The diagnosis of a mental disorder generally depends on clinical observations and phenomenological symptoms reported by the patient. The definition of a given diagnosis is criteria based and relies on the ability to accurately interpret subjective symptoms and complex behavior. This type of diagnosis comprises a challenge to translate to reliable animal models, and these translational uncertainties hamper the development of new treatments. In this review, we will discuss how depressive- like behavior can be induced in rodents, and the relationship between these models and depression in humans. Specifically, we suggest similarities between triggers of depressive-like behavior in animal models and human conditions known to increase the risk of depression, for example exhaustion and bullying. Although we acknowledge the potential problems in comparing animal findings to human conditions, such comparisons are useful for understanding the complexity of depression, and we highlight the need to develop clinical diagnoses and animal models in parallel to overcome translational uncertainties. Keywords: RDoC, stress, resilience, vulnerability Introduction One of the greatest challenges of our society is to prevent and research is hampered by the fact that the psychiatric diagnos- treat mental disorders. Despite major breakthroughs in neuro- tic procedure is open for subjective interpretation of reported science, only limited progress in the treatment of mental disor- symptoms and that it lacks objective measures of behavior or ders has been made in the last 30 years. This can be explained biomarkers. Regarding depression, the collection of disparate partly by a lack of compatibility between neuroscience and clin- symptoms is particularly troublesome (Spitzer et al., 1978). The ical practice. wide variability of symptoms and clinical presentations among The diagnosis major depression was introduced in the mid- subtypes of depression clearly indicates that depression is not a 1970s and incorporated in the third edition of the American homogenous disorder but is actually a spectrum of related dis- Psychiatric Association’s Diagnostic and Statistical Manual orders (Lux and Kendler, 2010). of Mental Disorders (DSM) (Spitzer et al., 1978). Since then, We argue that animal models are necessary for understand- the diagnostic system has been improved and advances have ing the neurobiology of the disease. However, it is futile to seek been made, including increased awareness of mental health one single animal model for the diagnosis depression. Instead, issues and the development of specific forms of psychother - we should use various animal models, where depressive-like apy. However, the translation between clinical work and animal behavior has been induced in different ways, to learn more about Received: May 31, 2017; Revised: February 22, 2018; Accepted: 13 April 2018 © The Author(s) 2018. Published by Oxford University Press on behalf of CINP. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http:// creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, 668 provided the original work is properly cited. For commercial re-use, please contact firstname.lastname@example.org Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Söderlund and Lindskog | 669 the mechanisms underlying the psychiatric conditions they What do we talk about when we talk about model (Table 1). When we compare animal models with clinical depression in rodents? conditions, we need to use less well-established clinical descrip- The translational uncertainties described above, including the tions. These conditions are often associated with an increased lack of functional biomarkers for major depression, compli- risk of developing major depression (Figure 1), but are not major cate the definition of an animal model of depression. Thus, the depression per se. It becomes clear that the animal models are evaluation of the model needs to be based on clinically relevant not diagnostic models, but rather models of risk and vulnerabil- symptoms, and these symptoms must be detectable and quan- ity factors of depression. Importantly, this is not a drawback of tifiable in the animals’ behavior. In this text we refer to “tests” the animal models, but again illustrates that the current diag- as behaviors that can be evaluated, whereas an “animal model” nostic system is neither compatible with neuroscience nor is it is an animal that has been manipulated as to score higher in sufficient to describe the underlying mechanisms of depression. these tests. We will also discuss how these weaknesses can be overcome. Table 1. Animal Models of Depression Based on their Mode of Induction as well as Functional Characteristics Animal Model References Induction Possible Disease Relevance Stress-Induced Chronic mild stress Katz, 1982; Willner et al., 1987 Unpredictable repeated stress Studying of risk-factors for burn-out Unescapable stress Maier, 1984; Telner and Acute, intense stress Possible overlapping mechanisms between Singhal, 1984 PTSD and depression Social defeat Golden et al., 2011; Forced to subordination Bullying as riskfactor for depression Krishnan and Nestler, 2011 Social isolation Grippo et al., 2007; Djordjevic Individual housing et al., 2012 Maternal separation Matthews and Robbins, 2003; Maternal separation Early separation/insecure attachment as Vetulani, 2013 riskfactor for depression Selectivly Bred FSL Gómez-Galán et al., 2013; Sensitivity to cholinergic agents Tests for antidepressant treatments Overstreet and Vulnerability for stress Wegener, 2013 Wistar-Kyoto Paré, 1989; Sensitivity to hypertension Vulnerability for stress Solberg et al., 2001; β-blockers effect on behavior Nam et al., 2014 Other selectively bred Sensitivity to stress, anxiety, Vulnerability factors and risk-behavior models subordination… associated with depression Genetic Manipulations SERT-KO Lira et al., 2003 Total knock-out of serotonin Increased anxiety, serotonergic syndrome, transporter NET-KO Haenisch and Bönisch, 2011 Total knock-out of noradrenaline Protective against depression transporter BDNF modification Kaufman et al., 2006 Knock-out of BDNF or TrkB Study of antidepressant effects vGlut1-KO Garcia-Garcia et al., 2009 Total knock-out of vesicular Depressive-like behavior glutamate transporter Vulnerability to depression DISC1 KO Shen et al., 2008 Total knock-out of DISC1 Overlapping symptoms between schizophrenia and depression Other transgenic Renoir et al., 2013 Specific genetic deletions in Systemic studies on vulnerability to animals targeted organs depression Network Alterations Stimulation of raphe Warden et al., 2012 Optogenetic activation of PFC The role of serotonin in depression nucleus projections to Raphe Nucleus Inhibition of VTA Tye et al., 2013 Optogenetic inhibition of VTA The role of dopamine in depression projections to Nucleus Neuronal activity pattern as vulnerability to Accumbens depression D2 receptors in basal Francis et al., 2014 Targeted modification of Dopamine pharmacology to treat ganglia Dias et al., 2014 D2-containing neurons in the depression striatum Striatal microcircuit involvement in depression Stimulaion of central Tye et al., 2011 Optgenetic activation of central Protective against development of amygdala nucleus of amygdala depression Stimulation of Hsu and Wang, 2014 Optogenetic activation of habenula Reinforcement and aversion in depression Hanbenula Neuroinflammation LPS injection Yirmiya, 1996 Stimulation of inflammation Relationship inflammation - depression Injections of IL6, IL1 or Fleshner et al., 1995; Stimulation of specific inflammatory Relationship inflammation – depression kynurenine Smagin et al., 1996 pathways Systemic signals involved in depression Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 670 | International Journal of Neuropsychopharmacology, 2018 an animal model or the response to a drug; in particular, tests for memory and anxiety are commonly used (Lapiz-Bluhm et al., 2008). Stress-Induced Models of Depression In humans, depression can occur in the absence of any not- able life stressor, while conversely many individuals who are exposed to chronic life stressors never develop clinical depression (Feder et al., 2009). Nevertheless, many patients with depression often describe stress-inducing life events as contributing factors, and stress is commonly used to induce depressive-like behavior in animals (Slattery and Cryan, 2017). Studies using these stress-induced models of depression have generally reported an increase of plasma corticosterone, and the depressive-like behavior is not elicited when the stress response is reduced (such as in adrenalectomized or genetic- ally modified animals; Keeney et al., 2006; Goshen et al., 2008) . Likewise, major depressive disorder has been associated with increased production of corticotropin-releasing hormone and cortisol as well as increased size of the pituitary and the Figure 1. Scheme of exposures (arrows) and vulnerabilities (circles) that adrenals, indicating a general increased activity of the hypo- increase risk of depression. By combining risk factors for depression identified thalamic–pituitary–adrenal axis (Dinan, 1994) together with in the clinic with specific neurobiological manipulations in animal models, we will achieve a better understanding of the pathophysiology of depression. possible glucocorticoid receptor resistance and defective nega- tive feedback (Pariante, 2004). The best established and most commonly used tests for depressive-like behavior in animals have been developed for Chronic Stress predictive value; that is, an animal’s response to a given treat- ment can predict if the treatment will have an antidepressant The most common model, and one of the best-validated, is effect in humans. The Porsolt swim test (Porsolt et al., 1977) the chronic mild stress model (Katz, 1982; Willner et al., 1987) and the tail suspension test (Cryan et al., 2005) measure an in which animals are subjected to stressors several times a animal’s struggle to escape an unpleasant situation (i.e., being day at unpredictable time points. This protocol minimizes the in water or hanging upside-down, respectively), and these impact of acute stress, as each intervention (loud noise, tilting tests are commonly used by the pharmaceutical industry to of the cage, wet bedding, etc.) causes only moderate stress. One predict how patients will respond to a given antidepressant major advantage of the chronic mild stress model is long-last- drug. It can be argued that these tests also have face value, as ing effects allowing investigation of long-term administration of a decreased struggle can be interpreted as a lack of motivation antidepressant drugs. Although this model has a reputation of or despair—behaviors that are common among patients with being unreliable, a recent survey suggested that the unreliability depression. may not be worse in this model compared with other depression Another commonly used test for depression is to measure models (Willner, 2017). anhedonia (a loss of the ability to derive pleasure from an activ- By design, the chronic mild stress model could be viewed as ity that usually produces pleasure) by comparing how much an resembling exhaustion disorder, in humans caused by difficul- animal prefers sweetened water to unsweetened water. A lost ties in coping with life and occupational stress. Exhaustion dis- preference for sweet water is interpreted as anhedonia, a depres- order is characterized by a long prodromal phase that can include sive symptom (Willner et al., 1987). To avoid any possible bias reduced energy and increased emotional instability, eventually due to metabolic factors, anhedonia can be tested more directly progressing to an acute phase that includes exhaustion, hope- using a self-stimulating paradigm where a stimulating electrode lessness, and apathy. Recent studies suggest that exhaustion is implanted in a brain area mediating reward. The animal can disorder is separate from anxiety and major depression (Besèr then self-stimulate by pressing a lever or poking its nose into a et al., 2014), whereas others report considerable comorbidity hole; a decreased propensity to self-stimulate is considered to between exhaustion disorder and major depression (Glise et al., reflect anhedonia (Vogel et al., 1986). 2012). This may indicate either that exhaustion corresponds to A newly developed test is aiming for another core symp- one aspect of depression or that exhaustion increases the risk tom in depression: negative bias. In this test the propensity of depression. Regardless of the interpretation, the chronic mild for a rat to choose a reward rather than to avoid something stress animal model is an important tool for understanding the unpleasant is measured (Hales et al., 2014). Other tests that relationship between depressive-like behavior and sustained, are gaining in popularity are based on natural behaviors, such unpredictable stress. as social interaction and/or the motivation to explore, rather than examining the response to a given stimulus or stressful Acute Stress situation. For example, our group and others have shown that Another well-characterized model of depression displaying a decreased exploratory behavior is associated with a depres- sion-like phenotype (Kasahara et al., 2007; Li et al., 2010; wide variety of depressive-like behavior is the learned help- lessness model. This model is based on repeated exposure to Gómez-Galán et al., 2013; Magara et al., 2015). Several auxil- iary tests can be used to obtain a more complete evaluation of an inescapable acute threat, such as a foot-shock or a pinch to Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Söderlund and Lindskog | 671 the tail in a box where the animal learns that it has nowhere Neurobiological Models of Depression to escape (Maier, 1984; Telner and Singhal, 1984). Social defeat can be considered a variant of the learned helplessness model, Selectively Bred Animal Models where the stressor is chosen to be innate to the rodent, namely Selectively bred animal models are used to study depression, subordination to a dominant peer (Golden et al., 2011; Krishnan with the 2 most commonly used being the Flinders sensitive line and Nestler, 2011). Typically, animals are put as intruders in (FSL) and the Wistar-Kyoto line. The FSL rat was selectively bred the home cage of a large dominant male and thus forced to for sensitivity to acetylcholinesterase inhibitors (Overstreet subordination. et al., 1986) and displays several depressive-like behaviors, In addition to being well characterized models for depres- including increased immobility time in the Porsolt’s swim test, sion, the learned helplessness model, as well as the social decreased cognition, and increased anxiety (Gómez-Galán et al., defeat model, produce physiologic and behavioral symptoms 2013; Overstreet and Wegener, 2013). Thus, one could hypothe- that models aspects of PTSD (Pulliam, 2010; Schöner, 2017). size that an increased sensitivity in the acetylcholine system The exposure to an inescapable acute threat, as in the learned is associated with depressive-like symptoms, although FSL helplessness model, corresponds to the first DSM criterion for rats also have deficiencies in many other systems, including posttraumatic stress disorder (PTSD). Interestingly, situations of dopaminergic and glutamatergic transmission (Overstreet and power imbalance such as workplace bullying, often including Wegener, 2013). sexual harassment, are also associated with an increased risk The Wistar-Kyoto line was developed as a control group of development of PTSD (Spence Laschinger and Nosko, 2015) for hypertensive rats and was subsequently found to have as well as depression (Lund et al., 2009). Moreover, patients increased stress responses, altered sleep patterns, and depres- exposed to trauma or stressful events often exhibit anhedonia sive-like behavior in the Porsolt’s swim test (Paré, 1989; Solberg and dysphoric symptoms (among others) rather than anxiety- et al., 2001; Nam et al., 2014). These rats have a mixed response or fear-based symptoms. The learned helplessness and social to antidepressants and have been selectively bred further to defeat models may thus be useful tools to study how differ- establish lines of “more immobility” and “less immobility,” ent types of acute, severe stress affects the risk of acquiring which respond differently to antidepressants (Will et al., 2003; depressive-like behavior. Andrus et al., 2012). Although we still do not understand the underlying mecha- Social Stress nisms of depressive-like behavior in the FSL and Wistar-Kyoto lines, the transmitter systems used for selection in the devel- Another way to use social stress in animal models is social iso- opment of these two lines are paralleled with observations in lation. Individual housing for 4 weeks has been shown to induce human depression. Cholinergic signaling, which was the tar - depressive-like behavior (Grippo et al., 2007; Djordjevic et al., get system in the development of the FSL rat, regulates social 2012). A specific form of social isolation is maternal separation, stress, anxiety, and depressive-like behavior (Mineur et al., another commonly used model of depression (Matthews and 2013). The adrenergic system, which regulates blood pres- Robbins, 2003; Vetulani, 2013). In this model, the lactating dam sure and was targeted in the development of the Wistar-Kyoto is taken out of the cage for about 3 hours every day from the sec- rats, has not been primarily associated with depression, and ond to the twelfth day postpartum. This treatment is complex results obtained from clinical trials using antihypertensive and goes beyond pure isolation (Lehmann and Feldon, 2000), drugs vary. Although ß-blockers, and propranolol in particular, and it leads to depressive-like behavior that persists into adult- have been found to be associated with depression, it has been hood. This agrees with the fact that childhood separation and argued that the behavioral changes associated with propran- parental neglect can lead to long-lasting attachment insecurity olol do not resemble classic depressive syndrome (Patten and in humans (Waters et al., 2000). Both attachment insecurity and Barbui, 2004). early separation have been associated with depression (Roberts In addition to the well-characterized FSL and Wistar-Kyoto et al., 1996; Otowa et al., 2014), and emotional dysregulation has lines, several groups have bred animals selecting for a specific been proposed as the link between these triggering events and depression-like behavior. In addition to the behaviors that they depression (Malik et al., 2015). were selected on, these animals show disturbed sleep pattern, increased plasma corticosterone, and reduced levels of base- Vulnerability to Stress line serotonin and brain-derived neurotrophic factor (BDNF) (El Yacoubi et al., 2003; Will et al., 2003; Gersner et al., 2014). In addition to these commonly used models of stress-induced Selectively bred models can focus on one particular behav- depression, many other paradigms of stress have been used. ior among the depressive-like symptoms, thus allowing study Nevertheless, we still lack a comprehensive picture of how of the neurobiology or potential treatment of that particular different depression-triggering mechanisms are induced by construct. In contrast to models where selection is based on a different modalities of stress, or differences in when, and for known mechanism, models based on a specific behavior allow how long, stress is applied. Comparing different stress-induced us to discover unexpected mechanisms of relevance to depres- models will help us to get a clearer picture of the differential sion. The drawback of the unknown etiology is that it does not effect of different types of stress and its timing. Experimental allow us to draw causative conclusions. animal models can also be studied to identify factors of vulner- ability and resilience to different types of stressors (Dias et al., 2014). Vulnerability factors and environment may interact to Genetic Models produce depression, as has been shown for a specific serotonin transporter allele that renders an individual more sensitive or Twin studies suggest that the genetic contribution to depression less sensitive to stressful life events (Caspi et al., 2010; Wang ranges from 10% to 50% (Silberg et al., 1990; Edvardsen et al., et al., 2011). 2009), and several genes have been associated with depression; Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 672 | International Journal of Neuropsychopharmacology, 2018 however, no gene has emerged as the main gene underlying the in the ventral tegmental area‒nucleus accumbens connection disorder (Flint and Kendler, 2014). The lack of conclusive genetics can reduce or increase depressive-like behavior, respectivly (Tye data in animal models and human patients can have two pos- et al., 2013), and phasic firing of these projections increases sus- sible explanations: either depression is a highly complex trait, or ceptibility to social stress (Chaudhury et al., 2013). Stimulating depression is not actually one disease but rather a disease spec- the dopamine D -receptor expressing neurons in the basal trum that can be induced by a wide variety of biological causes ganglia increases the resilience to developing depressive-like (Andrus et al., 2012; Gómez-Galán et al., 2016). behavior, whereas activating the D -receptor expressing neu- Nevertheless, attempts have been made to use genetic mod- rons increases the vulnerability (Francis et al., 2015), suggest- ifications in mice to study depression, and the serotonergic and ing a differential role of these two populations in mediating noradrenergic systems have provided obvious targets for such resilience to depressive-like behavior. Moreover, overexpressing modifications. Interestingly, genetic deletions of the norepin- of the signaling molecule β-catenin in D -expressing neurons 2- ephrine transporter are protective against depression induced only specifically increases resilience to the social defeat proto- by restrained stress or social defeat (Haenisch and Bönisch, col (Dias et al., 2014). Finally, stimulating the central nucleus of 2011), whereas deleting the serotonin transporter leads to a the amygdala reduces anxiety-like behavior (Tye et al., 2011), phenotype that resembles the acute effect of selective sero- whereas specific molecular manipulations have revealed that tonin reuptake inhibitor treatment, including increased anxiety the habenula regulates reinforcement and aversion (Hsu and or even serotonergic syndrome (Lira et al., 2003). Genetic down- Wang, 2014) that can explain part of the depressive-like behav- regulation of tryptophan hydroxylase, the rate-limiting enzyme ior (Hsu et al., 2016). in serotonin synthesis, increases depressive-like behavior These experiments using specific genetic manipulations (Savelieva et al., 2008). have been extremely fruitful in terms of elucidating the neuro- A single nucleotide polymorphism in the human BDNF biological mechanisms that underlie various dimensions of gene increases vulnerability to stress-induced depression depressive-like behavior. Together with refined imaging in (Kaufman et al., 2006), making this gene a target for depres- animals as well as humans, we will get an increased under- sion models. However, when BDNF function is reduced standing of how a specific neuronal population or network can through genetic modification of BDNF or its receptor TrkB, be involved in the pathophysiology of depression (Chen et al., no unequivocal link to depression is found. Instead, reduced 2017), and they provide valuable information that can be used BDNF function leads to cognitive impairment and obesity to predict the clinical response to pharmacotherapy (Haenisch (Lindholm and Castrén, 2014). and Bönisch, 2011). However, it is striking to note how differ - Recent work describes depression as a misregulated glu- ent manipulations in different areas of the brain can induce a tamate transmission or an unbalance in excitation and similar phenotype. This may be interpreted either as overlap- inhibition, leading to studies targeting genes involved in the ping systems or that brain circuits interact: disrupting any one regulation of glutamate. For example, reduced levels of the of them trigger similar pathology. Thus, in addition to identify- vesicular glutamate transporter lead to increased depressive- ing the pieces of the puzzle, we must also assemble these pieces like behavior and increased vulnerability to stressors (Garcia- correctly to form a coherent picture. Garcia et al., 2009). In contrast to selectively bred animals, a model that has been Neuroinflammation modified with targeted genetic manipulation could allow to There is a well-established link between inflammation and establish a causative link between a biological function and the depression (Krishnadas and Cavanagh, 2012), and injection of depressive-like behavior. However, it is important to note that the inflammation-inducing agent lipopolysaccharide (LPS) is a the specificity of a genetic mutation can be deceiving. A gen- popular model for inducing depression (Yirmiya, 1996). Specific etic change will affect many compensatory and downstream pathways in the inflammatory response have also been targeted pathways, and the most relevant mechanism may be hard to to study the mechanisms of inflammation-induced depression, pinpoint. In addition, many pathways converge on the same including IL-6, IL-1, and kynurenine. The injection of proinflam- behavior; for example, the 4 distinct genetic models targeting matory cytokines activates the hypothalamic–pituitary–adrenal serotonin, glucocorticoid, glutamate, and cannabinoid signal- axis as seen through increased cortisol levels, depletion of hypo- ing were all recently reported to share the behavioral changes thalamic noradrenaline, and increased extracellular levels of nor - (Hoyle et al., 2011). For an extensive review of targeted genetic adrenaline (Fleshner et al., 1995; Smagin et al., 1996). Activation of modifications that lead to depressive-like behavior, see (Renoir inflammatory pathways is, however, not enough, since cytokine- et al., 2013). induced depressive-like behavior was abolished in the serotonin transporter knock-out mice; this implies the involvement of the Animal Models with Specific Network Alterations serotonin modulatory system (van Heesch et al., 2013). In add- The development of targeted genetic manipulations and ition to soluble signaling molecules, the vagus nerve plays a major optogenetics, where a light-activated ion-channel can be select- role in the communication between the peripheral inflammatory ively inserted and activated in a population of neurons, has response and the brain, and vagotomy protects from IL-1 and LPS- made it possible to study specific circuits and/or signaling path- induced depressive behavior (Konsman et al., 2000). ways in behavioral dimensions relevant to depression (Berton In humans, depression has been associated with increased et al., 2012; Deisseroth, 2014). For example, stimulation of the serum levels of the proinflammatory cytokines IL-1β, IL-6, and raphe nucleus through direct activation of projection neurons TNFα (Howren et al., 2009; Dowlati et al., 2010). Furthermore, in the prefrontal cortex reduces immobility in the forced swim during severe infections and inflammatory diseases, patients test (Warden et al., 2012) as does inhibition of NMDA receptors often show a typical sickness behavior defined by lack of energy, in prefrontal projections to the thalamus (Miller et al., 2017). apathy, lack of interest, reduced intake of food, etc., thus closely In addition, activating or inhibiting the dopaminergic neurons resembling major depression (Dantzer et al., 2008). Recent Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Söderlund and Lindskog | 673 work reveals tryptophan catabolism as a converging point for immediately useful clinical tool.” However, as discouraging as immunological factors and neurotransmission. The conversion this may sound to any clinical psychiatrist, we argue that such of L-tryptophan to N-formyl-kynurenine is governed by the rate- a common context can be achieved without dramatic paradig- limiting IDO/TDO enzyme (Rafice et al., 2009). IDO is modulated matic shifts in the clinic. during immune responses (Babcock and Carlin, 2000), and major The current diagnostic criteria for major depression do not depression has been associated with increased levels of the emphasize how core symptoms of depression like anhedonia, tryptophan metabolite quinolinic acid in the cerebrospinal fluid motor retardation, and despair have developed. However, it is (Bay-Richter et al., 2015). only when the reported symptoms are combined with a thor - Overall, there is ample evidence from both animals and ough analysis of exposures to risk factors and vulnerability humans that inflammatory pathways are involved in depres- factors that the clinical picture emerges. Identification of the sion. Evaluation of their cross-talk is an important field of future relevant constructs of RDoC overlaps with the taking of such work (Feder et al., 2009). a clinical history (Table 2) and will be directly relevant for the diagnosis. From a clinical point of view, RDoC could be used to improve Discussion the diagnostic specificity and specificity of treatment by increas- ing the theoretical ground on which the clinical decisions are We have reviewed animal models used in depression research, based on how the depressive-like behavior is based. RDoC may thus guide the clinician to more information from the relevant animal models. For example, if the clinical induced. We argue that it is impossible to find one valid ani- mal model of depression that reflects all the aspects of the picture is dominated by the construct “response to threat” of the negative valence domain, the clinician may find more infor - disease; rather, different animal models should be used to study different aspects of depression. Experimental models mation from the corresponding animal models of social defeat and learned helplessness, and if the clinical picture involves offer the opportunity to expose the animals to one factor at a time, thus allowing us to study vulnerability factors and the construct “disrupted attachment” of the social processes domain, the clinician may gather more information from the objectively measure and quantify proposed mechanisms involved in depression. A critical step to develop better corresponding animal model of maternal separation rather than without further notice giving the patient a major depressive dis- treatment and prevention of depression is now to improve the understanding between clinical practitioners and basic order diagnosis. From a neuroscience perspective, a common context of researchers through the development of a common context for depression. In our point of view, such a context must depression, where clinicians put more emphasis on risk fac- tors and vulnerability factors specified as constructs of RDoC, involve symptoms (regardless of whether these fulfill all the necessary criteria for major depressive disorder according to would lead to a stronger validation of the animal models, gen- erate more specific hypothesis, and enhance development of the DSM or not) as well as etiologic factors, exposure, and vulnerabilities. more specific drugs and better tools to uncover the neurobiology behind symptoms (Figure 2). To enhance the interaction between clinic and neuro- biology, the NIMH has proposed to use Research Domain In summary, the gap between the clinical practice and neuroscience may be smaller than initially thought, and the Criteria (RDoC) as a novel approach to categorizing psychiatric conditions (see http://www.nimh.nih.gov/research-priorities/ use of RDoC may indeed be a way to bridge this gap. Clinical experience is necessary to develop fruitful animal models, but rdoc/constructs/rdoc-matrix.shtml). The intention is to cre- ate a system based on well-defined constructs (e.g., response the animal models are also needed to delineate the complex patterns of different exposures and vulnerability factors char - to threat, responsiveness to reward, long-term memory within the larger domains: negative valence system, positive acterizing the clinical situation, indicating the potential mutual benefit between research and clinic (Figure 2). RDoC may facili- valence system, cognition, social response, and arousal) that will facilitate communication between research and clinic. At tate these iterations. If constructs are studied in patients and in animal models, this will shorten the length of the iteration cycle their homepage, the NIHM describes RDoC as “a framework to guide classification of patients for research studies, not as an and facilitate early identification of potential breakthroughs or Table 2. Research Domain Criteria as Risk Factors for Depression, Their Clinical Manifestations, and the Coresponding Animal Behavior Domain Constructs Clinical Manifestations Experimental Animal Tests Negative valence Acute threat Powerlessness Lack of social interaction Porsolt swim test Sustained threat Exhaustion, powerlessness Porsolt swim test Lack of social interaction Frustrative non-reward Exhaustion Lack social interaction Responses to potential harm Anxiety Elevated plus maze, Open field, Novelty suppressed feeding Positive valence Reward valuation Anhedonia Sucrose preference, self stimulation Arousal Arousal, interaction with Appetite, sleep, sex, locomotors Sleep-pattern, social interaction, modified valence activity serial-5 choice Social processes Disrupted attachment Disorganized attachment Aggressive behavior Disrupted affiliation Victimization Subordination, Social withdrawal Social withdrawal Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 674 | International Journal of Neuropsychopharmacology, 2018 Besèr A, Sorjonen K, Wahlberg K, Peterson U, Nygren A, Asberg M (2014) Construction and evaluation of a self rating scale for stress-induced exhaustion disorder, the Karolinska exhaus- tion disorder scale. Scand J Psychol 55:72–82. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE (2010) Genetic sensitivity to the environment: the case of the serotonin transporter gene and its implications for studying complex diseases and traits. Am J Psychiatry 167:509–527. Chaudhury D, et al. (2013) Rapid regulation of depression-related behaviours by control of midbrain dopamine neurons. Nature 493:532–536. Chen T, Tanaka M, Wang Y, Sha S, Furuya K, Chen L, Sokabe M (2017) Neurosteroid dehydroepiandrosterone enhances activ- ity and trafficking of astrocytic GLT-1 via σ1receptor-mediated PKC activation in the hippocampal dentate gyrus of rats. Glia 65:1491–1503. Cryan JF, Mombereau C, Vassout A (2005) The tail suspension test as a model for assessing antidepressant activity: review of pharmacological and genetic studies in mice. Neurosci Biobehav Rev 29:571–625. Figure 2. To be able to study relevant mechanism of depression, we need to find Dantzer R, O’Connor JC, Freund GG, Johnson RW, Kelley KW (2008) a common context for clinical and experimental work. For most, if not all, dis- From inflammation to sickness and depression: when the eases it will be impossible to generate animal models that can serve as faithful immune system subjugates the brain. Nat Rev Neurosci 9:46–56. models of a specific diagnosis that could be used to test treatments. Instead, Deisseroth K (2014) Circuit dynamics of adaptive and maladap- animal models should be seen as experimental tools where hypothesis regard- tive behaviour. Nature 505:309–317. ing biological correlations or mechanisms can be investigated and where new Dias C, et al. (2014) Β-catenin mediates stress resilience through hypothesis regarding the clinical condition can be tested. It is only by combining findings from the clinical and experimental fields that we will achieve better Dicer1/microRNA regulation. Nature 516:51–55. diagnosis and treatment. Dinan TG (1994) Glucocorticoids and the genesis of depressive illness. A psychobiological model. Br J Psychiatry 164:365–371. Djordjevic J, Djordjevic A, Adzic M, Radojcic MB (2012) Effects of dead-ends. Furthermore, the different animal models of depres- chronic social isolation on wistar rat behavior and brain plas- sion, elucidating different aspects of depression, may be used ticity markers. Neuropsychobiology 66:112–119. to refine these constructs and make them even more relevant Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim for the clinic. EK, Lanctôt KL (2010) A meta-analysis of cytokines in major depression. Biol Psychiatry 67:446–457. Edvardsen J, Torgersen S, Røysamb E, Lygren S, Skre I, Onstad S, Acknowledgments Øien PA (2009) Unipolar depressive disorders have a common The authors thank the students at the graduate course genotype. J Affect Disord 117:30–41. Neurobiology of Psychiatric Disorders for interesting discus- El Yacoubi M, Bouali S, Popa D, Naudon L, Leroux-Nicollet I, sions on the topic of this review. Hamon M, Costentin J, Adrien J, Vaugeois JM (2003) Behavioral, neurochemical, and electrophysiological characterization of a genetic mouse model of depression. Proc Natl Acad Sci U S Statement of Interest A 100:6227–6232. None. Feder A, Nestler EJ, Charney DS (2009) Psychobiology and molecu- lar genetics of resilience. Nat Rev Neurosci 10:446–457. Fleshner M, Goehler LE, Hermann J, Relton JK, Maier SF, Watkins References LR (1995) Interleukin-1 beta induced corticosterone elevation Andrus BM, Blizinsky K, Vedell PT, Dennis K, Shukla PK, Schaffer and hypothalamic NE depletion is vagally mediated. Brain DJ, Radulovic J, Churchill GA, Redei EE (2012) Gene expression Res Bull 37:605–610. patterns in the hippocampus and amygdala of endogen- Flint J, Kendler KS (2014) The genetics of major depression. ous depression and chronic stress models. Mol Psychiatry Neuron 81:484–503. 17:49–61. Francis TC, Chandra R, Friend DM, Finkel E, Dayrit G, Miranda J, Babcock TA, Carlin JM (2000) Transcriptional activation of Brooks JM, Iñiguez SD, O’Donnell P, Kravitz A, Lobo MK (2015) indoleamine dioxygenase by interleukin 1 and tumor necro- Nucleus accumbens medium spiny neuron subtypes medi- sis factor alpha in interferon-treated epithelial cells. Cytokine ate depression-related outcomes to social defeat stress. Biol 12:588–594. Psychiatry 77:212–222. Bay-Richter C, Linderholm KR, Lim CK, Samuelsson M, Träskman- Garcia-Garcia AL, Elizalde N, Matrov D, Harro J, Wojcik SM, Bendz L, Guillemin GJ, Erhardt S, Brundin L (2015) A role for Venzala E, Ramírez MJ, Del Rio J, Tordera RM (2009) Increased inflammatory metabolites as modulators of the glutamate vulnerability to depressive-like behavior of mice with N-methyl-D-aspartate receptor in depression and suicidality. decreased expression of VGLUT1. Biol Psychiatry 66:275–282. Brain Behav Immun 43:110–117. Gersner R, Gal R, Levit O, Moshe H, Zangen A (2014) Inherited Berton O, Hahn CG, Thase ME (2012) Are we getting closer to behaviors, BDNF expression and response to treatment valid translational models for major depression? Science in a novel multifactorial rat model for depression. Int J 338:75–79. Neuropsychopharmacol 17:945–955. Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Söderlund and Lindskog | 675 Glise K, Ahlborg G Jr, Jonsdottir IH (2012) Course of mental symp- Krishnan V, Nestler EJ (2011) Animal models of depression: toms in patients with stress-related exhaustion: does sex or molecular perspectives. Curr Top Behav Neurosci 7:121–147. age make a difference? BMC Psychiatry 12:18. Lapiz-Bluhm MD, Bondi CO, Doyen J, Rodriguez GA, Bédard- rd Golden SA, Covington HE 3 , Berton O, Russo SJ (2011) A stand- Arana T, Morilak DA (2008) Behavioural assays to model cog- ardized protocol for repeated social defeat stress in mice. Nat nitive and affective dimensions of depression and anxiety in Protoc 6:1183–1191. rats. J Neuroendocrinol 20:1115–1137. Gómez-Galán M, De Bundel D, Van Eeckhaut A, Smolders I, Lehmann J, Feldon J (2000) Long-term biobehavioral effects of Lindskog M (2013) Dysfunctional astrocytic regulation of maternal separation in the rat: consistent or confusing? Rev glutamate transmission in a rat model of depression. Mol Neurosci 11:383–408. Psychiatry 18:582–594. Li Y, Zheng X, Liang J, Peng Y (2010) Coexistence of anhedonia Gómez-Galán M, Femenía T, Åberg E, Graae L, Van Eeckhaut A, and anxiety-independent increased novelty-seeking behav- Smolders I, Brené S, Lindskog M (2016) Running opposes the ior in the chronic mild stress model of depression. Behav effects of social isolation on synaptic plasticity and transmis- Processes 83:331–339. sion in a rat model of depression. Plos One 11:e0165071. Lindholm JS, Castrén E (2014) Mice with altered BDNF signaling Goshen I, Kreisel T, Ben-Menachem-Zidon O, Licht T, Weidenfeld as models for mood disorders and antidepressant effects. J, Ben-Hur T, Yirmiya R (2008) Brain interleukin-1 mediates Front Behav Neurosci 8:143. chronic stress-induced depression in mice via adrenocortical Lira A, Zhou M, Castanon N, Ansorge MS, Gordon JA, Francis activation and hippocampal neurogenesis suppression. Mol JH, Bradley-Moore M, Lira J, Underwood MD, Arango V, Kung Psychiatry 13:717–728. HF, Hofer MA, Hen R, Gingrich JA (2003) Altered depression- Grippo AJ, Gerena D, Huang J, Kumar N, Shah M, Ughreja R, Carter related behaviors and functional changes in the dorsal CS (2007) Social isolation induces behavioral and neuroendo- raphe nucleus of serotonin transporter-deficient mice. Biol crine disturbances relevant to depression in female and male Psychiatry 54:960–971. prairie voles. Psychoneuroendocrinology 32:966–980. Lund R, Nielsen KK, Hansen DH, Kriegbaum M, Molbo D, Due Haenisch B, Bönisch H (2011) Depression and antidepressants: P, Christensen U (2009) Exposure to bullying at school and insights from knockout of dopamine, serotonin or noradren- depression in adulthood: a study of Danish men born in 1953. aline re-uptake transporters. Pharmacol Ther 129:352–368. Eur J Public Health 19:111–116. Hales CA, Stuart SA, Anderson MH, Robinson ES (2014) Modelling Lux V, Kendler KS (2010) Deconstructing major depression: a val- cognitive affective biases in major depressive disorder using idation study of the DSM-IV symptomatic criteria. Psychol rodents. Br J Pharmacol 171:4524–4538. Med 40:1679–1690. Howren MB, Lamkin DM, Suls J (2009) Associations of depres- Magara S, Holst S, Lundberg S, Roman E, Lindskog M (2015) sion with C-reactive protein, IL-1, and IL-6: a meta-analysis. Altered explorative strategies and reactive coping style in the Psychosom Med 71:171–186. FSL rat model of depression. Front Behav Neurosci 9:89. Hoyle D, Juhasz G, Aso E, Chase D, del Rio J, Fabre V, Hamon M, Maier SF (1984) Learned helplessness and animal models of Lanfumey L, Lesch KP, Maldonado R, Serra MA, Sharp T, Tordera depression. Prog Neuropsychopharmacol Biol Psychiatry R, Toro C, Deakin JF (2011) Shared changes in gene expression 8:435–446. in frontal cortex of four genetically modified mouse models of Malik S, Wells A, Wittkowski A (2015) Emotion regulation as depression. Eur Neuropsychopharmacol 21:3–10. a mediator in the relationship between attachment and Hsu YW, Wang SD, Wang S, Morton G, Zariwala HA, de la Iglesia depressive symptomatology: a systematic review. J Affect HO, Turner EE (2014) Role of the dorsal medial habenula in Disord 172:428–444. the regulation of voluntary activity, motor function, hedonic Matthews K, Robbins TW (2003) Early experience as a determin- state, and primary reinforcement. J Neurosci 34:11366–11384. ant of adult behavioural responses to reward: the effects of Kasahara M, Groenink L, Breuer M, Olivier B, Sarnyai Z (2007) repeated maternal separation in the rat. Neurosci Biobehav Altered behavioural adaptation in mice with neural cortico- Rev 27:45–55. trophin-releasing factor overexpression. Genes Brain Behav Miller OH, Bruns A, Ben Ammar I, Mueggler T, Hall BJ (2017) 6:598–607. Synaptic regulation of a thalamocortical circuit controls Katz RJ (1982) Animal model of depression: pharmacological depression-related behavior. Cell Rep 20:1867–1880. sensitivity of a hedonic deficit. Pharmacol Biochem Behav Mineur YS, Obayemi A, Wigestrand MB, Fote GM, Calarco CA, 16:965–968. Li AM, Picciotto MR (2013) Cholinergic signaling in the Kaufman J, Yang BZ, Douglas-Palumberi H, Grasso D, Lipschitz hippocampus regulates social stress resilience and anx- D, Houshyar S, Krystal JH, Gelernter J (2006) Brain-derived iety- and depression-like behavior. Proc Natl Acad Sci U S A neurotrophic factor-5-HTTLPR gene interactions and envir - 110:3573–3578. onmental modifiers of depression in children. Biol Psychiatry Nam H, Clinton SM, Jackson NL, Kerman IA (2014) Learned help- 59:673–680. lessness and social avoidance in the Wistar-Kyoto rat. Front Keeney A, Jessop DS, Harbuz MS, Marsden CA, Hogg S, Blackburn- Behav Neurosci 8:109. Munro RE (2006) Differential effects of acute and chronic Otowa T, York TP, Gardner CO, Kendler KS, Hettema JM (2014) The social defeat stress on hypothalamic-pituitary-adrenal impact of childhood parental loss on risk for mood, anxiety axis function and hippocampal serotonin release in mice. and substance use disorders in a population-based sample of J Neuroendocrinol 18:330–338. male twins. Psychiatry Res 220:404–409. Konsman JP, Luheshi GN, Bluthé RM, Dantzer R (2000) The vagus Overstreet DH, Wegener G (2013) The flinders sensitive line rat nerve mediates behavioural depression, but not fever, in model of depression–25 years and still producing. Pharmacol response to peripheral immune signals; a functional ana- Rev 65:143–155. tomical analysis. Eur J Neurosci 12:4434–4446. Overstreet DH, Janowsky DS, Gillin JC, Shiromani PJ, Sutin EL Krishnadas R, Cavanagh J (2012) Depression: an inflammatory (1986) Stress-induced immobility in rats with cholinergic illness? J Neurol Neurosurg Psychiatry 83:495–502. supersensitivity. Biol Psychiatry 21:657–664. Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018 676 | International Journal of Neuropsychopharmacology, 2018 Paré WP (1989) Stress ulcer susceptibility and depression in Spence Laschinger HK, Nosko A (2015) Exposure to workplace bully- Wistar Kyoto (WKY) rats. Physiol Behav 46:993–998. ing and post-traumatic stress disorder symptomology: the role Pariante CM (2004) Glucocorticoid receptor function in vitro in of protective psychological resources. J Nurs Manag 23:252–262. patients with major depression. Stress 7:209–219. Spitzer RL, Endicott J, Robins E (1978) Research diagnostic crite- Patten SB, Barbui C (2004) Drug-induced depression: a system- ria: rationale and reliability. Arch Gen Psychiatry 35:773–782. atic review to inform clinical practice. Psychother Psychosom Telner JI, Singhal RL (1984) Psychiatric progress. The learned 73:207–215. helplessness model of depression. J Psychiatr Res 18:207–215. Porsolt RD, Le Pichon M, Jalfre M (1977) Depression: a new ani- Tye KM, Prakash R, Kim SY, Fenno LE, Grosenick L, Zarabi H, mal model sensitive to antidepressant treatments. Nature Thompson KR, Gradinaru V, Ramakrishnan C, Deisseroth K 266:730–732. (2011) Amygdala circuitry mediating reversible and bidirec- Pulliam JV, Dawaghreh AM, Alema-Mensah E, Plotsky PM (2010) tional control of anxiety. Nature 471:358–362. Social defeat stress produces prolonged alterations in acous- Tye KM, Mirzabekov JJ, Warden MR, Ferenczi EA, Tsai HC, tic startle and body weight gain in male Long Evans rats. J Finkelstein J, Kim SY, Adhikari A, Thompson KR, Andalman Psychiatr Res 44:106–111. AS, Gunaydin LA, Witten IB, Deisseroth K (2013) Dopamine Rafice SA, Chauhan N, Efimov I, Basran J, Raven EL (2009) neurons modulate neural encoding and expression of Oxidation of L-tryptophan in biology: a comparison between depression-related behaviour. Nature 493:537–541. tryptophan 2,3-dioxygenase and indoleamine 2,3-dioxyge- van Heesch F, Prins J, Konsman JP, Westphal KG, Olivier B, Kraneveld nase. Biochem Soc Trans 37:408–412. AD, Korte SM (2013) Lipopolysaccharide-induced anhedonia is Renoir T, Pang TY, Hannan AJ (2013) Effects of environmen- abolished in male serotonin transporter knockout rats: an intra- tal manipulations in genetically targeted animal models of cranial self-stimulation study. Brain Behav Immun 29:98–103. affective disorders. Neurobiol Dis 57:12–27. Vetulani J (2013) Early maternal separation: a rodent model of Roberts JE, Gotlib IH, Kassel JD (1996) Adult attachment security and depression and a prevailing human condition. Pharmacol symptoms of depression: the mediating roles of dysfunctional Rep 65:1451–1461. attitudes and low self-esteem. J Pers Soc Psychol 70:310–320. Vogel GW, Minter K, Woolwine B (1986) Effects of chronically Savelieva KV, Zhao S, Pogorelov VM, Rajan I, Yang Q, Cullinan administered antidepressant drugs on animal behavior. E, Lanthorn TH (2008) Genetic disruption of both trypto- Physiol Behav 36:659–666. phan hydroxylase genes dramatically reduces serotonin and Wang Z, Baker DG, Harrer J, Hamner M, Price M, Amstadter A affects behavior in models sensitive to antidepressants. Plos (2011) The relationship between combat-related posttrau- One 3:e3301. matic stress disorder and the 5-HTTLPR/rs25531 polymorph- Shen S, Lang B, Nakamoto C, Zhang F, Pu J, Kuan SL, Chatzi C, He ism. Depress Anxiety 28:1067–1073. S, Mackie I, Brandon NJ, Marquis KL, Day M, Hurko O, McCaig Warden MR, Selimbeyoglu A, Mirzabekov JJ, Lo M, Thompson KR, CD, Riedel G, St Clair D (2008) Schizophrenia-related neural Kim SY, Adhikari A, Tye KM, Frank LM, Deisseroth K (2012) A and behavioral phenotypes in transgenic mice expressing prefrontal cortex-brainstem neuronal projection that con- truncated disc1. J Neurosci 28:10893–10904. trols response to behavioural challenge. Nature 492:428–432. Schöner J, Heinz A, Endres M, Gertz K, Kronenberg G (2017) Post- Waters E, Merrick S, Treboux D, Crowell J, Albersheim L (2000) traumatic stress disorder and beyond: an overview of rodent Attachment security in infancy and early adulthood: a stress models. J Cell Mol Med 21:2248–2256. twenty-year longitudinal study. Child Dev 71:684–689. Silberg JL, Heath AC, Kessler R, Neale MC, Meyer JM, Eaves LJ, Will CC, Aird F, Redei EE (2003) Selectively bred wistar-kyoto rats: Kendler KS (1990) Genetic and environmental effects on self- an animal model of depression and hyper-responsiveness to reported depressive symptoms in a general population twin antidepressants. Mol Psychiatry 8:925–932. sample. J Psychiatr Res 24:197–212. Willner P (2017) Reliability of the chronic mild stress model of Slattery DA, Cryan JF (2017) Modelling depression in ani- depression: a user survey. Neurobiol Stress 6:68–77. mals: at the interface of reward and stress pathways. Willner P, Towell A, Sampson D, Sophokleous S, Muscat R (1987) Psychopharmacology (Berl) 234:1451–1465. Reduction of sucrose preference by chronic unpredictable Smagin GN, Swiergiel AH, Dunn AJ (1996) Peripheral adminis- mild stress, and its restoration by a tricyclic antidepressant. tration of interleukin-1 increases extracellular concentra- Psychopharmacology (Berl) 93:358–364. tions of norepinephrine in rat hypothalamus: comparison Wook Koo J, Labonté B, Engmann O, Calipari ES, Juarez B, Lorsch Z, with plasma corticosterone. Psychoneuroendocrinology Walsh JJ, Friedman AK, Yorgason JT, Han MH, Nestler EJ (2016) 21:83–93. Essential role of mesolimbic brain-derived neurotrophic fac- Solberg LC, Olson SL, Turek FW, Redei E (2001) Altered hormone tor in chronic social stress-induced depressive behaviors. Biol levels and circadian rhythm of activity in the WKY rat, a Psychiatry 80:469–478. putative animal model of depression. Am J Physiol Regul Yirmiya R (1996) Endotoxin produces a depressive-like episode Integr Comp Physiol 281:R786–R794. in rats. Brain Res 711:163–174. Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/668/4982723 by Ed 'DeepDyve' Gillespie user on 03 July 2018
International Journal of Neuropsychopharmacology – Oxford University Press
Published: Apr 23, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera