Animal Models of Psyc h i a t r i c Dis o r ders and Their Rel e va n c e to Alcoholism

Robert Hitzemann, Ph.D.

Animal models are important tools in the study of psychiatric disorders, including alcoholism,1 because they allow the use of research methods that cannot be used for ethical reasons in humans. Consequently, scientists have developed numerous approaches to evaluate the validity and reliability of animal models for studying human behavior and human disorders. Researchers have developed animal models of , fear and anxiety, depression, and alcoholism, all of which are being used to study the relationship between alcoholism and co-occurring psychiatric disorders. These models may help researchers and clinicians determine how best to treat patients with alcoholism and co-occurring psychiatric disorders. KEY WO R D S : animal model; behavioral and mental disorder; AOD (alcohol or other drug) dependence; comorbidity; validity (research methods); reliability (research methods); schizophrenia; emotional and psychiatric depression; fear; anxiety; animal strains

cientists are increasingly being asked can create genetically altered mice (e.g., connections), and theref o r e influence both to develop and to defend the mice with a foreign gene added to their be h a v i o r , are probably similar in organ- Suse of animal models for studying genetic makeup) to examine the influence isms as “si m p l e ” as the worm and as ps y chiatric disorders that appear to be of specific gene products on behavior. “co m p l e x ” as the human. In higher ver - uniquely human, such as alcohol and Recent advances in genetic engineering tebrates, howeve r , these basic proc e s s e s other substance abuse, schizop h re n i a , al l o w investigators to activate or inacti- combine into highly sophisticated and de p r ession, and anxiety. These disorde r s vate specific genes in specific regions of complex arrays that no computer system either are not observed in animals or the brain. Other res e a r ch techniques has yet been able to duplicate. Thus, cannot be directly measured (i.e., the that scientists can only use in animals th r oughout the foreseeable future, animals cannot verbally indicate what include injecting materials directly into they think or feel). This rev i e w exam- the brain and implanting brain electrod e s . ines some of the animal models used Critics of animal models in behavioral ROB E R T HIT Z E M A N N , PH.D . , is a prof e s - to study these disorders, with a special res e a r ch have argued that these studies sor and chairman of the Dep a r tment of emphasis on the use of these models in could be conducted using animals with Beh a v i o r al Neu r oscience, Oregon Hea l t h the search for genes that contribute to simpler systems, such as worms, or Sciences Uni ve r s i t y , Por tland, Oreg o n , the disorde r s . computer simulations. This argument and a res e a r ch pharmacologist at the The general argument for using ani- is based on the theory that the basic Res e a r ch Ser vice, Vet e r ans Affairs mal models in behavioral res e a r ch is that pr ocesses that increase or decrease the Medical Cen t e r , Por tland, Oreg o n . such models allow res e a r chers to test st r ength of communication betwee n specific hypotheses under highly con- and among cells (i.e., the synaptic This res e a r ch was supported in part by tr olled conditions using methods that Public Health Ser vice Gran t s ar e either impossible or unethical to use 1For the purposes of this article, alcoholism is considered AA–11034 and MH 51372 and a gran t in humans. For example, res e a rc h e r s a psychiatric disorder. fr om the Dep a r tment of Vet e r ans Affairs.

Vol. 24, No. 3, 2000 149 res e a r chers investigating complex behav- ma r kers that may indicate a susceptibility be measured under similar conditions in io r a l traits and characteristics (i.e., to a disorder). Endophenotypes generally la b o r a t o r y animals and humans). How- phenotypes) of humans will continue ar e not immediately visible, but they eve r , as previously stated, face val i d i t y to be forced to use animals with wel l - may contribute to the susceptibility to for an endophenotype does not justify de v eloped central nervous systems rather de v elop a particular behavior or syn- its use. than simpler organisms or computer dr ome. The use of endophenotypes is Pred i c t i v e val i d i t y refers to the extent models. Although rodents are not as helpful in that the underlying neurob i - that an animal model can allow a re- ap p r opriate as are primates for modeling ology (i.e., the mechanisms or mecha- se a r cher to make predictions about the all aspects of a specific human behavior, nism by which the endophenotype human phenotype. In many cases, pre- rodents will carry the main workload of di c t i v e validity refers to how useful animal animal res e a r ch for some time to come. models are for predicting the efficacy This article describes animal models and safety of drugs for treating psych i a t r i c of schizop h r enia, fear and anxiety, and di s o r ders. For example, new antipsy- de p r ession. It also discusses the use of chotic drugs wer e screened for years animal models in studying alcoholism Rodents will by their ability to block (in rod e n t s ) and co-occurring psychiatric disorde r s . ca r r y the main amphetamine-induced behaviors (e.g., First, howeve r , validity and rel i a b i l i t y — in c r eased licking, gnawing, and chewi n g ) . the criteria used to evaluate animal wo r kload of This test was based on the hypothesis mo d e l s — a r e defined. that because amphetamine increa s e s animal res e a r ch le v els of dopamine (a key brain chemical for some time in vo l v ed in nerve cell communication) Criteria for Animal Models and is associated with increa s e d of Human Behavior to come. le v els of dopamine, drugs that block amphetamine-induced behavior would Validity and reliability are the main cri- be effective as antipsych o t i c s . teria for evaluating animal models (for Howeve r , the drugs selected using a rev i e w, see Gey er and Mar kou 1995). this test also produced extra-pyramidal This article first examines the four aspects in c r eases susceptibility) is freq u e n t l y symptoms (EPS), such as drug - i n d u c e d of validity—face, pred i c t i v e, etiological, kn o wn and thus res e a r chers have the Par kinsonism. EPS was one of the major and genetic—and then focuses on the ad v antage (i.e., in the case of genetic reasons that people did not comply with issue of rel i a b i l i t y . studies) of linking specific gene(s) that dr ug treatment. This problem even t u a l l y Face val i d i t y refers to the similarity may influence the behavior with spe- led to the development of new atypical be t w een the animal model and the spe- cific brain mechanisms. A disadvan t a g e an t i p s y chotic drugs (i.e., drugs that cific human behavior of interest. Sup e r - of endophenotypes, howeve r , is their maintain antipsychotic efficacy but do fi c i a l l y , it would seem that the animal lack of specificity. For example, prep u l s e not produce EPS). New drugs wer e behavior should mimic the related human inhibition of the acoustic startle res p o n s e sc r eened for a particular receptor pro- behavior as much as possible. For exam- (A S R ) 2 se r ves as an operational measure, file (i.e., either high or low affinity for pl e , mouse and man display similar or endophenotype, for the deficits in a particular receptor—that is, a prot e i n responses to a loud noise (i.e., st a rt l e the mechanisms that allow normal that binds to a specific brain chemical, responses). Both respond with an eye - individuals to filter or block most of or neurotransmitter). Once these com- blink, a rapid change in heart rate, an d the sensory and cognitive stimuli that pounds wer e identified, the amphetamine some form of whole body movem e n t they rec e i v e (i.e., abnor- test was used to determine if the new (e.g., a mouse jumps, a person jerks ) . malities). Such abnormalities are often compounds could block the drug - Howeve r , when studying the reg u l a t i o n seen in schizop h re n i a (see table 1). induced behaviors by a novel mechanism. of the startle response, the observed Howeve r , whereas schizop h re n i c s sh o w Sc r eens such as this led to the devel o p - response need not be obviously similar. poor , deficits also ment of a new generation of atypical For inves t i g a t i v e purposes, the res u l t s ar e seen in Tou re t t e ’s syndrome, bed- an t i p s y chotics (e.g., risperidone and would be just as informative if the mouse wetting incidents, and during the men- ol a n ze p i n e ) . yawned or curled its tail, providing that st r ual cycl e . Etiological val i d i t y exists when one these phenotypes could be rel i a b l y Furt h e r m o r e, the popularity of can determine—to the extent possi- me a s u r ed and had pred i c t i v e, etiological, endophenotypes may derive partly from ble—that the mechanisms invol v ed and genetic validity (described later in the fact that they generally have face in a behavior observed in the animal this arti c l e ) . validity (i.e., a complex human behavior model are similar to those invol v ed Res e a r chers studying animal models has been reduced to something that ca n in the associated human behavior. of human psychiatric disorders are His t o r i c a l l y , res e a r chers have established in c r easingly examining intermediate 2The degree of startle response is reduced when a brief etiological validity by using the animal phenotypes or endophenotypes (i.e., “warning” stimulus precedes the startle-eliciting stimulus. model to determine the underlying

150 Alcohol Research & Health Animal Models of Psychiatric Disorders and Their Relevance to Alcoholism

mechanisms of behavior and then by 1999), we obtained results essentially se n s o r y deficits when using inbred mouse testing the role of such mechanisms identical to those obtained by Cra b b e strains. Num e r ous strains, including in human behavior. In coming yea r s , and colleagues (1983). some of those most frequently used (e.g., ho weve r , the rev erse process prob a b l y Stability of the phenotype also can DBA/2J, C57BL/6J, and BALB/cJ) will also be increasingly used: res e a rc h e r s refer to test-retest rel i a b i l i t y , a construc t ha v e marked high frequency hearing will first examine the underlying mech- fr equently used in studies on humans. deficits, which in the DBA/2J can be anisms in humans and then examine Scientists working with animals, how- detected as early as 4 to 6 weeks of age. animal models to determine whether eve r , have been reluctant to use a test- Other strains (e.g., the C3H/HeJ) show similar mechanisms are at work. retest design; arguing—and freq u e n t l y ma r ked retinal degeneration at an early For example, in a group of normal demonstrating—that the first test will age. Obvi o u s l y , such animals would be human controls, Vol k o w and colleagues significantly affect the second test. When unsuitable for tests that req u i r e the ani- (1999) found that human subjects with wo r king with inbred (i.e., genetic a l l y mal to remember spatial cues. In addi- naturally low D2 dopamine rec e p t o r identical) animal strains, the extent to tion, albino rats and mice show impaired density rep o r ted a euphoric effect in which this is a problem can be meas- vision under bright lights, which they response to an intravenous injection ur ed easily by comparing animals tested ar e exposed to in the standard open-field of methylphenidate (Ritalin™). These once with animals tested multiple times. test (see table 1), a deficit that must be data led Thanos and colleagues (1999) Howeve r , when using genetically segre- accounted for when interpreting their to investigate whether elevating D2 gating populations (i.e., in which each pe r formance. Although the avai l a b i l i t y receptor density in rats would red u c e individual is genetically unique), such of transgenic mice is an advantage for alcohol pref e r ence. An increased rec e p - a solution is not possible. In order to res e a r ch, this technology presents dis- tor density was associated with red u c e d make a sound decision, res e a r chers must ad v antages as well (see the article in alcohol pref e r ence. Thus, the res e a rc h e r s weigh the advantages and disadvan t a g e s this issue by Bowers, pp. 175–184). would test the relationship betwee n of repeatedly testing animals. Bec a u s e receptor density and drug pref e re n c e the main effect of poor test-retest rel i a - under controlled laboratory conditions. bility is to decrease statistical power , Models of Psychiatric Genetic val i d i t y exists when the risk in c r easing sample size and testing only Disorders for a disease is known to have a similar once is a solution to the prob l e m . genetic component in both humans and Inv estigators use animal models to study the animal model. At least 50 perce n t Suitability of the Animal Species the mechanisms invol v ed in sever a l of the risk for psychiatric disorders such for a Specific Experimental Need ps y chiatric disorders and to inves t i g a t e as schizop h r enia, depression, and alco- the relationship between alcoholism holism can be attributed to inherited His t o r i c a l l y , behavioral studies have and co-occurring disorders. This section genetic factors. Thus, for any animal largely relied on the use of the albino describes animal models for schizop h r e- model of these disorders, it would be ou t b r ed rat (i.e., each animal is genetically ni a , de p r ession, and fear and anxiety. The useful—but not necessary—to demon- unique) as the primary experimental use of animal models to study al c o h o l i s m strate a significant genetic component species. For genetic studies, howeve r , and co-occurring psych i a t r i c di s o rd e r s to the natural phenotypical var i a t i o n . mice offer advantages over rats, such is discussed in the subsequent section. Various strategies can be used, with the as the availability of numerous inbred ev entual goal of detecting rel e v ant genes. mouse strains, animals in which genes Schizophrenia Res e a r chers have made considerable ha v e been added or deleted (i.e., trans- pro g r ess in this area for behavioral phe- genic animals), and a wel l - d e s c r i b e d Sc h i zo p h r enia is one of the most dis- notypes (Crabbe et al. 1999), although genome. When rat paradigms are adapted abling of all psychiatric disorders. The results still lag behind the detection of to the mouse, especially paradigms lifetime risk of developing schizop h re n i a genes for other complex phenotypes in v olving higher order learning, mice is approximately 1 percent, a risk rate (e.g., diabetes, obesity, and hyperte n - tend to do poorly (Crawley et al. 1997). that appears to be largely independent sion) (e.g., Pomp 1997). Whether or not this difference ref l e c t s of race, culture, and socioeconomic sta- a difference in “intelligence” or the need tus. Schizop h r enia is characterized by Reliability of the Model for new protocols specifically suited to both positive symptoms (e.g., delusions, the mouse’s abilities and temperament hallucinations, and thought disorde r ) The reliability of the animal model is unclear. and negative symptoms (e.g., withdrawal refers to the stability and rep ro d u c i b i l i t y Mice present other disadvantages and a dementia-like state). In addition, of the phenotype. Some animal pheno- to res e a r chers as well. For example, the mechanisms of schizop h r enia appear types appear to show rem a r kable stability. res e a r chers frequently encounter marke d to be highly diverse. Giv en this level of For example, by using alcohol-induced co m p l e x i t y , the difficulty in devel o p i n g locomotion as a measure of alcohol’s an animal model that can capture the 3 3When some mice are administered low doses of alcohol, acute stimulant res p o n s e in a series of their locomotor activity increases. This measure is often sc h i zo p h r enic pathology is apparen t . in b r ed mouse strains (Dem a r est et al. used to test for initial sensitivity to alcohol. Prog r ess has been made in this effort,

Vol. 24, No. 3, 2000 151 Table 1 Summary of Animal Models1

Behavior Modeled Phenotype Description

Startle Reactivity Acoustic Startle Both the ASR and TSR measure startle reactivity and, in general, show a Response (ASR) good correlation to each other. The ASR is elicited with a loud noise; the TSR Tactile Startle is generally elicited with an air puff. The response (reflex) is measured by cou- Response (TSR) pl i n g the startle platform to a strain gauge transducer or some similar device. In humans, the startle response is generally measured by the strength of the eye-blink reflex. Habituation to the startle response is abnormal in schizophrenia.

Sensorimotor Prepulse Inhibition PPI refers to a reduction in the startle response that is observed when a brief Gating (PPI) (e.g., 20 millisecond) sensory stimulus (the prepulse stimulus) is delivered prior (100 to 1000 milliseconds) to the stimulus that induces the ASR or TSR. The prepulse stimulus inhibits the magnitude of the ASR or TSR. The prepulse can be different from the startle stimulus. In general, treatments that increase brain dopaminergic activity decrease PPI. PPI is reduced in schizophrenia. Ho w e v e r , any disorder (e.g., drug abuse) that is likely to involve brain dopamine systems is a candidate to affect PPI.

Associative Latent Inhibition LI refers to the inhibition of a conditioned response (e.g., avoidance of a shock) Processes (LI) caused by preexposure to the conditioned stimulus (CS) (e.g., a loud tone). The preexposure phase consists of random exposure to a CS. In the test phase, the CS is presented with an aversive unconditioned stimulus (US) (e.g., a scrambled footshock). Preexposure to the CS delays acquisition of the conditioned avoidance response. Numerous human paradigms are available for measuring LI. LI is reported to be abnormal in schizophrenics (i.e., they more quickly make the CS–US association). In general, the animal data sug- gest that PPI and LI are measuring different processes.

Conditioned Fear Fear Potentiated In a typical FPS paradigm, the degree of conditioned fear is reflected by the Startle (FPS) increased amplitude of the ASR or TSR elicited in the presence of a CS previ- ously paired with an aversive US. In humans, the ASR is enhanced both in anticipation of aversive shock and in association with certain disorders (e.g., posttraumatic stress disorder).

Contextual Fear Contextual fear conditioning is measured by noting the change in locomotor Conditioning activity (generally bouts of freezing in place) when the animal is placed in the chamber where the CS was paired with an aversive US.

Cued Fear Cued fear conditioning is conceptually similar to FPS. Rather than measuring Conditioning the change in ASR, the change in locomotor activity is measured after presen- tation of the CS. To avoid confusing cued fear with contextual fear, researchers must ensure that the test chamber and conditioning chamber are viewed differently.

Unconditioned Open-Field The open-field test assesses some aspects of unconditioned fear and anxiety. Fear Activity The standard open field consists of a brightly lit, circular arena with opaque walls. The floor of the open field is scored in a grid. The animal is placed on a small stage in the open field. Outcome is measured by total activity, activity in peripheral versus central grids (i.e., thigmotaxis), rearing, and defecation. High anxiety and emotionality is evidenced by low activity, especially in the central grids, and increased defecation. Automated devices that retain some aspects of the historical open-field apparatus are available.

152 Alcohol Research & Health Animal Models of Psychiatric Disorders and Their Relevance to Alcoholism

Table 1 Summary of Animal Models1 continued

Behavior Modeled Phenotype Description

Light-Dark The light-dark transition test forces the animal to make a choice between explor- Transition in g the brightly lit area of the open field or staying in the dark area (the natural tendency). Anxiety-reducing drugs increase exploration in the bright area.

Elevated Plus The elevated plus maze is similar in concept to the light-dark test except that Maze the animal must choose between exploring a novel environment and the tendency to escape from the elevated open arms of the maze.

Depression Reserpine- This is one of the earliest models using drugs to mimic a psychiatric disorder. Induced Behaviors It is based on the clinical observation that 10 to 20 percent of patients admin- istered reserpine (generally for hypertension) report symptoms of depression. In rodents, moderate doses of reserpine induce locomotor depression. The model is no longer used.

Learned Learned helplessness refers to a variety of paradigms in which the animal is Helplessness exposed to an inescapable aversive stimulus (e.g., a foot shock). Eventually, a certain proportion of the animals will give up and make no attempt to escape. The model has numerous problems but does show predictive validity in that antidepressants and electroconvulsive shock reverse the learned helpless- ness in susceptible animals.

Behavioral Conceptually, behavioral despair is related to learned helplessness. In the Despair typical test, the mouse or rat is forced to swim in a tank from which it cannot escape. At the outset, the animals swim vigorously; however, over time some of the animals become immobile, a condition Porsalt (1981) termed behavioral despair.

Psychosis Chronic Stimulant This model builds from the repeated observation that the chronic administra- Treatment tion of central nervous system stimulants can induce paranoid psychosis. In animals, chronic stimulant administration will generally induce sensitization to some behaviors (e.g., stereotyped activity) and habituation to others (e.g., exploratory activity).

N-methyl-D- Clinically, drugs that act by blocking NMDA receptors (i.e., NMDA receptor aspartate (NMDA) antagonists), such as phencyclidine (PCP), ketamine, and dizoclipine induce Receptor temporary psychotic symptoms and cognitive disturbances somewhat mimick- Antagonists ing the negative symptoms of schizophrenia. In animals, the effect of these drugs is generally measured by the temporary increase in locomotor activity.

Psychotomimetic Similar in concept to the NMDA receptor antagonist model, this model generally Drugs refers to drugs, such as LSD and mescaline, that induce hallucinations, particu- larly visual hallucinations. Although visual hallucinations are uncommon in schizophrenia and related psychotic disorders, understanding this relatively rare phenotype would be of value. Animal models generally focus on acute drug administration; response is measured as changes in locomotor activity or behavioral disruption.

1This list of animal models is not intended to be comprehensive but to provide the reader with an exposure to models that have been important in understanding psychiatric disorders. References to the models are found in the text.

Vol. 24, No. 3, 2000 153 and a few of the approaches used are pr epulse inhibition of the acoustic st a r ting points for future inves t i g a t i o n s — described here. st a r tle response and latent inhibition st a r ting points that would not have been Sc h i zo p h r enia is characterized by (re v i e wed in Kline and colleagues 1998). possible without the use of animal studies. deficits in attention-related proc e s s e s , These data led us to investigate prep u l s e In addition to the endophenotype information processing, and the sensing inhibition and latent inhibition in lines ap p r oach, res e a r chers have used other and filtering (i.e., sensorimotor gating) of mice that had been selectively bred strategies to model aspects of schizop h r e- of environmental stimuli (Braff 1993). to respond or not respond to the drug ni a , including administering hallucino- Prepulse inhibition of the acoustic star- haloperidol, a typical antipsych o t i c gens and acute and chronic administration tle response (described in table 1) has (also termed neuroleptic). Catalepsy, of stimulants (e.g., amphetamine). A been used to measure the deficits in number of problems occur with these sensorimotor gating associated with ap p r oaches (see Gey er and Mark o u sc h i zo p h r enia. The data obtained led 1995). Because hallucinations cannot to increased interest in the related animal The sensitization be observed in mice and rats, res e a rc h e r s model, an approach that has sever a l ha v e used endpoints, such as behavioral ad v antages. Mea s u r ements of prep u l s e model has been di s r uption or locomotor stimulation inhibition are easily adapted to mice adopted by alcohol (many hallucinogens are also stimulants). and rats, and the underlying mecha- Furt h e r m o r e, the hallucinations induced nisms regulating prepulse inhibition and other drug by drugs like LSD are primarily visual, appear to be similar, if not identical, abuse res e a rc h e r s to wh e r eas those seen in schizop h r enia are in animals and humans. Antipsych o t i c mo r e likely to be auditory or olfactory. dr ugs improve poor prepulse inhibi- in v estigate the adaptive Stimulant models of psychosis aros e tion, and dopamine agonists (i.e., com- after rep o r ts that stimulant abuse res u l t e d pounds that combine with dopamine and pathological effects in paranoid delusions. Res e a r chers con- receptors) decrease prepulse inhibition. of chronic intoxi c a t i o n . ducted controlled chronic administra- Fin a l l y , the regulation of prepulse inhi- tion studies to demonstrate that the bition in animals is markedly influenced paranoia was not associated with a pre- by genetic factors. Thus, prepulse inhi- existing psychiatric disorder but res u l t e d bition has face, pred i c t i v e, construc t , the mouse equivalent to extra pyrami- fr om a maladaptive drug response. The and genetic val i d i t y . dal symptoms, was used to measure ch r onic administration of stimulants to In addition to prepulse inhibition, response to the drug (see table 1). Two most strains of rats and to some strains poor latent inhibition is also associated lines wer e created, neuroleptic res p o n s i v e of mice leads to drug sensitization (i.e., with schizop h r enia (Braff 1993) and can (NR) and neuroleptic nonres p o n s i v e in c r eased sensitivity) characterized by be easily measured in animals. Latent (NNR), which differ about twentyfold in an increase in certain behaviors (e.g., inhibition refers to the inhibition of a their sensitivity to haloperidol. Num e ro u s patterned locomotion, highly rep e t i t i v e conditioned response as a result of pre- di f f e r ences have been observed in brain gr ooming, and gnawing and licking ex p o s u r e to the conditioned stimulus. dopamine systems between the NR mo vements). Thus, res e a r chers devel o p e d In our laboratory, we have measured and NNR lines, including the number an animal model in which increa s e d latent inhibition in mice as follows. Mic e of dopamine neurons and the density motor responses substituted for the wer e randomly divided into two grou p s . of D2 dopamine receptors (Hit ze m a n n paranoid delusions observed clinically. One group was given 50 trials, each et al. 1995). Despite the lack of face val i d i t y , the consisting of a 10-second conditioning Giv en these differences, unsurpris- argument was made that the model stimulus (CS) (i.e., a 2,000 hertz, 85 in g l y , Kline and colleagues (1998) was likely to have etiological val i d i t y . decibel tone); each trial was separated found marked deficits in both prep u l s e Furt h e r m o r e, the stimulant sensitiza- by 5 to 75 seconds. The second grou p inhibition and latent inhibition in the tion model both developed from and (i.e., the control group) did not rec e i v e NNR mice when compared with the added to the hypothesis that psych o s i s the auditory CS. On the second day, NR line. From these data, we can con- is related to excesses in dopamine activ- both groups wer e assessed for the acqui- clude the following, at least in mice: it y . In recent years, res e a r chers inves t i - si t i o n of a conditioned avoidance res p o n s e (1) some of the same genes reg u l a t i n g gating psychosis and schizop h re n i a (CAR) in which the CS was associated the catalepsy response also regulate pre- ha v e favor ed other theories over the with a mild footshock. Success on the pulse inhibition and latent inhibition, dopamine hypothesis. Howeve r , the test was measured by how quickly the and (2) elucidating these genes will likely sensitization model has been adopted by animals learned to avoid the mild shock. be important to understanding the alcohol and other drug abuse res e a rc h e r s Latent inhibition was observed in the underlying mechanisms of these behav- to investigate the adaptive and patho- pr eexposed group (i.e., they acquired iors. The question of whether or not logical effects of chronic intoxi c a t i o n . the CAR more slowl y ) . such studies will illuminate schizop h re n i c The validity and heuristic value of this Considerable evidence suggests that pathology remains unclear. Howeve r , ap p r oach is detailed in this issue and in dopamine has a reg u l a t o r y role in both the information obtained provides new volume 24, number 2.

154 Alcohol Research & Health Animal Models of Psychiatric Disorders and Their Relevance to Alcoholism

Another approach to modeling Thus, depression was thought to devel o p is not a common adaptive mechanism. sc h i zo p h r enia, which has elements of fr om an imbalance between high cholin- Furt h e r m o r e, among inbred strains of both the hallucinogen and stimulant er g i c activity on one side and low sero- mice, some will exhibit helplessness in strategies, examines drugs that act by tonergic or noradrenergic activity on the absence of training (i.e., the B6 blocking receptors for N-m e t h y l - D- the other. mouse free zes in response to shock and as p a r tate (NMDA) (i.e., NMDA antag- Learned helplessness and the rel a t e d does not attempt to escape). These data on i s t s ) , such as phencyclidine (PCP), phenotypes of behavioral despair and suggest that some of the differen c e s ketamine, and MK-801. Soon after ch r onic unpredictable stress have been be t w een helpless and nonhelpless animals the introduction of PCP in the 1950s, used to model some aspects of depres - may relate to natural differences in the res e a r chers rec o g n i z ed that the drug pattern of motor responses. Furt h e r m o re , could produce both positive and negative other investigators have emphasized the symptoms of schizop h r enia in normal The importance of difficulty in unraveling helplessness from co n t r ol subjects and intensify psych o t i c the learned helplessness a shock-induced reduction in motor symptoms in chronic schizop h re n i c s . activity (see Willner 1991), which is Animal models developed from these model in understanding commonly seen in rodents in res p o n s e ob s e r vations have largely focused on to any chronic unpredictable stres s NMDA antagonist-induced increa s e s the pa t h o l o g y of (e.g., crowding, irregular feeding, and in locomotor activity and disruption of de p r ession builds from in t e r r upted sleep). In contrast with se n s o r y gating (e.g., prepulse inhibition these concerns, the model has had rea - of the acoustic startle response) (Geye r th e argument that sonably good pred i c t i v e validity from a and Mar kou 1995). ps y chopharmacological perspective; a helplessness is a wide variety of antidepressants and Depression co m m o n ch a r acteristic el e c t ro c o n v u l s i v e shocks reduce the helpless behavior. Dep r ession is a common disorder (i.e., of depres s i o n . Behavioral despair is related concep- the lifetime prev alence is 15 to 25 per- tually to learned helplessness. In this cent) and is frequently encountered both sion. Mar tin Seligman originally devel - model, res e a r chers forced rodents to in the psychiatric clinic and the primary oped the learned helplessness model by swim in a tank from which they could ca r e setting. In addition to symptoms exposing dogs to shocks from which they not escape. At the outset, both the mice such as depressed mood, failing to find could not escape. The dogs even t u a l l y and rats swam vigorou s l y . Over time, pl e a s u r e from normally pleasurable ga v e up and made no effort to escape ho weve r , the animals become immobile, activities, and feelings of worth l e s s n e s s the shocks (i.e., they became helpless). a condition Porsalt (1981) called behav- and guilt, typical depression is charac- In addition, other behaviors wer e affected ioral despair. The underlying construc t te r i z ed by a number of veg e t a t i v e signs, (e.g., the dogs appeared apathetic and validity of this model was questionable including slowed activity, insomnia (or had poor appetites). (e.g., the immobility could simply have hypersomnia), and weight loss (or gain). The importance of the learned help- been an efficient coping strategy). How- One of the earliest models of lessness model in understanding the eve r , the model had high pred i c t i v e de p r ession was built from the clinical pathology of depression builds from the validity in identifying antidepres s a n t s , ob s e r vation that more than 10 perce n t argument that helplessness is a common especially after chronic administration of patients taking the antihyperte n s i o n characteristic of depression and that (Willner 1991). Thus, the swim test and dr ug reserpine develop some symptoms de p r ession can improve if the clinician modifications of it (e.g., coupling the of depression. In animals, reserpine causes instills in the patient a sense of mastery test with chronic unpredictable stres s ) locomotor depression. The observat i o n and control over the environment. Rather wer e well-used models. that the locomotor depression could than debating the value of this model be rev ersed by 5-hydr oxy t ry p t o p h a n (if any) to understanding depres s i o n , Fear and Anxiety (5–HTP) (a precursor to serotonin) or we will discuss the observance of learned di h yd r oxyphenylserine (DOPS) (a pre- helplessness in animals. The model has Both fear and anxiety are alerting signals cursor of norep i n e p h r i n e ) led to the se v eral problems. For example, only a that warn the individual against impend- de v elopment of the theory that depres - rel a t i v ely small prop o r tion of the ani- ing danger and enable the individual to sion occurred as a result of a de f i c i e n c y mals exposed to the inescapable shock take defensive measures. For animals, in serotonin or norep i n e p h r i n e (a theory become truly helpless. Among outbred the distinctions between fear and anxi- that has changed little in more than 40 stocks of rats, such as the Spr a g u e - Da w l e y ety are vague. For the purposes of this years). Res e a r chers have expanded the or Wis t a r , only 10 to 50 percent devel o p discussion, the animal models are th e o r y based on the observations that the helplessness syndrome (Wil l n e r divided into those that measure condi- inhibitors of the enzyme cholinesterase 1991). The low incidence may be tioned fear (i.e., fear that is learned by caused clinical symptoms of de p re s s i o n attributable to genetic factors, but the association with an aver s i v e stimulus) and psychomotor depres s i o n in animals. data also suggest that the phenomenon and unconditioned fear (i.e., fear that is

Vol. 24, No. 3, 2000 155 not learned). Unconditioned fear is one foot shock). On the second day, the by bouts of free zing, the B6 and D2 of the behavioral components measured animals are placed in the acoustic startl e ha v e essentially identical responses (Owen in the standard open-field test, one of apparatus (i.e., a device that measures et al. 1997). Thus, depending on the the most reliable measures in behav- acoustic startle response). The startl e conditioning paradigm used, the B6 ioral science. The test, which has been response is then measured in the pres e n c e response is greater than, less than, or de v eloped and refined over the past 70 and absence of the CS. The degree of equal to the D2 response. It is importa n t years (Hall 1934), consists of placing conditioned fear is measured by the to rec o g n i z e that such differences exist the animal in a brightly lit novel envi- in c r ease in the degree of the acoustic and that these differences must be kept ronment in which it cannot return to st a r tle response when the CS is pres e n t . in mind as one extends from the ani- the home cage. The novel environ m e n t In humans, the acoustic startle res p o n s e mal model to the clinic. and the light provide the stres s o r s , because rodents are photophobic. Inc r eased emotionality and anxiety is Animal models may Models of Psychiatric associated with a decrease in locomotor Disease and Alcoholism activity and rearing, increased activity in help determine which the periphery versus the central part of Animal models of alcoholism have the testing area (thigmotaxis), and defe- tr eatments are most been discussed elsewh e r e in this issue cation (see table 1). ef f e c t i v e in red u c i n g and in volume 24, number 2. Thus, Res e a r chers also measure uncondi- this article focuses on models that inte- tioned fear and anxiety in the light-dark alcohol consu m p t i o n grate alcoholism and other psych i a t r i c transition test and in the elevated plus diseases and describes some examples ma z e, both of which present approa c h - among people of such models. The significance of the av oidance conflicts. The light-dark tran- with an underly i n g models is attributable to the high comor- sition test is a modification of the stan- bidity of alcoholism with other disorde r s . da r d open-field test in which the animal ps y chiatric disorde r . The most common comorbid psych i a t r i c is provided a dark en c l o s u r e in the open- diagnoses with alcoholism are other field environ m e n t . The animal must is enhanced in anticipation of aver s i v e substance abuse-related disorders, anti- choose between its natural tendency to shock and in posttraumatic stress disor- social personality disorde r , mood disor- ex p l o r e the open field with the tendency der (see McCaughran et al. 2000). In de r s , and anxiety disorders. The question to escape from the bright light. The ele- rats, the FPS model appears to have of which di s o r der arises first rem a i n s vated plus maze pr esents the animal good pred i c t i v e validity in that the star- co n t r oversial. Animal models offer the with a similar conflict; the animal must tle response is weakened by anxiety- ad v antage of allowing res e a r chers to choose between the tendency to explore reducing drugs, such as diazepam, and test the interaction from both direc t i o n s a novel environment and the tendency enhanced by anxiety-inducing drug s , under highly co n t r olled conditions. to escape fr om the elevated open arms of such as yohimbine. The rat model also Furt h e r m o r e, the issues being addres s e d the maze. (See table 1 for further descrip- appears to have high construct val i d i t y go beyond underlying mechanisms of ti o n of these tests.) For both tests, anxiety- (see Davis et al. 1997), and res e a rc h e r s the disorde r s to the development of new reducing drugs increase exploratory ac t i v - ha v e described the neuronal mecha- tr eatment modalities. Thirty to forty ity (Crawley et al. 1997). Howeve r , when nisms that regulate FPS. pe r cent of people diagnosed with alcohol the res e a r chers compared the genetically Co m p a r ed with rats, the situation in abuse or dependence are likely to devel o p identical mice based on the two tests, the mice appears to be significantly more a major depres s i v e episode during their co r r elations between the two tests on both complicated. Falls and colleagues (1997) lifetime (two to three times the normal baseline activity and drug-induced activ- found that a marked FPS res p o n s e incidence). Twen t y - f i v e to fifty perce n t ity wer e weak, suggesting that differen t could be detected in the DBA/2J (D2) of people with alcohol use disorde r s mechanisms wer e invol ve d . but not in the C57BL/6J (B6) inbred also meet the criteria for anxiety disor- Conditioned fear, which combines strain. This finding was confirmed by ders. Res e a r ch with animal models may elements of learning and memory with McCaughran and colleagues (2000). help determine which treatments are fear demonstration, has been studied in These data wer e of interest, because in most effective in reducing alcohol con- se v eral contexts. Fear-potentiated star- a contextual fear-conditioning paradigm, sumption among people with an under- tle (FPS) has been extensively studied in which the animal is trained to asso- ly i n g ps y chiatric disorde r , if in fact this in rats (Davis et al. 1997) and more ciate the CS with a US in a parti c u l a r can be determined with animal models. recently in mice (Falls et al. 1997; en v i r onment (i.e., the context), the B6 Res e a r chers are accumulating data McCaughran et al. 2000). In the basic and many other inbred strains perfo r m to address some of these issues. For the paradigm, animals are trained on day 1 quite well and the D2 performs poorly BXD recombinant inbred (RI) series4 to associate the conditioned stimulus (C rawley et al. 1997; Owen et al. 1997). of mice, data have been collected on (CS) (e.g., a light or sound cue) with On a cued paradigm, in which the alcohol consumption, pref e r ence, and the unconditioned stimulus (US) (e.g., conditioned fear response is measured acceptance (Phillips et al. 1994;

156 Alcohol Research & Health Animal Models of Psychiatric Disorders and Their Relevance to Alcoholism

Rod r i q u e z et al. 1994); open-field activ- Recent findings (Hit z emann et al. associated with pref e r ence. Knapp and it y (K oyner et al. 2000); FPS (McC a u g h r a n in pres s ) sh o w that in crosses derived colleagues (1997) examined the rel a - et al. 2000); context fear conditioning fr om other inbred strains, the genes tionship between emotional state and (O w en et al. 1997); and cued fear con- associated with alcohol phenotypes are alcohol intake in several rat lines and ditioning (Owen et al. 1997). The cumu- di f f e r ent from those detected in the strains, using as a measure of emotional la t i v e na t u r e of the RI data allows inves - B6xD2 intercr osses. Thus, the possibil- state the tendency to emit ultrasonic ti g a t i o n into whether genetic rel a t i o n s h i p s ity still remains that in some cros s e s , vocalizations in response to an aver s i v e exist between the alcohol phenotypes pre f e r ence and anxiety may be rel a t e d . but nonpainful stimulus (e.g., an air and the animal models of fear and anx- This argument could be extended from pu f f ). The authors concluded that “th e ie t y . The results are summarized in table di f f e r ent inbred mouse strain crosses to relationship between emotional st a t e 2 and illustrate that no significant cor- di f f e r ent species (e.g., rats). Howeve r , and ethanol drinking is complex and relations exist between the alcohol phe- the data in rats show the same com- cannot be attributed to a simple el e - notypes and the fear and anxiety phe- plexity observed in mice. For example, vated state of anxiety or emotionality.” notypes. Thus, we conclude that the McKinzie and colleagues (2000) have Fin a l l y , res e a r chers should not ignore genes that predispose the BXD RI ani- ob s e r ved that alcohol-preferring (P) rats the possibility that the lack of correl a t i o n mals to high alcohol consumption (i.e., sh o w significantly greater FPS res p o n s e s in some studies between pref e r ence and a major featu r e of alcoholism) are not the than alcohol nonpreferring (NP) rats. fear/anxiety could reflect weaknesses in same genes that predispose the animal These data suggest that some of the either the models for alcohol pref e re n c e to demonstrate high levels of uncondi- genes associated with alcohol pref e re n c e or consumption and the models for tioned and conditioned fear and anxi- also affect FPS, theref o r e linking a fe a r / a n x i e t y . For example, strain or et y . Howeve r , these results should be me a s u r e of fear conditioning to alcohol selected line differences in taste or caloric vi e wed ca u t i o u s l y . The results described pre f e r ence. Howeve r , other rat-res e a rc h - factors could confound differences in table 2 refer only to this parti c u l a r de r i v ed data linking anxiety and alcohol in pref e r ence. Howeve r , considerable RI series, derived from the B6 and D2 responses are less clear. Hall and col- data (described in part elsewh e r e in this in b r ed mouse strains. Although the B6 leagues (1998) examined Fawn Hoo d e d issue) indicate that this is not the case and D2 strains are highly polymorphic (i.e., P rats) and Wistar rat strains and for both the mouse and rat strains and and differ markedly in a number of concluded that measures of anxiety wer e lines previously discussed. Overall, behavioral phenotypes, the data obtained independent of alcohol’s stimulatory the data rev i e wed in this section force ar e valid only for understanding the and anxiety-reducing effects. McMi l l e n res e a r chers to question whether an ani- di f f e r ences between the B6 and D2 strains. and colleagues (1998) compared the P mal model exists that mimics the high rats with Wistar rats, the strain from comorbidity of alcohol-related disorde r s 4A recombinant inbred (RI) series is a set of animal strains which the P rats wer e derived . Using a and anxiety and impulsiveness observed derived by inbreeding the offspring of two parent strains. variety of measures, including the ele- in the type II alcoholic.5 The curren t Each strain in an RI series has a unique combination of genes from the parent strains, and all animals in a single vated plus maze, the authors concluded models appear less than ideal. Imp r ove- strain have the same gene combination. that anxiety and impulsiveness are not ments in the animal models will stem

Table 2 Relationships Between Alcohol Consumption, Preference, and Acceptance and Measures of Unconditioned and Conditioned Fear in the BXD Recombinant Inbred Series

Open-Field Fear-Potentiated Fear-Conditioning Fear-Conditioning Activity 3 Startle 4 Contextual 5 Cued 5

Alcohol consumption1 0.16 0.12 0.26 -0.09

Alcohol preference1 0.24 0.05 0.23 -0.04

Alcohol acceptance (F)2 0.11 -0.40 -0.15 -0.09

Alcohol preference (F)2 0.42 -0.01 0.21 -0.10

Alcohol acceptance (M)2 -0.08 -0.01 0.12 0.26

Alcohol preference (M)2 0.37 -0.02 0.30 0.19

NOTE: These results show that no significant correlations exist between the alcohol phenotypes and the fear and anxiety phenotypes. SOURCES:1Phillips et al. 1994 (female mice only); 2Rodriquez et al. 1994 (male [M] and female [F] mice); 3Koyner et al. 2000 (male mice only); 4McCaughran et al. 2000 (male mice only); 5Owen et al. 1997 (sex not specified).

Vol. 24, No. 3, 2000 157 fr om two sources. One, alcoholism will FAL L S , W.A.; CAR L S O N , S.; TUR N E R , J.G.; AN D MCKIN Z I E , D.L.; SAJ D Y K , T.J.; MCBRI D E , W.J.; need to be defined into ever more rig- WIL L O T , J.F. Fear potentiated startle in two strains MUR P H Y , J.M.; LUM E N G , L.; LI, T.K.; AN D of inbred mice. Behavioral Neuroscience 111: 855– SHE K A R , A. Acoustic startle and fear potentiated or ous and homogeneous phenotypes 861, 1997. startle in alcohol preferring (P) and -nonpreferring and the new classifications will need to (NP) lines of rats. Pharmacology Biochemistry and include informative and easily measured GEY E R , M.A., AN D MAR K O U , A. Animal models of Be h a v i o r 65:691–696, 2000. endophenotypes. Two, scientists worki n g psychiatric disease. In: Bloom, F.E., and Kupfer, D.J., eds. Psychopharmacology: The Fourth Generation MCMIL L E N , B.A.; MEA N S , L.W.; AN D MAT T H E W S , with animals will need to be vigorou s of Progress. New York: Raven Press, 1995. pp. 787–798. J.N. Comparison of the alcohol-preferring P rat to in building new models that incorporate the Wistar rat in behavioral tests of and the new clinical classifications but at HAL L , C.S. Emotional behavior in the rat. I. De- an x i e t y . Physiology and Behavior 63:71–375, 1998. fecation and urination as measures of individual the same time maintain ethological differences in emotionality. Journal of Comparative OWE N , E.H.; CHR I S T E N S E N , S.C.; PAY L O R , R.; AN D pe r s p e c t i v e. ■ Ps y c h o l o g y 18:385–403, 1934. WEH N E R , J.M. Identification of quantitative trait loci involved in contextual and auditory cued fear HAL L , F.S.; HUA N G , S.; FON G , G.W.; PER T , A.; conditioning in BXD recombinant inbred mice. 5The characteristics of the type II alcoholic include impul- AN D LIN N O I L A , M. Effects of isolation-rearing on Behavioral Neuroscience 111:292–300, 1997. si v e n e s s , aggressive tendencies, criminality, and other locomotion, anxiety and responses to ethanol in antisocial behaviors. Fawn Hooded and Wistar rats. Ps y c h o p h a r m a c o l o g y PHI L L I P S , T.J.; CRA B B E , J.C.; MET T E N , P.; AN D 139:203–209, 1998. BEL K N A P , J.K. Localization of the genes affecting alcohol drinking in mice. Alcoholism: Clinical and References HIT Z E M A N N , R.; QIA N , Y.; KAN E S , S.; DAI N S , K.; Experimental Research 18:931–941, 1994. AN D HIT Z E M A N N , B. and the organization BRA F F , D.L. Information processing and attention of the basal ganglia. International Review of Neuro- POM P , D. Genetic dissection of obesity in polygenic dysfunctions in schizophrenia. Schizophrenia Bulletin bi o l o g y 38:43–94, 1995. animal models. Behavior Genetics 27 ( 4 ) : 2 8 5 – 3 0 6 , 19(2):233–259, 1993. 19 9 7 . HIT Z E M A N N , R.; DEM A R E S T , K.; KOY N E R , J.; CIP P , CRA B B E , J.C.; KOS O B U D , A.; YOU N G , E.R.; AN D POR S A L T , R.D. Behavioral despair. In: Enna, S.J.; L.; PAT E L , N.; RAS M U S S E N , E.; AN D MCCAU G H R A N , JAN O W S K Y , J.S. Polygenic and single-gene determi- Malick, J.B.; and Richelson, E., eds. JR., J. Effect of genetic background on the detection nation or responses to ethanol in BXD/Ty recom- Antidepressants: Neurochemical, Behavioral and of quantitative trait loci. Pharmacology Biochemistry binant inbred mouse strains. Neurobehavioral Toxi- Clinical Perspectives. New York: Raven Press, 1981. and Behavior, in press. cology and Teratology 5:181–187, 1983. pp. 121–139. KLI N E , L.; DEC E N A , E.; HIT Z E M A N N , R.; AN D CRA B B E , J.C.; PHI L L I P S , T.J.; BUC K , K.J.; CUN N I N G - ROD R I G U E Z , L.A.; PLO M I N , R.; BLI Z A R D , D.A.; MCCAU G H R A N , JR., J. Acoustic startle, prepulse HA M , C.L.; AN D BEL K N A P , J.K. Identifying genes JON E S , B.C.; AN D MCCLE A R N , G.E. Alcohol accep- for alcohol and drug sensitivity: Recent progress inhibition, locomotion and latent inhibition in the tance, preference and sensitivity in mice. I. Quanti- and future directions. Trends in Neurosciences neuroleptic responsive (NR) and neuroleptic non- tative genetic analysis using BXD recombinant 22:173–179, 1999. responsive (NNR) lines of mice. Ps y c h o p h a r m a c o l o g y inbred strains. Alcoholism: Clinical and 139:322–331, 1998. Experimental Research 18:1416–1422, 1994. CRA W L E Y , J.N.; BEL K N A P , J.K.; COL L I N S , A.C.; CRA B B E , J.C.; FRA N K E L , W.; HEN D E R S O N , N.; KNA P P , D.J.; KAM P O V -P OL E V O Y , A.B.; OVE R S T R E E T , THA N O S , P.K.; VOL K O W , N.D.; FRE I M U T H , P.; HIT Z E M A N N , R.J.; MAX S O N , S.C.; MIN E R , L.L.; D.H.; BRE E S E , G.R.; REZ V A N I , A.H. Ul t r a s o n i c HIT Z E M A N N , B.; UME G A K I , H.; IKA R I , H.; RIT H , SIL V A , A.J.; WEH N E R , J.M.; WYN S H A W -B OR I S , A.; vocalization behavior differs between lines of ethanol- G., ING R A M , D.K.; AN D HIT Z E M A N N , R. Up r e g u - AN D PAY L O R , R. Behavioral phenotypes of inbred preferring and nonpreferring rats. Alcoholism: lation of the D2 receptors in the nucleus accum- mouse strains: Implications and recommendations Clinical and Experimental Research 21 : 1 2 3 2 – 1 2 4 0 , bens of the rat using a D2 adenovirus vector for molecular studies. Psychopharmacology 13 2 : 19 9 7 . (AsCMV.DopD2R) and its effect on alcohol and 107–124, 1997. cocaine abuse. Sociological Neuroscience 25:1326, 1999. KOY N E R , J.; DEM A R E S T , K.; MCCAU G H R A N , J.; DAV I S , M.; WAL K E R , D.L.; AN D LEE , Y. Roles of CIP P , L.; AN D HIT Z E M A N N , R. Identification and VOL K O W , N.D.; WAN G , G.J.; FOW L E R , J.S.; the and bed nucleus of the stria terminalis time dependence of quantitative trait loci for basal LOG A N , J.; GAT E L Y , S.J.; GIF F O R D , A.; HIT Z E M A N N , in fear and anxiety measured with the acoustic star- locomotor activity in the BXD recombinant inbred R.; DIN G , Y.S.; AN D PAP P A S , N. Prediction of rein- tle reflex: possible relevance to PTSD. Annals of the series and a B6D2 F2 intercross. Behavior Genetics forcing responses to psychostimulants in humans New York Academy of Sciences 821:305–331, 1997. 30:159–170, 2000. by brain D2 receptor levels. American Journal of Ps y c h i a t r y 156:1440–1443, 1999. DEM A R E S T , K.; MCCAU G H R A N , JR., J.; MAH J U B I , MCCAU G H R A N , J.; BEL L J.; AN D HIT Z E M A N N , R. E.; CIP P , L.; AN D HIT Z E M A N N , R. Identification of Fear potentiated startle response in mice: Genetic WIL L N E R , P. Animal models of depression. In: an acute ethanol response quantitative trait locus on analysis of the C57BL/6J and DBA/2J intercross. Behavioral Models in Psychopharmacology: mouse chromosome 2. Journal of Neuroscience 19 : Pharmacology Biochemistry and Behavior 65 : 3 0 1 – Theoretical, Industrial and Clinical Perspectives. 549–561, 1999. 312, 2000. New York: Cambridge University Press, 1991.

158 Alcohol Research & Health