SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Objective

Rheumatoid arthritis is associated with an excess of agalactosylated (G0) IgG that is considered relatively proinflammatory. Assessment of this association in juvenile idiopathic arthritis (JIA) is complicated by age-dependent IgG glycan variation. The aim of this study was to conduct the first large-scale survey of IgG glycans in healthy children and patients with JIA, with a focus on early childhood, the time of peak JIA incidence.

Methods

IgG glycans from healthy children and disease-modifying antirheumatic drug–naive patients with JIA were characterized using high-performance liquid chromatography. Agalactosylated glycans were quantitated with reference to monogalactosylated (G1) species. Associations were sought between the G0:G1 ratio and disease characteristics.

Results

Among healthy children ages 9 months to 16 years (n = 165), the G0:G1 ratio was highly age dependent, with the ratio peaking to 1.19 in children younger than age 3 years and declining to a nadir of 0.83 after age 10 years (Spearman's ρ = 0.60, P < 0.0001). In patients with JIA (n = 141), the G0:G1 ratio was elevated compared with that in control subjects (1.32 versus 1.02; P < 0.0001). The G0:G1 ratio corrected for age was abnormally high in all JIA subtypes (enthesitis-related arthritis was not assessed), most strikingly in systemic JIA. Glycosylation aberrancy was comparable in patients with and those without antinuclear antibodies and in both early- and late-onset disease and exhibited at most a weak correlation with markers of inflammation.

Conclusion

IgG glycosylation is skewed toward proinflammatory G0 variants in healthy children, in particular during the first few years of life. This deviation is exaggerated in patients with JIA. The role for IgG glycan variation in immune function in children, including the predilection of JIA for early childhood, remains to be defined.

Juvenile idiopathic arthritis (JIA) is characterized by a heterogeneous mixture of arthritis phenotypes of unknown etiology. HLA associations have been observed for multiple subtypes of JIA, suggesting a role for adaptive immune mistargeting (1, 2). The participation of antibodies in disease pathogenesis is supported by several observations, including the prevalence of autoantibodies such as antinuclear antibodies (ANAs) and anti-DEK antibodies, circulating immune complexes, and complement consumption in the blood and joint fluid of some patients (3–6).

One remarkable feature of the epidemiology of JIA is age at onset. The peak incidence of JIA occurs at ∼2–3 years of age (1, 7, 8). This peak is represented primarily by patients in the subgroup with oligoarticular disease but also by patients with seronegative polyarticular JIA, psoriatic JIA, and potentially systemic JIA (7, 9–11). In contrast, onset before age 1 year is unusual. The basis for this epidemiologic pattern is unknown. One possibility is that it reflects the time point of first contact between a genetically susceptible host and a specific environmental trigger. Supporting this concept is the observation that younger patients with JIA and older patients with JIA exhibit different HLA associations, even within the same International League of Associations for Rheumatology (ILAR) subtype (1, 2).

A complementary possibility is that some feature of early childhood immunity may favor the initiation of arthritis. Children and adults differ immunologically in multiple respects (12). For example, the proportion of circulating lymphocytes that are naive is initially high and falls gradually. The level of circulating IgG reaches the lowest point at ∼3–4 months of age with the decay of maternal antibodies, then increases to 60% of adult levels by 1 year of age and to adult levels by 10 years of age (13). Humoral immunity is also functionally immature in young children, with the most striking evidence being poor responses to polysaccharide antigens in the first 18–24 months of life (14).

Another potentially important difference between pediatric and adult immunity concerns IgG glycosylation. Approximately 3% of the mass of IgG is carbohydrate, representing principally 2 branched glycans that attach to a canonical asparagine (Asn297) in each heavy chain (Figure 1A) (15). These oligosaccharides reside within the Fc region, where they help to maintain its 3-dimensional conformation (16). Polymorphisms in IgG glycans modulate the ability of the Fc domain to bind Fc receptors and fix complement and thus are highly determinative of antibody effector function (15, 17). In particular, IgG glycans lacking galactose (G0) bind mannose-binding lectin, thereby facilitating activation of complement (Figure 1B) (18–20). Interestingly, one analysis of IgG glycans in healthy individuals demonstrated that children, like older adults, exhibited an excess of proinflammatory agalactosylated IgG glycoforms (21), although a smaller survey did not identify this pattern (22).

thumbnail image

Figure 1. Structure of IgG-associated N-glycans. A, Biantennary (2-armed) structure of the IgG Fc region of asparagine-linked glycans, with core pentasaccharide (stem composed of 2 N-acetylglucosamines [GlcNAc] and 3 mannoses) elaborated variably with additional sugars as noted, drawn using the Oxford nomenclature (47). B, Major structures present in agalactosylated (G0) and monogalactosylated (G1) fractions. Note core fucosylation but the absence of bisecting GlcNAc or terminal sialic acid, while a single galactose (if present) may be attached to either the 2,3 arm or the 2,6 arm.

Download figure to PowerPoint

Hypogalactosylation of IgG may be particularly relevant for arthritis. Studies over the past 30 years have demonstrated that adult rheumatoid arthritis (RA) is associated with an increased prevalence of circulating G0 IgG (23–29). Furthermore, elevated G0 IgG expression can predate the diagnosis of RA by more than 3 years, varies directly with disease activity, and is particularly pronounced for antibodies directed against citrullinated peptides (28, 30). Finally, animal studies have suggested that antibodies rendered G0 by enzymatic techniques are especially arthritogenic (31). Taken together, these studies support the notion that aberrant IgG glycosylation has a role in the pathogenesis of RA.

A similar mechanism may be operative in JIA. Several groups of investigators have studied IgG glycans in patients in whom the onset of arthritis occurred before age 16 years (24–26, 32, 33). Using diverse methodologies, these series have generally (although not invariably [32]) led to the conclusion that IgG hypogalactosylation is a feature of childhood arthritis. However, the data have important limitations. First, the number of patients studied was relatively small (n = 100) and was divided among multiple phenotypic classes using diverse nomenclatures. Second, most patients provided samples while they were receiving treatment, which is known to alter IgG glycosylation (27, 29). Finally, and most critically, patients were not always matched to control subjects by age. Indeed, very few samples from healthy children younger than age 6 years, the peak of JIA onset, have been characterized, rendering IgG glycan data from young patients with JIA exceptionally difficult to interpret (21, 22).

In the present study, we set out to characterize IgG galactosylation in JIA, with a particular focus on early childhood. To achieve this goal, we collected samples from healthy children and from patients with JIA who were naive to disease-modifying antirheumatic drug (DMARD) treatment, enabling age-normalized analysis of IgG glycan variation by disease subtype, laboratory features, and age at onset. Our results establish the association of JIA with aberrant IgG glycosylation and demonstrate markedly elevated expression of agalactosylated IgG in healthy young children, suggesting a possible mechanism favoring onset of JIA in early childhood.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Patients.

Samples from healthy children without JIA were obtained from 4 cohorts: 1) patients without inflammatory disease who were evaluated at the Rheumatology Program at Children's Hospital Boston (CHB), 2) patients recruited from the primary care clinic at CHB, 3) patients undergoing lead screening at primary care practices referring to the CHB clinical laboratory, in whom measured lead levels were normal (<6 μg/dl) (discard specimens), and 4) healthy children recruited for a biorepository of pediatric biosamples at Cincinnati Children's Hospital Medical Center (CCHMC). For some of the healthy children, age data available for review were limited to integer year of age (e.g., 1 year, 2 years). For these patients, an integer year of age plus 0.5 years (e.g., 1.5 years, 2.5 years) was assumed in calculations and plots.

JIA samples were obtained from biorepositories at CHB and CCHMC. Entry criteria for JIA patients included a definitive diagnosis of JIA according to ILAR criteria (34), no DMARD therapy within 6 months of sampling, and no concurrent systemic illness. Almost all of the patients with JIA had recent-onset disease (the mean interval from symptom onset to sampling was 0.5 year). Samples from healthy adults were obtained from consenting volunteers. These studies were performed in accordance with the requirements of the respective institutional review boards.

Glycan characterization.

Glycans were analyzed as described previously (28, 35). Briefly, N-glycans were liberated enzymatically from 5 μl of whole serum or plasma, labeled with a fluorescent tracer, and analyzed by normal-phase high-performance liquid chromatography (HPLC), which provides precise relative quantitation of molecular species separated by size and charge. The area under the peaks corresponding to G0 glycans were normalized to the area under the peaks corresponding to G1 structures after confirming that the percent G1 value is independent of age, including into early childhood (data not shown). Because the large majority of neutral biantennary glycans in serum are from IgG, the G0:G1 ratio in whole serum or plasma is a proxy for the IgG G0:G1 ratio (R2 = 0.83) (35).

Inflammation markers.

C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) studies were performed at the clinical laboratories of each institution. S100A12, myeloid-related protein 8 (MRP-8), and MRP-14 were assessed in serum from 69 CCHMC patients with JIA by enzyme-linked immunosorbent assay (36).

Statistical analysis.

Population means were compared with Student's 2-tailed t-test, using GraphPad Prism version 4.0. Data are presented as the mean ± SD. Linear regression to correct for age was performed using PASW Statistics 18 (release 18.0.0).

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Equivalence of sampling conditions.

To ensure comparability of the G0:G1 data obtained under different sampling conditions, blood was drawn from healthy adult volunteers and processed in duplicate using a range of collection conditions mirroring those used for the study samples. As expected, given the known stability of IgG glycans, no difference among conditions was apparent (Figure 2). In 2 of 100 tests, one replicate diverged by >10% while the second was consistent with all other results from the same donor, corresponding to an error rate (2%) concordant with our previous data on the reproducibility of this technique (28). We concluded that samples from all pediatric sources could be used together in our analysis.

thumbnail image

Figure 2. IgG glycans are stable across different collection conditions. Blood obtained from 5 healthy adult donors (shown as A, B, C, D, and E) was drawn into the collection tubes used for the control subjects and arthritis patients, including plain glass and EDTA-coated, sodium heparin–coated, and lithium heparin–coated plasma separator tubes (PSTs). Blood in plain glass tubes was allowed to clot for 2 hours at room temperature, followed by centrifugation to separate serum. For the other tubes, plasma was obtained by centrifugation either after 2 hours or after storage at room temperature for 24 hours (1d) or 48 hours (2d). All samples were frozen at −80°C and were thawed together for simultaneous glycan analysis by high-performance liquid chromatography. G0/G1 = agalactosylated IgG:monogalactosylated IgG ratio. Bars show the mean ± SD.

Download figure to PowerPoint

Age-dependence of normal IgG glycosylation in childhood.

One hundred sixty-five samples were available from children without known inflammatory disease, ranging in age from 9 months to 16 years. The G0:G1 ratios demonstrated a clear age-dependent decrease over time, from a mean ± SD ratio of 1.19 ± 0.22 in children younger than age 3 years to a nadir at 0.83 ± 0.14 after age 10 years (Spearman's ρ = 0.60, P < 0.0001) (Figure 3). When we compared these results with our published glycan data from healthy adult blood donors (28), the mean G0:G1 ratios of children younger than age 3 years were equivalent to those of adults ages 60–70 years, while older children (age ≥10 years) exhibited glycosylation equivalent to that observed in adults younger than age 40 years. These results confirmed the earlier observations that the proportion of IgG glycans that are G0 exhibits a U-shaped distribution with age, highest in children and older adults, extending these observations with high resolution to early childhood (21). No overall difference between sexes was evident, although both the highest and lowest G0:G1 values were observed in female children.

thumbnail image

Figure 3. Hypogalactosylation of IgG glycans in healthy children. The agalactosylated IgG:monogalactosylated IgG (G0:G1) ratios in healthy children, as determined by high-performance liquid chromatography, varied according to age (Spearman's ρ = 0.60, P < 0.0001). Each data point represents a single subject (n = 165).

Download figure to PowerPoint

IgG glycosylation in JIA.

Samples were obtained from 141 patients with JIA, representing all ILAR subtypes except enthesitis-related arthritis. Overall, IgG glycosylation in these samples differed markedly from that in controls (Figure 4A). The mean ± SD G0:G1 ratio in samples from patients with JIA was comparable with that in samples from adult patients with RA (1.32 ± 0.40 versus 1.36 ± 0.43) (28). Aberrant IgG glycosylation was observed across the age spectrum and was particularly marked in older children with JIA (Figure 4B).

thumbnail image

Figure 4. IgG glycosylation in juvenile idiopathic arthritis (JIA). A, Agalactosylated IgG:monogalactosylated IgG (G0:G1) ratios in all patients with JIA compared with those in control subjects ages 1 year to <16 years. Control subjects younger than age 1 year were excluded, because no case of JIA began before age 1 year. The mean ± SD G0:G1 ratios in JIA and control samples were significantly different (1.32 ± 0.40 versus 1.02 ± 0.21, respectively; P < 0.0001). Bars show the mean. B, Distribution of G0:G1 values among patients with JIA across the age spectrum (up to age 16 years). Each data point represents a single subject (n = 141 patients with JIA and n = 165 controls).

Download figure to PowerPoint

ILAR criteria and alternate classification methods.

We next categorized patients according to ILAR classification criteria (34). Compared with pediatric controls as a whole, elevation of the G0:G1 ratio was evident in all subtypes analyzed with the exception of extended oligoarthritis, for which few samples were available (Figure 5A). Given the divergent age distributions among patients and control subjects, we used linear regression analysis to adjust for age, confirming that an elevated G0:G1 ratio is a feature of JIA overall and of all JIA subtypes tested (Table 1).

thumbnail image

Figure 5. IgG glycosylation in subsets of patients with JIA. A, G0:G1 ratios among patients categorized according to the International League of Associations for Rheumatology classification criteria (enthesitis-related arthritis was not included). ∗ = P < 0.05; ∗∗∗ = P < 0.001 versus control, by Mann-Whitney U test. B, G0:G1 ratio as a function of age at onset of JIA among patients with oligoarticular (oligo) JIA, polyarticular (poly) JIA, and psoriatic JIA compared with control subjects, according to age at onset (<6 years and ≥6 years). Both subgroups of JIA were distinct from the respective control groups, and the 2 control groups differed from each other (all P < 0.001); the G0:G1 ratios in the 2 patient groups were similar (P not significant [NS]). C, G0:G1 ratio as a function of antinuclear antibody (ANA) status among patients with oligoarticular JIA, polyarticular rheumatoid factor–negative (RF−) JIA, and psoriatic JIA with available data on ANA status (n = 98). The G0:G1 ratio did not differ between the group with and the group without ANA (P NS). Each data point represents a single subject. Bars show the mean. pers = persistent; ext = extended (see Figure 4 for other definitions).

Download figure to PowerPoint

Table 1. Association of JIA and JIA subtypes with an elevated G0:G1 ratio compared with healthy controls after adjustment for age, by linear regression analysis*
JIA subsetβ95% CIP
  • *

    Data reflect unstandardized β coefficients. Extended oligoarticular juvenile idiopathic arthritis (JIA) was omitted because of an insufficient sample size. The ages used were those at the time of sample acquisition. G0:G1 = agalactosylated IgG:monogalactosylated IgG; RF = rheumatoid factor; 95% CI = 95% confidence interval.

All JIA0.3170.245–0.389<0.001
JIA subtype   
 Oligoarticular, persistent0.3370.225–0.389<0.001
 Polyarticular   
  RF negative0.3380.241–0.435<0.001
  RF positive0.3250.203–0.447<0.001
 Psoriatic0.2420.116–0.368<0.001
 Systemic0.4670.331–0.599<0.001

However, classification of JIA remains a work in progress. Although rheumatoid factor (RF)–positive polyarticular JIA and systemic JIA are readily distinguished, substantial uncertainty surrounds the categories of oligoarthritis (persistent and extended), RF-negative polyarthritis, and psoriatic JIA (37). We therefore grouped these patients together and assessed whether age at onset (age <6 years versus age ≥6 years) or ANA status was informative with regard to IgG glycosylation in JIA. We observed similar elevations of the G0:G1 ratio in both age groups and in both ANA-positive and ANA-negative patients (Figures 5B and C). Of note, when the analysis was restricted to patients with JIA who presented at age <3 years (n = 27), the time of peak JIA incidence, no clear difference compared with control subjects ages 1 year to less than 3 years (n = 38) was observed (mean ± SD G0:G1 ratio 1.27 ± 0.40 versus 1.18 ± 0.24; P not significant).

Association of the G0:G1 ratio with markers of inflammation.

Whereas patients with systemic JIA showed the most deviation in the G0:G1 ratio, we considered the possibility that the G0:G1 ratio might be a proxy for systemic inflammation. A weak correlation could be identified between the G0:G1 ratio and the CRP level (n = 34; ρ = 0.17, P = 0.03). However, the significance of the correlation between the G0:G1 ratio and the ESR was borderline (n = 120; ρ = 0.17, P = 0.06), and no correlation was observed with S100A12 (n = 69; ρ = 0.10, P = 0.42) and MRP-8/MRP-14 (n = 69; ρ = 0.11, P = 0.37).

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Juvenile idiopathic arthritis exhibits a remarkable predilection for early childhood; this tendency is most pronounced in oligoarticular JIA but also is evident in other subtypes. Our results identify one potential contributor to this epidemiology: an elevation in the expression of proinflammatory hypogalactosylated IgG glycoforms in healthy young children, simultaneous with the peak incidence of JIA. Furthermore, we demonstrate conclusively that hypogalactosylation of IgG, corrected for age, is evident in JIA generally and across specific JIA subtypes.

The basis for the elevation in hypogalactosylated IgG levels in early childhood is unknown. Glycosylation of IgG occurs in the endoplasmic reticulum and Golgi apparatus, and variation in glycan structure is believed to reflect variation in the expression of enzymes that add or trim sugar residues. However, the regulation of this process is poorly understood. It is possible that different populations of B cells elaborate antibodies with different glycosylation patterns (38). Alternatively, glycosylation may reflect the cytokine environment in which B cells mature (39). Because the subset distribution, surface phenotype, and effector activity of T cells and B cells all vary with age, it will be an interesting challenge to identify the specific factors involved in the pediatric IgG glycan phenotype (12).

The “skew” toward hypogalactosylated IgG in young children is likely to have physiologic relevance. Hypogalactosylation of IgG may be a mechanism by which young children compensate for immature humoral immunity. Conversely, enhanced effector potency might predispose young children to IgG-mediated inflammatory diseases, potentially including JIA. If confirmed, this novel hypothesis could have important implications for the understanding of pediatric immune function and dysfunction.

We observed that elevated expression of agalactosylated IgG was relatively similar in all JIA subtypes tested, and that the G0:G1 ratio for JIA overall was essentially identical to that previously determined for RA (28). This result is consistent with several possible explanations. First, aberrant IgG glycosylation could be an epiphenomenon of synovitis. Our data do not exclude this possibility, but there are several reasons to doubt that it is the case. Direct experimental evidence supports the potentially enhanced arthritogenicity of agalactosylated antibodies (20, 31, 40, 41). In adult RA, glycosylation becomes aberrant more than 3 years prior to diagnosis, suggesting that glycosylation precedes overt joint disease rather than resulting from it (28). Finally, we observed that correlations between the G0:G1 ratio and markers of systemic inflammation were either weak or absent.

A second basis for the parallel elevation of the G0:G1 ratio across JIA subtypes may pertain to classification. The ILAR classification criteria imperfectly delineate homogeneous patient groups, as assessed by HLA associations, peripheral blood messenger RNA patterns, and clinical and laboratory phenotypes (1, 2, 10, 42–44). The similarity of G0:G1 ratios among JIA subtypes could therefore reflect scrambled classification of patients with respect to optimal biologic subgroups.

A final explanation for our results could be termed the “threshold hypothesis.” Hypogalactosylation may push a pool of potentially pathogenic IgG across a threshold required to engender synovitis, and this threshold may be similar on average across individuals. This hypothesis makes the interesting prediction that the incidence of arthritis should vary with the average G0:G1 ratio in the healthy population, and indeed inflammatory arthritis peaks both in early childhood and in the seventh decade of life (21, 22, 45). Furthermore, in the youngest children with JIA (age <3 years at onset), we observed no significant difference in the G0:G1 ratio compared with age-matched controls. An intriguing corollary of the threshold hypothesis is that a decline in the G0:G1 ratio with age might in some cases favor amelioration of disease in young children. Indeed, the subset of JIA that is most prevalent in this age group—persistent oligoarticular JIA—is the only type of idiopathic inflammatory arthritis with a spontaneous remission rate in excess of 50% (46). These speculations raise interesting opportunities for hypothesis-directed investigation.

Our study has several limitations. The G0:G1 method measures total serum N-glycans, and although serum G0 and G1 are primarily derived from IgG, it is possible that other serum proteins could have influenced our results. Our technique does not address other variable IgG residues that could also affect IgG function, including core fucose, an additional N-acetylglucosamine at the junctional mannose, or sialic acid (Figure 1A). A further technical limitation is that V–D–J recombination and somatic mutation result in the appearance of a Fab region N-glycan attachment site in ∼20% of IgG molecules, contributing to “background” signal and reducing the specificity of our findings for Fc glycans (15). However, these exposed glycans are generally sialylated and therefore highly charged, migrating outside the HPLC-determined peaks used for our G0:G1 calculations (28, 35).

Most importantly, we have not directly examined the ability of JIA IgG to fix complement, engage Fc receptors, form immune complexes, or otherwise promote inflammation. Additional study is needed to determine the net effect of aberrant galactosylation and other IgG glycan changes on the effector potency of the IgG pool and the role of these changes as contributory to the pathogenesis of JIA.

Taken together, our analyses represent the first detailed examination of glycan variation as a function of age in children, in particular in young children during the period of highest vulnerability for JIA. The results suggest that the antibody pool in young children may be more potent at triggering effector responses than that in older children and young adults. This observation has important ramifications for pediatric immunity, potentially including susceptibility for JIA, a disease family that we have confirmed to be characterized by a marked elevation in the expression of hypogalactosylated IgG. Additional work will be required to define the physiologic and pathophysiologic importance of these observations.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Nigrovic had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Ercan, Rudd, Nigrovic.

Acquisition of data. Ercan, Hazen, Tory, Henderson, Dedeoglu, Fuhlbrigge, Grom, Holm, Kellogg, Kim, Adamczyk, Son, Sundel, Foell, Glass, Thompson, Nigrovic.

Analysis and interpretation of data. Ercan, Barnes, Hazen, Adamczyk, Rudd, Foell, Thompson, Nigrovic.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES