| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
The Journal of Neurophysiology Vol. 80 No. 3 September 1998, pp. 1598-1604
Copyright ©1998 by the American Physiological Society
RAPID COMMUNICATION
Divison of Biology 216-76, California Institute of Technology, Pasadena, California 91125
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Hartmann, Mitra J. and James M. Bower. Oscillatory activity in the cerebellar hemispheres of unrestrained rats. J. Neurophysiol. 80: 1598-1604, 1998. We recorded multiunit neural activity in the granule cell layer of cerebellar folium Crus IIa in unrestrained rats. Seven- to 8-Hz oscillatory activity was seen during behavioral states in which the animal was immobile; any movement the animal made coincided with termination of the oscillations. However, nearly one-third of oscillatory episodes appeared to cease spontaneously, in the absence of any observable sensory input or movement. Oscillations were synchronized both within and between cerebellar hemispheres, demonstrating precise temporal coordination among multiple, bilateral levels of the somatosensory system. We interpret these data in the context of similar oscillations observed in other brain structures and suggest that the oscillations are an underlying dynamic property of the entire somatosensory network.
In the awake rat, 7- to 12-Hz synchronous activity has been seen at multiple levels of the sensory trigeminal system, including the spinal trigeminal nucleus (SpV), ventrobasal thalamus (VB), and primary somatosensory (S1) cortex (Buzsáki 1991 Five female albino Sprague-Dawley rats, aged 4-10 mo, were implanted with microwire electrode arrays, either in left Crus IIa (n = 3) or in both right and left Crus IIa (n = 2). During implantation animals were anesthetized with xylazine/ketamine-hydrochloride delivered intramuscularly (70 mg/kg ketamine, 3.5 mg/kg xylazine, 0.7 mg/kg acepromazine) and pentobarbital sodium delivered intraperitoneally (20 mg/kg). During the surgery five or six stainless steel screws were placed over neocortical areas and covered with dental acrylic to form a stable base. Next, Crus IIa was exposed and the grid of wire electrodes (either 18-µm-diam platinum-iridium or 50-µm-diam stainless steel) was fixed with acrylic above the exposure. The electrodes were lowered and fixed in position, and the receptive field at each recording site was determined. During implantation surgery, we confirmed that the responses were physiologically characteristic of the Crus IIa GCL. The reference electrode was a stainless steel (76-µm diam) wire laid flat over the entire length of Crus IIa. All animal procedures were approved by Caltech's Animal Use Committee.
We recorded oscillatory activity in Crus IIa from all five implanted rats. Figure 1A shows simultaneous recordings from three electrodes in left Crus IIa of an awake animal during an oscillatory episode. The three electrodes were arranged in a mediolateral line and spaced ~500 µm apart. As shown in the enlarged timescale in Fig. 1B, each oscillation in the local field potential was accompanied by a burst of multiunit activity. Each oscillatory peak usually had two or more smaller subsidiary peaks (arrowheads).
We report for the first time the presence of 7- to 8-Hz oscillatory activity in the GCL of the rat lateral cerebellar hemispheres. These oscillations are similar in frequency to oscillations previously described in rat S1, VB, and SpV (Buzsáki 1991
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
; Kandel and Buzsáki 1993, 1997; Nicolelis et al. 1995; Semba et al. 1980, 1984). In addition, it is well established that 7- to 12-Hz spindle oscillations occur in many neocortical areas during the early stages of drowsiness and sleep (Contreras et al. 1997; Hammond et al. 1979; Kandel and Buzsáki 1993, 1997; Steriade and Deschenes 1984; Steriade and Llinás 1988). In this study, we examined oscillations in tactile regions of the granule cell layer (GCL) of the lateral cerebellar hemispheres (Crus IIa) in unrestrained rats. This cerebellar folium receives projections from S1 via the pontine nuclei (Bower and Woolston 1983; Brodal 1983; Mihailoff et al. 1981) as well as direct tactile input from the trigeminal sensory nuclei (Huerta et al. 1983; Watson and Switzer 1978; Woolston et al. 1981, 1982). Our results demonstrate clear 7- to 8-Hz field potential oscillations in the GCL of Crus IIa, synchronized both within and between cerebellar hemispheres. Oscillations were present only during periods of immobility, and any movement invariably disrupted the oscillatory activity. However, a significant percentage of oscillatory episodes ceased spontaneously, in the absence of observable movement or sensory input. We therefore propose that the oscillations reflect the baseline dynamic state of the entire tactile sensory system; although they can be interrupted by sensory input or movement, they do not under natural conditions appear to anticipate such interruption.
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
). Maximum amplitude responses were found between 400 and 700 µm below the pial surface. A high-input-impedance preamplifier (Microprobe, CFP-1020) mounted directly on the animal's head carried neural signals to a custom-built amplifying system with a minimum of mechanical and electrical artifact. Signals were amplified and filtered in analog between 1 Hz and 5 kHz and collected at a 
10-kHz sampling rate. Neural data were synchronized with behavioral data in real time with the use of a custom-built video mixer (Rasnow et al. 1997). During subsequent computer analysis, data were digitally filtered either between 1 and 300 Hz (field potential activity) or between 300 and 3,000 Hz (multiunit burst activity). To detect and quantify periods of oscillation, continuous neural recordings were divided into 1-s trials and processed through a standard fast Fourier transform (FFT; Matlab v5.0.0 1996, the MathWorks). The power spectrum was taken to be the square of the absolute value of the FFT. Any individual trial was considered to include oscillatory activity if any peak in the power spectrum accounted for >8% of the total power.
), but in brief, the small size (5-6 µm) of granule cells precludes single-cell isolation in awake behaving animals. It is very likely, however, that some component of the multiunit activity reflects the activation of granule cells because these same signals have been shown to precede and predict short-latency simple spike responses in overlying Purkinje cells (Bower and Woolston 1983; Jaeger and Bower 1994).
; Huang et al. 1991; Jaeger and Bower 1994; Joseph et al. 1978; Morissette and Bower 1996; Shambes et al. 1978; Welker 1987). Depth profiles in anesthetized animals have demonstrated that these type of responses are not found in the Purkinje cell or molecular layers and that GCL responses are well isolated with the use of either 20- or 50-µm wires. We confirmed that neural responses were localized to the GCL at three different stages during experiments: first, during the implantation surgery; second, during recording sessions after recovery from the surgery; and, third, in several rats, immediately before euthanasia.
). Prior to euthanasia, the animal was anesthetized and electrolytic lesions (
5.0 µA, 10 s) were made at recording sites. The animal was then perfused with phosphate buffer solution and a 4% formaldehyde solution. The cerebellum was extracted, the hemispheres sectioned parasagitally, and the slices stained either with neutral red or cresyl violet. Lesion sites were centered in the middle of the GCL.
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
View larger version (45K):
[in a new window]
FIG. 1.
Oscillatory activity seen in left Crus IIa during awake immobility. Positive potentials (relative to ground) are upward in this and all subsequent figures. A: 3 s of typical oscillatory activity recorded from 3 electrodes spaced ~500 µm apart mediolaterally. Data were filtered between 1 and 300 Hz. Scale bar: 250 µV top trace, 500 µV bottom 2 traces. B: expanded timescale of the data from the most medial electrode in A. Each field potential oscillation (bottom trace, filtered 1-300 Hz) was accompanied by a burst of multiunit activity (top trace, filtered 300-3,000 Hz). Most oscillatory peaks were found to contain two or more subsidiary peaks ( 
). Scale bar: 50 µV top trace, 200 µV bottom trace.
View larger version (52K):
[in a new window]
FIG. 2.
A: oscillation frequency compared among all rats. For each rat, we recorded episodes of oscillatory activity over several days, then divided these episodes into n 1-s trials and averaged the Fourier spectra of these trials. Because field potential amplitude depends on many factors, including electrode impedance, each average spectrum was normalized such that the amplitude of the central (8 Hz) peak exactly equals 1. Oscillatory activity in all rats exhibited a central peak ~8 Hz, with subsidiary peaks near 1.2 and 15.0 Hz. Rat 3 shows the frequency spectra from electrodes implanted bilaterally ( 
, left Crus IIa; - - -, right Crus IIa). B: examples of granule cell layer (GCL) activity from 2 different rats comparing the oscillations observed during immobility (top trace in each graph) with the periodic activity seen during active whisker movements (bottom trace in each graph). See text for details.
) activity. Both cross-correlations were normalized by the variance in the autocorrelation of each signal alone, demonstrating a doubling of the correlation coefficient from 0.35 to 0.70. The increase in synchrony during oscillatory episodes was found to hold uniformly throughout trials taken over several days. Over a span of 13 1-s trials from one rat, the average correlation coefficient between hemispheres during nonoscillatory periods was 0.31 ± 0.16, whereas during oscillatory periods it increased to 0.60 ± 0.09 (Student's t-test significance level <1 × 10
5).
View larger version (65K):
[in a new window]
FIG. 3.
Interhemispheric synchronization during oscillatory episodes and bilateral interruption of oscillatory activity. A: 8 s of continuous recording from right and left cerebellar hemispheres of a bilaterally implanted rat. The 1st episode of oscillatory activity was interrupted when the rat twitched its nose, and the 2nd oscillatory episode terminated with a head movement. B: expanding the timescale of the oscillations in the last 2 s of A shows clearly that the 2 hemispheres are oscillating in phase. C: cross-correlations of the data shown in A, comparing periods of oscillatory and nonoscillatory activity. All correlations were normalized by the autocorrelations of each of the two signals alone. , average cross-correlation for the data between 1-2 and 6.3-8.3 s (periods of oscillation). - - -, cross-correlation for the data from 2 to 6 s (period of nonoscillatory activity). Correlation coefficient during oscillatory periods = 0.70; during nonoscillatory period = 0.35.
). Finally, oscillations at the same frequencies were observed during the early stages of sleep and drowsiness, during recording sessions in the middle of the afternoon (when the rat would normally be asleep). During these sessions the room lights were fully on, the rat was curled into a small motionless ball, and respiration dropped to ~1-1.5 breaths/s. The time course and frequency of oscillations were essentially identical in each of the three behavioral states. During the later stages of sleep we also observed bursts of large, irregular 7- to 12-Hz waves, but these were not included in this analysis.
View this table:
TABLE 1.
Frequency of behaviors coinciding with termination of oscillatory activity
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
; Kandel and Buzsáki 1993, 1997; Nicolelis et al. 1995; Semba et al. 1980, 1984). In addition, the cerebellar oscillations occur during the same behavioral states as in these earlier studies, including both immobile "crouching" or "attentive resting" (Nicolelis et al. 1995; Semba et al. 1980, 1984) and drowsiness (Coenen et al. 1991, 1995; Drinkenburg et al. 1991 1993; Steriade and Llinás 1988; Steriade et al. 1993). The majority of oscillatory episodes, however, occurred during behavioral states between these two extremes that could not be satisfactorily classified as alert or drowsy. Under these conditions, the animal had usually been sitting quietly for 1-15 min. Oscillations occurred only when the animal was immobile and the perioral regions represented in Crus IIa (Bower and Kassel 1990) were not being stimulated. Oscillations immediately ceased on any movement but also (in 32% of trials) appeared to cease spontaneously in the absence of observable movement or sensory input.
), there has been considerable debate concerning both their origin and functional significance (reviewed in Kaplan 1985). Researchers studying thalamus and neocortex generally proposed that they originate in bursting thalamic cells and thalamocortical loops (Buzsáki 1991; Inoue et al. 1993; Steriade et al. 1985; Steriade and Llinás 1988). The discovery of similar patterns in SpV (Nicolelis et al. 1995) and now the cerebellum suggests that a much larger network of somatosensory structures is involved in this oscillatory activity. Because Crus IIa receives projections from both the trigeminal nuclei (Huerta et al. 1983; Woolston et al. 1981, 1982) and S1 (Bower and Woolston 1983; Morissette and Bower 1996), the coherence seen in the cerebellar oscillations implies a coordination in timing throughout different levels of the somatosensory pathway. It is possible that the subsidiary peaks noted within each larger oscillatory peak reflect coordinated activity between these two pathways. In addition, the high degree of bilateral synchrony observed during oscillatory episodes again suggests widespread involvement of the entire somatosensory system. That the oscillations are so precisely timed, over multiple, bilateral levels of the somatosensory system, makes it unlikely that they emanate from a single brain structure and more likely that they have multiple sources and/or are an emergent property of the somatosensory network. Recent work in culture and computer modeling indeed suggested that the cerebellar GCL itself may participate in the generation of oscillations near these frequencies (Maex and DeSchutter 1998; Nuñez et al. 1996).
; Welker 1964), several authors have speculated that they may represent an internal model for whisking or a mechanism to increase sensory reception during whisking (Nicolelis et al. 1995; Semba et al. 1980, 1984). Consistent with these studies, we did on occasion observe whisker tremor (<2°) during some oscillatory episodes, but the majority of oscillations occurred without visible tremor. Furthermore, oscillatory activity did not predict the onset or occurrence of any particular type of movement, including that of the whiskers. Movements that resulted in the termination of oscillations included nose and vibrissae twitches, lip smacks, and head and body movements. Vibrissal twitches, defined as a single back-and-forth deflection (<5°) of the large vibrissae, coincided with only 10% of oscillation terminations. In no instance did sustained whisking activity coincide with the termination of an oscillatory episode.
). Movements of 0.1° would not be observable in standard video recordings, but there are three reasons why the oscillations are highly unlikely to be the result of rhythmic peripheral input or motor activity. First, the oscillations were not present when the animal was sniffing the air rhythmically, even when the sniffing was accompanied by whisker tremor or whisker movements. Second, as shown in Fig. 2B, the periodic GCL activity during low-amplitude whisker movements is different from the GCL oscillations associated with immobility, even when the whisker movements occur in the same frequency range as the oscillations. Finally, previous studies have established that these type of oscillations occur in central sensorimotor structures in the absence of oscillatory activity in more peripheral structures. Semba et al. (1980) showed that 7- to 8-Hz cortical and thalamic oscillatory activity can occur without oscillations in the vibrissal EMG. Nicolelis et al. (1995) demonstrated that this type of synchronous activity occurs in the SpV, VB thalamus, and S1 without oscillatory activity in either the trigeminal ganglion or the principal trigeminal nucleus.
). Consistent with this work, we found GCL oscillations to occur only during immobility and to terminate immediately with any movement. However, unlike the reaching monkeys, our rats were not engaged in a timed behavioral task; rather, they were unrestrained and freely moving. While this paradigm limits our ability to time the oscillations relative to an invariant sensory or motor event, it also puts us in a much better position to examine the correlations of these oscillations with natural behavior. It is clear from both studies that movement or sensory input always disrupt the oscillations, but it is important to distinguish between the following two possibilities. 1) Are the oscillations "anticipatory of" the sensory input or movement that coincides with their termination? 2) Does sensory input or movement simply interrupt a continuous baseline of oscillatory activity that might otherwise cease spontaneously? Our data are more consistent with the latter hypothesis; we found that oscillations were associated with a wide range of behavioral states, and the presence of oscillations did not predict imminent movement or sensory input. Oscillatory episodes had highly variable durations, and the durations did not predict whether, or which type of, movement was to follow. In fact, a large percentage of oscillatory episodes ceased spontaneously, in the absence of any observable movement or sensory input.
; Coenen 1995; Coenen et al. 1991; Drinkenburg et al. 1991, 1993; Robinson and Gilmore 1980; Vergnes et al. 1987). However, it is our view that the tendency of the rat somatosensory system to oscillate at these frequencies does not reflect a pathological state as in the case of epilepsy, but is rather a fundamental and important property of the system as a whole.
(4-12 Hz) frequency of the olfactory system, the oscillations described here reflect a baseline clocking activity of the entire somatosensory system, important for the temporal segmentation of incoming data. We previously proposed such a function for theta oscillations based on our computer models of the olfactory cerebral cortex (Bower 1996; Wilson and Bower 1992). These models suggest that oscillations in olfactory cortex emerge from network properties of the cortex itself and that the olfactory cortex is tuned to oscillate at the frequencies of input it naturally receives (Wilson and Bower 1992). In awake rats oscillations are coincident with sniffing behavior, and this oscillatory activity thus reflects the way in which the olfactory system segments its sensory data stream.
; Morissette and Bower 1996; Simons 1978; Simons and Carvell 1989). The current study shows that oscillatory activity near 10 Hz arises in the somatosensory system of awake animals specifically in the absence of movement and tactile sensory input. We suggest that these oscillations, observed during periods of immobility, directly reflect the mechanism for the ~10-Hz segmentation, and further that this segmentation is an inherent part of computation within cerebral-cortically related systems. As resonant modes of physical systems are often most obvious in the absence of forcing functions, so the baseline oscillations of neural circuits may be most apparent when the network is not under barrage by sensory data. This hypothesis leads to the prediction that there must exist inhibitory feedback loops to prevent uncontrolled oscillations; the epileptic activity observed in some inbred strains of rats may result from deficiencies in this feedback regulation.
). Whisker movements frequency matched to the resonance might result in enhanced signal amplitude, processing speed, and efficiency, but whisking frequencies unmatched to the resonance may result in more complicated functions, as suggested especially by the bottom traces of both graphs in Fig. 2B. Thus, although our hypothesis is consistent with the earlier suggestion that the oscillations might serve to increase sensory reception at the 7- to 8-Hz whisking frequency, the oscillations may be more precisely said to reflect the temporal structure inherent to cortical information processing. In other words, thalamocorticocerebellar loops may serve to ensure that incoming data are processed near 10 Hz, independent of the frequency of the incoming data: the oscillations are simply what we observe when recording the baseline state of this resonant system. That the oscillations do not occur at all times when the rat is immobile suggests that they may be additionally modulated by internal variables without externally observable behavioral correlates, such as subtle fluctuations in arousal.
; Llinás and Yarom 1986; Sasaki et al. 1989) than with the GCL. For example, it was recently suggested that oscillatory activity within the inferior olivary nucleus is the source of periodic timing information controlling motor coordination (Welsh et al. 1995). In our view, however, the tendency for the inferior olive to oscillate near 10 Hz may simply reflect the tendency for a much larger network of cerebral-cortically related structures to oscillate at these frequencies. In this study we have suggested that these underlying dynamics are most apparent in the absence of sensory input or motor behavior; consistent with our data, Keating and Thach (1997) recently reported that climbing fibers do not exhibit oscillatory activity while monkeys are actively performing arm movements. It may very well be, then, that olivary oscillations, like the GCL oscillations described here, are not as apparent when the sensorimotor system is actively receiving input or generating output.
). The results presented here are consistent with this idea.
| |
ACKNOWLEDGEMENTS |
|---|
We thank K. Neville, C. Chan, and A. Poole for help during this project and C. Chee and C. Assad for useful discussions and readings of the manuscript.
This work was supported by the Human Frontier Science Program.
| |
FOOTNOTES |
|---|
Address reprint requests to M. J. Hartmann.
Received 27 January 1998; accepted in final form 5 June 1998.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
origin mainly from n. interpolaris and n. principalis.
Neurosci. Lett.
10: 77-82, 1978.This article has been cited by other articles:
|
|
 |
|
  L. Forti, E. Cesana, J. Mapelli, and E. D'Angelo Ionic mechanisms of autorhythmic firing in rat cerebellar Golgi cells J. Physiol., August 1, 2006; 574(3): 711 - 729. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. S. Soteropoulos and S. N. Baker Cortico-Cerebellar Coherence During a Precision Grip Task in the Monkey J Neurophysiol, February 1, 2006; 95(2): 1194 - 1206. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Lu, M. J. Hartmann, and J. M. Bower Correlations Between Purkinje Cell Single-Unit Activity and Simultaneously Recorded Field Potentials in the Immediately Underlying Granule Cell Layer J Neurophysiol, September 1, 2005; 94(3): 1849 - 1860. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Fontanini and D. B. Katz 7 to 12 Hz Activity in Rat Gustatory Cortex Reflects Disengagement From a Fluid Self-Administration Task J Neurophysiol, May 1, 2005; 93(5): 2832 - 2840. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Courtemanche and Y. Lamarre Local Field Potential Oscillations in Primate Cerebellar Cortex: Synchronization With Cerebral Cortex During Active and Passive Expectancy J Neurophysiol, April 1, 2005; 93(4): 2039 - 2052. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Cheron, D. Gall, L. Servais, B. Dan, R. Maex, and S. N. Schiffmann Inactivation of Calcium-Binding Protein Genes Induces 160 Hz Oscillations in the Cerebellar Cortex of Alert Mice J. Neurosci., January 14, 2004; 24(2): 434 - 441. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Hartmann, N. J. Johnson, R. B. Towal, and C. Assad Mechanical Characteristics of Rat Vibrissae: Resonant Frequencies and Damping in Isolated Whiskers and in the Awake Behaving Animal J. Neurosci., July 23, 2003; 23(16): 6510 - 6519. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. B. Katz and J. E. Steinmetz Psychological functions of the cerebellum. Behav Cogn Neurosci Rev, September 1, 2002; 1(3): 229 - 241. [Abstract] [PDF]
|
|
|
|
|
|
R. Courtemanche, J.-P. Pellerin, and Y. Lamarre Local Field Potential Oscillations in Primate Cerebellar Cortex: Modulation During Active and Passive Expectancy J Neurophysiol, August 1, 2002; 88(2): 771 - 782. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. O'Connor, R. W. Berg, and D. Kleinfeld Coherent Electrical Activity Between Vibrissa Sensory Areas of Cerebellum and Neocortex Is Enhanced During Free Whisking J Neurophysiol, April 1, 2002; 87(4): 2137 - 2148. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Dumoulin, A. Triller, and S. Dieudonne IPSC Kinetics at Identified GABAergic and Mixed GABAergic and Glycinergic Synapses onto Cerebellar Golgi Cells J. Neurosci., August 15, 2001; 21(16): 6045 - 6057. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Ermentrout, J. W. Wang, J. Flores, and A. Gelperin Model for Olfactory Discrimination and Learning in Limax Procerebrum Incorporating Oscillatory Dynamics and Wave Propagation J Neurophysiol, April 1, 2001; 85(4): 1444 - 1452. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. D'Angelo, T. Nieus, A. Maffei, S. Armano, P. Rossi, V. Taglietti, A. Fontana, and G. Naldi Theta-Frequency Bursting and Resonance in Cerebellar Granule Cells: Experimental Evidence and Modeling of a Slow K+-Dependent Mechanism J. Neurosci., February 1, 2001; 21(3): 759 - 770. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Armano, P. Rossi, V. Taglietti, and E. D'Angelo Long-Term Potentiation of Intrinsic Excitability at the Mossy Fiber-Granule Cell Synapse of Rat Cerebellum J. Neurosci., July 15, 2000; 20(14): 5208 - 5216. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. G. Carter and W. G. Regehr Prolonged Synaptic Currents and Glutamate Spillover at the Parallel Fiber to Stellate Cell Synapse J. Neurosci., June 15, 2000; 20(12): 4423 - 4434. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Dieudonne and A. Dumoulin Serotonin-Driven Long-Range Inhibitory Connections in the Cerebellar Cortex J. Neurosci., March 1, 2000; 20(5): 1837 - 1848. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
|
