Signal to nurture: remodelling the pregnant brain

Successful parenting is an essential skill in many species. Since it can be undertaken with minimal experience, and entails considerable sacrifice with no immediate benefit, parenting behaviour is likely innate, driven by evolutionarily shaped, hard-wired circuits.

While he was a postdoc in Catherine Dulac’s lab at Harvard, Jonny Kohl began tracing these parenting circuits in the mouse brain, focussing on a small group of neurons in the medial preoptic area (MPOA) of the hypothalamus which express the neuropeptide galanin. These MPOAGal neurons had previously been shown by the Dulac group to be essential for parenting behaviour in both sexes, and Jonny and his colleagues extended this work to show that this small group of neurons—10,000 out of a total of 100 million—act as the hub of a brain-wide, dedicated parenting circuit. MPOAGal neurons are organised in distinct clusters which all receive input from about 20 brain areas; each individual cluster then projects to a distinct downstream destination in the brain and manipulating specific clusters can induce discrete aspects of parenting behaviour [1].

After opening his lab at the Crick in 2019, Jonny began to pursue a remarkable aspect of the parenting problem—how it is that during pregnancy, the brain prepares for the future behavioural needs of parenthood, without “knowing” in advance what those needs might be. The physiological adaptations of pregnancy were known to be accompanied by behavioural adaptations, but how and when pregnancy hormones remodel the brain to instruct these changes was unclear. In this paper, the Kohl lab showed in mice that MPOAGal neurons are the key to the puzzle: the hormones oestradiol and progesterone act together to modify these neurons and also rewire the circuits the neurons control, creating the profound changes in behaviour that prepare mice for motherhood [2].

Newborn pups need not just milk from their mother, but also protection, warmth and grooming. Although a small number of classical studies in rats suggested otherwise, these behavioural responses were widely thought to kick in post-birth. To explore the issue of timing, the Kohl lab looked at pup-directed behavioural changes during the whole 3-week gestation period of pregnant mice. They found mothers-to-be were more maternal towards pups throughout their pregnancies than virgin controls, performing sequences of pup retrieval, crouching, nest building and grooming that by late pregnancy rivalled the parenting performance of mothers. Persistent pup exposure was not required to drive the change, meaning that it was induced by the hormonal milieu of pregnancy. Crucially, most changes persisted even when oestradiol and progesterone levels returned to baseline, suggesting the adaptations likely resulted from long-lasting remodelling of the brain.

Oestradiol and progesterone exert their long-term effects by changing gene expression. They do this by binding and activating their respective intracellular receptors estrogen receptor 1 (Esr1) and progesterone receptor (PR), which then act as transcription factors. Whether activation of Esr1 and PR might mediate the known effects of MPOAGal neurons on maternal behaviour was not known. To study this, the lab first showed that expression of both Esr1 and PR was broadly enriched in many of the MPOA neurons, and then used a sophisticated knockout strategy to ablate each receptor gene in just the small population of MPOAGal neurons. When virgin mice with either of these highly specific knockouts were mated, there was a complete loss of nurturing behaviour during pregnancy, which also persisted once their litters were born. Therefore, direct action of oestradiol and progesterone on MPOAGal neurons through their intracellular hormone receptors is necessary for pregnancy-mediated increases in maternal behaviour, and if hormone action is blocked during pregnancy, maternal behaviour cannot be rescued by post-birth endocrine events.

What was happening to the MPOAGal neurons? Rather than their simply becoming more active during pregnancy, there were more subtle changes. By looking in brain slices from late-pregnant females, the team could see that the MPOAGal neurons were being fine-tuned by increasing the signal-to-noise ratio: their spontaneous activity was reduced, but they were more active than normal if electrically stimulated. In parallel with this, new excitatory inputs were being recruited, correlated with an increase in the excitatory signal receivers—the dendritic spines—of the MPOAGal neurons. This remodelling only occurred in MPOAGal neurons, not in the wider population of Gal-negative MPOA neurons.

To determine whether these changes were a result of direct hormone action, the experiments were repeated using MPOAGal neurons from pregnant mice in which either the Esr1 or PR genes had been deleted. Loss of these receptors had distinctly different effects: Esr1 deletion specifically prevented the increase in signal-to-noise ratio but had no effect on synaptic inputs and spine density, while PR deletion did exactly the opposite—there was no up-regulation of synaptic inputs and spine density, but the signal-to-noise ratio was still improved. Oestradiol and progesterone therefore control separate aspects of pregnancy-induced plasticity in MPOAGal neurons: oestradiol tunes up intrinsic excitability, and progesterone connects new excitatory inputs, changing the circuit context in which MPOAGal neurons function.

Looking at MPOAGal neurons from mothers shortly after they had given birth, and from those whose pups were weaned, it was clear there was a difference in the durability of the changes. The Esr1-mediated fine-tuning faded as the pups matured, but the PR-mediated circuit changes endured, and promoted nurturing behaviour even after the pups were weaned and relevant hormone levels had returned to normal. This change in the wiring of the MPOAGal neurons, if general, could explain why mothers of many species go on behaving more maternally, even towards offspring not their own.

Brain slices in a dish cannot react to behavioural inputs, so the lab next examined MPOAGal neuronal activity in live mice. This can be indirectly measured by expressing a genetically encoded calcium sensor in MPOAGal neurons to look at transient calcium signalling, and monitoring this in individual MPOAGal neurons via a glass lens implanted into the brain. As pregnancy progressed, there was a decrease in spontaneous MPOAGal neuronal activity, but those MPOAGal neurons still active were more intensely activated during pup interactions, but not in response to other stimuli. Mirroring the situation in brain slices, after pups were weaned, the fine-tuning of MPOAGal neurons returned to normal, but pup stimulus selectivity showed a long-lasting increase.

This honing of neurons to adapt and respond strongly to particular stimuli is known as population sparsening, and it is a well-studied developmental phenomenon. As the brain of a young animal matures, responses to external cues such as visual stimuli change so that fewer neurons fire but those that do are more intensely excited: the brain's circuits are being adaptively rewired. The work here shows that this is not confined to early life, and that pregnancy hormones can open up a similar window of plasticity in the adult brain to prepare the animal for motherhood.

This paper is an important advance in understanding how hormones can remodel the brain, but there are now new mysteries to solve. Firstly, how is it that loss of PR or Esr1 in just the MPOAGal neurons is sufficient to block the pregnancy-driven induction of maternal behaviour? This is unexpected given the broad hormonal milieu of pregnancy which includes increased prolactin, but also because many other neuron types express PR and Esr1. How is specificity achieved? The question speaks to the wider issue of how steroid signalling can regulate competing behavioural repertoires, such as aggression, mating and parenting by activating only a small subset of the larger population of hormonally responsive neurons. One theory proposed by the authors is that differential gene activation may depend on differences in chromatin access—MPOAGal neurons may have a different set of genes "open for business" to PR and Esr1 than do other hormone-responsive neuronal cell types.

Secondly, what is the origin of the newly rewired input the MPOAGal neurons are receiving? From Jonny's previous paper, it is known that MPOAGal neurons receive long-range inputs from about 20 areas, but it is not clear which connectivity is being remodelled, and indeed, whether new input sources, perhaps even from the MPOA itself, are coming online. This question is part of a wider one that the lab is currently working on: as the MPOAGal neurons are getting information from the sensory periphery, from which they are distant, are hormones remodelling a cascade of upstream neuronal connections, and if so, which hormones are involved and what are they doing?

Finally, parenting can also be taught by socialisation, as males and ovariectomised female mice can still become highly parental via social experience (i.e. repeated exposure to pups): does this route to good parenting act on the same parts of the brain as the hormonal route, and if so, are the same plasticity mechanisms being used to affect the circuitry? The Kohl lab's approach means that the answers to these and many other fascinating questions are now within reach.