Brain Research, 172 (1979) 217-228 © Elsevier/~Nolth-HollandBiomedical Press
217
RECEPTIVE FIELDS OF T H A L A M I C N E U R O N S P R O J E C T I N G TO T H E M O T O R C O R T E X IN T H E CAT
H. ASANUMA, K. D. LARSEN and H. YUMIYA The Rockefeller University, New York, N. Y. 10021 (U.S.A.)
(Accepted December 21st, 1978)
SUMMARY The locations and receptive fields of thalamic neurons projecting to the motor cortex were examined and the following results were obtained. (1) Neurons located at the border area between nucleus ventralis lateralis (VL) and nucleus ventralis posterolateralis (VPL) could be activated antidromically from the motor cortex. (2) These neurons received topographically organized somesthetic inputs arising from skin and deep receptors. (3) The receptive fields of neurons in the small area of the motor cortex where these thalamic neurons projected could be examined in 8 instances. In 6 instances, the cortical neurons and the thalamic projection neurons were activated by exactly the same stimuli in the periphery. (4) Removal of the sensory cortex did not significantly change the characteristics of afferent inputs from the periphery to the motor cortex. (5) It is concluded that the motor cortex receives somesthetic inputs directly from the thalamus. The functional role of these inputs was discussed in relation to the known cortical reflexes.
INTRODUCTION In the preceding paper 24 it was shown that neurons in the border region between nucleus ventralis lateralis (VL) and ventralis posterolateralis (VPL) of the thalamus were labeled by retrograde transport of horseradish peroxidase injected into the motor cortex. Neurons in this region receive projections from spinothalamic and cervicothalamic tracts as well as from nucleus Z10,17,2a. The feline spinocervical tract is known to be distinct from the dorsal column-lemniscal tract it and carries afferent impulses from hair and skin as well as deep receptors which are similar to those carried by the
218 dorsal column12,13. Since the results reported in the first paper of this series suggested that the motor cortex receives somesthetic inputs directly from the thalamus, neurons in this border region were suspected of transferring the peripheral inputs directly to the motor cortex. We have examined the above possibility using electrophysiologieal techniques, i.e. first, by antidromically activating these projection neurons from the motor cortex and then by examining their peripheral receptive fields. These neurons will be shown to transfer finely grained and topographically organized peripheral inputs to neurons in the motor cortex. METHODS Altogether 17 cats of either sex weighing between 2.5-3.5 kg were used. Operations were carried out under inhalation anesthesia composed of a mixture of oxygen (50 ~ ) and nitrous oxide (50 ~ ) supplemented with 1.5 ~ halothane. The cat was mounted on a stereotaxic instrument and the skull was opened over the motor and the occipital cortices. In 13 of the cats, two closed chambers were attached over the openings; one was directed to the motor cortex, and the other toward the ventrobasal thalamus using stereotaxic coordinates. Further details of this procedure have been described elsewhere (see ref. 5). In the remaining 4 cats, one chamber was attached over the motor cortex. After the installation of the chambers, a few screws were drilled into the skull and covered with dental cement. Additional screws were embedded in this cement mass and they were used to fix the head. A cannula was inserted into the cisterna magna to prevent swelling of the brain during the experiments. All the wound areas were infiltrated with a long lasting local anesthetic (Zyljectin, Abbott). At the end of the operation, a tranquilizing dose of sodium pentobarbital (Nembutal, 10 mg/kg) was injected intramuscularly, the ear bars and the snout clamp were removed and inhalation anesthesia was discontinued. A photograph of the exposed motor cortex was taken to map the location of later electrode insertion. The experiments were started 1-2 h after the operation. By giving milk before and during the experiments, the animals were usually quiet and cooperative.
Stimulation and recordings Glass insulated tungsten electrodes of the type described by Stoney et al. 81 were used for stimulation as well as recording unitary spikes. An array of 6 electrodes, 2.0 mm apart and aligned in a U-shape, were inserted into the pericruciate cortex surrounding the cruciate sulcus through metal tubes attached to the lid of one of the chambers (Fig. 1). Each electrode was marked with black ink at a distance of 1.5 mm from the tip and manually inserted into the brain to an approximate depth of 1.0 mm or less under the dissection microscope. The electrode was then fixed to the guide tube with dental impression wax and left at the same position throughout the experiment. The sites of cortical insertion were classified as fore- or hindlimb area by the effect of ICMS delivered through each electrode. The ink marks, which were placed 0.5 mm above the cortex, served as indicators of the cortical level. When the brain shrank, the electrodes were re-inserted but when it swelled, the experiments were usually abandoned. These 6
219
"~
lr.
T
Fig. 1. Experimental set-ups. Pulses produced by the window discriminator trigger the oscilloscope. All the signals are displayed through analog delay, therefore, the unitary spikes which produced the trigger pulses can be displayed on the oscilloscope.
electrodes were used for intracortical microstimulation (ICMS) and in some experiments, were also used for recording unit spikes. Recording electrodes were inserted into the ventrobasal area of the thalamus through the second closed chamber attached to the occipital bone (Fig. 1). The insertions were made in sagittal planes with 34° to the frontal plane and were directed toward the border area between VPL and VL as suggested by the preceding experiments 24. The spikes were displayed on the oscilloscope through an analog delay (AD-5, Bak) which enabled visuafization of the same spike on the scope after being used as a trigger pulse (Fig. l). Whenever unit spikes were evoked antidromically by ICMS, the receptive fields of the cells were examined by listening to the tone of pulses generated by the window discriminator, the accuracy of which was confirmed by the display through the analog delay. Lesions were made at all spots where antidromic spikes were recorded by passing negative current of 10 #A for 10 sec.
Histological examinations At the end of the experiments, lesions were made through those cortical electrodes which elicited antidromic spikes in the thalamus. Probe insertions were made into the thalamus near the recording sites. The animals were deeply anesthetized by Nembutal and the brain was perfused with saline followed by 10 % formalin and removed. The motor cortex and the thalamus were frozen and sectioned sagittally in 50/~m thickness. The cortical sections were stained with Kliiver and Barrera's 22 method and the thalamic sections were stained alternately with Kltiver and Barrera's method or with cresyl violet.
220 RESULTS
Sites of cortical stimulation Since the primary purpose of the experiments was to determine whether the thalamic neurons projecting to the motor cortex carry somesthetic information arising from the periphery, it was important to ascertain that ICMS activated only those fibers terminating within the motor cortex. It has been shown that the effective current spread (x, mm) for the thalamic projection fibers in the motor cortex is proportional to the square root of the current (y, #a), i.e., x = V'y/k and the constant (k) is approximately 10004. Since we used a maximum current of 30 #A to prevent etching of the electrode and to protect the tissue from damage z, the maximum effective spread was expected to be 0.17 mm. Since the locations of all of the effective cortical electrodes were confirmed by histological examination (see Methods), and they were never inserted deeper than 1.2 mm, there was almost no possibility that ICMS activated fibers in the white substance. All the sites of effective ICMS were located in the area where there were large pyramidal cells and there was no layer IV (area 47: Hassler and Muhs-Clement19). Identification of thalamocortical projection neurons The recording electrode was inserted into the thalamus while ICMS of 30/tA was delivered through each of the 6 electrodes. It is already known that the projection neurons from VL to the motor cortex do not have discrete somesthetic receptive fields in the periphery 4. Therefore, special attention was focused on those neurons located at the borderline between VL and VPL, i.e. neurons located at the rostral border of the VPL. When the electrode picked up spikes from this region which responded to the ICMS, the cortical electrode responsible for evoking the spikes was determined and the threshold current was measured. The antidromicity of the activation was examined following the criteria suggested by Abzug and Petersonl: (1) latency fluctuation at threshold stimulation is small (less than 0.1 msec); and (2) latency reduction by increasing the stimulus intensity to twice the threshold is small (less than 0.1 msec). In addition, we examined (3) the refractory period (1.3 msec or less) following the previous criteria by Asanuma et al. 4 and whenever possible, (4) collision of antidromic impulses with spontaneous discharges, an example of which is shown in Fig. 5.
% ...I ...I
4
W
u 2¸
0.7
0 9
1I
1.3
1,5
17
2,6
2.8
LAT E NC Y (msec) Fig. 2. Latency distribution of thalamic neurons antidromically activated from the motor cortex. Shaded bars indicate neurons which were not activated by peripheral inputs.
221 A total of 20 neurons satisfied all 4 criteria and an additional 20 neurons satisfied all but criterion no. 4. The latter neurons were not tested for collision because of the lack of spontaneous discharges or difficulty in isolating the cell from simultaneously recorded cells. However, since all the cells which satisfied the other 3 criteria also satisfied the fourth criterion when the collision test was possible, we are confident that all 40 cells were antidromically activated. Fig. 2 illustrates the latency distribution for antidromic activation of these 40 neurons. The majority of the cells were rather evenly distributed between 0.7 and 1.7 msec. Although all of these 40 cells were recorded at the V L - V P L border area, 8 could not be activated by any natural stimulation applied to the body as will be described later. The distribution of latencies of neurons located at VL-VPL border area was wider than that of VL neurons projecting to the motor cortex and sampled by similar criteria 4. The mean latency of VL-VPL neurons carrying peripheral information was 1.27 msec (S.D. 4-0.5) whereas the mean latency of VL neurons projecting to the motor cortex was 0.84 msec (S.D. 4-0.27). This difference may reflect a difference in the size3z of cells in the V L - V P L border areas and VL. Distribution o f cells projecting to the motor cortex The first penetration was usually directed toward the area of L: 5, A: 112o and recordings were started at the level of H: + 5 and continued until the electrode reached the internal capsule, indicated by the appearance of positive spikes. Subsequent penetrations were made at 0.5 mm steps to form a grid around the target area. Fig. 3A shows a histological reconstruction of electrode tracks of a typical experiment. In this example, 4 penetrations were made stereotaxically on the same sagittal plane and 3 of them appeared on the same histological slide. The other penetration (no. 4) was found on the slide 50/~m medial to this section. Altogether 55 cells could be isolated during these penetrations and 3 of them were activated antidromically from the motor cortex. Lesions were made at these 3 recording sites and at the bottom of track no. 3. The projection cell on track no. 4 could not be driven from the periphery, but the cells on tracks 2 and 3 responded to passive movements of joints. These lesions on tracks nos. 2 and 3 may appear to be located inside of the VPL, but it should be noted that the medial border of VPL in this area curves down steeply, hence, they were actually located at the border near VL. Fig. 3C illustrates the receptive fields of neurons encountered during the penetrations. As has already been reported 29, this region of the thalamus is highly topographically organized. The hindlimb was represented in the rostrolateral portion of the nucleus and the representation shifted gradually towards the forelimb when the electrode moved to the mediocaudal region. All the cells encountered in the VPL could be activated by natural stimulation somewhere in the body, and cells located close together were activated from the same part of the body by stimulation of the same modality, several examples of which are shown in Fig. 3. This feature of the similarity of the inputs to adjacent cells allowed estimation of receptive field of a cell even when the spikes of the target cell could not be isolated from the neighboring cells. Table I summarizes the type of adequate stimuli for all the projection neurons
C
A
]
B
1
2
2
3
4
34
,lmm
f
Fig. 3. Histological reconstruction of electrode tracks and distribution of cells along the penetration. A: specimen sagittal section of ventral thalamus at the level of lat. 5.0 mm showing electrode tracks and lesions. The left-most track and hemorrhage at the bottom was due to probe insertion. B: reconstruction of tracks from the histological preparations. Short bars on the tracks are location of neurons with no receptive fields. Long right side bars on the tracks are location of neurons activated from skin receptors. Long left side bars are those activated from deep receptors. Circles indicate locations of lesions made during experiment. Arrows indicate neurons activated by ICMS. C: receptive fields of neurons encountered during the penetrations. Blackened areas indicate receptive fields for hair bending or light touch. Circles indicate deep receptive fields such as pressure or passive joint movement. Numbers on the column correspond to track numbers and each inset corresponds to respective long bar on the track. Further details in the text. TABLE I
Adequate stimuli Parenthesis gives no. of cells judged from receptive fields of neighboring cells.
Site of stimulus
Forelimb area Hindlimb area
No. of cells
Forelimb
Hindlimb
Skin
Deep
Skin
Deep
31 9
9 0
16 (4) 1
0 3
1 2
Undriven
5 3
223
L 4.9 ~
V
'-,~ • w
"11
,," ° ,,'-LI , / 1 { ,,~ oOoX 5 " / / ,, _._,2//
L 5.2 ',, v~ \\ "",,/\ . v~ "lJ >.. N
L5.6 ~ VA ~\ II*"~ ' l /
VB ~
/
H 3 2
.-I
(...
217
RECEPTIVE FIELDS OF T H A L A M I C N E U R O N S P R O J E C T I N G TO T H E M O T O R C O R T E X IN T H E CAT
H. ASANUMA, K. D. LARSEN and H. YUMIYA The Rockefeller University, New York, N. Y. 10021 (U.S.A.)
(Accepted December 21st, 1978)
SUMMARY The locations and receptive fields of thalamic neurons projecting to the motor cortex were examined and the following results were obtained. (1) Neurons located at the border area between nucleus ventralis lateralis (VL) and nucleus ventralis posterolateralis (VPL) could be activated antidromically from the motor cortex. (2) These neurons received topographically organized somesthetic inputs arising from skin and deep receptors. (3) The receptive fields of neurons in the small area of the motor cortex where these thalamic neurons projected could be examined in 8 instances. In 6 instances, the cortical neurons and the thalamic projection neurons were activated by exactly the same stimuli in the periphery. (4) Removal of the sensory cortex did not significantly change the characteristics of afferent inputs from the periphery to the motor cortex. (5) It is concluded that the motor cortex receives somesthetic inputs directly from the thalamus. The functional role of these inputs was discussed in relation to the known cortical reflexes.
INTRODUCTION In the preceding paper 24 it was shown that neurons in the border region between nucleus ventralis lateralis (VL) and ventralis posterolateralis (VPL) of the thalamus were labeled by retrograde transport of horseradish peroxidase injected into the motor cortex. Neurons in this region receive projections from spinothalamic and cervicothalamic tracts as well as from nucleus Z10,17,2a. The feline spinocervical tract is known to be distinct from the dorsal column-lemniscal tract it and carries afferent impulses from hair and skin as well as deep receptors which are similar to those carried by the
218 dorsal column12,13. Since the results reported in the first paper of this series suggested that the motor cortex receives somesthetic inputs directly from the thalamus, neurons in this border region were suspected of transferring the peripheral inputs directly to the motor cortex. We have examined the above possibility using electrophysiologieal techniques, i.e. first, by antidromically activating these projection neurons from the motor cortex and then by examining their peripheral receptive fields. These neurons will be shown to transfer finely grained and topographically organized peripheral inputs to neurons in the motor cortex. METHODS Altogether 17 cats of either sex weighing between 2.5-3.5 kg were used. Operations were carried out under inhalation anesthesia composed of a mixture of oxygen (50 ~ ) and nitrous oxide (50 ~ ) supplemented with 1.5 ~ halothane. The cat was mounted on a stereotaxic instrument and the skull was opened over the motor and the occipital cortices. In 13 of the cats, two closed chambers were attached over the openings; one was directed to the motor cortex, and the other toward the ventrobasal thalamus using stereotaxic coordinates. Further details of this procedure have been described elsewhere (see ref. 5). In the remaining 4 cats, one chamber was attached over the motor cortex. After the installation of the chambers, a few screws were drilled into the skull and covered with dental cement. Additional screws were embedded in this cement mass and they were used to fix the head. A cannula was inserted into the cisterna magna to prevent swelling of the brain during the experiments. All the wound areas were infiltrated with a long lasting local anesthetic (Zyljectin, Abbott). At the end of the operation, a tranquilizing dose of sodium pentobarbital (Nembutal, 10 mg/kg) was injected intramuscularly, the ear bars and the snout clamp were removed and inhalation anesthesia was discontinued. A photograph of the exposed motor cortex was taken to map the location of later electrode insertion. The experiments were started 1-2 h after the operation. By giving milk before and during the experiments, the animals were usually quiet and cooperative.
Stimulation and recordings Glass insulated tungsten electrodes of the type described by Stoney et al. 81 were used for stimulation as well as recording unitary spikes. An array of 6 electrodes, 2.0 mm apart and aligned in a U-shape, were inserted into the pericruciate cortex surrounding the cruciate sulcus through metal tubes attached to the lid of one of the chambers (Fig. 1). Each electrode was marked with black ink at a distance of 1.5 mm from the tip and manually inserted into the brain to an approximate depth of 1.0 mm or less under the dissection microscope. The electrode was then fixed to the guide tube with dental impression wax and left at the same position throughout the experiment. The sites of cortical insertion were classified as fore- or hindlimb area by the effect of ICMS delivered through each electrode. The ink marks, which were placed 0.5 mm above the cortex, served as indicators of the cortical level. When the brain shrank, the electrodes were re-inserted but when it swelled, the experiments were usually abandoned. These 6
219
"~
lr.
T
Fig. 1. Experimental set-ups. Pulses produced by the window discriminator trigger the oscilloscope. All the signals are displayed through analog delay, therefore, the unitary spikes which produced the trigger pulses can be displayed on the oscilloscope.
electrodes were used for intracortical microstimulation (ICMS) and in some experiments, were also used for recording unit spikes. Recording electrodes were inserted into the ventrobasal area of the thalamus through the second closed chamber attached to the occipital bone (Fig. 1). The insertions were made in sagittal planes with 34° to the frontal plane and were directed toward the border area between VPL and VL as suggested by the preceding experiments 24. The spikes were displayed on the oscilloscope through an analog delay (AD-5, Bak) which enabled visuafization of the same spike on the scope after being used as a trigger pulse (Fig. l). Whenever unit spikes were evoked antidromically by ICMS, the receptive fields of the cells were examined by listening to the tone of pulses generated by the window discriminator, the accuracy of which was confirmed by the display through the analog delay. Lesions were made at all spots where antidromic spikes were recorded by passing negative current of 10 #A for 10 sec.
Histological examinations At the end of the experiments, lesions were made through those cortical electrodes which elicited antidromic spikes in the thalamus. Probe insertions were made into the thalamus near the recording sites. The animals were deeply anesthetized by Nembutal and the brain was perfused with saline followed by 10 % formalin and removed. The motor cortex and the thalamus were frozen and sectioned sagittally in 50/~m thickness. The cortical sections were stained with Kliiver and Barrera's 22 method and the thalamic sections were stained alternately with Kltiver and Barrera's method or with cresyl violet.
220 RESULTS
Sites of cortical stimulation Since the primary purpose of the experiments was to determine whether the thalamic neurons projecting to the motor cortex carry somesthetic information arising from the periphery, it was important to ascertain that ICMS activated only those fibers terminating within the motor cortex. It has been shown that the effective current spread (x, mm) for the thalamic projection fibers in the motor cortex is proportional to the square root of the current (y, #a), i.e., x = V'y/k and the constant (k) is approximately 10004. Since we used a maximum current of 30 #A to prevent etching of the electrode and to protect the tissue from damage z, the maximum effective spread was expected to be 0.17 mm. Since the locations of all of the effective cortical electrodes were confirmed by histological examination (see Methods), and they were never inserted deeper than 1.2 mm, there was almost no possibility that ICMS activated fibers in the white substance. All the sites of effective ICMS were located in the area where there were large pyramidal cells and there was no layer IV (area 47: Hassler and Muhs-Clement19). Identification of thalamocortical projection neurons The recording electrode was inserted into the thalamus while ICMS of 30/tA was delivered through each of the 6 electrodes. It is already known that the projection neurons from VL to the motor cortex do not have discrete somesthetic receptive fields in the periphery 4. Therefore, special attention was focused on those neurons located at the borderline between VL and VPL, i.e. neurons located at the rostral border of the VPL. When the electrode picked up spikes from this region which responded to the ICMS, the cortical electrode responsible for evoking the spikes was determined and the threshold current was measured. The antidromicity of the activation was examined following the criteria suggested by Abzug and Petersonl: (1) latency fluctuation at threshold stimulation is small (less than 0.1 msec); and (2) latency reduction by increasing the stimulus intensity to twice the threshold is small (less than 0.1 msec). In addition, we examined (3) the refractory period (1.3 msec or less) following the previous criteria by Asanuma et al. 4 and whenever possible, (4) collision of antidromic impulses with spontaneous discharges, an example of which is shown in Fig. 5.
% ...I ...I
4
W
u 2¸
0.7
0 9
1I
1.3
1,5
17
2,6
2.8
LAT E NC Y (msec) Fig. 2. Latency distribution of thalamic neurons antidromically activated from the motor cortex. Shaded bars indicate neurons which were not activated by peripheral inputs.
221 A total of 20 neurons satisfied all 4 criteria and an additional 20 neurons satisfied all but criterion no. 4. The latter neurons were not tested for collision because of the lack of spontaneous discharges or difficulty in isolating the cell from simultaneously recorded cells. However, since all the cells which satisfied the other 3 criteria also satisfied the fourth criterion when the collision test was possible, we are confident that all 40 cells were antidromically activated. Fig. 2 illustrates the latency distribution for antidromic activation of these 40 neurons. The majority of the cells were rather evenly distributed between 0.7 and 1.7 msec. Although all of these 40 cells were recorded at the V L - V P L border area, 8 could not be activated by any natural stimulation applied to the body as will be described later. The distribution of latencies of neurons located at VL-VPL border area was wider than that of VL neurons projecting to the motor cortex and sampled by similar criteria 4. The mean latency of VL-VPL neurons carrying peripheral information was 1.27 msec (S.D. 4-0.5) whereas the mean latency of VL neurons projecting to the motor cortex was 0.84 msec (S.D. 4-0.27). This difference may reflect a difference in the size3z of cells in the V L - V P L border areas and VL. Distribution o f cells projecting to the motor cortex The first penetration was usually directed toward the area of L: 5, A: 112o and recordings were started at the level of H: + 5 and continued until the electrode reached the internal capsule, indicated by the appearance of positive spikes. Subsequent penetrations were made at 0.5 mm steps to form a grid around the target area. Fig. 3A shows a histological reconstruction of electrode tracks of a typical experiment. In this example, 4 penetrations were made stereotaxically on the same sagittal plane and 3 of them appeared on the same histological slide. The other penetration (no. 4) was found on the slide 50/~m medial to this section. Altogether 55 cells could be isolated during these penetrations and 3 of them were activated antidromically from the motor cortex. Lesions were made at these 3 recording sites and at the bottom of track no. 3. The projection cell on track no. 4 could not be driven from the periphery, but the cells on tracks 2 and 3 responded to passive movements of joints. These lesions on tracks nos. 2 and 3 may appear to be located inside of the VPL, but it should be noted that the medial border of VPL in this area curves down steeply, hence, they were actually located at the border near VL. Fig. 3C illustrates the receptive fields of neurons encountered during the penetrations. As has already been reported 29, this region of the thalamus is highly topographically organized. The hindlimb was represented in the rostrolateral portion of the nucleus and the representation shifted gradually towards the forelimb when the electrode moved to the mediocaudal region. All the cells encountered in the VPL could be activated by natural stimulation somewhere in the body, and cells located close together were activated from the same part of the body by stimulation of the same modality, several examples of which are shown in Fig. 3. This feature of the similarity of the inputs to adjacent cells allowed estimation of receptive field of a cell even when the spikes of the target cell could not be isolated from the neighboring cells. Table I summarizes the type of adequate stimuli for all the projection neurons
C
A
]
B
1
2
2
3
4
34
,lmm
f
Fig. 3. Histological reconstruction of electrode tracks and distribution of cells along the penetration. A: specimen sagittal section of ventral thalamus at the level of lat. 5.0 mm showing electrode tracks and lesions. The left-most track and hemorrhage at the bottom was due to probe insertion. B: reconstruction of tracks from the histological preparations. Short bars on the tracks are location of neurons with no receptive fields. Long right side bars on the tracks are location of neurons activated from skin receptors. Long left side bars are those activated from deep receptors. Circles indicate locations of lesions made during experiment. Arrows indicate neurons activated by ICMS. C: receptive fields of neurons encountered during the penetrations. Blackened areas indicate receptive fields for hair bending or light touch. Circles indicate deep receptive fields such as pressure or passive joint movement. Numbers on the column correspond to track numbers and each inset corresponds to respective long bar on the track. Further details in the text. TABLE I
Adequate stimuli Parenthesis gives no. of cells judged from receptive fields of neighboring cells.
Site of stimulus
Forelimb area Hindlimb area
No. of cells
Forelimb
Hindlimb
Skin
Deep
Skin
Deep
31 9
9 0
16 (4) 1
0 3
1 2
Undriven
5 3
223
L 4.9 ~
V
'-,~ • w
"11
,," ° ,,'-LI , / 1 { ,,~ oOoX 5 " / / ,, _._,2//
L 5.2 ',, v~ \\ "",,/\ . v~ "lJ >.. N
L5.6 ~ VA ~\ II*"~ ' l /
VB ~
/
H 3 2
.-I
(...
