Fragmentation and high entropy of neonatal experience predict adolescent emotional outcome

Abstract

Vulnerability to emotional disorders including depression derives from interactions between genes and environment, especially during sensitive developmental periods. Across evolution, maternal care is a key source of environmental sensory signals to the developing brain, and a vast body of work has linked quantitative and qualitative aspects of maternal care to emotional outcome in children and animals. However, the fundamental properties of maternal signals, that promote advantageous vs pathological outcomes in the offspring, are unknown and have been a topic of intense study. We studied emotional outcomes of adolescent rats reared under routine or impoverished environments, and used mathematical approaches to analyze the nurturing behaviors of the dams. Unexpectedly, whereas the quantity and typical qualities of maternal care behaviors were indistinguishable in the two environments, their patterns and rhythms differed drastically and influenced emotional outcomes. Specifically, unpredictable, fragmented maternal care patterns translated into high-entropy rates of sensory signals to the offspring in the impoverished cages. During adolescence, these offspring had significant reductions in sucrose preference and in peer-play, two independent measures of the ability to experience pleasure. This adolescent anhedonia, often a harbinger of later depression, was not accompanied by measures of anxiety or helplessness. Dopaminergic pleasure circuits underlying anhedonia are engaged by predictable sequences of events, and predictable sensory signals during neonatal periods may be critical for their maturation. Conversely, unpredictability maternal-derived signals may disrupt these developmental processes, provoking anhedonia. In sum, high-entropy and fragmented patterns of maternal-derived sensory input to the developing brain predicts, and might promote, the development of anhedonia in rodents, with potential clinical implications.

Figure 1

Sucrose consumption, a measure of anhedonia, is reduced in adolescent male rats reared in cages with limited bedding/nesting material (LBN). (a) Relative consumption of sucrose (% of total liquid intake) as well as (b) sucrose consumption (ml), measures of pleasure/reward seeking, were reduced over a 2-week period in individual rats reared in LBN cages during a sensitive early postnatal period as compared with those reared in routine laboratory cages (CTL) (n=24 per group). Data are also presented as daily consumption for individual rats as well as box and whisker plot (10th and 90th percentiles) for the groups. Horizontal bars represent mean values. Black asterisks denote statistical significance using the two-way RM-ANOVA; *P<0.05; **P<0.01. RM-ANOVA, repeated-measures analysis of variance.

Figure 2

Peer-play, a second measure of pleasure, is significantly altered by the rearing environments. (a) Total duration of social interactions was not different in rats reared in limited bedding/nesting material (LBN) vs control cages. However, (b) the fraction of total time spent playing with peers, a measure of pleasure or joy, as well as (c) time spent in following or other play behaviors were significantly lower in these adolescent males vs the controls (P<0.05, Student's t-tests; n=5 per group). (d) Immobility time in the Porsolt forced swim test, a measure of learned helplessness, did not differ significantly between groups. In two measures of anxiety-like behaviors, (e) open field and (f) elevated plus maze, durations of behaviors considered indicative of anxiety did not differ significantly among groups (all P>0.05, Student's t-tests; values are provided in seconds; n=12 per group for (d–f). Data presented as box and whisker plot show the 10th and 90th percentiles. Horizontal bars represent mean values. Black asterisks denote statistical significance using the Student's t-test; *P<0.05, ^†P=0.056. CTL, control.

Figure 3

Several measures of the quantity and quality of maternal care do not distinguish dams in routine cage environment (CTL) from those in limited bedding/nesting (LBN) cages. (a) Total duration of nursing was not significantly different. (b) Durations of arched-back nursing, considered a measure of optimal quality of maternal care, were similar between groups. (c) Duration of time spent licking/grooming pups were comparable in CTL and LBN dams (all P>0.05, Student's t-tests). When separate analyses were performed for nurturing behaviors during the light phase or the dark phase, (d) nursing, (e) arched-back nursing and (f) licking/grooming times did not differ between groups (all P>0.05, Bonferroni's post hoc test). Values are provided in seconds, and are sum of observations over two 50-min periods per day for 8 days. (n=6 per group). Data presented as box and whisker plot show the 10th and 90th percentiles. Horizontal bars represent mean values. CTL, control.

Figure 4

Fragmentation, unpredictability and high-entropy rates of maternal care in limited bedding/nesting (LBN) cages. (a) Mean duration of individual bouts of licking/grooming was drastically shorter in the LBN group, creating a fragmented sensory input to the pups. (b) Heat-map depicting behavioral sequences of an individual dam. The probability that a given behavior will follow another is depicted along a color scale. Controls tended to have either high (orange–red) or very low (white and light blue) probabilities of behavioral sequences. In contrast, LBN mothers had mid-range probabilities (bright blues and greens) to engage in any sequence of nurturing behaviors, suggesting that these sequences were random. (c) Entropy was employed to mathematically define the unpredictability of maternal behavior patterns (see Materials and methods). Entropy rates in LBN dams were significantly higher on average than those of CTL dams. (d and e) When broken down for entropy rates during light phase and dark phase, the group differences persisted, with a stronger effect during the light phase period. Data presented as box and whisker plot show the 10th and 90th percentiles. Horizontal bars represent mean values. Black asterisks denote statistical significance using a linear model; ***P<0.001, **P<0.01, ^†P=0.08. CTL, control.