Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Neurosci. Author manuscript; available in PMC 2010 December 1.
Published in final edited form as:
PMCID: PMC2918390

Dopaminergic terminals in the nucleus accumbens but not the dorsal striatum co-release glutamate


Coincident signaling by dopamine and glutamate is thought to be crucial for a variety of motivated behaviors. Previous work has suggested that some midbrain dopamine neurons are themselves capable of glutamate co-release, but this phenomenon remains poorly understood. Here, we expressed the light-activated cation channel Channelrhodopsin-2 (ChR2) in genetically defined midbrain dopamine neurons to stimulate exocytosis specifically from dopaminergic terminals in both the nucleus accumbens (NAc) shell and dorsal striatum (DS) of brain slices from adult mice. Optical activation resulted in robust glutamate-mediated excitatory postsynaptic currents in all medium spiny neurons examined in the NAc shell. In contrast, optically evoked glutamatergic currents were nearly undetectable in the dorsal striatum. Further, we used a conditional knockout mouse lacking vesicular glutamate transporter 2 (VGLUT2) specifically in dopamine neurons to determine whether VGLUT2 is required for the exocytotic release of glutamate from dopamine neurons. Our data shows that conditional knockout completely abolished all optically evoked glutamate release. These results provide definitive physiological evidence for VGLUT2-mediated glutamate release by mature dopamine neurons projecting to the NAc shell, but not to the dorsal striatum. Thus, the unique ability of NAc-projecting dopamine neurons to synchronously activate both dopamine and glutamate receptors may have crucial implications for the ability to respond to motivationally significant stimuli.

Keywords: optogenetics, dopamine, glutamate, corelease, VGLUT2, Ventral tegmental area, nucleus accumbens


The role that a neuron plays within a circuit depends to a large extent on the neurotransmitter(s) released. Midbrain dopamine neurons project to a variety of forebrain targets including the dorsal striatum and nucleus accumbens where the release of dopamine is thought to contribute to the generation of motor function and motivated behaviors (Wise, 2004). As a variety of adaptive behaviors are ascribed to specific striatal subregions, it is possible that the striatal micro-circuits that underlie these specific behavioral processes have distinct properties, such as differences in afferent or efferent connectivity.

Accumulating evidence suggests that a subset of midbrain dopamine neurons may also corelease the excitatory neurotransmitter glutamate both in vitro (Sulzer et al., 1998; Joyce and Rayport, 2000; Dal Bo et al., 2004) and in vivo (Chuhma et al., 2004; Lavin et al., 2005; Hnasko et al., 2010). However, this possibility has generated controversy due to difficulty stimulating selectively dopamine neurons or their terminals.

Vesicular glutamate transporters package glutamate into synaptic vesicles and are both necessary and sufficient for the exocytotic release of glutamate from neurons (Reimer and Edwards, 2004; Takamori, 2006). In culture, dopamine neurons immunostain for the vesicular glutamate transporter VGLUT2 (Dal Bo et al., 2004). However the histochemical evidence for VGLUT2 expression by dopamine neurons in vivo remains less clear. One study found substantial colocalization of mRNAs encoding the dopaminergic marker tyrosine hydroxylase (TH) and VGLUT2 in the ventral midbrain, with a higher incidence of colocalization in medial and caudal regions of the A10 dopamine cell group (Kawano et al., 2006). A second report detected co-localization only very rarely, but medial aspects were not examined (Yamaguchi et al., 2007). Using single-cell RT-PCR (Mendez et al., 2008) and immuno-electron microscopy (Dal Bo et al., 2004; Descarries et al., 2008), Trudeau and colleagues detected substantial co-localization of VGLUT2 transcript in dopamine neurons and VGLUT2 protein in TH+ terminals of the nucleus accumbens. However, the level of co-localization declined significantly across development (Mendez et al., 2008; Berube-Carriere et al., 2009), suggesting that glutamate corelease may similarly down regulate in adulthood.

To determine whether dopamine neurons release synaptically relevant amounts of glutamate in the adult brain, a method to selectively stimulate exocytotic release from dopamine terminals is required. We therefore used a multifaceted strategy, combining optogenetics and conditional gene disruption to selectively stimulate dopaminergic terminals while recording post-synaptic currents from medium spiny neurons (MSN) in either the ventral or dorsal striatum of control or conditional knockout mice lacking VGLUT2 specifically in dopamine neurons.

Materials and Methods

Experimental subjects

Age-matched, sex-matched adult male and female mice (~25g) were used as subjects. Control mice carried one copy of cre recombinase driven by dopamine transporter regulatory elements (Zhuang et al., 2005) and one conditional VGLUT2 allele (Hnasko et al., 2010) [Slc6a3+/cre; Slc17a6+/lox].Conditional VGLUT2 knockout mice were identical but carried two conditional VGLUT2 alleles [Slc6a3+/cre; Slc17a6lox/lox].Mice were group housed in a colony maintained with a standard 12-hr light/dark cycle and given food and water ad libitum. Experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals, as adopted by the NIH, and with approval of UCSF Institutional Animal Care and Use Committee.

Stereotaxic recombinant adeno-associated virus (AAV) injection

Methods were adapted from (Tsai et al., 2009). Briefly, mice were anesthetized with ketamine/xylazine, placed in a stereotax (Kopf), the skull leveled, small holes drilled, and 1 µl (0.25 µl/min) of AAV5-EF1α-DIO-ChR2-mcherry (~3*1012 genomes/ml) was injected using a Hamilton syringe bilaterally into the VTA (coordinates in mm relative to bregma: −3.25 AP, −4.50 DV, & 0.5 or −0.5 ML). Mice were allowed to recover for at least 3 weeks prior to further procedures.


Mice were perfused with cold PBS followed by 4% PFA, brains were removed, post-fixed overnight, cryoprotected in 30% sucrose, frozen in superchilled isopentane, and 35-µm sections were cut on a cryostat and floated in PBS. Sections were rinsed with PBS containing 0.2% Triton-X-100, blocked 1 h w/ 4% NDS, incubated overnight at 4°C in sheep-anti-TH (Pel-Freze) and rabbit-anti-DsRed (Clontech), both at 1:2000 dilution in blocking solution. Sections were rinsed, incubated 2 hr with Cy2-conjugated anti-sheep and DyLite549-conjugated anti-rabbit secondaries (1:500, Jackson Immunoresearch), rinsed again, mounted on slides, dehydrated through alcohol/xylenes and coverslipped w/ DPX.

Images were collected using a Nikon Eclipse Ti-E motorized inverted microscope equipped with epifluorescece and a PhotometricsCoolsnap HQ2 camera or a Nikon FN1 upright C1si spectral confocal microscope. For quantification of TH/ChR2-mcherry+ neurons in the VTA, confocal images were taken of using a 60x objective. ChR2-mcherry+ neurons were identified and then scored for TH content.

Brain slice preparation

Mice were anesthetized and rapidly decapitated. Dissected brains were transferred to ice-cold ACSF containing the following (in mM): 75 sucrose, 87 NaCl, 2.5 KCl, 7 MgCl2, 0.5 CaCl2, 25 NaHCO3, 1.25 NaH2PO4, 1 ascorbic acid (saturated with 95% O2 and 5% CO2). Coronal sections of the striatum (200 µM) were cut with a vibratome (VT1200, Leica). Slices were incubated for at least 30 min in a holding chamber containing the following (in mM): 126 NaCl, 2.5 KCl, 1.2 MgCl2, 2.5 CaCl2, 26 NaHCO3, 1.2 NaH2PO4, 11 glucose, 1 ascorbic acid (saturated with 95% O2 and 5% CO2). During recording, slices were superfused (2 ml/min) with this same ACSF at ~32°C but with picrotoxin (100 µṂ to block GABAA receptor-mediated synaptic currents) and without ascorbic acid.

Patch-clamp electrophysiology

Whole-cell voltage-clamp recordings from MSNs located in the nucleus accumbens (NAc) shell and dorsal striatum (DS) were obtained under visual control on a differential interference contrast, upright microscope with infrared illumination. Recordings were obtained using 3–6 MΩ resistance pipettes backfilled with internal solution containing the following (in mM): 120 CsCH3SO3, 20 HEPES, 0.4 EGTA, 2.8 NaCl, 5 N(CH2CH3)4Cl, 2.5 Mg-ATP, 0.25 Mg-GTP, pH 7.3. Currents were measured using either an Axopatch 1D or 200A amplifier (2 kHz low-pass Bessel filter) with a DigiData 1440 interface (5 kHz digitization) and pClamp software (Molecular Devices). MSNs, identified by their morphology and hyperpolarized resting membrane potential, were voltage clamped at −70 mV. An optical fiber (200 µm core diameter, 0.2 N.A.) coupled to a diode pumped solid-state 473 nm laser was placed 200 µm from the site of recording, and was used to deliver optical stimulation at a frequency of 0.1 Hz. Stimulus intensity ranged from 1 to 30 mW with a pulse duration of 5 ms. Series resistance (5–20 MΩ) was monitored on-line with a 5 mV hyperpolarizing step (50 ms) given after each stimulus.

To quantify EPSC amplitudes, 6 sweeps were collected from each cell at each light intensity (0, 1, 2, 5, 10 mW for input/output experiment, 30 mW for maximal stimulation and pharmacology experiments). Sweeps were then averaged offline to determine EPSC amplitude at a particular intensity. Light stimulus intensities were presented in a random order and counter balanced across recordings. For AMPA receptor antagonism experiments 10 µM 6,7-dinitroquinoxaline-2,3-dione (DNQX) was used, and for DA receptor antagonism experiments 2 µM SCH23390 and 2 µM raclopride were used to respectively block D1 and D2 receptors. Experimenters were blind to the animal’s genotype during data acquisition.

Fast-scan cyclic voltammetry

Methods were adapted from previous work (Stuber et al., 2008). T-650 carbon fiber microelectrodes (100 – 200 µm in length) were used for detection of dopamine. Electrodes were placed in the NAc shell or dorsal striatum of Slc6a3+/cre; Slc17a6+/lox mice. The potential applied to the electrode was ramped from −0.4 V to +1.3 V to −0.4 V vs. a Ag/AgCl reference every 100 ms at a rate of 400 V/s and resulting electrochemical data was acquired using custom written software in LabVIEW. Electrochemical data was low-pass filtered at 1 kHz offline. Background subtracted cyclic voltammograms were generated immediately after optical stimulation of the slice, and were characteristic of dopamine (peak oxidation potential 600–700 mV). Current resulting from the oxidation of dopamine (monitored at the peak oxidation potential determined from cyclic voltammograms) were then converted into concentration changes using a calibration factor of 10 nA/µM DA.


An adeno-associated virus encoding a conditional allele of the light-activated cation channel channelrhodopsin-2 (ChR2) activated by cre-mediated recombination (AAV5-DIO-ChR2-mcherry) (Tsai et al., 2009) was bilaterally injected into the ventral tegmental area (VTA) of adult mice (2–4 months old) expressing cre recombinase under the control of dopamine transporter (DAT) regulatory sequences (Zhuang et al., 2005). Four weeks after injection, immunohistochemistry revealed robust, selective expression of ChR2-mcherry in VTA dopamine neurons (Fig. 1A,B). Indeed, 99.5% of ChR2-mcherry+ neurons identified in the VTA (n = 409 cells) also labeled for the catecholamine biosynthetic enzyme tyrosine hydroxylase (TH). Furthermore, essentially all ChR2-mcherry labeled fibers and puncta in striatal terminals co-stained for TH (Fig. 1C,D).

Figure 1
Expression of ChR2-mcherry in midbrain dopamine neurons and their projections to the striatum

We then prepared coronal sections through striatal regions from injected mice and light-evoked stimulation was used to depolarize and stimulate neurotransmitter release from the dopaminergic terminals selectively expressing ChR2. Whole-cell voltage-clamp recordings from visually identified MSNs revealed that optical stimulation of dopaminergic terminals in the NAc shell resulted in excitatory postsynaptic currents >10 pA in 24/24 neurons, which increased in amplitude as a function of light stimulus intensity (Fig. 2A, B). Light-evoked EPSCs in the NAc shell were blocked by α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate receptor (AMPAR) antagonism (Fig. 3A,C). The rapid response to light suggests a monosynaptic event, but to ensure that the glutamate-mediated currents did not reflect the postsynaptic action of dopamine, we applied D1- and D2-type dopamine receptor antagonists to the bath. The addition of SCH23390 and raclopride did not significantly affect light-evoked EPSCs (Fig. 3B,C), indicating that they do not require the light-evoked release of dopamine. Rather, dopamine neurons themselves appear to release glutamate.

Figure 2
Dopaminergic terminals in the nucleus accumbens shell but not dorsal striatum corelease glutamate
Figure 3
EPSCs evoked by dopamine terminal stimulation in the nucleus accumbens shell are blocked by AMPAR antagonism but unaffected by D1R/D2R antagonism

The ability of dopamine neurons to release glutamate implies that they express a vesicular glutamate transporter. To determine whether VGLUT2 expression by dopamine neurons is required for their release of glutamate, we performed the same experiments in conditional knockout (cKO) mice [Slc6a3+/cre; Slc17a6lox/lox] where cre recombinase drives both selective excision in dopamine neurons of the gene encoding VGLUT2 (Hnasko et al., 2010) and, following virus injection, the selective expression of ChR2 in dopamine neurons. In contrast to control mice, light never evoked EPSCs in cKO mice (Fig. 2A,B), demonstrating that VGLUT2 is required for glutamate packaging and release from dopaminergic terminals.

Although previous results have suggested that mesolimbic dopamine neurons release glutamate (Chuhma et al., 2004; Hnasko et al., 2010), it is unknown whether dopamine neurons that project to the dorsal striatum (DS) also have the capacity to release glutamate. Therefore, we also examined brain slices through the DS. Optical stimulation of dopaminergic terminals in the DS resulted in ~10 fold smaller mean glutamate-mediated currents than in the NAc shell of control mice, with no detectable release in the DS from mice lacking VGLUT2 in dopamine neurons (Fig. 2C,D). Importantly, light-evoked dopamine release as measured by fast-scan cyclic voltammetry was detected in both the DS (Fig 4A,C) and NAc shell (Fig. 4B,D) of virally injected mice. Within the striatum, glutamate co-release is thus relatively restricted to dopamine terminals in the ventral striatum.

Figure 4
Optical stimulation of dopamine terminals in dorsal striatum and nucleus accumbens shell results in dopamine release


Our results provide direct physiological evidence that mature dopamine neurons projecting to the NAc shell co-release glutamate. Previous studies have suggested that mesolimbic dopamine neurons release glutamate (Chuhma et al., 2004; Lavin et al., 2005; Chuhma et al., 2009; Hnasko et al., 2010), but electrical stimulation non-specifically activates all VTA neurons, making it difficult to exclude glutamate release from non-dopaminergic neurons. Indeed, there exist VGLUT2-positive neurons in the VTA that do not co-label for TH (Kawano et al., 2006; Yamaguchi et al., 2007; Nair-Roberts et al., 2008). Although some of these neurons appear to project locally (Dobi et al., 2010), it is unknown whether they may also project to striatal and/or limbic brain regions. Our ability to drive ChR2 expression selectively in dopamine neurons of the VTA and thus to selectively activate dopamine terminals in striatal regions with light eliminates the possibility of inadvertently activating these non-dopaminergic VTA glutamate neurons. Essentially all ChR2-mcherry-expressing neurons in the midbrain and puncta in the striatum also labeled for TH by confocal microscopy (Fig. 1B,D). Additionally, no ChR2-mcherry-expressing neurons were seen in brain regions known to send glutamatergic projections to the NAc shell such as cortex, thalamus, hippocampus, amygdala, or lateral hypothalamus. Surprisingly, we did observe small clusters of ChR2-expressing neurons in both the lateral habenula and in the ventral premamillary nucleus (data not shown), but these neurons are not known to project to the NAc shell.

Due to technical limitations, earlier electrophysiological studies of glutamate co-release in the NAc shell have reported relatively small AMPAR currents limited to brain slices made from young (~3 wk old) animals (Chuhma et al., 2004; Hnasko et al., 2010). Given that VGLUT2 expression by dopamine neurons appears to upregulate in culture (Dal Bo et al., 2004; Mendez et al., 2008) and downregulate over development in vivo (Mendez et al., 2008; Berube-Carriere et al., 2009), it has remained unclear whether mature dopamine neurons also corelease glutamate. However, we now find that VGLUT2 in dopamine neurons is required for glutamate release from dopaminergic terminals in the NAc shell of adult mice, demonstrating that VGLUT2-dependent glutamate corelease persists into adulthood. Although some evidence suggests that only a small fraction of dopamine neurons express VGLUT2 in adult rodents (Yamaguchi et al., 2007), we see ubiquitous light-evoked glutamatergic currents in MSNs of the NAc shell. Therefore, it is likely that a significant number of dopamine neurons in the adult VTA express VGLUT2 at levels too low to detect by conventional means, or alternatively, a small percentage of VGLUT2-expressing dopamine neurons may send axon collaterals to essentially all MSNs in the NAc shell.

Although we were able to observe substantial ChR2-mcherry expression and light-evoked dopamine release in the DS, glutamate currents were much smaller and less frequent in the DS relative to the NAc shell. These results are consistent with a previous report (Kawano et al., 2006) showing the absence of VGLUT2 in the substantia nigra pars compacta compared to the VTA. More recently, we reported that cKO mice lacking VGLUT2 selectively in dopamine neurons show a presynaptic reduction in dopamine storage and release that is restricted to the ventral striatum (Hnasko et al., 2010). Together, these findings strongly suggest that within the striatum, VGLUT2 expression and glutamate corelease is restricted to mesolimbic projections, that may also extend to other limbic structures (Lavin et al., 2005).

In conclusion, our findings provide unequivocal evidence that a subset of dopaminergic terminals release glutamate capable of postsynaptic signaling. Mice lacking VGLUT2 in dopamine neurons show deficits in psychostimulant-induced locomotion (Hnasko et al., 2010; Wallen-Mackenzie et al., 2010), apparently due to a presynaptic role for glutamate in vesicular monoamine filling(Amilhon et al., 2010; Hnasko et al., 2010). Although we still understand little about the role for glutamate released by monoamine neurons as an independent signal, rapid, synaptic signaling mediated by glutamate may provide the temporal specificity required for reward prediction and/or incentive salience (Lapish et al., 2006; Lapish et al., 2007).


We thank Kurt Thorn and the Nikon Imaging Center at UCSF for assistance and Karl Deisseroth for the AAV-DIO-ChR2-mcherry construct. This work was supported by ABMRF and NARSAD (GDS), the A.P. Giannini Foundation (TSH) and the State of California for Medical Research on Alcohol and Substance Abuse through UCSF (AB).


  • Amilhon B, Lepicard E, Renoir T, Mongeau R, Popa D, Poirel O, Miot S, Gras C, Gardier AM, Gallego J, Hamon M, Lanfumey L, Gasnier B, Giros B, El Mestikawy S. VGLUT3 (vesicular glutamate transporter type 3) contribution to the regulation of serotonergic transmission and anxiety. J Neurosci. 2010;30:2198–2210. [PubMed]
  • Berube-Carriere N, Riad M, Dal Bo G, Levesque D, Trudeau LE, Descarries L. The dual dopamine-glutamate phenotype of growing mesencephalic neurons regresses in mature rat brain. J Comp Neurol. 2009;517:873–891. [PubMed]
  • Chuhma N, Choi WY, Mingote S, Rayport S. Dopamine neuron glutamate cotransmission: frequency-dependent modulation in the mesoventromedial projection. Neuroscience. 2009;164:1068–1083. [PMC free article] [PubMed]
  • Chuhma N, Zhang H, Masson J, Zhuang X, Sulzer D, Hen R, Rayport S. Dopamine neurons mediate a fast excitatory signal via their glutamatergic synapses. J Neurosci. 2004;24:972–981. [PubMed]
  • Dal Bo G, St-Gelais F, Danik M, Williams S, Cotton M, Trudeau LE. Dopamine neurons in culture express VGLUT2 explaining their capacity to release glutamate at synapses in addition to dopamine. J Neurochem. 2004;88:1398–1405. [PubMed]
  • Descarries L, Berube-Carriere N, Riad M, Bo GD, Mendez JA, Trudeau LE. Glutamate in dopamine neurons: synaptic versus diffuse transmission. Brain Res Rev. 2008;58:290–302. [PubMed]
  • Dobi A, Margolis EB, Wang HL, Harvey BK, Morales M. Glutamatergic and nonglutamatergic neurons of the ventral tegmental area establish local synaptic contacts with dopaminergic and nondopaminergic neurons. J Neurosci. 2010;30:218–229. [PMC free article] [PubMed]
  • Hnasko TS, Chuhma N, Zhang H, Goh GY, Sulzer D, Palmiter RD, Rayport S, Edwards RH. Vesicular glutamate transport promotes dopamine storage and glutamate corelease in vivo. Neuron. 2010;65:643–656. [PMC free article] [PubMed]
  • Joyce MP, Rayport S. Mesoaccumbens dopamine neuron synapses reconstructed in vitro are glutamatergic. Neuroscience. 2000;99:445–456. [PubMed]
  • Kawano M, Kawasaki A, Sakata-Haga H, Fukui Y, Kawano H, Nogami H, Hisano S. Particular subpopulations of midbrain and hypothalamic dopamine neurons express vesicular glutamate transporter 2 in the rat brain. J Comp Neurol. 2006;498:581–592. [PubMed]
  • Lapish CC, Seamans JK, Judson Chandler L. Glutamate-dopamine cotransmission and reward processing in addiction. Alcohol Clin Exp Res. 2006;30:1451–1465. [PubMed]
  • Lapish CC, Kroener S, Durstewitz D, Lavin A, Seamans JK. The ability of the mesocortical dopamine system to operate in distinct temporal modes. Psychopharmacology (Berl) 2007;191:609–625. [PubMed]
  • Lavin A, Nogueira L, Lapish CC, Wightman RM, Phillips PE, Seamans JK. Mesocortical dopamine neurons operate in distinct temporal domains using multimodal signaling. J Neurosci. 2005;25:5013–5023. [PubMed]
  • Mendez JA, Bourque MJ, Dal Bo G, Bourdeau ML, Danik M, Williams S, Lacaille JC, Trudeau LE. Developmental and target-dependent regulation of vesicular glutamate transporter expression by dopamine neurons. J Neurosci. 2008;28:6309–6318. [PubMed]
  • Nair-Roberts RG, Chatelain-Badie SD, Benson E, White-Cooper H, Bolam JP, Ungless MA. Stereological estimates of dopaminergic, GABAergic and glutamatergic neurons in the ventral tegmental area, substantia nigra and retrorubral field in the rat. Neuroscience. 2008;152:1024–1031. [PMC free article] [PubMed]
  • Reimer RJ, Edwards RH. Organic anion transport is the primary function of the SLC17/type I phosphate transporter family. Pflugers Arch. 2004;447:629–635. [PubMed]
  • Stuber GD, Klanker M, de Ridder B, Bowers MS, Joosten RN, Feenstra MG, Bonci A. Reward-predictive cues enhance excitatory synaptic strength onto midbrain dopamine neurons. Science. 2008;321:1690–1692. [PMC free article] [PubMed]
  • Sulzer D, Joyce MP, Lin L, Geldwert D, Haber SN, Hattori T, Rayport S. Dopamine neurons make glutamatergic synapses in vitro. J Neurosci. 1998;18:4588–4602. [PubMed]
  • Takamori S. VGLUTs: 'exciting' times for glutamatergic research? Neurosci Res. 2006;55:343–351. [PubMed]
  • Tsai HC, Zhang F, Adamantidis A, Stuber GD, Bonci A, de Lecea L, Deisseroth K. Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning. Science. 2009;324:1080–1084. [PubMed]
  • Wallen-Mackenzie A, Wootz H, Englund H. Genetic inactivation of the vesicular glutamate transporter 2 (VGLUT2) in the mouse: what have we learnt about functional glutamatergic neurotransmission? Ups J Med Sci. 2010;115:11–20. [PMC free article] [PubMed]
  • Wise RA. Dopamine, learning and motivation. Nat Rev Neurosci. 2004;5:483–494. [PubMed]
  • Yamaguchi T, Sheen W, Morales M. Glutamatergic neurons are present in the rat ventral tegmental area. Eur J Neurosci. 2007;25:106–118. [PMC free article] [PubMed]
  • Zhuang X, Masson J, Gingrich JA, Rayport S, Hen R. Targeted gene expression in dopamine and serotonin neurons of the mouse brain. J Neurosci Methods. 2005;143:27–32. [PubMed]