Raft localization of DAT or direct interaction with cholesterol?
The definition of lipid rafts, or membrane rafts, has been a source of contentious debate (Pike 2006). Rafts are generally thought of as specialized microdomains comprised of sphingolipids and glycerophospholipids, which facilitates the concentration of cholesterol in lipid rafts (Pike 2006; Allen et al. 2007). This composition makes rafts resistant to solubilization in non-ionic detergents, but the concentration of detergent that should be used to define ‘raft-localized’ DAT is not agreed on (Shogomori and Brown 2003; Pike 2006; Foster et al. 2008; Hong and Amara 2010). However, strong evidence for localization of a portion of DATs with membrane rafts comes from its association with other markers of rafts, such as GM1 (Foster et al. 2008), flotillin-1 (Hong and Amara 2010; Cremona et al. 2011), and syntaxin 1a (Binda et al. 2008). What are the differences in functionality between non-raft and raft-localized DATs? In answering this question, investigators have used the oligosaccharide mβCD as a tool to efficiently remove membrane cholesterol and thereby disrupt lipid rafts (Adkins et al. 2007; Zidovetzki and Levitan 2007; Foster et al. 2008). However, as pointed out by Hong and Amara (2010), Cremona et al. (2011) and by others (Zidovetzki and Levitan 2007), mβCD removes cholesterol from the membrane, reducing its content, and thereby not only disrupts lipid rafts but also interferes with the direct interactions between cholesterol and the DAT, as well as removal of specific protein regulators, extraction of phospholipids, and cytoskeletal rearrangements (Ohvo and Slotte 1996; Kwik et al. 2003). Thus, the reduced function in DA uptake seen with mβCD does not necessarily imply that it is caused by raft disruption. In fact, Cremona et al. (2011) show it is more likely because of the loss of a direct effect of cholesterol on the DAT, as nystatin, which disrupts rafts but maintains cholesterol in the membrane, did not affect uptake of [3H]tyramine, another substrate for the DAT. In addition, depletion of flotillin-1, which is required to maintain raft-DAT, was also found not to alter [3H]tyramine uptake (Cremona et al. 2011). Our results agree with this conclusion, as nystatin did not appreciably affect DA uptake under our conditions. To address the latter more closely, we exchanged cholesterol with desmosterol, a direct a precursor of cholesterol, which has been shown to be inefficient at replacing cholesterol in ordered membrane domains, that is, lipid rafts (Megha and London 2006; Vainio et al. 2006). Indeed, we provide biochemical evidence to show that desmosterol repletion failed to reassemble lipid rafts as indicated by the inability to restore DAT (and flotillin-1) to lighter fractions in sucrose gradient experiments (Fig. 2). It has been shown that destabilizing lipid rafts by sterol exchange with desmosterol impaired insulin signaling, ligand binding function of the 5-HT1A serotonin receptor, and caveolar structure without interfering with non-raft functions of cholesterol (Wechsler et al. 2003; Xu et al. 2005; Lu et al. 2006; Vainio et al. 2006; Jansen et al. 2008; Singh et al. 2009). In comparison, DAT function as measured under the present conditions was not compromised by desmosterol repletion, but it was impacted by direct non-raft functions of cholesterol.
Given that desmosterol can be converted to cholesterol, it should be considered that this may occur within the time frame of our experiments. However, cholesterol-depleted HuH7 hepatoma cells (Vainio et al. 2006) or hippocampal membranes (Singh et al. 2009) replenished with desmosterol showed ≤ 10% increase in cholesterol within 1 h. Therefore, the contribution of metabolism of desmosterol to cholesterol is likely negligible in our case. It must be noted, however, that repletion with sterols in our system resulted in ~1.5–2 fold increase over control levels (Fig. 1). We do not have the data comparing over-repletion with repletion at original starting levels, and therefore, more work is need to describe the changes over the entire range of cholesterol (desmosterol) levels.
Cremona et al. (2011) did point to the importance of raft-DAT in AMPH-induced DA efflux, based on the observed decrease in efflux upon loss of flotillin-1 in neuronal cultures. We did not come to the same conclusion based on our efflux experiments. Besides the difference between neuronal cultures and HEK cells heterologously expressing DAT in the two studies, the efflux measurements themselves were quite different. In this study, AMPH is used to trigger reversed transport of DA. Clearly, if changes in the uptake rate of AMPH occur by a certain experimental manipulation, it will reflect upon the efflux observed, as AMPH-induced DA efflux as measured by RDEV is primarily exchange based (Chen et al. 1998). In this study, we corrected for this by dividing the efflux Vmax by the uptake Vmax. It must be noted, however, that the underlying assumption is that changes in the transport rate of DA equal changes in AMPH transport. Although this is a reasonable assumption, circumstances may exist, whereby transport rates for different substrates can respond differentially to a given manipulation. To resolve this for the present conditions, it will be required to prepare and apply [3H]amphetamine, which is not currently available on the market and would have to be prepared by custom synthesis. Conversely, Cremona et al. (2011) monitored AMPH-induced efflux under voltage clamp, starting at a holding potential of −60 mV with subsequent stepping in 20-mV intervals in depolarizing direction. It can be seen that the difference between efflux in flotillin-1 depleted neurons and that in control neurons was not statistically significantly different below −20 mV, but became increasingly so at positive potentials (+40 mV). Under our conditions, we found the membrane potential of HEK cells to be ~ −80 mV (Chen and Reith 2004), and therefore, the lack of effect on AMPH-induced efflux following raft disruption (by nystatin) at this voltage agrees with the Cremona et al. study.
Under physiological conditions, AMPH acts on neurons going through varying membrane potentials during impulse flow. At depolarized or positive potentials, it appears that AMPH-induced efflux starts recruiting raft-DAT, as indicated by the amperometric measurements under voltage clamp (Cremona et al. 2011). However, at resting potential, it appears AMPH-induced efflux does not require raft-DAT, as judged from both RDEV and classical radiotracer approaches (this study). It is plausible that both mechanisms may operate differentially on DAT function dependent on whether neurons are in a resting or depolarized state (Hoffman et al. 1999).
Importance of conformational states of DAT in cholesterol effects
The work of Hong and Amara (2010) strongly suggests that membrane cholesterol promotes outward-facing DAT. Thus, a scenario in which cholesterol levels modulate DATs conformational equilibrium is plausible. The data presented in the last two subsections of Results on WT and W84L treated with mβCD or Zn2+ suggest, additionally, a mechanism by which cholesterol depletion increases the proportion of DAT in an inward-facing conformation: a slowing down of transitions between inward- and outward-facing conformations, with the conversion from out-to-in slightly less impacted than the conversion from in-to-out conversion rate. Importantly, the reduced conversion rates explain a dual effect on DA uptake and AMPH-induced DA efflux; of note, the uptake and efflux Km and Vmax changes observed here upon mβCD treatment of hDAT are similar to the changes described upon mutation of Thre62 to Ala in hDAT by Guptaroy et al. (2009). Therefore, we propose that cholesterol, by a direct action on DAT, keeps transitions between inward- and outward-facing states at equilibrium.
In this context, it is instructive to consider the steps in DA translocation: binding of DA to DAT on one side of the membrane, translocation of DA to the other side of the membrane, and DA's release. In this cycle, the return of empty carrier is the rate-limiting step in transport, which suppresses Km to a value below the substrate binding affinity for the carrier (Zhang et al. 1997; Erreger et al. 2008). With a slower return step, Km becomes smaller, and with a faster return step, the Km value becomes higher increasing toward substrate binding affinity (this can also be seen in the formulas for transport derived by us previously (Chen et al. 2001; Zhang et al. 1997). In the presence of cholesterol (or desmosterol), the carrier returns easier when loaded with DA as speculated to occur during enhanced AMPH-induced efflux (Fig. 4b and c). If this also applies to the return of the empty carrier in a general conformational bias (Hong and Amara 2010), it would explain why cholesterol (or desmosterol) repletion has a slight tendency to increase the Km for DA uptake (Fig. 4a). In addition to such a kinetic mechanism, another factor to consider is the effect of cholesterol in increasing the Ki of DA binding to DAT (Table 1), which also contributes to an increased uptake Km (Fig. 4a). This Km change agrees with the reported reduction in Km upon cholesterol depletion with mβCD (Adkins et al. 2007; Cremona et al. 2011). In a different approach, DA affinity was assessed by competing for [3H]CFT binding in whole cells and in raft-enriched fractions obtained by sucrose density gradient separation. The IC50 of DA was found to be twofold higher in rafts than whole cells (Foster et al. 2008). This raises the question of the importance of the distribution of surface DAT across non-raft and raft domains in the plasma membrane.
Although these data suggest a possible link between cholesterol and transporter orientation or conformational interconversion rates, other factors may be in play. DAT has been shown to exhibit channel like activity under certain conditions. In particular, Zn2+ has been found to facilitate amphetamine-induced substrate efflux by DAT (Scholze et al. 2002), subsequently shown to be as a result of facilitation of an uncoupled (i.e., independent of translocation of charged substrate) Cl- transport (Meinild et al. 2004). It could therefore be thought that Zn2+ affects amphetamine-induced DA efflux in the present RDEV experiments by such an uncoupled mechanism. However, the RDEV method has been shown to primarily measure exchange-based DA efflux (Chen et al. 1998), and therefore, uncoupling of uptake from efflux is not detected with this technique. Zn2+-facilitated ion current could play a role in the present experiments on amphetamine-induced [3H]DA release, but the contribution of such a mechanism appears minor as the results of these experiments were similar to those obtained with RDEV. However, the possibility that cholesterol manipulation alters DAT channel function remains intriguing, and this could contribute to changes in DAT function under as yet to define conditions.
Also of note is the observed lack of effect on [3H]CFT binding. Although our findings agree with Foster et al. (2008) in that raft-DATs displayed similar cocaine binding as whole cells, the outward-facing state of the DAT is reported to exhibit increased Bmax of [3H]CFT binding (Norregaard et al. 1998; Liang et al. 2009; Hong and Amara 2010; Schmitt and Reith 2011). Therefore, a slight increase of the inward-facing state of DAT induced by cholesterol depletion would be expected to display a reduced Bmax, which was not seen in the present experiments (Table 1). However, for reasons not entirely understood [and discussed in (Schmitt and Reith 2011; Liang et al. 2009)], the effects of Zn2+ on the Bmax have not been consistent between published reports. In fact, two earlier studies (Wu et al. 1997; Chen et al. 2004a) found, in the presence of Na+, an impact of Zn2+ on Kd rather than Bmax, whereas Liang et al. (2009) observed the Bmax effect only in the absence of Na+. Be that as it may, factors other than conformational changes participate in the outcome of [3H]CFT binding experiments as evidenced by an unexpected lack of effect of cholesterol manipulation not only on the Bmax but also Kd (Table 1). This result differs from that reported by Hong and Amara (2010), and likely results from differences in the experimental conditions for measuring [3H]CFT binding.
The present results along with observations by others discussed above indicate that cholesterol does play an important role in DAT function, presumably through a direct action on DAT. Evidence for cholesterol binding with integral membrane-bound proteins particularly with GPCRs is mounting (Hanson et al. 2008). Numerous studies provide evidence of direct interactions of cholesterol with receptors and appear to be required for structure and function (Fernandez-Ballester et al. 1994; Klein et al. 1995; Gimpl et al. 2000; Eroglu et al. 2002; Nguyen and Taub 2003). In support, studies regarding neurotransmitter proteins have suggested that cholesterol plays a direct role for optimal function although conclusive evidence is lacking (Scanlon et al. 2001; Eroglu et al. 2002; Butchbach et al. 2004; Magnani et al. 2004; Liu et al. 2009).
Finally, the work on cholesterol is conceptually important in linking cholesterol content of brain membranes with alterations in DAT-mediated DA neurotransmission. Given that statins are widely used to treat hypercholestemia and are among the most prescribed drugs in the United States, this could have major ramifications in understanding side effects of statins in relation to DA, such as symptoms of depression and cognitive impairment (Kirsch et al. 2003; While and Keen 2010; Beydoun et al. 2011). Taken together, understanding how the membrane lipid environment influences DAT function may provide important information as to one of the molecular mechanisms that participates in modulating dopaminergic activity under physiological and diseased states.