Brain Research, 149 (1978) 347 364 © Elsevier/North-Holland Biomedical Press
347
COMPARISON OF FIR1NG PATTERNS AND SENSORY RESPONSIVENESS BETWEEN SUPRAOPTIC AND OTHER HYPOTHALAMIC NEURONS IN THE UNANESTHET1ZED SHEEP
D. P. J E N N I N G S ,
J. T. HASKINS and J. M. ROGERS
Department 0[" Physiological Sciences, College of Veterinary Medicine, Oklahoma State University, Stillwater, Okla. 74074, The Marine Biomedical lnstitate, 200 University Boulevard, Galveston, Texas 77550 and DepJrtment Of Anatomy, Colorado State University, Fort Collins, Colo. (U.S.A.) (Accepted October 27th, 1977)
SUMMARY
Adult Southdown ewes were surgically prepared with pituitary stimulating electrodes, carotid and jugular cannulae, and a cranial platform-cylinder arrangement for chronic single unit recording. Isolated neurons (n z 112) in the region of the supraoptic nucleus (SON) were identified by pituitary stalk stimulation as AD'~ (antidromically invaded) SON neuroendocrine cells (n = 75) or AD-- (not antidromically invaded) SON neurons (n -- 37). Spontaneous firing pattern distribution and sensory evoked behavior of these SON region neurons were compared with activity recorded from 112 randomly located non-identified neurons of extra-SON areas of the hypothalamus. Spontaneous discharge activity was categorized into six distinct firing pattern types: continuously active slow (CAS), continuously active fast (CAF), continuously active bursting (CAB), continuously active regular (CAR), low frequency bursting (LFB), and high frequency bursting (HFB). These 6 firing pattern types were characterized by computer analysis and their mean order independent statistical parameters compared. Bursting discharge patterns (LFB, HFB, and CAB) were compared with respect to mean burst duration, burst mean firing rate, and interburst intervals. Ninety-three per cent of all neurons maintained a stable discharge pattern in the absence of apparent stimuli. Occasionally CAS and CAF neurons spontaneously generated spike clusters sufficient to give the transient appearance of a bursting discharge pattern and LFB neurons lapsed spontaneously into CAS activity. All 6 firing pattern types recorded from non-identified extra-SON neurons were also recorded in the SON region. However, spontaneously discharging A D + SON neurons exhibited only continuously active slow (CAS), continuously active fast (CAF), and low frequency bursting (LFB) activity. The total absence of high frequency bursting (HFB), continuously active regular (CAR), and continuously active bursting (CAB)
348 patterns of discharge from A D ..... SON neurons suggests that A D - - SON neurons exhibiting these firing patterns may function as mterneurons, pacemaker neurons. or receptor neurons. A significant number of LFB discharging neurons were recorded in widespread extra-SON regions of the hypothalamus, indicating this discharge pattern may not be unique to magnocellular neuroendocrine cells. AD-~ SON LFB neurons sampled in this study demonstrated a significantly longer mean interburst interval (20.86 sec) compared to extra-SON LFB neurons (12.43 sec). No A D - - SON neuron tested was significantly sensitive to non-specific sensory arousal or sleep-waking state changes. In extra-SON areas of the hypothalamus. I l of 75 neurons tested to sensory arousal and 6 of 19 neurons tested to sleep-waking changes responded with significant changes in mean firing rate (MFR): no significant difference between firing pattern types was demonstrated in arousal or sleep-waking sensitivity. Significantly more A D .... SON neurons (66 90) responded to rapid intracarotid injections or slow intrajugular infusions of hypertonic NaC1 than extra-SON neurons (19~). It was concluded that extra-SON neurons could have been affected nonspecifically by the behavioral and E E G arousal effects of rapid intracarotid injections of hypertonic NaC1. Fifty-one per cent of the A D SON neurons were excited, 12 O/o inhibited, and 38 9/00unaffected by slow intrajugular infusions of hypertonic NaCI. No one spontaneous firing pattern was significantly more osmosensitive than another; however. 8 0 ~ of the CAS or C A F discharging neurons excited by this osmotic loading became low frequency bursters during, or immediately, following the infusion.
INTRODUCT|ON Experiments on unanesthetized monkeys TM, urethane anesthetized rats 29 and. more recently, unanesthetized sheep s, have demonstrated that antidromically identified cells in the supraoptic and paraventricular nucleus may be silent or discharged spontaneously in one of two basic patterns, continuously active (irregular), or low frequency bursting (phasic). It has been suggested that the functional significance of these different firing pattern types may be attributable to different thresholds for sensory input 2, and to different functional neuroendocrine cell types, each having 'specific' input connections and associated with one hormone, such as A D H or oxytocin7,10. The low frequency bursting (LFB) pattern in particular has stimulated the interest of neuroendocrine investigators; it may be unique to neuroendocrine cells and responsible for the release Of hormone 'spurts' into the blood vascular system. This implies functional mechanisms controlling the firing behavior of neuroendocrine cells which may be unique in the mammalian central nervous system (CNS) to the functions o f a neuroendocrine system. Recent evidence m urethane anesthetized lactating rats suggests that this pattern is associated primarily with vasopres~n neuro-
349 endocrine cells and that oxytocin cells discharge continuously irregular with the exception of occasional spike bursts associated with milk ejection 21,3°. In vivo and in vitro electrical stimulation studies 6,1s suggest that a given number of action potentials are more effective in causing the release of a hormone if they occur close together in time. Thus it appears that recruitment of continuously irregular firing neuroendocrine cells into a bursting pattern of discharge could be an essential mechanism for releasing appropriate amounts of hormone from the posterior pituitary. Electrophysiological evidence11,15,17 suggest the presence of recurrent inhibition which may involve interneurons similar to Renshaw cells of the spinal cord. Other electrophysiologic studies by Hayward and Vincent 13 in the unanesthetized monkey suggest that osmoreceptor neurons in the perinuclear zone of the SON exist separately from neuroendocrine cells, and that they respond monophasically to an intracarotid pulse injection of hypertonic NaC1. Thus, it appears likely that the supraoptic nucleus region, including its perinuclear zone, may contain a functionally heterogeneous population of receptor neurons, interneurons, and magnocellular neuroendocrine cells of at least two types (oxytocin and vasopressin secreting). The purpose of this paper is to describe the firing pattern distribution and sensory evoked behavior of antidromically identified neuroendocrine cells and other neurons of the supraoptic nucleus region in the unanesthetized sheep. In order to determine the functional uniqueness of this neuroendocrine associated pool of neurons, these characteristics were compared to similar information from a randomly selected population of unidentified neurons recorded from widespread extra-SON portions of the hypothalamus. In order to avoid possible influences from chemical anesthesia, behavioral levels of consciousness, or acute surgical complications, it was deemed essential to record from an unanesthetized animal, well conditioned to its recording environment and with sleep-waking behavior carefully monitored. Sheep were chosen as the experimental animal for numerous reasons. Because of their large size and placid behavior, they tolerate, with minimal restraint, the necessary cranial platform and microdrive assembly required for single unit recording in the unanesthetized animal, and are easily prepared for multiple blood sampling. Although single unit recording has not previously been reported in this species, considerable information is available concerning anteriorS, 19 and posterior1,14, 22 pituitary function as well as sleep-waking behavior 23 in goats and sheep. This makes them particularly attractive as a research animal for future neuroendocrine studies. MATERIALS AND METHODS Southdown ewes (4-9 years of age) previously conditioned to a restraining stanchion and recording chamber were surgically fitted with biparietal epidural EEG electrodes, periorbital eye movement electrodes, and a cranial platform-cylinder arrangement similar to that described by Hayward and Jennings 10 for the monkey. A lateral radiograph of the head was taken immediately following surgery, and frontal coordinates determined with reference to the implanted cylinder for location of the
350 pituitary fossa. Two or three days following this surgery, the ewe was again anesthetized for permanent implantation of bipolar electrodes across the pituitary stalk and placement of silicone rubber cannulae in each jugular vein. The carotid artery was cannulated via the thyrolaryngeal artery in 3 sheep utilizing a technique described by Buck ~. Correct placement of the pituitary stimulating electrodes was later confirmed when electrical stimulation through them evoked an antidiuresis in the experimentally hydrated animal. The jugular and carotid cannulae were filled with heparimzed saline (1000 units/ml) and led subcutaneously to the back of the sheep, where they were exteriorized and terminated into capped hypodermic needles within a canvas pouch fastened to the skin of the sheep. Following recovery from the second surgery, as indicated by eating at preoperative levels, the animal was placed in a restraining stanchion located within the recording chamber. The stanchion was designed with a head restraining platform and nose band which painlessly steadied the head. allowing the animal to doze intermittently during single unit recordings and to behave otherwise normally. Tungsten microelectrodes were hydraulically driven into the hypothalamus at coordinates which could be determined by positioning a Trent Wells, Inc. titanium micropositioner fastened to the cylinder. Extracellular single unit activity was amplified and channeled into a storage oscilloscope monitor (Tektronix 5103N/013), a magnetic tape recorder (Sony TC-366), an audiomonitor, and a pulse height discriminator. Two outputs from the pulse height discriminator (a one-to-one pulse output and an analog output proportional to the rate of discharge) were simultaneously recorded on an ink writing polygraph (Grass Model 7) with the cortical EEG and eye movement potential. A television camera placed inside the recording chamber permitted observation of the animal via a closed circuit television monitor. On the basis of EEG, eye movement. body movement, and observed behavior, judgement was made as to whether the animal was sleeping, drowsy, or alert awake. A written record provided by the polygraph allowed the investigators to correlate sleep-waking and sensory arousal behavior with the sheep's single neuron firing behavior. Recording periods to be analyzed were replayed from magnetic tape and the spike activity displayed on a storage oscilloscope in order to establish the stability of each unit's waveform. Single cell spike trains clearly separable from base line activity and neighboring units were led into the pulse height discriminator which emitted 0.5 msec pulses triggered by the unit's action potentials. These pulses were recorded onto a second magnetic tape which was used to generate rate meter and dot raster a displays for visual analysis of spike train firing patterns. The same tapes were played back onto a digital computer (lnterdata 167 programmed to calculate mean interspike interval, standard deviation of the intervals, coefficient of varmnce, and mean firing rate, as well as plot interspike interval histograms, accumulative rate plots, autocorrelograms and serial correlograms 2°. Burst analysis of those spike trains exhibiting definite short term periodicities was accomplished with the aid of a digital counter which was reset and triggered manually to count spikes during a burst. The sheep were provided water ad libitum between experiments and were in a state of normal hydration. The recording time for each neuron averaged 30 min,
351 but ranged from 5 rain to 3 h. Once the spontaneous firing pattern of a neuron was established, it was tested for sensitivity to non-specific sensory arousal, sleep-waking behavior (if possible), and intrajugular or intracarotid injections of hypertonic NaCI solutions. The location of neurons outside or within 1 mm of the histologic boundaries of the SON was determined by plotting their relative location with respect to Prussian blue spots 9laced by a steel microelectrode at several known levels along the recording tracts prior to terminating an animal. Physiologic confirmation of each neuron's location was accomplished by recording the distance it was recorded above the light sensitive optic chiasm and by observing the presence or absence of an evoked electric field resulting from single pulse stimulation of the pituitary stalk. RESULTS
Spike trains from a total of l l2 units located within the SON region (an area defined to be within 1 mm of the SON's histologic boundaries) were studied. Of these, 75 were invaded antidromically following stimulation of the pituitary stalk ( A D + SON), and assumed to be neuroendocrine cells; 37 were not invaded antidromically and referred to as A D - - SON neurons. The spontaneous firing patterns and sensitivity of SON neurons to sleep-waking behavior, non-specific sensory arousal, and intrajugular or intracarotid infusions of hypertonic NaCI were determined and compared to similar data from a population of 112 unidentified neurons recorded from randomly located areas of the hypothalamus outside the defined SON region. The spontaneous activity of all individual neurons was categorized into one of the 6 following categories: continuously active slow (CAS), continuously active fast (CAF), continuously active regular (CAR), continuously active bursting (CAB), high frequency bursting (HFB), and low frequency bursting (LFB). These firing pattern 'types' were further characterized by computer analysis of their component spike trains (interspike interval histogram (ISIH), accumulative rate, autocorrelogram, and TABLE 1
Comparison o f mean statistics between firing pattern types Abbreviations used are: CAR, continuously active regular, CAF, continuously active fast, HFB, high frequency bursting, CAB, continuously active bursting, LFB, low frequency bursting, and CAS, continuously active slow. Any two means underscored by the same solid line are not significantly different. Dotted lines act to connect solid lines. Any two means not underscored by the same solid line are significantly different (P < 0.05).
Comparison o f ~pontaneous firing patterns
CAR
CAF
HFB
CAB
LFB
CAS
Mean firing rates
5.02
4.84
3.24
7.13
1.57
0.42
Mean interspike intervals
0.25
0.34
0.39
0.15
0.86
7.05
Mean interspike interval modes
0.27
0.10
0.02
0.05
0.15
0.44
Mean coefficients of variation
3.19
0.92
0.36
0.42
0.46
0.80
352 TABLE II Comparison o f mean burst parameters between hring pattern type,s
Abbreviations used are: CAB, continuously active bursting, HFB, high frequency bursting, A D - - LFB.
non-identified low frequency bursting units located outside the supraoptic nuclear region, and A D ÷ LFB, antidromicatly identified tow frequency bursting. Any two means underscored by the saree solid line are not significantly different. Any two means not underscored by the same solid line are significantly different (P < 0.05). Comparison oj'spontaneous bursting patterns
Mean spikes per burst Mean burst firing rates (spikes sec) Mean burst durations tsec) Mean firing rates (spikes 'se:)
CAB
HFB
AD-- LFB
A D ~ LFB
7.09
9.22
23.23
33.99
I 1.37 0.66
19.19 0.57
4.05 6.22
4.03 9.76
6.62
L14
1.83
1.26
serial correlogram). Table I summarizes order independent statistical parameters associated with each firing pattern type and Table I1 summarizes bursting statistics for representative units exhibiting CAB, HFB. and LFB patterns of discharge.
Definition of firing pattern types Continuously active slow (CAS). CAS neurons discharged spontaneously with intermittently occurring spikes having an average interval greater than 1 sec. The ISIH has a broad asymmetrical modal peak with a mode less than the mean These spike trains exhibited predominantly long intervals on the dot raster plot, although numerous short intervals were generally present indicating occasional clustering of spikes. The auto- and serial correlation plots tended to be flat with no short term (period less than 60 sec) repeating periodicities. Continuously active fast fCAF). C A F neurons were typified by irregularly discharging trains of predominantly singly occurring spikes having a mean interval tess than l sec. The ISIH appeared as a Poisson distribution with broad, asymmetrical modal peak. and a mode less than the mean. Spike trains from these neurons may exhibit long cyclic or irregular fluctuations in mean firing rate (MFR) and, like CAS neurons, have occasional clusters of action potentials. However, the autocorre~ation function and serial correlation coefficient plots tend to be fiat, indicating no regularly occurring short term periodicities and the overall visual impression is not one of periodic activity. The C A F category tended to serve as 'catch-all' for units exhibiting ambiguous spontaneous firing patterns without consistent bursting or regularity. Continuously active regular (CA R). These neurons exhibited an unusual regularit y of interspike intervals. They were defined as a train of single spikes occurring at relatively regular time intervals. The ISIH had a relatively symmetrical modal peak with the mode close to the mean; unlike all other firing pattern types, there were few intervals scattered outside a normal distribution around the mode and none shorter than 200 msec. indicating a total absence of 'cluster' firing. The interval standard deviation is small; during occasional slow shifts in the MFR, spike intervals appeared to be
353 sequentially related with progressively shorter or longer intervals. The serial correlation coefficients were all between 0.3 and 1.0. Autocorrelation plots demonstrated numerous sharp peaks separated by a time interval (bin width × number of bins) approximately equal to the mean spike interval. CAR neurons in this study exhibited mean firing rates ranging from 2.0 to 9.4 spikes/sec. Because of C A R pattern resemblance to that sometimes generated with physical microelectrode damage to neurons, care was taken to include only units which met the following criteria: (1) did not change M F R with movement of electrode; (2) negative component of spike significantly greater in amplitude than the positive component, and (3) not killed when electrode was moved beyond the recording site and then back. Low frequency bursting (LFB). Low frequency bursting was characterized by relatively long duration (2-15 sec) and low frequency (2-8 spikes/sec) bursts having a period of from 6 to 50 sec; occasional single spikes or spike clusters may be present during interburst intervals, although bursts were usually followed by a period of silence. The overall M F R was greater than 1/sec, and their ISIH exhibited a narrow modal peak (mode greater than 60 msec) with several random long intervals. Sequential spike intervals during a burst were randomly distributed with occasional high frequency clusters unlike the 'parabolic burster' of invertebrates 25. The autocorrelation plot for 87 % of these spike trains demonstrated periodic peaks separated by time intervals having a mean of 29.5 ± 15.9 sec. The serial correlation coefficients were all close to zero. Mean burst characteristics are listed in Table I1. High.fi'equency bursting (HFB). These neurons were characterized by firing in short duration (0.3-0.8 sec) high frequency bursts (greater than 10 spikes/second), with irregular periods not less than twice the burst duration. The ISIH has a narrow short modal peak with a random distribution of longer intervals. Auto- and serial correlograms were flat in all instances, indicating no regular periodicities. The standard
30
-~ 2c
72 g n~ ~c
DS
CAS
DSON AD+
LFB
CAF
~]SON A D - -
HFB
CAB
CAR
~ E x t r a SON
Fig. 1. Percentages of firing pattern types recorded in the supraoptic nucleus and other areas of the hypothalamus. Abbreviations: S, silent; CAS, continuously active slow ; LFB, low frequency bursting; CAB, continuously active bursting; CAR, continuously active regular; SON AD4 , antidromically identified supraoptic neuroendocrine cells; SON AD--, non-identified cells in the supraoptic nucleus region; Extra-SON, neurons recorded from hypothalamic areas exclusive of the supraoptic nucleus region.
354 deviation of the interspike interval was always greater than 1000 sec. Mean burst characteristics are listed in Table I 1. Continuously active bursting (CAB). These neurons discharged in short, rapid and regular bursts giving a superficial appearance of continuous activity although few, if any, individual spikes were observed. The ISIH was bimodal with a short interval mode of 10-70 msec. The standard deviation was always less than 1000 msec. The autocorrelation plots of these spike trains all demonstrated periodic peaks separated by time intervals having a mean of 1.16 ~ 0.45 sec. The serial correlation coefficients were all close to zero. Mean burst characteristics are listed in Table I1.
Spontaneous firing patterns of ADq- SON. A D - - SON, and extra-SON neurons Fig. 1 demonstrates the spontaneous firing pattern distribution recorded between SON and extra-SON regions of the hypothalamus. Note that all 6 described firing pattern types were recorded from both regions. However, spontaneously discharging A D - - S O N neurons exhibited only CAS, C A F and LFB patterns of activity. Although HFB, CAR, and CAB patterns of discharge were recorded from A D - - SON neurons, none of the A D neurons discharged in this manner. Eight percent of the A D SON neurons were silent, and discovered only following pituitary stalk stimulation. Although the bar graph in Fig. 1 demonstrates a rough approximation of firing pattern occurrence recorded in SON and extra-SON areas of the hypothalamus, it was not generated from a truly random sampling procedure. Aside from the usual sampling bias toward larger more easily isolated units, an additional bias for selecting less prevalent and more interesting discharge patterns for firing pattern analysis influenced the extra-SON distribution of firing pattern types. Thus, in extra-SON areas. LFB and C A R discharging neurons are over represented and C A F discharging neurons are under represented. This latter type of sampling bias did not exist for the population of A D SON neurons where all 75 well isolated units identified by pituitary stalk stimulation were studied. Chi-square analysis did not demonstrate a significant difference in the per cent total neurons discharging CAS. LFB, and C A F between these sample populations of A D ~ SON and unidentified extra-SON neurons. The t-test for non-paired experiments was used to test for significant differences between mean firing pattern statistical parameters (interspike intervals, modes, and coefficient of variation) found in A D SON neurons versus unidentified extra-SON neurons exhibiting similar firing patterns (Table IIl). AD4- SON neurons exhibiting CAS activity had significantly greater mean interspike intervals (P < 0.05) and interval mode (P < 0.001) than CAS discharging neurons found outside the SON region. C A F A D - - neurons exhibited a significantly greater (P < 0.01) coefficient of variation than C A F extra-SON neurons: this may be a reflection of greater tendencies toward cluster firing of AD4- SON C A F neurons. A D LFB neurons sampled in this study demonstrated a significantly longer mean interburst interval (20.86 sec) compared to unidentified extra-SON LFB neurons (t2.43 sec).
Firing pattern stability Recording time for each neuron during spontaneous firing averaged 20 min but
355
o ~5 q4
it
c5 = ~
c5 -d
=c5
c~
z ~,
©
r-
c~
tt S I~E
c~ = d
=
Lt
o
=N
Z
©
~5 -q a',q4 a~ +t
ti p_. tt~
÷
il
347
COMPARISON OF FIR1NG PATTERNS AND SENSORY RESPONSIVENESS BETWEEN SUPRAOPTIC AND OTHER HYPOTHALAMIC NEURONS IN THE UNANESTHET1ZED SHEEP
D. P. J E N N I N G S ,
J. T. HASKINS and J. M. ROGERS
Department 0[" Physiological Sciences, College of Veterinary Medicine, Oklahoma State University, Stillwater, Okla. 74074, The Marine Biomedical lnstitate, 200 University Boulevard, Galveston, Texas 77550 and DepJrtment Of Anatomy, Colorado State University, Fort Collins, Colo. (U.S.A.) (Accepted October 27th, 1977)
SUMMARY
Adult Southdown ewes were surgically prepared with pituitary stimulating electrodes, carotid and jugular cannulae, and a cranial platform-cylinder arrangement for chronic single unit recording. Isolated neurons (n z 112) in the region of the supraoptic nucleus (SON) were identified by pituitary stalk stimulation as AD'~ (antidromically invaded) SON neuroendocrine cells (n = 75) or AD-- (not antidromically invaded) SON neurons (n -- 37). Spontaneous firing pattern distribution and sensory evoked behavior of these SON region neurons were compared with activity recorded from 112 randomly located non-identified neurons of extra-SON areas of the hypothalamus. Spontaneous discharge activity was categorized into six distinct firing pattern types: continuously active slow (CAS), continuously active fast (CAF), continuously active bursting (CAB), continuously active regular (CAR), low frequency bursting (LFB), and high frequency bursting (HFB). These 6 firing pattern types were characterized by computer analysis and their mean order independent statistical parameters compared. Bursting discharge patterns (LFB, HFB, and CAB) were compared with respect to mean burst duration, burst mean firing rate, and interburst intervals. Ninety-three per cent of all neurons maintained a stable discharge pattern in the absence of apparent stimuli. Occasionally CAS and CAF neurons spontaneously generated spike clusters sufficient to give the transient appearance of a bursting discharge pattern and LFB neurons lapsed spontaneously into CAS activity. All 6 firing pattern types recorded from non-identified extra-SON neurons were also recorded in the SON region. However, spontaneously discharging A D + SON neurons exhibited only continuously active slow (CAS), continuously active fast (CAF), and low frequency bursting (LFB) activity. The total absence of high frequency bursting (HFB), continuously active regular (CAR), and continuously active bursting (CAB)
348 patterns of discharge from A D ..... SON neurons suggests that A D - - SON neurons exhibiting these firing patterns may function as mterneurons, pacemaker neurons. or receptor neurons. A significant number of LFB discharging neurons were recorded in widespread extra-SON regions of the hypothalamus, indicating this discharge pattern may not be unique to magnocellular neuroendocrine cells. AD-~ SON LFB neurons sampled in this study demonstrated a significantly longer mean interburst interval (20.86 sec) compared to extra-SON LFB neurons (12.43 sec). No A D - - SON neuron tested was significantly sensitive to non-specific sensory arousal or sleep-waking state changes. In extra-SON areas of the hypothalamus. I l of 75 neurons tested to sensory arousal and 6 of 19 neurons tested to sleep-waking changes responded with significant changes in mean firing rate (MFR): no significant difference between firing pattern types was demonstrated in arousal or sleep-waking sensitivity. Significantly more A D .... SON neurons (66 90) responded to rapid intracarotid injections or slow intrajugular infusions of hypertonic NaC1 than extra-SON neurons (19~). It was concluded that extra-SON neurons could have been affected nonspecifically by the behavioral and E E G arousal effects of rapid intracarotid injections of hypertonic NaC1. Fifty-one per cent of the A D SON neurons were excited, 12 O/o inhibited, and 38 9/00unaffected by slow intrajugular infusions of hypertonic NaCI. No one spontaneous firing pattern was significantly more osmosensitive than another; however. 8 0 ~ of the CAS or C A F discharging neurons excited by this osmotic loading became low frequency bursters during, or immediately, following the infusion.
INTRODUCT|ON Experiments on unanesthetized monkeys TM, urethane anesthetized rats 29 and. more recently, unanesthetized sheep s, have demonstrated that antidromically identified cells in the supraoptic and paraventricular nucleus may be silent or discharged spontaneously in one of two basic patterns, continuously active (irregular), or low frequency bursting (phasic). It has been suggested that the functional significance of these different firing pattern types may be attributable to different thresholds for sensory input 2, and to different functional neuroendocrine cell types, each having 'specific' input connections and associated with one hormone, such as A D H or oxytocin7,10. The low frequency bursting (LFB) pattern in particular has stimulated the interest of neuroendocrine investigators; it may be unique to neuroendocrine cells and responsible for the release Of hormone 'spurts' into the blood vascular system. This implies functional mechanisms controlling the firing behavior of neuroendocrine cells which may be unique in the mammalian central nervous system (CNS) to the functions o f a neuroendocrine system. Recent evidence m urethane anesthetized lactating rats suggests that this pattern is associated primarily with vasopres~n neuro-
349 endocrine cells and that oxytocin cells discharge continuously irregular with the exception of occasional spike bursts associated with milk ejection 21,3°. In vivo and in vitro electrical stimulation studies 6,1s suggest that a given number of action potentials are more effective in causing the release of a hormone if they occur close together in time. Thus it appears that recruitment of continuously irregular firing neuroendocrine cells into a bursting pattern of discharge could be an essential mechanism for releasing appropriate amounts of hormone from the posterior pituitary. Electrophysiological evidence11,15,17 suggest the presence of recurrent inhibition which may involve interneurons similar to Renshaw cells of the spinal cord. Other electrophysiologic studies by Hayward and Vincent 13 in the unanesthetized monkey suggest that osmoreceptor neurons in the perinuclear zone of the SON exist separately from neuroendocrine cells, and that they respond monophasically to an intracarotid pulse injection of hypertonic NaC1. Thus, it appears likely that the supraoptic nucleus region, including its perinuclear zone, may contain a functionally heterogeneous population of receptor neurons, interneurons, and magnocellular neuroendocrine cells of at least two types (oxytocin and vasopressin secreting). The purpose of this paper is to describe the firing pattern distribution and sensory evoked behavior of antidromically identified neuroendocrine cells and other neurons of the supraoptic nucleus region in the unanesthetized sheep. In order to determine the functional uniqueness of this neuroendocrine associated pool of neurons, these characteristics were compared to similar information from a randomly selected population of unidentified neurons recorded from widespread extra-SON portions of the hypothalamus. In order to avoid possible influences from chemical anesthesia, behavioral levels of consciousness, or acute surgical complications, it was deemed essential to record from an unanesthetized animal, well conditioned to its recording environment and with sleep-waking behavior carefully monitored. Sheep were chosen as the experimental animal for numerous reasons. Because of their large size and placid behavior, they tolerate, with minimal restraint, the necessary cranial platform and microdrive assembly required for single unit recording in the unanesthetized animal, and are easily prepared for multiple blood sampling. Although single unit recording has not previously been reported in this species, considerable information is available concerning anteriorS, 19 and posterior1,14, 22 pituitary function as well as sleep-waking behavior 23 in goats and sheep. This makes them particularly attractive as a research animal for future neuroendocrine studies. MATERIALS AND METHODS Southdown ewes (4-9 years of age) previously conditioned to a restraining stanchion and recording chamber were surgically fitted with biparietal epidural EEG electrodes, periorbital eye movement electrodes, and a cranial platform-cylinder arrangement similar to that described by Hayward and Jennings 10 for the monkey. A lateral radiograph of the head was taken immediately following surgery, and frontal coordinates determined with reference to the implanted cylinder for location of the
350 pituitary fossa. Two or three days following this surgery, the ewe was again anesthetized for permanent implantation of bipolar electrodes across the pituitary stalk and placement of silicone rubber cannulae in each jugular vein. The carotid artery was cannulated via the thyrolaryngeal artery in 3 sheep utilizing a technique described by Buck ~. Correct placement of the pituitary stimulating electrodes was later confirmed when electrical stimulation through them evoked an antidiuresis in the experimentally hydrated animal. The jugular and carotid cannulae were filled with heparimzed saline (1000 units/ml) and led subcutaneously to the back of the sheep, where they were exteriorized and terminated into capped hypodermic needles within a canvas pouch fastened to the skin of the sheep. Following recovery from the second surgery, as indicated by eating at preoperative levels, the animal was placed in a restraining stanchion located within the recording chamber. The stanchion was designed with a head restraining platform and nose band which painlessly steadied the head. allowing the animal to doze intermittently during single unit recordings and to behave otherwise normally. Tungsten microelectrodes were hydraulically driven into the hypothalamus at coordinates which could be determined by positioning a Trent Wells, Inc. titanium micropositioner fastened to the cylinder. Extracellular single unit activity was amplified and channeled into a storage oscilloscope monitor (Tektronix 5103N/013), a magnetic tape recorder (Sony TC-366), an audiomonitor, and a pulse height discriminator. Two outputs from the pulse height discriminator (a one-to-one pulse output and an analog output proportional to the rate of discharge) were simultaneously recorded on an ink writing polygraph (Grass Model 7) with the cortical EEG and eye movement potential. A television camera placed inside the recording chamber permitted observation of the animal via a closed circuit television monitor. On the basis of EEG, eye movement. body movement, and observed behavior, judgement was made as to whether the animal was sleeping, drowsy, or alert awake. A written record provided by the polygraph allowed the investigators to correlate sleep-waking and sensory arousal behavior with the sheep's single neuron firing behavior. Recording periods to be analyzed were replayed from magnetic tape and the spike activity displayed on a storage oscilloscope in order to establish the stability of each unit's waveform. Single cell spike trains clearly separable from base line activity and neighboring units were led into the pulse height discriminator which emitted 0.5 msec pulses triggered by the unit's action potentials. These pulses were recorded onto a second magnetic tape which was used to generate rate meter and dot raster a displays for visual analysis of spike train firing patterns. The same tapes were played back onto a digital computer (lnterdata 167 programmed to calculate mean interspike interval, standard deviation of the intervals, coefficient of varmnce, and mean firing rate, as well as plot interspike interval histograms, accumulative rate plots, autocorrelograms and serial correlograms 2°. Burst analysis of those spike trains exhibiting definite short term periodicities was accomplished with the aid of a digital counter which was reset and triggered manually to count spikes during a burst. The sheep were provided water ad libitum between experiments and were in a state of normal hydration. The recording time for each neuron averaged 30 min,
351 but ranged from 5 rain to 3 h. Once the spontaneous firing pattern of a neuron was established, it was tested for sensitivity to non-specific sensory arousal, sleep-waking behavior (if possible), and intrajugular or intracarotid injections of hypertonic NaCI solutions. The location of neurons outside or within 1 mm of the histologic boundaries of the SON was determined by plotting their relative location with respect to Prussian blue spots 9laced by a steel microelectrode at several known levels along the recording tracts prior to terminating an animal. Physiologic confirmation of each neuron's location was accomplished by recording the distance it was recorded above the light sensitive optic chiasm and by observing the presence or absence of an evoked electric field resulting from single pulse stimulation of the pituitary stalk. RESULTS
Spike trains from a total of l l2 units located within the SON region (an area defined to be within 1 mm of the SON's histologic boundaries) were studied. Of these, 75 were invaded antidromically following stimulation of the pituitary stalk ( A D + SON), and assumed to be neuroendocrine cells; 37 were not invaded antidromically and referred to as A D - - SON neurons. The spontaneous firing patterns and sensitivity of SON neurons to sleep-waking behavior, non-specific sensory arousal, and intrajugular or intracarotid infusions of hypertonic NaCI were determined and compared to similar data from a population of 112 unidentified neurons recorded from randomly located areas of the hypothalamus outside the defined SON region. The spontaneous activity of all individual neurons was categorized into one of the 6 following categories: continuously active slow (CAS), continuously active fast (CAF), continuously active regular (CAR), continuously active bursting (CAB), high frequency bursting (HFB), and low frequency bursting (LFB). These firing pattern 'types' were further characterized by computer analysis of their component spike trains (interspike interval histogram (ISIH), accumulative rate, autocorrelogram, and TABLE 1
Comparison o f mean statistics between firing pattern types Abbreviations used are: CAR, continuously active regular, CAF, continuously active fast, HFB, high frequency bursting, CAB, continuously active bursting, LFB, low frequency bursting, and CAS, continuously active slow. Any two means underscored by the same solid line are not significantly different. Dotted lines act to connect solid lines. Any two means not underscored by the same solid line are significantly different (P < 0.05).
Comparison o f ~pontaneous firing patterns
CAR
CAF
HFB
CAB
LFB
CAS
Mean firing rates
5.02
4.84
3.24
7.13
1.57
0.42
Mean interspike intervals
0.25
0.34
0.39
0.15
0.86
7.05
Mean interspike interval modes
0.27
0.10
0.02
0.05
0.15
0.44
Mean coefficients of variation
3.19
0.92
0.36
0.42
0.46
0.80
352 TABLE II Comparison o f mean burst parameters between hring pattern type,s
Abbreviations used are: CAB, continuously active bursting, HFB, high frequency bursting, A D - - LFB.
non-identified low frequency bursting units located outside the supraoptic nuclear region, and A D ÷ LFB, antidromicatly identified tow frequency bursting. Any two means underscored by the saree solid line are not significantly different. Any two means not underscored by the same solid line are significantly different (P < 0.05). Comparison oj'spontaneous bursting patterns
Mean spikes per burst Mean burst firing rates (spikes sec) Mean burst durations tsec) Mean firing rates (spikes 'se:)
CAB
HFB
AD-- LFB
A D ~ LFB
7.09
9.22
23.23
33.99
I 1.37 0.66
19.19 0.57
4.05 6.22
4.03 9.76
6.62
L14
1.83
1.26
serial correlogram). Table I summarizes order independent statistical parameters associated with each firing pattern type and Table I1 summarizes bursting statistics for representative units exhibiting CAB, HFB. and LFB patterns of discharge.
Definition of firing pattern types Continuously active slow (CAS). CAS neurons discharged spontaneously with intermittently occurring spikes having an average interval greater than 1 sec. The ISIH has a broad asymmetrical modal peak with a mode less than the mean These spike trains exhibited predominantly long intervals on the dot raster plot, although numerous short intervals were generally present indicating occasional clustering of spikes. The auto- and serial correlation plots tended to be flat with no short term (period less than 60 sec) repeating periodicities. Continuously active fast fCAF). C A F neurons were typified by irregularly discharging trains of predominantly singly occurring spikes having a mean interval tess than l sec. The ISIH appeared as a Poisson distribution with broad, asymmetrical modal peak. and a mode less than the mean. Spike trains from these neurons may exhibit long cyclic or irregular fluctuations in mean firing rate (MFR) and, like CAS neurons, have occasional clusters of action potentials. However, the autocorre~ation function and serial correlation coefficient plots tend to be fiat, indicating no regularly occurring short term periodicities and the overall visual impression is not one of periodic activity. The C A F category tended to serve as 'catch-all' for units exhibiting ambiguous spontaneous firing patterns without consistent bursting or regularity. Continuously active regular (CA R). These neurons exhibited an unusual regularit y of interspike intervals. They were defined as a train of single spikes occurring at relatively regular time intervals. The ISIH had a relatively symmetrical modal peak with the mode close to the mean; unlike all other firing pattern types, there were few intervals scattered outside a normal distribution around the mode and none shorter than 200 msec. indicating a total absence of 'cluster' firing. The interval standard deviation is small; during occasional slow shifts in the MFR, spike intervals appeared to be
353 sequentially related with progressively shorter or longer intervals. The serial correlation coefficients were all between 0.3 and 1.0. Autocorrelation plots demonstrated numerous sharp peaks separated by a time interval (bin width × number of bins) approximately equal to the mean spike interval. CAR neurons in this study exhibited mean firing rates ranging from 2.0 to 9.4 spikes/sec. Because of C A R pattern resemblance to that sometimes generated with physical microelectrode damage to neurons, care was taken to include only units which met the following criteria: (1) did not change M F R with movement of electrode; (2) negative component of spike significantly greater in amplitude than the positive component, and (3) not killed when electrode was moved beyond the recording site and then back. Low frequency bursting (LFB). Low frequency bursting was characterized by relatively long duration (2-15 sec) and low frequency (2-8 spikes/sec) bursts having a period of from 6 to 50 sec; occasional single spikes or spike clusters may be present during interburst intervals, although bursts were usually followed by a period of silence. The overall M F R was greater than 1/sec, and their ISIH exhibited a narrow modal peak (mode greater than 60 msec) with several random long intervals. Sequential spike intervals during a burst were randomly distributed with occasional high frequency clusters unlike the 'parabolic burster' of invertebrates 25. The autocorrelation plot for 87 % of these spike trains demonstrated periodic peaks separated by time intervals having a mean of 29.5 ± 15.9 sec. The serial correlation coefficients were all close to zero. Mean burst characteristics are listed in Table I1. High.fi'equency bursting (HFB). These neurons were characterized by firing in short duration (0.3-0.8 sec) high frequency bursts (greater than 10 spikes/second), with irregular periods not less than twice the burst duration. The ISIH has a narrow short modal peak with a random distribution of longer intervals. Auto- and serial correlograms were flat in all instances, indicating no regular periodicities. The standard
30
-~ 2c
72 g n~ ~c
DS
CAS
DSON AD+
LFB
CAF
~]SON A D - -
HFB
CAB
CAR
~ E x t r a SON
Fig. 1. Percentages of firing pattern types recorded in the supraoptic nucleus and other areas of the hypothalamus. Abbreviations: S, silent; CAS, continuously active slow ; LFB, low frequency bursting; CAB, continuously active bursting; CAR, continuously active regular; SON AD4 , antidromically identified supraoptic neuroendocrine cells; SON AD--, non-identified cells in the supraoptic nucleus region; Extra-SON, neurons recorded from hypothalamic areas exclusive of the supraoptic nucleus region.
354 deviation of the interspike interval was always greater than 1000 sec. Mean burst characteristics are listed in Table I 1. Continuously active bursting (CAB). These neurons discharged in short, rapid and regular bursts giving a superficial appearance of continuous activity although few, if any, individual spikes were observed. The ISIH was bimodal with a short interval mode of 10-70 msec. The standard deviation was always less than 1000 msec. The autocorrelation plots of these spike trains all demonstrated periodic peaks separated by time intervals having a mean of 1.16 ~ 0.45 sec. The serial correlation coefficients were all close to zero. Mean burst characteristics are listed in Table I1.
Spontaneous firing patterns of ADq- SON. A D - - SON, and extra-SON neurons Fig. 1 demonstrates the spontaneous firing pattern distribution recorded between SON and extra-SON regions of the hypothalamus. Note that all 6 described firing pattern types were recorded from both regions. However, spontaneously discharging A D - - S O N neurons exhibited only CAS, C A F and LFB patterns of activity. Although HFB, CAR, and CAB patterns of discharge were recorded from A D - - SON neurons, none of the A D neurons discharged in this manner. Eight percent of the A D SON neurons were silent, and discovered only following pituitary stalk stimulation. Although the bar graph in Fig. 1 demonstrates a rough approximation of firing pattern occurrence recorded in SON and extra-SON areas of the hypothalamus, it was not generated from a truly random sampling procedure. Aside from the usual sampling bias toward larger more easily isolated units, an additional bias for selecting less prevalent and more interesting discharge patterns for firing pattern analysis influenced the extra-SON distribution of firing pattern types. Thus, in extra-SON areas. LFB and C A R discharging neurons are over represented and C A F discharging neurons are under represented. This latter type of sampling bias did not exist for the population of A D SON neurons where all 75 well isolated units identified by pituitary stalk stimulation were studied. Chi-square analysis did not demonstrate a significant difference in the per cent total neurons discharging CAS. LFB, and C A F between these sample populations of A D ~ SON and unidentified extra-SON neurons. The t-test for non-paired experiments was used to test for significant differences between mean firing pattern statistical parameters (interspike intervals, modes, and coefficient of variation) found in A D SON neurons versus unidentified extra-SON neurons exhibiting similar firing patterns (Table IIl). AD4- SON neurons exhibiting CAS activity had significantly greater mean interspike intervals (P < 0.05) and interval mode (P < 0.001) than CAS discharging neurons found outside the SON region. C A F A D - - neurons exhibited a significantly greater (P < 0.01) coefficient of variation than C A F extra-SON neurons: this may be a reflection of greater tendencies toward cluster firing of AD4- SON C A F neurons. A D LFB neurons sampled in this study demonstrated a significantly longer mean interburst interval (20.86 sec) compared to unidentified extra-SON LFB neurons (t2.43 sec).
Firing pattern stability Recording time for each neuron during spontaneous firing averaged 20 min but
355
o ~5 q4
it
c5 = ~
c5 -d
=c5
c~
z ~,
©
r-
c~
tt S I~E
c~ = d
=
Lt
o
=N
Z
©
~5 -q a',q4 a~ +t
ti p_. tt~
÷
il
