This study uses two independent genetic strategies to explore the requirement for phosphoinositide-dependent kinase-1 (PDK1) in the development of mature T cell populations from CD4/CD8 double-positive thymocytes. The data show that CD4/CD8 double-positive thymocytes that do not express PDK1 or express a catalytically inactive PDK1 mutant fail to produce mature invariant Vα14 NKT cells but can differentiate to conventional CD4, CD8, or regulatory T cell subsets in the thymus. The PDK1 requirement for Vα14 NKT cell development reflects that these cells require the PDK1 substrate protein kinase B to meet the metabolic demands for proliferative expansion in response to IL-15 or AgR stimulation. There is also constitutive PDK1 signaling in conventional α/β T cells that is not required for lineage commitment of these cells but fine-tunes the expression of coreceptors and adhesion molecules. Also, although PDK1 is dispensable for thymic development of conventional α/β T cells, peripheral cells are reduced substantially. This reflects a PDK1 requirement for lymphopenia-induced proliferation, a process necessary for initial population of the peripheral T cell niche in neonatal mice. PDK1 is thus indispensable for T cell developmental programs, but the timing of the PDK1 requirement is unique to different T cell subpopulations.
T cell development in the thymus is a process that aims to populate the peripheral immune system with nonautoreactive T lymphocytes that can control adaptive immune responses to pathogens. T cells produced by the thymus include conventional α/β T cells expressing T cell AgR complexes composed of highly variable TCRα and β subunits that recognize peptide–MHC complexes. The different developmental stages of thymocytes can be identified by the expression of CD8 and CD4, the receptors for MHC class I and class II molecules, respectively. T cell progenitors are CD4/CD8 double-negative (DN) and progress to a CD4/CD8 double-positive (DP) stage if they successfully rearrange their TCRβ locus and express a pre-TCR complex. In the CD4/CD8 DP compartment, cells undergo TCRα-chain rearrangements, and if they express a functional α/β TCR complex, they are then subjected to selection processes that generate CD4 or CD8 single-positive (SP) cells that can populate the periphery. CD4-positive T cells can be further subdivided into conventional CD4 T cells, regulatory T cells (Tregs) (1), and NKT cells (2). A significant proportion of NKT cells express an invariant Vα14 TCR that recognizes glycolipid–CD1d Ag complexes [invariant Vα14 NKT (iNKT) cells] and play a role in immune surveillance and immune homeostasis (2, 3).
The balanced production of peripheral T cell subpopulations is essential to ensure the function and the homeostasis of the peripheral immune system. Accordingly, one key issue for T cell developmental biology is the nature of the signals that determine why CD4/CD8 DP thymocytes can produce different T cell subpopulations each with unique functions. One insight is that the strength and/or duration of signaling plays a key role in lineage commitment. For example, it has been suggested that Tregs may derive from thymocytes selected to express α/β TCRs with relatively high affinity for self-peptide–MHC complexes (4). The commitment of DPs to either the CD4 or CD8 lineage is also linked to signal strength in the sense that persistent TCR signaling drives cells to the CD4 lineage (5). Other important insights come from the differential requirements for transcriptional factors for the different peripheral T cell lineages. For example, ThPOK and Runx transcription factors cross-regulate each other in a balancing act that controls T cell commitment to either CD4 helper or CD8 cytotoxic T cell lineages (6, 7) whereas the transcription factor Foxp3 is critical for Treg differentiation (1). It is also evident that there are distinct signaling requirements for the differentiation of iNKT cells. For example, DP thymocytes lacking expression of Fyn, SAP, T-bet, NF-κB, and c-myc fail to produce iNKT cells, despite normal development of conventional α/β T cells (8–15).
One explanation for the apparently unique molecular requirements for iNKT cell development could be that these cells undergo a large proliferative burst as they differentiate from DPs (14). This is in contrast to conventional α/β T cells where a major proliferative burst occurs as TCRβ-chain–selected cells transit from DNs to DPs (16). In DN T cell progenitors, a key signaling pathway that supports the biosynthetic demands of rapid cell division is mediated by phosphoinositide-dependent kinase l (PDK1) (16, 17). This serine/threonine kinase phosphorylates and activates the PI3K-controlled serine/threonine kinase protein kinase B (PKB), also called Akt (18). PDK1 also phosphorylates and activates the 70-kDa ribosomal protein S6 kinase-1 (S6K1) and the 90-kDa ribosomal protein S6 kinase (RSK) and members of the protein kinase C family (18). T cell progenitors lacking PDK1 or PKB arrest at the CD4/CD8 DN stage of thymocyte development because they cannot increase their metabolism to support the pre–TCR-induced burst of proliferation required for the DN to DP transition of conventional T cells (16, 17, 19–21). Is PDK1 similarly required for the proliferative burst required for iNKT cell development at the DP to SP stage? In this context, it has been reported that deletion of PDK1 in CD4/CD8 DP thymocytes did not prevent the formation of CD4 SP thymocytes (22). The caveat was that the CD4 SP cells that developed in the absence of PDK1 did not activate in response to AgR/CD28 costimulation (22). It was also reported that the loss of PDK1 in DPs caused fewer CD8 SP thymocytes to develop (22). However, the impact of PDK1 loss on the production of different subsets of α/β T cells, such as Tregs or iNKT cells, has not been explored. This is pertinent because it has been described recently that PKB signaling pathways control the differentiation of peripheral effector T cells following immune activation (23–25).
Accordingly, the object of the present study was to analyze the role of PDK1 in the DP to SP transition of thymocytes and to compare the PDK1 requirements for the development of T cell subpopulations. The data show that PDK1 is indispensable for T cell developmental programs, but the timing of the PDK1 requirement, either thymic or peripheral, is unique to different T cell subpopulations.
Materials and Methods
PDK1fl/fl (26), CD4-Cre (27), and PDK1K111A/WT mice were bred and maintained in the University of Dundee (Dundee, U.K.) in compliance with U.K. Home Office Animals (Scientific Procedures) Act 1986 guidelines. PDK1fl/fl CD4-Cre mice were generated crossing mice with floxed PDK1 alleles and mice expressing Cre recombinase under the control of the proximal CD4 promoter (CD4-Cre) (27). PDK1K111A/fl CD4-Cre mice were generated by backcrossing PDK1fl/fl CD4-Cre mice with PDK1K111A/WT mice.
Generation of catalytically inactive PDK1 knockin mice
A targeting construct was generated as previously described (28) to replace the wild-type (WT) exon 4 of the gene encoding PDK1 with a mutant form of exon 4 encoding for alanine in the position of Lys111 (K111A). A neomycin resistance “cassette” flanked with Cre recombinase-removable loxP sites was present in the targeting construct to permit selection of targeted embryonic stem cells. The targeting vector was introduced into E14 mouse embryonic stem cells by electroporation and embryonic stem cell clones generated. Targeted embryonic stem cells were identified by Southern blot analysis using probes external to the targeting vector. The neomycin cassette was removed from the targeted embryonic stem cells by transient expression of Cre recombinase. Heterozygous PDK1K111A/WT knockin embryonic stem cell clones were microinjected into C57BL/6 × BALB/c blastocysts, which were then reimplanted into recipient female mice. Chimeric mice that had a high degree of embryonic stem cell contribution were identified by coat color and were then crossed to BALB/c mice. This allowed germline transmission of the knockin allele to be identified by the gray coat of the resulting pups. Genotyping for the PDK1K111A knockin allele was carried out by PCR of genomic DNA isolated from ear clips using the same methodology and primers used to genotype PDK1155E/155E knockin mice described previously (28).
Flow cytometric analysis
Abs conjugated to FITC, PE, allophycocyanin, and biotin were obtained from either BD Pharmingen (San Diego, CA) or eBioscience (San Diego, CA). TriColor, PE-Cy5.5, and allophycocyanin Alexa Fluor 750-conjugated Abs were obtained from Caltag Laboratories (Burlingame, CA). Cells were stained for surface expression of the following markers using the Abs in parentheses: CD4 (RM4-5), CD8 (53-6.7), CD24 (M1/69), CD25 (7D4), CD44 (IM7), L-selectin (also called CD62L) MEL-14), Thy1.2 (53-2.1), TCRβ (H57-597), CD122 (TM-β1), and CD16/CD32 (2.4G2) (Fc block). Intracellular Foxp3 staining was performed using a Foxp3 intracellular staining kit (eBioscience). Cells were stained with saturating concentrations of Ab in accordance with the manufacturer’s instructions. For CCR7 staining, cells were labeled with mouse rCCL19-Fcγ and detected using PE-conjugated anti-human Fcγ (both from eBioscience). For Vα14 TCR staining, CD1d-DimerX (BD Pharmingen) was loaded with the α-galactoceromide (αGalCer) synthetic analog KRN7000 (Enzo Life Sciences, Exeter, U.K.): 0.14 μg αGalCer/μg CD1d-DimerX at 37°C overnight (CD1d-αGalCer). Cells were labeled with this CD1d-αGalCer and detected with allophycocyanin-conjugated rat anti-mouse IgG1 (BD Pharmingen). For NKT cell MACS enrichment, cells were labeled with CD1d-αGalCer and IgG1-allophycocyanin, as described above, then bound to anti-allophycocyanin magnetic beads and enriched on an autoMACS cell separator (Miltenyi Biotec, Auburn, CA). For intracellular flow cytometry of promyelocytic leukemia zinc finger (PLZF), thymocytes were fixed and permeabilized using the fixation/permeabilization set from eBioscience. Intracellular PLZF was detected using 4 μg/ml mouse mAb D-9 (Santa Cruz Biotechnology, Santa Cruz, CA) and PE- or allophycocyanin-conjugated rat anti-mouse IgG1 (BD Pharmingen).
Thymocytes were lysed on ice in lysis buffer [10 mM Tris (pH 7.05), 50 mM NaCl, 0.5% Triton X-100, 50 mM sodium fluoride, 30 mM sodium pyrophosphate, 1 mM DTT, 50 nM calyculin A, 5 μM ZnCl2, 10% glycerol, and complete protease inhibitors (Roche, Basel, Switzerland)]. Protein was separated by SDS-PAGE, transferred to nitrocellulose membrane, and detected by western blot analysis using standard techniques. Phospho-RSK (Ser227) and total RSK Abs were purchased from Santa Cruz Biotechnology, total PDK1 from Millipore (Bedford, MA), and all other Abs from Cell Signaling Technology (Beverly, MA).
PDK1 kinase assay
E11.5-12.5 embryo cell lysates (500 μg) were used to immunoprecipitate and assay PDK1 kinase activity. The lysates were incubated at 4°C for 1 h on a shaking platform with 10 μg Ab coupled to 10 μl protein G-Sepharose (Amersham Biosciences, Piscataway, NJ). The immunoprecipitates were washed twice with 1 ml lysis buffer and once with 1 ml buffer A [50 mM Tris (pH7.5), 0.1% 2-ME, and 0.1 mM EGTA]. The standard assay (50 μl) contained: washed protein G-Sepharose immunoprecipitate, 50 mM Tris/HCl (pH 7.7), 0.1 mM EGTA, 0.1% 2-ME, 2.5 μM PKI (TTYADFIASGRTGRRNAIHD, peptide inhibitor of cAMP-dependent protein kinase), 10 mM magnesium acetate, 0.1 mM [γ-32P]ATP (∼200 cpm/pmol), and 1 mM peptide substrate (KTFCGTPEYLAPEVRR). One mUnit of activity was that amount of enzyme that catalyzed the phosphorylation of 1 pmol substrate in 1 min.
Calcium flux analysis
Thymocytes from WT and Pdkfl/fl CD4-Cre mice were isolated and loaded with 4.5 μM indo-1 in the presence of 0.045% (v/v) pluronic F-127 (Molecular Probes, Eugene, OR) for 45 min at 37°C. Cells labeled with Abs (CD3-biotin, CD4-PE, CD8-FITC) on ice for 30 min. After staining, cells were place on ice (1 × 106 cells/ml). Cells were warmed to 37°C for 5 min immediately before analysis, stimulated with avidin (20 μg/ml), then subjected to flow cytometric analysis.
Thymocytes from WT (CD45.1) and Pdkfl/fl CD4-Cre (CD45.2) mice were mixed at a 1:1 ratio and injected into the tail vein of Rag2−/− mice. After 4 wk, mice were sacrificed and analyzed for the presence of WT and Pdkfl/fl CD4-Cre T cells.
Thymocytes were depleted of CD8+ cells by magnetic cell separation (Miltenyi Biotec). Cells were stained with 10 μM CFSE at 37°C for 20 min in RPMI 1640 containing l-glutamine (Invitrogen), 50 μM 2-ME (Sigma-Aldrich), and penicillin-streptomycin (Life Technologies, Rockville, MD). Cells were then cultured in RPMI 1640 containing l-glutamine, 10% (v/v) heat-inactivated FBS (Life Technologies), 50 μM 2-ME, 10 mM HEPES buffer and penicillin-streptomycin for 4 d. Cell were cultured; ± IL-15 (10–100 ng/ml), with CD1d (1 μg/106 cells), with or without αGalCer, and with or without AKT-1/2i (1 μM) (Calbiochem, San Diego, CA). Cell proliferation was assessed by flow cytometric analysis of CD1d-αGalCer+/TCRβmid NKT cells or CD1d-αGalCer−/TCRβhigh CD4SP T cells as dilution of CFSE staining.
iNKT cell culture
iNKT cells were purified from WT thymocytes by magnetic separation using streptavidin-coated magnetic beads (Miltenyi Biotec) following labeling with αGalCer-loaded CD1d-DiamerX (BD Biosciences) bound to biotinylated rat anti-mouse IgG1 (BD Biosciences). iNKT cells were cultured in RPMI 1640 containing l-glutamine, 10% (v/v) heat-inactivated FBS (Life Technologies), 50 μM 2-ME, 10 mM HEPES buffer, and penicillin-streptomycin for up to 3 wk with IL-2 (0.25 μg/ml). Cells were starved of IL-2 for 24 h before stimulation with IL-15 or CD1d-αGalCer.
PDK1 is dispensable for the DP/SP transition of conventional α/β T cells
To explore the role of PDK1 as thymocytes transit the DP/SP stage of thymocyte development, we backcrossed mice with floxed alleles of PDK1 to a mouse expressing cre recombinase under the control of the CD4 promoter CD4-cre. Thymocytes were present at normal numbers in Pdk1fl/fl CD4-cre mice and comprised both CD4/CD8 DP and CD4 or CD8 SP subpopulations (Fig. 1A). The deletion of PDK1 alleles in Pdk1fl/fl CD4-cre DPs was confirmed by PCR analysis (data not shown), and the loss of PDK1 protein was confirmed by Western blot analysis of purified Pdk1fl/fl CD4-cre DPs (Fig. 1B). Analysis of purified SP thymocytes also revealed that Pdk1fl/fl CD4-cre SPs cells have deleted PDK1 (Fig. 1C). There is a basal level of PDK1 signaling in WT SP thymocytes evident through the high basal phosphorylation of RSK on the PDK1 substrate sequence (Ser227) (Fig. 1D). The phosphorylation of RSK Ser227 is lost in Pdk1fl/fl CD4-cre SP thymocytes (Fig. 1D).
Mature SP thymocytes express high surface levels of α/β TCR complexes, and Fig. 1E shows that the frequency of TCRβhigh SP thymocytes is normal in Pdk1fl/fl CD4-cre mice. It was, however, evident from the CD4/CD8 staining profiles that the CD8 SPs in the Pdk1fl/fl CD4-cre thymus have reduced surface expression of CD8 compared with WT CD8 SP thymocytes (Fig. 1A). This phenotype of CD4/CD8 expression has been described previously and interpreted to mean that Pdk1fl/fl CD4-cre mice have reduced number of CD8 SP thymocytes (22). However, the level of CD8 expression on SPs is upregulated following positive selection during a process termed “coreceptor tuning” (29). Accordingly, it is conceivable that Pdk1fl/fl CD4-cre thymi do contain normal numbers of CD8 SP cells, but these cells express low levels of CD8. Hence, a more accurate way to quantify CD8 SP thymocytes and in particular to distinguish these cells from the CD8-positive thymocytes that are intermediate between DNs and DPs is to first identify cells expressing high levels of TCRβ subunits and then to quantify expression of either CD4 or CD8 on the TCRβhigh cells. Using this analysis, the data show that CD4 and CD8 mature SP populations are present in Pdk1fl/fl CD4-cre thymi at equivalent frequency to WT thymi (Fig. 1F, 1G). However, although the frequency of CD8 SP thymocytes was normal in Pdk1fl/fl CD4-cre mice, the level of CD8 expressed on the PDK1-null TCRβhigh cells was reduced ∼2-fold (Fig. 1H). Analysis of the expression of CD122, CD24 (also called heat-stable Ag), and CD8αα confirmed that these cells are conventional CD8 SP thymocytes rather than unconventional innate-like CD8+ T cells that normally develop at very low frequency in the thymus (data not shown) (30, 31). Therefore, CD4 and CD8 SP thymocyte numbers are normal in the Pdk1fl/fl CD4-cre mouse, but PDK1 regulates the quantity of CD8 expressed on SPs.
Do PDK1-null SP thymocytes mature normally? First, we assessed whether mature PDK1-null SP thymocytes expressed a functional α/β TCR complex by determining whether the cells could respond normally to TCR triggering and initiate downstream signaling. In this context, Fig. 2A shows that Pdk1fl/fl CD4-cre SP thymocytes can normally increase intracellular Ca2+ concentrations in response to cross-linking TCR complexes. We also examined PDK1-null SP thymocytes for the expression of CD24, a cell surface molecule that is downregulated as SP thymocytes mature in the thymus. In this respect, the frequency of CD24 low mature CD4 and CD8 SP thymocytes in Pdk1fl/fl CD4-cre mice was normal (Fig. 2B).
One other defining phenotype of mature SP thymocytes is that they express high levels of the adhesion molecule CD62L and the chemokine receptor CCR7. We have shown previously that strong activation of PDK1 signaling in T cells downregulates expression of CD62L and CCR7 and upregulates the expression of the adhesion molecule CD44 (32). PDK1 is a constitutively active kinase, but it is not known whether the basal levels of PDK1 activity found in SPs (Fig. 1D) play any role in determining the basal levels of CD62L, CCR7, and CD44 expressed by T cells. Fig. 2C compares the CD62L, CCR7, and CD44 staining profiles of WT and Pdk1fl/fl CD4-cre SP thymocytes. The data show that loss of PDK1 causes increased expression of CD62L and CCR7 in SP thymocytes but causes loss of CD44 expression. Collectively, these data argue that PDK1 is not required for the DP to SP transition of conventional α/β T cells. However, the basal levels of PDK1 activity control the repertoire of adhesion molecules expressed by mature SP thymocytes.
CD4+ SP thymocytes include populations of Tregs that are characterized by the expression of the transcription factor Foxp3 and the expression of the IL-2Rα subunit CD25. Fig. 2D shows CD25 and Foxp3 staining of CD4 SP thymocytes and reveals that Treg frequency in the thymus of Pdk1fl/fl CD4-cre mice is similar to that of WT mice.
PDK1 is required for the development of iNKT cells
Another important CD4+ thymocyte subpopulation is the NKT cell subset that has an invariant Vα14 TCR that recognizes glycolipid Ags presented by the MHC-like molecule CD1d. iNKT cells are usually present at a low frequency in the thymus (∼0.4–0.6% of total thymocytes) but can be distinguished from conventional thymocytes because they can bind CD1d molecules loaded with the iNKT cell Ag αGalCer (33). Other surface markers can be used to distinguish the different stages of iNKT cell development such as CD24, which is expressed on the earliest iNKT cells progenitors (stage 0) but not thereafter (stages 1–3) (34). Accordingly, WT and Pdk1fl/fl CD4-cre thymocytes were stained with CD1d-αGalCer and CD24. An initial analysis of total thymocytes (Fig. 3A) indicated a substantial defect in the frequency of stage 1–3 iNKT cells (CD1d-αGalCer+/CD24−) in Pdk1fl/fl CD4-cre thymocytes. The earliest iNKT cell subset (stage 0 iNKT cells) is present at a very low frequency in the thymus and arises following positive selection of Vα14 TCR expressing DP iNKT progenitors. Rearrangement of the TCRα-chain locus to generate the Vα14-Jα18 TCRα-chain gene segment is crucial to the generation of iNKT cells, and this process is successfully completed in Pdk1fl/fl CD4-cre CD69− DP (preselection) thymocytes (Supplemental Fig. 1). Stage 0 iNKT progenitors express the Vα14 TCR complex that can bind CD1d–αGalCer complexes and can be characterized as NK1.1−/CD44−/CD24+, and unlike conventional α/β thymocytes, they express the PLZF protein transcription factor (35, 36). Detailed analysis using these key markers revealed normal numbers of stage 0 iNKT cells in the Pdk1fl/fl CD4-cre thymus (Fig. 3B), indicating that initial positive selection of iNKT progenitors proceeds normally in the absence of PDK1. Further analysis quantified the numbers and frequencies of iNKT cells at each of the developmental stages 1–3 (Fig. 3C, 3D). These data show that NK1.1−/CD44− stage 1 cells were reduced 4-fold in numbers with a more severe reduction in numbers of stage 2 and stage 3 iNKT cells (Fig. 3C). Fig. 3E shows that PDK1-null stage 1 iNKT express high levels of PLZF, confirming that these are indeed stage 1 iNKT cells.
During stage 1, iNKT progenitors undergo rapid growth and proliferation, thus implicating PDK1 in regulating these processes. Analysis of WT iNKT progenitors confirms that as they transit through stage 1 they increase in size (Fig. 3F) and reveals that during stage 1 iNKT cells upregulate the expression of important nutrient receptors: the iron transporting transferin receptor (CD71) and CD98, a key component of the l-amino acid transporter (Fig. 3F). However, in the absence of PDK1, stage 1 iNKT cells fail to upregulate the expression of CD71 and CD98 and remain small cells (Fig. 3G). Collectively, the data show that Pdk1fl/fl CD4-cre thymocytes have a block in iNKT cell development as the cells transit stage 1. The results argue that PDK1-null cells cannot meet the metabolic demands required for growth and proliferation.
Analysis of peripheral lymphocyte subpopulations in Pdk1fl/fl CD4-cre mice
iNKT cells exit the thymus, and in the periphery, they are found at high frequency in the liver. Fig. 4A, top panels, shows that there was a distinct population of iNKT cells in the liver from WT mice, but no clear population was discernable in the Pdk1fl/fl CD4-cre liver. The livers of Pdk1fl/fl CD4-cre mice did, however, contain conventional NK cells at normal frequencies (Fig. 4A, bottom panels). In fact, iNKT cells were largely absent from all Pdk1fl/fl CD4-cre peripheral tissues (liver, spleen, and blood) compared with WT (Fig. 4B). These data confirm the block in the development of iNKT cells in Pdk1fl/fl CD4-cre mice. The data presented in Fig. 1 show that Pdk1fl/fl CD4-cre thymi contain normal numbers of conventional α/β SP thymocytes. However, the livers of Pdk1fl/fl CD4-cre mice had not only lost iNKT cells but also had reduced levels of conventional α/β T cells (Fig. 4A, 4C). There were similarly reduced numbers of conventional α/β T cells in the spleen, blood, and lymph nodes of Pdk1fl/fl CD4-cre mice (Fig. 4C). Hence, although the intrathymic development of conventional α/β T cells is normal in Pdk1fl/fl CD4-cre mice, these PDK1-null α/β cells fail to effectively populate peripheral lymphoid tissues. A defect in T cell egress from the thymus, peripheral survival, or homeostatic proliferation induced in the lymphopenic state in neonates might all account for the observed decrease in peripheral α/β T cell numbers. In this paper, it is relevant that it has been previously demonstrated that PDK1-null T cells display normal IL-7–dependent peripheral survival (22), and we confirmed this observation using our mouse model (data not shown).
In a thymus where T cells are unable to egress to the periphery, an accumulation of CD24low mature TCRβhigh CD4/CD8 SP T cells occurs (37–39). There was no such accumulation of mature T cells in Pdk1fl/fl CD4-cre thymi (Fig. 2B). We therefore considered that possibility that the peripheral defect in Pdk1fl/fl CD4-cre T cells occurred, because these cells failed to undergo lymphopenic-induced proliferation. In this context, in neonatal mice, there is a physiological process of lymphopenia-induced T cell proliferation that populates the peripheral lymphoid tissues (40). To simulate this process, we mixed congenically marked WT and Pdk1fl/fl CD4-cre thymocytes 1:1 and injected them into the tail vein of a lymphogenic host (Rag2−/− mice) and determined the relative engraftment after 4 wk. Fig. 4D demonstrates deficient engraftment of Rag2−/− mice by Pdk1fl/fl CD4-cre T cells relative to WT T cells. PDK1-null T cells are thus defective in their ability to undergo homeostatic proliferation in a lymphopenic host.
PDK1 catalytic function in DPs is required for iNKT cell development but not for the thymic development of conventional α/β T cells
PDK1 is a kinase with multiple interaction domains, both protein-protein and protein-lipid binding domains, and it has been proposed that PDK1 may have a role as a scaffolding protein that is independent of its catalytic activity (41). PDK1 also has a reported role in the assembly of a multiprotein complex that controls NF-κB activation at the membrane following TCR/CD28 triggering of mature T cells (42). To assess whether iNKT cell development required PDK1 catalytic function, we analyzed iNKT cell development in mice where WT PDK1 alleles were substituted with a kinase dead mutant of PDK1. For these experiments, mice with WT PDK1 alleles substituted with a PDK1 allele with a point mutation at the key lysine residue required for ATP binding (K111A) in the PDK1 catalytic domain were generated. Mice homozygous for catalytically inactive PDK1 did not complete embryogenesis, and of 254 pups born from matings between Pdk1WT/K111A mice, none was homozygous for the PDK1K111A mutation. At E10.5 and E11.5, Pdk1K111A/K111A embryos were found at normal Mendelian frequency, but no embryos homozygous for the PDK1K111A mutation were found after E12.5 (data not shown).
To confirm that the K111A mutation renders PDK1 catalytically inactive, protein lysates were generated from Pdk1WT/WT, Pdk1WT/K111A, or Pdk1K111A/K111A E11.5-12.5 embryos. In vitro kinase assays demonstrate that PDK1 kinase activity is reduced by ∼50% in Pdk1WT/K111A embryos, whereas no PDK1 catalytic activity is detected in Pdk1K111A/K111A embryo lysates (Fig. 5A). PDK1 phosphorylates RSK on Ser227 and PKB on Thr308. Western blot analysis demonstrates that Pdk1K111A/K111A embryos express PDK1 (Fig. 5B), but these cells fail to phosphorylate RSK or PKB on their PDK1 substrate sequences (Fig. 5C). We also assessed PDK1 activity in Pdk1K111A/K111A cells by examining the phosphorylation of the S6 ribosomal protein by S6K1. S6 ribosomal protein phosphorylation is exquisitely sensitive to PDK1 activity, because it requires dual PDK1 inputs. S6K1 is itself phosphorylated and activated by PDK1 but only following phosphorylation by mammalian target of rapamycin, which itself is activated by PDK1/PKB signaling. The data (Fig. 5C) show that the phosphorylation of the S6 ribosomal protein was absent in Pdk1K111A/K111A cell lysates. Collectively, these data show that Pdk1K111A/K111A cells express a catalytically inactive PDK1.
As mice that are homozygous for the PDK1 K111A mutation do not survive beyond developmental stage E12.5, a conditional strategy was needed to produce thymocytes that selectively express Pdk1K111A. This was achieved by backcrossing mice that express a single Pdk1K111A allele and a single Pdk1fl allele (Pdk1K111A/fl) with mice expressing cre recombinase under the control of the CD4 promoter. The presence of the single PDK1 floxed allele in all tissues allows normal mouse development. However, as Pdk1K111A/fl CD4-cre T cells develop to DP thymocytes and express cre recombinase, the floxed PDK1 allele is deleted, thereby generating DPs that express a single Pdk1K111A allele. Fig. 5D shows that thymocyte numbers were normal in Pdk1K111A/fl CD4-cre thyme, and the thymocytes comprised both CD4/CD8 DP and CD4 or CD8 SP subpopulations. The frequency and total numbers of TCRβhigh CD4/CD8 SP cells was normal in Pdk1K111A/fl CD4-cre thymi (Fig. 5E, 5F), indicating that the catalytic activity of PDK1 in DP thymocytes is not required for the development of conventional α/β T cells in the thymus. However, the data show that in the absence of PDK1 catalytic activity, the surface expression of the CD8 coreceptor on CD8 SP thymocytes is reduced (Fig. 5D, 5F, Supplemental Fig. 2A). In addition, the expression of CD62L is elevated, whereas CD44 expression is decreased in SP thymocytes in Pdk1K111A/fl CD4-cre mice (Supplemental Fig. 2B). Therefore, these data confirm a role for basal PDK1 signaling in regulating the expression of CD8, CD44, and CD62L on mature thymocytes. Also consistent with the Pdk1fl/fl CD4-cre mouse phenotype, Pdk1K111A/fl CD4-cre mice have a profound defect in peripheral T cells (data not shown) (Fig. 5G); a stage 1 block in iNKT cell development with small stage 1 iNKT cells lacking the expression of CD71 and CD98 (Fig. 5H, Supplemental Fig. 2C), and iNKT numbers in peripheral tissues are severely decreased (Supplemental Fig. 2D). Taken together, these data confirm that the observed T cell and iNKT cell phenotype of the Pdk1fl/fl CD4-cre mouse is due to the loss of PDK1 kinase activity rather than any scaffold function of PDK1.
PKB is required for iNKT cell proliferation
In DN T cell progenitors, PDK1 is essential because it is required to couple activation of PI3K to the PKB-controlled signaling pathways that support the energy demands of proliferating TCRβ-selected thymocytes (16, 17, 19–21). In this respect, the loss of the p110δ isoform of PI3K results in decreased iNKT cell but not conventional T cell numbers (43). Therefore, there is a requirement for both PI3K and PDK1 signaling for iNKT cell versus conventional thymocyte development. In this context, PKB needs to bind phosphatidylinositol-3,4,5-trisphosphate, the lipid product of PI3K, to be phosphorylated and activated by PDK1 (44, 45). We therefore considered whether the PDK1 dependency for iNKT cell development was explained by a PKB requirement for the robust proliferative burst of AgR-primed cells that accompanies iNKT cell differentiation. The greatly reduced numbers of iNKT cells in the Pdk1fl/fl CD4-cre thymus made it technically very difficult to examine whether there was a failed proliferative burst or a failure of iNKT cell survival. Accordingly, we used a pharmacological strategy in vitro to test the hypothesis that PKB is required to support the proliferation of iNKT cells. In initial experiments, we examined the PKB requirement for IL-15–mediated iNKT cell proliferation. The rationale for these experiments is that IL-15 is a key cytokine for iNKT cells, and in its absence, there is a block in normal iNKT cell development in the thymus, whereas the development of conventional α/β T cells is normal (46, 47). To assess IL-15–driven iNKT cell proliferation, WT thymic iNKT cells were stained with CFSE, cultured in various doses of IL-15, and then iNKT cell proliferation assessed by CFSE dilution. Fig. 6A shows that WT iNKT cells but not CD4 SP thymocytes proliferate in a dose-dependent manner in response to IL-15. However, it is not known whether IL-15 stimulates PKB activity in iNKT cells. Therefore, we examined whether iNKT cells activate PKB in response to exposure to IL-15. The data show that IL-15 stimulation of iNKT cells increased the phosphorylation of PKB on the two key residues required for activity, Ser473 and the PDK1 substrate site Thr308 (Fig. 6B). Analysis of the phosphorylation of the PKB substrate tuberous sclerosis protein 2 (TSC2) (Ser1462) confirmed increased PKB kinase activity following IL-15 stimulation (Fig. 6B). We next examined the role of PDK1/PKB in IL-15–mediated iNKT cell proliferation by using a selective allosteric inhibitor of PKB, AKT-1/2i (48). This compound prevents the conformational change that occurs when the PKB pleckstrin homology domain binds phosphatidylinositol-3,4,5-trisphosphate, thus inhibiting the PDK1-mediated phosphorylation essential for PKB activation (49). Inhibition of PKB does not compromise iNKT cells survival (Fig. 6C) but strongly inhibits in vitro proliferation of iNKT cells in response to IL-15 (Fig. 6D).
Would failure of IL-15–mediated proliferation explain the deficient iNKT cell development in PDK1-null mice? Partially, although the defect in iNKT cell development seen in the absence of the IL-15R is less severe to that observed in the Pdk1fl/fl CD4-cre mice (46, 47). This discrepancy indicates that PDK1 must mediate the response of iNKT cells to additional exogenous signals. One possibility is that PDK1 is required for the AgR-induced priming of iNKT cells that is required to initiate their proliferative burst in the thymus. To test this hypothesis, we examined the PKB requirement for Vα14 T cell AgR-induced proliferation of iNKT cells. Fig. 7A shows that WT iNKT cells proliferate following in vitro Vα14 TCR stimulation with αGalCer-loaded CD1d. CD1d-αGalCer specifically stimulates the invariant Vα14 TCR because there was no proliferation of conventional α/β CD4 SP T cells in this assay (Fig. 7B). We then assessed whether triggering of the Vα14 TCR in iNKT cells activates PKB. Fig. 7C shows that CD1d-αGalCer stimulation of in vitro-cultured iNKT cells causes increased phosphorylation of PKB and increased PKB activity as judged by increased phosphorylation of the PKB substrate TSC2. To determine whether PKB is required for TCR-stimulated proliferation of iNKT cells, experiments with the specific inhibitor AKT-1/2i were performed. The data show that AKT-1/2i treatment abrogates CD1d-αGalCer stimulated iNKT cell proliferation (Fig. 7D). Therefore, TCR-induced proliferation of iNKT cells requires PKB activity.
This study uses two independent genetic strategies to study the role of PDK1 in the differentiation and maturation of DP thymocytes. One strategy is to delete PDK1 floxed alleles in DPs with cre recombinase. The second strategy is to study the development of DP thymocytes where WT PDK1 alleles are substituted by a PDK1 allele with a single point mutation that prevents ATP binding to the catalytic site. This latter approach ablates PDK1 kinase activity while retaining any scaffold function of the protein. It thus provides a more precise model for studying the role of PDK1 kinase activity in T cell development. It is also an approach that avoids some of the fundamental caveats of gene deletion models, namely that removal of a protein can alter the balance of multiple signaling pathways. A key finding is that loss of PDK1 catalytic function in DP thymocytes causes a profound defect in intrathymic iNKT cell development but allows the maturation of conventional α/β thymocytes and Tregs in the thymus.
Why would the development of iNKT cells be dependent on PDK1? Although initial positive selection of iNKT cell progenitors proceeds normally in the absence of PDK1, it is clear that PDK1 signaling is required for subsequent iNKT cell development involving robust proliferative expansion of stage 1 progenitors. Thus, PDK1-null stage 1 iNKT cells are small and have failed to upregulate the expression of crucial nutrient receptors CD71 and CD98. The data in this study show that the PDK1 substrate PKB is essential for the proliferative responses of iNKT cells to multiple signals that are crucial iNKT cell development, namely IL-15 and the Vα14 TCR. The molecular basis for the failed iNKT cell development caused by loss of PDK1 catalytic activity can thus be attributed to the essential role of the PDK1/PKB signaling pathway for the proliferative burst of iNKT cells that is driven by Vα14 TCR/IL-15 signaling and accompanies the positive selection of these cells. The data identify PDK1 as a kinase that integrates cytokine and AgR signaling to support the proliferative expansion of iNKT cells in the thymus. This is reminiscent of a similar PDK1/PKB requirement for the pre–TCR/Notch-induced changes in thymocyte metabolism that support the proliferative expansion that occurs as T cell progenitors transit from DNs to DPs (16, 17, 19–21).
The PDK1 independence of conventional α/β T cells at the DP to SP stage is indicative that there are distinct metabolic requirements for the positive selection of conventional α/β T cells versus iNKT cells, probably reflecting that conventional α/β T cells only proliferate modestly at this stage (50). However, the PDK1 independence of the DP/SP transition of conventional α/β T cells and Tregs was unexpected because this stage of thymus development is controlled by the strength of TCR signaling and triggering of the T cell AgR activates multiple PDK1-controlled kinases. Moreover, PDK1/PKB signaling in peripheral T cells modulates their differentiation: for example, strong activation of PKB can impair Th17 and Treg differentiation (24, 25). Conversely, premature termination of PKB activity promotes the differentiation of Tregs (23). Nevertheless, PDK1 was dispensable for the production of regulatory and conventional α/β T cells in the thymus. This does not mean that there is no role for PDK1 in SP thymocytes. In this respect, we could clearly demonstrate that constitutive PDK1 signaling programs the biology of SPs. The data in this study thus show that PDK1 “fine-tunes” expression of the coreceptor CD8 in peripheral T cells and controls the expression of adhesion molecules and chemokine receptors including CD44, CD62L, and CCR7. PDK1-null SPs thus fail to express CD44 but show increased expression of CD62L and CCR7. The expression of CD62L and CCR7 is controlled by the transcription factor Foxo1, which has high basal activity in quiescent cells. Foxo1 is phosphorylated and inactivated by the PDK1 substrate PKB following T cell activation. The data show that in naive T cells, Foxo1 transcriptional activity can be increased by loss of PDK1. This result affords the insight that Foxo activity in SP T cells is dynamic and restrained by constitutive PDK1 signaling pathways.
The importance of PDK1 signaling for conventional T cells was emphasized by the universal deficit of T cell subsets in peripheral lymphoid tissues in both Pdk1fl/fl CD4-cre and Pdk1K111A/fl CD4-cre mice. The inability of PDK1-null T cells to undergo homeostatic proliferation induced proliferation in a lymphopenic environment may explain why the PDK1-null α/β SP thymocytes fail to effectively populate the periphery. Thus, the physiologic lymphopenic environment existing in neonatal mice supports the proliferative expansion of T cells that then populate the peripheral lymphoid tissues (40). In the absence of PDK1, thymic emigrants might not be able to meet the metabolic demands of this proliferative phase. In this respect, it has recently been shown that basal PI3K/PKB/Foxo signaling initiated by the BCR is crucial for peripheral homeostasis of mature B cells (51). Although PDK1 signaling is not required for IL-7–mediated T cell peripheral survival, weak TCR signaling, TCR “tickling,” is also important for T cell peripheral homeostasis (52). Therefore, basal TCR-mediated PDK1/PKB signaling may also have a role in maintaining T cell homeostasis in the periphery.
The data in this study are consistent with a model that PDK1 is necessary for T cells whenever there is a metabolic demand on the cells imposed by a phase of rapid proliferation. PDK1 catalytic activity is thus essential for all T cells but at precise developmental stages depending on the metabolic demands exerted upon the developing T cell.
Disclosures The authors have no financial conflicts of interest.
This work was supported by a Welcome Trust Principal Research Fellowship and Program grant (065975/Z/01/A0).
The online version of this article contains supplemental material.
Abbreviations used in this paper:
invariant Vα14 NKT
phosphoinositide-dependent kinase 1
protein kinase B
protein kinase C
promyelocytic leukemia zinc finger
90-kDa ribosomal S6 kinase
70-kDa S6 ribosomal protein kinase 1
regulatory T cell
tuberous sclerosis protein 2