Bioelectricity Buzz

Recent Bioelectricity-Related Articles Selected by Ann M. Rajnicek, Media Editor of Bioelectricity

The year has started out with a scientific bang, which is reflected in this instalment of Bioelectricity Buzz. The hot topics this time range from the role of electricity in making the perfect cup of espresso, a role for dynamic calcium oscillations in mechanosensitivity of diabetic bone, a bioinspired ion channel for proximity detection, a brain organoid system that can (a little creepily) “learn” to recognize speech patterns, ion channel control of temperature-sensitive leaf folding in plants, the link between seasonal electrical brain activity in reindeer, and their feeding behavior. Hopefully, there is something to suit everyone.

In a Shocking Grind: Electrostatic Charging of Coffee During Grinding

As you settle down to read this first Buzz of 2024, you might be enjoying a hot cup of coffee. But have you ever considered the role electricity may play in generating that perfect brew? Harper and colleagues have quantified how moisture content influences particle charging and clumping during the grinding process and the consequences on the resulting cuppa.

As coffee beans fracture and rub together during grinding, the resulting particles acquire substantial surface charges by physical cracking and rubbing processes, called fractoelectrification and triboelectrification, respectively. Their influence is especially pronounced when grinding for espresso preparation, which requires a finer grind texture. Accumulated charges cause the grounds to clump together into aggregates or to disperse onto the grinding machinery, impacting the efficiency and consistency of coffee preparation. Although this is important in domestic kitchens, it is even more important at industrial scales.

Initial experiments quantified the charge on whole coffee beans rolled down a vibrating ramp lined with materials used in the coffee industry, including metal and plastic. The beans were collected in a Faraday cup at the bottom of the ramp and their charge was recorded. Intact beans became positively charged as they rubbed against plastics such as polyvinyl chloride and Mylar, but became negatively charged when encountering glass and nylon, and acquired virtually no charge against office paper.

Coffee is necessarily ground before use, and this process imparts significant charge, which Harper and coworkers quantified for coffee ground under controlled conditions. Typically, the resulting particle diameters range from 100 to 2 mm depending on the burr type and grind settings. The finer grinds required for espresso yield particles with more “fines” (<100 μm) and smaller “boulders” (>100 μm), which would have experienced more fracture events and burr grinding contact time, and, therefore, are likely to accrue more surface charges. This was measured by collecting the grounds into a Faraday cup and calculating the charge-to-mass (Q/m) ratio (total charge per cup) to account for differences in grind yield.

When ∼30 different coffee samples of different Agtron (color scale) roasts were assessed, there was a weak correlation with charge, with positive charging at darker Agtron values. This was slightly better when internal water content (after roasting) was considered (>2% water by mass). Using two subparallel plates (held at potential of ∼8.2 kV) at the end of the collection chute as an electrostatic separator coupled with imaging and laser diffraction particle sizing, it was possible to determine whether charge polarity impacted size distribution, and to quantify the Q/m for each particle class.

Some general trends emerged. The resulting Q/m ratios of positive coffee samples were generally smaller (<50 nC/g) than those of negative coffee samples (up to 120 nC/g). The larger “boulders” tended to carry a negative charge regardless of roast or the bean's geographic origin, and boulders emerged from the grinder chute first. Fine (<100 μm) particles (as from dark roasts) tended to have a negative charge but midsized (100–300 μm) particles, which typify an espresso grind profile, had a mainly positive charge. Indeed, fine darker roasts had Q/m ratios akin to charged pumice particles in volcanic plumes and thunderclouds.

Customized roasting of Ethiopian green coffee beans (12% starting water content) yielded 12 distinct roast types, ranging from light to dark, which were then ground at uniform burr settings. The charge profile of the resulting coffee revealed that roasting conditions were more important than the characteristics of the green coffee beans in determining the charge accumulated during grinding. The trend was positive charges for short roasts, transitioning to negative with increased roasting times. Negative charging was favored when water content was <2%.

Baristas sometimes deal with the pesky problem of static by incorporating a small amount of water with the whole beans when grinding. Harper and team found that this strategy resulted in less clumping and practically no grounds being retained in the grinder, validating the barista's technique, and suggesting that it is the most efficient way to extract ground coffee. Harper and colleagues' tests included systematically adding water and determining the Q/m ratio of the resulting grounds. Incorporating water of ∼20 μL/g approached zero net charge (0 nC/g) on the grounds, but atmospheric humidity levels had to exceed 60% for a similar effect. Furthermore, charge was passivated, and clumping was reduced by adding as little as 5 μL/g water.

To deduce whether the triboelectrification and fractocharging arise mechanistically through nuclear ion transfer, electronic ionization, or a combination, the team added either pure water or sodium chloride (0.5 or 1 M) during grinding. If ion transfer is the principal mechanism, then the salty solution should reduce charging, but this did not occur, suggesting that electron transfer underpins charging.

This is interesting in an academic sense, but much scientific endeavor is driven by coffee consumption, so practically minded scientists want to know what impact this has on our final cup of espresso. When brewing parameters were kept constant, using either dry ground beans or beans ground with water added, the wet coffee preparation produced consistently slower shots through the espresso puck and increased beverage strength. So next time you need a “pick me up,” add a drop or two of water to the grinder to boost your brew (and to help reduce the mess caused by statically charged particles).

Under Pressure: Mechanoresponsiveness of Diabetic Mouse Bone Is Mediated by Calcium Oscillations

The incidence of type 2 diabetes (T2D) and its accompanying health challenges are on the rise, including bone fragility and elevated risk of fracture. Shao and colleagues explored the mechanism underpinning this fragility, implicating osteocyte calcium dynamics in mechanical responsiveness and density.

Bones become porous in individuals with T2D, increasing the risk of fracture, particularly for the load-bearing bones of the legs and hips. Mechanical loading (e.g., weight-bearing exercise) increases bone formation and prevents bone resorption, but mechanistic understanding of the link between force transduction and bone density under T2D conditions is limited, so few therapies target bone density directly.

Structurally, the bone matrix consists of narrow fluid-filled channels (the lacunar–canalicular system, LCS) populated by osteocytes. In addition to their role in controlling bone homeostasis, osteocytes are the main mechanosensory cells in bone. Shao and colleagues showed previously that skeletons respond to mechanical loading by LCS fluid shear stress, Ca²⁺ oscillations in osteocytes, actin contractions, and secretion of vesicles containing important bone regulatory proteins. These observations linked mechanical loading of bone with Ca²⁺ oscillations in osteocytes, suggesting that manipulation of these oscillations could improve bone mechanoresponsiveness. The team hypothesized that under T2D conditions, mechanosensitivity would be attenuated in osteocytes due to changes in the Ca²⁺ oscillation pattern.

Initial experiments compared bones of a nondiabetic (male) mouse with those of a genetically spontaneous (KK-Ay mice) T2D model, and with those of mice in which T2D was induced experimentally using high-fat diet (HFD) and streptozotocin (STZ) injection. In each group, axial cyclic compressive loading of the tibia of the left leg was performed over 2 weeks, with no compressive loading (control) of the right leg. Microcomputed tomography (micro-CT) and hematoxylin and eosin staining showed that mechanically loaded tibiae in male nondiabetic mice had 30% more trabecular bone mass and thickness than the unstimulated contralateral tibiae, and they had thicker less porous cortical bone.

However, the T2D mice showed no such improvements in tibial bone mass or porosity in response to loading. Biomechanical tests demonstrated increased mechanical properties for nondiabetic bone after cyclical loading, but the mechanical properties of T2D bones were comparatively reduced.

To understand the mechanism(s) underpinning these differences, metabolic profiles of osteocytes, osteoblasts, and osteoclasts were compared for bones of nondiabetic and T2D mice in response to cyclic loading. Relative to control tibiae of nondiabetic animals, there was decreased LCS density, lower osteocyte osteoprotegerin (OPG) protein expression, but increased osteocyte survival, receptor activator of nuclear factor kappa-B ligand (RANKL) expression, and sclerostin expression in T2D mouse tibiae. Cyclic loading increased LCS density, osteocyte survival, and OPG expression, and loading reduced RANKL and sclerostin expression in osteocytes of nondiabetic mouse tibiae.

Interestingly, cyclic loading of T2D tibiae did not induce changes for those same parameters. In nondiabetic mice, loading increased the number of osteoblasts on the bone surface and mineral deposition, whereas in T2D mice, these differences were not observed in response to loading.

Real-time Ca²⁺ imaging of oscillatory signaling in tibiae in situ revealed that osteocytes in nondiabetic (control) mice responded to mechanical loading with robust repetitive Ca²⁺spikes, in contrast to T2D osteocytes, in which loading produced only a few weak spikes. The T2D mouse skeletons also had fewer Ca²⁺ responsive cells, lower number of spikes, and greater relaxation time. In vitro experiments using fluid shear stress (FSS) as a stimulus revealed similar findings for osteocytes (large Ca²⁺ oscillations for nondiabetic and attenuated responses for T2D conditions), suggesting that reduced Ca²⁺ oscillations in T2D are linked mechanistically to reduced mechanoresponsiveness in T2D bone.

To probe the cellular mechanism for this attenuation of T2D responsiveness, Shao and team targeted the sarcoplasmic/endoplasmic reticulum (ER) Ca²⁺-ATPase (SERCA), a pump in the ER. After the ER releases stored Ca²⁺, SERCA is responsible for transporting the Ca²⁺ back into the ER. Treating osteocytes with SERCA antagonists (thapsigargin or cyclopiazonic acid) attenuated osteocyte Ca²⁺oscillations in vitro upon FSS stimulation and in situ upon mechanical loading of tibiae.

Exposure to a high-fat/high-glucose environment reduced both gene and protein expression of SERCA2 in osteocytes in vitro, and SERCA2 expression (gene and protein) was decreased in T2D mice compared with that in nondiabetic mice. Collectively, the series of experiments reduced SERCA2 expression with reduced osteocyte mechanoresponsiveness.

As a counterpoint to the SERCA2 antagonist experiments, the team used istaroxime (ISTA), a selective SERCA2 agonist, which accelerates Ca²⁺ uptake into the ER through SERCA2, demonstrating that it augmented Ca²⁺ oscillations in response to compressive loading. Ca²⁺ number of spikes and intensity were both increased in the T2D osteocytes upon cyclic loading in situ with similar enhanced responses observed in vitro upon FSS stimulation.

Since ISTA-augmented Ca²⁺ signaling, its influence was tested on bone mechanoresponsiveness in T2D mice in situ. After 2 weeks of cyclic loading (1200 cycles/day), micro-CT revealed increased tibial bone mass and thickness in the ISTA-treated and mechanically loaded T2D tibiae, but not in the non-ISTA-treated tibiae. The ISTA-treated (mechanically loaded) group also had fewer osteoclasts on the tibial trabecular bone surfaces and more osteoblasts with enhanced mineral deposition. Immunohistology also revealed upregulation of SERCA2 expression after ISTA treatment in T2D mouse tibia.

Having implicated SERCA2 mechanistically in the deterioration of T2D bone mechanoresponsiveness, the team generated a conditional osteocyte SERCA2 knock-in mouse. The mice were then rendered T2D by the HFD/STZ method followed by 2 weeks of cyclic loading of tibiae. Micro-CT analysis of the SERCA2 knock-in mice under T2D conditions revealed increased bone thickness and density compared with wild-type mice treated with HFD/STZ. The conclusion was that SERCA2 overexpression in osteocytes interferes with the bone deterioration usually associated with T2D.

The nuclear transcription factor peroxisome proliferator-activated receptor alpha (PPARα) is a metabolic regulator that aids glucose regulation, and its dysregulation underpins diseases states, including obesity and diabetes. Shao and colleagues identified that under high-fat/high-glucose conditions, reduction of SERCA 2 pump activity in osteocytes is associated with attenuation of PPARα activity. Since SERCA2 is important for Ca²⁺oscillation dynamics, it connects mechanistically with mechanoresponsiveness and homeostasis of diabetic bone, suggesting new targets to prevent bone loss in T2D.

You Are Getting Closer: A Hygroelectric Proximity Sensor Inspired by Biological Ion Channel Function

Minute changes in the environmental electric field allow some animals to detect prey. This biological phenomenon has inspired a proximity detector for robotic devices based on humidity changes emitted by approaching biological objects that drives ion transport through an artificial membrane to generate electricity.

Sensory systems in animal skin can be extremely sensitive to external stimuli, including heat, touch, and humidity. The humidity generated by biological organisms enables some robotic proximity sensors to distinguish between biological and nonbiological entities by sensing changes in atmospheric water vapor as the object approaches. Detection of environmental object proximity is important for robots too, so they can plan movement trajectories and avoid obstacles.

At millimeter scales, robotic proximity sensors tend to be based on magnetic, pneumatic, and capacitive systems, each of which has unique technical limitations. In addition, they cannot distinguish between living biological and inert entities, which restricts their capacity for use in scenarios that require noncontact human–machine interaction (e.g., a touchless computer screen).

There are clear examples in nature of proximity sensing by animals, such as the fresh water tropical aquarium fish Apteronotus albifrons (black ghost knifefish), which responds to prey proximity by altering properties of sensory cell membrane potential and ion transport. This concept inspired Zhang and colleagues to devise a self-powered proximity sensor (SPS) based on hydroelectric power generation. As a living organism approaches the SPS, a shift in the humidity field induces ion transport across a membrane film, with the short circuit current indicating its proximity to the sensor.

The team used negatively charged carbon nanotube (CNT) films and positively charged anodic aluminum oxide (AAO) films to create selective ion transport channels in the hybrid membrane. Plasma treatment of the upper CNT electrode creates abundant oxygen-containing sites for binding atmospheric moisture, with the lower electrode being a metal, such as indium gallium liquid metal. The CNT films are ∼2.3 μm thick and have pores ranging from 10 to 200 nm, whereas the AAO films are ∼47 μm thick, with pores ∼90 nm in diameter. The nanoscale of the pores permits adsorption and ionization of water, with ion transport controlled by the electric double layer, endowing it with ion selectivity.

The oxygen-containing COOH and OH groups on the surface of the CNT film yield a negative zeta potential after ionizing protons. Conversely, the neighboring AAO surface has a positive zeta potential after the adsorption of protons, creating an electric field in which the cathode is at the CNT film and the anode is at the AAO membrane. In response to humidity, the water molecules, together with their ionized H⁺ and OH⁻ radicals, migrate toward the lower electrode, creating a spatial ion gradient. Redox reactions occur at the electrodes, transforming the current from ionic to electronic current.

No external power source is needed for this device to detect humidity changes. It derives energy from latent heat released by the gas–liquid phase transition of water, the gradient energy of the humidity difference between the upper and lower electrodes, and the chemical energy released by the redox reaction of the electrodes. Because an increase in humidity results in relatively more ion accumulation at the electrodes and increased redox reactions, this generates a larger electrical signal, permitting proximity of living organisms to be estimated.

The response time of the SPS during transition from 15% to 70% relative humidity was an impressive 0.3 s, with a rapid recovery time of 2.5 s. Importantly, the approach of a human finger was detected, at 12, 6, and 2 mm, with maximum SPS response at 2 mm distance, returning to baseline as the finger is moved away, demonstrating the reversibility of the system. Proximity testing with other wet objects (sponge, damp wood) also yielded increased signal as the object approached.

Proof-of-concept experiments imagined future usage scenarios. For example, the device could be built into a face mask to monitor breathing rate, it could be used to monitor deterioration of protective (insulating) gloves, to create noncontact switches and, therefore, prevent transmission of viruses or bacteria. Combining several SPS units in an array permits the two-dimensional shape of the approaching object to be identified (tested here for triangles, squares, hexagons, and parallelograms). By passing a finger over such an array, it is possible to activate units (therefore, switches) in sequence, thus permitting a sort of contactless interactive touch screen.

Challenges moving forward are to improve temporal and spatial resolutions and to make sure the SPS detector can adapt to the natural variation between different people's usual finger “sweat” states (some people are sweatier than others), especially in relation to atmospheric humidity changes. Nonetheless, this technology has a lot of potential (ahem) for future applications. I hope our community can get creative with it. For example, contact sensors for prosthetic limbs?

Brain Training: Electrical Stimulation of a Brain Organoid Trains It to Recognize Speech Patterns

My facility for languages other than English is (embarrassingly) pitiful, so the tiny three-dimensional brain organoids used in this study put my relatively complex brain to shame. After electrical training, the cultured tissues can sculpt neuronal connectivity without further supervision to enable speech recognition. This impressive technology provides new horizons for biological–electronic artificial intelligence (AI) applications.

The technological shift to AI has inspired attempts to recapitulate neural network function in silico as artificial neural networks (ANNs), but this approach is limited by its general reliance on electronic principles for AI hardware. Ultimately, AI systems seek to emulate the low energy demands (∼8 W for the brain compared with 20 W for a comparable ANN) and high processing efficiency of biological neural networks. Cai and colleagues have created a brain organoid–high-density multielectrode array hybrid they named “Brainoware.”

The starting point is a brain organoid created in vitro by the self-assembly of an aggregate of human pluripotent stem cells. This brain organoid is composed of cells with multiple characters such as early stage neurons, mature neurons, astrocytes, and progenitor cells. In addition, the organoid was shown to develop structures akin to brain regions of the ventricular and subventricular zones, also developing complex organoid neural networks (ONNs). After 2 months in culture, the resulting organoids were mounted onto an electrode array that allowed recording of up to 1024 channels and the stimulation of up to 32 electrodes simultaneously. Experiments were performed 7 days after plating onto the array. External electrical stimulation of the ONNs evoked neural activity that was then recorded and decoded.

The Brainoware system was tested as a reservoir computing framework, in which input systems map onto higher computational frameworks through a “black box” (the reservoir) comprising the dynamics of the physical system. When input is fed into the reservoir, an output is generated that maps the state of the reservoir onto a desired output, performing the computational task. In a traditional system, the dynamics of the reservoir are fixed so training of the readout function is not possible. Brainoware offers more flexibility in that the reservoir is a biological entity (the organoid), so “unsupervised learning” of what Cai and colleagues term the “adaptive living reservoir” occurs as neural connections are made and modified.

In practical terms, specific spatiotemporal electrical stimulation patterns are projected onto the organoid, and neural output signals provide a readout function for tasks. Training with specific electrical stimulation sequences allows Brainoware to become more computationally efficient. But perhaps more impressively, this study demonstrates that Brainoware adapts the living reservoir through unsupervised learning, responding to the electrical stimulation by altering functional neural connectivity within the organoids.

This was tested in an elegant set of experiments. Having demonstrated the reservoir properties of Brainoware (nonlinear dynamics, fading memory, and spatial information processing) using pulses of differing duration and voltages, it was tested on a real-world task of speech recognition. The aim was to demonstrate that Brainoware could identify a particular speaker's vowels from a database of isolated Japanese vowel sounds. For the test, 240 audio clips of eight different male speaker's isolated vowel sounds (one epoch) were converted to spatiotemporal electrical sequences of bipolar pulses delivered to Brainoware. The ONN output was processed and converted to a logistic regression algorithm, and the resulting stimulation pattern was used to train the organoid.

A naive organoid was trained over 2 days, with one epoch delivered every 12 h. A confusion matrix assessment performed before any training showed ∼51% accuracy, suggesting baseline levels of existing neural connections used to complete the task. However, after four training epochs, the same organoid improved its accuracy to 78%, with accuracy of recognition improving over training epochs, indicating that training gradually improved performance of the speech recognition task. This was supported by evidence that synaptic plasticity (a blend of weakened, strengthened, new, and pruned synapses) was increased in trained organoids. This notion was proven by pharmacological block of activity-dependent synaptic plasticity in organoids, which prevented unsupervised learning.

The ability to merge human and electronic technologies in a way that can also display biological adaptation of a learning response is both a great achievement and a wee bit eerie. But fears of a fully functioning cyborg are premature. There are several hurdles that need to be addressed if Brainoware is to achieve high efficiency, reproducible performance, and inexpensive high-throughput production. First, biology is pesky. It is not trivial to create large number of viable (uninfected, generally healthy) brain organoids with standard characteristics, and, therefore, optimal computational abilities.

And such cultures, especially at scale, require costly materials, growth factors, and media. Second, the electrode array used by Cai and colleagues was flat, which means only a tiny percentage of the surface of the organoid was connected to the electronics. Devising a softer malleable electrode will allow it to conform to the shape of the organoid, potentially improving recording and stimulating parameters and efficiency. The enhanced signal output will require improved computational methods for encoding and decoding, but hopefully development of new algorithms and computational methods will keep pace.

A Chilled Salad Samanea pulvini Outward-Rectifying K⁺: Ion Channel Control of Leaf Folding in Cold Wet Environments

For living organisms, the ability to sense environmental temperature is essential for survival. Temperature sensors in animals may cause them to move to shade when too hot and to find shelter when too cold. But what about relatively sedentary plants, which do not have the same temperature sensors as animals, and which have fewer ways to alter their immediate environment?

Animals, algae, and fungi use thermosensitive transient receptor potential (TRP) cation channels as thermosensors, but land plants lack TRP channel genes. The legume tree Samanea saman responds to changes in light, temperature, and rainfall by folding its leaves, a circadian behavior known as foliar nyctinasty. Muraoka and colleagues tested the hypothesis that S. saman leaf folding is in response to a drop in leaf temperature (e.g., accompanying rainfall) and, therefore, nyctinasty could be used to explore the mechanism for temperature sensitivity in land plants.

Water was sprayed onto S. saman leaves to simulate rain. Cold (4°C) but not warm (30°C) water reduced leaf temperature (measured with an infrared camera) and induced leaf folding. Environmental cold also caused leaves to fold, with unfolding upon warming. Collectively these observations demonstrate that leaf chilling stimulates folding. Mechanistically, folding is caused by volume changes in flexor motor cells located at the base of the leaf. Therefore, flexor protoplasts were prepared and subjected to temperature changes to determine whether volume changes could be detected directly in response to cooling.

Swelling was detected within 2 min of a temperature drop, which is consistent with the flexor cell response to darkness, another nyctinastic trigger. These data indicate that flexor cells sense temperature directly and respond with changes in cell volume.

Since ion efflux has been linked previously to nyctinasty and flexor cell swelling, whole cell patch-clamp experiments were performed on flexor cell protoplasts. Outward K⁺ currents were identified, with significantly higher current detected at warm temperatures, implicating deactivation of the K⁺ current upon chilling in cell swelling. To identify the ion channel genes underpinning leaf folding, gene expression was quantified for the Samanea pulvini outward-rectifying K⁺ channel 2 (SPORK2) and two of the silent slow-type anion channel subunits (SsSLAH1 and SsSLAH3) in flexor cells at warm and cold temperatures. Cold-induced swelling was shown to be independent of transcriptional regulation of these main ion channel genes.

Since gene expression was not the most important factor, Muraoka and colleagues next explored the role of temperature activation or deactivation of ion channel activity by expressing SPORK2 or the SsSLAH1–SsSLAH3 complex in Xenopus oocytes. Using two-electrode voltage-clamp techniques, they discovered that the outward-rectifying K⁺ currents carried by SPORK2 were significantly increased at warm (30°C) temperatures compared with cooler (18°C) temperatures. This change was reversible and independent of extracellular K⁺ concentration.

In contrast, the Cl^- transport activity of SsSLAH1–SsSLAH3 was not affected by a temperature shift, and intracellular Ca²⁺ chelators had only minimal influence, suggesting that SPORK2 was the major thermosensing ion channel in flexor cells. Exploration of the protein sequence and thermosensitivity of various chimeric proteins identified the transmembrane helix 3 and extracellular loop 2 (TM3) and C-linker region at the carboxy terminus as the temperature sensing regions of SPORK2.

Because genetic modification of S. saman is impossible, the team switched to Arabidopsis thaliana, in which the gated outwardly rectifying K⁺ channel (GORK) present in guard cells regulates stomatal opening and closing by cell swelling or shrinking, respectively. GORK maintains K⁺ transport activity at lower temperatures in S. samani, so it was hypothesized that if SPORK2 replaced GORK in A. thaliana, it would have defective stomatal closure at lower temperatures because of defective ion channel function. Heterologously expressed SPORK2 induced stomatal closure in response to temperature.

Overall, the study demonstrates that SPORK2 activity is regulated by temperature directly and that it controls leaf folding in response to cold, suggesting mechanistic parallels for temperature sensing by ion channels in plants (SPORK2) and animals (TRP channels).

A Late-Night Snack for Santa's Team? Seasonal Sleep–Wake Rhythms Affect Reindeer Rumination

Just as we move into spring, putting thoughts of winter and its holiday overindulgences behind us, here is an interesting tidbit about the electrical activity of reindeer brains and its relationship with seasonal food intake.

The daily rhythms of our bodies, including eating and sleeping patterns, are driven by an inbuilt circadian clock. Conversely, Arctic reindeer (Rangifer tarandus) have a weak (or nonexistent) circadian clock, with activity and melatonin secretion responding predominantly to light conditions. Consequently, if light (or dark) conditions are constant, reindeer do not show circadian cycles of melatonin secretion or daily behaviors (e.g., eating). Furrer and colleagues tested for the first time using electroencephalographic (EEG) recordings whether reindeer sleep cycles show a similar absence of 24-h rhythm.

EEG recordings were made from female reindeer in winter, summer, and autumn. Lighting conditions in the stable followed the local twilight in Tromsø, Norway, except in winter, where stable windows were covered to simulate constant darkness (conversely, it was naturally constantly light in summer). EEG brain activity and electromyogram muscle activity (rumination) recordings were made over 4 days per season, incorporating two 2-h sleep deprivations (midday and midnight) daily with 24-h baseline recordings between 5 am and 5 am.

Across all seasons, the time spent in classical vigilance states was consistent, with ∼5.4 h in nonrapid eye movement (NREM) sleep, 0.9 h in rapid eye movement (REM) sleep, and 2.9 h ruminating. Furthermore, there was no significant 24-h rhythm identified in inactive states under constant lighting (winter and summer), but in autumn reindeer, perhaps unsurprisingly, spent more time in inactive states when it was dark than when it was light.

Sleep quality was assessed by measuring the slow wave activity (SWA) in the EEG during NREM sleep before and after the sleep deprivation episodes. After 2-h sleep deprivation, SWA was elevated during NREM sleep in all seasons, indicating elevated sleep pressure (the cumulative physiological urge to sleep), but SWA elevation in winter was larger than in autumn and summer. Since sleep deprivation increased sleep pressure, the research team wondered whether reindeer used strategies other than NREM sleep to prevent accumulation of sleep pressure, such as rumination, an activity that shares behavioral quiescence and EEG oscillation features with NREM sleep.

When behavior and EEG of NREM sleep were assessed, animals were found to be in a “sleep-like state” during rumination, with temporal coupling between rumination slow waves and the slow waves characteristic of NREM sleep. The more reindeer ruminated, the less time they spent in NREM sleep. This invited exploration of the differences between sleep recovery during rumination and NREM sleep.

During rest periods (animals awake for ≤5 min), SWA of REM sleep was lower after rumination than before it, and during prolonged rumination episodes (>30 min) the SWA increased initially then decayed. The dynamics of SWA during REM sleep were modeled to determine whether the decline in SWA during rumination correlated with dissipation in sleep pressure. Data were fitted to two opposing models: one in which rumination decreases sleep pressure akin to NREM sleep, and the other in which rumination increases sleep pressure, as in the awake state. There was a striking preference for the model in which sleep pressure is decreased during rumination.

Indeed, reindeer were found to spend less time in NREM sleep the more they ruminate. This suggests that reduced sleep pressure may drive additional feeding during summer months, linking the Artic extreme environmental conditions with substantial seasonal changes in electrical EEG activity, sleep pressure, overall activity, and food intake.

Now I am off for a salad (with reindeer-inspired carrots) and a cup of coffee. Until next time, that is the Buzz!