Mask family proteins ANKHD1 and ANKRD17 regulate YAP nuclear import, stability and phase separation

The Mask family of multiple ankyrin repeat and KH domain proteins were discovered in Drosophila to promote the activity of the transcriptional coactivator Yorkie (Yki), the sole fly homolog of mammalian YAP (YAP1) and TAZ (WWTR1). The molecular function of Mask, or its mammalian homologs Mask1 (ANKHD1) and Mask2 (ANKRD17), remains unclear. Mask family proteins contain two Ankyrin repeat domains that bind Yki/YAP as well as a conserved nuclear localisation sequence (NLS) and nuclear export sequence (NES), suggesting a role in nucleo-cytoplasmic transport. Here we show that Mask acts to promote nuclear import of Yki, and that addition of an ectopic NLS to Yki is sufficient to bypass the requirement for Mask in Yki-driven tissue growth. Mammalian Mask1/2 proteins also promote nuclear import of YAP, as well as stabilising YAP and driving colloidal phase separation into large liquid droplets. Mask1/2 and YAP normally colocalise in a granular fashion in both nucleus and cytoplasm, and are co-regulated during mechanotransduction. Our results suggest that Mask family proteins promote YAP nuclear import and phase separation to regulate YAP stability and transcriptional activity.


Introduction
The YAP/TAZ family of oncoproteins has a single homolog in Drosophila named Yorkie (Yki) that was discovered to control tissue growth in proliferating epithelia (Huang et al., 2005). Genetic analysis of YAP and TAZ in mice is revealing an important role for both proteins in driving cell proliferation during tissue regeneration as well as during formation of several tumour types (Cai et al., 2015;Cai et al., 2010;Camargo et al., 2007;Dong et al., 2007;Elbediwy et al., 2016;Gruber et al., 2016;Reginensi et al., 2015;Schlegelmilch et al., 2011;Vincent-Mistiaen et al., 2018;Zhang et al., 2011). Yki/YAP/TAZ were shown to function as transcriptional co-activators of the nuclear DNA binding transcription factors TEAD1-4 (named Scalloped in Drosophila) (Koontz et al., 2013;Wu et al., 2008). The molecular mechanisms by which Yki/YAP/TAZ are physiologically regulated are still being determined.
One key mechanism regulating Yki/YAP/TAZ subcellular localisation is phosphorylation by the Hippo pathway kinase Warts/LATS on five Serine residues, which induces retention in the cytoplasm by binding to 14-3-3 proteins (Huang et al., 2005;Irvine, 2008, 2009;Zhao et al., 2007). In response to mechanical stretching/flattening of cells, both Yki (Fletcher et al., 2018) and YAP/TAZ (Dupont et al., 2011;Wada et al., 2011;Zhao et al., 2007) can translocate from the cytoplasm to the nucleus. Both Yki (Manning et al., 2018) and YAP (Ege et al., 2018) undergo dynamic and continuous nuclear-cytoplasmic shuttling which must involve specific nuclear import and export factors. Several regulators of YAP nuclear export have been proposed (Furukawa et al., 2017;Lee et al., 2018). In contrast, no nuclear import factors for the Yki/YAP/TAZ family have been identified and these proteins lack a conventional nuclear localisation sequence (NLS) (Kofler et al., 2018;Wang et al., 2016).
Here we show that the Mask family of ankyrin-repeat domain proteins, which feature conserved NLS and NES (nuclear export sequence) motifs, mediate nuclear import of both Drosophila Yki and mammalian YAP. Previous work identified an essential requirement for Drosophila Mask and its mammalian homologs Mask1 (ANKHD1) and Mask2 (ANKRD17) in promoting Yki/YAP transcriptional activity, but the mechanism by which Mask family proteins act has remained unclear (Dong et al., 2016;Machado-Neto et al., 2014;Sansores-Garcia et al., 2013;Sidor et al., 2013). We find that loss of Mask family proteins prevents nuclear import of Yki/YAP in both mammalian cells and Drosophila. Furthermore, while Mask is normally required for Yorkie to drive tissue growth, addition of an ectopic NLS to Yki is sufficient to bypass this requirement in Drosophila. Double conditional knockout of Mask1/ANKHD1 and Mask2/ANKRD17 in mouse intestinal organoids, together with siRNA knockdown of these proteins in human intestinal cells, confirms an essential requirement for Mask proteins in YAP nuclear import and stability. Finally, we show that overexpression of Mask1/2 is sufficient to stabilise YAP protein levels and can also drive phase separation of YAP into liquid droplets, suggesting that colloidal phase separation may contribute to the regulation of YAP activity.

Results
We began by examining whether Mask family proteins have a role in regulating the subcellular localisation of Yki, as we were unable to identify a direct transcriptional activation function for Mask ( Fig S1). Previously, we ruled out a possible role for Mask in promoting Yki nuclear import based on antibody staining for Yki in mask 10.22 null mutant clones in the Drosophila wing disc, where Yki is mostly cytoplasmic (Sidor et al., 2013). Recently, a Yki-GFP knock-in line revealed robust nuclear localisation of Yki in the mechanically stretched cells of the Drosophila ovarian follicle cell epithelium (Fletcher et al., 2018). We therefore induced mask 10.22 null mutant clones induced in the developing follicle cell epithelium, in which an endogenously tagged Yki-GFP knock-in is cytoplasmic at stage 10 but becomes strongly nuclear during stage 11 as the columnar cells are stretched mechanically (Fletcher et al., 2018) (Fig 1D,E). We find that Yki-GFP is lost from the nucleus and accumulates in the cytoplasm in mask mutant cells (Fig 1F,G).
These findings indicate that Mask proteins are required for normal nuclear localisation of Yki.
The above observations suggest a potential role for Mask in nuclear import of Yki. Mask family proteins have a conserved nuclear localisation signal (NLS) located just C-terminal to the second ankyrin repeat domain (Fig 2A). To test whether this NLS is required for the function of Mask in vivo, we generated a mask CRISPR-knockin lacking the NLS motif (mask∆NLS), which was homozygous lethal. Consistent with a role for Mask in nuclear import, clones of follicle cells homozygous for the mask∆NLS allele show a strong decrease in nuclear Yki-GFP and a corresponding increase in the level of cytoplasmic Yki-GFP (Fig 2B-D). These results demonstrate the essential requirement for the NLS motif in Mask function and support a key role for Mask in Yki nuclear import.
Since the phenotypic characterisation of mask mutants focused on the proliferating epithelia of Drosophila, such as the developing wing (Sansores- Garcia et al., 2013;Sidor et al., 2013), we sought to examine the role of Mask in regulating Yki-GFP in this tissue. Yki is known to be primarily cytoplasmic in the developing wing, which is composed of densely packed columnar epithelial cells (Oh and Irvine, 2008) (Fig 3A). Since clones overexpressing Yki (MARCM clones expressing tub.Gal4 UAS.Yki) were shown to require Mask in order to drive cell proliferation in the wing (Sidor et al., 2013), we examined whether Mask affects the nuclear localisation of overexpressed Yki.
We find that overexpressed Yki is readily detected in both nucleus and cytoplasm in control MARCM clones, but not in mask mutant MARCM clones, where the level of nuclear Yki is reduced relative to the level of Yki in the cytoplasm ( Fig 3A). To test the function of Mask in nuclear import of endogenous Yki, we examined the peripodial epithelium, which features strongly nuclear Yki-GFP. We find that silencing of Mask expression by RNAi in the peripodial epithelium with Ubx.Gal4, prevents the normally strongly nuclear Yki-GFP localisation in these cells ( Fig 3B). These results confirm that Mask is required for the nuclear localisation of both endogenous and overexpressed Yki in the developing wing of Drosophila.
The above results suggest that the reduced level of nuclear Yki may account for the failure of overexpressed Yki to drive cell proliferation in mask mutant clones in the developing wing. To test this notion, we linked an ectopic nuclear localisation sequence (NLS) and an epitope tag (HA) to Yki and expressed it in wild-type and mask mutant MARCM clones in the wing. We confirm that MARCM clones expressing tub.Gal4 UAS.Yki-NLS-HA are able to restore both nuclear localisation of Yki and cell proliferation in mask mutant cells (Fig 4A-G;(Sidor et al., 2013)). We note that even Yki-NLS-HA is still mildly less nuclear and more cytoplasmic when expressed in mask mutant cells compared to wild-type cells, and that this may therefore explain the mild difference in growth between these two types of clones (Fig 4E-G). We find similar results in the eye imaginal disc (Fig 4H-K). Thus, linkage of a nuclear localisation sequence to Yki can largely bypass the requirement for mask in Yki-driven cell proliferation in vivo. These results indicate that the primary function of Mask is to promote nuclear localisation of Yki in Drosophila.
To extend these findings to mammals, we performed siRNA knockdown experiments in human HEK293T and Caco2 cells. In both cases, silencing of Mask1 (ANKHD1) was sufficient to reduce both nuclear localisation of YAP and total YAP levels, as measured by immunostaining or western blotting of cell lysates (Fig 5A-C). Notably, double siRNA against Mask1/2 causes apoptosis of transfected cells, indicating that the two proteins act redundantly ( Fig 5C; Fig S2). Furthermore, we generated double conditional floxed mice for Mask1 (ANKHD1) and Mask2 (ANKRD17). The double floxed Mask1/2 mice were used to generate intestinal organoids, which were then infected with GFP-tagged Adenoviral Cre (AdCre-GFP) to drive deletion of Mask1/2 in clones. Clonal deletion of Mask1/2 resulted in a strong reduction in YAP levels compared to surrounding wild-type cells ( Fig 5D) and clone sizes were typically only 1-2 cells, with cells frequently undergoing extrusion from the epithelium and death ( Fig S3). These findings show that mammalian Mask1/2 have a conserved function in nuclear import of YAP but are additionally required to stabilise the YAP protein.
We next tested whether overexpression of Mask1/2 might be sufficient to cause mis-regulation of YAP in human cells. We find that overexpression of either Mask1 or Mask2 caused strong stabilisation of YAP protein levels in Caco2 cells (Fig 6A,B). Furthermore, overexpression of Mask1 in HEK293T cells, which are known to generate high levels of expression, caused formation of abnormally large droplets containing both Mask1 and YAP proteins (Fig 6). These results suggest that Mask proteins can cluster/polymerise YAP to drive colloidal phase separation -a classic mechanism of cellular compartmentalisation (Hardy, 1899;Iborra, 2007;Walter and Brooks, 1995;Wilson, 1899). Recent work has shown that cells expressing GFP-tagged YAP can exhibit colloidal phase separation and formation of YAP liquid droplets in both nucleus and cytoplasm after treatment with 25% PEG to induce macromolecular crowding (Cai et al., 2018). Our findings indicate that endogenous YAP can also phase separate when it becomes stabilised and concentrated by clustering with Mask family proteins.
To further explore the possible physiological role of colloidal phase separation, we examined the subcellular localisation of Yki/YAP and Mask family proteins. Yki-GFP localises to intracellular granules in either the nucleus and cytoplasm of follicle cells or cultured Drosophila S2 cells (Fig 7A).
Similarly, antibody staining for Yki and Mask reveals co-localisation in the same intracellular granules in Drosophila S2 cells (Fig 7B). In human cells in culture, immunostaining for YAP and Mask1 reveals a similarly granular pattern in both cytoplasm and nucleus (Fig 7C). These findings indicate that endogenous Yki/YAP and Mask proteins co-localise and cluster/polymerise to form intracellular granules (or 'hubs') and therefore provides evidence for physiological 'colloidal phase separation' of these proteins as they interact with cytoplasmic 14-3-3 or nuclear TEAD (Fig 7D).
We previously showed that human Mask1/2 proteins can relocalise with YAP from the cytoplasm to the nucleus of cultured human cells in a densitydependent fashion (Sidor et al., 2013). We later found that integrin signalling via Src family kinases is the primary mechanism of YAP mechano-regulation in these cells and is crucial to regulate wound healing Elbediwy et al., 2016). We therefore tested whether the density-dependent subcellular localisation of Mask1/2 proteins was affected by loss of integrin adhesions (by removing Calcium), by loss of mechanical forces (by treatment with the actin drug Latrunculin), by loss of Src family kinase activity (by treatment with the Src inhibitor Dasatinib), or after scratch wounding ( Fig   8A,B). In all cases, alterations in the localisation of Mask1/2 and YAP before and after treatment were the same, indicating a common mechanism of regulation (8A-C). Finally, we previously identified a physiological role for YAP (and TAZ) in skin wound healing, which causes YAP to localise to the nucleus in leading edge skin epithelial cells (Elbediwy et al., 2016). Mask1/2 are similarly nuclear localised during skin wound healing (Fig 8D), supporting the notion that these proteins are co-regulated in response to both mechanical cues and tissue damage in vitro and in vivo.

Discussion
Our results shed light on the molecular mechanisms governing the activity of Yki/YAP family of transcriptional co-activators. Dynamic and continuous nucleo-cytoplasmic shutting of both Yki and YAP is well established (Ege et al., 2018;Manning et al., 2018). Dynamic shutting enables the subcellular localisation of the bulk of Yki/YAP protein to be governed simply by the availability of binding partners such as cytoplasmic 14-3-3 proteins or nuclear Sd/TEAD. Indeed, shuttling enables Hippo signalling to control bulk Yki/YAP subcellular localisation by regulating phosphorylation of Yki/YAP and thus its binding to 14-3-3 proteins. In the absence of a Hippo signal, unphosphorylated Yki/YAP binds primarily to nuclear Sd/TEAD and thus adopts a bulk nuclear localisation while maintaining shuttling to allow continuous sensing of Hippo activity in the cytoplasm. Our results reveal a novel molecular mechanism for Yki/YAP nuclear import by Mask family proteins.
We have established the essential requirement for the canonical nuclear Although the above findings demonstrate the primacy of the Mask NLS in mediating nuclear import, it is possible that non-canonical 'RaDAR' nuclear import motifs that bind RanGDP within the ankyrin repeat domains of Mask family proteins may also contribute, although the corresponding residues in Mask proteins are not strongly hydrophobic so may only function weakly, if at all . Furthermore, it is possible that Importin-alpha binding non-canonical nuclear import motifs within Yki (Wang et al., 2016) and a Ranindependent nuclear import sequence in YAP/TAZ (Kofler et al., 2018) may also contribute to import. However, these non-canonical motifs are not sufficient to maintain normal nuclear localisation of either Yki or YAP in the absence of Mask proteins.
Our findings also establish a novel role for mammalian Mask1/2 in stabilising YAP protein levels. Unlike Drosophila Yki, YAP/TAZ stability is regulated by two SCF-type E3 ubiquitin ligases, beta-TrCP and FBXW7, which earmark YAP/TAZ for degradation after phosphorylation by LATS and CK1 on a specific phosphodegron containing Ser381 (Tu et al., 2014;Zhang et al., 2016;Zhao et al., 2010). Consequently, while loss of Drosophila Mask does not reduce total Yki levels, loss of Mask1/2 causes a dramatic loss of YAP protein, suggesting that binding of Mask1/2 to YAP is essential to prevent YAP degradation.
An interesting consequence of the role of Mask1/2 in stabilising YAP protein is that overexpression of Mask proteins is sufficient to dramatically raise the concentration of YAP within the cytoplasm, even causing formation of colloidal YAP liquid droplets in human cells. Similar YAP liquid droplets have recently been shown to occur by phase separation after treatment of cells expressing GFP-tagged YAP with 25% PEG to drive macromolecular crowding (Cai et al., 2018). Whether phase-separation of YAP has a physiological role is still unknown, but the possibility is supported by the fact that endogenous YAP can also be observed to phase separate when in a complex with Mask1 at high concentrations. In particular, phase-separation has been proposed to support formation of transcriptional enhancer complexes ("transcription factories") in the nucleus (Boehning et al., 2018;Boija et al., 2018;Hnisz et al., 2017;Iborra, 2007;Jackson et al., 1993;Lu et al., 2018;Sabari et al., 2018). Furthermore, Mask family proteins contain a conserved RNA binding KH-domain, and several RNA binding proteins have key roles in colloidal phase separation (Anderson and Kedersha, 2006;Lin et al., 2015;Maharana et al., 2018;Mao et al., 2011;Molliex et al., 2015;Ramaswami et al., 2013;Weber and Brangwynne, 2012). Finally, the fact that Mask family proteins contain two YAP-binding ankyrin repeat domains provides a plausible multivalent interaction mechanism by which Mask proteins can promote clustering of multiple YAP molecules in either the nucleus or cytoplasm ( Fig   7D).
In conclusion, our results identify novel molecular mechanisms regulating the Yki/YAP/TAZ family of transcriptional co-activators. Mask family proteins promote nuclear import of both Yki and YAP, while also acting to stabilise YAP protein levels, which can lead to phase separation into large droplets at high concentrations. Endogenous Mask1/2 and YAP co-localise in intracellular granules in both the nucleus and cytoplasm, and the nucleocytoplasmic distribution of Mask1/2 and YAP is co-regulated in response to mechanical cues in human cells. These functions of Mask family proteins are crucial to enable dynamic regulation of these co-activators by Hippo signalling and integrin signalling to control cellular behaviour.

Drosophila genetics
All Drosophila strains have been previously described in (Sidor et al., 2013) and (Fletcher et al., 2018) or are available from Bloomington Drosophila Stock Centre. Mosaic tissues were generated using the FLP/FRT and the MARCM system with a heat shock promoter (hs) to drive the expression of the FLP recombinase. Clones in imaginal discs were induced by heat shocking larvae at 60 hr (± 12 hr) of development and larvae were dissected at the third instar stage. Clones in ovarian follicle cells were induced by heat shocking adult females fed with yeast for 3 days before dissection at various times after heat shock.

Drosophila
Ovaries and imaginal discs were dissected in PBS, fixed for 20 mins in 4% paraformaldehyde in PBS, washed for 30 minutes in PBS/0.1% Triton X-100 (PBST) and blocked for 15 minutes in 5% normal goat serum/PBST (PBST/NGS). Primary antibodies were diluted in PBST/NGS and samples were incubated overnight at 4°C (Fletcher et al., 2018). The following primary antibodies were used: Secondary antibodies (all from Molecular Probes, Invitrogen) were used at 1:500 for 2-4 hours prior to multiple washes in PBST and staining with DAPI at 1µg/ml for 10-30mins prior to mounting on slides in Vectashield (Vector labs).

Human cells
Cells were fixed and stained by standard procedures using the following antibodies:

Luciferase assay in S2 cells
The luciferase assay was performed in sextuplicates in a 96 well plate. Cells were transfected with 20 ng of each DNA plasmid per well. 48 hours after transfection, cells were lysed and tested for Luciferase and Renilla activity using the Dual-luciferase reporter assay system kit (Promega) and a luminometer.

Mouse strains
All experiments were carried out in accordance with the United Kingdom Animal Scientific Procedures Act (1986) and UK home office regulations under project license number 70/7926.

Organoid cultures
Intestinal crypts were isolated from the proximal part of the small intestine of Passages were performed by resuspending Matrigel-embedded organoids in cold DPBS and transferring them to a Falcon tube using a 2 mL syringe with a 27 G ½ needle (BD Microlance #300635) to break them up. After two low speed washes in ADF-12, organoids were resuspended and plated in Matrigel.
(K) Eye imaginal disc containing null mutant mask MARCM clones also expressing nlsYki HA (nlsGFP+, green) are rescued for their growth and survival to a level similar to wild-type clones. Note the dramatic reduction in YAP protein levels (n=12 organoids).