Lymphoid EVA1 Expression Is Required for DN1-DN3 Thymocytes Transition

Background Thymus organogenesis and T lymphocyte development are accomplished together during fetal life. Proper development and maintenance of thymus architecture depend on signals generated by a sustained crosstalk between developing thymocytes and stromal elements. Any maturation impairment occurring in either cellular component leads to an aberrant thymic development. Gene expression occurring during T lymphocyte differentiation must be coordinated in a spatio-temporal fashion; one way in which this is achieved is through the regulation by cell-cell adhesion and interactions. Principal Findings We examined the role played by Epithelial V-like Antigen 1 (EVA1), an Ig adhesion molecule expressed on thymus epithelial cells (TEC) and immature thymocytes, in T cell development by employing RNA interference in vitro and in vivo models. Fetal liver derived haematopoietic progenitors depleted of Eva1, displayed a delayed DN1-DN3 transition and failed to generate CD4CD8 double positive T cells in OP9-DL1 coculture system. In addition, we could observe a coordinated Eva1 up-regulation in stromal and haematopoietic cells in coculture control experiments, suggesting a possible EVA1 involvement in TEC-haematopoietic cells crosstalk mechanisms. Similarly, Rag2-γc double knock out mice, transplanted with Eva1 depleted haematopoietic progenitors displayed a 10-fold reduction in thymus reconstitution and a time delayed thymocytes maturation compared to controls. Conclusions Our findings show that modulation of Eva1 expression in thymocytes is crucial for lymphocyte physiological developmental progression and stromal differentiation.


Introduction
The thymus is the primary lymphoid organ where T cells are generated [1] and, unlike the bone marrow (BM), it does not contain self-renewing progenitors [2]. Lymphopoiesis depends on the input of new progenitors from either BM in the adulthood or the fetal liver in the embryo [3] and thymic environment is required for differentiation into T lymphocytes [4]. Direct contact of haematopoietic precursor cells with thymic epithelial cells (TEC) allows the initiation of signalling events, that are relevant to the development of either cell types. In particular, lympho-stromal interactions are necessary for thymocyte development, thymus homeostasis and proper selection of immunocompetent T cell repertoire [5]. Interactions between TEC and developing thymocytes, named ''thymic crosstalk'' [6], are fundamental for the development of a normal thymic architecture [7,8,9].
Epithelial V-like Antigen1 (EVA1) is an immunoglobulin-like, epithelium-specific, adhesion molecule expressed early in ontogeny [11]. It is strongly expressed in the emerging thymus, where it is down-regulated co-temporally with the emergence of the first DP thymocytes, [12]. In thymus EVA1 is expressed in both epithelial cells and developing DN thymocytes [13]. Transgenic overexpression of EVA1 in the thymus cortex resulted in a modified stromal environment, which elicited an increased absolute number of heterotypic complexes as thymic nurse cells (TNC), that are thought to be sites of lymphostromal interactions. The effect of EVA1 ectopic expression on the stromal microenvironment, together with the observation that stroma/haematopoietic ratio did not change, suggested that the increase in both cellularity and organ dimension was the result of an enhanced early stromathymocyte cross-talk.
Here we looked at the role of EVA1 in early T cell development by using fetal liver (FL) derived haematopoietic EVA1 depleted precursors for both in vitro and in vivo models. In OP9-DL1 coculture system [14], we observed Eva1 up-regulation in haematopoietic cells during DN1-DN3 transition and a coordinated up-regulation in OP9-DL1 stromal cells. Conversely, Eva1 interference caused a delay in DN1-DN3 transition and no upregulation either in haematopoietic or in OP9-DL1 stromal cells, was observed. Similar Eva1 depleted haematopoietic precursors were used in thymic repopulation experiments in alymphoid Rag2-cc double knock out (dKO) mice [15]. These mice reconstituted with Eva1 interfered haematopoietic cells, showed a strong reduction in thymic reconstitution compared with controls, and a delay of DN1-DN3 progression with a consequent delay of thymocytes maturation.
Collectively our findings suggest that EVA1 plays a key role in the early phases of T cell development and that coordinated Eva1 expression is necessary so that DN1-DN3 progression takes place correctly, suggesting a possible role in TEC-haematopoietic cells crosstalk.

Results and Discussion
Eva1 expression is finely regulated during thymus development in intrathymic haematopoietic cells at DN1-DN3 transition In a previous work aimed at defining the cell lineages expressing EVA1 during thymus organogenesis [13], we demonstrated that EVA1 is expressed in both epithelial and haematopoietic cells. While expression in TEC appeared to be stable, the one in haematopoietic cells was specific for DN subsets [13]. To characterize timing of Eva1 expression during thymus ontogeny in the different cellular components, quantitative RT-PCR (qRT-PCR) was performed in flow cytometry purified haematopoietic cells (CD45 + ) and TEC (CD45 2 ), from wild type mice at different stages of embryonic development (from embryonic day 13.5-E13.5 to newborn -NB). As shown in figure 1A, in CD45 + cells, Eva1 showed an expression peak between E14.5 and E16.5 stages, then was strongly downregulated. Eva1 modulation was probably related to DN subsets appearance and progression toward later stages of development, such as DP cells. We also confirmed that, in TEC, Eva1 expression level was stable from E13.5 to birth, then was moderately downregulated to a steady state which was maintained throughout adult life (Fig. 1A) [13]. Notably, a slight increase of Eva1 expression could be observed at the stage immediately after the downregulation seen in the haematopoietic counterpart (i.e., E17.5) (Fig. 1A), suggesting a coordinated EVA1 modulation between TEC and developing thymocytes.
We then used mutant mice with an early block in thymus development to define EVA1 requiring stages. We observed a marked increase of Eva1 expression in sorted CD45 + thymocytes from SCID (with a spontaneous mutation in protein kinase, DNAactivated, catalytic polypeptide Prkdc scid ,) and Rag2-cc dKO mice, but not from Foxn1 KO nude or Tg-calcineurin mice (DCam) (Fig. 1B left panel). Interestingly, in both SCID and Rag2-cc dKO thymi, the majority of CD45+ cells are DN1-DN3 stage cells [15,16]. In contrast, Foxn1 ko nude mice have a pre-DN cells block [17,18,19] and Tg-calcineurin mice express a constitutively active form of calcineurin, have an arrest of thymocyte development around DN4 stage [20].
Tydell et al., using mutant mice, demonstrated that also adult DN subsets such as T cell precursors sorted cells express Eva1 [21]. To verify that the above observation was not mutation-dependent, we looked at Eva1 expression in sorted DN subpopulations from two-month-old C57Bl6 thymi. As shown in Figure 1B (right panel), we indeed observed up-regulation of Eva1 during DN1-DN3 transition, with an expression peak at DN3 stage and a sharp decline at DP stage, confirming that Eva1 positive regulation takes place during adult DN1-DN3 transition, too.
All together these data confirm that Eva1 is transcriptionally regulated during DN1-DN3 transition. LSK-EVAi cells had a delayed DN1-DN3 transition and failed to generate CD4 and CD8 SP cells in in vitro T cell development T lymphocyte development, can be studied in vitro by coculturing haematopoietic progenitors with OP9 stromal cells expressing Notch ligand Delta-like-1 (OP9-DL1). We applied this system together with Eva1 lentiviral RNA interference (RNAi) to evaluate the role of EVA1 in T cell differentiation [14]. Haematopoietic precursors isolated from E14,5 fetal liver (FL) by cell sorting (Lin 2 ; Sca1 + ; c-Kit + )(LSK) were used. Interference with Eva1 was performed by infection with lentiviral vectors (see material and methods for details). The efficiency of RNA interference was assessed by qRT-PCR and immunofluorescence analysis (IF). As shown in Figure 1C (upper panel), LSK cells interfered for Eva1 (EVAi) showed a strong reduction of Eva1 expression, while LSK infected with control lentiviral vector (CT) showed an expression level of Eva1 comparable to uninfected LSK-WT cells. Interference was confirmed by IF (Fig. 1C, lower panel) using a rabbit polyclonal serum anti-EVA1. We tested three different interfering sequences obtaining comparable results (data not shown). To address whether EVA1 depletion had an effect on T cell development, progressive phenotype acquisition based on CD44/CD25 and CD4/CD8 expression, was evaluated by flow cytometry in cocultures experiments. As described [14], LSK-CT cocultured with OP9-GFP cells did not show the CD44/CD25 progression and did not give rise to T cells ( Fig. 2A, left and right panels). Otherwise, after 4 days of coculture, LSK-CT cells cultured on OP9-DL1 cells showed a differential surface expression of CD44/CD25 molecules ( Fig. 2A, left panel). The progression in CD44/CD25 maturation pathway was observed until 18 days of coculture ( Fig. 2A, left panel). This progression correspond to CD4 and CD8 expression ( Fig. 2A, right panel). LSK-CT cells gave rise to CD4 + CD8 + immature double positive (DP) T cells after 8 days of coculture ( Fig. 2A, right panel), with an increase of DP cells after 18 days of coculturing ( Fig. 2A, right panel). The temporal kinetics of LSK-CT differentiation cocultured with OP9-DL1 cells was similar to that observed with LSK-WT cells (data not shown). A delayed progression into the CD44/CD25 maturation pathway was observed when LSK-EVAi cells were cultured on OP9-DL1 cells and compared to LSK-CT progression ( Fig. 2A, left panel). After 18 days, LSK-EVAi cells were arrested at DN2-DN3 stages and failed to generate DP cells ( Fig. 2A, right panel). These data confirmed that Eva1 upregulation occurs at DN1-DN3 transition during physiological thymocyte development and maturation in vitro.
Furthermore, we looked at Eva1 regulation in LSK and stromal cells, during in vitro differentiation. qRT-PCR analysis showed that no Eva1 regulation was detectable in both cellular partners of negative control cocultures LSK-CT/OP9 (Fig. 2B, upper panels), while Eva1 expression was increased in LSK-CT cells at day 4 and 8 of coculture, with a peak on day 4, corresponding to the DN1-DN3 transition (Fig. 2B, centre panels). Eva1 upregulation, was also observed in stromal cells with a peak at day 8, in agreement with data obtained on ex-vivo cells (Fig. 1A). The great early lymphoid Eva1 upregulation observed in the positive control cocultures, seemed to be driven by the contact with the stromal partners and accompained physiological differentiation steps, while no variations in Eva1 expression levels were observed in LSK-EVAi/DL1 (Fig. 2B, lower panels). Sequential modulation in the two cellular partners underlines the role of EVA1 in thymocyte-stroma cross-talk (Fig. 2B, centre panels), demonstrating that EVA1 expression is required for DN1-DN3 transition and that its lymphoid positive regulation needs haematopietic/stromal interactions. Furthermore, to define if EVA1 expression in stromal cells was sufficient for DN differentiation, we expressed EVA1 in OP9 cells. Coculture experiments with LSK-CT and OP9-EVA cells did not elicited T cell differentiation and the tremendous increase (twenty thousand fold) of Eva1 expression after four days of coculture (Fig. 2C) was not enough to generate DN cells. These results indicated that EVA1 expression per se is not sufficient to drive early T cell maturation and needs other molecules/signal derived by Notch/Delta-ligand-1 interactions.
Thymus reduction in Rag2-cc double knockout mice reconstituted with LSK-EVAi T cell development in the thymus depends on the continuous recruitment of haematopoietic precursors from the bone marrow (BM) via the blood [3]. Mice with deletion of the common cytokine receptor c chain (cc) and recombinase activating gene 2 (Rag2-cc dKO) have a stable phenotype characterized by the absence of all T lymphocytes, B lymphocytes and NK cells. We performed BM transplantation experiments to assess the ability of LSK-CT and LSK-EVAi cells to reconstitute a functional thymus in Rag2-cc dKO sublethally irradiated mice. After 8 weeks from the transplant, recipient mice injected with LSK-CT cells showed a marked thymic reconstitution if compared with not transplanted mice (Fig. 3A, left panel). Mice transplanted with LSK-EVAi cells showed a smaller thymus compared with the LSK-CT-injected mice, indicating a less efficient thymic reconstitution (Fig. 3A, left panel). This result was confirmed by total cell numbers that were significantly smaller in LSK-EVAi than in LSK-CT reconstituted mice (Fig. 3A, histogram). Similar results were obtained at 6 weeks after transplant (data not shown).
Reconstituted thymi were analyzed by flow cytometry 6 and 8 weeks after transplantations for CD44/CD25 and CD4/CD8 subsets distribution. Thymi from mice transplanted with LSK-CT cells displayed higher percentages of DP and SP cells compared with thymi from mice transplanted with LSK-EVAi cells (Fig. 3B); the difference in DP and SP subpopulations distribution between the two groups was still evident after 8 weeks (Fig. 3B). LSK-EVAi cells reconstituted thymi showed a delayed in cell differentiation, with accumulation of DN thymocytes already evident at 6 weeks from transplant (Fig. 3B). Thereafter, there was an increase of DP cells and the percentage of DN cells still remained significantly (p,0.01) higher than in LSK-CT reconstituted thymi (Fig. 3B). Comparison of DN counts obtained from the three different mouse groups, showed that spontaneous generation of DN subset was very modest even at eight weeks from irradiation, indicating that endogenous progenitors are not implicated in reconstitution events. These results confirm that EVA1 is important for normal developmental processes.
Patterning of thymic epithelial compartment is defined by differential keratin (K) expression: in particular thymic cortex is characterized by the expression of K8 and K18, while thymic medulla expresses K5 and K14 isoforms. While initial patterning does not depend on inductive signals from haematopoietic cells, thymocyte derived signals are required during late phases of fetal thymus development to generate and sustain a normal thymic epithelial compartment in newborn and adult mice [22]. Moreover, thymic stroma consists of TEC organized in a three-dimensional (3-D) structure [23], which is driven by developing T cells via TEC crosstalk [6]. Haematoxylin and eosin staining of thymi from Rag2-cc dKO transgenic mice revealed thymocyte paucity (Fig. 4A), while immunohistochemistry showed an aberrant K8 + K5 + phenotype (Fig. 4A, 4B), [24] and a 2-D thymus organization (Fig. 4B). (Fig. 4A). LSK-CT cells transplanted into Rag2-cc dKO mice led to redistribution of keratins and reappearance of 3-D thymic architecture (Fig. 4B). Moreover, haematoxylin and eosin analysis revealed that a significant amount of haematopoietic cells repopulated the thymus after transplantation (Fig. 4A). Conversely, in LSK-EVAi transplanted mice we observed only a partial reappearance of 3-D thymic architecture and modest keratin redistribution (Fig. 4B), with a few lymphocytes repopulating thymus (Fig. 4A). Possibly, the absence of EVA1 homotypic interactions leads to DN1-DN3 progression delay, which prevents stromal thymus reorganization. In addition we further show that DN1-DN3 transition controls thymic epithelium morphogenesis [25,26].
Taken together, our results demonstrate that EVA1 is important in early T cell differentiation, affecting DN maturation and is a key molecule in thymocyte-stroma cross-talk.

Materials and Methods
All experiments were approved by the Italian Health Ministry, according european legislation.

Mice
C57/Bl-6 (Charles River, Calco-I) and C57/Bl-6 Rag2-cc dKO (Taconic Europe, DK) were maintained under specific pathogen free (SPF) conditions at Fondazione Parco Biomedico S. Raffaele, Rome. The day of first vaginal plug detection was designated as gestation day 0.5 (E0.5). All experiments were approved by the Italian Health Ministry, according european legislation.

RNA extraction and quantitative RT-PCR
Total RNA was extracted from sorted cell populations using the RNeasy plus mini kit (Qiagen GmbH, D-40724 Hilden, Germany), and reverse transcribed with oligo-dT primer and superscript RT II reverse transcriptase (Invitrogen corporation, 92008 CA, USA). Real time PCR was performed on diluted cDNA samples using TaqMan Universal PCR Master Mix in ABI PRISM 7500 Real Time PCR System. All materials, including primers for EVA (EVA-FAM) and GAPDH standard (GAPDH-VIC) were purchased from Applied Biosystem (94404 CA, USA). Samples were examined in triplicate. The DELTA Ct method was used for normalization to GAPDH mRNA.

Cocultures
OP9 coculture assays were performed as described [23]. Stromal cells were plated 48 hours before use in 24-well plates (Falcon, BD) to achieve a confluent monolayer. All cocultures were initiated with 4610 3 FL Lin 2 Sca1 + c-Kit + (LSK) in alpha MEM-20% FBS in the presence of 5 ng/ml Flt3L and 5 ng/ml IL7 (Immunotools). Cocultures were harvest by forceful pipetting at the indicated time points. LSK were separated from stromal cells by labeling with CD45-coated magnetic beads (Miltenyi Biotec). Labeled cells were subsequently separated and recovered by magnetic column separation (Miltenyi Biotec).

Immunohistochemistry
Thymuses for sectioning were washed in PBS, embedded in OCT compound (Bio-Optica, 20134 Milano, Italy) and snapped frozen. Five mm frozen sections were cut and plated onto Poly-llysine slides (Menzel-Glaser, Menzel GmbH, D-38116 Braunschweig, germany). For immunostaining, sections or cells were permeabilized in PBS-1%BSA 0.2%TritonX100 and blocked in 20% normal goat serum, then incubated with primary antibody for 1-2 hours followed by incubation with the appropriate secondary antibody. Immunoreaction was analyzed and photographed using fluorescent microscope (Nikon Eclipse TE 2000-E) and Metamorph 7.5 software.
In vivo thymus reconstitution 5610 5 infected and selected LSK cells were intravenously injected into irradiated Rag2-cc dKO(4 Gy) mice. Thymi were analyzed 6 or 8 weeks after transfer by flow cytometry. Donor cells were distinguished from host cells by their ability to differentiate in lymphoid cells.

Statistics
Statistical comparison was performed by the Student's t-test using Excel software.