A Neuroscience Primer for Integrating Geroscience With the Neurobiology of Aging

Abstract

Neuroscience has a rich history of studies focusing on neurobiology of aging. However, much of the aging studies in neuroscience occur outside of the gerosciences. The goal of this primer is 2-fold: first, to briefly highlight some of the history of aging neurobiology and second, to introduce to geroscientists the broad spectrum of methodological approaches neuroscientists use to study the neurobiology of aging. This primer is accompanied by a corresponding geroscience primer, as well as a perspective on the current challenges and triumphs of the current divide across these 2 fields. This series of manuscripts is intended to foster enhanced collaborations between neuroscientists and geroscientists with the intent of strengthening the field of cognitive aging through inclusion of parameters from both areas of expertise.

Aging is the single greatest risk factor for numerous diseases, with no organ system escaping the impact of time. Unlike many pathological conditions, however, barring a catastrophic event, aging affects 100% of the population. Over the past several decades, there have been remarkable increases in the average life expectancy in the United States and other developed countries (1). Unfortunately, the cognitive health span does not frequently accompany this enhanced longevity, and impaired cognition that could threaten an individual’s ability to live independently is a hallmark of advanced age (2). Thus, the discovery and translation of effective therapies for promoting optimal cognition across the entire human life span are of fundamental importance. Designing these types of therapies, however, is impeded by an incomplete understanding of the neurobiological mechanisms that are responsible for cognitive aging and how peripheral health directly impacts the brain. The neurobiology of aging field has a long history of linking brain changes to cognitive status, but this research has largely developed independently of research that examines how aging impacts peripheral health and the biological variables that control life span.

As outlined in a companion perspective in this issue by Hernandez et al. (3), the fields of geroscience and the neurobiology of aging have advanced in relative isolation. While there are a number of reasons for the figurative barrier at the neck when examining the impact of advanced age on organ systems and to move forward with improving overall health outcomes in older adults, it is going to be critical to unite the fields of geroscience and neuroscience. The purpose of this primer and minireview is to introduce geroscientists to the neurobiology of aging and highlight how neuroscience approaches to understanding cognitive aging could compliment research into the biology of aging. This is intended to compliment a companion primer that introduces neuroscientists to the field of geroscience by Hoffman et al. (4). It is our goal to bridge these 2 disciplines to promote new hypothesis and interdisciplinary collaboration.

A Historical Perspective on the Neurobiology of Aging

In 1905, evaluating data from hundreds of brains, Raymond Pearl documented a reduction in brain weight with age in both males and females (5). In 1955, Brody was the first to suggest that age-related reductions in brain weight could be explained by reduction in neuron number (6). These early observations were followed by a series of reports that documented an overall degradation of brain morphology with age (7). The data obtained from these early reports, however, were confounded by various technical and methodological issues, such as tissue processing and sampling design, that later called into question the veracity of their findings (8). In the 1980s, when new stereological principles for unbiased estimation of cell numbers were developed, it became evident that in humans (9), nonhuman primates (10), and rodents (11,12), significant neuron loss is not characteristic of normal aging. Moreover, around the same time, it was also found that in the absence of dementia or frank neurodegeneration, general neuron morphology is largely intact in the aged brain (13). Thus, the age-related impairments in memory, processing speed, and other cognitive functions that became widely reported by the 1970s (14,15) had to result from more subtle neurobiological changes that affect how neurons communicate and dynamically change in response to experience (16).

Two seminal observations laid the foundation for what would evolve into the neurobiology of aging field. The first, was the 1957 case study of patient H.M. (now known as Henry Molaison; Figure 1A) who suffered a complete anterograde amnesia and retrograde amnesia for more recent events following a bilateral medial temporal lobe resection to treat intractable medial temporal lobe epilepsy. Brenda Milner (Figure 1B), who is considered by many to be the founder of the field of neuropsychology, conducted the initial cognitive assessments of Mr. Molaison. Critically, Milner observed that not all memory functions were equally impaired (17). Molaison was able to acquire new procedural skills similar to normal controls, but he had lost all ability to acquire new memories for events (ie, episodic memory). The clinical evaluation of Molaison both highlighted that there were different forms of memory with distinct neural substrates and identified the hippocampus and medial temporal lobe structures (eg, the entorhinal cortex) as critical brain regions for supporting episodic memory (17). This put the hippocampus at the center of investigation for researchers interested in mechanisms of memory and why it declines in old age (see Annese et al. (18) for anatomical and histological images of patient H.M.’s hippocampal lesion).

Figure 1.

The anatomy of patient H.M. revealed the hippocampus as a structure critical to memory. (A) Patient H.M. also known as Henry Molaison courtesy of Wikipedia (19). (B) Researcher, Brenda Milner, PhD, who treated H.M. and pioneered the effort to understanding the role of the hippocampus in memory (20).

The second observation that was foundational for shaping the neurobiology of aging field was the 1971 discovery of long-term potentiation (LTP) (21). Neuroanatomists had long postulated that physical changes in brain structure could underlie new learning and memory (22,23). Working in the laboratory of Per Andersen in Oslo, Norway, Tim Bliss and Terje Lømo observed a lasting enhancement of synaptic responses recorded from the dentate gyrus (DG) granule cell region of the rabbit hippocampus following high-frequency stimulation of the perforant pathway fiber projection, which is the major input into the hippocampus originating in the entorhinal cortex (see Figure 2A for a hippocampal subfield reference). This enhanced synaptic efficacy was sufficiently long lasting to underlie a biological basis of memory (24). The final terminology for this phenomenon of modified synaptic strength following patterned electrical or modulatory stimulation became “long-term potentiation” (ie, LTP (25)). While LTP is an experimentally induced form of synaptic change that may not be directly relevant to the smaller and more selective modifications that follow learning and memory, it is still believed by many to reflect the plasticity capabilities of modifiable neural networks that support higher cognitive functioning.

Figure 2.

Long-term potentiation (LTP). (A) Anatomical representation of stimulating and recording electrodes in the hippocampal subfields, specifically, the dentate gyrus (DG). (B) Example traces of the change in millivolts over time induced by stimulation (observed by stimulation artifact) indicating the displacement of voltage due to movement of ionic currents from the outside to the inside of neuronal synapses. Top trace represents the synaptic field potential prior to high-frequency stimulation, and the bottom trace represents the synaptic field potential post high-frequency stimulation (HFS). Note the increase in slope steepness and amplitude of the synaptic field potential is indicative of synaptic strength. The threshold for inducing an increase in the field potential and the durability of the field potential change (or the length of time the change lasts) is considered the capacity and longevity of the network’s ability to be modified by experience. Scale bar represents voltage potential (mV) and time (ms). (C) Example of data showing the LTP of synapses. The field excitatory postsynaptic potentials is normalized to baseline, and the shaded area represents the standard error of the mean.

In the mid-1970s, working in the laboratory of Graham Goddard, Carol Barnes had the hypothesis that alterations in the plasticity capabilities of the aged hippocampus may be related to memory deficits in old animals. To test this, Barnes developed a test of spatial memory in which young and aged rats learned over the course of days to find the location of a small dark escape tunnel to avoid being exposed in an open field with a bright light. As rats are nocturnal, they prefer small dark spaces to open, bright areas. Aged rats were impaired at recalling the location of the escape tunnel. This test, known as the Barnes maze, is still widely used today to test spatial learning and memory. The same rats were then chronically implanted with recoding electrodes in the hilus of the hippocampus DG subregion and stimulating electrodes were placed in the perforant pathway fiber bundle that contains axons collaterals from the entorhinal cortex. This allowed evoked excitatory postsynaptic responses (ie, excitatory postsynaptic potentials) to be recorded over the course of weeks within the same animals. Following high-frequency stimulation of the perforant pathway repeated on 3 consecutive days, the older animals showed faster decay of LTP over 14 days. This decay in synaptic response amplitude was significantly correlated with the rats’ abilities to recall the location of the escape tunnel (26). This work was the first demonstration that hippocampal plasticity was related to cognitive function and that both hippocampal-dependent memory and LTP were impaired in old animals. Since this early report, many experiments have linked memory deficits to altered LTP and disrupted synaptic function across different hippocampal cell populations (16). Furthermore, the hippocampus and its closely connected structures in the parahippocampal region continue to be a focus of research into the neurobiology of cognitive aging.

Behavioral Testing Allows for Inferences of Brain Structure and Function

Cross-Species Neuroanatomical Homology

The cognitive evaluation of Molaison (17) as well as the work of other neuropsychologists with soldiers that suffered brain damage during World War II (27) established the important precedent that behavioral deficits following localized lesions to the brain could be used to link brain structure to function. Many of the same behavioral assessments that were developed to evaluate the impact of lesions on cognition have been applied to studies of aging to identify structures that become vulnerable in the later stages of life. Moreover, these tools have been readily adapted to nonhuman primate and rodent models of aging to highlight those functions relying on the hippocampus and prefrontal cortical structures are particularly sensitive to decline in old age. An important factor to consider in these studies and the translational potential of findings from animal models is the extent to which different brain regions show homology across species. The rodent and primate hippocampus and medial temporal lobe, which also includes the amygdala and parahippocampal region, show extensive homology with similar neuron populations, developmental profiles, and connectivity patterns being observed across species (28). The prefrontal cortex is also critical for higher cognitive functions and vulnerable to dysfunction in advanced age (29). In primates, this region has evolved substantially more than in rodents and occupies a larger proportion of total brain mass than it does in the rodent. Nonetheless, some researchers have argued that there is some homology between the lateral prefrontal cortex of primates and the rodent medial prefrontal cortex, which show similar functions, connectivity patterns, and receptor distributions across species (30), but see Laubach et al. (31) for a recent discussion of the nuances in relating the rodent and primate prefrontal cortex.

Dissociating Cognitive and Motoric Deficits in Aging

In addition to structural homology across species, when designing or evaluating behavioral experiments that assess cognitive function in older animals, there are several factors one needs to consider. First is the difference between cognitive deficits versus performance impairments. The best example of how performance differences could confound the interpretation of behavioral data is the influence of motor slowing, which ubiquitously occurs in older ages across all species, on tests of spatial memory. There are several age-related explanations for motor slowing that can be dissociated from cognition, for example, pain as well as atrophy from disease states. Indeed, motor slowing can precede age-related cognitive decline (32) and confound results. Spatial memory is often tested by an animal’s ability to find a learned target location in a maze. If one uses latency to infer an animal’s ability to recall a target location, then a slow aged animal with accurate memory may erroneously lead the experimenter to believe that the animal has memory impairments. For this reason, it is critical to use quantitative measures that control for differences in motor performance and movement speed, such as path length or cumulative search error (33). Relatedly, it is also critical to disentangle declines in basic sensory function from potential impairments in higher cognitive functions. For example, it is common for Fischer 344 rats, which are a widely used model in neurobiology of aging studies, to go blind at advanced ages (34). If the behavioral task involves the use of visual stimuli or cues, an aged rat with poor vision and accurate memory may be evaluated as having cognitive impairments. Thus, it is critical to include experimental controls or parameters that can evaluate and account for potential differences in sensory function. Additionally, aging may result in the use of alternative strategies to complete a task (35), and therefore, researchers should consider designing control trials or behavioral measurements that allow for the assessment of strategy implementation differences between age groups.

Motivational Factors in Behavioral Testing

Most cognitive tasks involve some form of motivated behavior by using food rewards in animals undergoing restriction or an aversive stimulus, such as a mild shock, that animals want to avoid. It is therefore critical that motivation is matched across age groups. For example, the Fischer 344 × Brown Norway F1 hybrid rat, which is available aged from the National Institute of Aging, shows dramatic differences in body mass between young (4–8 months) and old (18–30 months) age groups, with older ad libitum-fed rats commonly being morbidly obese. Most approved protocols from Institutional Animal Care and Use Committees to implement food restriction for encouraging appetitive behaviors involve a 15% reduction in weight. An overconditioned aged animal, however, may still be obese after a 15% weight reduction and not at the equivalent level of motivation as its younger counterpart. In our experience, it is more optimal to match young and aged rats by body condition (36) rather than % weight loss, which results in comparable levels of motivation to seek reward (37).

Maze-Based Spatial Tasks

A number of behavioral tasks have been developed to measure function of medial temporal lobe structures across the life span that control for the potential motor, sensory, and motivational confounds discussed above. Spatial episodic memory supported by medial temporal lobe structures, such as the hippocampus, is impaired in older adults (38). Several maze-based procedures have been employed to investigate age-related declines in spatial learning and memory in rodents and monkeys. Mazes typically utilize visual tracking software that can accumulate recordings for post-hoc data extraction and analysis so that measurements of path length or search error can be quantified. In the Barnes maze, rats or mice are trained over days to find the location of an escape tunnel located in the perimeter of a circular platform under a bright light. While the maze itself is typically devoid of any proximal spatial cues, distal cues located around the environment containing the maze allow for orientation and navigation (26). Aged rats and mice typically take longer to acquire the location of the escape tunnel (39) and are more likely to forget the correct location over time (26). In 1982, a comparable task was developed to examine the impact of hippocampal lesions on spatial learning and memory, referred to as the Morris water maze (40). In this task, rats or mice are placed in a large cold pool of opaque water and trained over days to find the location of a submerged escape platform based on distal spatial cues. Over the past several decades, many variants of the Morris water maze have been implemented that produce the consistent observation that aged animals are impaired at both learning and recalling the location of the escape platform (41,42). One note regarding the water maze is that in our experience, rats are only motivated to find the escape platform if the water is cold (<78 °F/25 °C). Similar tasks have been implemented in monkeys in which the location of a baited food well in the perimeter of a testing arena needs to be retained over a delay of up to 4 hours. While young (6.6 years) and aged monkeys (23–33 years) perform similarly at a short delay of 5 minutes, aged monkeys make significantly more errors than young animals at long delays (43).

Object Recognition Tasks

The ability to recall if a stimulus is familiar or novel (44), and to discriminate a novel stimulus from similar stimuli that have been previously experienced (45) also critically relies on medial temporal lobe structures. There are many types of tasks that assess this ability, all of which can be broadly grouped into tests of recognition memory. An effective way to assess this cognitive process in rodents is with the spontaneous object recognition task. This task tests rats’ abilities to differentiate a novel object from one that has been experienced and has the explicit advantage of not requiring any pretraining. This spontaneous object recognition task, first described in 1988, capitalizes on a rodent’s natural tendency to spend more time exploring novel objects relative to stimuli that have been previously encountered (46). In this task, a rodent is placed in a familiar testing arena with 2 identical novel objects and given several minutes to explore (Figure 3). After this familiarization phase, there is a variable time delay, and then during the test phase, the rat is placed back in the same testing arena but with 2 different objects in the arena. One object is a triplicate copy of the objects used during the familiarization phase (ie, a familiar object), while the other object is novel. The animal is given several minutes to explore the objects, and normal young rats will show significant novelty discrimination, spending more time exploring the novel object relative to the familiar object. Aged rats show a delay-dependent deficit on this task compared to young animals. Specifically, both age groups show a significant preference for the novel over the familiar object when the delay between the familiarization and the test phase is less than 2 minutes, indicating that sensory processing and motivation to explore are similar between age groups. When the delay is longer than 15 minutes, however, only young rats are able to discriminate between the novel and familiar object (47). Critically, this deficit arises from the aged rats exploring the novel object less, relative to the young rats, indicating that the old animals behave as if novel objects are familiar at delays greater than 2 minutes. In fact, control experiments have supported the notion that during long delays, rats encounter stimuli or other forms of interference that share common features with the novel object that will be presented during the test phase. The aged brain is less able to disambiguate the common features that were observed during the delay from those of the novel object, and therefore, the novel object is identified as being familiar (48). This idea is supported by 2 additional empirical findings. First, it is well documented that aged individuals are more vulnerable to the effects of interference relative to young subjects (49). Second, in a mouse model of Alzheimer’s disease in which behavioral deficits and false memory for novel objects has also been observed, deprivation of sensory input during the delay reinstates an ability to discriminate the novel from the familiar object (50). Interestingly, the misidentification of novel objects as familiar is also seen in scan paths of aged monkeys’ saccadic eye movements. When young monkeys are presented with a novel and familiar image, more spontaneous eye movements are made to the novel image. This is not the case in aged monkeys, which show a reduction in saccades to the novel image (51).

Figure 3.

Example of object familiarity task. Each box represents the same maze at different time points. After a period of habituation to the maze, rats are exposed to 2 objects. The rats are given a fixed amount of time to explore and become familiar with the objects. After a fixed delay (anywhere from 30 s to 24 h), the rats are then placed back in the maze and a novel object is placed alongside a familiar object. The time on each object is measures and because of rats’ preference for exploring novel objects, there should be a disproportionate amount of time spent exploring the novel object.

The notion that longer delays cause an inability of aged animals to discriminate between novel and familiar objects lends to the hypothesis that aged animals, like aged humans, have difficulty discriminating between objects that are similar even when mnemonic demand is limited. In fact, there is support for this idea. When young and aged rats are tested on a variant of the spontaneous object recognition task in which the familiar and novel objects share features, but the delay between familiarization and test phases is only 30 seconds, only young rats have a preference for the novel object. Similar to when the delays are long, the aged rats explore the novel object less, behaving as if it is familiar (52). The impact of feature similarity on object discrimination has also been modeled in monkeys and rats using LEGO blocks to construct objects in which the amount of overlap between pairs of objects can be parametrically manipulated. While aged and young animals are similarly able to discriminate between objects pairs with low feature overlap, at greater levels of overlap, aged monkeys (52) and rats (37) are impaired at discrimination. This same observation has also been made with human study participants (53). Together, these data from different object recognition tasks point to a consensus that, across species, aged animals have declines in object discrimination particularly when these discriminations are difficult.

Tests of Executive Function

Neuropsychological data from patients with brain damage unequivocally indicated that the prefrontal cortices were critical for executive functions, which include working memory, decision making, response inhibition, and other control or planning mechanisms that mediate and guide goal-directed behavior (29). The observation that older adults are impaired at many of these behaviors highlighted that the prefrontal cortex is vulnerable to dysfunction in advanced age (54). Similar to test of memory function, many of the tasks used to assess executive functions in humans have been adapted for monkeys. In nonhuman primates, it has consistently been reported that, compared to young animals, aged monkeys perform worse at tasks of working memory (55), set-shifting and reversal learning (56), and reward-based outcome evaluation (57).

Nonhuman primates have been an important model for understanding the neurobiology of age-related declines in executive functions due to homologous prefrontal cortical areas across monkeys and humans. Executive functions broadly refer to a class of behaviors that involve working memory, decision making, set-shifting, response inhibition, and motivation. Many of these functions rely on the dorsal lateral prefrontal cortex (58–60), and this area of the prefrontal cortex is consistently observed to show age-associated neurobiological changes that include morphological alterations and reduced synapse numbers (61–64). Decades of research has shown that aged monkeys (>20 years) are impaired relative to their younger counterparts on spatial working memory, which is the ability to keep a target location in mind over a brief period, and the magnitude of the deficit increases with the length of the delay (65,66). Another executive function that declines with age is conceptual set-shifting, or the ability to update a response based on changing task contingencies. For example, a monkey learns that an image of a certain color (eg, red) is rewarded over other colors. After a certain number of trials, the rule changes and the correct choice is based on shape (eg, triangle is correct) rather than color. While aged monkeys acquire the initial rule at a similar rate to young, they perseverate and do not update their responses as quickly as young animals when the rule changes (67). For scholarly reviews on the selective vulnerability of the prefrontal cortex in aged monkeys, please see Morrison and Baxter (68) and Upright and Baxter (69).

In rodents, operant-based tasks have been developed to examine how age impacts executive functions (Figure 4A–C). Operant chambers can be fully automated by software that also records task-related events, which are typically lever presses. The outcome of events of interest are recorded and analyzed as a measure of behavioral performance. One example of assessing prefrontal-dependent cognitive function with operant training is the delayed match-to-sample task (Figure 4B). This task assesses working memory dependent upon the prefrontal cortex (70). The task consists of 3 phases: sample, delay, and test. In the sample phase, one of 2 levers is extended into the chamber and the rodent must press the lever to proceed to the next phase. After the rat presses the lever, it retracts, and the delay phase begins. Delays can be set by the experimenter and can range between 0 and 24 seconds (or sometimes longer). Finally, in the test phase, both levers extend into the chamber, and the rat must choose the lever presented to it in the sample phase to be rewarded. Several studies have shown that with longer delays between the sample and test phases, aged rats have worse performance than young adult rats (71,72).

Figure 4.

Examples of operant tasks. (A) Schematic of the delay discounting task, illustrating the choices and trial blocks across which the duration of the delay to the large reward decreased. (B) Schematic of the delayed response task, illustrating the 3 phases of each trial. (C) Schematic of the set-shifting task, illustrating both initial (visual) discrimination and set-shift (left/right) discrimination trials.

Other tasks can be designed to test other aspects of brain function in aging, for example, cognitive flexibility (71) or decision making (73). Like humans and monkeys, aged rats show impairments in the ability to shift from one rule set to another. For example, when learning that a cue light over a lever indicates a correct choice to receive a food reward, both young and aged rats can acquire the rule within the same number of trials (71,72). However, when the rule shifts so that a cue light no longer represents the correct lever and, for example, lever location indicates the correct choice, aged rats take more trials to acquire the rule (Figure 4C). Additionally, aged rats (72,73), like older adults (74), have been shown to have a greater ability to delay gratification with the use of operant-based tasks. For example, when given the option between a lever that gives an immediate food pellet reward or a lever that gives a larger food reward but after a longer period of time, aged rats prefer larger, delayed rewards (Figure 4A).

Goals for Behavioral Testing

In contrast to spatial and episodic memory and executive functions, other cognitive domains such as crystallized intelligence and verbal memory appear to improve across the life span (75). Thus, the central goal of research examining the neuroscience of aging is to link cognitive decline to the underlying neurobiological mechanisms that account for these changes. A consistent observation that has been leveraged to identify underlying mechanisms of cognitive aging are the profound individual differences in how brain regions are impacted and how cognition is affected in old age in humans, monkeys, and rodents (76). This allows for cellular mechanisms to be directly linked to cognitive output within the same age group. Moreover, the fact that some individuals retain normal cognitive function across the life span while others do not highlights the potential to mitigate the development of behavioral impairments with age. While these are only a few examples of tools that have been used in cognitive aging, there are many other ways to implement mazes and operant chambers to study geroscience and to incorporate more sophisticated physical performance measures to examine the bidirectional relationship between cognition and peripheral health.

Methodological Approaches for Identifying the Neurobiology of Age-Related Cognitive Decline

Noninvasive Functional and Structural Imaging

Neuroscientists have used in vivo imaging techniques, such as structural and functional magnetic resonance imaging (fMRI) and diffusion tensor imaging, which can be used to assess white matter integrity, to understand mechanisms of cognitive aging. There is an extensive body of work that has identified age-related changes in the global brain connectome (77), major white tracks (78), gray matter volume (79), and activation patterns measured by blood oxygen levels in humans (80). Although several structures supporting specific cognitive domains are functionally impaired with aging, studies have shown that aging is also associated with a shift in the structure recruited during task performance (81). For example, when performance is equated between aged individuals and their young counterparts in a decision-making task, aged individuals appear to use their brain differently to perform the same task compared to young individuals (82). This highlights the brain’s adaptive ability, and this type of adaptive compensation may account for why some aged individuals are more resilient to cognitive impairment than others. More recently, magnetic resonance spectroscopy has been implemented to measure the levels of neurotransmitters, markers of neuroinflammation, oxidative stress, and metabolism in the brains of older adults (83).

While noninvasive imaging is always going to be limited by poor spatial and temporal resolution, these imaging approaches are starting to be widely used in rodent models of aging (84,85). This offers the opportunity to then perform invasive assays in the same animals to directly link imaging measures to underlying cellular mechanisms of brain aging. Recently, it was observed that increased functional connectivity between the medial prefrontal cortex and dorsal striatum in aged behaviorally impaired rats was associated with increased expression of the immediate-early gene Arc which is critical for synaptic scaling and cognition (85). The opportunity to enhance the translational potential of discovery science in animals by combining cellular measures with noninvasive imaging is just beginning to be fully realized. New advances in invasive 3-dimensional imaging with tissue clearing and lightsheet microscopy (86), which can be applied to both the brain as well as peripheral organs, will provide unprecedented spatial resolution that can be integrated with imaging to build a solid pathway to translation in humans. The major barrier to opening this new research frontier is the development of bioinformatic tools needed to derive meaning from large and elaborate multimodal datasets. Thus, geroscientists and neuroscientists should forge new collaborations with computer science and machine learning to pursue this new horizon of investigation.

In Vitro Electrophysiology

The primary means by which neurons communicate across the synapse is with electrical potentials that can be measured. Prior to the 1940s, not much was known about the electrical properties of the neuron. That would change when Alan Lloyd Hodgkin and Andrew Fielding Huxley developed the action potential theory by inserting electrodes into the nerves of a giant squid and recording the ionic currents generated (the squid giant axon (87)). Since then, neuroscientists have applied electrophysiological techniques to brain slices of animals to determine the intrinsic properties of healthy neurons and better understand how these characteristics deviate in pathology. Beyond understanding intrinsic biophysical properties, the contribution of different receptor types to neuronal signaling can also be assessed. This can be done at the level of single neurons via patch clamping techniques in which a recording electrode is physically attached to a single neuron, or in a larger population of neurons via recordings of the evoked voltage fields generated by stimulation. It was the latter approach that led to the seminal discovery of LTP described above (Figure 2).

Several studies using in vitro electrophysiological methods have contributed to the understanding of the neurobiology of aging. A remarkable aspect of this work is that most of the intrinsic biophysical properties of neurons in the hippocampus remain intact in old age (88). Importantly, the hippocampus is comprised of different subregions that include the DG, CA3, and CA1 (see Figure 2A). Each of these subregions has distinct neuron populations and connectivity patterns and do not show the same alterations as a function of age. Specifically, it has been shown that there is divergence in the intrinsic excitability of each region. For example, studies have shown that pyramidal neurons in CA1 are less excitable with age (89), whereas CA3 pyramidal neurons have increased excitability (90). Across all subregions, however, the dynamic properties of neurons, that is, their ability to undergo plastic change in response to stimulation, appears to decline in old compared to young rats (91). Recordings obtained from hippocampal brain slices have shown that LTP (a proposed mechanism for memory) is attenuated at many different synapses in aged animals relative to their younger counterparts (88). While less research on the intrinsic activity properties of neurons in the prefrontal cortex has been conducted, there is evidence for increased neuronal inhibition as a function of age (92).

In Vivo Electrophysiology

Neuroscientists have also used in vivo electrophysiology methods to correlate neural activity with behavior. In vivo electrophysiology makes use of surgically implanted microelectrodes in a brain region of interest to gather single unit recordings (ie, recordings of action potentials from a single neuron) or local field potential recordings, which is the aggregated synaptic activity across a local population of ≈1 000 neurons (93). The local field potential is comparable to the electroencephalogram that can be recorded from the scalp of humans (94). Since the first report of neuron recordings in an intact animal, in which Renshaw et al. recorded from the cortex of anesthetized cats with glass-insulated microelectrodes (95), the field of neurophysiology has vastly grown with the aid of new tools and techniques.

In the 1970s, O’Keefe and Dostrovsky discovered that neurons recorded from the hippocampus of freely moving rats had increased activity in certain locations of an open arena (96). They termed this pattern of activity the “place field,” and O’Keefe was awarded the Nobel prize in 2014 (along with May Brit Moser and Edvard Moser) for this discovery. In the decades to follow, in vivo electrophysiology increased in scope allowing hundreds of well-isolated neurons to be simultaneously monitored in behaving animals (97). Neurophysiological recordings have also been applied to studies of cognitive aging in both aged monkeys (98) and rodents (99,100), leading to many new insights regarding how alterations in the dynamics of neuron activity lead to cognitive decline. While a comprehensive description of the findings in this area are beyond the scope of this paper, a few key findings serve as examples of how this tool has been leveraged to better understand the neurobiology of aging. Firstly, it has been shown that relative to young animals, the activity properties of CA1 neurons recorded from aged rats are not as influenced by experience (101), and do not update in response to changes in the environment (16). Secondly, the firing rates of aged pyramidal cells in the CA3 are higher than in young rats (102). Notably, this observation has been replicated in monkeys using intracranial single-neuron recordings (103), and in humans with fMRI (53). Lastly, it has been reported that neuron firing in the prefrontal cortex during the delay phase of a working memory task is attenuated in aged compared to young monkeys (104).

Optogenetics

Although it is powerful, electrophysiological tools only allow for neural activity to be correlated with behavior. In contrast, optogenetics offers the ability to test the causal role of neuronal circuits in awake behaving animals by leveraging anatomical and temporal specificity. Optogenetics was pioneered by Karl Deisseroth and Edward Boyden (105). In this technique, a light-sensitive opsin gene is spliced into a plasmid between a neuron-specific gene promoter and a reporter gene, and the plasmid is then packaged into an adeno-associated viral vector (or lentivirus) for delivery into the brain region of interest. These opsins can allow for either activation or inhibition of neuronal communication. Specifically, channelrhodopsins are membrane channel proteins that allow the passage of cations (Na⁺ and Ca²⁺) upon activation with light (in the range of 450–500 nm). This light-evoked current depolarizes the neuronal membrane, causing the neuron to fire action potentials. Alternatively, halorhodopsins are membrane channel proteins acting as pumps that allow the passage of anions (Cl^–) upon activation with light (in the range of 550–600 nm). This anionic current hyperpolarizes the neuronal membrane making it less excitable and therefore less likely to fire an action potential. Once delivered to a brain structure, approximately 2 weeks is needed for complete expression of the opsin gene. Then, an optic fiber is implanted into the same brain structure to allow light delivery from either a laser or LED (106). In awake behaving animals, light delivered to a nuclei of neurons can either turn on or off these neurons to cause behavioral changes in a temporally specific manner that allows neuroscientists to determine function in real time (106). Importantly, the actions of one structure onto another, or the communication between regions and their behavior outputs, can be dissected by modulating the activity of the fiber terminals of neurons transduced with opsins (106).

Studies using optogenetics to understand brain aging should account for the aged brain’s tendency for seizures (107). Indeed, while epileptic onset can be restrained by use of halorhodopsins to inactivate neuronal communication, stimulation frequency and duration when using channelrhodopsins to activate neuronal communication should be carefully titrated to avoid epileptogenesis. To date, only one study has employed this optogenetic approach to understanding potential differences in the temporal dynamics of neural activity that shift as a function of age. Specifically, the role of amygdala activity in distinct components of decision making was evaluated in aged (26 months) and young (6 months) rats by inhibiting activity during different phases of an intertemporal choice task in which animals had to choose between a small immediate reward or a large reward that was given after a delay. Because optogenetics allows for precise temporal control of activity, the amygdala could be inactivated during choice deliberation or during reward outcome. This study reported that when the amygdala was inactivated during the deliberation phase, both young and aged rats preferred the large delayed reward, that is, they were less impulsive. In contrast, when the amygdala was inactivated during the delivery of the small immediate reward, only the young rats showed an increase in impulsive choice, while the behavior of the aged rats was unaffected (108). Notably, this is an example of aging resulting in the differential recruitment of a brain structure when performance is equated between young and aged rats, as had been previously observed in humans (82). While this remains the only study to our knowledge done in young and aged awake behaving animals, there are others soon to be published leveraging optogenetics to dissect the contributions of neurons to cognitive function across the life span. More recently, an optogenetic approach was also used in a genetically engineered strain of Escherichia coli that caused this bacterium to secrete colanic acid under the quantitative control of light. Green light was then used to control the expression of colonic acid in the Caenorhabditis elegans gut, which protected intestinal mitochondria and extended life span (109), thus exemplifying how a tool that was developed to manipulate brain circuits could be applied to understanding the biology of aging.

Pharmacological Approaches Are a Bridge Between Translational and Basic Research

In neuroscience, there are several pharmacological approaches to promote basic discovery in drug research. For one, using drugs to directly target a specific brain region in awake behaving animals can help neuroscientists determine how the receptor distribution contributes to regional functionality. This approach requires the placement of infusion cannula during brain surgeries to facilitate direct delivery of drug. After recovery, the animal subjects can begin behavioral testing. Prior to a test session, the drug can be delivered directly to the region of interest, and depending on the experimental design (ie, between-subjects vs within-subjects), the effects of drug dose on the measure of interest is compared with the effect of the drug’s vehicle on the same measure. This intracerebral delivery of drug has allowed neuroscientists to understand how receptors in a specific brain area contribute to age-related cognitive alterations. For example, directly infusing a drug that activates receptors involved in increasing neuronal inhibition into the prefrontal cortex improved aged rats’ cognitive performance relative to a vehicle control infusion on a set-shifting task (110).

Though intracerebral delivery of drug is advantageous to the understand receptor function in a targeted structure, a systemic approach is more relevant to potential clinical interventions and can make the results more generalizable to humans. Thus, studies in aged animals have also administered drugs systemically with an injection or orally in the food or water. For example, one study using rats to model age-related cognitive decline reported that systemic injection of a drug that antagonizes receptors associated with neuronal inhibition improved aged rats performance relative to vehicle on a working memory task (111). With systemic administration of drug paired with behavioral testing, many studies have shown improved cognitive outcomes (112). Although systemic drug delivery methods are standardized to milligrams of drug per kilogram of body weight, it is possible that decreased hepatic and kidney function coupled with the increased adiposity of aged animals affects the functional outcome of administered drug doses (113–115). As such, systemic drug efficacy observed in young animals may not translate to aged animals, but not necessarily due to drug inefficacy. These types of systemic drug studies have commonly been used in geroscience to examine the modulation of life span by different drugs, for example rapamycin (116), and they have translational potential for studies in humans.

Potential Areas for Synergy Between Geroscience and Neuroscience

Neurobiological Dysfunction and Aging Biology Are Driven by Similar Insults: The Role of Metabolism and Oxidative Damage

Both in the brain and the periphery, aging is associated with oxidative stress (117), inflammation (118), and cell senescence (119). Typically, these insults do not necessarily occur uniformly throughout the brain. For example, markers of oxidative stress and inflammation are predominant in the prefrontal cortex and hippocampus (120). Specifically, several studies investigating the role of oxidative stress and damage on hippocampal-dependent cognitive aging have shown that while oxidizing agents can reduce responses at CA3–CA1 synapses in young rats, reducing agents can enhance these responses in aged rats, and these responses are dependent upon receptors involved in synaptic plasticity and LTP (121). Additionally, there is some evidence suggesting cell-specific senescence in the prefrontal cortex (122), whereas cell senescence appears to play a smaller role in the hippocampus (123).

The brain is particularly vulnerable to energy imbalances, as it requires over 20% of the body’s energy demands (124). Because the relative rate of increase in oxygen and glucose consumption is greater for the brain than it is for the rest of the body, the evolution of increased brain volume has a disproportionate impact on the body’s metabolic budget in total (125). Cognitive decline in advanced age and Alzheimer’s disease is tightly linked to metabolic dysfunction (124) and glucose hypometabolism is implicated in epilepsy (126), which shares biological features with advancing age. Interestingly, age-related changes to brain activity have been observed in humans (53), monkeys (103), and rats (102) and antiepileptic drugs have been shown to rescue the behavioral deficits associated with activity imbalances in aging (127). Indeed, energy imbalances may be a common mechanism of dysfunction across nonpathological brain aging. Given the vulnerability to metabolic dysfunction in older individuals, it is not surprising that other common problems associated with advancing age include the high occurrence of diabetes type II (128) and obesity (129). Notably, obesity and diabetes are both associated with cognitive decline (130).

Rodent Models of Pain in Aging

Understanding the neural pathways involved in chronic pain in aging is another research area that could benefit greatly from an intersection between neuroscience and geroscience. Chronic pain disproportionately affects older adults and is associated with increased inflammation and cognitive decline (131). Several brain nuclei form an intricate network of pain perception. This network includes cortical and subcortical structures that receive afferent fibers from the peripheral nervous system (132). It is not surprising that structural and functional changes in the aged brain alter chronic pain outcomes. Indeed, it is thought that chronic pain in older adults may result from a reduction in the efficacy of analgesic systems in the body and brain (133), although descending inhibitory fibers also decrease with age (134).

Neuroscientists are currently studying the effects of neuropeptide signaling on pain including endorphins (eg, β-endorphin, enkephalin, dynorphin) and their opioid receptors located in several nuclei of the pain network. Rodent models of pain in aging have validated mechanisms for age-related increases in pain. For example, the concentration of opioid receptors declines with age in several brain structures associated with pain (135). Furthermore, the concentration of opioids endogenously made in the nervous system (136), specifically enkephalins and substance P (137), and beta-endorphins (138) also decline in advanced age. Recently, endocannabinoids have also shown promise in pain research as an alternative to the highly addictive opioids in pain medication (139). In this regard, greater communication between neuroscientists and geroscientists could broaden our understanding of treating pain in aging.

Epigenetics

One key area of study that could greatly benefit from more overlap between neuroscience and geroscience is epigenetics. While the genome contains the hereditary information of an organism, the epigenome is the modifiable access to that information (140–143). As such, aging biologists have categorized epigenetics as one of the hallmarks of aging due to the nature of its plasticity (144). That is, the modifications to the genome that can occur over an organism’s life span is a key mechanism of cellular aging. In early life, cells of the same type have similar patterns of gene expression, and through the process of aging, these patterns of gene expression are subject to change via transcriptional drift or genomic instability (145–147), and the degree of the change is dependent on both internal (ie, biological) and external (ie, environmental) factors (for review, see Pal and Tyler (148)). Considering that neuronal number is, with little exception, unchanged in the aged brain, and that other theories of brain aging such as deficits in synaptic remodeling and neuronal activity have emerged as more appropriate explanations, the brain is an environment ripe for epigenetic modifications. Indeed, the processes underlying learning and memory are subject to epigenetic modifications (149).

Several studies have linked epigenetic changes in the brain due to aging with increased risk of neurodegenerative disorders like Alzheimer’s disease. Furthermore, early-life stress or psychosocial stress result in modifications to DNA methylation that are predictors of life span and health span. In older adults, several epigenetic markers have been linked to brain volume and hyperintensities during brain imaging (150). Recently, rodent models of aging and neurodegeneration have demonstrated that targeting epigenetic mechanisms can have protective effects via histone deacetylases (151–153). These data suggest that experience-based changes in the brain accumulated over time impact the overall trajectory of brain aging as an active mechanism instead of a passive chronological aging event.

Brain and Peripheral Nerve Stimulation

Since the mid-1900s, the implantation of electrodes into the brain has been used for deep brain stimulation to modify clinical outcomes in patients with various diseases (154). Several disorders are still treated by the use of deep brain stimulation including chronic pain (155), neuropsychiatric disorders (156), and neurodegenerative disease (157). While this invasive approach is only feasible in clinical cases where a patient’s daily living is severely disrupted (eg, Parkinson’s disease), noninvasive brain stimulation techniques have been used to probe the function and plasticity of neural networks. These approaches have shown promise for improving cognitive function in older adults without pathology by normalizing activity and metabolism in brain areas affected by aging.

Transcranial magnetic stimulation (TMS) uses an externally placed device that pulses an electric current through a copper coil to generate a magnetic field that is perpendicular to the plane of the coil. The rapid rate of change of the magnetic field induces a secondary electric field in the underlying brain tissue with the potential to depolarize neuronal membranes to directly induce action potentials (158). The majority of TMS studies in relation to aging have been focused on examining motor cortex control over muscle contraction. This approach has identified age-related differences in motor plasticity (159), and could provide utility for examining brain–muscle interactions across the life span. Repetitive TMS has also been used to treat cognitive deficits in advanced age. In older adults with mild cognitive impairment (MCI, which is a preclinical stage of Alzheimer’s disease), repetitive TMS over the left dorsal lateral prefrontal cortex for 5 days improved inhibitory control relative to sham stimulation (160). In another study, older adults with MCI that received repetitive TMS over the left dorsal lateral prefrontal cortex for 10 sessions had significant improvements in episodic memory and executive function poststimulation compared to individuals that received sham stimulation, and this improvement was retained at a follow-up assessment 1 month later (161).

While TMS directly leads to the generation of action potentials in cortical tissue, transcranial direct current stimulation (tDCS) is an alternate noninvasive brain stimulation method that is hypothesized to achieve neuromodulation through subthreshold alteration of resting membrane potentials using direct electrical stimulation. This approach is safe to use and has shown potential for enhancing cognitive processes (158). In healthy older adults, tDCS paired with working memory training in which the anode electrode was placed over the right dorsal lateral prefrontal cortex and the cathode was placed over the contralateral cheek, led to improvements on the trained verbal and visual working memory tasks. Moreover, there were significant transfer benefits to cognitive domains that were not trained with tDCS (ie, processing speed, cognitive flexibility, and arithmetic) at 1 month after the intervention. This was not observed in individuals that received sham stimulation (162). In a similar study, healthy older adults received 2 weeks of tDCS 5 days per week during computer-assisted cognitive training for 30 minutes per day. Using 2 stimulators, tDCS current was delivered over the bilateral prefrontal cortex and 2 cathodes were placed on the nondominant arm. Significant improvements in digit span forward tests were seen in the stimulated compared to sham group that persisted for 28 days postintervention (163).

The mechanisms by which TMS and tDCS may lead to cognitive improvements in older adults remain to be determined. Researchers have postulated, however, that one potential means by which these treatments are effective for improving cognition is through the activation of peripheral nerves that have extensive projections to the brain as well as other organ systems (164). In line with this notion, vagus nerve stimulation (VNS) has also been proposed as a potential intervention for cognitive dysfunction. The vagus nerve contains efferent fibers with cell bodies in the motor nucleus of vagus in the medulla that widely project to the viscera. This nerve also contains afferent fibers from peripheral ganglia that synapse in the nucleus of the solitary tract, which then projects to several neuromodulatory structures that include the locus coeruleus (165). The locus coeruleus is the primary brain structure that synthesizes and releases noradrenaline throughout the brain, which is a critical neurotransmitter for arousal and plasticity. The locus coeruleus is also among the first brain regions to show Alzheimer’s disease pathology (166), and its function is reported to decline in advanced age even in the absence of pathology (167). For this reason, it is hypothesized that VNS might improve cognition in older adults. Notably, VNS has been shown to reduce inflammation (168) and promote plasticity (169), both of which become dysregulated in advanced age. While VNS has not yet been evaluated as an effective treatment for aging, it is used to treat irritable bowel syndrome (170). Moreover, this nerve is at the interface of the gut–brain axis and targeting it could restore homeostasis in the microbiota–gut–brain axis (see below) that becomes comprised in old age (170).

One challenge with each of these stimulation approaches is narrowing down the expansive space of different stimulation parameters that could be effective. This endeavor is further complicated by potential individual variability. Specifically, subcutaneous fat, skull thickness, and peripheral nerve anatomy are all going to impact how much current reaches the brain or a target nerve. Thus, it would be incredibly advantageous to have a peripheral marker of target engagement that could be used to precisely tune stimulation parameters for each individual. Given that cranial nerve axons project directly to the brain, targeting structures that are critical for higher-order behavior, such as the locus coeruleus, as well as peripheral organ systems, potent effects on cognition, inflammation, and metabolism are all plausible. Thus, therapeutic noninvasive stimulation is an area of synergy between neuroscience and geroscience that is likely to be high yield for improving the health span.

The Gut–Brain Axis

One area of study that has gained traction in recent years is the link between the gastrointestinal and nervous systems. There is a reciprocal link between the gut and the brain, commonly referred to the gut–brain axis (171–173). For example, the microbiome directly influences the generation of neurotransmitters (174,175). Furthermore, it has been shown that mice lacking a gut microbiome have cognitive impairment (176). These mice also have decreased expression of brain derived neurotrophic factor (177), and often, though not always, express decreased anxiety-like phenotypes and have decreased social cognition (178), thus showing cognitive changes and neuropsychiatric disorders are subject to gut microbiota alterations. Indeed, there have been several recent reports on the strength of the association between the composition of the gut microbiome and cognitive function (171), and notably, while there are links between gut microbiota producing markers of cognitive aging (179), it has been shown that age-related cognitive decline is associated with gut microbiota disorders (180). Moreover, transplant of an aged microbiome is sufficient to cause cognitive impairment in young mice (181). Furthermore, deviations from healthy gut microbiota are implicated in Alzheimer’s (182) and Parkinson’s disease (183).

While there are many potential avenues for the aging gut to impact the aging brain (or cognitive aging), this relationship is not unidirectional. There is also evidence that impaired brain function can facilitate poor gut health. For example, injuries to the forebrain result in gastrointestinal tract dysfunction via increased permeability (184). As such, one key area geroscientists may want to expand into is the effects the aging brain on the aged gut and the periphery in general.

Immunosenescence

The immunologic changes that occur during immunosenescence have been well characterized and include increased susceptibility to infection, autoimmune and inflammatory reactions, and reduced immune cell numbers and responses (185). For most of the history of neurobiology, the brain had been considered immune privileged (186–188). The blood–brain barrier and lack of access to the lymphatic system are 2 properties that help shield the central nervous system (CNS) from peripheral immune responses and the triggers that induce these responses. Recent studies, however, have leveraged more sensitive techniques to begin characterizing a homologue to the lymphatic system in the brain, termed the glymphatic system (189,190). As such, novel areas of study would benefit from integrating what is known about the aging immune system with the potential impact of immunosenescence on brain aging.

Although the brain lacks a comprehensive immune system, resident microglia represent a unique myeloid cell population that direct innate immune responses in the CNS. Microglia monitor the local CNS environment during their quiescent state by actively extending and retracting their processes which allows movement within the CNS. Interestingly, in a mouse model of accelerating aging (senescence-accelerated mice), these processes were reduced in length and number (191), which is a morphological defect shown to precede neurodegeneration (192). Similar to young adult immune cells of the periphery such as macrophages, young microglia maintain a balance in pro- and anti-inflammatory responses with the capacity for transient and robust responses to CNS insults (193). However, unlike senescent macrophages that undergo age-related decreases in functional proinflammatory responses, senescent microglia have an increased and sustained proinflammatory response characterized by production of cytokines such as tumor necrosis factor-α, interleukin (IL)-1β, and IL-6 and concomitant suppression of anti-inflammatory factors like IL-10 (193–195). Astroglia represent another cell population in the CNS that are immunocompetent and produce proinflammatory responses when activated (196). Although less explored than microglia, recent evidence indicates that aging astrocytes acquire a reactive and proinflammatory phenotype (197). These immunosenescent shifts are thought to contribute to increased neuronal toxicity, diminished neuronal protection, and neuronal plasticity (192). Though several researchers have already begun to study the aging brain within the context of immunosenescence, the field is primed for, and can be advanced by, greater integration of neurobiology of aging and immunology; a strategy that can benefit from increased collaboration between neuroscientists and geroscientists.

Conclusions

There is a deep and rich history of aging neurobiology. Throughout that history, neuro- and geriatric sciences have remained on 2 banks of a river. Along with its companion articles (3,4), this primer and minireview seeks to encourage the construction of a bridge across these metaphorical banks through greater communication and collaboration that will inevitably produce new hypotheses to enrich both the fields of neuroscience and geroscience. Many of the techniques and methodologies presented here have been fruitful assets leveraged by neuroscientists. It is reasonable that many, if not all, of these methods and approaches can be equally fruitful in the field of geroscience, and one way to address the gap between the fields is for neuroscientists to begin to assist geroscientists in the translation and application of these approaches potentially via the founding of local (or global) institutes or centers that seek to facilitate the integration of neuroscience and geroscience.

Funding

The work was supported by the National Institutes of Health (NIH)/National Institute of Child Health and Human Development (NICHD), 2T32HD071866-06 (C.M.H. and A.R.H.) and the Evelyn F. McKnight Brain Research Foundation, NIH/National Institute on Aging (NIA) RF1AG060977 (S.N.B.).

Conflict of Interest

None declared.