- Email: [email protected]
Rhythms in Cholesterol, Cholesteryl Esters, Free Fatty Acids, and Triglycerides in Blood of Lactating Dairy Cows
Rhythms in Cholesterol, Cholesteryl Esters, Free Fatty Acids, and Triglycerides in Blood of Lactating Dairy Cows JOEL BITMAN, D. L. WOOD, and A. M. LEFCOURT Milk Secretion and Mastitis Laboratory Livestock and Poultry SCiences Institute US Department of Agriculture, ARS Beltsville. MD 20705 ABSTRACT
Blood samples from six lactating dairy cows were analyzed to determine whether circulating neutral lipids exhibit rhythmic variations. Plasma neutral lipids were measured by quantitative TLC on every fourth integrated IS-min blood sample taken over 48-h periods. Cows were housed in an environmental chamber at 20·C with 16 h light:8 h dark (lights on at 0700 h). fed daily at 0900 h. and milked at 0830 and 2000 h. Other variables monitored included: body temperature. ammonia nitrogen, urea nitrogen. glucose. triiodothyronine, thyroxine. somatotropin. insulin, cortisol. and prolactin. Mean concentrations of cholesterol. cholesteryl esters, free fatty acids. and triglycerides were 21.4. 175.4. 3.1, and 6.3 mg/dl, respectively. Visual and power spectral analysis of the pulsatile fluctuations in lipids indicated rhythms with periods of 2 to 3 h. Amplitudes of rhythms for free fatty acids and triglycerides were 60% of mean concentrations and for cholesterol and cholesteryl esters were 20% of mean concentrations. The presence of these rhythms was conserved when data were averaged across time by cow. However, because of nonstationary conditions. rhythms identified by spectral analysis were not statistically significant. There was no evidence of circadian patterns in circulating neutral lipid components. All other metabolic and hormonal variables except cortisol exhibited distinct circadian rhythms.
Received May 24,1989. Accepted October 12. 1989.
1990 J Dairy Sci 73:948-955
(Key words: rhythms)
cows,
plasma
lipids.
INTRODUCTION
The amounts and composition of the lipids circulating in the blood of dairy cattle are dependent upon a number of physiological variables. The nature of the diet, time since feeding, age, breed. pregnancy, and stage of lactation may all affect lipid content and composition (5). Although a number of studies have characterized some of the changes occurring in blood lipids during lactation (5), there has been little investigation of variations in plasma lipids of cattle over 24-h periods. One objective of this study was to determine whether the circulating lipids in lactating dairy cows exhibited circadian rhythms. A continuous study of the amounts of lipids circulating in the blood was undertaken to aid in understanding the control and stability of these energy components during the process of milk fat synthesis. As part of an intensive study of blood hormones in early lactation, we simultaneously measured plasma concentrations of selected hormones, metabolites, and neutral lipids. A preliminary report on hormone and metabolite rhythms has been published (13). MATERIALS AND METHODS Animals, RaUons, and Housing
Six Holstein cows (4 to 7 mo into lactation) were housed in stanchion stalls (1.22 )( 2.59 m) in an environmental chamber (6.40 x 9.14 m; 20.0 ± .soC; relative humidity: 45 to 60%). The cows had been acclimated to the chamber for at least 6 wk. The lighting cycle was 16L:8D with lights on at 0700 h. The cows were fed a total mixed ration once daily at 0900 h. Water was available ad libitum. The cows were milked at 0830 and 2000 h. The pattern of feed consump948
BLOOD LIPID RHYTHMS IN COWS
tion was assessed by weighing the amount of ration remaining in th individual feed trough at I, 3, 5, 7, 9, II, 13, IS, and 24 h after placing the total ration in the trough. Body temperature was monitored continuously by radiotelemetry (14). Radiob"ansmitters were implanted between rear udder quarters and temperatures were recorded every 1.4 min (1024 readings per transmitter/d). Blood sampling
An automated blood sampling system (12) was used to collect continuous, integrated blood samples over IS-min intervals for 48 h (196 blood samples). Samples of blood (15 ml) diluted with heparin saline were collected from two cows simultaneously into test tubes containing sodium fluoride (to inhibit enzymes). Plasma was obtained by centrifugation and stored at -20·C until analyzed. Sample Analysis
A 1-ml aliquot of every fourth integrated IS-min plasma sample from the six cows was extracted with 9 ml of chloroform: methanol (2: 1) according to Folch et al. (6) as described earlier (2). Duplicate 5o-J.LI aliquots of the chloroform extract containing the lipids were applied to a Baker preadsorbent silica gel plate (20 x 20 cm, J. T. Baker Chemical Co., Phillipsburg, NJ 08865). After TLC development (2), the separated lipids were visualized by dipping them in 10% cupric sulfate in 8% phosphoric acid and charred by heating. Neutral lipid classes were determined by quantitative densitometric scanning of the charred TLC plates using the Shimadzu CS-930 Dual Wavelength TLC Scanner (Shimadzu Scientific [nstruments, Inc., Columbia, MD). Plates were scanned in a linear mode at 370 nm with a tungsten lamp and sample amounts determined as described earlier (2). Glucose and urea nitrogen were analyzed with a Technicon Autoanalyzer as described previously (8). The thyroid hormones and cortisol were analyzed by radioimmunoassay (10). Analysis
Identification of episodic or rhythmic components of biological signals is a complex and
949
difficult problem (15). Difficulties arise because: 1) biological waveforms are often periodic but not sinusoidal, 2) the waveform is not sampled frequently enough, and 3) the time interval between peaks or episodes varies. Not sampling a waveform frequently enough can cause aliasing problems whereby artificial frequencies are produced (4, 24). If the time between peaks varies, the waveform is considered nonstationary. The problem in analyzing a nonstationary waveform is evident when the autocorrelation function is examined. If the time interval between events varies, the autocorrelation function will tend toward zero (4, 24). Frequency spectrums, or periodograms, can be expressed in terms of the autocorrelation function. A frequency spectrum is the decomposition of a waveform into its component frequencies with the y-axis showing magnitude or power and the x-axis showing frequency (or the inverse of frequency, which is period). There are two common methodologies for identifying rhythms in biological signals. The first method involves identifying individual peaks and using this information to calculate peak heights and time intervals between peaks. In this study, peaks were identified visually using the technique of Santen and Bardin (23), who defined peaks as rises of greater than 20% from preceding minimum values. Alternatively, waveforms can be analyzed using spectral analysis or autocorrelation. With spectral analysis, individual peaks are not identified, but rather, just the characteristics of the entire waveform. In this study, periodograms were calculated for linearly detrended data and resulting frequency estimates smoothed over three sequential estimates using the smoothing function: .23, .54, and .23 (25). To ease comparisons, resulting spectrums were normalized to individual variances. Attempted analyses of biological waveforms using spectral analysis techniques often produce results that are not statistically significant because the waveforms are nonstationary (15). Often data from similar experiments can be averaged across time to improve the signal to noise ratio (15, 24). This technique, sometimes called signal averaging, is especially useful when analyzing data from animals that have been entrained to the same environmental cues, e.g., lighting, feeding. For this reason, data from all six cows were averaged by time and analyzed. Journal of Dairy Science Vol. 73. No.4, 1990
950
BITMAN ET AL.
TABLE 1. Distribution of plasma lipids in various classes in six lactating Holstein cows. Cow number
lipid class l
11
83
91
97
55
65
3.8 2.4 25.6 205.0 236.9
5.1 9.3 5.5 8.5 3.2 2.3 3.3 3.4 16.5 23.8 19.3 24.6 129.4 195.5 181.4 178.8 211.3 154.2 231.1 213.5 Percentage of tOla1 NL in lipid class 4.4 2.6 3.4 3.7 2.1 1.4 1.6 I.l 9.1 11.7 10.3 10.7 84.0 85.0 84.6 84.6
Mean
SE
n2
6.3 3.1 21.4 175.4 206.3
.2 .1 .1 12.4
50 49 50 50 50
.4 .2 .4 .4
50 49 50 50
Amounts (mgl100 ml)
TG
6.7 4.0 FFA 18.6 C 161.4 CE Total NL 190.7
TG
3.5 2.1 9.7 84.6
FFA C CE
1.6 1.0 10.8 86.6
= Triglycerides; C = cholesterol; CE = cholesteryl = Nwnber of samples per cow (l samplcJb).
ITO
~
RESULTS AND DISCUSSION
Plasma Neutral Lipid
Mean values for total neutral lipids and the distribution into lipid classes are shown in Table I. Total neutral lipids (NL) varied within a fairly narrow range of 154 to 237 mg/loo ml. The individual lipid classes also exhibited rather narrow ranges of variation. This can be seen readily when lipid in each class is expressed as percent of total lipid. Cholesteryl ester (CE), the principal neutral lipid class, contained a constant 85% of the total NL (range 83.95 to 86.55%). An additional 10% of the plasma NL was free cholesterol (C). Triglycerides (TO) comprised about 3% and FFA about 1.5% of total plasma NL. Comparison of mean values of plasma lipids (Table 1) with values in the literature (7, 11, 19, 20, 26) indicate a high degree of agreement both in absolute amounts and relative distribution by class. Plasma Neutral Lipid Rhythms
Plasma concentrations of C, CE, TO, and FFA in the six cows for the 2-d measurement period are shown in Figure 1. Triglyceride and FFA data for two cows were excluded from analysis because of excessively elevated concentrations at the stan-up of the experiment; Journal of Dairy Science Vol. 73. No.4. 1990
esters; and NL
1.1
3.2 1.6 10.4 84.9
= neutral
lipid.
these elevations were never seen during the second 24 h of sampling. It was thought that these values represented a response to unavoidable handling at the onset of each trial. In addition, large transients precluded the use of times series techniques for analyses. Visual analyses (23) indicated a time between peaks of approximately 2.5 h for all lipid classes (Table 2). Ampliwdes of peaks were: 20% of mean for C and CE and 60% of mean for FFA and TO. Because the cows were housed in an environmental chamber at constant temperature (20cC), changes in lipid concentrations with time were not due to fluctuations in ambient temperawre. There was no evidence of either a circadian peak or displacement of the pulsatile pattern by the light-dark cycle in any of the lipid classes (Figure 1). However, the presence of the ultradian peaks with 2.5-h periods was conserved when data were averaged across cows by time, indicating some form of entrainment. Recently, Blum et al. (3) measured a number of hormones and metabolites over a 24-h period in lactating cows fed high and low energy diets. They did not observe diurnal changes in blood TO, C, or phospholipids; these findings are in accord with our results. Pattern of Feed Consumption
The cows were fed once daily at 0900 h. In nonrwninants, the onset of feeding generally results in an increase in blood lipids. In rumi·
951
BLOOD LIPID RHYTHMS IN COWS TABLE 2. Pulse analysis of plasma neutral lipids by visual analysis (23). From visual analysis
Lipid class I
Average
CE C TO FFA
- - - - - - - (mgldI) - - - - - - 206.3 176.0 33.8 19.2 21.4 21.1 3.7 17.5 55.8 6.3 5.9 3.3 3.1 2.7 1.8 68.2
Baseline
Amplitude
% of Mean
Period (h)
2.6 2.6 2.7 2.7
ICE ::: Cholesteryl esters; C ::: cholesterol; TO ::: triglycerides.
nants. feeds have a relatively long residence time in the rumen, resulting in a buffering effect with no abrupt rises in plasma nutrients (5). In addition, although the total daily ration was provided to the animals at one time (0900 h), the cows self-regulated their intake in patterns probably closely related to ruminal fill (Table 3). During the 1st h after presentation of feed, the cows consumed 36% of total ration, whereas during the next 6 h, they consumed only about 20%. This was followed by a period of greater consumption: 35% was eaten during the next 6 h. Only 2.5% of the feed was eaten from midnight to 0900 h. This feeding pattern suggests that the initial high consumption of 36% of the ration fills the rumen, and, by a feedback mechanism, limits ingestion of additional feed during the next 6 h. Passage of this food from the rumen then permits ingestion of additional feed. The daily pattern of feed consumption, however, was not reflected in the plasma concentrations of the neutral lipids. Frequency Analysis
Spectral analysis of neutral lipid data identified no significant circadian rhythms. Representative periodograms for cow 91 for each lipid component are shown in Figure 2. The 2 to 3-h patterns identified by visual analysis were also indicated for C and CE in five of the six cows and for TG in three of four cows. The sensitivity and reliability of the spectral analyses were compromised by two factors. First, spectral analysis assumes a stationary time series, e.g., that the time between peaks is always identical. Second, samples must be taken at twice the highest frequency component of the time series and, because of noise, samples should be taken at three to five times this
frequency to prevent aliasing problems. Visual and autocorrelation analyses show that the time between peaks for NL varies. In addition, a 9Q-min pattern was demonstrated for most of the metabolic and hormonal variables measured (13). If this higher frequency rhythm exists for any of the NL, it could not be detected unless all samples, and not every fourth sample, were analyzed, and its existence would create an aliasing problem (4, 24) that could obscure true rhythms or create nonexisting rhythms. The inability to identify any statistically significant rhythms with periods greater than 200 min using power spectral analysis probably is due primarily to the nonstationary nature of the data. Still, power spectral analysis did indicate rhythms at the same frequencies identified by visual analysis, which adds credibility to these findings. Comparisons with Hormonal and Metabolic Variables
In contrast with our finding for NL, all other metabolic and hormonal variables except cortisol exhibited distinct circadian rhythms (13). In agreement with previous findings (1), body temperature exhibited a circadian rhythm as well as a higher frequency rhythm with a period of 90 min. In addition, spectral analysis of hormone and metabolite data showed a 90 to 120 min rhythm for most variables. This 90- to 120-min rhythm cannot be identified in the lipid data because of the longer effective sampling interval that resulted when only every fourth sample was analyzed (see preceding discussion of aliasing). For comparison purposes only and to indicate that rhythms can be observed in certain parameters, plasma urea nitrogen, triiodothyroJournal of Dairy Science Vol. 73, No.4, 1990
952
BITMAN ET AL,
CHOLESTEROL ESTER
TRIGLYCERIDE 15,
250,
12
200 "
,
.',..
9
. - ' . - .' \ ; "
6\
'.'
150 i
3
".-
250
.\
200
, ',,: I," .'
150
j1J
I
'/..... i .
j
1 .'
~
I
.
/'
",
,
i ~J \, '-.-.. '. ~"
~_-'j"
1\ '
J '. . '", ."" "
\/
.
..
t...-
\,,j..... \."')
" .\ . . . . .' ,...,.,~"", 'i~\1 ."'\ ,.. ',: -,..J, ~ V~'
.
~.
I(
~
.-
/
,tlo
\,.
250 .
-.
200'
~ ~ ~
200
j
150
1
~~
1
12
c5
.,
,
_ '!
, "" J ' - 1 " . ' , I...
~
"
'
~
.... \:.""
.......
" .........
f
9 t
~
6 ~" .. '" ,r""'- ...,..'
~
3'
\
, \""
15 :
../
12 200
9
i
r'.,
j
150
9
200
'\
.\'
",Vi
I2
18
24
06
12
18
24
t
1 !', T'--o. ~~ ".. . . ",,""'-,;"V'~r 3 ......../V'~ ''v"" 0----------------6
i L"V............l\.--""'\ \ '-"'\v""\./ \oJ V L
100
\r'
~j
100
150
","., /\\. Iir'j".' [ \ ~'i'
/ ..~.;'./.
-'
6 j
06
12
18
24
06
12
I8
24
06
W.AWL::::J
HOURS
HOURS
FREE FATTY FREE CHOLESTEROL 35 30 25
./
". , \ '
20
~~
20
I
"',..
; "--'/
'\
~\~'"
. '-~'\.'''''''-'''' ."",-
,\
-'''~ '\. .....
6 •
-
4
6 .
30
c5
I
2 ;.
4 t ...
25
--'
20
'-
~ ~~ I
i
0
2
'"
6 t
~
. 1
1
1
2 ~
15 \
,
ii, 1,1. ,,'". I 1\ j
i
4
~
O· 6
t
4 ,
-.
12
"l .. ,.; - - "....... ...... , .....
18
24
'"It'' 06
'12
2 18
24
06
HOURS
t
_'-" ..
"
.........""
. .
-
o ~---------------'2
18
24
06
12
18
24
06
HOURS
Figure 1. Temporal patterns of plasma concentrations in cholesteryl esters, free cholesterol, triglycerides and free fally acids, measured in six cows over 48 h. Plasma concentrations in milligrams per deciliter are plotted against time of day in hours. Horizontal black bars = lights off; white bars = lights on.
Journal of Dairy Science Vol. 73, No.4, 1990
953
BLOOD LIPID RHYTHMS IN COWS TABLE 3. Percentage feed COIISwnption by time interval for six lactating cows fed once daily. Percentage of total ration conswned at time interval. h
0-1
Item
Mean
35.7 \.9 18
SE n
2-3
4-5
8-9
10-11
12-13
5.2 1.1 18
10.0
6.4
10.3
12.4
12.6
4.9
13
9
15
1~
I~
I~
18
fr.7
18
18
nine (T3), and corusol for cow 91 are shown in Figure 3. Cow 91 was representative of the other five cows. and complete data for these measures will be reported at a later date. Circadian rhythms were clear and consistent for plasma urea nitrogen and T3. The plasma cortisol pattern was characterized by rapid episodic fluctuations with no evidence of a circadian pattern. Figure 4 compares the periodograms for urea nitrogen, T3, corusol, one lipid component, and cholesterol. The cholesterol periodogram, in contrast to those of urea nitrogen, T3. and corusol, gave little evidence of periodicity
17
14-15
17
16-24 2.5 .8 15
15
and the powers were very low, being only .1 to .3% of total power. Although the lipid data did not exhibit a circadian rhythm, pulsatile fluctuations were evident every 2 to 3 h. The repetitive nature of this pattern could possibly reflect a limit cycle control mechanism balancing rate of lipid entry and rate of removal in the circulatory system. This is possibly related to regulation of lipid supply (diet, production, synthesis) and utilization by peripheral tissues. The relationship of circadian metabolic and honnonal processes in the other variables studied to the observed 2-h higher frequency mythm in lipids cannot be assessed readily.
::E
E 12
9
cr
2,5
w
~
TRIIODOTHYRONINE
0
a.. 1.0 - - - - - - - - - - - - - - o
w N
:::J
E
2.0
'"c
I.S
"-
...:
:::li
cr
o
z
1.0
0 1.0 - - - - - - - - - - - - - - - FREE FATTY ACID
"lilllJlli:JJ ' 6
E "go
12
8 4
o 24.0
6.0 3.0 PERIOD (HOURS)
2.0
Figure 2. Periodograms of neutral lipids of cow 9 I.
a
12
18
24
06
12
18
24
06
HOURS
Figure 3. Temporal patterns of plasma concentrations of urea-nitrogen. triiodothyronine and cortisol of cow 9 I. Journal of Dairy Science Vol. 73, No.4, 1990
954
BITMAN ET AL.
f-
z
18·
w
u
'"D.-
15 .
W
'"w3= a
12
9
D.-
o
5
UREA NITROGEN
W
N
~
~
t
;t ':ODo;~"O'::E 3
te
on 24.0 5.0
•
3.0
-
s
CORTISOL
FREE CHOLESTEROL
2.0
1.5
22). The critical basic premise of metabolic profile or blood profile research was that abnormalities in blood chemistry would be indicators of metabolic disorders. In practice, however, blood metabolites are poor indicators of nutritional and physiologic status, since homeostasis is a basic biological control mechanism adapted by living organisms to prevent gross chemical imbalances (9). In addition, infrequent sampling, usually weekly or monthly in metabolic profile research (9, 16, 22), is not designed to readily reveal physiological variations. In contrast, intensive sampling, as demonstrated in our study of circulatory lipids, hormones, and metabolites, can reveal the presence of both ultradian and circadian variations.
1.0
PERIOD (HOURS)
Figw-e 4. Periodograms of urea-nilrogen. triiodothyronine, conisol and cholesterol. Power expressed as percentage (normalized for variance) is plotted against period in hours.
The lack of a circadian rhythm in the lipid data contrasts with the presence of definite circadian rhythms in a variety of other components in the same blood samples. The lack of a circadian rhythm in the cow has a parallel in findings in the rat for the lipid component, cholesterol (18). There was no apparent diurnal rhythm in cholesterol in either rat hepatic tissue or peripheral serum in spite of a marked circadian rhythm in cholesterol synthesis (17, 18). There is a marked circadian rhythm in the activity of hepatic hydroxymethylglutaryl-Coenzyme A reductase; the maximum synthetic rate occurred at 2400 h and was 3 to 10 times higher than the minimum at 1200 h (17, 18, 21). The lack of corresponding rhythmic changes in blood and liver C, the lipid product of the enzymatic synthesis, complicates interpretation of the physiological significance of the hepatic circadian rhythm in C biogenesis. Similarly, circadian rhythms of lipid synthesis (or utilization) may occur in the cow without circadian alterations in lipid components of the serum. The potential use of blood chemistry for diagnosis of clinical and nutritional problems was explored extensively in recent years (9, 16, Journal of Dairy Science Vol. 73. No.4. 1990
REFERENCES 1 BiU"nan, J., A. Lefcourt, D. L. Wood. and B. Stroud. 1984. Circadian and ultradian temperature rhythms of lactating dairy cows. J. Dairy Sci. 67:1014. 2 BiU"nan. J., D. L. Wood. M. Hamosh. P. Hamosh, Y.• and N. R. Mehta. 1983. Comparison of the lipid composition of breast milk from mothers of term and preterm infants. Am. J. Clin. Nutr. 38:300. 3 Blum, 1. W .• F. Jans, W. Moses. D. Frohli. M. Zemp, M. Wanner, 1. C. Hart, R. Thun, and U. Keller. 1985. Twenty-four-hour pattern of blood hormone and metabolite concenlrations in high-yielding dairy cows: effects of feeding low or high amounts of starch or crystalline fat. Zenlralbl. Veterinaermed. Reihe A. 32:401. 4 Chatfield. C. 1984. The analysis of time series: an introduction. 3rd ed. Olapman and Hall, New York. NY. 5 Christie. W. W. 1981. The composition. structure and function oflipids in the tissues of ruminant animals. Page 95 in Lipid metabolism in ruminant animals. W. W. Christie, ed. Pergamon Press. New York. NY. 6 Folch, J .• M. Lees, and G. H. Sloane-Stanley. 1957. A simple method for the isolation and purification of total lipids from animal tissues. 1. BioI. Chern. 226:497. 7 Hartmann. P. E., and A. K. Lascelles. 1965. Variation in the concentration of lipids and some other constituents in the blood plasma of cows at various stages of lactation. Aust. J. BioI. Sci. 18:114. 8 Huntington, G. B. 1984. Net absorption of glucose and nitrogenous compounds by lactating Holstein cows. 1. Dairy Sci. 67:1919. 9 Jones. G. M., E. E. Wildman, H. F. Troutt, Jr., T. N. Lesch. P. E. Wagner. R. L. Boman. and N. M. Lanning. 1982. Metabolic profiles in Virginia dairy herds of different milk yields. 1. Dairy Sci. 65:683. 10 Kahl. S.• A. V. Capuco. and J. BiU"nan. 1987. Serum concentrations of thyroid hormones and extrathyroidal thyroxine-5' -monodeiodinase activity during lactation in the rat. Proc. Soc. Exp. BioI. Med. 184:144. 11 Leal. W.M.F.• andJ. G. Hall. 1968. Lipid composition of lymph and blood plasma of the cow. 1. Agric. Sci. Camb. 71:189. 12 Lefcourt, A. M.. and J. BiU"nan. 1985. Method for automatic, continuous blood sampling with remote alarm
BLOOD LIPID RHYTHMS IN COWS for clogged catheters. J. Dairy Sci. 68:2108. 13 Lefcoun, A. M.. J. Biunan, S. Kahl. R. M. Akers. G. B. Huntington, and D. L. Wood. 1985. Continuous endocrine. metabolic and deep-body temperature profiles for six cows housed in an environmental chamber. Fed. Proc. 44:1362. 14 Lefcoun, A. M.• J. Biunan. D. L. Wood, and B. Stroud. 1986. Radiotelemetry system for continuously monitoring temperature in cows. 1. Dairy Sci. 69:237. 15 Merriam.G. R.. and K. W. Wachter. 1982. Measurement and analysis of episodic hormone secrelion. Pages 325-346 in Compulers in endocrinology. D. Rodbard and G. Forti, ed. Raven Press, New York. NY. 16 Payne. J. M.• S. M. Dew. R. Manston. and M. Faulks. 1970. The use of a metabolic profile lest in dairy herds. Vet. Rec. 87:150. 17 Ramasarma. T. 1976. Biogenetic interrelationship of ubiquinone and cholesterol. Biochem. Soc. Symp. 35: 245. 18 Rao, G. S" and T. Ramasarma. 1971. Rhytlunic activity of biogenesis of cholesterol. Environ. Physiol. 1: 188. 19 Raphael, B. C.• P. S. Dimick. and D. L. Puppiooe. 1973. Lipid characterization of bovine serum lipoproteins throughout gestation and lactation. J. Dairy Sci. 56:1025.
955
20 Remond. B.. R. Toullec. and M. Joumet. 1973. Changes in dairy cows in content ofdifferent blood constituents at the end of gestation and the beginning of lactation: relationships with secretion ofmilk fat. Ann. BioI. Anim. Biochem. Biophys. 13:363. 21 Rodwell. V. W.• D. J. McNamara. and D. J. Shapiro. 1973. Regulation of hepatic 3-hydroxy-3-methylglutaryl-Coenzyme A reductase. Adv. Enzymol. 38:373. 22 Rowlands. G. J. 1980. A review of variations in the concentrations of metabolites in the blood of beef and dairy cattle associated with physiology, nutrition and disease. with particular reference 10 the interpretation of metabolic profiles. World Rev. Nutr. Diet. 35:172. 23 Santen. R. J.• and C. W. Bardin. 1973. Episodic, luteinizing hormone secretion in man. Pulse analysis. clinical interpretation. physiologic mechanisms. J. Clin. Invest. 52:2617. 24 Shumway. R. H. 1988. Applied statistical time series analysis. Prentice Hall. Englewood Cliffs, NJ. 25 Statgraphics. 1987. Statistical graphic system. STSC, Inc.• Rockville. MD. 26 Starry, J. E. and J.A.F. Rook. 1964. Lipids in the blood plasma of cows of the Friesian and Channel Island breeds. Nature 201 :926.
Journal of Dairy Science Vol. 73, No.4, 1990
Blood samples from six lactating dairy cows were analyzed to determine whether circulating neutral lipids exhibit rhythmic variations. Plasma neutral lipids were measured by quantitative TLC on every fourth integrated IS-min blood sample taken over 48-h periods. Cows were housed in an environmental chamber at 20·C with 16 h light:8 h dark (lights on at 0700 h). fed daily at 0900 h. and milked at 0830 and 2000 h. Other variables monitored included: body temperature. ammonia nitrogen, urea nitrogen. glucose. triiodothyronine, thyroxine. somatotropin. insulin, cortisol. and prolactin. Mean concentrations of cholesterol. cholesteryl esters, free fatty acids. and triglycerides were 21.4. 175.4. 3.1, and 6.3 mg/dl, respectively. Visual and power spectral analysis of the pulsatile fluctuations in lipids indicated rhythms with periods of 2 to 3 h. Amplitudes of rhythms for free fatty acids and triglycerides were 60% of mean concentrations and for cholesterol and cholesteryl esters were 20% of mean concentrations. The presence of these rhythms was conserved when data were averaged across time by cow. However, because of nonstationary conditions. rhythms identified by spectral analysis were not statistically significant. There was no evidence of circadian patterns in circulating neutral lipid components. All other metabolic and hormonal variables except cortisol exhibited distinct circadian rhythms.
Received May 24,1989. Accepted October 12. 1989.
1990 J Dairy Sci 73:948-955
(Key words: rhythms)
cows,
plasma
lipids.
INTRODUCTION
The amounts and composition of the lipids circulating in the blood of dairy cattle are dependent upon a number of physiological variables. The nature of the diet, time since feeding, age, breed. pregnancy, and stage of lactation may all affect lipid content and composition (5). Although a number of studies have characterized some of the changes occurring in blood lipids during lactation (5), there has been little investigation of variations in plasma lipids of cattle over 24-h periods. One objective of this study was to determine whether the circulating lipids in lactating dairy cows exhibited circadian rhythms. A continuous study of the amounts of lipids circulating in the blood was undertaken to aid in understanding the control and stability of these energy components during the process of milk fat synthesis. As part of an intensive study of blood hormones in early lactation, we simultaneously measured plasma concentrations of selected hormones, metabolites, and neutral lipids. A preliminary report on hormone and metabolite rhythms has been published (13). MATERIALS AND METHODS Animals, RaUons, and Housing
Six Holstein cows (4 to 7 mo into lactation) were housed in stanchion stalls (1.22 )( 2.59 m) in an environmental chamber (6.40 x 9.14 m; 20.0 ± .soC; relative humidity: 45 to 60%). The cows had been acclimated to the chamber for at least 6 wk. The lighting cycle was 16L:8D with lights on at 0700 h. The cows were fed a total mixed ration once daily at 0900 h. Water was available ad libitum. The cows were milked at 0830 and 2000 h. The pattern of feed consump948
BLOOD LIPID RHYTHMS IN COWS
tion was assessed by weighing the amount of ration remaining in th individual feed trough at I, 3, 5, 7, 9, II, 13, IS, and 24 h after placing the total ration in the trough. Body temperature was monitored continuously by radiotelemetry (14). Radiob"ansmitters were implanted between rear udder quarters and temperatures were recorded every 1.4 min (1024 readings per transmitter/d). Blood sampling
An automated blood sampling system (12) was used to collect continuous, integrated blood samples over IS-min intervals for 48 h (196 blood samples). Samples of blood (15 ml) diluted with heparin saline were collected from two cows simultaneously into test tubes containing sodium fluoride (to inhibit enzymes). Plasma was obtained by centrifugation and stored at -20·C until analyzed. Sample Analysis
A 1-ml aliquot of every fourth integrated IS-min plasma sample from the six cows was extracted with 9 ml of chloroform: methanol (2: 1) according to Folch et al. (6) as described earlier (2). Duplicate 5o-J.LI aliquots of the chloroform extract containing the lipids were applied to a Baker preadsorbent silica gel plate (20 x 20 cm, J. T. Baker Chemical Co., Phillipsburg, NJ 08865). After TLC development (2), the separated lipids were visualized by dipping them in 10% cupric sulfate in 8% phosphoric acid and charred by heating. Neutral lipid classes were determined by quantitative densitometric scanning of the charred TLC plates using the Shimadzu CS-930 Dual Wavelength TLC Scanner (Shimadzu Scientific [nstruments, Inc., Columbia, MD). Plates were scanned in a linear mode at 370 nm with a tungsten lamp and sample amounts determined as described earlier (2). Glucose and urea nitrogen were analyzed with a Technicon Autoanalyzer as described previously (8). The thyroid hormones and cortisol were analyzed by radioimmunoassay (10). Analysis
Identification of episodic or rhythmic components of biological signals is a complex and
949
difficult problem (15). Difficulties arise because: 1) biological waveforms are often periodic but not sinusoidal, 2) the waveform is not sampled frequently enough, and 3) the time interval between peaks or episodes varies. Not sampling a waveform frequently enough can cause aliasing problems whereby artificial frequencies are produced (4, 24). If the time between peaks varies, the waveform is considered nonstationary. The problem in analyzing a nonstationary waveform is evident when the autocorrelation function is examined. If the time interval between events varies, the autocorrelation function will tend toward zero (4, 24). Frequency spectrums, or periodograms, can be expressed in terms of the autocorrelation function. A frequency spectrum is the decomposition of a waveform into its component frequencies with the y-axis showing magnitude or power and the x-axis showing frequency (or the inverse of frequency, which is period). There are two common methodologies for identifying rhythms in biological signals. The first method involves identifying individual peaks and using this information to calculate peak heights and time intervals between peaks. In this study, peaks were identified visually using the technique of Santen and Bardin (23), who defined peaks as rises of greater than 20% from preceding minimum values. Alternatively, waveforms can be analyzed using spectral analysis or autocorrelation. With spectral analysis, individual peaks are not identified, but rather, just the characteristics of the entire waveform. In this study, periodograms were calculated for linearly detrended data and resulting frequency estimates smoothed over three sequential estimates using the smoothing function: .23, .54, and .23 (25). To ease comparisons, resulting spectrums were normalized to individual variances. Attempted analyses of biological waveforms using spectral analysis techniques often produce results that are not statistically significant because the waveforms are nonstationary (15). Often data from similar experiments can be averaged across time to improve the signal to noise ratio (15, 24). This technique, sometimes called signal averaging, is especially useful when analyzing data from animals that have been entrained to the same environmental cues, e.g., lighting, feeding. For this reason, data from all six cows were averaged by time and analyzed. Journal of Dairy Science Vol. 73. No.4, 1990
950
BITMAN ET AL.
TABLE 1. Distribution of plasma lipids in various classes in six lactating Holstein cows. Cow number
lipid class l
11
83
91
97
55
65
3.8 2.4 25.6 205.0 236.9
5.1 9.3 5.5 8.5 3.2 2.3 3.3 3.4 16.5 23.8 19.3 24.6 129.4 195.5 181.4 178.8 211.3 154.2 231.1 213.5 Percentage of tOla1 NL in lipid class 4.4 2.6 3.4 3.7 2.1 1.4 1.6 I.l 9.1 11.7 10.3 10.7 84.0 85.0 84.6 84.6
Mean
SE
n2
6.3 3.1 21.4 175.4 206.3
.2 .1 .1 12.4
50 49 50 50 50
.4 .2 .4 .4
50 49 50 50
Amounts (mgl100 ml)
TG
6.7 4.0 FFA 18.6 C 161.4 CE Total NL 190.7
TG
3.5 2.1 9.7 84.6
FFA C CE
1.6 1.0 10.8 86.6
= Triglycerides; C = cholesterol; CE = cholesteryl = Nwnber of samples per cow (l samplcJb).
ITO
~
RESULTS AND DISCUSSION
Plasma Neutral Lipid
Mean values for total neutral lipids and the distribution into lipid classes are shown in Table I. Total neutral lipids (NL) varied within a fairly narrow range of 154 to 237 mg/loo ml. The individual lipid classes also exhibited rather narrow ranges of variation. This can be seen readily when lipid in each class is expressed as percent of total lipid. Cholesteryl ester (CE), the principal neutral lipid class, contained a constant 85% of the total NL (range 83.95 to 86.55%). An additional 10% of the plasma NL was free cholesterol (C). Triglycerides (TO) comprised about 3% and FFA about 1.5% of total plasma NL. Comparison of mean values of plasma lipids (Table 1) with values in the literature (7, 11, 19, 20, 26) indicate a high degree of agreement both in absolute amounts and relative distribution by class. Plasma Neutral Lipid Rhythms
Plasma concentrations of C, CE, TO, and FFA in the six cows for the 2-d measurement period are shown in Figure 1. Triglyceride and FFA data for two cows were excluded from analysis because of excessively elevated concentrations at the stan-up of the experiment; Journal of Dairy Science Vol. 73. No.4. 1990
esters; and NL
1.1
3.2 1.6 10.4 84.9
= neutral
lipid.
these elevations were never seen during the second 24 h of sampling. It was thought that these values represented a response to unavoidable handling at the onset of each trial. In addition, large transients precluded the use of times series techniques for analyses. Visual analyses (23) indicated a time between peaks of approximately 2.5 h for all lipid classes (Table 2). Ampliwdes of peaks were: 20% of mean for C and CE and 60% of mean for FFA and TO. Because the cows were housed in an environmental chamber at constant temperature (20cC), changes in lipid concentrations with time were not due to fluctuations in ambient temperawre. There was no evidence of either a circadian peak or displacement of the pulsatile pattern by the light-dark cycle in any of the lipid classes (Figure 1). However, the presence of the ultradian peaks with 2.5-h periods was conserved when data were averaged across cows by time, indicating some form of entrainment. Recently, Blum et al. (3) measured a number of hormones and metabolites over a 24-h period in lactating cows fed high and low energy diets. They did not observe diurnal changes in blood TO, C, or phospholipids; these findings are in accord with our results. Pattern of Feed Consumption
The cows were fed once daily at 0900 h. In nonrwninants, the onset of feeding generally results in an increase in blood lipids. In rumi·
951
BLOOD LIPID RHYTHMS IN COWS TABLE 2. Pulse analysis of plasma neutral lipids by visual analysis (23). From visual analysis
Lipid class I
Average
CE C TO FFA
- - - - - - - (mgldI) - - - - - - 206.3 176.0 33.8 19.2 21.4 21.1 3.7 17.5 55.8 6.3 5.9 3.3 3.1 2.7 1.8 68.2
Baseline
Amplitude
% of Mean
Period (h)
2.6 2.6 2.7 2.7
ICE ::: Cholesteryl esters; C ::: cholesterol; TO ::: triglycerides.
nants. feeds have a relatively long residence time in the rumen, resulting in a buffering effect with no abrupt rises in plasma nutrients (5). In addition, although the total daily ration was provided to the animals at one time (0900 h), the cows self-regulated their intake in patterns probably closely related to ruminal fill (Table 3). During the 1st h after presentation of feed, the cows consumed 36% of total ration, whereas during the next 6 h, they consumed only about 20%. This was followed by a period of greater consumption: 35% was eaten during the next 6 h. Only 2.5% of the feed was eaten from midnight to 0900 h. This feeding pattern suggests that the initial high consumption of 36% of the ration fills the rumen, and, by a feedback mechanism, limits ingestion of additional feed during the next 6 h. Passage of this food from the rumen then permits ingestion of additional feed. The daily pattern of feed consumption, however, was not reflected in the plasma concentrations of the neutral lipids. Frequency Analysis
Spectral analysis of neutral lipid data identified no significant circadian rhythms. Representative periodograms for cow 91 for each lipid component are shown in Figure 2. The 2 to 3-h patterns identified by visual analysis were also indicated for C and CE in five of the six cows and for TG in three of four cows. The sensitivity and reliability of the spectral analyses were compromised by two factors. First, spectral analysis assumes a stationary time series, e.g., that the time between peaks is always identical. Second, samples must be taken at twice the highest frequency component of the time series and, because of noise, samples should be taken at three to five times this
frequency to prevent aliasing problems. Visual and autocorrelation analyses show that the time between peaks for NL varies. In addition, a 9Q-min pattern was demonstrated for most of the metabolic and hormonal variables measured (13). If this higher frequency rhythm exists for any of the NL, it could not be detected unless all samples, and not every fourth sample, were analyzed, and its existence would create an aliasing problem (4, 24) that could obscure true rhythms or create nonexisting rhythms. The inability to identify any statistically significant rhythms with periods greater than 200 min using power spectral analysis probably is due primarily to the nonstationary nature of the data. Still, power spectral analysis did indicate rhythms at the same frequencies identified by visual analysis, which adds credibility to these findings. Comparisons with Hormonal and Metabolic Variables
In contrast with our finding for NL, all other metabolic and hormonal variables except cortisol exhibited distinct circadian rhythms (13). In agreement with previous findings (1), body temperature exhibited a circadian rhythm as well as a higher frequency rhythm with a period of 90 min. In addition, spectral analysis of hormone and metabolite data showed a 90 to 120 min rhythm for most variables. This 90- to 120-min rhythm cannot be identified in the lipid data because of the longer effective sampling interval that resulted when only every fourth sample was analyzed (see preceding discussion of aliasing). For comparison purposes only and to indicate that rhythms can be observed in certain parameters, plasma urea nitrogen, triiodothyroJournal of Dairy Science Vol. 73, No.4, 1990
952
BITMAN ET AL,
CHOLESTEROL ESTER
TRIGLYCERIDE 15,
250,
12
200 "
,
.',..
9
. - ' . - .' \ ; "
6\
'.'
150 i
3
".-
250
.\
200
, ',,: I," .'
150
j1J
I
'/..... i .
j
1 .'
~
I
.
/'
",
,
i ~J \, '-.-.. '. ~"
~_-'j"
1\ '
J '. . '", ."" "
\/
.
..
t...-
\,,j..... \."')
" .\ . . . . .' ,...,.,~"", 'i~\1 ."'\ ,.. ',: -,..J, ~ V~'
.
~.
I(
~
.-
/
,tlo
\,.
250 .
-.
200'
~ ~ ~
200
j
150
1
~~
1
12
c5
.,
,
_ '!
, "" J ' - 1 " . ' , I...
~
"
'
~
.... \:.""
.......
" .........
f
9 t
~
6 ~" .. '" ,r""'- ...,..'
~
3'
\
, \""
15 :
../
12 200
9
i
r'.,
j
150
9
200
'\
.\'
",Vi
I2
18
24
06
12
18
24
t
1 !', T'--o. ~~ ".. . . ",,""'-,;"V'~r 3 ......../V'~ ''v"" 0----------------6
i L"V............l\.--""'\ \ '-"'\v""\./ \oJ V L
100
\r'
~j
100
150
","., /\\. Iir'j".' [ \ ~'i'
/ ..~.;'./.
-'
6 j
06
12
18
24
06
12
I8
24
06
W.AWL::::J
HOURS
HOURS
FREE FATTY FREE CHOLESTEROL 35 30 25
./
". , \ '
20
~~
20
I
"',..
; "--'/
'\
~\~'"
. '-~'\.'''''''-'''' ."",-
,\
-'''~ '\. .....
6 •
-
4
6 .
30
c5
I
2 ;.
4 t ...
25
--'
20
'-
~ ~~ I
i
0
2
'"
6 t
~
. 1
1
1
2 ~
15 \
,
ii, 1,1. ,,'". I 1\ j
i
4
~
O· 6
t
4 ,
-.
12
"l .. ,.; - - "....... ...... , .....
18
24
'"It'' 06
'12
2 18
24
06
HOURS
t
_'-" ..
"
.........""
. .
-
o ~---------------'2
18
24
06
12
18
24
06
HOURS
Figure 1. Temporal patterns of plasma concentrations in cholesteryl esters, free cholesterol, triglycerides and free fally acids, measured in six cows over 48 h. Plasma concentrations in milligrams per deciliter are plotted against time of day in hours. Horizontal black bars = lights off; white bars = lights on.
Journal of Dairy Science Vol. 73, No.4, 1990
953
BLOOD LIPID RHYTHMS IN COWS TABLE 3. Percentage feed COIISwnption by time interval for six lactating cows fed once daily. Percentage of total ration conswned at time interval. h
0-1
Item
Mean
35.7 \.9 18
SE n
2-3
4-5
8-9
10-11
12-13
5.2 1.1 18
10.0
6.4
10.3
12.4
12.6
4.9
13
9
15
1~
I~
I~
18
fr.7
18
18
nine (T3), and corusol for cow 91 are shown in Figure 3. Cow 91 was representative of the other five cows. and complete data for these measures will be reported at a later date. Circadian rhythms were clear and consistent for plasma urea nitrogen and T3. The plasma cortisol pattern was characterized by rapid episodic fluctuations with no evidence of a circadian pattern. Figure 4 compares the periodograms for urea nitrogen, T3, corusol, one lipid component, and cholesterol. The cholesterol periodogram, in contrast to those of urea nitrogen, T3. and corusol, gave little evidence of periodicity
17
14-15
17
16-24 2.5 .8 15
15
and the powers were very low, being only .1 to .3% of total power. Although the lipid data did not exhibit a circadian rhythm, pulsatile fluctuations were evident every 2 to 3 h. The repetitive nature of this pattern could possibly reflect a limit cycle control mechanism balancing rate of lipid entry and rate of removal in the circulatory system. This is possibly related to regulation of lipid supply (diet, production, synthesis) and utilization by peripheral tissues. The relationship of circadian metabolic and honnonal processes in the other variables studied to the observed 2-h higher frequency mythm in lipids cannot be assessed readily.
::E
E 12
9
cr
2,5
w
~
TRIIODOTHYRONINE
0
a.. 1.0 - - - - - - - - - - - - - - o
w N
:::J
E
2.0
'"c
I.S
"-
...:
:::li
cr
o
z
1.0
0 1.0 - - - - - - - - - - - - - - - FREE FATTY ACID
"lilllJlli:JJ ' 6
E "go
12
8 4
o 24.0
6.0 3.0 PERIOD (HOURS)
2.0
Figure 2. Periodograms of neutral lipids of cow 9 I.
a
12
18
24
06
12
18
24
06
HOURS
Figure 3. Temporal patterns of plasma concentrations of urea-nitrogen. triiodothyronine and cortisol of cow 9 I. Journal of Dairy Science Vol. 73, No.4, 1990
954
BITMAN ET AL.
f-
z
18·
w
u
'"D.-
15 .
W
'"w3= a
12
9
D.-
o
5
UREA NITROGEN
W
N
~
~
t
;t ':ODo;~"O'::E 3
te
on 24.0 5.0
•
3.0
-
s
CORTISOL
FREE CHOLESTEROL
2.0
1.5
22). The critical basic premise of metabolic profile or blood profile research was that abnormalities in blood chemistry would be indicators of metabolic disorders. In practice, however, blood metabolites are poor indicators of nutritional and physiologic status, since homeostasis is a basic biological control mechanism adapted by living organisms to prevent gross chemical imbalances (9). In addition, infrequent sampling, usually weekly or monthly in metabolic profile research (9, 16, 22), is not designed to readily reveal physiological variations. In contrast, intensive sampling, as demonstrated in our study of circulatory lipids, hormones, and metabolites, can reveal the presence of both ultradian and circadian variations.
1.0
PERIOD (HOURS)
Figw-e 4. Periodograms of urea-nilrogen. triiodothyronine, conisol and cholesterol. Power expressed as percentage (normalized for variance) is plotted against period in hours.
The lack of a circadian rhythm in the lipid data contrasts with the presence of definite circadian rhythms in a variety of other components in the same blood samples. The lack of a circadian rhythm in the cow has a parallel in findings in the rat for the lipid component, cholesterol (18). There was no apparent diurnal rhythm in cholesterol in either rat hepatic tissue or peripheral serum in spite of a marked circadian rhythm in cholesterol synthesis (17, 18). There is a marked circadian rhythm in the activity of hepatic hydroxymethylglutaryl-Coenzyme A reductase; the maximum synthetic rate occurred at 2400 h and was 3 to 10 times higher than the minimum at 1200 h (17, 18, 21). The lack of corresponding rhythmic changes in blood and liver C, the lipid product of the enzymatic synthesis, complicates interpretation of the physiological significance of the hepatic circadian rhythm in C biogenesis. Similarly, circadian rhythms of lipid synthesis (or utilization) may occur in the cow without circadian alterations in lipid components of the serum. The potential use of blood chemistry for diagnosis of clinical and nutritional problems was explored extensively in recent years (9, 16, Journal of Dairy Science Vol. 73. No.4. 1990
REFERENCES 1 BiU"nan, J., A. Lefcourt, D. L. Wood. and B. Stroud. 1984. Circadian and ultradian temperature rhythms of lactating dairy cows. J. Dairy Sci. 67:1014. 2 BiU"nan. J., D. L. Wood. M. Hamosh. P. Hamosh, Y.• and N. R. Mehta. 1983. Comparison of the lipid composition of breast milk from mothers of term and preterm infants. Am. J. Clin. Nutr. 38:300. 3 Blum, 1. W .• F. Jans, W. Moses. D. Frohli. M. Zemp, M. Wanner, 1. C. Hart, R. Thun, and U. Keller. 1985. Twenty-four-hour pattern of blood hormone and metabolite concenlrations in high-yielding dairy cows: effects of feeding low or high amounts of starch or crystalline fat. Zenlralbl. Veterinaermed. Reihe A. 32:401. 4 Chatfield. C. 1984. The analysis of time series: an introduction. 3rd ed. Olapman and Hall, New York. NY. 5 Christie. W. W. 1981. The composition. structure and function oflipids in the tissues of ruminant animals. Page 95 in Lipid metabolism in ruminant animals. W. W. Christie, ed. Pergamon Press. New York. NY. 6 Folch, J .• M. Lees, and G. H. Sloane-Stanley. 1957. A simple method for the isolation and purification of total lipids from animal tissues. 1. BioI. Chern. 226:497. 7 Hartmann. P. E., and A. K. Lascelles. 1965. Variation in the concentration of lipids and some other constituents in the blood plasma of cows at various stages of lactation. Aust. J. BioI. Sci. 18:114. 8 Huntington, G. B. 1984. Net absorption of glucose and nitrogenous compounds by lactating Holstein cows. 1. Dairy Sci. 67:1919. 9 Jones. G. M., E. E. Wildman, H. F. Troutt, Jr., T. N. Lesch. P. E. Wagner. R. L. Boman. and N. M. Lanning. 1982. Metabolic profiles in Virginia dairy herds of different milk yields. 1. Dairy Sci. 65:683. 10 Kahl. S.• A. V. Capuco. and J. BiU"nan. 1987. Serum concentrations of thyroid hormones and extrathyroidal thyroxine-5' -monodeiodinase activity during lactation in the rat. Proc. Soc. Exp. BioI. Med. 184:144. 11 Leal. W.M.F.• andJ. G. Hall. 1968. Lipid composition of lymph and blood plasma of the cow. 1. Agric. Sci. Camb. 71:189. 12 Lefcourt, A. M.. and J. BiU"nan. 1985. Method for automatic, continuous blood sampling with remote alarm
BLOOD LIPID RHYTHMS IN COWS for clogged catheters. J. Dairy Sci. 68:2108. 13 Lefcoun, A. M.. J. Biunan, S. Kahl. R. M. Akers. G. B. Huntington, and D. L. Wood. 1985. Continuous endocrine. metabolic and deep-body temperature profiles for six cows housed in an environmental chamber. Fed. Proc. 44:1362. 14 Lefcoun, A. M.• J. Biunan. D. L. Wood, and B. Stroud. 1986. Radiotelemetry system for continuously monitoring temperature in cows. 1. Dairy Sci. 69:237. 15 Merriam.G. R.. and K. W. Wachter. 1982. Measurement and analysis of episodic hormone secrelion. Pages 325-346 in Compulers in endocrinology. D. Rodbard and G. Forti, ed. Raven Press, New York. NY. 16 Payne. J. M.• S. M. Dew. R. Manston. and M. Faulks. 1970. The use of a metabolic profile lest in dairy herds. Vet. Rec. 87:150. 17 Ramasarma. T. 1976. Biogenetic interrelationship of ubiquinone and cholesterol. Biochem. Soc. Symp. 35: 245. 18 Rao, G. S" and T. Ramasarma. 1971. Rhytlunic activity of biogenesis of cholesterol. Environ. Physiol. 1: 188. 19 Raphael, B. C.• P. S. Dimick. and D. L. Puppiooe. 1973. Lipid characterization of bovine serum lipoproteins throughout gestation and lactation. J. Dairy Sci. 56:1025.
955
20 Remond. B.. R. Toullec. and M. Joumet. 1973. Changes in dairy cows in content ofdifferent blood constituents at the end of gestation and the beginning of lactation: relationships with secretion ofmilk fat. Ann. BioI. Anim. Biochem. Biophys. 13:363. 21 Rodwell. V. W.• D. J. McNamara. and D. J. Shapiro. 1973. Regulation of hepatic 3-hydroxy-3-methylglutaryl-Coenzyme A reductase. Adv. Enzymol. 38:373. 22 Rowlands. G. J. 1980. A review of variations in the concentrations of metabolites in the blood of beef and dairy cattle associated with physiology, nutrition and disease. with particular reference 10 the interpretation of metabolic profiles. World Rev. Nutr. Diet. 35:172. 23 Santen. R. J.• and C. W. Bardin. 1973. Episodic, luteinizing hormone secretion in man. Pulse analysis. clinical interpretation. physiologic mechanisms. J. Clin. Invest. 52:2617. 24 Shumway. R. H. 1988. Applied statistical time series analysis. Prentice Hall. Englewood Cliffs, NJ. 25 Statgraphics. 1987. Statistical graphic system. STSC, Inc.• Rockville. MD. 26 Starry, J. E. and J.A.F. Rook. 1964. Lipids in the blood plasma of cows of the Friesian and Channel Island breeds. Nature 201 :926.
Journal of Dairy Science Vol. 73, No.4, 1990