Eur. J. Immunol. 1989.19: 441-446 Modulation of the murine IgE response by interferon-gamma 441

Michael Hagen', Naeem A. EssaniA and Gill H. Strejan

Role of interferon-gamma in the modulation of the IgE response by 2,4-dinitrophenyl-BordeteZZa pertussis vaccine in the mouse* Department of Microbiology and

Immunology, University of Western Ontario, London Immunization of mice with 2,4-dinitrophenyl-Bordetellu pertussis (DNP-BP) failed to

induce anti-DNP IgE responses. Administration of DNP-BP induced, however, the formation of anti-DNP IgE B memory cells, as demonstrated by adoptive transfer. Furthermore, mice pretreated with DNP-BP and primed with 2 pg DNP-ovalbumin (OA) in alum 2 weeks later produced high day-7 anti-DNP IgE levels. These subsided to near undetectable levels by day 12-14. The transient expression of serum IgE levels was accompanied by normal levels of anti-DNP IgG. The anti-OA response induced as a result of priming with DNP-OA in alum was not affected by pretreat- ment with DNP-BP. IgG subclass analysis revealed that mice pretreated with DNP- BP had elevated levels of IgGza and reduced levels of IgGl as compared to control (TNP-keyhole limpet hemocyanin-pretreated) mice. Treatment of mice with an anti- interferon-y monoclonal antibody, shortly after immunization with DNP-BP, not only reduced anti-DNP IgGza levels, but prevented the sharp anti-DNP IgE decline that occurred after priming with DNP-OA in alum. These results suggest that DNP-BP- induced interferon-y production modulates Ig isotype expression in vivo in an anti- gen-specific manner.

1 Introduction

It is widely accepted that the IgE antibody response is under strict T cell regulatory influences. Work from several laboratories led to the formulation of two distinct views on how this immunoregulation operates. One has developed as a result of the isolation and characterization of T cell-derived IgE-binding factors (IgE-BF), which either potentiate or sup- press IgE responses (reviewed in [l, 21, whereas the other invokes immunoregulation through a mechanism of isotype switching induced by distinct subpopulations of helper T (Th) cells. TH1 cells produce interferon-y (IFN-y) which is directly involved in selectively stimulating IgG2, and inhibiting IgGl and IgE synthesis by B cells, while TH2 cells produce inter- leukin 4 (IL4) and IL5 which stimulate, IgGl, IgE and IgA, and inhibit IFN-y-induced IgG2, synthesis (reviewed in [3, 41).

In experimental animals the production of IgE antibodies reaches detectable serum levels only if the immunogen is administered with Bordetellu pertussis (BP) vaccine or Al(0H)S (alum) as adjuvants. On the other hand the adminis- tration of antigens in association with complete Freund's

[I 71391

* This work was supported by a grant from the Medical Research Council of Canada. Current address: Department of Pathology, University of Iowa, College of Medicine, Iowa City, I 0 52242, USA.

A Current address: Department of Immunology, St. Jude Children's Research Hospital, Memphis, TN 3801, USA.

Correspondence: Gill H. Strejan, Department of Microbiology and Immunology, University of Western Ontario, London, Ontario, Canada N6A 5C1

Abbreviations: B P Bordetella pertussis mAb: Monoclonal antibody(ies) KLH: Keyhole limpet hemocyanin BSA Bovine serum albumin O A Ovalbumin TNP 2,4,6-Trinitrophenyl PCA: Passive cutaneous anaphylaxis Tb: Helper T (cell) IFN-1: Interferon-y

adjuvant promotes good delayed hypersensitivity and IgG antibody responses but poor IgE production (reviewed in [5, 61). The administration to mice of DNP-conjugated mycobacteria, an essential component of CFA, resulted in the production of T suppressor (TJ cells and soluble suppressor molecule's, selective for the IgE isotype [7].

The initiation of the present experiments was based on the premise that coupling of hapten to an adjuvant known to potentiate IgE production might result in enhanced anti-DNP IgE antibody production. Here we show that contrary to this prediction, mice immunized with DNP-BP failed to produce detectable serum levels of anti-DNP IgE antibodies. The ina- bility to stimulate IgE production coincided with the appear- ance in the serum of very high levels of anti-DNP IgG2, and depressed IgGl antibodies.

2 Materials and methods

2.1 Animals

C B N J mice (8-12 weeks old) were purchased from Jackson Laboratories (Bar Harbor, ME) and bred in our animal facility by brotherkister matings. Sprague-Dawley female rats (retired breeders) were obtained from Charles River Canada, Ottawa, Ontario, Canada.

2.2 Antigens, lymphokines and antibodies

Bovine serum albumin (BSA), fraction V, was purchased from Sigma Chemical Co. (St. Louis, MO), crystallized oval- bumin (OA) was from Miles Laboratories (Rexdale, Ontario, Canada) and keyhole limpet hemocyanin (KLH) was from Biomarine Pacific Supply Co. (Venice, CA). 2,CDinitroben- zene sulfonic acid, sodium salt (Eastman-Kodak Co., Roches- ter, NY) was coupled to proteins as previously described [8]. DNPzzBSA and DNPsOA were stored at -20 "C. 2,4,6-trini- trophenyl (TNP)&LH was stored at 4°C. The subscripts

0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1989 0014-2980/89/0303-01$02.50/0

442

refer to the average number of mol DNP/mol of carrier, assuming a molecular mass of 40 kDa for OA, 72 kDa for BSA and 800 kDa for the dissociated form of KLH. BP vac- cine (Armand Frappier Institute, Laval, Quebec, Canada) contained 20 OU/ml (U.S. Standard). Ascaris suum antigen was prepared in this laboratory as previously described [9]. DNP-BP was prepared according to Kishimoto et al. [lo]. Briefly, BP vaccine containing 4.5 x 10" killed organisms was centrifuged at 10000 x g for 20 min. The pellet was resus- pended in cold isotonic saline and the cells were washed by centrifugation five times at 4°C. The washed organisms were resuspended in 0.25 mlO.1 M Na2C03 to which 2.5 p1 l-fluoro- 2,4-dinitrobenzene (Eastman-Kodak Co.) was added and the mixture was stirred gently at 37 "C for 60 min, protected from light. The cells were then washed to remove unreacted hapten, suspended in borate-buffered saline, pH 8.4, and stored at 4°C at a concentration of 2 x 10'' organisms/ml. The number of DNP groups was determined by lyzing the cells in boiling 1 N NaOH and determining the absorbance at 360 nm. Assuming a molar absorptivity of 17500, there were 5 x 106-20 x lo6 DNP groupdcell. Affinity-purified rat IgG, monoclonal antibody (mAb) specific for IFN-y (XMG6) and for (4-hydroxy-3-nitropheny1)acetyl (NP;J4.1) were kindly provided by Dr. F. Finkelman, USUHS, Bethesda, MD.

M. Hagen, N. A. Essani and G. H. Strejan Eur. J. Immunol. 1989.19: 441-446

2.3 Immunization

Groups of 4-6 mice were injected i.p. with lo9 DNP-BP (50 pg dry weight equivalent) alone or mixed with 2 mg Al(OH)3, in a 0.5-ml volume. Mice injected with other hapten-carrier con- jugates received 50 pg antigen in saline or 2 pg antigen mixed with 2.5 mg alum. To examine the effects of IFN-y on the in vivo anti-DNP IgE and IgG responses, groups of 4 mice were immunized with lo9 DNP-BP i.p., followed 36 h later by the i.v. injection of 1 mg mAb XMG6, J4.1 or saline. The mice were boosted 12 days later with 2 pg DNP-OA in alum. Mice were bled by cardiac puncture at various intervals, the blood from all animals in a group was pooled and the sera were stored at -20°C until needed. Additional details are given in Sect. 3.

2.4 Antibody determinations

IgE antibodies were measured on rats by passive cutaneous anaphylaxis (PCA) with a 48-h latent period between the i.d. injection of serum and the antigen challenge [8]. Fourfold dif- ferences in PCA titers between serum samples were con- sidered statistically significant. IgG antibodies and subclass determinations were made utilizing an enzyme immunoassay as previously described [ 111. Affinity-purified biotin-labeled goat anti-mouse IgG (Fc specific; Jackson Immunoresearch, BiolCan Scientific, Mississauga, Ontario, Canada) was utilized for the detection of antigen-specific IgG levels whereas the IgG isotype distribution in serum samples was determined using affinity-purified goat anti-mouse IgGl or IgG2, (South- ern Biotechnology Associates, Birmingham, AL), followed by incubation with biotinylated mouse anti-goat IgG (Jackson) and alkaline phosphatase-labeled streptavidin. All plates were read in a Titertek Multiscan plate reader (Flow Laboratories, Mississauga, Ontario, Canada) at 405 nm. Total anti-DNP IgG was expressed in pg/ml following extrapolation from a curve constructed with a mo&e anti-DNP serum standard. Anti-DNP IgGl and IgG2, antibodies were expressed in arbi-

trary units per well. One unit was defined as equal to an absorbance reading 405 nm of twice background. Therefore, X units equal the dilution of the serum giving an b5 reading of twice background, multiplied by that reading. Background readings ranged between 0.08 and 0.15.

2.5 Adoptive transfer

Mouse spleens were removed aseptically and single-cell sus- pensions were prepared in Hank's balanced salt solution. Cells were washed by centrifugation, treated with 0.85% ammonium chloride to remove red blood cells and counted. Viability was determined by trypan blue exclusion. Donor mice were injected with lo9 DNP-BP 4 weeks prior to transfer, or with 0.5 pg OA in alum one week prior to transfer. On the day of transfer, 2 x lo7 cells from each type of donor were mixed and transferred to X-irradiated (660 rads) syngeneic recipients, which were then boosted with 2 pg DNP-OA in alum.

3 Results

3.1 Immunogenicity of DNP-BP

To determine whether DNP-BP could stimulate anti-hapten IgE responses, mice were injected i.p. with lo9 DNP-BP (50 pg) either alone or adsorbed onto 2.5 mg alum. Fig. 1 shows that DNP-BP failed to stimulate a detectable primary anti-DNP IgE response. Low PCA titers were detected in sera obtained 7 days after booster (day 35) only in the group immunized with DNP-BP in alum; these decreased to unde- tectable levels by day 14 after booster (day 42). In contrast, IgG antibodies were elevated in both groups. For comparison, the PCA and IgG levels of sera from mice immunized with 2 pg DNP-OA in alum are also shown. No IgE antibodies to BP were detected, and IgG anti-BP antibodies were not deter- mined.

The inability of DNP-BP to stimulate an IgE response could not be attributed to lack of immunogenicity of the conjugate since an excellent IgG response was detected. Moreover, as seen in Fig. 2, the adoptive transfer of spleen cells from DNP- BP-primed donors together with cells from OA-primed donors to X-irradiated recipients resulted in a typical secondary IgE response (PCA titer 1 : 1600), 7 days after booster with 2 pg DNP-OA in alum. In contrast, the transfer of spleen cells from OA-primed donors together with normal spleen cells resulted in a day -7 anti-DNP PCA titer of 1 : 80, typical for an accel- erated primary response in the presence of carrier-primed helper T cells. These results clearly show that spleen cells from mice primed with DNP-BP contain DNP-specific IgE B mem- ory cells, and suggest that the failure to respond to immuniza- tion with DNP-BP may be due to an active mechanism of IgE- selective suppression.

3.2 Pretreatment with DNP-BP modulates anti-DNP IgE

Fig. 3 shows that mice pretreated with DNP-carrier conjugates exhibit an accelerated primary anti-DNP IgE response (day -7 PCA titer of 1 : 200-1 : 400), while the BP- or saline- pretreated groups presented the typical primary IgE response on day 14 (PCA titer 1 : 400). By this time, in contrast to all

Eur. J. Immunol. 1989.19: 441-446 Modulation of the murine IgE response by interferon-gamma 443

I I I

I

1 I 0 7 14 21 Zd 1 4 2 t DAY t

Figure 1 . Induction of IgE and IgG anti-DNP antibodies after immunization with DNP-BP. Mice were injected i.p. on days 0 and 28 (arrows) with lo9 (50 pg) DNP-BP in saline (A-A), lo9 DNP-BP in alum ( H a . . H) or 2 pg DNP-OA in alum (0-4). (A) IgE. (B) IgG .

other groups, mice pretreated with DNP-BP exhibited drasti- cally decreased anti-DNP IgE levels which continued to decline until day 28. A booster with 2 pg DNP-OA in alum on day 28 resulted in a seconary IgE response, but the anti-DNP PCA titers were 4-10 times lower than in any of the controls. Anti-DNP IgG antibodies were not appreciably affected by DNP-BP pretreatment.

It is interesting to note that the sharp IgE-selective decline was limtied to the hapten; the anti-OA PCA titers were not affected regardless of whether the immunization on day 0 was

ant i -DNP IgE (PCA t i te r , log.)

CELLS TRANSFERRED

DONOR 1 DONOR 2 ~~

( 1 600) N o r m a l DNP-BP

OA D N P - B P

OA N o r m a l

Figure 2. Adoptive transfer of anti-DNP IgE responses demonstrat- ing DNP-specific IgE B cells in mice immunized with DNP-BP. Donor 1 mice were primed with 0.5 pg OA in alum 7 days before transfer or were left unprimed. Donor 2 mice were primed with lo9 DNP-BP 28 days before the transfer or were left unprimed. On the day of transfer, 2 x lo7 spleen cells from each donor were mixed and transferred i.v. to sublethally irradiated recipients which were boosted on the same day with 2 pg DNP-OA in alum i.p. PCA titers were determined on day 7 (hatched bar) and day 10 (solid bar) after boost- ing. PCA titers are in brackets.

3.3 Anti-DNP IgG isotype distribution

Since anti-DNP IgG levels were always detected after pre- treatment with either DNP-BP or TNP-KLH, we focused our attention on a possible difference in IgG subclass distribution between DNP-BP- and TNP-KLH-pretreated mice. Fig. 4 clearly shows that while the predominant anti-DNP isotype in anti-DNP-BP-pretreated mice was IgG2, (Fig. 4A), IgGl and IgGh were evenly represented in mice pretreated with TNP- KLH (Fig. 4B). In mice that were not pretreated, but were primed with 2 pg DNP-OA in alum, IgGl anti-DNP antibodies were significantly higher than IgG2, (results not shown). As obseved with PCA titers, the anti-OA IgGl and IgG2, subclass distribution was not affected by pretreatment, and followed the expected pattern of IgGl and IgE expression as observed in mice. immunized with DNP-OA in alum alone (Table 2). The anti-OA IgGl and IgE levels in mice primed with 2 pg OA in alum were considerably higher than those of mice primed with DNP-OA in alum, since the predominant antibody response in the latter is against the hapten.

Table 1. Effect of DNP-BP pretreatment on anti-OA IgE antibodies

hetreatmenp) Primingb) PCA titer Day 7 Day 14

Saline TNP-KLH DNP-BP Saline TNP-KLH DNP-BP

OA 40 320 OA 40 320 OA 40 320

DNP-OA ND') 160 DNP-OA ND 160 DNP-OA ND 160

a) Groups of 4 CBNJ mice were injected on days -16, - 14 and -12 with saline, 50 pg TNP-KLH or lo9 (50 pg) DNP-BP i.p.

b) Mice were injected on day 0 either with 2 pg OA or with 2 pg DNP-OA, both adsorbed onto 2.5 mg A1(OH)3.

with OA or DNP-OA (Table 1). c) Not determined.

444 M. Hagen, N. A. Essani and G. H. Strejan Eur. J. Immunol. 1989.19: 441-446

P

7 14 21 20 35 DAY t

(6)

10 m 0 7 14 21 20 35

t DAY t Figure 3. Suppression of the anti-DNP IgE antibody response after treatment with DNP-BP. Groups of CBNJ mice were pretreated by i.p. injection with lo9 DNP-BP on days -16, -14, -12 (A), with lo9 DNP-BP on day -14 (A), with 50 pg TNP-KLH on days -16, -14, -12 (W), with 50 pg DNP-ascaris antigen on days -16, -14, -12 (O), with lo9 BP on days -16, -14, -12 (0) or with saline (0). All mice were primed on day 0 and boosted on day 28 with 2 pg DNP-OA in alum (arrows). The results are from a typical experiment. (A) IgE. (B) IgG.

3.4 Modulatory role of IFN-y

These results strongly suggested that the inability of DNP-BP to promote anti-DNP IgE antibody production was attributed to the preferential stimulation of TH1 cells and may operate through increased production of IFN-y. To test this assump- tion, three groups of CBA/J mice were pretreated with lo9

Table 2. Effect of DNP-BP pretreatment on anti-OA isotype distnbu- tion

Pretreatmenta) IgG,b) IgGh IgE0

Normal*] Saline TNP-KLH DNP-BP None")

1 1 0 10 3 160

110 11 160 110 35 160 900 18 rn

Groups of 4 CBNJ mice were injected with saline, 50 pg TNP- KLH or lo9 DNP-BP i.p. on day -14, and with 2 pg DNP-OA in alum on day 0. Anti-OA IgGl and IgG,, levels were determined on day 14 and are expressed as units/well. Anti-OA PCA titer on day 14. Mice were not immunized. Mice were not pretreated but were injected with 2 pg OA in alum instead of DNP-OA, on day 0.

. .

'I 1% -7 1:'

Jo5 21

. . . . 0 1oomm4M)

anti-DNP lgG (Units)

Figure 4. Distribution of anti-DNP IgG, and IgG2. isotypes. Groups of 4 mice were treated with lo9 DNP-BP on days - 16, - 14, - 12 (A) or with 50 pg TNP-KLH on the same days (B). The sera obtained on the days indicated were tested for anti-DNP IgG, (hatched bars) and IgGb (solid bars) antibodies.

DNP-BP on day -14, and 36 h later were injected i.v. with 1 mg purified anti-IFN-y XMG6, 1 mg purified anti-NP anti- body 54.1 or saline.On day 0, all mice were immunized with 2 pg DNP-OA in alum, bled on days -7,0,7 and 14, and the sera were tested for &GI, IgG2, and IgE anti-DNP antibodies. Fig. 5 shows that treatment with anti-IFN-y determines a dras- tic decrease in anti-DNP IgG2, levels (A) and the elimination of the characteristic spike and suppression of PCA titers which is otherwise observed in DNP-BP-pretreated animals, after priming with DNP-OA in alum (B). The groups that were injected either with 1 mg 54.1, an isotype-matched but irrele- vant mAb, or with saline, exhibited a 4-fold drop in PCA titers from day 7 to day 14. Anti-DNP IgGl levels were low in all cases as DNP-BP does not stimulate significant levels of IgGl (Fig. 4).

Eur. J. Immunol. 1989.19: 441-446 Modulation of the murine IgE response by interferon-gamma 445

(APC) is the production and release by T cells of several lym- phokines which influence the activation, proliferation and differentiation of both T and B lymphocytes. IFNy and IL4 have recently been identified as the major mutually antagonis- tic molecules controlling B cell isotype expression [ 12-15]. The present results also establish a causal relationship between the decrease in anti-DNP IgE levels and IFN-y production, as a consequence of DNP-BP treatment. The administration of anti-IFN-y mAb (XMG6) to DNP-BP-treated mice resulted in the abrogation of the IgE decline, since the accelerated (day 7) anti-DNP IgE response was not followed by a decrease in PCA titers on day 14. The day 7 PCA titers of this group were lower than those of the groups injected with an irrelevant isotype-matched antibody (54.1) or with saline. It is important to note, however, that while the PCA titers in the two control groups decreased 4-fold from day 7 to day 14, the PCA titers of the group receiving anti-IFN-y antibody remained unchanged. This group also showed markedly decreased anti- DNP IgG2, levels. IFN-y exhibits a broad spectrum of immunoregulatory functions. Among these is the ability to activate macrophages and to induce the expression of MHC class I1 molecules on the surface of APC [16, 171. The drastic reduction in IgG2, (the predominant isotype produced in response to DNP-BP stimulation) and the slight reduction of anti-DNP IgE on day 7 may be a consequence of deficient antigen presentation by macrophages. Alternatively, anti- IFN-y may be involved in preventing activation of Th cells or in the generation of Ly-2’ T, cells as described by others [18]. The kinetics of the IgE spike and suppression as well as the strict hapten specificity of the phenomenon suggest that the administration of DNP-BP preferentially induces BP-specific Thl cells. These may release IFN-y which in turn promotes the switch to IgG2, antibody production.

The suppression of the anti-DNP IgE response which follows the day 7 “spike” could be explained in several ways: (a) the appearance of IgE-selective suppressor cells or IgE-suppres- sive factors; (b) the production of regulatory auto-anti-idioty- pic antibodies; (c) feedback regulation by anti-DNP IgGZ,, or (d) exhaustion of the DNP-specific B cell repertoire as a result of the IgG2, switch. Utilizing a variety of techniques we were unable to detect either suppressor cells or anti-idiotypic anti- bodies. Furthermore, we believe that feedback regulation is unlikely because ongoing IgE responses are difficult to sup- press by passively administered IgG [19]. Moreover, unless we postulate that IgG2, is much more effective than IgG, in reg- ulating ongoing IgE production, we cannot explain the failure of equivalent concentrations of anti-DNP IgG in the serum of animals pretreated with TNP-KLH to mediate the IgE-sup- pressive effect. We suggest that the most likely explanation for the modulation of the IgE response described in this study is that a large segment of the anti-DNP B cell repertoire may have been induced to express and secrete IgG2, antibodies. Once committed to IgG2,, it is unlikely that these cells can switch to other isotypes [15]. Immunizition with DNP-OA in alum would activate OA-specific Th cells in which the TH2 subset predominates. TH2 cells, in turn, would collaborate with the limiting number of DNP-specific IgE B cell precursors induced by the DNP-BP pretreatment. As no new DNP- specific B cells can be recruited, the anti-DNP IgE response rapidly disappears after the day 7 peak. According to this pro- posed mechanism, anti-OA IgE responses should remain unaf- fected. Indeed, anti-OA PCA titers were the same, regardless of whether or not the mice were pretreated with DNP-BP (Table 1). Moreover the IgGI/IgG2, isotype distribution of the

1280 - 0 Y w .-

320 0 a v

- $ 1 80 a Z ? : 20 .- w

0 L-s=a- O L -7 0 7

DAY DAY

- I4

Figure 5. Effects of anti-IFN-y antibody on anti-DNP IgG,, IgGh and IgE production in vivo. Groups of 4 mice were pretreated with lo9 DNP-BP on day -14 and primed with 2 pg DNP-OA in alum on day 0. Thirty-six hours after DNP-BP pretreatment, separate groups of mice received 1 mg anti-IFN-y mAb XMG6 (..El.., I), 1 mg anti-NP mAb 54.1 (. . A. ., -A-) or saline ( 0 .O. ., -C) i.v. (A) Anti- DNP IgGl (closed symbols) and IgGz, (open symbols) are expressed in units. (B) Anti-DNP IgE (closed symbols) are expressed as PCA ti- ters.

4 Discussion

The objective of the present investigation was to determine whether immunization of mice with a hapten coupled to B. pertussis vaccine, an adjuvant known to potentiate IgE responses, would result in enhanced anti-DNP IgE antibody production. The results showed that CBNJ mice, immunized with DNP-BP either in saline or adsorbed on alum, failed to produce detectable anti-DNP IgE levels after a primary stimu- lation, and had only very low levels after booster.

Pretreatment with DNP-carrier conjugates followed by immunization with DNP-OA in alum led to the appearance of two distinct response patterns, depending on the nature of the carrier used for pretreatment. Thus, injection of DNP-BP resulted in depressed IgE and IgGl and increased levels of anti-DNP IgG,, while treatment with TNP-KLH determined good IgE and almost equal levels of IgG, and IgGz, anti-DNP antibodies. In both instances, the administration of 2 pg DNP- O A in alum 14 days after pretreatment resulted in an acceler- ated anti-DNP IgE response, indicating the induction of DNP- specific IgE B cells by either type of pretreatment. In DNP- BP-pretreated groups, a sharp decline of the anti-DNP IgE levels consistently followed 72-96 h after the peak of the accel- erated primary response. Similar treatments with TNP-KLH never resulted in an IgE decline, suggesting therefore that the abrupt decrease in anti-DNP IgE levels should be attributed to intrinsic properties of the BP vaccine. It is interesting to note, however, that no such decrease was observed after pretreat- ment with BP alone (Fig. 3), or when lo9 BP were mixed with 50 pg TNP-KLH (not shown).

One consequence of the interaction between T cells and MHC-linked antigen on the surface of antigep-presenting cells

i“

446

anti-OA response was also unaffected by pretreatment (Table 2). These results clearly argue against a role for iso- type-selective, antigen-nonspecific suppressor factors.

M. Hagen, N. A. Essani and G. H. Strejan Eur. J. Immunol. 1989.19: 441-446

The ability of a bacterial vaccine to stimulate IgGza responses preferentially, and its failure to produce IgE was recently reported after the administration of Brucella abortus [20]. This effect was attributed to the ability of the vaccine to stimulate IFN-y production. Treatment with anti-IFN-y antibody strongly suppressed IgG2, and stimulated IgGl, but not IgE production. It must be noted, however, that total isotype levels and not antigen-specific antibodies were measured in that study. The present findings are, to our knowledge, the first to demonstrate the control by IFN-y of an antigen-specific response, as a result of deliberate antigen administration, and the dependence of the isotype profile of this response on the nature of the antigen injected.

The relevance of this mechanism for human disorders of the IgE system is difficult to assess but recent reports seem to suggest that similar, lymphokine-dependent regulatory mechanisms are involved [21, 221.

The authors wish to thank Dr. F. D. Finkelman for kindly supplying the monoclonal antibodies XMG6 (anti-IFN-y) and 54.1 (anti-NP). We thank Mr. D. Surlan for his technical assistance, and Ms. T. M. Hagen and Ms. W. Dodds for their assistance in the preparation of this manu- script.

Received October 3, 1988; in revised form December 7, 1988.

5 References

1 Ishizaka, K., Annu. Rev. Immunol. 1984.2: 159. 2 Ishizaka, K., Annu. Rev. Immunol. 1988. 6: 513. 3 Coffman, R. L., Seymour, B. W. P., Lebman, D. A., Hiraki, D.

D., Christiansen, J. A., Schrader, B., Cherwinski, H. M., Savel- koul, H. F. J., Finkelman, F. D., Bond, M. W. and Mosmann, T. R., Immunol. Rev. 1988. 102: 5.

4 Snapper, C. M., Finkelman, F. D. and Paul, W. E., Immunol. Rev. 1988. 102: 51.

5 Tada, T., Prog. Allergy 1975. 19: 122. 6 Katz, D. H., Immunology 1981. 41: 1. 7 Kishimoto, T., Prog. Allergy 1982. 32: 265. 8 Colby, W. D. andstrejan, G. H., Eur. J . Immunol. 1980.10:602. 9 Hussain, R., Bradbury, S. M. and Strejan, G. H., J . Immunol.

10 Kishimoto, T., Hirai, Y., Suemura, M. and Yamamura, Y., J.

11 Hagen, M. and Strejan, G. J., J . Immunol. Methods 1987.100: 47. 12 Paul, W. E. and Ohara, J., Annu. Rev. Immunol. 1987. 5: 429. 13 Mosmann, T. R. and Coffman, R. L., Immunol. Today 1987.

8: 223. 14 Cherwinski, H. M., Schumacher, J. H., Brown, K. D. and Mos-

mann, T. R., J . Exp. Med. 1987.166: 1229. 15 Stevens, T. L., Bossie, A., Sanders, V. M., Fernandez-Botran, R.,

Coffman, R. L., Mosmann, T. R. and Vitetta, E. S., Nature 1988. 334: 255.

1973. 111: 260.

Immunol. 1976. I1 7: 396.

16 Beller, D. I . and Unanue, E. R., J . Immunol. 1981. 126: 263. 17 Wong, G . H. W., Clark-Lewis, I., Hams, A. W. and Schrader, J.

18 Frasca, D., Adorini, L., Landolfo, S. and Doria, G., J . Immunol.

19 Ishizaka, K. and Okudaira, H., J . Immunol. 1972. 109: 84. 20 Finkelman, F. D., Katona, I. M., Mosmann, T. R. and Coffman,

R. L., J. Immunol. 1988.140: 1022. 21 Pene, J . , Rousset, F., Briere, F., Chretien, I., Paliard, X., Ban-

chereau, J., Spits, H. and De Vries, J. E., J . Immunol. 1988.141: 1218.

22 Del Prete, G., Maggi, E., Parronchi, P., Chretien, I., Tin, A., Macchia, D., Ricci, M., Banchereau, J., De Vries, J. E. and Romagnani, S., J. Immunol. 1988. 140: 4193.

W., Eur. J . Immunol. 1984. 14: 52.

1988. 140: 4103.