Eur. J. Immunol. 1988.18: 313-316 The half-lives of serum immunoglobulins 313

Short paper

Paulo Vieira and Klaus Rajewsky

Institute for Genetics, University of Cologne, Cologne

1 Introduction

The half-lives of serum immunoglobulins in adult mice*

We determined the half-lives of several sets of munne monoclonal antibodies span- ning all immunoglobulin isotypes in the serum. The antibodies in each set possess the same V region. With this approach, the differences in half-life observed between the different isotypes are independent of the V region carried by the monoclonal anti- bodies and therefore must relate to each other in the same way as the half-lives of each class of serum immunoglobulins. The half-life of a monoclonal antibody of the yZa isotype is identical to the average half-life of serum IgG2, as previously determined (6-8 days; P. Vieira and K. Rajewsky, Eur. J . Zmmunof. 1986.16: 871). Therefore, the half-lives determined with monoclonal antibodies possessing the same V region represent the half-life of the serum immunoglobulins. In this way we calculated the half-life of IgM as 2 days, IgG3 and IgGI as 6-8 days, IgGzb has a half-life of 4-6 days. IgE has a half-life of 12 h. A polymeric form of IgA was found to be eliminated from the serum with a half-life of 17-22 h.

We recently determined the half-life of IgG2, in the serum of adult allotype-suppressed mice [l-31. The decay measured was that of the endogenous population of molecules and we argued that the true half-life, in the serum, of IgG2, is 6-8 days on average [l].

A similar system of allotype suppression is not available for the other Ig isotypes. Also, as argued by Jerne [4], it is not correct to extrapolate from the decay of individual monoclonal antibodies the haif-life of serum immunoglobulins.

However, an antibody of the yza isotype with the same half-life as that of the total IgG2, in the serum has a V region that does not influence its decay to a greater extent than the total set of V regions expressed by the antibodies of that class. The same argument applies for that particular V region when expressed together with other isotypes.

In order to obtain good estimates of the half-lives of the differ- ent classes of serum immunoglobulins, we determined the half-lives of several sets of monoclonal antibodies, in each of which the antibodies carry the same V region, but differ in isotype. One of the monoclonal antibodies, of the yza isotype, has the same half-life as the overall population of serum IgGz,. Therefore, the values obtained represent reliable estimates for the true average half-life of serum immunoglobulins.

[I 64241

2 Materials and methods

2.1 Mice

Adult female C57BL16 and (CBA x BALB/c)F1 mice were used. The animals were bred in our own colony in the Institute for Genetics, Cologne.

2.2 Antibodies

Ac146 ( y l , x ) , Ac146.2a (yza,x), Ac146.2b ( Y ~ ~ , x ) and Ac146E ( E , x ) all belong to the same family of class switch variants [5, 61. Each animal received, i.p., approximately 100-200 pg of purified protein, except for Ac146E where ascites fluid from tumor-bearing mice was used.

S24/63/63 (y&) and S24/1-47 (yl,hl) are NP binding mono- clonal antibodies with identical V regions [7]. Each animal received approximately 50 pg of purified antibody i.p. B1-8 (p,hl) [8], N1G9 (yl,hl) [9] and 233.1.3 (a,hl) [lo] are inde- pendently derived monoclonal anti-NP antibodies bearing the same V region but differing in their isotypes. Each animal received 300-400 pg of purified protein, i.p.

Class-specific goat antisera used in the enzyme-linked immunosorbent assay (ELISA) were purchased from South- ern Biotechnology Associates Inc., Carshalton, Surrey, GB. Goat anti-e serum, kindly provided by Dr. A. Radbruch, was extensively absorbed with Sepharose-coupled N1G9 and Ac146 [ l l ] until no more reactivity against the N1G9 or Ac146 V regions could be detected. Biotinylated rabbit sera specific for goat IgG was purchased from Vector (Burlingame, CA).

* This work was supported by the Deutsche Forschungsgemeinschaft through SFB-74. 2.3 ELISA

The concentration of antibodies possessing the appropriate specificity and class in the serum of the animals was deter-

Correspondence: Paulo Vieira, Institut fur Genetik der Universitat Koln, Weyertal 121, D-5000 Koln 41, FRG

Abbreviations: ELISA: Enzyme-linked immunosorbent assay PBS: Phosphate-buffered saline HPLC: High performance liquid chroma- tography

mined by ELISA as described [121. Briefly, 96-well Plates were coated with a preparation Of purified protein diluted to 10 pg/ml in phosphate-buffered saline (PBS; 50 @/well). After

0014-2980/88/0202-0313$02.50/0 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1988

314

a second incubation with PBS containing 2% bovine serum albumin (PBS-2% BSA; 200 yl/well), test sera serially diluted in PBS-2% BSA were applied to the plates (30 pl/well) and bound antibodies determined by incubating the plates with biotinylated developing antibody (30 yVwell). The assays were standardized with purified antibody of the appropriate class and specificity. All samples from the same animal at various time points were titrated in the same assay plate. Thus, the antibody concentrations at various times were determined on the basis of the same standard curve. The concentration of the antibodies belonging to the Ac146 family was determined by coating the plates with purified Fab fragments of NlG9, to which Ac146 is an anti-idiotypic antibody [9], and developing with biotinylated class-specific goat antisera. The concentra- tion of Ac146E antibodies in the sera was determined using as a second-step reagent a biotinylated rabbit anti-goat IgG serum, to reveal the binding of the goat anti-e antibodies to the Ac146E molecules. The values obtained were compared to a dilution of 1/50 of ascites to which an arbitrary concentration of 100 unitdm1 was attributed. The results are expressed as unitdml.

P. Vieira and K. Rajewsky Eur. J . Immunol. 1988.18: 313-316

The concentrations of S24/1/47, S24/63/63, N1G9 and 233.1.3 were determined on NIP-coated plates developed with goat sera specific for the appropriate class. We found that the back- ground of NIP-binding IgM antibodies in the serum of normal mice was approximately 10 pg/ml. Therefore, the concentra- tion of B1-8 was determined using Ac38-coated plates, developed with goat anti-y sera. In this case, the background ranged from 2-2.5 yg/ml. Ac38 (yl,hl) is an anti-idiotypic anti- body to B1-8 [5].

2.4 High performance liquid chromatography (HPLC)

It was performed using a I300 (7.8 mm I.D. x 30 cm) tandem column (Waters, Konigstein, FRG). The running buffer was PBS containing 0.1% NaN3, pH 7.5. The flow rate was 1 ml/ min.

Table 1. Half-lives of monoclonal antibodies in individual mice')

B1-8 familyb) B1-8 N1G9 1.8 days 6.7 days 2.1 days 6.8 days 2.2 days 7.1 days 2.3 days 7.3 days

S24 family') S24163163 S2411147 5.7 days 6.1 days 6.0 days 7.5 days 6.1 days 7.7 days 6.7 days 9.1 days

Ac146 familvd) Ac146 Ac146.2b Ac146.2a Ac146E

5.5 days 3.8 days 5.8 days 7.33 h 6.6 days 3.9 days 7.7 days 12.15 h 6.6 days 4.1 days 7.9 days 12.18 h 7.8 days 5.2 days 9.2 days 16.14 h

5.5 days 5.7 days

2.5 Statistical analysis

The regression lines and the half-lives were determined by the least squares method.

3 Results and discussion

3.1 The Ad46 family

Ac146 is a monoclonal anti-idiotypic antibody raised in CBA mice. It is an IgGl and binds the anti-NP antibody B1-8 [5 ] . Several spontaneous class-switch variants, of the y2a, Y2b and E

isotypes, have been isolated from the original cell line [6]. The half-life of each of these was determined by injecting into groups of (CBA X BALB/c)F1 mice a preparation of each of the antibodies and measuring the decay of NlGPbinding material in the serum. (NlG9 (91 carries the same V region as B1-8). Representative decay curves are shown in Fig. 1, and the half-lives calculated in individual mice in Table .l. Ac146.2a has a half-life of 6-9 days, not different from the value calculated for the total IgGz, in the serum [l]. There- fore, the values calculated for Ac146E (7.3 to 16 h), Ac146.2b (4-6 days) and Ac146 (6-8 days) reflect the average half- lives, in the serum, of IgE, IgG2b and IgGl, respectively.

3.2 The S24 family

This family consists of two monoclonal anti-NP antibodies of C57BL16 origin: ,524163163 (y3,hl) and S2411147 (yl,hl) [7]. Both antibodies decay with similar half-lives, (6-9 days) in (syn- geneic) C57BLI6 mice (Fig. 2 and Table 1). If the half-life of Ac146 represents the half-life of the total serum IgGl, then S24/1/47, whose half-life is the same, has a V region that does not influence its decay. Therefore, the half-life in the serum of IgG3 is 6-7 days, the same as S24/63/63.

233.1.3 16.9 h 18.8 h 22.2 h 22.8 h

a) The values shown represent half-lives calcu- lated in single mice in each of the groups.

b) B1-8 (IgMJl), N1G9 (IgG,,hl) and 233.1.3 (IgA,hl) have the same V region. Half-lives determined in adult syngeneic C57BL16 mice.

c) S24/1/47 (IgG1,hl) is an isotype switch variant with the same V region as S24/63/63 (IgG3,hl). Half-lives determined in adult syngeneic C57BL16 mice.

d) Ac146 (IgG,,x), Ac146.2b (IgG,,,x), Ac146.2a (IgG2,,x) and Ac146E (IgE,x) com- prise a family of class switch variants of CBA origin. The half-lives were determined in (CBA x BALB/c)F, adult mice.

Eur. J. Immunol. 1988.18: 313-316 The half-lives of serum immunoglobulins 315

- E lo3 m . 3.

102

10

1

I

1 -

i 2 i 8 1;o 264 408 hours

Figure 1. Representative decay curves obtained for the following anti- bodies: Ac146 (y,) (A), Ac146.2a (y2J (A), Ac146.2b (y2b) (0) and Ac146E (E) (W). The ordinate represents concentration in the serum in pg/ml or unitdm1 in the case of Ac146E.

3.3 The B1-8 family

B1-8 (p,Al), N1G9 (yl,Al) and 233.1.3 (a,hl) are indepen- dently derived anti-NP antibodies isolated from C57BL/6 mice. The sequences of their H and L chain V regions are identical [8-lo]. This family enabled us to determine (Fig. 3 and Table 1) the half-life of IgM in C57BL/6 mice to be 2 days. IgA decays with a half-life of 17-23 h. Note that NlG9, an IgGl antibody, has a half-life of 7 days, not different from the half-lives of Ac146 and S24ill47.

In an attempt to determine whether antibody 233.1.3 is a monomer or a polymer, the same antibody preparation was run through a HPLC column and the fractions obtained were

z 102

3 l

1- 1 2 4 6 10 20

days

Figure 2. Representative decay curves for the antibodies belonging to the S24 family. S24/63/63 (y3) (H) and S24/1/47 (n) (0).

- E 103 B

102

10

1

days

Figure 3. Representative decay curves for the antibodies of the B1-8 family. N1G9 (y,) (H), B1-8 (p) (0) and 233.1.3 (a) (A). In the case of B1-8 the concentration of Ac38'lp' antibodies in the serum before injection of the antibody was 2.6 pg/ml. Therefore the last point in the B1-8 curve (day 20: 2 pglrnl) was excluded from the calculation of the half-life.

titrated to determine the concentration of NIP-binding IgA antibodies. In parallel B1-8 and N1G9 were also run. In this way we estimated the molecular weight of 233.1.3 as being intermediate between B1-8 (a pentameric IgM) and N1G9 (an IgG1). As shown in Fig. 4,233.1.3 elutes from the column in a fraction distinct from the fractions were B1-8 or N1G9 elute. The precise molecular weight of the antibody cannot be deter- mined on this column because B1-8, a pentameric IgM of appr. 1 x lo3 kDa, is eluted from the column with the void volume.

This result was confirmed by sodium dodecyl sulfate-poly- acrylamide gel electrophoresis (5-10% gradient gel, under nonreducing conditions) where 233.1.3 migrates to a position between N1G9 and B1-8 (not shown).

Thus, the half-life calculated for the 233.1.3 antibody is likely to represent that of an IgA dimer, although we cannot exclude that this antibody is a higher polymer of IgA. It should be noted, however, that, with a molecular weight intermediate between IgG and IgM, it can only be a polymer of 2, 3 or 4 units. Note also that the material titrated in the assay is homogeneous in this respect because it elutes from the HPLC column as a single peak (see Fig. 4B).

The values calculated for IgM are in good agreement with previous estimates [13, 141. The half-life of IgE is somewhat longer than the value obtained by Haba et al. [15], namely 6 h, as compared to 12 h for Ac146E. Their case may be an exam- ple of a monoclonal antibody decaying with a half-life different from that of the total serum Ig. The half-lives of IgG are longer than former estimates with iodinated antibodies [13, 141 but compatible with more recent estimates obtained with noniodinated monoclonal antibodies [15, 161, (but see below the discussion of the case of Ac38).

316 P. Vieira and K. Rajewsky

E 1950

600 300 150

mO,=Nr

time

fraction no.

Eur. J. Immunol. 1988.18: 313-316

IC)

time

3000 -

Figure 4. Results of the separation by HPLC of B1-8 (A), 233.1.3 (B) and N1G9 (C) . The upper curves represent the plot of the absorbance at 280 nm

m o - ~ m l t ~ ) w r - m ---,------ r ; ~ g f g g for each of the preparations. The bars reoresent the concentration of NIP+/

1500- 1200 - 900 -

fraction no. p'(B1-8), NIPt/at (233.1.3) or NIP+/ yIt (NlG9) in each of the fractions ob- tained from the column.

The half-lives in Table 1 represent, we believe, the true aver- age half-lives of serum immunoglobulins of the various iso- types. (In other words, if it were possible to measure the half- life of the total population of serum immunoglobulins, as it is in the case of IgG2, [l], then the values obtained would not be different from the values given in Table 1).

Because we calculate the half-lives by comparing all serum samples from the same animal to the same standard curve (see Sect. 2.3), the experimental error in the individual half-lives is minimized. By calculating 95% confidence limits of the best fitting lines we found our determinations of half-lives to be accurate within roughly k 20% (not shown). Therefore, great- er differences in half-lives are likely to reflect true differences in the metabolic rates of the individual animals. Such differ- ences, or even differences between strains, might explain the slight slower half-lives calculated by Bosma [ 171 for hemag- glutinins in the serum (- 1 day for the 19 S and - 4-5 days for the 7 S class).

All of the above arguments do not exclude that a particular antibody may have a half-life different from that of the main fraction of serum Ig of the same isotype. This could be the case, as argued above, for the monoclonal antibodies used by Haba et al. [15] to determine the half-life of IgE. It would also explain the discrepancy between the values given in this report and the half-life of 4.5 days calculated for an IgGl antibody (Ac38), in this laboratory by Takemori and Rajewsky [18]. Alternatively the discrepancy could be due to the difference in the age of the animals used to calculate these half-lives: Takemori and Rajewsky [18] calculated a half-life of 4.5 days between 4 and 6 weeks of age, and our decays are all calcu- lated in 8-10-week-old animals.

However, the half-life of most antibodies in the serum appears to depend mainly on their isotype, as suggested by the consis- tency of the present data within themselves and the earlier

observation [l] that > 95% of IgG2, in the serum have essen- tially the same decay.

We wish to express our gratitude to Claudia Uthoff-Hachenberg, to W . Miiller and to A . Schmitz for helping to bleed some of the mice, run the HPLC and help us with the SDS gels. We also thank M. Cramer for advice and Udo Ringeisen for preparing the figures.

Received September 14, 1987; in revised form December 19, 1987.

4 References

1 Vieira, P. and Rajewsky, K., Eur. J. Immunol. 1986. 16: 871. 2 Herzenberg, L. A. and Herzenberg, L. A., Contemp. Top.

3 Bosma, M. J. and Bosma, G. C., Nature 1976.259: 313. 4 Jerne, N. K., Immunol. Rev. 1984. 79: 5. 5 Reth, M., Imanishi-Kari, T. and Rajewsky, K., Eur. J. Immunol.

6 Miiller, C . E. and Rajewsky, K., J. Immunol. 1983. 131: 877. 7 Baumhackel, H., Liesegang, B., Radbruch, A., Rajewsky, K. and

8 Reth, M., Hammerling, G. J. and Rajewsky, K., Eur. J. Immunol.

9 Cumano, A. and Rajewsky, K., Eur. J. Immunol. 1985. 1.5: 512. 10 Siekevitz, M., Kocks, C . , Rajewsky, K. and Dildrop, R., Cell

11 Axen, R., Porath, J. and Ernback, S . , Nature 1967. 214: 1302. 12 Kendall, C., Ionescu-Matiu, I . and Dreesman, G. R., J. Immunol.

13 Spiegelberg, H. L., Adv. Immunol. 1974. 19: 259. 14 Waldmann, T. A. and Strober, W., Prog. Allergy 1969.13: 1. 15 Haba, S. , Ovary, Z . and Nisonoff, A., J. lmmunol. 1985. 134:

3291. 16 Tokuhisa, T., Oi, V. T., Gadus, F. T., Herzenberg, L. A. and

Herzenberg, L. A., in Pernis, B. and Vogel, H. J. (Eds.), Regulat- ory Tlymphocytes, Acadmic Press Inc., New York 1980, p. 3150.

17 Bosma, M. J., in Mirand, E. A. and Back, N. (Eds.), Germ-free Biology, Experimental and Clinical Aspects, Plenum Press, New York 1969, p. 249.

Immunobiol. 1973. 3: 41.

1979. 9: 1004.

Sablitzky, F., J. Imrnunol. 1982. 128: 1227.

1978. 8: 393.

1987. 48: 757.

Methods 1984. 56: 329.

18 Takemori, T. and Rajewsky, K., Immunol. Rev. 1984. 79: 103.