Effects of early life adversity on immediate early gene expression: Systematic review and 3-level meta-analysis of rodent studies

Early-life adversity (ELA) causes long-lasting structural and functional changes to the brain, rendering affected individuals vulnerable to the development of psychopathologies later in life. Immediate-early genes (IEGs) provide a potential marker for the observed alterations, bridging the gap between activity-regulated transcription and long-lasting effects on brain structure and function. Several heterogeneous studies have used IEGs to identify differences in cellular activity after ELA; systematically investigating the literature is therefore crucial for comprehensive conclusions. Here, we performed a systematic review on 39 pre-clinical studies in rodents to study the effects of ELA (alteration of maternal care) on IEG expression. Females and IEGs other than cFos were investigated in only a handful of publications. We meta-analyzed publications investigating specifically cFos expression. ELA increased cFos expression after an acute stressor only if the animals (control and ELA) had experienced additional hits. At rest, ELA increased cFos expression irrespective of other life events, suggesting that ELA creates a phenotype similar to naïve, acutely stressed animals. We present a conceptual theoretical framework to interpret the unexpected results. Overall, ELA likely alters IEG expression across the brain, especially in interaction with other negative life events. The present review highlights current knowledge gaps and provides guidance to aid the design of future studies.


Study Protocol
Registration number (if applicable) 9.
Stage of review at time of registration Completed preliminary searches, started with piloting of the study selection process. B. Objectives Background 10.
What is already known about this disease/model/intervention? Why is it important to do this review?
Exposure to adversities during childhood (early-life adversities, ELA) increases the risk to develop psychiatric disorders in adulthood. Building upon the compelling epidemiological evidence, rodent studies have investigated the mechanistic effects of ELA on the brain, ultimately leading to changes in behavior. Modelled as alterations in maternal care, ELA alters brain development on multiple levels, including synaptic organization.
Important contributors to synaptic development and cognition are immediate early genes (IEGs). IEGs are expressed directly but transiently upon cell activity; hence, they can be considered a marker for information processing in the brain. Since IEG proteins vary from transcription factors to post-translational proteins, their different functions can highlight different aspects of synaptic development.
Systematically reviewing the current literature on the topic can provide insights on long-term changes of ELA on IEGs throughout the brain, thereby providing possible mechanisms for ELA-induced changes in information processing. Research question 11. Specify the disease/health problem of interest Childhood maltreatment; Early life adversity; Stress-related psychopathology; Healthy animals 12.
Specify the population/species studied mice and rats, because they are the most frequently used animal models in stress research; female and male 13.
Specify the intervention/exposure 1) Early life adversity starting before P14; early life adversity defined as alteration in maternal care; models included are maternal separation/deprivation, isolation, limited bedding and nesting, licking and grooming (as measure of variation in maternal care, with final comparisons between offspring receiving low vs high maternal care (for reference: Liu et al., 1997)), handling 2) Acute stressors applied to adult animals; will be restricted to most common ones as based on results from formal screening Both, 1) and 2) need to be applied. 14.
Specify Type of animals/population (e.g. age, gender, disease model) Inclusion criteria: adult mice or rats (older than 8 weeks, but younger than 1 year); female and male Exclusion criteria: any other species than mice or rats; sexes are pooled; sex is not specified; ovariectomized females; specific pathogen free animals; genetic manipulations; animals bred for high/low anxiety or novelty response or sensitivity/resilience to depression; animals separated in high/low performance; any manipulations to earlier generations; animals with any comorbidities 25.
Type of intervention (e.g. dosage, timing, frequency) Inclusion criteria: 1. Early life adversity starting before P14; early life adversity defined as alteration in maternal care; models included are maternal separation/deprivation, isolation, limited bedding and nesting, licking and grooming (as measure of variation in maternal care, with final comparisons between offspring receiving low vs high maternal care (for reference: Liu et al., 1997)); handling is also considered early life adversity, but will be included only at a systematic review level 2. Acute stressors applied to adult animals; the types will be restricted to most common ones as based on results from formal screening Exclusion criteria: pharmacological intervention ("control" injections (of any pharmacological intervention) such as vehicle, saline, sesame oil are instead included); communal nesting as early life adversity model; maternal separation with early weaning, unless early weaning is also applied to control group; the same acute stressor has been applied earlier in life and is therefore not new to the animal Change: we include also at rest measures. I.e. lack of acute stress is not an exclusion criteria.

26.
Outcome measures Inclusion criteria: IEGs expression as measured by mRNA or protein expression in one of the following brain regions: amygdala, hippocampus, hypothalamus, medial prefrontal cortex, nucleus accumbens, striatum; IEGs expression in other brain regions will be included only at a systematic review level Exclusion criteria: brain regions not specified 27.
Publication date restrictions None 29. Other The second hypothesis-confirming research question asks about the effects of multiple hits on the relationship between ELA and IEGs expression.
The following events are considered second hits: Other (e.g. drop-outs) / Assessment risk of bias (internal validity) or study quality 37.
Specify (a) the number of reviewers assessing the risk of bias/study quality in each study and (b) how discrepancies will be resolved Risk of bias will be assessed by two independent researchers. Risk of bias is assessed following SYRCLE guidelines, and it will be distinguished between experimental and study bias. Discrepancies will be resolved by discussion between two experimenters. Should no conclusion be reached between two experimenters, a third researcher (expert in the field of early life adversity), will be consulted for a solution. 38.
Define criteria to assess (a) the internal validity of included studies (e.g. selection, performance, detection and attrition bias) and/or (b) other study quality measures (e.g. reporting quality, power) Specify (a) the number of reviewers extracting data and (b) how discrepancies will be resolved (a) One reviewer will complete data extraction, with a second reviewer checking random samples for agreement. Any numbers presented in the article or supplementary material will be extracted. If data is only presented graphically, then 'WebPlotDigitizer' will be used to extract data from graphs. If results of statistical analyses are given, these will be used to infer summary statistics. Two authors per publication will be contacted in case of missing data, followed by a reminder in case of no reply. Should authors not answer within two months, the comparisons will be reported as missing and will be excluded from analyses.
(b) Discrepancies will be resolved by discussion between two experimenters. Should no conclusion be reached between two experimenters, a third researcher (expert in the field of early life adversity), will be consulted for a solution. Data analysis/synthesis 42.
Specify (per outcome measure) how you are planning to combine/compare the data (e.g. descriptive summary, metaanalysis) A quantitative synthesis is planned for results concerning the immediate early genes c-fos, arc, and egr1. Data will be split by sex, and the meta-analysis will be conducted for each dataset separately, since we consider males and females to be two different biological systems that should not be grouped together. 43.
Specify (per outcome measure) how it will be decided whether a meta-analysis will be performed The decision on which brain regions and acute stressors to include in the quantitative analysis will be made after study selection, with frequency being the determining factor. Remaining immediate early genes, brain regions and acute stressors, as well as the early life adversity model of handling, will be covered in a narrative/descriptive synthesis. If a meta-analysis seems feasible/sensible, specify (for each outcome measure): 44.
The effect measure to be used (e.g. mean difference, standardized mean difference, risk ratio, odds ratio) The standardized mean difference Hedge's g ( (mean(Control) -mean(Experimental)) / pooled SD ) will be used for all outcome measures.

45.
The statistical model of analysis (e.g. random or fixed effects model) 3-level mixed effect meta-analysis (in case of multiple outcomes from the same animals) otherwise random effects meta-analysis, with early life stress predicting IEG mRNA and protein expression. IEG identity, presence of second hits and brain region (if applicable) will be moderators in the model. 46.
The statistical methods to assess heterogeneity (e.g. I 2 , Q) Cochranes Q-test; I 2

47.
Which study characteristics will be examined as potential source of heterogeneity (subgroup analysis) Type of IEGs, ELA models, species, types of acute stressors, brain area and outcome measure (mRNA vs protein) will be used for subgroup analyses. These will be considered exploratory.

48.
Any sensitivity analyses you propose to perform Specified prior to the analysis, we will assess the influence of the following factors on the outcome measure: 1. Influential cases and outliers. 2. Blinded and randomized studies. 3. Risk of potential bias (bias will be assessed with the SYRCLE Risk of bias tool and an overall score will be used for sensitivity analysis). 49.
Other details meta-analysis (e.g. correction for multiple testing, correction for multiple use of control group)

Extracted Variables
To increase subjectivity during data extraction, variables to be extracted were determined a priori.
The spreadsheet containing all extracted variables and variable coding is available at https://osf.io/qkyvd/. IEG name and product; measurement (technique, unit of recording (e.g., counts, expression, optical density) and unit of comparison (e.g., raw data, fold change, averages across slices)); brain area and hemisphere Data mean, variance and n of control and experimental groups; significant effect

Variables' grouping
A) Brain areas as named in publications and as grouped for the analysis.

Grouped for analysis Named in publications Amygdala
Central

Bias assessment
Bias assessment. A) Risk of bias assessment according to SYRCLE's risk of bias tool. B) Funnel plot for publication bias.

Sensitivity analysis species
Given that data from both rats and mice were combined in the meta-analysis, we performed a sensitivity analysis to confirm that the findings are robust to the effect of species. The effect size remains unchanged when analyzing data from rats only ( Given fundamental biological differences between males and females [1], we a priori chose to evaluate female cFos data separately from males'. Only ten publications reported on cFos expression in female rodents (ncomp = 77). The majority of these studies found no significant differences between cFos levels of ELA versus controls at rest or after an acute stress challenge (ncomp = 55; [2][3][4][5][6]).
Only five studies performed the same experiments in both male and female rodents. Among these, Desbonnet et al. [5], Gaszner et al. [6] and Renard et al. [2] reported the same null effects for both male and females. In contrast, James et al. [3] and Genest et al. [4] found no significant ELA effects on cFos levels in females, while they did report significant differences in males under the same conditions. These sexually dimorphic results could have methodological origins, such as male-focused behavioral paradigms (for ELA or acute stress) or could reflect true biological differences between the sexes [1,7].
The remaining five studies investigated exclusively females, and all reported at least one significant difference between ELA and control rodents. Auth et al. [8] found significantly increased cFos levels in female mice at rest, but not after acute stress exposure. Interestingly, across two independent baseline cohorts, increased cFos was observed once in the dorso-lateral periaqueductal gray and once in the lateral amygdala, suggesting that the effects do not easily replicate within the same lab. Similarly, Rivarola and colleagues [9,10] observed an increase in cFos levels in the anteriordorsal thalamic nucleus of animals with a history of multiple hits in a first [9] but not a second publication [10].
Finally, O'Leary et al. [11] reported decreased cFos levels in the dorsal dentate gyrus and ventral CA3 of female ELA mice after restraint stress, but not in other hippocampal, hypothalamic, prefrontal cortical or amygdalar areas. Banqueri et al. [12] demonstrated differential directionality of effects after the Morris water maze, with ELA females showing increased cFos levels in hippocampal structures, and decreased expression in prefrontal areas. All in all, ELA effects on cFos in females appeared limited. Whether the results are truly sexually dysmorphic remains to be elucidated.

cFos and other brain areas
Five studies investigated the effect of ELA on cFos expression in brain areas of male rodents other than those reviewed in the meta-analysis, including the striatum, sensory cortices, hindbrain nuclei and the cerebellum of male rodents. Out of 24 comparisons, 16% displayed a significant difference between ELA and control animals (ncomp = 4) at systematic review level. Troakes et al. [13] showed that cFos levels of ELA males are significantly decreased in the piriform cortex in comparison to controls after acute exposure to a mild stressor, but not at rest. Early research indicated that cFos levels in the piriform cortex are highly responsive to acute stressors, and its role in the sensory integration of olfactory stimuli suggests that the reduced cFos expression could correspond to decreased information processing abilities under stressful [14,15].
In addition, Menard et al. [16] found decreased cFos expression in ELA males in the lateral septal complex and the ventral subiculum after performing a shock-probe burial task, but not in other striatal areas or hindbrain nuclei. Given that the lateral septal complex relays reward and fear information for contextualization of the experience, the decreased cFos expression here potentially presents a task-specific effect related to spatial mapping of the buried probe [17]. However, Shin et al. [18] report upregulation of cFos after ELA in the lateral septal complex as well as the ventral tegmental area in a social interaction task, suggesting a broader task-specific involvement of striatal areas.
Finally, neither Clarke et al. [19] nor Desbonnet et al. [5] could find significant differences between ELA males and controls in the bed nuclei of the stria terminalis, neither at rest nor after acute stress, suggesting that cFos expression in this area is not or only minimally changed after ELA.
All in all, these results suggest that areas with task-specific effects are worth exploring, and that cortical areas involved in sensory processing and information integration potentially display altered transcriptional activity as well. Yet, considering that the most frequently areas under investigation are also those areas considered to be sensitive to the effects of stress, it is likely that the main results of interest are covered by the meta-analytic outcomes.

cFos and alternative behavioral paradigms.
Acute stressors that included a strong memory, reward or social component were excluded from the meta-analysis. They involve cognitive processes other than the response to stress, which recruit brain-areas depending on the task requirements.
Daskalakis et al. [20] investigated cFos expression in rats placed back into a fearful context after a fear-conditioning paradigm, thereby probing memory processes in addition to stress-related functions. cFos expression in the medial amygdala and basolateral amygdala was increased in rats placed into a novel cage during the maternal separation (MS) procedure, while an increase was only observed in the medial amygdala in MS animals that remained in the home cage [20]. This study highlights how that choices of study characteristics (i.e., home cage vs novel cage) can influence the outcome investigated.
Two studies further investigated the effects of ELA on cFos expression after exposure to a rodent version of the Iowa Gambling Task [21,22]. This task depends not only on spatial memory, but also contains a strong reward component [22]. In 2012, van Hasselt et al. [21] correlated percentage of licking and grooming with cFos expression in a wide range of brain areas in male and female rats and found a negative correlation in the shell of the nucleus accumbens and the agranular insular cortex when sex was pooled. However, using the same task, MS did not alter cFos expression in these areas in the 2017 study, but rather decreased cFos expression in the right CA1, right CA3, left infralimbic area and left agranular insula [22]. While inconsistent, these studies highlight that rewardbased processes also likely result in differential activation of IEGs after ELA exposure, thus, warranting further investigations in the future.
Under several social paradigms, no differences between ELA and control animals were observed in medial PFC areas [18,23,24], the central amygdala [23], the dorsal raphe nucleus [25], or striatal and hypothalamic areas [18]. On the other hand, Benner et al. [23] observed an increase in the basolateral amygdala and a decrease in CA1 of cFos expression in ELA mice compared to controls after 40-days social competition task. A possible explanation is that the differences observed in the study by Benner and colleagues are due to the memory component, rather than the stress/social component of the task. In addition, Shin et al. [18] observed an increase in cFos expression in the lateral septal complex and the ventral tegmental area after social interaction in mice previously exposed to social isolation, suggesting that multiple adverse experiences may be required to observe altered IEG expression after ELA in social tasks. Overall, social behaviors in isolation seem less inducive of activity-regulated transcription than the above-discussed reward-based and memory-based paradigms.

ELA and IEG other than cFos
Arc is a post-synaptic protein, which plays an essential role in regulating the homeostatic scaling of AMPA receptors, thereby directly modifying plasticity at the synapse [26]. Arc expression has been investigated in five publications under varying conditions in male and female mice and rats.
While two publications did not find any alterations in the mPFC, hippocampal, or amygdaloid areas at rest or after acute stress [23,27], another study reported a significant decrease in CA1, CA3 and dentate gyrus Arc levels in male ELA animals at rest [28]. Interestingly, animals in this study were exposed to maternal separation for one week longer (PND 1-21) than animals in the studies reporting no significant alterations, suggesting that the duration of the ELA experience could be essential in causing long-term effects on Arc expression. It is noteworthy that a decrease in Arc expression results in increased synaptic plasticity [26], thus, following in line with the findings of increased cFos expression at rest in the male meta-analysis.
In contrast, McGregor et al. [29] found increased Arc expression at rest in the dorsal striatum of male rats with and without a history of second hits. As this publication is the only one reporting on IEG levels in the dorsal striatum, it is unclear whether the finding is a result of the study design or presents a genuine area specific IEG response. Rincel et al. [30] suggest that ELA effects on Arc expression are sex-specific, showing evidence that ELA leads to decreased Arc expression in the mPFC of male mice, but to increased Arc levels in the mPFC of female mice. These contradictory findings could be a strain-specific, as C3H/HeNRj mice were used [30]. All in all, reported Arc levels appear to be in coherence with cFos effects on synaptic plasticity at rest, and thereby further support the notion of at rest sensitization of activity-regulated transcription.
Early-growth response (Egr) proteins are a family of transcription factors with a zinc-finger motif, which allows all Egr factors to connect to identical DNA binding sites [31]. We identified three studies investigating Egr expression after ELA exposure at rest; one investigated Egr-1 [32], another investigated Egr-4 only [30], and one other investigated Egr-2 and Egr-4 [29].
Egr-1 mRNA expression was decreased in the cortex, but only in Balb/c and not C57Bl/6 male mice [32]. This is in line with the general notion of Balb/c mice as a stress-sensitive strain [33]. In contrast, McGregor et al. [29] report increases in Egr-2 and Egr-4 expression in the dorsal striatum, with Egr-2 levels only significantly increased in animals experiencing a second hit during adolescence.
Finally, Rincel et al. [30] highlight that ELA alters Egr-4 expression in a sex-specific manner in the mPFC of mice, observing an downregulation in males but an upregulation in females.
Since it is expected that proteins of the Egr-family behave similarly [31], the discrepancy between findings are likely the result of differences in study design, such as the brain areas investigated. Considering that IEGs of the Egr-family, and in particular Egr-1, have been shown to be associated with the development and treatment of those psychiatric disorders, which individuals with a history of early life stress are more likely to develop, Egr-family proteins are an understudied, yet important candidate for investigating activity-regulated transcriptional alterations after ELA in the future [34].
FosB is an IEG of the Fos family, and -similarly to cFos -if binds to members of the Jun family to form the AP1 transcription factor [35]. Of particular interest in stress research is its isoform FosB, whose extended half-life makes FosB an exceptional marker for chronic stress [35].
Three publications reporting on the expression of FosB at rest in ELA and control animals were identified. Kim et al. [36] reported a reduction of FosB expression in the nucleus accumbens of ELA females in comparison to controls, whereas Wang et al. [37] report elevated FosB levels in the mPFC of ELA rats of unspecified sex, pointing towards opposite effects of ELA in these two areas.
Interestingly, and in line with these findings, previous results suggest that overexpression of FosB in the nucleus accumbens accompanied by reduced expression of FosB in mPFC promote a phenotype resilient to the effects of chronic stress [38,39]. It should still be highlighted, that Lippmann et al. [40] found no significant alterations in either of these areas in male rodents, neither induced by maternal separation nor by handling. Due to the low number of studies investigating FosB, we cannot conclude whether these null findings are attributable to sex or a result of study design heterogeneity. Yet the outlined potential of a more stable IEG in researching chronic alterations in transcriptional activity emphasizes the relevance of investigating ELA modifications on FosB expression.