In this Chapter a flow cytometry-based autorosetting assay was developed to investigate the molecular basis of murine autorosetting. Using the assay the main findings were: (i) autorosetting cells are primarily DP and to a lesser extent CD8 SP thymocytes, whereas autorosetting by peripheral T lymphocytes is negligible, (ii) autorosetting is mediated by the CD8αβ heterodimer on thymocytes, with CD8β making a major contribution to erythrocyte binding, (iii) CD8 interacts with highly sulfated HS on the erythrocytes surface, this interaction specifically requiring the presence of 6-O-sulfate groups, (iv) sialylation of CD8β on peripheral CD8+ T lymphocytes results in the masking of the HS binding site on CD8 molecules, a process that can be reversed by neuraminidase treatment and, finally, (v) it was confirmed that HRG in serum inhibits autorosetting although other, biochemically similar, molecules in plasma appear to be inhibitory. Consistent with previous reports (Charreire and Bach, 1975, Kolb, 1977, Sia and Parish, 1980a, Sia and Parish, 1980b), the data shown in FIG. 3.4 confirmed that the thymus is the primary depot for autorosetting lymphocytes, 60-80% of DP and 20% of CD8 SP thymocytes autorosetting, whereas few peripheral CD8+ T lymphocytes (i.e., less than 5%) autorosetted. At face value, it is conceivable that autorosetting is developmentally controlled, being enabled at a certain point during DP thymocyte maturation, but eventually being disabled in CD8 SP thymocytes and severely impeded in peripheral
Autorosetting: Mechanisms & Regulators
lymphocytes. Although previous reports claimed that up to 20% of peripheral lymphocytes autorosetted, only 6% of peripheral lymphocytes autorosetted in this study, a discrepancy that is likely to be due to the technical advantages of the flow cytometry- based autorosetting assay. Firstly, Hoechst-33342 labelling allowed the analysis of autorosetting to be entirely on viable lymphocytes, eliminating the artifactual binding of erythrocytes by dead/dying cells which could have occurred in earlier studies. Secondly, the shear force of the flow cytometer sheath fluid would disperse weakly formed autorosettes. It should also be noted that the subset-specific analysis of autorosetting in the current assay eliminated the dilution effect of non-rosetting populations, presumably encounted in earlier studies. Thus, the data presented in this Chapter were generated by a highly objective assay that only counted stable autorosettes formed by viable lymphocytes and, consequently, was ideally placed to facilitate the identification of the autorosetting receptor on thymocytes and its ligand on erythrocytes.
Earlier studies suggesting that autorosetting is MHC-I restricted (Primi et al., 1979, Charreire and Bach, 1975, Sia and Parish, 1980a), which indirectly implicated CD8 in autorosetting as this molecule binds MHC-I. Indeed, based on the ability of anti-CD8 mAbs to inhibit autorosetting and the failure of CD8-deficient thymocytes to autorosette (FIG. 3.6), it was established that the CD8αβ heterodimer is the autorosetting receptor, with CD8β being the key autorosetting component of the heterodimer. Although partial inhibition of autorosetting by an anti-TCRβ mAb (FIG. 3.5) would suggest that TCRβ
participates in autorosetting, this results is probably due to steric hindrance as the TCR complex is located proximal to the CD8αβ molecule (Devine et al., 2006). Unfortunately, the study of autorosetting by TCR-deficient thymocytes is not feasible as TCR-deficiency results in aberrant thymocytes development (Mombaerts et al., 1992). Moreover, MHC-I molecules on erythrocytes are not the main autorosetting ligand interacting with the CD8αβ heterodimer on thymocytes as their absence in KO mice did not affect autorosetting (FIG. 3.7). A possible explanation for this discrepancy is that the high shear forces in the flow cytometry assay used in this thesis to measure autorosetting eliminated the contribution of weak TCR-MHC-I interactions to the formation of autorosettes.
Consistent with an earlier report (Parish et al., 1984), it was confirmed that the sulfated polysaccharides heparin and DxS completely inhibited autorosetting (FIG. 3.8A). These
Chapter 3
data suggested that analogous molecules on the erythrocyte surface, namely HS, could function as an autorosetting ligand, a role ultimately confirmed by the failure of HPNSE and HPSE treated erythrocytes to autorosette (FIG. 3.8C). In addition, inhibition of the binding of anti-CD8 mAbs to lymphocytes by heparin and DxS (FIG. 3.9) suggested that HS may interact with the CD8αβ heterodimer, particularly CD8β. Indeed, the impaired ability of CD8 deficient thymocytes to bind FITC-heparin was consistent with the CD8αβ heterodimer directly binding heparin/HS (FIG. 3.10B and C). Furthermore, this HS interaction is 6-O-sulfate dependent and is optimal when maximum negative charges are available on the heparin/HS chains (FIG. 3.8D).
CD8β is developmentally α2-3 sialylated (FIG. 3.11A and B), the ST3Gal-1 sialyltransferase being upregulated in CD8 SP thymocytes and peripheral CD8+ T
lymphocytes, with this modification being previously reported to alter the conformation of the CD8αβ heterodimer (Moody et al., 2003). Additional to the blocking effect on MHC-I binding (Daniels et al., 2001), results presented in this Chapter indicate that α2- 3 sialylation of CD8αβ impedes the interaction of CD8 on thymocytes and peripheral CD8+ T lymphocytes with HS. Thus, biochemical desialylation of CD8 by neuraminidase treatment strikingly augmented thymocyte and restored peripheral CD8+
T lymphocyte autorosetting (FIG. 3.11C and D). Hence, α2-3 sialylation not only limits the ability of CD8 molecules on peripheral CD8+ T lymphocytes to bind heparin/HS, as desialylation of DP thymocytes also enhances heparin/HS binding by CD8 as well as autorosetting. Moreover, data shown in FIG. 3.13 unequivocally show that autorosetting by neuraminidase treated peripheral CD8+ T lymphocytes is HS-dependent.
The data shown in FIG. 3.14A confirmed that serum-derived HRG strongly inhibits autorosetting, consistent with earlier reports from the Parish Laboratory (Rylatt et al., 1981, Sia et al., 1982). Although not tested in this study, HRG is likely to block autorosetting by binding to erythrocyte HS (Jones et al., 2004), thus preventing it from binding to CD8β. Whilst HRG is the major serum component that is responsible for autorosetting inhibition by WT serum, the enhanced ability of HRG deficient serum to inhibit autorosetting suggests the dominance of a substitute inhibitor in the HRG.KO mice (FIG. 3.14C). Western blotting revealed that HRG was absent from HRG.KO sera but Co2+-affinity chromatography, which depleted HRG from WT serum, also removed the unknown autorosetting inhibitor from HRG.KO serum. Thus, it is possible that the
Autorosetting: Mechanisms & Regulators
autorosetting inhibitor in HRG.KO serum is an isoform of HRG not recognised by the generic anti-HRG polyclonal Ab used for Western blotting, being produced from an artifactual transcript encoded by the altered HRG gene in HRG.KO mice. Further work is required to resolve this issue.
In summary, using a new flow cytometry-based autorosetting assay, the molecular mechanisms that facilitate and control murine autorosetting have been characterised. The CD8αβ heterodimer, but mainly CD8β, on DP thymocytes binds directly to highly negatively charged HS chains, carrying 6-O-sulfate groups, on erythrocytes. During development in the thymus, and in the periphery, α2-3 sialylation of CD8β impedes the ability of CD8 to bind HS, an inhibitory effect that is reversed by enzymatic desialylation. Thus, together with circulating HRG, the ability of CD8 to interact with HS is constantly and tightly suppressed in peripheral CD8+ T lymphocytes.