Epithelia are essential tissues in all multicellular organisms to establish a barrier against the external environment. To exert this vital function, epithelial cells need to be tightly connected to each other while at the same time they allow for complex tissue and cell shapes to arise. Occluding Junctions (OJs) have evolved to become epithelia’s gatekeepers- they seal intercellular spaces but also provide selective permeability. Recent studies have also proposed that OJs regulate epithelial morphogenesis and polarity, but the mechanisms are still poorly understood. Here we explored novel functions of OJs by investigating the function of Neurexin-IV (Nrx-IV), a conserved transmembrane protein of the Neurexin/Contactin family localized at invertebrate OJs, using the Drosophila melanogaster larval epithelium as a model system. By knocking down the expression of this protein in a specific subset of epithelial cells and in a time-controlled manner, we show that Nrx-IV is required for proper epithelial tissue organization and survival. This suggests that OJs play a more important role in epithelial morphogenesis than previously anticipated. As Nrx-IV is also present in invertebrate glial OJs and in vertebrate axo-glial junctions this work not only increases our knowledge about epithelia but can also have implications in nervous system biology.

OJs are essential in multicellular organisms to regulate the barrier function in epithelia. In vertebrates, this role is played by Tight Junctions (TJs) whereas invertebrates do it through the Septate Junctions (SJs)^[1][2]. TJs and SJs share key components, such as proteins of the Claudin family. SJs also contain conserved molecules present in vertebrate paranodal junctions, which link myelinated glial cells to axons^[3].

Studies in the fruit fly Drosophila melanogaster (hereafter called Drosophila) have allowed the discovery of more than 20 conserved proteins present at the SJ that seem to form a stable molecular complex^[4][5]. The core components of the SJs include both transmembrane, such as Claudins, the Caspr-domain Neurexin (Nrx-IV), Neuroglian and Na+/K+-ATPase, and cytoplasmic proteins, such as Coracle and Varicose.

Although the paracellular barrier is the most studied function of SJs, recent studies suggest they also play a role in epithelial morphogenesis independently of their barrier function^[6][7][8]. For example, impairing SJ function during embryonic development leads to defects in diverse processes such as epithelial tube size, extracellular matrix deposition and cellular rearragements. Nevertheless, the molecular mechanisms involved are still unknown. Furthermore, most of the knowledge about these junctions comes from studies performed in the embryo and, hence, their contribution to the biology of a post-embryonic epithelium has remained largely unaddressed. This has motivated us to investigate the role of SJs in the Drosophila larval wing imaginal disc epithelium.

The main objective of this work was to explore novel functions of SJs in epithelial morphogenesis in a post-embryonic tissue.

To understand the function of SJs in the Drosophila wing imaginal disc epithelium, we first looked at the subcellular localization of the SJ component Nrx-IV using a GFP-tagged protein trap line previously shown to reflect Nrx-IV endogenous localization^[9][5]. SJs are typically localized at the apical region of the cell, below Adherens Junctions (AJs) and its associated cortical actin belt^[1]. Therefore, we compared Nrx-IV localization to F-actin in XZ sections. As expected, Nrx-IV is apically localized, just below F-actin (Fig. S1A-C).

To investigate the function of Nrx-IV, we knocked down the expression in the wing imaginal disc. For that we used the Gal4/UAS system to express double-stranded RNA (dsRNA) against Nrx-IV specifically in the posterior compartment of the wing disc using the hedgehog-Gal4 (hh-Gal4) driver^[10][11]. Using this approach, the anterior compartment of the same wing disc serves as an internal control.

The SJs form a stable protein complex in which the knockdown of a single component leads to mislocalization of the other components, thus compromising SJ function^[4][5]. Thus, to validate the Nrx-IV RNAi we checked whether its expression affects the localization of another SJ core component, Coracle, known to bind Nrx-IV^[12] (Fig. S1D-L’). As expected, in control wing discs Coracle is localized mostly at the apical region of cells (Fig. S1D’). We knocked-down Nrx-IV using two different RNAi lines. In both cases we observed a reduction of Coracle levels in the posterior compartment, where Nrx-IV RNAi was expressed (Fig. S1G, J). Additionally, Coracle localization was impaired: it was localized evenly along the lateral membrane and not mostly apically as in the control (more evident with the Nrx-IV RNAi dsRNA-HMS00419 line, Fig. S1G’, white arrowhead).

To assess the impact of Nrx-IV knockdown on epithelial integrity and polarity, we examined the localization of F-actin and the AJ protein E-cadherin on fixed wing discs. E-cadherin localization does not seem to be affected in Nrx-IV knockdown: both in control and Nrx-IV knockdown wing discs, E-cadherin was mostly localized at the apical-most region of cells (Fig. S2). F-actin staining revealed that, in contrast to control wing discs (Fig. 1A, A’), the posterior region of Nrx-IV knockdown wing discs displayed an abnormal shape (Fig. 1F, F’, K, K’, see wing disc pouch region, yellow dashed lines). The Nrx-IV RNAi posterior compartment typically presented a depression close to the anterior-posterior boundary (Fig. 1F’, K’, yellow arrowheads). Moreover, in some RNAi expressing cells, F-actin accumulated in more basolateral regions, whereas in control cells F-actin localized at the apical region (Fig 1F’, K’, white arrows). These data suggest that Nrx-IV is required for epithelial sheet organization but not for AJ localization.

Looking at the localization of the nuclear marker we noticed that, in contrast to controls, several Nrx-IV RNAi expressing cells presented fragmented nuclei and had delaminated from the epithelial layer (Fig. 1H’, M’, Fig. S2F’, white arrowheads), but did not seem to express E-cadherin (Fig. S2D’-F’, white arrowheads). This suggests that the absence of Nrx-IV might induce cell death. To explore this, we stained controls and Nrx-IV RNAi wing discs for the apoptosis marker activated Caspase-3^[13]. We found that, indeed, the delaminating cells expressing Nrx-IV RNAi are positively stained for activated Caspase-3 (Fig. 1I-I’, J-J’, N-N’, O-O’, white arrows). This indicates that Nrx-IV knockdown induces cell death. Consistent with this, the posterior compartment in Nrx-IV RNAi wing discs appears to have less cells than controls (compare Fig. 1A’-E’ to F’-J’).

Altogether, our results suggest that SJs, and in particular Nrx-IV, are essential to maintain epithelial organization and survival in the wing imaginal disc. Recent studies have already proposed that SJ components might have an important role in epithelial morphogenesis independently of their barrier function. Namely, it has been shown that Nrx-IV, together with other SJ molecules (Yurt, Coracle, and the Na+/K+-ATPase), acts to maintain apical-basal polarity during early embryonic stages, before the SJ barrier function is established, but not during late embryogenesis, when the SJ barrier function is fully established^[14]. Other SJ proteins have also been shown to regulate cell shape and rearrangements during salivary gland formation and wound closure^[15][8]. Finally, SJs regulate the mechanical properties of the embryonic epidermis^[8]. On the other hand, we show that Nrx-IV loss of function does not seem to affect AJ localization, which is also supported by previous studies^{[16][17][18][8]}. Therefore, future experiments should further address how apical-basal polarity is influenced by Nrx-IV or other SJs components in the wing imaginal disc.

In what concerns post-embryonic epithelia, our study is the first showing that a core SJ component might influence the actin cytoskeleton. Whereas in vertebrates a link between OJs and F-actin has been clearly demonstrated, in invertebrates this connection is still poorly understood^[19][20][8]. More studies are thus needed to understand the molecular mechanisms that regulate the interactions between OJs and the cytoskeleton.

The cell death phenotype we observed upon Nrx-IV knockdown also points to a role of SJs in cell survival and tissue integrity. Previous studies have reported that clones of mutant cells for several SJ components do not survive to adult stages^[17][18]. Although cell death has not been assessed in these studies, their results support the hypothesis that SJs are required for cell survival. Interestingly, one study has also reported that the SJ transmembrane protein Neuroglian stabilizes epithelial integrity in the ovarian follicular epithelium and that its loss of function leads to abnormal cell polarity and delamination, although the molecular mechanisms are still not completely understood^[21].

This report shows that the SJ core component Nrx-IV is required for epithelial tissue organization and integrity in the Drosophila wing imaginal disc. In particular, Nrx-IV loss of function induces defects in F-actin localization and cell death.

The functional analysis in this work has been performed by knocking down Nrx-IV function using RNAi, which likely only induces a partial loss-of-function. Although we used two different RNAi transgenic lines with no predicted off-targets, more experiments should be performed using a null mutant condition.

Our functional analysis is only based on the knockdown of Nrx-IV, one of the described core components of SJs in invertebrates. We observed a decrease in Coracle levels and its mislocalization upon Nrx-IV knockdown, suggesting that the effect we observe is due to SJ impairment and not a Nrx-IV specific phenotype. However, the function of other SJ components should be analyzed to confirm this possibility.

In addition, the observed basolateral distribution of F-actin in Nrx-IV knockdown cells could be related to the cell extrusion/delamination and cell death^[22] and not to a direct link between F-actin and Nrx-IV.

In future studies, it would be important to determine whether Nrx-IV knockdown also affects the localization of other epithelial polarity components and how it induces apoptosis. In addition, the functional analysis of other SJ components should tell us whether the phenotypes observed in this study are specific to Nrx-IV or whether they are a general feature of SJ proteins. Finally, as SJ loss-of-function affects the integrity of the wing disc epithelium (this study) and the mechanical properties of embryonic epidermis^[8] it would be interesting to address whether SJ loss-of-function also affects the response of the wing disc epithelium to mechanical stress.

Experimental design and fly stocks

All strains used were purchased from the Bloomington Stock Center (Indiana, USA).

Nrx-IV localization was assessed using a GFP protein trap line developed by the FlyTrap Project (CA06597). This line consists of a GFP-tagged protein expressed from its endogenous locus, allowing the visualization of protein subcellular localization^[9]. The line used in this work has been previously shown to reflect the localization of the endogenous Nrx-IV^[5].

NrxIV knockdown was performed by expressing two different RNAi lines (UAS-Nrx-IV-dsRNA-HMS00419 and UAS-Nrx-IV-dsRNA-HMS01991, TRiP at Harvard Medical School), in a tissue-specific manner, taking advantage of the Gal4/UAS system^[23][24]. We used the hedgehog-Gal4 (hh-Gal4) driver, which drives gene expression in the posterior compartment of the larval wing imaginal disc^[11]. We also expressed UAS-GFP to visualize the region where the RNAi was expressed. UAS-luciferase (control for Valium 20 RNAi lines, TRiP at Harvard Medical School) or W1118 flies, not expressing RNAi, were used as controls. To restrict the Nrx-IV knockdown to the larval stages, we temporally restricted RNAi expression using the TARGET technique in which Gal4 expression is inhibited by a temperature-sensitive version of Gal80 (Gal80ts) at restrictive temperatures (18ºC) but active at permissive temperatures (29ºC)^[25].

Eggs were collected for 12–24 h and embryos and early larvae were kept at restrictive temperature (18ºC) for 3–5 days. To activate the expression of Nrx-IV RNAi, larvae were incubated at 29ºC for 48–72 h and mid-third instar (Mid-L3, wandering stage) larvae were collected to perform immunofluorescence staining.

Immunofluorescence protocol

Mid-L3 larvae were dissected as described^[26]. Dissected larvae were fixed for 20 min in 4% Paraformaldehyde (Sigma-Aldrich) in Phosphate Buffered Saline (PBS) 1× at room temperature (RT) and washed 5×-with PBST [PBS 1× + 0,1% Triton-X, (Acros Organics)]. Larvae were incubated in blocking solution [0,5% Bovine Serum Albumin (Sigma-Aldrich) in PBST] for a minimum of 45 min, followed by an overnight incubation at 4ºC with primary antibodies (mouse anti-GFP, 1:500, Roche; rabbit anti-GFP, 1:2000, Invitrogen; rat anti-E-cadherin, 1:20, DSHB; mouse anti-Coracle C-terminal region, 1:50, DSHB; rabbit anti-Cleaved Caspase-3 (Asp175), 1:150, Cell Signalling) diluted in blocking solution. Larvae were rinsed 5× with blocking solution and incubated with secondary antibodies (Alexa Fluor® 647 Donkey Anti-Rabbit, 1:250, Jackson ImmunoResearch; Alexa Fluor® 488 Goat Anti-Mouse, Cy5® Goat Anti-Rat; both 1:250, Life Technologies) for 2 h at RT. Alexa Fluor® 568 Phalloidin (1:100, Life Technologies) was added to the secondary antibody incubation to label F-actin. DAPI solution (4', 6-diamidino-2-phenylindole, 1:500, Sigma-Aldrich) was added in the last 10 min of secondary antibody incubation to label nuclei. Larvae were rinsed 5× with PBST before adding anti-fading mounting medium [2% DABCO (1,4-Diazabicyclo[2.2.2]octane, Sigma-Aldrich) + PBS 1× (1:4) + glycerol]. Stained wing discs were detached from the remaining larval tissues and mounted in a drop of mounting medium between two 24×60 millimeter coverslips. The coverslips were separated by one-coverslip-high bridge to prevent tissue damage and sealed using nail polish.

Imaging and image analysis

Wing disc imaging was performed on a LSM 710 (Carl Zeiss) confocal microscope using a 40X water objective. Images were analyzed using Fiji^[27]. Caspase-3 channel was processed for Noise correction, using the Remove Outliers function (radius 5 pixels, threshold 10) using Fiji. Figures were made using Adobe Photoshop and Adobe Illustrator.

This research was supported by Fundação para a Ciência e a Tecnologia (SFRH/BPD/84569/2012 to L.C., PD/BD/106058/2015 to S.P., and PTDC/BIA-BID/29709/2017_MechanOJunctions), European Research Council (2007-StG-208631 and ERC-2015-PoC-713735-EMODI), Marie Curie Intra-European Fellowship (PIEF-GA-2009-255573) and Congento LISBOA-01-0145-FEDER-022170, co-financed by FCT (Portugal) and Lisboa2020, under the PORTUGAL2020 agreement (European Regional Development Fund).

We are grateful to T. Pereira for invaluable support on imaging and data analysis; C. Mendes for sharing Caspase-3 antibody; past and present members of the Tissue Repair and Inflammation Lab for helpful discussions and technical help; the microscopy and fly facilities at CEDOC; the Bloomington Drosophila Stock Center (NIH P40OD018537), the TRiP at Harvard Medical School (NIH/NIGMS R01-GM084947), the FlyTrap project and the Kyoto Stock Center (DGRC) for providing Drosophila lines.

The authors declare no conflicts of interest.

Not applicable.

No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.