The effect on immunoglobulin glycosylation of altering in vivo production of immunoglobulin G

Abstract

The effect on murine immunoglobulin G (IgG) glycosylation of altering IgG production in vivo was assessed in interleukin (IL)‐6 transgenic and CD4 knockout mice. C57BL/6 mice carrying the IL‐6 transgene showed increased levels of circulating IgG. This was associated with decreased levels of galactose on the IgG oligosaccharides. No decrease in β4‐galactosyltransferase mRNA or in enzyme activity was seen in IL‐6 transgenic mice. MRL‐lpr/lpr mice normally have elevated levels of circulating IgG, again accompanied by decreased levels of IgG galactose. Disruption of the CD4 gene in MRL‐lpr/lpr mice led to a substantial decrease in the concentration of circulating IgG, but IgG galactose levels remained low. Thus, an enforced decrease in IgG levels in the lymphoproliferative MRL‐lpr/lpr mice did not alter the percentage of agalactosyl IgG in these mice, suggesting that agalactosyl IgG production is not simply caused by excessive IgG synthesis leading to an insufficient transit time in the trans‐Golgi, but rather to a molecular defect in the interaction between galactosyltransferase and the immunoglobulin heavy chain.

Introduction

In certain disease states such as tumours (e.g. colonic adenocarcinoma, see reference ¹), autoimmunity (e.g. rheumatoid arthritis; see reference ²) and infectious disease (e.g. tuberculosis; see reference ³), there are alterations in the glycoforms present on one or more cell surface or secreted molecules. The factors that finely regulate the glycosylation of glycoproteins and glycolipids have yet to be fully understood. Glycosylation is controlled enzymatically by the action of glycosyltransferases and glycosidases. For example, β1,4‐galactosyltransferase‐1 (β4GalT1) adds galactose to N‐acetylglucosamine (GlcNAc) residues during the biosynthesis of complex‐type oligosaccharides. Immunoglobulin G (IgG) has a conserved N‐linked glycosylation site at Asn297 in the Cγ2 domain of the heavy chain and the glycoform of IgG which lacks galactose on both arms of this biantennary oligosaccharide is referred to as agalactosyl IgG.

There is a shift to the agalactosyl IgG glycoform in certain autoimmune diseases, such as rheumatoid arthritis in humans.² MRL‐lpr/lpr mice develop a spontaneous autoimmune disease with some features similar to rheumatoid arthritis, including an increased percentage of agalactosyl IgG.⁴ These mice also have an increased concentration of circulating IgG⁵ and elevated interleukin‐6 (IL‐6) levels.⁶ IL‐6 belongs to a family of glycosylation‐modifying cytokines⁷ and is present in the synovium in patients with rheumatoid arthritis.⁸ CD4⁺ T lymphocytes are an important source of this cytokine. Injection of recombinant IL‐6 into normal CBA/Ig^b mice leads to an increase in serum agalactosyl IgG levels⁹ and IL‐6 transgenic mice have also been reported to have an increased proportion of this IgG glycoform in their sera.¹⁰

Agalactosyl IgG is known to be pathogenic in a CD4⁺ T‐cell‐primed collagen‐induced model of rheumatoid arthritis.¹¹ It has been possible to delay progression of the autoimmune disease in MRL‐lpr/lpr mice using antibody directed against CD4^12,13 or by disrupting CD4 expression using gene targeting to create CD4^–/– mice.^14,15 However, mice lacking CD4⁺ T cells are still able to undergo immunoglobulin class switching from immunoglobulin M (IgM) to IgG.^16,17 We have therefore used IL‐6 transgenic C57BL/6 mice and CD4^–/– MRL‐lpr/lpr mice as models to investigate the importance of altered IgG production on IgG galactosylation in vivo.

Materials and methods

Mice

Spleens and peripheral blood were taken from 3–4‐month‐old mice. The C57BL/6 IL‐6 transgenic mice have been described previously¹⁸ and were produced using the human IL‐6 genomic gene fused with the human immunoglobulin heavy chain enhancer. The CD4‐deficient MRL‐lpr/lpr mice have recently been reported¹⁵ and were derived through a series of backcrosses of MRL‐lpr/lpr with CD4‐deficient C57BL/6 × 129/SV hybrid mice. MRL‐lpr/lpr mice heterozygous for the disrupted CD4 gene were produced by backcrossing homozygous CD4‐deficient MRL‐lpr/lpr mice with wild‐type MRL‐lpr/lpr. These mice, which express CD4 normally, were then crossed with homozygous CD4‐deficient MRL‐lpr/lpr mice so that CD4‐deficient MRL‐lpr/lpr offspring could be compared with their heterozygous CD4‐expressing MRL‐lpr/lpr littermates.

Measurement of IgG concentration

IgG levels were measured using an enzyme‐linked immunosorbent assay (ELISA), essentially as described by Engvall.¹⁹ Calibrated mouse serum (The Binding Site, Birmingham, UK) was diluted to a range of 0·003–3 μg/ml of IgG in phosphate‐buffered saline (PBS)/0·05% Tween‐20 (PBS/T). Sera (diluted 1 : 10 000) from the experimental mice and the standards were tested in duplicate.

Serum IgG galactosylation

Serum IgG galactosylation was measured using a previously described assay.²⁰ Mouse sera were diluted 1 : 25 in 0·1 m glycine containing 0·16 m NaCl, pH 8·0, and 50 μl was added in triplicate to two protein G′‐coated maxisorb plates. The captured IgG was heat denatured to expose the oligosaccharides, which were then probed with the biotinylated lectins (Vector Laboratories Inc., Peterborough, Cambridgeshire, UK) Bandeiraea simplicifolia II (BSII, used to detect terminal GlcNAc), at 40 μg/ml in PBS/T containing 1% bovine serum albumin (BSA) and 0·1 mm calcium chloride, and Ricinus communis agglutinin I (RCAI; used to detect galactose) at 2 μg/ml. Controls for the assay included sera with known ratios of BSII/RCAI binding previously determined in comparison with murine IgG, the oligosaccharide structures of which were characterized by Professor T. Rademacher and colleagues (Department of Biochemistry, University of Oxford, Oxford, UK), using hydrazinolysis.^21,22

β4GalT1 gene expression

Splenic lymphocytes were prepared and β4GalT1 mRNA levels were measured using a ribonuclease protection assay (RPA II kit; Ambion, Austin, TX), both as previously described.²³ The probe recognizing murine β4GalT1 corresponds to nucleotides 704–1263 of the sequence published by Shaper et al.²⁴ and was labelled to a specific activity of ≈ 8 × 10⁷ counts per minute (c.p.m.)/μg. The murine glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) and β‐actin probes (Ambion) were labelled to specific activities of ≈ 8 × 10⁶–1 × 10⁷ c.p.m./μg. Protected fragments were analysed on a polyacrylamide–urea gel. The levels of β4GalT1 mRNA and steady‐state control mRNA were assessed by phosphoimager analysis using a Bio‐Rad GS‐250 Molecular Imager (Bio‐Rad, Richmond, CA). Results are expressed as the ratio of β4GalT1/control mRNA signal.

β4GalT enzyme activity

Enzyme activity was measured using an ELISA, as previously described.²⁵

Results

IgG concentration and galactosylation in IL‐6 transgenic mice

The total IgG concentration was found to be 15‐fold higher in IL‐6 transgenics (83 ± 11 mg/ml) compared with littermates (5 ± 2 mg/ml) (P = < 0·0001, unpaired Student’s t‐test, Fig. 1). This increase in IgG production was associated with significantly decreased levels of IgG galactosylation (as measured by an increased BSII/RCAI ratio) in IL‐6 transgenic mice compared with the littermates (P = < 0·005, Fig. 2).

Figure 1.

Serum immunoglobulin G (IgG) concentration in interleukin‐6 (IL‐6) transgenic C57BL/6 mice (n = 12) and littermates (n = 14). Mean ± standard error of the mean (SEM) is indicated.

Figure 2.

Level of agalactosyl immunoglobulin G (IgG) in sera from interleukin‐6 (IL‐6) transgenic C57BL/6 mice (n = 12) and littermates (n = 14). Mean ± standard error of the mean (SEM) is indicated.

β4GalT expression and activity in IL‐6 transgenic mice

β4GalT gene expression and enzyme activity were examined in IL‐6 transgenic mice. The mRNA levels for β4GalT1 were comparable in IL‐6 transgenics and controls, regardless of whether GAPDH (P = > 0·5) or β‐actin (P = > 0·2) were used as internal controls for the amount of RNA in each lane (Fig. 3). The level of β4GalT enzyme activity was not significantly different between mice expressing the IL‐6 transgene and controls (P = > 0·19, Fig. 4).

Figure 3.

(a) RNase protection analysis of splenic β1,4‐galactosyltransferase‐1 (β4GalT1) mRNA in interleukin‐6 (IL‐6) transgenic mice (lanes 1 and 4–6) and their littermates (lanes 2, 3 and 7). Controls: yeast tRNA hybridized with β‐actin probe alone (A), β4GalT1 probe alone (B) and glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) probe alone (C), without RNase, showing the full‐length 350‐nucleotide (nt) β‐actin probe, 602‐nt β4GalT1 probe and 404‐nt GAPDH probe, respectively. Lanes D, E and F show yeast tRNA hybridized separately with β‐actin (D), GAPDH (E) and β4GalT1 (F) and treated with RNase to show the protected fragments for β‐actin (250 nt), GAPDH (316 nt) and β4GalT1 (559 nt), respectively. (b) Quantification of the β4GalT1 mRNA levels in spleen cells by phosphoimaging of RNase protection analysis experiments in IL‐6 transgenic mice (n = 6) and control littermates (n = 4), expressed as the ratio of β4GalT1 over β‐actin and as the ratio of β4GalT1 over GAPDH. Mean ± standard error of the mean (SEM) is indicated.

Figure 4.

β1,4‐galactosyltransferase (β4GalT) enzyme activity in spleen cells of interleukin‐6 (IL‐6) transgenic mice (n = 6) and control littermates (n = 6). Mean ± standard error of the mean (SEM) is indicated.

IgG concentration and galactosylation in CD4‐deficient MRL‐lpr/lpr mice

The total IgG concentration was determined in MRL‐lpr/lpr CD4 knockout mice and heterozygous age‐matched CD4‐expressing littermates. The concentration of IgG was significantly lower in mice lacking CD4 (2·2 ± 0·4 mg/ml) compared with the heterozygotes (9·1 ± 0·6 mg/ml) (P = < 0·0001, Fig. 5). No significant difference in the level of IgG galactosylation was observed between the CD4‐expressing and CD4‐deficient mice (P = 0·32, Fig. 6).

Figure 5.

Serum immunoglobulin G (IgG) concentration in CD4‐expressing (n = 6) and CD4‐deficient (n = 6) MRL‐lpr/lpr mice. Mean ± standard error of the mean (SEM) is indicated.

Figure 6.

Level of agalactosyl immunoglobulin G (IgG) in sera from CD4‐expressing (n = 6) and CD4‐deficient (n = 6) MRL‐lpr/lpr mice. Mean ± standard error of the mean (SEM) is indicated.

Discussion

The level of IgG galactosylation can affect the Fc‐mediated functional aspects of the antibody molecule.²⁶ Although the particular glycoforms that are borne by glycoproteins are known to be influenced by culture conditions in vitro,²⁶ it is unclear to what extent alterations in the production of glycoproteins in vivo affects the pattern of glycosylation. In particular, we wished to ascertain whether lowering the rate of production of IgG in MRL‐lpr/lpr mice would lead to a normalization of IgG glycosylation, i.e. correction of the agalactosyl IgG ‘defect’ seen in these mice.

To address the above issues we investigated IgG galactosylation in vivo in genetically modified mice. C57BL/6 mice, in common with most other normal strains of inbred laboratory mice, have an IgG serum concentration of ≈ 5 mg/ml. However, when an IL‐6 transgene is introduced into these mice, they exhibit increased levels of circulating IgG. The line of IL‐6 transgenic mice used in the present study develop autoantibodies and a mesangial cell proliferative glomerulonephritis, but do not show any sign of arthritis. Another line of IL‐6 transgenic mice, H2‐IL‐6 mice,²⁷ also do not develop arthritis, suggesting that overexpression of IL‐6 is not sufficient to cause this disease. IL‐6 was first identified for its action on B‐cell differentiation and immunoglobulin induction,²⁸ and the increase in IgG concentration observed in the IL‐6 transgenic mice used in the present study was associated with a concomitant increase in agalactosyl IgG levels. This observation confirms and extends the previous results of Rook and colleagues who examined IgG glycosylation using this line of mice.¹⁰ An increase of serum IgG concentration in IL‐6 transgenic mice has been reported by Suematsu and colleagues, with IgG1 being increased 120–400‐fold as compared with the levels of age‐matched normal control mice.¹⁸ The other classes of immunoglobulins (IgM, IgG2a, IgG2b, IgG3 and IgA) were found to be present at levels no higher than three times the concentration found in non‐transgenic mice. A recent study²⁹ has reported murine IgG isotype‐specific differences in galactosylation with IgG1 exhibiting the highest percentage of agalactosyl IgG (45–48%) followed by IgG2a (27–37%), IgG3 (20–32%) and IgG2b the lowest (13–17%). Therefore, the increase in agalactosyl IgG observed in the IL‐6 transgenic mice, as measured in this study by the binding of BSII/RCAI to IgG captured on protein G′, and by reactivity of a GlcNAc‐specific monoclonal antibody with protein A‐captured IgG by Rook and colleagues,¹⁰ is probably explained by increased production of the IgG1 isotype rather than by any IL‐6‐induced change within the B cell resulting in altered galactosylation of IgG per se. This would be supported by our current observations that neither β4GalT enzyme activity nor mRNA levels are altered in IL‐6 transgenic C57BL/6 mice.

With respect to the MRL‐lpr/lpr mice, the production of IgG was significantly decreased in the CD4‐deficient animals compared with heterozygous controls. However, even in the absence of CD4, these mice were able to produce moderate levels (≈ 2 mg/ml) of IgG. It has been shown previously by others that CD4 is not absolutely necessary for the effector function of major histocompatibility complex (MHC) class II‐restricted helper T cells.^16,17 Rahemtulla and colleagues have demonstrated that C57BL/6 × 129/sv (H2^b) mice lacking CD4 show in vivo immunoglobulin isotype class switching from IgM to IgG in response to sheep erythrocytes and vesicular stomatitis virus (VSV). CD4^– CD8^– T‐cell receptor (TCR)‐αβ⁺ T cells were shown to be responsible for providing help in the antibody response to VSV infection in these CD4‐deficient mice. A population of CD4^– CD8^– TCR‐αβ⁺ T cells that bear B220 is characteristic of MRL‐lpr/lpr mice. Recent evidence suggests that even αβ TCR‐deficient mice can produce strong IgG responses, in this instance possibly as a result of signals provided by γδ T cells.³⁰ Despite the fact that deficiency of CD4 in the MRL‐lpr/lpr mice led to a 76% decrease in IgG production, this did not influence the level of agalactosyl IgG. Although we have not assessed the level of β4GalT1 mRNA in CD4^–/– mice, we have previously observed that surface IgG⁺ B cells isolated from the secondary lymphoid tissues of MRL‐lpr/lpr mice have normal levels of mRNA for this enzyme.²³ In CBA mice, the degree of galactosylation of antigen‐specific IgG decreases as the specific IgG concentration increases during an immune response.³¹ The current study suggests that the lack of IgG galactose seen in the MRL‐lpr/lpr mice is not caused by the excessive IgG production leading to insufficient time for the β4GalT enzyme to function effectively during transit of the nascent immunoglobulin heavy chain through the trans‐Golgi. It would therefore seem that a more probable explanation for the increased level of agalactosyl IgG seen in MRL‐lpr/lpr mice lies within a yet to be defined aspect of the molecular interaction between enzyme and substrate within this cellular compartment.

Glossary

Abbreviations

β4GalT1

β1,4‐galactosyltransferase‐1

GlcNAc

N‐acetylglucosamine

IgG

immunoglobulin G

IL‐6

interleukin‐6

BSII

Bandeiraea simplicifolia II

RCAI

Ricinus communis agglutinin I

RPA

ribonuclease protection assay

GAPDH

glyceraldehyde 3‐phosphate dehydrogenase

VSV

vesicular stomatitis virus