DOCK8 deficiency impairs CD8 T cell survival and function in humans and mice

To explore the effects of DOCK8 deficiency on T cells, we first analyzed peripheral blood T cells from six DOCK8-deficient patients from five unrelated kindreds. Some of the clinical characteristics of these patients are detailed in Table I. Consistent with a previous study (Engelhardt et al., 2009), the frequency of CD8 T cells within the total lymphocyte population from DOCK8-deficient patients was comparable with that of normal age-matched controls (Fig. 1, A and B). However, when subsets of CD8 T cells were enumerated, there was a dramatic and highly significant loss of naive (CD45RA⁺CCR7⁺) cells (∼10% in patients compared with ∼45% in normal controls), together with a smaller but still significant decrease in memory (CD45RA⁻CCR7^−/+) CD8 T cells. Strikingly, the majority of CD8 T cells from DOCK8-deficient patients exhibited a CD45RA⁺CCR7⁻ phenotype (Fig. 1, A and B), which is typically associated with cell exhaustion or replicative senescence (Appay et al., 2008). Because of their reexpression of CD45RA, these cells have been termed effector memory CD45RA⁺ (T_EMRA) cells.

For functional analyses, CD8 T cells from the different patients were sorted into populations of either naive (CD45RA⁺CCR7⁺), memory (CD45RA⁻CCR7^−/+), or T_EMRA (CD45RA⁺CCR7⁻) cells, labeled with CFSE, and stimulated with T cell activation and expansion (TAE) beads, which were coated with mAb to human CD3, CD28, and CD2, in the absence or presence of IL-2. Because of the lymphopenia that is characteristic of DOCK8-deficient patients (Zhang et al., 2009), not all subsets of these cells could be recovered in sufficient numbers from all patients. Consequently, the data presented in Fig. 1 C are a composite derived from experiments performed using subsets of CD8 T cells isolated from different patients. Naive (CD45RA⁺CCR7⁺) and memory (CD45RA⁻CCR7^−/+) CD8 T cells from normal controls were induced to undergo multiple rounds of division in response to stimulation with TAE beads, whereas the corresponding subset from DOCK8-deficient patients failed to divide under these stimulatory conditions (Fig. 1 C). Notably, provision of exogenous IL-2 failed to improve proliferation of DOCK8-deficient memory cells (Fig. 1 C). In contrast, there was an absence of a proliferative response by T_EMRA cells from either normal donors or patients (Fig. 1 C). This is consistent with previous findings that have investigated the CD8 T_EMRA cell subset from normal donors as well as in other clinical settings (Champagne et al., 2001; Brenchley et al., 2003; Geginat et al., 2003; Papagno et al., 2004). The shift of the CD8 T cell compartment toward exhausted T_EMRA cells in DOCK8 deficiency reveals a separate effect that complicates interpretation of the role of DOCK8 in human CD8 T cell proliferation.

In contrast to the striking shift in CD8 T cell phenotypic subsets, when the CD4 T cell compartment of DOCK8-deficient patients was examined, although there was a significant reduction in the frequency of CD4 T cells within the lymphocyte population, we found that the distribution of naive (CD45RA⁺CCR7⁺) and memory (CD45RA⁻CCR7^−/+) subsets in DOCK8-deficient patients resembled that observed for normal controls (Fig. 1, D and E). Despite this, functional analysis of naive and memory CD4 T cells from patients 1 (Fig. 1 F) and 2 (Fig. 1 G) revealed markedly diminished proliferation after stimulation with TAE beads compared with the vigorous division of corresponding CD4 T cell subsets from normal donors. Furthermore, although division of the mutant DOCK8 T cells could be improved by exogenous IL-2, it remained substantially less than that of control cells (Fig. 1, F and G). Thus, DOCK8 deficiency in humans diminishes clonal expansion of both CD4 and CD8 T cells in response to engagement of TCR and co-stimulatory molecules.

T_EMRA cells have been found to have experienced extensive replication in vivo (Hamann et al., 1999; Dunne et al., 2002; Rufer et al., 2003) yet exhibit poor proliferation in vitro (Champagne et al., 2001; Geginat et al., 2003). Furthermore, in the setting of chronic viral infection, these cells may have undergone replicative senescence (Chen et al., 2001; Brenchley et al., 2003; Papagno et al., 2004; Plunkett et al., 2005). A feature of the T_EMRA cell population is expression of CD57 on a substantial proportion of these cells (Appay et al., 2008). Interestingly, CD57 expression correlates with both an inability to undergo proliferation in response to either polyclonal or antigen-specific stimulation and heightened susceptibility to apoptosis in vitro (Brenchley et al., 2003). To further analyze CD8 T cell subsets in DOCK8 deficiency and to understand the proliferative defects in these cells, we assessed expression of a range of molecules whose expression is altered during differentiation of naive cells into effector cells (Appay et al., 2008). Compared with naive cells, memory CD8 T cells from normal donors not only acquire expression of CD57, but also dramatically up-regulate CD244 (2B4), CD11a, CD11b, CX3CR1, CD95, and granzyme B and concomitantly down-regulate CD27, CD28, and CD127 (IL-7Rα; Fig. 2 and Fig. S1). Notably, expression of these molecules is substantially increased (CD244, CD11a, CD11b, CD57, CX3CR1, CD95, and granzyme B) or decreased (CD27, CD28, and CD127) on normal T_EMRA cells (Fig. 2 and Fig. S1). Remarkably, the levels of expression of these molecules on memory CD8 T cells from DOCK8-deficient patients resembled that of T_EMRA cells from normal individuals (Fig. 2 and Fig. S1). Indeed, unlike normal memory and T_EMRA cells, there was no significant difference in their levels of expression between DOCK8-deficient memory CD8 T cells and normal T_EMRA cells (Fig. 2 and Fig. S1). Furthermore, these molecules were up- or down-regulated to a greater extent on DOCK8-deficient T_EMRA CD8 T cells compared with corresponding cells from normal donors (Fig. 2 and Fig. S1). Thus, the vast majority (>70%) of DOCK8-deficient T_EMRA CD8 T cells exhibited a CD57⁺CX3CR1⁺CD27⁻CD28⁻CD127⁻ phenotype, compared with only 40–50% of normal T_EMRA cells which had this phenotype (Fig. 2). Remarkably, even naive CD8 T cells from DOCK8-deficient patients differed phenotypically from normal naive cells inasmuch that they expressed elevated levels of CD11a, CD11b, and CD95 and displayed a mild down-regulation of CD28, CD27, and CD127 (Fig. 2 and Fig. S1). Overall, these data suggest that DOCK8-deficient CD8 T cells have undergone aberrant differentiation in vivo, such that most non-naive cells have acquired an effector/exhausted phenotype, whereas naive cells exhibit features of early activation.

The cell-autonomous role of DOCK8 in T cells was further investigated in specific pathogen-free C57BL/6 mice homozygous for the inactivating Dock8^pri/pri mutation. These animals inherit a Ser>Pro substitution predicted to disrupt a key α-helix in the Rho/Rac-binding site of the DHR2 domain and have a comparable phenotype with Dock8^cpm/cpm mice homozygous for a truncating mutation that abolishes over half of the protein (Randall et al., 2009). We first sought to understand the depletion of naive CD8 T cells in human DOCK8 deficiency by focusing on the formation and numbers of circulating naive CD8 T cells in Dock8^pri/pri mice.

Paralleling the human phenotype, the number of circulating naive CD8 T cells was decreased in Dock8^pri/pri homozygous mutants (Fig. 3, A and B). CD8 and CD4 T cell subsets were subdivided into naive and activated/memory cells by flow cytometric staining for CD44. Naive CD44^lo T cell subsets within both CD8 and CD4 populations were decreased ∼50% in blood, spleen, and LNs of Dock8^pri/pri mice (Fig. 3, A and B; and Fig. S2). In contrast, the percentages and numbers of activated/memory CD44^hi CD8 and CD4 T cells were comparable or slightly increased in DOCK8 mutant animals. Further analysis of the expression of CD45RB, PD-1, CTLA-4, KLRG1, CD127, CD122, CD25, CD27, CD3, or Tbet on the CD8⁺CD44^hi subset revealed no difference in Dock8^pri/pri mice compared with normal controls (Fig. S3). Parallel analyses of Dock8^cpm/cpm mice yielded comparable results (Figs. S3 and S4). These data indicate that DOCK8 deficiency in specific pathogen-free mice also diminishes naive CD8 cells, but in this situation it does not result in accumulation of exhausted memory CD8 T cells.

Decreased naive CD8 or CD4 T cells in the periphery of Dock8^pri/pri mice was not explained by any consistent decrease in the numbers of CD8 or CD4 single-positive T cells in the thymus (Fig. 3 C). In some experiments, there were slightly fewer CD4 and CD8 T cells in the thymus of Dock8^pri/pri mice (Fig. S5 A), although this was not consistently observed. Further subsetting of mature CD4 and CD8 thymocytes based on expression of CD62L and CD69 revealed slightly decreased frequencies of less mature CD69^hiCD4⁺ and CD69^hiCD8⁺ cells in Dock8^pri/pri thymi as a percentage of the total lymphocytes (Fig. 3 D). There was no consistent evidence for accumulation or deficiency of mature CD62L⁺CD69⁻ single-positive cells (Fig. 3 D and Fig. S5) in the thymus as is observed in animals with defective thymic egress (Matloubian et al., 2004; Carlson et al., 2006; Shiow et al., 2008). Moreover, when the Dock8^pri/pri mutation was crossed into transgenic mice expressing an MHC I–restricted TCR, OT-I, it caused no discernable decrease in the positive selection of CD8 single-positive T cells bearing this TCR (Fig. 3, E and F). Interestingly, the frequency of naive peripheral CD8 T cells bearing the OT-I TCR was unaffected by the Dock8^pri/pri mutation. Collectively, these results indicate that DOCK8 deficiency causes only subtle changes in thymic T cell differentiation.

The decrease in naive peripheral CD8 and CD4 T cells could reflect either a cell-autonomous requirement of DOCK8 in these T cell subsets or a secondary effect of abnormalities in other cell types. To distinguish between these alternatives, we constructed mixed bone marrow chimeras by reconstituting irradiated mice with 50:50 mixtures of congenically marked bone marrow from Dock8^pri/pri CD45.2 and Dock8^+/+ CD45.1 donors. Peripheral B220⁺ cells were reconstituted in almost equal ratios by mutant and WT cells, consistent with previous results demonstrating no effect of the mutation on circulating B cell numbers (Randall et al., 2009) and thereby providing an internal measure for relative engraftment by the two marrow sources in each chimeric animal (reconstitution of individuals shown by bars in Fig. 4 A). In contrast, CD45.2⁺ T cells with defective DOCK8 were dramatically underrepresented in peripheral CD8 and CD4 T cell subsets so that they comprised only 5% of circulating T cells on average. In the same animals, DOCK8 mutant T cells averaged 16% of thymic CD4 and CD8 single-positive T cells, 22% of CD4⁺CD8⁺ double-positive cells, and 30% of CD4⁻CD8⁻ double-negative cells. The proportion of peripheral CD45.2⁺ CD8 T cells that were CD44^hi in the mixed chimeras was not elevated, comprising ∼20% of Dock8^pri/pri CD8 cells and 18% of Dock8^+/+ CD8 cells (Fig. 4 B). Collectively, these results indicate that DOCK8 deficiency has a subtle cell-autonomous effect on T cells that is compounded at progressive stages of thymic and peripheral T cell formation.

To investigate further the decreased number of naive peripheral CD8 T cells, we adoptively transferred mixtures of congenically marked normal and Dock8^pri/pri mutant spleen T cells into normal C57BL/6 mice. The survival of these adoptively transferred T cells was then determined by assessing the total number of adoptively transferred cells present as well as the ratio of WT to mutant donor-derived T cells at various time points compared with their ratios at the time of injection. Both WT and mutant T cells slowly declined in frequency up to 39 d after adoptive transfer (Fig. 5, A and B). The contribution of mutant T cells as a percentage of total donor T cells was consistently lower than the input percentage. Similar results were obtained when the transferred WT and Dock8^pri/pri T cells were obtained from OT-I TCR transgenic mice with a homogeneous TCR specificity: mutant OT-I T cells accounted for 39% of the OT-I cells at the time of transfer but only 20% after 39 d, representing a halving of survival over this time period (Fig. 5 B, bottom). We conclude that the persistence of naive peripheral CD8 T cells is intrinsically decreased, albeit modestly, when they lack normal DOCK8. Collectively with the aforementioned normal thymocyte subsets, the reduced peripheral survival provides the simplest explanation for the 50% lower numbers of naive CD8 T cells at steady-state.

We asked whether the decrease in survival of naive CD8 cells lacking normal DOCK8 was caused by a difference in the expression of the α chain of the receptor for IL-7, CD127, or in the T cell response to IL-7. Only a subtle but nevertheless consistent decrease in CD127 expression was observed on naive CD8 and CD4 T cells from Dock8^pri/pri mice relative to WT littermates (Fig. S6). DOCK8 mutant CD8 and CD4 T cells responded to optimal concentrations of IL-7 in culture with indistinguishable short-term survival relative to WT controls (Fig. S7 A), although they did exhibit a slight decrease in Bcl-2 induction under these conditions (Fig. S7, B and C).

We next examined how DOCK8 deficiency affected CD8 T cell activation. Naive CD8 T cells specific for ovalbumin peptide SIINFEKL presented by H2 K^b were purified from the spleens of OT-I TCR transgenic mice with WT or mutant DOCK8. The T cells were CFSE labeled and co-cultured with SIINFEKL-pulsed bone marrow–derived DCs, and their cell division was followed over a time course by flow cytometry (Fig. 6 A). After 28 h, most of the WT cells had completed their first cell division, whereas few mutant cells had divided at this time point and only completed their first division by 34–40 h. Subsequent divisions nevertheless occurred rapidly so that by 49 h there was little difference between the mutant and WT cells. Addition of exogenous IL-2 did not rectify the delayed first division of DOCK8 mutant T cells (not depicted). In vitro activated Dock8^pri/pri CD8 OT-I cells had normal cytotoxic activity in chromium release assays (Fig. S8), and normal TNF, IFN-γ, and IL-2 were produced by Dock8^pri/pri CD8 cells in response to PMA and ionomycin (not depicted). DOCK8-deficient nontransgenic T cells divided almost normally in response to anti-CD3 and CD28 antibodies, although there was again evidence of a slight decrease (not depicted).

To investigate earlier steps in CD8 T cell activation, DOCK8 mutant and WT naive OT-I T cells were purified and co-cultured with SIINFEKL-pulsed DCs as above, and cell conjugates formed after 1 h were fixed and immunofluorescently labeled. Polarization of the microtubule-organizing center (MTOC; identified by tubulin staining) to the T cell–DC interface was used as a marker of immunological synapse formation. Images were obtained by confocal microscopy, and the polarization of proteins important for immune synapse formation was scored, with 3 indicating that the protein of interest was highly polarized to the T cell/DC interface, 0 indicating no polarization, and −3 indicating that the protein of interest was highly polarized away from the T cell–DC interface. LFA-1 was highly polarized to the immune synapse of WT OT-I T cells, whereas Dock8^pri/pri OT-I T cells mostly showed an even distribution of LFA-1 on the cell surface, indicating a failure of LFA-1 polarization to the synapse (Fig. 6, B and C). DOCK8 mutant CD8 T cells also failed to polarize actin to the synapse, whereas polarization of PKC-θ occurred relatively normally (Fig. 6, D–G). Together these results indicate that DOCK8 is required to form a normal T cell immune synapse with antigen-presenting DCs.

To extend the aforementioned results and examine how DOCK8 mutation affects CD8 T cell clonal expansion and memory cell persistence in vivo after a viral challenge, OT-I T cells with either normal or defective DOCK8, distinguishable by CD45.1 and CD45.2 allelic markers, were mixed, and 1,000 of each were cotransferred into normal C57BL/6 mice. At the same time, the recipient mice were injected i.p. with a modified influenza virus that produces the SIINFEKL peptide incorporated within the neuraminidase stalk. Thus, the transferred OT-I T cells will receive help from, and compete with, normal T cells recognizing other components of the virus in the recipient mice so that DOCK8 deficiency will be specifically limited to half of the transferred OT-I T cells. In addition to testing the cell-autonomous function of DOCK8 in CD8 T cells, any mouse to mouse variation in viral exposure or kinetics of the response could be internally controlled by directly comparing the ratio of the cotransferred OT-I T cells within a single animal. Primary and secondary viral challenges of the recipients were performed with two separate influenza strains, H3N2 (HKx31) and H1N1 (PR8), both expressing the SIINFEKL peptide, to recall memory CD8 T cell responses against the virus and the ovalbumin peptide without interference by neutralizing antibodies against viral envelope proteins (Jenkins et al., 2006).

Flow cytometric analysis of lymphocytes from spleen, peritoneal cavity, and bone marrow was used to enumerate the transferred mutant and WT OT-I CD8 T cells (Fig. 7 A). Both DOCK8 mutant and WT OT-I CD8 T cells increased >1,000-fold in response to viral antigen challenge. The DOCK8 mutant T cells clonally expanded almost as well as the normal controls in the same animals 7 d after the first viral challenge, although there was a subtle but consistent decrease in each animal reproduced across many experiments (Fig. 7 B). Thus, the mutant OT-I CD8 T cells represented ∼35% of all donor-derived OT-I CD8 T cells in the spleen, peritoneal cavity, and bone marrow at day 7 after infection, compared with 45% of the donor cells at the time of adoptive transfer (Fig. 7 B, bottom). This result extends the aforementioned in vitro proliferation experiments and establishes that DOCK8 deficiency causes only a subtle decrease in the net proliferation of CD8 T cells in response to viral antigen in vivo.

By 28 d after the first viral challenge, when viral antigen had been cleared, the OT-I CD8 T cells with WT DOCK8 had decreased in number in the spleen to ∼30% relative to their peak numbers at day 7 (Fig. 7 B). In contrast, DOCK8 mutant OT-I T cells in the same animals decreased to 5% of their peak number so that they now represented only 10% of the overall donor-derived OT-I CD8 T cells (Fig. 7 B, bottom). When memory T cells against the virus and the SIINFEKL peptide were restimulated by a second viral infection on day 35, both WT and DOCK8 mutant OT-I CD8 T cells increased in numbers relative to day 28. Despite this, the recall response by the mutant memory T cells was less so that they failed to reach the numbers of the first exposure and accounted for only 5% of the OT-I CD8 T cells at the height of the recall response. The difference between the normal and DOCK8-defective T cells could not be explained by preferential movement of mutant OT-I cells into the bone marrow or peritoneum, as comparable differences occurred in the spleen, bone marrow, and peritoneum (Fig. 7 B). The possibility that Dock8^pri/pri OT-I T cells were selectively rejected because of minor histoincompatibility, either as a result of the DOCK8 mutation, other linked mutations, or the CD45.1 marker, was ruled out by performing parallel adoptive transfers with CD45.1 OT-I control cells that were either Dock8^pri/+ heterozygous or WT (Fig. 7 C). The persistence and recall of both types of control OT-I T cells was equal to their cotransferred CD45.1/2 WT cells, whereas again there was greatly reduced persistence and recall of CD45.1 OT-I T cells that were homozygous for the Dock8 mutation.

Another possible explanation for the failure of DOCK8-defective CD8 T cells to persist after antigen-induced clonal expansion would be that they selectively formed short-lived effector cells but not cells with the phenotypic markers of memory CD8 T cells. To investigate this possibility, we examined the expression of several markers of CD8 T cell activation and memory cell formation. Surface expression of CD44, CD62L, CD127, and KLRG1 was indistinguishable between mutant and WT OT-I cells at all time points (Fig. 8 A), indicating that there was no measurable difference in the proportion of effector cells or absence of any phenotypically defined memory cell subsets to explain the failure of DOCK8 mutant cells to persist. Slightly lower expression of Bcl2 and Bim was observed in mutant OT-I cells on day 7 of the primary influenza response, but this did not reach statistical significance in one of two experiments (not depicted).

To define the role of DOCK8 in the persistence and recall of memory CD8 T cells over longer time periods, the cotransfer of OT-I T cells with normal or defective DOCK8 was repeated with a 6-mo delay between the primary H3N2 and secondary H1N1 SIINFEKL influenza exposure. To directly relate the data from primary, memory, and recall phases, a single cohort of recipient mice was followed longitudinally by measuring the transferred OT-I T cell frequency in their blood (Fig. 8 B). As before, there was only a subtle decrease in expansion of mutant OT-I T cells 7 d after the first viral exposure and a much more rapid decline of mutant T cells after this first antigen was cleared. However, by 195 d, the numbers of DOCK8-deficient OT-I T CD8 cells fell below the limit of reliable enumeration in all of the animals analyzed. In contrast, at the same time point, OT-I T cells with normal DOCK8 were still readily detected in the blood of three of the five recipient mice. In these animals, a large CD8 T cell recall response was made 7 d after reexposure to SIINFEKL influenza (day 202), with the numbers of OT-I T cells increasing 100-fold relative to day 195 immediately before reexposure and 10-fold relative to the peak of the primary response on day 7. Although DOCK8 mutant T cells also increased in these animals, they comprised <1% of the secondary response. Thus, DOCK8 mutation in T cells causes only a subtle decrease in the initial virus-induced clonal expansion of CD8 T cells in vivo but severely compromises the long-term persistence and recall of their memory CD8 progeny.

Our findings define a qualitative deficit in CD8 T cells caused directly by DOCK8 deficiency. Interpreting the consequences of DOCK8 mutation in CD8 T cells in DOCK8-deficient humans is complicated by their humoral immunodeficiency, varying degrees of lymphopenia, concurrent infections, and antibiotic and antiviral medication. Our analysis has identified an important confounding factor, namely that many of the circulating CD8 T cells correspond to a CD45RA⁺CCR7⁻ exhausted T_EMRA cell subset that is poorly responsive even in normal individuals. Furthermore, many of the CD45RA⁻ memory CD8⁺ T cells in these patients exhibited phenotypic and functional features of such exhausted cells. This phenotype is likely to reflect the repeated viral exposure faced by these patients, as has been observed in patients with other chronic infections such as late stage HIV (Chen et al., 2001; Brenchley et al., 2003) and patients with EBV-positive X-linked lymphoproliferative disease (Plunkett et al., 2005). We also found that DOCK8 deficiency in human T cells abrogated the robust proliferative response observed for normal CD4 and CD8 T cells after stimulation through the TCR and co-stimulatory molecules. This most likely reflects the advanced level of activation of these cells in vivo as a result of chronic pathogen exposure, which can cause extensive proliferation resulting in replicative senescence (Brenchley et al., 2003; Papagno et al., 2004). Indeed, a study of human CD8⁺ T cell subsets found that expression of CD57 correlates with impaired proliferation and increased susceptibility to apoptosis in vitro (Brenchley et al., 2003). Similarly, CD8⁺ T cells that are CD57⁺ (Brenchley et al., 2003), CD57⁺CCR7⁻CD27⁻ (Papagno et al., 2004), or CD45RA⁺CCR7⁻CD27⁻CD28⁻ (Rufer et al., 2003; Romero et al., 2007) have undergone the greatest levels of expansion in vivo, as determined by quantitation of TCR excision circles or telomere lengths, yet fail to proliferate in vitro in response to TCR-mediated stimulation. Furthermore, exogenous cytokines failed to restore the proliferative defect of CD57⁺CD8⁺ human T cells (Brenchley et al., 2003). Thus, the defects in proliferation of all DOCK8-deficient CD8⁺ T cell subsets is consistent with an enrichment in the frequency of CD57⁺CD27⁻CD28⁻ cells within both the memory and T_EMRA cell subsets, as well as unusually high expression of CD95 by naive cells.

When DOCK8 deficiency was confined to a subset of the CD8 T cell repertoire in specific pathogen-free mice, we were able to establish an essential, intrinsic role for DOCK8 in CD8 T cells that has its most dramatic effects on the persistence and recall of antigen-stimulated memory CD8 T cells. These effects, together with the effects of DOCK8 deficiency on LFA-1 polarization at the CD8 T cell synapse and on initial entry into cell division, extend and translate to human immunity a growing body of evidence that normal CD8 T cell memory depends on the quality of stimuli received through the T cell synapse before the first cell division.

DOCK8 was found not to be essential for thymic production of T cells, with normal numbers of single-positive CD8 and CD4 cells, no evidence for a defect in CD8 cell–positive selection in OT-I transgenic mice, and modest competitive decreases in all thymic subsets in 50:50 chimeras. Although we have not directly measured thymic egress, there was no consistent evidence for either accumulation or deficiency of mature CD69⁻CD62L⁺ single-positive T cells in the thymus as occurs in animals with thymic egress defects (Matloubian et al., 2004; Carlson et al., 2006; Shiow et al., 2008). DOCK8 played a greater role in peripheral CD8 T cells, with mutation of DOCK8 decreasing the number of naive T cells by 50%. The numbers of activated/memory CD44^hi CD8 T cells were nevertheless slightly increased, presumably because these were sustained by antigen-driven or homeostatic proliferation (Surh and Sprent, 2008). In competitive bone marrow chimeras, 95% of the peripheral CD8 T cells were derived from the normal rather than the mutant bone marrow, and there was no relative shift toward CD44^hi T cells among the DOCK8 mutant T cells. Cell transfers show that DOCK8-deficient naive CD8 T cells persist less well in peripheral lymphoid tissues and that influenza-activated OT-I T cells exhibited a persistence defect (70% decline in contribution to transferred OT-I cells over 21 d between days 7 and 28; Figs. 7 B and 8 B) that appears more severe than the survival defect exhibited by unstimulated OT-I cells bearing the mutation (50% decline in contribution to transferred OT-I cells over 39 d; Fig. 5 B). A subtle decrease in CD127 expression and IL-7–induced Bcl-2 was observed in DOCK8 mutant naive T cells and could theoretically explain the subtle decrease in their survival in vivo, but future studies will be needed to address how DOCK8 deficiency affects this pathway. The more dramatic decrease in memory CD8 T cell survival was not accompanied by any measurable decrease in CD127 expression, and only a marginal decrease in Bcl-2. Elucidating the molecular pathway by which DOCK8 promotes survival of naive and memory CD8 T cells and how this may relate to the synapse defect will be challenging given the many gaps in current understanding of the signals that induce long-lived memory T cells. Regardless of its molecular basis, this intrinsic defect in naive and memory T cell persistence coupled with uncontrolled viral infections provides an explanation for the markedly decreased frequency of circulating naive CD8 T cells and the dramatic shift to CD8⁺CD45RA⁺CCR7⁻ T_EMRA cells in DOCK8-deficient patients.

Although the experiments here establish that DOCK8 deficiency causes a striking qualitative deficit in CD8 T cells, it is unlikely that the CD8 T cell defect alone accounts for the peculiar spectrum of viral infections and tissue-tropism seen in DOCK8 deficiency. Bone marrow transplantation resolved the cutaneous viral infections in two DOCK8-deficient patients, indicating that hematopoietic rather than epidermal defects are responsible (Bittner et al., 2010; Gatz et al., 2011). Deficiency of the skin-homing subset of CD8 T cells is unlikely to explain the cutaneous infections because we find normal frequencies of skin-homing CD103⁺ CD8 T cells in the skin and gut of DOCK8 mutant mice (unpublished data), as well as in the circulation of DOCK8-deficient patients (unpublished data). It is significant that cutaneous infections with HSV, HPV, and M. contagiosum are much less prevalent in severe combined immunodeficiency or selective CD8 T cell deficiency, indicating that the CD8 T cell defects defined here are unlikely to be sufficient to explain the prevalence of these viral infections in DOCK8 deficiency (Laffort et al., 2004; Cerundolo and de la Salle, 2006; Notarangelo et al., 2009). Uncontrolled cutaneous infections with many of the same viruses are associated with evidence for greatly exaggerated Th2 cell activity in otherwise normal individuals with atopic dermatitis, asthma, and food allergy (Wollenberg et al., 2003; Beck et al., 2009). In this setting, Th2 cytokines are believed to suppress Th1 cells and IFN-γ production in the skin, thereby reducing control of these cutaneous viral infections by CD8 T cells and NK cells. Thus, it is attractive to speculate that in AR-HIES caused by DOCK8 deficiency, the effects of excessive Th2 cells interact with the qualitative defect in CD8 T cells to produce the severe HSV, M. contagiosum, and other skin viral infections. In contrast, autosomal dominant HIES (AD-HIES) caused by STAT3 mutations results in mucocutaneous infections with Candida albicans and cutaneous infection with Staphylococcus aureus rather than uncontrolled viral infections. AD-HIES is associated with defective Th17 cells (de Beaucoudrey et al., 2008; Ma et al., 2008; Milner et al., 2008; Renner et al., 2008) and perhaps other cell or cytokine defects, which result in this specific infectious susceptibility.

Patients with another primary immunodeficiency that causes qualitative T cell abnormalities, Wiskott–Aldrich syndrome (WAS), share some clinical characteristics with DOCK8 immunodeficiency patients in that they have eczema, a similar range of infections with M. contagiosum and papilloma and herpes viruses (Atherton et al., 2004), and elevated levels of IgE (Bosticardo et al., 2009). The WAS protein is important for T cell synapse formation (Dupré et al., 2002; Badour et al., 2003; Cannon and Burkhardt, 2004), and its expression in DCs is important for interactions between naive CD8 T cells and DCs (Pulecio et al., 2008). It had been previously assumed that the cutaneous infections in WAS were secondary to the eczema (Bosticardo et al., 2009), but the similarities between the clinical manifestations and cellular consequences of WAS protein and DOCK8 deficiency support a link between susceptibility to these viral infections and the presence of a qualitative CD8 T cell synapse defect. A similar spectrum of viral infections (papillomavirus and M. contagiosum) is also seen in epidermodysplasia verruciformis (Casanova and Abel, 2007). This primary immunodeficiency occurs because of mutations in EVER1 and EVER2 (Ramoz et al., 2002), disrupting zinc transport in keratinocytes, with increased zinc concentrations in the nucleus aiding papillomavirus replication (Lazarczyk et al., 2008). However, the molecular mechanism underlying susceptibility to a similar spectrum of infections in DOCK8 deficiency and EVER-1/2 deficiency is distinct because DOCK8 patients do not have intrinsic abnormalities of their keratinocytes as bone marrow transplantation has been associated with resolution of chronic viral infections (Bittner et al., 2010; Gatz et al., 2011).

Although the initial clonal expansion of DOCK8-deficient OT-I CD8 T cells in response to influenza-produced antigen was almost equal to their paired OT-I WT T cells, the persistence of mutant CD8 T cells after antigen was cleared was greatly diminished, and they exhibited poor secondary expansion upon reexposure to a CD8 T cell cross-reacting influenza strain. The preferential effect of DOCK8 mutation on the persistence and recall of CD8 memory T cells has parallels with several other studies (Williams et al., 2006b; Scholer et al., 2008; Teixeiro et al., 2009). Like DOCK8-deficient CD8 T cells, IL-2Ra–deficient CD8 T cells also show normal primary expansion upon viral challenge in vivo but are greatly decreased in their secondary clonal expansion upon reexposure to antigen (Williams et al., 2006b). However, Il2ra deficiency differs because it does not accelerate the loss of antigen-specific CD8 T cells during the contraction phase after the primary response. OT-I T cells bearing a mutation in the TCR β-chain transmembrane domain resemble DOCK8-deficient OT-I cells more closely (Teixeiro et al., 2009). The TCR-β mutation selectively interfered with clustering of the TCR and PKC-θ at the T cell synapse with antigen-presenting cells and the activation of NF-κB. When the TCR-β mutant OT-I T cells were adoptively transferred into mice infected with Listeria monocytogenes expressing ovalbumin antigen, the primary clonal expansion was comparable with normal OT-I T cells, but the mutant T cells subsequently declined more rapidly and made little recall expansion upon a second challenge (Teixeiro et al., 2009). The behavior of Dock8^pri/pri OT-I T cells most closely resembles normal CD8 OT-I T cells stimulated by ICAM-1–deficient DCs (Scholer et al., 2008). Using normal and Icam1^−/− mice, Scholer et al. (2008) adoptively transferred normal OT-I T cells and then immunized them with anti-DEC205 mAb coupled to ovalbumin with or without antagonistic anti-CD40 mAb. Upon immunization, the primary clonal expansion of OT-I T cells was comparable in the two groups, but during the contraction phase, most of the OT-I T cells were lost in the Icam1^−/− mice, and little recall clonal expansion occurred upon rechallenge. Two-photon microscopy revealed a failure to form long-lived contacts between T cells and Icam1^−/− DCs, consistent with other evidence that prolonged contact with antigen is important for the formation of CD8 memory T cells (Bachmann et al., 2006). Based on the findings here and studies of other DOCK proteins (Côté and Vuori, 2007), DOCK8 deficiency in OT-I T cells may interfere with the inside-out signals needed to activate LFA-1 binding to ICAM-1 on DCs and to promote actin filaments that polarize LFA-1/ICAM-1 at the synapse.

The similarities between the effects of DOCK8 deficiency, ICAM-1 deficiency, and TCR-β mutation provide three independent lines of experimental evidence connecting the quality of the TCR synapse and the intracellular signals it generates with the persistence of memory CD8 T cells. Given the delay to first cell division caused by DOCK8 deficiency in mice and the impaired proliferation of T cells from DOCK8-deficient patients, it is conceivable that qualitative differences in TCR signals are imprinted or stored in the T cell during the period leading up to the first cell division and then passed on silently through the rapid successive divisions to become apparent only when the cells exit cycle and need to persist as nondividing memory cells (Mercado et al., 2000; Kaech and Ahmed, 2001; van Stipdonk et al., 2001). Further analysis of DOCK8 deficiency in man and mouse will help to establish general principles for persistence of CD8 T cell–mediated immunity, to guide new strategies for enhancing this arm of immunity in chronic viral diseases and cancer and for suppressing CD8 T cell memory in transplantation.

Mice were housed in specific pathogen-free conditions, and all experiments were approved by the Australian National University Animal Ethics and Experimentation Committee. Dock8^pri/pri mice were generated as previously described (Randall et al., 2009). C57BL/6 and C57BL/6.SJL-CD45.1 mice were obtained from Australian National University Bioscience Services. OT-I mice on the C57BL/6 background (Hogquist et al., 1994) were provided by F. Carbone (The University of Melbourne, Parkville, Victoria, Australia) and were crossed with Dock8^pri/pri or Dock8^+/+ B6.CD45.1 mice to obtain OT-I transgenic offspring that were either Dock8^pri/pri CD45.1 homozygous or Dock8^+/+ CD45.1/CD45.2 heterozygous.

Bone marrow chimeras were generated as previously described (Randall et al., 2009). For adoptive transfers, C57BL/6 recipient mice were injected intravenously with 10³ FACS-sorted OT-I CD8 T cells (50:50 mixture of Dock8^pri/pri CD45.1 homozygous and Dock8^+/+ CD45.1/2 heterozygous CD8 T cells). For influenza challenge experiments, the mice were also injected i.p. with H3N2 (A/HKx31) followed later by H1N1 (A/PR8) strains of recombinant influenza A viruses expressing SIINFEKL OVA257-264 (Jenkins et al., 2006) obtained from S. Turner (The University of Melbourne).

Single cell suspensions of spleen, LN, and bone marrow were obtained as described previously (Randall et al., 2009). Data were acquired on a FACSCalibur (BD), FACS Sort (BD), or LSRII (BD) flow cytometer and analyzed using FlowJo software (Tree Star). The following antibodies were used for surface staining of lymphocytes for flow cytometry (from BD unless otherwise stated; clones indicated): FITC-conjugated anti-CD4 (GK1.5), peridinin chlorophyll protein (PerCP)–cyanine (Cy) 5.5–conjugated anti-CD4 (RM4.5), PE-indotricarbocyanine (PE-Cy7)–conjugated anti-CD4 (RM 4.5), allophycocyanin (APC)-conjugated anti-CD4 (RM 4.5), APC-Cy7–conjugated anti-CD4 (GK1.5; BioLegend), PE-conjugated anti-CD4 (GK1.5), PE-conjugated anti-CD8a (53–6.7), PerCPCy5.5-conjuated anti-CD8a (53-6.7), Alexa Fluor 405–conjugated anti-CD8a (5H10; Invitrogen), APC–Alexa Fluor 750–conjugated anti-CD8a (53-6.7, 5H10), PE-Cy7–conjugated anti-CD8a (53-6.7; BioLegend), APC-conjugated anti-CD25 (PC61), APC-Cy7–conjugated anti-CD25 (PC61; BioLegend), PE-conjugated anti-CD27 (LG.3A10), APC-conjugated anti-CD44 (1M7), Alexa Fluor 405–conjugated anti-CD44 (1M7; Invitrogen), Pacific blue–conjugated anti-CD44 (IM7; BioLegend), PE-conjugated anti-CD45.1 (A20), Alexa Fluor 700–conjugated anti-CD45.1 (A20; BioLegend), Pacific blue–conjugated anti-CD45.2 (104; BioLegend), FITC-conjugated anti-CD45.2 (104), APC-conjugated anti-CD45.2 (104; eBioscience), FITC-conjugated anti-CD62L (MEL-14), APC-Cy7–conjugated anti-CD62L (MEL-4; BioLegend), PerCPCy5.5-conjugated anti-CD69 (H1.2F3), FITC-conjugated anti-CD122 (5H4), PE-indodicarbocyanine (PE-Cy5)–conjugated anti-CD127 (A7R34; eBioscience), PE-conjugated anti-CD152 (UC10-4F10-11), FITC-conjugated anti-B220 (RA3-6B2), APC–Alexa Fluor 750–conjugated anti-B220 (RA3-6B2; Invitrogen), APC-conjugated anti-KLRG1 (2F1; eBioscience), PE-conjugated anti-TCR Vα2 (B20.1), FITC-conjugated anti-CD45RB (16A), PE-conjugated anti-PD-1 (J43), and Alexa Fluor 700–conjugated CD3 (17A2; eBioscience). Cell viability was determined with 7 aminoactinomycin D (Invitrogen) and FITC-conjugated anti–Annexin V. When staining using Annexin V, Annexin V binding buffer (BD) was used. For FACS sorting, splenocytes were stained with PE-conjugated anti-CD4 and FITC-conjugated anti-B220 as described above, and the negatively selected populations were harvested for the adoptive cell transfers. The following antibodies were used for intracytoplasmic staining of lymphocytes: PerCP-conjugated anti-tbet (eBio4B10; eBioscience), FITC-conjugated Bcl-2 (3F11), and Dylight-conjugated anti-Blimp (3H2-E8). Anti-Bim (clone 3C5; a gift from A. Strasser, Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia) was conjugated to Alexa Fluor 647 using a Protein Labeling kit (A20173; Invitrogen).

Splenic WT and pri/pri OT-I T cells were isolated using a MACS column negative selection kit (Miltenyi Biotech) according to the manufacturer’s protocol. Naive CD8 T cells were labeled with a final concentration of 5 µM CFSE as per Quah et al. (2007). Bone marrow–derived DCs were derived as described previously (Oliaro et al., 2010) and seeded into 8-well chamber slides at 3 × 10⁴ cells/well. They were then pulsed with SIINFEKL peptide for 1 h and washed twice before the addition of 2 × 10⁴ CFSE-labeled T cells/well. Cells were harvested at the indicated time points.

For immunofluorescence staining, 2 × 10⁴ naive splenic WT and pri/pri OT-I T cells/well were co-cultured with 3 × 10⁴ SIINFEKL-pulsed bone marrow–derived DCs/well in chamber slides (Oliaro et al., 2010). The cell conjugates formed were fixed at 1 h after incubation with fixative solution (100 mM Pipes, 5 mM MgSO₄, 10 mM EGTA, 2 mM DTT, and 3.7% [wt/vol] paraformaldehyde, pH 7.0) for 10 min. The cells were then washed twice and incubated in permeabilization solution (PBS, 0.1% Triton X-100, and 0.5% BSA [wt/vol]) for 5 min before being washed and kept in blocking agent (PBS and 1% BSA [wt/vol]) overnight at 4°C. Primary antibodies against the proteins of interest (anti–LFA-1 [BD], anti–PKC-θ [Santa Cruz Biotechnology, Inc.], and rhodamine-conjugated actin/phalloidin [Invitrogen]), together with mouse α-tubulin (Rocklands) or rabbit α-tubulin (Sigma-Aldrich), were added and incubated for 45 min at 4°C in the dark. Secondary antibodies (anti–rat and anti–rabbit Alexa Fluor 488 and anti–mouse and anti–rabbit Alexa Fluor 546) were added after washing and incubated for another 45 min at 4°C in the dark. The cells were then washed and mounted in ProLong Gold antifade reagent with DAPI (Invitrogen). Cells were imaged using a confocal microscope (Fluoview FV1000; Olympus) equipped with a 12.9-mW 488-nm multi-ion argon laser, a 1-mW 543-nm multi-ion green HeNe laser, and an 11-mW 633-nm red HeNe laser. All images were captured using a PlanApoN 60× oil immersion objective (NA = 1.42). The images were subsequently processed using the Olympus Micro FV10-ASW program and displayed as z-stack projections (sections = 0.5 µm). To acquire images in which an immune synapse was formed, conjugates with the MTOC (as marked by α-tubulin) of T cells polarized to the interface with the DCs were chosen. Blinded scoring of the images was used as a means for quantifying the localization of proteins of interest. An integer score was given from 3 (very polarized to the DC–T cell interface), to 0 (not polarized), to −3 (very polarized away from the interface).

Splenocytes from C57BL/6 mice were pulsed with 1 µM SIINFEKL peptide for 2 h at room temperature before irradiation at 6 Gy for 30 min. Splenocytes from WT or pri/pri OT-I mice were added to the irradiated C57BL/6 splenocytes with the addition of 50 U/ml IL-2 and cultured for 5 d. DCs labeled with 100 µCi Cr⁵¹ were pulsed with SIINFEKL peptide for 1 h at 37°C. Activated WT and pri/pri OT-I splenocytes were added to the DCs at the indicated ratios and incubated at 37°C for 4 h. Supernatant was collected and analyzed for chromium release by an automatic gamma counter (Wallac Wizard 1470; PerkinElmer).

Spleen cells from control and mutant mice were isolated, red cells were lysed, and the cells were counted by trypan blue exclusion. Three million cells (per milliliter in a 24-well plate) from each animal were cultured in the presence or absence of 25 ng/ml IL-7 (R&D Systems) with the complete tissue culture medium at 37°C and 5% CO₂ for the indicated time points. Cells were stained as in Flow cytometry of murine cells.

Peripheral blood was collected from normal healthy donors (Australian Red Cross Blood Service) and DOCK8-deficient patients (Table I). All human experiments were approved by the Sydney South West Area Health Service and St. Vincent’s Hospital Darlinghurst Human Research Ethics Committees, as well as by the institutional review boards of the Necker Medical School, the Rockefeller University, and the National Institute of Allergy and Infectious Diseases (National Institutes of Health). For analysis, the following mAbs were used: FITC-conjugated anti–human CCR7 (150503; R&D systems), PerCP-Cy5.5–conjugated anti–human CD45RA (HI100; eBioscience), PE-Cy7–conjugated anti-CD4 (SK3; BD), Pacific blue–conjugated anti–human CD8 (RPA-T8; BD), eFluor450-conjugated anti–human CD127 (eBioRDR5; eBioscience), PE-conjugated anti–human CD244 (2B4; Beckman Coulter, c1.7), PE-conjugated anti–human CD57 (HCD57; BioLegend), APC-human–conjugated anti-CX3CR1 (2A9-1; BioLegend), APC-conjugated anti–human CD27 (O323; eBioscience), PE-conjugated anti–human CD95 (DX2; BD), APC-conjugated anti–human CD11a (HI111; BD), PE-conjugated anti–human CD11b (D12; BD), APC-conjugated anti–human CD28 (CD28.2; BioLegend), PE-conjugated anti–human perforin (dG9; eBioscience), and APC-conjugated anti–human Granzyme B (GB11; Invitrogen). CD4 and CD8 T cells were isolated from freeze thawed PBMCs using Dynal beads (Invitrogen) as per the manufacturer’s instructions. CD4⁺ T cells were labeled with anti-CCR7 and anti-CD45RA mAbs and then sorted into naive and memory populations using a FACSAria (BD). Naive and memory CD4 T cells were isolated as CD45RA⁺CCR7⁺ and CD45RA⁻CCR7^+/− cells, respectively, whereas naive and different central and effector memory CD8 T cells were isolated by sorting CD45RA⁺CCR7⁺, CD45RA⁻CCR7^−/+, and CD45RA⁺CCR7⁻ cells. CD4 and CD8 naive and memory T cell subsets were labeled with CFSE and cultured with TAE beads (one bead for every two cells; Miltenyi Biotech) with or without 20 U/ml recombinant human IL-2 (Millipore) for 5 d. After this time, proliferation was determined by measuring CFSE dilution.

Statistical analyses were performed using Prism software (GraphPad Software) or Excel (Microsoft). Statistical comparison was performed with unpaired Student’s t tests and χ² analysis as noted in the text.

Fig. S1 shows extended phenotypic analysis of human DOCK8-deficient CD8⁺ T cells. Fig. S2 shows T cell counts in blood and LNs for Dock8^pri/pri and C57BL/6. Fig. S3 shows that there is no evidence of an exhausted CD8 T cell phenotype in Dock8^pri/pri and Dock8^cpm/cpm mice. Fig. S4 compares T cell numbers in Dock8^pri/pri, Dock8^cpm/cpm, and C57BL/6 mice. Fig. S5 shows further analysis of thymic subsets in Dock8^pri/pri mice. Fig. S6 quantifies CD127 on the surface of T cell subsets from Dock8^pri/pri and C57BL/6 mice. Fig. S7 shows survival of Dock8^pri/pri and C57BL/6 T cells in culture with and without the addition of IL-7. Fig. S8 shows a chromium release assay using Dock8^pri/pri OT-I T cells.

We wish to thank D. Howard, H. Burke, and J. Wang (Australian National University [ANU], Canberra, Australian Capital Territory, Australia) for expert technical assistance, the staff at the ANU Bioscience Division for animal husbandry, the Australian Phenomics Facility genotyping and mapping team for genetic analysis, the Garvan Institute Flow Facility for cell sorting, F. Carbone for OT-I mice, S. Turner for recombinant influenza constructs, A. Strasser for Bim antibody, and Dr. H. Su (National Institutes of Health, Bethesda, MD) for genotyping patient #6.

This work was supported by contract U54 AI054523 from the National Institutes of Health (to C.C. Goodnow and E.M. Bertram), The Wellcome Trust (C.C. Goodnow and R.J. Cornall), a National Health and Medical Research Council (NHMRC) Program Grant (to C.C. Goodnow and S.G. Tangye), and an NHMRC Project Grant (to E.M. Bertram). C.C. Goodnow was supported by an Australian Research Council Federation Fellowship and is an NHMRC Australia Fellow. C.S. Ma was supported by a Peter Doherty Research Fellowship and is the recipient of a Career Development Award from the NHMRC. S.G. Tangye is a Senior Research Fellow of the NHMRC, and K.L. Randall was an NHMRC Medical Postgraduate Scholar.

The authors declare that they have no competing financial interests.