Activation of RHO-1 in cholinergic motor neurons competes with dopamine signalling to control locomotion

The small GTPase RhoA plays a crucial role in the regulation of neuronal signalling to generate behaviour. In the developing nervous system RhoA is known to regulate the actin cytoskeleton, however the effectors of RhoA-signalling in adult neurons remain largely unidentified. We have previously shown that activation of the RhoA ortholog (RHO-1) in C. elegans cholinergic motor neurons triggers hyperactivity of these neurons and loopy locomotion with exaggerated body bends. This is achieved in part through increased diacylglycerol (DAG) levels and the recruitment of the synaptic vesicle protein UNC-13 to synaptic release sites, however other pathways remain to be identified. Dopamine, which is negatively regulated by the dopamine re-uptake transporter (DAT), has a central role in modulating locomotion in both humans and C. elegans. In this study we identify a new pathway in which RHO-1 regulates locomotory behaviour by repressing dopamine signalling, via DAT-1, linking these two pathways together. We observed an upregulation of dat-1 expression when RHO-1 is activated and show that loss of DAT-1 inhibits the loopy locomotion phenotype caused by RHO-1 activation. Reducing dopamine signalling in dat-1 mutants through mutations in genes involved in dopamine synthesis or in the dopamine receptor DOP-1 restores the ability of RHO-1 to trigger loopy locomotion in dat-1 mutants. Taken together, we show that negative regulation of dopamine signalling via DAT-1 is necessary for the neuronal RHO-1 pathway to regulate locomotion.


Introduction
The small Rho GTPase RhoA is a well-characterized regulator of the actin cytoskeleton [1,2]. In the nervous system RhoA acts on actin dynamics to regulate neuronal polarity [3], axon outgrowth [4], dendritic branching and synaptic connectivity [5]. RhoA has more recently been shown to play a role in neuronal activity by regulating endocytotic events [6], synaptic plasticity [7], and by enhancing learning and memory in adult mice [8]. Malfunctions in RhoA-signalling pathways are related to mental retardation phenotypes, which can be restored in the adult implicating new treatments for these disorders [9,10]. In C. elegans the single Rhoortholog (RHO-1) is expressed throughout the nervous system [11] and is required for cell migration, cell shape regulation, and neurite outgrowth [12][13][14]. Our previous work together with the work of others demonstrates that RHO-1 acts via multiple pathways to regulate locomotion in adult animals [15][16][17][18][19]. Inhibition of endogenous RHO-1 signalling results in lethargic locomotion with shallow body bends [16,18,19], whilst activation of RHO-1 signalling triggers a neuronal hyperactivity and loopy locomotion associated with increased acetylcholine (ACh) release and exaggerated body bends [18]. RHO-1 regulates neuronal function downstream of a network of G-protein signalling pathways in cholinergic neurons [15][16][17], of which at least two regulate locomotion: RHO-1 regulates locomotion downstream of G q via the activation of the NALCN ortholog NCA cation channels [16] and downstream of G 12 through the regulation of diacylglycerol kinase (DGK-1) and the recruitment of the synaptic vesicle protein UNC-13 to release sites [18]. However, the evidence suggests that there are unidentified down stream targets of RHO-1 that are required for the regulation of locomotion [18].
Dopamine signalling controls mood, reward, motor control, and learning in humans [20]. It is negatively regulated by the dopamine transporter (DAT), which removes dopamine from the extracellular space reducing the concentration and spread of dopamine [21]. DAT is the target of legal (modafinil and Ritalin) and illegal (cocaine and amphetamines) psychostimulants [22], implicating its importance in regulating dopamine signalling. Polymorphisms in human DAT have been related to mental health disorders such as ADHD, drug abuse, bipolar disorder, Tourette syndrome, and autism [23][24][25]. Loss of function mutations of DAT have been associated with infantile parkinsonism-dystonia [26] and mutations in genes encoding DAT-associated proteins such as α-synuclein and Parkin are associated with inherited forms of Parkinson's disease [27,28]. Pharmacological interventions targeting DATs through cocaine or amphetamine application result in the modulation of DAT function and lead to severe behavioural effects [29][30][31]. Therefore, understanding how DAT function, trafficking and gene expression are regulated has significance both biologically and sociologically. Many of these underlying mechanism have been studied [22,24,[32][33][34], highlighting DAT as an integral component of the dopamine signalling pathway, which regulates behaviour.
Dopamine, which is negatively regulated by DAT, has a central role in modulating locomotion in both humans and C. elegans [35][36][37]. In C. elegans, dopamine signalling mutants fail to switch from swimming to crawling locomotion when they move from a liquid to a solid surface, they also fail to slow down when they encounter a bacterial lawn [37,38]. In contrast, increased dopamine signalling slows locomotion accompanied by shallow body bends [39][40][41]. Dopamine via DOP-3 receptors negatively modulates the activity of NCA channels to reduce locomotion whereas RHO-1 signalling activates them [16,17]. Thus, the dopamine and RHO-1 pathways have opposite effects on locomotion. The C. elegans DAT-1 is conserved with human DAT, shares trafficking mechanisms and has similar kinetics of dopamine reuptake and sensitivity to drugs [42][43][44]. Activation of RhoA has been shown to regulate DATinternalization in mouse midbrain slices [31], and DAT-surface availability in C. elegans [43] cell autonomously linking RhoA to DAT function. In this study, we show that dat-1 is required for RHO-1 to modulate locomotion in vivo. We propose a model whereby enhanced RHO-1 signalling in cholinergic motor neurons leads to a cell non-autonomous transcriptional upregulation of dat-1 in dopaminergic neurons. This decreases the dopaminergic signalling that normally acts to reduce locomotion and enables RHO-1 signalling to drive locomotion. Our data suggest that the neuronal RHO-1 pathway regulates dat-1 gene expression to fine-tune locomotory behaviour.

RHO-1 signalling regulates locomotion
Previously we have shown that RHO-1 signalling in adult cholinergic neurons is required for normal locomotion [18,45]. Reducing RHO-1 signalling, by expressing the RHO-1 inhibitor C3 transferase in adult cholinergic neurons or using the Rho GEF mutant unc- 73(ce362), results in lethargic locomotion characterized by flat, shallow body bends [18,19] while expressing the constitutively active RHO-1 (RHO-1 (G14V)) mutant from the unc-17 promoter in cholinergic neurons (nRHO-1 Ã ) drives hyperactive locomotion, best characterized by exaggerated loopy body bends (Fig 1A, S2 Movie) [18]. For a more detailed analysis of the nRHO-1 Ã mediated loopy locomotion behaviour we quantified body curvature using the Worm Tracker 2.0.4 system and the Worm Analysis Tool Box version 3 [46]. The degree of curvature of hyperactive loopy nRHO-1 Ã animals compared to wild type animals was significantly increased at each of the five equally spaced points along the body (all p< = 0.0001) (Fig 1B and 1C, S1 and S2 Movies), indicating that nRHO-1 Ã increases curvature along the entire body ( Fig 1C). For clarity one point on the worm, mean hip (point 4), was chosen to represent alterations in curvature on subsequent strains (See Material and Methods). Given that nRHO-1 Ã animals have altered body curvature we employed a second locomotion assay, the dispersal assay, to determine the effect of this change on the animal's ability to cover distance over time. Animals are placed in the centre of a large NGM plate that has a narrow ring of food around the outside edge. The number of animals that reach the ring of food, at given time points, are scored as a percentage of the total number of animals on the plate [47] (Fig 1D). We observed that 95.2% ±0.8 of the wild type animals reached the food after 60 minutes while within this time only 0.5% ±0.5 of nRHO-1 Ã animals did so (p< = 0.0001) (Fig 1E). Approximately 50% of the nRHO-1 Ã animals reached the food when plates were left overnight (data not shown) suggesting that despite being loopy nRHO-1 Ã hyperactive animals are still able to direct their locomotion to a certain extent, although at a far slower rate when compared to wild type animals.

dat-1 mutations suppress nRHO-1 Ã mediated changes in locomotion
Our previous work together with the recent work of others [16,18] demonstrates that RHO-1 acts via multiple pathways to regulate locomotion. One of these pathways involves RHO-1 negatively regulating diacylglycerol kinase (DGK-1) to increase diacylglycerol (DAG) levels at release sites resulting in the recruitment of UNC-13. However, other DGK-1 independent pathways acting downstream of RHO-1 remain to be identified [18]. To identify these other components of neuronal RHO-1 signalling pathways we performed a forward genetic screen for mutations that suppressed nRHO-1 Ã mediated increases in curvature. The original suppressor screen strain had both versions of active RHO-1, nRHO-1 Ã and heat shock hs::RHO-1 Ã , only alleles supressing both were chosen to be suppressor mutants. One mutant, nz99, showed a significant decrease in curvature from 33.1˚±0.5 in nRHO-1 Ã to 27.1˚±0.8 in nRHO-1 Ã ;nz99 animals (p< = 0.0001) (Fig 2B). Whole genome sequencing of the nRHO-1 Ã ; nz99 strain identified 42 mutations predicted to alter protein coding (S1 Table). Among these nRHO-1 Ã expression alters locomotion phenotypes. a Representative images of control strains (wild type, dat-1(ok157)) and different nRHO-1 Ã animals showing characteristic body curvature and motion tracks on a culture dish (8x magnification). b Worm tracking data of body curvature at 5 body points within the worm (as shown in c) comparing control animals (wild type) to nRHO-1 Ã animals. Statistical comparisons were performed using one-way ANOVA and presented as mean curvature ± SEM with Tukey's multiple test correction. Diamonds indicate outliers. Statistical significance indicated as ( ÃÃÃ P 0.001), n = 70 and 32 animals respectively. c Schematic of a C. elegans. Reference points to measure body curvature using worm tracker terminology and indicated as red dots.d Schematic presentation of dispersal assay as described in materials and methods. e Dispersal assay as described in materials and methods. Genotypes are color-coded as indicated below the panel. Hash keys represent data points 1%. Data presented as the mean percentage of animals on food at indicated time points ± SEM. Statistical comparisons were performed using two-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P 0.001), n = 3 experiments (~200 animals per assay). A dat-1 mutation suppresses nRHO-1 Ã locomotion phenotypes. a Schematic of the dat-1 gene with alleles ok157 and nz99 indicated. b Worm tracking data of body curvature measurements of control animals (wild type, dat-1(ok157)), nRHO-1 Ã animals, nRHO-1 Ã suppressor strains (nRHO-1 Ã ;dat-1(nz99) and nRHO-1 Ã ;dat-1(ok157)) and rescue lines. Statistical comparisons were performed using one-way ANOVA and presented as mean hip ± SEM with Tukey's multiple test correction. Diamonds indicate outliers. Statistical significance indicated as ( ÃÃÃ P 0.001, n.s. = not significant), n = 70, 14, 32, 19, 15, 18, and 17 animals respectively.c Dispersal assay as described in materials and methods. Genotypes are color-coded as indicated on the right side of the panel. Hash keys represent data points 1%. Data presented as the mean percentage of animals on food at indicated time points ± SEM. Statistical comparisons were performed using two-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P 0.001, n.s. = not significant), n = 3 experiments (~200 animals per assay).d Aldicarb assay as described in materials and methods. Control strains (wild type, dat-1(ok157)), nRHO-1 Ã animals and nRHO-1 Ã ;dat-1(ok157) double mutant assayed on 1mM Aldicarb. Genotypes are color-coded as indicated below the panel. Data presented as mean percentage of paralyzed animals scored every 10 minutes up to 100 min ± SEM, n = 3 (>30 animals per assay).

dat-1 is not required for nRHO-1 Ã increased ACh release
Constitutively active RHO-1 signalling increases the release of ACh onto body wall muscles, which leads to a faster rate of paralysis in the presence of the acetylcholinesterase inhibitor aldicarb (Fig 2D) [18]. We have previously assumed that the hyperactive loopy locomotion of nRHO-1 Ã animals is a result of increased ACh release hyper-activating muscles. Interestingly, we found that dat-1(ok157) has no impact on the increased ACh release (hyperactive phenotype) observed in constitutively active RHO-1 animals, as the nRHO-1 Ã ;dat-1(ok157) double mutants remain hypersensitive to aldicarb despite being less loopy. For example, at 50 minutes, 0% ± 0.0 of wild type animals were paralysed compared to 38.3% ±7.3 of nRHO-1 Ã (p< = 0.0001), or to 43.3% ±7.3 of nRHO-1 Ã ;dat-1(ok157) (p< = 0.0001) ( Fig 2D). These data indicate that increased ACh release at the neuromuscular junction is not solely responsible for the increase in body curvature and dispersal phenotypes in animals with enhanced RHO-1 signalling. Our data suggest additional DAT-1 dependent pathways are also required to cause the observed changes in locomotory behaviour.

Dopamine suppresses the locomotion phenotypes of constitutively active RHO-1 signalling
DATs negatively regulate dopamine signalling by removing dopamine from the extracellular space, thus DAT reduces the concentration and the spread of dopamine released from the dopaminergic neurons [21]. We predicted that dat-1 (ok157) animals, which lack DAT-1, could potentially have increased extracellular dopamine levels and dopamine signalling. We used two approaches to examine these possibilities. Firstly, if increased dopamine signalling is required to suppress nRHO-1 Ã locomotion then this suppression would be compromised in mutants defective for dopamine synthesis such as cat-2, which encodes a tyrosine hydroxylase, a rate-limiting enzyme required for dopamine synthesis [50]. Our results show that the cat-2 (e1112) mutation restores curvature and dispersal in nRHO-1 Ã ;dat-1(ok157) animals back to similar levels observed in nRHO-1 Ã animals (curvature of nRHO-1 Ã ;dat-1(ok157);cat-2(e1112) 34 3B). These data indicate that dopamine and nRHO-1 Ã co-suppress each other's locomotion phenotypes, suggesting that these pathways antagonistically regulate locomotion.

RHO-1 signalling regulates dat-1 expression
Our data suggest that RHO-1 and dopamine signalling antagonistically regulate locomotion, but by what mechanism does RHO-1 signalling reduce dopamine signalling to drive locomotion? DAT-1 is the major negative regulator of dopamine signalling and is required for RHO-1 mediated changes in locomotion. Our data indicates that nRHO-1 Ã animals paralyze at a slower rate compared to wild type animals when exposed to exogenous dopamine ( Fig 3D). We investigated whether the expression or localisation of DAT-1 was altered in nRHO-1 Ã animals. We first created a p.dat-1::mCherry::DAT-1 translational fusion using the same genomic fragment that we used for the dat-1 rescue (Figs 1A and 2B and 2C). mCherry signals were detected in the four pairs of dopaminergic cells (PDE, ADE, CEPD and CEPV) in both wild type and nRHO-1 Ã animals suggesting the expression pattern of dat-1 was unchanged by nRHO-1 Ã [41,51]. In nRHO-1 Ã animals we saw an obvious increase in the fluorescence of the p.dat-1::mCherry::DAT-1 reporter compared to wild type animals expressing the same transgene suggesting that RHO-1 increases mCherry::DAT-1 protein levels ( Fig 5A). Moreover, this increase could be induced in adult animals by expressing constitutively active RHO-1 from a heatshock-inducible promoter ( Fig 5B). In contrast, expression of a co-injected pan-neuronal GFP reporter (p.snb-1::GFP) showed no obvious change (Fig 5A and 5B). We then tested our observation, that dat-1 expression is increased in the presence of nRHO-1 Ã by analysing the mRNA level of the endogenous dat-1 gene with qPCR. dat-1 mRNA levels were almost four times higher in nRHO-1 Ã animals with 231.1 ±30.9 copy no/reaction compared to wild type with 60.8 ±7.0 copy no/reaction (p< = 0.0001) (Fig 6A). dat-1 is expressed exclusively in dopaminergic cells, an expression pattern shared with cat-2, which is required for dopamine synthesis [52]. cat-2 expression however, was not significantly altered by constitutively active nRHO-1 Ã (p = 0.998) (Fig 6B). We also analysed the expression levels of the 4 dopamine receptors (dop-1-4), and a general neuronal gene synaptobrevin (snb-1) to exclude the possibility that nRHO-1 Ã was affecting global transcription. We observed no significant change in expression of dop-1-4 or snb-1 in nRHO-1 Ã animals compared to wild type animals (Figure a-e in S3 Fig). These data indicate that RHO-1 signalling regulates dat-1 expression in particular, and not generally the expression of components of the dopamine signalling pathway such as dop-1-4 or cat-2. However, the transcription of other genes not tested here may also be altered in nRHO-1 Ã animals. The constitutively active RHO-1 construct is under the unc-17 promoter which drives expression in cholinergic neurons [17,18], therefore nRHO-1 Ã and dat-1's expression pattern best fits a cell non-automonous regulation model.

Increased dat-1 expression is not a consequence of increased body curvature, or increased ACh, or neuropeptide release
We investigated whether the increase in body curvature or the presence of DAT-1 protein itself could cause the increase in dat-1 expression in nRHO-1 Ã animals. Although the dat-1(ok157) mutant fails to produce functional DAT-1 protein, the dat-1 gene retains the promoter and transcriptional start sites. Using primers 5' of the dat-1(ok157) deletion ( Fig 6D) we detected an even higher increase of dat-1 mRNA in nRHO-1 Ã ;dat-1(ok157) animals of 480.4 ±49.5 copy no/reaction than we did in nRHO-1 Ã animals with 231.1 ±30.9 copy no/reaction (p< = 0.0001) ( Fig 6A). As nRHO-1 Ã ;dat-1(ok157) animals lack both functional DAT-1 protein and exaggerated body bends we concluded neither of these are required for nRHO-1 Ã mediated increases in dat-1 transcription. We have previously shown that neuronal RHO-1 negatively regulates the diacylglycerol kinase DGK-1 [18]. dgk-1 mutations partly mimic constitutively active RHO-1 Ã animals as they have elevated ACh release [18,53] and increase their body curvature to a level intermediate between that of wild type and nRHO-1 Ã animals [53] and similar to that of nRHO-1 Ã ;dat-1(ok157) animals (dgk-1 22.7˚±0.6 compared to nRHO-1 Ã ;dat-1 (ok157) 25.8˚±0.6, p = 0.0042) (Fig 6C). However, unlike either nRHO-1 Ã or nRHO-1 Ã ;dat-1 (ok157) animals dgk-1 mutants do not increase dat-1 expression (57.1 ±10.9 copy no/reaction compared to wild type 60.8 ±7.0, p> = 0.9999) ( Fig 6A). These data indicate that the upregulation of dat-1 expression represents a new RHO-1 neuronal signalling pathway acting independently of DGK-1, body curvature and elevated ACh release. In addition to neurotransmitters, neuropeptides play a key role in regulating locomotion [54][55][56]. We hypothesised that neuropeptides may be involved in our nRHO-1 Ã predicted cell non-autonomous regulation of dat-1 expression since a number of studies have shown that neuropeptides act in a cell non-autonomous manner [57] and can regulate gene expression [58,59]. Bhattcharya et al, showed that the neuropeptide NLP-12 modulates dopamine signalling via DOP-1 and that over expression of nlp-12 (NLP-12 OE) induced a hyperactive locomotion phenotype [60]. NLP-12 OE animals have a similar curvature (29.7 o ±1.4) to nRHO-1 Ã animals (33.1 o ±0.5, p = 0.01) compared to wild type animals (17.4 o ± 0.3, p< = 0.0001) ( Fig  6E). We analyzed the mRNA expression level of nlp-12 in nRHO-1 Ã animals and observed a significant increase in nlp-12 mRNA in nRHO-1 Ã animals with 28,954 ± 2211 copy no/reaction compared to wild type with 11,780 ± 354.2 copy no/reaction (p< = 0.0001) (Fig 6F). However, dat-1 mRNA levels in the NLP-12 OE animals (86.7 ± 8.4 copy no/reaction) were not significantly different from wild type animals with 60.8 ± 7.0 copy no/reaction (p = 0.979) (Fig 6G). These data suggest that although nRHO-1 Ã did significantly increase nlp-12 expression the over-expression of nlp-12 alone does not increase dat-1 expression. Since high expression of NLP-12 alone does not increase dat-1 expression we sought to determine whether any neuropeptides were involved in regulating dat-1 expression, by using mutants of two proteins required for the priming and the exocytosis of neuropeptide loaded dense core vesicles, protein kinase C (PKC-1) and the CAPS-1 protein UNC-31 [61,62]. We analysed dat-1 mRNA expression levels in a gain of function PCK-1 mutant (pkc-1(gf)nuIs131), which has elevated levels of neuropeptide release and is hypersensitive to aldicarb indicating a hyperactive locomotion phenotype [61], and also in the UNC-31 mutant (unc-31(e928)), which has reduced neuropeptide release and is lethargic [62,63] (Fig 6E). These mutants, however, did not show significant changes in dat-1 mRNA expression levels (pkc-1(gf)87.4 ± 17.3 and unc-31(e928) 129.7 ± 11.1 copy no/reaction) compared to wild type animals with 60.8 ± 7.0 copy no/reaction (p> = 0.05) (Fig 6G). We also analysed nRHO-1 Ã ;unc-31(e928) double mutants to see if reduced neuropeptide release caused by the unc-31 mutation would decrease the increased dat-1 expression caused by nRHO-1 Ã . We observed a slight decrease in dat-1 mRNA expression in nRHO-1 Ã ; unc-31(e928) animals with 167.4 ± 30.4 copy no/reaction when compared to nRHO-1 Ã animals with 231.1 ± 30.9 copy no/reaction however this decrease was not significant (p = 0.294) ( Fig  6G). The limitation of using the pkc-1(gf) and the unc-31 mutants is that we are targeting general neuropeptide release and therefore cannot distinguish between the approximately 250 individual neuropeptides that have been identified to date, many of which have opposing effects on behaviour [64]. Therefore our data do not fully rule out the possibility that a specific Fig 6. nRHO-1 Ã increases transcription levels of dat-1. a and b Transcription levels of dat-1 or cat-2 in nRHO-1 Ã single and nRHO-1 Ã ;dat-1(ok157) double mutants compared to control strains (wild type, dgk-1(sy428)) determined by quantitative PCR. Data presented as mean normalized copy number per reaction ± SEM. Statistical comparisons were performed using one-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P 0.001, n.s. = not significant), n = 6, 6, 3, 4 for a and n = 5, 5, 4 for b independent samples for each genotype respectively. c Worm tracking data of body curvature of control animals (wild type, nRHO-1 Ã , nRHO-1 Ã ;dat-1(ok157) the same data used in Figs 2B, 3A and 4A), or dgk-1 (sy428) animals. Statistical comparisons were performed using one-way ANOVA and presented as mean hip ± SEM with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P 0.001), n = 70, 32, 14, and 15 animals respectively.d Schematic of the dat-1 gene with the ok157 allele indicated. qPCR primer set for dat-1 within first and second exon shown as arrows.e Worm tracking data of body curvature of control animals (wild type, nRHO-1 Ã the same data used in Figs 2B, 3A, 4A and 6C), unc-31(e928), nRHO-1 Ã ;unc-31(e928) and NLP12 OE animals. Statistical comparisons were performed using one-way ANOVA and presented as mean hip ± SEM with Tukey's multiple test correction. Triangles indicate outliers. Statistical significance indicated as ( Ã P 0.05, ÃÃÃ P 0.001), n = 70, 32, 12, 16 and 14 animals respectively.f Transcription levels of nlp-12 in nRHO-1 Ã compared to control (wild type) and NLP12 OE strains determined by quantitative PCR. Data presented as mean normalized copy number per reaction ± SEM. Statistical comparisons were performed using one-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P 0.001) n = 6, 6 and 3 independent samples for each genotype respectively. g Transcription levels of dat-1 in control (wild type, nRHO-1 Ã the same data used in Fig 6A) compared to pkc-1(gf), unc-31(e928), nRHO-1 Ã ;unc-31(e928) and NLP12 OE strains determined by quantitative PCR. Data presented as mean normalized copy number per reaction ± SEM. Statistical comparisons were performed using one-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( Ã P 0.05, n.s. = not significant), n = 6 for all apart from n = 3 for NLP12 OE independent samples for each genotype. https://doi.org/10.1371/journal.pone.0204057.g006 RHO-1 requires DAT-1 to control C. elegans locomotion set or single neuropeptide transmits the signal from cholinergic nRHO-1 Ã expressing cells to dopaminergic neurons to upregulate dat-1.
The interpretation of these data depends on our assumption that the copy number of the transgenes is proportional to the expression of dat-1 from these transgenes and that the 1ng of p.cat-2::dat-1 produces a physiologically relevant amount of dat-1 protein. We measured dat-1 mRNA levels using dat-1 primers that anneal within the dat-1(ok157) deletion (Fig 7D and  7E); this ensures that we measure transcripts only from the transgene and avoid any mRNA expression from the defective endogenous dat-1 gene. In the absence of any transgene dat-1 mRNA levels in nRHO-1 Ã ;dat-1(ok157) were 7.2 ±0.3 copy no/reaction. In nRHO-1 Ã ;dat-1 (ok157) animals transgenes made with 1ng, 10ng, and 20ng of p.cat-2::dat-1 plasmid the expression of dat-1 was 43.1 ±19.2 (p = 0.99), 377.0 ±100.3 (p = 0.029), and 715.9 ±178 (p< = 0.0001) (Fig 7D), indicating that the increase in transgene concentration results in a proportional increased in dat-1 expression. The expression from the 1ng of p.cat-2::dat-1 transgene was 43.1 ±19.2 which is very similar to that of wild type dat-1 expression levels 53.5 ±15.9 (p = 0.99) (Fig 7D). Injection of 1ng of p.cat-2::dat-1 plasmid was able to rescue the previously described swimming induced paralysis (SWIP) phenotype of the dat-1(ok157) animals [65,66] (Fig 7Cc), indicating that these transgenes do express dat-1 mRNA at a level that produces sufficient functional DAT-1 protein to reduce this phenotype. However, the transgenes in the  -1 ( o k 1 5 7 ) d a t -1 ( o k 1 5 7 ) p . c a t -  Increased dat-1-levels essential for nRHO-1 Ã mediated loopy locomotion. a Worm tracking data of body curvature of control animals (wild type, nRHO-1 Ã or nRHO-1 Ã ;dat-1(ok157) the same data used in Figs 2B, 3A, 4A and 6C) versus nRHO-1 Ã ;dat-1(ok157) expressing increasing concentrations of an RHO-1 Ã independent dat-1-rescuing construct (p.cat-2::dat-1). One line of each concentration is shown as representative from two lines. Statistical comparisons were performed using one-way ANOVA and presented as mean hip ± SEM with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P < 0.001, n.s. = not significant), n = 70, 32, 15, 19, 20, and 20 animals respectively. b Dispersal assay as described in materials and methods. Genotypes are color-coded as indicated below the panel. nRHO-1 Ã ;dat-1(ok157) expressing increasing concentrations of an RHO-1 Ã independent dat-1-rescuing construct (p.cat-2::dat-1). One line of each dat-1-rescuing construct concentration is shown as representative from two lines. Data presented as the mean percentage of animals on food at indicated time points ± SEM. Statistical comparisons were performed using two-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃ P 0.01-0.001; ÃÃÃ P < 0.001), n = 3 experiments (~200 animals per assay). c RHO-1 Ã independent dat-1-rescue construct p.cat-2:: SWIP assays are independent from the 1ng of p.cat-2::dat-1 transgene in the nRHO-1 Ã ;dat-1 (ok157) animals and may not produce identical amounts of protein. Our data therefore suggest that dat-1 expression levels correlate with the increased ability of RHO-1 signalling to trigger locomotion with high curvature and a failure to disperse, however we have not fully confirmed this. Taken together our data suggest that dat-1 expression is necessary for RHO-1 signalling to drive locomotion and that dat-1 is a new component of a neuronal RHO-1 signalling pathway that regulates locomotion.

Discussion
In this study we describe our finding that the C. elegans ortholog of RhoA, RHO-1, modulates locomotory behaviour by repressing dopamine signalling through the upregulation of the dopamine re-uptake transporter DAT-1. The most striking phenotype of expressing constitutively active RHO-1 in cholinergic neurons (nRHO-1 Ã ) is a switch from normal sinusoidal locomotion to a hyperactive loopy locomotion [18]. This phenotype was quantified in two ways: worm tracking to analyse the body curvature and dispersal assay to analyse the animal's ability to coordinate their locomotion towards a food source. nRHO-1 Ã animals have increased body curvature with much larger body bends and they cannot coordinate their locomotion and fail to disperse, together we refer to this as loopy locomotion. The ability of nRHO-1 Ã to trigger loopy locomotion is strongly reduced in two separate dat-1 mutant alleles, nz99 and ok157. We rescued this phenotype by expressing a wild type dat-1 transgene suggesting that DAT-1 has a role in RHO-1 mediated loopy locomotion. The dat-1(ok157) also reduced RHO-1 Ã loopy locomotion following heat shock of hs::RHO-1 Ã animals, indicating this is not a strain specific effect. The reduction of loopy locomotion is not 100%, which could hint to dat-1 and nRHO-1 acting in parallel pathways or, more likely, due to the nature of activated RHO-1 Ã signalling being very complex and involving multiple pathways. DATs negatively regulate dopamine signalling by removing dopamine from the extracellular space reducing the concentration and spread of dopamine released from the dopaminergic neurons [21]. DAT mutations are predicted to increase extracellular dopamine abundance and signalling. However, it is also possible that DAT-1 acts as a dopamine-gated depolarizing ion channel or that chronically high levels of dopamine cause adaption such that dopamine signalling could be decreased in dat-1 mutant animals [67,68]. To test if increased dopamine signalling is involved in the suppression of nRHO-1 Ã loopy locomotion we used a cat-2 mutant background, in which dopamine synthesis is selectively reduced. The loopy locomotion phenotype was no longer suppressed by the dat-1 mutation in a nRHO-1 Ã ;dat-1;cat-2 triple mutant background supporting our prediction that dopamine signalling is involved. In contrast, if dat-1 mutants suppressed nRHO-1 Ã -mediated locomotion due to chronically decreased dopamine signalling or if DAT-1 functions as a dopamine-activated ion channel, the effect of the cat-2 mutation would have been minor. Furthermore, we exposed nRHO-1 Ã single mutant animals, which have increased levels of dat-1 expression, to exogenous dopamine and observed a suppression of their loopy phenotype. These data indicate that loss of dat-1 rescues SWIP of L4 stage dat-1(ok157) animals at low injection concentrations of 1ng/μl. Data presented as mean percentage ± SEM of animals swimming in water after 10 minutes. Statistical comparisons were performed using one-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( ÃÃÃ P < 0.001), n = 3 experiments (10 animals per assay). d Expression levels of dat-1 in control (wild type) animals and non-injected control strain nRHO-1 Ã ;dat-1(ok157) compared to three rescue lines with increasing plasmid concentrations of dat-1 under cat-2 promoter (p.cat-2::dat-1) determined by quantitative PCR. Data presented as mean normalized copy number per reaction ± SEM. Statistical comparisons were performed using one-way ANOVA with Tukey's multiple test correction. Statistical significance indicated as ( Ã P 0.05-0.01, ÃÃÃ P 0.001, n.s. = not significant), n = 6, 4, 4, 5, and 4 independent samples for each genotype respectively. e Schematic of the dat-1 gene with the ok157 allele indicated. qPCR primer set for dat-1 within the ok157 deletion shown as arrows.
https://doi.org/10.1371/journal.pone.0204057.g007 dat-1 results in an increase in dopamine signalling which supresses the loopy locomotion phenotype of nRHO-1 Ã . We reasoned that if increased dopamine were responsible for antagonizing nRHO-1 Ã then this would require a dopamine receptor. There are four well characterized dopamine receptors in C. elegans, DOP-1-4 [48]. Mutations in dop-1, and partially dop-4, reversed dat-1 suppression of the loopy locomotion in nRHO-1 animals. No change in expression was observed for any of the DOP receptors in the nRHO-1 Ã animals. Taken together these results suggest dat-1 suppression of nRHO-1 Ã acts through increased dopamine signalling and unlikely through the reported DAT-1 ion channel function or a change in dopamine receptor signalling [67]. The effect of the dat-1 mutation on dopamine signalling is likely to be complex with changes that will include both the strength of the dopamine signal received by a cell and the number of cells that respond to dopamine as well as other changes [39]. Our data indicate that mutations in dat-1 cause an increase in dopamine signalling via the DOP-1 receptor that prevents nRHO-1 Ã from triggering the loopy locomotion phenotype. This result is somewhat surprising as DOP-3 receptors have been shown to be involved in the basal slowing response upon encountering food, paralysis on exogenous dopamine and swimming induced paralysis [39,65]. Allen et al reports that DOP-3 signalling reduces ACh-release from motor neurons whereas DOP-1 signalling stimulates ACh-release [40]. nRHO-1 Ã ;dat-1 double mutants despite showing a clear reduction in loopy locomotion remain hypersensitive to aldicarb indicating no change in ACh release from motor neurons. DOP-1 receptors most likely act in an alternative pathway to repress loopy locomotion. It has been shown recently that RHO-1 acts to stimulate NCA channel activity to regulate locomotion [16], and that dopamine signalling acts via DOP-3 receptors to antagonize NCA-1 channel activity [17]. The mutation in dop-3, however, had no effect on the locomotion of nRHO-1 Ã ;dat-1(ok157) animals, therefore the underlying mechanism is likely to be a different one.
In dat-1 mutants the absence of DAT-1 prevents nRHO-1 Ã from triggering loopy locomotion so how does nRHO-1 Ã override the antagonistic dopamine signalling to trigger loopy locomotion in animals with a functioning dat-1 gene? Exogenously administered dopamine reduced loopy locomotion in nRHO-1 Ã animals leading to paralysis, however, the nRHO-1 Ã animals paralysed at a much slower rate in comparison to wild type animals. One interpretation of these data could be that there is reduced dopamine abundance at its site of action in the nRHO-1 Ã animals. We argue that this is achieved by increasing the levels of the dopamine reuptake transporter DAT-1 in the nRHO-1 Ã animals. We observed that nRHO-1 Ã increased expression of an p.dat-1::mCherry::DAT-1 fusion protein and showed that transcription from the endogenous dat-1 gene was increased in nRHO-1 Ã animals. Genes required for the synthesis and release of dopamine from dopaminergic cells share transcriptional control elements with dat-1 [52,69]. Of these only dat-1 and cat-2 are expressed exclusively in dopaminergic cells and transcription of cat-2 according to our qPCR data was unaffected by nRHO-1 Ã . We also analysed the expression levels of other dopamine pathway genes such as the dopamine receptors dop-1-4 and a general neuronal gene expressed in dopaminergic neurons, snb-1, none of which showed significant alteration in nRHO-1 Ã animals. These data indicate that the effects of nRHO-1 Ã on dopaminergic cells did not cause a general increase in transcription, suggesting that RHO-1 acts via DAT-1 to decrease dopamine signalling. We investigated whether the upregulation of dat-1 was necessary for the induction of loopy locomotion, by exploiting the observation that cat-2 mRNA level did not significantly increase in response to nRHO-1 Ã based on our qPCR data. We made a nRHO-1 Ã independent dat-1 transgene using the cat-2 promoter (p.cat-2::dat-1). nRHO-1 Ã ;dat-1(ok157) animals carrying these transgenes showed a correlation between increased expression of dat-1 through higher transgene concentrations and increased loopy locomotion. These data lead us to hypothesize that nRHO-1 Ã increases the transcription of the dat-1 gene in order to trigger loopy locomotion. However, we recognize the caveat that the low copy number transgene, even though expressed at the similar level to the endogenous dat-1 in wild type animals, might have been too low to rescue the nRHO-1 Ã phenotype as the function of an different low copy number plasmid was tested in the SWIP assay. Additionally, the expression level of dat-1 required to rescue the SWIP phenotype might be different to the one required to rescue the locomotion phenotype in a nRHO-1 Ã ;dat-1(ok157) double mutant. However, active RHO-1 was not able to increase expression levels of dat-1 when placed under the cat-2 promoter. As dat-1 mutations can only suppress nRHO-1 Ã in the presence of dopamine synthesis and signalling, we interpret our data as follows: RHO-1 increases dat-1 transcription resulting in increased DAT-1 protein at the cell surface and consequently in a reduction of extracellular dopamine. Our data also suggest that without a DAT-1-mediated reduction in dopamine signalling the RHO-1 signalling pathway is severely reduced in its ability to regulate locomotion. These data indicate that we have identified an additional branch of the neuronal RHO-1 signalling pathway, which modulates locomotion by regulating the expression of DAT-1 and thus dopamine signalling.
The increase in dat-1 expression is not just an indirect consequence of loopy locomotion as nRHO-1 Ã increases expression from the non-functional dat-1(ok157) gene in non-loopy nRHO-1 Ã ;dat-1(ok157) animals. The dat-1(ok157) allele cannot produce functional DAT-1 protein due to a deletion in the dat-1 3' end, however, the dat-1(ok157) deletion leaves the dat-1 promoter and transcription start sites intact and we observed an increase from this faulty dat-1 gene in the non-loopy nRHO-1 Ã ;dat-1(ok157) animals. Moreover, dgk-1 mutants do not increase dat-1 expression despite having a similar body curvature as nRHO-1 Ã ;dat-1(ok157), indicating that nRHO-1 Ã mediated increases in dat-1 expression are not an indirect effect of increased body curvature during loopy locomotion. How does RHO-1 regulate dat-1 expression when the cells expressing nRHO-1 Ã and dat-1 do not overlap? Although p.unc-17 is a promoter widely used to drive cholinergic neuron specific expressions of proteins [17,18] at this moment we cannot exclude that p.unc-17::nRHO-1 Ã may be expressed at a low level in cells other than cholinergic neurons such as the same cell as dat-1. RhoA has been shown to regulate DAT-internalization in mouse midbrain slices [31], and DAT-surface availability in C. elegans [43] cell autonomously linking RHO-1 to DAT function. However, our hypothesis is that the cholinergic cells which express nRHO-1 Ã release a RHO-1-regulated signal that acts on the dopaminergic cells to non-autonomously regulate dat-1 expression. The signal is unlikely to be ACh, although we cannot fully exclude the possibility that increased ACh release is necessary for dat-1 expression. nRHO-1 Ã induces the release of the neurotransmitter ACh from synaptic vesicles [18] but, ACh release is also stimulated in dgk-1 mutants [18] and pkc-1(gf) animals [61] neither of which show increased dat-1 expression. With RHO-1 playing a role in the regulation of synaptic vesicle release [18] we investigated the possibility that RHO-1 signalling could also regulate dense core vesicle release and therefore the release of neuropeptides. A number of neuropeptides have been shown to regulate locomotion [54][55][56]. The neuropeptide NLP-12 modulates dopamine signalling via DOP-1 and over-expression of nlp-12 induces a hyperactive locomotion phenotype [60]. Although NLP-12 OE animals have a very similar loopy locomotion as nRHO-1 Ã animals and we show that nRHO-1 Ã significantly increases nlp-12 expression, dat-1 expression however was not significantly increased in the NLP-12 OE animals. These data indicate that high levels of NLP-12 do not increase dat-1 expression, but these data do not fully exclude the involvement of nlp-12 in this RHO-1 signalling pathway. The use of an nlp-12 loss of function mutation may shed more light on its involvement. Targeting general neuropeptide release via UNC-31 or PKC-1 mutants did not deliver conclusive results on whether neuropeptides are involved in our pathway. We believe that one or more of the over 250 distinct neuropeptides [64] would be a good candidate for the RHO-1 mediated regulation of dat-1 expression and targeting individual neuropeptides represents a future task in order to understand how RHO-1 regulates dat-1 expression to control locomotion.
In this study, we propose that changes in dat-1 expression affect the behaviour of C. elegans, so how likely is it that changes in Dat expression can cause behavioural changes in other species? Mice heterozygous for a Dat deletion display changes in dopamine levels intermediate between that of wild type and full knockout mice [70]. In rats iron deficiency reduced Dat expression and this correlated with changes in motor behaviour [71]. In humans DAT polymorphisms can alter DAT expression and variants associated with human mental diseases can alter transcription over a 3 fold range, approximately the same range we observed nRHO-1 Ã to alter dat-1 expression [72]. The contribution of these human DAT polymorphisms to mental disorders is being intensely studied with the clearest connection being that between changes in DAT expression and ADHD, however, linkage between DAT alleles and a large range of mental disorders or general behavioural traits are being reported [23,73]. The C. elegans RHO-1/ DAT-1 pathway increases dat-1 expression almost fourfold, from the information above such changes in humans would alter many brain functions such as changes in mood, motivation, movement, and cognition. If the RHO-1/DAT-1 pathway is conserved in humans then any defects in this pathway could have effects ranging from subtle changes in behaviour to severe mental disorders. Drugs capable of targeting such a DAT regulatory pathway could have clinical applications for treatments of the wide range of mental conditions that DATs are increasingly being implicated in.

Plasmids
Plasmids (listed as QT# or SJN) were constructed by standard techniques, and verified by sequencing.
SJN617 pBS dat-1 genomic fragment starting 722bp 5' of the start ATG and 918bp 3' of the STOP of dat-1a was amplified by PCR with a SacI site added 5' and a KpnI site added 3' and subcloned into pBluescript SK.
SJN717 p.cat-2::dat-1 genomic, an EagI site 40bp 5' of the dat-1 start ATG was used to swap out the promoters. An 852bp fragment of cat-2 starting 901bp 5' and finishing 47bp 5' of the cat-2 start ATG was amplified with PCR with a SacI site 5' and an EagI site 3' and used to swap out the SacI-EagI fragment in the dat-1 promoter of SJN617. SJN628 p.dat-1::mCherry::dat-1 A NotI site plus one base GCGGCCGCc was inserted immediately 3' to the dat-1a start ATG in SJN617. Into this NotI site was subcloned mCherry amplified with flanking NotI sites such that it was in frame with dat-1 with 3 extra Alanines added at the N and C terminus by the flanking NotI plus a base sites.

Induction of heat-shock-inducible transgenes
In all cases, expression from the heat-shock promoter was achieved using two rounds of heat shock for 60 min at 30˚C, separated by 30 min at 20˚C, followed by recovery for 30 min at 20˚C [18].

Fluorescence microscopy
Transgenic lines were mounted on agarose pads in paralyzing agent (10 mg/ml 2,3-Butanedione monoxime, Sigma) and imaged on a Perkin Elmer Ultraview Vox spinning disc confocal using a 100x oil lens. Laser images and exposures times were the same per channel and images were obtained and processed to give maximum intensity projections of a z-series using Perkin Elmer Volocity 6.3 software. Several control and mutants animals were observed in the same session, and these sessions were repeated on different days.

Aldicarb assay
Sensitivity to 1 mM aldicarb (Greyhound Chromatography) on NGM plates in the presence of food was determined by analysing the onset of paralysis as described previously [53]. For each experiment, 30 animals were tested, and each experiment was repeated at least in triplicate.

Dopamine locomotion assay
Assays were performed as described in [39]. For the assay, approx. 30 young adult animals were transferred to 1.7% agar plates containing 2 mM glacial acetic acid plus or minus 15 mM dopamine, and the number of paralysed animals were scored every 10 min for 100 min. Animals were scored as paralysed if they no longer responded to harsh touch and each experiment was repeated at least in triplicate. For Worm tracking young adult animals were individually transferred to dopamine plates plus or minus 20 mM dopamine, incubated for 20 min, and then recorded on the Worm Tracker 2.0.4 system. A minimum of 16 animals were tracked per condition.

Calculation of curvature
The mean curvature of the hip (Fig 1C) was analysed by tracking single worms using the Worm Tracker 2.0.4 system as developed by the Schafer Lab [46]. In summary, 1-day-old adults were transferred to low peptone tracking plates, seeded the day before with a full lawn of OP50, and individually tracked following a 30 min acclimatization period. Animals were tracked for 5 min and at least 12 animals were assayed over different days for each strain, with each batch containing wild type animals. The videos were analysed using Worm Analysis Tool Box version 3 and Matlab (Matlab script see S1 Script). For clarity one point on the worm, mean hip (point 4), was chosen to represent alterations in curvature on subsequent strains. This was done firstly as a means to cut down on the presented data and mean hip was specifically chosen due to potentially less interference from head movements when foraging [76] which, could affect point 1 and 2. Point 3 could potentially be affected by the swollen vulva phenotype of nRHO-1 Ã animals [75] and point 5 could be affected by defecation and/or tail swelling another phenotype of nRHO-1 Ã animals [77].

SWIP assay
Swimming induced paralysis (SWIP) assays were performed as described in [66]. In summary 10 L4 animals were transferred to a 45ul spot of dH2O and scored for SWIP after 10 min. We defined paralysed animals in our studies as absence of movement for at least 20 sec. The assay was repeated at least in triplicate for each strain.

Dispersal assay
Assays were performed as described in [47]. In summary OP50 was seeded at the edge of a 10 cm NMG plate 24 hours before assay. A plate of non-starved adult animals were washed 3x in M9 buffer (animals were allowed to settle on ice, by gravity, between each wash, as centrifugation altered dispersal of animals). Approximately 200 animals were added to the centre of the plate and timed once the M9 buffer had been absorbed. The number of animals on food, were scored at 20, 40, 60, 120 and 240 minutes and the total number of adults were scored at the end of the assay. The assay was repeated at least in triplicate for each strain on different days to minimise batch effect (Fig 1D).

Quantitative PCR
RNA was isolated using Qiagen RNeasy mini kit following the manufacturers protocol. At least 3 independent samples were used per genotype. qPCR was performed and analyzed by qStandard (www.qstandard.co.uk, 61 Wolmer Gardens, HA8 8QB Edgware UK). Primer sequences are as follows; dat-1 set 1 forward agttggtgcctacagacgat reverse tccagcaagagaacagtggtc (Fig 6D), dat-1 set 2 (primers that anneal within the dat-1(ok157) deletion) forward gaaatgctcaagagaccatcgg reverse tcggagttctcatggcatactc (Fig 7E), cat-2 set forward caacaacggatccacgacat reverse gcctatctcgtcaccaaact, nlp-12 set forward gattaccggccacttcagttc reverse cgttttccgaactgcaatgga. Three reference genes were used based on unusually stable expression levels with relatively little variation between adults, dauers, and L3 larvae (qStandard): Y45F10D.4 forward aagcgtcggaacaggaatc reverse gagtcgacgataacggaaaaa, pmp-3 (C54G10.3.1) forward tcgctgaaacaattccatga reverse atggtcccttcacgacattg, cdc-42 (R07G3.1), forward tatgtgccgacagtcttcgacaatta, reverse ctctatcgtatccacagaccga. The data were normalized by dividing the copy number per reaction by the normalization factor of three reference genes from geNorm. geNorm (https://genorm.cmgg. be/) is a popular algorithm to determine the most stable reference (housekeeping) genes from a set of tested candidate reference genes in a given sample panel. From this, a gene expression normalization factor can be calculated for each sample based on the geometric mean of a userdefined number of reference genes.

Screen and whole genome sequencing
nRHO-1 Ã animals were mutagenized using a standard C. elegans EMS protocol. 1930 haploid genomes were screened, 116 strong suppressors of loopy locomotion (15 were unreproducible, 15 contained inactivated nzIs29 transgene) so 86 suppressor mutants were identified. The QT677 non-loopy nRHO-1 Ã ;nz99 animals were detected by eye, singled and their progeny were shown to be non-loopy. This non-loopy phenotype was maintained following heat shock activation of the second integrated nRHO-1 Ã transgene, nzIs1. QT677 was chosen as a neuronal-specific suppressor mutant because while it suppressed loopy locomotion driven by both transgenes, it failed to suppress the sterility phenotype of hs::RHO-1 Ã , which is thought to have a non-neuronal origin [75]. Whole genome sequencing (WGS) and bioinformatics of un-backcrossed nRHO-1 Ã ;nz99 animals was performed by GeneService (http://www.sourcebioscience. com/) using Illumina 76bp paired ends sequencing that resulted in an average coverage of greater than 10x. The WGS data of the QT677 strain (nRHO-1 Ã ; nz99) was compared against the reference nRHO-1 Ã strain and 2809 mutations were identified, 1278 of which were induced by EMS. These 1278 mutations led to 42 protein coding altering mutations identified by analysis with MAQSeq [78] (S1 Table).

Statistical analysis
All raw data is provided as supplementary information (S1 Dataset). Statistical analysis was performed as specified in the figure legends, using either an unpaired two-tailed t-test, oneway ANOVA with Tukey's multiple test correction or two-way ANOVA with Tukey's multiple test correction using GraphPad Prism1. P values are indicated as follows Ã = 0.05-0.01 ÃÃ = 0.01-0.001 and ÃÃÃ = 0.001.