HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 97/5/3421    most recent 00184.2007v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Qin, L. Articles by Sato, Y. Search for Related Content PubMed PubMed Citation Articles by Qin, L. Articles by Sato, Y.

Comparison Between Offset and Onset Responses of Primary Auditory Cortex ON–OFF Neurons in Awake Cats

Department of Physiology, Interdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, Yamanashi, Japan

Submitted 18 February 2007; accepted in final form 7 March 2007

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Primary auditory cortex (A1) neurons are believed not to carry much information about tonal offsets because A1 neurons in barbiturate-anesthetized animals are usually described as having only onset responses. We investigated tonal offset responses in comparison with onset responses in the caudal part of A1 of awake cats. Cells responding to both onsets and offsets were commonly found (59.2% of recorded cells). Offset responses usually co-occurred with phasic onset responses or phasic components of sustained responses. These ON–OFF cells had diverse combinations of offset- and onset-frequency-receptive field (FRF): offset-FRF was similar to onset-FRF, or narrower, wider, lower, or higher than onset-FRF. The distribution of FRF patterns was diffuse with no boundaries between the different FRF-pattern groups. The onset- versus offset-FRF pattern of each cell remained unchanged across multiple stimulus intensities. Mean offset response showed similar peak latency (19.5 vs. 21.5ms), longer half-decay time (74.5 vs. 48.5 ms), and lower peak amplitude (20.4 vs. 35.9 spikes/s) compared with the mean onset response. Although offset responses were facilitated when preceded by the suppression of spike activity, they were still elicited without preceding spike suppression. It is concluded that neurons showing paired onset and offset responses are predominant in the caudal A1. Their frequency-filtering property is usually not static but dynamic, changing between sound onsets and offsets. Offset responses are similarly precise and salient as onset responses for effectively encoding sound offsets. They may be elicited as active spike responses to sound offset rather than simple rebound facilitation.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Earlier psychophysical experiments showed that sound offset serves as an important acoustic cue in some phenomena such as the acoustic startle reflex (Ison and Allen 2003), perception of sound duration (Schlauch et al. 2001), and consonant identification (Pind 1998). The presence of cortical responses to sound offset was clearly demonstrated by the auditory-evoked potential (Takahashi et al. 2004), neuromagnetic response (Gutschalk et al. 2002; Hamada et al. 2004; Hari et al. 1987), and functional magnetic resonance imaging activation (Harms et al. 2005; Okada et al. 2004).

In contrast, previous physiological experiments in barbiturate-anesthetized animals showed that primary auditory cortex (A1) neurons usually respond to tone bursts at onsets of stimuli (Phillips and Hall 1990). The following studies focused on onset response properties such as frequency tuning: Schreiner and Sutter (1992); Sutter and Schreiner (1991); level dependency: Phillips et al. (1994); Sutter and Schreiner (1995); response latency: Heil (1997a); Mendelson et al. (1997); and stimulus slope sensitivity: Heil (1997b). Phillips et al. (2002) concluded that offset responses are relatively uncommon in the central auditory system and, when they do occur, they are rebounds from preceding inhibition.

Experiments in ketamine-anesthetized cats, on the other hand, showed that A1 neurons responded to a tone burst in more complex and various manners than in deep barbiturate-anesthetized animals: response time courses were phasic (57%) or tonic (26%) and phasic-response neurons exhibited ON responses (57%, responses only to sound onsets), ON–OFF responses (36%, responses to both sound onsets and offsets), and OFF responses (7%, responses only to sound offsets), indicating that responses to sound offsets are common in A1 neurons (Volkov and Galazjuk 1991). A recent study in halothane-anesthetized cats also reported that only 9% of the units recorded in A1 had pure onset responses, roughly 40% of the units had sustained responses throughout the stimulus duration (115 ms), and 27.7% of the units had significant responses after stimulus offsets (Moshitch et al. 2006). It appears that the onset response is the dominant response in A1 neurons of barbiturate-anesthetized animals; in contrast, A1 neurons show diverse response time courses in ketamine- and halothane-anesthetized animals.

In awake animals, a number of previous studies reported that A1 neuron responses to tonal stimuli show diverse response time courses from phasic to sustained patterns (Barbour and Wang 2003a,b; Brugge and Merzenich 1973; Chimoto et al. 2002; Evans and Whitfield 1964; Pfingst and O'Connor 1981; Qin et al. 2003; Recanzone 2000; Shamma and Symmes 1985; Volkov et al. 1985). Some of those studies in awake animals described the presence of offset responses in addition to onset responses with an incidence of 10–30% of recorded neurons (Brugge and Merzenich 1973; Chimoto et al. 2002; Evans and Whitfield 1964; Pfingst and O'Connor 1981; Recanzone 2000). It should be emphasized that the studies in awake animals mentioned the presence of offset responses only anecdotally: they did not focus on the details of their response properties such as frequency tuning and the response time course. At present there are no reports describing offset response characteristics compared with onset responses of A1 neurons in awake animals except for an abstract form in awake monkeys (Tian and Rauschecker 2003).

The aim of this study was to examine 1) the properties of tonal offset responses including frequency tuning and response time courses compared with onset responses and 2) the possible neural mechanism underlying offset responses. We will show details of tonal offset responses compared with onset responses of A1 neurons in awake cats. The neural mechanisms underlying offset responses will be discussed.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Animal preparation, recording, and histology

Experiments were performed in a manner consistent with the Guidelines for Animal Experiments, University of Yamanashi, and the Guiding Principles for the Care and Use of Animals approved by the Council of the Physiological Society of Japan. Animal preparation, recording, and histology procedures were as in previous reports (Chimoto et al. 2002; Qin et al. 2004a,b, 2005). Nine cats were chronically prepared for single-cell recordings from the auditory cortex. Under pentobarbital sodium anesthesia (initial dose 40 mg/kg) and aseptic conditions, an aluminum cylinder (ID 12 mm) was implanted bilaterally in the temporal bone for microelectrode access, at an angle of 10–20 ° from the sagittal plane. A metal block was embedded in the dental acrylic cap to immobilize the head. After >1 wk of postoperative recovery, the cat's body was gently wrapped in a cloth bag. The head was restrained with holding bars for a short period. In successive daily sessions, the period was lengthened and they were familiarized with sitting in an electrically shielded, sound-attenuated chamber. The animals were given food and drink during the sessions and, after each session, they were returned to their home cages. The conditioning procedure lasted at least 2 wk. When recording experiments began, they sat with no sign of discomfort or restlessness. One day before beginning the recording session, the bone (diameter 1–2 mm) at the bottom of the cylinder was removed, leaving the dura intact under ketamine anesthesia (initial dose 15 mg/kg).

The recording session began the following day. The dura was pierced with a sharpened probe and a glass microelectrode (tip diameter, 1.8–2.5 µm; resistance, 2–3 M; filled with 2 M NaCl) was advanced into A1 with a remote-controlled micromanipulator (Narishige, MO-951). Tone bursts of variable single frequency and sound pressure level (SPL) were presented as search stimuli. Extracellular single-unit activity was discriminated using a window discriminator. Spike-occurrence outputs from the window discriminator were captured directly through a digital-to-digital interface (Cambridge Electronic Design Limited 1401), using a Pentium-based data-input computer with a time resolution of 0.002 ms as the digital input for later data analysis. Data were stored on hard disk.

The cat's face, particularly the eyes, was continuously observed on a monitor connected to a charge-coupled-device camera. In our preliminary experiments, we studied the relationship between the status of the eyes and electroencephalography (EEG). Slow waves in EEG were observed when the eyes were closed and when the eyes were open but drifting, which were judged as a sleep state. Saccadic eye movements and eye fixation were judged as signs of an awake state. Rapid eye movements of paradoxical sleep, in which slow EEG waves were absent, were easily identified by their characteristic appearance of half-opened eyelids and were judged as signs of sleep. When drowsiness was suspected, the cat was alerted by gently tapping the body with a remote-controlled device or by briefly opening the door.

The cats sometimes moved during recording sessions, producing artifacts in the recording. By carefully checking the monitors and the spike train, artifacts were noted on the recording computer in real time during recording. Data with artifacts could therefore be rejected.

Daily recording sessions lasted 3–5 h for 2–6 mo in each animal. At the end of each daily recording session, the recording chamber was rinsed with sterile saline and antibiotic fluid and sealed with Exafine (GC, Tokyo, Japan) and an aluminum cap. The animal was returned to its cage. All cats remained healthy throughout the experimental period. At the end of the experiment, some recording sites were marked with electrolytic lesions (25 µA, 10 s). The animal was deeply anesthetized with sodium pentobarbital and perfused with 10% formalin before the brain was removed. The brain surface was photographed. The cerebral cortex was cut in coronal sections and stained with neutral red. Based on the lesion locations and electrode tracks, the recording sites were reconstructed.

Sound generation and delivery

Sound generation and delivery were as in previous reports (Chimoto et al. 2002; Qin et al. 2004a,b, 2005). In brief, sinusoids were generated using custom-written programs in a MATLAB (The MathWorks) environment on a Pentium-based computer. The signals were fed into a 12-bit D/A converter (PCI-MIO-16E-4) at a sampling interval of 100 kHz and to an eight-pole Chebyshev filter (NF Electric Instruments, P-86) with a high cutoff frequency of 20 kHz. The output was attenuated and sent to a low-output–impedance power amplifier (Denon, PMA2000III) and tones were then played through a speaker (AKG, K1000) placed 2 cm from the auricle contralateral to the recording site. We equalized and calibrated the sound-delivery system between 128 and 16,000 Hz and the output varied by ±1.5 dB. Harmonic distortion was < –60 dB. Pure tone stimuli, 500 ms in duration with a rise/fall time of 5 ms, were presented at an interstimulus interval of 1,640 ms. The intensity was set at 20- to 80-dB SPL by 10-dB steps. At each SPL, 125 different frequencies were presented in pseudorandom order. Each frequency was presented once if the frequency responses were well obtained; otherwise, this procedure was repeated. The frequency range was typically 128–16,000 Hz at linear steps of 128 Hz, respectively. Finer steps (64, 32, 16, or 8 Hz) were also used in some units. Because the time to continuously isolate a single unit was limited, we first attempted to complete the procedures at 30-, 50-, and 70-dB SPL then recorded data at other stimulus intensities. Thus the tested sound intensities of each cell were different. All cells in this study were tested at least at 30-, 50-, and 70-dB SPLs.

Data analysis

During recording of a single neuron, we presented pure-tone stimuli with variable frequencies at a given SPL. Spike activities during a stimulus period (0.5 s) in addition to pre- and poststimulus periods (each 0.5 s) were analyzed off-line with custom-written programs in a MATLAB (The MathWorks) environment. Figure 1 shows the procedures for estimating spectral and temporal response features of A1 cells. First, raster plots of the spike trains in response to 125 frequency stimuli were aligned at stimulus onsets in ascending order of frequency (Fig. 1A).

View larger version (56K): [in this window] [in a new window]   FIG. 1. Procedures for estimating frequency-receptive field (FRF) and the response time course in an example ON–OFF cell (phasic-response cell). A: raster display of spike discharges in response to pure tone stimuli at 125 different frequencies. Two vertical lines mark the onset (0 s) and offset (0.5 s) of sound stimuli. B: spectrotemporal spike-activity diagram constructed by calculating the spike density of each 5-ms time window along each spike train in A. Spike densities of various grades are shown in different colors. Two white vertical lines mark the onset (0 s) and offset (0.5 s) of sound stimuli. C: spike density histogram (0.1 spikes/s in bin width) during 0.5-s prestimulus period in B. Red line indicates 1% probability level at the high tail of the spike density distribution, representing the boundary between the background level and excitation. Similarly, the blue line shows the boundary between the background level and suppression. D: spectrotemporal spike-activity diagram colored for 3 activity levels: excitation (brown), background (green), and suppression (blue) levels. Three horizontal bars show onset, termination, and offset time windows, respectively. White arrows show low- and high-edge frequencies of onset-FRF E: onset-response time course (1-ms bin width, smoothed by Gaussian with 5-ms SD) across the frequency range of onset-FRF (3,200–15,872 Hz). F: offset-response time course across the frequency range of offset-FRF (512–16,000 Hz).

Third, background spike rates during the prestimulus period were analyzed by constructing a spike-density histogram (0.1 spikes/s in bin width) (Fig. 1C). Excitatory spike activity was defined as a spike rate of >1% probability level at the high tail of the spike-density histogram (red line in Fig. 1C), whereas suppressive spike activity was defined as a spike rate of <1% probability level at the low tail (blue line in Fig. 1C). Thus pixels with excitatory spike activities were colored brown; those with suppressive spike activities, blue; and the remaining pixels with activity around the background level, green, in the spectrotemporal spike-activity diagram (Fig. 1D). There should be <125 brown/blue pixels of the 12,500 pixels in the prestimulus background area.

Fourth, we evaluated the statistical significance of excitatory responses by counting the number of brown pixels in the during and poststimulus areas. To analyze the responses, we selected "onset" (15–115 ms after stimulus onset), "termination" (385–485 ms), and "offset" (515–615 ms) time windows (horizontal bars in Fig. 1D). The time windows were shifted 15 ms from the stimulus onset or offset to avoid the effect of the zero-padded edge. The lowest and highest five frequencies were also removed from each time window; that is, there were 2,300 pixels (20 time bins x 115 frequency bins) in each analysis area. When the number of brown pixels was >35, the increase was significant (P < 0.01); that is, excitation was significant. Similarly, when the number of blue pixels was >35, suppression was significant (P < 0.01).

Once a significant response was identified, the lowest and highest frequencies of the brown area (lowest and highest five tested frequencies were also considered) during the onset time window were designated as "low-edge frequency" and "high-edge frequency" (white arrows in Fig. 1D). The frequency range between them (high-edge frequency–low-edge frequency) was defined as "onset-FRF." The "termination-FRF" and "offset-FRF" were similarly defined. When more than two responsive areas were separated by unresponsive areas, the response parameters were evaluated only from the largest area, excluding other smaller areas. The "suppressive FRF" (sFRF) was the frequency range (high-edge frequency–low-edge frequency) of the blue area during the termination time window.

The fine onset-response time course was investigated by constructing a firing rate histogram along the time axis in 1-ms bin width across the frequency range of the onset-FRF (3,200–15,872 Hz in this example cell). The mean firing rate of the 0.5-s prestimulus period was subtracted from the firing-rate histogram to convert to the driven-rate histogram and was smoothed by Gaussian function with a 5-ms SD (Fig. 1E). The "best onset-SPL" was defined as the stimulus SPL at which the largest brown area during the stimulus period was obtained across the tested SPL range.

The offset-response time course (Fig. 1F) was similarly constructed across the frequency range of offset-FRF (512–16,000 Hz in this cell). The "best offset-SPL" was defined as the stimulus SPL at which the largest brown area after stimulus offset was obtained. The response properties of onset and offset responses were compared at the best offset-SPL unless otherwise specified.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The results reported here are based on 638 single cells recorded from nine cats. The recording site was determined with reference to the electric lesion and the electrode track in the brain section. Histological reconstruction of the recording sites showed that cells were sampled from the caudal part of the middle ectosylvian gyrus, the banks of the dorsal tip of the posterior ectosylvian sulcus, and a small portion of the adjacent posterior ectosylvian gyrus, i.e., the caudal part of A1 (Reale and Imig 1980).

Tonal response pattern versus offset responses

Of the 638 recorded cells, 610 showed significant excitatory responses during tonal stimuli. As reported previously (Qin et al. 2003, 2004a), the excitatory responses of A1 cells in awake cats had various response time courses. Three typical examples are shown in Figs. 1 and 2. Some cells showed phasic excitatory responses to tonal onset: FRF rapidly disappeared with the stimulus time (Fig. 1). Other cells showed sustained excitatory responses to tonal stimuli: FRF was unaltered throughout the stimuli duration (Fig. 2, A, C, and E). The other cells showed phasic responses to some tonal frequencies, whereas sustained responses to other frequencies had a relatively wide FRF soon after tonal onset, after which FRF gradually diminished with time, resulting in a relatively narrow FRF at the termination of stimuli (Fig. 2, B, D, and F).

View larger version (50K): [in this window] [in a new window]   FIG. 2. Tonal responses of 2 other example ON–OFF cells (sustained-response cells). A and B: spectrotemporal spike-activity diagrams in continuous color grades. C and D: spike density histograms during 0.5-s prestimulus periods in A and B, respectively. Note that both distributions are Poisson with a low spike density, making it impossible to find the 1% probability level at the low tail. E and F: spectrotemporal spike-activity diagrams in 2 activity levels: excitation (brown) and background (green) levels. See Fig. 1 for other captions.

We then investigated the occurrence rate of offset responses in different tonal response types: most (204 of 280 cells, 72.9%) phasic-response cells had offset-FRF, whereas roughly half (178 of 330 cells, 53.9%) sustained-response cells had offset-FRF at the best onset-SPL.

We further investigated whether the onset- versus termination-FRF pattern is different between sustained-response cells with (178 cells) and without (152 cells) offset-FRF. For this purpose we compared onset- and termination-FRF by calculating high- and low-edge frequency differences (see METHODS) at the best onset-SPL; that is, the high-edge difference was calculated as the log₂ (high-edge frequency of onset-FRF/high-edge frequency of termination-FRF) and low-edge difference was defined as the log₂ (low-edge frequency of onset-FRF/low-edge frequency of termination-FRF). Both high- and low-edge differences are expected to be around 0 when onset- and termination-FRF are similar, like the cell in Fig. 2E. The high-edge difference is positive and the low-edge difference is negative when the width of termination-FRF becomes narrower than that of onset-FRF, as shown in Fig. 2F.

The high-edge difference was plotted against the low-edge difference in Fig. 3A for the 330 sustained-response cells. The plot was characterized by a continuum concentrating on the zero point and spreading in the second quadrant, indicating that onset-FRF is generally similar to or wider than termination-FRF; that is, sustained-response cells usually had an unaltered or damped pattern of onset- versus termination-FRF. Figure 3B illustrates the distribution of edge differences for such sustained-response cells without offset responses: most cells (117/152 = 78.0%) showed little difference between onset- and termination-FRF (both low- and high-edge differences were <1 octave). In contrast, for sustained-response cells with offset responses (Fig. 3C), the majority (110/178) of cells had a damped pattern of onset- versus termination-FRF. It appears that the offset response is common among phasic-response cells and sustained-response cells with salient phasic onset responses. For sustained-response cells with less salient phasic onset responses, the offset response is uncommon.

View larger version (17K): [in this window] [in a new window]   FIG. 3. Distributions of onset- vs. termination-FRF patterns at the best-onset sound pressure level (SPL). A: high-edge difference between onset- and termination-FRF was plotted against the low-edge difference between them in all 330 sustained-response cells. Pair of vertical bars in the corner of each quadrant indicates a model pattern between onset- (left bar) and termination-FRF (right bar) of the cells located in the quadrant. I, II, III, and IV are quadrant numbers. B: distribution of 152 sustained-response cells without offset responses. C: distribution of 178 sustained-response cells with offset responses.

Of the 638 recorded cells, 406 were significantly excited by tonal offsets. Twenty-eight of the 406 cells responded to offsets but not to onsets of tone bursts. These so-called offset response cells (He et al. 1997) are beyond the scope of this study. In total, 378 cells (59.2%) showed excitatory responses to both onsets and offsets of tone bursts. We focused on these 378 "ON–OFF cells" in the following analysis to compare the offset and onset response properties.

Offset-FRF was investigated relative to onset-FRF at the best offset-SPL. Offset-FRF showed diversity in width and frequency range relative to onset-FRF. To quantitatively analyze, the high-edge difference was calculated as the log₂ (high-edge frequency of onset-FRF/high-edge frequency of offset-FRF) for each ON–OFF cell. The low-edge difference was calculated similarly. The high-edge difference was plotted against the low-edge difference in Fig. 4A for 378 ON–OFF cells. The plot was characterized by continuous distribution without clear boundaries between the different FRF-pattern groups. Cell distribution was densest in the second quadrant, characterized by the damped pattern of onset- versus offset-FRF (onset-FRF was wider than offset-FRF). Some cells were located in the first quadrant, which represents a descended pattern (both low and high edges of onset-FRF are higher than those of offset-FRF). Inversely, other cells showed an ascended pattern (both low and high edges of onset-FRF are lower) in the third quadrant. The cells in the fourth quadrant were sparse, indicating that cells with a ramped pattern (onset-FRF is narrower than offset-FRF) were relatively uncommon. There was also a concentration of cells on the axes themselves (cells with low-edge difference of 0 on the y-axis, cells with high-edge difference of 0 on the x-axis). This was caused by the limitation of our stimulus frequency range of 128–16,000 Hz. For example, the low-edge difference is zero in our experimental paradigm even if the real low-edge difference is not zero in the frequency range <128 Hz such as low-edge frequencies 100 Hz at onsets versus 120 Hz at offsets.

View larger version (26K): [in this window] [in a new window]   FIG. 4. Distributions of onset- vs. offset-FRF (A) and termination- vs. offset-FRF (B) patterns at best offset-SPL. A: high-edge difference between onset- and offset-FRF was plotted against the low-edge difference between them in all 378 ON–OFF cells. Pair of vertical bars in the corner of each quadrant indicates a model pattern between onset- (left bar) and offset-FRF (right bar) of the cells located in the quadrant. B: high-edge difference between termination- and offset-FRF was plotted against the low-edge difference between them in 174 sustained-response cells with termination-FRF.

Termination- versus offset-FRF patterns

Offset-FRF was also investigated relative to termination-FRF in sustained-response cells at best offset-SPL. In this case, the high-edge difference was calculated as the log₂ (high-edge frequency of termination-FRF/high-edge frequency of offset-FRF) and the low-edge difference was calculated as the log₂ (low-edge frequency of termination-FRF/low-edge frequency of offset-FRF). The high-edge difference was plotted against the low-edge difference in Fig. 4B for the 174 cells with termination-FRF (sustained-response ON–OFF cells) at the best offset-SPL. The plot was characterized by continuous distribution excluding the second quadrant. This suggests that offset-FRF is usually lower (first quadrant), higher (third quadrant), or wider (fourth quadrant) than termination-FRF in sustained-response cells. Only 34.5% of sustained-response cells had an offset-FRF similar to termination-FRF (high- and low-edge differences are both <1 octave). For these cells, offset-FRF might also be considered as the prolongation of sustained responses rather than responses regenerated by the sound offset.

Effects of stimulus intensity on FRF patterns

Best onset-SPL and best offset-SPL were investigated at the range of 20- to 80-dB SPLs in the 378 ON–OFF cells. The former was 67.6 ± 16.8 dB (mean ± SD), whereas the latter was 65.3 ± 14.3 dB SPL and, although the difference was small, it was statistically significant (P < 0.05, t-test). On the other hand, the threshold of offset responses was significantly higher than that of onset responses (44.0 ± 14.8 vs. 32.3 ± 9.3dB, P < 0.01, t-test), indicating that more sound energy is necessary to evoke offset responses than onset responses.

We examined the effects of the sound level on the onset- versus offset-FRF patterns of 253 ON–OFF cells, in which both onset- and offset-FRF were present at the same stimulus intensity and such onset–offset response pairs were found at multiple intensities. Example responses of a representative cell are illustrated in Fig. 5A. In this cell, although changing the sound level resulted in a change in the width and frequency range of onset- and offset-FRF, the pattern of onset- versus offset-FRF, which is specific for the unaltered pattern at the best offset-SPL, remained largely unchanged across 40- to 80-dB SPL.

View larger version (16K): [in this window] [in a new window]   FIG. 5. Effects of stimulus intensity on FRF patterns. A: onset- (white bars) and offset-FRF (shaded bars) at different stimulus intensities in an example cell. Asterisk marks the best offset-SPL of the cell. Note that the onset- vs. offset-FRF pattern in the cell remains unchanged at the shift of stimulus intensity. B: percentage overlapping between onset- and offset-FRF at different SPLs. C: deviation of percentage overlapping at different stimulus intensities. Percentage overlapping at the best offset-SPL (asterisks in A) was normalized as 0% to calculate the amount of deviation (ordinate), and the best offset-SPL was normalized as 0 dB (abscissa). D: population average (253 cells) of the deviation of percentage overlapping between onset- and offset-FRF at different stimulus intensities. Dots represent SD of data in each SPL. E: population average (139 cells) of the deviation of percentage overlapping between termination- and offset-FRF at different stimulus intensities.

We then subtracted the percentage overlapping at each tested SPL from that at best offset-SPL, calculating the "deviation of percentage overlapping." The deviation of percentage overlapping was plotted against the stimulus level relative to best offset-SPL (Fig. 5C). The maximum deviation of percentage overlapping was <10% in the example cell. This was also the case for all 253 cells tested: mean and SD of the deviation of percentage overlapping was plotted against the stimulus intensity in Fig. 5D, where the deviation of overlapping was very small regardless of the stimulus intensity. The results indicate that the onset- versus offset-FRF pattern was relatively sound level invariant.

There were also 139 cells showing both termination- and offset-FRF at multiple intensities. The "percentage overlapping" and "deviation of percentage overlapping" between termination- and offset-FRF were also investigated at each SPL tested. The mean and SD of the deviation of percentage overlapping in these cells were also very small (Fig. 5E). Thus the termination- versus offset-FRF pattern was relatively independent of the sound level.

Response time course

As shown in Fig. 1, E and F, we constructed driven-rate histograms showing the time courses of onset and offset responses in a given cell. To compare the onset and offset responses, the onset-response time course was constructed at the cell's best onset-SPL, whereas the offset-response time course was constructed at the cell's best offset-SPL. Then, the onset- and offset-response time courses of all 378 ON–OFF cells were averaged, respectively. The mean responses were superimposed by aligning the stimulus onset and offset at time 0 in Fig. 6A. Both onset and offset responses were characterized by a quick rise in the "peak driven rate" in a relatively short period ("peak latency": 21.5 ms after sound onset and 19.5 ms after sound offset). The peak driven rate of the offset response (20.4 spikes/s) was about half that of the onset response (35.9 spikes/s). After the peak response, the onset response sharply decayed to half of its peak, taking 48.5 ms ("half-decay time"). On the other hand, the half-decay time of the offset response (74.5 ms) was longer than that of the onset response. We noted that the onset response amplitude never fell to 0 during the period of sound stimulus because some cells had sustained responses during tonal stimuli, whereas the offset response gradually decayed close to 0 with the increase of time.

View larger version (14K): [in this window] [in a new window]   FIG. 6. Population average of onset- and offset-response time courses at the best onset- and offset-SPL, respectively. A: onset- (blue) and offset- (red) response time courses averaged from 378 ON–OFF cells. Vertical broken line and 0 in abscissa represent stimulus offset for the offset response and stimulus onset for the onset response. B: onset-response time courses averaged from 204 phasic-response cells (P cells, red), 174 sustained-response ON–OFF cells (S cells with offset, blue), and 152 sustained-response cells without offset responses (S cells without offset, black). C: offset-response time courses averaged from 204 phasic-response ON–OFF cells (P cells, red line) and 174 sustained-response cells with offset responses (S cells, blue line).

The time course of sustained-response cells with offset responses (sustained-response ON–OFF cell, blue line in Fig. 6B) was quite salient in the early section of stimuli (<100 ms) and later became flat. In contrast, the mean onset-response time course of the 152 sustained-response cells without offset responses (black line in Fig. 6B) was relatively flat throughout the duration of sound stimuli. This is in line with the result of Fig. 3, B and C, suggesting that offset responses usually co-occur with phasic onset responses.

We further plotted the mean offset-response time courses of 204 phasic- and 174 sustained-response cells separately in Fig. 6C, which showed a similar shape of offset-response time course, except that the time course of phasic-response cells had higher peak amplitude (25.3 vs. 18.1 spikes/s). Because the phasic-response cells also had higher onset response amplitude (Fig. 6B), the offset response amplitude correlated with the onset response amplitude. This was confirmed by the results that the peak driven rate of onset responses was correlated with that of offset responses in the 378 individual cells (r = 0.64, P < 0.01; not illustrated).

Effects of stimulus intensity on response time course

The response time courses at different stimulus intensities were averaged when the driven-rate amplitudes were >2 SD of the background discharge level (351 cells at 70 dB, 332 cells at 50 dB, and 238 cells at 30 dB SPL). By comparing the population average, we found that the change of stimulus intensity modulated the peak amplitude of the onset response: the peak amplitude was, as mentioned earlier, 36.0 spikes/s at 70 dB SPL, and gradually decreased to 31.3 and 27.2 spikes/s at 50 and 30 dB SPL, respectively (Fig. 7A). The shift of stimulus intensity resulted in little change of peak latency (range: 3 ms) and half-decay time (range: 13 ms).

View larger version (16K): [in this window] [in a new window]   FIG. 7. Effects of stimulus intensity on the response time course. A: onset-response time courses averaged when driven rates were >2 SD of the background discharge level (351 cells at 70 dB, 332 cells at 50 dB, and 238 cells at 30 dB SPL). B: offset-response time courses (354 cells at 70 dB, 192 cells at 50 dB, and 122 cells at 30 dB SPL).

Suppression of spontaneous firings preceding offset responses

A1 neurons in awake cats usually have spontaneous firing activities (mean and SD, 7.7 ± 8.4 spikes/s in 5-ms time window, 378 cells). We noted that offset responses were often preceded by suppressive responses under the spontaneous firing level during tonal stimuli (i.e., Fig. 1A). We then investigated the relationship between offset responses and suppression activity. Of the 378 ON–OFF cells, 202 had a low-tail level of 1% probability (blue line in Fig. 1C) >1 spike/s in the background spike-density histogram; such cells with relatively high background spike rates were further analyzed for spike-activity suppression. In the remaining 176 cells, the level of 1% low-tail probability was <1 spike/s (i.e., presented in Fig. 2, C and D), making it impossible to analyze spike-activity suppression. Spike suppression under the spontaneous activity level was statistically examined by measuring sFRF just before the end of the sound stimulus (see METHODS). Offset-FRF was compared with significant sFRF in the time window of 15–115 ms preceding the sound offset. The analysis time window of sFRF was shifted 15 ms ahead of the sound offset to avoid the effects of the zero-padded procedure on constructing the spectrotemporal spike-activity diagram.

We found that 39.6% (80/202 cells) ON–OFF cells had significant sFRF preceding the offset responses at best offset-SPL. The other 60.4% (122/202 cells) cells had no significant sFRF or the sFRF did not just precede the sound offset. The mean response time course across the frequency range of offset-FRF is plotted in Fig. 8A for cells with sFRF (red line) and without sFRF (blue line), respectively. The mean driven rate for cells with sFRF was <0 from 58 to 500 ms poststimulus onset, reflecting that spike activity was continuously suppressed during sound stimuli. The peak driven rate of the offset response was obviously higher in these cells than in cells without sFRF (31.7 vs. 23.6 spikes/s), whereas the peak latency and decline of offset response in these two groups were similar. This suggests that sFRF preceding the sound offset facilitated the cell's response to sound offset.

View larger version (17K): [in this window] [in a new window]   FIG. 8. Suppression of spontaneous firing preceding offset responses. A: response time courses averaged from 80 ON–OFF cells with preceding spike suppression (red) and 122 cells without suppression (blue). Vertical broken lines indicate stimulus onset and offset at 0 and 0.5 s, respectively. Note that the offset-response amplitude of cells with suppression is higher than that of cells without suppression; B: distribution of suppressive FRF (sFRF)- vs. offset-FRF patterns. High-edge difference between sFRF and offset-FRF was plotted against the low-edge difference between them in 80 ON–OFF cells with suppression. Pair of vertical bars in the corner of each quadrant indicates a model pattern between sFRF- (left bar) and offset-FRF (right bar) of cells located in the quadrant.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
As mentioned in the INTRODUCTION, the responses of A1 neurons to pure tones in barbiturate-anesthetized animals are usually described as having onset responses (Heil 1997a,b; Mendelson et al. 1997; Phillips and Hall 1990; Phillips et al. 1994, 2002; Schreiner and Sutter 1992; Sutter and Schreiner 1991, 1995). Thus A1 neurons are believed not to carry much information about tonal offsets (Phillips et al. 2002). In awake animals (Brugge and Merzenich 1973; Chimoto et al. 2002; Evans and Whitfield 1964; Moshitch et al. 2006; Pfingst and O'Connor 1981; Recanzone 2000) and in ketamine- (Volkov and Galazjuk 1991) and halothane- (Moshitch et al. 2006) anesthetized animals, however, tonal offset responses were commonly found in A1 neurons. Nevertheless, systematic studies focused on tonal offset responses never reported in awake animals. This study is the first to describe the details of tonal offset response characteristics compared with onset responses in awake animals.

Technical consideration of pure-tone stimulus paradigm

Previously, pure-tone stimuli were usually presented in short durations of 50–200 ms to investigate the response properties of auditory cortex neurons. In this study, we used a longer duration (500 ms) of pure-tone stimuli for the following reasons. First, a number of A1 neurons of awake cats like the cell in Fig. 2F showed different response durations to different tonal frequencies. To investigate the relationship between the tonal response pattern and the occurrence of the offset responses, we adopted relatively a long stimulus duration of 500 ms. If we use a short stimulus duration (i.e., 100 ms), there are no more differential response durations in response to differential stimulus frequencies in the sustained-response cell in Fig. 2F. Second, a long stimulus duration is necessary to clearly distinguish between tonal offset and onset responses. Considering the mean response time courses of onset (peak-latency + half-decay time = 70 ms) and offset (peak-latency = 19.5 ms) responses (Fig. 6A), if we use short stimulus duration (i.e., 50 ms), onset responses would merge into offset responses, making it difficult to distinguish between onset and offset responses. In our preliminary experiments, we observed that offset responses merged with onset responses in all 17 phasic cells tested at 50-ms duration (unpublished observation). Third, relatively long stimulus duration is also useful to detect spike-activity suppression between onset and offset responses (Fig. 1D).

Previously, pure-tone stimuli were usually presented on the logarithmic frequency scale, which requires a relatively small number of stimulus frequencies to investigate the frequency range of FRF. In this study, we presented tones in fine linear steps (usually 128 Hz, sometimes 64, 32, 16, or 8 Hz) to calculate the frequency ranges of onset- and offset-FRF in fine sample intervals.

Cells with dynamic FRF show offset responses

This study has shown that the offset response is found in most phasic-response cells (72.9%) but in half of sustained-response cells (53.9%). This does not mean that the sustained response interferes with offset responses. Rather, the offset response may be related to the occurrence of phasic responses at tonal onsets because sustained-response cells without offset responses tended to have stable FRF throughout the stimulus period, whereas those with offset responses tended to have salient phasic-response components at tonal onsets (Fig. 3). Thus the offset response is commonly found in cells showing dynamic FRF but is uncommon in cells showing static FRF.

ON–OFF neurons have dynamic filtering properties

We have found that onset- and offset-FRF are usually located at different frequencies in a given cell. Our findings are in good accordance with a previous report in abstract form (Tian and Rauschecker 2003), which described the coexistence of offset and onset responses in the same A1 neurons and inconsistent tunings of offset and onset responses in the majority of such ON–OFF neurons in awake monkeys.

For the first time, we showed the details of frequency-tuning patterns of offset responses compared with onset responses in the same single cells (Fig. 4). Offset-FRF was similar to onset-FRF (unaltered pattern) in about one third of cells. The other two thirds of cells showed diverse combinations of offset- and onset-FRF; offset-FRF was narrower (damped pattern), higher (ascended pattern), lower (descended pattern), or wider (ramped pattern) than onset-FRF. One noticeable trend is that the damped pattern was most common and the ramped pattern was most rare when offset-FRF was compared with onset-FRF (Fig. 4A). This tendency suggests that ON–OFF neurons tend to prefer a complex sound with relatively wide spectral width at sound onset and relatively narrow spectral width at offset. At present, however, there is no evidence for such an interpretation of the findings and this suggestion should be investigated in future studies.

This trend was reversed when offset-FRF was compared with termination-FRF (Fig. 4B). This could be expected because termination-FRF was usually narrower than onset-FRF if they coexisted in the same ON–OFF neurons (Fig. 3B). Systematic shift of the stimulus level did not change the onset- versus offset-FRF-pattern of a given cell (Fig. 5).

This large and diffuse diversity of onset- versus offset-FRF patterns and level tolerance suggests a functional role of ON–OFF neurons, which may serve as dynamic filters for natural sounds. ON–OFF neurons are not just simple filters limited to sound onsets but are complex filters operating on both sound onsets and offsets. The filtering property changes depending on the stimulus time in individual neurons. The ranges of filtering frequencies are diffusely prepared among the ON–OFF neuron group. A1 may have a strategy to prepare a variety of ON–OFF neurons with various combinations of time-dependent FRF for detecting offsets and onsets of the natural sounds that momentarily vary their acoustic parameters.

Time resolution of ON–OFF neurons

We showed the details of the time courses of offset responses compared with onset responses (Fig. 6). The difference between the mean peak latencies of offset and onset responses was very small (2 ms), suggesting that A1 cells had similar precision to detect sound offset and onset. The half-decay time was longer in offset responses (74.5 ms) than in onset responses (48.5 ms), which could be attributable to the presence of suppression mechanisms after onset responses in some cells (red line in Figs. 6B and 8A). It is possible that this finding has direct relevance to backward masking, wherein the onset response of a new sound would disrupt the normally prolonged neural processing associated with the offset of a preceding sound.

Although offset responses were characterized as having lower peak amplitude than that of onset responses, spike activity at sound offset was highly lifted from the background firing rate (driven rate was 20.4 spikes/s). The offset response in A1 was salient enough to effectively encode sound offset signals.

Considering the 21.5-ms peak latency and 48.5-ms half-decay time of the mean onset response (21.5 + 48.5 = 70 ms), the detection of sound offsets at 50 ms after sound onset (50-ms sound duration + 19.5-ms offset peak latency = 69.5 ms) is around the time limit of resolution resulting from the fusion of offset- and onset-response time courses. It appears that the offset responses are precise and salient enough to effectively encode sound offsets when the sound duration is relatively long (500 ms). When the sound duration is relatively short (<50 ms), the offset responses would merge with onset responses, showing 50-ms time resolution of ON–OFF cells.

Active spike-generation mechanism generates offset responses

Offset responses, if any, were suggested to reflect the rebound from inhibition but not active spike evocation (Phillips et al. 2002). Thus this study tested for the presence of suppression preceding offset responses. We found that spontaneous spike activities during the stimulus period preceding offset responses were suppressed in some cells (40%) with a high spontaneous firing rate and the preceding suppression facilitated the amplitude of offset responses (Fig. 8A), suggesting that some offset responses relate to the preceding suppression; however, the same results were also interpretable as that offset responses were generated without preceding suppression in the remaining cells (60%). In addition, by investigating the frequency range difference between sFRF and offset-FRF in each cell, we found that offset responses were generated irrespective of the preceding spike suppression (Fig. 8B). Thus the preceding suppression, although it contributed to the facilitated amplitude of offset responses, did not account for all cases of offset response generation. Our findings support the presence of active spike-generation mechanisms underlying offset responses. Considering that offset responses are similar in latency to onset responses, and that the offset response amplitude correlated with the onset response amplitude, offset and onset responses may be generated by the same neural mechanism responsible for the detection of acoustic parameters, such as the rapid change of signal amplitude (rising or falling).

	GRANTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
This work was supported by a grant from the Ministry of Education, Science, Culture, Sports and Technology, Japan.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
We thank N. Yaguchi for technical assistance.

	FOOTNOTES

 
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Address for reprint requests and other correspondence: Y. Sato, Department of Physiology, Interdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, Chuo, Yamanashi 409-3898, Japan (E-mail: yu_sato{at}yamanashi.ac.jp)

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Barbour DL, Wang X. Auditory cortical responses elicited in awake primates by random spectrum stimuli. J Neurosci 23: 7194–7206, 2003a.[Abstract/Free Full Text]

Barbour DL, Wang X. Contrast tuning in auditory cortex. Science 299: 1073–1075, 2003b.[Abstract/Free Full Text]

Brugge JF, Merzenich MM. Responses of neurons in auditory cortex of the macaque monkey to monaural and binaural stimulation. J Neurophysiol 36: 1138–1158, 1973.[Free Full Text]

Chimoto S, Kitama T, Qin L, Sakayori S, Sato Y. Tonal response patterns of primary auditory cortex neurons in alert cats. Brain Res 934: 34–42, 2002.[CrossRef][ISI][Medline]

Evans E, Whitfield I. Classification of unit responses in the auditory cortex of the unanaesthetized and unrestrained cat. J Physiol 171: 476–793, 1964.[Free Full Text]

Gutschalk A, Patterson RD, Rupp A, Uppenkamp S, Scherg M. Sustained magnetic fields reveal separate sites for sound level and temporal regularity in human auditory cortex. Neuroimage 15: 207–216, 2002.[CrossRef][ISI][Medline]

Hamada T, Iwaki S, Kawano T. Speech offsets activate the right parietal cortex. Hear Res 195: 75–78, 2004.[CrossRef][ISI][Medline]

Hari R, Pelizzone M, Makela JP, Hallstrom J, Leinonen L, Lounasmaa OV. Neuromagnetic responses of the human auditory cortex to on- and offsets of noise bursts. Audiology 26: 31–43, 1987.[ISI][Medline]

Harms MP, Guinan JJ Jr, Sigalovsky IS, Melcher JR. Short-term sound temporal envelope characteristics determine multisecond time patterns of activity in human auditory cortex as shown by fMRI. J Neurophysiol 93: 210–222, 2005.[Abstract/Free Full Text]

He J, Hashikawa T, Ojima H, Kinouchi Y. Temporal integration and duration tuning in the dorsal zone of cat auditory cortex. J Neurosci 17: 2615–2625, 1997.[Abstract/Free Full Text]

Heil P. Auditory cortical onset responses revisited. I. First-spike timing. J Neurophysiol 77: 2616–2641, 1997a.[Abstract/Free Full Text]

Heil P. Auditory cortical onset responses revisited. II. Response strength. J Neurophysiol 77: 2642–2660, 1997b.[Abstract/Free Full Text]

Ison JR, Allen P. A diminished rate of "physiological decay" at noise offset contributes to age-related changes in temporal acuity in the CBA mouse model of presbycusis. J Acoust Soc Am 114: 522–528, 2003.[CrossRef][ISI][Medline]

Mendelson JR, Schreiner CE, Sutter ML. Functional topography of cat primary auditory cortex: response latencies. J Comp Physiol A Sens Neural Behav Physiol 181: 615–633, 1997.[CrossRef][Medline]

Moshitch D, Las L, Ulanovsky N, Bar-Yosef O, Nelken I. Responses of neurons in primary auditory cortex (A1) to pure tones in the halothane-anesthetized cat. J Neurophysiol 95: 3756–3769, 2006.[Abstract/Free Full Text]

Okada T, Honda M, Okamoto J, Sadato N. Activation of the primary and association auditory cortex by the transition of sound intensity: a new method for functional examination of the auditory cortex in humans. Neurosci Lett 359: 119–123, 2004.[CrossRef][ISI][Medline]

Pfingst BE, O'Connor TA. Characteristics of neurons in auditory cortex of monkeys performing a simple auditory task. J Neurophysiol 45: 16–34, 1981.[Free Full Text]

Phillips DP, Hall SE. Response timing constraints on the cortical representation of sound time structure. J Acoust Soc Am 88: 1403–1411, 1990.[CrossRef][ISI][Medline]

Phillips DP, Hall SE, Boehnke SE. Central auditory onset responses, and temporal asymmetries in auditory perception. Hear Res 167: 192–205, 2002.[CrossRef][ISI][Medline]

Phillips DP, Semple MN, Calford MB, Kitzes LM. Level-dependent representation of stimulus frequency in cat primary auditory cortex. Exp Brain Res 102: 210–226, 1994.[ISI][Medline]

Pind J. Auditory and linguistic factors in the perception of voice offset time as a cue for preaspiration. J Acoust Soc Am 103: 2117–2127, 1998.[CrossRef][ISI][Medline]

Qin L, Chimoto S, Sakai M, Sato Y. Spectral-shape preference of primary auditory cortex neurons in awake cats. Brain Res 1024: 167–175, 2004a.[CrossRef][ISI][Medline]

Qin L, Kitama T, Chimoto S, Sakayori S, Sato Y. Time course of tonal frequency-response-area of primary auditory cortex neurons in alert cats. Neurosci Res 46: 145–152, 2003.[ISI][Medline]

Qin L, Sakai M, Chimoto S, Sato Y. Spectral-edge sensitivity of primary auditory cortex neurons in alert cats. Brain Res 1014: 1–13, 2004b.[CrossRef][ISI][Medline]

Qin L, Sakai M, Chimoto S, Sato Y. Interaction of excitatory and inhibitory frequency-receptive fields in determining fundamental frequency sensitivity of primary auditory cortex neurons in awake cats. Cereb Cortex 15: 1371–1385, 2005.[Abstract/Free Full Text]

Reale RA, Imig TJ. Tonotopic organization in auditory cortex of the cat. J Comp Neurol 192: 265–291, 1980.[CrossRef][ISI][Medline]

Recanzone GH. Response profiles of auditory cortical neurons to tones and noise in behaving macaque monkeys. Hearm Res 150: 104–118, 2000.[CrossRef]

Schlauch RS, Ries DT, DiGiovanni JJ. Duration discrimination and subjective duration for ramped and damped sounds. J Acoust Soc Am 109: 2880–2887, 2001.[CrossRef][ISI][Medline]

Schreiner CE, Sutter ML. Topography of excitatory bandwidth in cat primary auditory cortex: single-neuron versus multiple-neuron recordings. J Neurophysiol 68: 1487–1502, 1992.[Abstract/Free Full Text]

Shamma SA, Symmes D. Patterns of inhibition in auditory cortical cells in awake squirrel monkeys. Hear Res 19: 1–13, 1985.[CrossRef][ISI][Medline]

Sutter ML, Schreiner CE. Physiology and topography of neurons with multipeaked tuning curves in cat primary auditory cortex. J Neurophysiol 65: 1207–1226, 1991.[Abstract/Free Full Text]

Sutter ML, Schreiner CE. Topography of intensity tuning in cat primary auditory cortex: single-neuron versus multiple-neuron recordings. J Neurophysiol 73: 190–204, 1995.[Abstract/Free Full Text]

Takahashi H, Nakao M, Kaga K. Cortical mapping of auditory-evoked offset responses in rats. Neuroreport 15: 1565–1569, 2004.[CrossRef][ISI][Medline]

Tian B, Rauschecker JP. On- and off-responses in Rhesus monkey auditory cortex. Soc Neurosci Abstr 182. 15, 2003.

Volkov IO, Dembnovetskii OF, Galaziuk AV. Characteristics of the responses of auditory cortex neurons in the cat to tonal stimulation during nembutal anesthesia and after recovery from it. Neirofiziologiia 17: 728–737, 1985.[Medline]

Volkov IO, Galazjuk AV. Formation of spike response to sound tones in cat auditory cortex neurons: interaction of excitatory and inhibitory effects. Neuroscience 43: 307–321, 1991.[CrossRef][ISI][Medline]

This article has been cited by other articles:

				L. Qin, J. Y. Wang, and Y. Sato Representations of Cat Meows and Human Vowels in the Primary Auditory Cortex of Awake Cats J Neurophysiol, May 1, 2008; 99(5): 2305 - 2319. [Abstract] [Full Text] [PDF]

This Article Abstract