Severe Defect in Thymic Development in an Insertional Mutant Mouse Model¹

The thymus plays a central role in the immune system. It provides an architecturally organized microenvironment for the development of mature T cells. T cell development is dependent on a two-way interaction between thymocytes and thymic stromal cells (reviewed in Refs. 1 and 2). The thymic stroma comprises thymic epithelial cells (TECs),⁴ macrophages, dendritic cells, and fibroblasts. The thymus is divided into two main compartments, the cortex and the medulla, defined by morphology and distribution of specific TEC subsets (3, 4). Different TEC subsets generate discrete microenvironments that are specialized for different steps in T cell development and selection. Therefore, thymocytes at different stages of differentiation reside in spatially restricted domains of the thymus (5, 6).

A number of spontaneous as well as induced mouse mutants have been characterized by having defects in thymus development. For example, nude (nu/nu) mice have a homozygous gene mutation in the transcription factor Foxn1 known to regulate TEC differentiation (7, 8). This mutation results in an early arrest in TEC expansion and defective recruitment of T cell precursors, which completely prevents the production of functional T cells. Transgenic (tg) mice expressing a human CD3ε transgene (tgε26) have a defect in the lineage choice between B and T cells leading to an early arrest of T cell development combined with an accumulation of B cells in the thymus (9, 10). As a consequence, these mice have a hypoplastic thymus with disorganized architecture (11). These mutations illustrate the reciprocal interactions between TECs and the developing thymocytes required for normal thymus and T cell development. Several other transcription factors are known to contribute to thymus organogenesis. Null mutations of the Hoxa3 (12, 13) or Eya1 (14) genes lead to lack of the parathyroid glands and thymus. Furthermore, Pax1- (15) and Pax9-deficient mice (16) both display thymic hypoplasia. Differentiation of the medullary TEC subset has been shown to be regulated by avian reticulo endotheliosis viral oncogene-related B, TNFR-associated factor 6 (TRAF6), lymphotoxin β receptor, and NF-κB-inducing kinase (17).

In the present study, a new transgenic mouse strain (Tg66) with a severe thymic defect is described. Thymi from these mice have abnormal architecture and are reduced in size. The corticomedullary junction (CMJ) is indistinct and there is a preferential loss of cortical TECs. Consequently, T cell production is severely impaired. The defect resides primarily in nonhemopoietic cells, because bone marrow from the tg mice readily repopulated thymi of irradiated wild-type (wt) as well as RAG-deficient mice. Furthermore, the thymic phenotype is not caused by the transgene expression per se. Severity is dependent on whether the mice carry integration on one or two chromosomes. Two integration sites lead to an earlier and more severe thymic phenotype, consistent with a semidominant effect. These data are indicative of a random transgene insertion in the genome that alters expression of one or more genes important for thymic development and/or maintenance. In this study, we provide a detailed characterization of heterozygous animals.

C57BL/6 (B6), B6 RAG-1^−/− mice (The Jackson Laboratory), and B6 RAG-2^−/− × B6 common γ-chain^−/− mice (Taconic Farms), here denoted as RAG × γC^−/−, were bred and housed under standard conditions at the Department of Cell and Molecular Biology and Department of Microbiology, Tumor Biology and Cell Biology (Karolinska Institutet, Stockholm, Sweden). All animal procedures were approved by the local departments as well as by the Committee for Animal Ethics (Stockholm, Sweden).

A 695-bp NK1.1 (NKRP-1C) cDNA was cloned from IL-2-activated B6 NK cells into the VACD2 vector (18) using the following primers: upper, ATAGAATTCGCACCATGGACACAGCAAGTATCTACCTCG and lower, GTGGAATTCATGGGATTCGCAGTCAGGAGTCATTACTCG. The fragment for microinjection was isolated from the vector by restriction with KpnI and NotI followed by gel purification. Fractions were collected and analyzed by pulse-field gel electrophoresis to assess the purity of the insert. A 12.3-kb fragment was used in pronuclear injection of B6 embryos to produce VACD2.NK1.1 tg mice. Tg founder mice were identified by staining PBLs with anti-NK1.1 and anti-TCRαβ Ab and analyzed by flow cytometry and/or by PCR using the primers described above. When not else noted, all experiments in this study were performed with heterozygous Tg66 mice and non-tg littermates as controls (referred to as wt).

All reagents were purchased from DakoCytomation unless otherwise specified. Anti-cytokeratin (CK) 5 Ab was purchased from Babco, anti-CK14 Ab was purchased from Biosite, and goat anti-rat Alexa 488 and goat anti-rabbit Alexa 568 were purchased from Molecular Probes. Anti-CK8 Ab (Troma1) was a gift from Dr. I. Mikaelian (The Jackson Laboratory) (19).

Organs and tissues were fixed in neutral-buffered formalin for 24 h at room temperature, dehydrated, and embedded in paraffin according to standard procedures. Four-micrometer-thick sections were either stained by H&E or processed for immunohistochemistry. For preparation of frozen sections, unfixed thymi were embedded in OCT compound (Miles), snap-frozen in liquid N₂, and stored at −80°C until sectioning in a cryostat. Seven-micrometer-thick sections were collected on Plus slides (Fisher Scientific) and processed for immunofluorescent staining. Paraffin sections were deparaffinized, Ag unmasked by microwaving 10 min in citrate buffer (pH 6), rinsed in TBS buffer (pH 7.6), with 0.1% Tween 20), and blocked in 0.1% BSA. After incubation with Avidin/Biotin Blocking kit (Vector Laboratories), the sections were further incubated overnight with the primary Ab at 4°C. For single stainings, sections were incubated with a biotin-labeled secondary Ab. After incubation with streptavidin-HRP, sections were developed with diaminobenzidine/H₂O₂ and counterstained with hematoxylin. For double stainings, sections were incubated with goat anti-rabbit Alexa 594 and a donkey anti-rat biotin followed by streptavidin-Alexa 488. Double-stained sections were examined using a Leica TCS NR ArKr Laser Confocal Microscope (Leica Microsystems).

Six- to 8-wk-old mice (wt, Tg66, or RAG × γC^−/−) were sublethally irradiated with 500 rad followed by an additional 600 rad 8 h later and injected with 10⁶ bone marrow cells (from wt or Tg66 mice) i.v. in the tail vein. Six weeks later, thymi and spleens were analyzed macroscopically, by histology and by flow cytometry. Thymocytes were stained for NK1.1 to confirm chimerism when appropriate.

Primary cultures were set up from Tg66 embryos. Eight to 10 days postcoitum (dpc), embryos were minced and the cell suspension was filtered and explanted. Chromosomes were prepared from first passage cultures (5 days postexplant) and blocked by colcemide according to the standard cytogenetic procedure. The VACD2.NK1.1 construct was linearized, labeled with biotin-14-dATP using a BioNick Labeling System (Invitrogen Life Technologies), and used as a probe. FISH was performed as previously described (20). The hybridization signal was detected using avidin-FITC conjugate (Vector Laboratories) and the carrier chromosome (Chr) 2 was identified by 4′,6′-diamidino-2-phenylindole (DAPI) banding. The FISH was repeated in a dual color system with the same probe and a rhodamine-labeled mouse Chr 2-specific painting probe (Cambio).

All Ab were purchased from BD Pharmingen. FcRs were blocked to exclude nonspecific staining by incubating the cells with anti-CD16/32 Ab, after which the cells were stained with the specified Ab or its isotype control at 10 μg/ml. The fluorescence intensity was measured on a FACScan flow cytometer (BD Biosciences) and analyzed using CellQuest computer software (BD Biosciences).

Peptide (100 μg) was dissolved in 100 μl of H₂O, mixed with 100 μl of IFA by sonication and injected s.c. at the base of the tail. Cytotoxicity assays: on day 10 postimmunization, 25 × 10⁶ immune splenocytes were cultured in complete medium (MEM, 10 mM HEPES, 2 × 10⁻⁵ M 2-ME, 10% FCS, 100 U/ml penicillin, 100 U/ml streptomycin) and 1 μg/ml peptide. Five days after restimulation, these cells were used as effectors in a standard 4-h ⁵¹Cr release assay. RMA-S cells pulsed with peptide (as indicated) were used as target cells. Anti-CD3-induced proliferation assays: CD4⁺ and CD8⁺ splenocytes were sorted according to a MACS purification system (Miltenyi Biotec), CFSE labeled, and cultured at 10⁵ cells/well in six-well plates precoated with anti-CD3 Ab (1 μg/ml) for 3 days and analyzed by flow cytometry. For intracellular IFN-γ staining, mice were immunized as above and splenocytes were cultured in the presence of peptide as indicated for 6 h in the presence of monensin (BD Pharmingen) after which IFN-γ production was detected using an Intracellular Cytokine Staining kit (BD Pharmingen) and analyzed by flow cytometry.

TRIzol reagent (Invitrogen Life Technologies) was used to extract total RNA from thymus from 9-day-old wt and Tg66 mice according to the manufacturer’s recommendations. The TaqMan Reverse Transcription Reagent kit (Applied Biosystems) with random hexamers was used for cDNA synthesis according to the manufacturer’s instruction.

PCR amplification for Pax1 was performed in a 50-μl volume containing 10 pM of each primer, 1× PCR buffer, 2.5 mM MgCl₂, 0.5 mM dNTPs, and 1.25 U of Taq polymerase (Invitrogen Life Technologies). Primers 5′-GCTGCCTACTCCCCCAAGA and 5′-CGCTGTATACTCCGTGCTG were used to amplify Pax1 and primers 5′ATTGTTGCCATCAACGACCCCTTC and 5′GTTGCTGTTGAAGTCACAGGAGAC were used to amplify G3PDH.

Quantitative PCR was performed on an ABI PRISM 7700 Sequence Detector using the ABI PRISM 7700 Sequence Detector Software 1.9.1 (Applied Biosystems). For each run, 50 cycles of a two-step PCR amplification (15 s, 95°C; 1 min, 60°C) were conducted after initial UNG enzyme inactivation (2 min, 50°C) and DNA polymerase activation (10 min, 95°C): all reactions were performed in MicroAmp Optical 96-well Reaction Plates with MicroAmp optical caps (Applied Biosystems). The amplification reactions were performed in triplicates in 25-μl reaction volumes with TaqMan Universal PCR Master Mix (Applied Biosystems), TaqMan Primer-probe for 18S rRNA endogenous control VIC/TAMRA Probe or mPax1 FAM (Applied Biosystems). Comparative cycle threshold (C_T) method was used for relative quantification of Pax1 mRNA in thymus tissue of 9-day-old Tg66 mouse compared with wt.

The Student two-sample t test was used assuming unequal variances when comparing the data derived from different mice within each experiment.

To study the role of the NK cell-associated receptor, NK1.1 (NKR-P1C), on T cells during development and in the periphery, we expressed NK1.1 cDNA under the control of a human CD2 promoter. Following pronuclear injections, six positive founders were identified that exhibited varying transgene expression levels on expected lymphocyte subsets, predominantly on T cells and NK cells (Fig. 1; data not shown). Three founders were crossed with B6 mice to establish lines of tg mice. All matings produced offspring with germline transmission. Mice from these three lines reproduced normally and showed no signs of generalized disease. However, one founder line, Tg66, had fewer peripheral T cells than the wt and the other founder lines, Tg51 and Tg71 (Fig. 1). Upon further examination, the thymus of the Tg66 mice was markedly reduced. Because, only Tg66 exhibited this phenotype, it was likely that transgene integration was the cause of the defect observed. To localize the integration site, FISH analysis was performed. This revealed a single integration site on Chr 2, band G2 (Fig. 2,A). The chromosomal localization was confirmed using Chr 2-specific paint probes (Fig. 2 B). The unexpected phenotype of Tg66 mice and the localization of the integrated tg to Chr 2 prompted us to study the development of the thymus and the distribution and phenotype of T cells in different lymphoid compartments of this line.

To follow postnatal thymus and T cell development, we performed necropsies of Tg66 mice at several time points from birth up to 16 mo of age. When observed macroscopically, wt and Tg66 mice had thymi of similar size up to 11 days of age (Fig. 3, A, B, D, and E). Strikingly though, by 28 days of age, thymi of Tg66 mice were greatly reduced in size (Fig. 3, C and F). Cell counts of dissociated thymi revealed a progressive loss of cells in Tg66 thymi occurring mainly between days 11 and 19 after birth (Fig. 4 M). On day 19, Tg66 thymi had, on average, only 23 × 10⁶ (±11 × 10⁶) cells compared with 167 × 10⁶ (±18 × 10⁶) in wt thymi (n = 3). In contrast, thymocyte counts of the founder lines Tg51 and Tg71 revealed cell numbers similar to that of wt mice (data not shown).

Histological analysis revealed progressive changes in Tg66 thymi resulting in poorly defined cortical and medullary structures seen as early as 3 days of age (Fig. 3, G–L). This characteristic became even more pronounced with increasing age. By 28 days of age, the thymi of Tg66 mice were severely atrophied, and cortical and medullary areas were no longer discernable (Fig. 3,L). The morphological changes were virtually the same in thymi isolated from animals up to 16 mo of age (data not shown). In contrast, histopathologic examination of the founder lines Tg51 and Tg71 up to 15 mo of age revealed a normal thymic architecture (Fig. 3,L, inset; data not shown). Because the surface expression of the transgene (NK1.1) was similar or even higher in Tg51 and Tg71 mice (Fig. 1), it was likely that the Tg66 phenotype was caused not by transgene expression per se, but rather by transgene insertion at a critical locus (Chr 2, band G; see above).

To study additional transgene effects that might occur later in life, we performed necropsies and more extensive histopathological analyses of Tg66 mice up to 16 mo of age. However, no changes were visible in other organs, including the spleen, lymph node (LN), liver, lungs, kidneys, intestine and skin when compared with those of age-matched wt mice (data not shown). No sign of autoimmune reactions was found either.

To study whether the thymic defect affected T cell development, the expression of CD4 and CD8 on thymocytes was determined by flow cytometry. Thymi of day 9 and day 16 Tg66 mice had dramatically decreased frequencies of CD4⁺ single-positive (SP) thymocytes compared with wt thymi (Fig. 4, B, C, E, and F). In contrast, the frequency of CD8⁺ SP thymocytes was affected less at these ages. CD3 expression levels (mean fluorescence intensity (MFI)) on the CD8⁺ and CD4⁺ SP thymocytes were reduced by 30 and 42%, respectively, in the Tg66 mice (Fig. 4,N). This comparison showed that SP thymocytes from Tg66 mice were more immature than those from wt mice (21). In addition, at all time points studied, a low CD4 expression remained on the CD8⁺ SP thymocytes from Tg66 mice (Fig. 4, D–F).

To further analyze the early steps in T cell development, immature triple-negative (TN; TCRβ⁻CD4⁻CD8⁻) thymocytes were analyzed for the expression of CD44 and CD25 (IL-2Rα). Although CD25/CD44 expression was roughly the same in thymi from wt and Tg66 mice on day 3 (Fig. 4, G and J), by day 9, the CD44⁺CD25⁻ (TN1) and CD44⁺CD25^low (TN1-TN2) populations were slightly increased in Tg66 thymi (Fig. 4, H and K). On day 18, the Tg66 thymocytes were equally divided into TN1 (CD44⁺CD25⁻) and TN4 (CD44⁻CD25⁻) populations (Fig. 4,L), whereas the wt thymocytes were mainly of the TN2-TN3 or TN4 phenotypes (Fig. 4,I). This CD25/CD44 expression pattern remained stable up to 6 wk of age in the thymi of wt and Tg66 mice, and correlated well with the progressive morphological disturbances observed (Fig. 3, G–L). Overall, these results indicate a partial block at an early stage of T cell development in the Tg66 mice. However, it cannot be excluded that selective cell death during certain developmental stages may have altered the proportions of some subsets due to the rapid decrease in thymocyte numbers.

The tissue distribution of CD25- and CD44-expressing cells in the thymus was determined by immunohistochemistry on day 28. As previously described (5), in wt thymi, CD25-expressing cells were dominant in the subcapsular zone, and CD44-expressing cells were mostly present in the CMJ and in the medulla. In contrast, in the Tg66 mice, all CD44⁺ cells as well as a small number of CD25⁺ cells were more evenly distributed in the thymic parenchyma (data not shown).

Although the thymus is the major site for T cell development, some T cells do not require thymus for their development. These thymus-independent T cells develop mainly in the intestine and liver and include certain subsets of TCRαβ⁺ and γδ⁺ T cells (22). It has previously been shown that the numbers of thymus-independent T cells increase in parallel with thymic involution (23, 24). Therefore, it was interesting to study how different T cell subsets were affected by the thymic phenotype described in this study. By 28 days of age, the total number and frequency of TCRαβ⁺ T cells in the spleens of Tg66 mice was reduced by 55 and 70%, respectively, compared with those of wt mice (Table I). In contrast, the TCRγδ⁺ T cells seemed less affected by the thymic defect, because the frequency remained the same in the spleens of wt and Tg66 at that age. Furthermore, the number of TCRγδ⁺ T cells in the ear epidermis of the Tg66 mice did not differ from that of wt mice (data not shown). Unlike TCRγδ⁺ T cells, the frequency and total numbers of invariant NKT cells (defined as αGalCer-loaded CD1d-dimer⁺B220⁻) were clearly reduced in the spleens of Tg66 mice by day 28 of age compared with that of wt mice (Table I). This suggested that different T cell populations were differently affected by the thymus defect.

Despite the disrupted architecture seen by day 3 of age and the progressive reduction in thymocyte numbers, older Tg66 mice (6–8 wk) retained a fairly normal T cell repertoire in the periphery. When CD4⁺ and CD8⁺ TCR Vβ and Vα expression was analyzed by flow cytometry, no notable change was evident in TCR usage by Tg66 mice compared with wt mice (data not shown). To address potential phenotypic changes of the peripheral T cells from Tg66 mice, the surface expression of a number of memory-associated molecules was determined. On day 10, no notable change was evident in the surface expression of CD44, CD62L or CD122 in either spleens or mesenteric LN. However, by 5–6 wk of age, Tg66 mice had a significant increase in the fraction of CD44⁺ or CD122⁺CD8⁺ T cells in the spleens and LN compared with those of wt mice (data not shown). Even more evident, by 3 mo of age, the vast majority of the peripheral CD8⁺ T cells in Tg66 mice were of a memory-like phenotype, defined as CD25^lowCD44^highCD122^high and had lower TCR expression than that of wt mice (Fig. 5).

To analyze T cell function in the Tg66 mice, we measured polyclonal and Ag-specific responses. Anti-CD3-mediated stimulation of purified CD8⁺ and CD4⁺ splenocytes showed that Tg66 T cells proliferated in response to polyclonal stimuli at levels not significantly different from wt T cells (Fig. 6,A). To study Ag-specific responses, wt and Tg66 mice were s.c. immunized with a vesicular stomatitis virus-derived peptide (RGYVYQGL) or lymphocytic choriomeningitis virus-derived peptide (KAVYNFATM). Ten days later, splenocytes from the immunized mice were restimulated in vitro and Ag-specific IFN-γ release and cytotoxicity were measured. Tg66 T cells produced IFN-γ (Fig. 6,B) and killed specific peptide-loaded target cells at levels not markedly different from the wt T cells (Fig. 6 C).

To address whether other lymphocytes were affected by the deprivation of thymic function and distortion of T cell repertoire, we examined the frequencies and total numbers of NK cells and B cells. Numbers of these cells were not markedly altered in the spleens of mice between 10 days and 6 mo of age (Table I; data not shown). Thymi from 19 day-old Tg66 mice had an increased frequency of B220⁺CD3⁻ B cells compared with wt mice (3.0 ± 0.2% vs 0.2 ± 0.0%, p < 0.001), but without any significant increase in the total number of B cells (data not shown). The cytotoxic potential of NK cells was not altered either as IL-2-activated NK cells from Tg66 mice killed YAC-1 cells as efficiently as did wt NK cells (data not shown). Furthermore, the cell surface phenotypes of Tg66- and wt-derived NK cells were similar (data not shown).

Development of T cells is dependent on intimate interactions between thymocytes and TEC (reviewed in Ref. 2). To address whether the observed thymic defect primarily involved hemopoietic or nonhemopoietic lineages, bone marrow reconstitution assays were performed. Efforts to reconstitute thymi (i.e., to restore a normal size and architecture of the thymi) of irradiated Tg66 with wt bone marrow cells failed, whereas Tg66 bone marrow cells readily repopulated wt thymi and restored a relatively normal size and arcitechture of the thymi (Fig. 7, A and B). The phenotype of the T cells in wt→Tg66 chimeras was similar to that observed in the Tg66 mice, i.e., reduced TCR expression and a memory phenotype (data not shown). This indicated that the primary defect did not reside in the T cell precursors or in other hemopoietic cells. To further demonstrate that Tg66 bone marrow could reconstitute the size of the thymi, irradiated RAG × γC^−/− mice (which lack T cells and so do not have normal thymus development) were reconstituted with bone marrow from wt or Tg66 mice. Reconstituted thymi from both groups of mice showed a histologically normal architecture, with defined cortical and medullary areas (Fig. 7, C and D). Additional immunohistochemical analysis showed CK14 to be expressed solely in the medullary areas, indicating correct distribution of TEC (data not shown). In addition, proportions of CD4⁺ and CD8⁺ T cells were similar in thymi from mice reconstituted with either wt or Tg66 bone marrow (Fig. 7, E and F). These results demonstrated that Tg66 bone marrow precursor cells were competent to develop into T lymphocytes and further suggest that the major defect in Tg66 resides within thymic stromal cells, such as TEC. Notably, with respect to the enumeration of thymocyte populations, the number of thymocytes in the tg→CγRAG bone marrow chimera mice did not completely reach the levels found in wt→CγRAG bone marrow chimera mice (96 × 10⁶ ± 25 × 10⁶ vs 157 × 10⁶ ± 27 × 10⁶ cells). At this stage, we cannot fully explain these differences.

The notion that the defect causing the present thymic phenotype primarily resided in stromal cells prompted a more careful analysis of TEC subsets. For that purpose, we performed immunohistochemical analyses of thymi from animals at 13.5 dpc to 16 mo of age using Ab directed against CK5 and CK8. These markers define the major TEC populations, including the cortical (CK5⁻CK8⁺) and medullary (CK5⁺CK8⁻) subsets (25). No differences could be seen during the prepartum stages (13.5 and 17.5 dpc) of wt and Tg66 mice (data not shown). However, clear differences were visible already on postnatal day 3 (data not shown).

Subsequent double immunofluorescence stainings on paraffin sections using anti-CK5 and -CK8 Ab revealed that wt mice, as expected, had a large number of CK8⁺ TEC throughout the cortex with only a minor positive component in the medulla (Fig. 8, A–C). CK5⁺ cells dominated the medulla with a few cells scattered in the cortex. In addition (26), some double-positive (DP) TEC were focused around the CMJ. In Tg66 thymi, however, there was no clear distinction between these different TEC compartments (Fig. 8, D–F). The disorganized thymic structure observed by histology (Fig. 3, J–L) correlated well with the distribution of CK5- and CK8-expressing cells. Thymi from 3-day-old Tg66 mice displayed a dispersed pattern of a few cortical CK8⁺ cells (Fig. 8,D). CK8⁺ cells were also present to a large extent in the medulla-like areas. CK5⁺ TEC were predominant in the medulla, with extension into cortical areas. Images combining the green and red channels clearly demonstrated that CK5⁺ and CK8⁺ cells were intermixed with a substantial amount of cells expressing both CK (Fig. 8,D). By 11 days of age, thymi from Tg66 mice had lost most of the CK5⁺ and CK8⁺ cells from the peripheral cortical zone (Fig. 8 E). The fact that both CD25- and CD44-expressing cells were found in the subcapsular zone of Tg66 thymi (data not shown) suggests that the peripheral cortical zone was now dominated by thymocytes.

The structural changes seen on day 11 were even more prominent on day 28. Epithelial cells were present as a dense carpet, in which CK8⁺ cells aggregated in smaller, sometimes ring-shaped, structures. The CK14 expression pattern (Fig. 8, G–L) was nearly identical with that of CK5. Thus, the major medullary epithelial subset (CK5⁺CK8⁻CK14⁺) appeared to be the dominating component by day 28. Overall, these findings demonstrate a close to complete loss of normal areas of cortical TEC and a general destruction of the thymic architecture in Tg66 mice.

The detailed characterization of the Tg66 mice has this far focused on heterozygous mice. When mice were bred to homozygosity for the transgene insertion, an accelerated and more pronounced phenotype was observed (Fig. 8, M–O). Compared with heterozygotes (Fig. 8,N), the thymic disruption at postnatal day 9 was much more pronounced in homozygous animals (Fig. 8,O), showing a distribution similar to what was shown for 28-day-old heterozygotes (Fig. 8 L).

The currently observed thymic phenotype bear some resemblance of Pax1 mutant mice (15, 27). Because of this, we analyzed Pax1 expression in the homozygous Tg66 mice. RT-PCR analysis was performed to compare expression levels between wt and homozygous Tg66 mice. Data indicate no major difference with respect to expression of Pax1 (Fig. 9). This was further substantiated by real-time PCR analysis. The amount of Tg66 Pax1 mRNA was normalized to 18S rRNA and compared with that of wt. This analysis suggested no major difference as the ΔC_T for Tg66 compared with wt was 0.37–0.60. Next, the gene product was cloned from Tg66 thymi and the sequencing revealed 100% identity to that of wt cDNA (data not shown). Furthermore, homozygous Tg66 have no obvious skeletal phenotype as observed in Pax1 mutant mice (27).

The new tg insertional mouse mutant, characterized here, had a phenotypically abnormal thymus, which already in adolescents was characterized by severely disrupted architecture and small size. As early as postnatal day 3, the CMJ was indistinct with a preferential loss of cortical TEC, a feature that became more apparent with age. At the same time, the CD3 expression levels were significantly reduced on SP thymocytes in the Tg66 mice, indicative of a delay in the development. Analysis of CD25/CD44 expression showed that thymocyte development was partially blocked at an early stage (TN1-TN2 transition) and that total thymic numbers dramatically decreased between 10 and 20 days of age. Reconstitution experiments, in which bone marrow chimeric mice were generated, indicated that the primary defect resided in nonhemopoietic cells and that the T cell precursors were competent to repopulate the hemopoietic system, including the thymus, of recipient wt animals. The data suggest that the transgene insertion led to a block/change in regulation/transcription of one or more genes important for thymic development and/or maintenance. Interestingly, these thymic changes were seen in animals heterozygous for the insertion and were exaggerated in animals homozygous for the transgene. To our knowledge, none of the existing mutant mouse strains with a thymus phenotype replicate the findings described here.

The dramatic reduction in thymocyte numbers, seen early in life in the Tg66 strain, may result either from an increased rate of cell death as a consequence of the disorganized architecture and/or a block in import of prothymocytes into the thymus (28). Despite the early and progressive decrease of thymocytes in the Tg66 mice, the peripheral T cell pool was relatively large. This pool may have arisen from homeostatic proliferation of a small number of mature T cells exported from the thymus and/or an increase in the production of thymus-independent T cells with the onset of thymic destruction (23, 24, 29). Both scenarios are supported by the expression of memory-like markers on the majority of the peripheral Tg66 T cells, similar to that seen on homeostatically expanded CD8⁺ T cells (30), as well as on thymus-independent T cells (31, 32). The CD4⁺ peripheral T cells were more affected by the deprived thymus, because the decrease in the number of CD4⁺ T cells was more dramatic than that of CD8⁺ T cells (Table I). This correlates well with the reduced numbers of CD4⁺ SP thymocytes in the Tg66 mice from 9 days and onward (Fig. 4, E and F).

The molecular pathways that underlie the differentiation of TECs and thymus compartmentalization are unclear. Current data suggest that the different TEC subsets are derived from the same precursor cell (33, 34, 35, 36, 37, 38, 39, 40, 41). Similar to the situation in Tg66 mice, gene ablation of TNFR-associated factor (TRAF) 6 results in a disorganized CMJ and thymic atrophy. In contrast to our findings, the TRAF6^−/− mice have a disorganized distribution of the medullary TECs and a selective reduction of the medullary regions (42). These mice also show several signs of autoimmune reactions not seen in Tg66 mice. One may speculate that a reduction of the medullary compartments (as seen in the TRAF6^−/− mice) results in the maturation of autoreactive T cell clones due to deprived negative selection (29, 43), whereas a reduction in the cortical compartments (as seen in the Tg66 mice) may primarily affect positive selection. Localization of the transgene integration site makes possible a number of genetic comparisons. Numerous genes involved in thymic development or function localized to Chr 2, in the proximity of band G. Mutants in Pax1 (band H1), among them the undulated series of mice, display a thymus phenotype. However, all these mice also have extensive skeletal changes (27) not seen in homozygous Tg66. Furthermore, the Pax1 gene sequence was identical in homozygous Tg66 and wt and there did not appear to be any dramatic decrease in Pax1 expression. Keratinocyte growth factor (FGF7; band F2) is the ligand for FGFRIIIb. Genetic ablation of this receptor results in a block of thymic development on embryonic day 12.5 (44). However, the expression of keratinocyte growth factor also appeared to be normal in the Tg66 mice (data not shown). In addition to the genes discussed above, two other insertional mutants encompassing the chromosomal region, 2G, have been reported. One of them, kkt, has a phenotype similar to that of the Pax1 mutants (45). The other mutant was characterized by lymphoepithelial thymomas developing at an early age at very high penetrance (46). BMP2-deficient embryos die in utero before day 9, precluding studies on its effect on thymic development. Recent results suggest that termination of BMP2/4 signaling is important for differentiation of DN1 thymocytes (47). Recently, the pre- and postnatal effect of BMP signaling interruption was characterized in a tg model that expressed noggin under the Foxn1 promoter. Noggin expression in TEC led to a small ectopically located thymus (48). However, Tg66 mice have relatively normal expression of BMP2 and identical sequence (data not shown), suggesting that the phenotype of Tg66 was not due to loss of BMP2 expression. Similar to our results, CCR7^−/− thymi are reduced in size and cellularity and exhibit deprived structure as well as a delayed thymocyte development (26). However, all of these characteristics were much more pronounced in the Tg66 mice. IL-7-deficient mice have been reported to have a 20-fold reduced cellularity in thymus (49). These mice, however, have normal distribution of the different thymocyte subsets as well as normal cortical and medullary areas; thus, they differ clearly from the Tg66 mice. Thus, while several genes and genetic regions are relevant to our new insertional mutation, none of them replicates the findings we have described. A search of the Mouse Genome Informatics (www.informatics.jax.org) yields a number of other mouse mutations, targeted, spontaneous or tg, associated with either thymic atrophy or hypoplasia. Most of these genes have been mapped to other chromosomes or are located far from the mapping position of our mutant. However, all are highly relevant in terms of different signaling pathways that may be involved or affected by the mutation observed in the Tg66 mice.

Several pieces of data argue that the observed phenotype is caused by transgene insertion and not by the transgene expression as such. First, the reconstitution experiments with the bone marrow chimeras suggested that the primary defect was in nonhemopoietic cells, while the transgene is expressed only in lymphocytes. Second, only one of the transgene-expressing founder mice exhibited the thymus phenotype, whereas other tg founder mice (with even higher transgene expression) were normal. Third, the transgene integrated in a region of Chr 2 around which several other genes have been described to be important in thymic function.

The Tg66 mice may serve as a useful model for identifying genes that regulate TEC differentiation and thymic development. These mice may also be a useful tool for intervention strategies aimed at restoring impaired thymic function, a phenotype frequently observed in many disease settings (i.e., during certain viral infections and after a course of steroid treatment). In addition, further insights into thymic development and maintenance are important for alleviating many problems associated with aging and a faltering immune system. For example, increased thymic size has been observed upon suppression of sex steroids in aging rats (50) and after administration of growth hormones to HIV-infected individuals (51). Clearly, new strategies are needed to improve thymic function in the treatment of many clinical conditions.

We thank Maj-Britt Alter and Margareta Hagelin for technical assistance and Dr. Peter Berglund for intellectual discussions.

The authors have no financial conflict of interest.