Transformations in oscillatory activity and evoked responses in primary somatosensory cortex in middle age: A combined computational neural modeling and MEG study
Introduction
Healthy aging is accompanied by widespread and systematic neuroanatomical changes (Raz and Rodrigue, 2006, Walhovd et al., 2009) that have been linked to alterations in perceptual and cognitive abilities in older adults (Ziegler et al., 2008). In addition, MEG and EEG studies have revealed age-related changes in brain dynamics on fast (ms) time-scales, such as changes in ongoing neural oscillations and in stimulus-locked evoked responses (d'Onofrio et al., 1996, Kolev et al., 2002, Kononen and Partanen, 1993, Marciani et al., 1994, Roubicek, 1977, Babiloni et al., 2006, Niedermeyer, 1997, Rossini et al., 2007). Less is known about the effects of middle age on these phenomena.
Most of the work examining age related changes in neural rhythms has focused on the spontaneous or resting state ‘alpha rhythm’ measured with EEG, typically defined as 7–14 Hz activity. In EEG sensor data, a generalized decrease in alpha power occurs with increasing age in parietal, temporal, and occipital cortices (Babiloni et al., 2006, Niedermeyer, 1997, Rossini et al., 2007). Further, older participants show decreases in alpha and low beta power (13–25 Hz), but increased high beta power (25–30 Hz) (Roubicek, 1977). A contradictory pattern emerges from studies of task-related rhythms, which show age-related decreases in alpha power in posterior sources, but increases in alpha power in anterior sources (Niedermeyer, 1997). Such a shift has been documented in the auditory (Yordanova et al., 1998) and visual (d'Onofrio et al., 1996, Kolev et al., 2002, Kononen and Partanen, 1993, Marciani et al., 1994) systems, with many of these changes seen in middle age. Few studies have examined the effects of aging on the commonly observed somatosensory mu rhythm, which is composed of a complex of alpha and beta components (Hari, 2006, Jones et al., 2009). One EEG study used a finger extension task and found age-related increases in mu-alpha and high mu-beta (22–23 Hz) power in anterior sensorimotor electrodes (Sailer et al., 2000).
Other studies have examined age-related changes in evoked responses driven by median nerve stimulation and have found increases in the amplitudes and latencies of somatosensory-evoked potentials with advancing age (Adler and Nacimiento, 1988, Desmedt and Cheron, 1980, Ferri et al., 1996, Kakigi and Shibasaki, 1991, Luders, 1970, Ogata et al., 2009, Stephen et al., 2006). These age-related increases in peak magnitudes occur in the first 100 ms following median nerve stimulation, with the greatest differences observed between 40 and 90 ms. Many of the changes are apparent by middle-age. The neural mechanism underlying these changes is typically attributed to a decline in cortical inhibition with advancing age (Drechsler, 1978, Simpson and Erwin, 1983, Stephen et al., 2006), but no direct evidence supports this conclusion.
An expanding body of literature has shown that ongoing brain rhythms are causally related to changes in evoked response activity (Jones et al., 2009, Mazaheri and Jensen, 2008, Nikulin et al., 2007, Zhang and Ding, 2009), and that these modulations are correlated with changes in behavioral states, such as attention and perception (Fries et al., 2001, Linkenkaer-Hansen et al., 2004, Zhang and Ding, 2009). Despite the growing accumulation of studies showing age-dependent changes in neural rhythms, evoked response activity, and cognitive abilities, an open question is whether these measures change over the lifespan (e.g., from young to middle age). Further, there is little mechanistic understanding of the underlying neurophysiological changes related to human aging. An obvious difficulty in achieving a mechanistic understanding is that the microscopic neural activity underlying macroscopically measured MEG/EEG signals is largely unknown and difficult to derive in humans without invasive recordings. Biophysically principled neural models can be used as powerful non-invasive tools to study the underlying neural dynamics generating these signals (Jones et al., 2007, Jones et al., 2009, Murakami et al., 2003, Murakami and Okada, 2006, Okada et al., 1997).
We have recently developed a neurophysiologically grounded laminar neural model of primary somatosensory cortex (SI) that accurately reproduces spontaneous somatosensory mu rhythms and tactile evoked responses measured with human MEG (Jones et al., 2007, Jones et al., 2009). The model contains a network of morphologically and physiologically principled excitatory pyramidal neurons and inhibitory interneurons spanning multiple cortical laminae. The model includes extrinsic feedforward and feedback excitatory synaptic inputs, defined by their laminar location of postsynaptic effects. These inputs represent feedforward (FF) lemniscal thalamic input to the granular layers of SI, and feedback (FB) input to the supragranular layers from intracortical sources or nonspecific thalamic nuclei (Felleman and Van Essen, 1991, Jones, 2001). This model can be used to identify the changes in network parameters, which have a direct neurophysiological interpretation, that produce observed changes in MEG-measured SI dynamics.
Here, we describe novel initial analyses of the impact of healthy aging on somatosensory dynamics, using a data set collected to address more general questions of dynamics, perception and rhythmogenesis in neocortex (Jones et al., 2007, Jones et al., 2009). We then applied computational modeling to disambiguate possible interpretations of these data. Specifically, we used MEG to examine age-related changes in spontaneous mu rhythms and tactile evoked responses from localized SI activity in healthy young adults (YA) and early middle aged adults (MA). Our main findings from the MEG data are that the beta component of the spontaneous mu rhythm is greater in MA and that the ∼ 70 ms peak (M70) in the tactile evoked response is also enhanced. These results provide initial evidence that both prestimulus SI mu rhythms and tactile evoked response magnitudes are increased in MA, compared to YA.
We then applied our biophysically realistic computational model of SI to predict the neural mechanisms underlying these changes. Our previous model results showed that the SI mu-rhythm could be reproduced by driving the SI network with two alternating ∼ 10 Hz inputs that contacted SI in FF and FB connection patterns, respectively. The relative alpha to beta power expressed in SI depended on the delay between and relative strength of the inputs (Jones et al., 2009). Based on these previous model results and our current MEG findings, we set forth to test two alternative predictions for the observed increase in mu-beta in MA: (A) that mu-beta dominance in MA, relative to YA, results from a decrease in the delay between 10 Hz FF and FB inputs, or that (B) the FF–FB delay is the same in both groups and the mu-beta dominance arises from “stronger” FB inputs. Comparison of model results and MEG data supported prediction (B), leading to specific hypotheses about the cellular-level neural events that intimately link modulations in ongoing rhythmic activity with evoked response magnitudes in MA.
Section snippets
MEG experiment
MEG data were collected from 10 right-handed healthy adults aged 23–43 years (mean = 31 years; six females) with no known neurological or psychiatric conditions during performance of a tactile detection paradigm. All of the participants in our study had either obtained a PhD degree or were enrolled in a doctoral-level graduate program at the time of the study, yielding a relatively homogeneous cohort in terms of educational demographics. In addition to MEG recordings, a high-resolution
Mu rhythm changes with age
To examine the effects of middle age on the spontaneous mu rhythm in SI, we computed the average power in the calculated TFRs over all trials for the entire mu range (7–29 Hz) from the 1000 ms prestimulus time period, as well as for the mu-alpha (7–14 Hz) and mu-beta (15–29 Hz) subcomponents. All references to power in our results were derived from this TFR calculation and the same analysis was used in all participants, except in Fig. 1C where power spectral densities (PSDs) were calculated
Discussion
The present study compared the spontaneous SI mu rhythm and tactile evoked responses in YA and MA using MEG, and used a computationally realistic laminar model of SI to make specific predictions about the underlying neurophysiological bases for the observed age differences. To document changes that occur during middle age, we restricted our study to healthy participants between the ages of 23 and 43 years of age. Linear regression and group comparison analyses of the MEG data revealed a
Acknowledgments
The authors wish to thank Michael Hines for excellent technical support in NEURON software code. This work was supported by NIH: P41RR14075, K25MH072941, 1RO1-NS045130-01, T32 GM007484, NSF: 0316933, the Athinoula A. Martinos Center for Biomedical Imaging, the McGovern Institute for Brain Research.
References (72)
- et al.
Age-related synaptic changes in sensorimotor cortex of the Brown Norway × Fischer 344 rat
Brain Res.
(2000) - et al.
Aging of subcortical nuclei: microstructural, mineralization and atrophy modifications measured in vivo using MRI
Neuroimage
(2009) - et al.
Somatosensory evoked potentials to finger stimulation in healthy octogenarians and in young adults: wave forms, scalp topography and transit times of parietal and frontal components
Electroencephalogr. Clin. Neurophysiol.
(1980) - et al.
Age-related changes in neural activity during performance matched working memory manipulation
Neuroimage
(2008) - et al.
Neural correlates of auditory recognition under full and divided attention in younger and older adults
Neuropsychologia
(2006) - et al.
Scalp topographic mapping of middle-latency somatosensory evoked potentials in normal aging and dementia
Neurophysiol. Clin.
(1996) - et al.
Thalamic relay functions and their role in corticocortical communication: generalizations from the visual system
Neuron
(2002) - et al.
Prestimulus oscillations predict visual perception performance between and within subjects
Neuroimage
(2007) Action–perception connection and the cortical mu rhythm
Prog. Brain Res.
(2006)The thalamic matrix and thalamocortical synchrony
Trends Neurosci.
(2001)