Brain Research, 161 (1979) 411-429 © Elsevier/North-HollandBiomedicalPress
411
INTERACTION OF NEURAL SYSTEMS WHICH CONTROL NUTRITIONAL BALANCE
RAIMOND EMMERS Department of Physiology, College of Physicians and Surgeons, Columbia University, New York, N. Y. 10032 (U.S.A.)
(Accepted June 8th, 1978)
SUMMARY The functional relationships of 5 nuclear masses of the cat brain were analyzed under conditions of general anesthesia by a combination of neurophysiological methods. The nuclear masses are known as (1) the area lateralis hypothalami (ALH), (2) the nucleus entopeduncularis (nEp), (3) the nucleus semilunaris accessorius (nSA), (4) the nucleus ventromedialis hypothalami (nVmH), and (5) the posterolateral portion of the zona incerta (ZI). Neurons of the ALH, nEp, and nVmH were tested for their responsiveness to electrical stimulation of the thalamic taste nucleus (nSA) and to intracarotid infusions of glucose and insulin. Following this, lesions were placed in some of the nuclei, and the responsiveness of the same neurons to glucose and insulin was re-evaluated. Additional experiments were performed to determine the influence of certain peripheral receptors on the 5 nuclear masses. Results indicated the following. (1) ALH-Ep neurons which are influenced by the nSA stimulation receive information about changes in plasma glucose from two sources: taste receptors and receptors located in the splanchnic area. The former are excited during the application of glucose to the tongue or with an increase in the plasma glucose concentration (intravascular taste); the latter, with a lowering of the plasma glucose level. (2) Destruction of the nSA leaves the ALH-Ep neurons unresponsive to glucose and insulin. (3) A negative feedback loop interrelates certain neurons of the ALH-Ep with the nVmH. (4) Interactions of the neurons of the 5 nuclear masses can account for the major events of the feeding cycle and for the specific hungers for some substances. Moreover, the interactions can resolve several seemingly discordant experimental findings which have presented obstacles in previous attempts to link the control of nutrition with the availability of glucose to tissues.
412 INTRODUCTION The series of experiments described in this report was begun with an attempt to answer a relatively simple question: does the gustatory system influence hypothalamic neurons involved in the control of food intake? When such an influence was observed 14, additional experiments were designed to study some details of this influence. As work progressed, it became clear that besides the well known ventromedial hypothalamus, taste buds and certain receptors in the splanchnic region monitor the availability of glucose to tissues is. The importance of taste in feeding appeared to stem from the ability of taste buds to detect substances not only on the tongue but also in the vasculature15,19. The latter is known as 'intravascular taste '9. Glucoreceptors in the splanchnic region are excited when the availability of glucose to them is decreased, and the thalamic taste nucleus relays the excitations from the taste buds and the receptors of the splanchnic region to the lateral hypothalamus is. In other laboratories, work on the hypothalamic feeding and satiety systems indicated that the control of feeding was not as simple as initially envisioned ~1. In several experiments interference with the activity of the ventromedial satiety neurons did not lead to the expected changes in feeding behavior 3°. Unaware of the ability of peripheral glucose detectors to compensate for the experimental manipulations, some authors were quick to label the involvement of the ventromedial hypothalamus 'a myth 'z2. Others 25 cautiously expressed doubts that these neurons represented a singular mechanism for satiety. Moreover, the ability of the ventromedial satiety neurons to inhibit feeding neurons of the lateral hypothalamus was questioned 3°. No clear anatomical evidence existed for a pathway between the feeding and the satiety neurons. Although electrical stimulation of the ventromedial neurons could inhibit some neurons of the lateral hypothalamus, others in the same area were excited 35. Stimulation of the lateral hypothalamus could either excite or inhibit neurons in the ventromedial area 29, but the excitatory influence was disregarded in favor of the slightly more numerous instances of inhibition; in subsequent publications the two hypothalamic loci were thought to inhibit each other ~0. Furthermore, most data dealing with the hypothalamic control of feeding tacitly assumed that hypothalamic neurons which were sensitive to glucose were the ones controlling food intake. Whether such an assumption is tenable has been recently questioned when the electrophysiologically tested glucoreceptors of the lateral hypothalamus could not be linked to feeding behavior 21. To be certain that neurons of the present series of experiments could be influenced during feeding, attention was directed toward those hypothalamic neurons that responded to stimulation of the gustatory system. On occasion, however, the activity of neurons that could not be influenced by gustatory stimulation was also examined. After the gustatory test, the activity of each neuron was recorded under a variety of conditions and experimental manipulations (Tables ! and II). This work led to the reconstruction of a complex system of neuronal interaction (Fig. 5). It provided a neurophysiological basis for the main sequence of events of spontaneous feeding cycles and assisted in explaining several puzzling phenomena: progressive satiety
413 during feeding, distribution of meal frequency, specific hunger for sweets, spontaneous recovery of feeding after a period of aphagia, and still others. This system is most likely modulated by structures of the limbic system and some of the catecholamine fiber networks 24 shown to be involved in the regulation of energy balance 1,2. Up to this time, portions of Fig. 5 have been described only in abstracts 6,~6-1a. These can now be unified, merged with additional research findings, and related to the existing body of knowledge. METHODS
Cats, the experimental animals, were anesthetized with an intraperitoneal injection of 35 mg/kg of Nembutal and prepared for the insertion of stimulating and recording electrodes in several subcortical structures: the thalamic taste nucleus, known anatomically as the nucleus semilunaris accessorius (nSA) 13, the area lateralis hypothalami (ALH), the nucleus entopeduncularis (nEp), and the nucleus ventromedialis hypothalami (nVmH). Prior to the placement of the head of the animal in the Stoelting Quadruple Stereotaxic Instrument,the common carotid artery on the side of the recording was cannulated for the infusion of test substances, and the left femoral vein was cannulated for the removal of blood samples. Stereotactic localization of the nSA was then performed. Since this nucleus is located immediately ventral to two layers of cells which separately relay pressure and cooling from the cat's tongue 13, electrophysiological correction for the stereotactic placement of a stimulating electrode in this nucleus was simple. Tonic activity of the pressure relay neurons on the oscilloscope was a sufficient indication that the electrode must be advanced 0.8 mm to reach the nSA. A concentric bipolar steel electrode with a total diameter of 1 mm and an inner core of 0.1 mm was substituted for the recording electrode. This electrode was used not only to deliver single square wave pulses to the nSA but also on occasion to destroy it by passing a 0.6 mA direct current between the tip of the core and the rim of the barrel of the electrode for 25 sec. After the stimulating electrode was inserted into the nSA of the left cerebral hemisphere, a tungsten microelectrode with a 1-3 /zm tip diameter was used to approach single neurons of the ALH-Ep also on the left side of the brain. When a neuron was electrophysiologically isolated, its spontaneous firing was allowed to stabilize over a period of 15 min. A count of its firing was then accumulated in a computer (CAT 400 C of Technical Measurements, Inc.) programmed for digital accumulation of events in 2.5 msec wide bins. This count was regarded as the initial control count. Depending on the experimental design, the counts were accumulated over 250-1000 sec periods. For the sake of uniformity and convenience of comparison, however, all firing was expressed as counts/1000 sec. Stability of the firing frequency was checked with additional sample counts. If for some reason the activity of the neuron was continuously drifting in one direction, the neuron was abandoned. Neurons with stabilized firing frequency did not vary more than -4- 400 counts/1000 see.
Test substances of a constant volume (2 ml) were infused into the common
414 carotid artery at a rate of I ml/min, and accumulation of the experimental counts was begun 1 min after the completion of any infusion. In addition to glucose (5~i) and regular insulin (Iletin, 80 USP units/ml) physiological saline was infused in some animals. But since its effect on the activity of hypothalamic neurons was nil, its use was discontinued to shorten the experimental procedure. A plastic tray, drained continuously, supported the cat's tongue. Either the 5 7ooglucose solution or distilled water at 37 °C was run gently over the surface of the tongue while the computer accumulated the experimental firing count. After each experimental manipulation, sufficient time was allowed for the count to return to its initial control value, or until a new stable control value was re-established. Consequently, all changes in counts could be expressed as an increase or a decrease in counts/1000 sec respective to the appropriate control count. The counts were typed out on a model 535 teletype punch-read unit of Technical Instruments for construction of spike density histograms, or they were totalled on-line by using a model 5214 L Hewlett-Packard Preset Electronic Counter. In order to have matched control and experimental counts, several tests had to be performed on the same neuron. For example, to obtain the appropriate values for comparing the response of a neuron to glucose before and after the destruction of both the nSA and the lateral portion of the n V m H the following procedure was carried out. (1) A stimulating electrode was placed in the nSA. (2)Another stimulating electrode was placed at the lateral border of the nVmH. (3) A neuron was electrophysiologically isolated in the A L H or the nEp, and its firing was accumulated several times to obtain a stable control count. (4) The site at the n V m H was stimulated to test for its influence on the A L H - E p neuron. If the stimulation produced no effect, another A L H - E p neuron was localized, or the stimulating electrode was repositioned to obtain a change in the activity of the neuron. (5) The stimulated site at the n V m H was destroyed electrolytically. (6) The nSA was stimulated to register its influence on the A L H - E p neuron. (7) A new control count of the activity of the A L H - E p neuron was established, and its response to glucose was tested by an IC infusion or by rinsing of the tongue. (8) The nSA was electrolytically destroyed. (9) A new control count was established, and the A L H - E p neuron was retested for its responsiveness to an IC infusion of glucose. (10) The recording site was marked with a microlesion. Although an experimental design of this type permitted study of only one A L H Ep neuron per animal, data obtained in this manner reached statistical reliability with relatively few neurons in a group (Table II). In addition, the electrophysiological monitoring of the placement of the stimulating and lesion making electrodes obviated errors inherent in the stereotactic method. To hold a neuron at the tip of the recording electrode throughout the period of the various manipulations, however, was not always technically possible. This in part is reflected in the time lapse between the undertaking of the project 14 and the completion of this report. Additional details about the technical arrangements of the experiments can be obtained by consulting previous publications 15,19. Blood samples collected immediately after the accumulation of certain counts were analyzed for glucose concentration by the method of Washka and Rice 36. The brain of the animal was perfused with a 10 Yo commercial formaldehyde solution via
415 the aorta, and the stimulation, recording, and lesion sites were reconstructed anatomically from 40/~m sections cut according to the freezing technique and stained with cresylecht violet 12. RESULTS
Effects of nSA stimulation All neurons that responded to i.c. infusion of glucose or insulin (55 in ALH and 127 in nEp) were influenced also by electrical stimulation of the nSA. As the stimulus intensity was increased from a subthreshold value, the probability that an ALH-Ep neuron would fire following the application of a single stimulus (0.8 V, 0.5 msec) decreased to zero. This inhibition could be detected not only on the poststimulus spike density histograms (Figs. 1C, 4A,B), but also on the oscilloscope. The inhibitory period was followed by excitatory-inhibitory (E-I) phasing of the spontaneous activity of the neuron. The E peaks occurred at intervals ranging from 70 to 150 msec and became indistinct within 1 sec. With 13 of the ALH-Ep neurons a narrow E peak reaching to a probability of 200/1000 samples followed the nSA stimulus at a latency of 6-8 msec. This stimulus-evoked activity was critically dependent on the position of the stimulating electrode; changing its position for a distance of 0.2 mm within the nSA by delicate rotation of the microdrive abolished only the short-latency firing. To x
A
0 0 0 [
0 O~ n
CO a
1
sec
n,. Q Z
i'" T T
"I-
ill
Z
i/t/
t,9
~Z
1.
]"
/ I io, I second
C
mv
LLI
lad ._1 I--0 t--
1
,
i
sec
1
Fig. 1. A: computer read-out of the spontaneous firing of an ALH-Ep neuron accumulated during 1000 sec on a 1 sec time base. B: activity of A in spike form. C: excitatory-inhibitory phasing of the read-out of A by stimulation of the nSA time locked at arrow, 1000 ×.
416 TABLE I Changes in the firing rates o f 2l A L H and52 nEp neurons in response to glucose or insulin (each entry contains a combination o f A L H and nEp neurons) Control average counts/lO00 sec
8728 8021 2811 * 13,188" 5395* ! 1,460" 7826
Experimental manipulation
Number o f neurotts out o f 73
2 ml, 5 ~ glucose, i.c.*** 40 160 USP units insulin, i.c.*** 10 160 USP units insulin, i.c., after 2 ml, 5 ~ glucose, i.c. 16 2 ml, 5 ~ glucose, i.c., after 160 USP units insulin, i.c. 6 2 ml, 5 ~o glucose, i.c., after same 8 160 USP units insulin, i.c. after same 4 Rinsing the tongue with 5 ~ glucose 8
Average int-test; crease or deP values* * crease in counts/ 1000 sec
--4487 -~ 4607
< 0.01 < 0.01
+4305
411
INTERACTION OF NEURAL SYSTEMS WHICH CONTROL NUTRITIONAL BALANCE
RAIMOND EMMERS Department of Physiology, College of Physicians and Surgeons, Columbia University, New York, N. Y. 10032 (U.S.A.)
(Accepted June 8th, 1978)
SUMMARY The functional relationships of 5 nuclear masses of the cat brain were analyzed under conditions of general anesthesia by a combination of neurophysiological methods. The nuclear masses are known as (1) the area lateralis hypothalami (ALH), (2) the nucleus entopeduncularis (nEp), (3) the nucleus semilunaris accessorius (nSA), (4) the nucleus ventromedialis hypothalami (nVmH), and (5) the posterolateral portion of the zona incerta (ZI). Neurons of the ALH, nEp, and nVmH were tested for their responsiveness to electrical stimulation of the thalamic taste nucleus (nSA) and to intracarotid infusions of glucose and insulin. Following this, lesions were placed in some of the nuclei, and the responsiveness of the same neurons to glucose and insulin was re-evaluated. Additional experiments were performed to determine the influence of certain peripheral receptors on the 5 nuclear masses. Results indicated the following. (1) ALH-Ep neurons which are influenced by the nSA stimulation receive information about changes in plasma glucose from two sources: taste receptors and receptors located in the splanchnic area. The former are excited during the application of glucose to the tongue or with an increase in the plasma glucose concentration (intravascular taste); the latter, with a lowering of the plasma glucose level. (2) Destruction of the nSA leaves the ALH-Ep neurons unresponsive to glucose and insulin. (3) A negative feedback loop interrelates certain neurons of the ALH-Ep with the nVmH. (4) Interactions of the neurons of the 5 nuclear masses can account for the major events of the feeding cycle and for the specific hungers for some substances. Moreover, the interactions can resolve several seemingly discordant experimental findings which have presented obstacles in previous attempts to link the control of nutrition with the availability of glucose to tissues.
412 INTRODUCTION The series of experiments described in this report was begun with an attempt to answer a relatively simple question: does the gustatory system influence hypothalamic neurons involved in the control of food intake? When such an influence was observed 14, additional experiments were designed to study some details of this influence. As work progressed, it became clear that besides the well known ventromedial hypothalamus, taste buds and certain receptors in the splanchnic region monitor the availability of glucose to tissues is. The importance of taste in feeding appeared to stem from the ability of taste buds to detect substances not only on the tongue but also in the vasculature15,19. The latter is known as 'intravascular taste '9. Glucoreceptors in the splanchnic region are excited when the availability of glucose to them is decreased, and the thalamic taste nucleus relays the excitations from the taste buds and the receptors of the splanchnic region to the lateral hypothalamus is. In other laboratories, work on the hypothalamic feeding and satiety systems indicated that the control of feeding was not as simple as initially envisioned ~1. In several experiments interference with the activity of the ventromedial satiety neurons did not lead to the expected changes in feeding behavior 3°. Unaware of the ability of peripheral glucose detectors to compensate for the experimental manipulations, some authors were quick to label the involvement of the ventromedial hypothalamus 'a myth 'z2. Others 25 cautiously expressed doubts that these neurons represented a singular mechanism for satiety. Moreover, the ability of the ventromedial satiety neurons to inhibit feeding neurons of the lateral hypothalamus was questioned 3°. No clear anatomical evidence existed for a pathway between the feeding and the satiety neurons. Although electrical stimulation of the ventromedial neurons could inhibit some neurons of the lateral hypothalamus, others in the same area were excited 35. Stimulation of the lateral hypothalamus could either excite or inhibit neurons in the ventromedial area 29, but the excitatory influence was disregarded in favor of the slightly more numerous instances of inhibition; in subsequent publications the two hypothalamic loci were thought to inhibit each other ~0. Furthermore, most data dealing with the hypothalamic control of feeding tacitly assumed that hypothalamic neurons which were sensitive to glucose were the ones controlling food intake. Whether such an assumption is tenable has been recently questioned when the electrophysiologically tested glucoreceptors of the lateral hypothalamus could not be linked to feeding behavior 21. To be certain that neurons of the present series of experiments could be influenced during feeding, attention was directed toward those hypothalamic neurons that responded to stimulation of the gustatory system. On occasion, however, the activity of neurons that could not be influenced by gustatory stimulation was also examined. After the gustatory test, the activity of each neuron was recorded under a variety of conditions and experimental manipulations (Tables ! and II). This work led to the reconstruction of a complex system of neuronal interaction (Fig. 5). It provided a neurophysiological basis for the main sequence of events of spontaneous feeding cycles and assisted in explaining several puzzling phenomena: progressive satiety
413 during feeding, distribution of meal frequency, specific hunger for sweets, spontaneous recovery of feeding after a period of aphagia, and still others. This system is most likely modulated by structures of the limbic system and some of the catecholamine fiber networks 24 shown to be involved in the regulation of energy balance 1,2. Up to this time, portions of Fig. 5 have been described only in abstracts 6,~6-1a. These can now be unified, merged with additional research findings, and related to the existing body of knowledge. METHODS
Cats, the experimental animals, were anesthetized with an intraperitoneal injection of 35 mg/kg of Nembutal and prepared for the insertion of stimulating and recording electrodes in several subcortical structures: the thalamic taste nucleus, known anatomically as the nucleus semilunaris accessorius (nSA) 13, the area lateralis hypothalami (ALH), the nucleus entopeduncularis (nEp), and the nucleus ventromedialis hypothalami (nVmH). Prior to the placement of the head of the animal in the Stoelting Quadruple Stereotaxic Instrument,the common carotid artery on the side of the recording was cannulated for the infusion of test substances, and the left femoral vein was cannulated for the removal of blood samples. Stereotactic localization of the nSA was then performed. Since this nucleus is located immediately ventral to two layers of cells which separately relay pressure and cooling from the cat's tongue 13, electrophysiological correction for the stereotactic placement of a stimulating electrode in this nucleus was simple. Tonic activity of the pressure relay neurons on the oscilloscope was a sufficient indication that the electrode must be advanced 0.8 mm to reach the nSA. A concentric bipolar steel electrode with a total diameter of 1 mm and an inner core of 0.1 mm was substituted for the recording electrode. This electrode was used not only to deliver single square wave pulses to the nSA but also on occasion to destroy it by passing a 0.6 mA direct current between the tip of the core and the rim of the barrel of the electrode for 25 sec. After the stimulating electrode was inserted into the nSA of the left cerebral hemisphere, a tungsten microelectrode with a 1-3 /zm tip diameter was used to approach single neurons of the ALH-Ep also on the left side of the brain. When a neuron was electrophysiologically isolated, its spontaneous firing was allowed to stabilize over a period of 15 min. A count of its firing was then accumulated in a computer (CAT 400 C of Technical Measurements, Inc.) programmed for digital accumulation of events in 2.5 msec wide bins. This count was regarded as the initial control count. Depending on the experimental design, the counts were accumulated over 250-1000 sec periods. For the sake of uniformity and convenience of comparison, however, all firing was expressed as counts/1000 sec. Stability of the firing frequency was checked with additional sample counts. If for some reason the activity of the neuron was continuously drifting in one direction, the neuron was abandoned. Neurons with stabilized firing frequency did not vary more than -4- 400 counts/1000 see.
Test substances of a constant volume (2 ml) were infused into the common
414 carotid artery at a rate of I ml/min, and accumulation of the experimental counts was begun 1 min after the completion of any infusion. In addition to glucose (5~i) and regular insulin (Iletin, 80 USP units/ml) physiological saline was infused in some animals. But since its effect on the activity of hypothalamic neurons was nil, its use was discontinued to shorten the experimental procedure. A plastic tray, drained continuously, supported the cat's tongue. Either the 5 7ooglucose solution or distilled water at 37 °C was run gently over the surface of the tongue while the computer accumulated the experimental firing count. After each experimental manipulation, sufficient time was allowed for the count to return to its initial control value, or until a new stable control value was re-established. Consequently, all changes in counts could be expressed as an increase or a decrease in counts/1000 sec respective to the appropriate control count. The counts were typed out on a model 535 teletype punch-read unit of Technical Instruments for construction of spike density histograms, or they were totalled on-line by using a model 5214 L Hewlett-Packard Preset Electronic Counter. In order to have matched control and experimental counts, several tests had to be performed on the same neuron. For example, to obtain the appropriate values for comparing the response of a neuron to glucose before and after the destruction of both the nSA and the lateral portion of the n V m H the following procedure was carried out. (1) A stimulating electrode was placed in the nSA. (2)Another stimulating electrode was placed at the lateral border of the nVmH. (3) A neuron was electrophysiologically isolated in the A L H or the nEp, and its firing was accumulated several times to obtain a stable control count. (4) The site at the n V m H was stimulated to test for its influence on the A L H - E p neuron. If the stimulation produced no effect, another A L H - E p neuron was localized, or the stimulating electrode was repositioned to obtain a change in the activity of the neuron. (5) The stimulated site at the n V m H was destroyed electrolytically. (6) The nSA was stimulated to register its influence on the A L H - E p neuron. (7) A new control count of the activity of the A L H - E p neuron was established, and its response to glucose was tested by an IC infusion or by rinsing of the tongue. (8) The nSA was electrolytically destroyed. (9) A new control count was established, and the A L H - E p neuron was retested for its responsiveness to an IC infusion of glucose. (10) The recording site was marked with a microlesion. Although an experimental design of this type permitted study of only one A L H Ep neuron per animal, data obtained in this manner reached statistical reliability with relatively few neurons in a group (Table II). In addition, the electrophysiological monitoring of the placement of the stimulating and lesion making electrodes obviated errors inherent in the stereotactic method. To hold a neuron at the tip of the recording electrode throughout the period of the various manipulations, however, was not always technically possible. This in part is reflected in the time lapse between the undertaking of the project 14 and the completion of this report. Additional details about the technical arrangements of the experiments can be obtained by consulting previous publications 15,19. Blood samples collected immediately after the accumulation of certain counts were analyzed for glucose concentration by the method of Washka and Rice 36. The brain of the animal was perfused with a 10 Yo commercial formaldehyde solution via
415 the aorta, and the stimulation, recording, and lesion sites were reconstructed anatomically from 40/~m sections cut according to the freezing technique and stained with cresylecht violet 12. RESULTS
Effects of nSA stimulation All neurons that responded to i.c. infusion of glucose or insulin (55 in ALH and 127 in nEp) were influenced also by electrical stimulation of the nSA. As the stimulus intensity was increased from a subthreshold value, the probability that an ALH-Ep neuron would fire following the application of a single stimulus (0.8 V, 0.5 msec) decreased to zero. This inhibition could be detected not only on the poststimulus spike density histograms (Figs. 1C, 4A,B), but also on the oscilloscope. The inhibitory period was followed by excitatory-inhibitory (E-I) phasing of the spontaneous activity of the neuron. The E peaks occurred at intervals ranging from 70 to 150 msec and became indistinct within 1 sec. With 13 of the ALH-Ep neurons a narrow E peak reaching to a probability of 200/1000 samples followed the nSA stimulus at a latency of 6-8 msec. This stimulus-evoked activity was critically dependent on the position of the stimulating electrode; changing its position for a distance of 0.2 mm within the nSA by delicate rotation of the microdrive abolished only the short-latency firing. To x
A
0 0 0 [
0 O~ n
CO a
1
sec
n,. Q Z
i'" T T
"I-
ill
Z
i/t/
t,9
~Z
1.
]"
/ I io, I second
C
mv
LLI
lad ._1 I--0 t--
1
,
i
sec
1
Fig. 1. A: computer read-out of the spontaneous firing of an ALH-Ep neuron accumulated during 1000 sec on a 1 sec time base. B: activity of A in spike form. C: excitatory-inhibitory phasing of the read-out of A by stimulation of the nSA time locked at arrow, 1000 ×.
416 TABLE I Changes in the firing rates o f 2l A L H and52 nEp neurons in response to glucose or insulin (each entry contains a combination o f A L H and nEp neurons) Control average counts/lO00 sec
8728 8021 2811 * 13,188" 5395* ! 1,460" 7826
Experimental manipulation
Number o f neurotts out o f 73
2 ml, 5 ~ glucose, i.c.*** 40 160 USP units insulin, i.c.*** 10 160 USP units insulin, i.c., after 2 ml, 5 ~ glucose, i.c. 16 2 ml, 5 ~ glucose, i.c., after 160 USP units insulin, i.c. 6 2 ml, 5 ~o glucose, i.c., after same 8 160 USP units insulin, i.c. after same 4 Rinsing the tongue with 5 ~ glucose 8
Average int-test; crease or deP values* * crease in counts/ 1000 sec
--4487 -~ 4607
< 0.01 < 0.01
+4305
