A polymorphism in a phosphotyrosine signalling motif of CD229 (Ly9, SLAMF3) alters SH2 domain binding and T‐cell activation

Summary Signalling lymphocyte activation molecule (SLAM) family members regulate activation and inhibition in the innate and adaptive immune systems. Genome‐wide association studies identified their genetic locus (1q23) as highly polymorphic and associated with susceptibility to systemic lupus erythematosus (SLE). Here we show that the Val602 variant of the non‐synonymous single nucleotide polymorphism (SNP) rs509749 in the SLAM family member CD229 (Ly9, SLAMF3) has a two‐fold lower affinity compared with the SLE‐associated Met602 variant for the small adaptor protein SAP. Comparison of the two variants in T‐cell lines revealed the Val602 variant to be significantly more highly expressed than CD229 Met602. Activation was diminished in cells expressing CD229 Val602 compared with CD229 Met602 as measured by up‐regulation of CD69. There was no correlation between homozygosity at rs509749 and activation in peripheral blood mononuclear cells from healthy donors. These findings identify potential mechanisms by which a single SNP can perturb fine‐tuning in the immune system with significant functional consequences.

SLAM family receptors signal through one or more intracellular immunoreceptor tyrosine-based switch motifs (ITSMs). 4,5 This motif (TxYxxI/V) binds SH2 domains of the adaptor molecules, signalling lymphocyte activation molecule-associated protein (SAP; SH2D1A) and EAT-2 (SH2D1B) in a tyrosine-phosphorylationdependent manner. 4,[6][7][8] The importance of SAP became clear from studying patients with X-linked lymphoproliferative disease who exhibit an abnormal immune response to Epstein-Barr virus infection. This rare immunodeficiency is typically associated with the absence or mutation of SAP. 1,2,7,9 SAP is expressed in leucocytes and links the kinase Fyn with SLAM family receptors in an atypical SH2-SH3 domain interaction. [10][11][12] Absence of functional SAP in T cells and natural killer cells severely compromises cellular immunity. 8,9,13,14 SLAM family receptors manifest activating and inhibitory effects, and the term 'switch motif' is derived from the dual specificity of ITSMs for activating and inhibitory SH2-domain-containing proteins. The switch from activation to inhibition is observed in the absence of SAP. 4 Direct competition between SAP and SH2-domain-containing phosphatases for ITSMs has been proposed as a molecular mechanism for inhibition. 4,15 The cytoplasmic tail of CD229 contains two ITSMs and an interaction between CD229 and SAP has been shown. 16 In common with all the SLAM family receptors that signal through ITSMs, activating effects of CD229 are dependent on SAP in primary human cells. 17 Studies with genetically manipulated mice lacking SLAM family adaptors revealed evidence of both activating and inhibitory effects of CD229. [18][19][20][21] CD229 and other SLAM family members have been linked with susceptibility to systemic lupus erythematosus (SLE), a complex autoimmune disease. 22,23 The human SLAM locus of 1q23 is polymorphic and is syntenic with the SLE-associated Sle1 locus in mice. 24 A linkage study of families based on the SLAM locus identified an association between susceptibility to SLE and a single nucleotide polymorphism (SNP) rs509749 that causes a non-synonymous exchange in exon 8 of CD229. 23 Susceptibility correlated with a Val 602 (TVYAQV) to Met 602 (TMYAQV) switch in the first ITSM of CD229. 23 We have previously characterized differences in fine specificity of ITSMs, which correlated with functional data. 6 Here we compare the two variants of Ly9 and reveal differences in binding properties and functional effects that can explain how the association came to be identified.

Cells and antibodies
Media were purchased from Sigma-Aldrich (St Louis, MO). Jurkat Clone 20 (JC20) and HEK-293T cells were grown in RPMI-1640 and Dulbecco's modified Eagle's medium (4Á5 g/l glycerol, 110 mM sodium pyruvate), respectively, supplemented with 2 mM glutamine, 10% fetal calf serum (FCS), 100 U/ml penicillin and 100 lg/ ml streptomycin. Puromycin (1 lg/ml) was added to maintain protein and small hairpin (sh)RNA expression in stable cell lines. Peripheral blood mononuclear cells (PBMCs) were isolated from the blood of healthy donors by Ficoll density gradient centrifugation and frozen at À80°in FCS containing 10% DMSO.

T-cell activation and flow cytometry
Jurkat T cells (5 9 10 4 ) or PBMCs (0Á2 9 10 5 ) were plated in each well of 96-well flat-bottomed plates precoated with the indicated concentration of anti-CD3e and stimulated for 6 hr or 16 hr, respectively. Surface molecules were stained with the appropriate antibodies directly before and after stimulation and expression levels were analysed using the BD FACSCaliburTM and Beckman Coulter CyAnTM ADP Analyzer. Cells were analysed based on live, single cells according to forward light scatter/side scatter characteristics unless stated otherwise. Data were evaluated using FLOWJO, MS EXCEL and GRAPHPAD PRISM 6 software.

Western blot analysis
Western blotting was performed as previously described. 6 Briefly 10 7 Jurkat T cells were lysed in 200 ll Triton X-100 buffer and 2Á5 9 10 5 to 7Á5 9 10 5 cells were resolved by SDS-PAGE under reducing conditions. Protein bands were detected by the LI-COR Odyssey Sa system after developing with rabbit anti-SAP antibody (clone FL-128 Santa Cruz Biotechnology (Santa Cruz, CA)), mouse anti-phosphotyrosine (clone PT-66, Sigma, St Louis, MO), goat anti-HA (biotinylated, Vector Labo-

Results
The SLE-associated variant of CD229 ITSM1 (Met 602 ) binds more strongly than the Val 602 variant to SAP We first investigated the effect of rs509749 on interactions of the CD229 ITSM. As it was reported that rs509749 affected T-cell populations 23 we focused on measuring the interaction between CD229 and the adaptor SAP because  both are expressed in T cells 5 and an interaction of CD229 ITSMs with SAP has been shown. 16 We measured the interaction between SAP and the ITSMs of the two variants of CD229 by surface plasmon resonance at 37°. Increasing concentrations of soluble monomeric recombinant human SAP were injected over immobilized peptides representing the CD229 Val 602 or Met 602 ITSMs phosphorylated on the tyrosine residues (Fig. 1a). Plotting the equilibrium binding data, we observed a two-fold higher K D for binding of SAP for the variant CD229 Val 602 (K D = 2Á8 lM) compared with the risk allele, Met 602 of CD229 (K D = 1Á5 lM) (Fig. 1b). The same two-fold weaker affinity for the allele Val 602 was consistently observed in three separate experiments (Table 1). ITSM peptides of the same length from CD244 served as a control for protein activity (Table 1). 6 We compared binding of SAP to tyrosine phosphorylated peptides of the same length representing the CD229 ITSM containing the rs509749 SNP (TV/MpYAQ) and the more distal ITSM (TIpYCS). SAP bound to the second non-polymorphic sequence (TIpYCS) with up to an order of magnitude greater strength (K D = 0Á1 lM at 37°C; n = 3 data not shown).
The CD229 Met 602 to CD229 Val 602 substitution does not alter specificity for an SH2 domain from an inhibitory phosphatase The ITSMs have dual specificity for the adaptor proteins and for SH2-domain-containing inhibitory enzymes including the inositol phosphatase, SHIP. 6,29 To test whether the Val 602 to Met 602 modification in ITSM1 in CD229 alters specificity, we measured binding of the SH2 domain from the phosphatase SHIP-1 to both variants. Binding of the SHIP SH2 domain was detected but the affinity was at least an order of magnitude weaker than SAP binding (Fig. 1). The Val 602 to Met 602 variation did not alter specificity for an activating adaptor SH2 domain compared with an SH2 domain from an inhibitory enzyme. Equilibrium dissociation constants for binding of SHIP SH2 domain to CD244 ITSM1 and FccRIIb ITIM are within the range for published data 6 and show that the protein is active (Table 1).

CD229 Val 602 is expressed more highly than the SLE-associated CD229 Met 602 variant
To examine how rs509749 SNP affects the surface expression of CD229 receptor in cells, we expressed the Val 602 and Met 602 variants in Jurkat cells. Jurkat cells express CD229 and SAP (Fig. 2a and see Fig. 4b). We used a bicistronic vector to down-modulate endogenous CD229 by targeting its 3 0 UTR and simultaneously express CD229 Val 602 or CD229 Met 602 under the control of the pEF1a promoter. Flow cytometry confirmed that endogenous CD229 was reduced by transduction of shCD229 and not by a control scrambled shRNA construct (Fig. 2a,b). Transduced CD229 Val 602 and CD229 Met 602 were concomitantly expressed (Fig. 2a,b)  Paired t-test was used to calculate differences between anti-CD3e and anti-CD229 stimulation. ns: non-significant, *P < 0Á05, **P < 0Á01. Results represent mean of two experiments. In each experiment two independently generated CD229M and CD229V cell lines were analysed. Error bars in (a) and (c) represent +SEM.
CD229 Val 602 was consistently expressed at a higher level compared with the CD229 Met 602 variant and both were more abundant at the cell surface than the original endogenous CD229 (Fig. 2a,b).
Activation through CD3 is diminished in Jurkat T cells expressing CD229 Val 602 compared with CD229 Met 602 The biochemical and expression data revealed two differences between the CD229 variants; weaker binding to SAP and higher expression at the cell surface by the CD229 Val 602 . We compared the functional effects of endogenous CD229 and the two variants in the Jurkat cells by measuring up-regulation at the cell surface of the activation marker CD69 after anti-CD3e stimulation.
There was no significant difference in the dose-dependent response to anti-CD3e cross-linking between cell lines expressing endogenous CD229 (shCTR), shCD229, CD229 Met 602 or CD229 Val 602 (Fig. 3a). In these experiments there was no significant difference between the levels of CD3 or CD229 in CD229 Met 602 or CD229 Val 602 expressing cell lines (Fig. 3b). These data show that the manipulation of CD229 expression has no deleterious effect on the cells. To investigate specific signalling effects of the two CD229 variants, we used a monoclonal antibody that has been shown to be effective in cross-linking and triggering through CD229. 30 We compared the effects of cross-linking CD3 and CD229 or CD3 and CD28. Co-stimulation by CD28 was observed in CD229 Met 602 but not CD229 Val 602 cells (Fig. 3c). There was also a difference in response between CD229 Met 602 and CD229 Val 602 cells when using anti-CD229. Cross-linking CD229 Val 602 had an inhibitory effect on CD3-induced CD69 expression levels, which were unaltered in CD229 Met 602 cells. These data revealed a functional difference between CD229 Val 602 -and CD229 Met 602 -expressing cells.
The cytoplasmic tail of CD229 Val 602 is more inhibitory than the SLE-associated CD229 Met 602 cytoplasmic tail To focus on detecting a potential signalling difference between CD229 Val 602 and CD229 Met 602 , the extracellular region of the receptor was replaced by an HA-tag. Flow cytometry confirmed down-modulation of the endogenous full-length receptor by its shRNA and concomitant expression of HA-tagged CD229 (Fig. 4a). Similarly to the full-length receptors, the HA-CD229 Val 602 (HAV) was more highly expressed than HA-CD229 Met 602 (HAM) (Fig. 4a,b) confirming that this expression effect was due to the cytoplasmic region. There were equal levels of SAP in both cell lines and receptors were phosphorylated to the same extent as judged by Western blot analysis (Fig. 4b).
There is no correlation between expression of SLEassociated CD229 Met 602 and increased activation of healthy primary human cells CD229 was up-regulated on cells from SLE patients but the genotype of CD229 was not determined in that study. 17 The A/G variants of rs509749 SNP are present in nearly equal measure (A = 55%, G = 45%) in the European population whereas in other populations, G is the major allele. 31 We genotyped five healthy donors and identified two heterozygotes and three homozygotes, two CD229 Met 602 and one CD229 Val 602 . We compared levels of expression of CD229 and CD3 on PBMCs from the five human donors (Fig. 5a,b). CD229 was expressed at low levels with no significant differences between the genotypes nor was there any evidence that CD229V was expressed more highly than CD229M on T blasts ( Fig. 5a and data not shown). We stimulated PBMCs from the five donors with anti-CD3e and measured up-regulation of CD69. The two heterozygous donors responded more strongly compared with the homozygotes. Increased responses by the heterozygotes did not correlate with CD3 levels, which were lower for the heterozygotes compared with the homozygotes.

Discussion
The SNP rs509749 was associated with susceptibility of SLE and with altered T-cell populations in a family-based study. 23 The molecular mechanism of genome-wide associations with disease is rarely known. We identified biochemical and functional differences between two variants of CD229 that differ at rs509749, a non-synonymous SNP that alters the sequence of the membrane proximal ITSM1 of CD229 from TMYAQV to TVYAQV.
The ITSMs in the variants of rs509749 in CD229 differ at the Tyr-1 position. The amino acid at the Tyr-1 position is not crucial but influences the three-pronged binding mechanism of SH2 domains to ITSMs. 32 The CD229 ITSM1 alleles differ only at the Tyr-1 position and a twofold difference in affinity for SAP between CD229 Met 602 (K D = 1Á7 lM) and CD229 Val 602 (K D = 3Á4 lM) for SAP was measured. There was no evidence of a switch in specificity as suggested previously. 23 A higher affinity of SAP for the membrane distal ITSM2 might explain why the polymorphism at Tyr-1 of ITSM1 does not cause greater differences in CD229 signalling and impact on SLE.
The recombinant variant, CD229 Val 602 was more highly expressed than the SLE-associated CD229 Met 602 in Jurkat T cells. Several SLE-associated SNPs have been shown to change epigenetic imprints by modulating gene transcription. 33 Transcriptional alterations can be excluded in the comparison of CD229 variants transduced into Jurkat cells because the promoter as well as coding regions are not native. Nevertheless messenger RNA and/ or protein may be more stable for CD229 Val 602 compared with CD229 Met 602 . Analysis of flow cytometric data indicated that CD229 Val 602 was more readily internalized after CD3 monoclonal antibody-induced activation compared with CD229 Met 602 , suggesting that slower internalization of CD229 Val 602 is unlikely to explain the difference in expression (unpublished data). Significant differences in CD229 expression levels were not observed in PBMCs from healthy donors. There was up-regulation of SLAM family receptors, CD229 and of CD352 (NTB-A, SLAMF6) in T cells from patients with SLE compared with healthy controls. 17 However, among the 11 SLE patients studied, there was a lower correlation between expression and disease severity for CD229 compared with CD352. 17 The association between CD229 expression levels and SLE susceptibility is not straightforward. An imbalance between stimulus strength, expression levels of receptor and signalling molecules can change the outcome of cellular activity, as has been shown for the SLAM receptor CD244 (2B4). 34 Over-expressing CD229 will alter the balance between receptor and intracellular adaptor. Increased expression correlated with an inhibitory phenotype suggesting that SAP, on which activation depends, may be limiting. Indeed, we found comparable expression levels of SAP in cell lines expressing the two CD229 variants, implying a lower adaptor to receptor ratio in cells expressing the CD229 Val 602 variant. The lower affinity of SAP for the CD229 Val 602 variant as well as the imbalance between adaptor and receptor might account for diminished activity of cells transduced with the SLE susceptible CD229 variant in comparison with CD229 Met 602 .
In conclusion, amplification of differences at the molecular and cellular levels between the CD229 variants might be sufficient to explain how rs509749 was identified as being associated with susceptibility to SLE specifically in a family-based study. 23,35,36