Developmental Arrest of T Cells in Rpl22-Deficient Mice Is Dependent upon Multiple p53 Effectors

Development of αβ-lineage T cell progenitors from the CD4^–CD8^– double-negative (DN) stage to the CD4⁺CD8⁺ double-positive (DP) stage requires traversal of the β-selection checkpoint, which ensures that only progenitors that have productively rearranged the TCR β locus will survive. DN thymocytes can be further subdivided based on the surface expression of CD25 and CD44 into four subsets: DN1, CD44⁺CD25^–; DN2, CD44⁺CD25⁺; DN3, CD44^–CD25⁺; and DN4, CD44^–CD25^–. TCR-β rearrangement is initiated as thymocytes develop from the DN2 to the DN3 stage. If during this transition rearrangement of the TCRβ locus fails to preserve the translational reading frame of TCR-β, the cells die by apoptosis (1); however, if the translational reading frame is preserved and produces a functional TCR-β protein, it assembles with the remaining subunits of the pre-TCR complex (pre-Tα, along with CD3γ, δ, ε, and ζ) and transduces ligand-independent signals, resulting in a number of developmental outcomes, including termination of TCR-β rearrangement (i.e., allelic exclusion), rescue from cell death, proliferation, and differentiation to the DP stage (2, 3). The pre-TCR complex orchestrates these developmental outcomes by regulating the expression or function of numerous transcription factors including early growth response genes (Egr1-3) and NF-ATc, which cooperatively yield increased expression of inhibitor of DNA binding 3 and traversal of the β-checkpoint (4–6).

We recently made the surprising finding that the ribosomal protein L22 (Rpl22) is required for traversal of the β-selection checkpoint and also appears to be an important molecular effector of the developmental outcomes orchestrated by pre-TCR signaling (7). Rpl22 is a ubiquitously expressed RNA binding protein that is a component of the 60S ribosomal subunit, but is not essential for global or CAP-dependent translation (8, 9). Strikingly, Rpl22 ablation does not affect health or size of the mice but does result in a profound T lymphopenia, with most αβ lineage T cells arresting at the β-selection checkpoint at the DN3 stage (7). The developmental arrest at the DN3 stage in Rpl22-deficient mice results from an αβ lineage-restricted induction of the p53 tumor suppressor, as epistasis analysis reveals that the developmental arrest is completely alleviated by eliminating p53 through gene ablation (7). Interestingly, p53 induction in Rpl22-deficient DN3 cells appears to result from increased translation, implicating Rpl22 as a regulator of p53 synthesis. Other ribosomal protein defects have been implicated in impairing hematopoietic cell development (e.g., RpS19 mutations disrupting erythroid development in Diamond-Blackfan Anemia) (10); however, no ribosomal protein mutations had been shown previously to selectively impair T cell development (11).

Whereas ribosomal proteins are critical components of cellular ribosomes on which all proteins are synthesized, an increasing number of reports have revealed additional roles for ribosomal proteins in regulating fundamental cellular processes, such as survival, from outside of the ribosome (12). Among these extraribosomal functions is the regulation of p53 expression. For example, when ribosomal protein S7 fails to be incorporated into ribosomes because of impaired small ribosomal subunit assembly, the translation of the ribosomal protein L11 (Rpl11) is enhanced, which in turn activates p53 by preventing its degradation by the ubiquitin ligase MDM2 (13). Likewise, ribosomal proteins Rpl5 and Rpl23 are also able to activate p53 by binding to and impairing MDM2 function following release from the nucleolus under conditions of nucleolar stress (14, 15). Finally, Rpl26 is able to directly increase the rate of p53 synthesis via binding to the 5′ untranslated region (UTR) of p53 mRNA (16, 17). Therefore, some ribosomal proteins control p53 by regulating its stability, but other ribosomal proteins (Rpl26 and Rpl22) do so by altering p53 translation (7, 13).

The tumor suppressor p53 regulates a complex and multilayered network of proapoptotic and cell cycle inhibitory factors and mediates life-or-death decisions in response to stresses. One way in which p53 activation regulates survival is through the transcriptional induction of the proapoptotic Bcl-2-homology domain 3-only (BH3) family of proteins (18). This family of proteins includes Bcl-associated X protein (Bax), NADPH oxidase activator 1 (Noxa), and p53-upregulated modulator of apoptosis (PUMA). An additional BH3-only family member, Bim, which is not a direct p53 target, is nevertheless induced in T cells in response to stresses such as cytokine withdrawal and DNA damage and ultimately leads to compromised mitochondrial integrity and apoptosis (19, 20). Several of these proteins have been identified as critical effectors of p53-mediated apoptosis or growth arrest. PUMA-deficient T cells are highly resistant to ultraviolet and chemically induced apoptosis, despite large increases in p53 protein (21). Similarly, Bim or Noxa-deficient T cell precursors or MEF cells exhibited enhanced resistance to cell death in response to similar stimuli (19, 22, 23). In addition to its role in regulating cell survival, p53 can also control cell-cycle progression via the G1 checkpoint protein p21^waf, which allows for reversible cell cycle arrest in the event of DNA damage (24–26) or for a permanent arrest through activation of an oncogene-induced senescence program. In addition to the transactivation of particular downstream regulators of growth or survival, p53 can also induce the micro-RNA miR-34a, which is able to regulate a large battery of individual p53 effectors and is thereby able to replicate many of the consequences of p53 induction, including G1 arrest and apoptotic cell death (27).

Many downstream targets are induced upon p53 activation, but the particular effectors responsible for the diverse developmental outcomes differ depending on the stimulus responsible for p53 induction. For example, PUMA orchestrates thymocyte death in response to DNA damage, whereas Noxa is more important in the context of oncogene activation (28). Although our observations indicated a link between Rpl22 and p53 protein expression in developing αβ lineage T cells, the p53 targets responsible for the arrest of thymocyte development were unknown. In this study, we identified the targets upregulated after p53 activation in Rpl22-deficient αβ-lineage precursors and then assessed their role in the resulting p53-mediated developmental block. We found that many direct targets of p53 activation were induced, including miR-34a, PUMA, p21^waf, Bax, and Noxa, as well as the proapoptotic BH3 only gene Bim, which is not considered to be a direct p53 target. Interestingly, whereas a gain-of-function analysis indicated that overexpression of miR-34a is sufficient to impair development in a similar fashion to that observed after p53 induction in Rpl22-deficient T cells, few proapoptotic factors (PUMA and Bim) exerted sufficiently potent effects such that development was detectably rescued by their individual elimination. Importantly, co-elimination of the proapoptotic factors PUMA and Bim resulted in a nearly complete rescue of the development of Rpl22^−/− thymocytes. In contrast, simultaneous elimination of the primary p53 regulators of cell cycle arrest (p21^waf) and apoptosis (PUMA) actually afforded less rescue in terms of cell frequency and cellularity than ablation of PUMA alone, suggesting that loss of the G1 checkpoint protein p21^waf rendered thymocytes more sensitive to arrest by other p53 effectors.

All mouse strains were housed in the Laboratory Animal Facility at Fox Chase accredited by the Association for Assessment and Accreditation of Laboratory Animal Care and were handled in accordance with protocols approved by the Institutional Animal Care and Use Committee. The following mouse strains were used: Rpl22-deficient (7); PUMA-deficient (21); Rag2-deficient (29); p21^waf-deficient (Jackson Research Laboratories, Bar Harbor, ME) (24); pre-Tα-deficient (a gift from Dr. H von Boehmer, Dana Farber Cancer Institute) (30); and p53-deficient (a gift from Dr. Maureen Murphy, Fox Chase Cancer Center) (31). Genotypes were confirmed by performing PCR analysis on tail DNA at weaning and again before use in experiments.

Total RNA was isolated from primary cells using the RNA-Easy system (Qiagen, Valencia, CA). Purified mRNA was converted to cDNA using Superscript II reverse transcription PCR with oligo dT_(12–18) primers (Invitrogen, Carlsbad, CA). Quantitative real-time PCR was performed on the Prism 7700 thermocycler (Applied Biosystems, Carlsbad, CA) using TaqMan real-time PCR primer/probe sets specific to murine PUMA/Bbc3 (Mm00519268_m1), Noxa/Pmaip1 (Mm00451763_m1), and Bim/(Bcl2l-11) (Mm01333921_m1). P21^waf/Cdkn1a and Bax primer/probe sets were custom generated by the Fox Chase Cancer Center Genomics Core Facility. miR RNA was isolated using the miR Easy system in conjunction with the RNeasy MinElute Cleanup Kit (Qiagen, Valencia, CA). Isolated RNA was converted to cDNA by the Multiscribe RT-PCR kit (Applied Biosystems). miR expression was determined by the TaqMan Micro RNA assay for miR-34a and sno-202. Real-time PCR was performed as indicated above.

Mice were injected i.p. with affinity column purified anti-CD3 (145-2C11) at 10 μg/g body weight in a total volume of 200 μL of PBS per mouse or were injected with PBS alone. Mice were then harvested at the indicated times. For each experiment, four mice received PBS (control) treatment and two mice per time point received anti-CD3 injection. CD4^–CD8^–CD44^– (DN3⁺DN4) thymocytes were isolated by cell sorting and used to generate cDNA for real-time PCR analysis or to produce detergent extracts for immunoblotting.

Cells were lysed in NP-40 Lysis Buffer (1% NP-40, 50 mM TRIS [pH 8.0], 150 mM NaCl) with a complete mini-tab inhibitor tablet. Samples were resolved on NuPage Novex Bis-Tris gels (Invitrogen) and blotted with the following Abs: anti-p53 (Clone #IMX25; Leica Microsystems, Newcastle Upon Tyne, U.K.), anti-Bim/BOD (Clone AAP-330; Assay Designs, Ann Arbor, MI), anti-Noxa (FL-103, Santa Cruz Biotechnology, Santa Cruz, CA), anti-myc tag (EQKLISEEDL) and anti-Bax (Cell Signaling Technologies, Beverly, MA).

Viral particles were produced by transient calcium-phosphate transfection of Phoenix cells with shRNAs expressed in the MSCV-based vectors, LMP or LMS, as described previously (32). Fetal livers were harvested on day 14 of gestation and genotyped as described above. Hematopoietic precursors were expanded on OP9-DL1–expressing monolayers containing IL-7 (5 ng/ml) and Flt-3 (5 ng/ml). After 4 days, cells were spin-infected with retrovirus containing shRNA targeting Bax (MLS-Bax), Bim (MLS-Bim), Noxa (MLS-Noxa), or an irrelevant control (MLS-T-1052). For p53 knockdown, precursors were infected with a control vector (LMP) or LMP-p53 (provided by S. Lowe, Cold Spring Harbor Laboratory). Cells were cultured for the indicated times and analyzed by flow cytometry.

Single-cell suspensions of thymic cells were stained with Abs to CD4 (clone GK1.5), CD25 (clone 7D4), CD44 (clone IM7), γδTCR (clone GL3), and CD8 (Clone 53-6.7) (BioLegend, San Diego, CA). For cell death assays, cells were stained with annexin V PE (BD Biosciences, San Jose, CA). Data were collected using the LSRII (BD Biosciences, San Jose, CA) and were analyzed using FlowJo software (Tree Star, Ashland, OR).

Two consecutive Myc-Tag sequences encoding EQKLISEEDL were appended to the 5′ end of the Noxa mRNA sequence, along with the relevant 3′ UTR binding region for the Noxa shRNA. The PCR-generated product was directionally cloned into PMIY dsCherry II and then used to generate virus for stable retroviral transduction into Scid.adh cells. Cells transduced with either control shRNA or Noxa shRNA were sorted for GFP expression and used in Immunoblot analysis for Myc-tag expression (anti-Myc Tag [71D10], Cell Signaling Technologies).

Cell cycle analysis was performed using the FITC BrdU Flow Kit (BD Pharmingen, San Jose, CA). Mice were injected i.p. with 1 mg/ml BrdU in sterile PBS and rested for 4 h prior to harvesting. Thymocytes were surface stained with T cell markers and then fixed and permeabilized with BD Cytofix/Cytoperm buffer. Each sample was treated with 30 μg DNAse for 1 h prior to intracellular staining with anti-BrdU FITC and 7-amino-actinomycin D (7-AAD) and analysis on the LSR II.

Germline ablation of the Rpl22 gene in mice results in a severe T lymphopenia caused by selective impairment of αβ-lineage T cell development in the thymus, with few T cell precursors maturing beyond the CD4^–CD8^– DN stage (Fig. 1A). Further subsetting of DN thymocytes, based on differential expression of CD44 and CD25, revealed that the developmental block occurs at the CD44^–CD25⁺ DN3 stage (Fig. 1B). The impairment in T cell development in Rpl22^−/− mice was associated with increased apoptosis, as indicated by elevated annexin V binding in the majority of T cell subsets (Fig. 1C). Interestingly, the developmental blockade was preferentially manifested in αβ lineage precursors, as development of γδ lineage cells was only mildly impaired (Fig. 1B). Previous analysis revealed that the development of Rpl22-deficient thymocytes is impaired by activation of a p53-dependent checkpoint (7). Indeed, development of Rpl22-deficient thymocytes is completely restored by p53 deficiency, as illustrated by the total and complete rescue of T cell development in Rpl22^−/−Trp53^−/− mice. These mice demonstrated both restored thymic cellularity and representation of T cell populations (Fig. 1A, 1B) as well as reduced apoptosis (Fig. 1C).

The relative resistance of developing γδ lineage cells to Rpl22 deficiency results from the absence of p53 induction in these cells (7). Accordingly, we compared the expression of p53 effector molecules in αβ and γδ lineage precursors to gain insight into those responsible for the observed developmental block. We performed quantitative real-time PCR analysis on sorted αβ-lineage DN3 and γδ T cells from Rpl22^+/+ and Rpl22^−/− mice as well as from pre-Tα-deficient (Ptcra^−/−) mice, which are similarly arrested at the DN3 stage, albeit by a distinct mechanism (30, 33). We observed a 2- to 6-fold increase in both cell cycle arrest and proapoptotic factors in Rpl22^−/− αβ-lineage thymocytes, including p21^waf, Bax, Bim, and PUMA, and a consistently robust increase in Noxa (Fig. 2). The increased expression of p21^waf and the proapoptotic factors was restricted to Rpl22^−/− αβ-lineage cells, as it was not observed in either Rpl22^−/− γδ lineage cells or in Ptcra^−/− mice, where the developmental arrest results from the absence of pre-TCR signaling (30). Finally, we observed Noxa induction in Rpl22^−/− γδ lineage cells, but to a much lesser extent than in Rpl22^−/− αβ lineage DN3 cells (Fig. 2). These data suggest that the impairment in αβ lineage T cell development in Rpl22-deficient mice is correlated with the induction of numerous direct and indirect p53 targets capable of inducing cell cycle arrest (p21^waf) and apoptosis (Bax, Noxa, PUMA, and Bim).

Developing T lineage precursors that undergo apoptosis in the thymus are rapidly cleared by phagocytosis (34). Accordingly, it is possible that the expression analysis in Fig. 2 could have been affected by the relative efficiency with which different populations are phagocytically removed. In an attempt to mitigate this concern, we used a model system in which the traversal of the β-selection checkpoint could be induced in a synchronous fashion, using anti-CD3 stimulation of pre-TCR deficient (Rag2-deficient) thymocytes (35). Rag2-deficient thymocytes that were either Rpl22^+/+ or Rpl22^−/− were stimulated in vivo by i.p. injection with anti-CD3 Ab, which mimics a pre-TCR stimulus and induces thymocytes to mature from the DN3 to the CD44^–CD25^– DN4 stage within 24 h (35) (Fig. 3A). We observed that by 24 h after stimulation, the extent of CD25 downmodulation was beginning to lag in the Rpl22^−/− mice, even though both Rpl22^+/+ and Rpl22^−/− thymocyte populations had received equivalent stimulation, as shown by identical induction of CD5, which is an indicator of signal intensity (36) (Fig. 3A). Rpl22^−/− thymocytes exhibited elevated levels of p53 protein before anti-CD3 stimulation that were robustly induced upon anti-CD3 treatment (Fig. 3B). These data indicate that the induction of p53 began to manifest as a developmental arrest by 24 h after stimulation (Fig. 3A, 3B). Surprisingly, we also observed a modest induction of p53 in anti-CD3–stimulated Rpl22^+/+ thymocytes (Fig. 3B). Moreover, it was transient, as it abated by 48 h in Rpl22^+/+ but not Rpl22^−/− thymocytes, where its sustained expression leads to cell death (Fig. 3B). The cause of p53 induction in anti-CD3–stimulated Rpl22^+/+ thymocytes is not clear but could be due to a transient stress response associated with the extensive proliferation caused by this mitogenic signal. Having identified an early stage in the p53-mediated developmental block of Rpl22^−/− thymocytes (24 h), we performed quantitative real-time PCR analysis to determine whether the set of p53 effectors induced under these conditions was similar to those above. As in asynchronously developing Rpl22^−/− thymocyte populations, we found 2- to 6-fold increases in expression of the proapoptotic factors PUMA, Bax, and Noxa compared with PBS-treated Rpl22^+/+ cells (Fig. 3C). In contrast to what we observed in asynchronously developing Rpl22^−/− thymocytes (Fig. 2), expression of p21^waf1 and Bim decreased in both the anti-CD3 stimulated Rpl22^+/+ and Rpl22^−/− cells (Fig. 3C). Currently, the basis for this difference in p21^waf and Bim expression between mitogenically stimulated and asynchronously developing Rpl22^−/− thymocytes is unknown; however, this may reflect a temporal delay in induction relative to that of other p53 targets.

miRs are small noncoding RNAs that can regulate gene expression by altering mRNA stability or translation (27, 37, 38). miR-34a was recently identified as a direct p53 target, whose induction is capable of independently replicating many of the effects of p53 activation, including G1 arrest and apoptosis (27). Therefore, we asked whether miR-34a was induced by p53 in Rpl22-deficient thymocytes. Both control Rag2^−/−Rpl22^−/− cells and those stimulated to differentiate with anti-CD3 exhibited a 2-fold increase in miR-34a expression over control treated Rpl22-sufficient T cells (Fig. 4A). Because miR-34a is induced in Rpl22-deficient T cells, we asked whether miR-34a expression alone was sufficient to replicate the effects of p53 induction and arrest T cell development. To address this question, we used a gain-of-function approach by retrovirally transducing miR-34a into adult DN thymocytes and monitoring development in OP9-DL1 stromal cell cultures. OP9-DL1 cells support the development of immature T cells through the DP stage in vitro (39). Cells transduced with empty vector or mutant miR-34a displayed a marked increase in the number of T lineage precursors that had developed to the immature DP stage between days 1 and 4 (Fig. 4B). Conversely, transduction with miR-34a significantly impaired development to the DP stage, both regarding the frequency and number of cells (Fig. 4B, C). Our findings suggest that induction of the p53 target miR-34a has an important role in the p53-mediated blockade of T cell development in Rpl22^−/− mice.

Gain-of-function analysis revealed that enforced expression of miR-34a caused a developmental arrest reminiscent to that resulting from p53 activation in Rpl22^−/− thymocytes. As a result, we asked whether loss-of-function analysis would reveal which of the induced p53 effectors have a sufficiently important role that their individual ablation alleviates the developmental block. p53 activation induces targets that fall into two broad functional classes: those that regulate cell growth and those that regulate survival. Among these factors, the ones most frequently implicated in regulating cell growth and survival are p21^waf and PUMA, respectively (21, 24, 40). Because both of these p53 targets are induced in Rpl22^−/− thymocytes (Figs. 2, 3C), we used loss-of-function analysis to investigate their importance in the p53-mediated impairment of thymocyte development. Specifically, we assessed whether p21^waf deficiency (i.e., Cdkn1a ablation) would alleviate the development block and restore development of Rpl22^−/− thymocytes to the DP stage. Surprisingly, rendering Rpl22-deficient thymocytes deficient for p21^waf (Rpl22^−/−Cdkn1a^−/−) failed to alleviate the developmental arrest, as the vast majority of thymocytes remained at the DN stage, primarily at the β-selection checkpoint at DN3, just as in p21^waf-expressing Rpl22^−/− mice (Fig. 5A, 5B). Moreover, p21^waf deficiency did not suppress the increased apoptosis routinely observed in Rpl22-deficient thymocytes (Fig. 5C). These data indicate that elimination of p21^waf alone was not sufficient to restore development of Rpl22^−/− thymocytes, perhaps because proapoptotic p53 targets were still present and might be inducing cell death.

Because ablation of the cell cycle regulator p21^waf failed to rescue development, we attempted to rescue the p53-mediated developmental arrest by disabling apoptotic pathways. PUMA is a proapoptotic BH3 family member and a major p53-activated inducer of apoptosis, whose ablation can abrogate p53-mediated death induced by cellular insults or stressors, including gamma-irradiation, dexamethasone, or cytokine withdrawal (21). We therefore asked whether PUMA deficiency would alleviate the p53-induced developmental block of Rpl22^−/− thymocytes. Relative to mice singly deficient for Rpl22, those that were doubly deficient for both PUMA and Rpl22 exhibited a significant increase in the frequency and absolute number of thymocytes that had developed to the DP stage (Fig. 6A). Nevertheless, this represented only a partial rescue, returning thymic cellularity to only ∼10% of wild type levels (Fig. 6A) and failing to cause a significant increase in the absolute number of cells that have developed beyond the DN3 stage to DN4 stage cells (Fig. 6B). Finally, an examination of cell viability by annexin V staining indicated that whereas PUMA deletion substantially reduced the level of apoptosis in nearly all T cell subsets, it appeared that the suppression of apoptosis was incomplete, as indicated by the elevated apoptosis manifested in total Rpl22^−/−Puma^−/− thymocytes and the CD4 SP thymocyte subsets, relative to littermate controls (Fig. 6C). These data indicate that PUMA has an important role in the p53-mediated promotion of apoptosis in Rpl22^−/− thymocytes; however, unlike thymocyte apoptosis following gamma-irradiation, which is fully rescued by PUMA deficiency (21), the apoptosis accompanying Rpl22 deficiency appears to involve additional proapoptotic effectors.

To determine whether proapoptotic effectors other than PUMA were contributing to the death of Rpl22^−/− thymocytes, we investigated the other death effectors identified in our expression analysis (Bax, Bim, and Noxa; Fig. 2). Their role in the developmental block of Rpl22^−/− thymocytes was examined by knocking down their expression in Rpl22^−/− fetal progenitors and monitoring the effect on development on OP9-DL1 monolayers, as was done for adult progenitors in Fig. 4. Analysis was restricted to GFP⁺ cells that had been transduced with the shRNA retroviral vector (MLS) (32). A comparison of GFP⁺ cells reaching the DP stage in shRNA-treated versus the control shRNA-expressing cells reveals the extent to which knockdown of the death effectors blunts p53-mediated apoptosis and alleviates the developmental blockade of αβ lineage Rpl22^−/− thymocytes (Fig. 7). Knockdown of p53 in Rpl22^−/− progenitors restored development to approximately half the level observed in control-transduced Rpl22^+/+ progenitors and was used to normalize results from knockdown of proapoptotic effectors (Fig. 7A). Surprisingly, while expression of multiple proapoptotic effectors was induced in Rpl22^−/− thymocytes (Bim, Noxa, and Bax), knocking down the expression of Noxa and Bax failed to alleviate the developmental blockade and promote development to the DP stage (Fig. 7A, 7C); however, knockdown of Bim resulted in a partial rescue of development, restoring development approximately half as effectively as did knocking down p53 (Fig. 7A). Importantly, when Bim was knocked down in Rpl22^−/−Puma^−/− T cells, the rescue of development to the DP stage was nearly equivalent to that resulting from knocking down p53 itself (defined as 100%; Fig. 7B, 7C). In summary, our loss-of-function analysis of proapoptotic factors suggests that Bim and PUMA have significant roles in p53-mediated developmental blockade, while Bax and Noxa do not, despite the large induction in Noxa expression.

Our analysis of the roles of proapoptotic effectors revealed that the combined elimination of Bim and PUMA resulted in nearly complete rescue of development, suggesting that simultaneous elimination of other p53 effectors might be informative. Because p53 regulates both cell cycle through p21^waf and apoptosis through PUMA and Bim, we asked how simultaneously disabling the ability of p53 to mediate cell cycle arrest (p21^waf deficiency) and promote apoptosis (PUMA deficiency) would affect development of Rpl22-deficient progenitors. That is, we asked whether Rpl22/PUMA/p21^waf triple deficiency would more completely rescue development. Surprisingly, we observed that p21^waf deficiency eliminated even the partial rescue in development of Rpl22^−/− thymocytes afforded by PUMA deletion alone (Fig. 8A). Both total thymic cellularity as well as the proportion and absolute numbers of DP thymocytes were decreased to levels exhibited by mice singly deficient in Rpl22 (Fig. 8A). DN subpopulations were not significantly altered, as expected because they remained unchanged even in Rpl22/PUMA-deficient mice (Fig. 8B). Evaluation of BrdU incorporation, as a means of measuring the number of cells in S phase, revealed that despite their increased expression of p21^waf, Rpl22^−/− DN3 thymocytes actually incorporated more BrdU than Rpl22^+/+ littermate controls (Fig. 8C). Therefore, for reasons that are unclear, Rpl22 deficiency appears to increase the proliferation of DN3 progenitor cells. It is possible that the increased proliferation of these cells make them more sensitive to p53-induced apoptosis. Consistent with this notion, we detected only modest changes in the extent of BrdU incorporation between Rpl22/PUMA-double deficient and Rpl22/PUMA/p21^waf-triple deficient thymocytes (DN4; Fig. 8C), but triple-deficient thymocytes exhibited greater levels of apoptosis than did thymocytes from Rpl22/PUMA-double deficient mice in many of the subsets analyzed (Fig. 8D). These surprising findings suggest that p21^waf was actually functioning to facilitate development in the Rpl22/PUMA-double deficient mice, either through subtly restraining proliferation or by promoting survival, because the additional loss of p21^waf resulted in increased cell death.

These data indicate that the p53-mediated developmental block in Rpl22-deficient thymocytes results from the coordinate induction of a constellation of p53 targets, with the proapoptotic factors PUMA and Bim having predominant roles in enforcing the p53-mediated developmental arrest. Collectively, these p53 effectors are limiting proliferation and survival in a manner distinct from that observed following induction of p53 through other means, such as DNA damage.

Rpl22-deficient thymocytes fail to progress beyond the β-selection checkpoint because of the activation of a p53-mediated checkpoint. The importance of this checkpoint in mediating the developmental block is demonstrated by the complete restoration of T cell development upon ablation of Trp53 in Rpl22^−/− thymocytes (Fig. 1). In this study, we analyzed the downstream p53 effectors responsible for the p53-mediated developmental arrest of Rpl22^−/− αβ-lineage precursors. Expression profiling revealed that many p53 effectors capable of regulating cell growth and survival were induced, but few played sufficiently important roles to alleviate the developmental block upon removal. Indeed, we found that the expression of a cell cycle inhibitor, p21^waf, as well as the proapoptotic factors PUMA, Bim, Bax, and Noxa were all induced in Rpl22^−/− thymocytes. Nevertheless, only elimination of PUMA and Bim resulted in detectable restoration of T cell development, with nearly full restoration of development occurring upon simultaneous elimination of both. Therefore, our results indicate that even within the same tissue, different inductive stimuli (e.g., DNA damage versus Rpl22 deficiency) result in the utilization by p53 of a somewhat distinct set of molecular effectors to limit cell growth and survival.

Our loss-of-function analysis of p53 effectors revealed that elimination of individual p53 molecular effectors had a limited ability to alleviate the p53-mediated block and restore development. While not a direct target of p53, Bim expression was elevated in Rpl22^−/− DN3 thymocytes. The ability of Bim knockdown to partially restore development of Rpl22^−/− thymocytes beyond the β-selection checkpoint to the DP stage is consistent with a report indicating that Bim is an important death effector at this checkpoint, in that Bim deficiency was able to partially restore development of pre-TCR–deficient thymocytes (41). Furthermore, we observed that combined reductions in Bim and PUMA resulted in nearly complete rescue of the development of Rpl22^−/− T cell progenitors in vitro. The results are in agreement with recent findings that indicate combined Bim, PUMA, and Bid loss prevents direct activation of Bax/Bak-mediated apoptosis (42). Likewise, the failure of Noxa and Bax knockdown to restore development or Rpl22^−/− thymocytes is consistent with previously published analyses (43). Noxa is reported to induce apoptosis indirectly by promoting Bax and Bak function and to require oncogene expression (e.g., oncogene E1A) for maximal effect (23). Moreover, because Noxa was also upregulated in Rpl22^−/− γδ-lineage cells, which do not show p53 induction or arrested development, we reasoned that Noxa induction might be p53 independent and therefore unlikely to have a major role in the p53-mediated developmental block at the β-selection checkpoint. Lastly, knockdown of the Bcl-2 family member Bax also failed to rescue development, perhaps because the proapoptotic function of Bax is reported to be less potent in thymocytes than that of Bim (44), and because Bax deficiency is reported to produce a mild rescue of the developmental block caused by the absence of IL-7 signaling, but is not known to have a major role at the β-selection checkpoint (45). Alternatively, the related proapoptotic protein Bak can also form homodimers that result in mitochondrial disruption and cytochrome c efflux, and thus might compensate for Bax loss (46).

It was surprising that ablation of the two major p53 effectors controlling cell growth and survival in the thymus—p21^waf and PUMA, respectively—failed to rescue development completely and behaved antagonistically when eliminated simultaneously. Indeed, PUMA-deficiency is able to render thymocytes completely resistant to p53-mediated death after exposure to DNA damage (21); however, PUMA-deficiency failed to fully rescue the development of Rpl22^−/− thymocytes, instead restoring thymic cellularity to ∼10% of normal levels. Whereas co-elimination of Bim and PUMA was able to provide nearly complete rescue of the development of Rpl22^−/− thymocytes in vitro, presumably because of complete suppression of p53-mediated apoptosis, co-elimination of PUMA with cell cycle regulator p21^waf failed to alleviate the developmental block of Rpl22^−/− thymocytes and rendered the developmental block more penetrant, interfering with the ability of PUMA deficiency to restore development to the DP stage (Fig. 8). The mechanistic basis by which p21^waf deficiency antagonizes the rescue of development by PUMA-deficiency remains unclear, because we detected only modest changes in BrdU incorporation caused by further elimination of p21^waf (Fig. 8C). Nevertheless, it is possible that even subtle changes in proliferation could be biologically relevant, because cells in cycle have been reported to be more sensitive to p53-mediated apoptosis, and PUMA deficiency may be unable to rescue cells from p53-mediated death in this context (47, 48). Consistent with this possibility, 21^waf deficiency resulted in increased apoptosis in many thymocyte subsets from Rpl22/PUMA/p21^waf-deficient mice. This outcome may be a consequence of the aforementioned subtle effects on proliferation. Alternatively, recent studies have demonstrated an antiapoptotic role for p21^waf under certain cell stress conditions (49–52). The greater sensitivity of proliferating cells to cell death may also have an important role in rendering αβ lineage cells more sensitive to developmental block in Rpl22^−/− mice (7). Indeed, we also made the surprising observation that, despite expressing higher levels of p21^waf, Rpl22^−/−, thymocytes incorporated more BrdU than their wild type littermates. The mechanistic basis for this is unclear but is under active investigation. Moreover, αβ lineage progenitors are thought to undergo much more extensive proliferation than do γδ lineage progenitors (53, 54), and this may contribute to their increased sensitivity to development arrest due to loss of Rpl22.

The cause of p53 induction in Rpl22^−/− thymocytes remains unclear, but we have shown that it is associated with increased p53 synthesis, not a change in turnover (7). Because we observe p53 induction in Rpl22^−/−Rag2^−/− thymocytes, it does not result from the Rag2-mediated DNA cleavage events associated with TCR gene rearrangements. Moreover, the absence of a change in p53 stability suggests it is not due to other forms of DNA damage, which is consistent with the incomplete dependence of the resulting cell death on PUMA (21). Instead, our current hypothesis is that induction of p53 in Rpl22^−/− αβ lineage cells results from stress that is a consequence of the proliferative burst accompanying pre-TCR stimulation at the DN3 stage. This idea is consistent with the developmental block in Rpl22^−/− thymocytes being selective to those of the αβ lineage, because these cells undergo a more robust proliferative expansion following pre-TCR signaling than that which occurs during development of γδ lineage progenitors (53, 54). In agreement, we observed a transient induction of p53 after acute, mitogenic anti-CD3 stimulation of Rpl22-expressing Rag2^−/− thymocytes, which mimics pre-TCR stimulation (7) (Fig. 3). We think the reason that p53 induction is normally transient is because pre-TCR signaling induces Rpl22 expression, which then returns p53 levels to baseline by repressing p53 synthesis. We hypothesize that p53 induction is sustained in Rpl22^−/− αβ lineage cells, because Rpl22 is not present and so is incapable of repressing p53 translation. Nevertheless, it is possible that there are normal circumstances where p53 induction might be sustained, such as when the Notch signals upon which development of αβ lineage cells depend are unavailable. Because Notch signaling has been reported to repress p53 (and induce Rpl22) (55, 56), it is tempting to speculate that the absence of Notch signals could selectively eliminate αβ lineage cells through sustained p53-induction, because γδTCR⁺ thymocytes develop in a Notch-independent manner (57).

Our findings reveal that p53 induction in Rpl22^−/− αβ lineage thymocytes impairs their development through multiple molecular effectors controlling growth and survival. The particular effectors used appear to differ depending on the circumstances surrounding p53 activation. Although the p53 effector Noxa functions in the context of oncogene induction, PUMA is almost solely responsible for p53-mediated death following DNA-damage. In contrast, multiple direct and indirect p53 effectors (PUMA, miR-34a, Bim, and perhaps others) are used to impair development of Rpl22^−/− thymocytes. Efforts to address the basis for the selective induction of p53 in Rpl22^−/− αβ lineage cells and how this influences the constellation of effector molecules controlling growth and survival are in progress.

We thank Dr. Maureen Murphy for critical review of the manuscript and the following core facilities at Fox Chase Cancer Center for vital support: Biomarker, Cell Culture, DNA Sequencing, DNA Synthesis, Flow Cytometry, and Laboratory Animals.

S.J.A. and T.O. are all employees of, and have received stock options from, Lexicon Pharmaceuticals, Inc. M.B. and M.O.C. were both employees of Rosetta Inpharmatics, LLC at the time this work was performed. The other authors have no financial conflicts of interest.