|Home | About | Journals | Submit | Contact Us | Français|
Extracellular signal-regulated kinase (ERK) 1/2, a well known regulator of gene expression, is likely to contribute to signaling events underlying enduring neural adaptations. Phosphorylated (phospho)-ERK was examined immunohistochemically after both single and repeated (i.e., sensitizing) doses of the partial D1-dopamine (DA) receptor agonist SKF-38393 (2,3,4,5-tetrahydro-7,8-dihydroxy-1-phenyl-1H-3-benazepine HCl) to adult rats lesioned as neonates (neonate lesioned) with 6-hydroxydopamine. Remarkably, prolonged phospho-ERK accumulated primarily in layers II–III of medial prefrontal cortex (MPC), where it declined gradually yet remained significantly elevated for at least 36 d after repeated doses of SKF-38393. Sustained (≥7 d) phospho-ERK was observed for shorter periods in various other cortical regions but was not detectable in striatum or nucleus accumbens. At 36 d, an additional injection of SKF-38393 to sensitized rats restored phospho-ERK to maximal levels only in MPC when examined 7 d later. Phosphorylated cAMP response element-binding protein (CREB), examined 7 d after the sensitizing regimen, was observed exclusively in MPC, where it was abundant throughout all layers. Systemic injections of SL327 (α-[amino[(4-aminophenyl)thio]methylene]-2-(trifluoromethyl)benzeneacetonitrile), an inhibitor of the upstream ERK activator mitogen ERK kinase, attenuated both ERK and CREB phosphorylation in layers II–III of MPC. Pretreatment with the D1 antagonist SCH-23390 ((R)-(+)-8-chloro-2,3,4,5-tetrahydro-3-methyl-5-phenyl-1H-3-benzazepine-7-OL maleate) inhibited the prolonged increase in MPC phospho-ERK, whereas the 5-HT2 receptor antagonist ketanserin (3-[2-[4-(4-fluorobenzoyl)-1-piperidinyl]ethyl]-2,4(1H,3H)-quinazolinedione tartrate) was ineffective. Competitive and noncompetitive NMDA receptor antagonists also blocked sustained ERK phosphorylation. Collectively, the present results demonstrate coupling of D1 and NMDA receptor function reflected in sustained activation of the ERK signaling pathway in MPC of SKF-38393-sensitized neonate-lesioned rats. Ultimately, long-lasting phosphorylation of ERK and CREB in MPC may play a pivotal role in any permanent adaptive change(s) in these animals.
Administration of the neurotoxicant 6-hydroxydopamine (6-OHDA), with desmethylimipramine pretreatment, results in the selective destruction of dopamine-containing neurons (Smith et al., 1973). Rats bilaterally lesioned with 6-OHDA as neonates have increased susceptibility for aggression, hyperexcitability, and self-injurious behavior in response to dopamine (DA) receptor agonists (Breese et al., 1984). These behaviors are similar to those observed clinically with Lesch–Nyhan syndrome (LNS), a developmental disorder characterized by reduced brain DA, choreoathetoid movements, and compulsive self-injurious behavior (Lesch and Nyhan, 1964). The susceptibility for self-injurious behavior observed in neonate-lesioned rats is not present in rats lesioned with 6-OHDA as adults, the latter of which mimic Parkinson's disease (Breese et al., 1984; Marsden, 1984). Thus, the neonatal rat brain must retain sufficient plasticity to allow for the development of compensatory mechanisms that attenuate, to some degree, the debilitating effects of DA depletion.
Repeated dosing with the D1-dopamine receptor agonist SKF-38393 (2,3,4,5-tetrahydro-7,8-dihydroxy-1-phenyl-1H-3-benazepine HCl) to neonate-lesioned rats results in long-lasting behavioral sensitization, which is seen as an agonist-induced increase in locomotor activity for at least 6 months after the sensitization (Breese et al., 1984; Criswell et al., 1989). Treatment with D1 receptor antagonists blocks the increasing motor responsiveness (Breese et al., 1985a,b; Criswell et al., 1989, 1990), a finding that demonstrates a requirement for D1 receptor activation in SKF-38393-mediated sensitization of these animals. Studies have shown that the sensitization is not dependent on contextual cues associated with SKF-38393 administration (Criswell et al., 1989), nor is it related to changes in density or binding affinity of D1 receptors in striatum (Breese et al., 1987).
Central to the notion that SKF-38393-mediated sensitization of neonate-lesioned rats relates to behavioral and neural plasticity is the discovery that pretreatment with the noncompetitive NMDA receptor antagonist MK-801 ((+)-5-methyl-10,11-dihydroxy-5H-dibenzo(a,d)cyclohepten-5,10-imine) abolishes this sensitization (Criswell et al., 1990). It is well established that NMDA receptor activation is critical for development of psychomotor stimulant-induced behavioral sensitization (Karler et al., 1989; Stewart and Druhan, 1993; Wolf and Jeziorski, 1993) and that NMDA receptor antagonists interfere with long-term potentiation, a model of learning and memory formation (Lynch et al., 1983; Davis et al., 1992; for review, see Riedel et al., 2003). These findings have established a critical role for NMDA receptor-dependent mechanisms that can lead to persistent cellular and behavioral adaptive changes. Thus, the finding that MK-801 antagonizes SKF-38393-mediated sensitization in neonate-lesioned rats supports consideration of this model as a type of “neuronal learning” (Criswell et al., 1989).
The discovery of specific neurobiological substrates of adaptive change(s) that accompany the enduring hyper-responsiveness of neonate-lesioned rats to repeated D1 receptor stimulation might provide insight into developmental and other disorders involving permanent DA-mediated sensitization, such as schizophrenia and psychostimulant abuse. A likely candidate, extracellular signal-regulated kinase (ERK) 1/2, is a cell-signaling molecule thought to be critical for various forms of neuroplasticity. Stimulation of D1 and NMDA receptors can recruit several second messenger systems to activate ERK (Xia et al., 1996; Sweatt, 2001). In this respect, Gerfen et al. (2002) demonstrated enhanced striatal ERK phosphorylation after acute SKF-38393 administration in unilateral adult 6-OHDA-lesioned rats. Furthermore, immediate-early gene transcription in DA-depleted striatum was ERK dependent, a finding that suggests ERK pathway involvement in adaptive changes occurring in these animals (Gerfen et al., 2002). Neuronal ERK phosphorylation has been linked to a number of other cellular signaling processes (for review, see Adams and Sweatt, 2002). Among these, activation of the transcription factor cAMP response element-binding protein (CREB) has emerged as a major regulatory mechanism for activity-dependent neuroplasticity (Lonze and Ginty, 2002).
The aim of the present study was to investigate whether repeated SKF-38393 administration to neonate-lesioned rats would result in enhanced ERK activation in the primary DA-terminal fields of the forebrain. Immunostaining for phosphorylated (phospho)-ERK revealed distinct temporal patterns in dorsal striatum (striatum), nucleus accumbens (accumbens), and cortex. The medial prefrontal cortex (MPC) was unique in that it demonstrated remarkably sustained phospho-ERK that was accompanied by increased phospho-CREB immunostaining. Moreover, the prolonged MPC phospho-ERK immunostaining was dependent on D1 and NMDA receptor coactivation. Our data strongly suggest that the sustained ERK phosphorylation observed in MPC reflects a neuroadaptive change that occurs with D1 agonist-induced sensitization of neonate-lesioned rats.
Pregnant Sprague Dawley rats obtained from Charles River Laboratories were individually housed, with Wayne Lab Blox laboratory chow and water available ad libitum.On day 3 after delivery, male and female rat pups were anesthetized with ether and then administered 100 μg (free base) of 6-OHDA intracisternally (i.c.), 60 min after desipramine (20 mg/kg, i.p.), to protect norad-renergic neurons (Breese et al., 1984). The bilateral lesion causes >90% loss of dopamine innervation into the striatum and disrupts basal ganglia– cortical system circuits (Smith et al., 1973). Some rats received desipramine (20 mg/kg, i.p.) and saline (i.c.) and served as unlesioned (sham-lesioned) controls. Rats treated with 6-OHDA or saline neonatally were weaned at day 30 and testing began at 40–60 d of age. All animal use procedures were in strict accordance with the Institutional Animal Care and Use Committee (2003), and all efforts were made to minimize the number of animals used.
Although several lines of evidence implicate gender differences in the activation of ERK and CREB (Cardona-Gomez et al., 2002; Bi et al., 2003; Wade and Dorsa, 2003), no significant SKF-38393-dependent differences in phospho-ERK or phospho-CREB immunoreactivity were found between male and female rats, regardless of neonatal lesioning (data not shown). Thus, SKF-38393-induced phospho-ERK, ERK, phospho-CREB, or CREB immunoreactivity was examined in the striatum, accumbens, MPC, and other selected cortical regions of both male and female rats.
The following sections describe all adult drug treatments, dosing regimens, and times of killing for animals used in this study. Data in supplemental Figure A (available at www.jneurosci.org) are provided as an additional guide for the experimental paradigm.
To assess the acute effects of a single dose of SKF-38393, naive neonate-lesioned and sham-lesioned rats were injected with 3 mg/kg SKF-38393 at 40–60 d of age and killed at 15 min (n = 4 per treatment group), 30 min (n = 4), 60 min (n = 4–5), 120 min (n = 5–12), and 360 min (n = 2) after agonist treatment. A separate group of adult neonate-lesioned or sham-lesioned rats were administered saline, rather than agonist, and killed at 15, 60, and 120 min to serve as controls (n = 2). Neonate-lesioned and sham-lesioned rats were also killed at 3 d (n = 4)and 7 d (n = 4–6) after this treatment regimen to examine the long-term effects of a single dose of agonist to these animals.
Neonate-lesioned rats do not show maximal sensitivity to D1 agonists unless exposed repeatedly to such agonists (Breese et al., 1985b; Criswell et al., 1989). Therefore, beginning at 40–50 d of age, neonate-lesioned rats in this treatment group received repeated treatments with SKF-38393 sufficient to allow the animals to reach a plateau of maximal behavioral supersensitivity (Criswell et al., 1989, 1990). To accomplish this sensitization process, lesioned animals were administered a total of 12 mg/kg SKF-38393, divided into three doses as follows: 6, 3, and 3 mg/kg, each spaced 1 week apart as described previously (Breese et al., 1985a,b; Criswell et al., 1989). A separate group of sham-lesioned rats received the same agonist-dosing regimen. To assess acute effects of repeated SKF-38393 administration, animals were killed at 15 min (n = 4–5 per treatment group), 60 min (n = 4–5), 120 min (n = 4), and 360 min (n = 2) after agonist treatment. To serve as controls, groups of neonate-lesioned and sham-lesioned rats were administered three consecutive injections of saline at weekly intervals and killed at 15 min (n = 2–4), 60 min (n = 2), 120 min (2–3), and 360 min (n = 2) after the final saline administration. To examine the chronic effects of repeated SKF-38393 administration, animals were killed at 3 d (n = 4–6), 7 d (n = 10–15), 14 d (n = 5–8), 21 d (n = 6–7), or 36 d (n = 5–6) after the final agonist or saline treatment. At 36 d, a separate group of previously treated neonate-lesioned and sham-lesioned rats were administered an additional dose of SKF-38393 (3 mg/kg) or saline and killed 7 d later (n = 5–6).
Immediately after the final dose of SKF-38393 or saline to neonate-lesioned or sham-lesioned rats, behavioral activity was assessed to assure maximal responsiveness of neonate-lesioned rats to the agonist. Rats were placed in a clear 17 × 17 inch computer-monitored activity chamber (Med Associates, St. Albans, VT), and horizontal, vertical, and stereotypical activity was recorded in 5 min bins overa3hr testing period. ANOVA F test of model fit for motor activity between treatment groups yielded F(3,206) = 54.90, p < 0.0001 for horizontal activity, F(3,206) = 17.68, p<0.0001 for vertical activity, and F(3,206) = 11.94, p<0.0001 for stereotypical activity. As determined by post hoc analysis using Fisher's PLSD test (mean ± SEM), neonate-lesioned rats receiving multiple SKF-38393 treatments demonstrated 120,486.24 ± 10,641.87 total horizontal counts, 2348.51 ± 263.30 total vertical counts, and 15,167.80 ± 535.27 total stereotypical counts compared with 20,681.32 ± 2357.34 total horizontal counts ( p < 0.0001), 799.09 ± 77.84 total vertical counts ( p < 0.0001), and 11,646.37 ± 476.32 total stereotypical counts ( p < 0.0001) observed for neonate-lesioned rats injected with saline, 20,389.82 ± 2751.24 total horizontal counts ( p < 0.0001), 1040.34 ± 118.37 total vertical counts ( p < 0.0001), and 12,371.19 ± 772.15 total stereotypical counts ( p = 0.0026) observed for sham-lesioned rats dosed repeatedly with SKF-38393, and 20,579.15 ± 1648.53 total horizontal counts ( p < 0.0001), 965.13 ± 71.12 total vertical counts ( p < 0.0001), and 10,864.39 ± 579.04 total stereotypical counts ( p < 0.0001) observed for sham-lesioned rats injected with saline. The novelty of the testing environment unlikely affected our results, because no significant differences were found in motor activity between the SKF-38393-sensitized neonate-lesioned animals and a separate group of neonate-lesioned animals (not used in this study) that were placed in activity chambers after each of four weekly treatments with SKF-38393 (supplemental Table A, available at www.jneurosci.org) ( p = 0.4229, horizontal activity; p = 0.3034, vertical activity; p = 0.5824, stereotypical activity indicating no difference with Fisher's PLSD test). This finding is consistent with previous data accumulated in neonate-lesioned rats showing that behavioral sensitization results in comparable levels of activity in response to SKF-38393 when the rats are repeatedly dosed in the same or a different environment from that in which the rats are finally tested (Criswell et al., 1989).
For this experimental series, SL327 (α-[amino[(4-aminophenyl)thio]methylene]-2-(trifluoromethyl) benzeneacetonitrile) (100 mg/kg), a selective inhibitor of the upstream ERK activator MEK, was administered to neonate-lesioned rats 30 min before each of three weekly doses of SKF-38393 (3 mg/kg), with the fourth weekly treatment consisting of only dimethylsulfoxide (DMSO) vehicle followed by SKF-38393 (n = 5). Control treatment groups included (1) four weekly treatments of DMSO followed by SKF-38393 (n = 4) and (2) four weekly treatments of DMSO followed by saline (n = 4). All rats thus received DMSO before the fourth and final dose of SKF or saline and were killed 7 d later.
Groups of neonate-lesioned and sham-lesioned rats each received four doses of SCH-23390 (( R)-(+)-8-chloro-2,3,4,5-tetrahydro-3-methyl-5-phenyl-1H-3-benzazepine-7-OL maleate; 0.3 mg/kg), ketanserin (3-[2-[4-(4-fluorobenzoyl)-1-piperidinyl]ethyl]-2,4(1H,3H)-quinazolinedione tartrate; 2 mg/kg), or saline 15 min before each of four weekly doses of SKF-38393 or saline. The primary treatment groups were as follows: (1) SCH-23390 followed by SKF-38393 (n = 6–10) and (2) ketanserin followed by SKF-38393 (n = 4). Control treatment groups included (1) saline followed by SKF-38393 (n = 2–10), (2) SCH-23390 followed by saline (n = 2–10), and (3) saline followed by saline (n = 4–5). Rats were killed 7 d after the final drug treatment.
For the noncompetitive NMDA receptor antagonist MK-801, male rats were dosed with 0.3 mg/ kg, whereas females received 0.14 mg/kg because female rats have much greater responsiveness to the motor effects of MK-801 (Fleischmann et al., 1991; Blanchard et al., 1992; Honack and Loscher, 1993; Haggerty and Brown, 1996; Frantz and Van Hartesveldt, 1999). MK-801, the competitive NMDA receptor antagonist CGS-19755 (cis-4-(phosphonomethyl)-2-piperidinecarboxylic acid; 10 mg/kg), or saline was administered 15 min before four weekly doses of SKF-38393 (3 mg/kg) or saline. The two primary treatments were (1) MK-801 followed by SKF-38393 (n = 13 per treatment group) and (2) CGS-19755 followed by SKF-38393 (n = 6–18 per treatment group). Control treatments included (1) saline followed by SKF-38393 (n = 5), (2) MK-801 followed by saline (n = 6–8), (3) CGS-19755 followed by saline (n = 6–15), and (4) saline followed by saline (n = 7–10). Rats were killed 7 d after the final drug treatment.
6-OHDA hydrobromide (ICN Biochemicals, Irvine, CA) was dissolved in saline containing 0.5% ascorbic acid and administered intracisternally. Desipramine hydrochloride (Sigma, St. Louis, MO), SKF-38393 (Sigma), SCH-23390 (Schering Corporation, Bloomfield, NJ), ketanserin (Sigma), MK-801 (a gift from Merck, Rahway, NJ), and CGS-19755 (a gift from CIBA-GEIGY Corporation, Summit, NJ) were dissolved in saline and administered intraperitoneally. The MEK inhibitor SL327 (kindly provided by DuPont Pharmaceuticals Company, Boston, MA) was dissolved in 100% DMSO and administered intraperitoneally (2 ml/kg) (Atkins et al., 1998; Selcher et al., 1999; Yamagata et al., 2002).
Rats were deeply anesthetized with an overdose of sodium pentobarbital (100 mg/kg), perfused transcardially for 4 min with PBS (150 mm NaCl, 100 mm sodium phosphate, pH 7.4) followed by 7 min of 4% phosphate–buffered paraformaldehyde (100 mm sodium phosphate), and their brains were collected and postfixed for 24 hr at 4°C. Forty-micrometer-thick sections were then cut with a vibrating microtome for the immunohistochemistry. Standard avidin–biotin–horseradish peroxidase methods were used as described previously (Knapp et al., 1998, 2001). After rinsing in fresh PBS three times (10 min each), free-floating tissue sections were blocked in 10% normal goat serum and 0.2% Triton X-100 in PBS for 1 hr. Affinity-purified polyclonal tyrosine hydroxylase (TH) (1:4000; Calbiochem, La Jolla, CA), phospho-ERK [phospho-p44/42 MAP kinase (thr202/tyr204); 1:500; Cell Signaling Technology, Beverly, MA], phospho-CREB, and CREB (both 1:500; Cell Signaling Technology) were used to detect protein expression. All sections were incubated in 3% normal serum, 0.2% Triton X-100, and antisera for 48–72 hr at 4°C with agitation. An antibody-blocking peptide to phospho-ERK containing phosphorylated amino acid residues threonine 202 and tyrosine 204 (Santa Cruz Biotechnology, Santa Cruz, CA) was used to verify specificity of the antibody. A 10-fold higher concentration of the blocking peptide was incubated with phospho-ERK primary antibody at room temperature for 30 min and then incubated with selected brain tissue as described above. Tissue sections were further processed using Vectastain Elite ABC kits (Vector Laboratories, Burlingame, CA) per the manufacturer's instructions with immunochemical detection using nickel-cobalt intensification of the diaminobenzidine reaction product. For analysis, great care was taken to match sections through the same region of brain at the same level. All visible positive nuclei or cell bodies within a 10 × 10 eyepiece reticule field were counted and expressed as number of cells per square millimeter of tissue for each brain site. For a single brain site, counts were averaged from three sections from each animal.
For Western blotting, rats were killed by decapitation, their brains were rapidly removed from the skulls, and the MPC and striatum were dissected on ice and stored at −80°C until use. Tissues were homogenized by sonication in solubilization buffer (10 mm Tris-Cl, 50 mm NaCl, 1% Triton X-100, 30 mm sodium pyrophosphate, 50 mm NaF, 5 nm ZnCl2, 100 μm Na3VO4,1 mm DTT, 5 nm okadaic acid, 2.5 μg aprotinin, 2.5 μg pepstatin, and 2.5 μg leupeptin). Insoluble material was removed by centrifugation (13,000 rpm for 20 min at 4°C), and protein concentration was determined using a BCA protein assay kit (Pierce, Rockford, IL). Samples were mixed with Novex 2× Tris-glycine SDS sample buffer (San Diego, CA) containing 5% 2-mercaptoethanol and heated to 90°C for 3 min. Aliquots of 20 μg of protein per lane were separated on 8–16% gradient Tris-glycine gels (Novex) under reducing conditions using the Novex Xcell II minicell apparatus. Proteins were transferred to polyvi-nylidene difluoride membranes (Immobilon-P, Millipore, Bedford, MA). Membranes were incubated in PBS with 0.05% Tween 20 (PBS-T), containing 1% milk powder for 2 hr at room temperature to block non-specific binding. Blots were probed with an antibody corresponding to the inactive form of ERK (p44/42 MAPK; 1:1000; Cell Signaling Technology) or to phospho-ERK (1:1000; Cell Signaling Technology), followed by goat anti-rabbit IgG conjugated with horseradish peroxidase (Chemicon, Temecula, CA) at a 1:20,000 dilution in blocking solution for 60 min. Membranes were then washed three times with PBS-T. Bands were detected using enhanced chemiluminescence (Pierce) apposed to x-ray film under nonsaturating conditions and analyzed by densitometric measurements using NIH Image 1.57 (public domain software developed by the National Institutes of Health and available at http://rsb.info.nih.gov/nih-image). Data are representative of four animals per treatment group and normalized on the basis of estimates obtained in the samples from sham-lesioned, saline-treated controls.
Details of brain region identification and cell counting strategy have been reported previously in our laboratory (Knapp et al., 1998). Average counts for a specific brain region for each animal were grouped by treatment and averaged for each time point to obtain the mean counts per square millimeter ± SEM for that brain site. Average cell counts for each defined brain region were compared within and between treatment groups using ANOVA. Statistical comparisons between control groups versus lesioned groups administered either SKF-38393 or saline were followed by post hoc tests. A more conservative significance level was set at p = 0.01 for all time-course experiments to correct for multiple comparisons, whereas the significance level for all other experiments was set at the traditional 0.05 level. The Fisher's PLSD test was performed when comparing combinations of means.
Previous studies (Smith et al., 1973; Breese et al., 1985a,b) documented that DA content within the striatum, accumbens, and cortex is drastically reduced in adult rats lesioned as neonates with 6-OHDA. Such loss of DA-containing terminals in the striatum of neonate-lesioned rats is illustrated in Figure 1 by the >90% reduction in TH immunoreactivity compared with the TH level in a sham-lesioned control animal. Because TH is the rate-limiting enzyme in the biosynthesis of dopamine (Nagatsu et al., 1964), TH immunoreactivity was determined for all lesioned and control animals in this investigation to establish that the neonate 6-OHDA lesioning induced an adequate loss of DA-containing neurons.
Phospho-ERK was evaluated in the primary forebrain DA-terminal regions (i.e., striatum, accumbens, and MPC) of neonate-lesioned rats at various time points after D1 agonist administration. The initial 15 min time point was chosen on the basis of Gerfen et al. (2002), who demonstrated maximum phospho-ERK immunoreactivity in striatum of unilateral adult-lesioned rats 15 min after a single SKF-38393 administration. In the present study, distinctive patterns of time-dependent ERK phosphorylation were observed across these brain regions in response to SKF-38393.
In the striatum, both single and repeated SKF-38393 treatments to neonate-lesioned rats produced acute, transient increases in phospho-ERK that peaked at 15 min after drug administration (Fig. 2A,D,E). Although slightly more phospho-ERK-positive cells were observed at 15 min in naive neonate-lesioned rats compared with rats sensitized with the agonist, the difference was not significant ( p = 0.2096). ERK activation was absent in the striatum by 60 min in neonate-lesioned rats administered a single dose of SKF-38393; however, phospho-ERK immunoreactivity remained significantly elevated at this time point in rats administered repeated doses of SKF-38393 ( p < 0.0001). This prolongation of ERK phosphorylation appeared to occur most prominently in the dorsomedial and dorsolateral quadrants of the striatum, with lighter staining occurring ventrally (Fig. 2B). Phospho-ERK was no longer detected at 120 min in striatum of SKF-38393-treated neonate-lesioned rats. No phospho-ERK immunoreactivity was detected in striatum of neonate-lesioned rats treated with saline or in sham-lesioned rats treated with single or repeated doses of SKF-38393 or saline at any of the time points examined (Fig. 2A,C,F).
In accumbens, a single dose of SKF-38393 to drug-naive neonate-lesioned animals or controls failed to produce significant change in phospho-ERK at any time point (Fig. 3A,C,E,F). Neonate-lesioned rats administered repeated doses of SKF-38393, however, exhibited significant phospho-ERK immunoreactivity at acute time points before 120 min (Fig. 3A,B,D) compared with neonate-lesioned animals administered saline ( p < 0.0001, 15 min; p < 0.0001, 60 min). This enhanced ERK phosphorylation was observed specifically in the rostral pole of the accumbens and was not present in the caudal shell or core of neonate-lesioned rats receiving repeated doses. Although the time course of accumbens phospho-ERK immunoreactivity was similar to that of striatum, total phospho-ERK-positive cell counts in the accumbens were ~20–40% of striatum, and the intensity of phospho-ERK immunostaining in the accumbens was much lower. It is unlikely that these effects were artifacts of tissue processing, because representative sections of the striatum and accumbens were always stained concurrently.
A baseline of ~18–25 phospho-ERK-positive cells per square millimeter was observed consistently in MPC and did not significantly differ among the sham or saline-treated lesioned groups (Fig. 4A). Neonate-lesioned rats that received a single dose of SKF-38393 demonstrated a transient (0–60 min) increase in ERK phosphorylation over baseline in MPC, which was similar in profile and duration to the striatum of these animals (compare Figs. Figs.2A2A,,4A).4A). Phospho-ERK immunoreactivity in this treatment group was significantly elevated over control groups only at 15 min after agonist administration ( p < 0.0001 for Lesioned SKF vs control groups).
Among the dopamine-rich regions, MPC displayed a unique, highly prolonged increase in phospho-ERK in neonate-lesioned rats previously administered repeated doses of SKF-38393. In these animals, phospho-ERK-immunoreactive cell counts were increased four- to sixfold, primarily within layers II–III of the MPC. These phospho-ERK-positive counts remained elevated over the course of the examination (up to 360 min) (Fig. 4A–H) ( p = 0.0085 at 15 min, p = 0.0015 at 60 min, p < 0.0001 at 120 min for Lesioned R-SKF vs Lesioned SKF groups; p < 0.01 at 15, 60, and 120 min, and p < 0.02 at 360 min for Lesioned R-SKF vs other control groups). The conspicuous persistence of ERK hyperphosphorylation in MPC, but not in striatum or accumbens, suggests that different processes of neuroadaptation occur among these brain regions and, furthermore, that the adaptive changes occurring in the MPC are more enduring.
Because the persistent increase in ERK phosphorylation was observed for up to 360 min in MPC of neonate-lesioned rats sensitized with SKF-38393, an additional evaluation of the time course of the sustained ERK response in this region was performed. As shown in Table 1, control groups did not significantly differ from each other across time points, with phospho-ERK-positive cell counts remaining at the baseline level of ~18–25 cells per square millimeter. Neonate-lesioned rats sensitized with repeated doses of SKF-38393 demonstrated a sustained increase in phospho-ERK immunoreactivity in MPC that was significantly elevated over neonate-lesioned rats injected with saline (as well as sham-lesioned control groups) across all extended time points ( p < 0.0001 at 3, 7, and 14 d; p = 0.0109 at 21 d; p = 0.0025 at 36 d comparing R-Saline and R-SKF groups). Prominent phospho-ERK in MPC of these animals persisted for >7 d after the final agonist treatment (Table 1). Beyond 7 d, sensitized phospho-ERK immunoreactivity began to diminish, yet remained significantly elevated from that observed in saline-treated neonate-lesioned and sham-lesioned rats for up to 36 d ( p < 0.01 for all comparisons).
As shown in Table 1, an additional dose of SKF-38393 administered at 36 d to previously sensitized neonate-lesioned rats resulted in significantly elevated levels of phospho-ERK immunoreactivity 7 d later (i.e., at 42 d) compared with levels observed at 36 d ( p = 0.0001). Phospho-ERK-positive cell counts in these animals were nearly identical to those found in neonate-lesioned rats killed 3 and 7 d after the initial SKF-38393 dosing regimen ( p ≥ 0.1110, indicating no significant difference). In neonate-lesioned rats injected repeatedly with saline, an additional dose of saline at 36 d did not result in elevated phospho-ERK immunoreactivity at 7 d after injection (data not shown), nor did a single dose of SKF-38393 to neonate-lesioned or sham-lesioned rats (Table 1). Sham-lesioned rats that received saline treatment at 36 d also did not exhibit a significant increase in phospho-ERK immunoreactivity at 7 d after injection (data not shown).
To ascertain that the observed immunohistochemical staining was phospho-ERK and not another phosphoprotein or nonspecific reaction, immunoblotting was performed with protein isolated from MPC of lesioned rats administered repeated doses of SKF-38393 or injected with saline (data not shown). Blots probed with the same concentration of phospho-ERK antibody used for immunohisto-chemistry revealed the typical doublet of bands near 42 and 44 kDa related to phospho-ERK1 and phospho-ERK2, respectively. No other bands were visualized, a finding that provides strong evidence that the elevated immunohistochemical staining observed in MPC of rats dosed repeatedly with SKF-38393 was specific for phospho-ERK.
A plausible explanation for our findings with sustained phospho-ERK would be that total ERK protein levels were elevated chronically in the neonate-lesioned animals after repeated exposure to SKF-38393. Therefore, we examined whether sensitization of neonate-lesioned rats to SKF-38393 would produce long-lasting changes in total ERK protein in MPC. Because immunohistochemistry for unphosphorylated ERK failed to produce a consistent signal, total ERK protein was analyzed by SDS-PAGE and immunoblotting of equally loaded total proteins from MPC and striatum (blot not shown). No significant differences were found for optical density measurements (mean ± SEM) of total ERK levels among neonate-lesioned rats repeatedly dosed with SKF-38393 (MPC, 101.86 ± 5.42; striatum, 91.0 ± 5.59), sham-lesioned rats repeatedly dosed with SKF-38393 (MPC, 95.60 ± 4.75; striatum, 91.56 ± 1.96), saline-treated neonate-lesioned rats (MPC, 105.30 ± 3.30; striatum, 94.00 ± 1.12), or saline-treated sham-lesioned rats (MPC and striatum standardized to 100%) at day 7 after the final injection of agonist (ANOVA F test of model fit: F(3,12) = 1.040, p = 0.4102 for MPC; F(3,12) = 1.865, p = 0.1893 for striatum). Because total ERK levels were not changed by the repeated D1 agonist treatments in neonate-lesioned rats, the sustained increase in phospho-ERK immunoreactivity in MPC could not be attributed to a sustained increase in total ERK protein.
Because of the extraordinarily sustained phospho-ERK response observed in MPC, we examined other cortical areas of neonate-lesioned rats dosed repeatedly with agonist to determine whether ERK phosphorylation was prolonged in these regions. (Fig. 5, Table 2). Phospho-ERK immunostaining was elevated in ventrolateral orbital cortex (VLOC) and in layer II–III cells of cingulate (CgC), motor (MC), somatosensory (SSC) and piriform (PirC) cortices when examined at day 7 after the final agonist treatment. As shown in Table 2, levels of phospho-ERK immunoreactivity for each cortical region were significantly higher at day 7 than levels observed at days 14–21 and 36 after the agonist treatment ( p < 0.01 for all comparisons). Levels at day 7 after agonist treatment were also significantly higher than levels observed at all time points in saline-treated neonate-lesioned rats and sham-lesioned controls for each cortical region examined (Fig. 5) ( p < 0.01 for all comparisons). Although apparent phospho-ERK-positive cell counts were still elevated (although not statistically significant) at 14–21 d after repeated agonist treatment, phospho-ERK in these cortical regions was clearly reduced by day 36. An additional dose of SKF-38393 at day 36 restored some phospho-ERK immunoreactivity in these cortical regions when examined 7 d later, but these levels were not significantly different from levels observed at day 36 before the additional dose (Table 2). Likewise, these levels were not different from those observed in saline-treated neonate-lesioned rats killed at day 36 or in saline-treated neonate- or sham-lesioned rat groups administered a single dose of SKF-38393 at day 36 and killed 7 d after injection (data not shown). In contrast to cortex, neither the striatum nor the accumbens of SKF-38393-sensitized rats exhibited phospho-ERK immunoreactivity at 7 d after the final sensitizing dose of agonist (Fig. 5).
Together, these data demonstrate that sustained phospho-ERK immunoreactivity occurs in multiple areas of the cortex, but not in the striatum or accumbens, of D1 agonist-sensitized neonate-lesioned rats. These changes are most remarkable in MPC, where ERK phosphorylation lingers longest and remains most sensitive to full reinstatement with an additional dose of agonist. That the ERK pathway could respond to repeated D1 agonist stimulation in such a prolonged and regionally selective manner suggests that specific neuroadaptive changes accompany D1 sensitization in the neonate-lesioned rat. With respect to the MPC, neuroadaptive changes in this region are thought to play a role in diseases related to dopaminergic dysfunction such as schizophrenia, attention-deficit–hyperactivity disorder, and drug abuse (Laruelle, 2000; Vanderschuren and Kalivas, 2000; Steketee, 2003; Sullivan and Brake, 2003). Because behavioral alterations in the neonate-lesioned rat have been suggested to model each of these diseases (Schwarzkopf et al. 1992; Stevens et al., 1996; Fahlke and Hansen 1999; Moy and Breese 2002; Davids et al., 2003), we chose to focus most of the remainder of our study on exploring this phenomenon in the MPC.
Phosphorylation of ERK can activate the transcription factor CREB to affect long-term gene expression (Lonze and Ginty, 2002). Thus, phosphorylated CREB potentially represents a functional product of the sustained ERK response after repeated dosing with SKF-38393 to neonate-lesioned rats. When sections from the same animals were immunostained concurrently, we observed a robust increase in phospho-CREB immunoreactivity accompanying phospho-ERK in MPC at day 7 after repeated SKF-38393 administration (Fig. 6A). To our surprise, however, increased phospho-CREB immunoreactivity was observed not only in layers II–III, but throughout all MPC layers (Fig. 6B,F). When quantified across all MPC layers, neonate-lesioned animals repeatedly dosed with SKF-38393 demonstrated approximately a four- to fivefold increase in phospho-CREB-positive cell counts (mean ± SEM; 795.07 ± 137.81) compared with neonate-lesioned rats injected repeatedly with saline (146.50 ± 40.28; p < 0.0001) and sham-lesioned rats treated repeatedly with SKF-38393 (127.13 ± 12.21; p = 0.0002) or saline (226.42 ± 78.00; p < 0.0003). Cell counts specifically in layers II–III in the agonist-sensitized rats exhibited nearly a ninefold increase in phospho-CREB-positive cell counts compared with control animals (Fig. 6A)( p < 0.0001 for Lesioned R-SKF vs Lesioned R-Saline, Sham R-SKF, or Sham R-Saline). The elevation in number of phospho-CREB-positive cells was not reflective of increased total CREB protein expression (mean values ± SEM were 343.75 ± 8.68 for Lesioned R-SKF, 319.67 ± 8.56 for Lesioned R-Saline, and 329.38 ± 16.55 for Sham R-Saline; ANOVA F test of model fit: F(2,13) = 1.475; p = 0.265). Notably, only scattered phospho-CREB-positive cells were noted throughout the striatum and accumbens at day 7, regardless of drug treatment (Fig. 6H,I). Sustained CREB phosphorylation was not observed in any other cortical regions examined (ANOVA F test of model fit: F(3,19) = 1.181, p = 0.345 for VLOC; F(3,19) = 0.575, p = 0.639 for CgC; F(3,19) = 1.086, p = 0.380 for MC; F(3,19) = 1.051, p = 0.394 for SSC; and F(3,19) = 0.506, p = 0.683 for PirC). Thus phospho-CREB immunolabeling, unlike phospho-ERK, was restricted to MPC. Furthermore, the multilamellar expression of phospho-CREB did not mirror the predominant appearance of phospho-ERK in layers II–III, suggesting that some, although perhaps not all, of the phospho-CREB immunolabeling was caused by mechanisms other than activation of ERK.
The involvement of MEK in sustained ERK phosphorylation was demonstrated using systemic injections of SL327, a MEK inhibitor that crosses the blood–brain barrier (Atkins et al., 1998; Selcher et al., 1999; Yamagata et al., 2002). Phospho-ERK immunoreactivity was suppressed in neonate-lesioned rats pretreated with SL327 before each SKF-38393 administration, compared with those pretreated with vehicle (Fig. 7A–C)( p < 0.0001). The number of phospho-ERK-positive cells in SL327-pretreated rats did not differ significantly from rats treated with vehicle alone. These data provide further evidence that the elevated immunostaining of MPC cells observed in SKF-38393-sensitized neonate-lesioned rats is indeed represented by MEK-dependent ERK phosphorylation.
We also examined the effects of SL327 pretreatment on phospho-CREB immunolabeling in MPC. Phospho-CREB immunoreactivity was reduced to some extent throughout all layers of MPC by pretreatment with SL327 (Fig. 7D,F)( p = 0.0238 for SL R-SKF vs R-SKF). The most dramatic suppression of phospho-CREB immunoreactivity by the MEK inhibitor, however, was produced in layers II–III (Fig. 7D–F)( p < 0.0001 for SL R-SKF vs R-SKF), in which cell counts did not significantly differ from rats that were not given the agonist-sensitizing regimen. These findings are consistent with direct involvement of both MEK and ERK activation in sustained CREB phosphorylation in layers II–III, but perhaps indirect, if any, involvement in CREB phosphorylation in deeper layers.
To determine whether the sustained phospho-ERK observed in MPC of neonate-lesioned rats sensitized with SKF-38393 was dependent on D1 receptors, rats were pretreated with the D1 antagonist SCH-23390. As shown in Figure 8, A and C, SCH-23390 pretreatment to neonate-lesioned rats abrogated phospho-ERK immunoreactivity at this brain site ( p < 0.0001 compared with Lesioned R-SKF group). The number of phospho-ERK-positive cells for this treatment group did not differ significantly from neonate-lesioned rats administered repeated saline injections or from sham-lesioned control groups. Moreover, repeated SCH-23390 treatment on its own did not enhance phospho-ERK immunoreactivity in either neonate-lesioned or sham-lesioned rats.
Rats were also pretreated with the 5-HT2 receptor antagonist ketanserin to control for the antagonist binding properties of SCH-23390 at 5-HT2 receptors (Fig. 8A) (Bischoff et al., 1988; McQuade et al., 1988). Ketanserin was ineffective at reducing the persistently increased levels of phospho-ERK ( p = 0.578 compared with Lesioned R-SKF group), indicating that the prolonged phospho-ERK observed in MPC of SKF-38393-sensitized neonate-lesioned rats is dependent on activation of D1 receptors.
Previous work in our laboratory has demonstrated that MK-801 inhibits SKF-38393-induced behavioral sensitization of neonatelesioned rats, a finding that implicates NMDA receptor function in adaptive processes underlying repeated SKF-38393 treatment to these animals (Criswell et al., 1990). In the present study, administration of MK-801 or CGS-19755 to neonate-lesioned and sham-lesioned animals before repeated injections of saline had no effect on basal phospho-ERK immunoreactivity (Fig. 9E). On the other hand, pretreatment with MK-801 or CGS-19755 before repeated doses of SKF-38393 to neonate-lesioned rats resulted in marked inhibition of sustained phospho-ERK immunoreactivity in MPC (Fig. 9A,C) ( p < 0.0001 for Lesioned MK-801 R-SKF group vs Lesioned R-SKF group, and for Lesioned CGS R-SKF group vs Lesioned R-SKF group). Consequently, these data point to an involvement of NMDA receptors in the sustained phospho-ERK response in MPC of neonate-lesioned animals.
This study demonstrates the remarkably protracted course of ERK phosphorylation in MPC and other cortical regions of neonate 6-OHDA-lesioned rats behaviorally sensitized to the effects of a D1 agonist in adulthood. Our observations suggest that prolonged MPC phospho-ERK is a neurobiological substrate of long-lasting adaptive change in these animals, as indicated by several key findings. First, MPC is unique among cortical and striatal regions, in that sustained ERK phosphorylation observed primarily in layers II–III declined gradually, yet remained significantly above control levels for at least 36 d. Second, maximal levels of phospho-ERK were restored fully in MPC, but not significantly in other cortical areas, on day 42 after an additional dose of agonist on day 36. Furthermore, an analogous increase in phospho-CREB in layers II–III of MPC, but not other cortical regions, at 7 d after the initial sensitizing regimen was MEK dependent. This finding suggests that a functional effect of sustained MPC ERK phosphorylation is activation of CREB-dependent gene transcription. Finally, the development of sustained phospho-ERK in MPC requires both NMDA and D1 receptor stimulation, a pattern of dependence resembling that identified in several cellular and behavioral neuroadaptive paradigms that have been described in cortical and striatal systems (for review, see Kelley and Berridge, 2002). These findings, together with the proposed contribution of the MPC in the progressive and enduring behavioral effects of drugs of abuse (Castner and Goldman-Rakic, 2003; Steketee 2003), schizophrenia (Laruelle, 2000; Tzschentke, 2001), and learning and memory (Castner et al., 2000; Elzinga and Bremner, 2002), suggest that the D1 agonist-sensitized neonate-lesioned rat presents an excellent model for the study of dopamine-dependent neuroadaptations and their functional consequences.
Several studies have demonstrated that activation of D1 receptors, via a protein kinase A (PKA)-dependent mechanism, can phosphorylate ERK to produce adaptive changes in brain (Vossler et al., 1997; Yao et al., 1998; York et al., 1998; Valjent et al., 2000). In unilateral adult 6-OHDA-lesioned rats, Gerfen et al. (2002) demonstrated transient ERK activation in striatum after acute D1 agonist administration, an effect attributed to sensitized D1 receptor-dependent responses. Consistent with this finding, the present study demonstrates a transient phosphorylation (<60 min) of ERK induced by a single dose of SKF-38393 in drug-naive neonate-lesioned rat striatum and MPC. Notably, the pattern and duration of ERK phosphorylation in striatum of naive neonate-lesioned rats administered a single dose of SKF-38393 were similar to that published by Gerfen et al. (2002) in unilateral adult 6-OHDA-lesioned rats. In agreement with the transient ERK phosphorylation, previous work has shown that a single dose of SKF-38393 can induce Fos expression in striatum of neonate-lesioned rats (Johnson et al., 1992). Together with the findings of Sgambato et al. (1998) and Gerfen et al (2002), it appears that transient ERK activation can drive immediate-early gene induction in striatum. It remains a question whether transient phospho-ERK observed in the striatum and MPC of neonate-lesioned rats after a single dose of SKF-38393 is sufficient to drive long-term gene regulation.
Several lines of evidence suggest that prolonged ERK activation (>60 min) is necessary to allow sufficient time for translocation of ERK to the nucleus, long-term gene transcription, and subsequent enhancement of enduring plasticity-associated changes in brain (for review, see Marshall, 1995). Repeated D1 agonist administration prolonged striatal phospho-ERK immunoreactivity nearly twofold. Conversely, in accumbens, ERK activation was absent with a single dose of agonist and was observed only after repeated treatment with SKF-38393. These findings, together with the prolonged presence of ERK phosphorylation in the MPC and various other regions of cortex (VLOC, CgC, MC, SSC, and PirC), point to differing mechanisms of adaptation among these regions. Furthermore, the reinstatement of ERK phosphorylation that occurs only in MPC supports a unique mechanism of adaptation in this region that can trigger rapidly the previous level of ERK activation on reexposure to the agonist.
A well characterized nuclear target for ERK action is the transcription factor CREB (Bourtchuladze et al., 1994; Yin et al., 1994), which is thought to play a central role in long-term plastic changes in brain by controlling the transcriptional expression of several genes (for review, see Curtis and Finkbeiner, 1999). In the present study, immunostaining for phospho-CREB was measured initially as a functional endpoint for sustained ERK activation in MPC. Surprisingly, although phospho-CREB emerged in parallel with ERK phosphorylation in layers II–III, the population of immunolabeled cells was more dense and widespread than those labeled with anti-phospho-ERK. In fact, phospho-CREB immunoreactivity appeared throughout all layers of MPC. At this time, we can only speculate as to the reasons for this seeming incongruence. One obvious possibility is that mechanisms independent of ERK drive CREB phosphorylation. Although phospho-CREB in layers II–III was sensitive to the MEK (and thus ERK) activation state, labeling in the deeper layers of MPC was less responsive to MEK inhibition. Because CREB is a common substrate of multiple kinase pathways in cell model systems (for review, see Herdegen and Leah, 1998; Curtis and Finkbeiner, 1999), other pathways, such as the PKA, CaM kinase, and stress-activated protein kinase cascades, might be responsible for the phospho-CREB effect. Another possibility could be related to functional integration of phospho-ERK-expressing neurons with nonexpressing cells, whereupon ERK-dependent activity in a small population of MPC layer II–III neurons could influence activation, signaling, and appearance of phospho-CREB in cortical and subcortical cells beyond simply those that contain phospho-ERK. The mild reduction in phospho-CREB immuno-reactivity observed throughout all MPC layers after MEK inhibition would seem to support a mechanism involving intercellular integration. Nevertheless, additional studies are necessary to resolve these complex issues.
Several lines of evidence suggest that sustained ERK activation may be indicative of neuronal insult, possibly leading to cell death (Stanciu et al., 2000; Kulich and Chu, 2001). Prolonged ERK activation in neurodegenerative processes has been linked to decreased CREB activation (Lee et al., 2002; Trentani et al., 2002). In the present study, sustained ERK phosphorylation in MPC was accompanied by increased, rather than decreased, CREB phosphorylation. On the opposite end of the spectrum, prolonged ERK activation accompanied by an increase in CREB has been linked to a continuum of processes that include cell survival and plasticity (Shen et al., 2001; Sweatt, 2001; Dawson and Ginty, 2002). Thus, it appears that the precise kinetics of ERK and CREB phosphorylation can ultimately determine the fate of a cell within a given region. Although neurodegenerative processes cannot be ruled out, we speculate that the sustained increases in ERK and CREB phosphorylation observed herein are likely to be indicative of cell survival and plasticity in MPC.
The sustained increase in phospho-ERK was dependent on D1 receptor function, because SCH-23390 antagonist pretreatment blocked the prolonged response to repeated administration of the D1-selective partial agonist SKF-38393. It is unlikely that SCH-23390 inhibited prolonged ERK phosphorylation through its interaction with 5-HT2 binding sites (Bischoff et al., 1988; McQuade et al., 1988), because systemic injections of the nonselective 5-HT2 antagonist ketanserin, before repeated doses of SKF-38393, had no effect on the sustained phospho-ERK response in MPC. In addition, the results with ketanserin further support the notion that long-lasting ERK phosphorylation does not drive expression of locomotor responsiveness to D1 agonists in SKF-38393-sensitized animals. The spontaneous hyperactive behavior of neonate-lesioned animals has been linked to serotonergic mechanisms (Bishop et al., 2004), and ketanserin blocks the locomotor sensitization observed in these animals (our unpublished data), without affecting the sustained activation of ERK in MPC.
Studies have demonstrated that pharmacological stimulation of NMDA receptors leads to activation of ERK in cortical neurons (Bading and Greenberg, 1991; Xia et al., 1996; Arvanov et al., 1997; Vanhoutte et al., 1999; Wang and O'Donnell, 2001). In this investigation, both the competitive NMDA antagonist MK-801 and the noncompetitive antagonist CGS-19755 eliminated sustained ERK phosphorylation in MPC, a finding suggestive of NMDA dependence for the ERK phosphorylation. Thus, MPC of neonate-lesioned rats sensitized with repeated doses of SKF-38393 might undergo persistent biochemical adaptations similar to other biological substrates of neuroplasticity such as long-term potentiation (Impey et al., 1998) and memory processing (Adams and Sweatt, 2002). In addition, accumulated evidence demonstrates an interaction of D1 and NMDA receptor functions (Konradi et al.; 1996, Pei et al., 2004) (for review, see Adriani et al., 1998; Salter, 2003) and dopamine modulation of glutamate neurotransmission within the prefrontal cortex (Goldman-Rakic and Selemon, 1997; Gonzalez-Islas et al., 2003; Otani et al., 2003; Sesack et al., 2003). The present findings further illustrate the presence of a dopamine– glutamate interaction in the MPC, whereby repeated D1 agonist administration maintains NMDA receptor-mediated responses in cells as reflected in ERK phosphorylation.
The neuromolecular mechanisms underlying sustained ERK hyperphosphorylation remain to be determined. Recent studies have established a critical role for protein phosphatases in coordinating neurotransmitter signaling (for review, see Greengard, 2001). Because phosphatases are presumed to rapidly deactivate phosphorylated proteins in brain, the measured increase in cells exhibiting phospho-ERK after SKF-38393 treatment could be related to altered phosphatase activity. Mitogen-activated protein kinase phosphatases (MKPs) 1–3 can directly control nuclear accumulation and persistent activation of ERK (for review, see Pouyssegur et al., 2002). Interestingly, MKP1 and MKP3 expression in frontal cortex and other brain regions is differentially altered by acute and chronic methamphetamine administration to rats (Takaki et al., 2001). Thus, a reduction in MKP1–3 or a related phosphatase might be a means by which ERK phosphorylation could be sustained for an extended period. The answer to this puzzle will have to be resolved in future experiments that examine sustained ERK phosphorylation after SKF-38393-mediated sensitization of neonate-lesioned animals.
This work was supported by National Institutes of Health Grants MH01896 and HD39795 and by the Lesch–Nyhan Foundation. We thank Edna Titus, Chris Halloran, and Carolyn Fleck for their expert technical assistance and Dr. William Pan for his statistical advice.