Reaction by Staphylococcal Protein a Cell Superantigen: Elicitation of An

Home , Arthus reaction

In Vivo Inflammatory Response to a Prototypic B Cell Superantigen: Elicitation of an Arthus Reaction by Staphylococcal Protein A

This information is current as Lisa M. Kozlowski, Weiping Li, Michael Goldschmidt and Arnold of October 1, 2021. I. Levinson J Immunol 1998; 160:5246-5252; ; http://www.jimmunol.org/content/160/11/5246 Downloaded from References This article cites 50 articles, 28 of which you can access for free at: http://www.jimmunol.org/content/160/11/5246.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/ • Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on October 1, 2021 Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1998 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. In Vivo Inﬂammatory Response to a Prototypic B Cell Superantigen: Elicitation of an Arthus Reaction by Staphylococcal Protein A1

Lisa M. Kozlowski,2* Weiping Li,* Michael Goldschmidt,† and Arnold I. Levinson3*

Staphylococcal protein A (SpA) is representative of a new class of Ags, the B cell superantigens (SAgs). These SAgs, unlike conventional Ags, bind to the Fab regions of Ig molecules outside their complementarity-determining regions. In addition, B cell SAgs can react with a substantial amount of a host’s serum Igs by virtue of their ability to interact with many members of an entire variable heavy chain (VH) or variable light chain gene family. For example, SpA reacts with the Fabs of most human Igs using ؉ heavy chains from the VH3 gene family (VH3 ). Members of this gene family are expressed on 30 to 60% of human peripheral B cells. We sought to determine whether the interaction of a B cell SAg with its reactive Igs can elicit immune complex-mediated Downloaded from tissue injury. Using the Arthus reaction in rabbits as an in vivo model of immune complex-mediated tissue inflammation, we demonstrated that untreated rabbits, which were administered SpA intradermally (i.d.), do not develop a cutaneous inflammatory response. However, when rabbits were pretreated i.v. with human IgG (hIgG), i.d. injections of SpA induced an inflammatory response with the classical histologic features of an Arthus reaction. To determine whether this Arthus-like response occurred via a B cell superantigenic mechanism, the rabbits were pretreated with VH3-depleted hIgG and then were administered SpA i.d. We ؉ found that the induction of a prominent inflammatory response by SpA was dependent upon the presence of VH3 molecules in http://www.jimmunol.org/ the hIgG pretreatment. These results provide compelling evidence that an interaction of the B cell SAg, SpA, with its reactive ؉ (VH3 ) IgGs leads to an immune complex-mediated inflammatory response in vivo. The Journal of Immunology, 1998, 160: 5246–5252.

taphylococcal protein A (SpA)4, a cell wall component of ability of this bacterial cell wall protein to activate human B cells

Staphylococcus aureus, binds to the Fc fragment of IgG. inaVH-selective manner (6). These properties are reminiscent of S In addition, an alternative site on SpA has been defined those of a T cell superantigen (SAg) and have led SpA to be char- that binds to the Fab region of Igs independently of the heavy acterized as a B cell SAg (9, 10, 14–18). Recently, several other chain isotype (1–6). Studies have mapped the Fab determinants to proteins have also been defined as B cell SAgs, including HIV by guest on October 1, 2021 framework regions 1 and 3 in the variable heavy chain (VH) region gp120, protein Fv (a human liver sialoprotein), protein L (a coat (7–9), with a possible contribution of residues in complementarity- protein of Peptostreptococcus magnus), and staphylococcal determining region 2 (8). The binding of this alternative site on enterotoxin D (19–22). SpA is restricted to human Igs using heavy chains from the VH3 ϩ Given its ability to react with a large amount of Ig molecules, a gene family (VH3 ) (9, 10). The VH3 gene family is the largest of B cell SAg could inflict tissue damage through a number of in- the seven human VH gene families and is expressed by 30 to 60% flammatory mechanisms. Its interaction with cytophilic IgG, IgE, of human peripheral B cells (11–13). The cross-linking of mem- or IgA molecules could lead to the cross-linking of the respective brane IgM by the alternative binding site on SpA accounts for the Ig FcRs on inflammatory cells, thereby resulting in the release of inflammatory mediators. Indeed, SpA, protein L, and protein Fv *Division of Allergy and Immunology, University of Pennsylvania School of Med- induce histamine release from human basophils by interacting with icine and †Laboratory of Pathology, University of Pennsylvania School of Veterinary the Fab region of IgE molecules that are bound to Fc⑀R on these Medicine, Philadelphia, PA 19104 cells (23–25); protein L and protein Fv also degranulate human Received for publication October 10, 1997. Accepted for publication January 29, 1998. mast cells (24, 25). The interaction of a B cell SAg with fluid- phase IgG could lead to immune complex-mediated tissue injury, The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance a possibility that has not been formally investigated. However, with 18 U.S.C. Section 1734 solely to indicate this fact. based on our recent findings (26), it is now known that the immune 1 This work was supported by a Biomedical Sciences Research grant from the Na- complexes that are formed by a B cell SAg with reactive serum Igs tional Arthritis Foundation and by National Institutes of Health Grant AI 22913. cause activation of the classical complement cascade. 2 Current address: Johns Hopkins University, 755 Ross Building, 1721 E. Monument Street, Baltimore, MD 21205. For many years, investigators have exploited the Arthus reaction 3 Address correspondence and reprint requests to Dr. Arnold Levinson, 726 Clinical as a model of in vivo immune complex-mediated tissue injury Research Building, 415 Curie Boulevard, Division of Allergy and Immunology, Uni- (27–30). In the classical model of the Arthus reaction, animals are versity of Pennsylvania School of Medicine, Philadelphia, PA 19104. immunized with an Ag until they have appreciable levels of pre- 4 Abbreviations used in this paper: SpA, staphylococcal protein A; VH, variable heavy cipitating IgG Abs. Intradermal (i.d.) injection of the same Ag chain; SAg, superantigen; HSA, human serum albumin; Mod SpA, staphylococcal protein A hyperiodinated to abrogate its IgG Fc-binding ability; Seph, CNBr-activated elicits a local inflammatory response (28). This response is char- Ϫ ϩ Sepharose 4B; hIgG, human IgG; VH3 hIgG, VH3-depleted human IgG; VH3 using ϩ ϩ acterized grossly by erythema, edema, and hemorrhage and mi- heavy chains from the VH3 gene family; VH3 /non-VH3 hIgG, human IgG con- ϩ ϩ croscopically by a prominent polymorphonuclear cell (PMN) in- taining both VH3 and non-VH3 IgGs; i.d., intradermal(ly); PMN, polymorphonuclear cell. filtrate that peaks at 8 h after cutaneous challenge.

In the present study, we sought to determine whether a B cell termined again as described above. The effluent fractions were then passed SAg could elicit immune complex-mediated tissue injury. We now over their respective columns a second time, and the dialysis and concen- show that rabbits injected i.d. with the model B cell SAg, SpA, do tration procedures were repeated. not develop a cutaneous Arthus reaction, as previously reported (30). However, when rabbits are pretreated i.v. with human IgG ELISA for determining the binding of IgG fractions to Mod SpA (hIgG) from healthy donors, they do develop a cutaneous reaction or SpA with the histologic features of the Arthus reaction at the sites that Half-area microtiter plate wells (Costar, Cambridge, MA) were coated with were injected i.d. with SpA. This reaction is mediated by the Fab- 100 ␮l Mod SpA, SpA, or BSA (Calbiochem) at 10 ␮g/ml in PBS over- binding site on SpA, since it was not induced in animals that were night at 4°C. Each well was subsequently saturated with 100 ␮l 1% BSA/ pretreated with (hIgG) depleted of V 3ϩ molecules (V 3ϪhIgG). PBSfor2hatroom temperature. The wells were washed three times with H H 0.05% Tween-20 (Sigma) in PBS. We incubated 100 ␮l aliquots of varying These data provide the first evidence that the interaction of a B cell ϩ concentrations of unfractionated hIgG or hIgG fractions from the Mod SAg, SpA, with its reactive (VH3 ) Igs leads to an immune com- SpA-Seph column or the Seph column for2hatroom temperature. The plex-mediated inflammatory reaction in vivo. wells were washed as described above, and peroxidase-conjugated goat Ј F(ab )2 anti-hIgG Fc Ab (Jackson ImmunoResearch Laboratories, West Grove, PA) was added for1hatroom temperature. Following this incu- Materials and Methods bation, the plates were washed as described above, the bound Ab was Animals detected by the addition of o-phenylenediamine substrate (Eastman Kodak, Male and female New Zealand white rabbits (2–4 kg) were obtained from Rochester, NY) in 20 mM citrate buffer (pH 4.0) plus 0.05% hydrogen Ace Animals (Boyertown, PA) and housed in the animal facility at the peroxide, and the OD was read spectrophotometrically at 450 nm.

University of Pennsylvania (Philadelphia, PA). Downloaded from Arthus reaction Administered reagents We adapted a previously described model (30) to determine whether the Lyophilized, pooled normal hIgG (kindly provided by Sandoz Pharmaceu- Arthus reactivity induced by SpA was mediated by the interaction of the tical, East Hanover, NJ) was reconstituted to 60 mg/ml with sterile 0.9% Fab-binding site on SpA with reactive IgG molecules. In this model, un- saline. rSpA (Repligen, Needham, MA) and human serum albumin (HSA) treated rabbits injected i.d. with SpA failed to develop inﬂammatory reac- (low endotoxin) (Calbiochem, La Jolla, CA) were sterile ﬁltered through tions at injection sites. However, animals pretreated i.v. with hIgG devel-

␮ http://www.jimmunol.org/ 0.2- m filters (Millipore, Bedford, MA). oped Arthus reactions at cutaneous sites that had been injected with SpA. Preparation of hyperiodinated SpA (Mod SpA) The rabbits in our experiments were sedated with 150 mg ketamine HCl (Ketaject, Phoenix Pharmaceuticals, St. Joseph, Missouri) and 10 mg xy- rSpA was hyperiodinated to abrogate its IgG Fc-binding activity, as pre- lazine (Xylaject, Phoenix Pharmaceuticals), and their backs were shaved. viously described (6). The animals were rested for 48 h before injections were performed to allow any inflammation that might have occurred from the shaving to subside. Conjugation of Mod SpA to Sepharose 4B After 48 h, the animals were sedated with ketamine/xylazine as described above and then injected i.v. with either 170 mg hIgG, 170 mg IgG effluent We suspended 300 ␮g of freeze-dried CNBr-activated Sepharose 4B from the Mod SpA-Seph column or the Seph column, or saline (5 ml), (Seph) (Sigma, St. Louis, MO) in 12.5 ml of 1 mM HCl; the Seph was followed by 3 to 5 ml of 0.9% saline. At 10 to 15 minutes after the i.v. subsequently rotated for 15 min at room temperature to swell the beads

injection, the dorsal skin of the rabbits was prepared with alcohol. The i.d. by guest on October 1, 2021 (300 ␮g of Seph yields 1 ml of gel.) The beads were then centrifuged at injections of 200 ␮lof900␮g of SpA or HSA (negative control) were 2500 rpm for 2 min at room temperature. The supernatant was aspirated, performed at duplicate sites. The same volume of saline was injected i.d. and the beads were resuspended in 10 ml of 1 mM HCl and centrifuged as at duplicate sites as an additional negative control. After 8 h, the dorsal skin described above. Following the repetition of these steps, the beads were was examined macroscopically for erythema, edema, and hemorrhage. The washed in coupling buffer (0.1 M NaHCO and 0.5 M NaCl (pH 8.3)). The 3 animals were then sacrificed, and skin biopsies were obtained from the i.d. supernatant was aspirated, and 4 mg Mod SpA plus coupling buffer or injection sites. The specimens were fixed in 10% buffered formalin, em- coupling buffer alone (for a total volume of 1.5 ml) was added to 1 ml of bedded in paraffin, and stained with hematoxylin and eosin. The histologic swelled Seph. Following rotation for 2 h, the beads were centrifuged as procedures were performed by the Laboratory of Pathology at the Univer- described above and washed once in coupling buffer. Next, 2 ml of 0.2 M sity of Pennsylvania School of Veterinary Medicine (Philadelphia, PA). glycine was added to 1 ml of swelled Seph and rotated for1hatroom temperature. The Seph beads were centrifuged, washed with an acetate buffer (0.1 M sodium acetate and 0.5 M NaCl (pH 4.0)), and then washed with a Tris buffer (0.1 M Tris-HCl and 0.5 M NaCl (pH 8.0)). This pro- Results cedure was repeated two more times. The Mod SpA-conjugated and un- Generation of a cutaneous inflammatory response by SpA conjugated Seph beads were stored in 0.1% BSA/PBS with 0.02% sodium To establish the Arthus model in our laboratory, three rabbits were azide at 4°C until needed. At the time of use, the Seph beads were washed extensively with the appropriate buffer. injected i.v. with 170 mg of hIgG and then injected i.d. with 900 ␮ ϩ g of SpA, HSA, or saline at duplicate sites 10 to 15 min later. Depletion of VH3 IgG from pooled hIgG Three additional rabbits were pretreated i.v. with saline and then The pooled normal hIgG was passed over either a Mod SpA-conjugated injected i.d. with the above reagents as described. All rabbits were Sepharose column (Mod SpA-Seph column) or an unconjugated Sepharose examined grossly and then sacrificed at 8 h after the i.d. injections column (Seph column). Polyprep chromatography columns (Bio-Rad, Her- were administered. cules, CA) were packed with 10 ml of either Mod SpA-Seph or Seph alone and washed with 100 ml of 1ϫ running buffer (1:5 dilution of 5ϫ running No macroscopic evidence of an inflammatory response was observed at any i.d.-injected sites of saline-pretreated rabbits, as pre- buffer (pH 7.0): 0.079 M KH2PO4,0.48MNa2HPO4, and 0.77 M NaCl). Next, 1 ml of pooled IgG (60 mg/ml) diluted with 250 ␮lof5ϫ running viously reported (30). By contrast, erythema and induration were buffer was added to each column and allowed to pass into the Seph. The both seen at the SpA-injected sites of animals pretreated i.v. with column was then closed, and the IgG was incubated in the column for 5 hIgG (data not shown). Such reactions were not observed at the min. This procedure was repeated twice for each column. Afterward, 10 ml of 1ϫ running buffer was added, and 1 ml fractions were collected. This HSA- or saline-injected sites of these hIgG-pretreated animals process was repeated until the A280 of the fractions was Ͻ0.100. The (data not shown). column was subsequently flushed with 50 ml of 1ϫ running buffer. The Histologic examination of the skin biopsies from the hIgG-pre- IgG concentrations of the collected fractions were determined by the fol- treated rabbits revealed a prominent inflammatory cell infiltrate lowing method: A280/extinction coefficient for IgG (1.43). Effluent fractions with a concentration of Ͼ1 mg/ml were pooled together and dialyzed only at the sites injected with SpA (Fig. 1, A and B). The venules two times against PBS at 4°C. The samples were concentrated in an Ul- were congested with PMNs. The PMNs were also observed within trafree-50 centrifugal filter (Millipore), and IgG concentrations were de- the walls of the venules and in a perivascular distribution. Several 5248 INDUCTION OF AN ARTHUS REACTION BY A B CELL SUPERANTIGEN

areas of hemorrhage and diapedesis of erythrocytes were also observed. The arterioles were not involved. Rare PMNs were observed in the lumen of the venules that were examined at the HSA- injected sites (Fig. 1C), whereas no inflammatory cells were observed at the saline-injected sites (data not shown). In the saline- pretreated animals, rare PMNs were observed in venule lumens at the SpA- (Fig. 2) and HSA-injected sites (data not shown), while no PMNs were seen at cutaneous sites injected with saline (data not shown). These results demonstrate that SpA induces a prominent cutaneous inflammatory response that has the features of an Arthus reaction, and that this response is only seen in rabbits pretreated with hIgG. The rabbit’s own IgG alone is not sufficient to sustain an Arthus reaction as demonstrated in the saline-pretreated animals challenged i.d. with SpA. Thus, these results confirm that SpA elicits an Arthus-like reaction in vivo, presumably via binding to hIgGs. Downloaded from ϩ Depletion of VH3 IgG molecules from hIgG To establish that the observed inflammatory response required ϩ both SpA and human VH3 IgGs, the hIgG preparation was sep- Ϫ arated into a VH3-depleted (VH3 ) fraction. To obtain this fraction (as described in Materials and Methods), hIgG was passed twice

over a Mod SpA-Seph column, and the effluent fractions were http://www.jimmunol.org/ ϩ ϩ collected. The control IgG, containing both VH3 and non-VH3 ϩ ϩ hIgGs (VH3 /non-VH3 hIgG, consisted of hIgG that was passed twice over a Seph column. Following fractionation over the Mod SpA-Seph column or the Seph column, the binding of the effluent IgG fractions to Mod SpA and SpA was examined by ELISAs (see Materials and Methods). As shown in Figure 3, the effluent IgG from the Mod SpA-Seph column had a 100-fold reduction in Mod FIGURE 1. Arthus reaction as seen in histologic sections from the skin SpA-binding IgGs as compared with equivalent concentrations of of an hIgG-pretreated rabbit. Rabbits were pretreated i.v. with 170 mg of unfractionated hIgG and the effluent IgG from the Seph column. by guest on October 1, 2021 ␮ ϫ ϫ ϫ hIgG, and 900 g of SpA (A, 31.2; B, 62.5), HSA (C, 31.2), or saline By contrast, IgG fractions from both columns bound SpA to a (data not shown) was administered i.d 10 to 15 min later. Animals were similar degree (data not shown). This finding demonstrates that the sacrificed after 8 h. Skin biopsies were fixed in 10% buffered formalin and stained with hematoxylin and eosin. The arrows in A demonstrate a prom- two types of IgG fractions contain roughly equivalent amounts of inent PMN infiltration in the venules. B shows a ϫ62.5 magnification of SpA-binding IgG molecules, and that the fractionation procedure the boxed area in A, demonstrating intense hemorrhage without the in- does not lead to nonspecific IgG degradation. Neither of the IgG volvement of the arteriole. The section in C shows a lack of cellular in- fractions bound BSA-coated wells (data not shown). Therefore, filtration and hemorrhage in the HSA-injected sites. Results are represen- fractionation of hIgG over a Mod SpA-Seph column removed a tative of duplicate injections from six rabbits. ϩ Ϫ sufficient amount of VH3 IgGs to use this fraction as a VH3

FIGURE 2. Lack of an Arthus reaction in a saline-pretreated rabbit. Rabbits were pretreated i.v. with saline, and 900 ␮gofSpA (ϫ31.2), HSA (data not shown), or saline (data not shown) was administered i.d 10 to 15 min later. The animals were sacrificed after 8 h. Skin biopsies were fixed in 10% buffered formalin and stained with hematoxylin and eosin. This figure demonstrates the rare PMNs observed in SpA-injected sites from rabbits pretreated with saline. Results are representative of duplicate injections from three rabbits. The Journal of Immunology 5249

ϩ ϩ Animals pretreated with VH3 /non-VH3 hIgG showed erythema, edema, and hemorrhage only at the SpA-injected, i.d. sites upon gross examination (Fig. 4, A and B). In contrast, animals Ϫ pretreated with VH3 hIgG demonstrated no erythema or edema at ϩ any injected sites (Fig. 4C). Thus, the depletion of VH3 IgGs from the hIgG preparation abrogated the macroscopic signs of an Arthus reaction at SpA-injected, i.d. sites. As expected from the gross examination, the SpA-injected sites ϩ ϩ in the VH3 /non-VH3 hIgG pretreated rabbits demonstrated a prominent PMN inflammatory response (Fig. 5, A and B). Dense infiltrates of PMNs were observed in a perivascular distribution surrounded by areas of hemorrhage. Again, the arterioles were not involved. The HSA-injected sites contained few PMNs in the lumens of the venules (Fig. 5C), while no PMNs were seen at the saline-injected sites (data not shown). FIGURE 3. Binding of IgG preparations to Mod SpA. hIgG was frac- ϩ ϩ In contrast to the results observed in V 3 /non-V 3 hIgG tionated over a Mod SpA-Seph column or a Seph column. The binding of H H pretreated rabbits, only rare inflammatory cells were seen at the these IgG fractions and unfractionated hIgG to Mod SpA was analyzed in Ϫ an ELISA as described in Materials and Methods. Various concentrations SpA-injected sites of animals pretreated i.v. with VH3 hIgG (Fig. of hIgG (ⅷ), the IgG fraction from the Seph column (⅜), or the IgG 6). The rare PMNs were confined to the lumens of the venules. The Downloaded from fraction from the Mod SpA-Seph column (f) were added to Mod SpA- HSA-injected sites showed rare PMNs in and around the venules, coated ELISA plates. Background absorbance was determined in Mod while no PMNs were seen at the saline-injected sites (data not SpA-coated wells in which no IgG proteins were incubated, and this ab- shown). These results indicate that SpA elicits a prominent Arthus- sorbance was subtracted from the raw data. OD values represent the means like reaction in rabbits only when they are pretreated with hIgG. of duplicate determinations from a representative experiment; similar re- Moreover, this inflammatory response is dependent upon the pres- sults were obtained for all fractionated IgG preparations. ϩ

ence of VH3 IgGs in the hIgG preparation. http://www.jimmunol.org/ population of IgGs. In addition, passage of hIgG over a Seph col- Discussion ϩ umn did not remove VH3 IgGs nor did it nonspeciﬁcally degrade The studies reported herein were undertaken to determine whether the IgG molecules. the interaction of a B cell SAg with its reactive IgG molecules

ϩ leads to immune complex-mediated tissue injury. Although the B Requirement of human VH3 IgGs for the SpA-induced Arthus cell SAgs interact with variable region determinants on Ig mole- reaction cules outside the complementarity-determining regions (7–9), the As shown in Figure 1, SpA induced an Arthus reaction in rabbits complexes that are formed can activate the classical complement pretreated with hIgG. To determine whether this reaction occurred cascade (26) and should be able to give rise to tissue inflammation. by guest on October 1, 2021 via a B cell superantigenic mechanism and consequently was de- Unlike a conventional Ag, a B cell SAg can react with sizable ϩ pendent on human VH3 IgGs, four groups of rabbits (two rabbits amounts of Ig molecules. Immunization is not a prerequisite for Ϫ per group) were injected i.v. with 170 mg of 1) VH3 hIgG, 2) this reaction to occur, since the reactive VH or variable light chain ϩ ϩ hIgG that had been passed over a Seph column (VH3 /non-VH3 Abs are found in the standing pool of serum Igs in the host. Ac- hIgG), 3) unfractionated hIgG, and 4) saline, respectively. At 10 to cordingly, there is great potential for a B cell SAg to engender 15 min after the i.v. injections, all animals were challenged i.d. at immune complex-mediated tissue injury. duplicate sites with 900 ␮g of SpA, HSA, and saline, respectively. We chose to examine the Arthus reaction as an in vivo model of After 8 h, the skin was examined grossly, and biopsies were ob- immune complex-mediated tissue inflammation. Although this tained for histologic analysis. The histologic results from repre- model of inflammation was formerly thought to be mediated by sentative skin biopsies from hIgG- and saline-pretreated rabbits are complement products liberated by the action of immune com- depicted in Figures 1 and 2, as described above. plexes (28, 29, 31), recent studies have underscored the importance

FIGURE 4. Hemorrhage in an 8-h Arthus reaction. A and B, Rabbits were pretreated i.v. with 170 mg of hIgG fractionated over a Seph column, and 900 ␮g of SpA (A), HSA (B), or saline (data not shown) was administered i.d. 10 to 15 min later. C, Rabbits were pretreated i.v. with 170 mg of hIgG fractionated over a Mod SpA-Seph column, and 900 ␮g of SpA (C), HSA (data not shown), or saline (data not shown) was administered i.d 10 to 15 later. The shaved dorsal skin of all rabbits was examined after 8 h. The small dark spots observed on the rabbit skin were used to mark the site of the i.d. injections. Results are representative of duplicate injections in two rabbits per each type of IgG administered. 5250 INDUCTION OF AN ARTHUS REACTION BY A B CELL SUPERANTIGEN

cause an Arthus reaction. However, when the rabbits were pretreated i.v. with hIgG, an Arthus reaction occurred at SpA-injected, i.d. sites. The investigators concluded that the Arthus reaction was induced via an interaction between SpA and the Fc region of hIgG molecules. However, the authors failed to explain why an interaction with hIgG Fc, but not rabbit IgG Fc, would lead to an Arthus reaction. One explanation for the inability of SpA to elicit an Arthus reaction in an untreated rabbit is that the binding of SpA to residues in the rabbit IgG Fc region blocks the binding sites for C1q and Fc␥R. The C1q-binding site includes residues 285 to 340 in the constant heavy chain 2 domain of IgG Fc (40–43). This region overlaps the SpA-binding site on IgG Fc (44), although it does not bind directly to the same residues. The Fc␥R-binding site on IgG Fc includes residues 234 to 237 in the lower hinge region and residue 331 in the constant heavy chain 2 domain (45, 46). These residues do not overlap the SpA-binding site, but the binding of SpA to IgG Fc may still sterically hinder the binding of the IgG

molecule to Fc␥R. If this explanation is correct for rabbit IgGs, Downloaded from then it should also apply to hIgGs. However, SpA does elicit an Arthus reaction when rabbits are pretreated with hIgGs. Thus, it is unlikely that SpA blocks the binding site on IgG Fc for Fc␥Ror C1q, at least not on hIgG molecules. In our opinion, a more appealing explanation is that SpA only

induces an Arthus reaction when it interacts with the Fab region of http://www.jimmunol.org/ ϩ reactive IgGs, such as human VH3 IgGs. SpA binds to IgG molecules from most species via their Fc and Fab regions. However, the rabbit is one of the few species in which SpA binds only to the Fc region of its IgG molecules (47, 48). Thus, it is possible that administrating SpA to a rabbit does not induce an Arthus reaction, because SpA cannot bind to the Fab region of rabbit IgGs. On the other hand, SpA elicits an Arthus reaction in rabbits pretreated with hIgGs, presumably because it can bind to the Fab region of a ϩ by guest on October 1, 2021 large proportion of hIgGs, namely VH3 IgGs. Proof of this hy- pothesis would strongly suggest that a B cell SAg can trigger an in vivo immune complex-mediated inflammatory response. FIGURE 5. Arthus reaction as seen in histologic sections. Rabbits were Accordingly, our initial experiments were performed to confirm pretreated i.v. with 170 mg of hIgG fractionated over a Seph column, and that SpA elicits an Arthus reaction in rabbits pretreated with hIgG 900 ␮g of SpA (A, ϫ39; B, ϫ93.7), HSA (C, ϫ31.2), or saline (data not but does not cause an Arthus reaction in untreated rabbits (30). We shown) were administered i.d 10 to 15 min later. The animals were sacri- injected hIgG or saline i.v. into several rabbits and then injected ficed after 8 h. Skin biopsies were fixed in 10% buffered formalin and the rabbits i.d. with SpA, HSA, or saline. Only animals pretreated stained with hematoxylin and eosin. A shows a prominent PMN infiltration with hIgG and subsequently injected i.d. with SpA developed er- ϫ in the venules and hemorrhage. B is a 125 magnification of the boxed ythema and edema at injected sites. Histologic analysis of the area in A, demonstrating an intense PMN infiltration with diapedesis of grossly inflamed, SpA-injected sites revealed a prominent inflam- erythrocytes. C shows rare PMNs in the venules. Results are representative matory response characterized by infiltrating PMNs (Fig. 1, A–C) of duplicate injections from two rabbits. and dermal hemorrhage. Animals pretreated with saline had rare luminal PMNs only at SpA- (Fig. 2) and HSA-injected sites. Thus, of Fc␥RIII-bearing cells (32–34). In particular, mast cells, neutro- these results demonstrate that SpA that is i.d.-administered to a phils, macrophages, and Langerhans cells were incriminated as rabbit pretreated with hIgG induces an inflammatory response that critical Fc␥RIII-bearing effector cells (33). These studies also sug- has the classical histologic features of an Arthus reaction (30). In gested that complement-activation products likely augment the addition, our results confirm that i.d. administration of SpA to Arthus reaction. We decided to investigate the Arthus-inducing untreated rabbits does not lead to the development of such an in- potential of SpA. Unlike other B cell SAgs, SpA has two Ig-bind- flammatory reaction. ing sites. One site binds to the Fc region of IgG, while the second To prove that this Arthus reaction reflected a B cell superanti- ϩ site binds to VH region residues on VH3 human Igs indepen- genic property of SpA, it was necessary to show that it was me- ϩ dently of their heavy chain isotype. It is this second site that en- diated by VH3 IgGs in the hIgG preparation. Therefore, we pre- Ϫ Ϫ dows SpA with its B cell superantigenic properties. treated rabbits with VH3 hIgG. The VH3 hIgG fraction had a ϩ Before the characterization of this Ig Fab-binding site on SpA, 100-fold reduction in Mod SpA-binding (VH3 ) IgG when com- ϩ ϩ several groups investigated the similarities between conventional pared with unfractionated hIgG and a VH3 /non-VH3 hIgG frac- Ag/Ab complexes and SpA/IgG complexes (35–39). Indeed, one tion which was passed over a Seph column (Fig. 3). Moreover, we Ϫ group hypothesized that SpA/IgG complexes could elicit an Arthus demonstrated that the VH3 hIgG fraction had an equivalent reaction based on the observation that Ag/Ab complexes and ag- amount of SpA-binding IgG molecules as compared with unfrac- ϩ ϩ gregated IgG elicit such an inflammatory reaction (30). These in- tionated hIgG and the VH3 /non-VH3 hIgG fraction (data not vestigators found that i.d. injections of SpA into a rabbit did not shown). The marked reduction in Mod SpA binding of this fraction The Journal of Immunology 5251

FIGURE 6. Lack of an Arthus reaction at the SpA-injected site in a rabbit pretreated Ϫ with VH3 hIgG and SpA. Rabbits were pretreated i.v. with 170 mg of hIgG fractionated over a Mod SpA-Seph column, and 900 ␮gof SpA (ϫ31.2), HSA (data not shown), or saline (data not shown) was administered i.d 10 to 15 min later. The animals were sacrificed after 8 h. Skin biopsies were fixed in 10% buffered formalin and stained with hematoxylin and eosin. This figure demonstrates the rare PMNs observed in SpA-injected sites Ϫ from rabbits that were pretreated with VH3 hIgG. Results are representative of duplicate injections from two rabbits. Downloaded from

ϩ attested to the completeness of the VH3 IgG depletion. Following Arthus reaction. It is likely that this reaction occurs via an inter- ϩ this pretreatment, we injected the rabbits i.d. with SpA, HSA, and action of SpA and the Fabs of human VH3 IgGs. Thus, these data

saline. The cutaneous responses were compared with the reactions provide the first evidence that the interaction of a B cell SAg, SpA, http://www.jimmunol.org/ ϩ Ϫ ϩ in animals pretreated with VH3 /VH3 IgG, hIgG, or saline. with its reactive (VH3 ) Igs leads to an inflammatory reaction in ϩ Inspection of the skin of the animals pretreated with VH3 /non- vivo. Such an in vivo response to a B cell SAg could have pro- ϩ VH3 hIgG revealed erythema, edema, and hemorrhage only at the found clinical significance. For example, SpA immunoadsorption SpA-injected, i.d. sites (Fig. 4, A and B). In contrast, no erythema has been used as a novel therapy in a variety of diseases that are Ϫ or edema was observed at any injected sites from the VH3 hIgG- mediated by pathogenic autoantibodies (50, 51). In addition, the pretreated animals (Fig. 4C). These results demonstrated that SpA therapy is being investigated as a treatment for cancer, thrombotic ϩ required the presence of human VH3 IgGs to elicit the macro- thrombocytopenic purpura, hemolytic uremic syndrome, myasthe- scopic appearance of an Arthus reaction. Histologic examination nia gravis, and HIV (52–54). However, a significant number of ϩ ϩ by guest on October 1, 2021 of skin biopsies from rabbits pretreated with VH3 /non-VH3 these patients have developed severe adverse side effects, most hIgG demonstrated a prominent inflammatory cell infiltrate and notably leukocytoclastic vasculitis (52, 55). Histologically, the dermal hemorrhage only at SpA-injected sites (Fig. 5, A–C). In dermal lesions are characterized by superficial and deep perivas- Ϫ striking contrast, the VH3 hIgG-pretreated rabbits contained only cular infiltrates containing mostly neutrophils (52). In addition, minimal PMN infiltration at the SpA-injected sites (Fig. 6). These diapedesis of erythrocytes was observed. These effects could be ϩ results demonstrate that SpA is only able to elicit a prominent due to an interaction between SpA/VH3 Igs and the formation of inflammatory response in rabbits that are pretreated with hIgG immune complexes that bind to and activate Fc␥R-bearing cells, ϩ containing VH3 IgGs. Even though SpA can bind with high af- such as macrophages and neutrophils. Further studies will be nec- finity to rabbit IgG Fc, the Arthus reaction does not occur when essary to determine whether this type of inflammatory reaction is ϩ human VH3 IgGs are not present. Therefore, an interaction be- associated with the in vivo response to other B cell SAgs. tween SpA, as a B cell SAg, and its reactive hIgGs is necessary to cause tissue inflammation in vivo. Acknowledgments Although the rabbit IgG does not appear to play a role by itself, it is possible that it might affect the complexes formed between We thank Julie Burns of the Laboratory of Pathology, University of Penn- SpA and hIgG. SpA could form complexes with IgG molecules by sylvania School of Veterinary Medicine for expert technical assistance with the histologic procedures. We also thank the veterinary technicians of binding to the Fab and Fc regions of hIgGs, in addition to the Fc the University Laboratory Animal Resources Department for assistance region of rabbit IgG. Such complexes would be large and possibly with the i.v. and i.d. injections into the rabbits, Jeanette Tasey and Paul very stable, since SpA binds to rabbit IgG Fc with high affinity (Ka Barrow of the University of Pennsylvania Biomedical Communications 9 10 Ϫ1 ϭ 10 -10 M ) (49). The presence of rabbit IgG might enhance Department for their excellent photography, and Dr. John Monroe for his the binding of these complexes to both rabbit Fc␥R and rabbit critical review of this manuscript. complement components. However, this explanation also suggests that SpA may need to bind to both the Fab and Fc regions of IgG References molecules to cause an inflammatory response, or that the additional binding of SpA to the rabbit IgG Fc region may not be required but 1. Inganas, M., S. G. O. Johansson, and H. H. Benninch. 1980. Interaction of human polyclonal IgE and IgG from different species with protein A from Staphylococ- Ј may enhance the response. Additional studies are required to de- cus aureus: demonstration of protein A-reactive sites located on the F(ab )2 frag- termine whether the Fc region of reactive or nonreactive IgG mol- ments of human IgG. Scand. J. Immunol. 12:23. 2. Inganas, M. 1981. Comparison of mechanisms of interaction between protein A ecules is needed, in addition to the Fab region of reactive IgGs, to from Staphylococcus aureus and human monoclonal IgG, IgA, and IgM in rela- ␥ Ј ⑀ interact with SpA and subsequently lead to tissue inflammation tion to the classical Fc and the alternative F(ab )2 protein A interactions. Scand. in vivo. J. Immunol. 13:343. Nevertheless, the results described in this paper demonstrate that 3. Young, W. W., Jr., Y. Tamura, D. M. Wolock, and J. W. Fox. 1984. Staphylo- ϩ coccal protein A binding to the Fab fragments of mouse monoclonal antibodies. SpA requires the presence of human VH3 IgGs to induce an J. Immunol. 133:3163. 5252 INDUCTION OF AN ARTHUS REACTION BY A B CELL SUPERANTIGEN

4. Vidal, M. A., and F. P. Conde. 1985. Alternative mechanism of protein A-im- 31. Cochrane, C. G. 1967. The Arthus phenomenon—a mechanism of tissue damage. munoglobulin interaction: The VH-associated reactivity of a monoclonal human Arthritis Rheum. 10:392. IgM. J. Immunol. 135:1232. 32. Sylvestre, D. L., and J. V. Ravetch. 1994. Fc receptors initiate the Arthus reac- 5. Romagnani, S., M. G. Guidizi, R. Biagiotti, F. Almerigogna, E. Maggi, tion: redefining the inflammatory cascade. Science 265:1095. G. Del Prete, and M. Ricci. 1981. Surface immunoglobulins are involved in the 33. Sylvestre, D. L., and J. V. Ravetch. 1996. A dominant role for mast cell Fc interaction of protein A with human B cells and in the triggering of B cell pro- receptors in the Arthus reaction. Immunity 5:387. liferation induced by protein A-containing Staphylococcus aureus. J. Immunol. 34. Hazenbos, W. L. W., J. E. Gessner, F. M. A. Hofhuis, H. Kuipers, D. Meyer, 127:1307. I. A. F. M. Heijnen, R. E. Schmidt, M. Sandor, P. J. A. Capel, M. Daeron, 6. Romagnani, S., M. G. Guidizi, G. Del Prete, E. Maggi, R. Biagiotti, J. G. J. van de Winkel, and J. S. Verbeek. 1996. Impaired IgG-dependent ana- F. Almerigogna, and M. Ricci. 1982. Demonstration on protein A of two distinct phylaxis and Arthus reaction in Fc␥RIII (CD16)-deficient mice. Immunity 5:181. Ig-binding sites and their role in the mitogenic activity of Staphylococcus aureus. 35. Forsgren, A., and J. Sjoquist. 1966. “Protein A” from Staphylococcus aureus: J. Immunol. 129:596. Pseudo-immune reaction with human ␥-globulin. J. Immunol. 97:822. 7. Hillson, J., N. S. Karr, I. Oplinger, M. Mannik, and E. H. Sasso. 1993. The 36. Langone, J. J., M. D. P. Boyle, and T. Borsos. 1978. Studies on the interaction structural basis of germline-encoded VH3 immunoglobulin binding to staphylo- between protein A and immunoglobulin I. Effect of protein A on the functional coccal protein A. J. Exp. Med. 178:331. activity of IgG. J. Immunol. 121:327. 8. Randen, I., K. Potter, Y. Li, K. M. Thompson, V. Pascual, O. Forre, J. B. Natvig, 37. Langone, J. J., M. D. P. Boyle, and T. Borsos. 1978. Studies on the interaction and J. D. Capra. 1993. Complementarity-determining region 2 is implicated in the between protein A and immunoglobulin II. Composition and activity of com- binding of staphylococcal protein A to human immunoglobulin VHIII variable plexes formed between protein A and IgG. J. Immunol. 121:333. regions. Eur. J. Immunol. 23:2682. 38. Hanson, D. C., and V. N. Schumaker. 1984. A model for the formation and Ј 9. Sasso, E. H., G. J. Silverman, and M. Mannik. 1991. Human IgA and IgG F(ab )2 interconversion of protein A-immunoglobulin G soluble complexes. J. Immunol. that bind to staphylococcal protein A belong to the VHIII subgroup. J. Immunol. 132:1397. 147:1877. 39. Ghetie, V., G. Mota, M.-A. Dobre-Ghetie, M. Laky, A. Olinescu, S. Dima, and 10. Sasso, E. H., G. J. Silverman, and M. Mannik. 1989. Human IgM molecules that J. Sjoquist. 1986. Modulation of IgG effector functions by a monovalent fragment bind staphylococcal protein A contain VHIII H chains. J. Immunol. 142:2778. of staphylococcal protein A. Mol. Immunol. 23:377. 11. Zouali M., and J. Theze. 1991. Probing VH gene family utilization in human 40. Burton, D. R., J. Boyd, A. D. Brampton, S. B. Easterbrook-Smith, E. J. Emanuel,

peripheral B cells by in situ hybridization. J. Immunol. 146:2855. J. Novotny, T. W. Rademacher, M. R. van Schravendijk, M. J. E. Sternberg, and Downloaded from 12. Huang, C., A. K. Stewart, R. S. Schwartz, and B. D. Stollar. 1992. Immunoglob- R. A. Dwek. 1980. The C1q receptor site on immunoglobulin G. Nature 288:338. ulin heavy chain gene expression in peripheral blood B lymphocytes. J. Clin. 41. Feinstein, A., N. Richardson, and M. J. Taussig. 1986. Immunoglobulin flexibil- Invest. 89:1331. ity in complement activation. Immunol. Today 7:169. 13. Huang, C., and B. D. Stollar. 1993. A majority of Ig H chain cDNA of normal 42. Tao, M.-H., S. M. Canfield, and S. L. Morrison. 1991. The differential ability of human adult blood lymphocytes resembles cDNA for fetal Ig and natural auto- human IgG1 and IgG4 to activate complement is determined by the COOH- antibodies. J. Immunol. 151:5290. terminal sequence of the CH2 domain. J. Exp. Med. 173:1025. 14. Silverman, G. J., M. Sasano, and S. B. Wormsley. 1993. Age-associated changes 43. Tao, M.-H., R. I. F. Smith, and S. L. Morrison. 1993. Structural features of human in binding of human lymphocytes to a V 3-restricted unconventional bacterial immunoglobulin G that determine isotype-specific differences in complement ac-

H http://www.jimmunol.org/ antigen. J. Immunol. 151:5840. tivation. J. Exp. Med. 178:661. 15. Seppala, I., M. Kaartinen, S. Ibrahim, and O. Makela. 1990. Mouse Ig coded by 44. Deisenhofer, J. 1981. Crystallographic refinement and atomic models of a human VH families S107 and J606 bind to protein A. J. Immunol. 145:2989. Fc fragment and its complex with fragment B of protein A from Staphylococcus 16. Zouali, M. 1995. B cell superantigens: implications for selection of the human aureus at 2.9- and 2.8-A resolution. Biochemistry 20:2361. antibody repertoire. Immunol. Today 16:399. 45. Canfield, S. M., and S. L. Morrison. 1991. The binding affinity of human IgG for 17. Sasano, M., D. Burton, and G. Silverman. 1993. Molecular selection of human its high affinity Fc receptor is determined by multiple amino acids in the CH2 antibodies with an unconventional bacterial B cell antigen. J. Immunol. 151:5822. domain and is modulated by the hinge region. J. Exp. Med. 173:1483. 18. Silverman, G. J. 1997. B cell superantigens. Immunol. Today 18:379. 46. Sarmay, G., J. Lund, Z. Rozsnyay, J. Gergely, and R. Jefferis. 1992. Mapping and 19. Berberian, L., L. Goodglick, T. J. Kipps, and J. Braun. 1993. Immunoglobulin comparison of the interaction sites on the Fc region of IgG responsible for trig- VH3 gene products: natural ligands for HIV gp120. Science 261:1588. gering antibody-dependent cellular cytotoxicity through different types of human 20. Silverman, G. J., P. Roben, J.-P. Bouvet, and M. Sasano. 1995. Superantigen Fc␥ receptor. Mol. Immunol. 29:633. ␥

properties of a human sialoprotein involved in gut-associated immunity. J. Clin. 47. Biguzzi, S. 1982. Fc -like determinants on immunoglobulin variable regions: by guest on October 1, 2021 Invest. 96:417. identiﬁcation by staphylococcal protein A. Scand. J. Immunol. 15:605. Ј 21. Bjorck, L. 1988. Protein L: a novel bacterial cell wall protein with afﬁnity for Ig 48. Erntell, M., E. B. Myhre, and G. Kronvall. 1986. Non-immune F(ab )2- and L chains. J. Immunol. 140:1194. Fc-mediated interactions of mammalian immunoglobulins with S. aureus and 22. Domiati-Saad, R., J. F. Attrep, H.-P. Brezinschek, A. H. Cherrie, D. R. Karp, and group C and G streptococci. Acta Path. Microbiol. Immunol. Scand. Sect. B P. E. Lipsky. 1996. Staphylococcal enterotoxin D functions as a human B cell Microbiol. 94:377. superantigen by rescuing VH4-expressing B cells from apoptosis. J. Immunol. 49. Boyle, M. D. P. 1990. The type I bacterial immunoglobulin-binding protein: 156:3608. staphylococcal protein A. In Bacterial Immunoglobulin-Binding Proteins, Vol. 1. 23. Marone, G., M. Tamburini, M. G. Giudizi, R. Biagiotti, F. Almerigogna, and M. D. P. Boyle, ed. Harcourt Brace Jovanovich, New York, p. 17. S. Romagnani. 1987. Mechanism of activation of human basophils by Staphylo- 50. Guthrie, T. H., Jr., and A. Oral. 1989. Immune thrombocytopenia purpura: a pilot coccus aureus Cowan I. Infect. Immun. 55:803. study of staphylococcal protein A immunomodulation in refractory patients. Se- 24. Patella, V., V. Casolaro, L. Bjorck, and G. Marone. 1990. Protein L: a bacterial min. Hematol. 26:3. Ig-binding protein that activates human basophils and mast cells. J. Immunol. 51. Jones, F. R., J. P. Balint, and H. W. Snyder. 1986. Selective extracorporeal 145:3054. removal of immunoglobulin G and circulating immune complexes: a review. 25. Patella, V., J.-P. Bouvet, and G. Marone. 1993. Protein Fv produced during viral Plasma Ther. Transfus. Technol. 7:333. hepatitis is a novel activator of human basophils and mast cells. J. Immunol. 52. Ainsworth, S. K., P. A. Pilla, S. H. Pepkowitz, and P. O’Brien. 1988. Toxicity 151:5685. following protein A treatment of metastatic breast adenocarcinoma. Cancer 61: 26. Kozlowski, L. M., A. M. Soulika, G. J. Silverman, J. D. Lambris, and 1495. A. I. Levinson. 1996. Complement activation by a B cell superantigen. J. Immu- 53. Messerschmidt, G. L., D. H. Henry, H. W. Snyder, Jr., J. Bertram, A. Mittelman, nol. 157:1200. S. Ainsworth, J. Fiore, M. V. Viola, J. Louie, E. Ambinder, F. R. MacKintosh, 27. Arthus, M. 1903. Repeated injections of horse serum in the rabbit. C. R. Soc. Biol. D. J. Higby, P. O’Brien, D. Kiprov, M. Hamberger, J. P. Balint, Jr., L. D. Fisher, 55:817. W. Perkins, C. M. Pinsky, and F. R. Jones. 1989. Protein A immunotherapy in the 28. Humphrey, J. H. 1955. The mechanism of the Arthus reaction I. The role of treatment of cancer: an update. Semin. Hematol. 26:19. polymorphonuclear leukocytes and other factors in reversed passive Arthus re- 54. Mittelman, A., J. Bertram, D. H. Henry, H. W. Snyder, Jr., G. L. Messerschmidt, action in rabbits. Brit. J. Exp. Pathol. 36:268. D. Ciavarella, S. Ainsworth, D. Kiprov, and Z. Arlin. 1989. Treatment of patients 29. Cochrane, C. G., W. O. Weigle, and F. J. Dixon. 1959. The role of polymorpho- with HIV thrombocytopenia and hemolytic uremic syndrome with protein A nuclear leukocytes in the initiation and cessation of the Arthus vasculitis. J. Exp. (Prosorba column) immunoadsorption. Semin. Hematol. 26:15. Med. 110:481. 55. Arbiser, J. L., J. S. Dzieczkowski, J. V. Harmon, and L. M. Duncan. 1995. 30. Gustafson, G. T., J. Sjoquist, and G. Stalenheim. 1967. “Protein A” from Staph- Leukocytoclastic vasculitis following staphylococcal protein A column immuno- ylococcus aureus II. Arthus-like reaction produced in rabbits by interaction of adsorption therapy: two cases and a review of the literature. Arch. Dermatol. protein A and human ␥-globulin. J. Immunol. 98:1178. 131:707.