Introduction: Sex differences in depression

Depression is the most prevalent mood disorder and a leading cause of mortality and morbidity worldwide. Despite its high incidence and socioeconomic impact, the etiology of depression remains poorly understood. It involves a combination of genetics and environmental factors as well as the dynamic interaction of a number of brain regions, however, it is not clear how these factors interact to trigger depression. Adding to this complexity are the differences observed between men and women. Women are nearly twice as likely as men to suffer from depression [1], and more than 2/3 of suicide attempts are by women [2]. Female depressed patients show greater severity, earlier age of onset, and increased duration of depressive episodes as compared to male patients [3]. The clinical presentation of symptoms of depression is sexually dimorphic as well, including differences in comorbid conditions [3]. Currently, the antidepressants commonly used for the treatment of depression are selective serotonin reuptake inhibitors (SSRIs) and tricyclics (TCA). Yet, these are successful in only a fraction of the population and take weeks to months to be effective in responders. Efficacy also differs between the sexes. Evidence put forth in an attempt to explain the disparity in depression and antidepressant response between males and females includes differences in neuronal circuitry, hormone levels, and metabolism [4,5,6]; however, the reason for these sex differences remains unclear.

Despite the evidence for greater prevalence of depression among women, there is considerably less attention devoted to studying depression in females or sex differences in depression. In fact, many of the animal models used to study biological mechanisms of depressive symptoms or of antidepressant response rely solely on male subjects ignoring a critically important study population. The development and verification of these assays in males has without a doubt been useful in reducing variability and producing meaningful data with respect to depression and antidepressant response in males. As sex differences result in variation in each step between animal behavior and clinical presentation of depression, the optimization of animal models for application in males only has made it exceedingly difficult to incorporate females into this same framework. This disconnect makes the already challenging study of the pathophysiology underlying depression even more difficult, obscuring and delaying the development of broadly effective antidepressants. This review aims to highlight the important neurobiological factors underlying sex differences in depression and antidepressant response. In delving into mechanisms that potentially explain these differences, we will dedicate our focus to neuronal circuits and synaptic transmission, as other important aspects, such as hormone regulation and pharmacokinetics have been reviewed extensively elsewhere [4, 7,8,9,10,11]. Investigation into the pathophysiology underlying depression has aided in our understanding of antidepressant response and efficacy. Similarly, knowledge regarding sex differences in depression is crucial to understanding differential antidepressant efficacy. Therefore, we will begin this review by discussing sex differences in depression in both humans and animal models.

Humans

Women have roughly twice the lifetime rates of depression as men [12, 13]. According to the latest NHANES survey data for 2013–2016 amongst adults 20 and over, women were roughly twice as likely to be suffering from depression as men in a given 2-week period (10.4% for women, 5.5% for men), with an overall rate of 8.1% [1]. However, this disparity is not an absolute throughout the lifespan. Depressive disorders through childhood have a relatively low prevalence estimated at <5% overall, and are reported at similar rates by sex, or at even greater rates in males than females (e.g. [14]). Beginning with puberty and on into young adulthood, incidence of depressive disorders rise sharply, with a greater increase in females compared to males. The greatest predictor of the disparity is pubertal development, specifically Tanner Stage III [15]. By age 13–15, females begin to suffer from dramatically higher rates of depression than males [16]. In a 10-year longitudinal study of 1037 children in New Zealand followed from ages 11 to 21, the lifetime prevalence of at least one major depressive episode meeting full diagnostic criteria increased from 1.8% to 20.7% in males and from 0.3% to 42.6% in females [17].

Patients who are depressed do not comprise a homogenous population of symptomatology. According to the latest Diagnostic and Statistical Manual of Mental Disorders (DSM-5), a diagnosis of major depressive disorder (MDD) requires at least 5 out of 9 possible symptoms present during the same 2-week period, and must include either (1) anhedonia or (2) depressed mood, or both. Other criterion symptoms include (3) insomnia or hypersomnia, (4) changes in appetite or weight, (5) changes in psychomotor status, (6) fatigue or loss of energy, (7) worthlessness or guilt, (8) diminished concentration or indecisiveness, or (9) suicidal thoughts or behaviors. Using these criteria, it is plain to see that there are a myriad of ways—227 in total—to qualify for a diagnosis of depression by checking combinations of anywhere from 5 to 9 symptoms. Furthermore, several of the criteria can be met in opposite ways, greatly increasing the range of clinical presentations to 14,528 possible combinations of criteria. It is thus possible for two patients diagnosed with MDD to share no common symptoms [18]. Zimmerman and colleagues examined 1566 patients who met MDD criteria, and observed 170 of these 227 possible combinations. The top 9 combinations only composed about 40% of the study population. This highlights potential problems that conflating diagnostic heterogeneity, either in a human study or an animal one, can yield for attempts to unravel pathophysiological and etiological mechanisms, as well as the search for biomarkers or broadly effective treatments for the entire depressed population.

Given this heterogeneity in presentation across the total population of patients that present with MDD, it is not surprising that the manner in which women suffer from depression tends to differ from men, in addition to the greater overall rate at which women suffer. One of the largest and longest concerted efforts to evaluate depression treatment to date provided important data on disease presentation. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D trial) enrolled over 4000 outpatients seeking depression treatment at 41 primary and specialty care sites to compare first-line and subsequent switches and add-on therapies towards an endpoint of remission [19]. At baseline, women enrolled in the study reported greater depression severity, greater rates of hyperphagia and weight gain, hypersomnia, interpersonal sensitivity, and other somatic symptoms like gastrointestinal disturbances as well as co-morbid somatoform disorder. Women were also more likely to report anxiety-related disorders, bulimia, and prior suicide attempts, but less current suicidal ideation [20]. Men, on the other hand, were more likely to report comorbid substance abuse. Both sexes were equally likely to report irritability. Although women make two-time to three-time more suicide attempts than men [2], they are less likely to be lethal. The greatest disparity in suicide attempt rates is driven by adolescent age females [21, 22].

Rumination, chronic negative circumstances or strain, and a low sense of mastery have each been found to be more common in women [23]. These individual components contribute to greater depressive symptoms in women, and depressive symptoms feed into greater rumination and less mastery over time, setting the stage for positive feedback and recurrence in women in particular [24]. Men have been found to be more likely to forget past episodes of depression [25], likely reflecting a lower likelihood of a ruminative coping strategy compared to women. Subclinical depression symptoms are also more common in women than in men [26]. Finally, heritability of MDD in women is greater than in men, suggesting a higher genetic vulnerability [27].

It is important to note that differences in disease presentation are not absolute, which might suggest an altogether separate etiology, but that these trends demonstrate key signs and symptoms that are more prevalent in women than in men. Understanding these differences as well as commonalities may shed light into the fundamental mechanisms which translate into altered mood and behavior in both sexes, as well as inform tailored diagnostic and treatment intervention strategies.

Many of these observations on variation in clinical presentation have recurred in the literature [28], resulting in efforts to categorize subtypes. For example, Silverstein described “somatic depression” associated with greater fatigue and disturbances in appetite and sleep, with resulting disparity in the rate of this somatic depression (females 15.2% lifetime vs. males 7.0%), but not in rates of depression that did not involve all three of these somatic categories (female 6.9% vs. male 6.0%) [29]. Atypical depression characterized by intact mood reactivity, as well as prevalence of hypersomnia and weight gain is associated with female gender, younger age of onset, greater severity and disability, and higher number of suicide attempts, but is no longer delineated in the DSM [30, 31].

Risk factors for depression include a combination of genetic and environmental components, such as family history of depression and exposure to stress, with interactions between traits and experience increasing likelihood of onset of a depressive episode. A study in randomly selected American adolescents suggested that greater early adolescent challenges contributed to differences in affect in high-school aged teenagers [32], findings which highlight a role for the so-called stress-diathesis model of depressive behavior in which exposure to stress interacts with with existing psychobiological traits. Classically in this context, stressful life events interact with a negative inferential cognitive style to generate hopelessness and predispose towards depression [33]. In a prospective study of adolescents, negative inferential styles at baseline interacted with negative events over a 5-month follow-up period to predict depression and anhedonia specifically in females [34]. The stress-diathesis model has been developed further beyond the interaction of stress and cognitive style to stress interactions with the biological framework in a genetic epigenetic context [35, 36]. A growing body of evidence supports roles of both prenatal and early life stress imparting vulnerability for mood-related symptoms in adulthood, however, caution should be taken when interpreting human data based on self-report and recollection.

No discussion on sex differences is complete without mention of the effects of the reproductive cycle on depressive disorders. As mentioned above, the sex-based disparity in depression prevalence emerges with puberty. For many women, severity of symptoms of existing depression increases in the premenstrual phase [37]. In data from the STAR*D trial, 66% of women reported premenstrual exacerbations in their symptoms [38]. Of note, this was associated with longer depressive episodes and shorter latency to relapse. The increased incidence of new onset depressive symptoms in women appears to be reduced dramatically with menopause [39]. Women with established history of depression retained an eight-fold greater risk of depressive symptoms after menopause and a 4-6-fold increased risk in relapse [39]. Regarding the effect of pregnancy and birth, DSM-5 includes a peripartum-onset specifier for a depressive episode in the third trimester to up to 4 weeks after childbirth [40]. In one study of women with postpartum depression identified by Edinburgh Postnatal Depression Scale, ~40% began post-partum, ~33% began during pregnancy, and ~26% had depressive symptoms before becoming pregnant [41]. A number of studies have also examined the interaction between depression and menstrual disturbances further linking sex hormones and depression [42, 43].

A myriad of specific effects of sex hormones will be discussed in further detail below. Furthermore there are a wide range of developmental effects to consider regarding biologically determined sex differences. These include effects of X and Y chromosomal genes, organizational effects of gonadal steroid exposure during development, activational effects of reproductive hormone exposure throughout the lifespan, as well as the effects of parental stress prior to conception that have differential effects in males and females on depression-related behaviors and stress regulation (for an excellent review, see [11]). No single manuscript is likely to be exhaustive in this regard, although we touch briefly on those most pertinent to our examination of current efforts to understand sex differences in the key neurobiological circuitry underlying the patho-etiology of depression and antidepressant treatment.

Animal models

Depression is a multifaceted, heterogeneous disease with limited knowledge regarding predisposing factors. Animal research is an essential method for improving our understanding of these factors and underlying neuronal substrates. Depression cannot be fully recapitulated in an animal model—recurrent thoughts of death or suicide or excessive thoughts of guilt are clearly not measurable in animals. Instead, focusing on and modeling one core sign or symptom of this disorder has the benefit of simplifying a complex disorder like depression into behaviors that are easier to measure and are likely to involve fewer genes [44, 45].

The ideal animal model would have identical symptoms (face validity), causal factors (construct validity), and response to treatments (predictive validity) to the human disease it is attempting to replicate. Strategies that have been used to model depression range from environmental to genetic manipulation. We will cover many, but by no means all, of these models and discuss sex differences observed as well as their roles in depression.

Stress-based models

Depressed patients report more negative or stressful life events than non-depressed individuals, with exposure to stress one of the most prominent environmental factors associated with onset of a depressive episode [35]. Women, to a greater degree, report interpersonal stressors as a contributing factor. This is particularly greater in adolescents and young adults, when the disparity in depression is greatest. The impact of stressors is dependent upon characteristics of the stressor itself (e.g. severity, chronicity, predictability), as well as an individual’s history and ability to cope with stress [46, 47]. This sensitivity to stress is greater in females, who show an increase in the magnitude of the stress response and longer recovery time as compared to males. There are clear sex differences in hypothalamic–pituitary–adrenal (HPA) axis regulation in rodents, and the connection between dysregulation of stress response and development of depression has been reviewed elsewhere [6, 48]. In this section, we will discuss some of the more widely used paradigms including observed sex differences.

Chronic stress

While stressors encountered on a daily basis are often acute in nature, it is often the case that individuals endure chronic stressors. These can be related to psychosocial issues, financial impositions, health, discrimination/stigma, or a combination of different factors. The stressors can be inconsistent, unpredictable, ambiguous, and uncontrollable. They may differ in severity or quality over time, making it difficult to establish adequate coping methods, or limiting the ability to take preparatory steps. Chronic stress paradigms used in animal models have sought to recapitulate this by exposing animals to stressors over the course of many days. For example, chronic unpredictable stress (CUS) (also referred to as chronic mild stress) exposes animals to different stressors daily over the course of 2–3 weeks [49]. While the exact stressor selection and order of rotation varies by investigator, some examples include forced swim, wet bedding, restraint, strobe light, white noise exposure, and cage tilt. Changing the stressor daily prevents animals from anticipating or adapting to stress exposure. As a result, male animals show anhedonic-like behaviors as measured by the sucrose preference test (SPT) and social interaction test (SIT). However, the use of sucrose preference to model anhedonia has been difficult to recapitulate in females [50, 51]. Additionally, sucrose palatability can be modulated by the estrous cycle [52].

Learned helplessness

One of the most widely used models of depression has traditionally been that of ‘learned helplessness,’ inspired by early observations in dogs thought to represent despair or hopelessness [53]. Animals are exposed to either escapable shock, yoked inescapable shock (in which shock exposure is dictated by the actions of a fellow “yoked” rodent), or no shock exposure. They are subsequently tested in a behavioral task, most often one that assesses escape performance. In this paradigm, the inescapably shocked animals subsequently show frequent failures to escape from shock, whereas rats or mice that had previously been exposed to either escapable shock or no shock do not exhibit such deficits [54]. Studies examining sex-dependent effects of this test found that while male animals exposed to inescapable shock showed deficits in escape behaviors, female behavior remained unchanged [55]. Additionally, inescapable shock alters exploratory and anxiety-like behaviors, which appeared to be longer lasting in males than females [56].

Social stress

The main source of stress in humans is social in nature and is thought to contribute to the development and expression of mood disorders [57]. Animal models that involve a social context may be more appropriate as they represent situations that may be presented to humans for which the response is evolutionarily conserved. This is salient in the context of sex differences as interpersonal stress, in particular, drives the increased exposure of females to stress and has been shown to partially mediate the increased prevalence of depression in females after puberty, while women in general are more likely to express affiliative behavior and seek support when faced with stress as a coping style [58, 59]. In animals, the most widely used social stressors are defeat, isolation, and crowding. These types of stressors are often integrated into CUS described above.

Social defeat

The chronic social defeat stress (CSDS or SDS) paradigm is based on the establishment of a territory by a male rodent and its defense against unfamiliar male intruders [60, 61]. In this paradigm, a male rodent (the intruder) is placed into the home cage of another male (the resident). The intruder is physically confronted and defeated by the resident. The resident is usually of a more aggressive and larger strain than the intruder. In rats, the resident is usually paired with a female while the intruder is individually housed. Conversely, with mice, the resident is kept isolated whereas the intruder is group-housed. The resident has also had previous experience of victory in such encounters. There are many variations of this model that involve different durations of exposure or include additional threats of attack. The greatest limitation of this model is that it is almost exclusively used in males as it is difficult to obtain strong dominance relationships or aggressive encounters in females. However, a recent study employed designer receptors exclusively activated by designer drugs (DREADDs) to promote male aggression toward females [62]. Using these males, Takahashi and colleagues were able to induce social defeat in females and begin to examine some of the underlying neurobiological changes [62]. Another female model exploits maternal aggression, in which a non-pregnant female is attacked by a lactating female [63].

Defeated animals show a variety of behavior changes following defeat. Regarding social behavior, defeated males show a decrease in social interaction with novel conspecific males [64, 65]. They also show a decrease in mating and aggressive behaviors [66]. These observations are thought to reflect social withdrawal, discussed above. Social defeat also alters many nonsocial behaviors including decreased locomotor activity, reduction in food and water intake, anxiety-related behaviors, and drug seeking behavior, as well as reduced brain-reward behaviors seen during ICSS [67,68,69,70,71]. Increased immobility in the FST has also been observed, while there have been mixed results detailing effects on sucrose preference [71,72,73,74].

Housing conditions

Isolation has been associated with numerous physical and mental health problems, and several studies have indicated that loneliness may play a role in the development and persistence of depression [75,76,77]. This can be studied in animal models by single housing animals for a prolonged period of time (on the order of weeks). Prolonged isolation leads to increased immobility in the FST, changes in reward-related behaviors as measured by decreased sucrose preference, changes in sexual behavior, and drug-associated behaviors, as well as increased anxiety-like behaviors [78,79,80,81]. Studies using male and female animals have found that both sexes show increased anxiety-related behavior and decreased sucrose preference in response to social isolation, though the effects appear to be more prominent in males [82, 83].

To introduce unpredictability to the housing condition, researchers have utilized a chronic social instability paradigm where isolation and crowding conditions are alternated [84,85,86]. Additionally, animals are rotated among social groups during those crowding phases. This has been found to lead to neuroendocrine changes, anhedonia, and increased anxiety-like behavior [84,85,86]. Many of these traits are inherited by offspring of these animals implicating the involvement in stress-induced epigenetic changes [87]. The advantage of this paradigm is that it has been found to be effective in both males and females, and is often incorporated into CUS approaches [84, 88].

Maternal deprivation

A specific form of social isolation is focused on isolation in young animals as evidence indicates that early life traumatic experiences are associated with psychopathology including depression later in life [89]. Maternal deprivation (MD) models early life stress through daily separation of new born pups from their mothers during the first few weeks of life. Behavior and neurobiological changes are then assessed in adolescent or adult animals. This early life isolation results in persistent neurobiological changes, and depression and anxiety-related behaviors can be observed in adulthood [90,91,92,93,94]. Moreover, this depression-like behavior can be transmitted across several generations reflecting epigenetic vulnerability to stress [95].

Adolescent mice exposed to maternal separation show sex-dependent alterations in behavior in response to the learned helplessness paradigm, with males demonstrating loss of controllability in an escapable shock condition. In contrast, females demonstrated motivational impairment in a no-shock condition. This effect, however, was absent in adulthood as females no longer displayed helpless behavior [96].

Although maternal deprivation induces the same endocrine changes in males and females, such as increased corticosterone levels, sexual dimorphisms have been observed in the neuronal and behavioral changes induced by maternal deprivation [97]. Several studies in adult rats have shown that separated male rats exhibit a higher immobility as measured by FST [98,99,100,101], as well as anhedonia [101, 102]. Studies using adult females have found that maternal separation has no effect on depression-related behaviors as measured by FST or SPT [102, 103].

Surgical/genetic models

Olfactory bulbectomy

Surgical removal of the olfactory bulbs (OBX) in rodents leads to a number of behavioral, cognitive, and neurochemical changes, which are normalized with antidepressant treatment [104]. While there appear to be some differences between the rat and mouse models, OBX has been a useful model in the study of antidepressant efficacy [104]. OBX increases the immobility time in the FST and TST in both rats and mice [105,106,107,108], and causes hyper activity and anhedonia as measured by the SPT in mice [109,110,111,112].

There have been very few studies involving female subjects in the OBX model and even fewer dedicated to comparing sex differences. However, male and female OBX animals that have also been gonadectomized/ovariectomized exhibit higher activity levels than intact OBX and control animals. The effect of gonadectomy was more robust in males. Female OBX rats exhibited lower sucrose preference levels than male OBX rats. This difference was true for both intact and gonadectomized/ovariectomized rats [113]. These results suggest that hormones may have activational effects on activity changes induced by OBX and organizational effects in the marked anhedonia exhibited by female OBX mice.

Genetic selection/selective breeding

There is a genetic component associated with depression and a number of studies have sought to determine these genetic determinants, with dozens of genetically modified mouse strains associated with depressed phenotypes [114]. Many studies have used transgenic or knock-out/knock-in mice as a targeted approach to study specific genes that might underlie genetic predisposition to depression. An alternative approach to studying the genetic component of depression is through selectively breeding for a specific phenotype. There are many examples of this approach, but one example is the Flinders sensitive line rats. These rats have been created by selectively breeding Sprague–Dawley rats for their hyper-sensitivity to cholinergic agonists; a characteristic that has been observed in depressed humans, as well [115]. These rats display depression-like behavior as shown by increased immobility in the FST, and reversal of depressive-like behavior has been observed in response to treatment with a variety of antidepressants [115]. Females do not exhibit enhanced immobility, but instead show decreased latency to become immobile, in comparison to Sprague–Dawley controls [116].

Additional models/considerations

Common cognitive deficits in depression include impairment of concentration and selective attention towards or processing of negative stimuli. It is believed that this negative bias serves to increase salience of depressive elements, thus prolonging and exacerbating depressive episodes [117], and may partly explain or interact with rumination, which has been shown to be more prevalent in women [118]. Despite the prevalence of cognitive dysfunction in depression, many studies do not incorporate any cognitive function assessment into their animal models [119, 120]. Even fewer studies have examined potential sex differences, although it does appear that females may exhibit more negative bias at baseline [121, 122] complementing evidence for increased negative recall bias in women [123].

In addition to cognitive dysfunction, there are many homeostatic changes associated with depression, such as altered sleep and circadian rhythms, as well as changes in feeding and metabolism. These somatic symptoms are less frequently incorporated into animals models, and more frequently observed in depressed women [20]. Incorporating these deficits into animal models would be a clear and necessary step to expand upon current models in order to gain better insight in the pathophysiology underlying depression and antidepressant response. In parallel, better characterization and understanding of sex differences in expression of depression symptomatology are key to developing more valid models in which to further dissect underlying mechanisms and develop the next generation of antidepressant treatments.

Antidepressant treatment

Tricyclic antidepressants (TCAs)

The TCAs are derived from antihistaminic compounds, and the first, imipramine was introduced in 1957 [124]. They block reuptake of serotonin and norepinephrine, which is thought to contribute to their therapeutic action (Fig. 1). The degree of selectivity of inhibition of serotonin versus norepinephrine transporters differs among the family of TCAs with desipramine and maprotiline being most potent at the norepinephrine transporter and clomipramine being most potent at the serotonin transporter [125]. In addition to their actions at serotonin and norepinephrine transporters, TCAs also exhibit anticholinergic, antiadrenergic, and antihistaminergic activity owing to their reported side effects [125]. Additionally, at high levels, they can inhibit sodium channels or cause serotonin syndrome making them potentially lethal [126] and increasing the risk of suicide by overdose [127]. This is particularly relevant for female patients as reports have shown increased suicide attempts by antidepressant overdose [128].

Fig. 1
figure 1

Antidepressant mechanism of action. Schematic of the proposed mechanisms of action for antidepressants at the synapse level. Serotonin and norepinephrine are represented together for simplicity. Ketamine has two proposed mechanisms, which are both shown. One is block of extrasynaptic NMDA receptors, leading to repression of eEF2 kinase and dephosphorylation of eEF2, thereby disinhibiting translation of BDNF. The second mechanism is block of NMDA receptors on inhibitory neurons, leading to increased excitation, BDNF release, and activation of mTOR through a TrkB-dependent signaling cascade. GABA-NAMs are negative allosteric modulators of alpha5 subunit-containing GABA receptors and thereby increase excitation and promote activity-dependent synaptic strengthening. SSRIs and TCAs act by blocking serotonin reuptake, increasing the synaptic concentration of serotonin, and thus increasing activation of serotonin receptors, thereby promoting activation of CaMK and PKA signaling and leading to increased synaptic strength via transcription and post translational modifications

The therapeutic use of TCAs for treating depression was not only a significant advancement in the treatment of this disorder, but it revealed the potential to understand the neurological basis of depression. Their use has fallen out of favor due to better safety profile of newer medications, however they are still employed in treatment-refractory cases. There is also evidence that they may be more effective in adult men and in older women than SSRIs [129, 130]. From the successful treatment of depression with TCAs and studies on serotonergic dysfunction emerged the serotonin hypothesis of depression.

SSRIs

The serotonin hypothesis of depression has driven antidepressant development for the better part of six decades, ultimately leading to SSRIs and later serotonin/norepinephrine reuptake inhibitors (SNRIs) (Fig. 1). Inhibition of serotonin reuptake leads to increased activation of the 14 subtypes of serotonin receptor, each with a unique pattern of expression and activation. An estimated 15.9% of adult women in the US take antidepressants, compared to 7.7% of men [131], reflecting the higher rate of depression overall. Six of the top ten prescribed antidepressants in the US are serotonin-based antidepressants.

Alongside explosive drug development leading to dozens of prescribed serotonin-based antidepressants, extensive and thorough basic research continued into the role of monoamine depletion in depressive symptomatology and has not fully satisfied the predictions of the hypothesis, with inconsistent and inconclusive evidence leaving considerable debate as its validity to explain depression as a single disease [132]. Mirroring the incompleteness of the underlying serotonin hypothesis is an incompleteness in patient response to this class of second-generation antidepressant medications. The response rate to this first line in antidepressant treatment is around 50%, with 70% of patient failing to fully remit after a 12-week course [133, 134]. These limitations have guided the most recent quest for the next generation of antidepressants, discussed in the next section. However, this too yields resolution in which to examine sex differences.

Fast acting antidepressants

Drugs targeting the monoamine system have been the standard of care over the past 50 years, however there are some significant limitations associated with this class of drugs. Only a fraction of patients experience relief of depression symptoms upon antidepressant administration, and among that fraction of responders (~30%), it can take weeks to months before patients feel those effects. Additionally, there are various side effects associated with these drugs that decrease compliance with drug regimens. However, there was not a significant shift in focus on novel antidepressants until the provocative finding that acute administration of ketamine could produce rapid (within hours) antidepressant effects [135]. This effect has been found to last for up to 2 weeks after a single administration [136, 137]. Furthermore, ketamine has also been found to be efficacious in treatment resistant patients, who failed to respond to monoamine-based antidepressants [136, 137]. This revolutionized the field of antidepressant research and pivoted the focus of the field to novel antidepressant mechanisms.

While the abuse potential for ketamine has restricted access to this treatment to medically staffed environments, these findings have aided much of the preclinical work on depression and antidepressant mechanisms. The mechanism of action for ketamine is a highly debated topic. Ketamine is an ionotropic glutamatergic N-methyl-d-aspartate receptor (NMDAR) antagonist and work has suggested that NMDAR blockade influences downstream intracellular signaling pathways (Fig. 1). In one model, this is shown to occur through inhibition of eukaryotic elongation factor 2 (eEF2) kinase ultimately leading to increased BDNF signaling, AMPAR insertion, and increased synaptic strength [138]. Another model suggests that NMDA receptor blockade increase mammalian target of rapamycin (mTOR) signaling, which increases synaptogenesis [139]. This was also shown to be the mechanism of action for another rapid acting antidepressant, scopolamine, a non-selective muscarinic receptor antagonist [140]. Recent work suggests that it may not be the action of ketamine on NMDAR that is important for its antidepressant effects, but rather the action of (2R, 6R)-hydroxynorketamine (HNK), the major ketamine metabolite found in plasma and brain [141]. Administration of HNK was shown to elicit rapid antidepressant effects and synaptic potentiation, similar to ketamine, however HNK action does not appear to be through inhibition of NMDARs [141].

Despite the differences between these proposed mechanisms, they all seem to converge on the idea that ketamine exerts its antidepressant effects through an increase in activity in neural circuits associated with mood regulation. This has been key in the development of novel rapidly acting antidepressants that might have fewer side effects than ketamine. One approach to minimize potential side effects has been through targeting receptor subtypes with region-specific expression. α5-containing γ-aminobutyric acid type-A (GABAARs) are most strongly expressed in PFC and hippocampus [142]. The negative allosteric modulators of this subtype (L-655,708 and MRK-016) elicit rapid antidepressant action and increased excitatory synaptic strength, similar to ketamine [143]. However, in contrast to ketamine, these drugs do not have psychotomimetic or sedative side effects [144].

Alternative neuromodulatory treatments

Despite the array of pharmaceuticals available for depression, a significant proportion of depressed individuals is treatment–resistant as discussed above. Alternative strategies have been sought to alleviate depression symptoms in these patients. Despite differences in methodology, these techniques are all based upon the idea that depression results from altered neuronal circuits and stimulating those circuits can at least partially reverse this altered activity and produce antidepressant effects. We will only discuss a few such approaches here.

Electroconvulsive therapy (ECT) is an intervention whereby electricity is delivered to the brain to induce generalized seizures. It is effective in treating depression with a roughly 60% remission rate and is even effective in treatment resistant populations [145, 146]. However, concerns including cognitive side effects, in particular temporary memory impairment, have limited the penetration of ECT into broad clinical practice. The induction of seizure activity alters the function of neuronal circuits, which is thought to underlie the beneficial effects of ECT in the treatment of depression. More recent strategies have sought to emulate ECT with stimulation better targeted to brain regions known to be involved in mood regulation, thereby reducing the negative side effects associated with altering whole brain activity.

Transcranial magnetic stimulation (TMS) uses alternating magnetic fields to induce electric currents in cortical neurons [147]. Multiple studies examining TMS targeting the prefrontal cortex have revealed positive antidepressant effects [148, 149]. Deep brain stimulation (DBS) involves implanting electrodes to stimulate specific brain regions [150]. These regions include subcallosal cingulate, ventral anterior internal capsule, and ventral striatum, which are all involved in regulation of mood and antidepressant response [150, 151]. Existing evidence is limited, a few studies using these neuromodulatory approaches have found equivalent efficacy in both males and females [152, 153]. However, response to TMS may be modulated by hormone levels as response in premenopausal women was correlated with estradiol/progesterone ratio while postmenopausal women were the least responsive to treatment [154].

Sex differences in response to antidepressants

There is not a clear consensus on whether there are sex differences in pharmacotherapy antidepressant efficacy, likely due to variability in methodology, as well as a dearth of specific investigation in this regard. A number of studies have shown that men experience a better therapeutic response to TCAs than women [130, 155, 156], whereas there is evidence that women respond better to SSRI treatment than men [129, 130, 157, 158]. This effectiveness was further exemplified in a study of patients given free choice of clinical interventions followed for up to 12 weeks: women were about 1.5× as likely to achieve remission as men, with most patients in the study taking either SSRIs or SNRIs [159]. In conjunction with findings that older women do not enjoy this superiority and that hormone replacement therapy has been found to eliminate a poor SSRI response [160], this suggests a role of female sex hormones in therapeutic response to SSRIs in particular. Estrogen has subsequently been shown to influence serotonin synthesis, as well as serotonin receptor binding and activity [161].

An alternative explanation is that the subtype of depression determines the antidepressant response. As discussed above, the presentation of symptoms differs between men and women, and women tend to show more somatic symptoms associated with atypical depression, which has been found to respond preferentially to SSRIs [162].

Because of the high incidence of poor or delayed response to antidepressant therapy, there has been extensive investigation into potential augmentation of traditional antidepressant pharmacotherapy. These include other medications with other indications such as lithium, triiodothyronine, and antipsychotics, as well as non-pharmacological interventions like bright light therapy. Such augmentation by estradiol treatment on SSRI effectiveness was mentioned above, but sex differences exist for other augmentation strategies as well. For instance, triiodothyronine (T3) has been found to accelerate response to TCA, and this effect is more pronounced in female patients [163]. Another study found that supplementing folic acid enhanced the effect of fluoxetine in women likely due to differences in resulting plasma folate concentrations [164]. Additional work into current and novel augmentation strategies may be useful in identifying approaches to optimize treatment in both males and females.

Currently, fast acting antidepressants are not widely used in humans; however, a few studies in rodents have suggested that females may be more sensitive to ketamine. Stress naïve females show an antidepressant response to a lower dose of ketamine as compared to males as measured by behavior in the FST [165, 166]. In mice exposed to chronic stress, the long lasting antidepressant effects of ketamine (7 days post injection) are more apparent in males [166]. This increased sensitivity is absent in ovariectomized females and is only restored in animals receiving estrogen and progesterone [165], suggesting both hormones are required for the increased sensitivity observed in females. This was also found to increase hedonic responses to low-dose ketamine as measured by increased sucrose preference [167]. In contrast, gonadectomy and testosterone had no effect on male responses, however, progesterone administration rendered males sensitive to low-dose ketamine [167]. This implicates an activational role for estrogen and progesterone in mediating sensitivity to ketamine.

It should be noted that many studies using the forced swim test (FST) for screening antidepressants or assessing depression-like phenotypes have relied on male subjects and assumed that the behavioral response would be the same in females. However, conflicting results have emerged from studies that have considered sex differences (see [168]). Hormone levels do not appear to play a major role in influencing behavior in normally cycling rodents, although some studies have found differences across the estrous cycle. In ovariectomized rats, hormone levels play a role in modulating FST behavior, with estrogen withdrawal leading to increased immobility [169], and estrogen administration leading to antidepressant-like effects [170].

Possible mechanisms underlying sex differences

If we consider depression as a result of altered neuronal circuitry, then differences in antidepressant response can emanate from sex differences in neuronal circuitry. Therefore, it is important to discuss sex differences in the neurobiology underlying depression as well antidepressant response. Here, we will focus on sex differences in synaptic transmission and include how hormones may contribute to these differences as the organizational and activational effects of hormones in generating these differences has been reviewed extensively elsewhere [8, 9, 11].

Sex differences in synaptic transmission

Dysfunction of a number of brain regions likely underlies the number of diverse symptoms associated with depression. Indeed, brain imaging, as well as postmortem anatomical studies have demonstrated changes in several brain areas including prefrontal cortex, cingulate cortex, hippocampus, striatum, amygdala, and thalamus [171,172,173,175,174]. Probing synaptic transmission in these areas has contributed to our understanding of mechanisms underlying antidepressant efficacy. Similarly, understanding sex differences in this synaptic transmission sheds light on the potential mechanisms underlying differences in antidepressant efficacy. Using the animal models described above has allowed researchers to delve deeper into the neuronal circuitry underlying depression and antidepressant response.

A number of hypotheses have been put forth in an attempt to explain the maladaptive changes found in depression. For instance, the monoamine hypothesis postulates that depression is caused by a deficiency in monoamines, yet incompletely explains depressive etiology as discussed above.

The hypotheses associated with depression appear to converge on a common factor that is a growing area of focus in understanding the pathophysiology underlying depression and antidepressant response: changes in synaptic strength. Particular focus has been dedicated to excitatory synaptic strength as there is increasing evidence to suggest that chronic stress, which induces depressive signs, weakens excitatory synaptic structure and function in multiple regions of the brain associated with depression. Conversely, antidepressants promote excitatory synaptic transmission in these same regions (Fig. 1). Differences in these synaptic changes may be associated with sex-differences in depression and antidepressant response. Though there are number of brain regions involved, reviewed in greater breadth elsewhere (e.g. [175]), we will limit our discussion to the hippocampus, prefrontal cortex, and nucleus accumbens (NAc) for the sake of brevity.

Hippocampus

The hippocampus plays critical roles in both mood regulation and higher cognitive functions. It is particularly stress sensitive due to high glucocorticoid receptor expression and its crucial role in HPA-axis feedback [176, 177]. Thus, the hippocampus is at a key intersection of neuronal mechanisms underlying depression. Patients with MDD exhibit decreased hippocampal volume, dendritic density, and spine number [178,179,180,181,181]. The hippocampus sends excitatory projections to regions of the brain important for reward processing including the NAc [182,183,184]. This is thought to be important in modulating the activity of NAc and providing contextual information associated with reward [185, 186]. Regarding stress, acute elevations of glucocorticoid levels potentiate hippocampal neurons [187], which inhibits HPA activity but also may enhance hippocampal function. However, chronically elevated glucocorticoids, as seen in chronic stress, may damage hippocampal neurons and impairs their synaptic protein expression and function [188] and consequently weaken excitatory synaptic strength [189].

In male rodents, chronic stress results in a retraction of apical dendrites of pyramidal cells in hippocampal area CA3 [190,191,192,193,194], decreased spine density on pyramidal cells in hippocampal area CA1 [195], a reduction in neurogenesis in the dentate gyrus [196,197,198]. Several classes of antidepressants reverse the stress-induced reduction in dendritic branching [199, 200] and neurogenesis [201, 202] of hippocampal neurons.

The deleterious effects of stress on the hippocampus are also reflected in alterations hippocampal function. Electrophysiological studies have revealed weakening of excitatory synapses [188, 189] or deficits in synaptic plasticity [203,204,205] in rodent models of depression and coincide with decreased expression of GluA1 and mGluR5 [188, 189, 206]. Several classes of antidepressants have been shown to reverse these changes [143, 189, 199, 200].

The same chronic stressor that produces a reduction in dendritic complexity in males does not affect apical or basilar dendritic length of neurons in CA3 pyramidal cells of females [191], and results in altered spine morphology but no change in spine density on CA1 pyramidal cells [195]. The hippocampus expresses high levels of the receptors for estrogen and progesterone, and these hormones have a profound effect on hippocampal structure and function. Administration of exogenous estrogen or progesterone increases spine density [207, 208], and the natural variation of estrogen levels induces changes in spine density and shape in CA1 [209, 210]. These estrogen-induced spines were found to comprise increased synaptic transmission of single presynaptic inputs onto multiple postsynaptic cells, suggesting the possibility that estrogen promotes the formation of new synapses [211]. It is possible that these effects occur at least in part as a result of estrogen influence on neurotrophin signaling. Neurotrophins are involved in modulation of dendritic morphology and spine density in many brain regions including the hippocampus [212]. Inhibiting downstream neurotrophin signaling blocks the beneficial effect of estrogen on hippocampal function [213]. Conversely, BDNF administration increases serotonin receptor expression [214], as well as 5-HIAA and serotonin turnover [215].

Estrogen itself has numerous modulatory effects on synaptic transmission. It enhances presynaptic function indicated by increased neurotransmitter release [216]. Additionally, estrogen and progesterone modulate the activity of a number of different neurotransmitter systems through a number of mechanisms including interactions with receptors, transporters, and enzymes involved in synthesis [217]. While too numerous to fully discuss these interactions here, some examples include upregulation of AMPA and NMDA receptor expression, decreased 5-HT1A receptor function, and altered SERT expression and activity [217, 218]. This is relevant for hippocampal function as excitatory synaptic strength in the hippocampus has been associated with depression and antidepressant action and serotonin signaling has been shown to be important in modulating the strength of excitatory synapses [132].

Prefrontal cortex (PFC)

The PFC is an important site at which cognitive evaluations associated with depression can influence affect and reward, such as the controllability of a stressor [219] or the pleasantness of a stimulus [220]. Patients with depression display decreased volume, spine density, and reduced activity in the PFC [221], and SSRI treatment restores normal activity levels [222,223,224]. These structural and functional changes are also reflected in many animal models of depression. Chronic stress results in atrophy of apical dendrites of pyramidal cells in the PFC and loss of dendritic spines [225,226,227,228], as well as downregulation of AMPA and NMDA receptor subunits and a decrease in synaptic excitation in layer V pyramidal cells in the PFC [229].

PFC neuron morphology appears to be sexually dimorphic at baseline. Unstressed females show decreased apical dendritic length and reduced apical branch number compared to unstressed males [230]. PFC morphology also appears to be modulated by estrogen levels as dendritic branching and spine density are decreased in ovariectomized rats and increased following estrogen treatment [231, 232].

Branch length and spine density increase in females in response to chronic stress, in contrast to males which showed a stress-induced decrease [233, 234]. This stress-induced increase in apical dendrite length appears to be dependent on estradiol as ovariectomy prevented this effect. This sexual dimorphism in response to stress seems to be developmentally regulated as no differences are observed between juvenile male and female animals exposed to stress [235]. This further supports a role for sex hormones in modulation of neuronal circuits associated with mood regulation and parallels clinical studies demonstrating the increase in prevalence of depression in females after the onset of puberty [236]. Another study comparing the effects of chronic stress on the PFC of males and females found that while chronic stress reduced glutamatergic transmission and AMPA and NMDA receptor expression in males, females remained unaffected [237]. This was due to a protective effect of estrogen, as blocking estrogen receptors rendered females susceptible to the effects of chronic stress [237]. Taken together, these studies highlight a possible neuroprotective effect for estrogens.

Estrogen levels have been correlated with activation of the prefrontal cortex and modulation of emotional processing and fear extinction [238,239,240]. Estrogen therapy increases frontal cortex activation and verbal recall, a process mediated by the prefrontal cortex, in peri-menopausal and post-menopausal women [241]. However, administration of a gonadotropin-releasing hormone agonist to pre- menopausal women decreases prefrontal activation and impairs verbal memory [242], revealing a nuanced role for estrogen action in the prefrontal cortex.

The modulation of structure and activity of PFC neurons may also be in part through modulation of neurotrophin signaling in this area. Similar to hippocampus, BDNF expression in the PFC is decreased in depressed patients, as well as animal models of depression [243,244,245], whereas antidepressants increase BDNF expression as well as TrkB activation [246, 247]. BDNF expression also seems to be modulated by estrogen as expression was found to be reduced in the PFC of female rats during proestrus [248].

Nucleus accumbens

A key brain region implicated in the etiology of depression is the NAc a critical area for regulating reward behavior. Patients with depression have decreased NAc volume, as well as decreased activity in response to reward [249, 250]. Decreased spine density and dendritic branching were observed in some animal models of depression [251,252,253], however many others have reported opposing findings [254, 255]. This may reflect differences in the models used and shed light on subtypes of depression. Accordingly, a recent study used neuroimaging biomarkers to identify subgroups found that increased NAc volume was only associated with a particular subgroup of depressed patients [256]. These human functional changes in NAc are reflected in rodent models in which decreased excitatory synaptic strength occurs in parallel to depression-like changes in behavior [257, 258]. Conversely, increased synaptic strength has been observed in other models [259]. This region incorporates excitatory information from a number of brain regions including the hippocampus and PFC, gating input from higher brain areas into subcortical reward networks. Further dissection of these inputs may provide keen insight into the neuronal circuitry underlying depression and reward regulation.

Sexual dimorphism is observed at NAc synapses. Spine density remains consistent through the rostral/caudal extent of the NAc in males, whereas a gradient exists in females with increased spine density in more caudal regions [260]. Sex differences are seen specifically at distal dendrites, with greater spine density, as well as a greater proportion of large spines in females [261]. Frequency of miniature excitatory post-synaptic currents (mEPSCs) is also greater in females suggesting, in conjunction with structural findings, that females have more excitatory synapses per cell in the NAc [262]. Sexual dimorphism observed in the NAc may not only be a reflection of sex differences specifically in this region but also of differences in hippocampal and PFC output.

The NAc receives dopaminergic input from the ventral tegmental area (VTA). The mesolimbic dopamine system is critical for reward processing, and rewarding stimuli is associated with dopamine release in regions like the NAc and PFC [263,264,265]. Sex differences in this system are thought to mediate sex differences in motivation. For instance, female rats self-administer low doses of cocaine at a faster rate as compared to males [266, 267]. Basal dopamine tone varies with the estrous cycle and parallels changes in estrogen levels with the nadir during diestrus [268]. This decreased dopamine tone is also observed in ovariectomized animals. In contrast, the basal dopamine levels in males are similar to that of females in proestrus or estrus and castration does not change these levels. Overall, basal dopamine tone is chronically higher in males than in females. Estrogen also modulates stimulated dopamine release in females [269, 270]. Taken together, this suggests that high basal dopamine levels result in downregulation of activity, such that when dopamine release is stimulated the relative increase is less in males than in females and the behavioral response is proportionally lesser, which could underlie sex differences in motivation. Yet another mechanism by which activity of the mesolimbic dopamine system differs between males and females is in dopamine receptor expression, which is greater in males as compared to females [271, 272]. Estrogen may also underlie this sexual dimorphism, as estrogen causes down-regulation of D2 receptor binding in female striatum [273].

The brain regions discussed here are profoundly interconnected, so sex differences in one area are likely to affect downstream regions. Much of the understanding we have regarding the changes in synaptic transmission in depression has been gleaned from studies in males. Further examination of synaptic transmission in females both in comparison to males at baseline, as well as in models of depression will provide key insight into understanding the pathophysiology underlying this disorder.

Sex differences in pharmacokinetics

In addition to differences in synaptic transmission, there are many other physiological factors that can lead to sex differences in antidepressant response. This has been reviewed extensively elsewhere [4, 7, 10], so we will only briefly review some of these differences. Differences in gastric environment, slower rate of gastric emptying, and longer colonic transit times observed in women can increase the rate of absorption of antidepressants [4, 7, 10]. In addition, women have a higher percentage of adipose tissue, which can prolong the half-life of lipophilic drugs, such as trazodone and bupropion [4, 7, 10]. Differences in metabolism or clearance of antidepressants can also contribute to higher concentrations of antidepressants in women due to differences in hepatic blood flow and cytochrome P450 enzymes [4, 7, 10]. Estrogen is also a substrate for some of the same cytochrome P450 isozymes as antidepressants, and the presence of both can shift the metabolism for both [4, 7, 10].

Sex differences in side effects and compliance

Sex differences have been reported in the side effects associated with antidepressant treatment. Women show decreased tolerability of TCAs and tend to report side effects, such as dizziness, nausea abnormal vision, constipation, and somnolence [130, 274, 275]. Men tend to report greater sexual dysfunction and urinary complaints [130, 275]. As to whether these side effects contribute to differences in compliance remains unclear. Non-compliance rates have been found to be higher for TCAs, though this was not found to be a sex-specific effect [276]. Another study found sex differences in compliance that varied by age but were not differentiated based on antidepressant type [277]. Further research into this topic as well as other sex-based differences in barriers to continuing care may yield further insight into ways to increase effective depression treatment programs.

Discussion

There are clearly many differences between males and females that impact depression and antidepressant efficacy. Despite the fact depression is more common in women, the vast majority of the basic research focus has been dedicated to studying males. While this narrowed focus has aided in attempting to simplify a complex disease and thus reducing variability due to baseline sex differences, it ignores a large patient population. Because the most widely used assays have been optimized in males, this provides a significant barrier to the inclusion of females in subsequent studies. As we have discussed above, sex differences have been reported in many of the factors associated with depression including precipitation and presentation of the disorder. Therefore, it may be neither pertinent to force males and females into the same model nor practical use the same behavioral outcomes to assess them. It may, instead, be more advantageous to model depression in males and females separately focusing on endophenotypes relevant to each sex. This approach has the potential to reveal new mechanisms and biomarkers associated with depression, as well as novel targets for antidepressant development. While unraveling sex differences in depression may seem to complicate our understanding of an already complicated disease, understanding the underlying neurobiology of sex differences may be a useful means to unraveling this debilitating disease in both men and women.