To determine whether there is a link between the frequency of natural killer T (NKT) cells and high levels of IgG in patients with systemic lupus erythematosus (SLE) and their relatives.
To determine whether there is a link between the frequency of natural killer T (NKT) cells and high levels of IgG in patients with systemic lupus erythematosus (SLE) and their relatives.
Blood samples were obtained from patients with SLE, their first-degree relatives, patients with rheumatoid arthritis (RA), and healthy control subjects. The frequency of NKT cells (defined as CD56+ T cells) was expressed as a percentage of total blood lymphocytes. Plasma levels of total IgG and IgM, and IgG antibodies to double-stranded DNA (dsDNA) were determined.
The frequency of NKT cells was lower in patients with SLE than in control subjects. No such decrease was observed in the relatives of patients with SLE or in patients with RA. High levels of IgG were observed in both patients with SLE and their relatives, while low levels of IgM were observed in these same groups. In relatives of patients with SLE, an inverse correlation between the frequency of NKT cells and IgG levels was observed. Moreover, raised levels of IgG in patients with SLE and their relatives and high levels of IgG anti-dsDNA in patients were associated with low frequencies of NKT cells.
These results suggest that NKT cells have an important role in the regulation of IgG production, although NKT cells with invariant T cell receptors may not necessarily be involved. NKT cells in the setting of SLE could lack the cytokine stimulus from NK or other cells that is needed to exert control on IgG production. Enhancement of NKT cell activity may provide a novel basis for therapy in SLE.
We previously reported a tendency toward low blood levels of natural killer (NK) cells in relatives of patients with the autoimmune disease systemic lupus erythematosus (SLE) (1). In that study, NK cells were defined as lymphocytes expressing surface CD56 but not CD3. However, a small group of cells in the CD3+ population express CD56 (2); these will be termed NKT cells. NKT cells form a heterogeneous population that, as will be discussed below, probably includes most of those cells expressing an invariant T cell receptor restricted by CD1d (invariant NKT cells) (3). Invariant NKT cells have been implicated in protection against a wide range of autoimmune diseases (4).
In the current study, we determined the frequency of CD56+ T cells in patients with SLE, first-degree relatives of patients with SLE, patients with rheumatoid arthritis (RA), and healthy control subjects. SLE is characterized by the production of high levels of IgG, including IgG autoantibodies that may give rise to widespread inflammation and tissue damage (5, 6). Therefore, we examined the relationship between the frequency of NKT cells and the level of total plasma IgG or IgM in the above-mentioned groups. In the group of patients with SLE, we also examined the relationship between the frequency of NKT cells and the level of IgG antibodies to double-stranded DNA (dsDNA). We present evidence that high levels of IgG occur only when the frequency of NKT cells is low, suggesting a role for these cells in the regulation of IgG production.
Blood samples were obtained from 65 patients with SLE (60 women and 5 men), 45 first-degree relatives of patients with SLE (31 female relatives and 14 male relatives), 21 patients with RA (17 women and 4 men), and 29 healthy control subjects (18 women and 11 men). Patients were attending clinics at St Mary's Hospital and the Centre for Rheumatology, University College London. Patients with SLE fulfilled the revised criteria of the American College of Rheumatology for the classification of SLE (7). The study was approved by the St Mary's Local Research and the University College University Trust ethics committees.
Details regarding the ages and ethnic backgrounds of the participants, as well as drug therapy for and the clinical disease activity of the patients, have been described previously (1). Disease activity in SLE was assessed using the system described by the British Isles Lupus Assessment Group (8), in which disease activity in each of 8 organs/systems (e.g., mucocutaneous, neurologic, renal) is assigned a category from A (most active) to D (currently inactive) or E (no current or previous involvement). A global disease activity score for each patient was obtained by allocating 9 points for A, 3 points for B (moderately active disease), and 1 point for C (stable mild disease).
Flow cytometric analysis of blood mononuclear cells was performed as described previously (1). NKT cells are defined as CD56+,CD3+ cells, and their frequency is expressed as a percentage of total lymphocytes (T cells including NKT cells, B cells, and NK cells). Occasionally, the percentage of NKT cells is converted to absolute numbers of circulating NKT cells by referring to the blood lymphocyte count (1).
The enzyme-linked immunosorbent assay (ELISA) procedures that were followed have been described previously (9). Flat-bottomed 96-well microtiter plates (M29AR; Sterilin, Feltham, UK) were coated with antiglobulin reagents or calf thymus dsDNA. In addition to the test plasma, dilutions of the appropriate standard were prepared on each plate. For total IgG and IgM, the standard was a normal serum (SPS-01; Protein Reference Unit, Sheffield, UK); for antibodies to dsDNA, plasma from a patient with SLE was used. Plates were developed with antiglobulin reagents conjugated to alkaline phosphatase in the presence of p-nitrophenyl phosphate. All optical densities were corrected for the background optical density obtained in uncoated wells. Results for total IgG or IgM are given as milligrams per milliliter of plasma, whereas results for anti-dsDNA are related to the standard values for SLE plasma (ELISA units [EUs]).
Values for the percentage of NKT cells and for total IgG and IgM in patients with SLE or their relatives were compared with control values, using the Mann-Whitney test. The same test was used when other group comparisons were made. Correlation analysis was carried out using Spearman's rank test. Fisher's exact test was used to compare the distribution of high and low values for the percentage of NKT cells in groups of patients with SLE with high or normal/low values for plasma IgG or IgG anti-dsDNA antibodies.
Figure 1 shows the values for NKT cells, expressed as a percentage of total blood lymphocytes, in male and female first-degree relatives of patients with SLE, male and female patients with SLE, male and female patients with RA, and male and female control subjects. The results are presented on a logarithmic scale to highlight low levels, and median values for the different groups are also shown. Values for all groups of patients and relatives showed considerable overlap with values for both female and male control subjects. Nevertheless, the percentage of NKT cells in female patients with SLE was significantly reduced compared with that in female controls (P = 0.05) or female relatives of patients with SLE (P < 0.02). Similarly, values for male patients with SLE were lower than those for male control subjects (P < 0.02). No differences were observed between values for relatives of patients with SLE and those for control subjects, although the median value for male relatives of patients with SLE was considerably lower than that for male control subjects. The frequency of NKT cells was not significantly increased in female patients with RA, but the relatively high values observed in that group were in marked distinction to those in female patients with SLE (P < 0.01).
Because of the reduced number of lymphocytes in patients with SLE (1), the observed reduction in the number of NKT cells compared with that in control subjects was accentuated when values for NKT cells were expressed as the number of circulating cells per milliliter (P < 0.005). However, no correlation was observed between the lymphocyte level and the frequency of NKT cells in patients with SLE (r = +0.13, P > 0.05). Only results based on the percentage of cells will be presented below, except in one instance in which absolute numbers provide additional insight.
Blood plasma levels of IgG in the various groups are shown in Figure 2. As expected for patients with SLE (5), the level of IgG was significantly increased compared with the levels in the corresponding female (P < 0.02) or male (P < 0.05) control subjects. Furthermore, some relatives of patients also had increased IgG levels. Seven female relatives of patients with SLE had values above the highest value for female control subjects, and 6 male relatives had values higher than those for their control subjects. Although these values did not represent significant increases for either the male or the female group separately, when data for both sexes were combined, the difference was significant (P < 0.05). This procedure seems valid in view of the absence of a sex difference in IgG levels in healthy control subjects (10). Two of the patients with RA had high levels of IgG.
In contrast to IgG, IgM levels were not significantly elevated in patients with SLE (Figure 3). Four female patients had values above the highest value for female control subjects, but 10 female patients had values below the lowest value for female control subjects. An unexpected finding was a reduced level of IgM in female relatives of patients with SLE (P < 0.01), and this did not correlate inversely with the higher IgG levels in that group (r = +0.07, P > 0.05). A positive IgG-versus-IgM correlation was demonstrated only in female control subjects (r = +0.82, P = 0.0001). The IgM data conform to the general rule that IgM levels are higher in women than in men (9).
Although no significant correlation was observed between patients and their relatives with respect to levels of NKT cells, IgG, or IgM, correlation coefficients were positive in each case, suggesting a degree of familial influence. For the percentage of NKT cells, r = 0.26 for male relatives, r = 0.10 for female relatives, and r = 0.12 for male and female relatives combined. The corresponding values for IgG were 0.16, 0.25, and 0.29, and for IgM were 0.47, 0.19, and 0.25. For this analysis, mean values were used when the same patient had more than 1 relative included in the study.
The relatives of patients with SLE comprised 12 mothers, 8 fathers, 9 sisters, 2 brothers, 10 daughters, and 4 sons (1). Five mothers, 4 fathers, 2 sons, 1 daughter, and 1 sister had IgG levels above control values. Two mothers, 2 fathers, 1 daughter, 1 sister, and 1 son had IgM levels below control values. In 1 father, 1 son, and 1 sister, the frequency of NKT cells was lower than that in control subjects.
In relatives of patients with SLE, there was an inverse correlation between IgG levels and the percentage of NKT cells (Figure 4). For male and female relatives combined, r = –0.48 (P = 0.001); for male relatives alone, r = –0.70 (P < 0.01); for female relatives alone, r = –0.39 (P < 0.05). Of the 8 relatives (7 of whom were female) with IgG levels above the highest level for control subjects (16.8 mg/ml), the percentage of NKT cells was <2% in 6 and was 6% in 1 (1 relative was not tested). No similar correlation was established for patients with SLE (r = +0.06, P > 0.05) (Figure 5A). However, 20 patients with SLE (including 1 male patient) had elevated levels of IgG (>16.8 mg/ml), and all of these patients had low values for the percentage of NKT cells (<2% in 17 patients, and 2–2.6% in 3 patients). Among the remaining 40 patients with normal IgG levels (<16.8 mg/ml), 8 had a high frequency of NKT cells (>2.6%), which represents a significantly increased proportion of high NKT cell values as compared with those for the group with high levels of IgG (P < 0.05). There was no correlation between the IgG level and the frequency of NKT cells in control subjects (r = +0.19) or patients with RA (r = +0.23). The graphed results for control subjects and patients with RA appeared similar to those for patients with SLE in that peak values for the percentage of NKT cells fell in the middle of their respective IgG ranges, but they lacked the tail at the top end of IgG values. Among patients with RA, 2 patients (1 man and 1 woman) had high values for IgG, with the values for the percentage of NKT cells being 2% and 5.1%, respectively.
No similar relationship was observed between IgM levels and the frequency of NKT cells. For example, for the group of male relatives of patients with SLE, r = +0.002 (P > 0.05). However, the association between high levels of IgG and low frequencies of NKT cells extended to IgG antibodies to dsDNA in patients with SLE (Figure 5B). Although overall there was no significant correlation (r = –0.10), the 7 highest antibody values (>1.0 EU) occurred with an NKT cell frequency of <1.5%. Conversely, each of the 6 patients with an NKT cell frequency of >3.5% had an antibody level of <0.5 EU. Using 2.6% as the cutoff for high/low NKT cell frequencies as described above, and a cutoff of 1.0 EU for anti-dsDNA antibodies, no significant difference was observed in the distribution of NKT cell frequencies in the group with high levels of anti-dsDNA antibodies as compared with the group with low levels of anti-dsDNA antibodies. On the other hand, for patients with anti-dsDNA antibody levels above the median value of 0.26 EU (the group of patients who may be considered positive for the autoantibody), the correlation with NKT cell frequency was significant (r = –0.38, P < 0.05). All control subjects had values <0.15 EU, and only 3 relatives had values >0.15 EU (maximum 0.36 EU).
We also examined the relationships between the levels of total IgG, total IgM, or IgG anti-dsDNA antibodies and NK cell frequency or lytic activity (1). No results paralleled the relationships seen with NKT cells, apart from a significant inverse correlation between the level of IgM and the percentage of NK cells in female relatives of patients with SLE (r = –0.48, P < 0.01). There were no correlations between the frequency of NKT cells and that of NK cells or their killing activity.
Among patients with SLE, there was no significant inverse correlation overall between the frequency of NKT cells and the global disease activity score (r = –0.10, P > 0.05) (Figure 6). However, it is striking that among the 12 patients with a disease activity score ≥8, only 1 had an NKT cell frequency >2%, and among the 7 patients with an NKT cell frequency >3.5%, only 1 had a disease activity score >5. No such tendencies toward correlation were observed in relation to IgG or IgM levels and disease activity. The group of 5 patients with high levels of anti-dsDNA antibodies (Figure 5B) had global disease activity scores of only 2–4.
The disease activity score is based on activity in different body organs/systems, and the results were analyzed to determine whether the involvement of particular systems was linked to levels of NKT cells, IgG, IgM, or IgG anti-dsDNA antibodies. The only significant result was an increased level of IgG in the group with cardiovascular/respiratory disease activity (n = 11) as compared with the rest of the patients (P < 0.05). The group of patients with musculoskeletal involvement (n = 36), in contrast, had reduced levels of IgG, with the difference between these patients and the remainder of the patients narrowly missing significance (P = 0.06).
Twelve female patients were being treated with azathioprine, alone or in combination with steroids (prednisolone) or steroids and hydroxychloroquine (HCQ) (Figure 6). In our previous study, patients who were treated with azathioprine showed markedly reduced levels of NK and B cells (1), but there was less effect on T cell numbers. The median level of T cells in female patients who were not receiving azathioprine was 1.06 × 106/ml, compared with 0.87 × 106/ml in the group receiving azathioprine (P > 0.05). In the current study, the corresponding values for NKT cells were 0.0161 × 106/ml and 0.0072 × 106/ml (P > 0.05). This translates to 1.3% and 0.7% in terms of the frequency of NKT cells. When patients receiving azathioprine were removed from the statistical analysis, the difference between female control subjects and the remaining female patients in terms of circulating NKT cells remained highly significant (P = 0.005). The difference in NKT cell frequency between female patients with SLE and female relatives of patients with SLE also remained significant (P < 0.05).
No effects of azathioprine on the levels of total IgG and IgM and IgG anti-dsDNA antibodies were observed. Similarly, we have assessed the overlapping patient groups receiving steroids or HCQ (1). The median level of anti-dsDNA antibodies was more than twice as high in the HCQ group as in the remaining patients (P < 0.02), but no effects of these drugs on NKT cell frequency or the other parameters were apparent.
In nearly all normal subjects in the present study, values for CD56+ T cells (NKT cells), expressed as a percentage of total lymphocytes, were in the range of 0.5–6.0%. The values in patients with SLE were significantly reduced, with several patients having values <0.5%. Unlike the well-documented decrease in the frequency of NK cells in SLE (1), there are few previous data for CD56+ T cells, although a similar decline in the percentage of CD56+,CD16+ T cells has been reported (11). The reduced frequency of NKT cells in patients with SLE contrasted strongly with our data for female patients with RA, for whom the median value was high compared with that for control subjects.
In contrast to our data regarding the frequency of NK cells (1), the current study revealed little evidence for abnormally low numbers of NKT cells in relatives of patients with SLE. In contrast, elevated levels of IgG, which is characteristic of SLE, were observed in both patients and their healthy relatives, particularly female relatives. These findings in relatives of patients with SLE, which are consistent with the notion of a genetic basis for this characteristic, extend our previous work (9) and are consistent with an early study by Larsen (10). Raised levels of IgG in the setting of SLE are clearly attributable to increased production (5). Indeed, catabolism of IgG is reported to be enhanced in SLE (12), leading to calculations that increases in the rate of production as great as 3–4-fold might be occurring (13). Relatives of patients with SLE also produce high levels of IgG in response to pokeweed mitogen (9).
The abnormally low levels of IgM that were observed in ∼20% of patients with SLE have also been noted by other investigators (14). Our further finding of significantly low IgM levels in female relatives of patients with SLE suggests that this might be a genetic determinant of SLE in at least some patients. As expected, most of the patients with RA had normal levels of IgG and IgM.
No correlation was observed between the frequency of NKT cells and natural killing activity or NK cell frequency, even in patients with SLE. However, an inverse correlation between the frequency of NKT cells and the level of IgG was observed in relatives of patients with SLE. Furthermore, with a single exception, whenever the level of IgG exceeded the normal range in either patients with SLE or their relatives, the percentage of NKT cells was at a low or moderately low level, usually <2%. The same was true for patients exhibiting high levels of IgG antibodies to dsDNA, and there was an inverse correlation between NKT cell frequency and the level of anti-dsDNA antibodies in patients who were positive for the autoantibody. A relatively low level of NKT cells thus appeared to be necessary, if not sufficient, for the development of high levels of IgG, and the possible significance of this in the causation of SLE is discussed below. Riccieri et al (15) reported an inverse correlation between IgG levels and the percentage of CD56+,CD16+ T cells in patients with systemic sclerosis.
In this and previous studies (1, 9), we observed that immunologic abnormalities associated with SLE are more likely to be seen in the parents and offspring of patients than in the siblings of patients. This was not the case in an earlier study of IgG, in which the age of relatives appeared to be an important factor (10).
It is of interest that female relatives of patients with SLE, who are the most likely of the healthy subjects studied to later develop SLE, displayed several unique characteristics. Aside from the results regarding the association of IgG levels and the frequency of NKT cells, female relatives of patients with SLE had low levels of IgM and an increased percentage of B lymphocytes relative to control subjects (1). Moreover, there was the above-mentioned inverse correlation between the level of IgM and the percentage of NK cells.
The lack of clear evidence of a low frequency of NKT cells in relatives of patients with SLE raises the question of whether the low levels seen in patients could be attributable to drugs used to treat SLE. As described above, this was not the case. Azathioprine, which was used to treat ∼20% of the patients, had previously been shown to have a dramatic lowering effect on the number of NK and B cells (1), but its effect on T cells, including NKT cells, was much smaller. The highest levels of disease activity were observed in association with low frequencies of NKT cells, and the relatively low median value for NKT cell frequency in patients receiving azathioprine may be attributable to this drug being used to treat more seriously affected patients. Of course, once a patient has been treated with a disease-modifying drug, clinical disease activity should be reduced, and immunologic abnormalities may be ameliorated. Such drug influences may explain our failure to show significant correlations between patients and their relatives with regard to the percentage of NKT cells and levels of IgG and IgM, despite probable genetic links.
The finding of an inverse relationship between IgG levels and NKT cell frequency suggests a role for NKT cells in regulating IgG levels. CD56+ T cells are a heterogeneous population consisting of CD4+, CD8+, and double-negative or double-positive cells (3). The majority of invariant NKT cells are reported to be distributed among this population (16, 3), although we are not aware of a definitive study of CD56 expression by these cells. Invariant NKT cells form 0.01–1% of total blood lymphocytes (17) or, at most, approximately one-half of the median level of CD56+ T cells found in our healthy controls. The numbers of invariant NKT cells are reported to be reduced in SLE and other autoimmune diseases (18, 19), but it seems unlikely that this could be responsible for more than part of the observed reduction in the number of total CD56+ T cells. We detected no decrease in the frequency of NKT cells in patients with RA, yet the number of invariant NKT cells was reduced to an extent similar to that in SLE (18, 19). In mouse models of SLE, both protective and pathogenic roles for invariant NKT cells have been reported (20). What is clear, however, is that these cells are strong secretors of interferon-γ, interleukin-4 (IL-4), and other cytokines (21, 22). Although presumably a glycolipid (4), the physiologic antigen for invariant NKT cells is unknown, and it is certainly possible that a more conventional CD56+ T cell carries this regulatory function.
The concept of defective suppressor/regulatory cell activity in SLE being the cause of enhanced IgG/ autoantibody production was first suggested many years ago (23), but convincing evidence was lacking. More recently, Horwitz and colleagues (24, 25) demonstrated reduced production of transforming growth factor β by NK cells in patients with SLE, which may in turn lead to a defect in the appearance of CD8+ T regulatory cells that are able to control IgG production. CD8+ regulatory cells may also express CD56 (26), which would classify them as NKT cells. It could thus be that a reduction in the frequency of regulatory NKT cells in patients with SLE diminishes the normal control on IgG production. High levels of expression of IL-10 in SLE (27) may be the basis for low activity of NK cells (1, 24) and also may have more direct positive effects on antibody production (28).
How, then, can one explain why control subjects and relatives of patients with SLE had a similar distribution of NKT values, while only the latter group had high levels of IgG and the correlation with NKT cell frequency? It could be that, as with NK cells (1), the activity of individual NKT cells as well as cell numbers are reduced in SLE. In control subjects, activity is normal and cell numbers may not be limiting, whereas activity in relatives may be impaired with regulation dependent on cell number.
Another possible interpretation of these findings is that the reduction in the percentage of NKT cells in patients with SLE is secondary to the disease process. It has been suggested, for example, that the lymphopenia associated with SLE is a consequence of the presence of autoantibodies against lymphocyte surface antigens (29), although we are not aware that antibodies specific for CD56+ T cells have been described. Evidence against this is that the clearest demonstration of an inverse correlation between NKT cells and plasma IgG levels was in healthy relatives of patients with SLE. Although lymphocytotoxic and other autoantibodies are found in some relatives (30), there was no evidence of lymphopenia (1).
In conclusion, high levels of plasma IgG in patients with SLE and their first-degree relatives and high levels of IgG anti-dsDNA antibodies in patients were observed to be associated with a low frequency of NKT cells. These results suggest that NKT cells, possibly through the intervention of NK cells, play an important role in the regulation of IgG production. Enhancement of NKT cell activity could provide a useful approach to therapy in SLE.
Dr. Salaman 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 design. Drs. Seifert, Isenberg, and Salaman.
Acquisition of data. Mr. Green, Ms Kennell, and Dr. Salaman.
Analysis and interpretation of data. Mr. Green, Ms Kennell, and Dr. Salaman.
Manuscript preparation. Drs. Isenberg and Salaman.
Statistical analysis. Dr. Salaman.
Provision of blood samples and clinical data from patients and relatives. Drs. Larche, Seifert, and Isenberg.
We thank Elena Kulinskaya for advice on statistical analysis and Charles Bangham, Stuart Berzins, and Keith Gould for helpful discussions.