Expansion and enhanced survival of natural killer cells expressing the killer immunoglobulin-like receptor KIR3DL2 in spondylarthritis




The spondylarthritides (SpA) are strongly associated with possession of HLA–B27. We hypothesized that the expression of abnormal forms of HLA–B27 in SpA may have a pathogenic role through interaction with cells bearing natural killer (NK) receptors, in particular, killer immunoglobulin-like receptor (KIR) KIR3DL2, a receptor for HLA–B27 homodimer (B272). We therefore undertook the present study to determine the number and function of NK and T cells bearing KIR3DL2 in SpA.


Expression of KIR3DL2 on NK and T cells was quantified in peripheral blood (PB) from 35 patients with SpA and 5 patients with juvenile enthesitis-related arthritis (juvenile ERA); samples were compared with samples from healthy and rheumatoid arthritis (RA) controls. Paired synovial fluid (SF) was studied where available. Expression of other KIRs as well as activation, memory, and homing markers on KIR3DL2+ NK and T cells was quantified. NK cell survival was assessed using the apoptotic markers annexin V and 7-aminoactinomycin D, and cytotoxicity by 51Cr release assay.


In SpA, an increased number of PB and SF NK and CD4+ T cells expressed the KIR3DL2 receptor compared with controls. In ERA, KIR3DL2 expression was increased in PB and SF CD4 T cells (and SF NK cells) compared with RA controls. KIR3DL2+ NK cells had an activated phenotype, and were protected from apoptosis by culture with a cell line expressing B272. SpA PB mononuclear NK cells from SpA patients showed greater cytotoxicity than those from controls.


KIR3DL2 expression on NK cells and CD4 lymphocytes is increased in SpA and ERA. These cells are activated and may have a pathogenic role.

Natural killer (NK) cells form a part of the innate immune system and play an important role in clearing malignant and virus-infected cells (for review, see ref. 1). The response of NK cells to target cells is dependent on the balance of signals transmitted through inhibitory and activating NK receptors (NKRs). In humans, NKRs can be divided into 3 classes: the immunoglobulin (Ig)–like receptors including killer Ig-like receptors (KIRs) and leukocyte Ig-like receptors (LIRs or LILRs), the lectin-like receptors (CD94/NKG), and the natural cytotoxicity receptors (NCRs) including NKp30, NKp44, and NKp46 (for review, see ref. 1). Healthy autologous cells are protected from NK cell lysis by the interaction of class I major histocompatibility complex (MHC)–specific inhibitory NKRs (for review, see ref. 1) with their cognate MHC.

KIRs are expressed on many NK cells and on subpopulations of T cells (2) and have been implicated in autoimmune conditions such as rheumatoid arthritis (RA) (3) and scleroderma (4), where associations have been found with 2-domain KIRs known to recognize HLA–C alleles. The 3-domain KIRs (KIR3DL1 and KIR3DL2) have been shown to bind to the HLA–Bw4 group of alleles including HLA–B27 (5) and HLA–A3,11, respectively (6). The combination of inheritance of HLA and KIR is important for protection from both infection and predisposition to certain autoimmune diseases including psoriatic arthritis (7), and the 2 have almost certainly coevolved (8). Two KIR haplotypes (A and B), which differ in number and content of KIR genes (9), exist in humans. KIR3DL2 is expressed in both haplotypes (10).

Ankylosing spondylitis (AS) is the most common form of spondylarthritis (SpA); 96% of AS patients carry HLA–B27 (11). HLA–B27 can fold incorrectly, forming a β2-microglobulin–free homodimer disulfide, bonded through the unpaired cysteine residue at position 67 (12). These homodimers, also known as B272, are expressed on peripheral blood mononuclear cells (PBMCs) from patients with SpA and also in HLA–B27–transgenic disease models (13, 14). B272 are ligands for a number of NKRs including KIR3DL1, KIR3DL2, LILRB2, and LILRA1 (13, 15). This pattern of recognition is distinct from that of B27 heterodimers, which are bound by KIR3DL1, LILRB1, LILRB2, and LILRA1 but not by KIR3DL2. While KIR expression is limited to NK and T cells, LIRs have a wider pattern of expression. LILRB1 is expressed on B, T, and NK cells. LILRB2 is more selectively expressed on myelomonocytic and dendritic cells. In contrast, LILRA1 is expressed only on myelomonocytic cells, and not on NK or T cells (16, 17).

Although both KIR3DL1 and KIR3DL2 transmit inhibitory signals through their immunoreceptor tyrosine-based inhibitory motifs, the expression of KIRs has been shown to correlate with increased longevity and resistance to apoptosis of memory T cells (18). Based upon these data, we hypothesized that the expression of KIR3DL2 on T cells (and perhaps on NK cells) might be associated with increased survival and possibly play a pathogenic role in SpA.

NK cells have been divided into 2 populations based on their CD56 expression, namely, CD56bright and CD56dim (19). KIRs are predominantly but not exclusively expressed on the CD56dim population (20). In peripheral blood, the majority (∼90%) of NK cells are CD56dim, whereas in synovial fluid (SF), CD56bright NK cells predominate (21).

Little is known about the KIR expression or function in NK cells in SpA. In this study, we showed that expression of KIR3DL2 is significantly increased on NK and CD4+ T lymphocytes from patients with SpA and on CD4+ T cells from patients with enthesitis-related arthritis (ERA). These cells have an activated phenotype. KIR3DL2+ NK cells are protected from apoptosis by culture with B272-expressing cells. Preliminary evidence is provided showing that NK cells from patients with AS have increased cytolytic function compared with those from RA patients and healthy controls.


Patients and samples.

The study included 35 SpA patients who fulfilled the European Spondylarthropathy Study Group criteria for disease (22). Twenty-four fulfilled the modified New York criteria for AS (23). Of the 24 AS patients, 22 were HLA–B27 positive (HLA–B*2702/2705) by molecular methods. Eight SpA patients (3 HLA–B27 positive) had psoriatic arthritis, and 3 (all HLA–B27 positive) had reactive arthritis. The main demographic and clinical features of patients are shown in Table 1. PB was obtained from all 35 SpA patients and paired SF from 10 SpA patients, all of whom had given consent. This study had ethical approval from the Oxford Radcliffe Trust local ethics committee (COREC C00.114). In addition, SF and PBMCs were obtained from 5 HLA–B27–positive patients with juvenile ERA (24). The collection of ERA samples had ethical approval from the Great Ormond Street Hospital local ethics committee. Controls included 13 healthy HLA–B27–negative and 5 HLA–B27–positive volunteers and 10 patients with RA who met the criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (25). PB was obtained from these study subjects, all of whom had given informed consent. Paired SF was also obtained from 10 RA patients. All SF samples in this study were obtained from knee effusions.

Table 1. Demographic, laboratory, and clinical features of the subjects studied*
 Healthy controls (n = 18)RA (n = 10)SpA (n = 35)ERA (n = 5)
  • *

    RA = rheumatoid arthritis; SpA = spondylarthritis; ERA = enthesitis-related arthritis; PBMCs = peripheral blood mononuclear cells; SFMCs = synovial fluid mononuclear cells; DMARD = disease- modifying antirheumatic drug.

  • Eight patients (80%) were rheumatoid factor positive.

  • Twenty-four patients had ankylosing spondylitis (AS) (22 HLA–B27+), 8 had psoriatic arthritis (3 HLA–B27+), and 3 had reactive arthritis (all HLA–B27+). AS patients' mean ± SD scores on the Bath AS Metrology Index, the Bath AS Functional Index, and the Bath AS Disease Activity Index, respectively, were 3.9 ± 2.4, 3.4 ± 1.9, and 4.2 ± 2.1.

Age, mean (range) years49.1 (23–60)52.0 (32–68)47.2 (28–69)10.4 (8–14)
Sex, male/female10/85/530/55/0
 SFMCs (knee SF)010105
HLA–B27+, no. (%)5 (27.7)0 (0)28 (80)5 (100)
Corticosteroid treatment, no. (%)0 (0)5 (50)2 (5.7)0 (0)
DMARD use, no. (%)0 (0)9 (90)15 (43)3 (60)
Infliximab use, no. (%)0 (0)1 (10)1 (2.9)0 (0)

Staining of cells for flow cytometry.

Synovial fluid mononuclear cells (SFMCs) and PBMCs were isolated using Lymphoprep (Nycomed Pharma, Oslo, Norway) gradient centrifugation. PBMCs and SFMCs were washed in phosphate buffered saline in the presence of 1% heat-inactivated fetal calf serum and 0.1% sodium azide and incubated for 30 minutes on ice with saturating amounts of directly conjugated monoclonal antibodies (mAb) to the following antigens: CD3, CD4, CD8, CD28, CD45RA, CD45RO, CCR7, CD62L, CD38, β7 integrin, perforin, and isotype controls (BD Biosciences, San Jose, CA). Antibody to the NK cell marker CD56 (Immunotech, Marseilles, France) was used. The mAb to KIR3DL1 (IgG1 DX9) and KIR3DL2 (IgG2a DX31) (6) were both kind gifts from J. Phillips (DNAX Research Institute, Palo Alto, CA). IgG1 and IgG2a isotype control staining did not reveal significant staining of samples from SpA patients or either RA or healthy controls (results not shown). The following mAb specific for NK receptors were purchased from BD Biosciences: anti-CD158a (anti-KIR2DL1 and KIR2DS1) and anti-CD158b (anti-KIR2DL2, KIR2DS2, and KIR2DL3). Cells were washed twice after incubation with mAb. A secondary phycoerythrin-labeled conjugated F(ab′)2 mAb (Dako, Carpinteria, CA) was added when unlabeled primary mAb DX31 and DX9 were used in the first step. Cells were fixed in 1% paraformaldehyde and analyzed with a FACSCalibur, using CellQuest software (Becton Dickinson, Mountain View, CA). In each experiment the viable lymphocyte pool was identified using forward and side scatter characteristics. Four-color analysis was performed and 100,000 to 200,000 cells were analyzed. Both fresh and cryopreserved frozen samples were compared on repeated stains, and consistent results were obtained (<5% difference).

Isolation of purified KIR3DL2+ NK cells and use in apoptosis assays.

NK cells were isolated by negative selection from 4 HLA–B27–positive patients with AS using the MACS NK cell isolation kit (Miltenyi Biotech, Bergisch Gladbach, Germany) according to the manufacturer's protocol. The negative fraction was >98% CD3−,CD56+ and was analyzed with a FACSort for KIR3DL1−,KIR3DL2+ cells. Purity was >99%.

Fifty thousand DX31+ NK cells were cultured in 0.1 ml complete medium (RPMI 1640/10% human AB serum with L-glutamine and penicillin/streptomycin) in flat-bottomed 96-well plates. Fifty thousand gamma-irradiated LBL721.220 or transfectants were added to the NK cells and were cultured for 7 days. Cells were harvested and incubated with mAb to annexin V and 7-aminoactinomycin D (7-AAD) according to the manufacturer's instructions (BD Biosciences) and analyzed by fluorescence-activated cell sorting (FACS). Annexin V–negative, 7-AAD–negative cells were considered to be fully viable, annexin V–positive, 7-AAD–negative cells undergoing early apoptosis with membrane integrity present, and annexin V–positive, 7-AAD–positive cells were considered to be undergoing end-stage apoptosis and committed to death.

51Chromium release assay.

NK cytotoxicity was assessed by incubating PBMCs (effector cells) with 51Cr-labeled K562 (target cells) at various effector:target (E:T) cell ratios as described previously (26). K562 is an NK cell–sensitive erythroleukemia cell line. The levels of radioactivity released from target cells into supernatants were assessed by gamma scintillation after 4 hours of incubation. All experiments were performed in triplicate in a 96-well microtiter plate. Spontaneous release was assessed in the wells containing 51Cr-labeled target cells in medium without effector cells. Maximum release was determined in the wells containing labeled target cells in the presence of 5% Triton X-100 to promote total lysis. The percentage of lysis was calculated as follows: percentage lysis = 100 × ([mean radioactivity of sample − mean radioactivity of spontaneous lysis]/[mean radioactivity of maximum lysis − mean radioactivity of spontaneous lysis]).

Statistical analysis.

For the data on KIR distribution, median values with interquartile ranges (IQRs) are shown. Sample means were analyzed using Student's 2-tailed t-test. When comparing more than 2 groups, one-way analysis of variance with Bonferroni correction was performed.


Increased percentage of KIR3DL2-expressing NK and CD4 T cells in SpA.

We determined the number of CD3−,CD56+ NK cells and CD3+ T cells expressing KIRs from 35 patients with SpA. Demographic details are shown in Table 1. PBMCs and SFMCs were stained with each KIR mAb separately, together with the relevant T and NK cell markers. Figure 1 shows the number of KIR3DL2+ NK, CD4+, and CD8+ T cells in each subject group. Table 2 shows the median and IQRs of KIR expression. We observed a striking increase in the number of KIR3DL2+ NK and CD4+ T cells in SpA compared with both healthy controls and RA controls (Figures 1A and B). Thus, in SpA PBMCs, a median 43.5% of NK cells (IQR 35.2–50.3%, mean ± SD 42.8 ± 13.5%) expressed KIR3DL2. This was significantly higher than for NK cells of healthy volunteers (median 24.3%, IQR 18.7–27.7%, mean ± SD 22.8 ± 10.2%) (P < 0.001) and RA controls (median 23.3%, IQR 17.9–25.5%, mean ± SD 20.2 ± 9.2%) (P < 0.001). HLA–B27–positive SpA patients expressed higher numbers of KIR3DL2+ NK cells (median 47.52%, IQR 41.3–51.8%, mean ± SD 46.4 ± 11.9%) than HLA–B27–negative SpA patients (median 23.5%, IQR 21.3–38.6%, mean ± SD 28.6 ± 10.38%) (P < 0.001). There was no difference in the absolute numbers of NK cells or CD4+ or CD8+ T lymphocytes in PBMCs or, indeed, the SF samples, between groups (data not shown). In ERA PBMCs, the number of NK cells expressing KIR3DL2 was lower than in SpA or RA PBMCs (median 16.3%, IQR 15.1–21.2%, mean ± SD 18.4 ± 5.0%).

Figure 1.

Increased KIR3DL2 expression by natural killer (NK) and CD4+ T cells in patients with spondylarthritis (SpA) and enthesitis-related arthritis (ERA). A, KIR3DL2+ expression on CD3−,CD56+ NK cells from different groups. A greater percentage of NK cells express KIR3DL2 in HLA–B27–positive SpA peripheral blood (PB) (group 3b) compared with healthy control PB, rheumatoid arthritis (RA) PB, and HLA–B27–negative SpA PB. An expansion of KIR3DL2+ NK cells was also seen in SpA synovial fluid (SF) (group 6) and ERA SF (group 7) compared with RA SF (group 5). # = P < 0.001 versus 1, 2, and 3a; + = P < 0.05 versus group 5. B, KIR3DL2 expression on CD4+ T cells. A greater percentage of CD4+ cells expressed KIR3DL2 in HLA–B27–positive SpA PB (group 3b) compared with healthy control PB, RA PB, and HLA–B27–negative SpA PB. ERA PB (group 4) also had a higher percentage of KIR3DL2+ CD4+ cells compared with healthy control PB and RA PB. In SpA SF (group 6) and ERA SF (group 7) there was also an expansion of KIR3DL2+ CD4+ cells compared with RA SF. # = P < 0.001 versus groups 1 and 3a and P < 0.005 versus group 2; = P < 0.005 versus groups 1 and 2; + = P < 0.01 versus group 5. C, KIR3DL2 expression on CD8+ T cells. There were no significant differences in KIR3DL2 expression on CD8+ cells between the patient groups. Horizontal bars show the means. PBMC = PB mononuclear cells; SFMC = SF mononuclear cells.

Table 2. Frequency of KIR expression on CD4, CD8, and NK cells in PBMCs and SFMCs*
Controls (n = 18)RA (n = 10)SpA (n = 35)ERA (n = 5)RA (n = 10)SpA (n = 10)ERA (n = 5)
  • *

    Values are the median (interquartile range) of KIR expression within natural killer (NK), CD4, and CD8 populations. See Table 1 for other definitions.

  • P < 0.001 versus control PB and RA PB.

  • P < 0.05 versus RA SF.

  • §

    P < 0.001 versus control PB; P < 0.005 versus RA PB.

  • P < 0.005 versus control PB and versus RA PB.

  • #

    P < 0.01 versus RA SF.

 KIR3DL224.3 (18.7–27.7)23.3 (17.9–25.5)43.5 (35.2–50.3)16.3 (15.1–21.2)11.5 (8.8–12.5)15.7 (13.8–17.1)15.3 (13.6–20.2)
 KIR3DL118.9 (3.5–31.9)24.7 (12.0–26.2)21.6 (7.4–35.4)8.9 (7.6–12.1)2.2 (2.1–3.9)2.5 (2.0–5.1)1.4 (1.3–1.5)
 CD158a22.7 (12.5–25.7)20.3 (18.5–25.3)26.5 (12.8–40.0)13.7 (13.1–16.8)3.6 (2.9–4.3)3.9 (3.2–4.3)1.9 (1.7–2.1)
 CD158b26.5 (24.7–34.5)26.2 (20.8–31.0)33.9 (27.3–40.8)10.8 (10.6–12.6)9.9 (8.1–11.2)9.3 (7.0–11.6)7.2 (6.0–9.1)
 KIR3DL25.3 (1.8–6.0)4.6 (3.2–7.2)11.2 (6.4–17.5)§10.2 (8.5–15.7)2.9 (2.1–4.3)4.8 (3.9–6.2)#5.3 (4.6–7.8)#
 KIR3DL10.1 (0.05–0.9)0.9 (0.4–1.5)0.5 (0.2–1.0)0.9 (0.7–1.0)0.8 (0.7–1.0)0.5 (0.1–1.1)0.7 (0.6–0.9)
 CD158a0.5 (0.4–1.1)0.5 (0.3–0.7)0.4 (0.1–1.2)0.1 (0.03–0.2)0.6 (0.4–0.7)0.9 (0.7–1.0)0.7 (0.6–0.9)
 CD158b0.8 (0.5–1.1)0.4 (0.1–0.8)0.3 (0.1–0.8)0.1 (0.08–0.1)0.4 (0.2–0.8)0.7 (0.6–1.3)0.2 (0.1–0.4)
 KIR3DL29.3 (4.6–13.7)8.6 (5.6–22.0)11.4 (5.9–28.1)14.9 (11.3–18.1)6.3 (5.3–9.6)5.8 (3.7–10.6)5.9 (4.9–7.7)
 KIR3DL10.9 (0.4–1.5)1.4 (1.0–2.5)1.0 (0.6–2.0)1.2 (0.9–1.4)2.1 (2.1–2.4)1.5 (1.1–1.7)0.9 (0.8–1.0)
 CD158a1.0 (0.3–1.2)1.1 (0.8–1.5)0.7 (0.2–1.3)0.4 (0.3–0.6)0.2 (0.1–0.3)0.07 (0.04–0.12)0.08 (0.06–0.1)
 CD158b1.2 (0.8–1.8)1.8 (1.6–2.0)1.6 (1.5–2.1)2.0 (1.5–3.1)0.7 (0.3–0.9)0.8 (0.7–1.1)0.4 (0.2–0.7)

Compared with SF from patients with RA (median 11.5%, IQR 8.8–12.5%, mean ± SD 11.7 ± 3.5%), a higher number of NK cells from patients with SpA (median 15.7%, IQR 13.8–17.1%, mean ± SD 16.7 ± 5.8%) or ERA (median 15.3%, IQR 13.6–20.2%, mean ± SD 15.6 ± 2.7%) expressed KIR3DL2 (uncorrected P < 0.05 for both). Figure 2A shows representative examples of paired PBMCs and SF KIR3DL2 staining on NK cells from patients with RA, SpA, and ERA. The overall levels of KIR expression in SF were lower than in PBMCs, largely because of the increase in the CD56bright population in all SFMC samples studied, which confirmed previous observations (21). There was, however, no significant difference in the numbers of either CD56bright or CD56dim cells between the patient groups (Figure 2A and data not shown). Thus, the increase in KIR3DL2 expression on NK cells in SpA and ERA patients was not due to an increase in the CD56dim population per se. Interestingly, Figure 2A also shows evidence of expression of KIR3DL2 on a CD56− (CD3−) population of cells in both SpA and ERA. These may be cells that have down-regulated CD3 or CD56.

Figure 2.

Expansion of KIR3DL2+ NK and CD4 cells and activated effector phenotype in SpA. A, Expression of KIR3DL2 on NK cells from representative patients with RA, SpA, and ERA. PBMCs and SFMCs are shown. There was an expansion of KIR3DL2+ NK cells in PB and SF of SpA and ERA patients. Fluorescence-activated cell sorting (FACS) plots were gated on CD3− lymphocytes. B, Percentage of KIR3DL2+ and KIR3DL2− SpA PBMC NK cells expressing perforin, CD38, and β7 integrin. Data are the mean and SEM from 15 SpA patients. = P < 0.05. C, Expression of memory and activation markers on SF KIR3DL2+ CD4+ lymphocytes from a patient with SpA. KIR3DL2+ CD4+ cells had an activated, largely CD38+ effector memory phenotype (largely CD45RA−, CD45RO+, CD28−, CCR7−, and CD62L−). FACS plots were gated on CD3+,CD4+ lymphocytes. Upper quadrants are KIR3DL2+ cells; lower quadrants are KIR3DL2− cells. Numbers show the percentage of gated cells in each quadrant. D, Percentage of KIR3DL2+ and KIR3DL2− CD4 cells expressing markers of memory (CD45RA, CD45RO, CCR7, CD62L, CD28) and activation (CD38). Data are the mean and SEM from 10 SpA patients. = P < 0.05; ∗∗ = P < 0.005; ∗∗∗ = P < 0.001. See Figure 1 for other definitions.

A greater number of CD4+ T cells expressed KIR3DL2 in both SpA PBMCs (median 11.2%, IQR 6.4–17.5%, mean ± SD 11.5 ± 5.8%) and ERA PBMCs (median 10.2%, IQR 8.5–15.7%, mean ± SD 11.7 ± 3.2%) than in PBMCs from healthy controls (median 5.3%, IQR 1.8–6.0%, mean ± SD 4.7 ± 3.1%) (P < 0.005 versus SpA and ERA PBMCs) or RA patients (median 4.6%, IQR 3.2–7.2%, mean ± SD 5.3 ± 2.7%) (P < 0.005 versus SpA and ERA PBMCs) (Figure 1B and Table 2). HLA–B27–positive SpA patients had a higher number of KIR3DL2+ CD4+ cells (median 12.4%, IQR 9.7–18.1%, mean ± SD 13.0 ± 5.5%) compared with HLA–B27–negative SpA patients (median 5.6%, IQR 4.55–6.32%, mean ± SD 5.5 ± 1.7%) (P < 0.001).

A higher percentage of CD4+ T cells expressed KIR3DL2 in SpA SFMCs (median 4.8%, IQR 3.9–6.2%, mean ± SD 4.6 ± 1.1%) and ERA SFMCs (median 5.3%, IQR 4.6–7.8%, mean ± SD 5.8 ± 1.4%) compared with RA SFMCs (median 2.9%, IQR 2.1–4.3%, mean ± SD 3.2 ± 1.4%) (P < 0.01 versus SpA and ERA SFMCs). No significant differences in the expression of KIR3DL1, KIR2DS1-2, and KIR2DL1-3 were seen between patient groups, in either PBMCs or SFMCs (Table 2). Although some patients with SpA had higher (∼40%) frequencies of CD8+ T cells expressing KIR3DL2 in SpA (Figure 1C), this was not statistically significant when compared with healthy volunteers and RA patients.

Activated cytotoxic phenotype and expression of β7 integrin in KIR3DL2+ NK cells.

In view of the increased percentage of KIR3DL2-expressing NK cells among SpA patients (Figure 2A), we studied the phenotype of these cells ex vivo. Figures 2B and 3 show that KIR3DL2+ NK cells from SpA PBMCs expressed CD38, a marker of activation, and intracellular perforin, important in NK cytotoxicity. PB KIR3DL2+ NK cells also expressed higher levels of β7 integrin, a homing marker for cells that have recirculated from the gut, than did KIR3DL2− cells (Figure 2B).

Figure 3.

High levels of perforin, β7 integrin, and CD38 expression in KIR3DL2+ NK cells. Representative fluorescence-activated cell sorting plots from an HLA–B27–positive SpA patient are shown, depicting higher levels of intracellular perforin (A) as well as cell surface CD38 (B) and β7 integrin (C) on KIR3DL2+ cells compared with the KIR3DL2− population of PB NK cells. Cells gated on CD3−,CD56+ lymphocyte population; numbers show the percentage of gated cells in each quadrant. See Figure 1 for definitions.

Effector memory phenotype in KIR3DL2+ CD4+ T cells.

We next studied the phenotype of KIR3DL2+ CD4+ T cells ex vivo in SpA. Figure 2C shows a representative staining of SF T cells, and Figure 2D shows data from PBMCs of 10 SpA patients. KIR3DL2+ CD4+ T cells in both PBMC and SFMC largely expressed CD45RO and were negative or low in their expression of CD28, CCR7, and CD62L. CD45RA expression was significantly lower than on KIR3DL2−,CD4+ T cells; some KIR3DL2+ cells coexpressed CD45RA and CD45RO. This phenotype is evidence that KIR3DL2+ CD4+ T cells are antigen-experienced.

Increased survival of KIR3DL2+ NK cells in the presence of B272.

In order to test the hypothesis that B272 promotes the survival of KIR3DL2+ NK cells, we sorted, by FACS, the KIR3DL2+ population of NK cells from PBMCs of 4 HLA–B27–positive patients with AS (Figure 4A) and cultured them in the presence of LBL721.220B27 cells expressing large amounts of B272 (27). Figure 4B shows that NK cells cultured with LBL721.220B27 were protected from apoptosis (assessed by expression of annexin V and 7-AAD) compared with controls. Although this cell line also expresses B27 heterodimers, protection from apoptosis was not seen with B27 Cys67Ser transfectants, which express cell surface B27 heterodimers but not homodimers (27), or with B7 transfectants. Protection from apoptosis mediated by B272 was inhibited by the KIR3DL2-specific mAb DX31 (Figure 4B) but not with isotype control antibody. This experiment was repeated 5 times with NK cells from 4 SpA patients with consistent results. Figure 4C shows values from 4 patients.

Figure 4.

Coculture with LBL721.220B*2705 cells expressing B272 promotes the survival of KIR3DL2+ NK cells. A, Left, fluorescence-activated cell sorting (FACS) showing purity (CD3−,CD56+, >98% pure) of NK cells from PB of an HLA–B27–positive SpA patient, negatively isolated with immunomagnetic beads from a magnetic-activated cell sorting kit. Right, KIR3DL1−,KIR3DL2+ cells (>99% pure) were isolated from this NK population for use in the apoptosis assay (cells gated on CD3−,CD56+ lymphocyte population). B, Representative FACS stains of KIR3DL2+,KIR3DL1− NK cells that were cocultured with LBL721.220 or transfectants for 7 days. LBL721.220B27, which expresses B272, protected NK cells from apoptosis (64.7% 7-aminoactinomycin D− [7-AAD−] annexin V−) compared with NK cells cultured alone in media (36.7%), LBL721.220 (40.9%), or HLA–B27 Cys67Ser or B7 transfectants. Protection from apoptosis by LBL721.220B27 was abrogated by the KIR3DL2-blocking monoclonal antibody DX31 but not with IgG2a isotype control antibody. FACS plots are gated on CD3−,CD56+ lymphocytes. Numbers show the percentage of gated cells in each quadrant. C, Percentage KIR3DL2+ NK cells protected from apoptosis (7-AAD−, annexin V−) when cultured in media alone or with LBL721.220 or its transfectants for 7 days. Data are the mean and SEM from 4 HLA–B27–positive SpA patients. See Figure 1 for other definitions.

Increased cytotoxic activity in PBMC NK cells from SpA patients compared with those from RA patients and healthy controls.

Figure 5 shows that PBMCs from 7 HLA–B27–positive SpA patients exhibited greater NK killing activity ex vivo against K562 targets than PBMCs from patients with RA (n = 7) or from healthy controls (n = 7). Killing was largely prevented by prior NK depletion (data shown for NK-depleted SpA PBMCs).

Figure 5.

Increased cytotoxicity of NK cells in SpA PB compared with RA and control PB NK cells. NK killing of K562 targets by PBMCs at different effector:target ratios is depicted. Samples from 7 HLA–B27–positive SpA patients, 7 RA patients, and 7 healthy controls were studied. Killing was prevented by prior NK depletion (NK−) in SpA, RA, and healthy control PB (shown for SpA). Data are the mean ± SEM from 1 representative experiment of 2 performed. See Figure 1 for definitions.


We have shown a substantially increased frequency of KIR3DL2 expression on NK cells from SpA peripheral blood and synovial fluid. Thus, a median of 43.5% of PBMC NK cells from SpA patients expressed KIR3DL2, compared with 24.3% of NK cells from healthy controls and 23.3% from RA patients (P < 0.001 for both comparisons). When SpA patients were subdivided into HLA–B27–positive and B27-negative groups, the increased KIR3DL2 expression was largely confined to the B27-positive individuals, with 50% of PBMC NK cells expressing KIR3DL2. There was no increase in the absolute number of NK cells in SpA or ERA patients, although we did find evidence of enhanced NK cytotoxicity in SpA compared with RA patients and healthy controls. No difference in the expression level of other KIRs studied (KIR2DS1-2, KIR2DL1-3, and KIR3DL1) was observed.

Although the percentages of NK KIR3DL2+ cells were lower in SF than in blood (in part because of the preponderance of CD56bright NK cells [28]), KIR3DL2+ cell numbers were significantly higher in SF NK cells from patients with adult SpA and in juvenile ERA than in patients with RA. We confirmed the findings of increased CD56 expression on NK SF cells reported by Dalbeth et al (21, 28), although KIR3DL2 expression was not the main focus of their studies.

Since T lymphocytes also express KIRs, although at lower levels than NK cells, we also studied KIR3DL2 expression on T cells. In blood, approximately twice as many SpA and ERA CD4 T cells expressed KIR3DL2 as did T cells from healthy and RA controls. In SpA and ERA SF, there were also a higher number of KIR3DL2+ CD4+ T cells compared with RA SF. This is consistent with the findings in PB. Because not all SpA PB had paired SF samples and in the absence of SF from healthy controls for comparison, we would place less weight on the observation that in SpA, the percentage of CD4 KIR3DL2+ cells was lower in SF than PB.

Our finding of an expanded KIR3DL2+ population on ERA PB and SF CD4 T cells suggests that similar pathogenic mechanisms operate in ERA and SpA. It also raises the possibility that differences in clinical phenotype, for example, the rarity of sacroiliac involvement at onset of ERA, might relate to the maturation status of the immune system and in particular of NK cells. The lower level of KIR expression on PB NK cells in children with ERA compared with adults with SpA may be age related. Although cord blood NK cells express lower levels of KIR than adult PB (29), little is known about NK repertoire and KIR expression during childhood. Unfortunately, we did not have access to blood or synovial fluid from healthy age-matched control children. Interestingly, NK cell dysfunction has been observed in children with severe systemic juvenile idiopathic arthritis (JIA) (24), where defective killing has been associated with a reversible defect in perforin expression in NK cells (30, 31). The impairment in NK activity in systemic JIA has been linked to the development of macrophage activation syndrome (for review, see ref. 32). While these findings implicate NK cells in the pathogenesis of childhood arthritis, no patients with ERA were included in these studies.

Our results suggest a common pathogenesis in adult SpA and juvenile ERA, through interaction of abnormal B272 with KIR3DL2. This is supported by the observation that the increase in KIR3DL2 expression is greater in the HLA–B27–positive SpA patients compared with the HLA–B27–negative SpA patients. The HLA–B27–positive group also had higher KIR3DL2 expression compared with healthy controls and RA patients. We speculate that disease might result from a specific trigger increasing B272 expression in susceptible individuals or from increased responsiveness to KIR3DL2 ligation.

NK cells mediate their cytotoxic effector actions through release of cytokines or perforin/granzyme A, and this is triggered via activating receptors such as CD16 (Fcγ receptor IIIa) and NCRs (33). In this study we show that the KIR3DL2+ NK cell population is largely activated with >70% expressing intracellular perforin and high levels of CD38. Interestingly, CD38 engagement triggers cytotoxic responses, including cytokine and granzyme release, by activated NK cells (34). This activating response has been likened to that of NK cell triggering via CD16 (35). In combination, the high perforin and CD38 expression on KIR3DL2+ NK cells would likely result in increased cytotoxicity, and we provide preliminary evidence that PBMC NK cells from SpA patients do indeed have enhanced killing ability. Similarly, KIR3DL2+ CD4+ T cells expressed high levels of CD38 and low levels of CD62 ligand (L-selectin), suggesting that they too are activated. KIR3DL2+ CD4+ T cells also largely expressed CD45RO which, together with low levels of CCR7 and CD28, indicates an antigen-experienced effector memory phenotype. These data suggest that KIR3DL2+ NK and CD4 T cells have experienced some form of peripheral activation.

Intriguingly, KIR3DL2+ NK cells also expressed increased levels of β7 integrin, a marker of migration from the gastrointestinal (GI) tract. Trafficking into GI mucosal tissues is aided by β7 integrin pairing predominantly with the α4 integrin, and α4β7 then binds to the mucosal vascular address in mucosal addressin cell adhesion molecule 1 (36). Increased β7 integrin expression has been observed in SpA synovium compared with RA synovium (37). Our results raise the possibility that levels of KIR3DL2+ NK and CD4 T cells may have been initially activated within the GI tract. It would be most informative to study KIR3DL2 expression on gut tissue from patients with SpA. This may be important in view of the association of SpA with inflammatory bowel disease. In contrast, expression of β7 integrin has been reported to be low in RA PB (38).

It is important to note that KIR3DL2 has a number of allelic forms (39), and it is possible that these differences might have functional consequences. However, KIR3DL2 is present on both KIR haplotypes, and all patients and controls studied here expressed KIR3DL2 as assessed using the KIR3DL2-specific mAb DX31.

We hypothesize that interaction with B272 gives a survival signal to T and NK cells expressing KIR3DL2. We provide evidence that this occurs ex vivo; coculture with B272-expressing cells protects KIR3DL2+ NK cells from apoptosis (and this can be blocked with mAb DX31). Although KIR3DL2 ligation provides an inhibitory signal, there is evidence that KIR expression is associated with enhanced memory T cell survival (18). Expression of KIR is thought to inhibit activation-induced cell death in T cells by blocking Fas ligand induction upon stimulation (40). Thus, KIR3DL2 ligation by B272 could protect T and NK cells from activation-induced cell death in the presence of a second signal. Interestingly, B272 has also been shown to bind to LILRA1, an activating receptor expressed on myelomonocytic cells (15). The overall pattern of B272 receptor recognition by myelomonocytic cells could result in a net proinflammatory response through activation by LILRA1 and/or by failure of recognition by the inhibitory receptor LILRB1. Such a proinflammatory effect could be amplified through enhanced survival of NK and T cells through KIR3DL2 ligation.

In preliminary experiments we have shown increased killing of K562 targets by NK cells from patients with SpA compared with those from both RA patients and healthy controls. These data would be consistent with high perforin expression in the NK KIR3DL2+ fraction in SpA. Defective NK killing has been described in RA (41) and systemic lupus erythematosus (42). Some studies of NK function in AS showed normal killing compared with controls (43, 44). Early studies also suggested that NK killing was lower in synovial fluid (45). In retrospect, this is consistent with the preponderance of CD56bright NK cells, which are perforin negative/low, in SF (21). More detailed study is required using the molecular markers for NK cells that are now available. We speculate that enhanced killing may also result from increased coexpression of stimulatory receptors such as the NCRs, together with enhanced perforin and CD38 expression shown here. Indeed it is possible that KIR3DL2 ligation by B272 results in up-regulation of such receptors. In addition to cytotoxicity, it is now clear that NK cells play a major role in immune response through a variety of other mechanisms including direct cytokine production, promotion of monocyte tumor necrosis factor α release, and interaction with other cells including dendritic cells (through ligands such as OX40 ligand). These interactions will require further investigation.

Thus, our study is the first to demonstrate indirect evidence for a role of cell surface B272 expression in SpA and ERA pathogenesis. B272-expressing cells can protect KIR3DL2+ NK cells from apoptosis. This suggests that expression of B272 may promote the survival of KIR3DL2+ NK and T cells during inflammation, and these cells could then have a pathogenic role either directly or through stimulation of other cell populations such as monocytes.