Changes of serum IgG glycosylation patterns in rheumatoid arthritis

Patient characteristics

Among RA patients, 21% (46/214) were male, and the median age was 55 years old. RF ( +), CCP ( +), and seropositive were presented in 84% (179/214), 81% (174/214), and 77% (164/214), respectively. 60 of RA patients (28%) were complicated with ILD. RA patients were characterized as 75 (35%) of remission, 29 (14%) of low disease activity, 61 (29%) of moderate disease activity, 48 (22%) of high disease activity according to DAS28-ESR. And based on DAS28-CRP, RA patients were divided into 102 (48%) of remission, 21 (10%) of low disease activity, 57 (27%) of moderate disease activity, 30 (14%) of high disease activity, respectively. Among DCs, 28% (14/50) were male and the median age was 44.5 in APS, 4% (8/50) were male with median age of 39.5 in TA, 7% (14/50) were male and the median age was 61 in VD. In the group of HC, 81% (81/100) were male and the median age was 65. For the three disease controls, which are chronic inflammatory diseases without RA antibody reactivity, as well as HC, those items related to RA are not applicable.

Analysis of serum IgG glycosylation in patients with RA by lectin microarray

Serum samples of 164 RA patients, 150 DC patients, and 100 HC were detected by lectin microarray (Fig. 1; Additional file 1: Fig. S1). Results of 56 lectins among all groups were performed cluster analysis and shown on a whole scope (Additional file 1: Fig. S2). In detail, 6 among 56 lectins showed significant differential signal intensities between the RA and DC/HC groups (Table 1). Serum IgG from RA patients had a higher affinity for SBA compared to HC, as well as higher affinities for STL, PHA-E, SNA, Jacalin, SBA compared to DC. Therefore, glycan levels of GalNAc (recognized by SBA), GlcNAc (recognized by STL), Galβ4GlcNAc (recognized by PHA-E), Sialic acid (recognized by SNA), Galβ3GalNAc (recognized by Jacalin) were increased characteristically in serum IgG from patients with RA (Fig. 2). Sugar specificity for lectins with significant differences between groups are listed in Table 2. Thus, the above lectins was chosen for verification in the later process.

Table 1 S/N data of lectin microarrayFig. 2figure 2

Specific changes of serum IgG glycosylation in groups. A left group in black refers to RA and right group in gray refers to HC; B each left group in black refers to RA and right group in gray refers to DC; C each left group in gray refers to seropositive and right group in black refers to seronegative; D each left group in black refers to RA-ILD and right group in gray refers to RA-nILD; E each left group in black refers to Remission and right group in gray refers to High Disease Activity. p values were showed on the top of each compared groups

Table 2 Sugar specificity for lectins with significant differences between groups

Lectin microarray results were further explored across different RA subgroups (Table 1), and results were illustrated in Fig. 2: (1) Significantly higher glycan levels of sialic acid (recognized by SNA-I), mannose (recognized by MNA-M and ConA), fucose (recognized by AAL), were observed for seropositive patients compared to the seronegative group (p < 0.05). (2) Significantly higher glycan levels of mannose (recognized by MNA-M and ConA), fucose (recognized by LCA) while lower glycan levels of Galβ4GlcNAc (recognized by PHA-E and PHA-L) were observed for RA-ILD patients compared to the RA-nILD group (p < 0.05). (3) Significantly higher glycan level of GalNAc (recognized by DBA) was observed for patients that categorized as remission (DAS28 ≤ 2.6) compared to the high disease activity (DAS28 > 5.1) group by using both standards of DAS28-ESR and DAS28-CRP (p < 0.05).

Validation of glycosylation changes of IgG by lectin blot

IgG heavy chains were selected in lectin blot to verify the microarray results. The intensity of the following lectins on serum IgG from related groups were analyzed: (1) STL, PHA-E, SNA, Jacalin in groups of RA patients and DC patients, (2) SNA-I and ConA in subgroups of RA-seropositive and RA-seronegative, (3) LCA and PHA-L in subgroups of RA-ILD and RA-nILD, (4) DBA in subgroups of remission and high disease activity, and no significant results were observed.

For groups of RA versus HC and RA versus DC patients, 24 serum samples were randomly selected for SBA lectin blot validation, and results showed that the intensity of SBA on serum IgG from RA patients was significantly increased compared to either HC or DC patients (p < 0.05) (Fig. 3A and B). For RA subgroups, at least 18 serum samples from each group were chosen for validation, and a new cohort of 50 RA-ILD patients was involved in the selection. The results were listed as follows: (1) MNA-M and AAL lectins were applied to recognize glycans of serum IgG in RA-seropositive and RA-seronegative groups, and increased intensities were observed in RA-seropositive samples (p < 0.05) (Fig. 3C); (2) Lectins of ConA, MNA-M and PHA-E were applied to recognize glycans of serum IgG in RA-ILD and RA-nILD groups, and increased intensities of ConA, MNA-M as well as decreased intensities of PHA-E were observed in RA-ILD samples (p < 0.05) (Fig. 3D). These results were consistent with those from lectin microarrays, which confirmed the reliability of lectin microarray analysis. The consistent summary of verification results was shown in Table 2.

Fig. 3figure 3

Lectin blot of lectins for serum IgG in RA/DC/HC groups and RA subgroups. Comparison of fluorescence intensity of lectin blot bands are showed in bar graph. R, reference; *p < 0.05, **p < 0.01, ***p < 0.001

Candidate biomarkers for the diagnosis of RA and RA-ILD

The prediction models of sensitivity and specificity were analyzed as described in the method, and the ROC curves were further constructed for the identified lectin biomarkers. Both data of lectin microarray and lectin blot were applied for the prediction model, and results showed that: (1) Based on the data of lectin microarray in the groups of RA/HC and RA/DC, the diagnosis of RA by lectin SBA showed a sensitivity of 66.46% and a specificity of 62% combined with an AUC of 0.65 (J = 1.362, p < 0.0001, Fig. 4A), and (2) a sensitivity of 65.24% and a specificity of 54.67% combined with an AUC of 0.61 (J = 1.376, p = 0.001, Fig. 4B), respectively. (3) By analyzing the data of lectin blot in the subgroups of seropositive and seronegative, the lectins of AAL (sensitivity = 62.1%, specificity = 73.33%, AUC = 0.70, J = 2.051, p = 0.01) and MNA-M (sensitivity = 50%, specificity = 93.33%, AUC = 0.70, J = 3.21, p = 0.01) could be used as alternative biomarkers for seropositive (Fig. 4C). (4) Data of lectin blot in the subgroups of RA-ILD and RA-nILD were applied for the prediction model, and the lectins of ConA (sensitivity = 65.38%, specificity = 95.83%, AUC = 0.87, J = 1.024, p < 0.0001), MNA-M (sensitivity = 79.17%, specificity = 75%, AUC = 0.75, J = 0.95, p = 0.003), PHA-E (sensitivity = 100%, specificity = 50%, AUC = 0.73, J = 1.096, p = 0.02) could be candidate biomarkers for the diagnosis of ILD in RA patients (Fig. 4D).

Fig. 4figure 4

The ROC curve of the biomarkers for the classification of A RA/HC B RA/DC C RA-seropositive and RA-seronegative D RA-ILD and RA-nILD

留言 (0)

沒有登入
gif