|Home | About | Journals | Submit | Contact Us | Français|
The predominance of dopamine (DA) receptors at extrasynaptic versus synaptic sites implies that dopamine signaling is by diffusion-based volume transmission. In this review, we compare characteristics that regulate extracellular DA behavior in substantia nigra pars compacta (SNc) and striatum, including regional differences in structure (a 40% greater extracellular volume fraction in SNc vs. striatum) and in dynamic DA uptake (a 200-fold greater DA uptake rate in striatum vs. SNc). Furthermore, we test the assumption of diffusion-based volume transmission for SNc and striatum by modeling dynamic DA behavior after quantal release using region-specific parameters for diffusion and uptake at 37 °C. Our model shows that DA uptake does not affect peak DA concentration within 1 μm of a release site in either SNc or striatum because of the slow kinetics of DATs versus diffusion. Rather, diffusion and dilution are the dominant factors governing DA concentration after quantal release. In SNc, limited DAT efficacy is reflected in a lack of influence of uptake on either amplitude or time course of DA transients after quantal release up to 10 μm from a release site. In striatum, the lack of effect of the DAT within 1 μm of a release site means that perisynaptic DATs do not “gate” synaptic spillover. This contrasts with the conventional view of DA synapses, in which DATs efficiently recycle DA by re-uptake into the releasing axon terminal. However, the model also shows that a primary effect of striatal uptake is to curtail DA lifetime after release. In both SNc and striatum, effective DA radius after quantal release is ~2 μm for activation of low-affinity DA receptors and 7-8 μm for high-affinity receptors; the corresponding spheres of influence would encompass tens to thousands of synapses. Thus, the primary mode of intercellular communication by DA, regardless of region, is volume transmission.
The nigrostriatal DA pathway extends from the substantia nigra pars compacta (SNc) to the dorsal striatum via the median forebrain bundle. Somatodendritic release of DA in the SNc, as well as axonal DA release in the striatum are necessary for basal ganglia-mediated motor behaviors (Robertson and Robertson, 1989; Timmerman and Abercrombie, 1996; Bergquist et al., 2003). Transporters for DA (DATs) are expressed exclusively by DA neurons and are found extrasynaptically on DA axons in striatum and on DA somata and dendrites in midbrain (Ciliax et al., 1995; Nirenberg et al., 1996; Hersch et al., 1997). Moreover, DA receptors are also predominantly extrasynaptic (Sesack et al., 1994; Yung et al., 1995; Hersch et al., 1995; Khan et al., 1998). This implies that intercellular communication by DA requires diffusion-based volume transmission (Fuxe and Agnati, 1991), which is most simply defined as “a functionally significant association of release and receptor sites via extrasynaptic diffusion” (Rice, 2000).
One would expect, therefore, that extracellular DA concentration ([DA]o) and lifetime after release would be regulated by diffusion, as well as uptake. This is indeed the case, although the relative contribution of each process differs between striatum and SNc. In striatum, [DA]o is often considered to be “uptake-limited”, with strong local regulation by the DAT (Stamford et al., 1988; Giros et al., 1996; Floresco et al., 2003), whereas in SNc, the influence of uptake is much less (Cragg et al., 1997), so that diffusion-based volume transmission is expected to dominate (Rice, 2000; Cragg et al., 2001). Recently, however, both of these basic expectations have been challenged. Evidence from physiological recording in midbrain DA neurons suggests limited diffusional contributions to DA transmission in the SNc (Beckstead et al., 2004; Beckstead et al., 2007), whereas modeling of quantal DA release from DA axons in striatum indicates minimal regulation of synaptic spillover by the DAT (Cragg and Rice, 2004).
In this review, we extend our previously published model of quantal DA release in striatum (Cragg and Rice, 2004) to include somatodendritic release in SNc. Other models that have been used to estimate the sphere of influence of released DA have been limited by the single distance examined (Gonon, 1997; Cragg et al., 2001) or by the evaluation of a population response in which DA diffusion, and thus diffusion distance, is irrelevant (Garris et al., 1994; Sulzer and Pothos, 2000). The model used here is based on experimentally determined parameters for diffusion and uptake in SNc and striatum and permits comparison of the duration and sphere of influence of [DA]o after single-vesicle release in both regions. Our findings confirm the comparatively limited role of uptake in regulation of DA behavior in SNc versus striatum and call for the revision of the current concept of synaptic DA release regulation, in which the DAT “gates” DA overflow by reuptake into the releasing presynaptic terminal.
Theoretical concentration-time profiles for quantal DA release at 37 °C were generated using the expression for diffusion from an instantaneous point source (Nicholson, 1985; Cragg et al., 2001) using the program VOLTORO by Dr. C. Nicholson (NYU School of Medicine, USA):
Quantal DA release was assumed to be instantaneous from a vesicle of volume U with a filling concentration of Cf, such that quantal size, Q = UCf. Diffusion distance, r, was varied from 1-20 μm.
The structural parameters that govern diffusion are the extracellular volume fraction (α) and the tortuosity (λ) of extracellular diffusion paths (Nicholson and Sykova, 1998). Experimentally derived values for α, λ, k′, Cf and U for SNc and striatum are given in Table 1. Macroscopic diffusion parameters have been shown to be appropriate to describe microscopic structure, including that surrounding a synapse (Rusakov and Kullmann, 1998). These parameters are assumed to be temperature independent. On the other hand, the diffusion coefficient, D, for DA is temperature dependent; the value used for D at 37 °C was 7.63 × 10-6 cm2 s-1 (Table 1), calculated from D at 32 °C (Nicholson, 1995), assuming a 2% increase per °C.
In experimental studies of DA diffusion and uptake in the SNc, we found that the linear uptake term k′ for specific DA uptake was sufficient to fit DA diffusion curves in this region (Cragg et al., 2001). To model nigral DA behavior at 37 °C, we increased the k′ value for SNc in accordance with the temperature-dependence of DA uptake evaluated in striatum (Cragg and Rice, 2004). Similarly, the k′ for non-specific DA uptake, determined from measurements of the diffusion of tetramethylammonium (TMA+) in midbrain (Cragg et al., 2001), was temperature-corrected to 37 °C (Table 1), and was used for “diffusion only” curves generated for both SNc and striatum.
Linear uptake was also used to model DA behavior in striatum (Cragg and Rice, 2004). The relationship between Michaelis-Menten uptake rate and concentration becomes linear when [DA]o is substantially below Km, i.e. C << 0.21 μM (Ross, 1991). Then, the uptake rate, V = VmaxC/(C+Km), simplifies to first order kinetics with respect to C, such that V = VmaxC/Km = k′ C, where k′ = Vmax/Km. Use of linear uptake for striatum in this model is justified because [DA]o remains in the nanomolar range following release of a single vesicle, except at distances very near to the site of release (Cragg and Rice, 2004). At those sites, the use of linear uptake overestimates uptake when [DA]o ≥ Km, because uptake rate increases linearly with increasing concentration (V = k′C), rather than approaching an asymptotic value (V = Vmax). However, this overestimation only strengthens the case for the limited DAT-dependent regulation of peak [DA]o after quantal release described below.
A further assumption of our model is that DA vesicle fusion at a DA synapse (or varicosity) (Descarries et al., 1996) in striatum or on a DA cell body or dendrite in SNc acts as an instantaneous point source (Cragg and Rice, 2004). The validity of this assumption for SNc is a matter of debate, given the uncertainty of whether release is mediated exclusively by conventional vesicle fusion (Cragg et al., 1997; Chen and Rice, 2001; Falkenburger et al., 2001). Nonetheless, quantal release from DA cells in the SNc has been recorded in acute midbrain slices (Jaffe et al., 1998). We used the reported quantal size (Q) of 14,000 DA molecules for simulations of DA behavior in SNc (Table 1), unless otherwise stated.
Because the extracellular volume fraction of midbrain is 40% larger than that in striatum (Rice and Nicholson, 1991; Cragg et al., 2001), the apparent source strength is 40% lower in SNc than in striatum. For initial comparisons of extracellular DA behavior in SNc and striatum, we normalized source strength between regions by a proportional decrease in striatal Q to 9,800 molecules (Table 1). The resultant values of Cf (350 mM, SNc; 250 mM, striatum) are within a range consistent with those determined experimentally for DA in cultured neurons (Pothos et al., 1998).
In the SNc, relatively low DAT activity compared to that in striatum leads to the surprising result that peak [DA]o is minimally affected by uptake even 20 μm from the site of release (Cragg et al., 2001) (Fig. 1, lower panel). By contrast, in striatum, where k′ is nearly 200-fold greater (Table 1), an effect of uptake on peak [DA]o is seen by r = 2 μm (Fig. 1, lower panel). Notably, however, whether SNc- or striatum-specific DA uptake is included or not, the dominant influence on [DA]o is the distance from a release site (note the order-of-magnitude decrease in concentration at each successive r in Fig. 1), as diffusing DA is diluted into an increasing volume defined by 4/3πr3 (Cragg and Rice, 2004). Overall, therefore, diffusion and dilution are the main factors that govern absolute [DA]o after quantal release in both SNc and striatum.
A striking finding from our simulations of DA behavior in striatum is the complete lack of uptake on peak [DA]o within 1 μm of a release site (Cragg and Rice, 2004) (Fig. 1, lower panel). Given that the cross-synaptic diameter of a striatal DA synapse is 0.3-0.6 μm (Pickel et al., 1981; Groves et al., 1994; Descarries et al., 1996), this means that DA must necessarily spill over from a synapse or other release site, with its maximum concentration unmodulated by uptake. We tested the hypothesis that this occurs because the rate of diffusion exceeds that of DA uptake near a release site (r < 5 μm; see Fig. 1) by “slowing” DA diffusion by doubling of the tortuosity factor, λ, which decreases the apparent diffusion coefficient, D*, by a factor of four (D* = D/λ2) (Cragg and Rice, 2004). The theoretical slowing of DA diffusion allows more time for DAT mediated uptake and led to a slight decrease in peak [DA]o, even 1 μm from the release site. This result confirms that DA uptake normally loses the kinetic competition against diffusion in mediating DA clearance within a synapse. Thus, perisynaptic DATs neither modulate maximum synaptic DA concentration nor gate initial DA spillover (Cragg and Rice, 2004).
The primary role of diffusion in governing [DA]o at all distances from a release site, whether in SNc or striatum, leads naturally to the question of how far DA can diffuse. In the absence of uptake (or metabolism) and with sufficient time, DA molecules can diffuse from a release site to the physical boundaries of the brain, i.e., a ventricle or the subarachnoid space. However, the better question is how far from a release site will DA have physiological consequences? This obviously depends on the concentration-sensitivity of DA receptors within a potential sphere of influence. Specifically, the EC50 values for activation of D1-like and D2-like receptors in vitro are on the order of 1 μM for low-affinity states and 10 nM for high-affinity states (Richfield et al., 1989; Neve and Neve, 1997). In striatum, the majority of D1-like receptors are suggested to be in a low-affinity state, whereas D2-like receptors are predominantly high affinity (Richfield et al., 1989). Assuming these relative affinities also hold true for SNc, the maximum “effective radius” can be defined as the distance within which release of one quantum of DA reaches high affinity D2 receptors at ≥ 10 nM above baseline (Cragg and Rice, 2004).
With this constraint and using the normalized Q values in Table 1, the maximum effective radius for spillover in SNc and in striatum would be 8.2 μm, in the absence of uptake (diffusion only) (Fig. 2). Unsurprisingly, given limited DA uptake in SNc, including DAT-mediated uptake in the simulation has no effect on maximum effective radius in SNc (Fig. 2, upper panel). However, in striatum, uptake would decrease the maximum effective radius to 7.0 μm (Cragg and Rice, 2004). Although the absolute difference in r seems negligible, the r3-dependence of spherical volume means that the sphere of influence of DA, and therefore the number of receptors activated, would be 40% smaller with normal uptake than under conditions of DAT inhibition.
The effective radius for DA acting at low-affinity, predominantly D1-like receptors is much smaller than for high-affinity receptors. For receptors requiring 1 μM DA for activation, the maximum effective radius would be < 2 μm in SNc or striatum (Fig. 2). At this distance from the release site, there is little influence of the DAT on peak [DA]o and consequently little influence on the sphere of influence in either SNc or striatum (Fig. 2). Consequently, these data indicate that DAT inhibition will have a greater influence on the activation of high-affinity DA receptors than on low-affinity receptors.
Thus far, we have considered the influence of the DAT on peak [DA]o and effective radius of DA transients. A third parameter of importance in DA transmission is active lifetime, i.e., the time DA is available at sufficient concentration to activate DA receptors. The maximal active lifetime is the time during which [DA]o remains above the EC50 for high affinity DA receptors, 10 nM, at a given site after quantal release (Cragg and Rice, 2004). This is measured as the difference between the time at which [DA]o first exceeds 10 nM (onset) and when it falls below 10 nM (offset) (Fig. 3, inset). In SNc, active lifetime is longest at r ≤ 1 μm, with a maximum lifetime of 100 ms for a single Q, then decreases progressively as r approaches the limit of the effective radius (Fig. 3, upper panel). This pattern is the same whether SNc-specific DA uptake is present or not. In striatum, a similar lifetime pattern is seen in the absence of uptake (Fig. 3, lower panel). However, under conditions of normal striatal DAT activity, uptake curtails the duration of striatal [DA]o transients within the effective radius by 50%. Thus, a major role of the DAT in striatum, but not in SNc, is to limit DA lifetime after release. Through this mechanism, the DAT can readily influence DA transmission.
The sphere of influence of DA after quantal release can be calculated from the maximum effective radius for a given set of conditions. We determined the sphere of influence for activation of high-affinity (e.g., D2 receptors) and low-affinity DA receptors (e.g., D1 receptors), based on the effective radii for EC50 values of 10 nM and 1 μM, respectively, for a range of Q in SNc and striatum (Fig. 4). The range of Q examined was based on available literature indicating that the number of DA molecules per vesicle can vary from 1,000 to 14,000 (Garris et al., 1994; Pothos et al., 1998; Jaffe et al., 1998), reflecting the plasticity of the volume and quantal content of catecholamine vesicles (Pothos et al., 1998; Colliver et al., 2000), as well as methodological differences in assessing Q. We did not normalize Q between SNc and striatum for this comparison, but rather used identical Q values for both regions to gain insight into the role of α, as well as the DAT, in the regulation of [DA]o after quantal release.
Our model shows that in SNc, the sphere of influence is directly proportional to the number of molecules released, whether specific DAT-mediated uptake is active or inhibited (Fig. 4A). For example, a seven-fold larger Q (14,000 vs. 2,000) will have a seven-fold larger sphere of influence. A similar linear dependence on Q also occurs in striatum in the absence of uptake (Fig. 4C), although the resulting spheres would be roughly 40% larger than those in SNc for equal Q, because of the smaller striatal volume fraction, α (compare Figs. 4A,B and 4C,D for “diffusion only”).
Normal DA uptake in striatum constrains the sphere of influence for activation of high-affinity DA receptors across a range of Q when compared to the spheres seen when clearance is by diffusion only (Fig. 4C). Importantly, the influence of the DAT increases as Q increases, because larger, longer-lasting [DA]o transients allow longer times for uptake to remove DA (Cragg and Rice, 2004). For example, uptake decreases the sphere of influence by ~20% for Q = 2,000 DA molecules, but by ~40% for Q = 14,000 (Fig. 4C). Strikingly, the spheres of influence for activation of low-affinity DA receptors are largely unaffected by uptake, regardless of Q, in both SNc (Fig. 4B) and striatum (Fig. 4D). Again, this reflects the rapid clearance of DA by diffusion and the consequently minimal influence of the DAT at r < 2 μm within which [DA]o is ≥ 1 μM.
The minimal effect of DA uptake on the spheres defined by activation of low-affinity DA receptors coupled with the smaller α of striatum versus SNc means that the absolute tissue volume affected for a given Q is larger in striatum than in SNc. In contrast, the spheres of influence for DA acting at high-affinity DA receptors in the presence of region-specific uptake are surprisingly similar between SNc and striatum after the release of equal Q (compare Figs. 4A and 4C for “diffusion + uptake”). This similarity reflects the opposing influences of uptake and α in each region: the larger α of SNc offsets the lower k′, whereas the higher k′ of striatum is balanced by the smaller α. Intriguingly, the relative effects of these two parameters on sphere of influence vary with Q. For small, fast transients generated by the release of small Q, α dominates over uptake, such that for Q = 2,000 the sphere of influence is smaller in SNc than in striatum, whereas for larger transients larger Q, uptake dominates, such that for Q = 14,000 the sphere is larger in SNc than in striatum.
Anatomical data suggest that there are on average 0.05 DA synapses per μm3 or 1 DA synapse per 20 μm3 in the striatum (Doucet et al., 1986), although recent estimations suggest that the density might be twice as high as that (Arbuthnott and Wickens, 2007). Even assuming a density of only 1 DA synapse per 20 μm3, when the average intersynaptic distance can be calculated to be 3.5 μm, synaptic “spillover” to neighboring synapses within these spheres of influence will readily occur, with gradients of efficacy that depend on receptor sensitivity (Cragg and Rice, 2004). How many synapses will be affected? In striatum, the spheres of influence for Q = 2,000-14,000 DA molecules acting at high-affinity receptors would contain ~20-100 DA synapses under condition of normal DAT activity (Cragg and Rice, 2004). Spillover of DA to neighboring synapses will therefore necessarily occur after single-vesicle release. However, the sphere defined by r ≤ 2 μm for low affinity DA receptors would affect 0-1.5 DA synapses (Cragg and Rice, 2004).
Comparable calculations cannot be made for SNc, however, given the absence of axo-dendritic DA synapses and only few dendro-dendritic synapses in that region (Wilson et al., 1977). Moreover, DA receptors are expressed peri- and pre-synaptically at non-DA synapses, as well as on neuronal soma in both SNc and striatum. A better sense of DA influence might therefore come from considering the number of receptors within a given sphere. However, such evaluation is hampered at present by a lack of quantitative data about receptor number for high and low affinity DA receptors in striatum or SNc. A practical alternative approach is to consider the total population of synapses (DA and non-DA) potentially available for dopaminergic regulation. Total synaptic density in striatum is ~1 synapse per μm3 (Pickel et al., 1981). Given the consistency of this synaptic density with that in other CNS regions, e.g., the CA1 region of hippocampus (Rusakov et al., 1999), we used this value to assess the influence of DA in SNc, as well as striatum. With normal uptake, released DA would encounter ~300 to 2,500 synapses within the spheres defined by [DA]o ≥ 10 nM for Q = 2,000-14,000 molecules in both regions (Fig. 4A,C). By contrast, the number of synapses in the spheres defined by [DA]o ≥ 1 μM is nearly 200-fold lower, with only ~5 to 35 synapses encountered (Fig. 4B,D). Notably, the minimal influence of uptake on DA transients in these smaller spheres unmasks the effect of the larger α of SNc versus striatum, with a lower number of synapses encountered in SNc than in striatum for all Q.
One final point to consider for both SNc and striatum is the active lifetimes of DA within these spheres of influence. Our model indicates that over the quantal range examined, the active lifetime during which [DA]o exceeds a minimum of 10 nM would be 10 to 100 ms. Notably, the duration of extracellular DA availability does not need to be shorter than this since the effector responses of DA receptors, which are G-protein coupled, range from 100 to 5000 ms (Cragg and Greenfield, 1997; Robinson and Caron, 1997; Gonon, 1997; Benoit-Marand et al., 2001; Phillips et al., 2002).
In striatum, the greater effect of the DAT on larger, longer-lasting [DA]o transients after the release of larger Q (Fig. 4C), would also hold true for the release of multiple vesicles, from either single or multiple sites. The temporal and spatial summation of the resultant transients determine net [DA]o in the striatum and would therefore be expected to be strongly regulated by the DAT. Such a role is indeed consistent with the increases in basal [DA]o known to occur after uptake blockade (Cragg et al., 1997; Bradberry et al., 2000) or in DAT knockout mice (Giros et al., 1996). Importantly, the relative roles of the DAT on striatal [DA]o during tonic and phasic activity of midbrain DA neurons have also been determined experimentally (Floresco et al., 2003). As predicted by our model, inhibition of the DAT was found to cause a greater increase in [DA]o during enhanced burst firing of DA neurons, which leads to quantal summation, than during a simple increase in asynchronous tonic activity in these cells (Floresco et al., 2003).
Our model of DA behavior after quantal release indicates that uptake does not limit initial DA overflow from a release site in either striatum or midbrain. This contrasts completely with regulation of the prototypical fast synaptic transmitter, glutamate (Rusakov and Kullmann, 1998; Barbour, 2001; Cragg and Rice, 2004). Moreover, diffusion, rather than uptake, is the key factor that regulates dynamic DA behavior, not only in the SNc, as predicted, but also in dorsal striatum in which regulation of [DA]o has been characterized as “uptake limited.” Nevertheless, this process constrains dopaminergic transmission in space and time: a DA transient produced by release of a single Q will have a maximum effective radius of 7-8 μm for activation of high-affinity DA receptors and an active lifetime of 10-100 ms. Thus, our model shows that diffusion and dilution, especially when aided by uptake, regulate dopaminergic transmission to a greater extent than sometimes imagined for signaling via volume transmission.
Evidence for DA volume transmission in the SNc (Fig. 5) includes the predominance of extrasynaptic DA receptors (Cameron and Williams, 1993; Sesack et al., 1994; Yung et al., 1995), the virtual absence of DA synapses (Wilson et al., 1977), and a limited influence of DA uptake on either exogenously diffusing (Cragg et al., 2001) or endogenously released DA (Cragg et al., 1997; Chen and Rice, 2001). Moreover, as discussed here and elsewhere (Rice, 2000; Cragg et al., 2001), the release of a single vesicle of DA could modulate the excitability of tens to thousands of synapses within distances a few micrometers of a release site, in both SNc and striatum. This broad influence of released DA meets the definition of volume transmission given in the Introduction, which is independent of the source of release, i.e., from synaptic or non-synaptic sites.
Recently, the view that DA acts by volume transmission in the SNc has been challenged, based on electrophysiological monitoring of D2-receptor mediated inhibitory “postsynaptic” currents (IPSCs) in DA neurons (Beckstead et al., 2004; Beckstead et al., 2007). In these studies, local electrical stimulation in the presence of a cocktail of glutamate and GABA receptor antagonists evoked a robust IPSC in most DA neurons examined. Beckstead, Williams and colleagues have argued against DA volume transmission in the SNc (Beckstead et al., 2004; Beckstead et al., 2007) based on their observation that the time course of DA-dependent IPSCs is unaltered under conditions that increased the amplitude of the IPSCs, including by increasing stimulation intensity (Fig. 6A,B) or increasing quantal content with l-DOPA. Their interpretation is based on the premise that, “If super-threshold concentrations of dopamine were typically traveling micrometers to reach target D2 receptors then increasing release would have prolonged the time course of the synaptic current” (Beckstead et al., 2007). There are two flaws in this premise, however. First, there are few dendrodendritic synapses in the SN (Wilson et al., 1977) that could mediate synaptic DA transmission; moreover, no synaptic D2 receptors have been identified in midbrain DA neurons (Sesack et al., 1994; Yung et al., 1995; Khan et al., 1998). Second, the authors’ premise does not take into account the time course of diffusion of DA from a release site to its site of action. As shown in this review, the time to peak and duration of [DA]o transients after quantal release are much shorter than those of recorded IPSCs. At a distance of 5 μm from a release site, the resulting [DA]o transient would reach a maximum ~10 ms after release, whereas the observed peak IPSC does not occur until ~500 ms after the stimulus (Fig. 6B; [DA]o data from Fig. 1 superimposed on IPSC records from Beckstead et al. 2004). Even 20 μm from a release site, peak [DA]o would occur only 250 ms after release, which is still half the time of the IPSC peak (Fig. 6B). Thus, the IPSC data lack the temporal resolution to provide insight into the role of volume transmission on DA communication on the SN.
The rate-limiting step in IPSC generation is activation of D2-receptor linked GIRK (G-protein coupled inward rectifier K+) channels, not DA diffusion (Beckstead et al., 2004). The IPSC data reported are consistent with previous studies of the time course of D2-autoreceptor mediated inhibition of DA release in striatum, in which autoinhibition begins 50-100 ms after DA single-pulse evoked DA release, then persists for up to 5 s afterwards, which greatly exceeds the lifetime of the DA transient that initiated the autoreceptor response (Phillips et al., 2002). It is unsurprising, therefore, that the time course of DA IPSCs would be independent of the amplitude of rapid [DA]o transients that initiate D2 receptor-linked GIRK channel opening. Indeed, the dependence of IPSC amplitude on stimulation intensity supports a role for DA volume transmission, rather than arguing against it. Local stimulation activates not just the recorded neuron, but also surrounding cells and dendrites, such that the resulting increase in [DA]o is a population response. As stimulus intensity is increased, the resulting increase in current spread will lead to action potential generation and somatodendritic DA release from increasingly distant neurons. The dependence of D2-autoreceptor mediated current amplitude on stimulus intensity (Fig. 6A) is therefore consistent with increases in [DA]o detected by the recorded cell as the summation of DA transients from both local and increasingly distant sources, reflecting volume transmission.
Given the likely contribution of distant release sites, and therefore multiple vesicles, to initiating [DA]o transient, it is also unsurprising that one condition that did increase IPSC time course in DA neurons was DAT inhibition by cocaine (Beckstead et al., 2004; Beckstead et al., 2007) (Fig. 6C). This response is consistent with previous studies of the effect of DAT inhibition on experimental DA diffusion profiles in the SNc recorded 100 μm from a pressure ejection pipette (Cragg et al., 2001) (Fig. 6D). The proportionately greater effect of DAT inhibition on the IPSC records than on DA diffusion curves reflects the primary contribution of uptake to DA clearance after release by electrical stimulation, in which a locally homogeneous increase in [DA]o is predominantly cleared by uptake rather than diffusion (Garris et al., 1994). As Beckstead et al. (2004) conclude, their effect of cocaine confirms “the crucial role of uptake transporters in limiting the time dopamine is in the extracellular space,” and is therefore available to mediate volume transmission.
The conventional representation of a striatal DA synapse shows vesicular release with reuptake of release DA into the presynaptic terminal by perisynaptic DATs, which restricts synaptic spillover of DA and regulates synaptic DA concentration (Fig. 7, left panel). However, our kinetic analysis of DA diffusion and uptake after quantal release shows that synaptic DA spillover cannot be gated by perisynaptic DATs, because the kinetics of DA transport cannot compete with the rapid escape of DA away from a release site by diffusion (Cragg and Rice, 2004). This further means that perisynaptic DATs can do little to alter the concentration of DA within a striatal synapse. It is therefore time for a significant revision of the depiction of a DA synapse that incorporates these inconvenient facts (Fig. 7, right panel).
Two main factors contribute to the dominance of diffusion over DAT-mediated uptake. The first is the rapidity of diffusion over micrometer distances and the low DAT cycle rate (2-5 DA molecules per sec per transporter at 37 °C) (Povlock and Schenk, 1997; Prasad and Amara, 2001). The second, less obvious factor is that DAT expression is limited to DA neurons, including DA axons in striatum (Nirenberg et al., 1996). Immediately after release, therefore, a DA molecule diffusing in three-dimensional space will be subject to transporter activity only if it encounters the presynaptic membrane of the releasing DA axon (Fig. 7, right panel). This geometric organization is analogous to that of the retina, in which DA is released from DAT-expressing amacrine cells oriented in a single plane in the inner plexiform layer: once a DA molecule escapes this plane, its behavior is governed only by diffusion (Witkovsky et al., 1993; Rice, 2000). Diffusing DA can readily interact with predominantly extrasynaptic DA receptors (Sesack et al., 1994; Yung et al., 1995; Hersch et al., 1995; Khan et al., 1998) (Fig. 7, right panel), until it reaches the boundary set by receptor sensitivities (e.g., Fig. 2).
The perisynaptic landscape of a glutamate synapse is very different from that for DA synapses (Cragg and Rice, 2004). Glutamate transporters are localized primarily on non-glutamatergic cells, including astrocytic processes that can ensheath a glutamate synapse, as well as post-synaptic neurons (Seal and Amara, 1999; Danbolt, 2001). Moreover, glutamate transport kinetics exceed those of the DAT by an order of magnitude, with transport-cycle rates of over 35 glutamate molecules per sec per transporter at 37 °C (EAAT2) (Wadiche et al., 1995). Most glutamatergic communication is mediated by fast-responding (1-2 ms), ionotropic glutamate receptors located immediately postsynaptically, with slower modulation via metabotropic receptors found near the perimeter of the synapse (Ottersen and Landsend, 1997) and possible additional contributions from extrasynaptic receptors (Galvan et al., 2006). Although there is experimental evidence to support glutamate spillover (Rusakov and Kullmann, 1998; Rusakov et al., 1999), released glutamate is typically constrained by uptake to remain perisynaptic, thus preserving synaptic independence (e.g. Barbour, 2001).
At the present time, DA synapses are most commonly illustrated as though they were glutamate synapses (Fig. 7, left panel), despite marked differences between the structural and functional anatomy of these transmitter release sites. To understand dopaminergic transmission requires that the classic representation of the DA synapse be updated to reflect the functional microanatomy of these synapses, which, unlike glutamate synapses, are designed for overflow and volume transmission (Fig. 7, right panel).
The slow kinetics and limited distribution of the DAT leads to predominant regulation of DA behavior by diffusion in both SNc and striatum. Following quantal release, nigrostriatal DA can readily spill out from release sites and act within effective radii of several micrometers. Thus, diffusion-based volume transmission is the predominant mode of DA signaling after both somatodendritic and synaptic release.
The authors gratefully acknowledge support from the Wellcome Trust (SJC, MER), NIH/NINDS grant NS-36362 (MER), the National Parkinson Foundation, USA (MER), a Paton Fellowship (SJC), and the Parkinson's Disease Society, UK (SJC).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.