Immunity induced shortly after DNA vaccination of rainbow trout against rhabdoviruses protects against heterologous virus but not against bacterial pathogens

Developmental and Comparative Immunology 26 (2002) 173±179 www.elsevier.com/locate/devcompimm

Immunity induced shortly after DNA vaccination of rainbow trout against rhabdoviruses protects against heterologous virus but not against bacterial pathogens Niels Lorenzen a,*, Ellen Lorenzen a, Katja Einer-Jensen a, Scott E. LaPatra b b

a Danish Veterinary Laboratory, Hangùvej 2, DK-8200 Aarhus N, Denmark Clear Springs Foods, Inc., Research Division, PO. Box 712, Buhl, ID 83316, USA

Abstract It was recently reported that DNA vaccination of rainbow trout ®ngerlings against viral hemorrhagic septicaemia virus (VHSV) induced protection within 8 days after intramuscular injection of plasmid DNA. In order to analyse the speci®city of this early immunity, ®sh were vaccinated with plasmid DNA encoding the VHSV or the infectious haematopoietic necrosis virus (IHNV) glycoprotein genes and later challenged with homologous or heterologous pathogens. Challenge experiments revealed that immunity established shortly after vaccination was cross-protective between the two viral pathogens whereas no increased survival was found upon challenge with bacterial pathogens. Within two months after vaccination, the crossprotection disappeared while the speci®c immunity to homologous virus remained high. The early immunity induced by the DNA vaccines thus appeared to involve short-lived non-speci®c anti-viral defence mechanisms. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: Rhabdovirus; Glycoprotein; Rainbow trout; Plasmid DNA; Genetic immunization; Non-speci®c protection

1. Introduction Vaccination of rainbow trout (Oncorrhyncus mykiss) against the rhabdoviruses viral hemorrhagic septicemia virus (VHSV) and infectious haematopoietic necrosis virus (IHNV) by intramuscular (im) injection of plasmid DNA encoding the viral glycoproteins (G proteins) elicits immunity to experimental * Corresponding author. Tel.: 145-89372429; fax: 14589372470. E-mail address: [email protected] (N. Lorenzen). Abbreviations: viral haemorrhagic septicaemia, VHS; viral haemorrhagic septicaemia virus, VHSV; infectious haematopoietic necrosis virus, IHNV; glycoprotein, G protein; intramuscular, im; days post vaccination, dpv; 50% tissue culture infective doses, TCID50; colony forming units, CFU.

infections with the homologous virus [1,2]. It was recently observed that protection of ®ngerlings against VHSV was evident as early as 8 days post vaccination (dpv) with a 1 mg DNA dose [3]. Since establishment of protection by a speci®c immune response in rainbow trout is expected to require more time, this ®nding suggests involvement of nonspeci®c defence mechanisms. By using a 10-fold higher dose of DNA for vaccination of small fry, Kim et al. [4] observed that plasmids encoding two heterologous ®sh rhabdovirus G proteins could induce immunity to IHNV when challenge was performed 30 dpv, whereas at challenge 70 dpv protection was obtained only with the homologous DNA vaccine. Protection by the heterologous vaccines at 30 dpv was observed to correlate with presence of the non-speci®c antiviral

0145-305X/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 0145-305 X(01)00 059-3

174

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

protein Mx [4]. Earlier work by Boudinot et al. [5] showed that increased transcription of the Mx gene was evident in muscle tissue from mature ®sh as early as 7 days post injection of 30 mg of a plasmid construct mediating expression of the VHSV G protein. The experiments presented here were performed in order to determine to what extent the early immunity elicited in rainbow trout ®ngerlings by the VHSV and IHNV DNA vaccines involved cross-protective mechanisms. Besides VHSV and IHNV, challenge experiments were also performed with the bacterial ®sh pathogens Yersinia ruckeri and Aeromonas salmonicida. Experiments with VHSV and Y. ruckeri were performed at the Danish Veterinary Laboratory in Denmark while experiments with IHNV and A. salmonicida were done at the Clear Springs Foods, Inc. Research Division in Idaho, USA.

2. Materials and methods 2.1. Propagation of viruses and bacteria The VHSV isolate DK 3592B and the IHNV isolate 220-90 were propagated and titrated in established ®sh cell cultures as described earlier [3,6]. A low passage isolate of Y. ruckeri originating from ®sh suffering from enteric red mouth disease was repassaged once in vivo by intraperitoneal injection into rainbow trout and subsequently stored as tissue material (kidney, spleen, brain) in aliquots at 2808C. After plating on blood±agar plates and incubating for 3±4 days at 158C a typical colony was selected and tested for agglutination with anti Y. ruckeri rabbit antiserum and for motility by wet mount phase contrast microscopy. The con®rmed colony was inoculated into 100 ml of meat extract boullion and incubated on a shaker for 4±6 h at 21±228C. Exponential growth was con®rmed by regular measurements of the optical density at 600 nm (OD600). This culture was used for inoculation of a large scale overnight culture incubated under similar conditions and harvested the following day at OD600~1 while still in the exponential phase. The overnight culture was centrifuged at 5000 g for 25 min at 48C and resuspended in 0.9% NaCl in 1/4 of the original volume. OD600 was measured and titration performed by

colony counts of serial 10-fold dilutions on blood± agar plates. Aeromonas salmonicida strain 206-85, kept in frozen stocks at 2758C, was struck for single colony isolation on brain heart infusion (BHI) agar and incubated at 158C for 72 h. The identity of the culture was con®rmed by slide agglutination with homologous polyvalent antisera and checked for typical colony and bacterium cell morphology and production of water-soluble brown pigment. Approximately 10 colonies were suspended in 2±3 ml of BHI broth and this suspension was then used to swab 150 mm plates of BHI agar that were incubated at 158C for 72 h. The bacterial colonies were harvested in 0.5% tryptone, centrifuged, and the clari®ed supernatant was discarded. Plate counts were done on serial 10fold dilutions of a 1% (w/v) bacterial cell suspension. 2.2. DNA vaccination Plasmid constructs included the pcDNA3 vector with or without the VHSV (isolate DK-3592B) G gene [7] and the pcDNA3.1(1) vector with or without the IHNV (isolate US-WRAC) G gene [8] (vectors were from Invitrogen, Groeningen, The Netherlands). The pcDNA3-VHSV:G and the pcDNA3.1-IHNV:G constructs will here be called pVHS-G and pIHN-G, respectively. Plasmid DNA was puri®ed from overnight cultures of transformed E. coli DH5a by the alkaline lysis protocol of Saporito-Irwin et al. [9] or by anion-exchange chromatography as outlined earlier [7]. Disease-free rainbow trout ®ngerlings with an average weight between 1.9 and 4.5 g were anaesthetized and given an intramuscular injection of 1 mg DNA in 25 ml neutral buffer as described earlier [3,8]. 2.3. Challenge experiments At different timepoints after vaccination, the ®sh were challenged with VHSV, Y. ruckeri, IHNV or A. salmonicida. The ®sh were kept in running water at 11±138C in experiments including VHSV and Y. ruckeri and at 158C in experiments including IHNV and A. salmonicida. Mortality was recorded and dead ®sh removed from the aquaria daily. Experiments were terminated three weeks after inoculation of the ®sh. Representative samples of dead ®sh were examined virologically from all aquaria challenged with

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

175

Table 1 Cumulative mortality following challenge with VHSV at different times post vaccination. (Values are the cumulated percentages of mortality 21 days after challenge. Numbers in brackets refer to deaths/total for each group (sum of triplicate aquaria with 25±31 ®sh in each). Groups with signi®cant survival (p , 0.01) compared to unhandled ®sh are marked with asterisks*. No mortality occurred in aquaria with mock-challenged control ®sh in any of the four challenge experiments. VHSV was re-isolated from representative samples of dead ®sh from all aquaria inoculated with virus) Vaccine

Plasmid vector pVHS-G p IHN-G Unhandled

Timepoint of challenge 4 dpv

7 dpv

60 dpv

84 dpv

96% (87/91) 21% (18/87)* 41% (37/90)* 96% (86/90)

95% (87/92) 19% (17/90)* 12% (11/90)* 91% (82/90)

91% (82/90) 1% (1/89)* 94% (85/90) 86% (78/91)

96% (72/75) 5% (4/75)* 96% (72/75) 99% (74/75)

virus as described earlier [3,8], while all dead ®sh from the aquaria exposed to Y. ruckeri and A. salmonicida were examined bacteriologically. The latter included inoculation of anterior kidney material onto blood agar plates (Y. ruckeri) or onto tryptic soy agar plates (A. salmonicida). After incubation at 158C for at least 3 days, the resulting colonies were identi®ed by fermentation characteristics, colony and cell morphology, and agglutination with speci®c rabbit antisera. 2.3.1. Comparison of VHSV and IHNV DNA vaccines in challenge with VHSV Rainbow trout ®ngerlings, vaccinated at an average weight of 3.0 g, were challenged in triplicates of 25± 31 ®sh with VHSV by immersion for 2 h in water containing approx. 4 £ 10 5 50% tissue culture infective doses (TCID-50) of virus according to earlier described procedures [3]. Challenges was performed on subgroups of vaccinated ®sh at four different time points including 4, 7, 60 and 84 dpv, respectively. Three aquaria with mixed vaccine groups were mock-challenged with cell culture medium without virus and served as controls for background mortality in each challenge experiment. 2.3.2. Testing of the protective effect of the VHSV DNA vaccine against Y. ruckeri For immersion challenge with Y. ruckeri 100 ml bacterial suspension in 0.9% NaCl was added to 900 ml of aquarium water in a 2 l container supplied with aeration. Rainbow trout vaccinated at an average weight of 1.9 g were immersed in triplicates of approx. 30 ®sh. Control ®sh were immersed in

900 ml of aquarium water plus 100 ml 0.9% NaCl. The exposure was conducted for 3 h in 6 £ 10 8 colony forming units (CFU) per ml at challenge performed 8 dpv and for 4 h in 9 £ 10 8 CFU per ml at challenge performed 30 dpv, respectively. Following immersion, the ®sh and the 1000 ml bacterial suspensions were transferred to aquaria containing 7 l of water and supplied with running tap water. In parallel, groups of 30 ®sh were challenged with VHSV under similar conditions. Exposure concentrations were approx. 5 £ 10 5 TCIDÐ50 per ml. 2.3.3. Testing of protective effect of IHNV and VHSV DNA vaccines against A. salmonicida Challenges of rainbow trout with IHNV were performed on duplicate 20 ®sh groups 18 dpv. Fish were immersed in 10 5 plaque forming units/ml of strain 220-90 for 60 min with aeration in a volume of water that was 10 £ the total weight of the ®sh (g). Fish challenged with A. salmonicida were anaesthetized by immersion in 100 mg/ml of MS-222 and 50 ml of a 1:1000 dilution of a 1% (w/v) cell suspension was injected intraperitoneally into duplicate groups of 15±20 ®sh. Mock infected control groups were treated identically, but were exposed to cell culture media or injected with 0.5% tryptone only. 2.4. Statistics Statistical analyses for signi®cant differences between mortalities were based on logistic regression analysis with adjustment for possible over-dispersion due to the variation between replicate aquaria [10].

176

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

Fig. 1. Alignment of the deduced G-protein amino acid sequences of VHSV strain DK-3592B (Genbank AC No X66134) and IHNV strain USWRAC (AC No L40883). Identical amino acid residues are marked with asterisks. Overall identity is 37%.

3. Results and discussion Both the VHSV and IHNV DNA vaccines induced signi®cant (p , 0.001) protection against VHSV in challenge experiments performed 4 and 7 dpv. While there was a slightly better protection with the homologous vaccine in the 4 dpv-challenge, no difference was evident in the 7 dpv challenge experiment (Table 1). As observed earlier [3], the protective effect of the pVHS-G increased with time and resulted in almost complete protection in the challenge trials performed 60 and 84 dpv. In these trials, however,

the mortality among ®sh injected with pIHN-G was as high as in the control groups where the ®sh had received the vector without any insert or were left unhandled (Table 1). A similar pattern was recently observed when ®sh given the two DNA vaccines were challenged with IHNV. The VHSV vaccine gave a protection comparable to that of the homologous (IHNV) vaccine in challenges performed at 4, 7 and 14 dpv, whereas no protection was evident with the heterologous (VHSV) vaccine in a challenge performed 28 dpv [11]. In the present work, the early protective effect of the VHSV DNA vaccine against

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

177

Table 2 Cumulative mortality following challenge with VHSV or Y. ruckeri at different times post vaccination. (Values are the cumulated percentages of mortality 21 days after challenge. Numbers in brackets refer to deaths/total for each group (sum of triplicate aquaria). Groups with signi®cant survival (p , 0.01) compared to ®sh injected with the plasmid vector are marked with asterisks*. Background mortality in mock-challenged controls was 4% in the 8 dpv-challenge. No pathogens were detected in the dead controls. In the 30 dpv-challenge elevated background mortality occurred in a few aquaria including one of the unchallenged controls within the ®rst 3 days after challenge. Fish dying within the ®rst 3 days after the 30 dpv-challenge were, therefore, excluded from the results. VHSV or Y. ruckeri were reisolated from representative samples of dead ®sh in all aquaria inoculated with the respective pathogens) Vaccine

Challenge 8 dpv

Plasmid vector pVHS-G

30 dpv

VHSV

Y. ruckeri

VHSV

Y. ruckeri

97% (90/93) 13% (12/89)*

26% (24/92) 44% (40/91)

75% (67/89) 8% (7/89)*

64% (58/90) 65% (51/78)

IHNV was demonstrated in a challenge performed 18 dpv (Table 3). In terms of the two different ®sh rhabdoviruses, the early protection induced by the DNA vaccines was thus non-speci®c or cross reactive. While fully conserved cysteine residues indicate that the three-dimensional structure of the VHSV and IHNV G proteins is high [12], the overall amino acid homology of the two viral genes is only about 37%. Short homologous stretches of identical amino acids are present (Fig. 1), but it appears unlikely that speci®c immune response mechanisms would elicit considerable cross-reaction between the two viruses. Table 3 Cumulative mortality following challenge with IHNV or A. salmonicida. (Values are the cumulated percentages of mortality 21 days after challenge. Numbers in brackets refer to deaths/total for each group (sum of duplicate aquaria with 15±20 ®sh in each). Groups with signi®cant survival (p , 0.01) compared to unhandled ®sh are marked with asterisks*. Among the examined dead ®sh that had been challenged with IHNV, 87% (60/69) were IHNV-positive. For ®sh challenged with A. salmonicida, 100% (33/33 analysed) were positive. There was no detectable virus or A. salmonicida in any of the small number of mock-infected control ®sh that died in these experiments) Vaccine

Plasmid vector p VHS-G p IHN-G Buffer Unhandled

Challenge 18 dpv IHNV

A. salmonicida

78% (31/40) 13% (5/40)* 18% (7/40)* 67% (26/39) 73% (29/40)

28% (11/39) 27% (11/41) 17% (6/35) 40% (12/40) 36% (11/30)

This was supported by the highly speci®c protection found in challenges performed at later time points (Table 1). Boudinot et al. [5] accordingly observed no cross reactivity in the antibody response induced by VHS and IHN DNA vaccines similar to the ones used in the present study. In order to determine whether the early crossreactive protection induced by the two DNA vaccines went beyond protection against a heterologous virus, vaccination trials including challenge with the bacterial pathogens Y. ruckeri and A. salmonicida were also performed. Although the mortalities following the exposures to the two bacteria were lower compared to those induced by the viruses, the results demonstrated that the two DNA vaccines induced no protection against the bacterial pathogens (Tables 2 and 3). The cross protection against heterologous virus combined with the lack of protection against the two extracellular bacterial infections suggests that the rhabdovirus G gene DNA vaccines activate early non-speci®c antiviral defence mechanisms with similarity to the effects known from type I interferons [13]. The dual-phase scenario with early activation of non-speci®c protection following DNA vaccination against VHSV and IHNV followed by a later phase of speci®c immunity parallels observations reported for live attenuated vaccines. Bernard et al. [14] found that while speci®c antibodies could be associated with long term protection following vaccination with an avirulent variant of VHSV, rainbow trout were protected against challenge with a virulent isolate as early as 1 dpv. These authors suggested that the early

178

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

Table 4 Prevalence of CpG motifs in vectors and viral genes Sequence

Number of RRCGYY motifs a

VHSV G gene IHNV G gene pcDNA3 plasmid vector pcDNA3.1(1) plasmid vector

3 2 26 25

a The nucleotide sequence RRCGYY include all 16 combinations of two purines-CG-2 pyrimidines. The two CpG motifs in the IHNV G gene were both present in the plasmid vectors in two and ®ve copies respectively. Among the three CpG motifs in the VHSV G gene, two were not present in the plasmid vectors.

protection induced by the avirulent vaccine virus could either be due to a rapid induction of interferon, or infection of individual target cells by the vaccine virus could interfere with a secondary infection by the virulent virus isolate [14]. Since the DNA vaccine plasmid is non-infectious, only a limited number of cells in the ®sh will be transfected. Direct interference with a viral infection at the target cell level therefore cannot explain the early protection phenomenon. Interferons remain to be characterized at the molecular level in ®sh, but early work with passive transfer of immunity indicated that a low-molecular protective component with interferon-like activity in cell culture was present in sera obtained from ®sh injected with live VHSV 2 days earlier [15]. Further support for a similarity between the early innate defence mechanisms induced by viral infections and by the DNA vaccines has come from experiments demonstrating increased expression of the interferon induced Mx proteins in ®sh exposed to IHNV [16] as well as in ®sh vaccinated with plasmid DNA mediating expression of the respective glycoproteins [4,5]. However, non-speci®c defence mechanisms other than interferon could also explain/be involved in the early protection. In a detailed study of immune mechanisms induced in mice by a DNA vaccine encoding the rabies virus G protein, interferon as well as NK cell activity in the blood were increased as early as 1 dpv., while the ®rst speci®c IgM were detected at 3 dpv followed by IL2, cytotoxic T cells and IgG at 7 dpv [17]. Since that study did not include challenge at these early time points, the protective effect of these immune parameters was not determined. In DNA vaccination experiments with mammals, few

reports include challenge shortly after vaccination. Hassett et al. [18] were surprised to ®nd that mice were well protected against lymphocytic choriomeningitis virus as early as 7 dpv. It was assumed that the protection was mediated by antigen-speci®c CD8 1 T cells, but the speci®city of the protection was not analysed. Although remaining to be analysed in detail, it appears likely that expression of the viral G proteins is required for induction of the early protection observed in the present study. Accordingly, neither early nor late protection was observed with a pcDNA3-IHNV:G construct which failed to induce expression of the G protein in transfected cells (not published). Certain sequences with CpG motifs in non-methylated bacterial DNA are known to induce/promote non-speci®c defence mechanisms [19]. Fish rhabdovirus G genes contain few such motifs, but the prevalence is high in the pcDNA3 plasmid backbone (Table 4). While such CpG induced effects alone cannot explain the results, as evident from the lack of protection with the vector without insert, it cannot be excluded that CpG motifs could play a role indirectly by inducing/ augmenting mechanisms involved in the immune response induced by expression of the viral G genes [19]. In conclusion, the early phase of the immune response induced in rainbow trout by DNA vaccination against VHSV and IHNV include induction of non-speci®c anti-viral defence mechanisms which are gradually replaced by a more speci®c immune response. The exact nature of these mechanisms remains to be determined. Acknowledgements The authors thank Lisbeth Troels and Helle Hermansen for skilful technical laboratory assistance and Torben Kjñr for assistance with the aquarium experiments. Torben Martinussen is thanked for assistance with statistical analyses and Serge Corbeil is thanked for supplying the IHNV (WRAC) DNA vaccine construct. The work was supported by a grant from the Danish Directorate for Development (FISK 97-2), and by the European Commission (Contract FAIR CT98-4087).

N. Lorenzen et al. / Developmental and Comparative Immunology 26 (2002) 173±179

References [1] Lorenzen N, Lorenzen E, Einer-Jensen K, Heppell J, Wu T, Davis HL. Protective immunity to VHS in rainbow trout (Oncorhynchus mykiss, Walbaum) following DNA vaccination. Fish Shell®sh Immunol 1998;8:261±70. [2] Anderson ED, Mourich DV, Fahrenkrug SC, LaPatra SE, Shepherd J, Leong J-AC. Genetic immunisation of rainbow trout (Oncorhynchus mykiss) against infectious hematopoietic necrosis virus. Mol Mar Biol Biotechnol 1996;5:114±122. [3] Lorenzen E, Einer-Jensen K, Martinussen T, LaPatra SE, Lorenzen N. DNA vaccination of rainbow trout against viral hemorrhagic septicemia virus: a dose±response and timecourse study. J Aquat Anim Health 2000;12:167±80. [4] Kim CH, Johnson MC, Drennan JD, Simon BE, Thomann E, Leong JC. DNA vaccines encoding viral glycoproteins induce nonspeci®c immunity and Mx protein synthesis in ®sh. J Virol 2000;74:7048±54. [5] Boudinot P, Blanco M, de Kinkelin P, Benmansour A. Combined DNA immunisation with the glycoprotein gene of viral hemorrhagic septicemia virus and infectious hematopoietic necrosis virus induces double-speci®c protective immunity and nonspeci®c response in rainbow trout. Virology 1998;249:297±306. [6] LaPatra SE, Lauda KA, Jones GR. Antigenic variants of infectious hematopoietic necrosis virus and implications for vaccine development. Dis Aquat Org 1994;20:119±26. [7] Heppell J, Lorenzen N, Armstrong MK, Wu T, Lorenzen E, EinerJensen K, Schorr J, Davis HL. Development of DNA vaccines for ®sh: vector design, intra muscular injection and antigen expression using viral haemorrhagic septicaemia virus genes as model. Fish Shell®sh Immunol 1998;8:271±86. [8] Corbeil S, LaPatra SE, Anderson ED, Kurath G. Nanogram quantities of a DNA vaccine protects rainbow trout fry against heterologous strains of infectious hematopoietic necrosis virus. Vaccine 2000;18:2817±24. [9] Saporito-Irwin SM, Geist RT, Gutmann DH. Ammonium acetate protocol for the preparation of plasmid DNA suitable

[10] [11]

[12]

[13] [14]

[15] [16]

[17] [18] [19]

179

for mammalian cell transfections. Biotechniques 1997; 23:424±7. McCullagh P, Nelder JA. Generalized linear models. New York: Chapman and Hall, 1989. LaPatra SE, Corbeil S, Jones GR, Shewmaker WD, Lorenzen N, Anderson ED, Kurath G. Protection of rainbow trout against infectious hematopoietic necrosis virus four days after speci®c or semi-speci®c DNA vaccination. Vaccine 2001;19:4011±19. Einer-Jensen K, Krogh TN, Roepstorff P, Lorenzen N. Characterization of intramolecular disulphide bonds and secondary modi®cations of the glycoprotein from viral hemorrhagic septicaemia virus, a ®sh rhabdovirus. J Virol 1998;72: 10189±96. Biron CA. Role of early cytokines, including alpha and beta interferons (IFN-a/b), in innate and adaptive immune responses to viral infections. Sem Immunol 1998;10:383±90. Bernard J, de Kinkelin P, Bearzotti-le Berre M. Viral hemorrhagic septicemia of rainbow trout: relation between the G polypeptide and antibody production in protection of ®sh after infection with the F25 attenuated variant. Infect Immun 1983;39:7±14. de Kinkelin P, Dorson M, Hattenberger-Baudouy AM. Interferon synthesis in trout and carp after viral infection. Dev Comp Immunol 1982(Suppl 2):167±74. Trobridge GD, Chiou PP, Kim CH, Leong JC. Induction of the Mx protein of rainbow trout Oncorhynchus mykiss in vitro and in vivo with poly I:C dsRNA and infectious hematopoietic necrosis virus. Dis Aquat Org 1997;30:91±8. Bahloul C, Jacob Y, Tordo N, Perrin P. DNA based immunization for exploring the enlargement of immunological crossreactivity against lyssaviruses. Vaccine 1997;16:417±25. Hassett DE, Slifka MK, Zhang J, Whitton JL. Direct ex vivo kinetic and phenotypic analyses of CD8 1 T-cell responses induced by DNA immunization. J Virol 2000;74:8286±91. Manders P, Thomas R. Immunology of DNA vaccines: CpG motifs and antigen presentation. In¯amm Res 2000;49:199± 205.