Coordinated crosstalk between microtubules and actin by a spectraplakin regulates lumen formation and branching

Subcellular lumen formation by single-cells involves complex cytoskeletal remodelling. We have previously shown that centrosomes are key players in the initiation of subcellular lumen formation in Drosophila melanogaster, but not much is known on the what leads to the growth of these subcellular luminal branches or makes them progress through a particular trajectory within the cytoplasm. Here, we have identified that the spectraplakin Short-stop (Shot) promotes the crosstalk between MTs and actin, which leads to the extension and guidance of the subcellular lumen within the tracheal terminal cell (TC) cytoplasm. Shot is enriched in cells undergoing the initial steps of subcellular branching as a direct response to FGF signalling. An excess of Shot induces ectopic acentrosomal luminal branching points in the embryonic and larval tracheal TC leading to cells with extra-subcellular lumina. These data provide the first evidence for a role for spectraplakins in single-cell lumen formation and branching.


25
Although epithelial in origin, TCs do not have a canonical apical-basal polarity, and, as 26 they migrate, extend numerous filopodia on their basolateral membrane, generating transient protrusive branches at the leading edge (6). As a consequence, they display a 1 polarity similar to that of a migrating mesenchymal cell (7).

2
While migrating and elongating, the TC invaginates a subcellular tube from its apical 3 membrane, at the contact site with the stalk cell (1). The generation of this de novo 4 subcellular lumen can be considered the beginning of the single-cell branching 5 morphogenesis of this cell, which continues throughout larval stages to generate an 6 elaborate single-cell branched structure with many subcellular lumina (3).

7
We have previously shown that centrosomes are key players in the initiation of

21
In the present study, we uncover a novel role for the spectraplakin Shot in subcellular 22 lumen formation and branching. Our results show that shot loss-of-function (LOF) leads 23 to cells deficient in de novo subcellular lumen formation at embryonic stages. We show 24 that Shot promotes the crosstalk between microtubules and actin, which leads to the 25 extension and guidance of the subcellular lumen within the TC cytoplasm. We observe 26 that Shot levels are enriched in cells undergoing the initial steps of subcellular branching 27 Ricolo and Araújo, 2020 5 as a direct response to FGF signalling. And an excess of Shot induces ectopic 1 acentrosomal branching points in the embryonic and larval tracheal TC leading to cells 2 with extra subcellular lumina. Furthermore, we find that Tau protein can functionally 3 replace Shot in subcellular lumen formation and branching.

6
Loss of Shot causes defects in de novo subcellular lumen formation 7 Shot is expressed during Drosophila development in several tissues such as the 8 epidermis, the midgut primordia, the trachea and the nervous system (25, 31). We began 9 by analysing the effect of shot LOF during TC subcellular lumen formation. To do so, we 10 analysed dorsal (DB) and ganglionic branch (GB) TCs at late stages of embryogenesis 11 (st. 16) ( Fig. 1 A, B).

12
The shot 3 null mutant TC phenotype consisted in subcellular lumen elongation defects 13 with a penetrance of 80% ( Fig. 1 C, D and F-I and E). This phenotype resembled the 14 previously reported for blistered (bs) mutants (9). bs encodes the transcription factor 15 DSRF that regulates TC fate induction in response to Bnl-Btl signalling (9, 32). However, 16 we observed that DSRF was properly accumulated in shot 3 TC nuclei ( Fig. 1 D), 17 discarding a possible effect of Shot in TC fate induction.

18
To analyse if the shot phenotype was cell autonomous, we expressed shot RNAi to 19 knock-down Shot in all tracheal cells and found that, like in null mutant conditions, 80% 20 of TCs analysed (n=300) at the tip of the dorsal branches (n=150) or ganglionic branches 21 (n=150) were affected in subcellular lumen formation ( Fig. 1 J, K). Of these, 20% did not 22 develop a terminal lumen at all (Fig. 1 L).

23
shot 3 embryonic TC lumen phenotypes range in expressivity from complete absence of 24 subcellular lumen to different lengths of shorter lumina ( Fig. 1 M, N). When quantified, 25 out of the 80% of embryos that showed a TC luminal phenotype, 36% of TCs did not 26 elongate a subcellular lumen at all and 64% failed to accomplish a full-length lumen (300 ganglionic TCs and 300 dorsal TCs) ( Fig. 1 M, N). Taken together, these results indicated 1 that Shot is involved in de novo subcellular lumen formation and elongation.

18
In all cases, we could detect more MT bundles in TCs, associated with the ESLs (  ShotC-GFP localized more to the MT/lumen region, together with MT-bundles (Fig. S1), 23 in agreement with the lack of actin-binding capability of ShotC isoform. Interestingly, we 24 observed a highly ramified subcellular lumen when higher amounts of ShotC were 25 expressed in tracheal cells (Fig. 2 J) suggesting that the effect of ShotOE in subcellular 26 lumen branching was dosage dependent.

28
Ricolo and Araújo, 2020 7 Tracheal overexpression of shotC phenocopied that of shotA in inducing ESLs (Fig. 2 D-1 F, G, H), suggesting that the ABD is not necessary for the induction of additional luminal 2 branching events. In order to clarify this, we used two other isoforms of Shot: shot∆Ctail, 3 lacking the C-terminal MT-binding domain, and shotCtail, a truncated form containing 4 only the C-terminal MT-binding domain (35) (Fig.2 M, N). Whereas overexpressing 5 shotΔ-Ctail in TCs we could only detect a branching phenotype in 7% of embryos 6 analysed (n=40), (Fig. 2 K), overexpression the C-tail domain alone induced TCs with 7 extra branching in 23% of the embryos (n=40) (Fig. 2 L), indicating that the C-tail alone 8 was sufficient to induce ESLs in TCs. Taken together these results using different Shot 9 isoforms, lead us to conclude that the Shot MT-binding domain alone is sufficient for the 10 extra branching events observed in ShotOE TCs.

11
ESLs were previously observed when higher numbers of centrosomes were present in 12 TCs (8). We therefore asked if the observed extra branching phenotypes could be due 13 to supernumerary centrosomes induced by ShotOE in TCs. Consequently, we quantified 14 the number of centrosomes in the TCs of ShotOE embryos. In control TCs we detected 15 an average of 2,3 ± 0,5 (n=33) centrosomes per TC, and in ShotOE 2,2 ± 0,2 16 centrosomes per TC (n=33) (Fig. 3 A, B, D). In both conditions, and as previously 17 described (8), this centrosome-pair was detected at the apical side of the TCs (Fig. 3 A, 18 B). Besides, analysing ShotOE TCs at embryonic st.15, (n=16) we could detect that the 19 ESL arose from the pre-existing subcellular lumen, distally from the centrosome pair ( Fig.   20 3 B' arrow). These data indicate that ShotOE did not change TC centrosome-number 21 and induced ESL by a distinct mechanism from centrosome duplication.

22
In contrast with ShotOE alone (Fig. 3 B'), in Rca1 mutants the bifurcated subcellular 23 lumen arose from the apical junction and continued to extend during TC development 24 (8). When we analysed the luminal origins in Rca1, ShotOE conditions, both types of 25 ESL where detected in the same TC in 25% of the cases (n=12). In the same TCs two 26 types of ESL were generated, one from the apical junction and another sprouted from 27 the pre-existing lumen distally from the junction (Fig. 3 C, asterisks). In addition, the effect of Rca1 LOF and ShotOE was additive in producing TCs with a multiple-branched 1 subcellular lumen (Fig. S2). These morphological ESL differences suggested that Rca1 2 and shot operate in different ways in the de novo formation and branching of the 3 subcellular lumen.

12
We analysed live embryos using time-lapse imaging, and observed that Shot localization 13 was extremely dynamic throughout subcellular lumen formation. We could detect Shot

18
In wt conditions F-actin and the actin-binding protein Moe concentrate strongly at the tip 19 of the TC, but are also detected in the TC cytoplasm, and these different actin

1
We followed these analyses, observing endogenous Shot in fixed and antibody stained 2 embryos. At early stages, when TCs started to elongate, we detected Shot co-localizing 3 with Moe at the tip of the TC (Fig. 4 A). The overlap between Shot and Moe was 4 maintained until late st.15 (Fig. 4 B). Then, we examined Shot localization in relation to

19
We then asked how actin and MTs were localized and organized in shot 3 mutant 20 embryos. We analysed the different types of TC mutant phenotypes ranging from cases 21 in which the TC did not elongate and the subcellular lumen was not formed, to cases in 22 which the TC was able to elongate and form the lumen albeit not to wt levels (

3
Regarding MT-bundles, we observed stable MTs organized in longitudinal bundles 4 around the subcellular lumen in control TCs (Fig. 5 E). In shot 3 TCs (n=20), we detected 5 MT-bundle defects. In particular, we observed that when the TC was not elongated, MT 6 bundles no longer localized to the apical region and seemed to be fewer than in wt (

24
We then proceeded to molecular dissect the function of Shot in TCs. To do so, we used 25 the three different constructs Shot: ShotC, Shot∆Ctail and ShotCtail (Fig. 2 M). When we 26 expressed ShotC in the tracheal TCs we found that 20% of TCs analysed (n=200), had 1 correct de novo luminal morphogenesis.

2
We next expressed shotC-tail in order to address whether the Shot MT-binding domain 3 alone could restore subcellular lumen formation. We observed that 24% of TCs analysed 4 at the tip of GBs and DBs (n=250) were still not able to form a subcellular lumen ( Fig. 6 5 E), suggesting that the tracheal expression of shotC-tail was not enough to rescue the 6 null phenotype. Finally, we expressed shot∆C-tail to test whether Shot without the MT-7 binding domain could restore subcellular lumen formation. We observed that 16% of TCs 8 analysed at the tip of GBs and DBs (n=250) were still unable to form a subcellular lumen 9 ( Fig. 6 F). Taken together, these analyses suggested that full-length isoform A, allowing 10 Actin-MT crosslinking is necessary for correct de novo subcellular lumen formation.

12
In order to further test the hypothesis that full-length Shot is needed to correctly form a 13 subcellular lumen, we analysed shot kakP2 mutant phenotype. This allele carries an

23
Since the actin and MT binding domains were shown to be necessary for the proper 24 formation of a subcellular lumen, we asked whether it was necessary to have both 25 domains in the same protein or if simply the independent presence of these domains 26 was enough to generate a subcellular lumen. To do so, we generated transheterozygous 27 flies expressing two different Shot isoforms, Shot kakP2 and Shot ∆EGC . Shot ∆EGC is a truncated protein, lacking the EF-hand, the Gas2 and the C-tail domains of Shot, leading 1 to complete loss of the MT-binding activity (38). The analysis of shot ∆EGC mutant TC 2 phenotypes revealed that 18% of TCs (n=400; 200 ganglionic and 200 dorsal TCs) did 3 not develop a TC lumen at all (Fig. 6 H) and that shot ∆EGC mutant TCs displayed MT and 4 actin disorganization phenotypes (Fig. S4). Interestingly, in shot ∆EGC mutant TCs, actin 5 was found to be disorganized throughout the cytoplasm (and not at the tip as in control 6 TCs) but in higher levels than in shot kakP2 TCs (Fig. S4). This suggests that the actin-7 binding domain present in shot ∆EGC is able to organize the actin in TCs albeit not to wt 8 levels.

9
In shot ∆EGC /shot kakP2 transheterozygous embryos, Shot molecules contained exclusively 10 either the CH1 or the C-tail, but neither molecule had actin-and MT-binding activity 11 simultaneously. These embryos displayed the same TC phenotype as either

25
We then asked whether tau null mutants displayed any TC luminal phenotypes. For this 26 we analysed a tau deletion mutant tau MR22 previously shown to have nervous system 27 defects (42). tau MR22 null mutant TCs showed defects in subcellular lumen directionality, but not in subcellular lumen formation (Fig. 8 D, F). We then proceeded to analyse TCs 1 in double mutants for shot 3 and tau MR22 (shot-tau). These double mutants showed higher 2 numbers of TCs without lumen (85%) than TCs from shot 3 (22%) or tau MR22 (3%) alone, 3 or a mere sum of these phenotypes, indicating a synergistic genetic effect between shot 4 and tau (Fig. 8 D-H). Despite the strong phenotypes, shot-tau mutants have the correct 5 number of cells per branch and express DSRF in TCs (Fig. S5). Furthermore, using a 6 mouse anti-Tau antibody, we could detect Tau colocalizing with the growing lumen in 7 TCs (Fig. 8 K). These results indicate that, as seen in neurons (42)  is filled by a subcellular lumen, in Shot-RNAi TCs these were reduced to 37% of the TCs 20 and even so absent in most branches (Fig. 9 B and G). Also, on average, each wt TC 21 develops 17 branch points, but Shot-RNAi TCs only developed an average of 6,5 branch 22 points each (n=8) (Fig. 9 B and I). We then overexpressed Shot full-length (ShotA-GFP Shot are required for TC extension and subcellular lumen formation (Fig. 10 A). In 5 neurons, like in tracheal cells, ShotC is unable to rescue the phenotype caused by shot 6 LOF, which is only rescued by expression of the full-length ShotA isoform (24). Shot has 7 also been shown to be required for sealing epithelial sheets during dorsal closure (38).

Shot expression is regulated by DSRF in TCs
Our results show that molecular levels of Shot are important for cytoskeletal 1 rearrangements, indicating that there is a dosage dependent effect in lumen formation 2 and extension as well as in luminal branching events. Shot is present in many cells during 3 development but Shot level regulation is likely to be more important in cells such as 4 neurons and tracheal terminal cells, due to their morphology (34). bs/DSRF is a TC-5 specific transcription factor, whose expression is triggered by Bnl signalling (9, 45), and 6 is required for TC cytoskeletal organisation (1). DSRF has also been shown to be 7 necessary not just for the establishment of TC fate, but to ensure the progression of TC 8 elongation (32). Cytoskeletal organisation and remodelling as well as TC elongation are 9 tightly coupled during subcellular lumen formation and in bs mutants actin accumulation 10 was impaired at the TC tip (1). We observe a similar actin phenotype in Shot mutants 11 ( Fig. 5 A-D) suggesting that the actin defects observed in DSRF mutants may be due to 12 a lower expression of Shot in these cells.

18
Spectraplakin roles have also been reported in cell-cell adhesion and cell migration (31).

19
Recently, attention has gone into the role of spectraplakins not only during normal 20 cellular processes but also in human disease, from neurodegeneration to infection and 21 cancer (53). However, not much is known about a role for spectraplakins neither during 22 lumen formation nor during subcellular branching events. Here, we provide evidence for 23 the involvement of the Drosophila spectraplakin Shot in subcellular lumen formation in 24 branching. Through its actin-and MT-binding domains, Shot is necessary for subcellular 25 lumen formation and branching (Fig.10). This function can be functionally replaced by 26 Tau, another microtubule associated protein which has been shown to be able to

34
We are grateful to M. Llimargas and our lab colleagues for comments on the manuscript.

35
We thank J. Casanova and F. Serras for the support given during throughout this study.

9
In all panels the TC outline is drawn in yellow. GFP staining is showed in grey and cell