Dopamine transporter trafficking and Rit2 GTPase: Mechanism of action and in vivo impact

Following its evoked release, dopamine (DA) signaling is rapidly terminated by presynaptic reuptake, mediated by the cocaine-sensitive DA transporter (DAT). DAT surface availability is dynamically regulated by endocytic trafficking, and direct protein kinase C (PKC) activation acutely diminishes DAT surface expression by accelerating DAT internalization. Previous cell line studies demonstrated that PKC-stimulated DAT endocytosis requires both Ack1 inactivation, which releases a DAT-specific endocytic brake, and the neuronal GTPase, Rit2, which binds DAT. However, it is unknown whether Rit2 is required for PKC-stimulated DAT endocytosis in DAergic terminals or whether there are region- and/or sex-dependent differences in PKC-stimulated DAT trafficking. Moreover, the mechanisms by which Rit2 controls PKC-stimulated DAT endocytosis are unknown. Here, we directly examined these important questions. Ex vivo studies revealed that PKC activation acutely decreased DAT surface expression selectively in ventral, but not dorsal, striatum. AAV-mediated, conditional Rit2 knockdown in DAergic neurons impacted baseline DAT surface:intracellular distribution in DAergic terminals from female ventral, but not dorsal, striatum. Further, Rit2 was required for PKC-stimulated DAT internalization in both male and female ventral striatum. FRET and surface pulldown studies in cell lines revealed that PKC activation drives DAT-Rit2 surface dissociation and that the DAT N terminus is required for both PKC-mediated DAT-Rit2 dissociation and DAT internalization. Finally, we found that Rit2 and Ack1 independently converge on DAT to facilitate PKC-stimulated DAT endocytosis. Together, our data provide greater insight into mechanisms that mediate PKC-regulated DAT internalization and reveal unexpected region-specific differences in PKC-stimulated DAT trafficking in bona fide DAergic terminals.

endocytic brake and that Ack1 inactivation is required for PKCmediated brake release (24).
Rit2 (also known as Rin) is a neuronal small GTPase that lacks a CAAX domain and associates with the plasma membrane in a phosphoinositide-dependent manner (28 -30). Rit2 gene expression is highly enriched in DA neurons (31), and several recent GWAS studies identified RIT2 SNPs and long tandem repeat variants associated with multiple DA-related disorders, including PD, ASD, and schizophrenia (32)(33)(34)(35)(36)(37). However, despite its disease association, relatively little is known about endogenous DAergic Rit2 function. We previously reported that DAT directly binds to Rit2 and that Rit2 activity is required for PKC-stimulated DAT internalization (38). Moreover, we recently found that in vivo DAergic Rit2 knockdown (Rit2-KD) differentially alters acute cocaine sensitivity in males and females (39). However, it remains unknown whether PKC-stimulated DAT internalization in DAergic terminals requires Rit2 and whether there are region-or sex-specific differences in DAT's reliance upon Rit2. Further, it is unclear how intrinsic DAT domains influence the DAT-Rit2 interaction or whether Rit2 and Ack1 coordinate to release the DAT endocytic brake. In the present study, we leveraged biochemical and genetic approaches, in both cultured cells and ex vivo mouse striatal slices, to directly address these salient questions.

Rit2 cellular expression and antibody specificity
In our previous study, in which we initially reported the DAT-Rit2 (Rin) interaction (38), there were several paradoxical findings regarding 1) how PKC activation impacted the DAT-Rit2 interaction and 2) the Rit2 expression profile across various cell lines. In cellular imaging studies, which used CFP-Rit2, Rit2 appeared to remain at the plasma membrane following PKC-stimulated DAT internalization, suggesting that PKC may drive DAT and Rit2 to dissociate. In contrast, parallel co-IP studies found that PKC activation increased the DAT-Rit2 interaction in PC12 cells. Furthermore, although Rit2 expression is reportedly restricted to neurons (28,29,31,40), we detected a single, ϳ20 kDa immunoreactive band by immunoblot analysis in all neuronal and nonneuronal cell lines tested, as well as Rit2 mRNA expression in these cell lines by standard RT-PCR. The previous studies utilized the anti-Rit2 mAb, clone 27G2, and in that report, we additionally confirmed that 27G2 specifically recognizes Rit2, but not Rit1 (the ubiquitously expressed Rit2 homologue), using fluorescently tagged Rit2 and Rit1 fusion proteins. Recently, we decided to take advantage of highly specific, Rit2-directed, real-time qPCR probes to reassess the Rit2 expression profile across a large panel of human, mouse, and rat cell lines, as well as in mouse and rat midbrain, all of which exhibit the single, 20 kDa immunoreactive band when probed with the anti-Rit2 27G2. To our surprise, Rit2 mRNA was undetectable in any of the mouse or rat cell lines tested, whereas a robust Rit2 mRNA signal was detected in both mouse and rat midbrain controls (Table 1). Moreover, among the human cell lines tested, Rit2 mRNA was only detected in SK-N-DZ cells, as reported previously (40), and at low levels in SH-SY5Y cells. However, SK-N-MC and HEK293T cells expressed markedly less/negligible Rit2 signal than SK-N-DZ cells, ranging from 20-to 300-fold less (Table 1).
Our current RT-qPCR results raised the possibility that although 27G2 recognizes Rit2, it also may cross-react with a ubiquitously expressed protein that has an electrophoretic mobility close to that of Rit2. To test this, we screened several newer, commercially available anti-Rit2 antibodies using cell lysates from HEK293T cells transfected with CFP-Rit2. Consistent with our previous report, clone 27G2 identified a single 20 kDa band in both transfected and nontransfected cells and also detected CFP-Rit2 selectively in transfected cells (Fig. 1A). However, when immunoblots were probed with ␣Rit2 clone 4B5, the low-molecular-weight immunoreactive band was not detected, whereas CFP-Rit2 was detected in transfected cells (Fig. 1A). These results are consistent with our RT-qPCR results and confirm that 27G2 detects an artifactual band with an electrophoretic mobility close to the Rit2 predicted size. We next used 4B5 to assess Rit2 in cells transfected with HA-Rit2 and in mouse DAergic tissues. Clone 4B5 detected a single, ϳ28 kDa immunoreactive band selectively in cells transfected with HA-Rit2, which was identical in mobility to an immunoreactive band detected with ␣HA in parallel (Fig. 1B). ␣HA also detected a higher-molecular-weight band (ϳ30 kDa) specifically in cells transfected with HA-Rit2, which we did not detect using 4B5 (Fig. 1B), suggesting that Rit2 may have multiple isoforms and that 4B5 may only detect one of these. 4B5 also detected bands at ϳ24 kDa in mouse lysates enriched for either dorsal striatum or ventral midbrain (Fig. 1B). These immunoblotting results were consistent with our RT-qPCR results and confirmed that most cell lines either do not express Rit2 or express Rit2 at negligible levels. They further support that the 27G2 antibody cannot reliably distinguish between Rit2 and a robust, artifactual, background band. It should also be noted that whereas 4B5 specifically detects Rit2, it does so with extremely low sensitivity, even when Rit2 is highly overexpressed. Indeed, using the 4B5 antibody, we could only detect Rit2 by immunoblotting in lysates from mouse tissues when a very high protein mass was loaded. Moreover, although both hRit2 and mRit2 are predicted to have an identical number of amino acids and a predicted mass of ϳ24 kDa each, the overexpressed hRit2 protein Table 1 Rit2 expression in mammalian cell lines and rodent brain regions ran slightly higher than the putative mRit2 band. There are two known hRit2 isoforms, variants 1 and 2, which are predicted to be ϳ24 and ϳ17 kDa, respectively, and our hRit2 cDNA codes for variant 1. To date, it is unknown whether there are tissuespecific Rit2 isoforms in mouse neurons. Given that there is no global Rit2 Ϫ/Ϫ mouse available, we therefore cannot say with absolute certainty whether the single immunoreactive band in mouse tissue is definitely Rit2 or possibly a smaller splice variant. Therefore, for the majority of our cell line studies, we opted to use HA-Rit2 for greater sensitivity in cell lines and did not further assess Rit2 protein in tissue.

Rit2 is required for PKC-stimulated, but not basal, DAT internalization
In our original DAT-Rit2 study (38), we used shRNA to silence Rit2 in SK-N-MC cells and to test whether Rit2 is required for PKC-stimulated DAT down-regulation. In light of our current findings that SK-N-MC cells do not appreciably express Rit2, we were prompted to 1) rescreen Rit2 targeted shRNAs, and 2) reassess whether Rit2 is required for PKC-stimulated DAT internalization in SK-N-DZ cells, which endogenously express Rit2. We screened several candidate human Rit2-directed shRNAs and identified two shRNAs (shRit2-104 and -107) that significantly silenced CFP-hRit2 protein expressed in Neuro2a cells ( Fig. 2A). Additionally, both shRit2-104 and -107 significantly silenced endogenous Rit2 mRNA expression in SK-N-DZ cells (Fig. 2B). We utilized shRit2-107 to ask whether Rit2 is required for PKC-stimulated DAT internalization in SK-N-DZ cells. Rit2-KD significantly blocked PKC-stimulated DAT endocytosis as compared with vectortransduced cells (Fig. 2C), consistent with a requisite role for Rit2 in PKC-stimulated DAT internalization. The ability of shRit2-107 to block PKC-stimulated DAT endocytosis was not likely due to off-target effects, as shRit2-107 did not decrease expression of Rit1, the closest homolog to Rit2 (Fig. 2D). To further ensure that shRit2-107 effects were specific, we repeated these studies with shRit2-104. Similar to our findings with shRit2-107, Rit2 silencing with shRit2-104 significantly blocked PKC-stimulated DAT internalization (Fig. 2E).

Rit2 is required for striatal steady-state DAT surface expression and PKC-stimulated DAT internalization in a region-and sex-specific manner
PKC-stimulated DAT internalization in response to phorbol ester treatment has been reported by numerous laboratories, in a variety of transfected cell lines (15,25,(41)(42)(43)(44)(45). Moreover, we previously reported that phorbol 12-myristate 13-acetate (PMA) treatment decreases DAT surface levels in DAergic terminals in ex vivo total striatal slices containing both dorsal (DS) and ventral (VS) striatum (26). However, it is unknown whether PKC-stimulated DAT internalization differs between DAergic terminal regions, such as DS and VS, or between males and females. Moreover, although Rit2 is required for PKC-stimulated DAT internalization in SK-N-DZ cells (Fig. 2), it is unknown whether Rit2 is required for PKC-stimulated DAT endocytosis in DAergic terminals. We recently leveraged the TET-OFF system to achieve conditional, inducible DAergic Rit2-KD in Pitx3 IRES2-tTA mice, in which AAV9-shRit2 injection into mouse VTA significantly silenced Rit2 expression in both VTA and substantia nigra pars compacta (39). We found that DAergic Rit2 silencing decreased total striatal DAT protein in males, but not females. Further, although total DAT protein decreased within male total striatum, the DAT surface/ intracellular ratio was unchanged in either DS or VS, resulting in less overall surface DAT in both male striatal subregions (39). In the current study, we extended our in vivo Rit2-KD studies and asked whether Rit2 impacts DAT basal distribution in female DS and VS. We further asked whether PKC-mediated DAT internalization requires Rit2 in male and female DS and VS. Male and female Pitx3 IRES2-tTA/ϩ mice VTA were bilaterally injected with either AAV9-TRE-eGFP or AAV9-TRE-shRit2, and DAT surface expression was measured by ex vivo slice biotinylation in the VS and DS, following treatment with or without 1 M PMA for 30 min at 37°C.
We first examined the effect of Rit2 silencing on DAT surface expression in female DS and VS, under both basal and PKC-stimulated conditions. Surprisingly, in DS, PKC activation did not decrease DAT surface expression in control female mice (Fig. 3A). Moreover, DAergic Rit2-KD had no effect on DS DAT surface expression, under either basal or PKC-stimulated conditions (Fig. 3A). In contrast, in female VS, PKC activation significantly reduced DAT surface levels, and DAergic Rit2-KD completely blocked further surface loss in response to PKC activation (Fig. 3B). Moreover, DAergic Rit2-KD significantly Rit2-dependent dopamine transporter trafficking reduced the basal DAT surface level in female VS, as compared with control mice (Fig. 3B).
Our previous study probed the impact of DAergic Rit2 KD on basal DAT surface levels in males (39). Therefore, we next asked whether Rit2 was required for PKC-stimulated DAT internalization in male DS and VS. Similar to females, PKC activation had no effect on DAT surface expression in DS, but significantly decreased DAT surface expression in VS, measured in control (eGFP-injected) male mice (Fig. 4A). Also similar to females, DAergic Rit2-KD completely abolished PKC-stimulated DAT internalization in male VS (Fig. 4B). Given that phorbol esters can stimulate a variety of signaling pathways in addition to PKC, we further tested whether PMA-induced DAT internalization in VS was PKC-mediated. Pretreatment with the PKCspecific inhibitor bisindoylmaleimide I (BIM I) (1 M, 15 min, 37°C) significantly abolished PMA-induced DAT internalization, and treatment with BIM I alone was not significantly different from BIM I/PMA (Fig. 4C), clearly demonstrating that PMA-mediated DAT internalization in DAergic terminals is PKC-dependent.

Releasing the PKC-sensitive DAT endocytic brake drives DAT-Rit2 dissociation at the plasma membrane
We next sought to decipher the molecular mechanisms by which Rit2 impacts DAT trafficking. We first asked whether driving DAT internalization, by disengaging the PKC-sensitive DAT endocytic brake, impacts the DAT-Rit2 interaction at the plasma membrane. To specifically interrogate the DAT surface population and its associated protein complex, we leveraged a bungarotoxin-binding site (BBS)-targeted surface labeling strategy (46 -48) to label DAT in intact cells. We engineered a BBS into the DAT extracellular loop 2, a site that we previously successfully targeted for bio-orthogonal DAT labeling (49) and which also tolerates an HA epitope (50). BBS-DAT expressed and functioned comparably with WT DAT, and PKC activation acutely decreased BBS-DAT function to 67.97 Ϯ 5.9% control shRit2-107 transduction did not significantly affect Rit1 mRNA expression, p ϭ 0.965, two-tailed Student's t test, n ϭ 3. E, DAT internalization assays. Stable DAT-SK-N-DZ cells were transduced with the indicated lentiviral particles, and DAT internalization rates were measured 96 h post-transduction as described under "Experimental procedures." Data are presented identically as in C. *, p ϭ 0.03, one-tailed Student's t test, n ϭ 5. levels ( Fig. 5A), which is comparable with PKC-mediated DAT down-regulation, as previously reported by our group and others (22,38,43,49,51). We first tested whether BBS-DAT could 1) specifically isolate surface DAT via bungarotoxin labeling and pulldown and 2) recover DAT-associated proteins. HEK293T cells expressing HA-Rit2 and either BBS-DAT or WT DAT were incubated with ␣-BTX-b, and DAT surface complexes were isolated from cell lysates by streptavidin pulldown. We specifically recovered BBS-DAT, but not WT DAT, following ␣-BTX-b incubation (Fig. 5B), demonstrating the selectivity of the BBS pulldown approach to label and isolate surface BBS-DAT. Importantly, Rit2 was recovered from pulldowns with BBS-DAT, but not in control pulldowns from cells expressing WT DAT, demonstrating that Rit2 is recovered specifically following surface DAT pulldown. Additionally, BBS-DAT pulldowns did not recover the Rit2 homolog, Rit1 (Fig.  5C), consistent with a specific association between DAT and Rit2. We further asked whether other proteins required for PKC-mediated brake release are also part of the DAT surface complex. We previously reported that the nonreceptor tyrosine kinase, Ack1 (TNK2) imposes the PKC-sensitive endocytic brake and that PKC-mediated Ack1 inactivation is required for PKC-stimulated DAT internalization (24). However, it is not known whether Ack1 is associated with DAT at the plasma membrane. Following ␣-BTX-b labeling and pulldown, Ack1 was recovered from cells expressing BBS-DAT, but not from control cells expressing WT DAT (Fig. 5D), demonstrating that Ack1 is part of the DAT surface complex. Thus, BBS-DAT has precise utility to interrogate surface DAT and its associated proteins, such as Rit2 and Ack1.
Given our previous cellular imaging results (38), we hypothesized that PKC activation causes DAT and Rit2 to dissociate. Because PKC-stimulated DAT internalization can only occur when the endocytic brake is disengaged, we first leveraged BBS-DAT pulldowns to ask whether PKC-mediated brake release alters the DAT-Rit2 surface association. PKC activation (1 M PMA, 30 min, 37°C) significantly decreased the DAT-Rit2 plasma membrane association (Fig. 6A). Additionally, DAT and Rit2 significantly dissociated when we directly released the DAT endocytic brake, by inactivating Ack1 with AIM-100 (20 M, 30 min, 37°C) (Fig. 6B). DAT and Rit2 may specifically dissociate at the cell surface in response to releasing the PKCsensitive endocytic brake or may do so following any stimulus that drives DAT internalization. To discern between these two possibilities, we measured the DAT-Rit2 surface association in response to AMPH treatment, which also accelerates DAT internalization but is Rho-dependent (52). In contrast to PKCstimulated DAT-Rit2 dissociation, AMPH treatment (10 M, 30 min, 37°C) significantly increased the DAT-Rit2 surface association (Fig. 6C). Thus, the DAT-Rit2 surface dissociation occurs specifically when the PKC-sensitive DAT endocytic brake is disengaged, either in response to PKC activation or direct Ack1 inactivation, but is not a general result of accelerated DAT endocytosis.

The DAT N terminus is integral to the DAT-Rit2 interaction and PKC-stimulated dissociation
The DAT-Rit2 interaction was originally identified in a yeast two-hybrid screen, using DAT C-terminal residues 587-596 (FREKLAYAIA) as bait (38). However, it is not known which DAT domains are required (or sufficient) for the DAT-Rit2 association in the context of intact DAT protein; nor is it known whether any DAT domains are specifically required for PKCstimulated DAT-Rit2 dissociation. Rit2 specifically binds DAT, but not SERT (38); thus, we hypothesized that replacing DAT N or C termini with those of SERT may define DAT domains required for Rit2 binding and/or PKC-stimulated DAT-Rit2 dissociation. To test this possibility, we leveraged a series of DAT/SERT chimeras we previously characterized (53), in which either the DAT N terminus, C terminus, or both termini were substituted with those of SERT. HEK293T cells were cotransfected with YFP-Rit2 and CFP-tagged versions of these chimeras, and we performed live FRET imaging to quantify the chimera-Rit2 interactions. As we reported previously, control CFP-DAT and YFP-Rit2 elicited a significant FRET signal as compared with soluble YFP/CFP expression (Fig. 7A). Interestingly, replacing the DAT N terminus with that of SERT (CFP-N-S/DAT) significantly increased the DAT-Rit2 interaction compared with CFP-DAT, whereas replacing the DAT C terminus (CFP-DAT/C-S) or both termini (CFP-S/DAT/S) did not affect the DAT-Rit2 interaction (Fig. 7A). We also observed a significant increase in the interaction between YFP-Rit2 and CFP-N-S/DAT using the donor recovery after photobleaching

Rit2-dependent dopamine transporter trafficking
(DRAP) approach (Fig. 7B), demonstrating that the FRET signal is a bona fide interaction between the fluorophores. Using the BBS pulldown approach, Rit2 was likewise recovered with BBStagged versions of each DAT/SERT chimera (Fig. 7C). We next asked whether PKC-stimulated DAT-Rit2 dissociation requires either the DAT N and/or C termini. Substituting the DAT C terminus with that of SERT (DAT/C-S) had no significant effect on PKC-stimulated DAT-Rit2 dissociation, as compared with WT DAT controls (Fig. 7D, one-way ANOVA with Dunnett's multiple-comparison test, p ϭ 0.69). However, substituting the DAT N terminus with that of SERT (N-S/DAT) completely abolished PKC-stimulated DAT/Rit2 dissociation, and there was a strong trend for attenuated DAT/Rit2 dissociation when both DAT N and C termini were replaced by SERT (S/DAT/S) (Fig. 7D, p ϭ 0.058). Taken together, these results indicate that the DAT N terminus is required for the PKC-stimulated DAT-Rit2 dissociation and that the SERT N terminus does not suffice. However, SERT N and C termini suffice to maintain the DAT-Rit2 association.

The DAT N terminus is required for PKC-stimulated DAT internalization
Because the DAT N terminus is required for PKC-stimulated DAT-Rit2 dissociation, this raised the possibility that the N terminus may also be required for stimulated DAT internalization, driven by release of the DAT endocytic brake. To test these possibilities, we measured DAT and DAT/SERT chimera internalization rates in response to either PKC activation or direct Ack1 inactivation (with AIM-100) in stably transfected SK-N-MC cells, in which we previously characterized both PKC-and AIM-100 -stimulated DAT internalization (24,26). PKC activation (1 M PMA, 10 min, 37°C) significantly increased WT DAT internalization, and substituting the DAT C terminus with the SERT C terminus (DAT/C-S) did not significantly affect PKC-stimulated internalization (Fig. 8A).  HEK293T cells were co-transfected with DAT (with or without BBS tag) and either HA-Rit2 (B), GFP-Rit1 (C), or Ack1-HA (D), and DAT surface complexes were labeled and isolated by streptavidin pulldown as described under "Experimental procedures." Representative immunoblots for pull-downs (left panels) and their respective inputs (one-fourth of total input, right panels) are presented (n ϭ 3 independent experiments).

Rit2-dependent dopamine transporter trafficking
However, PKC-stimulated DAT internalization was abolished when either the DAT N terminus or both the N and C termini were substituted with SERT termini (N-S/DAT and S/DAT/S; Fig. 8A). In contrast, direct Ack1 inactivation (20 M AIM-100, 10 min, 37°C) significantly stimulated WT DAT, N-S/DAT, and DAT/C-S internalization but had no effect on S/DAT/S

Rit2-dependent dopamine transporter trafficking
internalization (Fig. 8A). We further evaluated the basal endocytic rates of the chimeric DATs, as compared with WT DAT. As presented in Fig. 8B, none of the chimera basal internalization rates differed significantly from WT DAT. To ensure that these effects were not due to the lack of Rit2 expression in SK-N-MC cells, we further assessed basal and PKC-stimulated N-S/DAT internalization in stably transfected SK-N-DZ cells, which endogenously express Rit2 (Table 1). In SK-N-DZ cells, N-S/DAT internalized significantly slower than WT DAT under basal conditions (Fig. 8C). Additionally, whereas PKC activation significantly increased the WT DAT internalization rate, it had no effect on N-S/DAT internalization as compared with its own vehicle control (p ϭ 0.17, one-way ANOVA with Bonferroni's multiple-comparison test, n ϭ 4 -7). These results demonstrate that the DAT N terminus is required, and that the SERT N terminus does not suffice, for PKC-stimulated DAT internalization.

Rit2 and Ack1 independently converge on DAT in response to PKC
We next asked whether there is a mechanistic relationship between Rit2-, Ack1-, and PKC-mediated DAT endocytic brake release. Rit2 may be either upstream or downstream from Ack1 in the signaling cascade that leads from PKC to DAT. Alternatively, Rit2 and Ack1 may respond independently to PKC activation to stimulate DAT internalization. We first used Rit2-KD in SK-N-DZ cells to ask whether Rit2 was required for PKCmediated Ack1 inactivation, a requisite step for PKC-mediated release of the endocytic brake and for stimulated DAT internalization (24). As we previously reported, in vector-transduced cells, PKC activation significantly reduced pY284-Ack1, to levels comparable with that achieved with the Ack1 inhibitor, AIM-100 (Fig. 9A). In cells transduced with shRit2-107, pY284-pAck1 levels were also significantly reduced in response to either PKC activation or AIM-100 treatment, and there was no

Rit2-dependent dopamine transporter trafficking
difference in the magnitude change of pY284-pAck1 levels following PKC activation (Fig. 9A). These results indicate that Rit2 is not likely to be upstream of Ack1 in the signaling pathway that leads from PKC to Ack1 inactivation.
We next asked whether PKC-mediated Ack1 inactivation is required for, and therefore upstream of, PKC-stimulated DAT-Rit2 dissociation. To test this possibility, we measured DAT-Rit2 dissociation in cells co-transfected with DAT, Rit2, and either vector or the PKC-insensitive, constitutively active Ack1 mutant (S445P), which we previously reported blocks PKCstimulated DAT internalization (24). In control cells, PKC activation drove a significant DAT-Rit2 dissociation (Fig. 9B). In cells co-transfected with S445P-Ack1, PKC activation likewise drove DAT-Rit2 dissociation, at levels indistinguishable from vector controls (Fig. 9B). These results demonstrate that PKCstimulated DAT-Rit2 dissociation does not require Ack1 inactivation. Moreover, they demonstrate that even in conditions where DAT cannot internalize in response to PKC activation (i.e. because Ack1 is constitutively active), PKC activation still drives DAT and Rit2 to dissociate. Thus, PKC-stimulated DAT-Rit2 dissociation is likely to occur prior to Ack1-mediated release of the DAT endocytic brake.

Discussion
DAT is stabilized at the cell surface by a PKC-sensitive endocytic brake that requires 1) residues in both the DAT N and C termini to engage (23,27) and 2) Ack1 inactivation to be released (24). Here, we found that Rit2 is required to release the PKC-sensitive DAT endocytic brake in cell lines and tissues where Rit2 is endogenously expressed, such as SK-N-DZ cells (Fig. 2C) and in VS (Figs. 3 and 4), respectively. Curiously, over 2 decades of evidence has demonstrated that PKC activation stimulates DAT internalization in in a wide range of cultured cell lines (42,50,(54)(55)(56), which our results demonstrate have negligible, if any, Rit2 expression ( Table 1). Given that Ack1 is ubiquitously expressed and that Rit2 and Ack1 converge independently on DAT in response to PKC activation (Fig. 9), we conclude that PKC-stimulated Ack1 inactivation can suffice to release the endocytic brake when DAT is heterologously expressed in a context that does not express Rit2. However, when DAT is expressed in its appropriate context (i.e. striatal terminals), Rit2 expression is absolutely required for PKCstimulated DAT internalization, and PKC-mediated Ack1 inactivation alone does not suffice to release the endocytic brake (Figs. 3 and 4). These results further suggest that although Rit2 and Ack1 independently converge on DAT in response to PKC activation in cell lines (Fig. 9), there may be an, as of yet, unidentified DAergic-specific mechanism(s) that facilitate a Rit2/ Ack1 interdependence required for endocytic brake release.
Our previous study using total (i.e. not subdissected) striatal slices found that PKC activation decreased DAT surface expression by ϳ20% (26). Interestingly, in the current study, PKC activation drove ϳ35% DAT surface loss in both male and female VS but had no effect on DAT surface expression in DS from either sex (Fig. 3). These data suggest that the somewhat modest PKC-stimulated DAT surface loss detected in total striatum reflects robust DAT surface loss in VS, diluted by the lack of a net effect in DS. Is DAT endocytosis completely resistant to PKC activation in DS? It is currently unknown which PKC isoform(s) stimulate DAT internalization. We activated PKC with the phorbol ester PMA, which activates all diacylg-

Rit2-dependent dopamine transporter trafficking
lycerol-sensitive PKC isoforms, including PKC␣, -␤1, -␤2, -␥, and -␦. PKC␤ activity is required for D2 receptor-mediated DAT insertion into the plasma membrane, and selective PKC␤ activation rapidly delivers DAT to the cell surface in both cell lines and striatal synaptosomes (57). Thus, it is possible that in DS, Rit2-dependent, PKC-stimulated DAT internalization may be countered by PKC␤-mediated DAT insertion, resulting in no detectable net change within the 30-min timeframe we tested. In contrast to DS, DAT surface levels in VS were significantly decreased in response to PMA treatment, suggesting that the balance between PKC␤-mediated DAT insertion and Rit2-dependent DAT internalization may differ between DS and VS. A recent study by Blakely and colleagues (58) further supports this premise, in which they reported that D2R activation increased DAT surface expression in the DS, but not VS. Given that few studies, to date, have identified the receptormediated signaling pathways that lead to PKC-stimulated DAT internalization, or the temporal profile of DAT response to PKC activation in DAergic terminals, it remains unclear how DAT insertion and internalization balance may occur. Finally, although our current data indicate that Rit2-KD is required for PKC-stimulated DAT endocytosis in female and male VS, it is important to bear in mind that DAT basal surface expression in both male (39) and female (Fig. 3B) VS was reduced following DAergic Rit2-KD. Thus, there is a possibility that a general floor effect blocked PKC-mediated DAT internalization, rather than a requirement for Rit2. However, we do not believe that this is likely, as DAT is capable of undergoing additive degrees of internalization when cells are subjected to dual PKC activation and AMPH exposure (59).
We used an extracellular BBS tag to interrogate surface DAT and its associated protein complex. This approach was previously used successfully by several groups (46,47,60) and has distinct advantages over co-IP approaches, because 1) the ␣-BTX/BBS affinity is significantly higher than that of antibody/antigen interactions, and 2) ␣-BTX/BBS binding is maintained in detergent lysates. Moreover, using an extracellular labeling approach, in general, maintains intracellular protein complexes that might be disrupted using an intracellularly targeted antibody in lysates/solution. Interestingly, although we observed significant DAT-Rit2 dissociation in response to releasing the PKC-sensitive DAT endocytic brake (Fig. 6), AMPH-stimulated DAT internalization increased the surface DAT-Rit2 population. Surface DAT is distributed among several membrane microdomains (38,(61)(62)(63)(64)(65)(66)(67). Moreover, we previously reported that 1) PKC activation preferentially depletes DAT from cholera toxin-positive (Ctxϩ) microdomains (26) and 2) there is significantly more DAT-Rit2 co-localization in Ctxϩ microdomains (38). Conversely, AMPH treatment increases DAT localization to Ctxϩ domains (63). Taken together, these results suggest that DAT may dissociate from Rit2 and internalize, preferentially from Ctxϩ microdomains in response to PKC activation. In contrast, AMPH potentially drives DAT internalization from CtxϪ domains, in which there is less DAT-Rit2 interaction, thereby leaving an enriched DAT-Rit2 population at the cell surface.
We originally identified the DAT-Rit2 interaction via a yeast two-hybrid screen using the C-terminal DAT domain "FRE-KLAYAIA" as bait, and FRET studies revealed that Rit2 directly interacts with DAT, but not SERT, at the plasma membrane (38). However, the domains required for the DAT-Rit2 interaction and their requirement for either PKC-stimulated DAT internalization or DAT-Rit2 dissociation were not defined. To address these questions, we used a series of DAT/SERT chimeras previously reported by our laboratory (53) and found that the DAT N terminus was required for PKC-stimulated DAT-Rit2 dissociation (Fig. 8). Indeed, under basal conditions, N-S/ DAT and Rit2 interacted to a significantly higher degree than DAT-Rit2 controls (Fig. 6), consistent with a lack of DAT-Rit2 dissociation for the N-S/DAT chimera and higher steady-state interaction. N-S/DAT also basally internalized significantly more slowly than WT DAT selectively in SK-N-DZ, consistent with its inability to disengage the endocytic brake. Moreover, DAT/C-S retained both Rit2 interaction and the ability to release the endocytic brake in response to PKC activation. This was surprising, given that 1) the DAT C-terminal bait (FRE-KLAYAIA) used to identify the DAT/Rit2 interaction is highly conserved across the SLC6 gene family (23,44) and 2) fulllength SERT does not interact with Rit2 (38). So, whereas the FREKLAYAIA domain is sufficient to interact with Rit2, its context within full-length DAT or SERT appears to dictate ultimate Rit2-binding potential. Interestingly, although N-S/DAT was resistant to PKC-stimulated internalization, it retained AIM-100 -dependent internalization, whereas S/DAT/S did not (Fig. 8), indicating that the DAT N terminus is not required for direct Ack1-dependent endocytic brake release and further supports the hypothesis that Rit2 and Ack1 converge on DAT independently in response to PKC activation. It should also be noted that we, and others, reported that AIM-100 also binds noncompetitively to DAT (24,68), and recent reports suggest that AIM-100 can enhance DAT surface oligomerization (68,69). Because both Ack1-dependent and AIM-100 -stimulated internalization are specific for DAT, and not SERT, it is possible that S/DAT/S endocytic resistance to AIM-100 may be because the substituted SERT domains perturb the DAT/AIM-100 interaction. Likewise, it is possible that the ability of N-S/DAT to internalize in response to AIM-100, but not PKC activation, may be due to a direct AIM-100 effect on DAT.
In this study, we present one of the first descriptions of region-and sex-dependent differences in DAT trafficking regulation. Furthermore, we greatly extend our knowledge of the mechanisms by which Rit2 governs DAT surface expression in bona fide DAergic terminals. Future studies that examine the cell autonomous endogenous signaling events that drive striatal DAT trafficking, and require Rit2, will shed further light on mechanisms that influence DA clearance and DA-dependent behaviors.

Materials
PMA was from LC Laboratories (P-1680). GF 109203X (BIM I) and AIM-100 were from Tocris-Cookson. All other reagents were from either Sigma-Aldrich or Fisher Scientific and were of the highest possible grade.

Rit2-dependent dopamine transporter trafficking Animals
All studies were conducted in accordance with University of Massachusetts Medical School Institutional Animal Care and Use Committee Protocol A-1506 (H. E. M). Pitx3 IRES2-tTA mice on the C57Bl/6 background were the generous gift of Dr. Huaibin Cai (NIA, National Institutes of Health) and were continually backcrossed to C57Bl/6 mice. Mice were maintained in a 12-h light/dark cycle at constant temperature, and humidity and food and water were available ad libitum.

AAV production and stereotaxic viral delivery
pscAAV-TRE-eGFP and pscAAV-TREmiR33-shRit2-eGFP plasmids were generated as described previously (39), and AAV particles (AAV9 serotype) were produced and purified and titers were determined by the University of Massachusetts Medical School Viral Vector Core, as described previously (70). For intracranial stereotaxic AAV injections, male and female mice (minimum 3 weeks of age) were anesthetized with 100 mg/kg ketamine/10 mg/kg xylazine (intraperitoneally), and 20% mannitol was administered (intraperitoneally) 15 min prior to viral delivery, to increase viral spread (71). Mouse heads were shaved and placed in the stereotaxic frame, and bilateral 0.8-mm holes were drilled into the skull at the indicated stereotaxic coordinates. 1 l of the indicated viruses were infused bilaterally into the VTA (bregma: anterior/posterior, Ϫ3.08 mm; medial/lateral, Ϯ0.5 mm; dorsal/ventral, Ϫ4.5 mm) at a rate of 0.2 l/min. Syringes were left in place for a minimum of 5 min postinfusion prior to removal. Mice were housed for a minimum of 4 weeks before experiments were performed. Viral expression in each animal was confirmed by visualizing GFP expression in 0.3-mm coronal ventral midbrain slices.

Ex vivo slice biotinylation
Ex vivo striatal slices were prepared 4 -5 weeks following viral injection in Pitx3 IRES2-tTA/ϩ mice (for Rit2 KD studies), or from 5-7-week-old C57Bl/6J (for PKC specificity studies). Mice were sacrificed by cervical dislocation and rapid decapitation, and heads were immediately submerged in ice-cold, oxygenated cutting solution, pH 7.4 (20 mM HEPES, 2.5 mM KCl, 1.25 mM NaH 2 PO 4 , 30 mM NaHCO 3 , 25 mM glucose, 0.5 mM CaCl 2 ⅐4H 2 O, 10 mM MgSO 4 ⅐7H 2 O, 92 mM N-methyl-Dglucamine, 2.0 mM thiourea, 5.0 mM Na ϩ -ascorbate, 3.0 mM Na ϩ -pyruvate) for 1 min. Brains were removed, and 300-m coronal slices were prepared with a VT1200 Vibroslicer (Leica) in ice-cold, oxygenated cutting solution. Slices were hemisected along the midline and were recovered in ACSF, pH 7.4 (125 mM NaCl, 2.5 mM KCl, 1.24 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM glucose, 2.4 mM CaCl 2 ⅐4H 2 O, and 1.2 mM MgCl 2 ⅐6H 2 O) for 40 min at 31°C. Hemislices were treated with or without 1 M PMA in ACSF for 30 min at 37°C with constant oxygenation, using their contralateral hemislice as a vehicletreated control. Following drug treatment, slices were moved to ice, and surface DAT was labeled by biotinylation as described previously (24,26,39). Briefly, slices were biotinylated with membrane-impermeant Sulfo-NHS-SS-biotin (1 mg/ml) for 45 min at 4°C. Residual biotin was quenched with two 20-min washes of ice-cold ACSF supplemented with 100 mM glycine, and slices were subjected to a final wash with ice-cold ACSF. Hemislices were enriched for dorsal and ventral striatum, by subdissecting in a line from the anterior commissure to the lateral olfactory tract. Subdissected slices were lysed in RIPA buffer containing protease inhibitors, and tissue was disrupted by triturating sequentially through a 200-l pipette tip, 22-and 26-gauge tech-tips. Samples were rotated for 30 min at 4°C to complete lysis, insoluble material was removed by centrifugation, and protein concentrations were determined using the BCA protein assay. Biotinylated proteins were isolated by batch streptavidin chromatography, overnight with rotation, at 4°C, at a ratio of 20 g of striatal lysate to 30 l of streptavidinagarose beads, which was empirically determined to recover all biotinylated DAT. Recovered proteins were washed with RIPA buffer and eluted from beads in 2ϫ SDS-PAGE sample buffer for 30 min at room temperature with rotation. Eluted (surface) proteins and their respective lysate inputs were resolved by SDS-PAGE, and DAT was detected by immunoblotting as described above. Percentage DAT at the cell surface was calculated by normalizing surface signals to the corresponding total DAT input signal in a given hemislice, detected in parallel. Note that all slice data for Rit2-KD experiments in the current study were acquired during our previous study, in which we first achieved AAV-mediated Rit2 KD (39). Basal DAT surface levels in vehicle-treated male slices were compared and reported in that study and thus were not reanalyzed for the current study.

Rit2-dependent dopamine transporter trafficking
However, the DAT surface levels from vehicle-treated male hemislices were reused in the current study as controls to determine whether PMA treatment affected DAT surface levels in contralateral hemislices. Rit2 knockdown in females was confirmed by RT-qPCR (Fig. S1c of the previous study (39)). For males, successful viral expression (AAV9-eGFP and AAV9-eGFP-shRit2) was confirmed by visual detection of GFP reporter immunofluorescence in midbrain slices from each experimental animal. For PKC specificity studies, data were reported as percentage change in DAT surface levels in drugtreated hemislices, normalized to their vehicle-treated, contralateral hemislices.
Lentiviral particles were produced in HEK293T cells as described previously (24). For cell transduction, 1 ϫ 10 6 cells DAT-SK-N-DZ cells were seeded onto 6-well plates and were transduced with 3.0 ml of lentiviral supernatant, supplemented with 0.8 g/ml Polybrene, 16 -24 h postseeding. Cells were selected for transduction beginning 24 h postinfection (72 h total) with SK-N-DZ medium supplemented with 1.25 g/ml puromycin. Assays were conducted 96 h post-transduction.

FRET
FRET was measured using an iMIC inverted microscope (TILL Photonics GmbH). Samples were focused using a ϫ60 (numerical aperture 1.49) oil objective (Olympus). Fluorescence was excited using a 100-watt xenon lamp (Polychrome, Till Photonics GmbH). Excitation light was filtered through either a 436/20-nm (CFP) or 514/10-nm (YFP) excitation filter (Semrock) and directed to the sample via a 442/514 dual line dichroic mirror (Semrock). Emitted fluorescence light was filtered through a 480/40-nm-570/80-nm dual emission filter (Semrock) and directed to a beamsplitter unit (Dichrotom, Till Photonics). Briefly, the emission light was separated spatially according to the fluorescence wavelength using a 515-nm dichroic mirror (Semrock). The resultant two channels (Ͻ515 nm and Ͼ515 nm) were projected next to each other onto an EMCCD chip (iXon Ultra 897Andor) and recorded using Live Acquisition software (version 2.5.0.21; TILL Photonics GmbH). To guarantee the best signal/noise ratio and dynamic range, the camera was operated in 16-bit mode with a readout speed of 1 MHz. According to the manufacturer's recommendation, an EM-gain of 16 was applied to overcome the noise floor. To analyze FRET (see below), two images were taken per set (donor excitation 3 donor emission/acceptor emission and acceptor excitation 3 acceptor emission, respectively). Per condition, 10 sets were recorded each experimental day; the images were then analyzed using Offline Analysis software (version 2.5.0.2, TILL Photonics GmbH). One region of interest (part of the plasma membrane) per cell was selected in the CFP channel. Background fluorescence was subtracted from each image, and the average intensity of each region of interest was used for calculations. Spectral bleed-through (BT) for donor (0.57) and acceptor (0.04) was determined using HEK293 cells expressing a CFP or YFP signal only. Normalized FRET (NFRET) was calculated as follows.

RNA extraction and RT-qPCR
RNA was isolated from cell lines and rodent midbrain using RNAqueous-Micro Kit RNA isolation (Thermo Fisher Scientific). For ventral midbrain samples, 1.0-mm bilateral tissue punches, encompassing both the ventral tegmental area and Rit2-dependent dopamine transporter trafficking substantia nigra par compacta, were taken from 300-m acute mouse and rat midbrain slices. Reverse transcription was performed using RETROscript reverse transcription kit (Thermo Fisher Scientific). Quantitative PCR was performed and analyzed using the Applied Biosystems 7500 Real-Time PCR System Machine and Software, using Taqman gene expression assays for human Rit2 (Hs01046673_m1), Rit1 (Hs00608424_m1), and GAPDH (Hs99999905_m1); mouse Rit2 (Mm01702749_mH) and GAPDH (Mm99999915_g1); and rat Rit2 (Rn01760884_m1) and GAPDH (Rn01775763_g1). Assays proceeded for 10 min at 37°C and were terminated by three rapid washes with ice-cold KRH buffer. Cells were solubilized in scintillation fluid, and accumulated radioactivity was determined by liquid scintillation counting in a Wallac MicroBeta scintillation plate counter. Nonspecific DA uptake was defined in the presence of 10 M GBR12909.

Internalization assays and immunoblotting
Relative internalization rates over 10 min were measured by reversible biotinylation as described previously (24,25,44). Briefly, the indicated stably transfected cells were seeded into 6-well plates at 1.5 ϫ 10 6 cells/well 1 day prior to analysis. Cells were incubated twice with 2.5 mg/ml Sulfo-NHS-SS-biotin (15 min, 4°C) and quenched twice with PBS 2ϩ (PBS, pH 7.4, 1.0 mM MgCl 2 , 0.1 mM CaCl 2 ) supplemented with 100 mM glycine (15 min, 4°C). Internalization was initiated by rapidly warming cells in prewarmed PBS 2ϩ supplemented with 0.18% glucose, 0.2% protease-/IgG-free BSA, and proceeded for 10 min at 37°C in the presence of the indicated drugs. Parallel surfacelabeled controls remained at 4°C. Cells were rapidly cooled by washing three times with ice-cold NT buffer (150 mM NaCl, 20 mM Tris, pH 8.6, 1.0 mM EDTA, pH 8.0, 0.2% protease-/IgGfree BSA), and remaining surface biotin was stripped by reducing in 100 mM tris(2-carboxyethyl)phosphine in NT buffer twice (25 min, 4°C). Cells were rinsed rapidly in PBS 2ϩ and were lysed in RIPA buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 1.0 mM EDTA, 0.1% SDS, 1% Triton X-100, 1% sodium deoxycholate) containing protease inhibitors (1.0 mM phenylmethylsulfonyl fluoride and 1.0 g/ml each leupeptin, aprotinin, and pepstatin). Lysates were cleared by centrifugation, and protein concentrations were determined with the BCA protein assay (Thermo Fisher Scientific) using BSA as a standard. Biotinylated proteins were recovered by streptavidin batch chroma-tography from equivalent amounts of cell lysate and were eluted in 2ϫ Laemmli sample buffer for 30 min at room temperature with rotation. Eluted proteins were resolved by SDS-PAGE, and proteins were detected and quantified by immunoblotting: hDAT and DAT/C-S were detected with amino-directed rat anti-DAT (MAB369, Millipore; 1:2000), and N-S/DAT and S/DAT/S were detected with amino-directed mouse anti-hSERT (mAb Technologies; 1:2000). Immunoreactive bands were detected using a VersaDoc imaging station (Bio-Rad) and were in the linear range of detection. Internalization rates were calculated as percentage of surface protein internalized/10 min, as compared with their respective surface signal at t ϭ 0 min (controls that were biotinylated and kept at 4°C). Note that for all representative immunoblots shown throughout the study, all brightness/contrast manipulations were made uniformly across any given blot. For presentation purposes, immunoreactive bands were cropped from the same exposure of the same immunoblot.

BBS-DAT/chimera pulldowns
HEK293T cells were transiently co-transfected with HA-Rit2, and either BBS-DAT or BBS-DAT chimeras, at a DAT/Rit2 plasmid ratio of 1:4, as described above. Cells were washed three times with ice-cold PBS 2ϩ , and surface BBS-DAT chimeras were labeled with 120 M biotinylated ␣-bungarotoxin (␣-BTX-b, Thermo Fisher) in PBS 2ϩ for 2 h at 4°C. For drug treatments, cells were treated with the indicated drugs for 30 min at 37°C, prior to labeling with ␣-BTX-b. Following labeling, ␣-BTX-b solution was removed, and cells were washed three times with ice-cold PBS 2ϩ and lysed in ice-cold co-IP lysis buffer (50 mM Tris, pH 7.4, 100 mM NaCl, 1% Triton X-100, 10% glycerol, and 1.0 mM EDTA) containing protease inhibitors (1.0 mM phenylmethylsulfonyl fluoride and 1.0 g/ml each leupeptin, aprotinin, and pepstatin) and Phosphatase Inhibitor Mixture V (EMD Millipore) for 30 min at 4°C. Labeled proteins were recovered from equivalent amounts of protein by batch affinity chromatography with streptavidincoupled M280 Dynabeads (Thermo Fisher) at 4°C overnight with rotation. Lysate/bead ratios were empirically determined to ensure quantitative recovery of all labeled proteins in lysates. Beads were gently washed three times with ice-cold co-IP buffer, with magnetic recovery between washes, and isolated proteins were eluted from beads in an equal volume of co-IP lysis buffer and 2ϫ SDS-PAGE sample buffer (100 mM Tris, pH 6.8, 4.4% SDS, glycerol, 100 mM DTT, and 0.04% bromphenol blue). Isolated proteins were resolved by SDS-PAGE, and specific protein bands were detected by immunoblotting with antibodies for rat anti-DAT (1:2000), mouse anti-hSERT (1:2000), and rabbit anti-HA (1:2000) as indicated above. Immunoreactive HA-Rit2 band densities were normalized to their respective recovered DAT, or chimera, bands in each independent experiment.

Statistical analysis
All data were analyzed using GraphPad Prism software. Prior to analyses, statistical outliers within data sets were identified using either Grubb's or Rout's outlier test and were removed from further analysis. Specific statistical tests used are detailed Rit2-dependent dopamine transporter trafficking within each figure legend. For comparisons between two groups, a Student's t test was used. For comparison among more than two experimental groups, one-way ANOVA with appropriate post hoc multiple-comparison test was performed, as indicated within each figure legend.

Data availability
All data presented and discussed are contained within the paper.