Rodent models of post-traumatic stress disorder: behavioral assessment

Although the etiology and expression of psychiatric disorders are complex, mammals show biologically preserved behavioral and neurobiological responses to valent stimuli which underlie the use of rodent models of post-traumatic stress disorder (PTSD). PTSD is a complex phenotype that is difficult to model in rodents because it is diagnosed by patient interview and influenced by both environmental and genetic factors. However, given that PTSD results from traumatic experiences, rodent models can simulate stress induction and disorder development. By manipulating stress type, intensity, duration, and frequency, preclinical models reflect core PTSD phenotypes, measured through various behavioral assays. Paradigms precipitate the disorder by applying physical, social, and psychological stressors individually or in combination. This review discusses the methods used to trigger and evaluate PTSD-like phenotypes. It highlights studies employing each stress model and evaluates their translational efficacies against DSM-5, validity criteria, and criteria proposed by Yehuda and Antelman’s commentary in 1993. This is intended to aid in paradigm selection by informing readers about rodent models, their benefits to the clinical community, challenges associated with the translational models, and opportunities for future work. To inform PTSD model validity and relevance to human psychopathology, we propose that models incorporate behavioral test batteries, individual differences, sex differences, strain and stock differences, early life stress effects, biomarkers, stringent success criteria for drug development, Research Domain Criteria, technological advances, and cross-species comparisons. We conclude that, despite the challenges, animal studies will be pivotal to advances in understanding PTSD and the neurobiology of stress.


Introduction
Post-traumatic stress disorder (PTSD) is an incapacitating chronic disorder. With a 3.9% lifetime prevalence rate worldwide and a 6.4-7.8% rate in the USA, PTSD's health burden is substantial [1][2][3][4][5] . Based on the World Mental Health Surveys, 69.7% worldwide (82.7% in the USA) reported exposure to a traumatic experience. While trauma exposure is a required criterion for PTSD diagnosis, only 5.6% worldwide (8.3% in the USA) of those who experienced trauma developed the disorder 1 . This is due to numerous factors, including trauma type, variation in trauma response, social support, and endogenous factors of individuals. For adults, adolescents, and children older than six years, eight diagnostic criteria, defined in the 5th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), specify measures concerning the victim's perception of trauma and symptoms. For children six years and younger, Criteria C and D (described below) are combined, making for seven diagnostic criteria 6 .
Criterion A: Exposure to actual or threatened death, serious injury, or sexual violence; one or more ways (e.g., direct experience, witnessing others, learning of close family member's or friend's trauma, and repeated or extreme exposure to aversive details).
Criterion B: Intrusion; one or more symptoms (e.g., nightmares, flashbacks, intrusive thoughts, and physiological reactions to trauma reminders). throughout disorder development. Therefore, human research focuses on populations already exposed to different uncontrolled traumatic events. Animal models overcome these barriers through the ability to longitudinally monitor PTSD development pre-trauma through posttrauma with controlled stressors. In particular, rodent models are critical to understanding PTSD induction, facilitating target identification for therapies, and testing drugs for human treatment. However, these models are simplified representations of a complex condition. PTSD is challenging to model in rodents because susceptibility, an individual's likelihood to develop long term symptomology, is influenced by the above factors. While there is no gold standard paradigm, animal models are expected to capture PTSD symptomatology (face validity), etiology (construct validity), and treatment response (predictive validity) 7 . As such, Yehuda and Antelman defined five criteria for the face validity of translational models: (1) the stressor induces PTSD biological and behavioral responses, (2) responses are intensity-dependent, (3) biological alterations persist or progress over time, (4) biobehavioral alterations are bidirectional, and (5) responses have inter-individual variability caused from experience, genetics, or both 8 . PTSD rodent models have been previously reviewed [9][10][11][12][13][14][15] . This review offers a comprehensive assessment of stress model variants against multiple criteria with a focus on behavioral assays used for model validation. Although a comprehensive rodent PTSD model is challenging to develop, the demand for more effective PTSD treatment and prevention strategies drives further research. Therefore, various stress paradigms that replicate specific disorder aspects are utilized to understand the pathophysiology of PTSD. Through a survey of recent literature, this review discusses rodent PTSD models that utilize acute stress, evaluating their translational efficacies, advantages, and challenges. The review concludes by highlighting opportunities for future work in studying the neurobiology of PTSD.

Rodent models of PTSD
Although the etiology and manifestation of psychiatric disorders are complex, mammals show biologically preserved behavioral and neurobiological responses to valent stimuli which underlie the use of rodent models of PTSD. The translational benefit of designing stress studies in rodents is supported by extensive comparative neuroanatomical studies [16][17][18] . Accordingly, the rodent models discussed reflect core PTSD phenotypes and vary by differences in stress type, intensity, duration, and frequency. Paradigms model the disorder by applying physical, social, and psychological stressors individually or in combination (Table 1). This review focuses on acute stressors and does not extensively address sub-chronic and chronic models because acute stressors are less likely to drive the symptom co-morbidity seen in other models. Behavioral tests Table 1 Protocol and advantages/disadvantages summary of the reviewed stress paradigms.   112,113 mentioned in this survey are discussed further in Table 2.
In literature, behavioral tests are frequently referred to as models. This can be misleading, and scientists should recognize their distinction, because tests elicit acute responses for measurement while models elicit pathology 19 .

Physical stressors
Physical stressors used to develop PTSD models include electric shock, underwater trauma, restraint/immobilization stress, and single prolonged stress. These stressors are advantageous for their procedural simplicity, clear symptom impact, and ease of scaling. Although widely utilized, physical stressors have not reproducibly differentiated individual variability or sex differences because most subjects display behavioral consequences. Physical stressors can also elicit injuries, pain, and/or inflammatory responses that can confound behavioral test results, including measures of motivation and movement 20 .

Electric shock
Electric shock is employed in animal models of anxiety, depression 21 , and PTSD. Inescapable and unpredictable electric shock can be administered through the animal's tail or foot via a steel grid floor in a shock chamber. These studies combined high-intensity currents (1.0-3.0 mA) with long durations (2-20 s) to induce lasting symptoms. Currents and durations were typically larger than those applied in fear conditioning (0.5-1.5 mA, 0.5-2 s) aiming to investigate short-term fear learning. Currently, learning and trauma protocols are not clearly differentiated as there is no clear evidence for separating "nontraumatic" shocks that are within a rodent's coping capacity from "traumatic" shocks that are beyond its coping capacity 22 . However, this stressor is more frequently used to study learning and memory, its initial application, rather than to model PTSD. Electric shock may be combined with contextual and/or cued trauma reminders (associative fear) through re-exposure to shock chambers and conditioned stimuli, and with neutral stimuli in a novel environment (non-associative fear). In stress-enhanced fear learning, pre-exposure to repeated intense footshocks in context A increased rodent fear response in context B when only a single shock, a less intense stressor, was administered 23 . This non-associative sensitization effect elevated freezing in context B for three months, even after context A extinction training, reflecting resistance to exposure therapy (extinction) 24 . Upon similar situational reminders, rodents exhibited behavioral responses comparable to human PTSD intrusion symptoms, such as crouching against the chamber wall, as well as increased freezing, respiratory rate, and fecal boli [25][26][27] . Since reexposure to stressor-related environments/cues is the clinical analog of exposure therapies, the electric shock model can be studied to advance cue-based out-of-context (reminders in office) and in-context (virtual reality) therapies 14 . Advantages of electric shock as a PTSD model are its controllable delivery and shock parameters (current intensity, duration, number, and interstimulus interval), reproducible context and cues, adjustable environmental cues, and reducible stress habituation 28 . For review of electric shock, see Aliczki and Bali & Jaggi 22,29 .

Restraint/immobilization stress
Restraint and immobilization stresses include confining rodents in enclosed chambers to limit movement for an extended period of time. Restraint stress is generally conducted by placing animals in Plexiglas or wire mesh tubes 30 . Immobilization stress is achieved by either placing animals into rodent immobilization bags (often Decapicones) [31][32][33] or attaching the animal's limbs and head in a prone position to wooden boards 34,35 . Currently, restraint and immobilization protocols are not clearly differentiated and the two terms are often used interchangeably 36 . While restraint and immobilization are similar in that they are stressors driven by the limitation of motion, it is important to make the distinction between the two as restraint does not prevent, but only restricts, movement of the rodent's limbs, body, and head. Accordingly, immobilization elicits stronger responses than restraint 37 . There are no comparative studies favoring any one acute stress protocol. Furthermore, only two studies comparing the sex-dependent effects of immobilization stress have been reported. One study found that the attribution of incentive salience toward food reward location (goal-tracking) was increased in Long-Evans males exposed to acute immobilization stress, whereas a bias toward food rewardassociated cues (sign-tracking) was increased in females. This suggests sex-differences in the failure to appropriately assign motivational value as related to amotivation/anergia and anhedonia 38 . Another study, in agreement with one subgroup of human endocrine responses 39,40 , reported that immobilization stress evokes long-term hypothalamic-pituitary-adrenal (HPA) axis desensitization following re-exposure to the same (homotypic) stressor and sensitization to novel (heterotypic) stressors 35 . This effect was greater in female Sprague-Dawley rats 41 . Studies also noted delayed expression (10 days post-stress) of avoidance behavior and changes in dendritic spine density in the basolateral amygdala, a stress responsive brain region 31 . For review of restraint and immobilization stress, see Buynitsky 42 .

Underwater trauma
Underwater trauma, or submersion stress, involves 1 min of forced swim followed by 30 s of forced submersion in a water tank. Exposed rats demonstrated  Step-through inhibitory (passive) avoidance Rodents are placed inside a cylindrical water tank Time immobile 251,252 Tail suspension Rodents are hung by their tails Time immobile 253 Open field Square, rectangular, or circular enclosure that is bare or covered with a thin layer of bedding Total locomotor activity, ambulatory locomotor activity, distance traveled, frequency of rearing, frequency of sniffing 254 Flinch-jump test Operant chamber in a ventilated, sound-attenuated cubicle with a shock grid floor, speaker, and video camera. Shock titrations are delivered in a series of ascending and descending intensities based on the rodent's response Flinch threshold, vocalization threshold, jump threshold 53 Hot-plate test Rodents are placed in a glass beaker on a plate heated to a constant temperature Latency to flinch or raise hind paws 255,256 Tail flick test The distal third of a rodent's tail is thermally stimulated with radiant heat (e.g., focused light from a light bulb), immersion in hot water, or direct contact with a heated surface Latency to withdraw tail 257 von Frey test von Frey filaments are applied to the plantar region of a rodent's paw through a wire mesh floor. Stimulation continues in a series of ascending and descending filament forces based on the rodent's response Paw withdrawal threshold 258 Sucrose preference After habituation to two drinking bottles, one water bottle is replaced with a 1-2% sucrose solution. Bottle positions (left vs. right) are alternated each day to control for side-preference bias % sucrose preference, % sucrose intake, % water intake 259 Intracranial selfstimulation Operant chamber in a ventilated, light-attenuated and soundattenuated cubicle with a wheel or lever manipulandum on the wall. Bipolar electrodes are implanted into a brain region that is part of the reward system. Training: manipulandum response following noncontingent stimulation prompts an identical contingent stimulation. Testing: stimulations are delivered in a series of ascending and descending intensities based on the rodent's response % baseline current-intensity threshold, response latency, number of extra responses, number of time-out responses 260,261 Social interaction test Two novel conspecific rodents (same sex, size, and age) interact freely in an open field Time spent in active social interaction (sniffing, licking, close following, allogrooming, crawling over the partner), number of social interactions, time spent in social avoidance (escaping, keeping the partner at distance with upright forepaws) 262 Social preference/ avoidance test Open field with a wire cage centered against a wall of the arena. One trial without a rodent in the cage (no target) and one trial with a rodent in the cage (target) Time spent in the interaction zone with the target absent/ present, time spent in the two corner zones opposite the wire cage, social avoidance ratio 263,264 Partner preference test Two boxes linked to a starting box. One female and male rodent is tethered to the rear of a goal box  266 immediate and persistent (7-30 days post-stress) increased arousal in acoustic startle response (ASR) and anxiety-like behavior in the elevated-plus maze (EPM) tests compared to control rats that swam without submersion [43][44][45] . Learning deficits in Morris water maze (MWM) spatial memory tasks, observed three weeks after trauma exposure, demonstrated the stressor's lasting negative effects on cognition 44 . However, interpretation of MWM results can be confounded by the re-exposure to water, a reminder of underwater trauma, that may influence task performance. Decreased plasma basal corticosterone (CORT) levels were found seven days after stress, indicating a lasting depression of HPA axis signaling following the trauma 43 . Based on successive performance in EPM and ASR, cut-off behavioral criteria demonstrated that the prevalence of maladaptive response to stress dropped from 91.6% of exposed Sprague-Dawley rats on day 1, the acute phase, to a constant rate of 41.6% by day 7 through day 30. Concomitantly, the prevalence of welladapted rats rose from 0% to 25% over days 1-30 45 . This temporal pattern reaffirms that animals display an individual variation in response similar to humans 46 . That is, the initially large affected proportion of the exposed population decreases steadily with time as many individuals show a tendency toward symptom improvement. Some may present acute stress disorder and a minority develop PTSD. Behavioral profiling, validated by immunohistochemical assessments, uncovered three separate stress response phenotypes: an anxious, fear-based group (38%), a co-morbid, fear-anhedonic group (15%), and an exposed-unaffected group (47%). In accordance with the high anxiety trait of the posttraumatic depression model 47 , enhanced pretrauma freezing correlated with posttrauma saccharin preference among fear-anhedonic phenotype rats, predicting anhedonia one month after exposure 48 . Although its intensity cannot be easily controlled, advantages of underwater trauma as a PTSD model are its reproducible context and ethologically relevant stress.

Single prolonged stress
Single prolonged stress (SPS) involves sequential administration of three stressors-2-h restraint, 20-min forced swim, and diethyl ether anesthesia-with a 7-day or 14-day undisturbed sensitization before testing. According to time-dependent sensitization studies, the undisturbed incubation period is necessary for PTSD-like symptom manifestation 49 . Interestingly, studies observed most behavioral and cellular changes seven days after SPS or reexposure, indicating time-dependent and experiencedependent sensitization 50 . Consistent with findings in PTSD patients, neuronal apoptosis and autophagy dysregulation in the hippocampus, amygdala, and prefrontal cortex appeared one day after stress, suggesting that morphological changes precede behavioral alterations 51 .
Combination of the three stressors is required for the PTSD phenotype, as combinations of any two does not induce all effects observed in the SPS paradigm 52 . As in underwater trauma, interpretation of forced swim and MWM test results is confounded by the re-exposure to water. Although age effects on SPS susceptibility have not been evaluated to date, maternal separation was found to strengthen adult SPS-induced increases in anxiety and contextual fear 53 . Early life exposure to SPS caused anxiety-like and depression-like behavior at postnatal day 32 (human early-adolescence), anxiety-like behavior at postnatal day 60 (human late-adolescence), followed by depression-like behavior (stress-susceptible) or no behavioral deficits (stress-resilient) at postnatal day 90 (human adulthood), suggesting that adaptations such as behavioral and cognitive switching occur at postnatal day 60 54 . The two studies of sex differences in fear extinction retention following SPS report conflicting results, with one observing no effect 55 and the other noting deficits 56 . The finding that female rats express fear by darting rather than freezing indicates that freezing alone may be misleading and motivates reinterpretation of female rodent fear conditioning studies 57 . SPS increased the latency of pairhoused Sprague-Dawley females to approach a novel rat in the social preference/avoidance test, implying an anxious phenotype, but decreased the latency of single housed females, implying social support seeking 58 . The dexamethasone suppression test revealed an exaggerated negative feedback control of the HPA axis in SPS-exposed Sprague-Dawley males, but not in females 59 . SPS lacks ecological validity and its stressor intensity cannot be modified. However, an advantage of SPS as a PTSD model is its combination of stressors to produce a synergistic effect. For an extensive review of SPS, see Lisieski, Souza, and Yamamoto [60][61][62] .

Social and psychological stressors
Social and psychological stressors include social defeat and predator stress. Advantageous for their ecological validity and relevance to interpersonal assault traumas 63 , these stressors are widely utilized in differentiating individual variability. While predator stress can be conducted on both sexes, social defeat is challenging to model in female rodents because they tend not to defend territories with aggressive behavior. When using direct exposure to a resident rodent or predator, reproducibility of the stressor can be challenging due to variation in rodent and predator aggression. If the social and psychological nature of these stressors causes physical injuries, behavioral measures may then reflect physical rather than neurological driven effects.

Social defeat stress
Social defeat stress (SDS) is used in rodent models of depression 64 and PTSD. There are three variations of SDS: resident-intruder, witnessed social defeat stress (trauma witness or vicarious social defeat), and cagewithin-cage resident-intruder. While in the residentintruder paradigm, an experimental rodent (intruder) is exposed daily (5-10 days; 5-10 min per day) to a novel dominant conspecific (resident) 65,66 , witnessed social defeat stress is induced by daily inescapable sensory contact with novel rodents undergoing physical social defeat 67 . In both procedures, subordination may then be reinforced through 24 h of sensory (visual, olfactory, and auditory) contact with the resident, separated by perforated screen. The cage-within-cage resident-intruder paradigm introduces rodents to sensory contact with novel residents for 6-h sessions that include one to three 1-min unpredictable physical contact periods 68 . Since territoriality is established in sufficient living space and enhanced in the presence of a sexual partner and sexual experience, the intruder is either introduced to an individually housed resident or replaces a cohabitating partner in the resident cage 69 . Intruders with lower body weights than residents are frequently used to guarantee intruder defeat 66 . Aggressive behaviors, different between male and female rodents, are characterized by attacking, pushing, aggressive grooming, chasing, pinning, and upright or lateral dominant postures 70 . Depression models employ chronic SDS, with stress repeated for 10 days to five weeks. It has, however, been suggested that chronic SDS may not only be a depression model, but may be more appropriately used to simulate the depression, anxiety, and social avoidance dimensions of PTSD 71 . Male rodents exposed to resident-intruder social defeat manifest a persistent behavioral syndrome. The social preference/avoidance test categorized 50-70% of exposed C57BL/6 J mice as susceptible to stress 65,72 . Interestingly, social rank predicted this individual variability through a link between response strategy and outcome, showing that dominant mice were more susceptible than subordinate mice 73 . Preexisting individual differences in the peripheral immune system also predicted stress susceptibility 74 .
Morning CORT levels showed no changes among groups at day 11, but decreased in susceptible mice and increased in resilient mice at day 39 after social defeat 72 . In another study, stress resilience was associated with the emergence of the gram-positive bacteria Bifidobacterium. This finding is relevant to clinical applications as gut microbiota dysbiosis is found in patients with PTSD 75 . Many studies have also shown that environmental enrichment promotes adaptive behavior and brain function. Enriched environmental housing before social defeat conferred stress resiliency. Further investigation, through lesion of the infralimbic cortex, illustrated that the ventromedial prefrontal cortex was involved in the acquisition of protective effects 76 . Extending this observation, optogenetic modulation of neuron projections to/from the ventromedial prefrontal cortex 77 , ventral tegmental area 78,79 , nucleus accumbens 80 , and dorsal raphe nucleus 81 , key nodes of PTSD circuitry, exerted antidepressant-like effects in susceptible mice.
Female-female agoniztic behaviors show low levels of direct attack in resident-intruder confrontations compared to inter-male aggression, limiting studies exploring the effects of social defeat in female rodents 70 . Furthermore, male behavior towards female versus male intruders is not similarly motivated and results in fewer attacks. As such, resident-intruder pairings are frequently same-sex conspecifics. The rarity of female aggressive behavior has been addressed through using lactating dams 82 , aggressive species such as California mice and Syrian hamsters 83,84 , mediobasal hypothalamic lesions 85 , as well as testosterone treatment in neonatal 86 and ovariectomized adult female rodents 87,88 . Female CD-1 mice exposed to lactating dams displayed elevated anxiety in the EPM, with more pronounced effects two weeks, but not 2 h, after social defeat 71 . Although qualitatively different from male SDS, recent studies have induced male aggression toward females through application of male odorants on females 89 and chemogenetic activation of the ventrolateral subdivision of the ventromedial hypothalamus in males 90 . The limited number of female rodent studies using similar species, methodologies, and behavioral tests complicates result comparison. In addition, cross-sex comparison is contingent upon proof of equivalent attack magnitude toward males and females. While care is taken to minimize severe injuries, behavioral measures may be compromised when the social nature of resident-intruder and cage-within-cage resident-intruder paradigms is mixed with the stress of physical injury. SDS is also limited by the potential for adaptation to its predictable repeated stress. Although the stressor type cannot be varied, its predictability could be reduced by exposing rodents to novel residents at different times each day for different stress durations. For review of social defeat, see Hammels 91 .

Predator stress
Predator stress models, prioritizing ecological validity, expose rodents to species-relevant predators or their scents. Variations of this stressor include predator exposure stress, predator-based psychosocial stress (PPS), and predator scent stress (PSS).
In the predator exposure stress model, rodents are acclimated for 5 min in an inescapable exposure environment, followed by a 5-60 min exposure to an unprotected/protected (subject is free/caged) cat 92,93 or ferret 94 . While initial acute cat exposure studies assessed overall behavior through tests of risk assessment, anxiety, and arousal 92 , later reports also found avoidance of traumareminder in the open field test 95 , with spatial memory retention impairments in the MWM and radial-arm water maze 93,96 . Like SPS, exposure enhanced dexamethasone suppression of CORT in Sprague-Dawley males, but not in females 59 . In addition, acute ferret exposure in Sprague-Dawley rats produced sensorimotor gating abnormalities in prepulse inhibition 94 . Out of the exposed Sprague-Dawley rats, 25.3% developed PTSD-like behavioral and endocrine dysregulation 96 . Sex differences in vulnerability were test specific, noted in EPM, open field, and ASR measures. Sex differences were also identified in hole-board and light-dark box measures which were unaffected by exposure 95,97 . Persistent behavioral changes, some lasting at least three weeks after exposure, corresponded to changes in the amygdala, medial prefrontal cortex, and hippocampus. These structural and molecular changes are consistent with brain regions involved in PTSD 93,95,98,99 . However, given predator exposure's variable cat-rodent interaction and the challenge of controlling cat aggression, both cat and rodent behavior must be assessed. For review of predator exposure, see 100 .
The PPS model, composed of acute and chronic components, combines 1 h of acute immobilization during novel cat exposure on days one (during the light cycle) and 11 (during the dark cycle), with 31 days of chronic unstable housing conditions to produce a risk factor synergistic effect. Following observations that rodents directed their postures away from cats, providing them with an element of control over confrontations, immobilization was included in this model as an analog to the sense of helplessness prominent in PTSD. Repeated cat exposure was included to (1) apply to people who develop PTSD only after multiple traumas; (2) mimic intrusive trauma reminders by forcing rodents to re-experience the original stress; (3) mimic the unpredictability of re-experiences (light or dark cycle); and (4) augment stress-induced changes. Further increasing symptom prevalence, daily randomized housing conditions were included to mimic chronic mild stress and lack of social support 101 . Three weeks after the second predator exposure, all stressed Sprague-Dawley rats exhibited PTSDlike sequelae. PPS also caused reduced growth rate, thymus weight, and basal glucocorticoid levels, as well as increased adrenal gland weight and physiological reactivity to an acute stressor [101][102][103] . Some effects were still observed more than four months after stress onset 104 . Neurotransmitter changes, such as greater norepinephrine and reduced serotonin levels in the hippocampus and prefrontal cortex, are in concert with human PTSD research [105][106][107] . This paradigm was also shown to induce sex-dependent cardiovascular alterations with notable male and ovariectomized female increases in myocardial sensitivity to ischemic injury. Stressed female rats displayed anxiety-like behavior in the EPM and open field, irrespective of estrous stage or ovariectomy condition. However, stressed female rats did not demonstrate physiological effects other than a reduced growth rate 108,109 . For review of PPS, see Zoladz 110 .
The PSS model involves a 5 to 15-min inescapable exposure to fox/bobcat urine 111 , soiled cat litter 112 , a catworn collar/cloth 113 , a ferret cloth 114 , or trimethylthiazoline (TMT), a synthetic compound isolated from fox feces 115 . Similar findings in successive ferret and cat cloth exposures confirmed failure of stress-habituation 114,116 . Moreover, different amounts of TMT 117 and sizes of cloth impregnated with cat scents 118 elicited fear-related behavior in a dose-dependent manner. After application of median split criteria, the incidence of susceptible TMT-exposed Sprague-Dawley rats ranged from 14 to 21.8% 119 . Similarly, after cut-off behavioral criteria, extreme behavioral responses (PTSD-like) to soiled cat litter were observed among 50% of Lewis, 10% of Fischer F344, and 25% of Sprague-Dawley rats 112 . Although sensitive to strain differences, cut-off behavioral criteria did not distinguish PTSD prevalence among male and female Sprague-Dawley rats 120 . In both sexes, extreme behavioral response rats displayed inhibited cardiac autonomic system habituation and recovery after exposure 121 . Early life stress increased vulnerability to cardiac autonomic dysfunction, and blunted basal CORT pulse amplitude predicted post-exposure PTSD susceptibility 122,123 . Avoidance of bobcat urine-paired context predicted post-stress thermal hyperalgesia in Wistar rats with high stress reactivity 111 . Interestingly, resting-state functional magnetic resonance imaging (rsfMRI) noninvasively detected prolonged neuroadaptation within the amygdala-medial prefrontal cortex circuit in Long-Evans rats exposed to cat collar 113 . Another study found that preexisting functional connectivity in olfactory and stressrelated neural circuits might predispose animals to differential stress responses, linked to PTSD susceptibility, during fox urine exposure. Susceptible rats exhibited less freezing, but greater avoidance of fox urine, and displayed a prolonged CORT response, as well as higher anxiety long after exposure 124 . These findings indicate the importance of analyzing behavior during exposure. A high-dose of CORT injected subcutaneously 1 h before soiled cat litter exposure reduced the prevalence of extreme behavioral responses from 50% to 8% in Lewis rats, indicating that elevated CORT levels before acute stress prevent later stress effects 112 . Analogous to clinical trial results, high-dose CORT administered 1 h after stress reduced the prevalence of PTSD-like rats, reversing extreme behavioral disruptions along with molecular and morphological measures in the hippocampal dentate gyrus. This evidence supports the use of high-dose CORT in trauma care and suggests that there is a treatment "window of opportunity" early after trauma 125 . Combined, these observations indicate the protective effects of glucocorticoids against the development of PTSD. Sleep deprivation for 6 h during the first resting phase after PSS attenuated PTSD-like behaviors in EPM, ASR, and hippocampal expression of glucocorticoid receptors, demonstrating an avenue for secondary prevention of stress-related clinical disorders 126,127 . Since predator scent can impregnate testing rooms and influence control animals, careful handling and stressing under a fume hood is advised. Olfactory cues such as dosage of a scent are challenging to control because olfaction is variable and driven by perception. An advantage of PSS as a PTSD model is its ecologically relevant stress, as rodents use the olfactory sensory system for survival-related behaviors 128 . For review of PSS, see Cohen and Staples 129,130 .

Translational studies of PTSD
Established stress paradigms that induce PTSD-like behavioral and biological phenotypes are available (Table  1). However, since stressor severity varies, stressful experiences that are within a rodent's coping capacity should be distinguished from traumatic experiences that are beyond its coping capacity. Rodent behavioral tests, mimicking the tests conducted in humans, are used to assess stress effects and allow researchers to make inferences about rodent psychology ( Table 2). Although all stress models produce lasting general anxiety or depression effects, variety in behavior robustness makes each paradigm effective at targeting specific constructs. Behavioral tests and neurobiological changes are used to evaluate an animal model's representation of the human disorder, satisfying DSM-5 PTSD symptom clusters (Tables 3, 4), validity criteria (Table 5), and Yehuda and Antelman's criteria (Table 6). Unlike for the DSM-5, no criteria have been established to assess how well an animal model meets validity or Yehuda and Antelman's criteria. Gaps in literature were addressed as no one model has been proven to satisfy all criteria.
Many behavioral tests were developed and validated in rats, and only later were adapted for mice. Rats were traditionally the species of choice in preclinical research for their performance in operant tasks and their larger size that facilitates application of invasive techniques, as well as toxicity tests of compounds. Mice, on the other hand, are advantageous for their ease of genetic modification, breeding, and group housing. Following the increasing use of mice, behavioral tests were translated into mouse versions, but with mixed success 131,132 . Therefore, the utility of mouse models is contingent on the availability of more behavioral tests that are optimized for use in mice.
More specific behavioral tests should be incorporated into batteries to address gaps in PTSD research. Since sucrose preference does not distinguish between affective response and motivation, more specific tests are needed such as facial reactivity analysis and measurements of reward-related ultrasonic vocalizations 133,134 . Other possible attention assessments that have not been conducted in the context of PTSD include the signal detection test with blank trials for sustained attention and multiplechoice serial reaction time test for sustained and selective attention 135 . The resident-intruder test has been used to assess irritability and aggression. However, this test is not listed in Table 2 because it incorporates the unclear stress effects of novel social interaction and could influence later behavioral time points.
One unrealized strength of rodent models of PTSD is in the discovery, development, and testing of effective pharmacological treatments. Historically, pharmaceutical companies did not invest in innovative drug discovery programs because modifying approved medications, particularly selective serotonin reuptake inhibitor (SSRI) antidepressants, was less time-consuming and more profitable 136 . While the SSRIs sertraline (Zoloft) and paroxetine (Paxil, Paxil CR, Brisdelle, and Pexeva), are the only medications approved for PTSD by the Food and Drug Administration (FDA) to date, off-label pharmacological treatments include fluoxetine (Prozac, Prozac Weekly, and Sarafem) and venlafaxine (Effexor XR) [137][138][139][140] . These drugs were tested in PTSD patients because of their effectiveness in treating depression. Many provide relief in patients, yet no medications have been approved to promote resilience. Therefore, pharmacotherapy is currently recommended as an adjunctive or next-line treatment to trauma-focused psychotherapy 139,141 . Interestingly, the two medications that are indicated for PTSD treatment, sertraline and paroxetine, and the two most promising drug candidates, ketamine and 3,4-methylenedioxymethamphetamine-assisted psychotherapy, did Table 3 Evaluation of reviewed animal models against DSM-5 criteria for PTSD-effects in males 6 . not emerge from basic research 142 . In addition, only some drugs have been tested in rodent models to study their targets and mechanisms of action (Table 5). One explanation for their mechanism is that medications may enhance psychotherapy's efficacy to engage biological targets associated with recovery or resilience 142 . Alternatively, medications may distinguish the PTSD biological subtypes that respond to their targets 143,144 . Now that rodent models of PTSD are more validated and have been refined over the last 20 years, their use could elucidate human psychopathology and further reveal mechanisms driving recovery. In rodents, as in humans, different traumas can cause different PTSD-like symptoms. Each with their advantages and disadvantages, there is no single model that serves all purposes. All reviewed animal models have phenomenological similarities to PTSD and satisfy face validity. Peripheral biologic correlates of PTSD have been evaluated for all stressors, generally meeting construct validity. For example, there are strong hypotheses regarding the neural mechanisms involved in PTSD, such as hyperactivity of the corticotropin-releasing hormone system, that have been verified in rodents 145,146 . However, because the biological basis of PTSD is not well understood, work needs to be done to define clear conditions for construct validity. Although preclinical research on traumatic stress has not yet resulted in drastic improvements in PTSD treatments, most models test pharmacological treatment options, providing them with predictive validity and potential for use in the development of clinical therapy. Given these results, more wellstudied models such as electric shock and single prolonged stress may fit more criteria, but lack demonstration of multiplex behavioral outcomes in individual animals. While DSM-5, validity, and Yehuda and Antelman's criteria each capture some of PTSD's complexity, consideration of all three may be used to improve categorization of a rodent model's trauma effects and inform stress paradigm selection. More accurately capturing the disorder in any individual model is critical, but overly limiting the types of models studied may be counterproductive in developing a comprehensive understanding of such a variable disorder.

Symptom comorbidity
Symptom comorbidity complicates PTSD diagnosis and treatment. Although rodent models are useful for symptom analysis, there are limitations to generalizing stress in rodents to PTSD in humans since clinical diagnosis heavily relies on a patient interview rather than quantitative diagnostic measures. Individuals with PTSD are 80% more likely than those without PTSD to have symptom comorbidity. Symptoms such as avoidance, anhedonia, and exaggerated startle response may overlap with symptoms of other mental illnesses (e.g., depressive, bipolar, anxiety, or substance use Table 4 Evaluation of reviewed animal models against DSM-5 criteria for PTSD-Effects in females 6 . disorders), making it difficult to attribute them to a specific PTSD model 6 . This creates challenges in distinguishing animal models of PTSD from other psychiatric illness models. The criteria used in the development and study of PTSD models will continue to evolve as scientific understanding of PTSD grows and the DSM is updated. Nevertheless, longitudinal studies are used to develop animal models with relevant PTSD symptom domains and to test causal factors in symptom development.

Anthropomorphism
The anthropomorphizing that can occur when uniquely human characteristics are attributed to animals is a danger of animal models and complicates their translatability. Many diagnostic criteria of mental disorders are subjective, involving thoughts, memories, and their interpretations. Such criteria as intrusive thoughts, emotional blunting, and cognitive distortion cannot be measured in animals without the danger of excessive anthropomorphism. Therefore, research is limited to observable behaviors with quantifiable measures, many of which are detailed in Table 2. However, even these measures have limitations, since rodent behavioral tests often do not directly correlate to human tests. The urgency to perform "translational" work can cause research groups to exaggerate a treatment's effects and scientific foundation, ultimately hindering treatment development. Prolonged exposure, for example, is a cognitive behavioral therapy that only works for some patients and may not be based on fear extinction, yet extinction learning is so robust and explainable in animals that it is used to justify the treatment's mechanism 147 . From the "translational" research opportunity and ease of modeling fear acquisition and extinction, this single aspect of PTSD has become more significant than it is clinically 148 . Anthropomorphizing can also result in data overinterpretation. For example, rodents that demonstrate escape or freezing behavior have been referred to as anxious rather than, more appropriately, exhibiting anxiety-related or anxiety-like behavior 149 . Further, this anthropomorphizing can lead to misleading interpretations of behavior, as it has been shown that behaviors like freezing are not simple quantitative measures of fear or anxiety as they are often utilized 150 . Therefore, to optimize the study of PTSD, rodent models should be empirically based, without anthropomorphic inference. Table 5 Evaluation of reviewed animal models against validity criteria for animal models of human mental disorders 7 .

Non-standardized experimental designs
Experimental designs are often not standardized across laboratories, complicating inter-laboratory comparison and result replication. Studies differ in their methods of trauma exposure, behavioral test batteries, and data analysis. This variability is exacerbated through differences in experimenter sex, rodent sex, species/strain/stock, age, housing and noise conditions (e.g., stressed and control animals caged separately/together, individually ventilated cages), food, incubation time between trauma and testing, acclimation time to the testing room and its light intensity, location of trauma and testing rooms, transportation to the testing room, subject location during trauma and testing of other subjects, specific tests conducted and their order, test duration, and test timing (light or dark cycle). All of these factors have a non-trivial effect on results and illustrate the often-underestimated importance of attention to detail.

Insufficient reporting
Many publications have insufficiently reported experimental details, rendering them not fit-for-purpose and limiting their value to researchers, doctors, and policymakers 151 . This challenge is compounded by concerns of positive publication bias, which favors the reporting of positive results (data supporting an alternative hypothesis) and can lead to spurious claims, wasteful experimentation, and reduced meta-analysis validity. Some journals (e.g., Journal of Negative Results in BioMedicine, Journal of Pharmaceutical Negative Results, Nature Negative Results section) compensate publication bias by exclusively publishing negative results. Favoring negative results, however, can also introduce bias 152 . Therefore, publishing criteria should focus on a study's design and statistical power, regardless of its outcome. Guidelines for planning and reporting animal experiments (e.g., Gold Standard Publication Checklist, Animal Research: Reporting of In Vivo Experiments, and Planning Research and Experimental Procedures on Animals: Recommendations for Excellence) have been established, but have not yet been widely adopted [153][154][155] . The lack of reproducibility in preclinical research may contribute to the failure of drugs in clinical trials 156 .

Ethical complications
Effective rodent models of PTSD utilize stressors that exceed the animal's homeostatic regulatory capacity, but Table 6 Evaluation of reviewed animal models against Yehuda and Antelman's criteria for animal models of PTSD 8 . Predator-based psychosocial stress 101 110 Predator scent stress 129,130 118 348 112,124 ethical complications remain. Studies are expected to minimize rodent suffering, improve human health, and advance general scientific knowledge. Since exposure to a significant trauma is required, emphasizing reduction in animal distress is counterproductive in PTSD models. This may result in minimizing exposure type or severity such that the model's efficacy and scientific value is compromised 10,157 . Predator scent stress, for example, developed in response to safeguards limiting predatorprey interactions, is less severe and may not as effectively induce the PTSD-like phenotype as predator exposure. Therefore, stressful experiences that are within a rodent's coping capacity should be distinguished from traumatic experiences that are beyond its coping capacity. Investigators are encouraged to collaborate with ethical committees to ensure that their selected exposure is relevant to PTSD and maximizes the probability of clinically significant findings.

Future directions
While it is apparent that well-established PTSD animal models are available, refining behavioral and neurobiological understanding of these models is still needed. Future directions include study of (1) individual variability and behavioral test batteries; (2) sex differences; (3) strain and stock differences; (4) early life stress effects; (5) biomarkers; as well as use of (6) stringent success criteria for drug development; (7) Research Domain Criteria (RDoC); (8) technological advances; and (9) cross-species comparisons.

Individual variability and behavioral test batteries
Many studies regard groups of animals under a condition as homogeneous, overlooking individual variability, and rely on a small number of tests with singular time points. Some models (e.g., immobilization stress, predator-based psychosocial stress) create a ubiquitous response in which all exposed animals present PTSD-like symptoms without the response variability inherent to human PTSD. Trends in variability have been described as "bidirectional" (Table 6), but with different meanings across research groups. Bidirectional expression has been used to mean both the concurrence or alternation between increased (i.e., intrusive re-experiencing, hyperarousal) and decreased (i.e., avoidance, numbing) responsiveness across different behavioral tests or subgroupings of animals within an experimental group. For example, a model could satisfy Yehuda and Antelman's bidirectional criterion by showing that a group of animals display hyperarousal in ASR and anhedonia in sucrose preference or that stressed subgroups show opposite responses to the same stimulus. When only a subgroup is expected to display long-term PTSD-like phenotypes, analyzing group level statistics based on exposed and control groups can lead to weaker statistical power and would make results harder to translate. While animal behavioral studies often employ few tests post-exposure to detect symptoms, test batteries covering multiple diagnostic criteria can achieve a more reliable profiling of individual animals. With multiple tests and time points, behavioral changes measured before and after stress could provide insight into the direct effects of stress over time, controlling for subject variability. Interpreting results in a manner similar to human diagnosis (e.g., cut-off behavioral criteria 96 , median split criteria 119 , behavioral profiling 48 , or clustering methods) then allows for the classification of susceptible and resilient trauma-exposed individuals. By researchers considering individual variability, treatments can be evaluated on whether they decrease the proportion of maladapted animals instead of their effect on symptom severity.

Sex differences
Sex differences are another area of variability that has been poorly studied (Table 4). PTSD was introduced into the DSM-III following the high prevalence of male Vietnam veterans seeking treatment for posttrauma symptoms 158 . All psychiatric disorders display sex differences, yet male rodents have been predominantly studied given the historical perspective of past publications 159 . Overreliance on male animals and cells conceals sex differences and may contribute to the increase of nontranslationality in preclinical research 156 . Due to the push to address sex differences, in consideration of females' higher prevalence of PTSD and adverse drug reactions, preclinical studies are now studying sex differences more readily 6,160 . Noting discrepancies in diagnostic criteria between the DSM-IV, DSM-5, and 10th Revision of the International Classification of Diseases (ICD-10), studies suggest that sex differences in symptom endorsement may make females more likely to meet DSM-5 and ICD-10 criteria for PTSD. Females were more likely to endorse intrusive thoughts, avoidance of external reminders, emotional blunting, exaggerated startle response, and sleep disturbance, while males were more likely to endorse aggressive behavior symptoms 161,162 . Sex differences in coping strategies, possibly mediated by differential endocrine responses (e.g., cortisol and oxytocin), have also been proposed 163 . Furthermore, human and rodent findings have suggested that differences in interactions between gonadal hormones testosterone or estrogen with the HPA axis or hippocampus contribute to the increased disorder risk among females [164][165][166] . Evaluations of behavioral effects of estrous cycle phase in female rodents observed trait and strain differences in test performance stability across phases, concluding that choices of behavioral paradigms, testing conditions, and genetic backgrounds are critical to controlling for hormonal effects 167 . However, hormone variation should be monitored as it could confound data interpretation or behavioral phenotyping performed blind to estrous cycle phase. In addition, studies should be designed with an adequate sample size to homogenize the estrous phase distribution. Further research is needed to understand the effects of hormones on PTSD and to develop hormonesensitive treatments.

Strain and stock differences
Similar to humans, rodent genetic background modulates stress susceptibility and phenotype variation, yet strain and stock differences have not been sufficiently characterized. Inbred strains of mice (e.g., C57BL/6, BALB/c) and rats (e.g., Lewis, Fischer), genetically homogenous, are used to minimize intrastrain differences. In contrast, outbred stocks of mice (e.g., Swiss, CD-1) and rats (e.g., Long-Evans, Sprague-Dawley), genetically heterogenous, are used to model the variability of human populations. The C57Bl/6 strain is reported to be more resilient to early life stress compared to strains such as the anxious BALB/c strain 168,169 . In addition, strain and stock differences in baseline behavior and psychotropic treatment response have been identified 170,171 . Therefore, rodent strain/stock, vendor, and colony should be considered across experiments within a study to better control experimental conditions and increase result reproducibility. Further research may reveal phenotypic behavioral differences with underlying genetic bases relevant to PTSD susceptibility.

Early life stress effects
Although early life stress, such as childhood abuse, increases vulnerability to psychiatric disorders later in life, this relationship is not clearly understood 172 . To better understand this relationship, early life stress is modeled in rodents through interventions in mother-pup interaction time periods (e.g., early handling for 3-15 min, repeated maternal separation for 1-8 h, single maternal separation for 24 h), the quantity and/or quality of maternal care (e.g., repeated cross-fostering, naturally occurring differences in maternal care, impoverished postnatal environment), and pharmacology (e.g., postnatal glucocorticoids, postnatal lipopolysaccharide) 173,174 . Since the brain's neurocircuitry matures from adolescence to early adulthood, trauma response may change over time 175 . In addition to effects of early life stress on brain and behavior over the developmental course, open research areas include identification of gene-environment interactions, epigenetic processes, sensitive periods, methods for disorder prevention or reversal, and beneficial effects. Different hypotheses exist to explain the interaction between early and adult stress in mental disorder susceptibility. Whereas multiple hit models, such as the cumulative-(diathesis-)stress hypothesis, state that disease risk increases with more adversity, the match/mismatch hypothesis states that risk deceases when early and adult environments match (stress inoculation), but increases upon mismatch 176,177 . Similarly, the differential susceptibility to environmental influence ("for-better-and-forworse") hypothesis states that individuals carrying alleles that are reactive to the environment display heightened sensitivity to disease risk in stressful contexts, but also to beneficial effects in supportive conditions 178 . Integrating hypotheses and considering individual differences in early programming sensitivity, the three-hit concept of vulnerability and resilience incorporates genetic predisposition, early life environment, and later life environment 179,180 . Additional assessments across the life span should be conducted to evaluate these hypotheses and clarify the relationship between maltreatment in early life and mental illness in adulthood.

Biomarkers
Potential PTSD biomarkers have been identified, but the topic remains underdeveloped as there are no biomarkers of susceptibility, disease, or therapy in clinical use. An individual's susceptibility to developing PTSD can be evaluated before trauma through risk markers for primary prevention, to prevent exposure, and early after trauma through risk markers for secondary prevention, to begin preventative therapy before symptom expression. Reported primary risk markers in humans include: (1) high fear response to a single 35% CO 2 inhalation challenge 181 ; (2) gene × environment interactions (e.g., FK506 binding protein 5 (FKBP5) gene × childhood trauma) 182 ; (3) HPAaxis regulators 183 ; (4) frequent nightmares (possibly related to the hampered fear extinction memory consolidation associated with REM sleep) 184 ; and (5) low hippocampal volume 185 . Few secondary risk markers, apart from increased heart rate 186 , have been reported because of sample size and prospective design limitations. Prominent disease diagnostic markers include HPA-axis dysregulation 187 , FKBP5 expression 188 , sympathetic nervous system hyperreactivity 189 , low hippocampal volume 190 , and high amygdala activity 191 . Use of biomarkers as inclusion or exclusion criteria for patient enrollment into clinical studies increases a drug candidate's likelihood of FDA approval 192 . However, few PTSD therapy stratification/selection markers, used to predict treatment responses and stratify patients by therapy responder types, or progress markers, used to monitor therapy responses, have been determined. The field's attempts to identify PTSD biomarkers has not resulted in treatments, so the aim that animal models will provide targets that will inform treatments has not yet been accomplished 142 . This is because biomarkers are currently limited by their interdependence, contribution to other disorders, inconsistent effect distribution, and small effect 146 . It is, therefore, important to determine how to use animal models to successfully translate findings. With multimodal study designs, identification of biomarkers that are homologous across species and play similar roles in the model and clinical condition will compliment behavioral measures to inform clinical trials of diagnostic and treatment targets. For reviews of PTSD biomarkers, see DePierro and Schmidt 142,193 .

Stringent success criteria
Implementing stringent success criteria into the preclinical stage of testing may improve upon the traditional drug development approach of expediting treatments into clinical trials to collect human data as early as possible. The unmet need for PTSD treatments may have reduced regulatory restrictions to their development and allowed drugs with suboptimal preclinical validation to enter clinical trials. Consequently, central nervous system drug candidates now have the second lowest success rate among therapeutic areas and take longer to bring to market 192,194 . The majority of drugs that showed initial promise in animal models failed in phases II and III of clinical development due to a lack of efficacy [195][196][197][198] . For example, Bionomics' BNC210, a negative allosteric modulator of the α7 nicotinic acetylcholine receptor, failed to significantly improve symptoms compared to placebo in a Phase II PTSD trial because of insufficient exposure (bioavailability) 199 . Target site exposure, target binding, and expression of functional pharmacological activity (the "three pillars of survival") all determine the likelihood of candidate survival in Phase II and progression to Phase III trials 200 . This underscores the importance of acquiring preclinical pharmacokinetic and pharmacodynamic data to test drug mechanisms and success. In response to the high unsustainable cost of clinical trials, pharmaceutical companies, including Bionomics, have downscaled, outsourced, or closed their neuroscience research programs [201][202][203] . The FDA estimates that a 10% improvement in failure prediction before clinical trials could save $100 million in development costs per drug 204 . Stringent success criteria would improve failure prediction by requiring more preclinical validation, thus shifting risk from the clinical to the preclinical stage. This may then decrease drug development costs, times to market, and prices to the public 151 .

Research domain criteria
PTSD is a heterogeneous disorder and the field is shifting focus from generalized diagnostic criteria toward targeting specific symptom domains. In parallel to disease-specific DSM-5 criteria, the US National Institute of Mental Health's RDoC is a framework for subject characterization based on degrees of dysfunctions in psychological/biological systems. Therefore, the RDoC assumes that there is a dimensional continuum between health and pathology. The RDoC matrix consists of dimensional psychological constructs (or concepts), grouped into higher-level domains of human behavior and functioning, and distinguished by the units of analysis used to investigate constructs. Continuously evolving from more construct validity and utility data, the RDoC matrix currently has five domains (negative valence, positive valence, cognitive, social processes, arousal and regulatory systems) and eight units of analysis (genes, molecules, cells, circuits, physiology, behavior, self-report, paradigms) 205 . Additional domains that are relevant to PTSD and psychiatric research have been proposed, namely stress and emotional regulation and maintenance of consciousness 206 . Intended to inform future diagnostic systems, RDoC integrates multiple measures for a comprehensive understanding of matrix constructs and, consequently, their corresponding symptoms. Investigators should align studies to RDoC as it encourages scientists to seek beyond observable results to neurobiological measures. Using a more stringent criteria for what we consider as a rodent model based on the criteria discussed can improve the translatability of results.

Technological advances
Utilizing technological advances in neuroimaging and genetic manipulation in PTSD research is promising as it allows for longitudinal assessments and investigations of circuits governing PTSD etiology. In clinical practice, neuroimaging, specifically structural, is mostly used to exclude brain pathology from the cause of psychiatric symptoms. Since diagnosis is subjective, quantitative clinical measures are needed to validate the current qualitative criteria and give preclinical researchers confidence that they are modeling PTSD, rather than genetic variability. Recent technological advances may be employed to achieve this goal and improve information flow between the two fields. fMRI, for example, measures blood-oxygenation-level dependent signal as an indirect indicator of neuronal activity. It is a powerful noninvasive measure of systems-level brain function and has a higher spatial resolution than other noninvasive modalities such as electroencephalography. MRI has been used to demonstrate altered fractional anisotropy, focal neural activity, functional connectivity, as well as focal atrophy of gray matter 207 . Awake rat rsfMRI avoids confounding factors from anesthetics and can be correlated with behavioral measures to study functional networks implicated in stress response, discover new regions of interest, and define endophenotypes of susceptibility 208 . Since fMRI can be applied to both animals and humans, findings may be translated across species and the conservation of neural circuits may be compared 209 . Stress signaling pathways can also be selectively stimulated via optogenetics and chemogenetics (or pharmacogenetics). Optogenetics enables manipulation of neuronal activity in stress circuits through activation of light-sensitive proteins, such as microbial opsins, that are selectively inserted into intact living mammalian neurons. Using this technology, researchers can control cell firing patterns with high tissue specificity and temporal precision to study the causal relationship between circuits and behavior in PTSD 210 . In chemogenetics, instead of with light, cell populations and neural circuits are modulated by the injection of biologically inert ligands that activate engineered receptors. Designer Receptors Exclusively Activated by Designer Drugs are widely used in chemogenetics and act through G-protein coupled signaling cascades. Chemogenetics has a lower temporal resolution, but is relatively noninvasive compared to optogenetics, which requires fiber-optic implants 211 . Integrated with fMRI, these technologies can map the downstream effects of local manipulations on global brain activity [212][213][214] . The neurobiology of stress can also be studied through diffusion tensor imaging 215 , magnetoencephalography 216 , calcium-based fiber photometry 217,218 , and molecular fMRI 214 . Combined with fMRI, these multimodal methods allow researchers to collect complementary and multi-dimensional information across spatiotemporal scales on the same animal. Implementation of these tools in studies of PTSD animal models and integration with established techniques in behavioral analysis, electrophysiology, and pharmacology can unveil progressive changes after stress exposure and susceptibility endophenotypes. While reliance on qualitative assessment is still standard, the shift towards multimodal methods will allow for the determination of quantitative PTSD biomarkers that are predictive of qualitative criteria used to assess PTSD prevalence. This goal will likely require increased collaboration between clinical and preclinical researchers to be achieved.

Cross-species comparisons
Cross-species comparisons of stress responses over the phylogenetic scale, ranging from rodents to animals in the genus Mustela (e.g., weasels, ferrets, minks) and nonhuman primates, are required to truly understand the preservation of PTSD features. Neurobiological results from rodent studies have already shown relevance in other species, but with mild stress models 219,220 . For example, both rodent and primate research has identified brain regions, such as the bed nucleus of the stria terminalis (BNST), amygdala, prefrontal cortex, cingulate cortex, and hippocampus, that regulate stress-related behavioral mechanisms and are dysfunctional in PTSD patients 146 . The BNST, in particular, has been investigated for its role in fear memory neural circuitry 221 . In mice, optogenetic photoinhibition of neurons projecting between the BNST and ventral tegmental area during electric shock stress decreased freezing in the exposure context and closed arm entries in the EPM 222 . Consequently, the BNST was investigated in rhesus monkeys and its intrinsic functional connectivity with the central nucleus of the amygdala was found to be correlated with behavioral responses to no-eye-contact human intruder stress 223 . A possible translational roadmap for studying PTSD is to combine mechanistic and behavioral measures from rodents with multimodal neuroimaging measures from nonhuman primates and humans. Cross-species comparisons such as these show promise to elucidating the clinical relevance of stress-induced cellular, molecular, and microstructural discoveries in rodents.

Conclusion
Since a variety of traumas, producing specific symptoms, can trigger PTSD, a single rodent model will not perfectly capture a human individual's disorder complexity. While individual animal models can replicate lack of coping and ineffective adaptation to stress, this does not necessarily represent coverage of human PTSD variability. This shortcoming is in part rooted in the modeling of general trauma conditions such as life threats, while modeling effects of particular stimuli (e.g., burning odors, sounds of gunfire) or cognitive assessments (e.g., helplessness, self-blame, coping strategy) of the general trauma condition is not further explored. As studies of animal models continue to improve, incorporation of experimental designs that can highlight the effect of specific features of a trauma or drive specific (see figure on previous page) Fig. 1 Superimposition of reviewed animal models against validity criteria for animal models of human mental disorders (inner circle), Yehuda and Antelman's criteria for animal models of PTSD (middle circle), and DSM-5 criteria for PTSD (outer circle). Validity criteria abbreviations: face validity (FV), construct validity (CV), predictive validity (PV). Yehuda and Antelman's criteria abbreviations: biological and behavioral sequelae of PTSD (1), dose-dependent (2), lasting symptoms (3), bidirectional (4), interindividual variability (5). DSM-5 criteria abbreviations: physiological reactions to trauma reminders (B), increased avoidance (C), cognitive alterations (D 1 ), mood alterations (D 2 ), increased arousal (E 1 ), concentration problems (E 2 ), sleep disturbance (E 3 ), lasting symptoms (F). Behavioral test abbreviations: elevated plus maze (EPM), open field (OF), light-dark box (LDB), Morris water maze (MWM), radial arm water maze (RAWM), fear conditioning / fear extinction (FCFE), novel object recognition (NOR), social interaction (SI), forced swim test (FST), sucrose preference (SP), marble burying (MB), acoustic startle response (ASR), electroencephalogram (EEG). White: not reported. Shaded: reported. coping strategies will improve animal models' representation of human PTSD variability. Similarly, more objective measures are needed to evaluate a rodent's inability to self-regulate, also a cognitive aspect of responses to anxiogenic stimuli, to assess aspects of human PTSD variability further in preclinical studies. This, along with the availability of physical, social, and psychological rodent stressors, varying in length and intensity, will improve translation of results to comparable clinical cases. Each model's strengths and limitations, reviewed here, should be considered for fit-for-purpose stress paradigm selection, hypothesis testing, and experimental design. For example, consideration of the DSM-5 risk and prognostic factors for PTSD in animal models (e.g., trauma reminders, female subjects, and early life stress) are likely to better replicate the human condition in patients with similar features. When evaluating paradigms, judging rodent models based on one set of criteria could be misleading because each focuses on different constructs. DSM-5 criteria, designed to facilitate diagnosis of human mental disorders, are used to compare stressors by their representation of PTSD symptomatology. Expanding the comparison to etiology and treatment response, validity criteria are used for animal models of human mental disorders. Yehuda and Antelman's criteria, more specific, are used for animal models of PTSD and include factors that are essential to the induction of PTSD-like phenotypes. Therefore, all three criteria should be superimposed for the optimal reflection of a stressor's relevance to PTSD (Fig. 1). As only a minority of individuals who experience traumatic event(s) develop the disorder, it is important to investigate animals' individual variability in stress response and subsequent development of PTSD-like behaviors to distinguish susceptibility. It is also important to remember that the goal of this research is to expand knowledge and improve the clinical treatment of PTSD. Despite the large financial investment and the thousands of papers published annually, no treatments have been approved for PTSD since Paxil in 2000. With the incorporation of the directions discussed, rodent models of PTSD can provide more utility and translational impact as they lead to shifts in practice.