RET/GFRα Signals Are Dispensable for Thymic T Cell Development In Vivo

Identification of thymocyte regulators is a central issue in T cell biology. Interestingly, growing evidence indicates that common key molecules control neuronal and immune cell functions. The neurotrophic factor receptor RET mediates critical functions in foetal hematopoietic subsets, thus raising the possibility that RET-related molecules may also control T cell development. We show that Ret, Gfra1 and Gfra2 are abundantly expressed by foetal and adult immature DN thymocytes. Despite the developmentally regulated expression of these genes, analysis of foetal thymi from Gfra1, Gfra2 or Ret deficient embryos revealed that these molecules are dispensable for foetal T cell development. Furthermore, analysis of RET gain of function and Ret conditional knockout mice showed that RET is also unnecessary for adult thymopoiesis. Finally, competitive thymic reconstitution assays indicated that Ret deficient thymocytes maintained their differentiation fitness even in stringent developmental conditions. Thus, our data demonstrate that RET/GFRα signals are dispensable for thymic T cell development in vivo, indicating that pharmacological targeting of RET signalling in tumours is not likely to result in T cell production failure.


Introduction
T cell development occurs mainly in the thymus [1]. However, by the time T cell precursors reach this primary lymphoid organ, they are not fully committed, and only later receive the cues that engage them on a T cell fate [1,2,3]. Thus, the thymic microenvironment is thought to provide appropriate signals that maintain a balance between thymocyte selection, proliferation and cell death [4,5]. These signals are dependent on thymocyte receptors and their cognate ligands, either soluble or membrane bound, which are obtained from the thymic microenvironment. Determinant factors to T cell precursor development have a mesenchymal or hematopoietic cell origin and are believed to trigger a gene expression program leading to specific cell fates [1,2,3]. Among major known molecular players in T cell development are Notch-Delta and TCR-MHC interactions [6,7]. However, identification of additional regulators of thymocyte development is still an unmet need in T cell biology.
T cell development starts in embryonic life [4,9]. Seeding of the embryonic thymus occurs around E13.5 and few thymocytes are beyond DN stage until E16.5 [4]. Full maturation of ab T cells is residual before E19.5, but some unique cd T cell populations are produced exclusively at defined foetal stages [2,4].
Previous studies showed expression of neurotrophic factors of the glial cell-line derived neurotrophic factor (GDNF) family (GFLs) in the thymus [10,11]. Productive signalling by GFLs is dependent on their association to a co-receptor (GFRa1 to 4), which also confers a degree of specificity to each GFL. Thus, GFRa1 is required to GDNF signalling, GFRa2 to NRTN, GFRa3 to ARTN and GFRa4 to PSPN [12]. GFRa molecules cooperate mainly with the transmembrane tyrosine kinase receptor RET for downstream signalling [12].
Activating mutations of Ret have been linked to cancer, i.e., somatic chromosomal rearrangements result in Papillary Thyroid Carcinoma, point mutations of RET lead to Multiple Endocrine Neoplasia 2 syndrome and RET is also differentially expressed in acute myeloid leukaemia [13,14]. Thus, RET inhibitors were recently developed for specific human cancer therapies [15,16].
RET signalling axes are critical to the neuronal system and kidney [12], but recent evidence indicates that RET signals are also key to intestinal lymphoid organ development [17,18].
Interestingly, it was shown that RET is expressed by mature lymphocytes [19] and GDNF promotes DN thymocytes survival in vitro [11]; thus, raising the exciting possibility that RET signalling may control thymocyte development in vivo.
In this study, we used cellular, molecular and genetic approaches to investigate the role of RET in foetal and adult thymic T cell development in vivo. We show that Ret, Gfra1 and Gfra2 are abundantly expressed in developing thymocytes, particularly in the earliest DN stages. Despite the developmentally regulated expression of these genes, analysis of E18.5 thymi from Ret 2/2 , Gfra1 2/2 or Gfra2 2/2 embryos revealed an insignificant impact of these molecules in T cell development. Sequentially, we used Ret conditional knockout mice in order to ablate Ret expression in T cell development. Similarly to foetal life, we found that RET is dispensable to thymocyte development in adulthood. This conclusion was further supported by the fact that RET gain of function mutations did not alter thymocyte differentiation. Finally, we employed competitive reconstitution chimeras to uncover subtle effects of Ret deficiency within the thymus. This very sensitive method revealed that the competitive fitness of developing Ret deficient thymocytes was intact. Thus, our data demonstrate that RET signalling is dispensable to thymic T cell development in vivo.

Results
Ret, Gfra1, Gfra2, Gdnf and Nrtn are expressed in the foetal thymus Previous reports have shown the expression of Ret, Gfra1 and Gdnf in the thymus [10,11]. Initially we investigated the expression of Ret and its co-receptors in E15.5 thymocyte subsets by RT-PCR. Although most E15.5 thymocytes are at the DN stage [4], due to minute cell numbers available at this developmental stage we sorted DN1+DN2 (pooling CD4 2 CD8 2 CD3 2 CD44 + CD25 2 and CD4 2 CD8 2 CD3 2 CD44 + CD25 + cells) and DN3+DN4 thymocytes (CD4 2 CD8 2 CD3 2 CD44 2 CD25 + and CD4 2 CD8 2 CD3 2 CD44 2 CD25 2 ) by flow cytometry. We found that while Ret, Gfra1 and Gfra2 were expressed in the foetal thymus, Gfra3 and Gfra4 were absent (Fig. 1A). Sequentially, quantitative RT-PCR analysis confirmed expression of Ret and Gfra1 in thymocytes at all DN developmental stages, a finding also confirmed at the protein level for RET (Fig. 1B, 1C). In contrast, Gfra2 was present in DN1+DN2 but absent from later DN stages (Fig. 1B). Sequentially, we evaluated the expression of the RETligands Gdnf and Nrtn in the thymic environment. We found that the main source of these transcripts were CD45 2 cells (Fig. 1D), while hematopoietic (CD45 + ) DN thymocytes only expressed minute levels of Gdnf and Nrtn (Fig. 1D, 1E). Thus, we confirmed that the molecules required for active RET signalling are expressed in the embryonic thymus, suggesting a role for these neurotrophic factor signalling axes in the early stages of foetal thymocyte development.
Since expression of Ret, Gfra1 and Gfra2 is higher in early DN thymocytes (DN1 and DN2) (Fig. 1B), we initially evaluated these differentiation stages in Ret, Gfra1 or Gfra2 deficient embryos. We found that both the percentage and cell number of DN1-4 subsets were similar between Ret, Gfra1 or Gfra2 deficient embryos and their respective WT littermate controls ( Fig. 2A; Fig. S1). Similarly, we found that total DN and ImmCD8 were equally represented in mutant embryos and their WT controls ( Fig. 2B;  Fig. S1).
Sequentially, we analyzed later stages of the ab TCR lineage development. Absolute numbers of DP thymocytes from Ret 2/2 , Gfra1 2/2 or Gfra2 2/2 embryos were identical to WT littermate controls ( Fig. 2B; Fig. S1). Similarly, the fraction and absolute numbers of cd TCR thymocytes, which are the majority of CD3 + cells at E18.5 [4], were unperturbed in Ret, Gfra1 or Gfra2 deficient animals ( Fig. 2C; Fig. S1). Consequently, absolute numbers of total thymocytes from Ret, Gfra1 or Gfra2 deficient embryos were similar to their WT littermate controls (Fig. 2D). Thus, we conclude that signals mediated by RET or by its co-receptors GFRa1 or GFRa2 are not required for foetal thymocyte development in vivo.

RET and its co-receptors are expressed in adult thymocytes
The thymic environment supports T cell development in embryonic and adult life. Nevertheless, T cell development in the foetus and adult thymus employs differential pathways, leading to different viability, proliferation and lineage commitment [4]. Thus, we investigated whether Ret related genes maintain their expression through adult thymopoiesis.
DN (CD4 2 CD8 2 CD3 2 ), DP, single-positive CD4 + T cells (SPCD4) and single positive CD8 + T cells (SPCD8) were FACS sorted and analyzed by quantitative RT-PCR analysis. RT-PCR analysis revealed that similarly to the foetal thymus only Ret and its co-receptors Gfra1 and Gfra2 were expressed in the adult thymus (Fig. S2). Quantitative RT-PCR confirmed that Ret, Gfra1 and Gfra2 expression was mainly expressed by DN thymocytes, although low levels of Gfra1 and Gfra2 expression were also expressed by DP thymocytes, a finding also confirmed at the protein level for RET (Fig. 3A, 3B). Sequentially, we evaluated the expression of the RET-ligands Gdnf and Nrtn in the adult thymus. While Gdnf expression was mostly found on CD45 2 cells, Nrtn was expressed both by CD45 2 and CD45 + DN and DP thymocytes (Fig. 3C).
Dissection of DN cells into DN1-DN4 subsets further revealed that DN1 thymocytes were the only DN subset that co-expressed appreciable levels of Ret, Gfra1 and Gfra2, while all other DN subsets expressed Gfra1 but only minute levels of Ret (Fig. 3D). Thus, we conclude that the expression of RET signalling partners in adult thymocytes mirrors to large extend the expression patterns of foetal thymocytes, ie, Ret, Gfra1and Gfra2 are most abundant in the earliest stages of T cell development, while Gdnf and Nrtn are mainly produced by non-hematopoietic thymic cells.

RET-mediated signals are dispensable for adult T cell development
Ret 2/2 animals die perinatally due to kidney failure, hindering analysis of adult T cell development [22]. Thus, in order to determine the role of RET signalling in adult thymopoiesis, we developed a Ret conditional knockout model (Ret fl/fl ) that allows a lineage targeted strategy for Ret ablation. These mice were bred to human CD2-Cre animals that ensure Cre activity from DN1 stage onwards [23] (Fig. S2). Analysis of the offspring of this breeding at 8 weeks of age showed that despite a marginal reduction in DN1 thymocyte numbers in CD2Cre/Ret null/fl animals, the subsequent DN stages were similarly represented in CD2Cre/Ret null/fl and CD2Cre/Ret WT/fl mice ( Fig. 4A; Fig. S3). Analysis of DN to SP ab T cell development showed similar fractions and absolute numbers of thymocytes within each subset (DN, DP, SP4 and SP8) in both CD2Cre/Ret null/fl and CD2Cre/Ret WT/fl mice ( Fig. 4B; Fig. S3). Finally, the fraction and absolute numbers of thymic cd TCR expressing T cells and total thymocyte numbers were not affected by Ret deletion (Fig. 4C, 4D). Thus, altogether our data indicate that RET-mediated signals are dispensable for foetal and adult thymic T cell development in vivo.

Ret gain-of-function mutations do not affect thymopoiesis
Over expression and expression of gain-of-function forms of RET are characteristic of proliferative cell phenotypes and oncogenic diseases [13]. Thus, in order to elucidate whether improved RET signals could affect thymopoiesis, we used a genetic model that drives a constitutively activated form of RET in Ret expressing cells (Ret MEN2B ) [24]. These mice harbour a single point mutation (Met919Thr) introduced into the endogenous Ret gene locus, thus resulting in improved ligand-dependent RET activation [24].
Analysis of Ret MEN2B/MEN2B and their WT littermate controls at 8 weeks of age revealed that DN (DN1-DN4) to SP mature ab T cell development had similar fractions and absolute numbers ( Fig. 5A-C; Fig. S4). Consequently, total thymocyte numbers were not affected by the Ret MEN2B gain-of-function mutation (Fig. 5D), demonstrating that this Ret gain-of-function mutation does not affect thymopoiesis.

RET signalling is dispensable for thymic reconstitution
Ret signalling has been implicated in the survival of a variety of peripheral and central neural cells, including motor, sensory and autonomic neurons [12]. Furthermore, it was shown that the RET ligand GDNF could promote thymocyte survival in foetal thymus organ culture (FTOC) systems [11]. Thus, it is plausible that subtle effects of RET signalling on thymocyte development might be masked in Ret conditional knockouts by compensatory mechanisms at later stages of thymocyte differentiation.

Discussion
Our data indicate that the neuroregulatory genes Ret, Gfra1 and Gfra2 are expressed in discrete DN thymocytes, while Gfra3 and Gfra4 transcripts were absent in thymocytes. Interestingly, the RET ligands Gdnf and Nrtn are predominantly produced by nonhematopoietic thymic cells. These gene expression patterns raised the exciting possibility that RET signalling axes could control T cell development, adding to the growing body of evidence that the nervous and immune systems share similar key molecular signals [11,14,17,18,19,25]. In line with this hypothesis, it was previously shown that GDNF could promote survival and maturation of thymocytes in vitro [11].
In order to test whether RET signalling axes control T cell development, we analyzed the thymus of genetically mutant embryos for Ret, Gfra1 or Gfra2 at E18.5 [20,21,22]. Despite expression of Ret in foetal thymocytes, Ret deficient embryos showed normal DN, immCD8 DP and cd T cell development. These findings were also consistent with normal T cell development in null mice for the RET co-receptors that provide specificity to the neurotrophic factors GDNF and NRTN, respectively Gfra1 2/2 and Gfra2 2/2 mice. Thus, we conclude that RET and its signalling partners GFRa1 and GFRa2, are dispensable for foetal T cell development.
Since T cell development in adulthood employs additional molecular mechanisms to foetal thymopoiesis, we investigated whether Ret related genes controlled adult T cell development.
Our data indicate co-expression of Ret, Gfra1 and Gfra2 in the early DN1 stage, and production of Gdnf and Nrtn in the adult thymic microenvironment. Both foetal and adult immature thymocytes co-express Ret, Gfra1 and Gfra2. These data are in line with a previous report indicating expression of these genes in the foetal thymus by in situ hybridization [10]. However, DP thymocytes expressed the RET co-receptors Gfra1 and Gfra2, despite absence of Ret expression, and similarly, adult DN2-4 expressed Gfra1 but lacked significant Ret expression. This observation is in line with previous reports showing that GFRas are more abundantly expressed than RET [26,27], which suggests that GFRas may modulate RET signalling in a non-cell-autonomous manner (signalling in trans) [28,29]. Accordingly, we have recently shown that Lymphoid Tissue initiator cells use unconventional RET signalling in which receptor activation is provided by soluble ligand and co-receptors in trans secreted from nearby cells [17]. Thus, it is possible that RET negative, GFRa positive thymocytes may modulate the activity of neighbouring non-hematopoietic RET expressing cells.
Analysis of adult conditional Ret mutant mice (Ret fl/fl ) bred to CD2Cre mice revealed a significant, but small impact on the absolute numbers of DN1 thymocytes. However, this reduction was not translated into consecutive developmental stages, indicating that RET mediated signals are dispensable to T cell Gfra1and Gfra2 deficient mice analyzed as in Figure 2A. Two-tailed student t-test analysis was performed between knockouts and respective WT littermate controls. No statistically significant differences were found. doi:10.1371/journal.pone.0052949.g002 production in vivo. The co-expression of Ret/Gfra1 in DN1 and the decreased number of DN1 cells in Ret fl/fl mice are consistent with a previous report indicating that GDNF promotes thymocyte survival in foetal thymic organ cultures (FTOCs) [11]. Neverthe-less, our data does not support a major role for this signalling axis in vivo, since Ret mutant thymocytes develop normally and positive modulation of RET-signalling (Ret MEN2B ) was not beneficial for T cell development. Thus, the contribution of RET signalling to T cell development in vivo appears to be insignificant. Moreover, while FTOCs reproduce several aspects of T cell development [30], they fail to mimic the exact events in T cell development [31,32], and therefore these different methodologies may also contribute to the observed discrepancies. GDNF/GFRa1 have been shown to activate the transmembrane receptor RET and the neural cell adhesion molecule (NCAM) in neurons [33,34]. Thus, although activation of a putative NCAM analogue by GDNF cannot be fully discarded in thymocytes, this is unlikely to have a significant physiological relevance since NCAM downstream signalling requires GFRa1 and Gfra1 2/2 embryos displayed normal thymopoiesis [33].
In order to overcome possible viability/proliferative compensatory mechanisms that may arise through T cell development, we performed sensitive competitive reconstitution assays in vivo with Ret deficient (CD2Cre/Ret null/fl ) and Ret competent (CD2Cre/ Ret WT/fl ) thymocytes. Our data demonstrate that even in a very sensitive competitive setting the fitness of Ret deficient T cell precursors is intact.
Finally, our findings indicate that pharmacological inhibition of the RET pathway in severe pathologies, such as medullary thyroid cancer, should not be confronted with undesirable T cell production failure [15,16].
In summary, our data demonstrate that RET signalling is dispensable to foetal and adult T cell development in vivo. Nevertheless, RET and its signalling partners are also expressed by mature T cells [19], thus, lineage targeted strategies will be critical to elucidate the contribution of RET signals to T cell function.

Generation of Ret conditional knockout mice
To generate mice harbouring a conditional Ret knock-out allele we engineered a targeting construct that firstly, included the introduction of a floxed 2.1 kb, Neomycin resistance (Neo r ) cassette under the control of the phosphoglycerate kinase-1 (PGK) promoter and a polyA tail (pA). This cassette (PGK- Figure 6. Competitive fitness and thymic reconstitution of Ret-null thymocytes. A. Experimental scheme: 9Gy irradiated hosts (Rag1 2/2 , CD45.1) received WT competitor precursors (CD45.1/2) together with hCD2Cre/Ret null/fl or control hCD2Cre 2 /Ret wt/fl precursors (CD45.2). B. 8 weeks after transplantation the thymus of the generated chimeras was analyzed by flow cytometry. Results show the ratio between hCD2Cre/Ret null/fl (grey bar) or hCD2Cre 2 /Ret wt/fl (black bar) and the third part WT competitor (CD45.1/2) through thymic T cell development. hCD2Cre/Ret null/fl precursor chimeras: n = 4; hCD2Cre 2 /Ret wt/fl precursor chimeras n = 4. Error bars show s.d. Two-tailed student t-tests were performed. No significant differences were found. doi:10.1371/journal.pone.0052949.g006 NEO r -pA) was inserted approximately 4.5 kb upstream at the Xho I site of the pBluescript KS (pBS KS) vector that carried approximately 13 kb of the 59 end of mouse Ret genomic locus flanking exon 1. The second modification included an insertion of a loxP ,2.5 kb downstream of exon 1, at the Hind III site in the intron between exons 1 and 2 of the mouse Ret locus. Finally, a viral thymidine kinase cassette (,3 kb) under the control of the PGK promoter (PGK-TK-pA) was inserted at the Hind III site ,5 kb downstream of the inserted LoxP site. To obtain homologous recombination, this targeting construct was linearised by Xho I, purified by gel elution and extraction using the Qiaquick gel extraction kit (Qiagen), prior to electroporation into 129SvJderived R1 ES cells grown on mouse embryonic fibroblast (MEF) feeder layers. Following double selection with 300 mg/ml Geneticin (G418, Invitrogen) and 2 mM Gancyclovir (Sigma), positive clones were identified by Southern blotting. Genomic DNA was digested with Hind III restriction enzymes and a 59 external probe of 500 bp was used to screen for positive clones. With the Hind III digest the WT and mutant alleles showed a band size of 16.5 kb and 6 kb respectively. Positive animals were subsequently crossed with transgenic mice expressing Vav1-iCre [23]  Anti hRET was performed with antibody from R&D (132507) and respective anti-mouse IgG1 isotype control. Flow cytometry analysis was performed on a LSR Fortessa (BD) and data was analyzed with FlowJo 8.8.7 software (Tree Star). Cell-sorting was performed on a FACSAria I or FACSAria III (BD), and purity of obtained samples was .97%. CD45 + and CD45 2 populations were sorted from the same samples.

Real-time PCR analysis
RNA was extracted from sorted cell suspensions using RNeasy Micro Kit (Qiagen). RT-PCR was performed as previously described [18] and quantitative Real-time PCR for Gfra1 and Gfra2 were done as previously described [18,36]. Hprt1 was used as housekeeping gene. For TaqMan assays (Applied Biosystems) RNA was retro-transcribed using High Capacity RNA-to-cDNA Kit (Applied Biosystems), followed by a pre-amplification PCR using TaqMan PreAmp Master Mix (Applied Biosystems). Taq-Man Gene Expression Master Mix (Applied Biosystems) was used in real-time quantitative PCR. TaqMan Gene Expression Assays bought from Applied Biosystems were: Gapdh Mm99999915_g1; Hprt1 Mm00446968_m1; Nrtn Mm03024002_m1; Gdnf Mm00599849_m1; Ret Mm00436304_m1.

Statistics
Statistical analysis was done using Prism. Variance was analyzed using F-test. Student's t-test was performed on homocedastic populations and student's t-test with Welch correction was applied on samples with different variances.

Supporting Information
Figure S1 Impact of Ret, Gfra1 or Gfra2 ablation in embryonic thymocytes. E18.5 thymocytes were analyzed by flow cytometry. A. Top: CD44 and CD25 expression profiles within the CD45 + Lin neg CD3 2 DN compartment for Ret 2/2 , Gfra1 2/2 , Gfra2 2/2 and respective WT littermate controls. Bottom: absolute numbers of DN1-DN4 in Ret, Gfra1and Gfra2 deficient mice. B. Dot plots: CD4 and CD8 expression profiles within the CD45 + Lin neg cdTCR 2 compartment from an example of Ret 2/2 and respective WT littermate controls. Similar gates were used in results shown. Note that within SPCD4 and SPCD8 gates .90% of cells were CD3 2 and are thus immature thymocytes. Results show percentage and absolute numbers of immature CD8 + thymocytes and absolute numbers of DN and DP thymocytes in Ret, Gfra1and Gfra2 deficient mice. C. Absolute numbers of cd TCR + thymocytes in Ret, Gfra1and Gfra2 deficient mice. In all panels: Null mice: open symbols; WT littermate controls: full symbols; Mean value: dash line. Two-tailed student ttest analysis was performed between knockouts and respective WT littermate controls. No statistically significant differences were found. (TIF) Figure S2 Generation of Ret conditional knockout mice. A. Adult (8 weeks old) DN, DP, single-positive CD8 (SP8) and single-positive CD4 (SP4) thymocytes were purified by flow cytometry. RT-PCR analysis was performed. B. (A) The floxed Neomycin cassette was inserted ,4.5 kb upstream of exon 1 of mouse Ret locus, a third loxP (LoxP3) was introduced downstream of exon 1 and ,5 kb downstream the PGK-TK-pA cassette was inserted to aid negative selection. Targeted events were identified by Southern analysis of either Hind III digests of genomic DNA using the 59 external probe. (B) The floxed allele was identified by PCR and the primers P1/P2 were used to identify the loxP that remained after excision of the Neomycin cassette (PGK-Neo-PA), while the loxP3 was identified using primers P3/P4. The primer sequences are in the methods section. (C) To screen for the null allele, primers P1 and P4 were used. C. Genotyping results from a litter of mice obtained from a cRet131WT/null6cRet131fl/fl breeding. In the loxP sites PCRs, upper band corresponds to the sequence with the loxP site and the lower band to the WT sequence. D. In order to evaluate the activity of Cre recombinase driven by hCD2, we bred hCD2Cre-expressing animals to Rosa26 eYFP animals. Histograms show flow cytometry analysis of eYFP expression in DN1 to DN4 thymocytes. Two-tailed student t-test analysis was performed between knockouts and respective controls. No statistically significant differences were found. (TIF)