The Role of ERp44 in Maturation of Serotonin Transporter Protein*

Background: ERp44 favors maturation of disulfide-linked oligomeric proteins. Results: ERp44 bound to SERT but preferentially to Cys mutants; in ERp44-silenced cells, 5-HT uptake was down-regulated; MTSEA-biotin labeled SERT with a higher affinity, indicating more free Cys on SERT in silenced cells. Conclusion: A disulfide link between Cys-200 and Cys-209 is a prerequisite for SERT oligomerization. Significance: This is the first study showing the involvement of ERp44 in maturation of SERT. In heterologous and endogenous expression systems, we studied the role of ERp44 and its complex partner endoplasmic reticulum (ER) oxidase 1-α (Ero1-Lα) in mechanisms regulating disulfide bond formation for serotonin transporter (SERT), an oligomeric glycoprotein. ERp44 is an ER lumenal chaperone protein that favors the maturation of disulfide-linked oligomeric proteins. ERp44 plays a critical role in the release of proteins from the ER via binding to Ero1-Lα. Mutation in the thioredoxin-like domain hampers the association of ERp44C29S with SERT, which has three Cys residues (Cys-200, Cys-209, and Cys-109) on the second external loop. We further explored the role of the protein chaperones through shRNA knockdown experiments for ERp44 and Ero1-Lα. Those efforts resulted in increased SERT localization to the plasma membrane but decreased serotonin (5-HT) uptake rates, indicating the importance of the ERp44 retention mechanism in the proper maturation of SERT proteins. These data were strongly supported with the data received from the N-biotinylaminoethyl methanethiosulfonate (MTSEA-biotin) labeling of SERT on ERp44 shRNA cells. MTSEA-biotin only interacts with the free Cys residues from the external phase of the plasma membrane. Interestingly, it appears that Cys-200 and Cys-209 of SERT in ERp44-silenced cells are accessible to labeling by MTSEA-biotin. However, in the control cells, these Cys residues are occupied and produced less labeling with MTSEA-biotin. Furthermore, ERp44 preferentially associated with SERT mutants (C200S, C209S, and C109A) when compared with wild type. These interactions with the chaperone may reflect the inability of Cys-200 and Cys-209 SERT mutants to form a disulfide bond and self-association as evidenced by immunoprecipitation assays. Based on these collective findings, we hypothesize that ERp44 together with Ero1-Lα plays an important role in disulfide formation of SERT, which may be a prerequisite step for the assembly of SERT molecules in oligomeric form.

The serotonin transporter (SERT 2 ; SLC64A) is a member of the Na ϩ -and Cl Ϫ -dependent monoamine transporter family, which includes the dopamine transporter and the norepinephrine transporter (1). These neurotransmitter transporters share extensive sequence homology (1)(2)(3)(4) with several common structural characteristics, including oligomeric properties (5-7), multiple sites for N-linked glycosylation (8 -10), and the cysteine residues connected by a disulfide bond on the second extracellular loop (EL2) (1,3,11,12). Proper post-translational modifications are essential regulatory factors in neurotransmitter uptake functions of SERT (9 -11), norepinephrine transporter (8), and dopamine transporter (6,12) and occur in a host-dependent fashion (13). Following these modifications transporters adopt more stable, lower energy conformations (14 -17) to initiate for their correct folding and assembly. For SERT, these alterations of its structure impact the extracellular uptake function of serotonin (5-HT) and hence its biological role in neurons and peripheral tissues (1)(2)(3). Consequently, identifying the mechanisms regulating post-translational modifications would advance our understanding of the importance of SERT conformations and oligomerization in biological processes.
Conformational maturation of monoamine transporter family members involves different pathways for post-translational modifications of the respective proteins (18 -21). Formation of intra-or intermolecular disulfide bonds is one of the major rate-limiting factors. SERT has three Cys residues on the second external loop. Disulfide bond formation between two residues, Cys-200 and Cys-209, is required for SERT folding, surface expression, and transport activity (11). Disruption of the disulfide bond by a single mutation of Cys-200 or Cys-209 produces mutants with a terminally exposed thiol that are retained intracellularly (11,22). However, mutation of both Cys in the pair allows SERT to reach the plasma membrane but compromises transport activity (22). Previous studies highlighting the importance of a disulfide bond for SERT folding and surface expression suggest a thiol-dependent quality control mechanism in SERT maturation (11,22). Moreover, the modification of exposed thiols with ␤-mercaptoethanol (MSH) altered the density of SERT molecules on the plasma membrane and the 5-HT uptake rates (10). Pretreatment of the cells with MSH led to much lower 5-HT uptake rates when compared with control cells, although MSH treatment released more SERT molecules to the plasma membrane. The association between SERT molecules also was obstructed by MSH treatment (10). Altogether, these findings indicate the involvement of SERT in a thiol-mediated retention mechanism. However, neither the mediators aiding in disulfide bond formation nor the role of disulfide bonds in SERT maturation has been vigorously studied yet.
To explore factors involved in SERT maturation, we examined whether the SERT maturation pathway utilizes ERp44 and ER oxidase 1-␣ (Ero1-L␣) because of their roles in the maturation and quality control of disulfide-containing oligomeric proteins (15,19,(23)(24)(25)(26)(27)(28)(29)). An exposed thiol for a protein localized to the ER favors formation of mixed disulfide bonds with thioredoxin family members, notably ERp44 (16,(23)(24)(25)(26)(27). During this process, ERp44 preferentially associates with unassembled subunits of disulfide-containing oligomeric proteins (15,26,27). In this study, we employed biochemical and molecular biological techniques using endogenous and heterologous expression systems to assess an association between ERp44 and SERT, the role of Ero1-L␣ in the maturation process, and the corresponding impacts on SERT localization and function at the plasma membrane. ERp44-silenced cells led to increased levels of SERT at the cell surface. The functional state of SERT was also compromised based on changes in the mechanism and maximal rate of 5-HT uptake by these cells. Moreover, SERT mutants (C200S, C209S, and C109A) with compromised disulfide bond formation and hence structure and function were shown to preferentially associate with ERp44. Taken together, these studies suggest that (i) disulfide bond formation may present a critical step in SERT folding to a fully active form, and (ii) SERT utilizes ERp44 as well as its partner Ero1-L␣, an oxidoreductase, in the disulfide bond formation process.

MATERIALS AND METHODS
JAR cells were provided by the American Type Culture Collection (Manassas, VA). Protein A-Sepharose beads and nonimmune rabbit serum were purchased from Zymed Laboratories Inc. (South San Francisco, CA). 3 H-Labeled 5-HT was purchased from PerkinElmer Life Sciences. HA-tagged and untagged forms of ERp44 and the Ero1-L␣ construct were a generous gift from Dr. Sitia Roberto (Salute San Raffaele, Milan, Italy). Lentiviral small hairpin RNA (shRNA) plasmid, anti-ERp44, and Ero1-L␣ antibodies were generous gifts from Dr. Scherer at the University of Texas Southwestern and used by Wang et al. (30). The second-generation packaging plasmid, psPAX2, and VSV-G were purchased from Addgene Inc. (Cambridge, MA). Expression vectors, cell culture materials, Lipofectin, and Lipofectamine 2000 were purchased from Invitrogen. ERp44 and Ero1-L␣ antibody (Ab) were purchased from Cell Signaling Technology (Beverly, MA). NHS-SS-biotin, the Micro BCA protein assay reagent kit, and Pico-West Supersignal ECL substrate were purchased from Pierce. Scintillation mixture was purchased from Fisher. A monoclonal SERT Ab recognizing amino acid residues 51-66 on the N terminus was purchased from Mab Technologies (Stone Mountain, GA).
Transporters with both glycosylation sites mutated to glutamine, QQ (N208Q and N17Q), were constructed utilizing a Stratagene QuikChange XL site-directed mutagenesis kit as described previously (7,10).
The three Cys residue (C109A, C200S, and C209S) mutations were introduced by site-directed mutagenesis using oligonucleotides 5Ј-CTT CCC CTA CAT AGC TTA CCA GAA TGG AG-3Ј, 5Ј-CTG CCC TGG ACC AGC TCC AAG AAC TCC TGG AAC AC-3Ј, and 5Ј-CCT GGA ACA CTG GCA ACT CCA CCA ATT ACT TCT CCG AG-3Ј, respectively, on SERT and the FLAG-and Myc-tagged forms of SERT. Using the same primers, the double mutant was generated.
We confirmed the subcloning processes by sequencing the genes at the University of Arkansas for Medical Sciences DNA Sequencing Facility. In addition, mutants with Cys-200 mutated to serine were prepared using the same method, and mutations were confirmed by sequencing.
These mutants were expressed in JAR cells by using the vaccinia-T7 transient expression system as described (10). Transfected cells were incubated for 16 -20 h at 37°C before they were used for transport or IP experiments. Protein concentration was obtained by means of the Micro BCA protein assay reagent kit (Pierce).
5-HT Uptake Assay-Before seeding the cells, a 24-well plate was coated with poly-D-lysine (0.1-0.5 mg/ml in sterile water) for 30 min and washed three times with sterile water. JAR cells were seeded 36 -48 h in a polylysine-coated 24-well plate prior to initiating the transport assay. Uptake assays were performed by incubation of cells (2 ϫ 10 5 cells/assay) in 20.5 nM [1,2-3 H]5-HT (3400 cpm/pmol) in PBS/CM (phosphate-buffered saline, 0.1 mM CaCl 2 , and 1 mM MgCl 2 ). The intact cells were washed quickly with ice-cold PBS to stop the activity, harvested in 2% SDS in PBS, and transferred to scintillation vials containing 5 ml of scintillation mixture, and the radioactivity was determined in a Beckman scintillation counter. An equal number of cells per cell line was confirmed by cell counting with a hemocytometer, and a group of cells was treated with a high affinity cocaine analog, 0.1 M 2␤-carbomethoxy-3-tropane, to monitor 5-HT influx in the background (2␤-carbomethoxy-3tropane was provided by the National Institute of Mental Health) (10).
The resulting data were fit to equations for two different models describing the relationship between the uptake rate and 5-HT concentrations. The traditional model describes a hyperbolic kinetic profile in which the uptake rate reflects contributions from a single transporter at a constant concentration and the transporter binds 5-HT in a 1:1 stoichiometry. When these conditions are not satisfied, the kinetic profile may deviate from a simple hyperbola, and thus we also fit data to the Hill equation describing cooperative effects of 5-HT concentrations on the uptake rate. Equation 1 depicts the Hill equation such that is the observed uptake rate, V max is the maximal uptake rate, K h is  at the midpoint of the curve, and n is the Hill coefficient or measure of cooperativity. When the Hill coefficient is 1.0, the equation reduces down to the one used for fitting the data to the traditional transport model. The fits of the data to these kinetic models were compared, and the most probable one was identified by the Akaike Information Criterion using GraphPad Prism software (San Diego, CA).
Preparation of shRNA Lentiviral Particles-Lentiviral shRNA expression constructs were created utilizing a second-generation packaging plasmid, psPAX2, obtained from Addgene. HEK-293FT cells were cultured for 48 h and transiently transfected with the respective shRNA construct graciously provided by Dr. Scherer (University of Texas Southwestern), psPAX2, and VSV-G in a 1:2 ratio of Lipofectamine 2000 to Opti-MEM medium. The medium was replaced 12 h later with full DMEM. After 48 h post-transfection, the medium was collected, filtered with a 0.45-m PDVF filter, and precipitated in PEG 8000 Na ϩ /Cl Ϫ solution for 20 h. The lentiviral particles were collected with centrifugation at 4°C and resuspended in PBS. Viral particles were stored at Ϫ80°C.
Transfection with Lentiviral Expression Constructs-JAR cells were grown to 60% confluence for 24 h in full medium. Cells were transfected with lentiviral constructs in full medium with 5 g/ml Polybrene and 1:1000 dilution with respective viral particles for 16 h. The medium was replaced the following day, and the cells were allowed to divide. The cells were sorted for positive GFP expression on a FACS Aria at the University of Arkansas for Medical Sciences Core Flow Cytometry Facility 72 h post-transfection, and the positive cells were cultured in full medium for further experiments. These cells (2 ϫ 10 5 cells/ assay) were used in the assays or transiently transfected with the three constructs, SERT, QQ, and C200S, under a CMV promoter as described previously (5,7).
WB Analysis-Cells (2 ϫ 10 5 cells/assay) were solubilized in PBS containing 0.44% SDS, 1 mM phenylmethylsulfonyl fluoride (PMSF), and protease inhibitor mixture, which contained 5 mg/ml pepstatin, 5 mg/ml leupeptin, and 5 mg/ml aprotinin. In lysis buffer, the alkylating agent N-ethylmaleimide at a final concentration of 5 mM was added to prevent oxidation and formation of nonspecific disulfide bonds during lysis and to retain the native monomeric structures in the gel (7). Samples were analyzed by 10% SDS-PAGE and transferred to nitrocellulose membrane. WB analysis was performed with anti-ERp44 (diluted 1:1000), anti-SERT (diluted 1:500), or anti-Ero1-L␣ (diluted 1:1000) Ab and then with horseradish peroxidase-conjugated anti-rabbit secondary Ab (diluted 1:7500), respectively. The signals were visualized using the ECL WB detection system. Blots were visualized using the VersaDoc 1000 gel visualization and analysis system and three separate readings in order to calculate the average intensity (Bio-Rad) (10). Statistical analysis of densitometric scans was done using a two-sided t test.
Cell Surface Biotinylation-The surface expression of the SERT and the mutant forms were monitored by biotinylation as described (5). In brief, cells (2 ϫ 10 5 cells/assay) were treated with the membrane-impermeant biotinylating reagent NHS-SS-biotin (Pierce) or 1 mM MTSEA-biotin (Toronto Research Chemicals, Downsview, Canada). These reagents selectively modify only the external lysine residues on membrane proteins. The unbound reagent was washed away and quenched with glycine (in the case of NHS-SS-biotin), and the cells were lysed in Tris-buffered saline containing 1% SDS, 1% Triton X-100, and protease inhibitor mixture/PMSF. The biotinylated proteins were recovered with streptavidin-agarose beads (Pierce). Quenching unreacted MTSEA-biotin was not required because of its rapid hydrolysis in aqueous buffers. The labeled proteins were resolved by 8% SDS-PAGE, were transferred to nitrocellulose, and were detected with anti-SERT, anti-FLAG, or anti-Myc antibodies as described (5,7,10). NHS-SS-biotin contains a sulfonic acid moiety with a fixed negative charge and was shown to selectively modify lysine amino groups exposed on the cell surface, and MTSEA-biotin was shown to selectively modify external Cys sulfhydryl groups under the conditions used (5, 10). The primary Ab was detected using horseradish peroxidase-conjugated secondary Ab and the ECL detection system.
Immunoprecipitation-JAR cells were used as an endogenous expression system to assay SERT, ERp44, and Ero1-L␣ association. Before collection, cells were treated with 10 mM N-ethylmaleimide for 30 min. Cells were lysed in IP buffer (55 mM triethylamine (pH 7.5), 111 mM NaCl, 2.2 mM EDTA, 0.44% SDS, 1% Triton X-100, 1 mM PMSF/protease inhibitor mixture) containing 5 mM N-ethylmaleimide (1,2). Cell lysates were first precleared by incubation with protein A beads. The precleared lysate was treated with primary Ab-coated protein A beads (polyclonal HA Ab in the heterologous and polyclonal SERT Ab in the endogenous expression system) overnight at 4°C. The lysate immune complexes were recovered by brief centrifugation in a bench top microcentrifuge (Beckman), washed several times with high and low salt IP buffers, and eluted in Laemmli sample buffer (50 mM Tris-HCl (pH 6.8), 2% SDS, 0.1% bromphenol blue, 10% glycerol, and 1% mercaptoethanol). Samples were separated on a 10% SDS-polyacrylamide gel. After electrophoresis, gels were analyzed by WB with either horseradish peroxidase-conjugated monoclonal FLAG Ab (diluted 1:750) or polyclonal anti-ERp44 or -Ero1-L␣-Ab (diluted 1:1000). The signals were developed with the ECL detection system.
Data Analysis-Analysis of variance was performed to assess group differences for 5-HT end points, and comparisons between groups were made using two-sided t tests based on the analysis of variance mean squared error. Data are presented as mean and S.D. of multiple experiments.

RESULTS
In order to evaluate ER thiol-mediated SERT retention at the plasma membrane and its role in SERT function, human placental JAR cells were pretreated with MSH at different concentrations, as it was previously studied for IgG (26). The surface expression of SERT in MSH-pretreated JAR cells increased in parallel with MSH concentrations (Fig. 1A), whereas their 5-HT uptake rates were significantly lower than the control cells (Fig. 1B). The results indicate the involvement of the disulfide bond formation in the maturation process of SERT in ER.  (5,7,10). Biotinylated PM proteins were resolved on SDS-PAGE followed by WB analysis using a polyclonal SERT and actin Ab. The level of actin in each lane shows the cell population in each group. B, the 5-HT uptake rates of MSH-treated JAR cells were measured as described previously (5,7,10). C, WB analysis of endogenous SERT, Ero1-La, and ERp44 expressions in JAR cells. D, the association between SERT and ERp44 or Ero1-L␣ was determined in JAR cells. JAR cell lysates were prepared and subjected to IP in the presence (ϩAb) or absence (ϪAb) of monoclonal anti-SERT Ab, as described under "Materials and Methods." The IP and corresponding supernatants (SERT-Ab-depleted cell lysate) were subjected to WB with polyclonal anti-ERp44, -Ero1-L␣, -SERT, and -actin Abs as indicated. A representative of three separate experiments is shown. All lanes contain protein recovered from the same number of cells equivalent to 30% of one well from a confluent 24-well dish. Three wells of each condition were pooled, and an aliquot of this mixture was run on the gel. E, the densitometric scanning of immunoblots in D shows the level of association between SERT-ERp44 and SERT-Ero1-L␣. Data with error bars are represented as mean Ϯ S.D. for triplicate samples.
ERp44 and Ero1-L␣ Interact with Endogenous SERT-JAR cells endogenously express SERT and have proven to be very useful in studies relating to the regulatory aspects of the placental transporter. We took advantage of this system in monitoring the processing of SERT. The JAR cell line in the transient transfection system offers an in situ environment in processing of the newly synthesized SERT. As reported previously (10,31), transient transfection produced an at least 50-fold higher level of protein than the endogenous system. Therefore, using JAR cells only mimics the endogenous condition for the transiently transfected plasmids. The endogenous expression of SERT, ERp44, and Ero1-L␣ were confirmed in JAR cells through WB assays with the corresponding Abs (Fig. 1C).
Associations between SERT and these chaperones were then demonstrated by IP assays. Cellular proteins pulled down by monoclonal SERT-Ab were analyzed with WB assays using anti-ERp44 or Ero1-L␣ polyclonal Abs (Fig. 1D). Densitometric quantification of the levels of ERp44 and Ero1-L␣ proteins on SERT-Ab showed that a significant level of chaperone proteins were bound to SERT-Ab compared with the bands when the primary Ab was omitted (Fig. 1E, gray filled bars). Although these results suggest the presence of an association between these ER proteins and SERT in the endogenous expression system, we further examined the specificity of our IP assay conditions by analyzing the levels of ERp44, Ero1-L␣, and SERT in SERT-Ab-depleted cell lysate (Fig. 1D). In SERT-Ab-depleted cell lysate, the levels of these three proteins appeared significantly less than their expression levels in the whole cell lysate (SERT-Ab-excluded IP). The intensities of the bands were quantified as representative of their expression level (Fig. 1E, black filled bars). These data show that 72.8% of ERp44 and 45.8% of EroL-1␣ of the whole cell were bound to SERT-Ab (gray bars versus black blocks in the graph). Actin was used as a loading control for protein lysate and was similar between lysate-loaded lanes (Fig. 1D).
Characterization of SERT in ERp44 shRNA or Ero1-L␣ shRNA Cells-The functional role of the ERp44 and Ero1-L␣ interaction with SERT was analyzed in JAR cells stably infected with shRNA for ERp44 or Ero1-L␣ as described under "Materials and Methods." We confirmed 60 and 90% reduction of ERp44 (lane 2) or Ero1-L␣ (lane 3) in cell lysates from their respective shRNA-expressing cells compared with their expressions in control JAR cells (Fig. 2A, lane 1). ERp44 is the primary ER localization mechanism of Ero1-L␣; in ERp44 shRNA-expressing cells, we observed a 30% reduction in Ero1-L␣ level (Fig. 2).
SERT Expression and Transport Activity in shRNA-expressing Cells-In order to evaluate the role of interactions between SERT and two ER chaperones, ERp44 and Ero1-L␣ we measured the plasma membrane and the whole cell expression of SERT in ERp44 or Ero1-L␣ shRNA-expressing JAR cells first and then compared the 5-HT uptake abilities of these cells. WB analysis of ERp44-and Ero1-L␣-silenced JAR cell lysates showed no difference in the total cellular expression of SERT compared with the control (scrambled plasmid) or JAR cells (Fig. 3A).
The density of SERT molecules on the surface of ERp44 or Ero1-L␣ shRNA-transfected JAR cells was measured by a cell surface biotinylation assay as described under "Materials and Methods." We observed a 50 and 100% increase in SERT expression on the plasma membrane in ERp44 and Ero1-L␣ shRNA-expressing cells, respectively (Fig. 3A). The quantification of the levels of SERT on the plasma membrane of these cells determined a 75 and 100% increase in SERT expression. Despite the increase in the cell surface of SERT expression in ERp44-or Ero1-L␣-silenced JAR cells, there was an unexpected greater than 50% reduction in 5-HT uptake rates (Fig. 3B).
SERT-ERp44 Protein Associations in Ero1-L␣ shRNA-expressing JAR Cells-SERT-ERp44 and SERT-Ero1-L␣ associations were tested with IP assays in Ero1-L␣and ERp44-shRNA expressing JAR cells, respectively. Detergent-soluble shRNAinfected JAR cell lysate was incubated with monoclonal SERT Ab-coated protein A beads. SERT Ab-bound cellular proteins were eluted and resolved on SDS-PAGE followed by WB analysis with polyclonal Abs recognizing ERp44 or Ero1-L␣ (Fig.  4A). Although the association between ERp44 and SERT was elevated by 100% in Ero1-L␣-silenced JAR cells compared with the control cell line (lane 2 versus lane 1 in the top blot), there was a 50% reduction in the level of association between Ero1-L␣ and SERT in ERp44-silenced JAR cells (Fig. 4A, lane 3 versus  lane 1 of the middle blot). The expression levels of SERT proteins in whole JAR cells did not show a difference from that in shRNA-transfected JAR cells (Fig. 4A, bottom blot).
Loss of ERp44 Compromises 5-HT Transport-We compared 5-HT transport by SERT expressed in JAR cells whether ERp44 was present or silenced (Fig. 5). When ERp44 was present, the kinetic profile for 5-HT uptake was qualitatively more sigmoidal (or cooperative) than hyperbolic (Fig. 5). A comparative Loss of ERp44 significantly altered the kinetic profile for 5-HT transport from a sigmoidal curve to an almost hyperbolic one. The fit of the data to the Hill equation was still favored but not as strong as observed previously (60% versus 99.9%). The source of this shift was the almost complete abolishment of cooperativity, as reflected in a Hill coefficient of 1.25 Ϯ 0.18. This value is close to 1.0, which corresponds to the traditional model. Despite an increase in SERT localization to the plasma membrane (Fig. 3A), the absence of ERp44 decreased V max to 0.52 Ϯ 0.04 (pmol/min/mg protein) and had no effect on the affinity of 5-HT for SERT (K h of 2.1 Ϯ 0.3 M). Thus, analysis of the data demonstrated that the loss of ERp44 activity impacted (i) the mechanism of transport and (ii) the maximal uptake rate for 5-HT.
ERp44 Preferentially Associates with SERT Mutants-SERT has three Cys residues, Cys-109, Cys-200, and Cys-209, on extracellular loops 1 and 2 (11). Here the Cys residues were changed one at a time, and a FLAG epitope was introduced into each mutant protein. We next investigated the association between ERp44 and FLAG epitope-tagged SERT mutants. Additionally, the form of the transporter with both N-glycosylation sites mutated, SERT-QQ, which is unable to be glycosylated and oligomerized (10), was included in these studies. JAR cells were transfected for 24 h with HA-ERp44 cDNA in a 1:1 ratio with FLAG-SERT, FLAG-QQ, FLAG-C200S, or FLAG-C109A, as described under "Materials and Methods" (2). The IP analysis showed a 50 and 100% increase in association between ERp44 and SERT mutant forms, -QQ (which is unable to be glycosylated and oligomerized) (lane 1) and -C200S (which cannot form a disulfide bridge) (lane 2), respectively, compared with the association between SERT and ERp44 (lane 4) (Fig. 6).
Next, the interaction between ERp44-C29S and SERT was tested in the same heterologous expression system. As reported previously, ERp44-C29S cannot interact with Ero1-L␣ (27,(32)(33)(34), The absence of a complex between ERp44 and Ero1-L␣ compromises the ability of ERp44 to interact with SERT as shown by the lack of SERT pulled down with ERp44-C29S (Fig.  6A, lane 3). The association between FLAG-SERT and HA-ERp44 was observed with less affinity than the SERT mutant interactions with ERp44 (lane 4).

Functional Role for Three Cys Residues on Second Extracellular Loop of SERT-Previous studies using a Cys-specific membrane-impermeable form of methanethiosulfonate (MTS)
reagents showed the involvement of Cys-200 and Cys-209 residues in the 5-HT uptake process (1,2). In this study, cells expressing these mutants and the wild-type SERT construct were assayed for 5-HT transport activity. Fig. 7 shows that 5-HT uptake rates of ERp44-silenced JAR cells were 3-fold lower than in physiological JAR cells. C109A is resistant to the MTS reagents as described previously (5). The 5-HT uptake rate of C109A is slightly lower than that observed for wild-type SERT. Expressing C109A in ERp44-silenced JAR cells reduced the 5-HT uptake rates to 3-fold the rate in JAR cells. Compared with the wild-type transporter, the relative 5-HT uptake rates for C200S, C209S, and the double mutant (C200S/C209S) were 18.76, 3.38, and 9.66%, respectively.
Glycosylation is an important prerequisite for the self-association of transporter protein (10), and the SERT monomers cannot be fully functional because they are in the oligomeric form (5). In relating these findings together with the kinetic characteristics of SERT in ERp44-silenced cells, next the role of Cys residues in the self-association ability of transporter protein was studied.
Self-association Abilities of C200S, C209S, and C109A and Role of ERp44 in This Process-We tested the self-association ability of SERT mutants as a measure of the possible role of the disulfide-bridge in SERT oligomerization.
JAR cells transfected with a 1:1 mixture of Myc-or FLAGtagged mutant plasmids and either FLAG-or Myc-tagged proteins were precipitated from the mixture by using protein A beads coated with polyclonal FLAG or monoclonal Myc Ab. FLAG-SERT or Myc-SERT and any associated proteins eluted from the beads were analyzed in WB assays. The FLAG-SERTassociated proteins were probed with monoclonal Myc Ab; Myc-SERT and associated proteins were blotted with polyclonal FLAG Ab.  (error bars)). The whole cell expression of SERT in these cell lines was also analyzed with WB for SERT and actin Ab as indicated B, 5-HT uptake rates of these four cells were measured as described previously (5,7,10). The asterisks indicate that 5-HT uptake rates of ERp44-or Ero1-L␣-silenced JAR cells are significantly different from the uptake rates of JAR cells (two-sided t test, p Ͻ 0.05). Results from three independent experiments are shown (mean Ϯ S.D.).
The self-association abilities of SERT molecules were shown under both IP assay settings, with either FLAG or Myc Abcoated protein A beads (Fig. 8, A or B, respectively); Myc-SERT was found in association with FLAG-SERT, indicating that although the two forms of Myc-SERT/FLAG-SERT and FLAG-C109A/Myc-SERT remained associated after detergent disruption of the cells, the two other Cys mutants, C200S and C209S, associated neither by themselves nor with wild-type SERT.
Surface Labeling of Cells with MTSEA-Biotin-When SERT is present at the cell surface, there are three exposed Cys residues that are susceptible to modification and thus can be interrogated to assess the presence of disulfide bonds. MTSEA-bio-tin in the external medium reacts with free Cys residues on SERT and the mutants containing exposed Cys residues (1,2,15). JAR and ERp44-silenced JAR cells were transfected with SERT or one of the Cys mutants (C109A, C200S, and C209S). The next day, cells were treated with MTSEA-biotin as described previously (5); after removing unreacted reagent, the cells were lysed in detergent, and Streptavidin beads were used to precipitate biotinylated surface proteins. Fig. 9A shows that JAR-C109A cells were not labeled with MTSEA-biotin at all, indicating the absence of free Cys residue on the external loop of the C109A form; however, JAR-SERT cells were labeled at a level 50% less than the labeling of JAR-C200S or JAR-C209S cells.
However, MTSEA-biotin labeling of ERp44-silenced JAR cells expressing SERT or one of the Cys mutant forms of SERT showed a different pattern. The level of MTSEA-biotin labeling of the JAR cells expressing C109A, C200S, and C209S was almost 50% of the labeling level of SERT in JAR cells.
These data show that in JAR cells, SERT must have only one Cys that can interact with MTSEA-biotin, but in ERp44-silenced cells, it has more free Cys residues to be labeled by MTSEA-biotin. In a similar way, in JAR cells, C109A should not have free Cys residues, but in ERp44-silenced JAR cells, C109A must have two free Cys residues.
Next, we wanted to verify these findings in an endogenous expression system. JAR, ERp44 or Ero1-L␣ silenced JAR cells were treated with MTSEA-biotin. Biotinylated proteins were pulled down and analyzed using anti-SERT-Ab. The level of biotin labeling of JAR cells was much less than that in ERp44-or Ero1-L␣-silenced JAR cells (Fig. 9B).

DISCUSSION
Despite a wealth of knowledge on key amino acid residues needed for SERT activity, there are limited data on the protein  mediators and quality control checkpoints in SERT maturation. Studies have shown that an exposed thiol at Cys-200 or Cys-209 on EL2 is sufficient for the intracellular retention of SERT, but SERT mutants without Cys residues on the second extracellular loop are able to reach the PM despite the lack of a disulfide bond (11). These studies suggest a quality control mechanism involved in SERT maturation, which recognizes exposed Cys in SERT molecules and retains them intracellularly. The ability of Cys mutants of SERT to reach the PM further implies that the quality control mechanism does not recognize non-native structure, such as hydrophobic patches or immature glycans, but rather, the retention of Cys mutants of SERT is entirely thiol-dependent.
The retention of proteins through reversible disulfide bonds has become the primary feature of thiol-mediated retention involving two ER resident proteins, ERp44 and Ero1-L␣ (15,26,27,30). ERp44 and Ero1-L␣ have been described as a quality control mechanism in the maturation of disulfide-containing oligomeric proteins (16,26,27,30). Given the importance of disulfide bond formation in the maturation of SERT and the nature of thiol-mediated retention in ER, herein, we tested a novel hypothesis of ERp44-mediated disulfide bond formation and the role of disulfide in oligomerization of SERT monomers.
In initial experiments, we were able to demonstrate that endogenous SERT interacts with ERp44 and Ero1-L␣ (Fig. 1C). ERp44 mediates retention of cargo proteins at the ER though a thiol-dependent process (15,26,27,30,32); thus, we examined the effects of a terminally reactive thiol on the association between SERT and ERp44 as a step in SERT maturation at the ER. A single mutation of Cys-200 or Cys-209 disrupts disulfide bond formation and produces an exposed Cys in SERT (11). In addition, C200S or C209S mutants of SERT are not able to reach the PM and are retained intracellularly (11). A 50% increase in association between a single Cys mutant of SERT and ERp44 demonstrates that an exposed Cys at Cys-209 is  Seventy-two hours postinfection, JAR cells were sorted for GFP expression at the FACS Flow Cytometry Facility, and equal numbers of GFP-positive cells and JAR cells (2 ϫ 10 5 cells/assay) were seeded. The next day, they were transfected with SERT, C109A, C200S, C209S, and double mutant C200S/ C209S plasmids. 5-HT uptake rates of these cell lines were measured as described previously (5,7,10). The rate of uptake is expressed as the mean and S.D. values (error bars) of triplicate determinations from three independent experiments. * and **, results of Student's t test with both p Ͻ 0.001 (compared with mutant versus wild-type SERT and SERT in JAR versus in ERp44-silenced JAR uptake rates, respectively). sufficient for the retention of SERT (Fig. 6A), and furthermore, ERp44 may be involved in the quality control of disulfide bond formation.
ERp44 also preferentially interacts with unassembled subunits of oligomeric proteins (27). This association enables ERp44 to selectively retain cargo proteins, like SERT, in an environment suitable for their maturation and concentrate subunits in a local environment to promote their polymerization (15,27,30). SERT mutants (e.g. QQ) that cannot undergo glycosylation demonstrate compromised oligomerizations and are unable to properly transport 5-HT (10); here, this mutant form of SERT was included in our studies to investigate if SERT undergoes disulfide bond-mediated oligomerization process. Data showed that if SERT molecules cannot associate in an oligomeric form, then this facilitates the association between ERp44 and SERT (Fig. 6).
Because ERp44 associates with cargo proteins through exposed thiols, our findings suggest that QQ mutants may not be able to form a disulfide bond, and further, glycosylation of SERT may contribute to its disulfide modification. Although not previously reported for SERT, glycosylation has been shown to facilitate disulfide bond formation in other proteins, such as epidermal growth factors and human insulin receptor (33,34). This result implies the requirement of disulfide bond formation in functional oligomerization of SERT monomers. Sequential modification would allow the stepwise maturation of SERT, and quality control checkpoints intimately associated with the folding process ensure that unmodified or misfolded proteins do not proceed to the next stage in maturation (15). In this manner, the ER folding machinery is able to couple the maturation and quality control of proteins within the secretory pathway (16,23). However, additional studies are needed in order to confirm with certainty the interdependence among post-translational modifications of SERT.
Once associated with ERp44, previous data have shown that cargo molecules are displaced by Ero1-␣ through binding with Cys-29 of ERp44 (36). Interactions with Cys-29 have also been shown to facilitate the association between ERp44 and other cargo proteins, such as adiponectin and IgM subunits (27,30). However, using a mutant form of ERp44, C29S, we showed that SERT does not interact with ERp44 through Cys-29, but rather disruption of interactions through Cys-29 through mutations increases its association with the transporter. Considering the role of Ero1-L␣ in displacement of other cargo proteins from ERp44, these findings suggest that Ero1-L␣ is required to interact with ERp44 for efficient release of SERT, and the relative ratio of the two chaperones, ERp44 and Ero1-L␣, to each other as well as SERT is a crucial determinant for proper SERT maturation.
Our co-IP assays suggested a possible association between ERp44 and SERT; however, studying protein-protein interactions in co-IP experiments with detergent-disturbed cell lysate suffers from drawbacks, such as possible nonspecific interactions and variations in the efficiency of the IP. Therefore, we performed an analysis of these associations on 5-HT uptake function of SERT proteins. The kinetic studies along with the MTSE-biotinylation assays strongly indicate that ERp44 and Ero1-L␣ contribute to the maturation of SERT. Consequently, FIGURE 9. MTSEA-biotin labeling of SERT and Cys mutants. A, JAR cells were infected with shRNA for ERp44 or Ero1-L␣. Seventy-two hours postinfection, JAR cells were sorted for GFP expression at the FACS Flow Cytometry Facility, and equal numbers of GFP-positive cells and JAR cells (2 ϫ 10 5 cells/assay) were seeded in 24-well plates. The next day, cells were transfected with SERT, C109A, C200S, and C209S. Twenty-four hours post-transfection, intact cells were treated with MTSEA-biotin (see "Materials and Methods" and Refs. 7 and 10). After removing unreacted reagent, the cells were lysed in detergent, and Streptavidin beads were used to precipitate biotinylated surface proteins. The biotinylated proteins were eluted and separated by SDS-PAGE, and WB analyses were performed with SERT Ab. Blots were stripped and reprobed with actin Ab. Densitometric scanning of immunoblots was calculated as the percentage of labeling of SERT in JAR or in ERp44-silenced JAR. Results from three different experiments are shown (mean Ϯ S.D. (error bars)). B, JAR cells expressing shRNA for ERp44 and Ero1-L␣ were prepared as described under "Materials and Methods." As described above, intact cells were labeled with MTSEA-biotin, and WB analysis was performed with SERT-Ab. The blot was stripped and reprobed with actin Ab.
we examined the functional contribution of the interaction. We were able to create stable knockout cell lines of ERp44 and Ero1-L␣ through the delivery of RNA interference (RNAi). Initial attempts at protein reduction with transient transfection of protein-specific shRNA to degrade targeted mRNA constructs were unsuccessful due to the long half-life of ERp44 and Ero1-L␣. As a solution, lentiviral particles containing shRNA targeted to ERp44 or Ero1-L␣ were produced and transfected into target cell lines for long term RNAi. The expression levels of ERp44 and Ero1-L␣ were successfully reduced to 60% of endogenous levels in JAR cells. Furthermore, as a primary ER localization mechanism of Ero1-L␣, we observed a 30% reduction in Ero1-L␣ when ERp44 protein levels were significantly reduced. Our data confirm the previously published studies (26,35) on the saturability of the ERp44-dependent retention mechanism. According to these studies, if the Ero1-L␣/ERp44 ratio exceeds its threshold for retention, a portion Ero1-L␣ escapes from the ER (26,35).
In our knockout expression system, SERT is endogenously expressed in JAR cells; thus, we are able to measure the functional consequence of reduced ERp44 or Ero1-L␣ protein expression on SERT activity. However, these could be an indirect effect of the gene silencing on other proteins which are required for the SERT folding and maturation. Therefore, the kinetic and biochemical studies together with the data from MTSEA-biotinylation assays are more specific in demonstrating the direct action of ERp44 on SERT.
For untreated cells, the kinetic profile for 5-HT uptake was strongly sigmoidal, indicating that increasing 5-HT concentrations led to positive cooperativity. In other words, 5-HT levels improved the rate of transport not simply by saturating the transporter but by altering its functional efficiency and/or concentration. Based on our previous studies (31,36,37), we favor the biphasic relationship between the extracellular 5-HT levels with the number of SERT molecules on the plasma membrane. Specifically, the number of SERT molecules on the plasma membrane and the 5-HT uptake rates of cells initially rise as extracellular 5-HT levels are increased but then fall below normal as the 5-HT level continues to rise. Indeed, our in vivo and in vitro studies confirm a dynamic relationship between extracellular 5-HT elevation, loss of surface SERT, and depletion of platelet 5-HT (31,36,37). The rise in concentration of oligomerized, active SERT improves 5-HT activity more than expected, as reflected in the sigmoidal kinetic profile.
This mechanism depends on ERp44 function to maintain the appropriate response of SERT to changing 5-HT levels and the resulting transport activity. Without ERp44 chaperone activity, relative levels of SERT at the cell surface were high prior to the addition of 5-HT. As 5-HT levels increased, the sigmoidicity of the kinetic profile was essentially abolished, which is probably due to the absence of any change in cellular SERT localization. Nevertheless, once SERT reached the cell surface, its activity was compromised based on a decreased maximal uptake rate. SERT maturation into an active transporter then requires ERp44 activity as well. Ero1-L␣ is another contributor to these processes through its interactions with ERp44. Taken together, these data suggest that ERp44 plays a critical role in the maturation and membrane trafficking of SERT within cells.
To identify the cause of reduced SERT activity, we measured the SERT surface expression by biotinylation in shRNA-expressing cells. Despite decreased transport activity, we discovered a greater than 1-and 2-fold increase in SERT expression on the PM in cells expressing ERp44 or Ero1-L␣ shRNA, respectively. A reduced transport activity can be attributed to an increase in nonfunctional SERT and/or a decrease in fully active SERT molecules at the cell surface, which was demonstrated by our biochemical studies in ERp44-silenced cells (Fig.  5). Therefore, significantly lowered ERp44 or Ero1-L␣ is sufficient to allow improperly folded SERT molecules to reach the PM, and ERp44 and Ero1-L␣ probably act as a quality control mechanism ensuring the fidelity of SERT maturation.
Previous data implicate ERp44 and Ero1-L␣ in the quality control of cargo proteins, and a dynamic relationship exists between ERp44 and Ero1-L␣ for the efficient retention and release cargo (27,30,32). It was previously shown that an interaction between Cys-29 of ERp44 and Ero1-L␣ aids the release of the cargo protein from ERp44 (15, 26 -28). We hypothesized that this mechanism is also required for SERT release from ERp44. However, the increased SERT surface expression in cells with 30% of endogenous Ero1-L␣ expression seemed contradictory; thus, we measured the association between ERp44 and SERT in JAR cells expressing Ero1-L␣ shRNA. We discovered an almost 100% association between SERT and ERp44 when Ero1-L␣ is not present at endogenous levels, and this result further implies the saturability of ERp44-mediated retention (Fig. 4). When Ero1-L␣ levels are not able to adequately release cargo proteins from ER retention as in our Ero1-L␣ shRNA-expressing cell line, ERp44 becomes saturated. This outcome causes the cargo/chaperone ratio to increase, and as a result, misfolded SERT molecules are unable to be properly retained and bypass quality control.
SERT is a complex oligomeric glycoprotein, which needs significant post-translational modification to fold into its native structure (5, 9 -11). The formation of such a large, multimeric complex is inherently error-prone and requires more time to fold. Whereas ERp44-mediated quality control recognizes improperly folded SERT molecules and prevents them from reaching the PM, ERp44-dependent ER retention may serve to increase the time SERT is in the secretory pathway and thus the opportunity for SERT to fold into its native structure. Furthermore, ERp44 is able to couple quality control and oxidative folding by directing non-native SERT to the ER machinery needed for disulfide bond formation.
In addition to a quality control mechanism, the regulation of ERp44 and Ero1-L␣ may offer an additional layer of post-translational control of SERT activity. Increased expression of both chaperones, ERp44 and Ero1-L␣, has been shown to stimulate the functional oligomerization and expression of other cargo proteins involved in thiol-mediated retention (30). However, up-regulation of only ERp44 has been shown to favor the retention of cargo molecules, and Ero1-L␣ overexpression stimulates their release (30). This is especially relevant due to the tissue-and sex-specific expression of ERp44 and Ero1-L␣. For example, there is a 50% reduction in ERp44 and Ero1-L␣ expression in male mice as compared with female mice and a greater than 80% reduction in ERp44 and Ero1-L␣ in ob/ob mice (30). Thus, our in vitro shRNA experiments are physiologically relevant and provide evidence that regulation of ERp44 and Ero1-L␣ expression and activity could affect SERT in vivo.