelling; 3=pronounced edematous swelling; 4=joint rigidity or deformation. Each limb was graded, giving a maximum score of 16 per rat. Clinical severity was also assessed by the quantification of hind-ankle swelling, which was obtained by measurement of the perimeter of the hind ankles with a dial-gauge caliper.

Histological analysis

Rats were killed at the end of the 21 days of therapy by diethyl ether narcosis; hind limbs were removed and fixed in 10% formalin buffer. The limbs were decalcified in 5% (v/v) nitric acid, processed for paraffin embedding, sectioned at 5 μm thickness, and subsequently stained with haematoxylin/eosin and Masson Trichrome for examination under an optical microscope. All joints were scored in a blind fashion for synovitis, bone erosions, cartilage damage and fibrous inflammatory hyperplasia using a predefined scoring system already reported [28]. Briefly, the severity of each kind of the four pathological changes was assessed using a semi-quantification method and classified as normal, mild, moderate or severe on the basis of the following criteria: 0=normal, no change; 1=mild, minimal pathological change limited to discrete foci; 2=moderate, pathological change present but normal joint architecture intact; 3=severe, severe pathological change and joint architecture disrupted. The histological score for each pathological change was calculated by multiplying its score by its weight (synovitis, bone erosions, and cartilage damage, ×1; fibrous inflammatory hyperplasia, ×3), on the basis of the change's importance in the pathology of rat CIA. The total histological score for each joint was obtained by addition of the histological score given for each of the four pathological changes.

Pro-inflammatory-cytokine assay

At the end of the experiment, samples of blood were drawn from a tail vessel and were collected in polyethylene tubes into which had previously been added 25 μl of heparin solution (3000 i.u.). The plasma samples obtained after centrifugation for 10 min at 3000 g and 4°C were stored at −80°C until assay. The plasma levels of TNF-α and IL-1β (interleukin 1β) were determined with a commercial ELISA kit (Rat TNF-α ELISA Kit, catalogue no. KRC0011; Rat IL-1β ELISA Kit, catalogue no. KRC3011; Biosource International) according to the supplier's protocol and are expressed as pg/ml.

Measurement of anti-type II collagen IgG

Circulating levels of anti-CII IgG were measured with a commercial ELISA kit (Arthrogen-CIA® ELISA; Chondrex) according to the supplier's protocol and expressed as units/ml (in the case of the Arthrogen-CIA® ELISA kit, 1 unit is approx. 1 ng of IgG antibody, corresponding to an A₄₉₀ value of 1 in the standard analysis).

Statistical analysis

Data are expressed as means±S.D. The difference between the means of two groups was evaluated with Student's t test and was considered significant at P<0.05.

RESULTS

Construction and expression of TNF-scFv and TNF-TeAb

The arrangement of the HSA–p53–c-Myc gene and its amino acid sequence are shown in Figure 1A. The tetramerization domain of human p53, comprising residues 319–360 of human p53 and capable of self-associating to a tetramer [23], was used as a multimerization device to tetramerize the scFv fragment. Oligomerization alone, however, does not automatically result in a dramatic increase in avidity of the self-assembled complex, because the four independent antigen-binding sites of the complex must be geometrically positioned to bind to separate antigen molecules simultaneously. The long and flexible HSA linker, forming a loop conformation and connecting helices 6 of domain IIIa and helices 1 of domain IIIb of HSA [21], was used to link the scFv fragment and the tetramerization domain of p53. The spatial separation of the scFv fragment from the tetramerization domain of p53 by the use of a flexible HSA linker was expected to provide for independent folding of the fused domains and, even more importantly, for a long and inflexible reach to bind to distinct antigens simultaneously. The construction scheme of TNF-scFv/pTS and TNF-TeAb/pTTA is shown in Figure 1(B). The HAS–p53–c-Myc gene assembled by an overlapping PCR with the nine synthesized primers (Table 1) was successfully fused to the C-terminus of the scFv fragment for the production of TNF-TeAb. Both the pTCM-derived vectors have a C-terminal His₆ tail to allow for purification of the constructs via IMAC. The pTCM-derived vector under the control of the T7 polymerase and Lac operator is a high-yield production system that allows the constructs to be overexpressed in the cytoplasm of E. coli. Both TNF-scFv and TNF-TeAb were expressed as inclusion bodies and accounted for approx. 40 and 30% of the sonicated pellet respectively, estimated from the gel-scanning quantitative system (Figure 2A). The target proteins were further identified by Western blotting (Figure 2B).

Figure 2. SDS/PAGE and Western-blotting analysis of the antibody constructs.

(A) SDS/PAGE analysis of the expression characteristics of the constructs. The arrows indicate the band of desired TNF-scFv (33 kDa) and TNF-TeAb (39 kDa). Lane M, the molecular-mass standards; lane 1, total proteins of E. coli cells harbouring pTCM; lane 2, supernatant of E. coli cells harbouring pTS; lane 3, pellet of E. coli cells harbouring pTS; lane 4; supernatant of E. coli cells harbouring pTTA; lane 5, pellet of E. coli cells harbouring pTTA. (B) Western blotting identification of the constructs. Lane 6, target protein TNF-scFv; lane 7, prestained molecular-mass standards; lane 8, target protein TNF-TeAb. Molecular masses in kDa are given on the extreme left and right.

Purification, refolding and gel filtration of TNF-scFv and TNF-TeAb

TNF-scFv and TNF-TeAb were routinely purified to greater than 95% purity by using IMAC under denaturing conditions (Figure 3A). After the refolding procedure we routinely obtained 14–16 mg of TNF-scFv from a 1 litre E. coli culture and, for TNF-TeAb, 10–12 mg/litre was obtained. On concentration by ultrafiltration, a concentration of at least 1 mg/ml could be obtained for both TNF-scFv and TNF-TeAb. The occurrence of multimerization of the constructs was demonstrated by gel-filtration analysis of the purified proteins on a TSK-G3000-SW_XL column. As shown in Figure 3(B), TNF-scFv was eluted at a retention time of 13.2 min, as expected for a monomeric species with a calculated molecular mass of about 33 kDa. TNF-TeAb was eluted at a retention time of 7.76, which corresponds to a calculated molecular mass of about 156 kDa consistent with a tetramer. No eluted peak corresponding to a monomer or dimer of TNF-TeAb was observed, which indicates that TNF-TeAb was homogeneous in the form of a tetramer. In no case were higher-molecular-mass aggregates detected, and the elution of single symmetric peaks indicated the homogeneity of the preparations.

Figure 3. Purification and gel-filtration analysis of the antibody constructs.

(A) Purity of TNF-scFv and TNF-TeAb. SDS/PAGE under reducing conditions shows the results of the purification of the constructs. Lane M, molecular-mass standards; lane 1, TNF-scFv; lane 2; TNF-TeAb. The molecular masses in kDa are given on the right of the gels. (B) Gel-filtration analysis of TNF-TeAb and TNF-scFv on a TSK-G3000-SW_XL column. The molecular-mass standards are as follows: IgG (150 kDa), BSA (66 kDa) and carbonic anhydrase (CA; 29 kDa).

In vitro bioactivity of TNF-scFv and TNF-TeAb

As shown in Figure 4(A), TNF-scFv and TNF-TeAb could specifically bind to rhTNF-α. TNF-α is cytotoxic to several murine and human cell lines expressing TNF receptors. Neutralizing antibody can protect TNF-sensitive cells from the cytotoxic effect by inhibiting binding of TNF-α to its receptors [29]. A solid ELISA assay was used to compare their ability to inhibit binding of TNF-α to its receptors. As shown in Figure 4(B), both TNF-scFv and TNF-TeAb produced a dose-dependent inhibition of binding of 1 μg/ml rhTNF-α to its receptors. A 50% level of inhibition was achieved at a TNF-scFv and TNF-TeAb input of approx. 0.5 μM and 60 nM respectively, and 100% inhibition was achieved at approx. 5 and 0.5 μM respectively. No inhibition was observed in the presence of control CD3-scFv. The constructs were also tested for their capacity to inhibit TNF-dependent L929 killing. As shown in Figure 4(C), the constructs inhibited 1 ng of cytolytic activity of rhTNF-α/ml in a dose-dependent manner. A level of 50% inhibition was obtained with approx. 16 and 2 nM of TNF-scFv and TNF-TeAb respectively, and 100% inhibition was obtained with approx. 256 and 32 nM respectively. No inhibition was observed in the presence of CD3-scFv. The cell morphology is shown in Figure 5. The data demonstrate that the avidity of TNF-TeAb increased enormously and revealed a stronger ability to inhibit both the receptor binding and cytolytic activity of TNF-α. As expected, the two constructs did neutralize the cytolytic activity of rat TNF-α. A 50% inhibition of 1 ng of activity of recombinant rat TNF-α/ml was achieved at a TNF-scFv input of approx. 64 nM or a TNF-TeAb input of about 8 nM (results not shown), which suggests that we can choose a disease model in rats to investigate the in vivo bioactivity of the constructs.

Figure 4. In vitro bioactivities of the antibody constructs.

(A) Antigen-binding ELISA of the antibody constructs. (B) Comparison of the ability of the constructs to inhibit receptor-binding of TNF-α. (C) Comparison of the ability of the constructs to neutralize the cytolytic activity of TNF-α.

Figure 5. Inhibition of rhTNF-α-induced cytolysis of L929 cells as shown by photomicroscopy.

Cells were incubated in the presence of: 256 nM TNF-scFv (A), 1 ng/ml rhTNF-α alone (B), 1 ng/ml rhTNF-α plus 16 nM TNF-scFv (C) or 1 ng/ml rhTNF-α plus 256 nM TNF-scFv (D). There was no difference: between (A) and the cells in medium alone (not shown) or the cells in 32 nM TNF-TeAb (not shown); between (C) and the cells in 1 ng/ml rhTNF-α plus 2 nM TNF-TeAb (not shown); between (D) and cells in 1 ng/ml rhTNF-α plus 32 nM TNF-TeAb (not shown).

Effect of TNF-scFv and TNF-TeAb on clinical signs of CIA

At 11 days after the second immunization, all the animals began to show the first signs of disease, predominantly in the hind paws and hind ankles. Anti-TNF-α treatment was carried out on this day, and the severity of arthritis was monitored by clinical score and observation of hind-ankle swelling. Dexamethasone, a commonly used anti-inflammation drug known to down-regulate the production of inflammatory cytokines such as TNF-α and IL-1β [30,31], was used as positive control in the treatment. Figure 6 shows the progression of arthritis during the treatment period. The disease developed with joint recruitment following the same pattern: tarsal, metatarsophalangeal and interphalangeal. The interphalangeal joints were never solely involved, and inflammation in these joints was invariably associated with inflammation in the tarsal joint. The mean clinical score of the arthritis in the saline-treated group was progressive and achieved a value of 13 in the last day. The same variations were also observed in the hind-ankle swelling of CIA rats given saline. In comparison with the saline-treated group, a significant reduction in clinical score in the TNF-TeAb-treated and the dexamethasone-treated groups was apparent throughout most of the treatment period, and a reduction in hind-ankle swelling was also observed in these two groups that reached statistical significance. TNF-scFv treatment, however, showed a lower level of suppression of arthritis progression, and significant differences in both the clinical score and the hind-ankle swelling only appeared towards the end of therapy. A significant difference was also observed in both clinical score and hind-ankle swelling between the two construct-treated groups throughout the latter half of the treatment period, which indicates that TNF-TeAb is more effective in suppressing the arthritis progression than is TNF-scFv.

Figure 6. Therapeutic effect of TNF-scFv and TNF-TeAb in established CIA.

(A) Arthritis severity scores in rats during a 21-day period of treatment. (B) Changes in hind-ankle swelling in rats during 21 days treatment. Day 1 corresponds to the first day that clinical arthritis was observed, and the antibody constructs were administered on days 1, 4, 7, 10, 13, 16 and 19. Results are means±S.D. *P<0.05 versus the group treated with saline; #P<0.05 versus the group treated with TNF-scFv.

Histology

As shown in Figure 7(A), the mean histological score given for the saline-treated group was at a level of 14.8 at the end of the experiment. Dexamethasone treatment, as expected, gave an obvious reduction in histological score. A significant reduction in histological score was also observed in the TNF-scFv-treated and TNF-TeAb-treated groups, whereas TNF-TeAb treatment resulted in a more pronounced reduction, which reached statistical significance in comparison with TNF-scFv treatment. Representative joint histopathology of the experimental groups is shown in Figure 7(B). The characteristics of arthritic joints in rats with CIA are synovial hyperplasia, pannus formation, exudation of cells into the joint space and erosion of bone and cartilage. A massive influx of inflammatory cells, synovial hyperplasia and accumulation of abundant monomorphonuclear and polymorphonuclear cells in the joint space of CIA rats given saline were evident compared with the normal control rats. By comparison, the rats treated with TNF-TeAb revealed an obvious reduction in inflammation or joint destruction to the extent that the synovial membrane in the joints was almost like normal synovium, except for few signs of synovial hyperplasia or other characteristic of inflammation. A lesser degree of reduction in arthritis severity was observed in the rats treated with TNF-scFv. Despite no visible damage to the joint, such as bone erosion and cartilage desquamation, the influx of inflammatory cells and synovial hyperplasia were somewhat pronounced. As expected, an obvious reduction of arthritis severity was observed in the rats that received dexamethasone treatment, and the joint histology was almost identical with that of normal rats.

Figure 7. Histological analysis of the joints among the experimental groups.

(A) The mean histological scores for the experimental groups at the end of the treatment. Results are means±S.D. *P<0.05 versus the group treated with saline; **P<0.01 versus the group treated with saline; #P<0.05 versus the group treated with TNF-scFv. (B) Representative joint histopathology in rats among the experimental groups: a, normal control; b and c, CIA plus saline; d, CIA plus TNF-scFv; e, CIA plus TNF-TeAb; f, CIA plus dexamethosone.

Effect of TNF-scFv and TNF-TeAb on pro-inflammatory cytokines

Figure 8 reports the changes in the concentrations of TNF-α and IL-1β assayed in plasma of rats at the end of the experiment. In the normal group, the plasma levels of TNF-α ranged from 20.0 to 40.0 pg/ml, and for IL-1β from 10.0 to 30.0 pg/ml. A marked increase in the plasma levels of both TNF-α and IL-1β was found in CIA rats given saline. In comparison with the rats given saline, the plasma levels of both TNF-α and IL-1β were significantly reduced in the rats treated with TNF-scFv and even more so in the rats given TNF-TeAb. The difference in plasma levels of both TNF-α and IL-1β between the two construct-treated groups was also statistically significant, suggesting that TNF-TeAb treatment was more effective than TNF-scFv treatment in down-regulating pro-inflammatory cytokines. As expected, there was a pronounced reduction in plasma levels of TNF-α and IL-1β in the dexamethasone-treated group.

Figure 8. Plasma levels of TNF-α and IL-1β in rats among the experimental groups.

Data are presented as means±S.D. *P<0.05 versus the group treated with saline; **P<0.01 versus the group treated with saline; #P<0.05 versus the group treated with TNF-scFv.

Plasma anti-CII IgG

One of the characteristic features of CIA is the presence of high levels of circulating anti-CII antibody, which specifically targets the cartilage of the joints and exacerbates the pathology of arthritis by formation of collagen–IgG complexes and activation of the complement cascade. Anti-TNF-α therapy has been shown to down-regulate the production of anti-CII antibody [32,33]. As shown in Figure 9, the plasma level of anti-CII IgG was 694000±10000 units/ml for the saline-treated group, indicating there was a strong antibody reaction to CII in the arthritic rats. Administration of TNF-scFv, TNF-TeAb, and dexamethsone significantly inhibited the increase of the plasma levels of anti-CII IgG in CIA rats. TNF-TeAb treatment was found to provide a stronger inhibitory effect than TNF-scFv treatment, and the difference between the two groups was statistically significant, suggesting that TNF-TeAb treatment was more effective than TNF-scFv treatment in down-regulating anti-CII antibody.

Figure 9. Circulating anti-cII IgG titres in rats among the experimental groups.

Results as means±S.D. *P<0.05 versus the group treated with saline; **P<0.01 versus the group treated with saline; #P<0.05 versus the group treated with TNF-scFv.

DISCUSSION

Two aspects of the multimerization are of importance for scFv fragments. First, multimerization leads to higher avidity by increasing the number of antigen binding sites. Secondly, the molecular mass is automatically increased by the presence of multiple copies of binding domains. This higher molecular mass extends the serum persistence of molecules in the circulation, because they do not pass, by filtration, into the kidney glomeruli [8,34]. Molecules that are too large to pass the filtration barrier will be retained in the blood pool for a longer time, which provides a greater chance for these molecules to bind to their target antigens. On the other hand, there is an inverse correlation between the molecular mass of these molecules and their ability to penetrate into the target tissue. Thus, in the design of targeting molecules with respect to the molecular mass, a compromise has to be sought. In consideration of the positive therapeutic effect of anti-TNF-α mAb and the impact of molecular mass on the tissue-penetration ability of such therapeutic molecules, we designed a tetravalent anti-TNF-α mini-antibody, TNF-TeAb, with a molecular mass of 156 kDa, which was almost identical with that of a mAb and above the threshold for renal filtration [34]. Multimerization was achieved by the use of the tetramerization domain of the human p53, which led to spontaneous assembly of the fused scFv fragment to a tetramer. The HSA linker had the lowest possible immunogenicity and served as the IgG hinge region, providing for independent folding of the fused domains and a long and flexible stretch for the antigen-binding domain to bind to separated antigens simultaneously. TNF-TeAb could be overexpressed as inclusion bodies in the cytoplasm of E. coli, and be produced in functional form by using an in vitro refolding system after purification with IMAC. The degree of the tetramerization was checked by gel-filtration analysis, and the presence of the expected tetrameric species was confirmed.

In vitro assays suggested that the constructs could inhibit binding of rhTNF-α to its receptors and neutralize the cytolytic acitivity of rhTNF-α. TNF-TeAb input for 50% inhibition of binding of 1 μg/ml rhTNF-α to its receptors showed an approx. 9-fold reduction compared with TNF-scFv, and this was undoubtedly due to the increased avidity of this molecule. The avidity gain was also confirmed by the fact that TNF-TeAb revealed a much stronger ability to neutralize the cytolytic activity of rhTNF-α. Its input for 50% inhibition of 1 ng/ml of activity of rhTNF-α showed an approx. 8-fold decrease compared with TNF-scFv. This result is consistent with the previous study, in which the mono-, di- and tetra-meric scFv constructs were tested for their antigen-binding behaviour, and increased avidity was found with increasing number of binding sites [18], and also supports the view that multimerization of scFv fragment with self-associating protein domain is an efficient way to increase its avidity.

Anti-TNF-α mAb has been well documented to effectively suppress the arthritis progression in both RA and CIA, but, up until now, there have been no reports as to whether antibody fragments have such activity under in vivo conditions. In the present study, type II CIA in rat, a classic model for human RA, was established for an in vivo bioactivity assay. CIA is a polyarthritis induced by sensitization of susceptible strains of animals with CII. There are several similarities with the human RA during the disease initiation and progression, including linkage of disease to genes residing in the histocompatability locus [35], mononuclear-cell infiltration, pannus development, fibrin deposition, erosion of cartilage and bone, and autoreactive T and B cells [36,37]. Cytokines such as TNF-α, IL-1β and IL-6 have been shown to display potent pro-inflammatory actions that are thought to contribute to the pathogenesis of RA [38–40]. Several lines of evidence indicate that TNF-α resides at the apex of this particular pro-inflammatory cytokine cascade and drives its own production and that of IL-1β and IL-6 [41,42]. These pro-inflammatory cytokines, which are increased in RA synovium and blood circulation, have also been shown to contribute to the development of arthritis in the CIA models [43–45]. Furthermore, clinical efficacy of anti-TNF-α agents has also been shown in rat and mouse CIA [46–48]. Since there are so many similarities between the CIA model and human RA, the therapeutic efficacy of an anti-TNF-α agent in the CIA models can directly provide insight into the feasibility of its employment in the treatment of human inflammatory diseases such as RA and IBD. To meaningfully interpret the results from the CIA model when investigating the therapeutic effect of the two constructs, a comprehensive assessment of clinical symptoms, histopathology of joints, pro-inflammatory cytokines and anti-CII antibody is clearly needed. The data obtained from the group treated with TNF-scFv and from TNF-TeAb were positive in all parameters. The findings presented here are consistent with the published work which reported the marked amelioration of established CIA by treatment with anti-TNF-α mAb [46,47], indicating that the antibody fragments also have the anti-TNF-α bioactivity under in vivo conditions. Comparing all the measured parameters between the two construct-treated groups, we can conclude that TNF-TeAb treatment provides a greater therapeutic effect than TNF-scFv treatment at the same therapeutic dose. TNF-scFv treatment showed only a little protection, and often the data were not significant or at the limit of significance, suggesting that monomeric scFv fragments are not the ideal candidates for in vivo targeting therapy. In contrast, TNF-TeAb treatment showed a more significant therapeutic effect in suppressing the CIA damage, with even further reductions in clinical score, hind-ankle swelling, joint damage, pro-inflammatory cytokines and anti-CII antibody. Its therapeutic effect was almost identical with that of dexamethasone at a dose of 0.1 mg/day per kg. The data demonstrate that TNF-TeAb exhibits a satisfactory in vivo anti-TNF-α bioactivity and has potential applicability for the treatment of the human inflammatory diseases, most probably owing to the avidity gain and molecular-mass increase. Unfortunately, there are no data available about the pharmacokinetic properties of these two antibody constructs, although a recent report about further elongation of the serum half-life for the multivalent mini-antibody by PEGylation [poly(ethylene glycol)ylation] is noteworthy [49]. Such studies remain to be carried out. Collectively, we have successfully constructed and produced a tetravalent anti-TNF-α mini-antibody, TNF-TeAb. In vitro and in vivo bioactivity assays demonstrate that TNF-TeAb has potential applicability for anti-TNF-α therapy. This provides a stable foundation for the development of a novel agent for inflammation targeting therapy.