The corticotropin-releasing factor (CRF) system is a peptide family [CRF and urocortins (Ucns)] and their G protein-coupled receptors (CRF₁ and CRF₂) implicated in stress responses. Since CRF was isolated, three mammalian prohormones that contain CRF-like sequences were identified (Ucn 1, 2, and 3). CRF is a potent CRF₁ agonist, but it is a lower affinity CRF₂ agonist. Ucn 1 is a potent agonist for both subtypes. In contrast, Ucn 2 and Ucn 3 have much greater potency at membrane CRF₂ than CRF₁ receptors. CRF and Ucn 1 also bind the corticotropin-releasing factor binding protein (CRF-BP) and soluble CRF_2(a) receptor; Ucn 2 and Ucn 3 bind these proteins with less (Ucn 2) or negligible (Ucn 3) affinity. These pharmacological and related primary structure differences led to Ucn 2 and Ucn 3 being termed “type 2 urocortins”, with a gene duplication event putatively originating the “type 1” CRF/Ucn 1 versus type 2 Ucn 2/Ucn 3 lineages (Fekete and Zorrilla, 2007).

CRF receptor subtypes show different anatomical distributions and ligand codistributions, suggesting different subtype functions. Extrahypothalamic CRF₁ activation is hypothesized to have arousing and anxiogenic-like properties (Zorrilla and Koob, 2004). However, anxiety-related functions of CRF₂ receptors remain unclear. Several results potentially support the hypothesis that brain CRF₂ receptor activation has anxiolytic-like effects, perhaps by counter-regulating CRF₁ action. Intracerebroventricular (i.c.v.) administration of the type 2 Ucns (Valdez et al., 2002, 2003; Ohata and Shibasaki, 2004; Venihaki et al., 2004; Zorrilla et al., 2004) or low doses of the CRF₂ antagonist antisauvagine-30 (Kishimoto et al., 2000) exert anxiolytic-like effects. In addition, some, but not all (Coste et al., 2000), studies of CRF₂ knock-out mice observed an anxiogenic-like phenotype of mice in the elevated plus-maze (Bale et al., 2000; Kishimoto et al., 2000), emergence (Kishimoto et al., 2000), open field (Bale et al., 2000), or, following stress, light/dark box exposure (see Fig. 6A in Henry et al., 2006). In some brain regions, CRF₂ receptor activation has electrophysiological effects opposite to those of CRF₁ activation (Liu et al., 2004). Finally, CRF₂ knockout mice exhibit increased CRF and Ucn 1 mRNA in the central amygdala and Edinger-Westphal nucleus, respectively (Bale et al., 2000), potential evidence that CRF₂ signaling tonically decreases the threshold for these molecular stress responses (Fekete and Zorrilla, 2007).

Yet, other findings (Takahashi et al., 2001; Pelleymounter et al., 2002, 2004; Risbrough et al., 2003) support the different hypothesis that CRF₂ receptor activation, especially in the lateral septum (Henry et al., 2006) or dorsal raphe (Hammack et al., 2003), has anxiogenic-like effects, promoting defensive responses to uncontrollable stressors. In this view, the anxiogenic-like phenotype of CRF₂ null mutants is not a direct action of CRF₂ deficiency. Rather, their increased expression of anxiogenic-like CRF₁ agonist molecules with affinity for their remaining CRF₁ receptors might (over) compensate for, rather than result from, CRF₂ deficiency. In addition, the CRF₂ knockout phenotype might be due to altered maternal rearing by mutant dams (Bale and Vale, 2004).

The net anxiety-related effects of i.c.v. (non-site-specific) administration of type 2 Ucns are of additional interest because synthetic CRF₂ agonists are candidate systemic therapeutics for cardiovascular disease. A limitation of studying Ucn 2 in this context is that it might produce CRF₁-mediated actions at pharmacological doses; Ucn 2 is a low-affinity, but direct CRF₁ agonist (EC₅₀ ∼360 nM) (Hoare et al., 2005) that also potently binds (K_i = 4.4 nM) the rodent CRF-BP (Fekete and Zorrilla, 2007), liberating sequestered CRF/Ucn 1. Ucn 3 does not share this CRF₁ agonist potential. Of physiological relevance, Ucn 3 is distributed with CRF₂ receptors in the lateral septum, medial amygdala, and bed nucleus of the stria terminalis (Fekete and Zorrilla, 2007).

To understand better the anxiety-related consequences of central CRF₁ versus CRF₂ activation, the present study compared the effects of i.c.v. administration of selective CRF₁ and CRF₂ agonists. Acute i.c.v. administration of the CRF₁ agonist stressin₁-A (Rivier et al., 2007), a CRF analog, was hypothesized to increase anxiety-like behavior. Conversely, equimolar i.c.v. doses of mouse Ucn 3 (mUcn 3), a CRF_2(a) agonist (EC₅₀ = 0.073–14.2 nM) that lacks CRF₁ activity (EC₅₀ >> 1500 nM) or CRF-BP affinity (K_i > 2000 nM), was hypothesized to lack anxiogenic-like activity and perhaps to have anxiolytic-like effects [see Tables 1 and 2 for comparison of stressin₁-A, mUcn 3, and ovine (o) CRF primary structure and activity, respectively]. CRF-related peptides were evaluated in the shock-probe, social interaction, and defensive withdrawal tests. To explore whether Ucn 3 treatment produced behavioral changes like those of a CRF₁ antagonist, the effects of systemic administration of MJL-109-2, a novel selective CRF₁ antagonist (Jagoda et al., 2003), were determined in the shock-probe test.

Materials and Methods

Animals. Male Wistar rats (Charles River, Raleigh, NC; 200–225 g on arrival) were group-housed (three rats/cage) in a temperature- and humidity-controlled vivarium on a reversed 12-h light/dark cycle (lights off, 10:00 AM) with food and water available ad libitum. Rats were allowed to acclimate to the vivarium for 1 week before experimental procedures. Behavioral tests were conducted during the dark cycle. All procedures adhered with the Institute of Laboratory Animal Resources (1996), and they were approved by the Institutional Animal Care and Use Committee of The Scripps Research Institute.

Intracerebroventricular Surgery. For intracranial peptide infusion, rats were surgically implanted with a stainless steel 22-gauge unilateral guide cannula (Plastics One Inc., Roanoke, VA). Anesthetized (isoflurane; 2–3% in oxygen) rats were secured in a stereotaxic frame (David Kopf Instruments, Tujunga, CA) and implanted, using sterile technique, with the cannula secured 1.2 mm above the final target injection site as follows: AP, –0.6; ML, ±2.0; V, –3.3 mm from the skull, with the interaural bar set at +5.0 mm (Pellegrino et al., 1979). Subjects were allowed at least 7 days recovery before testing, during which time they were handled frequently. Cannula placement and tissue integrity also were functionally confirmed at the conclusion of studies by a positive dipsogenic response (>5 ml of water intake in nondeprived rats within 30 min of injection) to i.c.v. angiotensin II (5 μg/5 μl).

Drugs and Injections. Murine Ucn 3 (Fekete and Zorrilla, 2007), stressin₁-A (cyclo_31–34[d-Phe¹², MeLeu^21,38, Ala²², Arg²³, Glu^25,39]-Achuman corticotropin-releasing factor_4–41) (Rivier et al., 2007), and angiotensin II were synthesized manually using solid-phase methodology, purified using high-pressure liquid chromatography, and fully characterized using capillary zone electrophoresis, high-pressure liquid chromatography, and mass spectrometry. Peptides were dissolved in 0.5× phosphate-buffered saline, pH 7.4, immediately before testing and kept on ice. Intracerebroventricular infusion was performed through a 28-gauge injector that extended 1.2 mm beyond the tip of the guide cannula. The injector was attached via polyethylene-20 tubing to a 10-μl Hamilton microsyringe (Hamilton, Reno, NV). Intracerebroventricular treatments were manually injected (5 μl) over 1 min. Injectors were left in place for 1 min after infusion to allow diffusion.

The selective CRF₁ antagonist MJL-1-109-2 (K_i for CRF₁ receptors = 1.9 nM; cLogP = 3), an analog of DMP696, was synthesized as described previously (Jagoda et al., 2003). It was administered i.p. (4 ml/kg) in 20% hydroxypropyl-β-cyclodextrin, pH 4.5.

Shock-Probe Test. The shock-probe test (De Boer and Koolhaas, 2003) was performed under room lighting (∼300 lux) in a polycarbonate cage with wood shavings (5.08 cm; 2 in.) along the bottom and a hole for the shock-probe 2.54 cm (1 in.) above the bedding. Rats were kept in the quiet, dark anteroom for at least 2 h before the beginning of treatments. On the day before the test, rats were group-acclimated to the test cage for 45 min with the probe absent. On the test day, the probe was inserted and connected to a Coulbourn E13-01 shocker (Coulbourn Instruments, Allentown, PA) to deliver a 1.5-mA a.c. shock when contacted. Rats were individually tested in a between-subjects design following pretreatment with equimolar doses of mUcn 3 or stressin₁-A (0, 0.04, 0.2, or 1 nmol, i.c.v.; 10 min before testing; n = 8–11 rats/group) or with MJL-1-109-2 (0, 5 or 10 mg/kg i.p.; 60 min before testing; n = 7–9 rats/group). The 10-min pretreatment interval was chosen based on previous studies that found behavioral effects of i.c.v. Ucn 3 administration (Valdez et al., 2003, 2004; Venihaki et al., 2004). To test the possibility that a longer pretreatment interval might reveal other acute effects of Ucn 3 administration, additional rats (n = 7 rats/group) received Ucn 3 (1 nmol) or vehicle 30 min before shock-probe testing. Upon delivery of a shock, the probe was deactivated, and behavior of the rat was recorded for a further 10 min. Bedding was changed, and the cage was wiped with water and dried between tests. A rater, unaware of the subject's treatment condition, scored the number of bouts and duration of the following behaviors from videotape: defensive burying, freezing, grooming, locomotion, and rearing. Reflexive shock reactivity also was rated on the following 1- to-4 scale (Pesold and Treit, 1992): 1, flinch of only head or forepaw; 2, whole-body flinch, with or without slow ambulation away from the probe; 3, whole-body flinch and/or jumping followed by immediate ambulation away from the probe; or 4, whole-body flinch and jump (all four paws in the air) followed by immediate and rapid ambulation (i.e., running) away from the probe. Stressors or anxiogenic-like compounds have been observed to increase defensive burying and/or freezing in this model (De Boer and Koolhaas, 2003).

Social Interaction Test. Conditions of the social interaction test were chosen to establish a moderate baseline level of anxiety-like behavior, thereby allowing detection of anxiogenic-like or anxiolytic-like effects, by performing the test in a brightly lit, but familiar arena, as described previously (File and Seth, 2003). Rats were familiarized to the walled test arena (a 106 × 92 × 77 cm open field made of black foamed polyvinylchloride) by placing them individually in the illuminated (∼300 lux) arena for 5 min on each of the 2 days before social interaction testing. Rats were single-housed for 5 days before the test day to promote social interaction (File and Seth, 2003).

Animals were kept in the quiet, dark anteroom for at least 2 h before the beginning of treatments. On test days, experimental rats were identified with a permanent marker, and they were assigned an experimentally naive, weight-matched, treatment-free, individually housed (5 days) male Wistar rat for social interaction testing. Experimental rats (n = 67) were pretreated (–10 min) with equimolar doses of mUcn 3 or stressin₁-A (0, 0.04, 0.2, or 1 nmol, i.c.v.; n = 8–11 rats/group) in a between-subjects design. The experimental rat and its untreated partner were then placed in opposite corners of the test arena, and the 4.5-min social interaction trial was videotaped and scored naive to treatment condition for the following classes of behavior: affiliative social interaction (conspecific sniffing, contact, and allogrooming), agonistic social interaction (mounting, sidling, attack, threat posture, flee/leave, defensive posture, and submission), and nonsocial behaviors (locomotion, rearing, and self-grooming). Only interactions initiated by the experimental rat were scored. Anxiolytics increase affiliative social interaction, whereas anxiogenics decrease affiliative social interaction in this model (File and Seth, 2003).

Defensive Withdrawal Test. Defensive withdrawal testing was performed in the identical walled open field as social interaction testing, but it contained a cylindrical chamber (a 2-l cylindrical Pyrex beaker) wrapped in dark brown packing tape. The chamber was open at one end and located 15.0 cm from a corner of the open field, aligned lengthwise, with the open end facing the corner. On the day of testing, rats were kept in the quiet, dark anteroom for at least 2 h before the beginning of treatments. To assess the effect of Ucn 3 on defensive withdrawal behavior in an unfamiliar setting, naive rats (n = 38; 9–10 rats/group) were pretreated with mUcn 3 (0, 0.02, 0.2, or 2 nmol i.c.v.; 10 min before testing) using a between-subjects design. For testing, the animal was placed in the small withdrawal chamber facing the rear of the chamber, and behavior was recorded via a video camera for 10 min. The apparatus was wiped clean with water and dried after each subject. Rats were tested under bright lighting (∼300 lux) during their dark (active) cycle. A rater, unaware of treatment condition, scored the following measures from videotape: initial latency to leave the chamber (i.e., placement of all four paws in the open field) and the number and duration of subsequent withdrawals back into the chamber. Anxiolytics and anxiogenics, respectively, decrease and increase the duration of time spent withdrawn in the chamber in this model (see Takahashi et al., 1990, and references therein).

Statistical Analysis. Effects of each compound were analyzed by separate one-way between-subjects analyses of variance (ANOVA). Linear contrast ANOVAs were used to identify monophasic dose-responsive effects, as defined by a log-linear dose-response function (Rosner, 1995). For the defensive withdrawal test, data were log₁₀-transformed due to skewed and unequal variance between groups. Because of extreme values, data for the defensive burying test still did not meet statistical assumptions of ANOVA after mathematical transformation, so the single most extreme values of each group were Winsorized, yielding homogenous and normal distributions. Significant omnibus tests were followed by Newman-Keuls or Dunnett's test, as appropriate, to evaluate pairwise group differences between treatment conditions. The statistical package used was SYSTAT 10.0 (SPSS Inc., Chicago, IL).

Results

Shock-Probe Test

mUcn 3. Administration (i.c.v.) of the selective CRF₂ agonist mUcn 3 did not significantly alter reflexive shock reactivity [vehicle, 1.5 ± 0.2 (mean ± S.E.M.); 0.04 nmol, 1.7 ± 0.3; 0.2 nmol, 1.3 ± 0.2; and 1 nmol, 1.4 ± 0.2] or the latency to initiate defensive burying or freezing in the shock-probe test (data not shown). As shown in Fig. 1, A and B, mUcn 3 also did not alter the total duration of defensive burying or freezing behavior. In contrast, mUcn 3 (0.2 and 1 nmol) significantly and in a log-linear dose-dependent manner increased the duration of grooming [linear ANOVA: F(1,26) = 5.59, p < 0.05] and reduced the duration of both locomotion [F(3,26) = 3.66, p < 0.05; linear ANOVA: F(1,26) = 8.91, p < 0.01] and rearing [F(3,26) = 8.43, p < 0.001; linear ANOVA: F(1,26) = 24.50, p < 0.001] (Fig. 1, C–E).

As shown in Table 3, mUcn 3 administration suppressed locomotion and rearing by potently and dose-dependently reducing the number, rather than average duration, of bouts of these behaviors, with a minimum effective dose of 0.2 nmol [for locomotion bouts: F(3,26) = 3.39, p < 0.05; linear ANOVA: F(1,26) = 9.43, p < 0.01; for rearing bouts: F(3,26) = 5.58, p < 0.01; linear ANOVA: F(1,26) = 14.98, p < 0.001]. The behavioral mechanism by which mUcn 3 increased grooming was less defined, because ANOVA did not show significant increases in either bout frequency or average bout duration for this behavior.

To explore the possibility that different behavioral outcomes might have resulted from longer Ucn 3 pretreatment intervals, separate rats also were tested in the shock-probe test 30 min after receiving vehicle or 1 nmol of mUcn 3. Similar to the acute 10-min pretreatment interval, rats that had received Ucn 3 30 min earlier did not differ in their reflexive shock reactivity (data not shown) or duration of burying and freezing behavior, but rather they showed significantly less rearing behavior relative to vehicle-treated controls [125.8 ± 19.7 (mean ± S.E.M.) versus 208.1 ± 14.9 s; n = 7 rats/group, p < 0.003], partly due to fewer bouts of rearing (31.7 ± 3.1 versus 39.4 ± 3.0, p < 0.02). The duration of locomotion also tended to be reduced by Ucn 3 (169.4 ± 13.2 versus 199.1 ± 13.0 s, p < 0.07). Unlike the more acute pretreatment interval, grooming was not altered following 30 min pretreatment with Ucn 3 (see also Supplemental Table 1).

Stressin₁-A. Administration (i.c.v.) of the selective CRF₁ agonist stressin₁-A did not significantly affect reflexive shock reactivity [vehicle, 1.8 ± 0.3 (mean ± S.E.M.); 0.04 nmol, 1.7 ± 0.2; 0.2 nmol, 1.1 ± 0.2; and 1 nmol, 2.3 ± 0.4] or the latency to initiate burying or freezing (data not shown). However, unlike mUcn 3, stressin₁-A differentially altered the duration of defensive burying [F(3,41) = 3.35, p < 0.05] and freezing behavior [F(3,41) = 2.86, p = 0.05] in the shock-probe test according to the administered dose (Fig. 1, A and B). Linear trend contrast indicated a significant log-linear relation of stressin₁-A dose to the duration of freezing behavior [F(1,41) = 5.66, p < 0.05]. Pairwise comparisons showed that the lowest dose (0.04 nmol) of stressin₁-A increased burying but not freezing behavior, whereas the highest dose (1.0 nmol) increased freezing but not burying behavior. As shown in Table 3, stressin₁-A administration (0.04 and 0.2 nmol) increased the total duration of burying by significantly increasing the average duration, rather than number, of burying bouts [F(3,39) = 3.90, p < 0.05]. Conversely, stressin₁-A increased the duration of freezing behavior by increasing the number, but not average duration, of freezing bouts [F(3,39) = 3.03, p < 0.05]. Linear trend contrast analysis indicated a significant log-linear relation of stressin₁-A dose to the number of bouts of freezing [F(1,39) = 5.86, p < 0.05]. Pairwise comparisons indicated significantly more bouts of freezing at the 1-nmol dose.

Stressin₁-A (0.2 and 1 nmol) also dose-dependently increased the total duration of grooming [F(3,41) = 3.12, p < 0.01; linear ANOVA: F(1,41) = 13.08, p < 0.01] (Fig. 1E). Stressin₁-A did so by potently increasing the number, but not average duration, of grooming bouts, with a minimum effective dose of 0.04 nmol [F(3,41) = 6.98, p < 0.001; linear ANOVA: F(1,41) = 19.21, p < 0.001] (Table 3). In addition, stressin₁-A decreased the duration of locomotor activity significantly at the 0.04-nmol dose, but not higher doses [F(3,41) = 4.34, p < 0.01], by decreasing the number, but not average duration, of bouts of locomotion [F(3,41) = 3.43, p < 0.05] (Fig. 1C; Tables 1 and 2). Stressin₁-A did not significantly affect rearing behavior (Fig. 1D; Table 3).

MJL-1-109-2. Systemic administration of the nonpeptide selective CRF₁ antagonist MJL-1-109-2 did not affect reflexive shock reactivity [vehicle, 1.8 ± 0.2 (mean ± S.E.M.); 5 mg/kg, 1.7 ± 0.3; and 10 mg/kg, 1.3 ± 0.2] or freezing behavior (Fig. 2B). However, 10 mg/kg MJL-1-109-2 pretreatment significantly increased the latency to initiate defensive burying [vehicle, 166.9 ± 58.3; 5 mg/kg, 155.5 ± 38.0; and 10 mg/kg, 443.3 ± 85.3 s; F(2,22) = 6.76, p < 0.01] and significantly decreased the total duration of burying [F(2,22) = 7.11, p < 0.01] (Fig. 2A). MJL-1-109-2 (10 mg/kg) reduced burying behavior by reducing the number, rather than average duration, of burying bouts [F(2,22) = 4.15, p < 0.05] (Table 4).

MJL-1-109-2 did not alter the total duration of locomotor activity, rearing, or grooming (Fig. 2, C–E). However, although the total duration of these behaviors was unaffected, MJL-1-109-2 at 10 mg/kg did significantly reduce the number of bouts of locomotion, rearing, and grooming [F(2,22) = 4.60, 10.69, and 4.00, all p values < 0.05], while reciprocally increasing their bout duration, significantly so for locomotor activity [F(2,22) = 4.74, p < 0.05]. Thus, MJL-1-109-2 pretreatment led to fewer, but more sustained, bouts of these active behaviors (Table 4).

Social Interaction

mUcn 3. Administration (i.c.v.) of the selective CRF₂ agonist mUcn 3 did not significantly alter the duration of affiliative (Fig. 3A, left), nonsocial (Fig. 3B, left), or agonistic behaviors (which totaled <39.2 s for all doses; data not shown) in the social interaction test.

Stressin₁-A. Unlike mUcn 3, i.c.v. administration of stressin₁-A decreased the duration of affiliative social behavior [F(3,28) = 4.28, p < 0.01] in a log-linear, dose-dependent manner [F(1,28) = 8.24, p < 0.01]. Pairwise comparisons showed significantly reduced affiliative interaction at the 0.2- and 1-nmol doses (Fig. 3A, right). Conversely, stressin₁-A significantly increased the duration of nonsocial behavior [F(3,28) = 5.40, p < 0.01], again in a log-linear dose-dependent manner [F(1,24) = 8.34, p < 0.01], with pairwise comparisons revealing more nonsocial behavior in rats treated with each dose of stressin₁-A (Fig. 3B, right). The decrease of affiliative social behavior was due to a potent (minimum effective dose = 0.04 nmol) reduction in social sniffing [vehicle, 73.3 ± 7.3 (mean ± S.E.M.); 0.04 nmol, 43.9 ± 6.8; 0.2 nmol, 39.7 ± 3.7; and 1 nmol, 44.7 ± 6.8], whereas the increase in nonsocial behavior reflected potent (minimum effective dose = 0.04 nmol) increases in locomotor activity (vehicle, 134.3 ± 14.8; 0.04 nmol, 174.9 ± 15.2; 0.2 nmol, 183.2 ± 3.7; and 1 nmol 177.7 ± 10.1). Higher doses of stressin₁-A (0.2 and 1 nmol) significantly increased the likelihood that a rat exhibited mounting behavior. Whereas only 4% (2/50) of rats treated with vehicle, Ucn 3 or the low dose (0.04 nmol) of stressin₁-A attempted to mount their partner, more than one third of rats treated with higher doses of stressin₁-A did so (5/14 or 36%; 29 and 43% at the 0.2- and 1-nmol doses, respectively) (Fisher's exact test, p = 0.004).

Defensive Withdrawal

Administration (i.c.v.) of mUcn 3 (2 nmol) significantly decreased the total time rats spent withdrawn in the defensive withdrawal chamber [F(3,26) = 3.82, p < 0.05] (Fig. 4B). As shown in Fig. 4, A and B, mUcn 3 did not reliably alter the latency to emerge from the chamber, but rather, it decreased the total duration of subsequent withdrawals back into the enclosure [F(3,26) = 4.30, p < 0.01], significantly so at the 2-nmol dose.

Discussion

The present studies indicate that brain CRF₁ and CRF₂ receptors differentially affect anxiety-like behavior, with the behavioral expression of anxiety qualitatively changing with increasing CRF₁ activation. In the shock-probe test, 40 pmol of stressin₁-A, a selective CRF₁ agonist, increased defensive burying, whereas 5- to 25-fold higher doses increased freezing and self-grooming. In the social interaction test, 40 pmol of stressin₁-A decreased social sniffing and increased locomotion, whereas higher doses also increased the frequency of conspecific mounting. The results resemble findings that i.c.v. CRF increases locomotor activation under low-stress conditions but facilitates freezing and aggression (Tazi et al., 1987; Zorrilla and Koob, 2004) under high-stress conditions. Likewise, CRF increases defensive burying in acclimated rats in the shock-probe test, but it decreases burying and increases freezing and grooming in unhandled rats (Diamant et al., 1992). Here, increasing doses of a CRF₁ agonist qualitatively shifted behavior from activation and “active coping” to “noncoping” and even aggressive, agonistic behaviors.

Ucn 3, CRF₂ Receptors, and Anxiety-Like Behavior. In addition to characterizing effects of a novel CRF₁ agonist, the present study sought to examine the net anxiety-related effects of brain CRF₂ receptor activation. Intracerebroventricular administration of up to 1 nmol of the selective CRF₂ agonist mUcn 3 did not influence anxiety-like behavior in the social interaction or shock-probe tests. A slightly higher dose of Ucn 3 (2 nmol) reduced defensive withdrawal behavior. Ucn 3 also dose-dependently (0.04–1 nmol) decreased locomotion and rearing behavior in the shock-probe test. However, Ucn 3 did not reduce locomotion in the social interaction test, suggesting that Ucn 3 does not nonspecifically alter motor function. Still, suppression of motor activity by Ucn 3 may complicate interpretation of anxiety-like behavior, for example, if Ucn 3 suppressed active behaviors, including burying, during shock-probe testing. Against this hypothesis, Ucn 3 increased self-grooming following a 10-min (but not 30-min) pretreatment interval, a “transition” behavior seen in response to diverse stimuli. Together, the results do not support the hypothesis that i.c.v. Ucn 3 potently increases anxiety-like behavior. This conclusion is consistent with previous studies that observed either no effect (Valdez et al., 2004) or decreased anxiety-like behavior of mice and of rats 5 to 10 min after i.c.v. Ucn 3 pretreatment (Valdez et al., 2003; Venihaki et al., 2004).

This conclusion seems at odds with reports from independent groups that i.c.v. administration of Ucn 2, another CRF₂ agonist, increased acoustic startle reactivity (Risbrough et al., 2003, 2004) and plus-maze anxiety-like behavior (Pelleymounter et al., 2002, 2004). Differences between those studies and the present study include the species (rat versus mouse), pharmacological properties of the injected agonist (Ucn 3 versus Ucn 2), pretreatment intervals (10–30 min versus 30 min–2 h), and specific tests of anxiety-like behavior used. Interestingly, Risbrough et al. (2004) concluded that effects of Ucn 2 on startle reactivity partly involved CRF₁ receptors, at which Ucn 3, unlike Ucn 2, lacks any activity. In the same studies, Ucn 3 (i.c.v.), unlike Ucn 2, did not affect plus-maze behavior (up to 3 nmol) (Pelleymounter et al., 2004) or acoustic startle reactivity (up to 2.4 nmol) (Risbrough et al., 2004). In other studies, i.c.v. Ucn 2 did not increase anxiety-like behavior of mice (1.2 nmol) in the open field, novel object, or light/dark box tests (Henry et al., 2006) or of rats (2.4 nmol) in the plus-maze (Valdez et al., 2002). Together, the findings do not strongly support the hypothesis that the singular activation of CRF₂ receptors by i.c.v. injection of selective CRF₂ agonists acutely increases anxiety-like behavior.

An important distinction here is that that CRF₂ activation within the dorsal raphe (Hammack et al., 2003) and lateral septum (Henry et al., 2006) reportedly exerts anxiogenic-like effects. Accordingly, intra-raphe/septal Ucn 3 infusion may very well elicit anxiety-like behavior; CRF₂ receptors in those brain regions also may mediate physiological anxiety-like responses. However, the present study explored the net anxiety-related effects of i.c.v. Ucn 3 administration, which also may activate CRF₂ receptors in other brain sites (e.g., amygdala) that potentially produce countervailing (Liu et al., 2004), anxiolytic-like effects. Similar to the present findings that i.c.v. CRF₂ agonist administration did not increase anxiety-like behavior, CRF₂ knockout mice do not exhibit an anxiolytic-like phenotype (Bale et al., 2000; Coste et al., 2000; Kishimoto et al., 2000; Henry et al., 2006).

Stressin₁-A, CRF₁ Receptors, and Agonistic Behavior. Stressin₁-A increased mounting behavior in a neutral arena, a context in which mounting is normally rare (McGinnis et al., 2002). Same-sex mounting by an adult male rat is a testosterone-enhanced, threat/attack-associated behavior, commonly interpreted as an “offensive” act of territorial dominance. Tail pinch or testing with a male undergoing tail pinch also increases mount frequency in neutral environments. Thus, it has been proposed that arousing stimuli increase the incentive salience of or sensitivity to potential social threats (McGinnis et al., 2002). Perhaps CRF₁ receptors mediate the ability of arousing stimuli to facilitate offensive agonistic responses to mildly threatening or ambiguous social stimuli. Supporting this hypothesis, antalarmin, a CRF₁ antagonist, reduced aggressive behavior in male rhesus macaques placed in a novel cage adjoining an unfamiliar male (Habib et al., 2000). A CRF₁ antagonist (SSR125543A) also decreased dominance/offensive behavior in Syrian hamsters exposed to smaller “intruders” (Farrokhi et al., 2004).

Yet, some data suggest that CRF receptors facilitate submissive/defensive agonistic behavior. Infusion of d-Phe-CRF_12–41 into the lateral ventricle or extended amygdala (Jasnow et al., 2004) reduced “submissive/defensive behavior” in previously defeated Syrian hamsters encountering an unfamiliar male. Intra-amygdala antalarmin similarly reduced “defensive posturing” in defeated mice exposed to an unfamiliar male (Robison et al., 2004). CRF receptor agonists (i.c.v.) increased defensive behavior and reduced attacks in isolated DBA/2 mice facing a male intruder (Mele et al., 1987).

Several explanations might reconcile the observations. Perhaps CRF₁ activation increases the evaluation or salience of social threat, but context or experience guides the specific behavioral response. Relatively innocuous social stimuli would be perceived as surmountable threats (facilitating offensive aggression), known perils would be avoided more readily (facilitating submissive behavior/defensive posturing), and unavoidable, injurious stimuli would be treated as matters of life and death (facilitating defensive fighting). Accordingly, shock-induced defensive fighting is reduced by CRF antagonists (Tazi et al., 1987). An alternative explanation is that intermale social conflict behavior falls on a continuum from offensive aggression (at one extreme) through submission/defensive posturing to defensive aggression (at the other extreme), with arousal/anxiety levels influencing the behavior exhibited along this dimension. Accordingly, intra-amygdala CRF increased offensive aggression at lower (10- and 50-pmol), but not higher (150-pmol) doses (Elkabir et al., 1990), and antalarmin reduced aggression in rhesus monkeys in relation to its anxiolytic-like activity (Habib et al., 2000). Thus, rather than directly subserving agonistic behavior, perhaps CRF₁ receptors indirectly influence social conflict behavior by modulating the evaluation of social threat or the general level of arousal or anxiety.

Unlike stressin₁-A, Ucn 3 (i.c.v.) did not influence intermale social interaction, consistent with normal intermale aggression of CRF₂ knockout mice (Gammie et al., 2005). The results suggest that CRF and stressors reduce social interaction via CRF₁ activation. Indeed, CRF₁ antagonists reverse reductions in social interaction due to ethanol withdrawal, maternal separation, and restraint stress (Zorrilla and Koob, 2004; Gehlert et al., 2005).

Whether CRF₂ receptor simulation alters behavior when CRF₁ receptors are concurrently activated was not addressed in the current study. CRF₂ stimulation has been proposed by some to elicit anti-CRF/CRF₁-like effects on behavioral outcomes (Valdez et al., 2002; Ohata and Shibasaki, 2004; Venihaki et al., 2004). Accordingly, functional antagonism between CRF₁/CRF₂ subtypes was observed in electrophysiological studies of septum and amygdala (Liu et al., 2004). In contrast, lateral septal CRF₂ stimulation produces anxiogenic-like effects in the presence of stressors via a CRF₁-dependent mechanism (Henry et al., 2006) (also see Risbrough et al., 2004). The results suggest complexity in the integrated output of these receptor systems that may be brain site- and experience-dependent.

In summary, i.c.v. Ucn 3 increased grooming and reduced locomotion and rearing in the shock-probe test, similar to reported locomotor-suppressing effects of i.c.v. Ucn 2 and Ucn 3 (Valdez et al., 2002, 2003; Ohata and Shibasaki, 2004; Zorrilla et al., 2004; Henry et al., 2006). At a higher dose, i.c.v. Ucn 3 reduced defensive withdrawal, similar to reported anxiolytic-like effects of Ucn 3 (Valdez et al., 2003; Venihaki et al., 2004), but anxiolytic-like activity was not observed in the social interaction or shock-probe tests up to a 1-nmol dose. Unlike Ucn 3, i.c.v. stressin₁-A increased behavioral arousal and active coping behaviors at lower doses, whereas it increased noncoping and offensive agonistic behavior at higher doses. MJL-1-109-2, a CRF₁ antagonist, reduced anxiety-like behavior in the shock-probe test, with a behavioral profile unlike that of Ucn 3. Thus, the behavioral expression of anxiety changes with greater levels of brain CRF₁ receptor activation, and stimulation of CRF₂ receptors by i.c.v. Ucn 3 does not mimic effects of either a CRF₁ agonist or antagonist. The results indicate differential, but not opposing, actions of brain CRF₁ and CRF₂ receptors on behavioral activation and anxiety-like behavior.