Metabolic demands, sensory deficits: tradeoffs in times of scarcity

Abstract

The brain requires a lot of energy to carry out its functions at peak performance. In times of energy deficit, something has to give. In this issue of Neuron, Padamsey et al. (2021) explore how metabolic demands impact cortical coding by demonstrating the effects of food restriction on visual processing.

The brain is a metabolically demanding organ, and the cerebral cortex, where much of our representation of the sensory environment is formed, is particularly greedy. Precise and accurate encoding of incoming sensory stimuli is a critical yet demanding task for the cortex, both in terms of real estate (i.e., the amount of tissue dedicated to sensory processing) and energy use. While estimates of the energy used by specific neural processes vary, the firing of action potentials and excitatory synaptic currents are significant contributors to cortical energy use. When an animal is facing an energy deficit, as arises during chronic caloric restriction, it would stand to reason that reducing energy consumed by the brain would be useful, even at the cost of optimal neural processing. In this situation, finding the correct balance between information processing and energy use could be critical for an animal’s fitness. Here, Padamsey et al. (2021) use in vivo recording approaches to demonstrate reduced ATP use by visual cortex neurons in chronically food-restricted mice and elucidate the effects of this energy deficit on visual information processing (Padamsey et al., 2021).

Recording in layer 2/3 of visual cortex in awake mice, Padamsey et al. (2021) use whole-cell voltage clamp recordings to show that excitatory synaptic currents triggered by visual stimulation were reduced in mice that had been food restricted to 85% of their normal weight, compared to ad libitum fed mice. This result suggests reduced ATP use in these neurons, a hypothesis supported by subsequent imaging of a fluorescence-based ATP sensor. Specifically, the authors locally apply an ATP-synthesis inhibitor to visual cortex and image the decay in ATP-dependent fluorescence as existing ATP is used, showing a faster rate of decay in ad libitum fed mice when compared to food-restricted mice. Despite this change in ATP use, there was no observable difference in action potential firing in these neurons during presentation of visual stimuli. These results cannot be attributed to acute hunger, since all mice were given access to food in the hour before all recordings, suggesting that the effects are the result of a longer-term energy deficit. Several controls demonstrated that the observed changes were also not due to changes in stress or arousal state, including measurements of stress hormones and pupil diameter, a sensitive readout of arousal and cortical excitability (McGinley et al., 2015).

Using ex vivo cortical slices acutely prepared from food-restricted mice, Padamsey et al. (2021) made patch clamp recordings and found that the change in excitatory currents was due to a reduced postsynaptic AMPA receptor-mediated current. Given this lower excitatory input current but no change in excitatory-inhibitory balance, alternative mechanisms would have to underlie the lack of change in spike rate. In vivo recordings demonstrated that both an increase in input resistance and depolarization of the resting membrane potential compensated for the reduced input current to maintain similar spike rates. If the brain can decrease energy use and maintain action potential firing in neurons, why isn’t this the default state? Is there a cost to this energy savings? A common approach to driving responses in neurons in visual cortex is to present gratings of different orientations. Particularly in layer 2/3 of primary visual cortex, pyramidal neurons are generally tuned to a specific orientation, with responses dropping off as the angle changes from the preferred orientation. Here, using single-cell recordings to track spiking activity in visual cortex of awake mice, the authors show that neurons exhibit less selective tuning for stimulus orientation in food-restricted mice compared to ad libitum fed mice. Furthermore, the authors show that this was a consequence of increased response variability due to increasing spike probability from the small depolarizations caused by non-preferred orientations.

As processing of more complex visual features is informed by low-level features like orientation tuning, the food restriction-induced broadening of orientation tuning should lead to impaired visual discrimination. To test this hypothesis, Padamsey et al. (2021) employ in vivo two-photon calcium imaging of visual cortical neurons during presentation of natural scenes and a maximum-likelihood decoder to assess how well one could discriminate between different scenes based on neural activity alone. The decoder was equally accurate in both the ad libitum fed mice and the food-restricted mice for scenes with large differences (i.e., scenes from different environments); however, for similar scenes (i.e., those from the same environment), the decoder accuracy was worse in the food-restricted mice, consistent with reduced visual acuity. This finding is supported by subsequent behavioral experiments showing that food-restricted mice do worse in a water maze task where they must discriminate between orientation gratings to find a hidden platform. Food-restricted mice underperform relative to the ad libitum fed mice once the target orientation and non-target orientation were within 10 degrees of each other.

Overall, these experiments suggest that a long-term energy deficit results in a shift in neuronal processing to save energy at the expense of visual acuity. Padamsey et al. (2021) then investigate potential physiological mechanisms relaying long-term energy balance to cortical neurons that could induce this change. The authors show that the visual deficits are present in food-restricted mice both before and after short-term ad libitum feeding, and that there is no difference in short-term regulators of energy balance, like glucose. However, the visual deficits corrected after several days of re-feeding, suggesting a longer-term regulator of energy balance. One key regulator of energy balance is circulating leptin, a hormone released from adipocytes that is significantly reduced during food restriction. Indeed, the authors show that leptin was reduced in food-restricted mice, even after short-term feeding. Furthermore, pharmacological restoration of leptin signaling was able to rescue the observed visual deficits.

These findings raise important considerations for the experimental design of studies investigating processing in sensory cortex. Animal studies, using electrophysiology or in vivo imaging techniques, have elucidated much about how the cortex processes sensory information, encodes cue-outcome associations, and builds a representation of the sensory environment. However, many of these studies utilize tasks where the animals are food or water restricted (resulting in chronic weight loss) and rewarded for sustained vigilance or task engagement. Padamsey et al. (2021) demonstrated that this type of chronic restriction can result in broader tuning in cortical neurons and reduced visual acuity. In light of these results, interpretations of previous studies of cortical processing should factor in the long-term energy state of the subject and its potential effects on neural processing.

Some questions not addressed by the present study include the fitness benefits of this energy-state-dependent change in cortical processing and whether there are “real-world” fitness deficits to the reduced visual acuity. Even mice rely on visual processing while hunting for food, at least in some contexts, when they are presumably in (or anticipating) a state of energy deficit (Hoy et al., 2016). Several recent studies have shown that neural representations of visual food-associated stimuli can be selectively enhanced when a human or mouse is hungry (Burgess et al., 2016; Huerta et al., 2014; Livneh et al., 2017). It is possible that this selective enhancement is able to compensate for the changes in processing in primary visual cortex that result from an energy deficit. While leptin was able to “rescue” visual acuity in the current study, more rapid, centrally regulated feeding signals may indirectly prime cortex to specifically respond to food-associated cues (Burgess et al., 2018). For example, short-term food restriction facilitates activity of neurons that provide motivation to seek food rewards (Calhoon et al., 2018; Mandelblat-Cerf et al., 2015), and activation of hypothalamic agouti-related peptide (AgRP) neurons, which causes a rapid increase in food seeking, was shown to selectively enhance food cue responses in insular cortex (Livneh et al., 2017). In this way, the sensory cortex could save energy to account for long-term energy deficits, at the expense of some information processing, but not hinder the ability to address that energy deficit through recognition and acquisition of food.

This study by Padamsey et al. (2021) provides novel insights into how the brain adapts in times of scarcity, but there remain a number of unanswered questions. Is there a role for glia and/or glia-neuron interactions, which play an active role in neural energy metabolism (Barros et al., 2018)? Are there short-term compensatory mechanisms to accurately process food-related stimuli during specific goal states (e.g., hunting, food-seeking)? Are cortical neurons encoding other sensory modalities or executive functions similarly affected? Despite these questions, it is evident that a greater consideration of brain state and physiological need when designing future studies investigating sensory processing will be important to fully understand cortical function.