The present study has shown that the NK cells were phenotypically altered in BD patients, especially those in inactive disease status. Features of the circulating NK cells in iBD patients included: downregulated gene expression of IL-12Rβ2
; upregulated gene expression of IL-13 in a subset of the patients; downregulated perforin and granzyme B gene levels; and impaired IL-12-induced Stat4 phosphorylation. Upregulated IL-13 and downregulated IL-12Rβ2 observed in NK cells from iBD patients were compatible with the NK2 phenotype. A serial NK phenotype analysis in aBD patients supported the association between NK2 bias and inactive disease status. Furthermore, NK2 cells obtained from iBD patients directly suppressed the IFNγ expression of Th1 cells derived from aBD patients in vitro
. These findings together suggest that the NK1/NK2 balance modulates disease flare/remission in BD patients by controlling the pathogenic Th1 response. This situation is analogous to multiple sclerosis, another Th1-mediated disease, in which NK2 bias is associated with disease remission [13
A major limitation of this study is the small number of patients analyzed, especially those with aBD. During 2 years of the study period, only 10 patients with aBD were enrolled in two major university hospitals in the Tokyo metropolitan area. In addition, there was a limited chance of obtaining peripheral blood samples from patients with aBD, because such patients required immediate introduction of treatment. Further multicenter studies involving a large number of patients with aBD are necessary to confirm our findings. Another limitation is the difficulty in classifying BD patients into those with active disease and those with inactive disease. We used a strict definition to select patients with aBD: flare of characteristic BD symptoms that required introduction of the intensive treatment, such as high-dose corticosteroids, cyclosporine, and infliximab. Patients with mild mucocutaneous manifestations or minor uveitis attack, which did not require intensive therapy, were therefore classified as having iBD. This clinical heterogeneity in the iBD subset may result in variability in the gene and protein expression profiles. Additional analysis according to individual clinical manifestations and/or treatment regimens would clarify these issues, but again the number of patients enrolled was too small to conduct subanalysis. Finally, we should recognize that a series of experiments involved only a subset of the patients and controls, which potentially bias the results.
Our results suggest that the NK2 cells in iBD patients can suppress the Th1 response through at least two distinct mechanisms. First, the NK2 cells in iBD patients were intrinsically hyporesponsive to IL-12 due to their downregulated expression of IL-12Rβ2
and impaired IL-12 signaling, resulting in deficient IFNγ production even in the Th1 environment. Second, the NK2 cells from iBD patients actively suppressed IFNγ expression in aBD-derived Th1 cells. A similar inhibitory effect of human NK2 cells on the production of IFNγ by T cells was also reported for healthy individuals' NK cells that were induced to express the NK2 phenotype [13
], and for NK2 cells obtained from multiple sclerosis patients in remission [21
]. Taken together, the NK cells and T cells - two major IFNγ producers - were deficient in IFNγ production in the NK2-biased immune environment observed in iBD patients.
How the NK2 cells from iBD patients suppress the IFNγ expression in Th1 cells, however, remains unclear. One potential soluble mediator in our cell-contact-free culture system is IL-13, a typical T-helper 2 cytokine that inhibits Th1 responses in vitro
and in vivo
], although upregulated IL-13 expression was detected only in one-third of the iBD patients. In addition, this IL-13-mediated inhibitory effect is reported to occur predominantly through the modulation of antigen-presenting cells rather than as a direct effect on T cells [22
]. Additional soluble factors secreted from NK2 cells are likely to be involved in this regulation, but the NK cells from iBD patients did not express IL-5, which plays a primary role in Th1 inhibition in multiple sclerosis patients in remission [13
]. Furthermore, it has been reported that NK cells modulate Th1 responses also by interacting directly with T cells, B cells, and dendritic cells though cognate cell-cell contact [24
Perforin and granzyme B, major cytoplasmic granule toxins, were downregulated in the NK cells from patients with iBD. Interestingly, this gene expression profile is analogous to that of the NK cells in patients with active pemphigus vulgaris, who also show NK2 bias [15
]. This phenomenon could be explained by the reduced IL-12Rβ2
expression and impaired IL-12 signaling, but the cytotoxic activity was the same among the NK cells of iBD patients, aBD patients, and healthy controls. The reason for this inconsistency is unknown, but the cytotoxic activity of NK cells might be regulated by more complicated mechanisms, involving a balance between activating and inhibitory NK receptors, as well as the expression of the ligands for death receptors on target cells [26
In aBD patients, the proportion of activated NK cells in the circulation was markedly increased. This is reasonable because IL-12 can activate NK cells in the Th1 environment, even though the nonspecific cytotoxic activity and gene expression profiles were similar between the NK cells from aBD patients and healthy controls. These activated NK cells would migrate to sites of inflammation and contribute to the ongoing tissue damage in aBD patients, but this appears to be just a bystander effect of the Th1 environment of aBD.