Visual function in humans degrades during the early stage of senescence beginning from middle 50s to 60s. To identify its underlying neural mechanisms, we investigated the aging effects on the primary visual cortex (V1) cells in early senescent (ES) monkeys (Macaca mulatta). Under anesthesia, receptive field properties of V1 cells were examined by extracellular single-unit recordings in the young adult (YA; 5–6 years old), ES (19–24 years old), and late senescent (LS; 28–32 years old) monkeys. We found clear indications of functional degradation in early senescence, including impaired stimulus selectivities, increased level of spontaneous activity and declined signal-to-noise ratio, and dynamic range of V1 cell responses. Importantly, the functional degradation in early senescence exhibited unique features that were different from the results for the LS animals, such as remarkable individual variability in orientation selectivity and unchanged peak response elicited by visual stimulation. Our results demonstrate that the function of V1 degrades during the early stage of aging in nonhuman primate, suggesting potential neural correlates for functional deficits observed in early senescence in human subjects. Moreover, these results provide new insight into the dynamics of the aging-related functional deterioration, revealing a more complex and heterogeneous picture of this process.
Visual abilities are affected adversely by aging. Old people exhibit a wide variety of functional deficits (for review, see Spear 1993; Owsley 2011), including, but not limited to, decreased contrast sensitivity (Owsley et al. 1983; Tulunay-Keesey et al. 1988; Elliott et al. 1990, 2009; Delahunt et al. 2008), degraded abilities of shape discrimination (Owsley et al. 1981; Wang 2001; Betts et al. 2007; Delahunt et al. 2008; Barnes et al. 2011), and motion perception (Trick and Silverman 1991; Norman et al. 2003, 2010; but see Betts et al. 2005; Bennett et al. 2007; Billino et al. 2008), as well as slowed visual processing speed (Salthouse 1993; Geldmacher and Riedel 1999).
It is known that the aging-related changes in optics of the eye cannot fully explain these functional degradations (Owsley et al. 1981; Spear 1993; Elliott et al. 2009; Owsley 2011). Also, previous studies have found that the retina and dorsal lateral geniculate nucleus of old nonhuman primate are relatively normal (Ahmad and Spear 1993; Spear 1993; Spear et al. 1994; Kim et al. 1996; Schmolesky et al. 2000). Thus, to fully understand the mechanisms underlying the declines in visual functions associated with senescence, it is necessary to examine functionality of visual cortex in the process of normal aging. Indeed, we have previously reported that a wide range of functions of cells in multiple visual cortices degrade in senescent monkeys and cats (Schmolesky et al. 2000; Leventhal et al. 2003; Wang et al. 2005; Yu et al. 2005, 2006; Hua et al. 2006; Yang et al. 2008; Zhang et al. 2008; Yang, Liang et al. 2009; Yang, Zhang et al. 2009; Fu et al. 2010; Liang et al. 2010). Specifically, in the primary visual cortex (V1) of monkeys, we have found that the selectivities for orientation, direction, spatial, and temporal frequencies, as well as the surround suppression of cells are all impaired in old animals. Additionally, these changes are accompanied by increased visual responsiveness, increased level of spontaneous activity, decreased signal-to-noise ratio (SNR) (Schmolesky et al. 2000; Leventhal et al. 2003; Hua et al. 2006; Zhang et al. 2008; Fu et al. 2010), prolonged response latency (Wang et al. 2005; Yu et al. 2005), and increased trial-by-trial variability (Yu et al. 2005; Yang, Liang et al. 2009). These aging-related changes in response properties of cortical cells provide possible neural mechanisms underlying visual deficits observed in aged humans.
The senescent monkeys used in most of previous studies (e.g., Schmolesky et al. 2000; Leventhal et al. 2003; Wang et al. 2005) were 26–32 years old and the degree of senescence, according to the lifespan, corresponded to 80–95 years of human age (Tigges et al. 1988). However, in human subjects, the degradation of visual abilities can be observed during the early stage of aging, that is, 60–70 years old or even earlier (Gittings and Fozard 1986; Wood and Bullimore 1995; Fahle and Daum 1997; Jackson et al. 1998; Tran et al. 1998; Porciatti et al. 1999; Betts et al. 2005; Del Viva and Agostini 2007; Billino et al. 2008; Kennedy et al. 2009; Weymouth and McKendrick 2012). It is important, both practically and theoretically, to investigate the neural mechanisms underlying the functional decline that occurs at the early stage of senescence. Practically, early senescent (ES) people consist of a much larger population (United Nations 2011), and they are more engaged in various daily activities that depend on vision, for example, driving, walking, and so forth (Sivak and Schoettle 2012). Therefore, examining the neuronal correlates of their visual degradation and identifying the effective ways to prevent or delay such decline will be essential for improving the everyday function and well-being for elder population. Theoretically, aging of the visual system, like aging in general, is a dynamical process (e.g., see Tran et al. 1998; Wist et al. 2000; Bennett et al. 2007; McBain et al. 2010). To look at different stages of this process will shed valuable light on the causal links among various aspects of the functional degradation and, eventually, provide better understanding about how aging affects vision.
Here, we studied functions of V1 cells, including the orientation/direction selectivities, spontaneous, and evoked activities, as well as the SNR, in monkeys during the early stage of senescence by extracellular unit recordings. For comparison, we also studied the same properties in young adult (YA) and late senescent (LS) monkeys.
Materials and Methods
All experiments were carried out in accordance with NIH guidelines for animal care and use. Subjects for this study were 14 male rhesus monkeys (Macaca mulatta) classified into 3 groups: the YA group consisted of 4 monkeys who were 5.8 ± 0.5 (mean ± standard deviation) years old; the ES group consisted of 7 monkeys at ages of 21.6 ± 1.5 years; and the LS group consisted of 3 monkeys at ages of 29.3 ± 2.3 years (cf. Table 1). According to the analysis based on sex maturation and whole lifespan (Tigges et al. 1988), the degree of senescence of 19–24 (ES) and 28–32 (LS) years old monkeys approximated to that of 60–70 and 80–90 years old humans, respectively. Monkeys were examined ophthalmoscopically and had no indication of optical or retinal problems that would impair visual function. Retinal blood vessels, lens clarity, and the maculae were all within the normal range.
|Subject||Age (years)||n||Orientation bias||Direction bias|
|Subject||Age (years)||n||Orientation bias||Direction bias|
n, number of cells; SD, standard deviation; Max, maximal value.
Animal Preparation and Recording
Subjects were sedated with Ketamine HCl (10–15 mg/kg, i.m., Parke-Davis, Morris Plains, NJ, USA) and then anesthetized with 3–5% halothane (Halocarton Laboratories, River Edge, NJ, USA) in a 70:30 mixture of N2O:O2. Intravenous and tracheal cannulae were inserted. Animals were then placed in a stereotaxic frame. All pressure points and incisions were infiltrated with a long-acting local anesthetic (2% lidocaine HCl; Copley Pharmaceuticals, Canton, MA, USA). A craniotomy was performed above the area in V1 corresponding to the representation of central vision. After surgical procedures were finished and sufficient level of anesthesia was assured, a mixture of d-tubocurarine (0.4 mg/kg/h; Sigma, St. Louis, MO, USA) and gallamine triethiodide (7 mg/kg/h; Sigma) was infused intravenously to induce and maintain paralysis. During recording, animals were artificially ventilated. End tidal CO2 partial pressure was monitored and maintained at ∼4–4.5%. Body temperature was maintained at 38–38.5°C by an automatically regulated heating pad. The anesthesia was maintained with a mixture of N2O (70%), O2 (30%), and halothane (0.25–1.0%, as needed). Electrocardiography was continuously monitored. The level of anesthesia was assessed by periodically applying nociceptive stimulation (toe pinch) and monitoring the changes in the heart rate. We paid close attention in experiments to maintain the similar level of anesthesia for all subjects through continuously monitoring and maintaining the similar level of the heart rate and end-tidal CO2 partial pressure, both of which have been demonstrated as good indicators of the level of halothane anesthesia (Villeneuve and Casanova 2003).
The nictitating membrane was prepared with neosynephrine. The pupils were dilated with atropine, and the eyes were covered with contact lenses for protection from desiccation. The optics of the eyes and retinal vasculature were monitored throughout the experiment. No visible deterioration in optics occurred during the experimental period in any of the animals. Action potentials of isolated units were recorded extracellularly by glass microelectrode (filled with 2-M NaCl, 1–3-MΩ impedance at 1000 Hz), which was advanced by a hydraulic microdrive (David Kopf Instruments, Tojunga, CA, USA).
For each isolated single unit, the eye preference was determined and all subsequent stimuli were presented monocularly to the dominant eye. Cell's receptive field was carefully plotted on a tangent screen by hand with the use of an ophthalmoscope. After that, a 17″ CRT color monitor (85-Hz refresh rate; Sony, Tokyo, Japan) was positioned at the distance of 57 cm in front of the animal's eye and centered on the receptive field of the cell. The program to generate the visual stimuli was written in MATLAB (MathWorks, Natick, MA, USA), using the extensions provided by the high-level Psychophysics Toolbox (Brainard 1997) and low-level Video Toolbox (Pelli 1997). To quantify the orientation and direction selectivities of V1 cell, drifting bars were used, whose width, length, and moving speed were adjusted to elicit strongest response from the recorded cell. The direction of motion of each presented bar was orthogonal to its orientation. We used moving bars at 24 randomly chosen movement directions, ranging from 0° to 360° in steps of 15° to compile the tuning curves for the cells studied. Each direction was presented 10 times. The inter-trial interval was 2–5 s. The luminance of the stimuli used was 39 cd/m2 for white and 0.95 cd/m2 for black.
Data Collection and Analysis
Signals recorded from the microelectrode were amplified (1000×), band-pass filtered (300–10 000 Hz), and then digitized (sampling frequency of 20 kHz) by using an acquisition board (National Instruments, Austin, TX, USA) controlled by IGOR software (WaveMetrics, Portland, OR, USA). Such original signals were stored in a computer for offline analyses. The responses to moving bars were defined as the maximal value in the peristimulus time histogram (PSTH, bin width of 10 ms) during the stimulation period. Before each drifting bar was presented, the spontaneous (baseline) activity was recorded during a period of 0.5–0.7 s while the screen with average luminance was presented.
Orientation and direction selectivities were quantified for each cell using the method established previously (Leventhal et al. 1995). Briefly, the responses of each cell to the different stimulus orientations and directions were represented as vectors, with the response amplitude (spikes/s) as the magnitude and the stimulus direction (degree) or twice the stimulus orientation as the angle. The vectors were first added and then divided by the sum of their absolute values. The angle of the resulting vector indicated the preferred orientation or direction of the cell. The magnitude of the resulting vector (taking values between 0 and 1)—termed the orientation bias (OB) or direction bias (DB)—provided a quantitative measure of the orientation or direction sensitivities of the cell. Zero indicates that the neuron responded equally to all orientations (directions) and one means that the neuron responded only to one stimulus orientation (direction).
Clustering analysis was conducted by using the k-means algorithm (MATLAB function kmeans). Clustering was based on normalized response measures (i.e., divided by the maximal value among all subjects). The number of clusters was set to be 3.
As some of the quantities analyzed in the present study (e.g., OB and DB, spontaneous, and evoked response) did not follow normal distribution (P < 0.05, Lilliefors test), we used the Mann–Whitney U-test or Kruskal–Wallis test to analyze the difference in median among various groups. Chi-square test was used for analyzing the difference in percentage. Kolmogorov–Smirnov (KS) test was used to assess the difference in distribution of various quantities across groups. To analyze the interactions between factors, a 2-way ANOVA was applied and the normality of the data was achieved by logarithmic transformation (Bland and Altman 1996; Olivier et al. 2008). In all analyses, a level of P < 0.05 was considered significant and P < 0.01 was considered highly significant.
We studied 140 neurons in 4 YA, 212 neurons in 7 ES and 52 neurons in 3 LS monkeys. Subjects for this study were abbreviated as YA1–4 (for YA monkey 1–4), ES1-7 (for ES monkey 1–7) and LS1-3 (for LS monkey 1–3) hereafter. Neurons studied in each age group were recorded from the same range of cortical depths and had similar eccentricities of receptive fields (<8°).
Orientation and Direction Selectivities
Previously, we reported that the orientation and direction selectivities significantly decreased in nonhuman primate V1 at the late stage of senescence (Schmolesky et al. 2000, Leventhal et al. 2003). The current results reconfirmed this finding. Figure 1A,D show examples of tuning curves obtained from the YA and LS monkeys, respectively. YA cell exhibited much stronger selectivities for both the orientation and direction. The impression conveyed by these examples was further supported by comprehensive analyses. First, we found that both OB and DB values showed no significant individual differences within the YA and LS groups (Kruskal–Wallis test; OB: P = 0.72 and 0.13 for the YA and LS monkeys, respectively; DB: P = 0.18 and 0.11), so the results for individual animals within each group were combined for further analyses. As shown in Figure 2A, the values for both OB and DB were much smaller for the LS group compared with the YA group. KS test revealed that the distributions of OB and DB were significantly different between the 2 groups (P < 10−4), which can be illustrated by clearly separated cumulative distribution functions (CDFs) for the 2 groups (Fig. 2C,D). Moreover, the medians of OB and DB values for the LS monkeys as a whole were significantly lower than those for the YA monkeys (Mann–Whitney U-test, P < 0.001) (Fig. 2E,F; see Table 1 for detailed OB and DB values for individual animals). We also found that the percentages of V1 cells biased (with OB or DB ≥ 0.1) or strongly biased (with OB or DB ≥ 0.2) for orientation and direction were significantly smaller for the LS monkeys than those for the YA monkeys (OB ≥ 0.1: χ2 = 17.58, P < 0.001; OB ≥ 0.2: χ2 = 10.36, P < 0.01; DB ≥ 0.1: χ2 = 38.51, P < 0.001; DB ≥ 0.2: χ2 = 18.24, P < 0.001; χ2 tests. See Table 2 for details).
|OB ≥ 0.1||OB ≥ 0.2|
|DB ≥ 0.1||DB ≥ 0.2|
|OB ≥ 0.1||OB ≥ 0.2|
|DB ≥ 0.1||DB ≥ 0.2|
n, number of cells.
*P < 0.05, **P < 0.01; comparing the YA group with either the ES or the LS group (the YA group has higher percentages); #P < 0.05, ^P < 0.05 and ##P < 0.01; comparing the LS group with the ES group. # or ##, the ES groups have higher percentages; ^, the ES group has lower percentage.
Unlike the other 2 groups, for the ES monkeys, significant individual differences were found on OB values (Kruskal–Wallis test, P < 0.001) (Table 1). Further analysis based on pairwise comparisons of OB suggested that the ES group could be further classified into 3 subgroups, including ES1-3 (ES1, 2 and 3), ES4-5 (ES4 and ES5), and ES6-7 (ES6 and 7). Both the median and the CDF of OB showed no significant difference between any pair of individuals within each subgroup (data not shown). We then combined the data for all individuals within each of the ES subgroups and then compared OB for these subgroups separately to the YA and LS groups. Both the median and CDF of OB in ES1-3 monkeys were not significantly different from those in the YA monkeys (Mann–Whitney U-test, P = 0.85; KS test, P = 0.50), but were significantly different (with higher median) from those in the LS monkeys (P < 0.001 for both tests). In contrast, the median and CDF of OB in ES4-5 and ES6-7 monkeys were significantly different (with lower median) compared with those in the YA monkeys (P < 0.01 for both tests), but, at the same time, did not show significant difference from those in the LS monkeys (Mann–Whitney U-test, P = 0.15 [ES4-5 vs. LS] and 0.052 [ES6-7 vs. LS]; KS test, P = 0.36 [ES4-5 vs. LS] and 0.12 [ES6-7 vs. LS]) (Fig. 2E).
In addition, the percentages of cells that were biased or strongly biased for orientation were computed for all 3 ES subgroups (Table 2). First, the percentages were compared between the YA group and each of the 3 ES subgroups. The percentages showed no significant difference between the YA and ES1-3 groups (OB ≥ 0.1: χ2 = 0.86, P = 0.36; OB ≥ 0.2: χ2 = 0.66, P = 0.42). However, the percentages were significantly smaller for ES4-5 and ES6-7 groups than those for the YA group (OB ≥ 0.1: χ2 = 11.18, P < 0.01 and OB ≥ 0.2: χ2 = 3.88, P < 0.05 for comparisons between ES4-5 and the YA groups; OB ≥ 0.1: χ2 = 32.75, P < 0.001 and OB ≥ 0.2: χ2 = 7.37, P < 0.01 for comparisons between ES6-7 and the YA groups). Second, the percentages were compared between the LS group and each of the 3 ES subgroups. The values were significantly higher for ES1-3 group than those for the LS group (OB ≥ 0.1: χ2 = 11.40, P < 0.01; OB ≥ 0.2: χ2 = 13.60, P < 0.001). For ES4-5 group, the percentages displayed no significant difference from those for the LS group (OB ≥ 0.1: χ2 = 0.77, P = 0.38; OB ≥ 0.2: χ2 = 2.05, P = 0.15). Interestingly, the percentage of cells that were biased for orientation was even lower for ES6-7 group than that for the LS group (χ2 = 4.60, P < 0.05), but the percentage of cells that were strongly biased revealed no significant difference between ES6-7 and the LS group (χ2 = 0.05, P = 0.83).
Importantly, the individual differences on OB we observed for the ES monkeys were not due to the larger sample size (i.e., number of subjects) for this group (n = 4, 7 and 3 for the YA, ES and LS groups, respectively). We chose 4 or 3 ES monkeys out of the whole ES population and examined individual differences for these downsampled ES groups. In total, there were 70 different ways to choose such downsampled groups. Among them, 84% (59 of 70) showed significant individual differences on OB (Kruskal–Wallis test, P < 0.05). Also, we note that the fact that ES1-ES3 had similar OB to the YA group and ES4-ES7 had similar OB to the LS group is not a result of any specific way of defining the subgroups (e.g., ES1-3, ES4-5, and ES6-7). We tested OB for individual ES monkeys, in comparison to the YA and LS groups. We found exactly the same results as obtained from the subgroup analysis. On one hand, ES1, ES2, and ES3, when compared individually, showed significant difference from the LS group (Mann–Whitney U-test, P = 0.001, 0.03, and 0.001, respectively) but no significant difference from the YA group (P = 0.78, 0.35, and 0.45). On the other hand, ES4-7, when compared individually, showed significant difference from the YA group (P = 0.004, 0.01, 0.001, and 0.001) but no significant difference from the LS group (P = 0.52, 0.10, 0.13, and 0.12). These control analyses excluded the possibility that some methodological issues might have biased our results.
In contrast to the remarkable individual variability in OB values found for the ES group, individual differences in DB values for the ES monkeys were not significant (Kruskal–Wallis test, P = 0.38). Similarly, in terms of DB values, no significant individual differences were found for either the YA group (P = 0.18) or the LS group (P = 0.11). The CDFs for all 3 age groups (i.e., YA, ES, and LS) showed significant difference among each other (P < 0.002) and, as a whole, the median of DB for the ES monkeys was significantly lower than that in the YA monkeys (P < 10−4) but higher than that in the LS monkeys (P < 0.002) (Fig. 2D,F). Additionally, for the ES monkeys, the percentages of cells that were biased or strongly biased for direction were lower than those for the YA monkeys (DB ≥ 0.1: χ2 = 27.86, P < 0.001; DB ≥ 0.2: χ2 = 15.71, P < 0.001) but higher than those for the LS monkeys (DB ≥ 0.1: χ2 = 8.21, P < 0.01; DB ≥ 0.2: χ2 = 5.60, P < 0.05) (Table 2).
The above analyses demonstrated that there was a subgroup of ES monkeys (i.e., ES1-3) that, when compared with the YA monkeys, exhibited an impairment in DB but not in OB. We found that, indeed, for this subgroup of subjects, the interaction between age (YA vs. ES) and feature (OB vs. DB) was significant (P < 0.01, 2-way ANOVA for logarithmically transformed data), suggesting the dissociation of aging-related changes in the orientation and direction selectivities during early senescence.
Spontaneous Activity, Peak Response and Signal to Noise Ratio
In normal aging, the decreased stimulus selectivities in V1 is often accompanied by elevated activity level in both the spontaneous and visually evoked conditions (Schmolesky et al. 2000; Leventhal et al. 2003). Consistent with the previous reports, we found that the LS group showed significantly increased medians of the baseline (i.e., spontaneous) activity and the peak visual response (P < 0.001 for both) compared with the YA group. In addition, as the increase in the baseline activity was more pronounced compared with the increase in the visually evoked activities, the median of ratio between the evoked and the baseline activity, here we termed SNR of the response, was significantly decreased in the LS group compared with that of the YA group (P < 0.001) (Fig. 3A–C).
The ES monkeys showed more complex, yet more interesting, pattern of changes. We first made the comparisons between the ES and YA groups. The median of baseline activity was higher (P < 0.001) for the ES monkeys but the peak response showed no significant difference between the ES and YA monkeys (P = 0.60). Consequently, the median of SNR was lower (P < 0.01) for the ES monkeys than that for the YA monkeys. The increased baseline activity, together with the unchanged visually evoked activity, resulted in a decreased SNR for the ES monkeys (P < 0.01). Second, the comparisons were performed between the ES and LS groups. Both the baseline activity and peak response were lower for the ES monkeys compared with those for the LS monkeys (P < 0.001 for both). The SNR was higher for the ES monkeys (P < 0.05). In all comparisons regarding the baseline, evoked activity and SNR among the 3 age groups, the results based on comparing the medians were consistent with the ones based on comparing the whole CDFs (Fig. 3A–C).
In addition, the percentage of cells that exhibited baseline activity >5 spikes/s was measured for each age group. The percentage for the ES group (39%, 78 of 202 units, for which the baseline activity was recorded) was significantly higher than that for the YA group (24%, 33 of 135; χ2 = 7.36, P < 0.01) but greatly lower than that for the LS group (69%, 35 of 51; χ2 = 14.84, P < 0.001). Moreover, the percentages of cells that displayed peak responses >150 spikes/s were computed for all 3 age groups. The percentage for the ES group (20%, 43 of 212) was not significantly different from that for the YA group (15%, 21 of 140; χ2 = 1.58, P = 0.21) but greatly lower than that for the LS group (73%, 38 of 52; χ2 = 54.72, P < 0.001).
Individual differences among the ES monkeys for either the peak response or SNR were not significant (Kruskal–Wallis test, P = 0.90 and 0.53, respectively). It was, however, significant for the baseline activity (P < 0.001). Moreover, we found significant individual differences among the YA monkeys on all 3 measures (i.e., baseline, peak response, and SNR). This was different from the cases for OB and DB, in which no significant individual variability was found for either the YA or LS groups. It suggests that these 3 measures may be 1) intrinsically more variable across individuals or 2) prone to be affected more by the method of sampling neurons randomly (e.g., the slight difference in types of cells sampled for individual animals). Therefore, to examine the overall trend of how the baseline, peak response, and SNR were affected by aging, we combined the data that belonged to the same age group for comparison. In light of the variability analysis that we conducted for OB and DB, the results regarding the baseline, peak response, and SNR should be interpreted as an average tendency. An even larger dataset might be needed to address the individual variability of these properties in the future.
Finally, we analyzed the quantitative relationship between the impaired feature selectivities (i.e., OB and DB) and the decreased SNR. We found that the means of the OB and DB were positively correlated with the means of the SNR for individual animals (for OB, the Pearson correlation coefficient, r, equals 0.43, with P = 0.13; for DB, r = 0.67, P = 0.008; Fig. 3D), indicating the tendency of covariation between these properties during aging.
Identifying Common Features for Early Aging with Multiple Measures
So far, our analyses were focused on individual properties (e.g., OB, DB, and baseline). Next, we considered multiple properties at the same time, trying to identify the common features for early senescence. To this end, we applied automatic clustering analysis (see Methods section for details). We found that based on DB, baseline, and peak response, the k-means clustering algorithm could correctly identify all the subjects that belonged to the same age group (Fig. 4A). It provides strong evidence that the groups defined on the basis of chronological age are also functionally distinct groups. In Figure 4B, such functional separation is visualized in the corresponding 3D space. It is clear that the combination of impaired DB, unchanged peak response and moderately increased baseline uniquely distinguishes the ES group from the other 2.
Anesthesia is an important factor to consider when interpreting our results. Besides paying close attention to maintain the similar level of anesthesia to all subjects in our experiment, there are several lines of evidence strongly suggesting that the present results cannot be an effect of anesthesia. In previous control experiments, we have systematically varied the level of halothane anesthesia while recording the activities from both young and aged monkeys (Wang et al. 2005; Yang et al. 2008). We found that the orientation/direction selectivities as well as the baseline were not affected by changing the level of anesthesia (see also Villeneuve and Casanova 2003). Moreover, although the peak response was inversely affected by the level of anesthesia, the effects were similar across young and aged monkeys. Finally, it has been reported that older subjects are more sensitive to anesthesia (Hoffman et al. 1985; Schwartz et al. 1989; Magnusson et al. 2000). So if anesthesia is a major factor, aged monkeys would exhibit decreased level of evoked activity. This is opposite to what we actually observed. Taken together, the evidence is compelling that the results in the present study cannot be attributable to factors related to anesthesia.
We have previously reported the functional degeneration of visual cortical cells in old monkeys during the middle and late stages of aging (Schmolesky et al. 2000; Leventhal et al. 2003). Given that a number of studies have demonstrated that the function of the visual system begins to degrade at the early stage of aging (Gittings and Fozard 1986; Wood and Bullimore 1995; Fahle and Daum 1997; Jackson et al. 1998; Tran et al. 1998; Porciatti et al. 1999; Betts et al. 2005; Del Viva and Agostini 2007; Billino et al. 2008; Kennedy et al. 2009; Weymouth and McKendrick 2012), it is expected that the changes in cell's response properties would also be observed during this stage. In the current study, we provide such empirical evidence and, for the first time, conduct systematic analysis about how functions of V1 cells are affected by different stages of aging. We found that ES cells exhibited clear functional degradations, including declined stimulus selectivities and elevated level of the spontaneous activity.
Overall, the changes we observed in the ES monkeys were less pronounced compared with that in the LS monkeys and such changes might represent an intermediate level of aging effects. This is consistent with observations that various visual functions decline progressively during aging (Gittings and Fozard 1986; Koss 1991; Weale 1992; Wood and Bullimore 1995; Tran et al. 1998; Schefrin et al. 1999; Hayashi et al. 2003). For example, an investigation that examined a group of people over a range of ages from 19 to 92 years showed that the perception and detection of motion deteriorated with age (Tran et al. 1998); another study, with human subjects with ages ranging from 21 to 92 years, demonstrated a gradual decline in visuospatial performance with age (Koss 1991). The present results provide physiological evidence, suggesting that the aging effect on some response properties of V1 cells may accumulate during the normal aging process.
Besides demonstrating that the functional degradation in V1 does occur, but with less severity, in early senescence, more interesting aspect of our finding is that those changes in ES subjects exhibit unique features, including the increased individual variability and different effects across various response properties (e.g., orientation vs. direction selectivity; baseline vs. evoked activity). Our results suggest that the aging of V1 cannot be described as a homogenous process, during which various functions gradually fade away more or less equally.
The increase of individual variability in orientation selectivity is a salient feature that we observed in our ES subjects. Among the 7 ES monkeys used in this study, 3 showed no aging effect when compared with the YA monkeys, but the other 4 displayed aging effect as strong as that in the LS monkeys. It has been suggested that the individual variability is an important aspect in studying the aging of the visual system as it may shed light on understanding how different genetic, physiological, and environmental factors will influence this process (Owsley 2011). Our results indicate that it may be particularly so for studying the visual degradation during early aging. In addition, the increased individual variability is observed for orientation selectivity but not for direction selectivity, suggesting that such variability may manifest itself differently with different functional domains.
In both the current results and the previously reported ones (e.g., Schmolesky et al. 2000; Leventhal et al. 2003), the orientation selectivity of V1 cells exhibits impairment invariably across the middle and late senescent monkeys. The finding that some of the ES monkeys did not show such deficits suggests the possibility that declines in orientation selectivity in V1 can be significantly delayed in early aging, but with the potential price that such deficits may eventually emerge in a more sudden manner. So far, it is not clear which psychophysical task is optimal for probing the functional deficits associated with the decreased orientation selectivity (e.g., see Delahunt et al. 2008), but human studies with other visual tasks indicate that such nonlinearity in functional loss does exist during aging (Wist et al. 2000; Bennett et al. 2007; Arena et al. 2012). Taken together, our results underline the importance to pay closer attention to the individual variability and dynamical changes in visual functions that rely on cortical representation of stimulus orientation.
In contrast to the orientation selectivity, we found that the impairment in direction selectivity can be invariably detected in the ES monkeys, suggesting that motion perception may be more vulnerable during early aging. Such result is in agreement with the previous human studies. Degradations in motion perception and related tasks are among the most documented functional declines associated with aging (Buckingham et al. 1987; Elliott et al. 1989; Trick and Silverman 1991; Whitaker and Elliott 1992; Wood and Bullimore 1995; Tran et al. 1998; Wist et al. 2000; Kline et al. 2001; Sosnoff and Newell 2006). These deficits can be observed both in the early stage of senescence in humans (Wood and Bullimore 1995; Tran et al. 1998) and with more severity compared with other visual functions (Kline 1987). A recent study reported that the deficit in motion perception is, indeed, one of the earliest signs indicating the aging of the visual system—it can be detected in persons' 50s–60s (McBain et al. 2010). The present results provide the possible physiological underpinning for this phenomenon. Direction selectivity of V1 cells is achieved mainly through the subtle balance of the excitation and inhibition in cortical networks (Priebe and Ferster 2005). There is increasing evidence for degradation of the inhibitory function during aging (McGeer and McGeer 1976; Gutierrez et al. 1994; Leventhal et al. 2003, Hua et al. 2008), suggesting the potential cause of the venerability of direction selectivity during early senescence. Our correlation analysis is also consistent with such a hypothesis, as we found stronger relationship between DB and SNR, while the latter is related to the level of inhibition in the cortex (Leventhal et al. 2003).
Another unique feature observed in the ES monkeys is the increased spontaneous activity coupled with the unchanged peak response level. Such a change will lead to the decreased dynamic range, that is, the range of neuronal activity that can be used to represent sensory information, for which the maintenance requires, again, the proper balance between the excitation and inhibition (Shew et al. 2009; Yizhar et al. 2011). Importantly, as we used the optimal stimuli to measure the peak response (i.e., not only with optimal orientation and direction but also optimal spatial and temporal frequencies, etc.), our result about the decreased dynamic range has implications beyond the response properties that we studied here. In other words, the V1 cells in the ES monkeys not only change their dynamic range in response to different orientations or directions but also change it for other features as well. As the dynamic range is a fundamental property for neuronal information processing (Friedrich and Korsching 1997; Wachowiak and Cohen 2001; Bhandawat et al. 2007; Yizhar et al. 2011), it would be important to investigate how the change of it during the early stage of senescence will affect the representation of visual information in the brain.
Finally, we emphasize that the current results, as well as the previously reported physiological changes in aged animals, demonstrate the neuronal changes during aging. Yet, to relate such physiological degradations to visual deficits that can be detected in psychophysical tests is not always straightforward (e.g., see Delahunt et al. 2008; Govenlock et al. 2010). In order to obtain further insight into the relationship between the neuronal changes and the visual degradations in aging, to combine physiological recordings and behavioral tests in alert animals would be necessarily the next step. Moreover, it is worth noting that the aging-related retinal changes in humans (e.g., Curcio et al. 1993; but see Spear 1993; Harman et al. 1997, 2000; Boulton and yhaw-Barker 2001; Curcio 2001; for review, see Owsley 2011) seem to be more significant than those in monkeys (Spear 1993; Kim et al. 1996; Schmolesky et al. 2000). Therefore, the visual system in aged humans may be subject to both the cortical alternations and the degraded input from the retina. Whereas the current investigations using senescent monkeys can delineate the aging-related changes intrinsic to the cortex, to better understand the effects of the degraded input as well as how it interacts with the cortical degradation would be an important future direction.
In summary, by comparing response properties of V1 cells in the YA, ES, and LS monkeys, we found clear indications of functional degradation in early aging, including decreased stimulus selectivities, elevated spontaneous activity and impaired SNR, and dynamic range of V1 cell responses when compared with the YA animals. Such changes are similar to, but less pronounced than those observed in the LS monkeys. Importantly, we also found features that are uniquely associated with early senescence. For example, the peak response level, which is affected in LS monkeys, is spared in early senescence. Moreover, the orientation selectivity is impaired for only some of the ES animals and largely unaffected for other individuals in the same age group. Our results provide 1) possible neural mechanisms underlying the overall trend of declining visual abilities in the early stage of aging and 2) new clues to understand the widespread individual variability associated with such visual deficits. In addition, these results suggest that V1, and probably other visual cortices, are affected by aging in a dynamic way, which could be more complex and heterogeneous than previously thought.
This work was supported by the National Basic Research Program of China (973 Programs, 2011CBA00405, 2011CB707800, and 2009CB941303), National Natural Science Foundation of China (31230032 and 30970978 to Y.Z.), Main Direction Program of Knowledge Innovation of Chinese Academy of Sciences (KSCX2-EW-J-23), the Training Program for Young Backbone Teachers of Yunnan University, the research foundation of Yunnan University (2011YB22), Natural Science Foundation of Yunnan Province (2011FZ012), and the special fund of the “211” third phase project of Yunnan University (21134018).
The authors thank Xiangrui Li, Pinglei Bao, Xiang Ye, Yun Yang, Zhen Liang, Guangxun Li, Xiusong Wang and Jie Zhang for their assistance in experiment. The authors are grateful to Audie Leventhal for his supervision at the early stages of this project. Conflict of Interest: None declared.