Recovery and recoverability in the central nervous system following long-term ethanol consumption


Material Information

Recovery and recoverability in the central nervous system following long-term ethanol consumption
Physical Description:
210 leaves : ; 29 cm.
King, Michael A
Publication Date:


Subjects / Keywords:
Ethanol -- adverse effects   ( mesh )
Central Nervous System -- drug effects   ( mesh )
Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- Neuroscience -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1985.
Bibliography: leaves 186-209.
Statement of Responsibility:
by Michael A. King.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000520851
oclc - 16878151
notis - ACU2372
System ID:

This item is only available as the following downloads:

Full Text







Copyright 1985


Michael A. King


We wish to thank Pat Burnett, Larry Ezell, Dot

Robinson, and Regina Davis for technical assistance with

these experiments. This work was supported in part by NIAAA

predoctoral fellowship AA05175 to H.A.K.. Other sources of

support include the Medical Research Service of the Veterans

Administration and NIAAA grant AA00200 to D.W.V.. Data

analyses were made using the Northeast Regional Data Center

IBM 3081 computer with MVS/XA at the University of Florida,

under release 82.4 of SAS (Statistal Analysis System, SAS

Institute Inc., Cary, N.C.).






. .. vii

LIST OF FIGURES . . v. iii




Brain Damage and Psychological Deficits Related
to Ethanol Abuse .. . 1
Alcoholism, Memory Deficits, and the Hippocampus 2
Recovery From Ethanol-Belated Disorders 3
Tissue and Cellular Mechanisms Related to Memory
Disorders . 4
Dendrites and Dementia, Alcohol and Aging 6
Alcohol Abuse and Brain Injury of Non-alcoholic

Origin . .
The Anatomy of the Rat Hippocampus .
Major Features .
Cell Types and Internal Features .
Dentate Gyrus Afferent Connections
Entorhinal-Dentate Afferents .
Septal-Dentate Afferents .
Brainstem and Hypothalamic Dentate
Afferents .
Hilar Afferents to Dentate .
Dentate Efferents a .
CA3 Affterents . .
CA3 Efferents . .
CA1 Afterents . a
CA1 Efferents . ..

. 20
. 20
. 22
S. 24
. 25
. 26
. 27

Reactive Synaptogenesis in the Dentate Gyrus
The Liquid Diet Chronic Ethanol Consumption
Model . .

. 27

. 32


Design and Methods .... 35
Golgi Histology . 36
Gross Brain Measurements . 37
Dendritic Spine Counts 37
Results . 41
Gross Measures . 42
Spine Density Measures ... 44
CA1 Spine Density . .44
Dentate Granule Cell Densities 45
Discussion . . 52
Experimental Considerations 52
Estimates of Total Neuron Alterations in
Spine Number . 59
Electrophysiological Consequences of Long-
term Ethanol Exposure . 61
Possible Mechanisms of Spine Density
Alterations .. 63
Somatic proximity 63
Presynaptic influences 65
Systemic and molecular influences 67
Spine Density and Recovery in CT Scans 73


Experimental Methods and Design . 74
Entorhinal Cortex Lesions . 74
Histological Preparation . 75
Verification of Lesion Placement and Size 77
Measurements of Histological Sections 79
Measurement software for Tima's material 81
Measurement software for AChE material 83
Stain Intensity Normalization 84
Results . . 87
Alcohol Consumption .- 87
Alcohol Effects: Coronal Timn-stained
Dentate . 89
Timma's Band Widths in Unlesioned Animals 89
Tiam's Stain Intensity in Unlesioned
Animals . 93
Alcohol Effects: The Contralateral Dentate 97
Contralateral Exposed Blade Widths in AChE
Material 105
Contralateral Exposed Blade AChE Stain
Intensity . 109
Contralateral Buried Blade Widths in AChE
Material . 112
Contralateral Buried Blade AChE Stain
Intensity . 116
Entorhinal Lesions . 120
Unlesioned vs. Contralateral AChE Patterns 132

Synaptic Reorganization Effects 139
Buried Blade Widths in AChE Material 139
Buried Blade AChE Intensity .. 140
Exposed Blade AChE Widths 150
Exposed Blade AChE Stain Intensity 153
Discussion . 161
Chronic Ethanol Consumption and Tima's
Patterns in Dentate 165
AChE Patterns in the Contralateral Dentate
Gyrus . 166
Inhibited Reactive Synaptogenesis in Alcohol
Animals . 170
Lesion-induced AChE Intensity Changes 172
Buried and Exposed Blade Differences Before
and After Lesions . 174
Possible Mechanisms of Long-term Ethanol.
Effects . 176
Experimental Considerations 178


Spine Density Recovery, but Inhibited Reactive
Synaptogenesis? .. .. 181





1. Gross Brain Measurements and Weights 43

2. CA1 Pyramidal Neuron Spine Densities . 46

3. Dentate Granule Neuron Spine Densities 49

4. Timm's Stain Band Widths and Intensity Measures 90

5. AChE Stain Band Widths 106

6. AChE Normalized Stain Intensity 107




1. The Hippocampus in Medial View. . 12

2. A Schematic Transverse Section Through the
Hippocampus . . 14

3. Dentate Gyrus Timm's and AChE Staining Patterns 21

4. Schematic Summary of Dentate Reactive Synaptogenesis 30

5. Spine Density Variability as a Function of the
Number of Segments Counted . 38

6. Dendritic Spines on Golgi-stained Dendrite 44

7. CA1 Pyramidal Neuron Dendritic Spine Densities 47

8. Dentate Granule Neuron Dendritic Spine Densities 50

9. Reconstruction of Entorhinal Cortex Lesions 78

10. Computer-assisted Tiaa's Width and Stain Intensity
Measurement . . 82

11. Computer-assisted AChE Band Width and Stain
Intensity Measurement . 84

12. Linearity of EyeCom II Gray Levels ... 86

13. Alcohol Consuartion Over the 20 Week Exposure Period 88

14. Buried Blade Coronal Unlesioned Tiam's Band Widths 91

15. Buried Blade Coronal Unlesioned Tima's Widths by A-P
Level . . 92

16. Exposed Blade Coronal Unlesioned Timm's Band Widths 94

17. Exposed Blade Coronal Unlesioned Timm's Band Widths
by A-P Level . . 95


18. Exposed Blade Coronal Unlesioned Timm's Band Widths
by M-L Level . . .. 96

19. Coronal Buried Blade Unlesioned Tiam's Band Raw
Stain Intensity . 98

20. Coronal Buried Blade Unlesioned Timm's Band
Normalized Stain Intensity . 99

21. Coronal Exposed Blade Unlesioned Timm's Band Raw
Stain Intensity . . 100

22. Tiaa's Intensity Means Along the Medial-lateral
Axis . . 101

23. Coronal Exposed Blade Unlesioned Timm's Band Stain
Intensity . . 102

24. Tiam's Unlesioned Exposed Blade Normalized Stain
Intensity by A-P Level . .. 103

25. Contralateral Exposed Blade Group Mean AChE Widths 108

26. Contralateral Exposed Blade Group Mean AChE Band
Widths by D-V Level . 110

27. Contralateral Exposed Blade Sucrose Group Mean
Normalized AChE Stain Intensities by D-V Level 111

28. Contralateral Exposed Blade Normalized C/A AChE
Intensity by D-V Level . 113

29. Contralateral Exposed Blade Normalized OML AChE
Intensity by D-V level . 114

30. Contralateral Exposed Blade Normalized OHL AChE
Intensity by D-V Level . 115

31. Contralateral Buried Blade AChE Band Widths 117

32. Contralateral Buried Blade AChE Band Widths by D-V
Level . . 118

33. Contralateral Buried Blade Sucrose Group Mean
Normalized AChE Stain Intensities by D-V Level 119

34. Contralateral Buried Blade C/A AChE Band Intensity
by D-V Level . 121

35. Contralateral Buried Blade OML AChE Band Intensity,
by D-V Level . . 122

36. Contralateral Buried Blade OEL AChE Band Intensity,
by D-V Level . 123

37. Contralateral Buried Blade OML AChE Band Intensity,
by D-V Level . . 124

38. Molecular Layer Shrinkage Following Entorhinal
Lesion . . 125

39. Molecular Layer Width Following Entorhinal Lesion 126

40. AChE C/A Band Width Following Entorhinal Lesion 127

41. AChE C/A Band Expansion Following Entorhinal Lesion 128

42. AChE Stain Patterns in Unlesioned and Lesioned
Animals . . 130

43. Tima's Stain Patterns in Unlesioned and Lesioned
Animals . . 131

44. Raw AChE Stain Intensity, Unlesioned vs.
Contralateral Dentate . 135

45. Corrected AChE Stain Intensity, Unlesicned vs.
Contralateral Dentate . 136

46. Normalized AChE Stain Intensity, Unlesioned vs.
Contralateral Dentate . 137

47. Normalized and Corrected AChE Stain Intensity,
Unlesioned vs. Contralateral Dentate .. 138

48. Buried Blade AChE Stain Band Width Changes in
Entorhinal Lesioned Animals . 141

49. Buried Blade Lesion-induced OHL Shrinkage by D-V
Level . . 142

50. Buried Blade AChE C/A Band Widths, Ipsi- and
Contralateral, by D-V Level . 143

51. Buried Blade AChE Band Stain Intensity Responses to
Entorhinal Lesions .. -. 145

52. Buried Blade AChE GCL and C/A Band Stain Intensity
Responses to Entorhinal Lesion 146

53. Buried Blade AChE GCL and C/A Band Stain Intensity
Responses to Entorhinal Lesions by D-V Levels 148

54. Buried Blade AChE OBL Band Stain Intensity Responses
to Entorhinal Lesions by D-V levels 150

55. Buried Blade AChE OM1 Band Stain Intensity Respcnses
to Entorhinal Lesions by D-V Levels 151

56. Buried Blade AChE OHL Band Stain Intensity Responses
to Entorhinal Lesions by D-V Levels 152

57. Exposed Blade AChE Band Width Responses to
Entorhinal Lesions . 154

58. Exposed Blade OHL Shrinkage in Response to
Entorhinal Lesion, by D-V Level . 155

59. Exposed Blade AChE C/A Band Width Responses to
Entorhinal Lesions, by D-V Levels ... 156

60. Exposed Blade AChE Stain Intensity Responses to
Entorhinal Lesions . 158

61. Exposed Blade AChE C/A Band Stain Intensity
Responses to Entorhinal Lesions, by D-V Levels 159

62. Exposed Blade AChE C/A Band Stain Intensity
Responses . a 160

63. Exposed Blade AChE OML Band Stain Intensity
Responses to Entorhinal Lesions . 162

64. Exposed Blade AChE 0ML Band Stain Intensity
Responses to Entorhinal Lesions . 163

65. Exposed Blade AChE OHL Band Stain Intensity
Responses to Entorhinal Lesions 164

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



Michael A. King

May 1985

Chairman: Don W. Walker
Major Department: Neuroscience

Recovery from central nervous system pathology consequent

to chronic ethanol consumption was assessed by comparing the

density of dendritic spines on two types of Golgi-stained

neurons in the rat hippocampus, in animals with or without a

post-consumption "abstinence" period. Spines are

morphological postsynaptic specializations on some nerve

cells and are theoretically related to intellectual

competence. A nutrition-controlled liquid diet was used to

expose male Long-Evans rats to ethanol. CA1 neurons

initially showed a nonsignificant reduction in spine

density, but significantly increased density after 20 weeks


abstinence. Dentate granule cells had significantly higher

spine densities after 20 weeks of exposure, but a

nonsignificant reduction during abstinence. Thus the

changes in gross brain morphology observed in CT scans of

human alcoholics before and after abstinence may be

associated with neuronal recovery.

Recoverability, the capacity of the central nervous

system to respond to acute invasive injury, was also

examined for the effects of long-term ethanol consumption.

Histochemical staining for hippocampal acetylcholinesterase,

(AChE) an enzyme involved in synaptic neurotransmission, and

the Tima's sulfide/silver stain for certain endogenous

metals, describe distinct afferent axon terminal fields in

the dentate gyrus; these form distinct bands in tissue

sections. The width and stain intensity of these bands

characteristically change following partial dentate

deafferentation, reflecting synaptic reorganization within

the dentate gyrus. Using an image analysis computer system,

this remodelling was quantified to compare the responses of

alcohol-exposed and control rats. Twenty weeks of alcohol

exposure may produce a residual impairment in the ability of

unlesioned axons to sprout new terminals in response to

unilateral destruction of entorhinal afferents, even when an

8 week ethanol-free period precedes the lesion. In

addition, evidence for pre-lesion damage was found in

unlesioned alcohol controls, using measurements of Timm's


stain bands, and on the side opposite the unilateral lesion.

Human alcoholics, predisposed to require neurological

treatment, may thus require different therapeutic approaches

than nonalcoholics.



Brain Damage and Psycholoqical Deficits Belated to Ethanol

Residual cognitive impairment and brain damage have long

been associated with chronic alcohol abuse. Recent reviews

affirm that a consistent pattern of clinical symptoms may be

exhibited by alcoholic patients (Begleiter et al., '80,

Goodwin and Hill, '75; Parsons, '77; Ron, *77; Ryan, '80;

Tarter, '75). The presence of physical brain damage in

human alcoholics has been inferred from

electroencephalographic (Coger et al., '79) and evoked

potential (Salamy et al., '80) abnormalities, and

pneumoencephalographs (Brewer and Perrett, '71; Haug, '68;

Tumarkin, '55). It has been confirmed directly in

postmortem examinations (Alling and Bostroa, '80; Courville,

'66; Lynch, '60; Miyakava et al., '77; Victor et al., '71),

and over 1000 cases of computed tomographic (CT) evidence of

damage in vivo have now been reported in the clinical

literature (Cala and Mastaglia, '81). Neurological

examination of chronic alcoholics commonly reveals actor

performance loss, and cerebral atrophy, if a CT scan is

obtained, as well as the evoked potential and EEG

abnormalities. Upon psychological testing, deficient

perception, subnormal conceptual and problem solving

abilities, and decremented learning and recent memory

capacity are commonly found. Additional complications often

include autonomic and endocrine imbalances, sleep disorders,

and emotional problems (Kissin, '79; Wagman and Allen, '77).

Because it is estimated that there are currently between ten

and twenty million "problem" drinkers in the United States

(Clark and Midanic, '82; Halin et al., '82), the clinical

neurobiological effects of chronic ethanol abuse constitute

an increasingly important health issue.

Alcoholism, Memory Deficits, and the Hippocampus

Human alcoholics, when neuropsychologically tested, often

exhibit disabilities of memory, especially recent, or short

term memory (Ryan, '80). Their performance is typically

sensitive to any interference or distraction they encounter

during testing (Butters et al., '77; Ryan, 80). In humans

who have undergone neurosurgery resulting in damage to the

hippocampal formation in the inferior temporal lobe, a very

similar memory disorder has been described (Butters and

Cermak, '75, Milner et al., '68, Sidman et al., *68). A

similar effect is also found in rodents after long term

consumption of ethanol (Freund and Walker, '71; Walker and

Hunter, '74, '78; Bond and DiGuisto, '76; MacDonnall and

Marcucella, '78; Denoble and Begleiter, '79; Smith et al.,

'79; Irle and Markowitsch, '83), or damage to the

hippocampal formation (Walker and Means, '73; Winocur, '79).

The consistent findings of hippocampal formation damage in

the brains of human alcoholics at autopsy (Brion '69; Victor

et al., '71; McLardy, '73a; Miyakawa et al, '77), and in

laboratory rodents (Riley and Walker, '78; McMullen et al.,

'84; Irle and Markowitsche, '83; Phillips and Cragg, '83)

and non-human primates (Montgomery et al., '79) exposed

repeatedly to ethanol have led to a working hypothesis that

alcoholic neuropathology in this structure is intimately

related to the functional deficits that are so consistently

observed (cf. Walker et al., '81; Irle and Markowitsch,

'83). Thus the hippocampus serves as the anatomical focus

in the experiments described below.

Recovery from Ethanol-Related Disorders

A recent issue in the study of central nervous system

(CNS) effects of human chronic alcoholism concerns the

propensity for recovery from the pronounced intellectual and

neurobiological consequences that are commonly observed

(Carlen and Wilkinson, '80; Ryan, '80). Evidence for the

operation of some form of recovery comes from both

neuropsychological improvement cn diagnostic tests during

abstinence (Guthrie and Elliot, '80; Hester et al., '80;

Schau et al., '80; Sharp et al., '77; Cermak and Ryback,

'76; Templer et al., '75; Clark and Houghton, t75; Farmer,

*73; Long and McLachlan, '74; Page and Linden, '74) and

indices of the morphological and physiological state of the

brain. The latter consist primarily of a tendency for

cerebral atrophy, as defined by the apparent enlargement of

extrabrain spaces, to spontaneously decrease in severity

during abstinence (Carlen et al., '78; Hill and Mikhael,

'79). In addition, measures of cerebral cortical blood

flow, and its regional activation during specific cognitive

activities, have been reported to improve (Berglund and

Risberg, '80), while abnormalities in stimulus-evoked

electrical waveform components recorded from the scalp have

been found to return toward normal (Salamy et al., '80;).

Sleep disorders may also abate during abstinence (Vagman and

Allen, '77). It appears likely, therefore, that functional

recovery from some of the deleterious effects of chronic

alcoholism may occur.

Tissue and Cellular Mechanisms Related to Memory Disorders

How could neuropathological changes subsequent to chronic

ethanol result in cognitive disabilities in mammals? The

tissue effects, in humans, are commonly demyelination

(Alling and Bostrom, '80), and various degrees of focal

encephalopathy (Lynch, '60; Victor et al., 71; Miyakava et

al., '77; Ron, '77; Nakada and Knight, '84); cell loss is

reported although little quantitative data exist to

substantiate this observation. Cell loss is also indicated


by shrinkage of cortical folds, and findings that chronic

ethanol treatment in rats results in hippocampal cell loss

in both of its main component substructures (Walker et al.,

'80) corroborate this view. Cell loss in other brain

regions, specifically the cerebellum, has also been

documented in animals exposed to alcohol for long periods

(Tavares and Paula-Earbosa, 82; Walker et al., '81; Irle and

Markowitsch, '83). A decrease in the number of neurons in

the CNS, however, is virtually permanent since there is

almost no cell division in neurons in the adult central

nervous system. If the apparent recovery of cerebral

atrophy and cognitive dysfunction really does occur, then

some other mechanism than replacement of lost neurons must

underlie this recovery. Several possible general mechanisms

for recovery from brain damage have been proposed (Finger

and Stein, '82), such as the takeover of functional

abilities by undamaged areas, or development of alternative

behavioral strategies for solving some task, using different

anatomical circuitry. One possible mechanism that is

experimentally related to relative intellectual capacity in

humans and animals, and consistent with current cellular

neurobiological evidence, is that the degree of

differentiation of the dendritic tree of neurons is flexible

(Morest, '69), and determinant of the number and perhaps

quality of synaptic contacts received from incoming axons.

If some condition results in a decrease in the number of


contacts a neuron maintains, then that cell cannot integrate

as much information (Preund, '82). From this it follows

that the information processing capabilities of neuronal

ensembles, and the nervous system as a whole, would be

compromised by any agent that reduced the number of

connections made by individual neurons.

Dendrites and Dementia, Alcohol and Aging

In several neurological conditions having the quality of

dementia (the decreased ability to think, remember, and

perceive), dendritic de-differentiation is common (Purpura,

'75, Buell and Coleman, *79), and takes two main forms.

The first is a decrease in the length of the treelike

dendritic branches, the second, a decrease in the density,

or morphological integrity, of dendritic spines. These are

small protrusions first described by Bamon y Cajal in 1891

that often cover the majority of dendrites, and contain

attachment sites for incoming axon terminals. Both changes

often occur together, resulting in substantial losses in the

number of connections a neuron can make. This met

degeneration of dendritic integrity is in contrast with what

appears to to the normal situation throughout most of life,

where both degeneration and synthesis occur, with some net

dendritic growth (Buell and Coleman, '79). Both fcras of

dendritic disintegration have been reported to occur in

senile and Alzheimer's deaentias (Hehraein et al., '75;


Buell and Coleman, '79), temporal lobe epilepsy (Scheibel et

al., '76), and mental retardation (Purpura, '74), in

humans, and in senescence (Bondareff and Genisman, 76;

Bondareff et al., 78; Hachado-Salas and Scheibel, '79; Cupp

and Uemura, '80; Uemura, '81; Hervis, '81) and experimental

chronic alcoholism in animals (Biley and Walker, '78,

Tavares et al., '83a,b; McMullen et al. '84; Phillips and

Cragg, '83). Following long term ethanol exposure in

animals, dendrites in the hippocampal formation and

elsewhere have been reported to become vacuolated and

contain various vesicular organelles (Goldstein et al.,

'83;), and display anomalous spine morphology (Tavares et

al., '83a,b; Phillips and Cragg, '83; Popova, '83), in

addition to outright decreases in spine numbers (Tavares et

al., '83b; Popova,'83; Riley and Walker, '78) and dendritic

extent (Tavares et al., '83a; McMullen et al., '84).

It numerical decreases in connections manifest in

intellectual disabilities, could increases in dendritic

complexity take place that might be associated with recovery

of such abilities? According to basic principles in the

cell biology of toxic substances, recovery of cell

processes, both chemical and physical, is often observed

unless a cell has been exposed to some critical, lethal

amount of the offending substance (Bridges et al., '83).

Such a restoration by the cellular elements of the CNS has

been proposed as a mechanism by which the ameliorization of


the cerebral atrophy diagnosed in CT scans night be possible

(Carlen et al., '78; Hill and Mikhael, '79), but

experimental discrimination among several possible

hypotheses is not yet possible in humans. Such a mechanism

could also account tor the lack of cognitive recovery in

alcoholics with extreme consumption histories (Guthrie,

'80). Recently, a translation of a report from the Soviet

Union describes qualitative recovery of dendritic branching

and spines damaged by long term ethanol exposure (Popova,

'83), and what must be considered a preliminary quantitative

study of hippocampal dendritic branching recovery has also

recently appeared (McMullen et al., '84). It has also been

reported that, following an experimental partial

deafferentation of dentate granule cells, both dendrites and

spines can be observed to undergo phases of de- and

regeneration (Steward and Vinsant, '83; Steward and Caceres,

'83), which consists of both replacement by new structures

and reconstruction by some of those dendrites and spines

actually having exhibited internal pathology- Presynaptic

axonal sprouting is known to accompany these changes (Cotman

et al., '81). According to the neuronal deafferentation

hypothesis of the Toronto researchers (Carlen et al., '78),

cerebral atrophy is at least partly a reflection of a

degenerative disconnection of neuronal circuitry due to

ethanol neurotoxicity. Recovery from this condition would

necessitate reconnection, which would require elaboration of

both pre- and postsynaptic structures.

The first experiment was designed to quantitatively

investigate one particular form of dendritic pathology in

neurons of the hippocampal formation of rats following long

term ethanol exposure, and the possibility that some

recovery from those effects might occur. To expose animals

to ethanol, a nutrition-controlled liquid diet model of

chronic ethanol consumption was used. This model of chronic

dietary ethanol consumption has been refined over the past

decade to address experimentally questions that are

unresolvable using clinical approaches. The effects of

ethanol on dendritic spine density were measured in two

types of hippocampal neurons, comparing animals with and

without postethanol abstinence periods. To determine

whether dendritic recovery might account for a restoration

of cerebral volume such as appears to occur in CT scans of

some abstinent human alcoholics, gross brain morphometrics

were also collected.

Alcohol Abuse and Brain Injury of Non-alcoholic Origin

It has recently come to be appreciated that alcoholics,

especially young alcoholics, are as much as three times as

likely to suffer strokes and other cerebrovascular disorders

compared to nonalcoholics (cf. Nakada and Knight, '84;

Koval', 78). Alcohol abuse can also be causally related to

accidental falls (Honkanen et al., '83), automobile

accidents, and other forms of violent injury. In one study,

in almost 40% of emergency medical cases involving 30-59

year old patients, chronic and/or acute alcohol abuse was

found (Kristensson-Aas et al., '81). A significant

percentage of these patients will require neurological

treatment at some time in their lives. Very little is

known, however, about the effects of long term alcohol

intake on recovery trom nervous system damage. Two factors

complicate attempts to investigate this problem

systematically in humans. First, alcoholics can often be

shown to have CNS (central nervous system) pathology

directly (neurotoxicity) or indirectly (poor nutrition,

peripheral organ dysfunction, etc.) related to alcoholism.

Since these variables vary considerably within the human

population, experimental control is extremely difficult to

obtain. Second, the CNS of adult mammals is often

notoriously intractable to recovery from pathological

insults. Together with the high degree of variation in

natural and accidental injuries to the CNS, and the lack of

technical means by which to assess human brain damage and

recovery noninvasively, the current problems in analyzing

neurobiological recovery in the clinic make it difficult if

not impossible to sort out the effects of another variable

such as alcohol use. It cannot presently be determined, for

example, how much of an observed functional recovery from

CNS injury represents a regeneration of damaged connections,

circuitry replacement by other brain systems, or simple but


subtle substitutions of behavioral strategies. Furthermore,

such hypothetical mechanisms are neither necessarily

mutually exclusive nor generalizable to all forms of injury

(Finger and Stein, '82).

To circumvent some of these difficulties, we have used an

animal model of chronic alcohol consumption and a well-

studied animal model system of adult mammallian CNS tissue

response to extrinsic damage to investigate the effect of

prior long-term ethanol exposure on CBS recovery. This is

part of a series of experiments to isolate effects of

ethanol prior to, following, or both before and after an

experimental brain lesion.

Before describing either the "recovery" or the

recoverabilityy" experiment in more detail, the anatomy of

the hippocampus and the experimental model of brain injury

will be reviewed, and the methods common to both experiments

(the liquid diet procedure) will be presented.

The Anatony of the Rat Hippocampus

Major Features

The hippocampal formation is a unique cortical structure

that consists of two sheets of neurons folded into one

another. In any transverse section these appear as

interlocked, forward and backward 'C's (figure 2). In the

rat, it lies curved over the top of the thalamus, and forms

the internal wall of much of the lateral cerebral ventricles


(Cajal, '68; Lorente de No, '34). Gross dissection of the

hippocampus (an excellent example is found in Fjerdingstad

et al., '74a) reveals some basic principles of hippocampal

anatomy. Looking like a pair of bananas (figure 1), with

the stem ends joined at the aidline dorsally and anteriorly,

the hippocampi diverge, descend, and finally turn forward

before blending in to other temporal lobe cortical

structures. The "stem" end is the septal pole, while the

ventral tip is the temporal pole.

If ?

T /r K1 '.T .' I
Legend: DG, dentate gyrus; f, fornix; fin, fimbria;
dhc, dorsal hippocampal commissure; vhc, ventral
hippocampal commissure. Anteriorly, the septal
complex is observed.

Figure 1: The Hippocampus in Medial View


At the septal poles, narrow bundles of white matter, the

fornices, consist of axons travelling tc and from each

hippocampus and the septal complex, the hypothalamus, and

midbrain nuclei. Between each tornix at about the point

where hippocampal gray matter causes the enlargement of the

"banana", dorsal and ventral hippocampal commissures carry

axons from each hippocampus to the other. As the

hippocampus gray matter begins, the fornices spread out over

it like the peel of the banana. This covering of white

matter on the dorsal surface of the hippocampus is called

the alveus, and where it forms a lateral ridge along the

lesser curvature, it forms the fiabria.

Cell Types and Internal Features

The elegant internal structure of the hippocampus

attracted the attention of many of the nineteenth and early

twentieth century neuroanatomists, who devised the

descriptive terminology still in use today (Lorente de Mo,

'34). As the most evolutionarily primitive form of cerebral

cortex, it has a relative structural simplicity that

underlies its use as an anatomical and physiological model

for investigating the neurobiology of cerebral cortex (cf.

Swanson et al., '82). In cross section, the two main

component structures of the hippocampal formation can be

easily discerned (figure 2). The ventral and medial

portion, the dentate gyrus, assumes approximately the shape

of a sideways V, with a blade exposed to the extrabrain

space ventrally and a blade buried in the other folded sheet

of neurons. This other sheet, resembling a C in transverse

section, is the hippocampus proper, also known as Ammon's

horn, for the Latin Cornu Ammonis, after an early

neuroanatomist fancied the similarity between the horn of a

ran and the hippocampus (Cajal, '68). Still visible between

the two components, the hippocampal fissure is a remnant of

the folding and growing together of the dentate and Asmmn's

horn during early development. Many hippocampal blood

vessels are situated along the fissure (Coyle, '76).


AMMON'S o' o0 .0
HORN T 0*-oo ^- o0



Figure 2: A Schematic Transverse Section Through the

The dentate gyrus is divisible into three subregions.

The easily visible layer of cell bodies of the predominant


neuronal type, the granule cell, may be 10 neurons or more

in thickness. The crest of the V-shaped granule cell layer

defines the transition between buried blade, where the

neurons abut the fissure, and the exposed blade, facing the

extraventricular cistern. The dentate gyrus is formed by

about 2,170,000 granule neurons, in the rat (West and

Anderson '80), that each extend a conical arborization of

dendrites exclusively to the outside surface of the dentate.

Between the granule cell body layer and the pia mater at

these external dentate borders, granule cell dendrites

extend into a sonata-poor zone called the molecular layer

(ML). Here, incoming afferentt) axons to granule cells form

synaptic contacts on their dendrites, which form conical

arborizations with many branches. The dendrites of granule

neurons are extensively covered with spines in normal


In the space enclosed by the granule cell layer, the

hilus of the dentate, over 21 morphological types of neurons

(Amaral, '78) form the most complex anatomical region in the

hippocampal formation. For this reason they are commonly

referred to collectively as comprising the polymorph zone,

where the transition from dentate to hippocampus occurs.

Granule cell axons descend into the hilus, making contacts

with many hilar neurons (Blackstad and Kjaerheim, 61;

Blackstad et al., '70). They then enter the Ammon's horn,

the ventral portion of which is known as regio inferior


(Lorente de No', '34). Regio inferior is populated by large

pyramidal neurons. These extend a thick, tree-like apical

dendritic arborization toward the fissure; a broad region of

the proximal dendritic tree courses through the stratum

radiatum, while the distal fifth extends into status

lacunosum/moleculare. Regio inferior pyramidal cells also

elaborate a basal dendritic tree, into stratum oriens. This

extends all the way from the cell body layer to the alveus.

About halfway along the cell body layer, a rapid and

dramatic transition to a smaller pyramidal cell occurs.

This defines the beginning of regio superior. In other

terminology, regio superior is called CA1 (for Cornu

Assonis), the narrow transition zone is CA2, the region

inferior neurons in a layer form CA3, and the scattered

hilar neurons constitute CA4.

Dentate Gyrus Afferent Connections

The pattern of connections between the hippocampal

formation and its inputs and outputs is astonishingly simply

ordered, compared to more recently evolved cortex. Not only

are the major cell types arranged in compact layers, but

many of the inputs terminate in distinct layers along

particular portions of hippocampal neurons.

Entorhinal-Dentate Afferents.

One major source of input to the hippocampus is the

entorhinal neocortical region. Layer II pyramidal neurons


from medial and lateral entorhinal areas send axons into the

medial and lateral perforant paths, respectively (Bjorth-

Simonsen, '73). These "perforate" the dentate molecular

layer after crossing through region superior, then the axons

course orthogonally to granule cell dendrites, making en

passage, asymmetrical Type I synapses (Nafstad, '67; Laatch

and Cowan, '66) on at least 85% (Matthews et al., '76a) of

the dendritic spines of the granule cells. Terminals from

lateral entorhinal afferents stratify exclusively in the

outer third of the molecular layer, while medial entorhinal

input restricts itself to the middle third (Steward, '76).

The projection is bilateral, although only about 5% of the

entorhinal output is believed to innervate the contralateral

hippocampal formation (Steward and Scoville, '76). The

crossed lateral entorhinal projection has been observed to

terminate more heavily in exposed than the buried blade,

especially along the most distal region of the granule cell

dendritic tree (Wyss '81; Hjorth-Simonsen, '72). In

contrast, the ipsilateral projection terminates more heavily

in the buried blade. The crossed medial entorhinal

projection apparently terminates equally in both the buried

and exposed blade. With the Tima's metal-sulfide stain, the

separation of the termination zones of the lateral and

medial entorhinal afterents to the dentate is revealed by

the presence of distinct dark (lateral) and light (medial)

bands that appear to correspond exactly to the afferent


terminal labelling pattern observed with tracing techniques

such as autoradiography or HRP (Wyss, '81; Hjorth-Simonsen

and Jeune, *72). The underlying difference in these two

afferents is not known; few electrophysiological differences

have been confirmed (McNaughton, '80; Abraham and

McNaughton, '84). The restriction of immunohistochemical

staining for enkephalins in the lateral entorhinal axcn

terminal zone (Gall et al., '84), and for cholecystekinin in

the medial entorhinal termination area (Stengaard-Pedersen

et al., '83) indicates that some specializations exist.

Both entorhinal inputs are of particular interest to

neurobiologists because they exhibit a form of physiological

plasticity known as long term potentiation. In brief, it

appears that the proper pattern of electrical stimulation of

these aftterents can result in a lasting increase in synaptic

efficiency in this pathway (Swanson et al., '82). Even more

exciting is the observation that some rapid change in the

shape of the dendritic spines of the granule cells is

associated with LTP (Desmond and Levy, '81; Fifkova and

Anderson, '82; Fifkova and Van Harreveld, '77). The

implications of such processes for learning and memory

research are enhanced by the long history of behavioral

learning and memory research focused on the hippocampus

(O'Keefe and Nadel, '78).

Septal-Dentate Afterents.


A more diffuse input to the dentate comes from the medial

septal nucleus and the nucleus of the diagonal band (of

Broca). These axons travel through the fornix, alveus, and

fimbria and synapse sparsely throughout the molecular layer

(Milner et al., '83; Swanson and Cowan, '79); in a narrow,

dense band along the margin of the granule cell layer and

the inner molecular layer (Raisman, '66a); and densely

throughout the hilus. Because destruction of the cell

bodies of origin of the septal projection results in greatly

reduced hippocampal concentrations of acetylcholinesterase,

the degradative enzyme for the putative neurotransmitter

molecule acetylcholine, the septohippocampal projection is

believed to employ ACh as its transmitter (Storm-Mathisen,

'74). Histochemical staining for AChE has been thought to

be coextant with the location of cholinergic septal afferent

terminals as localized with degeneration (Storm-Mathisen,

'64) and autoradiographic tracing methods (Swanson and

Cowan, '79; Rose et al., '76), but this correspondence may

be less than has commonly been assumed (Milner et al., '84).

In general, however, AChE stains heaviest where

neuroanatomical tracing methods indicate the heaviest

septal-dentate termination: the supragranular band in the

proximal molecular layer. Evidence is also beginning to

indicate that some and maybe most of the septal afferents do

not use acetylcholine (Baisden et al., '84) and may use the

peptide substance P (Vincent and McGeer, '81).

Brainstem and Bypothalamic Dentate Afferents.

Less dense inputs to the dentate gyrus originate in

hypothalamic and brainstem nuclei. Many of these appear to

use catecholamines as neurotransmitters. A diffuse

serotonergic plexus arises from the dorsal raphe (Moore and

Halaris, '75), while a sparse noradrenergic input is sent

from locus coeruleus and less so from other brainstem

noraderenergic cell groups (Swanson and Hartman, '75; Pickel

et al., '74; Koda and Bloom, '77). A dopaainergic

projection may arise in the midbrain raphe as well (Reyman

et al., '83). The catecholaminergic afferents terminate

most densely in the hilus, followed by the granule cell

layer; only scattered fibers and terminals are observable in

the molecular layer. The most obvious hypothalamic

projection appears to come from the supraaammillary nucleus

(Dent et al., '83; Segal '79; Wyss et al., '79), and

terminates primarily along the most proximal portion of the

granule cell dendrites. Lateral hypothalamic and zona

incerta neurons demonstrating immunoreactivity for oX-MSH,

dynorphin, and angiotensin II have been reported to project

to both the hippocampus and spinal cord (Kohler et al.,


Hilar Afferents to Dentate.

The polymorph cells in the hilus are of special interest

as such for their morphological variety (Amaral, '78; Cajal,

'68; Lorente de No, *34) as for the fact that they send

associational projections to ipsilateral granule cells

anterior and posterior to their own location, and

commissural projections to the contralateral dentate, in a

similar pattern (Swanson et al., '81). They form

asymmetrical synapses on granule cell dendrites and

dendritic spines in the proximal one-third cf the molecular

layer (Laatch and Cowan, '66; Kiishi et al., '80; Gottleib

and Cowan, '73; Amaral et al., '80), in a band apparently

defined by dark Tina's aetal-sulfide staining (Amaral et

al., '80), and a conspicuous lack of staining in AChE

histochemical preparations (figure 3). Thus the

commissural/associational projection constitutes a major

afferent source to the granule cells. A band of positive

staining for the peptide cholecystekinin, distinctly

separated from the one marking medial entorhinal input, has

also been described in this inner zone (Stengaard-Pedersen

et al., '83).


POLYMORPHE Staining Patterns



Figure 3: Dentate Gyrus Tima's and AChE Staining Patterns

The C/A projection diverges with an interesting

topography. Associational (ipsilateral) axons tend to

spread toward the ventral exposed blade in the temporal

hippocampus, and toward the septal buried blade in the

septal half (Zimmer, '71; Hjorth-Simonsen and Laurberg, '77;

Fricke and Cowan, '78; West et al., 79). The ipsilateral

projection is greater than the contralateral, and some of

the hilar neurons may give rise to both ipsilateral and

crossing collaterals (Laurberg and Sorensen, '81).

Electrophysiologically, the C/A projection appears to be

excitatory to granule cells ipsilaterally, but the

commissural fibers evoke inhibitory responses that appear to

dominate any excitatory effects (Douglas et al., '83). In

addition to excitatory input from the ipsilateral hilus,

putative GABAergic inhibitory basket cells distribute

synaptic terminals in the granule cell and molecular layers

(Cajal, '68; Amaral,'78; Ribak, '83); these may also send

commissural afferents to the contralateral dentate.

Dentate Efferents

As for the output of the dentate, the granule cells emit

one or few axons that descend into the hilus- Once in the

hilus, they may send rare collaterals back into the

molecular layer (Frotscher and Zimmer '83), but most

characteristically begin to exhibit 4-8 micrometer

enlargements that have been shown with the electron

microscope to be huge synaptic terminals (Blackstad, '61).

At light microscopic magnifications, these give the mossy

fibers their name. The thin, unayelinated mossy fibers

(Laatch and Cowan, '66) project to and synapse en passage

with spines on the proximal dendrites of CA3 pyramidal

neurons, and also non-pyramidal hilar neurons. The

translucent quality of these bundled axons in the light

microscope has led to the naming of the stratum lucidum

along the proximal apical CA3 dendrites. They do, however,

send terminals into the basal dendritic tree of region

interior pyramids, especially along the more hilar region.

This infrapyramidal mossy fiber terminal band exhibits

considerable genetic variation and appears particularly

sensitive to toxins during development (West et al., '83).

Behavioral correlates related to perseveration have teen

observed following experimental manipulation of the

infrapyramidal terminal band (Schwegler and lipp, '83).

The synaptic terminals of the mossy fibers are of special

interest for the high concentrations of the element zinc

they contain (Crawford and Conner, '72). Its purpose there

is currently a matter for investigation, but it is known

that zinc is a cotactor for the enzyme glutamic acid

dehydrogenase (Storm-Mathisen 174), which would be

consistent with evidence that mossy fiber terminals employ

an amino acid, either glutamate or its relative aspartate,

as their neurotransmitter molecule. Zinc is present

throughout the hippocampus, at levels about three times

those of the rest of the brain (Fjerdingstad et al., '74a);

this in part accounts for the relatively dense Tiaa's

staining in the hippocampus (Haug, '74; Danscher, '81).

Fjerdingstad et al., ('74b) also report that lead is present

at tenfold the concentration of the rest of the brain, but

its role is unknown as well.

CA3 Afferents

The large CA3 pyramidal neurons are notable for their

thick, branching dendrites covered with spines. Their

proximal portions exhibit larger protruberances, called

excrescences. Like spines, these are specialized

postsynaptic structures onto which mossy fibers synapse

(Blackstad and Kjaerheim, '61). In fact, they often

perforate completely mossy fiber terminals, creating

toroidal synaptic contacts. More distally, the apical

dendrites narrow and spread, and are contacted by axons from

entorhinal areas, hilar neurons, and other CA3 neurons, both

ipsilateral and contralateral. Layer III entorhinal neurons

project to stratum lacunosum-moleculare (Steward and

Scoville, '76), while the extensive associational and

commissural projections terminate in oriens and radiatum

(Gottleib and Covan, '72, '73). Septal, hypothalamic, and

brainstem input is diffuse and probably varies in exact

pattern depending upon the anterior-posterior hippocampal


level. In general, medial septum projects most strongly to

strata pyramidale and oriens, and much less so to radiatum

(Raisman, '66a). Locus coeruleus sends its most dense input

to strata lucidum and lacunosum-moleculare, but diffusely to

the other layers as well (Swanson and Hartman, '75; Loy et

al., '80). The raphe innervates predominantly stratum

lacunosum-moleculare (Moore and Balaris, '75; Wyss et al.,


CA3 Efferents

CA3 axons, in addition to making coaaissural and

associational connections within regio inferior, reach

several other target areas. In fact, individual CA3

pyramids may project to five or more areas (Swanson et al.,

'81). As the axons descend through stratum oriens into the

fiabria, they branch into several collaterals. These then

travel separately to the lateral septum and subiculum,

perhaps bilaterally, in addition to the bilateral CA3

collaterals (Swanson et al., '80, 81). Axons to

intrahippocampal sites form the Schaffer collateral system,

passing via stratum radiatum to form asymmetrical spine

synapes in that layer and stratum oriens of CA1 (Gottleib

and Cowan, *72, '73). Those that travel to contralateral

destinations exit the fimbria to cross via the hippocampal

commissures located toward the septal pole. It has been

suggested that the Schaffer system projects in a lamellar


fashion, that is, from a narrow transverse slice of CA3 the

axons will remain oriented in a narrow plane with little

anterior-posterior dispersion. While the

electrophysiological evidence for this organization is

persuasive, the difficulty in differentiating the

longitudinally oriented associational system from the

transversely oriented associational system has precluded

anatomical resolution of this question (O'Keefe and Nadel,


CA1 Afferents

The last subregion of the hippocampus is the regio

superior (Cajal, '68), or CA1 (Lorente de No, '34). These

neurons are smaller and have finer dendrites than the CA3

pyramids, but are similar in the tree-like shape of their

dendritic elaborations. In CA1 pyramidal cells, afferents

from ipsilateral CA3 form the Schaffer collateral system.

Ipsilateral regio interior terminals innervate both strata

oriens and radiatua, but appear to be more dense in the

latter (Swanson et al., '78; Laurberg, '79). Contralateral

CA3 sends a similar input to CA1, but synapses more densely

in oriens (Swanson et al., '78; Laurberg, '79). Other

inputs originate from entorhinal cortex, septum, and

brainstem loci that parallel those of CA3. Entorhinal

afferents stratify in stratum lacunosum-soleculare (Steward

and Scoville, '76). In addition, an input from thalamic


nucleus reuniens to stratum lacunosum-moleculare has been

described (Herkeuham, '78; Riley and Hoore, '81).

CA Effterents

These neurons project, via the alveus, to the lateral

septum, to the subiculum, and to medial and lateral

entorhinal areas (Hjorth-Simonsen, '73; Swanson and Cowan,

'77; Heiback and Siegel, '77; Raisman, '66b), completing

circuits that originate and end there. Thus the basic

anatomy of the hippocampus is two loops. One can be thought

of as originating in entorhinal cortex, relaying to dentate

granule cell, then to CA3, to CA1, and back to entorhinal

cortex. The other begins and ends in the septum and has the

sane interposed hippocampal neurons. Of course, the actual

anatomical organization is considerably more complicated,

but this model serves as a useful conceptual simplification

that is entirely adequate for the purposes of the

experiments described here.

Reactive Synaptogenesis in the Dentate Gyrus

The dentate gyrus is a useful structure for studying CNS

synaptic reorganization because when any of the major

afferent sources is removed, a compensatory restoration of

synaptic density can be observed to occur following

degeneration of the damaged axons and terminals. Because

the elimination of entorhinal afferents is technically

simplest, dentate reactive synaptogenesis has been best

described using this paradigm. The literature on dentate

reactive synaptogenesis will be considered only where

germane to the present experiment.

When the entorhinal cortex on one side of the brain is

destroyed by aspiration, electrolysis, or mechanical

separation, granule cells ipsilateral to the lesion lose

about 60% of their molecular layer synapses (85% of outer ML

synapses, Matthews et al., '76a). Among the early effects

that are observed are a decreased glucose utilization

(Steward and Smith, '80), which is noticeable by 1 day and

lasts about 5 days before returning rapidly to normal

between days 6 and 12. During this period, several pre- and

postsynaptic changes occur. The presynaptic terminals

degenerate and the debris is rapidly phagocytosed by

nonneuronal cells, probably aicroglia, which increase in

number by about 3 days postlesion (Matthews et al., '76a;

Rose et al., '76; Cotman and Radler, '78). Astroglia

hypertrophy and migrate to the deafferented zone by 3 days

postlesion (Matthews et al., '76a; Rose et al., '76; Cotman

and Nadler, '78). These reactive glia are evident until

about 15 days postlesion (Matthews et al., '80; Stora-

Mathisen, '74), but prevention of their proliferation does

not appear to be detrimental to the synaptic reorganization

that occurs (Avendano '83). Postsynaptically, granule cell

dendrites undergo a rapid change in shape during the first


few days (Steward and Caceres, '83). From the margin of the

deafferented zone outward, they appear to wilt and sag, as

if they were normally provided with structural support by

their afferents. Protein precursor incorporation increases

during the time reactive synaptogenesis is taking place

(Fass and Steward, '83). A reorganization of the

microvasculature appears to occur such that the preferred

orientation of capillaries parallels the granule cell layer

rather than the perpendicular dendritic axis (Scheff et al.,

'78); the significance of this is unknown. Degeneration

products are cleared by about 10 days postlesion, but

synaptic density does not reach normal levels until about 30

days after the lesion (Matthews et al., '76b).

New presynaptic terminals form from axons originating in

the contralateral entorhinal cortex (Cotman et al., '77;

Steward et al., *74, '76; Zimmer and Hjorth-Simonsen, '75);

these arise from neurons undistinguishable in type from

those that provide the ipsilateral afferents (Steward, '76;

Steward et al., '74; Steward and Scoville, '76; Steward and

Vinsant, *78). The contralateral entorhinal cortex normally

contributes about 5% of the synapses in the unlesioned

animal, so the proliferation of replacement synapses by the

intact crossed pathway represents what is known as homotypic

(same type) reinnervation (Cotman et al., '81). New

presynaptic endings also arise from the associational and

commissural afferents from the hilar polymorph neurons,

which expand the AChE C/A stain zone in both an absolute and
relative fashion (Lynch et al., '76) and from septal
afferents (Stanfield and Cowan, '82). The latter withdraw
from the inner molecular layer C/A zone, and form a compact
terminal band in the outer molecular layer. The overall
result of this reorganization, as schematically summarized
in figure 4, is an overall shrinkage of the molecular layer,
an expansion and lightening of the AChE C/A zone, a
contraction and darkening of the cholinesterase-positive
outer molecular layer terminal zone, and possibly an
expansion of the Timmas C/A band (Amaral et al., '80).
Photographs of these effects are presented in Chapter 3
(figures 39, 40). ENTORHINMAL


Figure 4: Schematic Summary of Dentate Reactive


The new presynaptic terminals attach to both new and old

postsynaptic sites (Steward and Vinsant, '83), and show

normal morphology (Matthews et al., '76b). They are also

able to support electrogenic neurotransmission (Steward et

al., '73, '74, '76), and exhibit at least some of the

electrophysiological properties of the normal input (Harris

et al., *78; Steward et al., '76; Wilson et al., '79).

There is also evidence the a behavior lost after entorhinal

lesion recovers with a tine course similar to that of the

synaptic reorganization (Loesche and Steward, '77), although

it is not yet known to what extent dentate reactive

synaptogenesis mediates this recovery. In any case, there is

a growing body of evidence that this synaptic reorganization

is in some sense functional.

It is also known that the number of spines, and the

geometric branching characteristics, of these neurons are

affected adversely in mental retardation in humans, chronic

ethanol treatment in rats and mice, behavioral impairment in

an animal model of mental retardation related to prenatal

alcohol exposure, and age-related demented states videe

supra). Degeneration, and later regrowth, of spines and

dendritic branches, have been found to occur after a partial

deafferenting of these cells (Parnevelas et al., '74,

Goldowitz et al., '79; McWilliams and Lynch, '78; Matthews

et al., '76a,b; Steward and Vinsant, '83; Steward and

Caceres, *83), like some, but not all, CYS neurons

(Frotscher, '83). Change in granule cell spine shape has

been associated with electrophysiological synaptic

efficiency increases that occur with certain stimulation

procedures, and which may be related to the formation and/or

saintainance of memories (Desmond and Levy, '81; Fifkova and

Van Harreveld, '77; Fifkova and Anderson, '82). Thus these

cells exhibit some natural morphological plasticity that may

serve some normal functional purpose, in addition to being

provokable via deafferenting lesions.

The Liquid Diet Chronic Ethanol Consumption Model

Over the past decade, an animal model of chronic alcohol

consumption has been developed in the University of Florida

laboratory of Drs. Don Walker and Bruce Hunter to

specifically address issues not currently amenable to

experimental analysis in human alcoholic patients. Age,

nutritional intake, duration and amount of consumption, and

to some extent, genetic heterogeneity, can be more tightly

controlled and/or monitored than is possible with human

beings. Experimental evidence obtained through the use of

this model, and others like it in use in other laboratories,

demonstrates its relevance to the study of the biology cf

alcohol. Perhaps the single most important result of

employing this animal model of chronic alcohol consumption

is the finding that CNS pathology following long-term

dietary alcohol intake is neither caused by concomitant


malnutrition nor preventable by proper nutrition (Walker et

al., '81; Freund, '82). This has helped to begin to put to

rest a longstanding dogma that has hindered research

progress in treating alcohol related disabilities. While

nutrition can evidently play an interactive role, data from

animals demonstrate that ethanol itself or its immediate

metabolites possess neurotoxic qualities. Both behavioral

and neuroanatomical findings obtained with this model

closely resemble those observed in human alcoholics (Walker

et al., '81).

Male Long-Evans rats were obtained at approximately 30

days of age and housed individually in stainless steel cages

in a colony room until they reached 60 days of age. During

this acclimation animals received ad libitum Purina Lab Chow

and tap water. Animals were paired by weight and pairs were

divided to form two diet treatment groups at the end of

acclimation. The alcohol diet consisted of the nutritional

formula Sustacal (Mead Johnson Co.), containing

progressively 8.1-9.8 1 v/v ethanol over the 20 weeks

animals were exposed. These ethanol concentrations provide

35-39% ethanol-derived calories in the diet, in increments

of 1% per month to compensate for the development of

tolerance. To this formula were added several times the

recommended daily requirements of vitamins and minerals (3.0

g/l Vitamin Diet Fortification Mixture, 5.0 g/1 Salt Mixture

XIV, Nutritional Biochemicals, Cleveland), as described

previously (Walker and Freund, '71). A control diet was

prepared by substituting sucrose isocalorically for ethanol.

Alcohol-receiving animals were given 120 al of the freshly

prepared diet each day; their weight-matched control animals

were given as such as their counterparts drank the previous

day. Thus, for 20 weeks, alcohol and sucrose animal pairs

had equivalent vitamin, mineral, protein, fat, carbohydrate,

and total caloric intake; the only difference in diet being

the presence or absence of ethanol. Individual daily

consumption data were collected throughout the diet

treatment to calculate average daily ethanci consumption,

and animals were weighed and inspected for general health

weekly. Prior to sacrifice of the animals, neutral code

numbers were assigned by an otherwise uninvolved third party

to assure the unbiased collection of all measurements.


Design and Methods

Previous studies in our laboratory have reported

electrophysiological evidence that a rearrangement of some

of the synaptic input to both the CA1 and dentate neurons

may occur consequent to chronic ethanol treatment (Abraham

et al., '84; Abraham and Hunter, '82; Abraham et al., '81).

In brief, commnissural afferents to CA1 radiatum, and lateral

entorhinal afferents to dentate, appeared to be reduced in

spatial distribution, number, and/or synaptic efficiency.

In addition, spine loss has been reported to occur after

chronic ethanol exposure in these neuronal types. Thus, we

chose to measure the linear density of dendritic spines

within several discrete regions of CA1 and granule cell

dendritic trees, and compare the spine densities measured in

animals sacrificed immediately after a 20 week ethanol

exposure with densities from animals allowed 20 weeks post-

exposure for any natural recovery to occur.

One ethanol and one sucrose control group were sacrificed

immediately following the 20 week exposure period, and one

ethanol and sucrose group were maintained for 20 weeks more


on ad libitum lab chow and water in individual cages before

being sacrificed. The two by two factorial design of this

experiment thus provides the means to test for initial

ethanol effects, residual effects, and any recovery that

might occur during abstinence. The use of two sucrose diet

control groups allows any effect of aging to be factored

into the analysis.

Golqi Histology

The rapid Golgi method of Scheibel and Scheibel ('78) was

modified to produce reliable staining in the hippocampal

formation. In brief, pentobarbital-overdosed animals were

transcardially perfused first with isotonic phosphate-

bufferred saline, followed by 1%:1.25% glutaraldehyde:

formaldehyde in phosphate buffered saline, followed by a

stronger 4%: 5% aldehyde solution. The brains were

removed, trimmed square at the frontal poles and behind the

cerebellum, weighed, and photographed on graph paper for

later measurements. The extracted whole brains were then

refrigerated in the second fix overnight, cut into coronal 2

am slabs the following day, and fixed until 48 hours from

sacrifice. They were then placed in aqueous 0.33% ossium

tetroxide, 2.7% potassium dichromate solution for 48 hours,

during which they were turned over once. Slabs were then

rinsed in 0.75% silver nitrate, and placed in fresh silver

nitrate for 24 hours in light-tight boxes. Dehydration and


celloidin embedding were carried out as quickly as possible.

Mounted blocks were cut on a sliding microtome at 80 to 120

microns and mounted on glass slides with Eukitt synthetic

medium (Calibrated Instruments, Ardsley, NY).

Gross Brain Measurements

Photographs of the freshly extracted, trimmed, coded

brains were measured with the aid of a Numonics 1224

digitizer. The scaled length, width, and area of each

cerebral hemisphere and the cerebellum was collected. These

measures correspond roughly to the type of measurements that

are made on CT scans to diagnose cerebral atrophy (Cala and

Mastaglia, '81; Ishii, *83).

Dendritic Spine Counts

Using oil immersion optics and a final magnification of

2250X, representative dendritic segments and the position of

spines thereon were traced using a drawing tube attached to

an Olympus BH-2 microscope. Dividing the number of spines

per segment by the scaled segment lengths, obtained using

the digitizer, yielded spine density estimates. The number

of samples necessary to produce per-animal estimates with

sufficiently low variability was calculated by plotting the

population standard error estimate by number of samples

(figure 5). Standard error estimates from our animal

samples averaged less than 6% of the mean density estimates.




S. .6
25 M. .5

0 2 .4
M 15 .3
K 10 .2

5 .1

1 2 3 4 5 6 7 8 9 10

Figure 5: Spine Density Variability as a Function of the
Number of Segments Counted

With further error checking it was found that the

critical measure for estimating linear density, i.e.,

segment length, varied less than 1% on two measurement

replications made for one animal. Across counting days,

spine density estimates varied by an average of less than

5%. Six animals were selected at random for measurement of

area, width, and height of the face of both tissue pieces,

from photographs taken at each stage in the histological

procedure. Because a high degree of variability in tissue

shrinkage was attributable to initial dehydration after the

staining steps, it was not possible to normalize each slab

by amount of shrinkage and some error in segment length can

be expected from this. Little or no distortion of the

tissue size was detected before the dehydration steps. As a


result, the gross brain size measurements are not likely to

contain this error. Mean tissue shrinkage in the slide-

mounted tissue was found to be about 151 relative to the

initial size of the fixed brains.

Dendritic segments selected for drawing were without

obvious distortion, cut or broken surfaces, and uneven or

overimpregnation. In general, the first n segments

encountered that were subjectively judged to be

representative for the section under observation were used,

with n being the number of samples to be collected for a

particular area. For all areas, not more than three samples

were obtained from the same neuron, and one sample per

neuron was considered desirable when enough cells were

stained to criterion. Optimal segments were stained red to

rust in color, with some transparency, and had 25-50

micrometers of dendritic branch in a relatively flat

orientation to and situated near the top of the tissue

section. This insured a narrow depth of field difference

and maximum resolution of individual spines. Spines were

counted wherever a discrete bulge from the dendritic surface

was observed, but only where connection to the dendrite

could be ascertained. In this respect, we say have missed

counting some of the spines with a "lollipop" shape and a

spine neck at or beyond the resolving power of the light

microscope. Spine localization onto segment tracings was

made once with the microscope condenser adjusted to the


numerical aperture of the principle objective, and once with

the condenser adjusted to decrease the depth of field and

increase image contrast, in order to check the decision to

identify the spines. Segments selected did not branch

within the distance measured, did not cross other

impregnated dendrites, glial processes, blood vessels, or

histological artifacts, and were chosen with an eye to

avoiding possible overlap with adjacent sampling regions.

Samples were also not taken on dendrites exhibiting any

grossly visible abberations of shape or size, and were thus

conservatively chosen as representative of the populations

under study.

For the CA1 pyramidal neuron, five samples were sought in

each of four regions of the dendritic tree (figure 2):

strata oriens, radiatum 1 proximall 1/2), radiatum 2 (distal

1/2), and lacunosum-moleculare. This last region was

defined by a conspicuous change in the density of spines and

the size and orientation of the dendrites. A marked

decrease in spine density and a turn to orient parallel to

the hippocampal fissure were observed. In the dentate gyrus

molecular layer, three regions were sampled, and 10 segments

per region per animal were sought: inner, middle, and outer

third, which correspond to the stratification of afferent

axons from three major sources of input to granule cell

dendrites (commaissural/associational, medial, and lateral

entorhinal cortex)(Anaral et al., '80). Division between


sampling regions was made subjectively, but conservatively,

such that it would be very unlikely for a spine density

estimate to be attributed to the improper sampling region.

within a sampling region, dendritic segments were selected

to form a sample corresponding subjectively to the average

size of dendrites in that area, e.g. in the stratum

radiatus region, the main shafts of CA1 pyramids were

avoided in favor ot the more numerous smaller caliber

branches. Sections to be used were taken at random through

the dorsal hippocampal formation, since the initial cutting

of the brain into slabs influenced the location of the best-

impregnated sections in the final product.

All data were analyzed with a two way analysis of

variance, for diet treatment, recovery time, and their

interaction. Additionally, means within diet or recovery

groups were tested using the student's t tests and one-way

analyses of variance. Analyses were performed using SAS

(Statistical Analysis Systems, SAS Institute Inc., Cary,



for clarity and convenience, the following shorthand will

be used to designate the individual treatment groups: AO,

alcohol/no recovery period; A20, alcohol/20 week recovery;

SO, sucrose/no recovery; S20, sucrose/20 week recovery.

Gross Measures

The gross measures data and statistical results are

summarized in Table 1. Body weight was not significantly

different between diet groups at either recovery time.

However, mean body weight for the AO group was slightly

lower than for the SO group, and body weight for the A20

group was slightly higher than that of S20 animals. Thus it

is possible that some rebound and overshoot of body weight

occurs in alcohol animals, although sample size may merely

have precluded the statistical significance of other group

differences. Both diet groups significantly increased mean

body weight during the recovery interval.

Brain weight was higher in both alcohol groups than their

respective sucrose control groups (n.s.). Both alcohol and

sucrose groups had significantly higher brain weights at 20

weeks than at 0 weeks. A significant 7% difference in the

brain weight of 0 and 20 week alcohol animals was found,

while 0 and 20 week sucrose group brain weights differred by

only 3% (n.s.).

Similarly, animals exposed to either diet treatment

showed significant increases in total cerebral hemisphere

area after 20 weeks. The means for the 0 week groups are

nearly identical, while the A20 group was slightly higher

than the S20 group (n.s.). The A20 animals were found to

have almost 9% more hemisphere area than AO animals, while

SO group hemisphere area differed from S20 group hemisphere


area by only 5%. Left hemisphere area patterns are

virtually identical to the combined hemisphere results; the

right cerebral hemisphere follows the same pattern but the

interaction between diet and recovery was significant.

Cerebellar area, which was only slightly lower in AO

animals than SO controls (1%), also increased more (13%)

over the 20 week recovery period in alcohol animals than in

sucrose controls (11%), although this interaction did not

reach significance. Although many of these effects failed

to reach the .05 significance level, their uniformity

supports the possibility that some grossly observable

recovery is occurring during abstinence.


Gross Brain Measurements and Weights

SO u. S20 SO v. AO AO v. A20 S20 v. A20
mEMISPHERE N=15 N=8 N=20 N.13 Diet: P<.9708
PkEA 1.868 1.962 1.867 2.005 Recovery: P<.003
(sm. cm) (.034) (.032) (.024) (.101) D*R: P<.1007
,.s. n..s n.s. P<.0007
.5769 .7722 .1517
LEFT HEM. N=15 N=8 N=20 N-13 Diet: P<.8623
kEA .927 .994 .926 1.003 Recovery P<.0027
(.018) (.015) (.015) (.013) D*R: P<.6985
n.s. n.s. n.s. P<.0073
.1455 .4326 .1017
;TIGHT HEM. N=15 N-8 N=20 N.13 Diet: P<.9938
AREA .94 .968 .941 1.002 Recovery: P<.1057
(.0231 (.025) (.018) (.018) D*R: P<.0434
n.s. n.s. n.s. P<.0182
.5398 .6146 .4739
CEREBELLAR N=:5 N-8 N=20 N-13 Diet: P<.5542
AEA .592 .662 .583 .657 Recovery: P<.0001
(.011) (.014) (.010) (.019) D*R: P<.3250
n.s. ** n.s. P<.0096
.9606 .3588 ** P<.0006
BRAIN WT. N=9 N=9 N11 N.12 Diet: P<.1861
,,: 1.604 1.647 1.622 1.709 Recovery: P<.0428
(.042) (.056) (.027) (.051) D*R: P<.4879
n.s. n.s. n.. n.s.
.4197 .2357 .0503 .1776
BODY NT. N=17 N=9 N=20 N*13 Diet: P<.6914
ramsm) 608 651 576 672 Recovery: P<.0008
(9) (20) (17) (22) D*R: P<.0852
r.s. n.s. n.s. P<.0038
.1145 .3549 .2211

Figure 6: Dendritic Spines on Golgi-stained Dendrite

Spine Density Heasures

As figure 6 shows, the rapid Golgi method demonstrates

the prominent spines along all levels of hippocampal neuron

dendrites. A total of 1790 dendritic segments were sampled

from 40 animals. The results of the spine density

comparisons complement the gross measures data.

CAl_aine Densit-.

Ethanol did not result in a significant lowering of spine

density in any of the CA1 zones, although a consistent

reduction was observed for the mean spine density in each

(table 2). In CA1 pyramids, the basal dendritic spine

density, in stratum oriens, was slightly lower in the 10

animals than SO controls (6%). This pattern held in the

apical dendritic tree; proximal stratum radiatum density was

3% and distal radiatum 10% lower than the same regions in


sucrose control animals, and a 6% reduction was observed in

stratum lacunosua-moleculare. Significant recovery period

(20 week) effects were observed only in proximal radiatus,

according to the analysis of variance for all 4 groups.

However, t-tests revealed that while no significant

differences in mean spine density were found between the SO

and S20 groups, in alcohol-treated animals the density means

in all CA1 areas (except lacunosum-aoleculare, P<.0749) were

significantly higher in 20 week than in 0 week animals. In

stratum oriens, A20 week group density was 13% higher than

in AO animals. Proximal radiatum density was 14%, and

distal radiatum 19% higher in 20 week than in 0 week alcohol

subjects, and a 19% greater density also was found in

stratum lacunosuma-oleculare. In fact, with the exception

of this last region, A20 CA1 linear spine densities were

higher in all regions than either of the sucrose control

groups! Conservative comparisons of the A20 densities with

the higher of the two obtained control values indicate at

least a 7% greater density in oriens, proximal, and distal

radiatum than in controls. Interactions between diet and

recovery time were significant, however, only in distal

radiatum. These results are presented in graphic form in

Figure 7.

Dentate Granule Cell Densities.

The dentate granule cells were found to be much

differently affected by CET (Chronic Ethanol Treatment) than


CA1 Pyramidal Neuron Spine Densities



SO u. S20 AO v. SO AO u. A20





































Diet: P<.9532
Recovery: P<.1568
D*R: P<.160f
n.s. P<.0345

Diet: P<.7199
Recovery: P<.0148
D*R: P<.1751
n.s. P<.0088

Diet: P<.8106
Recovery P<.0244
DR: P<.0294
** = .0303




Diet: P<.222E
Recovery: P<.0911
D*R: P<.6377

Diet: P<.6952
Recovery: P<.0844
D*R: P<.1377
* P<.0005

Diet: P<.695:
Recovery: P<.08A4
D*R: P<.1377
* P<.0079


A20 v. S20


1.5 .



S, I.o ,*.' -


1. 5


.75 T

5-0 +A ; ..'- a

SO S20 AO A20

Abscissae, linear spine density per micrometer, +/- S.E...
SO, stratum oriens; SRI, proximal stratum radiatum; SR2,
distal radiatum; SLH, stratum lacunosum-moleculare. SO,
S20, AO, A20 as in text.

Figure 7: CA1 Pyramidal Neuron Dendritic Spine Densities

the CA1 pyramids (table 3, figure 8). While significant

diet effects were detected only for the inner and middle

thirds of the dentate molecular layer, the mean linear spine

density along granule cell dendrites was uniformly highest

in the AO animals, lowest in the sucrose groups, and

intermediate in the A20 group. This increase in spine

density in granule cells was an unexpected finding.

Significantly more spines per micron were found in the AO

group than the SO group, in the commissural and medial

entorhinal terminal regions, and this difference disappeared

when the 20 week groups were compared.

Each of the three molecular layer regions sampled showed

increased density in AO animals: 6% in the outer molecular

zone, 14% in the middle, and 17% in the inner third, when

compared against SO values. A 6% increase in granule cell

soma area was found as well. Analysis of variance revealed

a significant diet effect in the C/A zone and the medial

entorhinal zone but not in the outer molecular layer.

Recovery effects and interactions failed to reach

significance. After 20 weeks, alcohol values returned

toward the sucrose levels. Although not statistically

different, the A20 mean spine densities were still slightly

higher than those of the S20 group. Dendritic spine density

in the outer third decreased 2%, the middle zone 6%, and the

inner region 7%, compared to 0 week alcohol animals. Thus

they all remained higher than either of the control groups


Dentate Granule Neuron Spine Densities

SO v S20

(sq. us > 106 .3
(8 9)

(inner third)

(middle third)

(outer third)

1 570
( 047)

1 541
( 061)

1 544
( 083)

LAYER (.058)



AO v. SO AO v. A20 A20 v. S20


1 596

1 470
( 158)

1 543
( 141)

1. 538
( 145)

n. s.



(4. 1)

1 835

1. 750


N-I 3
1 .753

n s.

n .

n. s.

n .



1 .713

1 653
( 130)

1 .612
( 072)

1 .659

n .
. 5045

n. .




Diet: P(.1764
Recovery: P(.5843
DR: P(.5487

Diet P(.0333
Recovery: P(.5039
D*R: P(.4419
* P(,0254

Diet: P(. 0436
Recovery P(.3926
D*R: P( 8972
* P( 0413

Diet P( 3963
Recovery: P( 8709
D*R, P( 8947

Diet: P( 0533
Recovery- P(.4087
D*R: P(.7174
* P( 0425

Hippocampal dendritic spine density results: group means +- S.E.M. by sampling region. The
rightmost column displays the results of two-way ANOVAs for diet treatment, recovery time and their
interaction. The four columns Interposed between the group means show the results of one-way ANOVAs
and T-tests performed within diet groups or recovery times. Asterisks mark effects significant above
the .05 level, but note that several other differences approach this level.

Abscissae, linear spine density
C/A, commissural/associational;
entorhinal terminal zones. SO,

per micrometer, +/- S.E.M..
MEC, LEC, medial and lateral
S20, AO, A20 as in text.

Figure 8: Dentate Granule Neuron Dendritic Spine Densities


at both recovery times. No change in granule cell soma area

was detected across the recovery interval. Thus, although

the initial effect of CET was opposite in these neurons from

CA1, at the end of a 20 week abstinence interval dendritic

spine densities for both neuronal types were more like those

of normal animals than they were immediately after the

exposure period.

When the counting regions were collapsed by cell type,

the only significant finding in CA1 pyramids was the

difference between spine densities of AO and A20 groups. In

granule cells, spine density was significantly higher in the

AO than the SO group, but this difference disappeared over

the 20 week recovery interval.

After these results were analyzed, qualitative

observations on the appearance of dendritic architecture

were followed up in the decoded tissue. While no valid

quantitative data support these observations, it can be

noted that elongated dendritic spines, dendritic

varicosities, increased tortuosity of dendritic paths, and

an increase in the size of stained astrocytes may be

associated with the consumption of ethanol for 20 weeks.

However, gross pathology was not often detected in the

alcohol animals, and noticeably abnormal dendrites were only

occasionally seen.


Experimental Considerations

Abberant dendritic changes following chronic ethanol

exposure have been reported before. In one of the earliest

reports, Riley and Walker found substantially greater spine

loss in both CA1 and granule cells, in female mice exposed

for 4 months. Since the same liquid diet paradigm was used

in that study as in the present one, our failure to

replicate their results can only be explained as a function

of the use of a different Golgi method, a different species,

the use of different measurement techniques, or some

difference in the maintainance of the experimental animals.

It is widely known that the different Golgi variations can

produce very different pictures of the dendritic

architecture of neurons, especially in the representation of

spines. Although the Golgi-Kopsch variant used by Riley and

Walker does impregnate spines well, it was decided during

pilot studies that the rapid Golgi variant yielded better

and more consistent spine impregnation and definition in our

hands. (It is not recommended for use where the branching

or lengths of dendrites are required, as unobstructed views

of completely impregnated neurons occur only infrequently.)

This raises the issue of whether Golgi stains accurately

reflect the number of spines in the tissue, one of several

methodological considerations discussed in detail by Mervis

(*81). If some deafferentation of hippocampal neurons

occurs with chronic ethanol treatment, and deafferented

spines are less amenable to Golgi impregnation (Powell,

'67), then spines might be present but not become stained.

We consider that a change in stainability reflects some

change in physical or chemical properties of the spine that

is indicative in its own right (Riley and Walker, '78). It

is possible, then, that the methods used by Riley and Walker

(*78) led to an overestimation of spine loss, if their Golgi

picture reflected not a loss but a difference in

stainability or visibility of spines in the alcohol animals.

Phillips and Cragg ('83) present some data that supports

this possibility; spine head area is significantly decreased

13% in the alcohol treated C57 black mouse, even after 4

months for recovery, to a size less than the wavelength of

blue visible light. In alcohol treated mice, therefore,

spines may simply not be resolvable using light microscopy.

Spine head area, estimated at .149+-.003 um squared in souse

basilar CA1 dendrites (Phillips and Cragg, '83), has been

estimated to be .72+-.02 us squared in the guinea pig,

(Turner and Schwartzkroin, '83) which is probably closer to

the size in the rat, and greater than the wavelength of

visible light.

Other differences between rats and mice could also

account for some of the differences in results. Mice have a

shorter life expectancy than rats, and a 4 month exposure

period in mice might be considerably more detrimental than a


longer duration in rats. Perhaps more significant is that

the mice in the previous study consumed an average of almost

25 g/kg ETOH/day, in contrast to less than 15 g/kg/d in the

rats. Although mice have a higher capacity for metabolizing

ethanol (Wallgren, '70), it is not understood how the doses

correspond between the species. If the relative exposure or

exposure duration was greater in mice in the Riley and

Walker study, the present results might not seen

inconsistent. Such a possibility is supported by data from

Kunz et al., (*76) who found an increase in rat CAI spine

density (.94/um v. 0.71, apical; 0.94 v. 0.65/um, basal),

using a chronic ethanol consumption model with a shorter

exposure period (52 days) and different dosing procedure

(6.7 g/kg/d intubation, once per day) than that used in the

present study. It is possible that dendritic degeneration

and recovery might be sensitive to ethanol exposure

parameters, considering that variables as apparently subtle

as the quality of the sensory environment (Connor et al.,

'80; Juraska, '84) may affect dendritic morphology.

Another possible factor might be the sex of the

experimental animals. In humans, sex differences in

psychometric performance by alcoholics have been described

(Fabian et al., '84; Ellenberg et al., '80), and sex

differences in dendritic spine changes consequent to

experimental environmental conditions have been observed

(Juraska, *84). Female mice were used by Riley and Walker

('78), while the present study used sale rats. The last

major difference between our rats and Riley and Walker's

mice is that the rats were individually caged, while the

mice were housed in groups. Housing, in combination with

other environmental conditions, has been shown to affect

dendritic spine density in mice (Connor et al., '80).

With the small and statistically non-significant

differences we find in many of our across-group means, it is

necessary to consider the significance of these effects.

Two points should ease the reader's mind. First, the

statistical analyses were made using animals within groups

rather than spine counts within groups. Good arguments can

be made for either approach; it was decided that the more

conservative method of averaging spine counts within animals

would be used. This may have sacrificed some of the

statistical weight of the results, since more degrees of

freedom would have been used in the analyses of variance if

spine counts per group were evaluated. Second, in other

studies of the effects of chronic ethanol treatment on

dendritic spine and branching parameters in cerebellar

Purkinje cells, it has been found that a non-significant 15%

decrease in the number of branches, and a significant 22%

loss of spines, after 3 months exposure (Tavares et al.,

'83a,b). Greater losses were consistently observed at 6

months exposure, with both measures reaching statistical

significance. In several other studies, alcohol treated

animals exhibit apparently detrimental although

nonsignificant differences in quantitative dendritic

parameters (Pentney, '82; Phillips and Cragg, '83; cHMullen

et al., '84), suggesting that initially nonsignificant

changes tend to become progressively more pronounced with

longer exposures, and may be considered to be real effects

even though failing to reach acceptable levels of

significance at early time points. The uniformity in the

direction of the differences, and the fact that many of our

hypothesis testing probabilities approach the .05 level,

argue that this sort of continuum may exist for the neurons

we measured. In addition, most of our nonsignificant

results still have a better than even chance of being real

effects. Although strict statistical interpretation of the

individual tests does not allow us to claim either an

initial ethanol effect in CA1 or a recovery effect in

granule cells, neither does post-hoc consideration of all

the results conclusively negate these possibilities. While

future experiments may have better statistical success by

using other durations of both diet and recovery, the present

results are supportive, if not unequivocally, of the

hypothesis that dendritic spine damage and recovery can

partly underlie CNS pathomorphology and its amelioration.

Our selection of 20 weeks as the sole diet duration was

made on the basis of earlier studies in our laboratory

indicating that a significant 15-20% loss of granule cells,

and a significant 10% loss of CA1 pyramids, was present


following this exposure in rats (Walker et al., '80), and

that a 50-60% spine loss occurred in dentate granule cells

and CA1 basilar dendrites in mice exposed for 16 weeks

(Riley and Walker, *78). Since it is a general principle of

toxicology that, with chronic sublethal exposures, cells

eventually reach a point of no return from which recovery is

impossible (Bridges et al., '83), we wanted to insure that

we used a diet period sufficiently long to result in

detectable dendritic pathology, but not so long that

recovery might be impaired. In contrast to the Riley and

Walker spine loss data, 18 weeks of ethanol consumption has

been reported by two groups not to affect spine density in

the rodent hippocampus at all, when electron microscopy is

used to estimate spine density (Phillips and Cragg, '83; Lee

et al., '81), although the average daily consumption of

ethanol by animals in the study that used rats may have been

less than our rats received (Lee et al., '81). Furthermore,

Lee et al. made their measurements in proximal radiatum,

where we find the smallest spine density changes.

A second possible source of apparent discrepancy between

spine estimates of different researchers is the finding that

spine density decreases on cerebral cortical pyramidal cells

by almost 50% between 90 and 414 days of age in normal rats

(Globus et al., '73), and then increases again by 630 days

to the levels of young animals. Similarly, primates undergo

an initial postnatal overproduction of spines (Duffy and

Rakic, '83), peaking at about 6 months of age, and then

spine density returns to neonatal levels by adulthood. Thus

the age of the animal at the time of sacrifice may be

especially important, and in a species (and possibly strain)

specific manner, to comparisons of spine density across

experiments. The last major possible cause is a difference

in the type of Golgi method used. Wide differences in spine

representation exist across the many Golgi procedures

available, and some variability can occur even within

individual stained blocks of tissue.

Other sources of error in estimating morphometric

parameters in populations from relatively small groups of

animals include the normal genetic heterogeneity of a

particular strain of animal. We consider that the

variability between our sucrose control groups is in part

indicative of the degree of heterogeneity encountered across

batches of Long-Evans rats obtained at different times.

Although all of our animals were obtained from the same

source, because of the labor-intensive nature of our liquid

diet consumption model it was necessary to use three sets of

animals obtained at different times. We can only explain

the intergroup variability by speculating that some genetic,

seasonal, or other unknowable source for this variability

exists apart from our own experimental error. Thus we

maintain that we in fact do see two small yet consistent

experimental effects that would be better statistically


validated if larger groups of animals were used. First, the

chronic dietary consumption of ethanol, in the face of

adequate nutrition, results in a decrease in dendritic spine

density in CA1 pyramidal cells and an increase in density in

dentate granule cells, in the sale Long-Evans rat. Second,

these changes at least partially revert during a post-

ethanol recovery period.

Estimates of Total Neuron Alterations in Spine Number

Using published data for the total length of the

dendritic trees of the two neuronal types measured in this

report, and the effects of chronic ethanol treatment

thereon, estimates can be derived of the degree of change in

connectivity that results from density alterations as found

here. For granule cells, protatly the best estimates for

total dendritic length are those of Desaond and Levy

('82) (rat) and Turner and Schwartzkroin ('83)(guinea pig),

who arrive at figures of 3,662+-88 and 3,700+-300 us,

respectively. Using our estimated average (control animal)

spine density for granule cells of 1.56/um, and a total

length estimate of 3,662 us, about 5,700 spine synapses are

probably observable in Golgi material per granule cell.

This is lower than estimates of 10,000 molecular layer

synapses per granule cell in the adult Wistar rat (West and

Anderson,*80) derived from electron microscopic

observations. Using a figure reported by Matthevs et al.


('76b) that 85% of synapses in the molecular layer occur on

spines, assuming 1 synapse per spine, and accepting that

only about .3 to .4 of the spines can be counted (Feldman

and Peters, '79, figure 6, Sd=.5, dendritic diameter<=4 un,

spine length=1 un), then a derived estimate of the true

number of spines per granule cell comes to about 16,000 and

the total number of molecular layer synapses per granule

cell may be as high as 19,000.

For CA1 pyramids in the rat, total dendritic branch

length has been calculated to be about 4,885 um at 90 days

of age (Pokorny and lamamoto, '81), (but 8,400 un in guinea

pigs (Turner and Schwartzkroin, '83)). Using their regional

total branch lengths with our regional spine count

estimates, we calculate that about 10,000 countable spines

occur along a normal CA1 pyramid of the control rats used

here. (Because of the greater range of dendritic diameter

in these neurons we do not attempt a corrected estimate of

the total number of spines per cell.) If a 10% decrease in

the branch length occurs after this duration of ethanol

treatment (McHullen et al., '84), and only a 10% decrease in

spine density on the remaining dendrites accompanies this

loss, then the CA1 pyramids would similarly support about

2,000 fewer countable spines per cell, at the end of 20

weeks of ethanol consumption. With the magnitude of

recovery of synaptic density that is reported here, CA1

pyramids in alcohol animals may gain 4-5,000 synapses during


the 20 weeks post-exposure, while sucrose animals may not

change significantly. Individual granule cells, in

contrast, might form about 3,000 more spine synapses with

their afferents, by the end of a 20 week ethanol exposure,

but lose about 2,000 of these during the post-ethanol


Electrophysioloqical Consequences of Long-term Ethanol

It night be expected that changes of this magnitude,

without even considering cell deaths, would noticeably

alter the electrophysiological synaptic transmission

characteristics recorded in response to a controlled

stimulation of afferent pathways; paradoxically, however,

significant effects have been difficult to detect for EPSPs

(excitatory postsynaptic potentials), or nearly any cther

measure of synaptic efficiency that has been obtained

(Abraham et al., '81; Lee et al., 81; but see Durand et al.,

'80 (impaired long-term synaptic potentiation)). However,

the degree to which an extracellularly recorded CA1

population action potential spike is increased in response

to a test stimulus applied to ipsilateral regio inferior

afferents shortly after a conditioning pulse (paired pulse

synaptic potentiation) has been found to be enhanced after

chronic ethanol treatment in previous studies in this

laboratory (in vivo, Abraham et al., '81; in vitro, Rogers

et al., unpublished data). The EPSP potentiation is

unchanged by the ethanol treatment. This indicates that

some adaptation of individual surviving synapses occurs in

the face of chronic ethanol treatment and withdrawal. Since

these animals were tested at least 8 weeks after exposure to

ethanol, this adaptation may include recovery of synaptic

structures. In support of the idea that spine density

recovery might be relfected in electrophysiological

recovery, Durand et al., ('84) described impaired long-term

potentiation in CA1 pyramidal neurons in chronic ethanol

animals before, but not after 2 month withdrawal periods.

In contrast to CA1, and possibly related to differences in

spine density effects, paired pulse potentiation of the

granule cell population action potential in response to

stimulation of the entorhinal-dentate pathway is diminished

in alcohol animals (Abraham et al., '84). Since the

postsynaptic potentials are unchanged in alcohol animals in

either neuronal type, the significance of the differential

spine density effects for understanding the differential

effects on the action potential is unclear. This electrical

event does not directly reflect synaptic (spine) properties

but the integration of synaptic potentials and their

conversion into action potentials. These contrasting

physiological findings could be an indication of specific

alcohol effects on interneurons. CA1 appears to have both a

greater population and variety of non-pyramidal neurons than

the dentate gyrus.

Possible Mechanisms of Spine Density Alterations

Somatic proximity.

The proxiaodistal pattern of spine density effects in

granule cells deserves some comment. Both the greatest

initial increases and the greatest subsequent "recovery"

occur proximally, and decrease progressively at sore distal

levels of the dendritic tree. Two possible explanations for

this pattern can be entertained. First, if the manufacture

and maintainance of spines requires a significant investment

of raw materials and energy by the soma, then those

dendritic regions close to the soma will be best able to

respond to metabolic influences on the soma: first served,

first cleared. Alternatively, the presynaptic axons

providing input to the granule cell dendritic tree, some of

which are known to sprout in response to deafferentation of

the outer dendritic tree (Cotman et al., '81), may exert

some control over the number of postsynaptic sites available

as the turnover of the terminals is affected by CET.

Factors as simple as the proximity of afferent sonata may

explain a more robust effect in the terminal region of hilar

associational afferents than in the termination zone of the

more distant entorhinal cortex projection neurons. A

slightly different explanation is that entorhinal neurons

are just more succeptible to the neurotoxic effects of

chronic ethanol exposure. In this context the findings of

Abraham et al. ('82a,b) are of particular interest.

Current source density analysis indicates that, in the

dentate molecular layer, some reorganization of afferents

occurs following chronic ethanol treatment, such that the

current sink representative of the entorhinal input shrinks

in relation to the height of the granule cell dendritic

tree. In CA1, a similar shrinkage occurs in the current

sink putatively attributed to afferent terminals arising in

the contralateral CA3 region. Since these are myelinated,

but their ipsilateral counterparts are not, (and neither are

they as affected in CSD measures) it must be considered

likely that alcohol may preferentially disrupt input carried

long distances, relayed mostly by myelinated axons. This is

consistent with the human clinical literature, and may be

related to our finding that both the initial decrease and

later increase in spine density are more pronounced in

distal than in proximal radiatum: It night be expected that

commissural fibers damaged by CET would be replaced by

collaterals from the same side, simply because of the

proximity of the new terminals to their perikarya. This

appears to be the reaction seen following neurotcxic

deafferentation of CA1 with kainic acid, which kills regio

inferior neurons preferentially (Nadler, '80). Since

dendritic spines density does not reveal the source of the

afferent fibers (the same number of spines may obtain even

if afferent reorganization occurs), quantitative

autoradiographic tracing experiments are currently being

conducted in our laboratory in order to bring more

resolution to this question.

Presynaptic influences.

It has been hypothesised that presynaptic terminals play

an inductive role in spine formation (Hamori, '73; Frotscher

et al., '77; but see Hirano, '83), and that the anomalous

long spines observed following ethanol treatment may

represent "seeking" of presynaptic terminals by deafferented

dendrites (Tavares et al., 83). A lower density of

presynaptic terminals in the vicinity of the dendrite say

lead to such spine elongation. There are several reports

that dendrites and spines become altered following

deafferentation (Jones and Thomas, *62; White and Westrum,

'64; Chen and Hillman, '82; Steward and Caceres, '83). If

spines are in fact extended in an effort to form synapses,

and CET produces preterminal degeneration of afferents to

the granule cell dendrites, then spine number would

increase. This would be especially advantageous if granule

cells not killed oft were to attempt to compensate for the

loss of whole cells and portions of the dendritic trees of

the remaining neurons, and in fact appears to exist as a

compensatory mechanism for aging-related neuronal lcss

(Cotman and Scheff, *79). To maintain the total number of

synapses within the structure, more spines would have to be

formed on the survivors. Such a response might also be more

likely to occur in the absence of a cytotoxin, i.e., during


the postethanol period, and to itself abate as the health of

damaged surviving neurons returned, allowing them to support

more synapses. This may also relate to the overproduction

of CAI spines in the 20 week alcohol group; different

initial and late responses to chronic ethanol by pyramidal

and granule cells may reflect differences in the time course

of toxic effects. If pyramidal cells die more slowly, and

continue to die after ethanol exposure, and granule cells

die early and rapidly but survivors exhibit greater

compensatory spine growth both during and post-exposure,

then a delayed increase in the number of spines per

surviving CAI neuron may still result from the effect of

decreases in the number of cells. In support of the

hypothesis that surviving neurons increase spine density in

response to neuron loss, Phillips and Cragg ('83) observed

no loss of CA1 pyramidal cells in animals sacrificed

immediately after ethanol exposure, but a 9% loss in animals

allowed to survive 4 months. Also, Kunz et al. ('76) have

described increased CA1 spine density in tissue where

electron microscopy reveals neuronal degeneration. Since

various neurotoxic agents and treatments are known to affect

these cell types differently (Nadler, '80; Walker et al.,

'80; Irle and Barkowitsch, '83; Kirino and Sano, '84 a,b)

the parsimony and probability of such an hypothesis merits

its further investigation. Our finding that spine density

remains higher in both neuronal types after 20 weeks of


recovery may then reflect the fact that both areas do suffer

permanent neuronal loss, with survivors maintaining a

compensatory increase in spine density.

Systemic and molecular influences.

Several factors known to affect neurons might be of

physiological importance for spine density changes resulting

from chronic ethanol consumption. In addition for any

direct effect upon neurons, such as the reported increase in

cholesterol content in alcohol-adapted neuronal membranes

(Crews et al., '83), alcoholism is known to produce many

peripheral organ effects (Cohen and Gallant, '81), many of

which may underlie pathological processes in the CNS. For

example, in human alcoholics, macrocythemia is very common,

and may result in the slowing of blood flow through

capillaries (Larkin and Watson-willians, '84). Since the

hippocampus is especially succeptible among CNS regions to

the adverse effects of anoxia, hypoxia, and ischemia

(Johansen et al., '84; Kirino and Sano, *84a), and acute

ethanol exposure results in ischemia specifically in

hippocampus and cerebellum (Goldman et al., *73), (two of

the most succeptible regions to alcohol-related

neuropathology), the role of hematological changes in

alcoholic hippocampal pathology may be very important. The

cytopathology of all four situations is very similar, with

degenerative processes that especially affect dendrites

(Johansen et al., *84; Kirino and Sano, *84b). Furthermore,


it has been observed that in humans, dentate gyrus alcoholic

neuropathology has been noted to be most severe in proximity

to blood vessels (McLardy, '74). Another line of evidence

that supports the role of altered blood supply in alcoholic

neuropathology is that patients with alcoholic organic brain

syndrome improve better on psychometric tests and EEG

diagnostics when treated with antihypoxidotic/nootropic

drugs, which allow nervous tissue metabolism rates to remain

normal while requiring less oxygen (Saletu et al., 83).

A second common finding in human alcoholics is elevated

circulating corticosteroid concentrations, and abnormal

regulation of steroid release (Bertello et al., '82; de La

Fuente et al., '83; Abou-Saleh et al., '84; Khan et al.,

'84). Abnormalities in the hypothalamic-pituitary-adrenal

axis, as reflected in dexamethasone suppression tests,

appear to rectify in the early weeks of abstinence (Abou-

Saleh et al., '84; Khan et al., '84), and say thus account

for some of the synaptic structural changes seen in chronic

alcohol consumption that recover with abstinence.

Physiological doses of corticosteroids, or disrupting the

normal circadian rhythm of release, have been shown to

adversely affect the magnitude of reactive synaptogenesis in

the rat dentate gyrus (DeKosky et al., '84), and have

thereby been implicated in the relative failure of old

animals to exhibit the same synaptic regenerative capacity

as young animals. Since chronic alcohol treatment also


reduces the amount of reactive synaptogenesis that rats can

produce in response to damage (West et al., '82; Walker et

al., '84), corticosteroid elevations might be a common

deleterious symptom of alcohol consumption and aging.

Corticosteroids have also been shown to play a role in the

regulation of synapse formation in developing retina (Puro,

'83). If there is natural turnover of hippocampal synapses,

then elevated steriod levels, or normal levels but

disordered circadian patterns might significantly reduce at

least the replacement component of turnover.

Another reported effect of chronic ethanol exposure is a

decrease in protein synthesis (Noble and Tewari, *74; Earvin

et al., '80; Lieber, '84). Since spines appear to contain a

specialized cytoskeletal network (Cohen et al., '83; Landis

and Reese, '83), composed of several protein species

(Fifkova and Delay, '82; Caceres et al., '83; see also Cohen

et al., '83), and have been reported to be spatially

associated with most of the protein synthetic apparatus,

polyribosomes, in the dendrites of hippocampal neurons

(Steward and Levy, '82), ethanol may affect the metabolic

and structural integrity of spines directly. It has been

shown that the organization of spine-associated

polyribosomes is responsive to experimental deafferentation

(Steward, '83); if CET produces a partial denervation, and

the normal polyribosome response is altered, then some

change in the morphology of dendritic spines might be

expected. As spines are currently under intense

investigation because of the possibility that they say

undergo morphological and enzymatic changes in response to

synaptic activation that may reflect functional modulations

in the synaptic efficiency of connections videe supra), this

possibility deserves more attention. However, the increased

spine density found on granule cells confounds such a simple

explanation. It may be important that, even though synaptic

connections are changed in number, population

neurotransmission characteristics are little changed from

normal. Can individual surviving synapses be modified in

their efficiency, so that normal I/O relationships hold even

in alcoholized animals? The evidence to date suggests that

EPSP generation and action potential thresholds are

unchanged, indicating that the answer may be "yes". In this

regard, membrane fluidity and composition changes may be

highly adaptive (Harris, '84; Crews et al., '83).

The primary metabolite of ethanol, acetaldehyde, is

itself reported to decrease the activity of microtubules and

protein synthesis, and also to cause cells to retain

proteins internally (Lieber, '84). These effects would be

detrimental to the cytoskeleton, and decrease a neuron's

ability to elaborate healthy dendrites and spines. In some

neurons, chronic ethanol treatment has been found to

increase intranucleolar bodies, aggregates of

ribonucleoproteins and REA that are believed to represent


'stored' protein synthetic components (Dunmire and LaVelle,

'83). Others have observed increased somatic rough

endoplasmic reticulum and multivesicular bodies, while

dendrites showed marked vacuolization and degeneration

(Goldstein et al., 83; Irle and Markowitsch, '83),

suggesting that distal protein synthetic material may become

sequestered in the soma at the expense of the peripheral

processes. Thus ethanol may produce disruption in the

extranuclear protein synthetic machinery while at the same

time lead to intranuclear sequestration of replacement

molecules. Such a process might facilitate cytological

recovery once alcohol is no longer present, since stcred

elements could be recruited more rapidly and with less

energy than required by the induction and execution of

translation processes. It may also represent an attempt by

the cell to limit its losses to the peripheral processes, in

order to save the soma, or be, simply, an aberrant

pathological condition.

Other molecules related to cytoskeletal dynamics that are

affected by chronic ethanol include cyclic AMP and

calmodulin. The former has been shown to regulate

synaptogenesis in culture, by influencing calcium channel

expression, glycoprotein modifications, and the levels of

certain proteins in the cell (Nirenberg et al., '83). In

mice, chronic ethanol treatment has been reported to

decrease the ability of neurotransmitter molecules to

increase adenylate cylcase activity (Tabakoff and Hoffman,

'79), and chronic ethanol treatment may reduce CNS

calmodulin levels (Towle et al., '81; but see Luthin and

Tabakoff, '84). Since calmodulin regulates adenylate

cyclase by associating with calcium ions, which enter the

spine upon the reception of neurotransmitter molecules, a

decrease in this enzyme could reduce the amount of cAHP

produced by activating a synapse. This might produce a

disruption in molecular communication between pre- and

postsynaptic structures necessary for maintaining the

functional properties of the connection, and would have

important implications for the hypothesized processes by

which synaptic potentiation might occur (Baudry and Lynch,

'80). Possibly related to the cAHP/calmodulin phenomena is

the finding that catacholamine metabolism is often abnormal

in chronic alcoholics. Norepinephrine, in particular, has

been linked to normally occurring synaptic plasticity (cf.

Cotman et al., '81), and found to be affected by alcohol

(Eisenhofer et al., 83).

Finally, many disorders of vitamin, mineral, and nutrient

metabolism, some of nutritional and some of

pathophysiological origin, are common in alcoholics (Lieber,

'84, Sherlock, '84). Thus differences in amino acid

availability, carbohydrate energy sources, electrolytes, or

biocatalysts may have important consequences for neurons

with spines.

Spine Density and Recovery in CT Scans

Since in living human alcoholics only gross morphology

can be quantified, the present results speak to those who

are interested in the mechanisms underlying the reversal of

cerebral atrophy diagnosed in CT scans. Although several

non-exclusive hypotheses have been proposed to explain the

neurobiology of this phenomenon, experimental evidence for

any one is minimal or nonexistent. We consider it

interesting, in light of the tissue dehydration/rehydration

hypothesis of alcoholic CNS damage (Carlen and Wilkinson,

'80), that we find an apparent difference in tissue density

between our groups, judging from the brain weight/area data.

This indicates that perhaps more than one symptom is

responsible for CT cerebral atrophy. The loss and regrowth

of dendritic spines on CNS neurons, although it may not be

extraordinarily large in magnitude, may be sufficient to

explain the occurrence and reversal of a change in tissue

volume, but other effects, such as hydration, may be partly

responsible for the effects of chronic ethanol consumption

on intellectual abilities.


Experimental Methods and Design

To study recoverability, or the relative capacity of CNS

tissue to respond to an extrinsic injury, the amount of

sprouting of spared afferent axons to the dentate gyrus was

quanitified in alcohol and control animals following a

partially deafferentating lesion. The dependent variables

are the widths and stain intensity of afferent-specific

stain bands in tissue prepared using AChE histochemistry,

and the amount of intra-animal difference in these bands

following unilateral entorhinal cortex lesions. Intra-

animal differences were calculated by comparing the lesioned

with the unlesioned sides. Baseline alcohol effects were

determined in unlesioned alcohol-treated animals and by

comparing the unlesioned side of alcohol and control


Entorhinal Cortex Lesions

Eight weeks after the end of the liquid diet treatment,

the entorhinal cortex of the left cerebral hemisphere was

destroyed by passing a 1 milliamp current through the

central carrier of a 75 micron insulated concentric bipolar

electrode (Frederick Haer) for 45 seconds at each of six

stereotaxically defined locations. With the skull mounted

level, lesions were made 2.5, 4.0, and 5.5 am below the

surface of the brain at 8.5 as posterior to the bregma skull

landmark and 3.7 ma lateral to the aidline. At 8.7 am

posterior and 4.6 as lateral, lesions were made 3.0, 4.5,

and 6.0 as beneath the surface. Current passage through the

electrode, which was introduced at an angle of 11 degrees

from vertical, was sufficient to produce a discrete

destruction of medial and lateral entorhinal cortex without

infringing upon the nearby dentate gyrus or other cortical

regions. Lesions were performed on animals anesthetized

with pentobarbital (50 mg/kg, 0.1 cc atropine and 0.2 cc

bicillin for postsurgical prophyllaxis).

Histological Preparation

During the post-liquid diet period and continuing during

the postlesion survival, animals were fed ad lib with

ordinary lab chow and tap water. Forty days after being

lesioned, they were sacrificed with intraperitoneal

pentobarbital overdose and perfused with gravity-fed

solutions delivered into the left cardiac ventricle through

18 guage hypodermic tips. To process these brains for both

Tima's and AChE histochemical reactions, animals were first

perfused with phosphate buffered physiological saline

(approx. 300 al, 30 degrees C.), followed by 500 al of a

solution made from 11.7 g sodium sulfide, 11.9 g sodium

phosphate, and 1.0 g sodium diphosphate in 1000 al distilled

water. This causes CNS cations like Zn+2, Pb+2, and Cu+2 to

form visible, insoluble sulfides. The brain was immediately

removed from the skull and placed for 1-2 hours in a 20%

sucrose fixative (975 al Sorensen's buffer (.066 H potassium

dihydrogen phosphate, 19.6%; .066 M disodium phosphate,

80.41) 10 g paraformaldehyde, and 25 al 50%

glutaraldehyde). The fixed trains were embedded in gelatin

albumin and cut at 40 microns on a freezing stage microtome.

Alternate sections were collected separately for subsequent

Tima's or AChE processing.

Sections for the Tima's stain were mounted on alum dipped

glass slides and developed in a photographic darkroom for

55-100 minutes in a mixture of 120 al gum arabic, 20 al

citrate buffer (25.5 g citric acid, 23.5 g sodium citrate,

100 al triple distilled water), 60 al 5.67% hydroquinone,

and 1.0 al 17% aqueous silver nitrate. Following this

intensification step, slides were dehydrated and

coverslipped with Eukitt medium (Calibrated Instruments,

Ardsley, NY).

Sections for AChE staining were collected into saturated

sodium sulfate over ice, then incubated at 21 degrees C. for

75 minutes in a solution of 7.2 al ethopropazine HCI, 115.6

mg acetylthiocholine iodide, 75.0 mg glycine, 50.0 mg copper

sulfate, 885 mg sodium acetate, and 100 al distilled water

(cf. Geneser-Jensen and Blackstad, '73). This reduces

nonspecific cholinesterase staining while forming a

chromogenic complex between the enzyme and its artificial

substate. The sections are mounted on glass slides after

being rinsed 6-7 times with distilled water and immersion

for 1 minute in 1.25% sodium sulfide in .1 H HCl (8.08 ml

HCI in 1000 ml water) in a fuse hood, followed by 6-7 more

distilled water rinses. The last rinse was 30% ethanol, and

began dehydation prior to clearing in xylene and

coverslipping with Eukitt.

Verification of Lesion Placement and Size

The location and extent of the entorhinal cortex lesions

were evaluated on tracings made using a microprojector.

Four dorsal-ventral levels (1, 4, 7, and 10) were compared

for evidence of complete entorhinal deafferentation via

either entorhinal cortex or perforant path/angular bundle

destruction. Animals were not included if lesions

encroached upon the dentate itself. Figure 9 shows the

lesion reconstruction of one representative animal. Ten

animals from each diet group were ultimately selected for

measurements. Measurements of the the extent and location

of the lesions are presented in the Results section.






Figure 9: Reconstruction of Entorhinal Cortex Lesions

Measurements of Histological Sections

Reactive synaptogenesis following unilateral entorhinal

lesion was evaluated by measuring the width and stain

intensity of the bands in the AChE-stained histological

sections. The band of principle interest was that

corresponding to the associational/commissural afferent

terminal field, in the inner third of the dentate molecular

layer. Measurements of other stain bands provided 1)data

corroborating the lesion reconstruction drawings, since

shrinkage should relfect the extent of aclecular layer

deafferentation at the point the band measurements are

obtained, 2) evaluation of reactive synaptogenesis by

contralateral entorhinal and septal afferents, and 3)data

revealing lesion effects previously unobserved or


Ten stained sections, spaced at approximately equal

intervals, were selected for measurement for each animal.

These were found between anatomical landmarks that were

common to all animals, defined an optimum sampling range of

the hippocampal formation for each plane of sectioning, and

could easily be replicated by other researchers. For

coronal sections, these landmarks were selected after

empirical evaluation of several possible features revealed

the two with the lowest interanimal variation in

interlandmark distance (King et al., in press). Anteriorly,

the first coronal section to exhibit the joining of the

buried and exposed blades of the dentate gyrus, and

posteriorly, the first section to show fibers of the

posterior commaissure crossing the aidline of the brain,

these landmarks defined a region of dorsal hippocampal

formation 1.56+/-.8 am thick. For horizontally sectioned

material, sections for measurement were selected from

between, dorsally, the section first demonstrating a clear

separation of CA3 pyramidal cell sonata from the dentate

gyrus, forming a separated anterior and posterior

hippocampus, and ventally, the loss of a V-shaped dentate

granule cell layer, and the occurrence of a grossly

distorted molecular layer, near the temporal pole of the

hippocampus. These correspond to Plates 63 and 58,

respectively, in the rat brain atlas of Paxinos and Watson

('82). This yielded a sampling region about 2.2 am in

animals whose brains were sectioned horizontally.

Measurements were obtained by projecting black and white

video images of histological sections from the microscope

(10X1.2512.5X) to the terminal screen of a Spatial Data

Systems EyeCom II image analysis terminal, then converting

the online images into arrays of digital values encoding

position (X,Y) and gray level (Z). Gray levels between 0

(black) and 255 (white) were then sampled at the desired

locations in the image to obtain indices of the optical

density of stain bands, with the use of Fortran programs

specifically designed for the patterns of staining in Timm's

or AChE histologies (figures 10 and 11).

Measurement software for Tiam's material.

For the Timma's material, where the aclecular layer

exhibits three bands, and particularly sharp borders between

bands, the program required the operator to first trace

along the granule cell layer/molecular layer border (figure

10). This measured the blade length, and this measurement

was used to calculate where to place markers on the screen

(but not in the measured image) at 101 intervals when the

operator retraced the blade. Next the operator moved the

cursor to a point at the distal border of the molecular

layer (pia mater for the exposed, and hippocampal fissure

for the buried blade) where a traverse from the first marker

to the point would be as perpendicular as possible to the

stain bands at that location. The computer generated this

line, and used it as a guide to sample a swath, 9 pixels

wide, at every pixel along the traverse (without "seeing"

the traverse line itself). From the average of the 9 pixel

samples at each point along the line, an optical density

profile along the traverse was collected. Using a half-

height algorithm to recognize the sharp Timm's band borders,

i.e. the positions in the profiles where the gray level

changes were greatest, the widths of each band were

automatically calculated. A scaling factor was included to

produce corrected terminal field widths for

commaissural/associational, medial entorhinal, and lateral

entorhinal afferents to the dentate. Summing these resulted


in an estimate of total molecular layer width. This process

was then repeated at each subsequent marker until 10

profiles were collected for the blade, then the program

stored the information. Both blades on both sides of the

brain were measured in the 10 sections for each animal,

resulting in the generation of a measurement sample 10

sections by 10 measurement locations for each blade of each


Figure 10: Computer-assisted Tima's Width and Stain
Intensity Measurement

Program testing, measurement replication, and testing of

this program proved it to be as reliable as making these

measurements by hand, with the advantages of speed,

objectivity (and thus reliability), and the ability to

collect a larger sample.

Measurement software for AChE material.

In the AChE material, the more diffuse nature of the

staining pattern required a less automatic approach (figure

11). Here, after measuring bladelength and placing 10%

interval markers (this time along the hilus/granule cell

layer border), the operator was requested to supply not only

the distal extent of the molecular layer, but the borders

between 1)the granule cell layer and the compact, AChe-rich

supragranular band, 2)the supragranular band and the less

dense commissural/associational band, 3)the C/A band and the

outer 2/3 region where the AChE is normally slightly acre

concentrated. This program sampled the Z values of all

pixels in boxes 20 pixels wide, along each traverse, for

each stain band. The outer 2/3 zone was subdivided into 5

equal boxes for closer examination of the AChE condensation

reaction, as well as quantitative description of the pattern

in the unlesioned animal and on the side contralateral to

entorhinal lesions.

All data were analysed with two-way analysis of variance

(CET and ECX) (diet v. lesion) on SAS (Statistical Analysis

System), with provisions for testing interactions resulting

from 1)aedial-lateral gradients or differences,

2)anterior/dorsal-posterior/ventral gradients or

differences, and 3)individual animal differences within and

between groups. Each measurement point in the 10 by 10

arrays can be compared individually, or collapsed as

Figure 11: Computer-assisted AChE Band Width and Stain
Intensity Measurement

desired. For example, because entorhinal lesion results in

a shrinkage of the molecular layer where it is deafferented,

we were able to accurately subset the points used for

analysis according to a criterion amount of molecular layer

shrinkage. This concept is illustrated in further detail in

the Results section.

Stain Intensity Normalization.

Because the microscope light level was adjusted to

produce optimum video image contrast prior to digitizing

each section, it was necessary to normalize the Z values to

a constant light intensity. The EyeCom II camera maintains

a linear relationship between Z and the light incident on

the lens (figure 12), for any particular optical density.

Because brain tissue stains appear to pass a greater

proportion of the incident light source at higher lamp

voltages, however, it was found that dark regions had

different light/Z ratios than lightly stained regions

(figure 12). Since the stain stands we measured covered a

range of optical densities, it was necessary to correct

values by first calculating the slope of the light/Z

relationship for the original Z values obtained, then

determining the (microscope) light-correction factor

according to this specific slope. The specific slope was

determined by dividing the raw Z by 57 (the Z obtained at

6.0 volts scope light intensity for an arbitrarily chosen

reference section where the optical density is such that Z

changes by 30 values per volt), raising the result to the .4

power (the exponential function relating the specific slope

to the particular optical density--inset, figure 12), and

multiplying by the slope for the reference density (30

Z/volt). Dividing this corrected slope by the difference in

scope light voltage (from 6.0) yeilds a normalization

quantity that was added (or subtracted) to the original Z

value. The following formula was used to normalize the

stain intensity raw data:

normalized=raw+((raw/57)**.4)*30 Z

normalized=voltalge normalized Z value
raw=raw Z value
voltage=Olympus BH-2 light voltage used to

Assuming that stain intensity variability averages out with

the use of many animals and several sections per animal,

200 -

c 150

E 100


z i00
30 40 50

6.0 6.5 7.0 7.5 8.0

Figure 12: Linearity of EyeCom II Gray Levels

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd