Amphetamine activates Rho GTPase signaling tomediate dopamine transporter internalization andacute behavioral effects of amphetamineDavid S. Wheelera,1,2, Suzanne M. Underhillb,2,3, Donna B. Stolzc, Geoffrey H. Murdochd, Edda Thielse,Guillermo Romeroa, and Susan G. Amarab,3

aDepartment of Pharmacology and Chemical Biology, University of Pittsburgh, Pittsburgh, PA 15213; bLaboratory of Molecular and Cellular Neurobiology,National Institute of Mental Health, National Institutes of Health, Bethesda, MD 20892; cCenter for Biologic Imaging, University of Pittsburgh, Pittsburgh, PA15261; dDepartment of Pathology, Division of Neuropathology, University of Pittsburgh, Pittsburgh, PA 15213; and eDepartment of Neurobiology, Centerfor the Neural Basis of Cognition, Center for Neuroscience, University of Pittsburgh, Pittsburgh, PA 15260

Contributed by Susan G. Amara, October 2, 2015 (sent for review September 3, 2015; reviewed by Aurelio Galli, Harald H. Sitte, and Roxanne A. Vaughan)

Acute amphetamine (AMPH) exposure elevates extracellular do-pamine through a variety of mechanisms that include inhibition ofdopamine reuptake, depletion of vesicular stores, and facilitationof dopamine efflux across the plasma membrane. Recent work hasshown that the DAT substrate AMPH, unlike cocaine and othernontransported blockers, can also stimulate endocytosis of theplasma membrane dopamine transporter (DAT). Here, we showthat when AMPH enters the cytoplasm it rapidly stimulates DATinternalization through a dynamin-dependent, clathrin-indepen-dent process. This effect, which can be observed in transfectedcells, cultured dopamine neurons, and midbrain slices, is mediatedby activation of the small GTPase RhoA. Inhibition of RhoA activitywith C3 exotoxin or a dominant-negative RhoA blocks AMPH-induced DAT internalization. These actions depend on AMPH entryinto the cell and are blocked by the DAT inhibitor cocaine. AMPHalso stimulates cAMP accumulation and PKA-dependent inactiva-tion of RhoA, thus providing a mechanism whereby PKA- andRhoA-dependent signaling pathways can interact to regulate thetiming and robustness of AMPH’s effects on DAT internalization.Consistent with this model, the activation of D1/D5 receptors thatcouple to PKA in dopamine neurons antagonizes RhoA activation,DAT internalization, and hyperlocomotion observed in mice afterAMPH treatment. These observations support the existence of anunanticipated intracellular target that mediates the effects of AMPHon RhoA and cAMP signaling and suggest new pathways to targetto disrupt AMPH action.

dopamine transporter | amphetamine | Rho GTPase | protein kinase A |endocytosis

Both cocaine and amphetamine (AMPH) are psychostimulantsthat induce euphoria and hyperactivity by increasing extra-

cellular dopamine. Despite the apparent commonalities in theeffects and mechanism of action of the two drugs, amphetamineexposure has behavioral, neuroadaptive, and neurotoxic conse-quences not associated with cocaine use (1–5). These dissimi-larities likely stem from several differences in the mechanism ofaction of the two drugs. Amphetamines, unlike cocaine andother nontransported inhibitors of the dopamine transporter(DAT), are transporter substrates that compete with dopaminefor transport, enter dopamine neurons, enhance efflux of dopa-mine, and also stimulate internalization of the DAT from the cellsurface (6, 7). To better understand how amphetamines mediatetheir effects within the cell, we undertook a series of studies toestablish the cellular pathways and signaling mechanisms thatunderlie the endocytosis of the DAT triggered by amphetamines.AMPH-mediated DAT internalization has been shown to be

dynamin-dependent (6). Dynamin-dependence is a key featureof clathrin-mediated endocytosis, and, indeed, previous reportsof clathrin-dependent constitutive cycling (8) and PKC-mediatedinternalization (9) of DAT make it reasonable to consider that

the same machinery would be involved in AMPH-mediated in-ternalization. A secondary pathway of dynamin-mediated inter-nalization of membrane-localized proteins involves small GTPasesof the Rho family. Rho proteins are composed of three subfam-ilies (Rho, Rac, and CDC42), each of which play a major role inactin cytoskeletal reorganization and vesicular trafficking (10). Tofurther characterize the dynamin-dependent mechanism of DATinternalization in response to AMPH, we examined if it was me-diated by clathrin or whether it was a Rho-dependent process.We also sought to resolve which signaling pathways activated

by AMPH trigger internalization, whether these effects reflect adirect intracellular action of AMPH, and whether they dependon the elevation of extracellular dopamine.Here, we demonstrate that AMPH [(+)-α-methyl phenylethyl-

amine] acts within the cytoplasm to stimulate Rho family GTPasesand trigger DAT internalization through a Rho- and dynamin-dependent pathway that does not involve clathrin. We also show

Significance

The dopamine transporter (DAT), a major target for psychosti-mulant drugs, including cocaine and amphetamines, clears ex-tracellular dopamine and restricts the temporal and spatialextent of neurotransmitter signaling. This study examines themechanism throughwhich amphetamines trigger internalizationof DAT and demonstrates that amphetamine activates the smallGTPases, Rho and Rac. Rho activation triggers endocytosis ofDAT by a dynamin-dependent, clathrin-independent pathway.Intriguingly, amphetamine must enter the cell to have theseeffects, and it also increases cAMP, which in turn inactivates Rhoand limits carrier internalization. Consistent with these obser-vations, the activation of receptors that couple to protein kinase Ain dopamine neurons also antagonizes the behavioral effectsof amphetamine in mice, suggesting new pathways to targetto disrupt amphetamine action.

Author contributions: D.S.W., S.M.U., G.H.M., E.T., G.R., and S.G.A. designed research; D.S.W.,S.M.U., D.B.S., G.H.M., and E.T. performed research; G.R. and S.G.A. contributed newreagents/analytic tools; D.S.W., S.M.U., G.H.M., E.T., G.R., and S.G.A. analyzed data; andD.S.W., S.M.U., G.H.M., G.R., and S.G.A. wrote the paper.

Reviewers: A.G., Vanderbilt University; H.H.S., Medical University Vienna, Center forPhysiology and Pharmacology; and R.A.V., University of North Dakota School of Medicine &Health Sciences.

The authors declare no conflict of interest.

See Commentary on page 15538.1Present address: Department of Medicine, Einstein Medical Center Philadelphia, Philadelphia,PA 19141.

2D.S.W. and S.M.U. contributed equally to this work.3To whom correspondence may be addressed. Email: susan.amara@nih.gov or smunderhill@yahoo.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1511670112/-/DCSupplemental.

E7138–E7147 | PNAS | Published online November 9, 2015 www.pnas.org/cgi/doi/10.1073/pnas.1511670112

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

that cytoplasmic AMPH stimulates a secondary pathway of cAMPproduction, which leads to Rho inactivation by PKA-dependentphosphorylation. This transient period of Rho activation corre-lates with ongoing DAT internalization, and this temporal windowof activation can be shortened by stimulating cell-surface Gs-coupled receptors to raise cAMP and increase the rate of Rhoinactivation. We also show in vivo that preemptive D1/D5 re-ceptor stimulation to activate PKA can reduce AMPH-evokedhyperlocomotion without altering the increase in locomotion in-duced by cocaine. Because these signaling events occur in trans-fected cell lines as well as in neurons, the results indicate that, onceAMPH enters the cell, it activates Rho, Rac, and PKA signalingindependently of any actions to elevate intracellular dopamine. Thedirect activation of cytoplasmic signaling cascades by AMPH withinthe cell can contribute to the behavioral effects of acute amphet-amine exposure and may explain some of the unique neurobiolog-ical consequences associated with amphetamine use and abuse.

ResultsAMPH-Induced DAT Internalization Is Clathrin-Independent. Previousstudies have indicated that the endocytosis of DAT activated byAMPH exposure is dynamin-dependent (6); however, the clathrindependence of this process has been established only for con-stitutive DAT internalization (8). Confocal examination of amonoaminergic neuroblastoma line, SK-N-SH, expressing GFP-tagged clathrin-light chain (GFP-CLC) and an mCherry-conju-gated DAT (mCherry-DAT) after AMPH treatment does notshow marked colocalization of the two proteins (Fig. 1A). Fur-ther analyses of AMPH-treated SK-N-SH cells using live-cell

total internal reflection fluorescence (TIRF) microscopy dem-onstrate that less than 0.5% of internalizing DAT colocalizeswith clathrin (Movie S1). In contrast, parallel experimentsmonitoring transferrin accumulation confirm previous studiesthat over 80% of internalizing transferrin colocalizes with clathrin(Fig. 1B). Thus, AMPH-stimulated DAT internalization inSK-N-SH cells is dynamin-dependent, but clathrin-independent.Dynamin-dependent, clathrin-independent internalization is gen-

erally divided into two categories depending on whether Rho ac-tivation is required (11). Coexpression of the DAT with a plasmidencoding GFP-tagged C3 exotoxin, a clostridial enzyme that ADPribosylates and inactivates Rho family GTPases, prevents AMPH-induced clustering and internalization of DAT in SK-N-SH cells(Fig. 1 C–E and Movie S2). Using TIRF microscopy we observedthat expression of C3 for 18–24 h in SK-N-SH cells reduces thenumber of focal adhesions, consistent with Rho inhibition. Thecontinuing spontaneous movement of the C3(+) cells and theirability to transport dopamine (Fig. 1F) indicate that the cells arestill viable. The presence of the Rho inhibitor prevented theincrease in DAT endocytosis observed with AMPH, supporting arole for Rho-GTPases in DAT internalization.Expression of a dominant-negative RhoA (T19N) also inhibits

the effects of AMPH on dopamine uptake (Fig. 1F), and DATcell-surface expression by TIRF microscopy (Fig. S1), an ob-servation that further confirms the significance of Rho activationto AMPH-stimulated DAT trafficking. Expression of either theC3 or the Rho (T19N) mutant did not alter baseline DAT ac-tivity, suggesting that RhoA activation does not play a role inconstitutive cycling of the DAT. In addition, the effects of AMPH

Fig. 1. AMPH-induced DAT internalization is clathrin-independent but Rho-dependent. (A) SK-N-SH cells were cotransfected with mCherry-DAT and aclathrin light chain-GFP fusion protein and stimulated with AMPH for 5 min. AMPH causes DAT (red) to internalize but, once endocytosed, DAT fails tocolocalize with clathrin (green). (B) SK-N-SH cells were transfected with a GFP-clathrin light chain and treated with transferrin conjugated to Alexa-Fluor-568for 5 min. These cells exhibit colocalization of clathrin with the transferrin-bound receptors. (C) SK-N-SH cells were cotransfected with mCherry-DAT (green)and GFP-C3 (red; note that pseudocoloring has been reversed for consistency throughout the figures). The field shown here is of a cell expressing DAT and C3(yellow) and a cell expressing DAT alone (green) at T = 0 (Movie S2). In the absence of C3 coexpression, AMPH causes robust clustering and internalization ofDAT at T = 15 and loss of surface DAT by T = 30 (Right, Top). However, AMPH fails to alter DAT at either T = 15 or T = 30 in a cell coexpressing C3 with DAT(Right, Bottom). The time course of AMPH-induced internalization in control DAT- and DAT-C3–expressing SKNSH cells assessed by (D) surface DAT fluo-rescence and (E) the number of internalizing puncta. The arrow indicates the start of AMPH treatment. (F) AMPH pretreatment decreases [3H]-dopamineuptake by DAT. Coexpression of a dominant-negative RhoA (T19N) or the Rho inhibitor C3 blocks this effect. The constitutively active form of RhoA (V14)decreases dopamine uptake and does not exhibit any further decrease in response to AMPH pretreatment. (G) Isolation of activated Rho and Rac by GST-RBDand GST-p21, respectively, from DAT-transfected SK-N-SH cells shows AMPH-mediated activation of Rho and Rac1 GTPases over time. (Scale bars: 25 μm inA and B, and the insets have been enlarged 3-fold; 30 μm in C.) ***P < 0.001, two-way ANOVA; #P < 0.001, one-way ANOVA compared with vehicle RhoA control.

Wheeler et al. PNAS | Published online November 9, 2015 | E7139

NEU

ROSC

IENCE

PNASPL

US

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

are mimicked by expression of a constitutively active RhoA (V14)(Fig. 1F). Rho family GTPases regulate internalization by con-trolling rearrangement of the actin cytoskeleton and activation ofcritical enzymes such as phospholipase D (12, 13). AMPH-inducedDAT internalization is sensitive to inhibitors of actin dynamics(latrunculin A and jasplakinolide) (Fig. S2A) and phospholipaseD2 (dominant-negative PLD2 isoform) (Fig. S2B), and it thusresembles other Rho-mediated internalization processes.

To directly assess the activation of Rho family members byAMPH, active (GTP-bound) Rho and Rac were affinity-purifiedfrom SK-N-SH cell lysates using GST-tagged constructs derivedfrom the Rho-binding domain of rhotekin (GST-RBD) and theGTPase-binding motif of p21-activated kinase 1 (GST-PAK1).These constructs selectively bind the activated forms of Rho andRac, respectively (14, 15). The bound activated GTPases wereidentified on immunoblots using specific antibodies to Rho or

Fig. 2. AMPH causes internalization of DAT in dopamine neurons. (A) In midbrain cultures, AMPH pretreatment causes a loss of DAT-mediated dopaminetransport in a time-dependent process. (B) Pretreatment with 10 μM AMPH for 30 min leads to a substantial loss of dopamine transport in primary cultures.Coapplication of the DAT inhibitor cocaine (100 μM) blocks this effect of AMPH pretreatment. (C) Primary murine midbrain neurons transfected with theGFP-AP-DAT construct (green) demonstrate DAT internalization by AMPH through decreased Alexa Fluor 568-streptavidin (red) surface labeling. (D) The ratioof surface GFP-AP-DAT (Alexa Fluor 568) to total GFP-AP-DAT (GFP) in TH(+) neurons is normalized to vehicle control. AMPH (10 μM) treatment reducessurface DAT by 58 ± 6%, and this effect is blocked by 100 μM cocaine. (E) Bar graph showing dopamine uptake in control primary midbrain cultures witheither vehicle control or the Rho inhibitor exotoxin C3 introduced into the cells by the scrape method (12). Approximately 18 h after the neurons were scrape-loaded in the presence or absence of the C3 exotoxin, neuronal cultures were treated with AMPH for 30 min. AMPH significantly decreases dopaminetransport in the sham-treated control cultures but has no effect on the cells treated with C3. (F) The ROCK inhibitor Y27632 also attenuates the AMPH-induced decrease in [3H]-dopamine uptake in primary cultures. (G) Primary midbrain neurons expressing GFP-AP-DAT labeled with streptavidin-conjugatedgold particles. Transmission electron micrographs of AMPH-stimulated neurons reveal gold particles (arrowhead) in tubulovesicular bodies (TV) and multi-vesicular bodies (MVB, Inset). Gold particles are absent from clathrin-coated pits or vesicles (CCV). (Scale bars: 50 μm in C, 100 nm in G, and 75 nm in the Insetof G). *P < 0.05, ***P < 0.001, two-way ANOVA.

E7140 | www.pnas.org/cgi/doi/10.1073/pnas.1511670112 Wheeler et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

Rac1. As shown in Fig. 1G, AMPH stimulates activation of bothof these GTPases in a time-dependent fashion.

AMPH-Mediated DAT Internalization in Neurons. To determinewhether a mechanism involving Rho-dependent endocytosis op-erates in dopamine neurons, we examined the effect of AMPHpretreatment on transport of radiolabeled dopamine into primarycultures of midbrain neurons. AMPH pretreatment caused a time-dependent down-regulation of endogenous DAT activity (Fig.2A). The decrease in dopamine transport activity after 30 min of10 μMAMPH pretreatment could be prevented by coapplicationof the nontransported inhibitor, cocaine, which blocks transportof AMPH into the cells (Fig. 2B). Pretreatment with cocainealone under the same conditions had no effect on transport ac-tivity (Materials and Methods). Taken together, these results areconsistent with the idea that the AMPH acts on an intracellulartarget to mediate its effects on DAT trafficking (7).To confirm that the decrease in transport activity reflects a

loss of membrane-localized DAT, we engineered a bicistronicplasmid that expresses DAT with a 15-amino-acid biotin ligaseacceptor peptide (AP) sequence (16) substituted in the secondextracellular loop of DAT (residues 193–203; GFP-AP-DAT)and the Escherichia coli BirA biotin ligase with an endoplasmicreticulum retention signal (BirA-KDEL). We have shown thatBirA-KDEL biotinylates the AP tag of the nascent membraneproteins efficiently and selectively within the lumen of the ERbefore they are trafficked to the surface (17). This biotinylatedDAT can be labeled at the surface with streptavidin conjugatedto fluorescent dyes (Fig. 2C). The presence of the biotinylatedAP tag did not significantly alter the ability of the carrier totransport dopamine. In GFP-AP-DAT–transfected primarymidbrain neurons, AMPH treatment reduces the proportion ofDAT on the surface (Fig. 2 C and D). Cocaine alone does notstimulate DAT internalization and yet blocks AMPH-mediatedinternalization (Fig. 2D).To further corroborate that AMPH acts within the cytoplasm

through a mechanism that does not depend on its ability to el-evate extracellular dopamine, we cotransfected HEK293 cells,which neither synthesize nor release dopamine, with GFP-AP-DAT and the norepinephrine transporter, NET. Because NETtransports AMPH with high affinity, these experiments allowedus to test directly whether transport through DAT is specificallynecessary for the actions of AMPH. Blocking AMPH transportthrough DAT with GBR12909 had no effect on AMPH-mediatedRho activation and internalization of DAT, nor did blockingAMPH transport only through NET with desipramine (Fig. S3A–C). However, blocking AMPH cytosolic access entirely bycoapplication of GBR12909 and desipramine prevented AMPH’seffects on Rho activation and DAT surface expression. Similarly,blocking both NET and DAT with cocaine also prevented AMPH-mediated DAT internalization. These data support previous ob-servations (7) that AMPH must enter the cell to exert an effect ontransporter trafficking; however, the means by which AMPH en-ters the cell, either directly through DAT or by another route (e.g.,NET), does not appear crucial.To examine the role of Rho activation in AMPH-mediated

DAT modulation in primary neurons, we again used the C3clostridial toxin to inhibit Rho activity; however, because primaryneurons are difficult to transfect efficiently with a plasmidexpressing the toxin (as in Fig. 1F), we used a scrape-loadingassay (12) to introduce the toxin itself into the cells. Parallelassays with the cell-impermeable fluorophore calcein confirmedthe efficient uptake of the C3 ribosyl transferase enzyme usingthis method. The presence of the C3 toxin prevented the de-crease in dopamine uptake mediated by AMPH in primaryneuronal cultures, but not in sham-treated control cultures,which indicates that DAT internalization in these cells also de-pends on the activation of Rho GTPases (Fig. 2E).

To further confirm the role of Rho activation on AMPH-mediated DAT internalization in primary neurons, we investi-gated a potential downstream effector of Rho activation. TheRho-associated coiled-coil containing kinase (ROCK) is acti-vated by Rho GTPases and plays a critical role in actin cyto-skeletal rearrangements. Coapplication of the ROCK inhibitor,Y27632, blocked the effects of AMPH pretreatment on dopa-mine uptake in primary midbrain cultures (Fig. 2F). These datafurther support a role for Rho activation in the mechanism ofaction of AMPH.We were also able to use the GFP-AP-DAT system to in-

vestigate DAT trafficking ultrastructurally using electron mi-croscopy (EM). In a representative experiment shown in Fig. 2G,we labeled surface GFP-AP-DAT with gold-conjugated strepta-vidin in cultured midbrain neurons. Ten minutes after AMPHtreatment, gold particles were observed on the inner leaflet ofcytoplasmic, irregularly shaped, single membrane structures80–100 nm in width (TV, Fig. 2G). These structures are consistentwith tubulovesicular bodies, which are a hallmark of clathrin-independent internalization (11). Intriguingly, in immuno-EMstudies of midbrain sections, DAT has been observed to be lo-calized not only on the surface, but also prominently in cyto-plasmic tubulovesicles (18). Constitutive DAT internalizationhas been characterized as a clathrin-mediated process (8). How-ever, we did not find gold particles in clathrin-coated pits orclathrin-coated vesicles (CCV), suggesting that constitutive in-ternalization is low relative to AMPH-induced internalizationunder these conditions.

AMPH Activates Rho Family GTPases in Acute Brain Slices. Althoughthe localization of GFP-AP-DAT in tubulovesicular bodies sup-ports a role for Rho activation in mediating the effects of AMPHon DAT trafficking in neurons, it was important to confirm thatAMPH could activate Rho in the adult mouse brain. We mea-sured endogenous Rho, Rac, and CDC42 activation in response toAMPH in acute midbrain slices and observed activation of bothRho and Rac, but not CDC42 as early as 5 min after AMPHaddition (Fig. 3 A and B). The precise anatomical localization ofactivated Rho was determined in acute brain sections by stainingwith biotinylated GST-RBD. In AMPH-treated sections, therewas a clear region of GST-RBD staining that overlapped withtyrosine hydroxylase (TH)-positive neurons of the substantia nigra(Fig. 3C). Intriguingly, rhotekin binding was observed pre-dominantly in neuronal processes and varicosities and not in thecell bodies, suggesting that Rho activation is spatially regulatedwithin dopaminergic neurons (Fig. 3D).

AMPH-Induced Increases in cAMP Lead to Rho Inactivation andTermination of DAT Internalization. The activation of Rho byAMPH is rapid but transient, peaking within 5–10 min in cells,neurons, and tissue (Figs. 1F and 3 A and B). Because AMPHincreases cAMP production (19), and because it has been shownthat PKA negatively regulates RhoA by phosphorylation of a serineresidue, S188 (20, 21), we hypothesized that AMPH-mediatedactivation of PKA pathways could also serve as brake on Rhosignaling. Fig. 4A shows the time course of cAMP accumulation inSK-N-SH cells after AMPH treatment. Peak phosphorylation ofRhoA at S188 occurs after the initial Rho activation and, as pre-dicted by the role of S188 phosphorylation in Rho inactivation,parallels the time course of inactivation of Rho (Fig. 4B). Thesedata suggest that the activation of RhoA by AMPH is terminatedby AMPH-mediated activation of PKA.To verify that Rho phosphorylation by PKA terminates the

DAT internalization, SK-N-SH cells were transfected with aconstruct encoding a mutant RhoA (S188A) that cannot bephosphorylated by PKA (22). In these cells, AMPH treatmentinduced significantly more DAT internalization, confirming thatphosphorylation of RhoA by PKA terminates RhoA-mediated

Wheeler et al. PNAS | Published online November 9, 2015 | E7141

NEU

ROSC

IENCE

PNASPL

US

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

internalization (Fig. 4C and Fig. S1). Taken together, these ob-servations suggest that AMPH signaling involves a wave of RhoAactivation that initiates DAT internalization followed by a waveof PKA-mediated inactivation of RhoA that terminates theendocytic process (Fig. 4F).This model predicts that the magnitude of DAT internaliza-

tion depends on the balance of RhoA activation and PKA-dependent RhoA inactivation. Pharmacological manipulation ofPKA activity should shift this equilibrium and alter the amountof DAT internalization. Consistent with this idea, pretreatmentof SK-N-SH cells with dibutyryl-cAMP (db-cAMP), a PKA ag-

onist, blocks AMPH-induced DAT internalization, whereaspretreatment with KT5720, a PKA inhibitor, enhances the ef-fects of AMPH on the DAT (Fig. 4D). Using the GFP-AP-DAT/BirA construct, we also confirmed that modulation of PKA ac-tivity alters AMPH-mediated DAT internalization similarly inprimary TH-positive midbrain neurons (Fig. 4E). Moreover,experiments using a radiolabeled cocaine analog, [125I]RTI-55,to identify surface carriers in acute brain sections also confirmedthat increases in PKA activity decrease AMPH-mediated DATinternalization, whereas PKA inhibitors enhance the effects ofAMPH (Fig. S4 A and B). Taken together, these observations fit

Fig. 3. AMPH activates Rho and Rac in dopamine neurons in midbrain slices. (A) Acute midbrain slices from adult Swiss-Webster mice were treated with10 μM AMPH for the indicated time. Activated GTP-bound Rho, Rac, and CDC42 were affinity-purified with the GST fusion proteins and analyzed by Westernblot. (B) Blots of activated GTPases were quantified by densitometry. AMPH treatment causes maximal activation of Rho and Rac at 5 min. (C) Frozen sectionsof vehicle- and AMPH-treated acute midbrain slices were stained with biotinylated GST-RBD (Rho-binding domain) to anatomically localize activated Rho. Invehicle-treated sections, staining with GST and GST-RBD were almost identical, indicating that Rho activity is minimal at baseline. In AMPH-treated sections,GST-RBD selectively stained a cluster of cells (arrowheads) in the midbrain. This staining overlaps with TH staining of dopamine neurons in adjacent sections.(D) Higher magnification of the region containing TH(+) cell bodies. Rho activation appears predominantly within neuronal processes. (Scale bar: 1 mm in Cand 75 μm in D.)

E7142 | www.pnas.org/cgi/doi/10.1073/pnas.1511670112 Wheeler et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

with the model presented in Fig. 4F. Although the precisemechanism by which AMPH stimulates Rho and cAMP signalingis unknown, our data indicate that AMPH acts on an intracellulartarget (or targets) to produce these effects.

Endogenous G Protein-Coupled receptors Modulate AMPH’s Effectson DAT Internalization and Locomotor Behavior. cAMP concentra-tions in dopaminergic neurons are modulated by multiple Gprotein-coupled receptors (GPCRs). SK-N-SH cells also expressmany of the GPCRs found in dopamine neurons, and we usedthis system as a model for how activation of endogenous re-ceptors might integrate PKA-signaling pathways with the actionsof AMPH on Rho signaling and DAT internalization. In thesestudies, we observed that stimulation of the Gs-coupled D1/D5dopamine receptors or the β2 adrenergic receptor, both whichare known to increase cAMP, are sufficient to block AMPH-induced DAT internalization as measured by dopamine uptake(Fig. 5A). We also tried to enhance Rho activation and DATinternalization by stimulating two Gi-coupled receptors, theD2 dopamine receptor or the α2-adrenergic receptor, but these

manipulations failed to modulate DAT internalization. Finally,stimulation of Gq-coupled muscarinic or endothelin 1 receptors,which do not directly affect adenyl cyclase, had no effect on DATinternalization. The D1/D5-selective agonist SKF38393 also blocksthe internalization of GFP-AP-DAT in primary dopamine neurons(Fig. 5B). Furthermore, in midbrain slices we found that AMPH-induced Rho activation could also be prevented by stimulationof D1/D5 Gs-coupled receptors by preapplication of SKF38393(Fig. 5 C and D). These data suggest that the acute in vivo actionsof AMPH might be mitigated by stimulating Gs-coupled receptorsexpressed on midbrain dopamine neurons.The ability of D1/D5 agonists to modulate AMPH-induced Rho

activation, and DAT internalization provides a useful tool to in-vestigate these signaling cascades in vivo. Adult male Swiss-Webstermice that received i.p. injections of AMPH (2 mg/kg) showed adecrease in surface DAT in midbrain slice preparations as mea-sured by biotinylation (Fig. 6A) and binding of [125I]RTI-55 (Fig.6B). As in primary neurons, SKF38393 blocked DAT internaliza-tion (Fig. 6B). The ability of SKF38393 to block AMPH-mediatedDAT internalization was paralleled by the blunting effect of

Fig. 4. AMPH stimulates cAMP, which terminates Rho signaling and DAT internalization. (A) Cumulative cAMP generation was measured in DAT-transfectedSK-N-SH cell lysates in the presence of 3-isobutyl-1-methylxanthine (IBMX). AMPH (10 μM) treatment of the cells causes a time-dependent increase in cAMPlevels with a τ1/2 of 12.5 min. (B) AMPH treatment increases the amount of phospho-Ser188 RhoA. (Inset) Immunoblot of SK-N-SH protein lysates probed withan antibody that recognizes P-Ser188 RhoA (Top) and total RhoA (Bottom). (C) Expression of RhoA S188A, a mutant that cannot be inactivated by PKA,enhances AMPH-induced DAT internalization as assessed by uptake of [3H]-dopamine. (D) Activation of PKA by 10 μM dibutyryl-cAMP blocks AMPH-mediatedinternalization of DAT. In contrast, inhibition of PKA with 600 nM KT5720 enhances DAT internalization. (E) In primary neurons, the actions of db-cAMP andKT5720 parallel the results seen in SK-N-SH cells as assessed by membrane labeling of the GFP-AP-DAT protein (F) A schematic illustrating the crosstalkbetween AMPH-induced Rho activation and cAMP signaling. AMPH activates Rho but also stimulates cAMP production, which in turn activates PKA toterminate Rho activation and DAT internalization. *P < 0.05, **P < 0.01, ***P < 0.001, two-way ANOVA; #P < 0.01 compared with RhoA + AMPH in C.

Wheeler et al. PNAS | Published online November 9, 2015 | E7143

NEU

ROSC

IENCE

PNASPL

US

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

SKF38393 on AMPH-induced hyperlocomotion (Fig. 6 C andD). Mice treated with SKF38393 in the absence of drug exhibitqualitatively similar movement patterns to those injected withvehicle (Fig. 6C). Importantly, SKF38393 did not inhibit the lo-comotor effects of cocaine, which demonstrates that this concen-tration of the D1/D5 agonist did not have a global effect onpsychostimulant-induced hyperlocomotion. Taken together, thesefindings suggest that regulated DAT internalization contributessignificantly to the locomotor effects of AMPH.

DiscussionUnlike cocaine and other nonsubstrate DAT blockers, amphet-amines have the capacity to enter neurons and stimulate endo-cytosis of the DAT and other surface membrane proteins. Tolearn more about the cellular signaling pathways activated byintracellular AMPH, we focused on the mechanism of AMPH-mediated DAT internalization, a process that intriguingly hasbeen shown to be disrupted in DAT variants associated withADHD (23) and autism (24). We show in the work presentedhere that cytoplasmic AMPH transiently activates Rho familyGTPases, which stimulate DAT internalization via a clathrin-independent mechanism. Cytoplasmic AMPH also stimulatescAMP accumulation and increases PKA activity, which feedsback to phosphorylate and inactivate Rho.Modulation of cAMP and PKA by pharmacological inhibitors

or endogenous ligands of Gs-coupled receptors alters AMPH-induced DAT internalization in a manner consistent with a neg-ative feedback loop in which PKA activation limits Rho-mediatedinternalization. This integration of Rho and PKA-signaling cas-

cades provides a mechanism through which dopamine neuronscan dynamically regulate plasma-membrane molecules in re-sponse to both intracellular and extracellular signals. The timingand relative activation of each pathway determines the net effectof Rho activation on trafficking of DAT and other potentialtargets. As a consequence of reduced DAT surface expression,an increase in dopamine receptor activation could further modifythe cascades initiated by AMPH. Taken together our data illu-minate a series of previously unidentified actions of AMPH onintracellular signaling pathways that are not mimicked by non-transportable DAT blockers, such as cocaine.

AMPH Activated Rho, Rac, and PKA. To examine the possible be-havioral consequences of these events, we stimulated D1/D5receptors with SKF38393 before AMPH treatment to inhibitRho activation and were able to block the effects of AMPH-induced hyperlocomotion as well as DAT surface expression.Consistent with this model, D5 knockout mice display enhancedlocomotor activity in response to methamphetamine (25). However,our data also suggest that there is a component of psychostimulant-induced hyperlocomotion that is not Rho-mediated becausethe activation of D1/D5 receptors does not bring the AMPH-induced hyperlocomotion down to control levels, but rather tothe level produced by cocaine. It is important to note that, inaddition to its effects on Rho-mediated transporter trafficking,AMPH has the potential to elevate extracellular dopaminethrough other mechanisms, such as facilitating efflux and inhib-iting the carriers. Thus, it will be critical to establish which of thevarious effects predominates at different times after AMPH

Fig. 5. Activation of endogenous Gs-coupled GPCRs prevents AMPH effects. (A) Stimulation of Gs-coupled receptors blocks the AMPH-mediated decrease indopamine uptake in SK-N-SH cells. D1/D5 dopamine receptors were stimulated with 100 nM SKF38393, and β2 adrenergic receptors were stimulated with10 μM isoproterenol. Activation of muscarinic (mAch; 10 μM carbachol), endothelin (ET1; 1 μM endothelin), D2 dopamine (D2; 10 nM bromocriptine), or α2adrenergic (α2; 100 nM norepinephrine) receptors does not alter the effect of AMPH treatment. Cells were treated with AMPH (10 μM) with or without GPCRagonist for 30 min, washed, and assayed for uptake of [3H]-dopamine. (B) D1/D5 receptor activation with 100 nM SKF38393 is able to suppress AMPH-inducedGFP-AP-DAT internalization in primary neurons. (C and D) AMPH-induced Rho activation is also inhibited by D1/D5 receptor stimulation. A 30-min exposure ofmidbrain slices to AMPH leads to increased GTP-bound activated Rho that is blocked by SKF38393. *P < 0.05 and ***P < 0.001, two-way ANOVA.

E7144 | www.pnas.org/cgi/doi/10.1073/pnas.1511670112 Wheeler et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

administration to understand the contribution of each of thedifferent mechanisms to the effects of AMPH in vivo.The effects of AMPH on intracellular signaling suggests a

physiological role of endogenous AMPH-like ligands, whichcould initiate Rho and PKA activation within dopamine neurons.A number of endogenous amine compounds such as β-phenyl-ethylamine have been proposed as potential “endogenous am-phetamines” (26). However, another plausible physiological Rhoactivator could be dopamine itself; consistent with this idea, wehave preliminary data demonstrating that cytoplasmic dopaminecan indeed activate Rho.The complex downstream effects of Rho-family GTPase-signaling

pathways on cellular functions such as actin remodeling andmembrane protein trafficking may explain some of the distinctfeatures of the brain’s response to AMPH compared with co-caine. For example, the activation of Rho modulates the actincytoskeleton and neurite extension (27), both of which have beenlinked to the plasticity and alterations in dendritic spine mor-phology observed in models of addiction (28). Our data using aROCK inhibitor to block the effects of AMPH pretreatment ondopamine uptake link ROCK activation to DAT internalization,complementing previous studies that suggest a role for ROCK insome aspects of AMPH’s behavioral effects (29). AMPH treat-ment also stimulates Rac, which influences neurite extension andbranching and oxidative stress-induced cell death (30). Further-more, the potential transcriptional consequences of AMPH’sability to increase cAMP and activate CREB signaling in the nu-cleus is also likely to contribute to longer-term adaptive changes indopamine neurons that would not be seen with cocaine.Here we focused on the actions of AMPH on RhoA-dependent

internalization of the DAT and explored how these effects on DATtrafficking might contribute to the acute behavioral response to the

drug. However, recently we also demonstrated that a neuronalglutamate transporter, EAAT3, can be internalized in response toAMPH through a process that also appears to require Rho acti-vation (17). This EAAT3 internalization in response to AMPHleads to a potentiation of glutamatergic synaptic responses in do-pamine neurons and reveals a previously undescribed action ofAMPH on glutamatergic signaling. A recent study has shownstriking compartmentalization of glutamatergic and dopaminergicrelease sites within the processes of TH-positive neurons, wherethe two vesicular transporter types, VMAT2 and VGluT2, aresegregated within distinct subcompartments (31). Intriguingly, weobserve Rho activation broadly distributed within the processes ofdopamine neurons. Taken together, these findings suggest a com-plex integration of dopaminergic and glutamatergic transmissionwithin the mesoaccumbens pathway.

Cytoplasmic Target(s) of AMPH. The activation of intracellularsignaling pathways by AMPH and the Rho-mediated internali-zation of DAT are also observed in nonneural cell lines trans-fected with DAT, which demonstrates that these effects do notrequire synaptic vesicles or endogenous dopamine. We alsofound that the presence of AMPH within the cytosol, and not thebinding and transport through DAT, was essential for stimulat-ing these signaling cascades; AMPH transport through NETcould support Rho activation and trigger internalization ofcoexpressed DAT. This observation is consistent with a previousreport showing that a DAT mutant incapable of AMPH trans-port could be internalized when AMPH was directly appliedintracellularly via a whole-cell patch pipette (7).The precise nature and the pharmacological properties of the

cytoplasmic target(s) of AMPH remain to be established, but atrace amine-associated receptor, TAAR1, that is expressed in

Fig. 6. Activation of endogenous Gs-coupled GPCRs prevents AMPH effects in vivo. (A) Western blot analysis of biotinylated tissue from adult male Swiss-Webster mice that were IP-injected with saline or AMPH (2 mg/kg). After 30 min, midbrain tissue was isolated, biotinylated, and assessed for plasmamembrane localization of the DAT. (B) Adult Swiss-Webster mice were IP-injected with saline or the D1/D5 agonist SKF38393 (3 mg/kg) followed 5 min later bya second injection of saline or AMPH (2 mg/kg). Animals were killed after 45 min, and acute midbrain sections were incubated with radiolabeled [125I]RTI-55.AMPH decreased [125I]RTI-55 binding. This effect was blocked by SKF38393 pretreatment. (C) Mice received IP injections of AMPH (2 mg/kg) or cocaine(20 mg/kg) with or without the D5 agonist, SKF38393 (3 mg/kg). Representative traces of animals’ movements in the test arena after injection show com-parable spatial distributions of activity. Variations in trace density reflect differences in total ambulation. (D) Quantification of ambulatory activity for allgroups. Pretreatment with SKF38393 significantly decreases AMPH-induced ambulation, but it does not affect cocaine-induced activity or have a significanteffect on its own **P < 0.0001 by paired t test; *P < 0.05, two-way ANOVA.

Wheeler et al. PNAS | Published online November 9, 2015 | E7145

NEU

ROSC

IENCE

PNASPL

US

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

dopamine neurons and has a predominantly intracellular distri-bution is a potential candidate. As a Gs-coupled GPCR thatresponds to a variety of endogenous and exogenous amines andneurotransmitter metabolites, TAAR1 has been shown to beactivated by AMPH and a variety of AMPH-like compounds andis likely responsible for the increases in cAMP generally ob-served following the application of AMPH to cells (32, 33).Whether TAAR1 also mediates the activation of the smallGTPases, RhoA, and Rac1 remains to be established.Together, the studies reported here indicate that the effects of

AMPH not only depend on the drug’s well-established actions onuptake and efflux through the DAT, but also require the acti-vation of multiple signaling pathways by acting on additionaltarget(s) within the cell. Cytoplasmic cAMP appears to integrateboth intracellular signals through GTPase activation and extra-cellular signals from GPCR-coupled pathways to shape responseof a dopamine neuron to AMPH. Thus, modulation of the Rhoactivation/inactivation sequence provides a mechanism by whichdrugs and endogenous neurotransmitters can influence the re-sponse of dopamine neurons to AMPH.

Materials and MethodsAll procedures were conducted in accordance with the NIH Guide for the Careand Use of Laboratory Animals and with the approval of the InstitutionalAnimal Care and Use Committee of the University of Pittsburgh and theAnimal Care and Use Committee of the NIH.

Cell Culture. SK-N-SH cells were maintained according to the distributor’sdirections (ATCC). Cells were transfected with Lipofectamine2000 (LifeTechnologies) and used 24–48 h later. GFP-C3 was a gift from DanielAltschuler University of Pittsburgh, Pittsburgh; GFP-RhoA was a gift fromChanning Der, University of North Carolina, Chapel Hill, NC (Addgene plas-mid #23224) (34); pcDNA3-EGFP-RhoA-T19N was a gift from Gary Bokoch,The Scripps Research Institute, La Jolla, CA (Addgene plasmid #12967) (35).For primary cultures, midbrains were isolated from E15 Swiss-Webster mice,dissociated, and plated on poly-D-lysine-coated glass coverslips. Astrocytegrowth was minimized by application of AraC at 2 d in vitro. Cells weregrown in MEM supplemented with 5% (vol/vol) FBS and 5% (vol/vol) horseserum for 2–4 wk before use. As assessed by immunolabeling, 5–8% of theneurons expressed both TH and DAT. Transient transfection of neurons wasperformed with NeuroMag (Oz Biosciences) according to the manufacturer’sdirections. Assays were performed on cultures from at least three differentplating procedures under blinded conditions.

Dopamine Uptake. After drug treatments, cultures werewashed and incubatedwith dopamine (100 nM [3H]-dopamine/20 μM cold dopamine) in PBS with1 mM Mg2+ and 100 μM Ca2+ for 10 min. Complete removal of drug pretreat-ments was assessed in cultures treated with cocaine and AMPH (incompletewashout of drugs would be reflected in residual cocaine blocking the transportof [3H]-dopamine). Background was determined by measuring [3H]-dopamineuptake in untransfected cells or DAT-specific inhibition by 10 nM GBR12909.

Microscopy. The cells were kept at 37 °C using an Open Perfusion Micro-Incubator (Harvard Apparatus) and imaged using an Olympus Fluoview 1000confocal microscope equipped with a SIM scanner. TIRF images were ac-quired with an Olympus 1 × 71 equipped with a 60× TIRF objective or anA1 Nikon confocal with a 100× TIRF objective. Images were acquired at 10-sintervals for 1–2 min before addition of treatments. Fluorescence intensitydata of the whole cell or of specific regions of interest were exported toPrism6 and analyzed by fitting to a single exponential decay.

cAMP Accumulation. SK-N-SH cells were labeled with 0.5 μCi of [3H]adeninefor 2 h. The cells were then treated with vehicle or 10 μMAMPH for 5–30 minin the presence of 250 μm of 3-isobutyl-1-methylxanthine (IBMX). The reactionwas terminated by addition of 1 M trichloroacetic acid followed by neutrali-zation with KOH. cAMP was isolated by the two-column method (36) andnormalized to total cellular protein.

GFP-AP-DAT Labeling. Drug treatments were applied to transfected cells for30 min at 37 °C in MEM. Cells were washed and chilled to 4 °C to minimizefurther trafficking events. Surface GFP-AP-DAT was labeled with streptavidin-conjugated Alexa Fluor-568 in 5% (wt/vol) milk for 20 min, rinsed, and fixed

with 4% (wt/vol) paraformaldehyde (PFA). Cells were immunolabeled forTH(+) with an Alexa Fluor-647 secondary antibody. TH(+)/GFP(+) cells wereidentified, and GFP and Alexa Fluor-568 fluorescence intensities were mea-sured with the photon-counting module of Fluoview software. Membraneexpression was determined as a ratio of Alexa Fluor-568/GFP fluorescence,normalized to vehicle-treated controls. Baseline GFP fluorescence indicated nosignificant effect on overall expression of the CMV-promoter–driven constructs.

Acute Brain Slices. Brains from naive or IP-injected adult male Swiss-Webstermice (Charles River) were removed and sectioned coronally into 1-mm slices incold HBSS. Midbrain slices were allowed to recover for 60 min at roomtemperature in artificial CSF (aCSF, in mM: 126 NaCl, 3.5 KCl, 1.3 MgCl2,2 CaCl2, 1.2 NaH2PO4, 25 NaHCO3, 10 glucose) with 50 μM kynurenic acid.The slices were then washed and incubated for 30 min in aCSF bubbled with95% O2/CO2 5% with the indicated treatments.

Radioligand Binding. After treatments, acute slices were washed three timeswith cold HBSS and incubated with 100 pM [125I]RTI-55 for 30 min at 4 °C. Non-DAT binding was determined with GBR12909, 10 nM. RTI binding was correctedfor radioisotope decay and nonspecific binding and normalized to wet weight.

Rho GTPase Activation. GST-conjugated Rho and Rac/CDC42-binding domainsof Rhotekin and p21-activated kinase (PAK1), respectively, were purified fromE. coli and bound to glutathione–Sepharose beads (14, 15). Acute brain slicesor transiently transfected cells were treated with vehicle or AMPH. Lysateswere prepared in binding buffer [50 mM Tris·Cl, pH 7.2, 1% (vol/vol) TritonX-100, 500 mM NaCl, 10 mM MgCl2]. Activated Rho proteins were capturedovernight at 4 °C. The beads were washed, and the protein was eluted withLaemmli’s buffer containing 200 mM DTT, separated on SDS gels, trans-ferred to Immobilon filters, and probed with antibodies to Rho, Rac, orCDC42. Both activated Rac and CDC42 bind to the GST-PAK1 fusion protein,and the two GTPases are differentiated by the primary antibodies used toexamine them by Western blot. Secondary HRP antibodies were used todetect the proteins, activated with Western Lightening Plus ECL chemi-luminescence and imaged on a Chem Doc-It Imaging System. Samples takenbefore GST-fusion purification were used to assess equal protein loading.Data were quantitated with Image J.

Activated Rho Staining of Acute Brain Slices. GST-RBD and GST were purifiedon a glutathione column, biotinylated with EZ-Link NHS-Biotin, 1 mg/mL, anddialyzed in PBS. Acute brain-slice cells were prepared and treated withAMPH and snap-frozen in isopentene. Ten-micrometer sections were madefrom this tissue, blocked, permeablized with 4% (wt/vol) BSA/1% (vol/vol)Triton X-100, and labeled with the biotinylated GST or GST-RBD. Sectionswere then washed and fixed with 4% (wt/vol) PFA. Endogenous peroxidasewas quenched, and streptavidin–HRP was bound to the biotinylated GSTcomplex and developed with a DAB reaction.

Biotinylation of Midbrain Slices. Cell-surface biotinylation was performed asdescribed previously (17) based on the methods of Daniels and Amara (37).Briefly, 39 min after IP of AMPH, 2 mg/kg, male Swiss-Webster mice werekilled and their brains were removed. One-millimeter coronal slices of themidbrain were made and rapidly chilled in ice-cold HBSS for 30 min. Cell-surface proteins were biotinylated with the cell-impermeable reagent 2 mg/mLsulfosuccinimidyl 2-(biotinamido) methyl-1,3-dithiopropionate (sulfo-NHS-SS-biotin) (Pierce) in biotinylation buffer (in mM: 2 CaCl2, 150 NaCl, and 10triethanolamine, pH 7.5) for 30 min. The biotin reagent was quenched witha 20-min glycine wash (100 mM in PBS). The tissue was homogenized in lysisbuffer [1% (vol/vol) Triton X-100, 150 mM NaCl, 5 mM EDTA, and 50 mM Tris,pH 7.5, containing a protease inhibitor mixture (Roche Molecular Biochem-icals)]. Biotinylated proteins were extracted by an overnight incubation withUltralink immobilized NeutrAvidin beads (Pierce) at 4 °C. The beads werewashed with a high-salt wash buffer (0.1% Triton X-100, 500 mM NaCl,5 mM EDTA, and 50 mM Tris, pH 7.5) followed by a no-salt wash buffer(50 mM Tris, pH 7.5). Proteins were analyzed by Western blot. Input samplesthat were set aside before Neutravidin isolation were also analyzed to assureequal loading of proteins as well as to assess for potential degradation ofproteins by these procedures. Intracellular proteins were not biotin-labeledas assessed by an absence of actin staining in the biotinylated fractions.

Behavioral Assessment. The behavioral tests were conducted in the RodentBehavior Analysis Core of the University of Pittsburgh Schools of HealthSciences. Adult male Swiss-Webster mice were maintained on a 12/12 h light-dark cycle with ad libitum access to food and water. Mice were handled daily

E7146 | www.pnas.org/cgi/doi/10.1073/pnas.1511670112 Wheeler et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1

during the light phase for 5 consecutive days. The next day, individual micewere placed into an arena (25 cm × 25 cm × 30 cm) fitted with infrared lightbeams and detectors for tracking of movement in the horizontal plane(Coulbourn Instruments). Each arena was housed in a sound-attenuatingchamber with ambient illumination level of ∼18 lx. After 30 min, mice wereinjected IP with physiological saline (vehicle solution; 3 mL/kg) or SKF38393(3 mg/kg) followed 5 min later with AMPH (2 mg/kg) or cocaine (20 mg/kg).Mice were returned to the test arenas, and their activity was monitored foran additional 60 min.

Statistical Analysis. All data presented as mean and SEM unless otherwisenoted. Statistical significance was calculated with Prism 6 software, typicallyby two-way ANOVAwith Bonferroni posttest or one-way ANOVA with Tukeyposttest unless otherwise noted (*P < 0.05, **P < 0.01, ***P < 0.001).

Drugs and Reagents. All drugs were from Sigma unless otherwise noted. An-tibodies used in this study were from Abcam (Rho, ab40673; RAC1+RAC2,ab68828; CDC42, ab41429; Rho-S188p, ab41435; Sigma TH, T2928) or Chem-icon (DAT, MAB369). Secondary antibodies were from Jackson Immuno-research (HRP conjugates) or Life Technologies (Alexa-Fluor reagents).

ACKNOWLEDGMENTS. We thank S. Watts for his intellectual contributions andM. Miller, M. Larsen, and P. Hullihen for their technical assistance. The behavioraldata were collected at the Rodent Behavior Analysis Core of the University ofPittsburgh Schools of the Health Sciences. This work was supported by AmericanRecovery and Reinvestment Act of 2009 (ARRA)-funded Grants DA07595 (toS.G.A.), F30DK083211 (to D.S.W.), ARRA-funded 1R56DK079864 (to G.R.),and NCRR-UL1RR024153 (to E.T.) and by grants from the Office of the SeniorVice Chancellor, Health Sciences, University of Pittsburgh (to G.R. and E.T.)and the Intramural Research Program of the National Institute of MentalHealth at the NIH (to S.G.A.).

1. Gonzalez Castro F, Barrington EH, Walton MA, Rawson RA (2000) Cocaine andmethamphetamine: Differential addiction rates. Psychol Addict Behav 14(4):390–396.

2. Kalechstein AD, et al. (2000) Psychiatric comorbidity of methamphetamine depen-dence in a forensic sample. J Neuropsychiatry Clin Neurosci 12(4):480–484.

3. Rawson R, et al. (2000) Methamphetamine and cocaine users: Differences in charac-teristics and treatment retention. J Psychoactive Drugs 32(2):233–238.

4. Mahoney JJ, III, Kalechstein AD, De La Garza R, II, Newton TF (2008) Presence andpersistence of psychotic symptoms in cocaine- versus methamphetamine-dependentparticipants. Am J Addict 17(2):83–98.

5. Yamamoto BK, Moszczynska A, Gudelsky GA (2010) Amphetamine toxicities: Classicaland emerging mechanisms. Ann N Y Acad Sci 1187:101–121.

6. Saunders C, et al. (2000) Amphetamine-induced loss of human dopamine transporteractivity: An internalization-dependent and cocaine-sensitive mechanism. Proc NatlAcad Sci USA 97(12):6850–6855.

7. Kahlig KM, et al. (2006) Regulation of dopamine transporter trafficking by intracel-lular amphetamine. Mol Pharmacol 70(2):542–548.

8. Sorkina T, Hoover BR, Zahniser NR, Sorkin A (2005) Constitutive and protein kinaseC-induced internalization of the dopamine transporter is mediated by a clathrin-dependent mechanism. Traffic 6(2):157–170.

9. Daniels GM, Amara SG (1999) Regulated trafficking of the human dopamine trans-porter. Clathrin-mediated internalization and lysosomal degradation in response tophorbol esters. J Biol Chem 274(50):35794–35801.

10. Croisé P, Estay-Ahumada C, Gasman S, Ory S (2014) Rho GTPases, phosphoinositides,and actin: A tripartite framework for efficient vesicular trafficking. Small GTPases 5:e29469.

11. Doherty GJ, McMahon HT (2009) Mechanisms of endocytosis. Annu Rev Biochem 78:857–902.

12. Malcolm KC, Elliott CM, Exton JH (1996) Evidence for Rho-mediated agonist stimu-lation of phospholipase D in rat1 fibroblasts. Effects of Clostridium botulinum C3exoenzyme. J Biol Chem 271(22):13135–13139.

13. Qualmann B, Mellor H (2003) Regulation of endocytic traffic by Rho GTPases. BiochemJ 371(Pt 2):233–241.

14. Ren XD, Kiosses WB, Schwartz MA (1999) Regulation of the small GTP-binding proteinRho by cell adhesion and the cytoskeleton. EMBO J 18(3):578–585.

15. Pellegrin S, Mellor H (2008) Rho GTPase activation assays. Curr Protoc Cell BiolChapter 14:Unit 14.18.

16. Howarth M, Ting AY (2008) Imaging proteins in live mammalian cells with biotin li-gase and monovalent streptavidin. Nat Protoc 3(3):534–545.

17. Underhill SM, et al. (2014) Amphetamine modulates excitatory neurotransmissionthrough endocytosis of the glutamate transporter EAAT3 in dopamine neurons.Neuron 83(2):404–416.

18. Nirenberg MJ, et al. (1997) Immunogold localization of the dopamine transporter: Anultrastructural study of the rat ventral tegmental area. J Neurosci 17(14):5255–5262.

19. Kennedy LA, Zigmond MJ (1979) The behavioral effects of D-amphetamine are cor-related with its effects on cAMP in different brain regions. Brain Res 168(2):408–413.

20. Forget MA, Desrosiers RR, Gingras D, Béliveau R (2002) Phosphorylation states ofCdc42 and RhoA regulate their interactions with Rho GDP dissociation inhibitor andtheir extraction from biological membranes. Biochem J 361(Pt 2):243–254.

21. Ellerbroek SM, Wennerberg K, Burridge K (2003) Serine phosphorylation negativelyregulates RhoA in vivo. J Biol Chem 278(21):19023–19031.

22. Andresen BT, Shome K, Jackson EK, Romero GG (2005) AT2 receptors cross talk withAT1 receptors through a nitric oxide- and RhoA-dependent mechanism resulting indecreased phospholipase D activity. Am J Physiol Renal Physiol 288(4):F763–F770.

23. Sakrikar D, et al. (2012) Attention deficit/hyperactivity disorder-derived coding vari-ation in the dopamine transporter disrupts microdomain targeting and traffickingregulation. J Neurosci 32(16):5385–5397.

24. Bowton E, et al. (2014) SLC6A3 coding variant Ala559Val found in two autism pro-bands alters dopamine transporter function and trafficking. Transl Psychiatry 4:e464.

25. Hayashizaki S, et al. (2013) Methamphetamine increases locomotion and dopaminetransporter activity in dopamine d5 receptor-deficient mice. PLoS One 8(10):e75975.

26. Janssen PA, Leysen JE, Megens AA, Awouters FH (1999) Does phenylethylamine act asan endogenous amphetamine in some patients? Int J Neuropsychopharmacol 2(3):229–240.

27. Negishi M, Katoh H (2002) Rho family GTPases as key regulators for neuronal networkformation. J Biochem 132(2):157–166.

28. Lüscher C, Malenka RC (2011) Drug-evoked synaptic plasticity in addiction: Frommolecular changes to circuit remodeling. Neuron 69(4):650–663.

29. Narita M, Takagi M, Aoki K, Kuzumaki N, Suzuki T (2003) Implication of Rho-associ-ated kinase in the elevation of extracellular dopamine levels and its related behaviorsinduced by methamphetamine in rats. J Neurochem 86(2):273–282.

30. Bosco EE, Mulloy JC, Zheng Y (2009) Rac1 GTPase: A “Rac” of all trades. Cell Mol LifeSci 66(3):370–374.

31. Zhang S, et al. (2015) Dopaminergic and glutamatergic microdomains in a subset ofrodent mesoaccumbens axons. Nat Neurosci 18(3):386–392.

32. Borowsky B, et al. (2001) Trace amines: Identification of a family of mammalian Gprotein-coupled receptors. Proc Natl Acad Sci USA 98(16):8966–8971.

33. Bunzow JR, et al. (2001) Amphetamine, 3,4-methylenedioxymethamphetamine, ly-sergic acid diethylamide, and metabolites of the catecholamine neurotransmitters areagonists of a rat trace amine receptor. Mol Pharmacol 60(6):1181–1188.

34. Roberts PJ, et al. (2008) Rho Family GTPase modification and dependence on CAAXmotif-signaled posttranslational modification. J Biol Chem 283(37):25150–25163.

35. Subauste MC, et al. (2000) Rho family proteins modulate rapid apoptosis induced bycytotoxic T lymphocytes and Fas. J Biol Chem 275(13):9725–9733.

36. Salomon Y, Londos C, Rodbell M (1974) A highly sensitive adenylate cyclase assay.Anal Biochem 58(2):541–548.

37. Daniels GM, Amara SG (1998) Selective labeling of neurotransmitter transporters atthe cell surface. Methods Enzymol 296:307–318.

Wheeler et al. PNAS | Published online November 9, 2015 | E7147

NEU

ROSC

IENCE

PNASPL

US

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 15

, 202

1