Overlapping expression patterns and functions of three paralogous P5B ATPases in Caenorhabditis elegans

P5B ATPases are present in the genomes of diverse unicellular and multicellular eukaryotes, indicating that they have an ancient origin, and that they are important for cellular fitness. Inactivation of ATP13A2, one of the four human P5B ATPases, leads to early-onset Parkinson’s disease (Kufor-Rakeb Syndrome). The presence of an invariant PPALP motif within the putative substrate interaction pocket of transmembrane segment M4 suggests that all P5B ATPases might have similar transport specificity; however, the identity of the transport substrate(s) remains unknown. Nematodes of the genus Caenorhabditis possess three paralogous P5B ATPase genes, catp-5, catp-6 and catp-7, which probably originated from a single ancestral gene around the time of origin of the Caenorhabditid clade. By using CRISPR/Cas9, we have systematically investigated the expression patterns, subcellular localization and biological functions of each of the P5B ATPases of C. elegans. We find that each gene has a unique expression pattern, and that some tissues express more than one P5B. In some tissues where their expression patterns overlap, different P5Bs are targeted to different subcellular compartments (e.g., early endosomes vs. plasma membrane), whereas in other tissues they localize to the same compartment (plasma membrane). We observed lysosomal co-localization between CATP-6::GFP and LMP-1::RFP in transgenic animals; however, this was an artifact of the tagged LMP-1 protein, since anti-LMP-1 antibody staining of native protein revealed that LMP-1 and CATP-6::GFP occupy different compartments. The nematode P5Bs are at least partially redundant, since we observed synthetic sterility in catp-5(0); catp-6(0) and catp-6(0) catp-7(0) double mutants. The double mutants exhibit defects in distal tip cell migration that resemble those of ina-1 (alpha integrin ortholog) and vab-3 (Pax6 ortholog) mutants, suggesting that the nematode P5Bs are required for ina-1and/or vab-3 function. This is potentially a conserved regulatory interaction, since mammalian ATP13A2, alpha integrin and Pax6 are all required for proper dopaminergic neuron function.

Introduction P-type ATPases are an ancient family of transmembrane proteins that use the energy derived from hydrolysis of ATP to actively transport substrates across membranes [1]. These transporters have four types of structural domains (Fig 1, Fig 2B): the actuator domain (A), the nucleotide binding domain (N), the phosphorylation domain (P) and the transmembrane domain (M) [2]. The signature characteristic of P-type ATPases is the highly conserved cytoplasmic DKTGT motif (P-domain, Fig 1) [1], which is autophosphorylated on aspartate during the catalytic cycle [3]. The P-type ATPases can be grouped into 5 subfamilies, P1-P5 [4]. The cellular functions and substrate specificities have been defined for one or more representatives of each of the P1-P4 subfamilies; however, specific substrates have not been definitively determined for either of the P5 subgroups, P5A and P5B [2,[5][6][7]. In this study, we focus on the P5B P-type ATPases, which have a putative substrate interaction motif of PPALP within transmembrane segment M4 [7,8].
The C. elegans genome encodes three P5B ATPases: CATP-5, CATP-6 and CATP-7. These proteins have a high degree of similarity, particularly in the M4 transmembrane domain (Fig  1, Fig 2D), which is thought to be critical for coordinating substrate in the binding pocket formed by M4, M5, M6, M8 and M9 [10] (Fig 1, Fig 2B). This suggests that CATP-5, CATP-6 and CATP-7 could have the same substrate specificity and therefore fulfill the same biochemical functions, but in different tissues and/or subcellular compartments. CATP-5::GFP has been shown to localize to the apical/surface of the intestinal cells and is required for the efficient uptake of polyamines from the gut lumen [11]. We previously showed that CATP-6 localizes to vesicular structures in multiple cell types, and that it acts to promote the function of the SLC16A transporter, GEM-1 [12]. No characterization of catp-7 has yet been reported in the literature. In this study, we use CRISPR/Cas9 to characterize all three C. elegans P5B P-type ATPases with regard to spatiotemporal expression pattern, subcellular localization and biological function in living animals.
The degron::mKate2 fusion was obtained by following the same procedure. But in addition we introduced the coding sequence of a degron motif [33] into the repair template, 5' of the coding sequence of mKate2 and flanked by two flexible aa linkers of the same sequence (GASGASGAS).

Imaging
Animals were mounted on 4% agarose pads for confocal, DIC and epifluorescence microscopy. Whole worms were immobilized with 10 mM levamisole and/or 1 mM sodium azide. Animals were imaged by using either Leica TCS SP5 II confocal microscope (Leica Application Suite LAS software) or Zeiss Axioskop 2 and MetaMorph software (Molecular Devices). Image processing was performed in Fiji/imageJ 2.0.0 [36], brightness and contrast was adjusted either in Fiji/imageJ 2.0.0 or Affinity Designer 1.6.0 (Serif).

Co-localization
Head neurons: Animals were immobilized with 1 mM sodium azide. Animals were mounted on 4% agarose pads. Spermatheca: Gonads were dissected in L-15/FBS medium [38] with 20 gauge hypodermic needles and mounted without any agar pad directly on microscope slides. Leica TCS SP5 II confocal microscope was used for imaging.

Brood size estimation
Triplicates of 10 L4 animals for each genotype were transferred to freshly seeded NGM plates (at 20˚C). To obtain the complete progeny of all 10 animals per triplicate, worms were transferred to new seeded NGM plates twice a day (morning and afternoon). Once the progeny on a plate reached young adult stage the plates were shifted to 10˚C. After six days, the complete progeny were washed off with MPEG (M9 plus 0.05% polyethylene glycol 8000 (PEG)) thereafter frozen at -80˚C. Water plus PEG (0.05%) was added to a volume of 30 ml. After shaking, 3 ml of worm suspension was taken to count the average brood size of 1 animal. Statistical analysis was performed by using RStudio Version 1.0.143 (https://www.rstudio.com). Shapiro-Wilk test in combination with graphical analysis was used to test for normality. Equality of variance was tested via Levene's test. We used an ANOVA (command: aov) and subsequent Tukey post-hoc test.

Growth rate estimation
Triplicates of 10 adult animals for each genotype were transferred to freshly seeded NGM plates (at 20˚C). All 10 animals were removed after exactly 1 h. 48 h later, the progeny were scored according their developmental stage. The progeny of all 3 replicates were pooled for analysis. Statistical analysis was performed by using RStudio Version 1.0.143 (https://www. rstudio.com). We used Fisher's exact test and subsequent Holm's correction for multiple comparison (command: pairwiseNominalIndependence).
The catp-6 locus produces three different mRNAs (WS262), each with a different 5' end region and encoding a distinct protein isoform. catp-6a and catp-6c encode isoforms that display the typical P5B ATPase structure (CATP-6a, 1256 aa, and CATP-6c 1207 aa). However, the isoform encoded by catp-6b is only 893 aa in length and lacks residues N-terminal to M3. We do not know whether CATP-6b is expressed and/or functionally significant in vivo; however, the RNA seq data summarized on WormBase (WS262), suggest that catp-6b is not a highly-abundant transcript.

Phylogenetic analysis of paralogous P5B ATPases
We obtained orthologous P5B ATPase aa sequences of five additional Caenorhabditis species (C. japonica, C. brenneri, C. briggsae, C. latens and C. remanei), more distant nematode species (Pristionchus pacificus, Onchocerca flexuosa and Trichinella spiralis) and Drosophila melanogaster as an outgroup from http://www.wormbase.org, http://www.flybase.org and from https:// www.ncbi.nlm.nih.gov. Since, the genomes of these species are thought to be fully sequenced [39][40][41][42], it is very likely that we have included all paralogous P5B ATPases for each species. We constructed a Maximum Likelihood tree by using the model LG+G [22] (Fig 3). We found that each Caenorhabditis species, except C. brenneri, has one CATP-5-like, one CATP-6-like and one CATP-7-like P5B ATPase. The CATP-6-like and CATP-7-like paralogs are more similar to each other then they are to the CATP-5-like paralogs, suggesting that catp-6 and catp-7 originated from the most recent gene duplication event. Although the genome of Pristionchus pacificus also codes for three paralogous P5B ATPases, these do not group with the paralogous P5B ATPases of the Caenorhabditis clade. The genomes of more distantly related nematodes, such as Onchocerca flexuosa and Trichinella spiralis, encode only a sole P5B ATPase.

CATP-5 expression pattern
In previous studies, the spatiotemporal expression patterns of CATP-5 and CATP-6 were characterized by the expression of tagged proteins using extrachromosomal arrays [11,12]. In the current study, we sought to verify and extend these findings by using both extrachromosomal arrays and by tagging the endogenous loci using CRISPR/Cas9.
In order to generate a tagged version of CATP-5 for use on extrachromosomal arrays, we fused the coding sequence of gfp to the 3' end of the catp-5 coding sequence, and drove expression using the putative catp-5 promoter (4.1 kb upstream of catp-5a.1). Consistent with the report of Heinick et al. [11], we found that CATP-5::GFP is expressed in the intestinal cells, where it localizes to the apical brush border (Fig 4A.5), and the excretory cell, where it appears to be distributed throughout the cytosol (Fig 4A.2,3). In addition, we observed CATP-5::GFP expression in the spermatheca (localized to the apical face, see below) and amphid sensory neurons (localized to the sensilla) (Fig 4A.1).
Since the presence of a C-terminal GFP tag could potentially interfere with proper subcellular localization, we used CRISPR/Cas9-mediated recombination to tag the endogenous CATP-5 protein at the N-terminus of CATP-5a (Fig 1) [31]. We refer to the protein encoded by the tagged locus (catp-5(dx187[gfp::catp-5 + loxP]) X), as native GFP::CATP-5. The expression pattern of native GFP::CATP-5 is nearly identical to that of CATP-5::GFP. However, we also detected expression on the plasma membrane of the distal tip cells (DTC) (Fig 4A.4) and on the plasma membrane of pachytene-stage germ cells (see below). We also generated a native  catp-5 allele in which gfp is inserted at the N-terminus of CATP-5c; however, we did not observe any GFP signal in these animals.

CATP-6 expression pattern
To create a red fluorescent protein tagged version of CATP-6 for use on extrachromosomal arrays, we fused the coding sequence for mKate2 to the 3' end of the catp-6 coding sequence, with expression driven by the putative catp-6 promoter (1135 bp 5' of exon 1 of catp-6a). The expression pattern of this transgene was essentially identical to that of the gfp-modified version of fosmid WRM067B_F08 used in our previous study [12]. However, we did observe that the signal from CATP-6::mKate was more finely resolved than that of CATP-6::GFP.
To obtain tagged native versions of CATP-6, we used CRISPR/Cas9 to produce two different tagged versions of catp-6, each with a fluorescent protein (FP) fused to the end of the coding sequence (Fig 1) [32]. One allele (catp-6(dx183[catp-6::gfp::3xFlag + loxP]) IV) utilized GFP, whereas the other (catp-6(dx179[catp-6::degron::mKate2::3xFlag + loxP]) IV) utilized mKate2 tag preceded by a degron motif [33]. (The degron segment is not functional in this genetic background). The expression patterns of the native FP-fusions are very similar to those observed using extrachromosomal transgenes. CATP-6::FP is expressed in a punctate pattern in many neurons in the head and tail, all body wall muscles, the posterior gut, the spermatheca, the DTC and the pharyngeal cells (Fig 4B and below). CATP-6::FP is also expressed in the gonadal sheath cells, where it appears to be closely associated with the plasma membrane and is not punctate (see below). Notably, native CATP-6::FP is strongly expressed in the germ line within the mitotic region, the transition zone and the pachytene region ( Fig 4B.5 and below), and appears to localize to the plasma membrane.

CATP-7 expression
To construct a tagged version of CATP-7 for use on extrachromosomal arrays, we inserted gfp at the 3' end of the catp-7 coding sequence, with expression driven by the putative catp-7 promoter (2377 bp 5' of exon1 of catp-7a). To tag the native locus, we inserted the coding sequence of either gfp or mcherry into the second exon of isoform catp-7a via CRISPR/Cas9 (catp-7(dx185[gfp::catp-7 + loxP]) IV and catp-7(dx191[mcherry::catp-7 + loxP]) IV), Fig 1,[31]). The expression and localization of the native N-terminal fusion GFP::CATP-7 closely resemble those of the C-terminally tagged extrachromosomal transgene. Native GFP::CATP-7 is expressed closely associated with the plasma membrane of the amphid sensory neurons (localized to the sensilla), the gonadal sheath cells, the spermatheca, the hypodermis and the excretory cell (Fig 4C and below). GFP::CATP-7 is expressed in most pharyngeal cells, where it localizes to the plasma membrane, plus internal membranous tubules (Fig 4C.2). GFP::CATP-7 is also expressed in the intestine, where it is associated with tubular structures in the basolateral domain and with vesicles in the apical domain immediately below the microvilli (see below; basolateral not shown). GFP::CATP-7 is strongly expressed in spermatocytes and spermatids (Fig 4C.4 and below). In the case of mCherry::CATP-7, we observed that the tagged protein was expressed and localized as for GFP::CATP-7, but it also accumulated in large aggregates within the cytosol (see below). This is most likely due to multimerization of mCherry, since this has been reported by others [43].
The CRISPR/Cas9-mediated single fp-fusion strains were not backcrossed after they were generated, but they did not show any obvious differences compared to wt N2. We estimated the average brood size for each single gfp-fusion strain in comparison with wt N2 (Table 1). There is no statistical difference detectable when the brood size of each CRISPR engineered strain is compared to N2. But the brood size of native catp-6::gfp differs from the brood size of native gfp::catp-5 ( Table 1, group b and a, ANOVA, Tukey post-hoc test, adj. p-value = 0.02327153).
Additionally, we estimated the growth rate of the progeny (1 h layoff) from 30 worms ( Table 2). 48 h after the layoff, 98-100% of the progeny of the single fp-fusion strain had reached the L4 stage, which is comparable to wt N2.
As expected, cytosolic GFP shows a calculated Mf of 92% (Fig 5E). Whereas CATP-5:: GFP excretory displays a Mf of only 0.33%, which is highly significantly different from cytosolic GFP (ANOVA, Tukey post-hoc test, adj. p-value = 0.0000307). Thus, it is very unlikely that the signal we observe is coming from cytosolic GFP in the excretory cell for transgene catp-5:: gfp excretory .
The During one imaging experiment, we serendipitously ruptured the basolateral membrane of the excretory cell and could thus observe consequences of cytoplasmic leakage into the extracellular fluid. In this case, we observed that CATP-5::GFP escaped from the cell in the form of small GFP-positive punctae, consistent with vesicular localization (Fig 5F). In addition, as cytosolic signal became depleted, it became possible to see that some of the CATP-5::GFP was tightly associated with the luminal membrane. We were able to reproduce the observation of persistent luminal association of CATP-5::GFP in two other experiments in which we used pressure on the coverglass to rupture the worm and sever the excretory canal.
Subcellular localization of CATP-6 in neurons and pharynx. In order to determine the subcellular compartment to which CATP-6 localizes, we imaged CATP-6::FP in relation to various compartment markers in head neurons via confocal microscopy.
As a marker for recycling endosomes we used immunostaining to assess colocalization of native CATP-6::GFP and RME-1 in neurons (Fig 6G). Again, although the pattern of RME-1 appears punctate, it is largely distinct from that of CATP-6::GFP.
In living spermathecal cells, CATP-6::GFP strongly colocalizes with LMP-1::RFP expressed from a transgene ( Fig 7A). However, the CATP-6::GFP positive vesicles are much larger than usual, so we suspected that the colocalization might be an artifact. Therefore, we used immunostaining to assess colocalization of native CATP-6::GFP and LMP-1 within the spermatheca (Fig 7B). In these preparations, the LMP-1 and CATP-6::GFP localization patterns are clearly different. LMP-1 is not associated with the basolateral cell membranes and the CATP-6 positive vesicles are relatively normal in size and shape.
In the case of RFP::RAB-5 we observed overlap with CATP-6::GFP in proximity to the plasma membrane of basolateral cell-cell junctions. This is consistent with the expected localization of RAB-5 to early endosomes (Fig 7C). The localization pattern of RFP::RAB-11 (expected to label recycling endosomes), is largely distinct from that of CATP-6::GFP ( Fig  7D). We used immunostaining to assess colocalization of native CATP-6::GFP and RME-1 as a second marker for recycling endosomes within the spermatheca (Fig 7E). Again, the pattern of RME-1 is largely distinct from that of CATP-6::GFP.

Overlapping expression patterns in developing reproductive tissues
All three paralogous P5B ATPases are expressed in the spermathecal epithelial cells. In order to directly compare their localization patterns, we constructed strains in which different pairs of proteins are tagged. Since the catp-6 and catp-7 loci are very tightly linked on chromosome IV, it was not possible to obtain a homozygous double-labeled strain by meiotic recombination. Therefore, we used CRISPR/Cas9 to generate a strain that expresses both native CATP-6::mKate2 and native GFP::CATP-7. The strain was not backcrossed but did not show any obvious difference compared to wt N2. We estimated the average brood size of this strain (Table 1) plus the growth rate of the progeny (1 h layoff) ( Table 2) and did not observe any difference compared to wt N2. In double-tagged animals, it is clear that the localization patterns of the two proteins within the spermathecal cells are distinct: CATP-6::mKate2 localizes to punctae in the basolateral domain, whereas GFP::CATP-7 localizes to the lateral and luminal cell borders, as well as to spermatids (Fig 8A).
By intercrossing, we obtained a strain that expresses both native CATP-6::mKate2 and native GFP::CATP-5 (Fig 8C). This double-tagged strain did not differ from wt N2 in terms of brood size or growth rate ( Table 1, Table 2). Native GFP::CATP-5 is expressed in a pattern qualitatively similar to that of native GFP::CATP-7, localizing to the apical lateral and luminal borders of the spermatheca (compare Fig 8A to Fig 8C), but GFP::CATP-5 is not detectable in spermatids. We obtained a strain that expresses both native mCherry::CATP-7 and native GFP::CATP-5 via crossing of the single fp-tagged strains (Fig 8B). Membrane-associated mCherry::CATP-7 signal localizes to the same lateral and luminal borders as GFP::CATP-5, confirming that both CATP-5 and CATP-7 localize to the same subcellular compartment in the spermatheca (Fig 8B). In the gonadal sheath cells, native GFP::CATP-7 localizes to large, raft-like patches on the plasma membrane, where it colocalizes with native CATP-6::mKate2 (Fig 8D). Moreover, we observe areas of the sheath cell that are positive for native CATP-6::mKate2, but not for native GFP::CATP-7 (Fig 8D). In favorable cross-section images, we were able to observe that native GFP::CATP-7 and native CATP-6::mKate2 are present on both the outer (pseudocoelom-facing) inner (germline-facing) plasma membrane surfaces of the sheath cell (see below).

Overlapping expression in the germ line
Native CATP-6::FP is strongly expressed in the adult germ line, and colocalizes with the plasma membrane marker ltIs44(P pie-1 mCherry::ph PLCδ ) within the mitotic, transition, and pachytene regions (Fig 9B). CATP-6::mKate2 is also associated with the plasma membrane of diakinesis-stage oocytes (Fig 9F). In L4 stage hermaphrodites, CATP-6::mKate2 localizes to vesicles in the rachis between the spermatocytes during spermatogenesis (Fig 9A and 9C). We also detect a weak cytoplasmic signal for CATP-6::mKate2 in spermatids (Fig 9E). Native GFP::CATP-7 is not expressed in the mitotic, transition or pachytene regions of the adult hermaphrodite germ line; however, it is strongly associated with the plasma membrane of spermatocytes and residual bodies in L4 larvae and young adults (Fig 9A and 9C, Fig 4C.4). In spermatids, GFP-CATP-7 localizes to punctae that probably correspond to fibrous body-membranous organelles (FB-MOs [44], Fig 9E). Native GFP::CATP-5 is expressed very weakly in the mitotic, transition and pachytene region of the germ line and also localizes to the plasma membrane ( Fig 9D).

Phenotypic analysis of single, double and triple null mutants
All three paralogous P5B ATPases are strongly expressed in the germ line and the somatic gonadal tissues (Fig 10A). Since the gonadal sheath cells and distal tip cell are known to be physiologically coupled to the germ line ( [45,46], Fig 10A and 10B), we considered it likely that the P5Bs would have essential functions during gonad development and germline proliferation. To test this hypothesis, we obtained and/or generated single, double and triple null mutant strains for catp-5, catp-6 and catp-7.
The single null mutants are all homozygous viable and fertile. catp-5(tm4481) is a 705 bp deletion that removes the coding sequences for 235 aa of M3 and M4 including the putative substrate binding site, the phosphorylation site (P domain) within the large intracellular loop, plus half of the N domain and also results in a frameshift (Fig 1). catp-5(0) animals have an overtly wt phenotype, as expected [11], with brood size and growth rate similar to those of wt N2 (Table 1, Table 2).
As one-day-old adults, catp-6(0); catp-5(0) animals are obviously different from wt N2. The DTCs inappropriately continue to migrate, leading to abnormal gonad arm positions ( Fig  11A). Although pachytene exit was not blocked, the growth of late stage oocytes was impaired and, unlike wild type worms, the first ovulation had not yet taken place (Fig 11A). A small fraction of the catp-6(0); catp-5(0) animals are fertile (1%), but these have a greatly reduced brood size, ranging from 1 to 19 progeny (Fig 11A and 11E).
Mid to late L4 catp-6(0) catp-7(dx189) worms exhibit a delay in DTC migration when compared to wt L4 of the same stage (S2 Fig), probably due to reduced germline proliferation. Young adult catp-6(0) catp-7(dx189) animals often show a fully reflexed gonad (S2 Fig). However, in contrast to wild type, no gametogenesis has yet taken place in the proximal arm (S2 Fig). Roughly half of analyzed gonads exhibit a DTC migration defect by omitting the second turn towards the midsection (S2 Fig and Fig 11B). 1 day old adult catp-6(0) catp-7(dx189) worms have often completed spermatogenesis, but oogenesis and ovulation are typically delayed (Fig 11B). The overall number of germ cells is usually much smaller than in wild type, suggesting a defect in germline proliferation.
We used DAPI staining of extruded gonads to determine whether gonadal sheath cells were still present in animals with defective germline proliferation. We were not able to unambiguously determine the exact number of sheath cell nuclei per gonad arm; however, we consistently observed multiple sheath cell nuclei (S3 Fig), thus indicating that the mutant phenotype does not result from a complete absence of gonadal sheath cells.
Among 1 day old adult triple mutants, some were able to produce both sperm and oocytes, but the oocytes were relatively small and abnormal in appearance. These animals also occasionally exhibited abnormal distal germline proliferation (Fig 11C), and some also had proximal germline proliferation. Most of the animals displayed severe defects in DTC migration with multiple inappropriate turns and perpetual DTC migration. Vul animals often had a large cavity in the midsection, consistent with the formation of at least a partial uterus in the absence of a vulva (Fig 11C).

Spatial and temporal expression patterns
We find that the C. elegans P5B ATPases are expressed in a wide range of tissues. Within these tissues, they exhibit both dynamic and overlapping expression patterns. These are summarized in Table 3 and Fig 10A. Depending upon context, each of the P5Bs may localize either to internal membranous structures or to the plasma membrane. However, each protein has distinct targeting signals, e.g., CATP-5 and CATP-7 both localize to the apical plasma membrane in spermathecal epithelial cells, but in the excretory cell CATP-7 is on the basolateral plasma membrane, whereas CATP-5 localizes to vesicles and the apical/lumenal membrane.

Overlapping functions in reproductive tissues
Each of the single mutants, catp-5(0), catp-6(0) and catp-7(0) is viable and fertile. However, we observed distinct synthetic sterile phenotypes in the catp-6(0); catp-5(0) and catp-6(0) catp-7  (Table 1, Table 2)), therefore we conclude that the fp-tagged alleles are functional. catp-6(0); catp-5(0) double mutants, have nearly normal somatic gonad development, but exhibit delayed germline proliferation and have very small brood sizes. Since both CATP-5 and CATP-6 are present on the plasma membrane of pachytene-stage germ cells, the simplest interpretation is that these two proteins function redundantly within this tissue to take up a common substrate from the extracellular fluid. However, CATP-6 is also expressed on the plasma membrane of the gonadal sheath cells, which are physiologically connected to the pachytene-stage germ line via numerous gap junctions [45]. Thus, the germ line should be able to obtain P5B transport substrates either by direct uptake or by transfer from the sheath cells (Fig 10B). Additional work will be necessary to resolve the basis for the sterility of the catp-6 (0); catp-5(0) double mutant.
catp-6(0) catp-7(dx189) double mutants exhibit variably severe defects in somatic gonad development, with uterine, spermathecal and sheath cells under-represented or absent. Given the known importance of the sheath cells for germline proliferation, their absence would be sufficient to explain the sterile phenotype of catp-6(0) catp-7(dx189) double mutants. However, we observed impaired germline proliferation even in animals where multiple sheath cells were present, so it is possible that sheath cell function is defective in catp-6(0) catp-7(dx189) animals.
The basis for the vulvaless phenotype of catp-6(0) catp-7(dx189) animals is also currently unresolved. The anchor cell could be missing and/or it could be defective in its ability to produce the inductive LIN-3 signal [47,48]. Furthermore, since CATP-6::mKate2 and GFP:: CATP-7 are both expressed in the developing vulval epithelial cells, it could be that the VPCs of double mutants are not competent to respond to LIN-3.

Distal tip cell migration phenotype
Although distal tip cell (DTC) migration is normal in each of the P5B single mutants, double and triple mutants show characteristic defects in migration of the DTCs. The perpetual DTC migration defect observed in catp-6(0); catp-5(0) double mutants is very similar to that observed in vab-3(lf) mutants, in which the alpha integrin ina-1 is not downregulated [49,50]. Furthermore, in catp-6(0) catp-7(dx189) double mutants, the DTCs frequently omit the second turn towards the center of the animal; this phenotype is very similar to that observed in mutants where either one of the integrin-associated proteins, SRC-1 or talin, is defective [51,52].
We speculate that catp-5, catp-6 and catp-7 function redundantly within the DTCs to allow correct localization and/or function of integrins and integrin-associated proteins. Given the known association between P5B ATPase function and Parkinsons disease, it may be significant that the human ortholog of vab-3 is Pax6, which has been shown to be important for dopaminergic neuron survival [53]. Moreover, the expression of integrin α5β1 is crucial for dopaminergic neurite outgrowth to the striatal cell region [54]. Since α5β1 integrin is known to be a target of Pax6 [55], we speculate that this is a conserved regulatory interaction, and that P5B ATPases may act coordinately to regulate integrin function.

P5B subcellular localization
Mammalian P5B ATPases have been reported to localize to lysosomes, late endosomes, multivesicular bodies, (ATP13A2; [9,[56][57][58] and ER (ATP13A4, [59]). While there are hints that ATP13A3 may localize to the plasma membrane [60], this has not been definitively established. We find that each of the nematode P5Bs can localize to the plasma membrane in certain tissues and stages of development. We observed that the CATP-6 localization pattern in spermathecal epithelial cells exhibits obvious similarity to that of the early endosomal marker RFP::RAB-5. Higher resolution imaging will be necessary to confirm that RAB-5 and CATP-6 colocalize at the level of individual vesicles. We did not observe substantial localization of CATP-6 to vesicular compartments marked by LMP-1 (lysosomes), RME-1 (recycling endosomes), RAB-11 (recycling endosomes), HGRS-1 (early endosomes), SS::GFP::CD4-HCIMPR (endosomes) or TGN-38 (trans-Golgi network). Therefore, we do not find strong evidence that the nematode P5Bs function primarily within lysosomes.

P5B substrate specificity
Although the transport specificity of P5B ATPases has not been unambiguously determined, various candidates have been proposed based on biological evidence. In the yeast, Saccharomyces cerevisaie, loss of P5B (Ypk9) function results in hypersensitivity to Mn 2+ , leading to the suggestion that Ypk9 detoxifies Mn 2+ by pumping it into the vacuole. In the slime mold, Dictyostelium discoideum, loss of P5B (Kil2) function prevents Mg 2+ -dependent killing of phagocytosed bacteria, leading to the suggestion that Kil2 pumps Mg 2+ into the phagosome. In mammalian tissue culture cells, loss of ATP13A2 function results in overaccumulation of Zn 2 + , suggesting that this P5B might play a role in Zn 2+ homeostasis [57]. On the other hand, ATP13A3 has recently been implicated as a candidate polyamine transporter [60]. With regard to the nematode P5Bs, the data of Heinick et al. [11] strongly suggest that polyamines are transported from the gut lumen into the intestinal cells via CATP-5 expressed on the plasma membrane. If CATP-6 and CATP-7 have the same specificity as CATP-5, then this would provide a simple explanation for their functional overlap within the germ line and the somatic sheath cells. This would also suggest that the sheath cells and the germ line are unable to autonomously produce sufficient quantities of polyamines to enable normal development and function of the reproductive system. Furthermore, given our original identification of CATP-6 as an interactor of the GON-2 Mg 2+ channel, it is possible that the P5Bs function as antiporters, exchanging polyamines and Mg 2+ between compartments. Some of the strains used in this study were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440), and the National BioResource Project (NBRP; https://shigen.nig.ac.jp/c.elegans/). This work was supported by a grant to E.J.L (LA 3380/2-1) from the Deutsche Forschungsgemeinschaft. Funding to B.C. was provided by the Center for Integrated Protein Science Munich.