Title: Behavior/stress interactions in diabetes mellitus
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00102847/00001
 Material Information
Title: Behavior/stress interactions in diabetes mellitus
Physical Description: Book
Language: English
Creator: Bellush, Linda L., 1943-
Copyright Date: 1986
 Record Information
Bibliographic ID: UF00102847
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 15682918
ltuf - AEL4669

Full Text








I decided early in my academic training to pursue a very

multidisciplinary course. This has not always been an easy

approach. I feel extremely fortunate to have found a

program such as the one at the University of Florida, the

Center for Neurobiological Sciences, which has the specific

goal of bringing together people from different fields of

study. My committee reflects this atmosphere of sharing,

and it has been a great joy to feel welcome in the

laboratories of neuroscientists and physiologists, as well

as those of psychologists.

To my committee members, Drs. Adrian Dunn, Melvin

Fregly, William Luttge, Merle Meyer and Carol

VanHartesveldt, I wish to express thanks for so many helpful

interactions as I planned and executed my dissertation work.

Special gratitude must be extended to my advisor, Neil

Rowland, who in addition to the usual difficult tasks

involved in training a new scientist, was called upon for

assistance in both professional and personal crises.

In addition to this wonderful professional family, I owe

a great deal to some loved ones. First, thanks are extended

to my children, Pam and Greg Bellush, who haven't had very

much of a mother for several years, but who have been

unfailingly supportive and proud of my endeavors. Thanks

are also extended to Dr. Bill Henley, my very best friend,

my partner. I'm quite certain that without his constant

optimism and encouragement and his unflagging faith in me,

this dissertation might not have become a reality.

Thanks must go, finally, to my parents, Dorothy and John

Jose, who supported me in ways too numerous to count

throughout my life, and in particular during my difficult

years in college. They have both died during my years in

school and therefore cannot be here to see the culmination

of all the years of struggle. This saddens me deeply, for

they so deserved to share in the rewards. This dissertation

is my tribute to their memory.



ACKNOWLEDGEMENTS ......................................




Clinical Perspectives........................

Experimental Perspectives....................

Summary.................................... ..



Experiment 1. Flinch-Jump Thresholds
of Diabetic and Nondiabetic Rats.............

Experiment 2. Open-Field Activity of
Diabetic and Nondiabetic Rats...............



Experiment 3. Retention for Passive
Avoidance Training in Diabetic and
Nondiabetic Rats ............................

Experiment 4. Plasma Corticosterone Con-
centrations Following Footshock Training
and Retention Testing in Diabetic and
Nondiabetic Rats ............................

Experiment 5. Twenty-Four Hour Catechol-
amine Excretion in Diabetic and Nondiabetic
Rats ............ ..........................


















Experiment 6. Neurochemical Correlates of
Aversive Learning in Diabetic and
Nondiabetic Rats............................. 107

TO CHRONIC COLD EXPOSURE.............. ......... 124

Experiment 7. Plasma Corticosterone
Concentrations, Catecholamine Excretion
and Central Catecholamine Concentrations
in Diabetic and Nondiabetic Rats
Exposed to Chronic Cold ..................... 126

V GENERAL DISCUSSION...... ........................ 164

REFERENCES........................................... 172

BIOGRAPHICAL SKETCH................................... 183

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Linda L. Bellush

December 1986

Chairman: Neil E. Rowland, Ph.D.
Major Department: Psychology

The responses of diabetic and nondiabetic rats to two

stressors were investigated. In the first series of

experiments, retention of passive avoidance was examined.

Possible state-dependent effects of glycemic state were

evaluated in separate diabetic and nondiabetic groups.

Additionally, retention of diabetic and nondiabetic rats was

compared. Also compared were corticosterone concentrations,

catecholamine excretion and catecholamine concentrations in

brain. Changes in glycemic state produced no state-

dependent effects in retention. However, diabetic rats

showed better retention for the task than nondiabetic rats.

Associated with retention differences, diabetics had higher

epinephrine excretion and nondiabetics lower excretion after

footshock training, relative to baseline measures.

Diabetics had higher norepinephrine excretion than

nondiabetics at both times. Diabetic rats had higher

norepinephrine concentrations in amygdala, hypothalamus and

brainstem, and higher dopamine concentrations in

hypothalamus. These data suggest that in diabetes mellitus,

alterations in peripheral and central catecholamine

functioning may result in greater responsivity to stressful

episodes. Another experiment evaluated metabolic

parameters in insulin- and saline-treated diabetic and

nondiabetic rats after 72 hours in a cold environment.

Corticosterone concentrations were significantly elevated

after three days of cold, and insulin had opposite effects

upon this parameter in diabetic and nondiabetic rats.

Diabetics had higher norepinephrine excretion than

nondiabetics during the first 24 hours of cold and lower

dopamine excretion during the last 24 hours. In the cold,

dopamine in frontal cortex was influenced in opposite ways

by insulin treatment, with concentrations in diabetics

increased and those in nondiabetics decreased. Diabetic

rats had higher norepinephrine concentrations in frontal

cortex and brainstem irrespective of insulin treatment.

These biochemical alterations in diabetic rats, and the

failure of insulin to normalize them, suggest that both

acute and chronic stresses may have different effects upon

the diabetic, because of different biochemical status


produced by diabetes and the failure of insulin to normalize

fully the physiological processes involved in adaptation to




Clinical Perspectives

Psychosocial Aspects of Diabetes

The streptozotocin-diabetic rat has provided a useful

model for the investigation of treatment, prevention, causal

factors and complications of insulin-dependent diabetes

mellitus (Mordes & Rossini, 1981). However, very little of

the animal research has focused upon questions related to

behavioral interactions with metabolic control in diabetes.

This is somewhat surprising, given the vast clinical

literature pertaining to psychological aspects of this

endocrine disorder in humans.

Menninger (1935), Dunbar, Wolfe and Rioch (1936) and

Daniels (1939) published some of the earliest formulations

concerning psychological aspects of diabetes. All of these

theorists emphasized the contribution of both personality

factors and environmental stresses in the precipitation of

diabetes, as well as in the course of the disorder. In

addition, they all leaned heavily on psychoanalytic concepts

such as that of unconscious conflict which could lead to

physiologic expression. Thus, Daniels (1939) noted that

certain personality "types" were predisposed to respond to

anxiety-provoking events by becoming clinically glycosuric

and hyperglycemic as a consequence of autonomic activation.

The environmental stressor most often linked to the onset of

diabetes was the loss of a loved one or object. The

psychoanalytic interpretation of the ensuing events involved

a mixture of hostility at the perceived abandonment and

sadness at the loss. These emotions, if not expressed in

other ways, could result in autonomic events ultimately

producing diabetes (Daniels, 1939).

Menninger (1935) and Dunbar et al. (1936) stated similar

theories relating personality factors and environmental

stresses to the onset and control of diabetes. Menninger

(1935) noted an increased incidence of glycosuria among

patients with other psychosomatic illnesses, particularly


Numerous studies have been published in the intervening

years from 1935 to the present which have attempted to

demonstrate either environmental stresses or personality

factors involved in the predisposition to diabetes. For

example, Stein and Charles (1971) suggested that a large

number of diabetics experienced the death of a loved one

prior to the onset of the disorder. However, these authors

included deaths occurring as long as several years prior to

the development of diabetic symptomology. Hinkle and Wolf

(1952) also argued that stress could be an important factor

in diabetes onset. Stress continues to be studied as a

variable of importance, both in the precipitation of

diabetes and in maintenance of metabolic stability (Johnson,

1980). Indeed, it has been suggested recently that diabetes

represents a state of uncontrolled stress (Zaidise &

Bessman, 1984).

As for predisposing personality factors, it has been

difficult to determine whether personality traits such as

dependence and anxiety cause, or are caused by, diabetes. A

number of studies have reported indications of increased

depression, anxiety or other adjustment problems among

diabetics (e.g. Swift, Seidman, & Stein, 1967; Simonds,

1977; Sullivan, 1979; Murawski, Chazan, Balodimos, & Ryan,

1970; Sanders, Mills, Martin, & DeL. Home, 1975). However,

a number of methodologic flaws such as subjective interview

techniques, invalid or unreliable tests (e.g. projective

tests such as the Rorschach), questionable selection of both

diabetic and control populations have marred the

interpretation of many of the results. Moreover, other

studies have found good adjustment, in general, among

diabetics (Kovaks, Feinberg, Paulauskas, Finkelstein,

Pollock, & Crouse-Novak, 1985). There is some evidence that

a relationship exists between diabetes and depression

(Lustman, Amado, & Wetzel, 1983) but the precise incidence

and prevalence of depression in diabetics and the impact of

depression upon blood glucose concentration have not been

established (Kymissis, Shatin, & Brown, 1979; Lilliker,


It has been questioned whether any consistent

personality differences or psychological problems actually

characterize diabetics as a group (Bruch, 1949; Johnson,

1980; Johnson & Rosenbloom, 1982). Nonetheless, there is a

continuing concern that psychosocial adjustment is a

relevant issue bearing upon metabolic status in the

treatment of diabetes (Johnson, 1980). For example, the

lifestyle of the diabetic is subject to numerous

constraints. It is necessary to self-inject insulin once or

several times each day, to test the urine or blood

frequently for determination of glucose concentrations, to

eat carefully specified types of foods at strictly specified

times, and to avoid many of the most desirable foods

altogether. No matter how carefully these restrictions are

observed, episodes of insulin shock occur from time to time.

In addition, these individuals live with the knowledge that

they are at greater risk than other individuals for the

development of nephropathy, retinopathy and neuropathy

(Johnson, 1980).

While good glycemic control is likely the best way to

avoid both short-term and long-term problems, compliance to

the diabetic regimen continues to be a major problem in

treatment (Johnson, 1980). This is not altogether

surprising. Given the numerous details of diabetes


management listed above, it is clear that even under the

best of circumstances such a demanding lifestyle could

become emotionally taxing. Moreover, there are often no

immediate consequences of the failure to comply with the

restrictions. In fact, some anecdotal evidence suggests

that the diabetic individual may actually feel more

energetic when somewhat hyperglycemic than when

normoglycemic (Bruch, 1949).

When levels of insulin in blood are inadequate, blood

glucose rises, and ketoacidosis and diabetic coma may also

occur in some diabetics. Overinsulinization, on the other

hand, can produce unsettling symptoms such as sweating and

dizziness, shock, coma or even convulsions, which can be

both embarrassing and life-threatening. Given the tedious

process involved in maintenance of appropriate glycemic

levels, it is not difficult to understand why compliance

problems often occur, particularly during difficult

developmental stages such as adolescence (Johnson, 1980).

Thus, rebellion against the environmental stringencies

imposed by diabetes may result in metabolic derangements

which may have both behavioral and physiological


There is evidence that stress may directly influence

metabolic parameters even when the diabetic regimen is

followed (Johnson, 1980). For patients referred to as

"brittle diabetics," who are particularly susceptible to the

development of hyperglycemia and ketoacidosis (Tattersall,

1981), establishing metabolic stability can be a constant

struggle. Even diabetics who are not particularly brittle,

and who have not omitted scheduled insulin doses, may

experience sudden ketoacidosis. In one report of 25 such

individuals (MacGillivray, Bruck, & Voorhess, 1981),

stressful events, including infection and psychological

stress, could be identified in 20 cases to have occurred

prior to the onset of the metabolic crisis. It was

postulated that stress overwhelmed the control imposed by

the prescribed dose of insulin, leading to hyperglycemia,

ketoacidosis and striking elevations of the stress-related

hormones, cortisol and epinephrine.

Much attention has been given both to compliance

problems and to the stresses inherent in family and

individual adjustment to diabetes, as they relate to

metabolic stability. It is clear that disturbed family

interactions are associated with disruptions in diabetic

control (Bruch, 1949; Johnson, 1980). However, much less

consideration has been given in the clinical literature to

the possibility that primary alterations in metabolism in

diabetic individuals may actually lead to behavioral changes

(Johnson, 1980). These metabolically induced behavioral

alterations could, in turn, cause further disruption in

psychosocial interactions, as well as in metabolic

stability. It is conceivable that hormonal or metabolic

changes in the diabetic individual may interfere with

comprehension of the instructions given by the physician.

Thus, compliance problems may not be due solely to rebellion

against an austere lifestyle. Altered metabolism may also

contribute to behavioral problems such as depression and

anxiety (Leigh & Kramer, 1984).

Numerous investigators have attempted to determine

whether there are behavior problems which arise specifically

in connection with diabetes (Bruch, 1949). This is, of

course, another version of the hypothesis of a diabetic

personality, with personality here seen as the effect rather

than the cause of diabetes. Just as no constellation of

personality traits seems to lead to onset of diabetes, no

such constellation can be identified consistently in

established diabetes (Dunn & Turtle, 1981; Johnson, 1980;

Bruch, 1949). Nonetheless, it may be that at least some

diabetics are vulnerable to depression or anxiety (Lustman

et al., 1983), and perhaps all diabetics may be more

sensitive to stress than are nondiabetics (Johnson, 1980).

Many studies reporting no difference in the psychosocial

adjustment of diabetics relative to that of nondiabetics

(e.g. Kovacs et al., 1985) have been careful to establish

"representative" samples of diabetics and to test subjects

under carefully controlled conditions. It is possible that

specific groups of diabetics, for example "brittle

diabetics" or diabetics with early onset diabetes are

uniquely at greater risk than other diabetics for

development of behavioral problems. Little investigation of

such high risk diabetics has been done, yet differences may

be found by comparing various subgroups of diabetic

individuals (Johnson, 1980), or by intentionally

manipulating the stress of the testing situation (Kemmer,

Bisping, Steingruber, Baar, Hardtmann, Schlaghecke, &

Berger, 1986).

In one recent study (Linn, Linn, Skyler, & Jensen,

1983), Type I (insulin-dependent) diabetic men reported a

larger number of stressful life events on the Holmes and

Rahe Social Readjustment Scale than Type II (non-insulin

dependent) diabetics. The Type I diabetics also perceived

more stress in each episode, experienced greater anxiety,

and suffered poorer immune function and metabolic control.

Neither family support nor compliance differed between the

two groups.

In another report (Jacobson, Rand, & Hauser, 1985) loss

of glycemic control was related to the reported incidence of

negative life events in diabetics experiencing early signs

of retinopathy. Diabetics with no retinopathy and diabetics

with long-standing retinopathy showed no such association

between stress and metabolic stability. Thus, there was a

relationship between glycemic control, psychological stress

and metabolic deterioration during a specific extended

period of time. Both this study and the study of Linn et

al. (1983) compared subgroups of diabetics. If all the

diabetics had been evaluated together as a single group and

compared with a control group of nondiabetic individuals,

the important interactions between stress and metabolic

stability may have been masked.

Finally, the majority of studies evaluating behavioral

interactions with hormonal control in diabetes have

emphasized the notion that the constraints of the diabetic

lifestyle lead to behavioral problems which in turn lead to

metabolic problems. It must also be considered that

hormonal imbalances created by diabetes mellitus may not be

totally normalized by insulin therapy, and are themselves

the primary causes of behavioral problems. This latter

possibility is a major impetus for using animal models of

diabetes, in which compliance is not a problem, and

metabolic state can be manipulated directly in order to

determine the effects of metabolic parameters upon

behavioral variables.

In summary, while there is not yet consensus as to the

specific role played by psychological factors in metabolic

stability in diabetic individuals, most current

investigators do feel that there are behavior-hormone

interactions that warrant further investigation (Johnson,

1980; Fisher, Delamater, Bertelson, & Kirkley, 1982).

Cognitive Changes in Diabetes


In addition to the psychosocial factors that may exert

influence upon the course of diabetes, there may also be

cognitive changes in diabetic individuals. A number of

neuropsychological functions have been compared in diabetic

and nondiabetic individuals, and some potentially important

differences have been noted.

For example, electroencephalograms (EEGs) have been

reported to be abnormal in many more diabetic than

nondiabetic individuals (Eeg-Olofsson, 1977; Haumont,

Dorchy, & Pelc, 1979). Such abnormalities in diabetic

brainwave patterns include lower alpha frequency and

increased paroxysmal activity. Abnormal paroxysmal activity

has been correlated with behavioral and perceptual

disturbances and with emotional liability in diabetic

individuals and with the occurrence of severe hypoglycemic

symptoms (Eeg-Olofsson, 1977). EEG abnormalities have also

been shown to correlate significantly with poor diabetic

control, as well as with retinopathy (Haumont et al., 1979).

Ack, Miller and Weil (1961) noted lower IQ scores on the

Stanford-Binet test in diabetics with onset prior to age 5,

when compared with IQ scores of nondiabetic siblings. No

difference in IQ scores between diabetics with onset after

age 5 and their nondiabetic siblings was found. A

relationship was also suggested between the magnitude of IQ

differences and the number of severe hyperglycemic and

hypoglycemic episodes experienced by the diabetic of early


These findings were confirmed and extended in a more

recent investigation by Ryan, Vega and Drash (1985)

utilizing a comprehensive neuropsychological test battery

which included the Wechsler Intelligence Scale for Children

(WISC-R). Adolescents with diabetes onset prior to age 5

performed more poorly that later-onset diabetics and matched

controls on virtually every test. These included measures

of intelligence, school achievement, visuospatial ability,

memory, motor speed and eye-hand coordination. Both age of

onset and duration affected performance on tests. Rovet,

Gore and Erlich (1983) obtained similar results when

comparing diabetics with onset prior to age 3 to later-onset

diabetics and nondiabetic siblings. Ryan et al. (1985)

believe the deficits in early onset diabetes derive from

the higher incidence of hypoglycemia leading to seizures and

subsequent brain damage in these youngsters.

Decreased cognitive functioning has also been reported

in older diabetics (Perlmuter, Hakami, Hodgson-Harrington,

Ginsberg, Katz, Singer, & Nathan, 1984). These diabetics

were type II diabetics (i.e. adult-onset diabetics).

However, 56 were receiving daily insulin injections and 34

others required oral hypoglycemic agents. The major deficit

found in these subjects when compared to nondiabetics was

impaired memory retrieval. Performance was poorer in

diabetics with neuropathy or with elevated hemoglobin Alc,

an indicator of poor metabolic control of long duration.

Not all studies have reported deficits in diabetic

cognitive performance. Fallstrom (1974) reported higher IQ

scores on the WISC in diabetics than in nondiabetics. This

was not the first study to obtain such results, but the

previous reports of higher IQ scores in diabetics had been

attributed to higher socioeconomic status of some samples of

diabetics. However, Fallstrom's (1974) sample was balanced

for socioeconomic status, and thus no simple confounding

variable could be identified which explained her results. A

relationship was found between age of onset of diabetes and

IQ in boys but not girls, presumably because few girls in

this sample had become diabetic before age 5.

In a study designed to evaluate the effects of maximal

metabolic control versus standard insulin control in

diabetics, 32 adults with duration of diabetes of about 20

years were administered, by telephone, the Wechsler Memory

Scale. No differences in performance were found as a

function of different treatment regimen. Thus, patients

exposed to hypoglycemic episodes due to the rigid glycemic

control of insulin pumps or multiple daily insulin

injections appeared to suffer no memory impairments relative

to diabetics receiving a single daily insulin injection. It

should be noted, however, that all subjects tested here had

had diabetes for 20 years and so may have suffered some

cognitive loss. No nondiabetic groups were included for

comparison in this study. Despite these shortcomings, the

study suggests that hypoglycemic episodes do not necessarily

cause impaired cognitive function.

Flender and Lifshitz (1976) evaluated motor

coordination, memory performance and concentration ability

in diabetic and nondiabetic children at different blood

glucose concentrations. Diabetics performed better than

nondiabetic controls on the memory test in the hyperglycemic

(150-300 mg/dl) and normoglycemic (91-150 mg/dl) ranges.

One potential problem with this study was the unusual

testing situation. Diabetics were tested hourly from 9 A.M.

to 3 P.M., a rather long session. The nondiabetics were

tested during glucose tolerance tests. Both of these

conditions may have exerted unknown effects upon


The effects of fluctuations in blood glucose

concentration upon cognitive performance of diabetic college

students were studied with the use of insulin infusions

which maintained blood glucose concentration at 60 mg/dl,

110 mg/dl and 300 mg/dl in successive time periods (Holmes,

Hayford, Gonzalez, & Weydert, 1983). Some impairment in

performance was observed at both high and low blood glucose

levels, with the greater impairment during hypoglycemia. It

would have been of interest to see if nondiabetic

individuals had suffered similar impairment at the lower

blood glucose levels, but no such subjects were studied.

The results do address the issue of fluctuations in blood

glucose experienced by diabetics routinely, and they suggest

that such excursions in glucose concentration may have some

cognitive impact.

Taken together, with the exception of Fallstrom (1974)

and Flender and Lifshitz (1976), these studies generally

agree that there are some cognitive deficits experienced by

diabetic individuals. These deficits appear to be more

prevalent in diabetics with early onset of the disease.

Metabolic instability during neural development may thus be

particularly damaging. The finding of alterations in EEG

patterns in some diabetic individuals suggests that there

may be physiological changes accompanying cognitive

impairment. Another indication of organic damage is found

in the lower flicker value of diabetics in the critical

flicker fusion test (Ryan, Vega, Longstreet, & Drash, 1985).

As with the issue of psychosocial adjustment of diabetic

individuals, diabetic cognitive functioning will require

further investigation to clarify the interaction between

metabolic state and behavior. However, it seems clear that

such interactions do occur in diabetes.

Experimental Perspectives

Hormonal and Neurochemical Correlates of Learning and Memory

In addition to the clinical findings which suggest a

relationship between the metabolic changes and behavioral

alterations occurring in diabetes, there are experimental

data which support the relevance of hormonal alterations in

diabetes to learning and memory. In order to explore these

experimental bases for investigating learning and memory in

the streptozotocin diabetic rat, I will first summarize some

of the known hormonal and neurochemical correlates of

learning and memory. Then I shall describe some of the

hormonal and neurochemical alterations in the streptozotocin

diabetic rat, as well as in human diabetics.

State-dependent retention

One major experimental paradigm utilized to examine

hormonal influences upon memory is that of state-dependent

retention. In this paradigm, it has been demonstrated that

if the internal state of subjects is different during

training than it is during retention testing, then retention

of the task may be attenuated relative to retention of the

task if the internal state during retention matches the

state existent during training (Overton, 1984). Training is

often carried out in rats, and the task is often one-trial

passive avoidance. Briefly, the animal is punished for

making a response and thus learns not to make the response

when given another opportunity to do so. The internal state

of the animal can be altered by administration of drugs,

including hormones. Izquierdo and Dias (1983) demonstrated

state-dependent retention of passive avoidance in rats with

exogenous administration of both ACTH and epinephrine.

Unfortunately, the notion of state dependency has become

a standard explanation for any decrement in performance of a

learned task, thus weakening its ability to contribute any

clearcut explanatory power. Moreover, it has been difficult

to defend the relevance of state dependent phenomena to real

world situations. One such situation often cited is the

loss of memory for events occurring during intoxication.

However, state dependent retention is not necessarily the

only, or even the best, explanation for such memory loss.

Diabetes mellitus, on the other hand, presents a

situation in which there can be a great deal of fluctuation

in glycemic state, and often in other metabolic parameters

affected by glycemic state as well. Thus, diabetics may

experience very high blood glucose which can be accompanied

by elevated ketones, epinephrine and cortisol (MacGillivray

et al., 1981). Alternatively, a number of diabetics also

experience episodes of hypoglycemia, sometimes so severe

that convulsions or coma ensue.

Such wide excursions in blood glucose concentration and

the accompanying alterations in metabolic status may provide

the basis for state-dependent retention. If so, performance

in school or in any area of life could be affected in subtle

ways. Experimental diabetes, induced in the laboratory, can

provide the basis for determining whether state dependent

retention occurs as a function of systematically altering

metabolic state, and neurochemical correlates can be

ascertained as well. The clinical relevance of such studies

is suggested in studies such as that by Holmes et al.,

(1983) in which altered glycemic state impaired cognitive

performances. In addition, experimental diabetes provides a

heuristic tool for the study of state dependent retention

with more naturally occurring forms of alteration of

internal state. Specifically, the loss of endogenous

insulin and its replacement with exogenous hormone to

restore normal blood glucose levels would allow modulation

of numerous metabolic and neurochemical parameters which

might underlie state dependency (Bartness & Rowland, 1983;

Lozovsky, Saller, & Kopin, 1981).

Hormonal modulation of memory processing

In addition to providing an internal milieu which may

serve as a contextual cue for remembering a training

episode, hormones may actually influence the mechanisms

involved in producing memory traces and in providing access

to such memories. Peripheral hormones of two major axes

have been implicated in learning and memory processes.

First, ACTH and glucocorticoids, hormones of the

hypothalamo-pituitary-adrenal (HPA) axis, have received a

great deal of attention for their potential role in

cognitive processes (McGaugh, 1983).

Hypophysectomy, which lowers plasma concentrations of

both ACTH and glucocorticoids, impaired learning of an pole-

jumping avoidance task (de Wied, 1964). Alternatively,

adrenalectomy, which elevates plasma levels of ACTH

facilitated acquisition of a two-way active avoidance

response, as did administration of ACTH. (Beatty, Beatty,

Bowman, & Gilchrist, 1970). In all cases, administration of

hormones to restore normal levels also restored performance

to control levels.

Administration of ACTH to intact animals also influenced

learning and memory. It delayed extinction of passive

(Levine & Jones, 1965) and active (de Wied, 1971) avoidance

responses and alley-running for food reward (Garrud, Gray, &

de Wied, 1974). Administration of ACTH prior to training

sessions led to an increase in the number of barpresses

during the latter period of the training session, as well as

during a subsequent extinction session (Guth, Levine, &

Seward, 1971). ACTH also attenuated carbon dioxide-induced

amnesia for a one-trial passive avoidance task (Rigter, van

Riezen, & de Wied, 1974). The effects of ACTH were often

noted when the hormone was administered immediately either

after training (Gold & van Buskirk, 1976) or prior to

retention testing (Rigter et al., 1974), suggesting that the

effects of exogenously administered ACTH might be related to

endogenous hormonal responses to training (Gold & Delanoy,


Thus, it was noted that following post-training

administration of ACTH, there were both dose-dependent and

time-dependent effects (Gold & Delanoy, 1981) upon


retention. Small doses enhanced, while larger doses

impaired, memory for a one-trial passive avoidance task.

Moreover, the effect of a given dose of ACTH interacted with

the footshock intensity utilized in training, suggesting

that hormone administration might be interacting with the

endogenous hormonal response to footshock.

The effectiveness of post-training ACTH also depended

upon administering the hormone within 10 minutes of the

training trial. Administration of the hormone more than 10

minutes after training was without effect upon retention.

This temporal gradient suggested the relationship of hormone

administration to normally occurring post-training hormone

release. Indeed, it was shown that footshock training

produced large increases in plasma ACTH (van Wimersma

Greidanus, Rees, Scott, Lowry, & de Wied, 1977). Moreover,

these elevations peak 5 minutes after stress and return to

baseline levels in 15 minutes. Interestingly, endogenous

ACTH levels increased even more upon re-exposure to the fear

cues associated with shock during retention testing than

they did to footshock itself. It is thus possible that

retention is to some degree related to the recurrence of

elevated hormone concentrations.

Glucocorticoids have been shown to have their own

effects upon learning, which are distinct from those of

ACTH. For example, post-training corticosterone injections

facilitated memory for a passive avoidance task in poorly


trained mice (Flood, Vidal, Bennett, Orme, Vasquez, &

Jarvik, 1978). Cycloheximide, which reduces plasma

corticosterone levels, attenuated memory for a passive

avoidance task in rats when given 30 minutes prior to

training (Cottrell & Nakajima, 1977). This attenuation

could be prevented if corticosterone was also administered

prior to training. Adrenalectomy facilitated extinction of

alley-running for food reward, and corticosterone

replacement normalized extinction (Micco, McEwen, & Shein,


The sympathoadrenal system has also been shown to

influence memory processing. Depletion of peripheral

norepinephrine (NE) impaired acquisition of a difficult

(two-way) active avoidance task (Oei & King, 1980).

Moreover, both epinephrine (EPI) and NE were elevated

immediately following a 3mA 2-second shock, which produced

good retention, but not following a 0.6mA shock, which

produced poor retention (McCarty & Gold, 1981). Thus, post-

training plasma levels of EPI and norepinephrine correlate

with retention performance.

Post-training injections of EPI and NE enhanced

retention for passive avoidance training (Gold & McGaugh,

1978), just as post-training ACTH did. It should be noted,

however, that ACTH administration did not elevate plasma NE

or EPI, nor did it affect the EPI response to footshock

(McCarty & Gold, 1981). Thus, these two sets of plasma

hormone responses may be independent, or they may interact

in the brain to modulate memory processing.

Glycemic state itself has been implicated in learning

and memory. For example, hypoglycemic rats chronically

treated with insulin (blood glucose=50 mg/dl) failed to

learn an appetitive T-maze task (Clayson, 1971). Insulin-

treated diabetic rats, which were essentially normoglycemic

(blood glucose=119 mg/dl) learned the task more quickly than

non-insulin-treated control animals. Unfortunately, no non-

insulin-treated diabetic group was included in this study.

Gold (in press) found that post-training subcutaneous

injection of 10 or 100 mg/kg of glucose enhanced retention

of a passive avoidance task. Because the timecourse and

dose-response of glucose modulation were similar to the

modulation produced by EPI administration, Gold (in press)

suggested that EPI-induced elevations of plasma glucose may

be involved in normally occurring memory storage.

Neurochemical Correlates of Learning and Memory

The alterations noted in peripheral hormone and glucose

concentrations following either training or testing episodes

are likely related to concurrent events within the central

nervous system related to the modulation of memory storage

and retrieval (Gold & McGaugh, 1978). While the mechanisms

underlying these processes are not yet fully understood, a

number of findings suggest that they entail neurohumoral


For example, the turnover of dopamine in the central

nucleus of the amygdala and the turnover of noradrenaline in

the hippocampal dentate gyrus were elevated in rats showing

good retention for passive avoidance (Kovacs, Versteeg, de

Kloet, & Bohus, 1981). Turnover was measured by

administering alpha-methyl-p-tyrosine and measuring

disappearance of either dopamine or norepinephrine 2 or 4

hours after treatment.

Reductions in forebrain and brainstem norepinephrine of

about 20%, measured 10 minutes after footshock training,

were indicative of memory storage required to later perform

a passive task (Gold & Murphy, 1980). Either larger or

smaller depletions were accompanied by attenuated retention.

Depletions of forebrain norepinephrine produced resistance

to extinction as well (Mason, Roberts, & Fibiger, 1979).

Concentrations of 3,4-dihydroxyphenylacetic acid (DOPAC)

have also been found to be affected by footshock training

(Herman et al., 1982). While footshock itself was

accompanied by elevations of DOPAC in frontal cortex,

olfactory tubercle, nucleus accumbens and amygdaloid

complex, returning the animals to the apparatus where they

had been shocked led to elevation of DOPAC only in the

anteromedial frontal cortex.

The hippocampus has also been implicated in learning and

memory, and its role seems to be related to interactions

with glucocorticoids (Micco et al., 1979). This system

seems particularly important in extinction, since

hippocampectomy was shown to retard extinction of alley

running for food, while adrenalectomy facilitated it.

Systemic corticosterone administration following

hippocampectomy did not restore extinction although hormone

replacement did normalize extinction after adrenalectomy.

Intrahippocampal hydrocortisone injection prevented amnesia

for passive avoidance in rats produced by cycloheximide

(Cottrell & Nakajima, 1977). Elevation of hippocampal

serotonin 24 hours after footshock training was found to be

associated with retrievability of the task (Ramaekers,

Rigter, & Leonard, 1978).

Hormonal and Neurochemical Alterations in Diabetes

Both human and experimental diabetes are accompanied by

a number of alterations in peripheral hormone

concentrations. In the clinical literature, there are

reports of alterations in both corticosteroid and

catecholamine function. For example, diabetic individuals

responded with a blunting of the normal suppression of

cortisol when given dexamethasone (Cameron, Kronfol, Greden,

& Carroll, 1984). In addition, plasma cortisol

concentrations in diabetics have been reported to be higher

than those of nondiabetics both in the early evening and in

the early morning (Lebinger, Saenger, Fukushima, Kream, Wu,

& Finkelstein, 1983). Asfeldt (1972) found normal day-night

rhythms and baseline concentrations of cortisol in plasma of


diabetics. However, elevations in plasma cortisol

concentration occurred in many diabetic individuals during

hyperglycemic states without acidosis, as well as during

perceived hypoglycemia, even when blood glucose

concentration actually was normal. Thus, the excursions in

blood glucose known to occur regularly in many diabetics

(Holmes et al., 1983) could be accompanied by cortisol


When challenged with infusions of cortisol and

epinephrine, diabetics experienced large increments in

plasma glucose concentration, which were sustained for

several hours (Shamoon, Hendler, & Sherwin, 1980). Controls

showed no increments at all in plasma glucose concentration

to cortisol infusion, and only transient increments to

epinephrine infusion. Glucose clearance was decreased by

cortisol and epinephrine infusions in both diabetics and

controls. The diabetics in this study were given continuous

insulin infusion during these hormone challenges, so that

the plasma glucose elevations occurred in spite of insulin


Twenty-four hour integrated plasma concentrations of

aldosterone, norepinephrine, epinephrine and growth hormone

were all found to be significantly higher in diabetics than

in nondiabetics (Zadik, Kayne, Kappy, Plotnick, & Kowarski,

1980). Diabetics in this experiment had received their

daily insulin. Plasma norepinephrine and epinephrine


measurements at a single time point have not demonstrated

differences between diabetics and nondiabetics (Christensen,

1974; Gustafson & Kalkhoff, 1981) as did integrated 24 hour

measures (Zadik et al., 1980). However, diabetics

experience excessive increments in plasma glucose as well as

in plasma epinephrine upon standing from a sitting position,

or during recovery from isometric exercises (Gustafson &

Kalkoff, 1981).

Animals with experimentally induced diabetes also have

have been shown to experience alterations in peripheral

concentrations of both corticosteroids and catecholamines.

Streptozotocin diabetic rats had elevated corticosterone

levels during the afternoon from about 1 P.M. to 6 P.M.

(with lights out at 8 P.M.), with a general flattening of

the normal day-night rhythm (Tornello, Coirini, & De Nicola,

1981b). In addition, residual pituitary ACTH activity under

resting conditions, as measured in a bioassay, was reduced

in diabetics relative to controls (De Nicola, Fridman, Del

Castillo, & Foglia, 1976, 1977). In response to the stress

of IP injection of cold water, residual ACTH activity was

reduced in the control rats but remained at the same level

in diabetic rats as during basal conditions (De Nicola et

al., 1977). Following withdrawal of insulin replacement in

alloxan diabetic rats, morning corticosterone concentrations

were elevated and the adrenocortical response to ACTH was

increased (L'Age, Langholz, Fechner, & Salzmann, 1974).


There has been considerably less investigation of

catecholamines in animal models of diabetes. Thus, little

has been published concerning sympathetic and

adrenomedullary function in the diabetic rat. However, a

recent report noted decreased norepinephrine turnover in

interscapular brown adipose tissue, heart and pancreas of

streptozotocin diabetic rats 9 weeks after diabetes

induction (Yoshida, Nishioka, Nakamura, & Kondo, 1985).

Another change noted in the streptozotocin-diabetic rat

was an increase in plasma dopamine-beta-hydroxylase (DBH)

activity (Berkowitz & Head, 1978; Berkowitz, Head, Joh, &

Hempstead, 1980; Schmidt, Geller, & Johnson, 1981). This

enzyme converts dopamine to norepinephrine, and thus is

related to synthesis of both norepinephrine and epinephrine.

The DBH elevations occurred within 24 hours of

streptozotocin administration (Schmidt et al., 1981) and

remained elevated when measured 7 months after diabetes

induction (Berkowitz et al., 1980). Plasma norepinephrine

was also found to be elevated in diabetic rats 4 weeks, but

not 16 weeks, after streptozotocin treatment. At 9 weeks,

diabetic rats had reduced plasma concentrations of both

norepinephrine and epinephrine despite the continuing

elevation in DBH (Berkowitz & Head, 1978). Clearly, changes

in peripheral catecholamine systems occur in the diabetic



A number of central nervous system (CNS) changes have

also been found in diabetic rats. For example,

streptozotocin diabetic rats had elevated NE concentrations

in hypothalamus and midbrain, decreased concentrations of

DOPAC and 5-hydroxyindole-acetic acid (5HIAA) in thalamus

and midbrain, increased alpha-adrenergic receptor

concentration (maximum binding capacity) in medial

hypothalamus and midbrain, and decreased Vmax of tyrosine

hydroxylase in hypothalamus, thalamus, medulla and midbrain

(Bitar, Koulu, Rapoport, & Linnoila, 1986). Brain serotonin

synthesis is also reduced in diabetic rats (Crandall &

Fernstrom, 1983). There are alterations in brain adenylate

cyclase activity, including reduced sensitivity to

stimulation by catecholamines, as well (Palmer, Wilson, &

Chronister, 1983).

Increased forebrain NE and decreased 3-methoxy, 4-

hydroxyphenylethylene glycol (MHPG) concentrations in

diabetic rats were normalized with insulin treatment

(Trulson & Himmel, 1985). Dopamine synthesis in striatum

and limbic forebrain is also decreased in diabetic rats, and

dopamine receptor number in striatum increased (Lozovsky et

al., 1981; Trulson & Himmel, 1983; Sailer, 1984; Serri,

Renier & Somma, 1985).

There was a decrease in the maximum number of binding

sites (Bmax) for [3H]-corticosterone in hypothalamus and

hippocampus of diabetic rats, while no changes in

corticosterone binding were noted in pituitary or cerebral

cortex (Tornello, Fridman, Weisenberg, Coirini, & De Nicola,


Clearly both humans and animals with diabetes experience

a number of changes in peripheral concentrations of both

steroids and catecholamines. In diabetic rats there are

also a large number of alterations in corticosteroid and

monoamine function in the central nervous system as well.

However, whereas there is a substantial clinical literature

concerning behavior/hormone interactions in diabetes, there

is almost no information about behavioral changes in

diabetic animals.

One behavioral deficit has been reported in the diabetic

rat. Stereotypy induced by dopaminergic agonists was

attenuated in the diabetic rat (Marshall, 1978; Bellush &

Rowland, in press). Both the duration and intensity of

stereotyped sniffing and oral behaviors induced by

peripheral injection of both apomorphine and amphetamine

were affected. It is quite possible that the attenuation

seen in the diabetic animals following amphetamine

administration is related to the reported decrease in

dopamine synthesis in limbic forebrain (Trulson & Himmel,

1983) and striatum (Lozovsky et al., 1981). However,

apomorphine-induced stereotypy would be expected to be

increased in diabetics, given the increased number of

dopamine receptors in these rats. Altered uptake of


apomorphine into brain (Saller, 1984) or alterations in

adenylate cyclase responsivity to catecholamine stimulation

Palmer et al., 1983) may account for this latter finding.

In any case, we were unable to normalize stereotyped

behavior in diabetic rats with dietary enrichment with

tyrosine, the amino acid precursor of dopamine, which

nonetheless increased brain tyrosine concentrations (Bellush

& Rowland, in press). While this attenuation of

apomorphine- and amphetamine-induced stereotypy in diabetic

rats is a heuristic tool for investigation of diabetes-

related behavioral changes, it has a limited applicability

to other behaviors because it is a pharmacologically induced

behavior. However, it does suggest that the neurochemical

alterations in diabetes may lead to functional consequences.


Given the numerous hormonal and neurochemical changes

which have been documented in diabetics, as well as the vast

clinical literature suggesting potential interactions of

hormones and behavior, it is somewhat surprising that no

investigations of cognitive performance in animal models of

diabetes have yet been reported. To our knowledge, the only

learning paradigm which included diabetic rats studied the

effects of insulin-induced hypoglycemia upon appetitive

learning and included an insulin treated diabetic group only

for the purpose of comparison, and not as a focus of the

study (Clayson, 1971). However, her results indicated that

insulin-treated diabetics learned an appetitive T-maze task

more quickly than untreated nondiabetic rats, while insulin-

treated diabetics failed to learn the task at all.

Motivational variables cannot be ruled out in this study,

since water deprivation was involved, and diabetic rats are

known to be hyperdipsic.

The following experiments were conducted to evaluate

possible changes in learning and memory in diabetic rats.

Because stress has been reported to be a major factor in

behavior/hormone interactions in the clinical literature,

the present studies utilized a stressful aversivee) learning

paradigm. This involved learning to avoid a place where

footshock punishment was delivered. Acute stresses

associated with a learning situation have been evaluated in

human diabetics (Kemmer et al., 1986).

In addition to the evaluation of a short-term stress,

experiments were conducted to compare the hormonal and

neurochemical effects of 72 hours of continuous exposure to

a cold environment (4C) in diabetic and nondiabetic rats.

The parallels to this chronic stress situation in the

clinical literature are adolescence and the period of

proliferation of retinopathy. Both of these periods

represent developmental periods when stress interacts for

extended lengths of time with metabolic and behavioral

variables in diabetes management. As with all

behavior/stress interactions in diabetes mellitus, it is

difficult to determine whether compliance problems are the

primary cause of hormonal changes and metabolic instability,

or whether some physiological factor first precipitates

behavioral problems. While no behavioral comparisons were

made in connection with chronic stress in the present

investigations, a number of physiological responses to cold

stress were compared in diabetic and nondiabetic rats. In

addition, some diabetic and nondiabetic rats were given

daily insulin injections so that it could be determined if

insulin treatment in diabetic rats would normalize any

altered responses found in untreated diabetics.



Chapter I outlined a large and diverse literature

concerning hormone/behavior interactions which occur both in

human diabetes and in experimentally induced diabetes in

animals. It was also emphasized that the hormones disturbed

by diabetes are the so-called stress hormones, and that

these hormones are importantly involved in the mechanisms of

memory formation and retrieval. Given these alterations in

diabetes mellitus, it is somewhat surprising that little

attention has been given to the investigation of learning

and memory in animal models of diabetes. The experiments in

this chapter were designed to investigate possible

alterations in either learning or memory in streptozotocin-

diabetic rats. They involved a comparison of retention for

one-trial passive avoidance training in streptozotocin-

diabetic and nondiabetic rats. An aversive task was chosen

because the stress hormones are more clearly involved in

such a task than in appetitive learning situations and

because motivation for food- or water-reinforced learning

might differ in diabetics, which are both hyperdipsic and


Before investigating learning and memory in the diabetic

rat, it was necessary to determine whether there would be

either motivational or perceptual differences between

diabetic and nondiabetic rats which would be mistaken for

associative differences. There is some basis for expecting

perceptual differences as a result of neuropathy which could

be present in diabetic rats (Jakobsen, 1978; Clements,

1979). Such neuropathy could result in reduced sensitivity

to footshock, the unconditioned stimulus to be used in

training, in the diabetic rats. Since retention of passive

avoidance is related to intensity of footshock (Ader,

Weijnen, & Moleman, 1972), diabetic rats might appear to be

impaired in their ability to learn the task when, in fact,

they simply perceive the stimulus as less intense.

Experiment 1 addressed this possibility by comparing flinch-

jump thresholds of diabetic rats to those of nondiabetic


Another potential problem in comparing diabetics to

nondiabetics involves possible differences in general

activity levels. Thus, utilization of metabolic substrates

for energy production is compromised in diabetic rats, and

they may therefore be less active than nondiabetics. Since

the learning task to be employed was a one-trial passive

avoidance task, inactivity could be mistaken for better


learning. This possibility was to be evaluated in the

learning paradigm itself by measurement of each animal's

latency to enter the learning chamber both prior to

footshock training and after footshock training. Still, the

issue of differential activity levels in diabetic and

nondiabetic rats can be more carefully determined by a

direct test. Experiment 2 evaluated diabetic and

nondiabetic rats in open field activity tests, which are

designed to reflect both activity and emotionality in a

novel environment (Veldhuis & deWied, 1984).

Experiment 1. Flinch-Jump Thresholds
of Diabetic and Nondiabetic Rats

A longstanding procedure for determining the sensitivity

(or responsivity) of rats to footshock is that of the

flinch-jump threshold test (Evans, 1961). It employs the

method of limits used in a number of psychophysical

measurements of sensory systems. Briefly, stimuli of

varying intensity are presented in alternating ascending and

descending series and the behavioral response is recorded.

If diabetic rats have a reduced capacity to respond to

footshock because of neuropathy, their flinch-jump

thresholds would be expected to be higher than those of

nondiabetics and it would be necessary to adjust the

footshock intensity used in passive avoidance training in

order to equate the responsivity of diabetics with that of


Another factor of concern in the learning situation is

the effect of glycemic state itself (Gold, in press). In

testing for such effects in the passive avoidance task, some

rats were to be given insulin during either training or test

sessions. Insulinization might influence the sensitivity of

diabetics or controls, or both, to footshock. Thus, in the

flinch-jump experiment, groups of insulin-treated diabetic

and control rats were included.

A final consideration in attempting to equate the

responsivity of diabetic and nondiabetic rats was the

disparity in body weight, often 100-200g, between diabetic

rats and age- and treatment-matched controls. Marks and

Hobbs (1972) found heavier rats to be significantly less

responsive to footshock than lighter animals (but see also

Beatty & Beatty, 1970). To rule out such a problem, a group

of nondiabetic rats which weighed the same as diabetic rats

was tested.


Animals and housing

All rats were adult male Sprague Dawley albino rats

(Zivic Miller, Pittsburgh). They weighed 350-425 g at the

beginning of the experiment. They were housed individually

in hanging wire cages in a vivarium maintained at 21+20C and

an artificial 12:12 hour light/dark cycle (lights on at 7:00

A.M.). Purina Laboratory Chow (Purina #5001) and water were

available ad libitum at all times unless otherwise stated.

Induction of diabetes

Diabetes was established in 20 rats by intraperitoneal

(IP) administration of 65 mg/kg streptozotocin (STP) in

citrate buffer, 0.1M, pH=4.5. Twenty additional rats were

given the citrate buffer. The remaining 9 rats were

untreated and were tested when they reached the same body

weight as diabetic rats (which typically weigh 100-200 g

less than their age- and treatment-matched controls).

Diabetes was confirmed 1 week after STP treatment by testing

the urine for glucose with Ketodiastix (Ames division of

Miles Laboratories). STP-treated rats which did not show

any glucose in the urine were reinjected at this time. Only

a few rats needed to be reinjected, and all the rats treated

with STP became severely diabetic within 2-3 weeks (blood

glucose concentrations > 400 mg/dl). One month after STP

administration, insulin replacement was begun in 10 diabetic

and 10 control rats. This consisted of 3 days of twice

daily (8:00 A.M. and 5:00 P.M.) subcutaneous (sc) injections

of 1 UNIT (U) protamine zinc insulin (PZI) followed by 7

days of twice daily injections of 2.5 U PZI.


Testing was conducted in an 8 in x 9 in x 8 in plexiglas

chamber (This was actually one side of a 2 compartment

shuttle box, the other half of which was not used in this

experiment). It had a grid floor of steel rods spaced 7/16

in apart. The grids were wired to a constant current


shocker scrambler (Lehigh Valley Electronics) and a 28 V

power supply (Lambda). A universal timer (Lehigh Valley

Electronics) controlled the duration of the shocks. A 25 W

light bulb was suspended 18 in above the chamber, providing

the only light in the room during testing.


All testing was conducted between 1:00 P.M. and 5:00

P.M. A rat was placed into the chamber and allowed 30 sec

to explore and to adapt to the new environment. At that

time, 10 series of 0.5 sec shocks were administered in

alternating ascending and descending sequence. The current

levels (in mA) ranged from 0.02 to 0.1 mA in 0.02 mA steps,

and from 0.1 mA to 1.OmA in 0.1 mA steps for a total of 13

shocks in each series. Two observers independently rated

the response of the rat to each shock as follows (Mactutus &

Tilson, 1984): O--No response; 1--Flinch: contraction of any

part of the body; 2--Flinch-shuffle: flinch plus movement of

one paw off the grid; 3--Shuffle: movement of two or three

paws off the grid; 4--Shuffle-jump: movement of paws plus

animal startles; 5-Jump: all paws leave grid simultaneously

or animal moves position at offset of shock. All the scores

of both observers were tabulated to produce a frequency of

occurrence of each behavioral response category for each

rat. Responses 2 (flinch-shuffle) and 4 (shuffle-jump) were

distributed equally between the categories to either side,

(e.g. responses 2 were equally distributed to response

categories 1 and 3). Flinch, shuffle and jump thresholds

were defined as the lowest current level (interpolated) to

which the response occurred on 50% of the trials. Animals

not reaching the 50% criterion for jump threshold at 1.0 mA

were assigned a threshold of 1.1 mA. The 3 thresholds were

determined for each subject. Median threshold scores were

then tabulated for each group--controls, insulin-treated

controls, diabetics, insulin-treated diabetics, and weight

matched controls. Flinch, shuffle and jump thresholds for

the 5 groups were compared with Kruskal Wallis analysis of

variance by ranks.

At the end of the flinch-jump testing session, blood was

taken from each rat by making a small nick in the tail with

a scalpel blade and stroking the tail to collect blood in

heparinized micro-hematocrit capillary tubes (Fisher

Scientific). Plasma glucose concentration was determined,

after centrifugation, with a Yellow Springs Instruments 23A

automated glucose analyzer. Mean plasma glucose

concentrations and mean body weights of the 4 age- and

treatment-matched groups were compared with analysis of

variance (ANOVA). Mean body weight of the body weight

control group was compared with that of saline treated

diabetics with a t-test.

Results and Discussion

Figure 1 shows the median threshold scores for the 5

groups. There were no significant differences among any of


0 .l .-
'0 4-

*- (10 0
0 *0

-- .1 4-
Xa (L(
-- -0
o 0 0)


*- .1

E 00

.j r 01
,rn (0 W

r- (

44 4-0

Q) *4-

ra (o

*4- 40 r-I
lO (0

033 OC
4 .0 0a
0a 0I 00

I rU

0 O!

D- r 4- 4 4 0
p 14 010

-H 44 *0 C
S(0 M 0






z c

Z 'c=

0 CO

0 w 0 o o


0 Cq Co Cu

(0 iu) A 0isualui > 0ois


the groups on any of the thresholds (Kruskall-Wallis H

values were flinch-1.7; shuffle-3.92; jump-7.62; Ps>O.10).

Figures 2 and 3 show mean body weights and mean blood

glucose concentrations respectively of the 5 groups. Body

weights varied significantly with main effects of diabetes

(F(1,36)=118.12, P<0.001) and of insulin replacement

(F(1,36)=15.91, P<0.001), as well as an interaction

(F(1,36)=4.84, P<0.05). Thus, the diabetic rats weighed

significantly less than did the nondiabetic rats, and

insulin replacement led to significantly higher body weight

in diabetic rats but did not affect the weight of

nondiabetics. As intended, the weight of the younger body

weight control group did not differ significantly from that

of the saline-treated diabetics (t(17)=1.77, P>0.05).

Plasma glucose concentrations also varied significantly with

main effects of diabetes F(1,36)=172.55, P<0.001) and of

insulin replacement (F(1,36)=203.75, P<0.001), as well as an

interaction (F(1,36)=142.22, P<0.001). That is, insulin-

treated diabetic and nondiabetic rats had significantly

lower blood glucose concentrations than did their saline-

treated counterparts. Insulin-treated diabetic rats had

blood glucose concentrations indistinguishable from those of

nondiabetic rats, while the saline-treated diabetic rats had

significantly higher blood glucose concentrations.

No differences occurred in any of the thresholds

measured in experiment 1 on the basis of diabetes, insulin

Figure 2. Mean body weights of of nondiabetic and diabetic
groups at the time of footshock threshold determinations.
(sal=saline-treated, ins=insulin-treated, bwt=body weight

sal ins bwt sal ins

Nondiabetics Diabetics

Figure 3. Mean plasma glucose concentrations of nondiabetic
and diabetic groups measured immediately following footshock
threshold determinations (sal=saline-treated, ins=insulin-
treated, bst=body weight controls).






sal ins bwt

sal ins


administration or body weight. In contrast to the findings

of Marks and Hobbs (1972), the present study did not detect

a difference in sensitivity to footshock in lighter versus

heavier control rats. Diabetics would therefore not be

expected to differ from controls in sensitivity to footshock

on the basis of differences in body weight. Further, the

fact that the diabetics did not differ from controls

suggests that any neuropathy existing in the diabetics had

no measurable effect upon responsivity to footshock.

Finally, in spite of a wide range of plasma glucose

concentrations from hypoglycemic to hyperglycemic ranges, no

difference in thresholds to footshock occurred. These

results made it possible to proceed with the learning

experiments with the assurance that any differences in

passive avoidance responding which might be found in

diabetic versus nondiabetic rats would not be the result of

differential responsivity to footshock, but could

confidently be ascribed to the effects of training.

Experiment 2. Open-Field Activity of
Diabetic and Nondiabetic Rats

Differences in either emotionality or activity level in

diabetic and nondiabetic rats when placed in a novel

environment might also lead to differences in performance of

passive avoidance unrelated to associative processes. The

passive avoidance paradigm to be used later was planned to

include measures of latency to enter the shock chamber prior

to footshock training; but it was necessary to evaluate

diabetic animals and determine prior to employing the

passive avoidance task whether differences in levels of

activity in diabetic and nondiabetic rats might produce

spurious differences in passive avoidance performance. To

evaluate potential differences in general activity levels of

diabetic and nondiabetic rats, an open field arena was

employed (Britton, Koob, Rivier, & Vale, 1982).


Subjects. Subjects were 50 adult male Sprague Dawley

rats, 27 of which had been made diabetic with streptozotocin

(STP), exactly as in Experiment 1. The remaining 23 rats

had been injected with vehicle. Beginning thirty days after

STP or vehicle injections, 14 diabetic and 6 control rats

were treated with insulin for 10 days, while the remaining

rats received injections of the vehicle as described in

Experiment 1.

Apparatus. The open field was a square enclosure,

consisting of 4 plywood walls measuring 27 in long by 14 in

high held together with L-braces and open at both top and

bottom. This open cube was placed on a white table upon

which were painted a black outline measuring 27 in on each

side and 9, 9-in black squares. Indirect lighting was

provided by an overhead light in another part of the room.

Procedure. The animals tested in the open field were

those to be trained in passive avoidance. Two days prior to

passive avoidance pretraining, each animal was brought from


the vivarium in its home cage and placed in the center

square of the open field. A running timer was activated and

the animals were observed for 3 minutes by two observers.

One observer counted and recorded the number of squares the

animal entered. The other observer recorded the number of

rears (i.e. raising up on hindlimbs), the number of grooming

episodes and the number of fecal boli. After each subject

was tested, the field was cleaned with 70% alcohol. Because

there were disparate numbers of subjects in the various

groups (17 insulin-treated controls, 6 saline-treated

controls, 7 insulin-treated diabetics and 7 saline-treated

diabetics), median scores were determined for each

behavioral category for each of the 4 groups. Kruskal-

Wallis analysis by ranks was used to evaluate the median


Results and Discussion

Results of the open field activity measurements are

shown in Figure 4. There were no significant differences

among the groups in any of the behavioral categories (Ps>

0.10). In these measures of activity and emotionality upon

exposure to a novel environment, there were no differences

between diabetics and nondiabetic controls. This finding in

the open-field test was in agreement with a previous finding

in this laboratory (Rowland, Joyce, & Bellush, 1985) using a

photocell activity-monitoring cage. Under these conditions,

diabetics were indistinguishable from controls during the

Figure 4. Median number of grooming episodes, boli
excreted, squares entered and rears in saline-treated and
insulin-treated nondiabetic and diabetic rats
(NS=nondiabetic saline-treated, NI=nondiabetic insulin-
treated, DS=diabetic saline-treated, DI=diabetic insulin-

w 4-

c o
E 2-



S20- .


w 15

, 10- n -




initial 5 minute exposure to the environment, but became

less active than controls during the latter portion of the

30 minute session. Both the photocell activity monitoring

results and the present results in the open field argue

against any activity or emotionality differences between

diabetic and nondiabetic rats in short term measurements.

Thus, there should not be a problem with inactivity in the

passive avoidance procedure, which entails evaluation of

behavior for a total of 10 minutes. Indeed, animals failing

to move from a lighted chamber into a darkened chamber

(where shock would later be delivered) in the first minute

of exposure to the shuttle box before footshock training was

given would be excluded from further testing. The results

of the present study and of the previous study in this

laboratory indicate that systematic activity differences in

diabetic rats would be very unlikely to interfere with the

evaluation of passive avoidance learning.



The clinical literature cited in Chapter I indicated

that a number of cognitive impairments may accompany

diabetes mellitus. In addition, a number of reports were

cited which documented changes in metabolic response to

catecholamine or cortisol challenge (Shamoon et al., 1980)

and elevated cortisol concentration in plasma of some

diabetics at certain times (Lebinger et al., 1983), although

the relationship of these hormonal alterations to cognitive

and psychological changes found in some diabetic individuals

remains to be clarified. It has been difficult to determine

whether cognitive and emotional alterations reported in

diabetics are in any way the result of hormonal and

metabolic changes, or whether the behavioral phenomena are

psychologically based. Negative attitudes toward the

restrictive lifestyle imposed by diabetes could evoke both

behavioral problems and problems of compliance which would

secondarily cause hormonal derangement. Such complexities

can be avoided by turning to animal models of diabetes for

further clarification of the relationship between hormonal


changes occurring in diabetes and changes in cognitive


In experimentally induced diabetes in rats, there are

alterations in concentrations of corticosterone in plasma,

both in basal and in stimulated states, and elevations in

plasma concentrations of DBH and norepinephrine, as well as

numerous functional changes in central nervous system (CNS)

monoamine systems. Moreover, the importance of both

peripheral and central actions of hormones of the

hypothalamo-pituitary-adrenal system and of the

adrenomedullary system in learning and memory in animals

have been demonstrated convincingly (Dunn & Kramarcy, 1984).

What remain to be investigated are the possible changes

in learning and memory in animal models of diabetes, given

that the hormonal systems affected by diabetes are those

clearly involved in processing of memory. The following

experiments deal with the evaluation of learning and memory,

as well as some related hormonal and neurochemical

measurements, in the streptozotocin diabetic rat.

The experimental learning paradigm selected for these

studies is one-trial passive avoidance, in which learning is

restricted to a specific point in time. It thus provides an

opportunity to investigate some of the physiological events

which accompany either the training episode or the retention

test. Specifically, these experiments compared retention

for a one-trial passive avoidance task in diabetic and

nondiabetic rats. In addition, plasma corticosterone

concentrations, which are thought to reflect stressful

stimuli (Bassett & Cairncross, 1975), as well as influence

learning and memory (Cottrell & Nakajima, 1977), were

measured following training and testing sessions, as were

regional CNS monoamine concentrations following the

retention test. Twenty-four hour urinary catecholamine

excretion, a measure of peripheral catecholamine activity

(Roy, Sellers, Flattery, & Sellers, 1977), was measured

during a baseline period and following footshock.

Experiment 3. Retention for Passive Avoidance Training
in Diabetic and Nondiabetic Rats

This experiment was divided into two separate

manipulations. In Experiment 3a, state-dependent retention,

based upon chronic changes in glycemic state, was evaluated.

Chronic (10 day) administration of insulin was utilized

because it corresponds with the duration of replacement

reported to normalize diabetes-induced increases in striatal

dopamine receptor number (Lozovsky et al., 1981; Serri et

al., 1985; Trulson & Himmel, 1983).

If state-dependent effects were related in any way to

known alterations in neurochemistry in the diabetic, such as

monoamine synthesis or change in receptor number, an insulin

replacement regimen which has been shown to normalize such

changes might also optimize any state dependent phenomena.

State-dependent effects were evaluated in both diabetic and

nondiabetic rats. The glycemic changes produced by insulin


are much more dramatic in diabetics, and insulin replacement

identical to that in diabetic rats did not alter dopamine

receptor number in striata of nondiabetic rats (Lozovsky et

al., 1981). However, insulin administration affects many

systems, any of which may be involved in state-dependency.

Thus, insulin may support state dependent retention in both

diabetic and nondiabetic rats. Alternatively, if diabetics

were subject to state-dependent effects while nondiabetics

were not, then the central dopaminergic alterations and

large excursions of plasma glucose concentration of diabetes

would be implicated in these behavioral effects. The

inclusion of nondiabetic rats along with diabetic rats in

this experiment thus would help to clarify the mechanisms

underlying state-dependency.

As noted in the introduction there was a report of

impaired cognitive performance in diabetics at high and low

plasma glucose concentrations (Holmes et al., 1983), while

another report demonstrated superior performance by

diabetics when either hyperglycemic or normoglycemic

(Flender & Lifshitz, 1976). In a study with rats,

hypoglycemia prevented the achievement of performance to

criterion in an appetitive T-maze task (Clayson, 1971). In

this study, insulin-treated diabetic rats learned the task

more quickly than normoglycemic control rats. Thus, results

of previous studies are conflicting as to whether diabetes

impairs or enhances performance. Indeed, different tasks


may lead to different results. The present experiment was

designed to examine retention for a relatively simple, one-

trial aversive task. One objective was to determine whether

retention for the task would be different if the glycemic

state during testing was the same as, or different from, the

glycemic state during training.

In addition to producing state-dependent effects upon

retention, diabetes could influence memory storage and

retrieval mechanisms as a result of producing alterations in

peripheral and central catecholamine and steroid activity

outlined in Chapter 1. In Experiment 3b, this possibility

was investigated. A method commonly used to evaluate

hormone-behavior interactions is that of assessing both

behavioral performance and the associated hormonal and

neurochemical changes, at different levels of footshock

(e.g., McCarty & Gold, 1981). In the present experiment,

separate groups of diabetic and nondiabetic rats were given

1- and 2-second shocks and compared for retention, as well

as for a number of physiological variables reported to be

sensitive to aversive learning. Longer durations rather

than higher intensity shocks were employed, because the

passive avoidance task involved 1.0 mA shocks throughout the

state-dependent experiment. This shock intensity was the

maximal intensity evaluated in flinch-jump testing in

Experiment 1. There was no way to assure that shocks of

higher current intensity would affect diabetics and


nondiabetics equally. However, it has been shown that

retention improves as a function of duration of shock as

well as its intensity (Ader et al., 1972).

Experiment 3a. State-Dependent Effects of Glycemic State


Subjects. Subjects were 116 adult male Sprague Dawley

rats weighing 250-400 g at the beginning of the experiments.

They were housed singly in hanging wire cages in a vivarium

maintained at 21+20C with a 12:12 hour light:dark cycle

(lights on at 7:00 A.M.). Purina lab chow and water were

freely available at all times.

Induction of diabetes. Diabetes was induced in half the

rats with administration of streptozotocin exactly as

described in Experiment 1, while the remaining rats were

given vehicle.

Insulin replacement. Thirty days after induction of

diabetes, rats were assigned to groups for the passive

avoidance task. There were 4 groups of control rats: one

group was to receive saline throughout all training and

testing (SS); the second group (SI) was to receive saline

for 10 days prior to passive avoidance training. On the

evening of the training session, these rats began receiving

insulin for 10 days. The first 3 days they received 1 UNIT

(U) of protamine zinc insulin (PZI) twice daily and the next

7 days they received 2.5 U PZI twice daily. The third group

(IS) received insulin as just described for 10 days prior to

training and were reversed to saline the evening of

training. The fourth group of controls was a departure from

the usual symmetrical design in state dependent learning in

which they would have received insulin throughout training

and testing. It was decided instead to train this group of

rats (SSTP) while normoglycemic and inject them during the

evening of training with STP so that they were hyperglycemic

at testing 10 days later.

Diabetic groups included SS, SI and IS identical to the

control groups with the fourth diabetic group (II) receiving

insulin throughout. Thus, in the diabetics a symmetrical

state-dependent design was employed.

Criteria for defining glycemic states were established

as follows. In diabetics the insulin treatment was required

to provide plasma glucose concentrations of less than 200

mg/dl during the training and/or testing procedure. Any

insulin-treated diabetic rats with concentrations of glucose

higher than this value were eliminated from the study. In

the controls, it was required only that plasma glucose

concentrations of the animals in the SI and IS groups be

lower during the insulin treatment than when saline was


Apparatus. Passive avoidance training was conducted in

a clear plexiglas two-chamber shuttle box. Both chambers

measured 8 in x 9 in x 8 in high. The clear side walls of

one chamber were covered with white paper, while those of

the other were covered with black paper. The grid floor of

the white chamber was covered with a clear plexiglas sheet.

A clear plexiglas lid was placed over the white chamber when

the rat was inside the shuttle box, and a 25 W bulb was

suspended 18 in above this chamber. A mirror was placed at

an angle above this chamber so that the activity of the rats

could be observed while the experimenter remained unnoticed.

A plywood lid was placed on top of the dark chamber. The

grid floor of the dark chamber was wired to a constant

current shocker-scramber as described in Experiment 1. The

two chambers were separated by a guillotine door which was

operated manually by the experimenter. The differences

between the two shuttle box chambers served two purposes.

First, they maximized the distinctiveness of the dark

chamber, where the rat was to be shocked, from the white

chamber, which was "safe." Second, the white, or "safe,"

chamber was designed to be brighter, and thus initially less

preferred by, or even slightly aversive to, the rats.

Footshock training was therefore designed to overcome a

natural aversion to the lighted chamber.

Procedures. Due to the large number of animals involved

in Experiment 3, the study was conducted in four separate

replications, in each of which animals from all the groups

were represented. In addition, in each of the replications,

the rats were divided into squads, each containing equal

numbers of diabetic and nondiabetic rats. All training and


testing sessions were conducted in the afternoon, between

1:00 and 5:00 P.M. After the initial ten days of

administration of either insulin or saline, each rat was

given a pretraining trial which consisted of a 10-minute

exposure to the shuttle box. Briefly, the rat was placed

into the white chamber, facing away from the guillotine

door, and a 10-minute timer was activated. Ten seconds

later, the guillotine door was opened and a hand-held

stopwatch was activated. The latency of the rat to enter

the dark chamber was recorded on the stopwatch. During the

remainder of the 10-minute session, the stopwatch was

activated whenever the rat was in the white chamber, and the

total time spent in this chamber (TTW) was also recorded.

Twenty-four hours after pretraining, the rats were again

placed into the white chamber of the shuttle box, the

guillotine door was opened and the stopwatch activated.

When the rat entered the dark chamber, latency to enter was

again noted. The door was then closed, confining the rat to

the dark chamber, and four 1-second, 1 mA footshocks were

delivered at approximately 10-second intervals, beginning 5

seconds after the door was closed. Within 5 seconds after

the last shock, the rat was removed from the dark chamber of

the apparatus and returned to the vivarium. Exactly 15

minutes later, a blood sample was collected from a tail nick

into heparinized micro-hematocrit capillary tubes,

centrifuged for three minutes and the plasma separated and

frozen until determinations of plasma glucose and

corticosterone concentrations were made. On the evening

after this training, those animals whose injections were to

be reversed were started on the new regime. Those control

rats who were to receive STP were given these IP injections

and then continued on twice daily saline.

Ten days after the training trial, each rat was again

placed into the white chamber of the shuttle box, and a 10

minute session identical to the pretraining trial was

conducted, with latency to enter the dark chamber and total

time in the white chamber once again recorded. At the end

of the 10 minute retention trial the rat was placed in a

holding cage in the training room for an additional 5

minutes to allow a total of 15 minutes to elapse after

exposure to the fear cues induced by the reexposure to the

shuttle box.

At this time the rat was taken to another room and

sacrificed by decapitation (the rats could not be

anesthesized in any way prior to decapitation, because this

would have interfered with many of the biochemical

measurements to be made). Trunk blood was collected in

heparinized 15 ml tubes and kept on ice until it was

centrifuged for 10 minutes. Plasma was collected and frozen

for later determination of plasma glucose and corticosterone

concentrations. The brain was quickly removed and placed on

ice, and the following regions were dissected and frozen at


-700C for later determinations of monoamine and metabolite

concentrations: frontal cortex, hypothalamus, amygdala,

hippocampus and brainstem. Plasma corticosterone

concentrations and neurochemical analyses will be discussed

in Experiments 4 and 6 below.

Data analysis. Body weights and plasma glucose

concentrations were compared with one-way analysis of

variance (ANOVA) and Duncan's post hoc evaluations of

individual differences. Pretraining and training latencies

for each rat were averaged and these average latencies were

considered training scores. Latencies and TTWs were tested

with Kruskal-Wallis one-way nonparametric analysis by ranks.

Results and discussion

Tables 1 and 2 show plasma glucose concentrations and

body weights of all the groups both at the time of training

and at the time of retention testing. As expected,

diabetics not receiving insulin had plasma glucose

concentrations well over 400 mg/dl, while insulin-treated

groups had concentrations in the normal range. Plasma

glucose concentrations of insulin-treated nondiabetic rats

were about half that of non-insulinized groups, and STP

treatment produced increments to severe hyperglycemic values

(SS2 groups will be described in Experiment 3b, below).

Body weights also followed expected patterns, with

diabetics weighing substantially less than nondiabetics.

Insulin replacement led to normalized weight gain in


Plasma Glucose Concentrations of Rats at
the Time of Passive Avoidance Training
and Retention Testing

Group Plasma Glucose (mg/dl)
Training Testing

Nondiabetics SS 131+4 146+3
SI 142+5 76+7+
IS 62+7+ 142+6
SSTP 135+2 449+27
SS2 141+5 151+3

Diabetics SS 568+29* 554+29*
SI 630+53* 93+12
IS 119+13 470+21+
II 93+11 90+9
SS2 541+17* 549+29*

Shown are X + SEM
* Significantly different from IS and II
+ Significantly different from all other groups


Body Weights of Rats at the Time of
Passive Avoidance Training and Retention Testing






Shown are X + SEM
* Significantly different
+ Significantly different

Bodyweight (g)
Training Testing





from IS and II
from all other groups


diabetics but produced no apparent effects in nondiabetics.

The SSTP group lost weight between training and testing.

This was an expected effect of the STP treatment.

The behavioral results are shown in Table 3. Kruskal-

Wallis analysis by ranks comparing training and test

latencies and total time spent in the white chamber (TTW)

revealed no significant differences among groups in either

the diabetic or nondiabetic state-dependent retention

manipulations (Ps>0.1). Thus, despite large changes in

blood glucose concentrations from training to testing in

several of the groups, there was no decrement in retention

of the footshock training in these groups relative to groups

whose glycemic state was not changed.

It was somewhat surprising that dramatic changes in

glycemic state, which were likely accompanied by numerous

other metabolic changes as well, failed to provide the basis

for state-dependent retention of the passive avoidance task.

In any case, on the basis of these negative results it might

therefore be concluded that the notion of state-dependency

is of little explanatory value.

However, two factors may have interfered with the

establishment of state dependency in the present study.

First, it was established in a pilot experiment that in

order to obtain robust retention in untreated rats over a 10

day period, it was necessary to deliver four footshocks in

the training session. It is, however, possible that this


Latencies to Enter the Dark Chamber and Total Time
Spent in the White Chamber (TTW) of State Dependent
Nondiabetic and Diabetic Groups

Group Pre-Training Post-Training
Latency TTW Latency TTW

Nondiabetics (sec) (se) (sec) (sec)
SS 7.0 34.7 114.2 302.9
SI 8.4 101.2 438.0 569.9
IS 10.3 46.8 166.3 367.5
SSTP 8.6 109.9 481.4 580.9

SS 16.6 134.7 409.6 495.8
SI 10.0 109.6 250.7 311.4
IS 13.3 47.3 600.0 600.0
II 8.4 69.4 439.0 563.9

Shown are the median scores

amount of shock provided strong enough learning to obscure

state-dependent effects. It is known that observation of

state-dependency requires provision of a weak learning event

(Overton, 1984). Our footshock training may have been too

intense to allow the emergence of state-dependent effects.

It should be noted, however, that the present results did

not reflect ceiling effects.

Another possible problem in the present experiment was

the chronic exposure to the glycemic states. The

traditional state-dependent paradigm establishes the

training and tests "states" only during the time of the

learning and retention trials (Overton, 1984). Such a

procedure allows the internal state to become a contextual

cue associated specifically with the learning events.

Longer exposure to insulin or to hyperglycemia in our

procedures may have decreased the effectiveness of metabolic

state in providing cues specific to the learning situation.

However, to the extent that we were modeling the

situation as experienced by diabetic humans, they too, have

chronic fluctuations in glycemic state whereby hyperglycemia

and hypoglycemia are experienced at length and likely with

some frequency as well. Thus, these altered glycemic states

may not have a major impact upon learning.

Experiment 3b. Comparison of Retention for Passive
Avoidance in Diabetic and Nondiabetic Rats After 1 Second or
2 Second Shocks


Subjects. Subjects for this experiment included the 26

rats from the diabetic and nondiabetic SS groups in

Experiment 3a. They received training footshocks of 1-

second duration, as previously described. Twelve additional

diabetic rats and 12 additional nondiabetic rats were also

included in the present manipulation. These rats (SS2)

received training footshocks of 2-seconds duration. They

were housed and maintained exactly as described in

Experiment 3a, and given saline injections twice daily

throughout the experiment. The additional 24 rats included

here were divided among replications 2-4 as described in

Experiment 3a, and trained and tested concurrently with the

other groups of rats.

Apparatus. The apparatus was the same apparatus

described in Experiment 3a.

Procedure. This study was conducted concurrently with

Experiment 3a. The only additional detail for this

experiment was the inclusion of the 2 additional groups

(SS2), one composed of diabetic rats and the other of

nondiabetic rats. In all other aspects, the experimental

procedure for this experiment was identical to that of

Experiment 3a. Plasma glucose concentrations and body

weights of these groups were compared with ANOVAs along with

those of the other groups in Experiment 3a. Latencies and

TTWs of the 4 groups in this experiment were compared using

separate Kruskal-Wallis analysis by ranks, with differences

among the individual groups determined with Mann Whitney U

tests (Siegel, 1956).

Results and discussion

Plasma glucose concentrations and body weights of the

rats in this experiment are shown in Tables 1 and 2

respectively. The 2 second shock groups (SS2) did not

differ in blood glucose concentration or body weight from

their 1 second shock counterparts (SS groups).

Behavioral results of Experiment 3b are shown in Figures

5-8. Latency data from pre-training and post-training

sessions are shown in Figure 5. Figure 6 shows individual

post-training latencies of rats in the 4 groups. There was

a significant difference in post-training latencies among

the groups as tested in the Kruskal-Wallis analysis

(H=16.16, P<0.01). Mann Whitney U tests indicated that the

diabetic (D) group given the 2-second shocks had a

significantly higher median latency than all the other

groups, while the D group given 1-second shocks had a

significantly higher median latency score than its 1-second

nondiabetic (N) counterpart. (Ps<0.05). The elevations

noted in the N group given 2-second shocks relative to the N

group given 1-second shocks approached significance

(P=0.06). Pre-training latencies were significantly

Figure 5. Median latencies of nondiabetic (N) and diabetic
(D) rats to enter the dark chamber of a shuttle box before
(textured bars) and after (clear bars) footshock training.
*Significantly different from all other groups.
+Significantly different from the 1-second N group.

1-sec shocks 2-sec shocks

Figure 6. Median latencies of nondiabetic (N) and diabetic
(D) rats to re-enter the dark chamber of a shuttle box where
they had received footshock 10 days earlier with individual
latencies indicated by solid circles.


1-sec shocks 2-sec shocks

Figure 7. Median total time spent by nondiabetic (N) and
diabetic (D) rats in the white chamber of a shuttle box
before (textured bars) and after footshock training (clear
bars). +Significantly different from the 1-second N group.





1-sec shocks

2-sec shocks








Figure 8. Median total time spent by nondiabetic (N) and
diabetic (D) rats in the white chamber of a shuttle box
during re-exposure to the box 10 days after receiving
footshock in the dark chamber with individual scores
indicated by solid circles.

a I a i & i -

1-sec shocks

2-sec shocks








different among the groups (H=10.8, P<0.05). However, all

pretraining median latencies were less than 20 seconds.

Moreover, when difference scores were calculated by

subtracting pre-training latencies from post-training values

and submitted to two-way ANOVA, diabetics showed

significantly longer adjusted latencies than did

nondiabetics at both shock durations (effect of diabetes,

F(1,49)=12.68, P<0.001; effect of shock duration,

F(1,49)=8.75, P<0.005).

There were also significant differences among groups in

the total time spent in the white chamber (TTW) during the

retention test session (H=15.28, P<0.01). Pretraining TTW

scores were not significantly different (H=3.02, P=0.404).

Median pre-training and post-training TTW scores are shown

in Figure 7. Figure 8 shows individual post-training TTW

scores. The D group given 2-second shocks spent

significantly more time on white after training than did all

other groups, and the N group given 2-second shocks had a

higher TTW score than the N group given 1-second shocks

(P<0.05). The difference between the two groups given 1-

second shocks (N and D) did not reach significance (P=0.09).

Adjusted TTW scores (post-training scores minus pre-training

scores) indicated an effect of diabetes which just missed

the usual accepted alpha level of 0.05 (F(1,49)=3.77,

P=0.0583). There was a significant effect of shock duration

on these adjusted TTW scores (F(1,49)=12.80, P<0.001).


These data demonstrated better retention for the passive

avoidance task than in diabetic rats than in nondiabetic

rats. Thus, while all the rats entered the dark chamber

with very short latencies prior to footshock training, and

spent little time in the white chamber, the four training

footshocks produced significantly longer latencies to re-

enter the dark chamber in diabetics than in nondiabetics.

Moreover, the diabetic rats spent significantly more time in

the white chamber than did the nondiabetics, despite the

initial preference for the dark chamber by all the rats.

These retention differences may be related to known

biochemical differences in diabetic and nondiabetic rats.

For example, diabetics have been shown to have elevated

resting corticosterone concentrations, as well as higher

peak increments following stress (L'Age et al., 1974;

Tornello et al., 1981b). Norepinephrine turnover is reduced

in interscapular brown adipose tissue, heart and pancreas of

streptozotocin diabetic rats (Yoshida et al., 1985), while

concentrations of DBH and norepinephrine in plasma are

elevated (Berkowitz et al., 1980). Thus, diabetic rats may

enter a learning situation in a very different basal

hormonal state than nondiabetic rats. Such differences are

known to influence subsequent responses to the training

episode (Leshner, 1975).

Differential hormonal states have been demonstrated to

be relevant to learning. It was shown that artificially


elevating post-training plasma hormone concentrations by

administration of either ACTH or epinephrine facilitated

retention for weak footshock training (Gold & McGaugh,

1978). Post-training administration of higher doses of ACTH

or epinephrine actually impaired retention. These treatments

were compared to the effect of employing stronger footshock

in producing better retention (Gold & van Buskirk, 1976).

Similarly, enhancement of learning due to increases in total

footshock delivered might be accompanied by higher

corticosterone concentrations.

Corticosterone concentrations have been found to be

responsive to both increased shock duration (Friedman, Ader,

Grota, & Larson, 1967) and increased shock intensity (Keim &

Sigg, 1976). In the present experiment, diabetics showed

better retention than nondiabetics following both 1-second

and 2-second shocks. One possible factor in this

superiority could have been that diabetics experienced

larger corticosterone increments in response to footshock

than did nondiabetics.

In order to evaluate the effects of our footshock upon

circulating hormone levels, plasma corticosterone

concentrations following training and testing for passive

avoidance were determined in Experiment 4.


Experiment 4. Plasma Corticosterone Concentratiors
Following Footshock Training and Retention Testing
in Diabetic and Nondiabetic Rats

A number of experimental stressors cause the release of

ACTH, which in turn stimulates the adrenal cortex to

increase synthesis and release of corticosteroids.

Elevations of plasma corticosterone concentrations have been

reported following cold, immobilization and footshock stress

in rats (Keim & Sigg, 1976; Lenox, Kant, Sessions,

Pennington, Mougey, & Meyerhoff, 1980; Odio & Maickel,

1985). ACTH concentration in plasma increased in response

to both footshock training and the re-exposure to the fear

cues during retention testing (van Wimersma Greidanus et

al., 1977). The hormonal response was much larger when rats

were replaced in the training situation than when they were

given footshock. Thus, there may be even greater

sensitivity of the hypothalamo-pituitary-adrenal system to

psychological stress (e.g. fear) than there is to some

physiological stresses (e.g. footshock).

This hormonal sequence of ACTH release followed by

corticosterone release has been associated with the storage

of memory. Some evidence has been reported which indicates

that the concentration of ACTH in plasma following training

is a determinant of the strength of the memory (Gold & van

Buskirk, 1976). In Experiment 3b, diabetics exhibited

better retention for passive avoidance training than

nondiabetics with both 1-second and 2-second duration


shocks. One possible explanation for this superiority would

be that diabetics had higher baseline corticosterone

concentrations (Tornello et al., 1981b) and/or experienced

greater corticosterone increments in response to footshock


To investigate these possibilities, corticosterone

concentrations were measured in plasma samples obtained 15

minutes after footshock training and 15 minutes after re-

exposure to the shuttle box during retention testing. The

15 minute timepoint was chosen because it was reported to be

the approximate time of maximal elevation of corticosterone

concentrations following footshock in some previous studies

(e.g., Madden, Rollins, Anderson, Conner, & Levine, 1971).

In addition this timepoint corresponds to the time when

neurochemical changes in response to footshock training

(Tanaka et al., 1982) and retention testing (Herman et al.,

1982) have been found. It was hoped that both neurochemical

and peripheral hormonal events linked to the training and

testing situations could be identified and thus a timepoint

most likely to accommodate all the various parameters was



The rats and behavioral procedures were as in

Experiments 3a and b. Blood samples were collected

following footshock training by tail nick, as described in

Experiment 1. Blood samples were collected following

retention testing by decapitation as described in Experiment

3. The samples were centrifuged and the plasma separated

and stored at -700C. Plasma was also collected from 7

diabetic and 7 nondiabetic rats within 3 minutes of removal

from their home cages, in order to establish baseline

corticosterone concentrations. Baseline plasma samples were

collected between 1:00 P.M. and 3:00 P.M., approximately the

same time at which training and testing were conducted.

Corticosterone concentrations were measured using the

competitive protein-binding assay of Murphy (1967).

Briefly, 10 ul samples of plasma (40 ul of plasma from

unstimulated rats) were extracted in 2 mis of methanol, and

duplicate 500 ul aliquots of the extract were dried under a

warm air stream along with corticosterone standards (.25,

25, 1.0, 1.5, 2.0, 2.5, 3.0 and 4.0 ng). After drying was

complete (approximately 20 minutes), 1 ml of a reaction

mixture containing dog plasma (from a dexamethasone-treated

dog) and [3H]-corticosterone was added to each tube. The

samples and standards were then incubated for 5 minutes at

40-450C and then placed on ice for 20 minutes to bring the

reaction to equilibrium. Dextran-coated charcoal (0.2 ml)

was then added and 10 minutes later the tubes were

centrifuged for 10 minutes at 40C. The supernatant was

decanted into scintillation vials and Beta activity was



The standards were fitted to a straight line using the

log-logit transformed data, and samples were evaluated by

comparison with the standard values along the linear

regression line. Group means were compared using two-way


Results and Discussion

The mean corticosterone concentrations of the various

groups are shown in Figures 9, 10 and 11. All values

measured 15 minutes after training and testing were elevated

relative to the afternoon baseline concentrations, although

the baseline measures were not compared statistically with

the learning corticosterone measures, since they were not a

part of the statistical design of the learning paradigms.

The mean baseline concentrations of diabetic and nondiabetic

rats were compared and were found to be indistinguishable

statistically. Among the diabetic state dependent groups

(Figure 9), the only significant effect was the effect of

session (F(1,35)=8.94, P<0.01), with test values being

higher than training values. Among nondiabetic state

dependent groups (Figure 10), there was an interactive

effect of treatment with session (F(1,34)=4.18, P<0.02).

Thus, values were generally higher after testing than after

training, but in addition, there were differential changes

between the sessions. In particular, the SI group had very

high corticosterone concentrations after testing, suggesting

an interaction of hypoglycemia and psychological fear. At

Figure 9. Mean plasma corticosterone concentrations of
diabetic state-dependent groups following footshock
(Training) and re-exposure to fear cues (Testing).
SS=saline prior to training and testing, SI=saline prior to
training and insulin prior to testing, IS=insulin prior to
training and saline prior to testing, II=insulin prior to
training and testing, Basal=non-stressed controls. *Testing
concentrations significantly higher than Training

Testing *



II Basal

Figure 10. Mean plasma corticosterone concentrations of
nondiabetic state-dependent groups following footshock
(Training) and re-exposure to fear cues (Testing).
SS=saline prior to training and testing, SI=saline prior to
training and insulin prior to testing, IS=insulin prior to
training and saline prior to testing, SSTP=saline prior to
training and testing and streptozotocin administration
following training, Basal=non-stressed controls. *Testing
concentrations significantly higher than training
concentrations. +Significantly different from all other
Testing groups.


Testing *

SS SI IS SStp Basal


Figure 11. Mean plasma corticosterone concentrations of
saline-treated nondiabetic and diabetic rats after 1-second
(N1 and Dl) or 2-second footshock (N2 and D2) after
footshock training and after retention testing and of groups
(N and D) not exposed to any stress (Basal). *Testing
concentrations significantly higher than Training

N1 D1 N2 D2

N1 D1 N2 D2


the other extreme were the SSTP rats which showed

essentially no difference in corticosterone response to

testing than to training.

Among the fear cue re-exposure groups (Figure 11), there

was a significant main effect of session (F(1,33)=5.34,

P<0.05), with concentrations at testing, again, higher than

those at training.

The present results comfirm the previous finding of

greater pituitary-adrenal activation following retention

testing than following footshock training (van Wimersma

Greidanus et al., 1977). However, the differences in

corticosterone values found here failed to provide any

hormonal basis for the superiority of diabetics in retention

for the avoidance task. It is possible that different ACTH

elevations occurred in the diabetics and nondiabetics

despite similar corticosterone increments (De Nicola et al.,

1977). Another possibility is that the timecourse of the

hormonal response to stressful stimuli is different in

diabetic and nondiabetic rats. Thus, maximal responses to

stress in diabetic and nondiabetic rats may be

indistinguishable while duration of the increments are

longer-lasting in diabetics. Such alterations in timecourse

have been noted in aging rats (Sapolsky, Krey, & McEwen,

1983). In addition, metabolic disturbances associated with

epinephrine or cortisol infusion endure much longer in

diabetics than in nondiabetics (Shamoon et al., 1980).


Finally, 15 minutes post-stress may not be the optimal

time for the measurement of peak elevations of

corticosterone in response to stress. There have been

studies which have reported maxima at 30 to 60 minutes

(Bassett & Cairncross, 1975; Keim & Sigg, 1976; Odio &

Maickel, 1985).

Experiment 5. Twenty-Four Hour Catecholamine Excretion
in Diabetic and Nondiabetic Rats

There have been reports of decreased norepinephrine

turnover in some peripheral tissues (Yoshida et al., 1985)

and decreased plasma epinephrine and norepinephrine

(Berkowitz & Head, 1978) in diabetic rats. However, DBH, an

important synthetic enzyme is elevated (Schmidt et al.,

1981), as is norepinephrine for at least the first two

months following streptozotocin administration, and thus the

diabetic may be either advantaged or disadvantaged in the

face of stress, depending upon whether existing plasma

catecholamine concentrations or synthetic capability is more

critical. In any case, plasma catecholamine responses to

footshock have been shown to be relevant to later

demonstration of good retention of passive avoidance

training (McCarty & Gold, 1981). To extend these previous

findings regarding peripheral catecholamine status in the

diabetic rat and its possible relationship to learning an

aversive task, the present experiment was conducted in which

24-hour excretion of catecholamines was measured in diabetic

and nondiabetic rats both under basal conditions and during

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.8 - mvs