Proc Natl Acad Sci U S A. 2010 February 2; 107(5): 2265–2270.
Abstract
Although there is an extensive amount known about specific sensory and motor functions of the vertebrate brain, less is understood about the regulation of global brain states. We have recently proposed that a function termed generalized arousal (Ag) serves as the most elemental driving force in the nervous system, responsible for the initial activation of all behavioral responses. An animal with increased generalized CNS arousal is characterized by greater motor activity, increased responsivity to sensory stimuli, and greater emotional lability. Implicit in this theory was the prediction that increases in generalized arousal would augment specific motivated behaviors that depend on arousal. Here, we address the idea directly by testing two lines of mice bred for high or low levels of generalized arousal and assessing their responses in tests of specific forms of behavioral arousal, sex and anxiety/exploration. We report that animals selected for differential generalized arousal exhibit marked increases in sensory, motor, and emotional reactivity in our arousal assay. Furthermore, male mice selected for high levels of generalized arousal were excitable and showed more incomplete mounts before the first intromission (IN), but having achieved that IN, they exhibited far fewer IN before ejaculating, as well as ejaculating much sooner after the first IN, thus indicating a high level of sexual arousal. Additionally, high-arousal animals of both sexes exhibited greater levels of anxiety-like behaviors and reduced exploratory behavior in the elevated plus maze and light-dark box tasks. Taken together, these data illustrate the impact of Ag on motivated behaviors.
One of the basic challenges facing all vertebrate animals is the necessity to activate large numbers of behavioral responses to large numbers of environmental conditions, some of which are threatening. We have recently proposed (
The following popper user interface control may not be accessible. Tab to the next button to revert the control to an accessible version.
Destroy user interface control1) that a function exists in vertebrate nervous systems that initiates behavioral activation of large numbers of responses to meet this basic challenge, a function we have termed generalized arousal. Neurons serving generalized arousal mechanisms would receive sensory inputs both from the external environment and the internal milieu and be able rapidly to activate arousal states that empower more specific, motivated behavioral responses. An increase in activity of this system would elevate generalized arousal, producing an animal with more motor activity, greater sensory responsiveness, and increased emotional lability. The purpose of the work reported here was to add to the evidence that a generalized arousal function exists (by beginning the breeding of high and low arousal lines of mice) and to test the impact of high or low generalized arousal states on specific motivated behaviors: male sexual behavior and anxiety/exploration.
So far, the existence of generalized arousal as a measurable and physiologically-relevant CNS state has been inferred from three separate approaches, the third of which is reported here.
First we used principal component analyses to extract the largest single factor that underlies behavioral arousal
One of the basic challenges facing all vertebrate animals is the necessity to activate large numbers of behavioral responses to large numbers of environmental conditions, some of which are threatening. We have recently proposed (1) that a function exists in vertebrate nervous systems that initiates behavioral activation of large numbers of responses to meet this basic challenge, a function we have termed generalized arousal. Neurons serving generalized arousal mechanisms would receive sensory inputs both from the external environment and the internal milieu and be able rapidly to activate arousal states that empower more specific, motivated behavioral responses. An increase in activity of this system would elevate generalized arousal, producing an animal with more motor activity, greater sensory responsiveness, and increased emotional lability. The purpose of the work reported here was to add to the evidence that a generalized arousal function exists (by beginning the breeding of high and low arousal lines of mice) and to test the impact of high or low generalized arousal states on specific motivated behaviors: male sexual behavior and anxiety/exploration.
So far, the existence of generalized arousal as a measurable and physiologically-relevant CNS state has been inferred from three separate approaches, the third of which is reported here.
First we used principal component analyses to extract the largest single factor that underlies behavioral arousal (2). To determine the relative contribution of this most elementary force for CNS arousal in mouse behavioral screens we performed a meta-analysis of five behavioral experiments with mice (3). The percent of the behavioral arousal data accounted for, by a forced one-factor solution, varied across the five experiments from 29% to 45%. Thus, we can say that generalized CNS arousal exists and accounted for about one third of the data in these experiments.
Second, there is a wealth of chemical and anatomical and physiological data on the proximate mechanisms that underlie CNS arousal. In neuroanatomical terms, ascending and descending pathways are well known. The neuronal regulation of arousal states is controlled by a distributed, bilateral, bidirectional neuronal network with interconnections between the lower brainstem and forebrain (reviewed in refs. 1, 4). Physiologically, brainstem arousal systems have evolved to maintain spontaneous (not necessarily task-related) activity in a default network in the cortex (5 –7). The sharp transition (8) from quiet wakefulness to active exploratory behavior is marked by decreased synchrony of low frequency oscillations (9) due to reduced correlations between membrane potentials in cerebral cortical neurons (10). Transitions from sleep to wakefulness are likely facilitated by increased electrical activity in medullary portions of the ascending reticular activating system and the medullary raphe groups (e.g., (11 –13). The neurochemistry of generalized arousal, also, is being worked out. Fibers from neurochemically distinct cell groups in the brainstem, e.g., the monoaminergic pathways, affect neurons in thalamic, hypothalamic, and other forebrain targets. In the hypothalamus, these inputs drive the activity of neurons that produce arousal-related neurochemicals such as hypocretin and histamine. Importantly, these connections are also bidirectional as histamine and hypocretin neurons in the hypothalamus project back to the brainstem arousal regions and are part of a network that regulate arousal and sleep (14). For instance, a “switch” for the control of global brain arousal has been described wherein GABAergic sleep-active neurons in the ventrolateral preoptic area (vlPO) promote sleep both by directly inhibiting the ascending arousal monoamine systems (norepinephrine, serotonin, histamine, etc.) and also by inhibiting hypocretin neurons in the lateral hypothalamus (LHA) that normally excite the monoamine pathways and the forebrain (15). The pathway is reciprocal because serotonergic, noradrenergic, and histaminergic inputs to the vlPO inhibit activity during wakefulness (16). vlPO activity thus inhibits ascending arousal activity and also removes tonic inhibition on its own activity (4). Similarly, the monoaminergic cell groups inhibit vlPO activity and in doing so disinhibit their own activity. In this manner, hypothalamic activity can regulate and gate the activity of ascending monoaminergic inputs to the forebrain. Finally, from a genomic perspective, there are likely more than 100 genes implicated in generalized CNS arousal, including, but not limited to, genes coding for synthetic enzymes for arousal-related transmitters, neuropeptides, and their receptors, e.g., the hypocretin system (17). Thus, mechanisms for a generalized arousal function are being documented at neuroanatomical, neurophysiological, and neurochemical levels.
These distributed ascending and descending systems are involved in both the global regulation of CNS arousal but also in a variety of specific effects on behavioral states, e.g., anxiety, fear, hunger, thirst, and sexual drive providing a substrate on which global changes in brain excitability can translate into alterations in specific behaviors (18 –20).
Here we report a third line of support for the existence of generalized arousal: data from a large scale selective breeding project that has been undertaken to produce lines of mice selected for either high or low generalized arousal. To that end, we have developed a quantitative and automated assay of generalized arousal and used that assay to select highly outbred mice based on an overall index of arousal behavior. As mentioned, a key component of the generalized arousal theory is that alterations in generalized arousal should be able to modulate the strength of specific motivated behaviors. Thus, we have investigated the relationship between levels of generalized arousal and male sexual behaviors, hormone-dependent behaviors that have previously been shown to be affected by some of the same neurochemicals that are involved in CNS arousal (e.g., histamine and hypocretin) (21, 22). We also used two assays of “anxiety,” the elevated plus maze and the light/dark transition test, assuming that increases in anxiety would translate into decreases in exploration. Results showed that animals bred for High versus Low levels of generalized arousal exhibit marked differences in tests of specific manifestations of arousal: sexual and anxiety/exploratory behaviors.
Results
We have developed a behavioral assay of generalized CNS arousal that takes into account the three proposed facets of arousal behavior, motor activity, sensory responsivity and emotionality (3). Genetically heterogeneous mice were tested in this assay ( SI Text ) for specific breeding details) and rank orders were established for high and low arousal on each of the three subscales of generalized arousal. The mice that exhibited the highest (and lowest) scores summed across the three subscales in each generation were then used as the founders of the next generation. Male and female mice derived from “high” parents are referred to as HM and HF, respectively, whereas the offspring of “low” parents are LM and LF. In all cases, we also separated mice based on the individual behaviors in the arousal assays to compare the effects of parental arousal versus offspring arousal on behavioral performance. Fig. S1 exhibits the arousal scores for the generation 5 (G5) mice that served as the stock from which the experimental mice (G6) in this manuscript were derived.
Generalized arousal measured by our assay was higher in mice from the high line. Both female and high generalized arousal mice exhibited more home cage activity (Fig. 1 A–D ), largely due to greater activity during the early part of the dark phase. Additionally, females and high arousal animals of both sexes exhibited more fearful behavior in a contextual-cued conditioning paradigm (Fig. 1E ). Finally, the behavioral reactivity to the administration of a neutral odorant was higher among High animals compared to those selected for Low arousal (Fig. 1F ); although no sex difference was apparent. These data indicate that even within 6 generations, divergent lines of mice exhibiting differential patterns of generalized arousal can begin to be generated.
Selective breeding alters behavior in the generalized arousal assay. All data are presented as mean (±SEM). (A) Total distance of home cage locomotor activity in 1-h bins and averaged across 4 consecutive days. Total distance traveled during the (B) light period, (C) first 4 h of the dark period, and (D) late in the dark period. (E) Change in total distance traveled from acclimation to posttone in the retrieval phase of the fear conditioning paradigm. Beam breaks (as measured by F) vertical activity, (G) horizontal activity, and (H) total distance in response to presentation of an odorant (air passed through 100% benzaldehyde). *Significantly different between high and low. Differences are considered statistically significant if P < 0.05. HF, n = 29, LF, n = 21, HM, n = 27, LM, n = 18.
Next, we sought to determine whether genetically-encoded differences in generalized arousal would translate into alterations in specific types of arousal-dependent motivated behaviors. To that end, we took the mice of G6 and divided them in two different ways by (i) parental arousal—whether their parents were in the high or low lines—and (ii) offspring arousal—whether the animal in question (G6) was at the top or bottom of the arousal distribution. First, male mice were exposed to a sexually naïve conspecific (of the Het8 strain) on consecutive days until they mated. Males from the high line and those offspring who exhibited high levels of generalized arousal exhibited a specific pattern of sexual behavior associated with a higher level of excitability and sexual arousal (Fig. 2). High arousal males exhibited more mounts before intromission (Fig. 2 A and E ), and then fewer intromissions before ejaculating (Fig. 2 B and F ), and they ejaculated more quickly after the first intromission (Fig. 2 C and G ). Additionally, the percentage of mount attempts that was successful in leading to intromission was significantly lower among male mice from the high arousal line (Fig. 2 D and H ). The pattern of sexual behavior indicates that high-arousal males were excitable in an inappropriate manner, as indicated by the very low intromission:total mount ratio. Importantly, the temporal structure of the mating bout was similar between the lines as there were no differences in the latency to mount, intromit, or ejaculate between the genetic lines and between offspring high and low arousal groups (Fig. 3 A–F ).
High generalized arousal is associated with high sexual arousal. All data are presented as mean (±SEM). Total number of mounts before first intromission broken out by (A) parental arousal, (E) offspring arousal, and (I) across both conditions. The number of intromissions before the first mount for (B) parental arousal and (F) offspring arousal. Latency to ejaculate after the first intromission in animals divided into (C) parental arousal and (G) and offspring arousal and intromission: mount ratio (number of successful intromissions/total number of mounts + intromissions) broken up by (D) parental arousal and (H) offspring arousal. Differences are considered statistically significant if P < 0.05. HM, n = 6, LM, n = 6.
Sexual behavior in high generalized arousal mice retains temporal structure. All data are presented as mean (±SEM). Latency to mount (A, D, and G), latency to intromit (B, E, and H), and latency to ejaculate (C, F, and I) do not differ between the lines or based on offspring arousal. Differences are considered statistically significant if P < 0.05. HM, n = 6, LM, n = 6.
Next, we asked the question, do increases in CNS arousal translate into increases in anxiety-like/exploratory behaviors? High and low mice of both sexes were tested on the elevated plus maze and light-dark transition tasks. Interestingly, differences were significant according to the parental level of arousal but were not systematically different among animals that differed in arousal themselves. In the elevated plus maze, mice from the high line exhibited an overall increase in anxiety-like behaviors (a decrease in exploration), as indicated by less time spent in the open arms (Fig. 4A ), a longer latency to first enter an open arm (Fig. 4B ), and an overall decrease in exploratory behavior as indicated by total arm entries (Fig. 4D ). In all cases, females, independent of breeding type, also spent less time in the open arm and exhibited a longer latency to first enter an open arm. There was not a consistent relationship between the offspring arousal scores and behavior in the elevated plus maze (Figs. 4 E–H ). In the light-dark test, mice from the high line did not differ in the time spent in the light (Fig. 5A ) but entered the dark side of the box after a longer interval (Fig. 5B ) and had significantly fewer transitions between the two sides of the box (Fig. 5C ). Again, only the parental arousal and not the offspring’s arousal were predictive of behavior in the light-dark task (Fig. 5 D–F ). Overall, animals from the high line exhibited more anxiety-like behavior and a reduction in overall exploration.
Selection for high generalized arousal induces anxiety-like behavior. All data are presented as mean (±SEM). Mice from the high Ag lines time spent in the both the open (A) and closed arms (B) and exhibited longer latency to enter an open arm (C) and exhibited fewer total arm entries (D). There was no relationship between the individual scores on the arousal assay (E–H). Differences are considered statistically significant if P < 0.05. HF, n = 22, LF, n = 22, HM, n = 24, LM, n = 15.
Selection for high generalized arousal alters light-dark transition behavior. The lines did not differ in total time spent in the light (or dark) (A) but high Ag animals exhibited a longer latency to enter the dark side of the chamber (B) and fewer overall light-dark transitions (C). There was no relationship between the individual scores on the arousal assay arousal (D–F). HF, n = 22, LF, n = 22, HM, n = 24, LM, n = 15.
Finally, to extract information about the most prominent feature of the data gathered from our generalized arousal assay, we used a mathematical method called principal components analysis. This method is used here to analyze the relative contributions of motor, sensory, and emotional (fear) measures as they influence the largest, most elementary dimension of arousal. That is, the most generalized, elementary force operating in our arousal assay is revealed by a forced one-component solution of our data set (2). The most interesting comparisons to come out of the principal components analysis are illustrated in Fig. 6. It demonstrates the separate contributions of Motor Activity, Olfactory Responsivity, and Fear to Principal Component #1, the component that quantifies the most generalized, powerful force generating behavior in these arousal assays. In Fig. 6, when a measure has a (–) sign, that means that it was, indeed, grouped with the forces on Principal Component #1 but in the reverse direction (low values of that behavior are strongly associated with Principal Component #1’s contribution to the production of arousal-related behaviors). Principal Component #1 reflects a high degree of motor activity. Our analysis raises the question of whether the structures of arousal functions are the same in males and females.
Mathematical structure of generalized arousal is different between males and females selected for divergent levels of generalized arousal. Differential contributions of motor, sensory (olfactory), and emotional (fear) measurements to the most generalized force driving behavior in the arousal assay, namely, Principal Component 1. The patterns of these contributions differed between HM and LM, between HF and LF, between HM and HF, and between LM and LF. For example, LM’s were low because motor measurements did not drive their Principal Component 1. Furthermore, HM had a high, positive contribution of fear to Principal Component 1 (HF did not), but HM lacked HF’s strong contribution of olfactory responsiveness. Controls using these same large sets of data scrambled and controls using random numbers failed to yield similar patterns and sharply reduced the percentage of the data explained by Principal Component 1.
Fig. 6 shows that the major differences between HM and LM come from the large contribution of motor activity of HM to Principal Component #1, as well as a difference in the contribution of fear. In fact, it is the failure of motor activity driven by Principal Component #1 that makes those males LM rather than HM. HM have high movement rates and are skittish. Females are different. The major difference between HF and LF comes from the strong reactivity of the HF to olfactory input. With respect to sex differences, between HM and HF there are large differences in the contributions of olfactory responsiveness to Principal Component #1, as well as fear. We speculate that an HF female ready to mate, having spent much time in her burrow, will emerge from her burrow just before ovulation. She must lack fear and locomote extensively, spreading the odor of vaginal secretions, a form of courtship behavior that encourages males to mate just as she is ovulating. In turn, her powerful olfactory response will help her choose healthy vigorous males as potential fathers of her litter (23, 24). Between LM and LF, the major difference is due to the fact that the LF had a large motor contribution to Principal Component #1, as well as a smaller difference between LM and LF in fear. From this application of principal components analysis, we infer that the structure of the primary arousal component is not the same in males as in females.
Discussion
These data provide evidence that genetically altering generalized arousal has a profound influence on sexual and exploratory/anxiety behaviors. The male sexual behavior in this study was characterized by HM animals being more excitable, showing more rapid movements, exhibiting many premature and unsuccessful mounts before their first successful penile insertion, and then ejaculating rapidly after a minimal number of additional intromissions. Thus, the ratio of intromissions to excited preintromission mounts was significantly lower in HM compared to LM animals. Interestingly, these results appeared whether animals were sorted by their parents’ or by their own arousal scores. Additionally, selection for high levels of generalized arousal resulted in reduced exploratory behavior in the light-dark and elevated plus maze tasks. This result, surprisingly, was a function of parental arousal scores and was independent of generalized arousal scores in the tested animals themselves.
If generalized arousal has implications for specific arousal subtypes, then it stands to reason that there should be physiological links between the two concepts. In the current work, we focused on two specific heightened states of arousal: sex and fear. We also have recently delineated a pattern of increases in female generalized arousal and sexual arousal following estrogen administration (3, 25). This occurs both directly and also indirectly, via the induction of specific and generalized arousal-related neurochemicals. Indeed, a number of neurochemical signals that promote generalized arousal throughout the CNS also promote sexual arousal in limbic structures (1, 26). For instance, the arousal-related neurotransmitter histamine is a potent arousal-promoting chemical that increases cortical activity and inhibits the sleep promoting neurons of the ventrolateral preoptic area. Male mice lacking the histamine synthesis enzyme histidine decarboxylase exhibited reduced mating behavior and prenatal exposure to antihistamines permanently impaired male sexual behavior (27, 28). Additionally, histamine in the ventromedial hypothalamus facilitates both electrical activity and lordosis behavior in female rodents (21, 29). Similarly, hypocretin peptides increase overall CNS arousal and microinjections of hypocretin into the medial preoptic area increase male sexual behavior (22). The inverse example also provides evidence for the same point, anesthetics administered before priming doses of estradiol prevent the expression of lordosis behavior and the induction of mating-related genes (30, 31) whereas amphetamine administration facilitates estrogenic induction of mating (32). Taken together, these data forge a link between sexual behavior and generalized arousal and indicate potential mechanisms by which genetic selection for high generalized arousal could impact specific arousal states such as sex.
Genetic selection for high levels of generalized arousal resulted in reduced exploratory behavior in the light-dark and elevated plus maze assays. A reduction in exploratory behavior in these tasks could be conceptualized as indicative of anxiety-like behavior (33). That is, high levels of generalized CNS arousal produced greater home-cage locomotor activity but reduced exploratory activity in a novel environment (light-dark box and elevated plus maze). We infer that the increased locomotor drive in High arousal animals is more than offset by arousal-induced increases in anxiety-like states that would serve to suppress exploration. Importantly, although sexual behavior was affected by both parental and offspring arousal, exploratory/anxiety-like behavior was only altered by the arousal state of the parents. Exactly how this parental strain effect occurred remains to be determined.
A linkage of the generalized arousal concept to specific behavioral states is also implied by results with other systems. For instance, a circadian function that has received intense scrutiny is the regulation of sleep. Currently, more than 15% of adults in the United States have some kind of sleep disorder. Although the biological functions of sleep remain controversial, the locations and some of the properties of neurons turned off during sleep have been charted (34). Ultimately, the regulation of sharp transitions between sleeping and waking may depend on negative feedback loops among neurons in the hypothalamus and basal forebrain (4). Problems in the controls over arousal mechanisms manifest as sleep problems are extremely common (35), as are arousal problems related to depression (36) and stress (37). Further, Aston-Jones et al. (38) have been able to implicate the activation of arousal-related hypocretin neurons in the lateral hypothalamus in reward-seeking behaviors. For each of these specific behavioral states—e.g., sleep, stress, mood and reward-seeking—generalized arousal provides the most elementary, primitive neuronal force for the activation of behavior, and its exact behavioral impact is shaped by the specific environmental circumstance.
Alternative Interpretations.
Our measures of arousal are all dependent on locomotor activity as the principal readout. Therefore, it could be argued that artificial selection for a gene that would tend to increase the overall level of locomotor activity would be interpreted (or misinterpreted) as an increase in CNS arousal. However, the specific details of our arousal assay do not support such a simplistic explanation. Indeed, for the sensory and fear components of the assay, we subtract background activity from the poststimulation responses to determine behavioral reactivity. As an additional issue of interpretation, it should be pointed out that differential sleep states are a potential random confound for the sensory component of the assay. However, that confound is minimized by the presentation of the rotational (vestibular) stimuli that is likely to awaken animals before the olfactory stimulation that is the particular subtest upon which selection was based. In any case, in novel contexts like the elevated plus maze and light-dark box, the selection for higher generalized arousal produces less overall exploratory behavior as compared to Low-selected mice. Rather than simply selecting for greater locomotor activity, the High mice appear to be exhibiting greater generalized CNS arousal in concert with increases in behaviors related to both fear and sex drives.
An additional alternative explanation is that we have produced differences in a single gene involved in the regulation of arousal-related neurotransmission. Because the arousal neurochemical systems are reciprocally regulated and overlapping the alteration in a gene for one of these systems could induce increases in generalized CNS arousal and behavior. Further, we recognize that we have presented results only from the most recent generation; consequently, the one that showed the greatest quantitative separation between High Arousal and Low Arousal lines. Data from additional generations will become available in the coming year. These lines may be useful to investigate the anatomical and genetic mechanisms that environmental pressures have engaged to adjust generalized arousal across developmental stages between sexes and over the 24 h light/dark time period. Future, studies will address the neurochemical, anatomical and genomic differences between the High and Low lines.
In summary, mice genetically selected for high levels of generalized arousal are producing a clear phenotype characterized by greater home cage activity, sensory responsivity, and emotional lability. Interestingly, a High Arousal phenotype is associated with exaggerated sexual arousal and reductions in exploratory behavior, when compared to mice selected for low levels of arousal. Taken together, these data provide support for a potential role for generalized arousal as a driver of specific motivated behaviors.
Methods
Methods.
All experiments were carried out in accordance with National Institutes of Health Guidelines and followed procedures approved by The Rockefeller University Institutional Animal Care and Use Committee.
The strain used in this study was derived from an extensively outbred stock, Het-8, that resulted from an extensive intercross of more than eight outbred strains followed by more than 60 generations of structured outbreeding (39). Mice were housed in groups of four to five same-sex siblings in standard laboratory conditions with ad libitum access to food and filtered tap water. All animals were housed in 12:12 light/dark cycles (lights on at 0600).
Breeding Procedure.
The animals in this manuscript represent our sixth generation of selection for both high and low levels of generalized arousal. In each of the subsequent generations, mice were tested in the generalized arousal assay and an overall arousal score was generated (see SI Text for descriptions of the behavioral tests). Briefly, the total distance traveled over the 24 h day in the home cage assay, the horizontal activity from the olfactory stimulus, and the relativized change in vertical activity on the fear training session were used as selection variables. Animals were rank-ordered for their scores on each of these variables, the scores were added and the animals with the most extreme scores (six highest and lowest) were selected as the founders of the next generation. Breeding is ongoing.
Supplementary Material
Footnotes
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/cgi/content/full/0914014107/DCSupplemental.
;
