Collective border cell migration requires the zinc transporter Catsup to limit endoplasmic reticulum stress

Collective cell migration is critical for normal development, wound healing, and in tumor progression and metastasis. Border cells in the Drosophila ovary provide a genetically tractable model to identify molecular mechanisms that drive this important cell behavior. In an unbiased screen for defects in border cell migration in mosaic clones, we identified a mutation in the catsup gene. Catsup, the Drosophila ortholog of Zip7, is a large, multifunctional, transmembrane protein of the endoplasmic reticulum (ER), which has been reported to negatively regulate catecholamine biosynthesis, to regulate Notch signaling, to function as a zinc transporter, and to limit ER stress. Here we report that catsup knockdown caused ER stress in border cells and that ectopic induction of ER stress was sufficient to block migration. Notch and EGFR trafficking were also disrupted. Wild type Catsup rescued the migration defect but point mutations known to disrupt the zinc ion transport of Zip7 did not. We conclude that migrating cells are particularly susceptible to defects in zinc transport and ER homeostasis.


Introduction
Collective cell migration has emerged as a key driver of normal organ development, wound repair, and tumor metastasis [1][2][3][4] . Border cell migration in the Drosophila ovary provides a powerful in vivo model of collective cell migration that is amenable to unbiased genetic screening. Drosophila ovaries are composed of ovarioles, which are strings of egg chambers progressing through 14 stages of development to mature eggs (Fig.1A). Each egg chamber is composed of 16 germ cells including 15 nurse cells and one oocyte, which are surrounded by epithelial follicle cells. During stage 9 ( Fig.1B and C), 4-8 border cells are specified at the anterior end of the egg chamber, delaminate from the follicular epithelium, and migrate posteriorly, reaching the anterior border of the oocyte by stage 10. Genetic screens have yielded insights into the molecular mechanisms that specify which of the ~850 follicle cells acquire the ability to migrate 5,6 , the developmental timing of the migration 7,8 , collective direction sensing, and cytoskeletal dynamics [9][10][11][12][13][14][15][16][17][18][19] . While much is understood, insights continue to emerge from border cell studies [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] . The gene catsup was identified both in a large-scale, ethyl methanesulfonate-induced mutagenesis screen for border cell migration defects in mosaic clones 35 and in a whole-genome gene expression profile 36 . The name catsup is an abbreviation of "catecholamines up", loss of which increases synthesis of aromatic amines including neurotransmitters such as epinephrine and dopamine 37 . Catsup is required for Drosophila tracheal morphogenesis, and in this context, it inhibits the Drosophila homolog of tyrosine hydroxylase (Ple) to limit dopamine synthesis 38 . In contrast, in wing imaginal disc cells, Catsup facilitates proper trafficking of Notch and EGFR 39 . Its mechanism of action in border cells is unknown. Catsup shares a 62% similarity and 53% identity with its mammalian homolog ZIP7 (also known as SLC39A7 or HKE4) 39 , a member of one of the two major families of zinc transporters 40 . ZIP7 is located within intracellular membranes including ER where ZIP7 transports Zn 2+ to the cytosol 41 . Zinc is a necessary trace element vital for many proteins to function, and zinc homeostasis requires 24 zinc transporters in humans, 14 of which are ZIPs 42 . Zip7 is a conserved protein found in the ER and Golgi in organisms as diverse as yeast, plants and animals [43][44][45][46][47][48] . In animal cells, loss of ZIP7 can lead to ER stress and in some cases cell death. Furthermore, increased ZIP7 expression is positively correlated with cancer cell proliferation, growth, invasion, and metastasis 49 of breast 50,51 , cervical 52 and colorectal cancer 53 . In the mammalian intestine, loss of ZIP7 causes an increase in ER stress and loss of stem cells 54 . Similarly, catsup loss-of-function causes ER stress in fly wing imaginal discs 39 . Thus, Catsup and ZIP7 are multifunctional proteins. However, the relationships between ER stress, zinc transport, and cell motility remain to be clarified.

Results
Using an endogenously-tagged Catsup::GFP fusion, we found that Catsup is expressed throughout oogenesis, including in all follicle cells ( Fig. 1A-C). Mammalian ZIP7 localizes predominantly to the ER 41 , so we investigated the subcellular localization of Catsup. Both overexpressed, tagged CatsupV5 ( Fig. 1D-L) and endogenously tagged Catsup::GFP ( Fig. 1M-U) significantly co-localized with the ER resident enzyme protein disulfide isomerase (PDI), but not with DNA or F-actin. Expressing a Catsup RNAi line in border cells using c306Gal4 ( Fig. 2A) and FruitlessGal4 led to incomplete migration in 80% of stage 10 egg chambers examined ( Fig.1B and C), a defect that was rescued by UAS-CatsupV5 (Fig. 1C). Using the FLP-FRT system, we generated clones of catsup mutant cells in genetically mosaic egg chambers. Compared to control clones ( Fig. 2D-D''), border cell clusters containing cells homozygous mutant for catsup exhibited migration defects ( Fig. 2E-E"), the severity of which was proportional to the percentage of mutant cells per cluster (Fig. 2F). In addition, in clusters containing both heterozygous and homozygous mutant cells, homozygous mutant cells tended to occupy rear positions ( Fig. 2G-G' and E), which is typical of mutations in genes required for motility 55 . One known function of Catsup is direct binding to and inhibition of the tyrosine hydroxylase Ple, which is the rate-limiting enzyme in catecholamine synthesis 56 . Ple and Catsup are both expressed in embryonic tracheal cells, where they contribute to achieving proper dopamine levels, which regulate Breathless (fibroblast growth factor receptor) endocytosis and signaling 38 . To test whether inhibition of Ple by Catsup was critical for border cells to migrate, we used an antibody to assess Ple expression in wild-type egg chambers. In contrast to tracheal cells, we detected no Ple protein in wild-type egg chambers (Fig. 3A). The antibody was effective because we could detect Ple ectopically expressed using c306Gal4 (Fig. 3B), as well as endogenous expression of Ple in neurons in the adult brain 57 (Fig. 3C). Furthermore, Ple overexpression in border cells caused no migration defect (Fig. 3B). Therefore, it is unlikely that negative regulation of Ple activity is the key function of Catsup in border cells, suggesting that the function of Catsup in border cell migration is likely distinct from its role in tracheal development.
In wing imaginal discs, Notch and EGFR signaling are disrupted by Catsup loss of function, and both of these pathways are required for border cell migration. Notch signaling facilitates initiation of border cell migration, specifically detachment from the anterior 58 , while EGFR is required for border cells to take a dorsal turn near the end of their migration to the oocyte 16,23 . We found abnormal intracellular accumulation of Notch in epithelial cells generally (Fig. 3D, D') and border cells specifically (Fig. 3E, E') upon Catsup knockdown ( Fig. 3D and E). Cells lacking Catsup also exhibited defective Notch signaling, detected by the Notch responsive element reporter 59 (Fig. 3E, E'). As in imaginal discs, EGFR also accumulated abnormally (Fig. 3G-I'). Since Notch signaling is essential for border cell migration and expression of constitutively active Notch (the Notch intracellular domain, NICD), which does not require intracellular trafficking or processing, rescues impaired Notch signaling in border cells 58 , we asked whether NICD expression might rescue Catsup knockdown. However, neither NICD expression nor overexpression of the Notch specific chaperone Ofucosyltransferase-1 60 was sufficient to rescue Catsup RNAi (Fig. S1). This suggests that multiple Catsup functions are essential for border cell migration.
Since suppression of ER stress is a conserved function of Catsup and ZIP7, we used the ER stress marker XBP1-EGFP 61 to compare heterozygous and homozygous catsup mutant cells in mosaic clusters (Fig. 4A). We found high levels of Xbp1 protein in homozygous cells compared to heterozygous border cells. Cells with elevated Xbp1 also exhibited reduced expression of Eyes Absent (Eya) (Fig. 4A), a nuclear protein that is required to repress polar cell fate and maintain border cell identity 6 . By contrast, the nuclear protein STAT, a transcription factor required for border cell fate specification 5 , was not decreased. Nuclear size was reduced by about half in mutant cells, possibly due to defective Notch signaling 62,63 (Fig. 4B). To test whether ER stress impairs migration, we genetically induced ER stress by overexpressing a misfolded rhodopsin protein RH1 G69D 64 . As expected, RH1 G69D induced Xbp1 expression in border cells (Fig. 4C). It also blocked migration (Fig. 4C), showing that high levels of ER stress are sufficient to inhibit border cell migration. In Xbp1-positive, RH1 G69Dexpressing cells, Eya expression was again reduced (Fig. 4C, insets), suggesting that ER stress, rather than another function of Catsup, was likely responsible for reduced Eya protein levels. ZIP7 transports zinc from the ER into the cytosol 65 . To test whether the zinc transporter activity of Catsup was likely required for border cell migration, we designed point mutations that alter amino acid residues that are conserved between Catsup, ZIP7 and a more distant family member from Arabidopsis IRT1, and that are required for zinc transport (Fig. 5B). Histidine is an amino acid that coordinates zinc 66 , and it appears in the highly conserved HELP domain and the CHEXPHEXGD motif that are important for Zinc transport 40 . The mutation Catsup H315A replaces a  conserved histidine within the HELP domain with alanine. Catsup H344A changes a histidine within the CHEXPHEXGD motif to alanine. We made transgenic flies expressing the mutants under Gal4 control and included a V5 tag so that we could monitor protein expression and localization. As a control, we made UAS-Catsup G178D , the point mutation present in the mutant line (Catsup 68E2 ) isolated from the screen for mutations that cause border cell migration defects in mosaic clones 35 . We then co-expressed each of the mutant lines with CatsupRNAi and quantified border cell migration (Fig. 5C). UAS-Catsup G178D provided no significant rescue compared to UAS-GFP-nls, though the mutation is likely a hypomorph and may have provided slight rescue that did not reach statistical significance (P=0.1). Castup H344A failed to rescue, as did expression of UAS-Catsup H315A , which though not statistically significantly different, might have caused an even more severe migration defect, possibly due to a dominant-negative effect (Fig. 5C). The point mutations did not destabilize the proteins or alter their localization (Fig. 5D, E), therefore the lack of rescue was likely a consequence of impaired activity rather than impaired expression.

Discussion
In this study we explored the roles of the multifunctional protein Catsup in border cell migration. Catsup is a conserved protein that goes by names including ZRT1 in yeast, IRT1 in plants, SLC39a7/Zip7/Ke4 in mammals. Despite the name Catsup (Catecholamines up), the most conserved features include subcellular localization to intracellular membranes including ER and Golgi, and bivalent cation transport. In Drosophila, Catsup has been implicated in direct binding and inhibition of tyrosine hydroxylase (TH/Ple), and thus in limiting dopamine production. This is important in the nervous system where as a neurotransmitter, dopamine levels must be tightly regulated. Somewhat surprisingly, the negative regulation of dopamine by Catsup is also important in tracheal development. Hsouna et al showed that Catsup and Ple are both required to achieve the appropriate level of dopamine, which regulates internalization of the Breathless (Btl) receptor tyrosine kinase. Excessive dopamine in catsup mutants leads to excessive endocytosis and thus downregulation of Btl, which inhibits tracheal cell migration. Ple mutations on the other hand result in reduced dopamine and Btl endocytosis, excess Btl signaling and therefore ectopic branching. Border cells also rely on chemotaxis via receptor tyrosine kinase signaling, so our first hypothesis was that the mechanism would be similar in tracheal and border cells. However, Ple is not detectable in border cells, so it is unlikely that negative regulation of dopamine synthesis is the primary function of Catsup in border cells. In Drosophila wing imaginal discs, Catsup was found to regulate trafficking of Notch and EGFR. Although the biological effects of Notch and EGFR are different in imaginal disc cells compared to border cells, both pathways are required in both cell types, and our results support a general role for Catsup in Notch and EGFR trafficking. Perhaps as a consequence of defective trafficking, Catsup knockdown also induced ER stress in border cells, supporting that this is a general and conserved function 54 . Moreover, we show for the first time that ER stress is sufficient to block migration and thereby regulate the abundance of the nuclear protein Eya, whereas STAT is unaffected. It is not immediately obvious how ER stress affects Eya abundance. Eya is an unusual protein in that it possesses protein phosphatase activity 67,68 , at least in vitro, and serves as a transcriptional activator 69 . Together the results demonstrate multiple essential functions for Catsup in border cells. The point mutations designed to disrupt zinc transport based on equivalent mutations in ZIP7 indicated that border cell migration requires not only expression and ER localization of Catsup but also its ability to transport zinc. Given the requirement for ZIP7 in cancer cell motility. its over-expression in numerous cancers, and its correlation with disease progression, invasion, and metastasis, the border cell system offers an excellent model for deciphering the key effects of this multifunctional protein on collective cell motility in vivo.

Design of UAS-RNAi-resistant Catsup point mutations
When generating UAS-Catsup-point-mutations, we designed the construct so it can not be targeted by the CatsupRNAi sequences. The region that is targeted by RNAi changed with the redundant codon for the same amino acids.

Immunostaining and confocal imaging
Female flies were fattened with yeast for 2 days at 29°C. Egg chambers are dissected from ovaries of female fly bodies in Schneider's medium with 10% FBS (pH=6.85-6.95) as described previously 71 . Freshly dissected egg chambers are fixed in 4% paraformaldehyde and then incubated overnight in 1xPBS with 0.4% triton with primary antibody to stain for ER mouse PDI (1:200) ADI-SPA-891-D Enzo Life Sciences, Inc., chicken GFP (1:200) ab13970 Abcam plc., Ple (anti-TH) antibody is a gift from Craig Montell lab, mouse anti-Notch intracellular domain (1:100) C17.9C6 DSHB, rat Ecadherin antibody DCAD2 (1:50) DSHB, V5 Tag Monoclonal Antibody-Alexa Fluor 555 (2F11F7) Invitrogen, mouse anti-dEGFR (1:2000) E2906 Sigma Aldrich. O-fut1 antibody was used to confirm O-fut1 overexpression, and it is a gift from Kenneth D. Irvine lab 72 . Secondary antibodies were incubated for 2 hours, together with Hoechst stains for nuclei, and Phalloidin stains for F-actin. Mouse anti-PDI and mouse anti-V5-555 co-staining was done by first stain with PDI primary and secondary, after through wash out, apply anti-V5-555 for overnight then wash out. Immunostained samples are mounted in VECTASHIELD Mounting Medium from Vector Laboratories. Zeiss LSM780 and LSM800 confocal microscopes were used to acquire images. Images were visualised by FIJI, rotated and cropped for presentation.

Sequence alignment
Catsup and ZIP7 amino acid sequences were acquired from NCBI in a FASTA format. The files were input into T-coffee http://tcoffee.crg.cat/apps/tcoffee/do:regular to generate multiple sequence alignment. The output was fed into Boxshade http://www.ch.embnet.org/software/BOX_form.html to generate the sequence alignment with black and grey shades to show conserved sequence region.

End Matter
Author Contributions and Notes X.G., W.D., and D.J.M. designed experiments and prepared the manuscript. X.G., and W.D., performed experiments. The authors declare no conflict of interest.