Glycan analysis of monoclonal antibodies secreted in deposition disorders indicates that subsets of plasma cells differentially process IgG glycans

Authors


Abstract

Objective

To compare the glycosylation of polyclonal serum IgG heavy chains in a patient with rheumatoid arthritis (RA) with that of monoclonal serum IgG heavy chains in the same patient during an episode of heavy-chain deposition disease (HCDD), to establish whether glycosylation processing is specific for subsets of B cells.

Methods

Serum IgG was purified using a HiTrap protein G column. Immunoglobulins were run on sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels, and IgG glycans were isolated from gel bands and fluorescently labeled. Glycans were analyzed by normal-phase high-performance liquid chromatography and by liquid chromatography–electrospray ionization–mass spectrometry.

Results

The glycosylation of serum immunoglobulins from a patient with seronegative RA and HCDD was analyzed. The predominant immunoglobulin was a truncated glycosylated γ3 heavy chain, and a small amount of polyclonal IgG was also present. The glycan profile showed that the monoclonal γ3 heavy chain contained fully galactosylated biantennary glycans with significantly less fucose but more sialic acid than in IgG3 from healthy controls. In contrast, the polyclonal IgG showed an RA-like profile, with a predominance of fucosylated biantennary glycans and low levels of galactosylation. The glycan profile of serum IgG obtained from the same patient during disease remission resembled a typical RA profile.

Conclusion

These data indicate that different types of B cells process a particular set of IgG glycoforms.

Monoclonal immunoglobulin deposition disease (MIDD) represents instances in which an excessive amount of immunoglobulin from a single clone of B cell lineage becomes pathogenic because it is deposited in tissue in a manner and quantity sufficient to compromise organ function (1). All forms of MIDD involve pathogenic deposition of immunoglobulins that have structural characteristics predisposing them to tissue deposition in either a fibrillar or nonfibrillar manner (2). The diagnosis of MIDD is based on identification and immunohistochemical characterization of deposits and Congo red staining of affected tissue. The major distinction is that only the fibrillar (amyloid) forms stain with Congo red (1).

The pathogenesis of the various forms of MIDD involves several processes during which a soluble cellular product becomes insoluble under physiologic conditions. Currently, most therapy is directed toward the proliferative step. Even in the absence of clinically bona fide multiple myeloma, the outcome of these diseases is often fatal. Nonamyloid immunoglobulin deposition may consist of either a monoclonal heavy chain together with light chains or an incomplete heavy chain devoid of a light chain (3). In all cases described thus far, the monoclonal free heavy chain had a lower molecular weight compared with its normal counterpart (4). Heavy-chain deposition disease (HCDD) is probably considerably underdiagnosed, and its true incidence and prevalence are uncertain (5). Regarding amyloids, the kidneys are consistently involved in HCDD, and renal function deteriorates rapidly. Other organs such as the heart and liver are also commonly affected (6).

The majority of biologically active proteins are glycosylated, and glycan processing can alter with disease. For example, in rheumatoid arthritis (RA) the oligosaccharide profile of the IgG molecule is altered, producing agalactosyl glycoforms that, when clustered, are able to be recognized by the mannose-binding lectin system and thus may provide a route to inappropriate activation of the complement system (7). All immunoglobulins have at least 1 conserved glycosylation site in their heavy chains. Human serum polyclonal IgG, which contains a conserved site at Asn297 in each heavy chain, consists of a mixture of at least 30 glycoforms (8). In healthy individuals, the molar ratios of these glycoforms fall within a narrow range (9). Myeloma IgG and normal IgG contain the same glycoforms, but their molar ratios differ very significantly (10, 11). This suggests that each B cell type processes a particular set of oligosaccharides (9), and that the ratio of B cells producing different sets of glycoforms in a healthy individual is constant. The reason for the altered glycosylation of IgG in RA is not yet fully understood. The altered glycosylation may have a biosynthetic origin in which there is an overall decrease in galactosyltransferase activity (12), or each B cell type may process a particular set of N-glycans such that the development of the RA profile may be attributable to a selected expansion of specific subsets of plasma cells with low expression levels or low activity of galactosyltransferase (9).

We analyzed the glycan profiles of monoclonal and polyclonal IgG from a patient with HCDD, together with the glycan profile of the patient's polyclonal IgG after her recovery. These data indicate that different types of B cells process a particular set of IgG glycoforms, and that the ratio of these B cells is relatively constant in a healthy individual but may be perturbed in diseases such as HCDD and RA as a result of a selected expansion of specific subsets of plasma cells.

PATIENTS AND METHODS

The patient.

The patient, a Caucasian woman, was born in 1948. In 1988, seronegative RA meeting the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 revised criteria (13) was diagnosed. Three years later, a monoclonal immunoglobulin component (free γ3 heavy chain) was detected in her serum and urine (4), but no anti-IgG antibodies reactive to the monoclonal γ3 heavy chain were found. The dimers of the free heavy chain consisted of the hinge region together with CH2 and CH3 domains and lacked both the variable region and the CH1 domain. This free heavy chain was found deposited in the synovial tissue. The HCDD was clinically considered to be relatively benign, and the patient received no specific treatment until 1996, when it was concluded that the disease was taking a malignant course. The patient was therefore started on a treatment protocol for multiple myeloma (4), which produced a sufficient response such that the patient's disease is now in complete remission.

Purification of the monoclonal immunoglobulin heavy chain from the patient during active HCDD and polyclonal IgG from the same patient after recovery.

Serum IgG from the patient with HCDD, samples of which were obtained during both a period of health and an active disease state, was purified using a HiTrap Protein G column (Amersham Biosciences, Uppsala, Sweden). The binding buffer was 20 mM sodium phosphate (pH 7.0), and the elution buffer was 0.1M glycine HCl (pH 2.7), used according to the manufacturer's protocol. The samples were desalted using a HiTrap column (Amersham Biosciences) with 0.2M NH4HCO3. Protein A–Sepharose beads (Amersham Biosciences) were used for further purification of IgG-HCDD. Briefly, the IgG was added to a protein A–Sepharose suspension, and the mixture was incubated for 3 hours. The beads, containing protein A–bound IgG, were then thoroughly washed. The supernatant from the first wash was collected (non–protein A–bound IgG).

Oligosaccharide release and analysis.

Glycan isolation from gel bands.

The immunoglobulins were run on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gels, the proteins were visualized by Coomassie brilliant blue staining, and the gel bands were excised (14). The gel pieces were washed alternately with acetonitrile and 20 mM NaHCO3 (pH 7), then dried in a vacuum centrifuge. The dried gel pieces were subjected to peptide N-glycosidase F (PNGase F; Roche Diagnostics, Mannheim, Germany) digestion at 37°C overnight. After incubation, the glycans were extracted from the gel by sonication, alternately in water and acetonitrile (14). The samples were desalted using AG-50W-X12, H+ form (Bio-Rad, Hercules, CA), and dried in a vacuum centrifuge.

Fluorescence labeling (15).

The released glycans were labeled with 2-aminobenzamide (2-AB) using a LudgerTag 2-AB labeling kit (Ludger Ltd., Oxford, UK). Excess label was removed by ascending chromatography in acetonitrile on Whatmann 3MM paper strips.

Exoglycosidase digestion.

The 2-AB–labeled glycan was digested with arrays of enzymes at standard concentrations in 10 μl 50 mM sodium acetate buffer (pH 5.5), overnight at 37°C. The enzymes used were Arthrobacter ureafaciens sialidase (EC 3.2.1.18; 1–2 units/ml), which removes sialic acids; bovine kidney α-fucosidase (EC 3.2.1.51; 1 unit/ml), which removes the core fucose; bovine testes β-galactosidase (EC 3.2.1.23; 2 units/ml), which removes galactoses; recombinant Streptococcus pneumoniae glucosaminidase in Escherichia coli (code GE31, EC 3.2.1.30; 4 units/ml), which removes N-acetylglucosamine (GlcNAc) except for the bisecting GlcNAc linked β1-4 to mannose. Enzymes were purchased from Prozyme (via Europa Bioproducts Ltd., Cambridge, UK). After digestion, enzymes were removed using protein-binding Micropure-EZ filters (Millipore, Watford, UK).

Glycan analysis by normal-phase high-performance liquid chromatography (NP-HPLC).

The labeled glycans were analyzed by NP-HPLC using a 4.6 × 250–mm TSK gel amide-80 column (Hichrom, Reading, UK) with a 20–58% gradient of 50 mM ammonium formate (pH 4.4) versus acetonitrile. The system was calibrated using an external standard of hydrolyzed and 2-AB–labeled glucose oligomers (16). Waters Alliance 2690 separation modules equipped with Waters 474 fluorescence detectors were used (Waters, Milford, MA).

Glycan analysis by liquid chromatography–electrospray ionization–mass spectrometry (LC-ESI-MS).

Glycans were analyzed using an LC Packings Ultimate HPLC system equipped with a Famos autosampler (Dionex, Leeds, UK) interfaced with a Q-Tof Ultima Global mass spectrometer (Waters Micromass, Manchester, UK). Chromatographic separation was achieved using a 2 × 250–mm microbore NP-HPLC TSK gel amide-80 column (Hichrom) with the same gradient and solvents as used with the standard NP-HPLC but at a lower flow rate of 40 μl/minute (17). The mass spectrometer was operated in positive-ion mode with 3-kV capillary voltage and the following parameters: 60-mm RF lens, source temperature 100°C, desolvation temperature 150oC, cone gas flow 50 liters/hour, desolvation gas flow 450 liters/hour.

RESULTS

Protein isolation.

We analyzed serum IgG isolated from the patient during active HCDD and from the same patient after recovery from HCDD. During active HCDD, the patient had an overproduction of a monoclonal truncated γ3 heavy chain (35–54 gm/liter) (4). Serum IgG was isolated using a protein G affinity column. This column interacts with the Fc region of IgG and, therefore, the predominant immunoglobulins purified were the monoclonal truncated γ3 heavy chain in samples isolated from the patient during active HCDD and polyclonal IgG in samples isolated from the same patient after recovery. Purified proteins were run on a 12.5% SDS-PAGE gel under reducing conditions and visualized by Coomassie brilliant blue staining. The gel showed that the major protein isolated from the patient during active HCDD had an apparent molecular mass of 50 kd (Figure 1a, lane 3). This molecular mass is lower than that of a normal IgG3 heavy chain (60 kd) and corresponds to that of the truncated immunoglobulin monoclonal heavy chain (γ3 heavy chain) that was overproduced in the patient (4). The polyclonal IgG purified from the patient after recovery from HCDD had IgG heavy chains (of which IgG3 comprises 5–10% [18]) of normal molecular weight (apparent molecular mass 54 kd) and normal light chains (30 kd) (Figure 1a, lane 2).

Figure 1.

a, Results of sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) (12.5% gel) showing heavy chains in serum samples obtained from the patient during active heavy-chain deposition disease (HCDD) (lane 3) and after recovery (lane 2). These gel pieces were excised, and N-glycans were released with peptide N-glycosidase F. b and c, SDS-PAGE (4–12% XT gel) of protein A–bound IgG from the patient with HCDD (lane 2) and non–protein A–bound IgG from the same patient (lane 3). The proteins were run under nonreducing (b) or reducing (c) conditions. In a, b, and c, lane 1 shows molecular weight markers.

Purified IgG obtained from the patient during an episode of HCDD was subjected to protein A–Sepharose purification to separate the monoclonal IgG (γ3 heavy chain) from the polyclonal IgG of subgroups 1, 2, and 4, which bound to the column. The protein A–bound material and 2% of the non–protein A–bound material were run on a nonreducing SDS-PAGE gel (Figure 1b) and on a reducing SDS-PAGE gel (Figure 1c). The protein A–bound IgG consisted mainly of a protein with an apparent molecular mass of 155 kd (Figure 1b). The non–protein A–bound IgG contained a protein with an apparent molecular weight of 83 kd (dimers of IgG in HCDD) together with a protein of a higher apparent molecular mass of 155 kd (polymers of IgG in HCDD) (Figure 1b).

N-glycosylation during active HCDD.

N-glycans were released from the heavy chain of the gel band during active HCDD by PNGase F digestion and fluorescently labeled with 2-AB. The released glycans were analyzed (before and after exoglycosidase digestions) using NP-HPLC and LC-ESI-MS. NP-HPLC of the glycans following exoglycosidase treatment (Table 1) showed that the dominant structures were the fully galactosylated structures A2G2 and A2G2S, with a sialic acid, and F(6)A2G2S, with a sialic acid and a core fucose. Sialidase digestion showed that ∼50% of the N-glycans during active HCDD were sialylated, and that 50% of them were neutral. Further digestion with both sialidase and fucosidase showed a reduced amount of core fucosylated structures compared with normal polyclonal IgG and IgG3 (19), because only 37% of the structures were fucosylated compared with 90% in normal IgG (20) and 84% in IgG3 (19). Further digestion with galactosidase and N-acetylglucosaminidase demonstrated that ∼10% of the N-glycans contained a bisecting N-acetylglucosamine residue. LC-ESI-MS analysis confirmed the preliminary oligosaccharide assignments (Figure 2a and Table 2).

Table 1. Exoglycosidase digestion of N-glycans from the patient during active heavy-chain deposition disease*
Peak no.StructureGUUndigestedExoglycosidase treatment
ABSABS + BKFABS + BTGABS + BTG + BKFABS + BTG + BKF + GUH
  • *

    Except where indicated otherwise, values are the percent area. All N-glycans have 2 core N-acetylglucosamines (GlcNAc) and a trimannosyl core. GU = glucose unit; ABS = Arthrobacter ureafaciens sialidase; BKF = bovine kidney α-fucosidase; BTG = bovine testes β-galactosidase; GUH = Streptococcus pneumoniae glucosaminidase; Man3 = the trimannosyl core alone; F(6) = core fucose linked α1–6 to inner GlcNAc; A = number of antenna (GLcNAc) on trimannosyl core; B = bisected GLcNAc linked β1–4 to inner mannose; G = number of galactose on antenna; S = number of sialic acids; G1[6] = galactose on the mannose 1–6 arm; G1[3] = galactose on the mannose 1–3 arm.

1Man34.42     84.28
2 4.55     2.31
3Man3B5.01   1.362.3813.41
4F(6)A15.26   3.253.83 
5A1B5.40      
6A25.508.448.3611.8648.2773.39 
7 5.63   1.452.37 
8A2B5.831.751.762.6510.1118.03 
9F(6)A25.921.992.15 26.84  
10F(6)A2B6.24   8.71  
11A2G1[6]6.284.945.176.25   
12a/bA2G1[3]/A2BG1[6]6.386.927.889.47   
13A2BG1[3]6.532.522.924.40   
14F(6)A2G1[6]6.690.710.94    
15a/bF(6)A2G1[3]/F(6)A2BG1[6]6.810.400.70    
16F(6)A2BG1[6]6.910.560.72    
17A2G27.1711.9129.7650.77   
18A2BG27.323.928.0014.59   
19F(6)A2G27.594.3323.69    
20F(6)A2BG27.701.947.96    
21A2G2S8.0913.03     
22A2BG2S8.322.52     
23F(6)A2G2S8.4812.92     
24F(6)A2BG2S8.683.47     
25A2G2S29.015.25     
26A2BG2S29.131.95     
27F(6)A2G2S29.387.82     
28F(6)A2BGS29.442.72     
Figure 2.

Relative abundance of the masses of N-glycans listed in Table 2 against retention time, on micro–normal-phase high performance liquid chromatography–electrospray ionization–mass spectrometry. a and b, Release of N-glycans during active heavy-chain deposition disease (a) and following recovery (b) of the patient.

Table 2. Masses of N-glycans detected by LC-ESI-MS*
Peak no.Mass [M + 2H]2+Calculated massOligosaccharide componentNo. of 2-AB labeling moleculesStructure
Active HCDDAfter recovery from HCDDNo. of HexNo. of HexNacNo. of FucNo. of NeuNAc
  • *

    Peak numbers correspond to those listed in Table 1. LC-ESI-MS = liquid chromatography–electrospray ionization–mass spectrometry; HCDD = heavy-chain deposition disease; Hex = hexose; HexNac = N-acetyl hexoseamine; Fuc = fucose; NeuNAc = N-acetyl neuraminic acid; 2-AB = 2-aminobenzamide; ND = not detected (see Table 1 for other definitions).

6719.23ND719.2934001A2
8820.82ND820.8335001A2B
9ND792.33792.3234101F(6)A2
10ND893.88893.8535101F(6)A2B
11, 12a800.34ND800.3144001A2G1
12b, 13901.90ND901.8545001A2BG1
14, 15a873.37873.37873.3444101F(6)A2G1
15b, 16974.92974.91974.8845101F(6)A2BG1
17881.37881.38881.3454001A2G2
18982.87ND982.8855001A2BG2
19954.39954.39954.3754101F(6)A2G2
201,055.951,056.001,055.9155101F(6)A2BG2
211,026.92ND1,026.8954011A2G2S
221,128.45ND1,128.4355011A2BG2S
231,099.931,099.941,099.9154111F(6)A2G2S
241,201.50ND1,201.4655111F(6)A2BG2S
251,172.44ND1,172.4354021A2G2S2
261,274.02ND1,273.9755021A2BG2S2
271,245.49ND1,245.4654121F(6)A2G2S2
281,346.99ND1,347.0055121F(6)A2BG2S2

The neutral glycans linked to IgG may be classified according to the number of terminal galactoses attached; that is, G0, G1, and G2, which contain 0, 1, and 2 terminal galactose residues, respectively (7). The percentages of G0, G1, and G2 structures were calculated from the results in Table 1, after A ureafaciens sialidase and bovine kidney α-fucosidase digestion. The N-glycans during active HCDD contained a high percentage of G2 structures (71%) and less G1 (16%) and G0 (12%) structures compared with typical values for healthy controls (for G2, 37%; for G1, 41%; for G0, 22%) (20). The glycosylation profile in the patient's urine isolated from the patient during active HCDD was the same as that in the patient's serum (21). N-glycans released from protein A–bound and nonbound samples (protein band with the apparent molecular mass of 155 kd) were profiled using NP-HPLC (Figures 3e and f), showing a high content of G0 structures in the protein A–bound IgG and a profile similar to that in the patient during active HCDD for the non–protein A–bound IgG3.

Figure 3.

Normal-phase high-performance liquid chromatography profiles of N-glycans from IgG heavy chains in a pool of healthy individuals (a), a patient with rheumatoid arthritis (RA; not the present study patient) (b), the present study patient following recovery (REC) from heavy-chain deposition disease (HCDD) (c), the present study patient during active HCDD (d), protein A–binding fraction from the present study patient (e), and protein A–nonbinding fraction from the present study patient (f). The N-glycan structures are represented by N-acetylglucosamine (squares), galactose (open diamonds), fucose (diamonds with a dot inside), mannose (circles), and sialic acid (stars). The linkages between sugars are represented by β linkage (solid lines), α linkage (broken lines), unknown linkage (wavy lines), 1–-2 linkage (vertical lines), 1–3 linkage (right-angled broken lines), 1–4 linkage (horizontal lines), and 1–6 linkage (left-angled broken lines).

N-glycosylation following the patient's recovery from HCDD.

The N-glycans released following the patient's recovery from HCDD were analyzed by NP-HPLC and LS-ESI-MS. The micro–NP-HPLC-ESI-MS analysis of samples obtained after recovery from HCDD (Figure 2b and Table 2) showed that the dominant structures were core fucosylated nongalactosylated structures (F[6]A2) and core fucosylated structures with 1 galactose (F[6]A2G1). Further analysis by NP-HPLC with fluorescence detection (Table 3) gave an estimated percentage of the 2 glycans of 47% and 18%, respectively. Digestion with sialidase demonstrated that ∼4% of the N-glycans following the patient's recovery from HCDD were sialylated. Further digestion with both sialidase and fucosidase showed that the glycans were extensively fucosylated (97%). The N-glycan profile consisted of a high percentage of G0 structures (for G0, 58%; for G1, 27%; for G2, 15%), which is very similar to an RA profile (8).

Table 3. Exoglycosidase digestion of N-glycans from the IgG heavy chain of the patient following recovery from heavy-chain deposition disease*
Peak no.StructureGUUndigestedExoglycosidase treatment
ABSABS + BKFABS + BTGABS + BTG + BKFABS + BTG + BKF + GUH
  • *

    Except where indicated otherwise, values are the percent area. All N-glycans have 2 core GlcNac and a trimannosyl core. See Table 1 for definitions.

1Man34.42     87.42
2 4.55     1.60
3Man3B5.01  1.67 2.72 
4F(6)A15.26   1.2331.590.60
5A1B5.40      
6A25.502.441.9245.626.0379.43 
7 5.63  0.97 1.61 
8A2B5.83  7.991.4112.97 
9F(6)A25.9247.0544.99 78.181.69 
10F(6)A2B6.248.419.09 13.15  
11A2G1[6]6.28  17.03   
12a/bA2G1[3]/A2BG1[6]6.38  7.74   
13A2BG1[3]6.53  4.33   
14F(6)A2G1[6]6.6917.7417.47    
15a/bF(6)A2G1[3]/F(6)A2BG1[6]6.816.307.67    
16F(6)A2BG1[6]6.913.354.10    
17A2G27.170.311.1312.17   
18A2BG27.32  2.49   
19F(6)A2G27.598.0011.49    
20F(6)A2BG27.702.372.13    
21A2G2S8.09      
22A2BG2S8.32      
23F(6)A2G2S8.484.03     
24F(6)A2BG2S8.683.47     
25A2G2S29.015.25     
26A2BG2S29.131.95     
27F(6)A2G2S29.387.82     
28F(6)A2BGS29.442.72     

DISCUSSION

The IgG purified from the patient during active HCDD consisted mainly of the monoclonal truncated IgG of subgroup 3, but polyclonal IgG was also present. The glycans from those 2 sets of IgG were substantially different, which is in contrast to myeloma sera, in which the N-glycans of both paraprotein and polyclonal IgG were very similar (22). The glycan profile of the monoclonal HCDD protein is similar to Fab glycosylation in that it has a predominant G2 structure and is sialylated to a greater extent than are normal Fc glycans. However, these glycans do not come from glycosylation of Fab, because the HCDD protein is devoid of Fab domains. In contrast, polyclonal IgG has a profile very similar to an RA profile (i.e., they have a high content of G0 structures). Consistent with this finding, lower amounts of the same extended sialylated N-glycans found on the monoclonal IgG were found on the polyclonal IgG (50% and 25% of the total glycans, respectively), reflecting their different content of galactosylated structures.

The 3-dimensional structure of the IgG molecule limits the complexity of its glycans. The finding that the monoclonal HCDD protein has a higher incidence of G2 structures therefore implies that there is more accessibility for the glycosyltransferases than in normal Fc regions, even though normal IgG3 has a higher incidence of G2 structures (23) compared with the other subclasses. In a study of IgG3, Jefferis et al (24) showed that disease-associated glycosylation changes in IgG may be associated with alterations in amino acid sequences. No unusual amino acid sequences were present in the chain during active HCDD (25), but it was a truncated, abnormal dimeric IgG. Therefore, it cannot be excluded that the 3-dimensional structure of the CH2 and CH3 domains may have been altered, providing more accessibility for the glycosyltransferases. However, traces of extended glycan structures were also observed on the full-size, polyclonal IgG, indicating an overall tendency for highly sialylated structures during the disease state. Furthermore, the accessibility of glycosyltransferases that add core fucose does not correlate with changes in the 3-dimensional protein structure, so that changes in core fucosylation may best be explained by differences in the levels (or types with different Km values for guanosine diphosphatefucose) of fucosyl transferases. The overall tendency for highly sialylated structures to be processed during the disease state, together with less core fucosylation, strongly indicates that the difference in galactosylation between the monoclonal and polyclonal IgG cannot be attributed to differences in the accessibility of the glycosidases to the glycosylation site.

The molar ratios of each glycan in IgG samples obtained from healthy individuals fall within a narrow range (9), while the molar ratio of oligosaccharide from myeloma IgG differs from person to person (10, 11). This suggests that each B cell type processes a particular set of N-glycans (9), and that the ratio of B cells producing different sets of glycoforms in a healthy individual is relatively constant. The reason for the altered glycosylation of IgG in RA is not yet fully understood. In peripheral blood B and T lymphocytes isolated from patients with RA, there is an overall decrease in galactosyltransferase activity (12, 26–28), which is consistent with the overall decreased galactosylation of serum IgG. This decrease is also seen in animal models of the disease (29, 30). It has been shown that B cells have different levels of galactosyltransferase activities (19), and that RA is associated with differential expression of GTase isoforms (27). Furthermore, it has been proposed that development of the RA profile may be attributable to a selected expansion of specific subsets of plasma cells with low expression levels of galactosyltransferases (9).

Our present study strongly supports the latter theory. The IgG purified from the patient during active HCDD consisted mainly of the monoclonal truncated IgG of subgroup 3, but polyclonal IgG was also present. The N-glycans from those 2 sets of IgG are substantially different, with monoclonal IgG having its unique profile and polyclonal IgG having a profile very similar to an RA profile, with a high content of G0 structures. After treatment with prednisolone and melphalan, there was no longer an overproduction of the plasma cells producing truncated heavy chains, but only a heterogeneous set of cells producing polyclonal IgG. The N-glycan profile after recovery from HCDD is similar to an RA profile. During HCDD, the patient had an overproduction of monoclonal B cells, with a specific set of glycosyltransferases giving an oligosaccharide profile with little fucosylation and more galactosylation than that from the polyclonal IgG together with polyclonal plasma cells, which produce an RA oligosaccharide profile. If the oligosaccharide profile in RA were attributed to an overall lower galactosyltransferase activity, one would expect that the monoclonal IgG (during active HCDD) would also have a low incidence of galactosylated oligosaccharides, but this is not the case.

Consequently, these results are consistent with our earlier data indicating that glycan profiling represents an opportunity for additional diagnostics in immunoglobulin-related diseases (31) as well as in diseases involving other glycoproteins, such as human pancreatic ribonuclease in prostate cancer (32) or the prion protein involved in spongiform encephalopathies (33).

Acknowledgements

We thank Azita Alavi for useful discussions.

Ancillary