We used a Nrg-GFP Flytrap line where a chimera between Nrg and GFP is expressed from the endogenous Nrg promoter as described in Morin et al. 19 and reflects the tissue and subcellular distribution of the endogenous protein 2021. GFP is inserted in the cytoplasmic tail close to the C-ter of Nrg. The homozygous line is viable and fertile without any apparent phenotype.

We analyzed the lateral mobility of Nrg using FRAP in whole living embryos at stage 15. At this stage, Nrg-GFP expression is restricted to the apico-lateral domain in epithelial cells indicating it is recruited at septate junctions. A photobleach was applied in areas of septate junctions and the fluorescence recovery was monitored in the bleached spot. These experiments were usually not extended for longer time than 160 s because of extensive epidermis movements. Photobleaching of a 3–5 μm² area was performed either in the apico-lateral (Fig. 1) or horizontal (Fig. 2) orientation. In either orientation, the recovery of fluorescence in the bleached area was low in wild-type embryos, and the mobile fraction of Nrg-GFP estimated to 25 % within 160 s (Fig. 3A). We did not observe any difference between the ventral and dorsal epidermis. We also examined the long-term recovery of Nrg-GFP after photobleaching and only a partial recovery (39 %) occurred even after 40 min [see Additional file 1]. Thus, our data indicates that Nrg-GFP is strongly stabilized at the apico-lateral membrane by forming an adhesion complex at septate junctions of wild-type embryos. As a matter of comparison, a similar stability was reported for the structural components of the adherens junctions, E-cadherin, α-catenin and Armadillo as analyzed by FRAP in live Drosophila embryos 22.

In nrx IV null mutant embryos, Nrg-GFP mislocalized to the baso-lateral membrane of epithelial cells and its mobile fraction strongly increased (44 % within 160 s) by comparison with wild-type embryos (Fig. 1B, 2B and 3B). The high mobility of Nrg-GFP in nrx IV mutant embryos may be correlated with disruption of the septate junction tripartite complex. It must be noted that Cont is diffusely distributed in the cytoplasm of epithelial cells in nrx IV mutants 6 implicating that both Nrx IV and Cont are lacking at the epithelial cell membrane. The loss of nrx IV is accompanied by an almost complete lack of junctional strands as analyzed by electron microscopy 12.

Similarly, in cont mutants, Nrg-GFP was expressed at the baso-lateral membrane of epithelial cells and the mobile fraction of Nrg-GFP (40 % within the time of 160 s) strongly increased when compared with wild type embryos (Fig. 1C, 2C and 3C). In cont or nrx IV mutant embryos, Nrg does not take part into the tripartite adhesion complex of septate junctions but may still be engaged in homophilic binding between two neighboring epithelial cells 2324. Under these last conditions, exchange between free and bound Nrg-GFP receptors may occur rapidly allowing lateral diffusion along the plasma membrane (Fig. 4B).

We used a Nrx IV-GFP Flytrap line where GFP is inserted into the first intron between the N-ter signal sequence and the discoïdin domain 25. This line is homozygous viable and fertile and Nrx IV-GFP expression is restricted to the apico-lateral domain in epithelial cells of stage 12–15 embryos indicating it is recruited at septate junctions as the endogenous protein (Fig. 5A). We performed FRAP analysis on wild-type embryos at stage 15 and determined that the mobile fraction of Nrx IV-GFP was 27 % within 160 s in septate junctions of epithelial cells (Fig. 6A).

Next, we analyzed the mobility of both Nrg-GFP and Nrx IV-GFP in a deficiency (Dfvari) for vari, which encodes a MAGUK that binds the cytoplasmic tail of Nrx IV 11. As shown in Fig. 1D and 5C, the loss of vari induces the mislocalization of both Nrg-GFP and Nrx IV-GFP at the baso-lateral membrane. In this deficiency, the recovery fraction of Nrg-GFP after photobleaching is very high (70 % within the time of 160 s) (Fig. 1D, 2D and 3D). Similarly, the mobile fraction of Nrx IV-GFP (65 % within the time of 160 s) strongly increased by comparison with wild-type embryos (Fig. 5C–D and 6B). The MAGUK protein Vari that contains L27, PDZ, SH3, HOOK, and GUK modules may act as a scaffolding molecule at septate junctions. The PDZ domain of Vari binds Nrx IV and its HOOK domain may be implicated in homo-multimerization or anchoring Dlg, another MAGUK associated with septate junctions 26 and interacting with Gliotactin 14. Indeed, mutations in the HOOK domain impair the accumulation of Vari at septate junctions 11. Therefore Vari may support oligomerization of transmembrane components of the septate junction complex or induce linkage with submembrane elements (Fig. 4A).

It must be noted that the mobile fraction of Nrg-GFP is significantly increased in Dfvari embryos by comparison with nrx IV null mutant embryos (70 % instead of 45 %). This is in accordance with genetic evidences indicating that Vari may display functions beyond interacting with Nrx IV and bridge together several elements of the septate junctions. As a matter of the fact, vari mutations can strongly enhance the phenotype caused by mutations of the claudin sinuous, whereas nrx IV mutations do not 17.

Finally, the mobility of Nrx IV-GFP was analyzed in a null allele for cora (Fig. 5E). Nrx IV-GFP localized at the baso-lateral membrane in the mutant embryos (Fig. 5F). As observed in Dfvari embryos, the mobile fraction of Nrx IV-GFP (64 % within the time of 160 s) strongly increased by comparison with wild-type embryos (Fig. 6C). The FERM domain of Cora is required for the recruitment of Nrx IV at septate junctions 10. However, how Cora would generate a scaffold is unknown since it does not contain an actin-spectrin binding domain like its related vertebrate homologue 4.1 adaptors. Our data suggest that Cora may participate to the septate junction scaffold beyond interacting with Nrx IV since its mutation has a similar effect as the loss of Vari.

Next, we compared the dynamics of the Nrx IV CAM when anchored or not at septate junctions in live embryos with the mobility of its vertebrate orthologue Caspr/Paranodin at the membrane of transfected cells without any junctional complex or intercellular contacts. We examined the lateral mobility of Paranodin fused with GFP at the C-ter in neurobastoma N2a cells. Paranodin-GFP was co-transfected together with Contactin, which is required for its cell surface expression in vertebrate cells 27. In this context, cell-cell contacts do not induce the formation of septate-like junctions as observed at the level of axo-glial paranodal junctions. FRAP experiments were performed by bleaching a section of cell membrane (Fig. 7A). The mean mobile fraction of Paranodin-GFP was 47 % within the time of 160 s (Fig. 7B).

We evaluated the role of the cytoplasmic region of Paranodin in its lateral mobility at the cell membrane. We analyzed the mobility of a mutant form, ParanodinΔcyto, with deletion of the cytoplasmic tail. The length of the C-ter of transmembrane proteins has been shown to influence their lateral mobility 28 since a minimal length should be required for the restriction by dynamic barriers. In the Paranodin-GFP constructs, GFP (265 aa) is fused at the C-ter and is expected to minimize the size effect of the cytoplasmic tail deletion (73 aa). This deletion mutant displayed a mobile fraction of 60 % (Fig. 7C), which is significantly enhanced by comparison with full-length Paranodin. Paranodin contains a juxta-membrane motif for the binding of 4.1 molecules, which are adaptors for linkage with the actin-spectrin cytoskeleton. Paranodin interacts with 4.1B at the paranodal junctions 29 similarly to Nrx IV that binds Cora in Drosophila septate junctions 10. However, in contrast to Nrx IV, Paranodin does not contain a C-ter PDZ-binding domain. When co-transfected in Cos cells, Paranodin forming a complex with Contactin interacts with the endogenously expressed 4.1 molecule Schwannomin and deletion of the 4.1-binding motif prevents this interaction 30. Thus, the association of the cytoplasmic tail of Paranodin with a 4.1 scaffolding molecule may reduce its lateral mobility.

Stocks were obtained from the Bloomington Drosophila Stock Center and published sources cont^ex956 6, nrx IV⁴³⁰⁴ 12 and cora¹ 33alleles have been described previously. The deficiency Dfvari (w¹¹¹⁸; Df(2L)Exel7079, P+PBac{XP5.WH5}Exel7079/CyO) was used as a null mutant for vari 11. The Flytrap lines Nrg-GFP (G00305) 19 and Nrx IV-GFP (CA06597) 25 obtained from the FlyTrap project contain the GFP exon inserted in the nrg and nrx IV genes, respectively. New mutant lines were generated with the following phenotypes: nrg-GFP; cont^ex956/TM3, nrg-GFP; nrx IV⁴³⁰⁴/TM3, nrg-GFP; Dfvari/CyO; cora¹/CyO; nrx IV-GFP, Dfvari/CyO; nrx IV-GFP.

Embryos were collected on yeast agar plates and aged to stage 15 (12 h–13 h 30 at 25°C). The embryos were dechorionated for 2 min with bleach (6% chlore), rinsed with water, blotted dry with paper towels, gently picked up with double-sided tape and placed in a Petri dish and covered with water for FRAP analysis.

The DNA construct pRc-CMV/F3 encodes the full-length sequence of F3/Contactin, Paranodin-GFP encodes the full-length sequence and Paranodin-Δcyto encodes the extracellular and transmembrane domains of Caspr/Paranodin (aa1-1297) fused with GFP in pEGFP-N1 34. Neuroblastoma N2a cells plated on coverslips were grown in DMEM containing 10% FCS and transiently transfected using jet PEI (Ozyme, Saint Quentin Yvelines, France). The coverslips were placed in a Petri dish in a temperature controlled chamber at 37°C for FRAP analysis, 48-h after transfection.

The FRAP analysis was performed on a Leica (Wetzlar, Germany) TCS SP2 laser scanning microscope equipped with 1.2 NA 63× water immersion objective. Imaging of GFP was performed using the 488-nm beam of an argon laser at 7% laser power, with the electronic zoom at 5×, scan speed of 400 Hz, box size 512 × 512 pixels. Images (two averages) were acquired with an open pinhole. These conditions were found to give minimal photobleaching over the observed time. The ROI for bleaching (3–5 μm²) was delineated and the FRAP sequence was started: 3–6 reference images were acquired first, then the sample was bleached by the laser at 100 % for 3 s (2 bleach scans), and fluorescence recovery was recorded for 160 s using the Leica FRAP software. Acquisition of z-stacks before and after bleaching indicates that the bleached spot can be detected over 2.8 μm in the z-axis [see Additional file 2]. Thus, even in case of epidermal movements in the z-axis, images of the recovery sequence are acquired within the bleached volume. In addition, all the images of the FRAP sequences were checked by eye to correct the ROI position for epidermal movements in xy. To create the recovery curves, the background-corrected fluorescence intensities were transformed into a 0–1 scale and plotted using Microsoft Excel. The individual traces were fitted using SigmaPlot through the data of the function I_m*(1-exp(-t/τ)) to calculate the mobile fraction I_m, and the recovery time τ.

Multiple embryos (2–5) for each genotype were analyzed. FRAP experiments were conducted in the same sample on heterozygous and homozygous embryos for each of the nrx, cont, cor, and vari mutations. Similar kinetics of fluorescence recovery were obtained for Nrg-GFP or Nrx IV-GFP in wild-type and heterozygous embryos.