Illuminating the impact of stress: In vivo approaches to track stress-related neural adaptations

Abstract

Stressful experiences can affect both daily life and long-term health outcomes in a variety of ways. Acute challenges may be adaptive, promoting arousal and enhancing memory and cognitive function. Importantly, however, chronic stress dysregulates the body's physiological regulatory mechanisms consisting of complex hormone interactions throughout the peripheral and central nervous systems. This disrupted signaling consequently alters the balance of synapse formation, maturation and pruning, processes which regulate neural communication, plasticity, learning, cognitive flexibility and adaptive behaviors - hallmarks of a healthy, functional brain. The chronically stressed brain state, therefore, is one which may be uniquely vulnerable. To understand the development of this state, how it is sustained and how behavior and neural function are transiently or indelibly impacted by it, we can turn to a number of advanced approaches in animal models which offer unprecedented insights. This has been the aim of my recent work within the field and the goal of my new independent research program. To achieve this, I have employed methods to uncover how key brain circuits integrate information to support motivated behaviors, how stress impacts their ability to perform this process and how best to operationalize behavioral readouts. Here I present an overview of research contributions that I find most meaningful for advancing our understanding of the impact of stress and propose new avenues which will guide my own framework to address the salient outstanding questions within the field.

1. Visualizing stress-sensitive circuitry

Animal models provide a valuable resource to better understand the dynamic structural and functional changes associated with stress exposure in varying brain regions and pathways, many of which are conserved across species (Bale et al., 2019; Nestler and Hyman, 2010; Parekh et al., 2022; Scarpa et al., 2020). Indeed, we have learned a great deal about the pathophysiology of stress-related disorders such as depression through basic research and preclinical findings (Drevets et al., 2008; Treadway and Zald, 2011; Williams, 2016). The active process by which the body maintains physiological functions (homeostasis) through coordinated activity of peripheral and central systems is known as allostasis. This process becomes allostatic load in the face of repeated stressful experiences (Goldstein and McEwen, 2002). Allostatic load is in turn associated with remodeling of the structures that facilitate neuronal communication, interfering with cognitive and emotional processes (McEwen, 2007; Dias-Ferreira et al., 2009; Arnsten, 2009; Holmes and Wellman, 2009). Snapshots of these changes have been captured using a number of paradigms involving varying stressor intensity, duration and modality. Cross-sectional studies have examined the effects of stress exposure across the lifespan, establishing critical neurodevelopmental periods during which elevated glucocorticoid levels produce long-lasting adaptations, wiring the brain for future vulnerability (Lupien et al., 2009; Liston and Gan, 2011; Levis et al., 2022). These studies laid the foundation for our understanding of how stress differentially impacts cortical, mesolimbic, hippocampal, thalamic and brainstem structures and nuclei (Parekh et al., 2022). Focusing on circuits is valuable when studying the effects of stress because changes may not be uniform across regions or cells. For example, within the nucleus accumbens, a critical integrator of cortical and limbic information, chronic stress elicits both regressive and stimulatory effects based upon whether the principal cells, medium spiny neurons (MSNs), express dopamine receptor subtype 1 (D1R) or 2 (D2R) (Parekh et al., 2022). These two cell populations are associated with distinct behaviors due to their long-range connectivity, underscoring the need to parse the circuit-specific adaptations caused by stress (Hikida et al., 2010). Intriguingly, activity of D1R-expressing MSNs correlating with social interaction was predictive of future resilience to stress-related social deficits – an effect which was not observed in the D2R population (Muir et al., 2018). One significant contribution made by myself and my colleagues towards advancing the circuit neuroscience of stress has been to apply in vivo multiphoton imaging and fiber photometry techniques to allow longitudinal assessment of morphological and functional changes (Grienberger et al., 2022a; Simpson et al., 2024; Gergues et al., 2024). Through this approach, we are able to establish an expanded timeline of the effects of stress exposure within the same individuals, extending upon findings from cross-sectional studies. It also enables the ability to correlate real-time behavioral measures with neural activity monitoring to understand how neural populations or individual cells are modulated by behavioral events and task features in the unstressed and stressed conditions. Tremendous progress has been made in the arena of tool development for quantifying the output of neural and non-neural cells in behaving animals (Guo et al., 2023; Hjort et al., 2024; Steinmetz et al., 2021; Wu et al., 2022; Zong et al., 2022). Biosensors for the major neuromodulator, neuropeptide and other signaling molecules have been optimized over time to capture and characterize their actions in the brain and role in supporting complex behavior (Muir et al., 2024). Targeted optogenetic manipulations serve to test the causal relationship between circuit function and behavior (Emiliani et al., 2022). We took advantage of these exciting tools, to ask what chronic stress does to the cells of the dorsomedial prefrontal cortex (dmPFC), a region long established as critical for executive functioning, decision-making, valuation, goal-directed behavior and emotional regulation (Carlén, 2017; Dalley et al., 2004). In a study published in Science in 2019, my co-authors and I applied this imaging technique to visualize individual neurons of the dmPFC labeled with the genetically encoded calcium indicator, GCaMP6s (Moda-Sava et al., 2019). GCaMP molecules change conformation when bound to calcium, emitting a fluorescent signal whose intensity is detected using laser scanning microscopy. These signals serve as a proxy measure of neuronal activation (Lin and Schnitzer, 2016). Multiphoton laser light is optimal to penetrate deep into brain tissue with minimal scattering from autofluorescence thus allowing for high resolution imaging of cells and subcellular structures such as individual dendritic spines (Grienberger et al., 2022b; Streich et al., 2021). We gained optical access to the dmPFC, which lies deep within the medial fissure, through implantation of glass microprisms containing a reflective surface. These microprisms, sometimes attached to small glass coverslips, create a stable window into the frontal cortex, to which we can return for repeated monitoring of circuit and synaptic changes (Andermann et al., 2013). A normative amount of turnover, marked by spinogenesis and spine elimination occurs in circuits throughout the brain as synaptic contacts change through plastic mechanisms (De Roo et al., 2008; Makino and Malinow, 2011; Zhou et al., 2020). We observed that chronic stress, through repeated restraint or neuroendocrine dysregulation via corticosterone (CORT) administration, diminishes motivated escape behavior and causes spine elimination in the dmPFC of adult male mice. This was altogether not surprising, however, tracking the same spines after a single dose of the noncompetitive N-methyl-D-aspartate (NMDA) receptor antagonist, ketamine, we found that approximately 40% of lost spines were restored in the original locations and a moderate amount of de novo spine formation also occurred. Untreated CORT-exposed mice, conversely, failed to show these reversal effects when imaged 10- and 21 days post-exposure, highlighting the ability of chronic stress to produce long-lasting structural adaptations in the brain (Moda-Sava et al., 2019). Ketamine, a dissociative anesthetic agent, has been extensively explored for its therapeutic potential in mood disorders (Berman et al., 2000; Riggs and Gould, 2021; Krystal et al., 2024; Miller et al., 2024). What is particularly interesting from our study is that ketamine-induced spinogenesis appears to be targeted, with the majority of restored spines clustered on specific dendritic branches, suggesting that certain inputs to dmPFC may be particularly stress-sensitive and responsive to antidepressant treatment. From where these inputs originate remains an open question. Additional in vivo studies have added to our understanding of the effects of ketamine on dendritic spine structure and function in the PFC (Ali et al., 2020; Phoumthipphavong et al., 2016).

This optical approach further provided us an opportunity to assess the functional effects of stress and ketamine in the frontal cortex. We observed that dmPFC neurons display coordinated ensemble activity at rest which is significantly disrupted by chronic stress. Ketamine treatment normalized this ensemble activity. The longitudinal nature of our studies offered novel temporal insights into how these adaptations manifest and to our surprise, functional connectivity- and behavior-level effects of ketamine precede synaptic changes. The importance of spinogenesis was established through our use of a photoactivatable Rac1 construct to eliminate newly formed spines associated with ketamine treatment (Hayashi-Takagi et al., 2015). The precisely timed blockade of spine formation impairs ketamine's ability to restore motivated escape behavior. We therefore identified a process by which antidepressant-dose ketamine exerts a rapid inducing effect in normalizing circuit function and a slower, potentially more sustaining effect of enhancing plasticity to promote remission. Thus, while previous studies have largely relied on cross-sectional measurements to correlate dendritic alterations with stress-induced behavioral deficits, our longitudinal, in vivo approach provides a causal mechanism linking structure with circuit function and antidepressant response. Inspired by what we had learned about these prefrontal mechanisms, I turned my focus to address a set of fundamental outstanding questions - how do distinct frontocortical circuits normally enable motivated behaviors which are compromised in stress related conditions? What is the nature of the information coded by these pathways? Which pathways are indispensable for learned associations between stimuli, outcomes and costs to maintain optimal behavioral responses?

2. Focusing the lens on translationally-relevant behavioral outcome measures

Stress related disorders such as depression are marked by symptom heterogeneity, which is mirrored by heterogeneity in the circuits and networks involved (Drevets et al., 2008; Drysdale et al., 2017). An individual may suffer from melancholic depression, characterized by anhedonia while another may experience atypical depression with fatigue and somatic symptoms (Downar et al., 2014; Gold and Chrousos, 2002; Miller et al., 2024). The etiology of distinct symptom clusters bears further exploration using appropriate preclinical models. Stress is closely tied with the development of anhedonia (Stanton et al., 2019). The chronicity of the stress exposure, however, may vary, suggesting that additional factors are at play in the response profile. Importantly, anhedonia is multifaceted, characterized by cognitive, motivational, anticipatory and consummatory components (Der-Avakian and Markou, 2012; Treadway and Zald, 2011). Disruption of any or all of these can underlie reward processing and learning deficits. Disentangling this and identifying the overlapping or unique contributions of various neural circuits is crucial. In the motivational domain, the regulation of effort in goal-directed behavior is of particular interest. The ability to integrate information to drive high-utility behavior is critical for survival - individuals must continually update and weigh the value of rewards available in the environment against real and perceived effort costs and select appropriate actions. A deficit in this process, known as effort valuation, is characteristic of amotivation in anhedonia (Der-Avakian and Markou, 2012; Yang et al., 2014; Winstanley and Floresco, 2016; Salamone and Correa, 2024). The underlying neurobiological bases for stress-related changes in cell activity and motivational regulation are incompletely understood. Frontocortical regions in animals and humans process a variety of signals to subserve aspects of decision making, including the incentive salience of stimuli, absolute and relative reward values, costs and outcome history (Hosokawa et al., 2013; Rudebeck et al., 2006; Scholl et al., 2015). Among these regions, the anterior cingulate cortex (ACC) is particularly well connected to integrate effort and reward information for action selection (Paus, 2001). ACC lesions result in apathy and a disinclination towards voluntary movements or speech in neurological patients (Grunsfeld and Login, 2006) and pharmacological inactivation of the region in animals biases behavior away from effortful reward seeking (Winstanley and Floresco, 2016). Thus, while we can appreciate its role, how precisely the ACC performs the computations to support effort-based decision making through its connections with other cortical and subcortical structures and how stress impacts these computations is still an open area of investigation.

To explore the mechanistic link between stress exposure and amotivation, new approaches should be considered which enhance the granularity of data collected. I addressed these questions in a recent study published in Neuron, once again enlisting an arsenal of in vivo tools to delineate circuit contributions to effortful reward seeking behavior (Fetcho et al., 2024). Using fiber photometry recordings, my co-authors and I discovered that nucleus accumbens (NAc)-projecting ACC neurons (corticostriatal pathway) encode an anticipatory reward signal which scales in magnitude with the amount of effort required when adult male mice choose to pursue a larger volume reward by climbing a taller barrier. Optogenetic Inhibition of these neurons during effort expenditure and reward consumption significantly reduces future high-effort reward seeking, underscoring the importance of this reinforcement or “confidence” signal. Moreover, the amplitude of the signal is dampened following chronic CORT administration and scales with the severity of anhedonia-like behavior as indexed by diminished effort valuation. Importantly, we measured serum CORT levels at the time of behavioral testing and found that they were returned to a normal range, suggesting that the effects of stress on circuit function in this context are enduring and not mediated acutely by circulating glucocorticoids. Stress effects on effort-based reward seeking behavior have been described using a variety of models (Dieterich et al., 2021; Kúkel'ová et al., 2018) however, our approach uncovered an important correlation between the strength of frontostriatal circuit function and individual variation in motivational impairment. In order to further delineate the cell- and circuit-level mechanisms through which animals effectively weigh effort costs against desired rewards, I have developed a task to enable simultaneous two-photon calcium imaging of cingulate neural populations defined by projection target. Using this paradigm, I have begun to uncover how stress may bias animals towards lower utility actions by diminishing the coding efficacy of reward- and effort-related stimuli within specific pathways (unpublished). Continued emphasis on the design and implementation of behavioral paradigms which probe distinct features of motivated responding offer greater translational potential. A focus on how brain and body mechanisms enable physical and cognitive effort expenditure is further supported by cross-species validation as these constructs can be measured effectively in animal and human subjects of both sexes (Bustamante et al., 2023; Treadway et al., 2009). Cleverly designed assays such as one recently proposed by Xeni and colleagues, offer the advantage of examining naturalistic behavior without deprivation-based incentives to study motivation (Xeni et al., 2024). Additionally, there is value in scaling up the number of variables measured while animals perform learned or naturalistic behaviors. Freely available machine vision and AI-based methods for pose estimation and unbiased behavioral classification provide the opportunity to build a richer picture of the development of pathological stress effects (Goodwin et al., 2024; Hu et al., 2023; Mathis et al., 2018).

3. Shining a spotlight on resilience

Understanding the contributing mechanisms which distinguish depressive and healthy brain states with attention to specific domains of function is integral to the development of effective and targeted therapies. Importantly, not all individuals who experience stress develop compromised reward processing and variation in stress susceptibility and resilience may be associated with differential circuit function and transcriptional profiles (Der-Avakian et al., 2014; Bagot et al., 2016; Nasca et al., 2019). Moreover, stress coping strategies and buffering stimuli or events may serve as protective elements against stress pathophysiology (Vila, 2021). Whether and how some neural pathways are spared from undergoing these adaptations is of particular interest. Behavioral and physiological biomarkers could hold value in predicting stress responsiveness. In a recent example, a correlation between early coping responses to social defeat and later social avoidance behavior was identified using behavioral classification methods as described in the previous section (Murra et al., 2022). Innovative in vivo and computational approaches shed new light on neural activity signatures of stress risk and resilience across networks (Hultman et al., 2018; Hing et al., 2024; Xia et al., 2024). Many potential molecular mediators may also contribute to stress-associated alterations in circuit function. Differentially expressed genes (DEGs) across various cell categories have been identified in prefrontal and striatal regions following stress (Labonté et al., 2017; Siemsen et al., 2022). Similarly, we may expect to observe DEGs in projection neurons which encode information relevant for effortful reward seeking as a consequence of stress exposure.

We are actively exploring this question using a spatial RNA sequencing platform which preserves the anatomical and microcircuit environment of brain regions of interest such as the ACC. By combining histological and transcriptome-level data from probe-containing “spots” and deconvolving cell type clusters, we can focus in on classes of neuronal and non-neuronal cells to reveal unique and shared transcriptional alterations in male and female mice vulnerable or resistant to motivational deficits (Rodriques et al., 2019). Our preliminary results reveal distinct patterns of overlapping DEGs, including genes up- or downregulated in both resilient and control animals compared to susceptible animals (‘susceptibility genes’), those up- or downregulated in both susceptible and control animals compared to resilient animals (‘resilience genes’), and a set which are up- or downregulated in both susceptible and resilient mice compared to controls (‘stress genes’). These categories of genes exhibit differential gene ontology enrichment, providing some insight into their biological functions. Enrichment for terms related to synaptic and glutamatergic transmission, and vesicle regulation may impact the ability of cells to maintain patterned activity and encode information (Arnsten, 2009). We might expect also that specific patterns will differ across projection populations given the variation in inputs and long-range connectivity. Thus, a distinct set of DEGs in resilient mice could confer a protective benefit against stress-related pathology and inform targets for future studies aimed at promoting resilience. Sequence data with a spatial dimension can be further leveraged to gain valuable information about cell-to-cell interactions and communication (Jin et al., 2021). As network level studies have done, spatial omics studies could help construct a more detailed picture of genetic and molecular signatures of stress exposure across connected cell classes such as inhibitory and excitatory neurons and glial cells.

4. Lighting the path ahead for the circuit and cellular neuroscience of stress

Looking ahead, I envision a landscape in which rapidly advancing techniques offer a fine-grained view into the regulatory mechanisms of stress in both the brain and body (Teckentrup and Kroemer, 2024). Tool development and refinement in this area is sometimes difficult to keep pace with but exciting to witness. I have had the privilege to train in methods that have expanded my perspective on what can be learned about genetic, molecular, circuit and behavioral effects of chronic stress exposure. A key goal of my research program is to integrate these levels of analysis to inform novel therapeutic candidates to address motivational deficits, a particularly intractable symptom class across diseases. A recent inspiring example in which basic research led to a breakthrough pharmacotherapy in psychiatry is the development of zuranolone, an FDA-approved treatment for postpartum depression (PPD). In a foundational study by Maguire and Mody, hippocampal GABA_A receptor $\delta$ subunit expression and function were identified as substrates regulating depressive and maternal behaviors during the postpartum period (Maguire and Mody, 2008). Targeting these specific receptors with positive allosteric modulators has proven promising in alleviating symptoms of this particular form of depression (Thompson, 2024). Importantly, as clinical observations underscore, depression is heterogenous and can be subtyped by symptom profiles and these profiles are associated with varying neural network signatures and functional connectivity which may involve pathology at distinct circuits (Drysdale et al., 2017; Lynch et al., 2024). Therefore, it will become imperative to leverage this information for precision mental healthcare. Exciting developments are also underway in the field of neuromodulation where deep brain stimulation (DBS) or non-invasive transcranial magnetic stimulation (TMS) is delivered in a targeted manner to alleviate symptoms in a subset of treatment-resistant depressed patients (Fujimoto et al., 2024). Preclinical stress models will continue to provide an avenue to test hypotheses about which circuits are most associated with behavioral domains affected in depression and how they may respond to specific neuromodulation approaches (Fig. 1). An overarching phenomenon elucidated by preclinical studies and supported by neuroimaging data is that synaptic alterations in depression and those induced by chronic stress are maladaptive. A complementary aim of my independent research program therefore is to evaluate the potential of different types of compounds, which specifically enhance plasticity, to restore adaptive behavior (Duman et al., 2016; Martin and Nichols, 2018; Aleksandrova and Phillips, 2021; Parekh et al., 2022; Davoudian et al., 2023). Our ultimate hope is to identify commonly regulated signaling pathways which can be targeted for pharmacological, stimulation-based or behavioral therapeutic intervention and to better understand natural resilience.

Fig. 1.

Preclinical studies can inform specialized therapies for stress-related disorders. A) Developments in optical physiological methods allow for the precise interrogation and manipulation of neural and non-neural activity throughout the brain in animal models of stress. B) Short- and long-term structural and functional consequences of stress can be determined within specific cell types. C) Sophisticated behavioral paradigms paired with neural data collection will continue to reveal the nuanced roles of brain structures and circuits in supporting processes which are disrupted in stress-related disorders. D) Circuit- and network-based studies can improve targeting and parameter optimization for therapies to treat the heterogenous symptoms of depression.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Puja K. Parekh reports financial support was provided by National Institute of Mental Health. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.