Diurnal

normal

rhythmicity
of human
pattern and adaptation
LYNN KESSLER
CELLA, EVE VAN CAUTER,
Committee for Human Nutrition
and Nutritional
University
of Chicago, Chicago, Illinois
60637

Cella,
Lynn
Kessler,
Eve Van Cauter,
and Dale A.
Schoeller.
Diurnal
rhythmicity
of human cholesterol
synthesis: normal pattern and adaptation
to simulated
jet lag. Am.
J. Physiol.
269 (Endocrinol.
Metab. 32): E489-E498,
1995xThe diurnal
rhythm
of cholesterol
synthesis
was determined
by deuterium
incorporation
from body water in five normolipemic men studied during a 24-h baseline period and on the lst,
Znd, and 4th days of a simulated
12-h time zone shift achieved
by delaying
sleep times and, starting
on the 2nd day, mealtimes. Profiles of plasma cortisol and thyrotropin
(TSH) were
obtained
simultaneously.
Under baseline
conditions,
cholesterol synthetic
rates varied
from essentially
zero in the
morning
to maximal
values around
midnight.
On the 1st
shifted
day, this diurnal
variation
was unaltered
despite
sleep-wake reversal. The diurnal pattern of cholesterol
synthesis, however, was shifted 5 h on the 2nd shifted day and N 12 h
on the 4th. The diurnal variation
of synthetic
rate cholesterol
fractional
synthesis and plasma cortisol levels was negatively
correlated
on both the baseline day and the 1st shifted day. A
positive correlation
with the TSH rhythm was found on the 1st
day only. During
the 2nd and 4th days, the rhythm
of
cholesterol
synthesis
adapted
faster than the rhythms
of
cortisol
and TSH. These findings
indicate
that cholesterol
synthesis is not acutely entrained
by the sleep-wake cycle nor
is it primarily
entrained
by the circadian clock.
deuterium;
mass spectroscopy;
sol; thyrotropin

circadian

rhythm;

sleep; corti-

that de novo cholesterol synthesis is a major source of cholesterol entering the body pool
in humans. Moreover,
it has been demonstrated
that
this rate of synthesis varies dramatically
across the day
(14,17,18,22).
Few studies, however, have investigated
the factors that regulate this diurnal variation
because
of a lack of methods for measuring
within-day
changes
in cholesterol
synthesis in vivo. This is unfortunate
because investigations
of cholesterol metabolism
in humans during
sleep deprivation
(30) and shift work
rotations (26) suggest that changes in sleep-wake and/or
light-dark
cycles may alter cholesterol homeostasis.
Among those who have investigated
the diurnal variation of cholesterol synthesis, Miettinen
(18) observed a
nocturnal increase in the plasma levels of the cholesterol
precursors squalene and lanosterol, with maximal levels
between midnight
and 0400. In addition, Parker et al.
(22) observed a nocturnal
increase in plasma levels of
the precursor
mevalonate,
consistent
with an overall
augmentation
in the biosynthetic
pathway. Direct measures of diurnal variations
in human cholesterol synthesis have only been recently achieved using the deuterium
incorporation
technique.
In one study (17)
measurements
of cholesterol synthesis were obtained at
4-h intervals
over a 48-h period in six young men.
IT IS WELL ESTABLISHED

0193-1849/95

$3.00 Copyright

cholesterol synthesis:
to simulated jet lag
AND DALE A. SCHOELLER
Biology and Department
of Medicine,

Maximal synthetic rates were found to occur in the early

morning hours or N 2-6 h later than estimated from the
above investigation
of precursor
levels. However,
in a
subsequent study by Jones et al. (15), maximum cholesterol synthesis
occurred
near that reported
for the
above-mentioned
precursor
studies. Moreover,
it was
found that profiles of the diurnal variation
in cholesterol
synthesis as measured by deuterium
incorporation
coincided with those of the plasma mevalonate
level method.
The lack of standardization
of sleep times, light-dark
exposure, and mealtimes and the relatively
infrequent
sampling intervals
could account for the discrepancies
in time of maximal synthesis observed in these studies.
A diurnal
rhythmicity
of cholesterol
synthesis has
also been clearly established in experimental
animals. A
diurnal variation
in synthesis has been demonstrated
in
the liver, and, with a somewhat lesser amplitude,
in the
intestinal
mucosa (8). In nocturnal
animals
fed ad
libitum, the rate of cholesterol synthesis is highest after
midnight
during the dark cycle and lowest during the
morning
and early afternoon
(8). This rhythmic
variation has been associated with changes in the activity of
3-hydroxy-3-methylglutaryl-coenzyme
A (HMG-CoA)
reductase, the rate-limiting
enzyme in cholesterol synthesis (l&25).
In the rat, the diurnal rhythm of cholesterol
synthesis appears to be intimately
related to food intake.
A shift in feeding sched ule without changi .ng the lighting cycl .e is followed by a shift in the rhythm that stays in
phase with feeding (8). However,
whereas intestinal
cholesterol synthesis shifts rapidly to stay in phase with
feeding (8, 9), hepatic synthesis does not come completely into phase when feeding times are changed (8),
suggesting that other factors, such as intrinsic circadian
rhythmicity,
may also play a role. Furthermore,
adrenalectomy has been shown to decrease (19) or abolish (7)
the nocturnal
rise in HMG-CoA
reductase
activity,
suggesting a role for corticosterone
in the regulation
of
the daily rhythm
of this enzymatic activity. More recently, Waurin and Schibler (31) have demonstrated
a
circadian rhythm in an HMG-CoA
reductase transcription activator protein DBP, which free-runs under constant conditions,
independently
of food intake, and
could be related to diurnal changes in cholesterol synthesis. Thus, at least in laboratory
animals, evidence exists
that both feeding schedule and circadian rhythm play a
role in regulating
the diurnal variation
of cholesterol
synthesis.
In view of the paucity of data regarding
diurnal
variations
in human cholesterol
synthesis and of the
uncertainties
regarding
their possible control mechanisms, the aim of the present study was to characterize
24-h changes in human
cholesterol
synthesis under
controlled feeding, lighting, and sleeping conditions, and

o 1995 the American

Physiological

Society

E489

E490

DIURNAL

Table 1. Physical

parameters

RHYTHM

and lipid

Age,

yr

23
23
22
25
24

Height,
cm

170
182
184
183
171

Weight,
kg

Body
Fat, %

Treatment

group

CHOLESTEROL

concentrations

Screening

Subjects

OF HUMAN

Plasma
mmol/l

Cholesterol

Experimental

20
5.15
2
92.0
19
3.93
3
65.4
17
2.72
4
70.1
8
2.87
5
87.6
20
4.60
Means
IL SD 23 2 1 178+ 7 77.6 2 11.5 17 + 5 3.85 + 1.06 0.88

1.40
1.00
0.55
0.82
0.61
+000.34

Control group
6
7

25
24

192
182

93.0
81.8

25
20

5.07
3.96

1.31
0.52

to test whether
they are entrained
to a sleep or the
circadian clock. The rate of adaptation
of the temporal
pattern of cholesterol synthesis was compared with that
of simultaneously
measured 24-h profiles of cortisol, a
hormone that is markedly influenced by circadian rhythmicity, and thyrotropin
(TSH), a hormone that is markedly influenced
by circadian rhythmicity
and sleep (29).
METHODS
Subjects
Volunteers
from the university
community
provided a medical history, and seven subjects were selected who reported that
they were healthy nonsmokers
without
cardiovascular,
liver,
renal,
endocrine,
lipid,
or psychiatric
disorders
and were
taking no medication
(Table 1). Shift workers or subjects who
had experienced
a transmeridian
flight within 6 wk before the
study were excluded.
Positive
factors for selection
included
regularity
of work, social, meal, and sleep schedules.
Fasting
lipid levels and blood glucose were within normal ranges (total
cholesterol
2.59-5.20
mmol/l,
fasting
triglyceride
0.1-1.8
mmol/l,
fasting
glucose 3.9-6.1
mmol/l)
in all individuals
studied.
Diet histories
indicated
a habitual
consumption
of
34 t 4% (SD) of daily energy intake from fat and a cholesterol
consumption
of 187 t 81 (SD) mg/l,OOO
calories. The nature
and intent of the study were explained,
and written informed
consent was obtained.
All procedures
were approved
by the
Institutional
Review Board at The University
of Chicago
Medical Center.

Fig. 1. Schematic representation

of the
5-day simulated jet-lag protocol. Complete darkness is indicated by the solid
horizontal
bar, simulated
daylight by
the open bar, and dim by the hatched
bar. Meals are indicated as breakfast
(B), lunch (Id, and dinner (D). DZO,
2Hz0 or deuterium
oxide; TBW, total
body water; IV, intravenous.

Protocol

Prestudy
habituation.
The study was conducted
at The
University
of Chicago.
For 1 wk before the study, all the
subjects carried out their usual work and social activities
but
were asked to adhere to a standardized
schedule of sleep (from
2300-0700)
and meals (at 0730,1230,
and 1730) to maximize
interindividual
synchronization
and fully establish
baseline
conditions.
During
this period the subjects
were asked to
abstain
from alcohol
or caffeine consumption.
For 3 days
before the study period
the subjects
were provided
with
standardized
breakfast,
lunch, and dinner meals composed of
Western-style
foods, which the subjects
consumed
in the
Clinical
Research Center (CRC) or were allowed to take with
them. Although
the time to stabilize
the diurnal
rhythm
of
cholesterol
synthesis
had not been investigated
before our
current
study, 3 days was selected because previous
studies
had demonstrated
that unlike plasma cholesterol
level, which
is slow to respond to a change in diet, the diurnal
variation
of
cholesterol
fractional
synthetic
rate (FSR) responds
very
rapidly
to intervention
(14, 16). Meal composition
was 40%
carbohydrate,
25% protein,
35% fat with 170 mg cholesterol/
1,000 calories, and a polyunsaturated-saturated
fat ratio of
0.8. Meals were prepared
in the CRC metabolic
kitchen.
Energy was provided
to maintain
weight as calculated
from a
measured postabsorption
resting metabolic
rate, using a Deltatrac metabolic
cart (Sensormedic,
Yorba Linda,
CA), multiplied by 1.1 for dietary-induced
thermogenesis
and 1.65 (outpatient) or 1.4 (inpatient)
activity
factor to meet daily energy
needs.
BaseLine day. The subjects spent the night before the study
in the CRC to habituate
to inpatient
sleeping conditions,
i. e.,
with a forearm venous catheter and polygraphic
sleep recording electrodes
in place. After
this habituation
night
the
subjects were released from the CRC but fed the habituation
diet. Subjects were readmitted
to the CRC on the same day at
1600 and remained
there for the next 5 days (Fig. 1). A
forearm venous catheter for blood sampling
was inserted and
kept patent with a saline drip (75 ml/h). Dinner was served at
1730. During the sleep periods the catheter was connected to a
plastic tube extending
to the adjacent room so that sampling
could continue
without
disturbing
the subject.
A baseline
blood sample was obtained
at 1800, after which a priming
dose
of deuterium
oxide (1.0 g 2H20/kg
estimated
body water, 99.8
atom% excess, tritium
depleted, MSD Isotopes, St. Louis, MO)
was orally administered.
Blood samples were collected hourly
for the next 72 h. Deuterated
oral fluids were provided
throughout
the study to maintain
constant
plasma
water

Levels,

Triglycerides

73.0

SYNTHESIS

& Loading
dose D20
lg/kgTBW
)-- Deuterated
drinking
water
I- I.V. saline
drip
)-- Hourly
blood
sampling
--I
1-1
Sleep
recording
1-1

Baseline

1800

Meals

Day
II

Day
I

1800

BLD

Day
I

1800

BL

I
I

BLD

4
I

1800

1800

LD

Day

4
i
I

BLD

1800

DIURNAL

RHYTHM

OF HUMAN

enrichment.
Because oral fluids only comprise
about onefourth of estimated
water input, all oral fluids contained
4.2 g
2H20/kg
total body water. The subjects slept in total darkness
from 2300-0700.
At 0700 the subjects were awakened and the
room was illuminated
with conventional
indoor and natural
outdoor lighting.
Breakfast
was served at 0730, lunch at 1230,
and dinner
at 1730. The subjects were allowed 30 min to
consume their meals and were not allowed any other foods or
fluids. During waking hours the subjects were allowed to walk
around
the unit, watch television,
listen to the radio, and
engage in conversation
with visitors and staff personnel.
Naps
were strictly prevented.
Two subjects stayed on this schedule
of bedtimes,
mealtimes,
and light exposure
to examine
the
stability
and reproducibility
of metabolic
and hormonal
profiles during prolonged
(i.e., 4 days) hospitalization
in the CRC.
The other five underwent
a simulated
jet lab as described
below.
Day 1, sleep deprivation12-h simulated
time zone shift. At
2300 the overhead lights in the room were dimmed
but the
subjects were not allowed to sleep. At 0700 overhead
lights
were turned on and natural
outdoor light was allowed into the
room. Breakfast
was served at 0730 (i.e., as on the baseline
day) and lunch at 1030 (i.e., 2 h earlier than on the baseline
day) while the subjects were kept continually
awake. At 1100,
i.e., 12 h later than during baseline conditions,
blackout
drapes
were drawn and the subjects were allowed to sleep in total
darkness.
At 1800 hourly blood samples had been obtained
continuously
for 48 h and day 2 began.
Days 2-4 of the 12-h shift. The subjects were awakened at
1900, i.e., 12 h later than under baseline conditions.
A mobile
panel of fluorescent
tubes providing
a light
intensity
of
5,000~5,500
lux at 2 ft were turned on and placed 4 ft from the
subject to simulate
daylight
(Sunbox, Rockville,
MD). Subjects
received breakfast
at 1930, lunch at 0030, and dinner at 0530,
i.e., 12 h later than under baseline
conditions.
At 0700 the
light panels were turned off and overhead lights in the room
were dimmed
to simulate
evening.
Subjects
slept in total
darkness from 1100 to 1900. Blood sampling
was terminated
at 1800, i.e., after 72 h of hourly collections.
The subjects were
maintained
on this 12-h shift schedule for an additional
48 h,
i.e., days 3 and 4. Blood sampling
at hourly
intervals
was
reinitiated
at 1800 on day 3 and continued
for the final 24 h of
the study.
Laboratory

Procedures

De novo choZesteroZ synthesis. De novo cholesterol

synthesis
was measured
by the deuterium
incorporation
method
(16).
Deuterium
enrichment
was measured
in plasma total cholesterol and plasma water. Neutral
lipids were extracted
from 1
ml plasma
samples
(4). Samples
were mixed with
10 ml
chloroform-methanol
2: 1 (vol/vol)
and filtered.
The filtrate
was back-extracted
with 2.5 ml of 0.1 mol/l KC1 and the upper
phase discarded. The lower phase was washed three times with
5.0 ml chloroform-methanol-O.1
mol/l
KC1 3:48:47 (vol/vol/
vol). Solvent was evaporated
at room temperature
in a vortex
evaporator.
Extracts
were saponified
at 50C in 0.4 mol/l
ethanolic
potassium
hydroxide
overnight.
After adding 1 ml
H20, the nonsaponifiable
fraction
containing
free cholesterol
and deesterified
cholesterol
was extracted
three times with 2
ml petroleum
ether, which was back-extracted
with 4.0 ml of
0.1 mol/l ethanolic
potassium
hydroxide.
Solvent was evaporated at room temperature
in a vortex evaporator.
The isolated
cholesterol
was transferred
to a 6-mm-OD
quartz tube using
three washes of chloroform.
Solvent was evaporated,
and 0.1 g
cupric oxide and a 2-cm length of silver wire were added before
flame sealing under vacuum. Samples were cornbusted
to CO2
and H20 at 750C for 2 h. The Hz0 was isolated via vacuum

CHOLESTEROL

E491

SYNTHESIS

distillation,
cryogenically
transferred
to a second quartz tube,
and reduced to H2 over 0.05 g zinc reagent (Friends of Geology,
Bloomington,
IN) at 500C for 30 min. The resulting
H2 was
then analyzed by gas isotope ratio mass spectrometry
using a
triple inlet Nuclide
3-60 HD-RMS
(PATCO,
Belefonte,
PA)
(16). Abundances
were expressed as parts per million
(ppm)
deuterium
(mol/mol)
based on the international
Standard
Mean Ocean Water (155.76 ppm). Isotope abundances
were
corrected for exchange during combustion-reduction
(unpublished data). Exchange
averaged 2.4 I_LM Hz with deuterium
abundance
of 148 ppm. To measure deuterium
enrichment
of
body water, postdeuterium
plasma samples were gravimetritally diluted
lo-fold with distilled
water to reduce the deuterium enrichment
to within the normal analytic range. Baseline
samples were not diluted. Triplicate
2-l_~1samples were vacuum
distilled,
reduced,
and mass spectrometrically
analyzed
as
previously
described.
Cortisol assay. Hourly plasma cortisol levels were measured
by radioimmunoassay
(Diagnostic
Products,
Los Angeles, CA)
with an average intra-assay
coefficient of variation
of 4.0%.
TSH assay. Hourly
TSH levels were measured
by radioimmunoassay
(Serono Diagnostics,
Coinsins,
Switzerland)
with
an average intra-assay
coefficient of variation
of 5.8%.
Plasma
lipid ZeveZs. Total and esterified
plasma cholesterol
levels were measured
enzymatically
(Lancers Kits, Division of
Sherwood
Medical,
St. Louis, MO). Free plasma cholesterol
levels were not directly measured but determined
by difference
from total and esterified
cholesterol.
Triglyceride
levels were
measured
enzymatically
(Boehringer-Mannheim,
Indianapolis, IN).
Cakulation

of CholesteroZ

FSR

FSR was calculated

for each 2-h interval
from plasma total
cholesterol
deuterium
enrichment
and plotted at the midpoint
of the time interval.
In previous
studies, plasma free cholesterol had been used to measure de novo cholesterol
synthesis
(16) because of the observation
that newly synthesized
cholesterol is released into the circulation
as free cholesterol
(24).
However,
we have recently
observed that the appearance
of
deuterium
in free cholesterol
demonstrates
a substantial
nonlinearity
after the first 24 h (unpublished
data). This
results in an underestimation
of FSR during subsequent
time
periods.
The sampling
of total cholesterol
substantially
reduces this problem.
FSR was calculated
assuming
labeled
hydrogen
incorporation
per carbon atom (H/C ratio) of 0.81
(6), i.e., 22 deuterium
atoms (D) out of 46 total hydrogen
atoms (H) in the 27 carbon (C) molecules
FSR (pools/day)
delta ppm
= excess ppm

(plasma

(cholesterol)
water)

x 12

x 22D/27C

x 27C/46H

where delta ppm (cholesterol)

is the difference
in the deuterium enrichment
across the 2-h sampling
interval
and excess
ppm (plasma water) is the average deuterium
enrichment
of
plasma water above baseline during the sampling
interval.
The
factor of 12 converts the 2-h FSR value to a 24-h rate.
Sleep Recording

and Analysis

The polygraphic
sleep records were scored visually
at 30-s
intervals
in stages wake, I, II, III, IV, and rapid eye movement
(REM) according to the standardized
criteria (23). Sleep onset
and morning
awakening
were defined,
respectively,
as the
times of occurrence of the first and last 30-s intervals
scored II,
III, IV, or REM. The sleep period was defined as the interval
separating
sleep onset from final morning
awakening.
Sleep

E492

DIURNAL

efficiency
was calculated
duration
of wake divided
Statistical

RHYTHM

as total time in bed

by the total time in bed.

OF HUMAN

minus

CHOLESTEROL

SYNTHESIS

total

Analysis

A 1:2:1 weighed 3-point moving average was used to calculate a smoothed

curve for each individual
FSR profile.
The
acrophase
(maximal
FSR), nadir (minimal
FSR), and amplitude (50% of the difference between the maximal
and minimal
rate expressed
as a percent
of the 24-h mean level) were
identified
on the individual
smoothed
curves within each 24-h
interval
(baseline,
day 1, day 2, and day 4). Reported
mean
values were calculated
from the individual
values and not from
the mean levels plotted in Figures 3-8. For cortisol and TSH
profiles,
a best-fit curve quantifying
the circadian
variation
was obtained
using a robust weighted
regression
procedure
(5). The best-fit curve was obtained
using the robust, locally
weighed, regression technique
described by Cleveland
(5) with
a window of 8 h, i.e., t4 h. This procedure involves smoothing
the data within a moving 8-h window by weighted least-square
polynomial
fit, with the weights being largest at the center of
the window and smallest
at the limits
of the window.
The
acrophases,
nadirs, and amplitudes
were calculated
from the
best-fit curve as for the individual
FSR profiles. The effect of
study day on the acrophase,
nadir, and amplitude
of the FSR
and hormonal
profiles and on the 24-h mean FSR and hormonal
levels was determined
using
analyses
of variance
(ANOVA)
for repeated-measure
analysis and a post hoc paired
t-test with a Bonferroni
correction.
The relationship
between
cholesterol
FSR and hormone
levels was quantified
by calculating
the coefficient
of cross
correlation
(13) for the individual
cholesterol
FSR and cortisolTSH profiles for each 24-h interval
(baseline, day 1, day 2, and
day 4) at time lags of 0 h (i.e., simultaneous
values of
cholesterol
FSR and cortisol-TSH),
of:1 h (i.e., cholesterol
FSR
leadingcortisol-TSH
by 1 h or visaversa),
22, +3, +4, 25, and
t6 h. Each cholesterol
FSR profile was determined
from 12
measured and 12 interpolated
values. For each pair of profiles,
the largest coefficient of cross correlation
was identified.
Time effects on plasma cholesterol,
plasma triglyceride,
and
plasma water deuterium
enrichment
were analyzed by regressing plasma
concentrations
against
time. Means
t SE are
presented unless otherwise stated.
RESULTS

Plasma Lipids, Body Weight,

and Precursor Enrichment
No significant
time effects were observed on plasmafree cholesterol
over the study period, but ANOVA
revealed that plasma total cholesterol
levels differed
with respect to study day (Fig. 2; P = 0.04). Post hoc
analysis indicated that baseline levels were 0.2 to 0.3
mmol/l lower than those of the other 3 days (P < 0.05).
No significant
changes in fasting plasma triglyceride
levels were observed over the study period. Mean body
weight was stable within 0.8%, with a mean change of
0.2 t 0.9 kg [not significant
(NS)]. The grand mean of
the plasma water deuterium
enrichments
was 841 t 11
ppm (Fig. 3). The within-subject
coefficient of variation
in enrichment
averaged 2.0%. No significant time effects
were observed on mean plasma water deuterium
enrichment over the study period.

3.00
6

a
r-

2.00

6
s
0

1.00

o~~~*~~~p-p-p-p-o~~-o-o-o------t

0.00

- - -o- o- o- 0

1 1
1800

1800

BL

Dl

1800

1800

1800

D2

D3

04

TIME

1800
DS

(time/day)

Fig. 2. Plasma free (--o--) and total (--->

cholesterol levels for the 5
simulated jet-lag subjects; means + SE. Bl, baseline; Dl-D5,
days
l-5.

Cholesterol

FSR, Cortisol,

and TSH Across the Study

ANOVA
indicated
that 24-h cholesterol
FSR was
stable across the study period, averaging 0.042 t O.Oll/
day on the baseline day, 0.051 t O.O07/day on shifted
day 1, 0.047 t O.Oll/day
on day 2, and 0.035 t
O.OOS/day on day 4 (NS). The 24-h mean cortisol level
was 157 t 11 nmol/l on the baseline day, 184 t 11
nmol/l on day 1,193 -+ 11 nmol/l onday2, and 204 t 17
nmol/l on day 4, but the trend was not significant
with
time (P = 0.17). The 24-h mean TSH level increased
with time (P = 0.01) from 1.9 t 0.4 mU/l on the
baseline day to 2.7 -+ 0.5 mu/l on day 1, 2.3 t 0.3 mu/l
on day 2, and 2.9 t 0.5 mu/l on day 4, although
only
day I and day 4 differed by post hoc t-test (P < 0.05).
Reproducibility

of Baseline

Profile

of Cholesterol

FSR

The profiles obtained in the two subjects who were

studied without
shift of sleep times, mealtimes,
and
exposure to light-dark
were examined to determine
the
day-to-day reproducibility
of the 24-h profile of cholesterol FSR and to test for the effects of prolonged
exposure t 1 hospitalization
and experimental
procez

1000

is
ii

900
_--------A
A.
ii
v----

Y
I0

800

!E
w

700

z
2
B

600

5
x

500
BL

,ss
--2-<--

55
--.

-------___
9-- .----,
-o----L---*9--

--. -k--c;,.
. .

8II----,-----z--z---

-A---

2
DAYS

---WC-

=s=-c,=_-c-c

---

-8
C-

Ir-

-B

,---------------A

Fig. 3. Deuterium
enrichment
for the 5 simulated jet-lag subjects for
the &day protocol. Dashed lines, individuals;
solid lines, means + SD.
Deuterium
abundance before labeling averaged 141 + 1 ppm.

DIURNAL

RHYTHM

OF HUMAN

dures. Figure 4 shows the profiles for these two subjects

on the baseline day and after 4 days in the CRC. These
two subjects displayed two nadirs, 0700 and 1700 at
baseline, but all within- and across-subject
differences in
the timings of the acrophase as well as nadirs were < 2
h. The overall waveshape of the diurnal profile appeared
stable. Individual
values for the 24-h mean FSR were
0.020 and O.O39/day on the baseline day and 0.014 and
O.O32/day after 4 days in the CRC, respectively.
Profiles of Cholesterol FSR, Hormonal
and Sleep on the Baseline Day

E493

SYNTHESIS

h
c
I

g:
9

Levels,

Figure 5 shows the mean profiles of cholesterol FSR,

plasma cortisol, and plasma TSH for the baseline day.
On the baseline day, the acrophase of cholesterol
FSR
occurred at 2325 t 0030 and the nadir at 0900 t 0040.
The mean amplitude of the rhythm was 83 t 19%.
The profiles of plasma cortisol and TSH levels conformed with classical descriptions
(29) on the baseline
day. The cortisol nadir occurred at 2325 t 0035 and the
acrophase at 0925 t 0200. Levels of TSH increased in
the late evening, reached an acrophase at 0005 t 0055,
i.e., shortly after sleep onset, and then declined continuously. The mean amplitude
of the rhythm was 27 t 8%.
Mean sleep efficiency on the baseline day was 89 t 7%.
The sleep period was 424 t 16 min with 50 t 16 min
awake, 243 t 13 min stages I + 11, 76 t 5 min stages
III + IV, and 110 t 5 min REM sleep. During daytime
recovery sleep on day 1, sleep efficiency was 87 t 10%

CHOLESTEROL

E
300
1

1900

0100
TIME

1300

1900

DAY

%l%i

1900

0100
TIME

-0.05

0700

1300

1900

TIME

0.15

-I
0700

0100

DAY

1300

I
DAY

1900
TIME

OF

0100

1300

1900

DAY

and the sleep period was 406 t 44 min, with 64 t 47 min

awake, 240 t 46 min stages I + 11, 106 t 22 min stages
III + IV, and 64 t 16 min stage REM.
Profiles of Cholesterol
and Sleep on Day 1

OF

0700

i\JI

OF

0700

Fig. 5. Baseline day values for cholesterol FSR (top), plasma cortisol
(midde),
and serum thyrotropin
(TSH) (bottom) in the 5 simulated
jet-lag subjects; means +_ SD. Stippled bar, period of sleep.

BASELINE

OF

T T

-0.05

0700

31

0700

DAY

Fig. 4. Cholesterol fractional synthetic rate (FSR) during baseline day

(top) and day 3 (bottom) in control subjects. Solid curves are the 121
moving averages. Stippled bar represents period of sleep.

FSR, Hormonal

Levels,

On day 1, which involved

12 h of sleep deprivation
during
the nighttime
and recovery
sleep during the
daytime,
the cholesterol
FSR acrophase
occurred
at
2230 t 0030 and the nadir occurred at 1000 t 0100, i.e.,
essentially at the same clock times as on the baseline day
(Fig. 6). The mean amplitude
of the rhythm
was not
different from baseline, averaging 93 t 22%.
On day 1, the cortisol profile was not different
from
baseline, with a nadir at 2310 t 0025 and an acrophase
at 0735 t 0030. The amplitude
of the diurnal variation
was also unchanged
(76 t 9% on day 1 vs. 69 t 7% on
the baseline day). In contrast, the TSH profile presented
a marked increase in amplitude
(from 27 t 4% on the

E494

DIURNAL

RHYTHM

OF HUMAN

CHOLESTEROL

SYNTHESIS

and that this change differed

from that of the two
controls
(P < 0.001). Post hoc paired t-tests demonstrated that both the acrophase and the nadir of the
profile of cholesterol
FSR had changed significantly
on
shifted day 2 compared with baseline (P < 0.015 and
P < 0.05, respectively),
with the delay in acrophase
averaging
5.0 t 0.7 h and the delay in the nadir
averaging
5.4 t 1.9 h. The amplitude of the rhythm
remained
similar to baseline (85 t 14%). Contrasting
with the significant
shift in diurnal variation
of cholesterol FSR, the profile of plasma cortisol remained
synchronized to the baseline, rather than shifted, schedule.
Indeed, the nadir occurred at 2400 t 0030 and the
acrophase at 0750 t 0040, which was not significantly
different
from baseline. The amplitude
tended to be
lower than on the baseline day, averaging
54 t 8%, but
the difference
was not statistically
significant.
A trend
for a partial adaptation
of the acrophase of the TSH
profile was apparent on day 2, as the acrophase occurred
h
F

1900

0100
TIME

6.0

0700
OF

1300

19(

0.15

a
9

0.10

d
ki
Iv)
3

l-

B
f%

0.05

0100
TIME

Fig. 6. Sleep deprivation

plasma cortisol (middle),
jet-lag subjects; means 2
for comparison. Stippled

day
and
SD.
bar,

0700
OF

1300

-0.05

1900
5oo

:B

400

0100
L

0700

1300

1900

1900

DAY

(day 1) values for cholesterol FSR (top),

serum TSH (bottom) in the 5 simulated
Baseline day is indicated as a dashed line
period of sleep.

baseline day to 51 t 4% on day 1, P < O.Ol), reflecting

the well-known
effects of nocturnal
sleep deprivation
on
TSH secretion (1, 3, 20). The timing of the acrophase
(0355 t 0100) was not quite significantly
different from
baseline (P = 0.1).
During daytime recovery sleep on day 1, sleep efficiency was 87 t 10% and the sleep period was 406 t 44
min, with 64 t 47 min awake, 240 t 46 min stages I +
II, 106 t 22 min stages III + IV, and 64 t 16 min stage
REM.
Profiles of Cholesterol
on Day 2

0.00

DAY

cO.

DAY

1900

FSR and Hormonal

1900

0100
TIME

0700
OF

1300

1900

DAY

Levels

Figure 7 illustrates
the mean profiles of cholesterol
FSR, plasma cortisol, and plasma TSH on the 2nd day of
shifted sleep-wake
cycle, light-dark
cycle, and mealtimes. ANOVA indicated
that the timing of both the
acrophase and nadir of cholesterol FSR in the experimental groups changed with dav of the studv (P < 0.001)

oo3
.
1900

0100
TIME

0700
OF

1300

1900

DAY

Fig. 7. Day 2 of simulated jet-lag values for cholesterol

FSR (top),
plasma cortisol (middle), and serum TSH (bottom) in the 5 simulated
jet-lag subjects; means f SD. Baseline day is indicated as a dashed line
for comnarison. Stinnled bar. neriod of sleen.

DIURNAL

A
r
B
a
2

0.15 .B
0.10

RHYTHM

OF HUMAN

~~1.
.,.,.: :;:.:,,.,
.:..;:

T
0

z.

6
a
r
8.

-005
. 5
1900
**

0100
0700
TIME OF

:B

1300
DAY

1900

CHOLESTEROL

0100

0700

1300

1900

. -A
1.0
0.0

1900

0100
0700
TIME OF

1300
DAY

/-

DAY

1900

Fig. 8. Day 4 of simulated jet-lag values for cholesterol

FSR (top),
plasma cortisol (middle), and serum TSH (bottom) in the 5 simulated
jet-lag subjects; means 2 SD. Baseline day is indicated as a dashed line
for comparison. Stippled bar, period of sleep.

at 0630 t 0130, i.e., 6.3 t 2.1 h later than on the

baseline schedule (NS, P = 0.1). The amplitude
was
similar to that observed on the baseline day, averaging

29?2%.
Profiles of Cholesterol
and Sleep on Day 4

E495

the baseline schedule, but the second acrophase was

delayed 10.1 t 1.0 h from baseline (P < 0.01). The
amplitude
of the TSH rhythm
was significantly
less
than under baseline conditions, averaging only 18 t 4%
(P < 0.05). Th e d ecrease in amplitude appeared to be
due to an increase in the nadir without
change in the
acrophase.
Sleep parameters
on day 4 were similar to those
observed on the baseline day. Mean sleep efficiency was
82 t 7%. The sleep period was 389 t 34 min with 79 t
30 min awake, 215 t 22 min stages I + II, 70 t 12 min
stages III + W, and 111 t 1 min REM stage.
Cross-Correlation

1900

SYNTHESIS

FSR, Hormonal

Levels,

Figure 8 shows the mean profiles of cholesterol FSR,

cortisol, and TSH on shifted day 4. The acrophase and
nadir of the FSR profiles were significantly
shifted from
baseline (P < 0.015 and P < 0.015, respectively),
with
delay averaging 9.6 ? 1.7 h for the acrophase and 14.8 t
0.5 h for the nadir. Partial adaptation
of the cortisol
profile to the simulated time zone shift was evident, as
the nadir was delayed by 5.4 t 1.5 (P < 0.06) and the
acrophase was delayed 5.6 t 2.1 h from baseline (NS).
Rhythm amplitude
was dampened compared with baseline conditions (averaging
49 t 6 vs. 69 t- 7%, NS). The
TSH profile exhibited two, rather than one, acrophases,
occurring at 2340 t 0015 and 1020 t 0030, respectively.
The first acrophase appeared to be a reestablishment
of

Analysis

ChoZesteroZ FSR and cortisol. On the baseline day

there was a consistent
negative
cross correlation
between the rhythms of cholesterol FSR and cortisol. The
r value averaged -0.75 t 0.03 at a lag of +1.2 t 0.9 h
(P < 0.05; cholesterol FSR leading cortisol). This strong
inverse relationship
persisted on shifted day 1, when the
sleep-wake
and light-dark
cycles had been reversed,
averaging
- 0.79 t 0.06 at a lag of +l.O t 0.9 h
(P < kOvs>. On day 2 and day $ cross correlations
between the rhythms
of cholesterol
FSR and cortisol
were no longer significant
(day 2 -0.23 t 0.06 at a lag
of +4.2 t 1.2 h;day4 -0.23 t 0.05at alagof +0.8 t 1.4
h), reflecting
the dissociation
of these two rhythms
during adaptation
to the 12-h simulated time zone shift
as well as differences in individual
rates of adaptation.
ChoZesteroZ FSR and TSH. Under baseline conditions,
temporal
changes in cholesterol
synthesis were less
consistently
correlated with the profile of plasma TSH
than with that of cortisol. The largest coefficient of cross
correlation
between
the cholesterol
FSR and TSH
rhythms on the baseline day was +0.53 t 0.09 at a lag of
- 3.8 t 1.4 h (TSH leading cholesterol FSR), which was
not statistically
significant.
However,
on shifted day 1,
i.e., during
sleep deprivation
and recovery
sleep, a
significant positive correlation
between the two rhythms
was detected, with a maximal coefficient of cross correlation of +0.71 t 0.06 at a time lag of -5.2 t 0.06 h
(P < 0.05). On day 2 and day 4, the cross correlations
between the two rhythms
were inconsistent
and no
longer significant,
indicating
a dissociation
of the two
rhythms.
On day 2, the maximum
coefficient
of cross
correlation
was +0.33 t 0.02 at a lag of + 1.3 t 0.8 h in 4
subjects and -0.28 at a lag of +6.0 in the fifth subject.
On day 4, the maximum coefficient was +0.30 t 0.04 at
a lag of -0.3 t 0.3 h in four subjects and - 0.30 with
zero lag in the fifth subject.
DISCUSSION
The present results provide an unequivoca
demonstration of the existence of a marked diurnal rh tihmicity in
human de novo cholesterol synthesis that displays little
between-individual
variability
under controlled
conditions. Under baseline conditions,
the intrasubject
variabilities
in the timing of the zenith and nadir were
remarkably
stable, ranging from 2200 to 0100 and 0700
to 1130, respectively.
The amplitude
was quite large but

E496

DIURNAL

RHYTHM

OF HUMAN

somewhat
more variable
(range 28-142%),
with the
mean diurnal
variation
in vivo cholesterol
FSR being
O.OO/day in the morning
nadir and O.OS/day at the
nighttime
acrophase. These minimal and maximal rates
of cholesterol
synthesis are consistent
with previous
estimations
obtained under less controlled
conditions
and with a lower sampling frequency (17). Furthermore,
the amplitude
of the diurnal
variation
in cholesterol
FSR by direct measurement
in the present study is
similar in magnitude
to that reported for plasma levels
of mevalonate
(15,22).
In the current
study, the mean maximal
rate of
cholesterol synthesis under baseline conditions occurred
at midnight-and
the minimal rate occurred in the midmorning. Our estimation
of the timing of the peak of the
diurnal variation
is thus considerably
earlier than the
0600 zenith reported
in the first investigation
of the
diurnal rhythm
of human cholesterol
synthesis using
deuterium
incorporation
(17). However,
our estimation
is nearly identical to that reported in a later study using
the same technique (15). Moreover,
the present findings
are compatible with two previous studies of the diurnal
variations
of plasma cholesterol precursor levels, which
both reported
maximal values between midnight
and
0400 (18, 22). These latter small discrepancies
may
reflect differences in sampling frequency. In the present
investigation,
the 2-h sampling
frequency
was more
frequent than in previous studies to obtain the highest
peak resolution
with the least amount of measurement
noise (unpublished
data). However, differences in study
conditions related to activity level, diet composition,
and
mealtimes could also account for discrepancies
in the
timing of the zenith and nadir. The current protocol was
specifically
designed to maximize interindividual
synchronization
with respect to these variables before our
controlled intervention.
A striking negative correlation
was found between the
rhythm
of cholesterol
synthesis and that of plasma
cortisol during both the baseline day and the first day of
simulated jet lag, suggesting that the control of cholesterol synthesis by circadian timing could be mediated by
cortisol. In this respect, our findings in the human are
quite similar to those reported in laboratory
animals.
Adrenalectomy
in rats has been demonstrated
to decrease (19) or abolish (7) the rise in enzyme activity that
occurs during
the dark phase. A single injection
of
hydrocortisone
administered
to adrenalectomized
rats 3
h before the expected zenith in enzyme activity resulted
in a twofold increase in the activity of the enzyme at the
zenith, to values similar to that observed in controls
(19). However,
administration
of hydrocortisone
3 h
before the expected minimum
did not result in a change
in activity of the enzyme at the nadir (19), suggesting
that glucocorticoids
are not the only factor triggering
induction of the enzyme. These animal findings, supporting a positive association between corticosteroids
and
cholesterol
synthesis, contrast with the observation
in
the present study of a negative correlation.
However, in
human skin fibroblasts,
physiological
concentrations
of
hydrocortisone
were reported to suppress, rather than
activate, the cholesterol synthetic pathway (12).

CHOLESTEROL

SYNTHESIS

The first day of the simulated

time shift included
nighttime
sleep deprivation
and daytime recovery sleep,
and a reversal of the light-dark
cycle. Under conditions
of shifted. sleep-wake
and light-dark
cycles without
concomitant
shift of mealtimes
(shifted
day I), the
diurnal pattern of cholesterol
synthesis was essentially
unaltered,
demonstrating
that the rhythm is not acutely
regulated by the sleep-wake cycle or lighting conditions.
In addition,
results indicate that hormones
that are
strongly dependent
on sleep (27, ZS), such as growth
hormone
(GH), are unlikely to have immediate
effects
on cholesterol synthesis. The observation
that the abrupt
displacement
of sleep, and therefore of sleep-related
GH
release, is not associated with detectable
changes in
cholesterol
synthesis is in agreement with the study by
Boyle et al. (2), who measured plasma mevalonate
and
GH levels in GH-deficient
and -sufficient
patients and
found no correlation
between peak nocturnal
mevalonate concentrations,
fasting mevalonate
concentrations,
and GH levels. It is possible, however, that the effects of
sleep-wake reversal on cholesterol
FSR rhythm
might
take longer than 1 day to appear. Over the 4 shifted
days, however, comparison
of the rates of adaptation
of
the rhythm of cholesterol synthesis and of the rhythm of
cortisol, which is primarily
driven by the central circadian signal and only slightly influenced
by mealtimes
(29), provided evidence that neither sleep-wake cycle nor
the circadian clock was the primary factor in synchronizing the rhythm
of cholesterol
synthesis. Conclusions
based on these comparisons
were further
strengthened
by examining the rate of adaptation
of the TSH rhythm,
which is influenced
by both circadian rhythmicity
and
sleep, but not by meal intake.
The dissociation
between the rhythms of cholesterol
synthesis from these of cortisol and TSH in the course of
adaptation
to jet lag does not conclusively
exclude a role
for the circadian
clock in the control of cholesterol
synthesis. To do so requires a demonstration
that the
rhythm of cholesterol
synthesis does not persist when
caloric intake is continuous,
rather than distributed
in
three meals. In this regard, Jones et al. (14) have
demonstrated
a diminution
of the rhythm when enteral
intake was divided over six meals taken at 4-h intervals.
This study by Jones et al., however,
only involved
a
small number of subjects, and a small rhythm may have
escaped statistical significance.
Further
studies during
continuous
enteral nutrition
or constant glucose infusions will be necessary to address this issue.
On the baseline day, the daily maximum of cholesterol
synthesis was coincident
with peak levels of TSH, a
hormone that is normally inhibited
by nocturnal
sleep.
As anticipated,
a more than twofold
increase in the
amplitude of the TSH rhythm was observed during sleep
deprivation
(1,3,21). This major alteration
in the profile
of TSH concentrations
was associated with a modest
elevation of the acrophase of the rhythm of cholesterol
synthesis,
and these concomitant
changes in both
rhythms were reflected in an increase in their coefficient
of cross correlation.
These findings raise the possibility
that TSH may exert an effect on cholesterol
synthesis,
particularly
during
sleep deprivation.
During
normal

DIURNAL

RHYTHM

OF HUMAN

sleep-wake conditions,
the diurnal variation
in thyroid
hormone secretion is of low amplitude
and may even be
undetectable
(1,29). However,
during sleep deprivation,
a well-defined
nocturnal
increase in thyroid
hormone
levels parallels the elevation of TSH levels (29). Because
the activity of HMG-CoA
reductase has been demonstrated to be affected by thyroid hormone (10, ZO), it is
conceivable
that the activation
of the pituitary-thyroid
axis associated with sleep deprivation
on the 1st day of
the simulated time zone shift exerted a modest modulatory influence on cholesterol synthesis.
In contrast to the absence of a shift in the diurnal
rhythm of de novo cholesterol synthesis on day 1 when
sleep and the light-dark
cycles were altered, the timing
of the nadir and zenith of cholesterol
synthesis was
altered on the 1st day after the meal schedule was
shifted by 12 h (day 2). This is similar to studies
performed
in rats (8). Furthermore,
because the diurnal
rhythm in humans is suppressed by feeding six meals at
4-h intervals around the clock (14) and cholesterol FSR
is suppressed on the 1st day of fasting (16), it is likely
that mealtime plays a major role in the regulation
of the
diurnal variation
of de novo cholesterol
synthesis. Further studies in which only mealtime is varied, however,
should be performed.
In conclusion, the present study demonstrates
that de
novo cholesterol
synthesis varies dramatically
across
the day, indicating that cholesterol synthesis in humans
is a tightly regulated
system. This 24-h rhythm is not
acutely dependent
on the sleep-wake cycle. Mealtime
appears to play a major role in determining
the timings
of the daily maximum
and minimum
synthetic rate.
Further studies are needed to determine the importance
of the mealtime for cholesterol homeostasis in response
to dietary cholesterol.
We thank Dr. Neil Scherberg and the staff of the Thyroid Function
Laboratory
for TSH determinations,
William Pugh for cortisol assays,
John Lukens
for measurements
of cholesterol
levels, and Rachel
Leproult
for help with computerized
data analysis. We gratefully
acknowledge
the skillful
assistance
of Jacqueline
Imperial,
Beena
Loharikar,
and the staff of the University
of Chicago Clinical Research
Center.
This work was partially supported by National Institutes
of Health
Grants HL-45574,
DK-26678,
DK-41814,
and RR-00055
and Grant
F49620-94-1-0203
from the Air Force Office of Scientific Research.
Address for reprint requests: D. A. Schoeller, Dept. of Medicine, MC
4080, Univ. of Chicago, 5841 South Maryland Ave., Chicago, IL 60637.
Received 30 November

1994; accepted in final form 6 April

CHOLESTEROL

5.
6.

7.

8.

9.

10.

11.
12.

13.
14.
15.

16.

17.

18.

19.

20.

1995.

REFERENCES
1. Allan,
J. S., and C. A. Czeisler.
Persistence
of circadian
thyrotropin
rhythms
under constant conditions and after lightinduced shifts of circadian phase. J. CZin. EndocrinoZ. Metab. 79:
508-512,1994.
2. Boyle, P. J., A. Avogaro,
L. Smith, D. M. Bier, A. S. Pappu,
R. Illingworth,
and P. E. Cryer.
Role of GH in regulating
nocturnal
rates of lipolysis and plasma mevalonate
levels in
normal and diabetic humans. Am. J. PhysioZ. 263 (EndocrinoZ.
Metab. 26): E168-E172,
1992.
3. Brabant,
G., K. Prank,
U. Ranft, T. Schuermeyer,
T. 0. F.
Wagner,
H. Hauser,
B. Kummer,
H. Feistner,
R. D. Hesch,
and A. von ziir Muhlen.
Physiological
regulation of circadian
and pulsatile thyrotropin
secretion in normal man and woman. J.
Clin. Endocrinol.
Metab. 70: 403-409,199O.
4. Chakrabarty,
K., J. Radhakrishnan,
R. Sharifi,
M. F.
Mozes, J. R. Manahgog,
and H. Jeffay. Lipogenic activity and

21.

22.

23.

24.

SYNTHESIS

E497

brown
fat content of human perirenal
adipose tissue. CZin.
Biochem. 21: 249-254,1988.
Cleveland,
W. S. Robust locally weighted regression and smoothing scatterplots.
J. Am. Stat. Assoc. 74: 829-836, 1979.
Deitschy,
J. M., and D. K. Spady. Measurements
of rates of
cholesterol
synthesis
using tritiated
water. J. Lipid Res. 25:
1469-1476,1984.
Edwards,
P. A. Effect of adrenalectomy
and hypophysectomy
on
the circadian rhythm of p-hydroxy+-methylglutaryl-coenzyme
A
reductase activity in rat liver. J. BioZ. Chem. 248: 2912-2917,
1973.
Edwards,
P. A., H. Muroya,
and R. G. Gould.
In vivo
demonstration
of the circadian rhythm of cholesterol biosynthesis
in the liver and intestine of the rat. J. Lipid Res. 13: 396-401,
1972.
Feingold,
K. R., G. Zsigmond,
S. R. Lear, and A. H. Moser.
Effect of food intake on intestinal
cholesterol
synthesis
in man.
Am. J. PhysioZ. 251 (Gastrointest.
Liver Physiol. 14): G362-G369,
1986.
Gruder,
W., I. Nolte, and 0. Wieland.
The influence of thyroid
hormones on P-hydroxy+methylglutaryl
coenzyme A reductase
on rat liver. Eur. J. Biochem. 4: 273-278, 1968.
Hamprecht,
B. C., C. Nussler,
and F. Lynen.
Rhythmic
changes of hydroxy-methyl-glutaryl-coenzyme
A reductase activity in livers of fed and fasted rats. FEBS Lett. 4: 117-121, 1969.
Henze, K., B. J. Kudchodkar,
A. Chait,
J. J. Albers,
and
E. L. Bierman.
The effect of hydrocortisone
on cholesterol
metabolism
of cultured
human skin fibroblasts.
Biochim. Biophys. Acta. 666: 199-204,198l.
Jenkins,
G. M., and D. G. Watts.
Spectral Analysis
and Its
Application.
San Francisco, CA: Holden Day, 1968.
Jones,
P. J. H., C. A. Leitch,
and R. A. Pederson.
Mealfrequency effects on plasma hormone concentrations
and cholesterol synthesis in humans. Am. J. CZin. Nutr. 57: 868-874, 1993.
Jones, P. J. H., A. S. Pappu,
D. R. Illingworth,
and C. A.
Leitch.
Correspondence
between plasma mevalonic acid levels
and deuterium uptake in measuring human cholesterol synthesis.
Eur. J. CZin. Invest. 22: 609-613,1992.
Jones, P. J. H., A. M. Scanu, and D. A. Schoeller.
Plasma
cholesterol synthesis using deuterated water in humans: effects of
short term food restriction.
J. Lab. CZin. Med. 111: 627-633,
1988.
Jones,
P. J. H., and D. A. Schoeller.
Evidence for diurnal
periodicity
in human cholesterol
synthesis.
J. Lipid Res. 31:
667-673,199O.
Miettinen,
T. A. Diurnal
variation
of cholesterol
precursors
squalene and methyl sterols in human plasma lipoproteins.
J.
Lipid Res. 23: 466-473, 1982.
Mitropoulos,
K. A., and S. Balasubramanian.
The role of
glucocorticoids
in the regulation of the diurnal rhythm of hepatic
P-hydroxy-P-methylglutaryl-coenzyme
A reductase
and cholesterol 7a-hydroxylase.
Biochem. J. 160: 49-55,1976.
Ness,
G. C., R. E. Dugan,
M. R. Lakshmanan,
C. M.
Neprokroeff,
and J. W. Porter.
Stimulation
of hepatic p-hydroxy-P-methylglutaryl
coenzyme A reductase activity in hypophysectomized rats by L-triiodothyronine.
Proc. NatZ. Acad. Sci. USA
70: 3839-3842,1973.
Parker,
D. C., L. G. Rossman,
A. E. Pekary,
and J. M.
Hershman.
Effect of 64-hour sleep deprivation
on the circadian
waveform
of thyrotropin
(TSH): further evidence of sleep-related
inhibition of TSH release. J. CZin. Endocrinol.
Metab. 64: 157161,1987.
Parker,
T. S., D. J. McNamara,
C. Brown,
0. Garrigan,
R.
Kalb, H. Batwin,
and E. H. Ahrens,
Jr. Mevalonic acid in
human plasma: relationship
of concentration
and circadian rhythm
to cholesterol synthesis rates in man. Proc. NatZ. Acad. Sci. USA
79: 3037-3041,1982.
Rechtschaffen,
A., and A. Kales. A ManuaZ of Standardized
Terminology,
Techniques and Scoring System for Sleep Stages of
Human Subjects. Washington,
DC: Government
Printing Office,
1968.
Schwartz,
C. C., R. Vlahcevic,
M. Berman,
J. G. Meadows,
R. M. Nisman,
and L. Swell.
Central role of high density
lipoprotein in plasma free cholesterol metabolism. J. CZin. Invest.
70: 105-116,1982.

E498

DIURNAL

RHYTHM

OF HUMAN

25. Shapiro,
D. J., and V. W. Rodwell.
Diurnal
variation
and
cholesterol
regulation
of hepatic HMG-CoA
reductase
activity.
Biochem. Biophys. Res. Commun. 37: 867-872,1969.
26. Theorell,
T., and T. Akerstedt.
Day and night work: changes in
cholesterol,
uric acid, glucose and potassium
in serum and in
circadian patterns of urinary catecholamines
excretion. Acta Med.
Stand. 200: 47-53,1976.
27. Van Cauter,
E. Hormones
and sleep. In: The Pharmacology
of
Sleep, edited by A. Kales. Berlin: Springer Verlag. In press.
28. Van Cauter,
E., M. Kerkhofs,
A. Caufriez,
A. Van Onderbergen, M. 0. Thorner,
and G. Copinschi.
A quantitative
estimation of GH secretion in normal man: reproducibility
and relation

CHOLESTEROL

SYNTHESIS

to sleep and time of day. J. CZin. EndocrinoZ.

Metab. 74: 14411450,1992.
29. Van Cauter,
E., and F. W. Turek.
Endocrine
and other
biological rhythms.
In: Endocrinology,
edited by L. J. DeGroot.
Philadelphia, PA: Saunders, 1994, vol. 3, p. 2487-2548.
30. Vondra,
K., V. Brodan,
M. Dobiasova,
V. Vitek,
and
J. Kopecka.
Effect of sleep deprivation
on cholesterol
metabolism and triglyceridaemia
in male volunteers.
Eur. J. AppZ.
Physiol. Occup. Physiol. 55: 83-87, 1986.
31. Waurin,
J., and U. Schibler.
Expression
of the liver-enriched
transcriptional
activator protein DBP follows a stringent
circadian rhythm. CeZZ 63: 1257-1266,199O.