Perturbation of thymocyte development in nonsense-mediated decay (NMD)-deficient mice

Abstract

The random nature of T-cell receptor-β (TCR-β) recombination needed to generate immunological diversity dictates that two-thirds of alleles will be out-of-frame. Transcripts derived from nonproductive rearrangements are cleared by the nonsense-mediated mRNA decay (NMD) pathway, the process by which cells selectively degrade transcripts harboring premature termination codons. Here, we demonstrate that the fetal thymus in transgenic mice that ubiquitously express a dominant-negative form of Rent1/hUpf1, an essential trans-effector of NMD, shows decreased cell number, reduced CD4CD8 double-positive thymocytes, diminished expression of TCR-β, and increased expression of CD25, suggesting a defect in pre-TCR signaling. Transgenic fetal thymocytes also demonstrated diminished endogenous Vβ-to-DβJβ rearrangements, whereas Dβ-to-Jβ rearrangements were unperturbed, suggesting that inhibition of NMD induces premature shut-off of TCR-β rearrangement. Developmental arrest of thymocytes is prevented by the introduction of a fully rearranged TCR-β transgene that precludes generation of out-of-frame transcripts, suggesting direct mRNA-mediated trans-dominant effects. These data document that NMD has been functionally incorporated into developmental programs during eukaryotic evolution.

Frameshift or nonsense mutations are a common cause of inherited and acquired genetic diseases, suggesting that nonsense-mediated decay (NMD) may have evolved to protect the organism from the deleterious consequences of truncated proteins that would result if nonsense transcripts were stable. Indeed, NMD modulates the severity of several disease phenotypes (1–3). Recent evidence suggests that the pathway also participates more broadly in the regulated control of gene expression. NMD has been shown to regulate the stability of physiological transcripts that mimic the architecture of nonsense mRNAs in both yeast and mammals (4–6). The nonsense surveillance machinery also appears to function in clearing premature termination codon (PTC)-harboring transcripts resulting from inefficient, alternative, or faulty RNA processing events occurring during transcription and splicing.

The Upf proteins (Upf1p, Upf2p, and Upf3p) have been shown to be essential trans-effectors of NMD in yeast (7–10). Human orthologs of each of these proteins have been identified, termed Rent1–3/hUpf1–3 (11–15). Complete loss of NMD in lower eukaryotes is well tolerated (16–19). However, homozygous targeted disruption of Rent1/Upf1 in mice resulted in stabilization of nonsense transcripts to WT levels and death at the periimplantation stage of development (20). Heterozygous targeted mice were completely competent in their ability to perform NMD and were fertile, had normal life spans, and showed no apparent phenotypic abnormalities (20). Recently, using a tissue-specific knock-out approach, Upf2 was shown to be essential for survival of hematopoietic stem and progenitor cells (21).

One process unique to higher eukaryotes where NMD is anticipated to be of paramount importance is maturation of the immune system. During normal lymphocyte development, T-cell receptor (TCR) and immunoglobin genes undergo a series of programmed gene rearrangements. For TCR-β, this process initiates with joining of a diversity (Dβ) segment to a joining (Jβ) segment. This is followed by a second recombination event in which the DβJβ unit is joined to a variable (Vβ) segment. One mechanism for generating diversity in the T-cell repertoire involves the addition and subtraction of nucleotides at the VβDβ and DβJβ junctions. Although these events function to increase the repertoire of TCRs capable of recognizing different antigens, the process is random and, as a result, two of three rearrangements result in the production of a frameshift and subsequent PTC (22). In the one-third of cells that have a productive rearrangement on the first try (β⁺β⁰), rearrangement of the second allele is inhibited through a poorly defined process termed allelic exclusion. The remaining two-thirds of cells (β⁻β⁰) rearrange the second allele; one-third of these events will be productive (β⁻β⁺), and two-thirds will be nonproductive (β⁻β⁻). The nonsense transcripts derived from out-of-frame TCR-β alleles (β⁻) are efficiently degraded by the NMD pathway (23). The physiological importance of this phenomenon was recently suggested by the absence of single positive T cells carrying β⁻ alleles in the periphery of mice conditionally deleted for Upf2, suggesting that complete loss of Upf2 is lethal to developing T cells that harbor a β⁻ allele (21). Only cells expressing a productively rearranged TCR-β allele (β⁺) undergo positive selection and survive.

Our understanding of the physiological importance of Rent1/hUpf1 and NMD in higher eukaryotes has been limited by the embryonic lethality of mice completely lacking function of the pathway. We have therefore generated a transgenic (Tg) mouse that ubiquitously expresses a dominant-negative form of human Rent1/hUpf1 containing a single amino acid change (R844C) in the highly conserved helicase domain of the protein that was previously shown to cause partial stabilization of nonsense transcripts in mammalian cells (24). Tg mice were viable and fertile with no gross phenotypic abnormalities. Here, we show that they demonstrate a crisis in development of the thymus coincident with the onset of TCR-β allele rearrangement. The phenotype included clonal dropout of cell populations, reduced total thymocyte cell number, a dramatic paucity of double-positive (DP) thymocytes with a corresponding increase in CD25^high double-negative (DN) cells, and reduced expression of TCR-β relative to WT littermates, suggesting arrest at the pre-TCR stage of development. These changes could be prevented by introduction of a fully rearranged TCR-β allele that effectively precludes the generation of out-of-frame TCR-β transcripts. Moreover, Tg mice demonstrated reduced frequency of Vβ-to-DβJβ rearrangements, which are subject to allelic exclusion, but normal frequency of Dβ-to-Jβ rearrangements, which are not. In summation, these data suggest that stabilized TCR-β nonsense transcripts may be sufficient to inhibit TCR rearrangement and, therefore, that NMD has been functionally incorporated into critical developmental programs during eukaryotic evolution.

Results

Generation and Characterization of Rent1/hUpf1 DN Tg Mice.

A human Rent1/hUpf1 cDNA encoding a mutant protein (R844C) with documented dominant-negative activity was introduced into fertilized murine oocytes using traditional Tg technology (Fig. S1). The founder that carried the highest copy number of the transgene was runted, produced no offspring, and died at ∼4 mo of age. A Tg littermate with a slightly reduced copy number was viable and fertile, allowing derivation of the line used for these studies (Fig. 1A). Importantly, mice derived from an independent founder showed concordant results, documenting that the observed effects relate to transgene expression rather than gene disruption at the insertion site (Fig. S2). Quantitative RT-PCR analysis revealed comparable levels of endogenous WT and mutant (Tg) Rent1/hUpf1 message in the fetal day (Fd) 16 thymus and all adult tissues except the kidney (Fig. 1B). To determine the transgene's effect on the efficiency of NMD, fibroblast cell lines generated from Tg and control littermates were transiently transfected with WT or nonsense-containing (PTC) forms of a TCR-β minigene construct (25). The normalized steady-state abundance of nonsense TCR messages was increased approximately twofold in Tg vs. WT cell lines (Fig. 1C). The transgene was also bred onto the gus^mps background, which carries a single base-pair deletion in the β-glucuronidase gene that generates a downstream PTC and initiates NMD (26). The steady-state abundance of the β-glucuronidase nonsense transcript was increased approximately threefold in the thymus of Tg animals relative to their WT littermates (Fig. 1D). These data document relative inhibition of NMD in Rent1/hUpf1 dominant-negative mice.

Fig. 1.

Generation and characterization of R844C Rent1/hUpf1 dominant-negative Tg (TG) mice. (A) Southern blot depicting inheritance of the Rent1/hUpf1 transgene. (B) Expression of the human Rent1/hUpf1 transgene relative to endogenous mouse (WT) message. The upper band represents endogenous Rent1/hUpf1 message, and the lower band is derived from the transgene. (C) Northern blot showing the steady-state abundance of WT or nonsense (PTC) TCR-β message in fibroblasts from Rent1/hUpf1 Tg or WT mice transiently transfected with the corresponding TCR minigene constructs. Neomycin resistance gene mRNA, encoded on the same plasmid, served as a control for loading and transfection efficiency. Neo, neomycin resistance gene. Ratio of normalized nonsense-to-WT TCR transcript levels is shown. (D) Northern blot analysis of endogenous thymus β-glucuronidase mRNA from gus^mps/gus^mps mice that did (TG) or did not (WT) carry the Rent1/hUpf1 transgene. Endogenous β-actin mRNA served as a loading control. β-gluc, β-glucuronidase.

Disruption of Fetal Thymic Development in Rent1/hUpf1 Tg Mice.

A dedicated histopathological analysis did not reveal gross or microscopic abnormalities in any tissues of mice harboring the R844C Tg allele, with the exception of the thymus. We hypothesized that the developing thymus would be particularly susceptible to deleterious consequences of NMD inhibition because of the high physiological burden of nonsense transcripts derived from the TCR-β locus. Indeed, we observed reduced numbers of thymocytes in fetal Tg animals relative to their WT littermates (7.8 ± 0.9 × 10⁵ vs. 25.0 ± 3.9 × 10⁵, respectively, at Fd17; P = 0.0008) at a developmental stage coincident with the onset of TCR-β gene rearrangement at approximately Fd16 in the mouse (27–30).

During fetal thymic ontogeny, large numbers of thymocytes pass through a series of well-defined developmental stages in synchrony. The successive stages of T-cell maturation can be defined by expression of surface markers (Fig. 2A). Flow cytometric analysis with CD4- and CD8-specific antibodies revealed a reduction of DP thymocytes in Tg animals at Fd16 and Fd17 (Fig. 2B). The majority of thymocytes in Tg animals demonstrated impaired transition from CD25^high to CD25^low and reduced expression of TCR-β (Fig. 2 C and D), similar to findings previously reported in mice incompetent for pre-TCR synthesis or signaling (31–37).

Fig. 2.

Abnormal T-cell development in fetal Rent1/hUpf1 dominant-negative Tg (TG) mice. (A) Normal thymocyte maturation. Thymocytes DN for CD4CD8 can be further subdivided into four discrete stages of differentiation defined by expression of CD25 and CD44 as shown. Only cells that express a functional TCR down-regulate CD25, up-regulate CD4 and CD8, and undergo selection. Flow cytometric analysis of thymocytes from fetal (at the indicated Fds) WT or Rent1/hUpf1 Tg mice with antibodies to CD4 and CD8 (B), CD25 and CD44 (C), TCR-β (D), and TCR-γδ (E). (Insets) Percentage of gated cells in each quadrant from a representative litter is given within each panel.

During T-cell development, precursor cells choose between αβ and γδ lineages. The decision to enter either the αβ or γδ pathway occurs at the DN stage of thymocyte maturation. We stained Fd16 thymocytes using an antibody directed against γδ TCR and observed no significant difference in the percentage of γδ-positive cells in Tg animals relative to WT controls (Fig. 2E). Taken together, these data suggest that Rent1/hUpf1 dominant-negative activity is associated with a block in αβ lineage development as a consequence of impaired pre-TCR function in fetal thymocytes. No increase in the γδ T-cell population is evident at this developmental stage, suggesting that the immature thymocytes accumulating as a result of Rent1/hUpf1 suppression are not shunted toward the γδ lineage.

Reduced Efficiency of NMD Disrupts Fetal Thymic Architecture.

Histological examination of thymic architecture in Rent1/hUpf1 Tg mice revealed distinct zones of cellular dropout that were first seen at Fd17 and were unique to Tg thymi (Fig. 3A). By Fd19, these lesions were more numerous, larger in size, diffusely distributed throughout the cortex, and associated with intense TUNEL staining at their periphery (Fig. 3 B and C). These data suggest that there is a population of cells in the Tg thymus that is able to remain viable and expand at very early stages of fetal development but then undergoes crisis after clonal proliferation in the cortex. Expression of CD4CD8, CD25CD44, and TCR-β in Tg thymocytes approaches WT levels as cellular dropout progresses in the Tg thymus. In adults, the thymocyte profiles for Tg and WT animals are nearly indistinguishable (Fig. S3 A–C). No histological differences were observed between adult Tg and WT thymi (Fig. S3D).

Fig. 3.

Distortion of thymic architecture in Rent1/hUpf1 dominant-negative mice. H&E stain of WT and Tg (TG) thymic lobes from Fd 7 (A) and Fd19 (B) mice. (Magnification: A, 40×; B, 10×.) (C) TUNEL assay for DNA fragmentation on histological sections from Fd19 WT and Tg thymus. (Magnification: 20×.)

Productively Pre-Rearranged TCR Allele Rescues the Phenotype Observed in Rent1/hUpf1 Tg Animals.

Although our data were consistent with a deleterious consequence for stabilized TCR-β nonsense transcripts, a nonspecific toxic effect of NMD inhibition on developing thymocytes could not be excluded. To address this issue, we introduced a fully rearranged TCR-2C transgene (38) that prevents endogenous TCR alleles from recombining by allelic exclusion, and thereby precludes nonproductive rearrangements and the generation of nonsense transcripts (39–41). Expression of the TCR-2C transgene completely restored CD4CD8 expression, CD25CD44 expression, and thymic cellularity in fetal Tg mice (Fig. 4 and Fig. S4), suggesting that accumulation of out-of-frame TCR-β transcripts is either directly or indirectly responsible for the thymic phenotype observed in Rent1/hUpf1 Tg mice.

Fig. 4.

Rescue of thymocyte development by the productively rearranged TCR-2C transgene. Flow cytometry of Rent1/hUpf1 Tg (TG) and WT thymocytes using CD4 and CD8 antibodies in the presence (+TCR-2C) or absence (no TCR-2C) of the pre-rearranged TCR-2C transgene at Fd17. (Insets) Percentage of gated cells in each quadrant from a representative litter is given within each panel.

Rent1/hUpf1 Transgene Leads to an Overrepresentation of Out-of-Frame TCR-β Alleles and a Reduction in Vβ-to-DβJβ Rearrangements but Not Dβ-to-Jβ Rearrangements in Fetal Thymi.

To determine the relationship between phenotypic abnormalities in the Rent1/hUpf1 Tg mice and the rearrangement status at the TCR-β locus further, we determined the representation of in- and out-of-frame alleles in WT and Tg mice using a previously described PCR and sequencing strategy (22). We observed a dramatic overrepresentation of β⁻ alleles in the Tg thymus at Fd16, compared with WT littermates (Table S1; 46% vs. 22%, respectively; P = 0.02). This difference was not seen in adult animals (Table S1; 25% vs. 30%, respectively; P > 0.4). A number of possibilities were considered. First, suppression of Rent1/hUpf1 might somehow cause prolonged survival of β⁻β⁻ cells, leading to relative overrepresentation of β⁻ alleles. However, we observed decreased cell number and increased cell death. Second, inhibition of Rent1/hUpf1 might impair pre-TCR–mediated proliferation of β⁺β⁰ cells, again resulting in apparent enrichment of β⁻ alleles. In contrast to this model, we observed that the rate of cellular proliferation is identical in WT and Tg thymi (Fig. S5). A final possibility was that suppression of Rent1/hUpf1 perturbs progression of TCR-β allele rearrangement in β⁻β⁰ cells. In keeping with this hypothesis, fetal Tg thymocytes demonstrated a markedly diminished frequency of Vβ-to-DβJβ rearrangements (Fig. 5). This difference was most striking at Fd16, still present but less pronounced at Fd19, and not apparent in 1-wk-old or adult Tg mice (Fig. 5 and Fig. S6). An equal frequency of Dβ-to-Jβ rearrangements in Tg and WT thymocytes suggested that the discrepancy in Vβ-to-DβJβ rearrangements is not simply a consequence of reduced thymic cellularity in Tg animals. These data suggest that inhibition of Rent1/hUpf1 either blocks thymocyte development before the onset of Vβ-to-DβJβ rearrangement or that it induces a premature shut-off of TCR-β rearrangement.

Fig. 5.

Diminished endogenous TCR-β gene rearrangements in Rent1/hUpf1 Tg (TG) mice. Southern blot analysis of PCR products derived from amplification of WT and Rent1/hUpf1 Tg thymocyte DNA from Fd16 or 7-d-old mice using a sense primer in Vβ 8, 10, or 11 (to detect Vβ-to-DβJβ rearrangements) or upstream of Dβ2 (to detect Dβ-to-Jβ rearrangements) and an antisense primer located downstream of Jβ2.7. Only rearranged alleles can be amplified using the Vβ and Jβ2.7 primer pairs, with each specific band corresponding to the particular J segment (1–5, 7) involved. Both unrearranged (indicated with arrow) and rearranged segments can be detected with the Dβ2 and Jβ2.7 primer set. The blot was hybridized with a probe specific for Jβ2.7. A segment of the Cμ gene was amplified as a loading control. Ratios of rearrangements of Rent1/hUpf1 Tg to WT Vβ-to-DβJβ (normalized for the amount of Dβ-to-Jβ rearrangements) are shown.

Rent1/hUpf1 Transgene Leads to Impaired B-Cell Maturation.

Like TCR-β, Ig heavy chain (HC) genes also undergo a series of programmed gene rearrangements that frequently results in production of nonsense transcripts normally degraded by the NMD pathway. These rearrangements occur at the CD19⁺ c-kit⁺ pro–B-cell stage. Once a productive HC gene has been generated, the μHC is expressed and triggers the transition from the μHC-negative pro–B-cell stage to the μHC-positive pre–B-cell stage of development. The fetal liver is the major site of B-cell development during embryonal life. Therefore, we analyzed the livers of newborn mice and observed an increased proportion of c-kit⁺ pro-B cells in the CD19⁺ B-cell populations of Tg mice compared with WT littermates (30.8% ± 0.9% vs. 19.8% ± 2.5%, respectively; P ≤ 0.0001; Fig. S7). Of these pro-B cells, slightly fewer cells produced an intracellular μHC in the Tg mice (8.9% ± 1.6% vs. 11.5% ± 1.7%; P = 0.0293), indicating a developmental impairment before or during VDJ recombination rather than an impairment in the differentiation of μHC⁺ pro-B to pre-B cells. As a consequence, Tg mice had fewer surface IgM⁺ cells (6.3% ± 1.6% vs. 9.2% ± 1.3%; P = 0.0046). These data suggest that deficiency of NMD impairs early B-cell development coincident with the timing of nonsense transcript production. Similar results are reported in the accompanying paper by Lutz et al. (42), which shows impaired B-cell development in mice expressing nonsense μHC transcripts that cannot be degraded by NMD.

Mice Expressing Truncated TCR-β Protein Do Not Recapitulate Any Aspects of the Rent1/hUpf1 Tg Phenotype.

The random addition and subtraction of nucleotides at the VβDβ and DβJβ junctions result in the generation of a PTC shortly downstream in the TCR-β message in two of three rearrangements. Normally, these transcripts are efficiently degraded by the NMD pathway. However, the truncated proteins these mRNAs encode might be expected to increase in abundance when NMD is disrupted, and therefore may contribute to the phenotype observed in the Rent1/hUpf1 Tg mice. To test this hypothesis, Tg mice [called leader (L) VDJ] were created that would express a truncated protein encoded by a construct containing only the L and rearranged VDJ exons of TCR-β (Figs. S1 and (S8A). This is representative of the truncated protein that would be predicted to be expressed from a stabilized nonsense TCR-β message. Mice expressing the LVDJ transgene showed no changes in CD4CD8, CD25CD44, or TCR-β expression in fetal or adult thymocytes compared with their WT littermates (Fig. S8 B–D). RT-PCR documented expression of the transgene in LVDJ mice at Fd17 (Fig. S8E). Therefore, expression of a stable truncated RNA that is fully translationally competent to express truncated TCR-β peptides representative of those encoded by nonproductively rearranged TCR-β alleles is insufficient to recapitulate the phenotype observed in Rent1/hUpf1 Tg mice. These data make it highly unlikely that the phenotype observed in Rent1/hUpf1 mice is a result of truncated TCR-β proteins that accumulate as a result of disruption of NMD.

Discussion

The NMD pathway has previously been recognized to play a central role in degrading out-of-frame TCR-β transcripts (23). However, the consequences of inhibiting this function in vivo are largely unknown. Initial attempts to address this question by targeted silencing of Rent1/hUpf1, an essential trans-effector of the pathway, were noninformative because complete loss of function was incompatible with embryonic viability (20). Conditional deletion of Upf2 revealed that loss of NMD results in death of T cells harboring β⁻ but not β⁺ alleles; however, the underlying mechanism responsible for this observation is not known. We have generated Tg mice with reduced efficiency of NMD by overexpressing a dominant-negative form of Rent1/hUpf1. These mice were viable, and a subset of fetal thymocytes in Tg animals arrests at a stage of development consistent with a defect in pre-TCR signaling, as evidenced by an increased population of CD4CD8 DN cells that express high levels of CD25 and low levels of TCR-β.

There are many lines of evidence to suggest that stabilized TCR-β nonsense transcripts contribute to the impaired thymocyte maturation and expansion seen in the Rent1/hUpf1 Tg mice. First, abnormal thymocytes in Rent1/hUpf1 animals arrested just before the pre-TCR stage of development (CD25^highCD44^negativeCD4CD8^DN). This correlates temporally with the onset of TCR-β rearrangement, an event known to initiate production of a high physiological burden of substrates for the NMD pathway (23). Second, it would be predicted that a nonspecific toxic effect imposed by NMD deficiency would affect all thymocytes equally. In contrast, histological examination of fetal Rent1/hUpf1 Tg thymi revealed discrete zones of cellular dropout, plausibly representing populations of β⁻β⁰ cells, which have an initial nonproductive rearrangement resulting in the production of potentially deleterious out-of-frame messages that are now stable because of inhibition of NMD. β⁺β⁰ cells, which productively rearrange a TCR-β allele on the first try, and therefore do not express any out-of-frame TCR-β message, would be exempt from any deleterious effect of NMD inhibition. Third, we see an increase in the frequency of out-of-frame alleles in fetal Tg thymi despite documentation that surviving thymocytes proliferate normally. There is no obvious explanation if one invokes a generalized and nonspecific toxic effect of NMD inhibition. Finally, arrest at the pre-TCR stage of development can be rescued by breeding a functionally pre-rearranged TCR-β allele onto the Tg background, presumably attributable to the absence of TCR-β nonsense transcripts because rearrangement of both endogenous alleles is inhibited through allelic exclusion.

The mechanisms regulating rearrangement at the TCR-β locus as well as thymocyte maturation and expansion remain an area of active research. We observed that thymocytes from fetal NMD-deficient mice demonstrated a normal frequency of Dβ-to-Jβ rearrangements, which are not subject to allelic exclusion, but reduced Vβ-to-DβJβ rearrangements, which are subject to allelic exclusion. Our data therefore suggest that stabilized TCR-β nonsense transcripts that only encode the variable domain, which cannot support pre-TCR assembly or signaling, may contribute to allelic exclusion. Although we cannot exclude the possibility that these effects may be mediated by other functions of the NMD pathway or Rent1/hUpf1 itself, this model is supported by recent observations that stable but nonproductive μHC transcripts can suppress VDJ recombination in pro-B cells (42). A prior study had failed to observe phenotypic consequence after expression of a TCR-β message with a physiological frameshift mutation in the DβJβ region that could not derive functional TCR-β protein (43). However, because of the downstream PTC, this message is predicted to be recognized and degraded by the NMD pathway. Recently, Schlimgen et al. (44) reported that initiation of monoallelic Vβ-to-DβJβ recombination events in developing thymocytes results from stochastic interactions of Tcrb alleles with repressive nuclear compartments. We propose that stable TCR-β RNA may delay further recombination until the outcome of rearrangement on the first allele is tested, reconciling the time lag paradox inherent to many prior models. In this view, allelic exclusion is a multiphasic process that is initiated by interaction of Tcrb alleles with repressive nuclear compartments, maintained transiently by stable TCR-β messages, and consolidated by pre-TCR signaling. Normalization of the thymic phenotype in adult dominant-negative Rent1/hUpf1 Tg animals indicates that this function of NMD may only be operative or discernible in fetal thymocytes undergoing maturation in a synchronous wave.

Although many studies have established an essential role for the mature TCR-β chain and TCR-β signaling in regulating TCR-β rearrangement and allelic exclusion, this study implicates a role for TCR-β message and NMD in these events. This model establishes a new paradigm that mRNAs can have activities independent of their contribution to protein production. Although well established for noncoding RNAs, this represents a new concept for RNAs that also encode protein products. An unanticipated function of NMD, therefore, may be to safeguard the intended correlation between mRNA and protein functions in these exceptional circumstances.

Materials and Methods

Mice.

The Rent1/hUpf1 Tg mice express the mutated (R844C) form of human Rent1/hUpf1 (24) from the β-actin promoter. The LVDJ mice express a transgene consisting of the L and rearranged VDJ exons of TCR-β from the lck promoter. Details of plasmid construction and genotyping methods can be found in SI Materials and Methods. Mice heterozygous for the gus^mps mutation (B6.C-H2^bml/ByBirgus^mps/+) were purchased from the Jackson Laboratory. Gus^mps and TCR-2C mice were genotyped as previously described (20) and by flow cytometry, respectively. All mice used in this study were bred and maintained under pathogen-free conditions. Experiments were approved by the Animal Care and Use Committee.

Cell Lines, Transfection Conditions, RNA Isolation, and Northern Blot Analysis.

Transfected fibroblast cell lines established from Fd13.5 Tg and WT embryos, and thymic tissue retrieved from Gus^mps mice, were analyzed by Northern blot analysis as described in SI Materials and Methods.

RT-PCR.

cDNAs from multiple tissues of Rent1/hUpf1 Tg adult mice, Fd16 thymi, and human and mouse fibroblast cell lines were used to amplify Rent1/hUpf1. Amplicons were digested with MboI and analyzed by Southern blotting as described in SI Materials and Methods. A similar approach was used to evaluate LVDJ transgene expression.

Flow Cytometry.

Pregnant females from matings between the two lines of Tg animals (Rent1/hUpf1 and LVDJ) and either C57.BL6 or TCR-2C mice were killed at various days postcoitum. Single-cell suspensions prepared from thymi or fetal liver were stained and analyzed as described in SI Materials and Methods. Multiple mice of each genotype were evaluated.

Histology.

Thymi were stained with H&E, and the TUNEL assay was performed per the manufacturer's instructions (Boehringer Mannheim) as described in SI Materials and Methods.

Frame Assay.

A PCR assay utilizing 500 ng of genomic thymocyte DNA isolated using the QIAamp DNA mini kit (Qiagen) was performed using Vβ2 and Jβ2.2 primers and conditions as described previously (22). PCR products were cloned (TOPO TA 2.1 kit; Invitrogen) and sequenced to determine frame. At least 45 unique sequences were evaluated to assess frame.

TCR-β Rearrangement Assay.

TCR-β DJ and VDJ gene rearrangements were determined by semiquantitative PCR analyses using primers and conditions as previously described (45).

Statistics.

Statistical comparisons were done using the Mann–Whitney U statistic, Student t test, or χ² test (Prism software; GraphPad). P values less than 0.05 were considered significant.