76
Circadian Variation in Platelet Imipramine Binding During the Day in Healthy Subjects P. M. Ellis 1, Ruth Beeston2, R. R. Cooke 2, G. Mellsopl Department of Psycho1ogical Medicine l • WeUington School ofMedicine. University ofOtago and Division ofLaboratory Services 2, Wellington Hospital, Wellington, New Zealand
Platelet tritiated imipramine binding values in healthy controls vary considerably from study to study. A possible contributor to such variation might be a circadian rhythm afTecting binding, although previous studies of this have been contradictory. Platelet imipramine binding was examined in 12 healthy, medication-free subjects studied at 8 a. m., 11 a. m., 4 p. m., and 10 p. m. during one day. lmipramine binding was determined on platelet membranes, using 0.8 - 8 nM 3H-imipramine, and nonspecific binding was defined by 50 f.LM desipramine. All sampIes from a given individual were assayed simultaneously. Tbe intra-assay coefficient of variation was 6.3 percent. Tbere was no evidence of significant differences in binding capacity or affinity (Bmax or Kd) at different times of day. Circadian variation was explored using COSINOR analysis (DeMet et al., 1989). Tbere was no evidence of circadian variation in binding using this model, even when only the variable portion of binding was considered for each individual. Intraindividual variation in binding was substantiaI, with a mean coefficient ofvariation of29 percent for Bmax and 38 percent for~. Tbe possible basis of this variation is unclear, but may retlect the presence of "occult" binding sites in the membrane, or the effect of endogenous modulators ofbinding. Tbe interrelationship of Bmax and Kd may also be a factor. It was considered that low-affinity binding did not account for a significant part of the variation in ~ in this assay. Tbe utility of imipramine binding as a biological marker of depression may be limited by such levels of intraindividual variation in binding parameters.
Introduction A decrease in the level of binding of tritiated imipramine to platelets has been repeatedly found in groups of depressed patients, and it has been suggested that this may prove to be a useful biological marker of depression (Langer and Schoemaker, 1988). However, despite the achievement of Pharmacopsychiat. 24 (1991) 76 - 80 C Georg Thieme Verlag Stuttgart· New York
Samenvatting Oe bindingswaarden van 3H-imipramine aan bloedplaatjes in gezonde kontrolle groepen varieren aanzienlijk in verschilIende onderzoeken. Een mogelijke bijdrage tot zo'n variatie kan een circadisch ritme zijn dat de binding beinvloedt. Vorige studies echter zijn hierover niet eensluidend. Oe bindings-kapaciteit van imipramine aan plaatjes werd onderzocht in 12 gezonde proefpersonen, die geen geneesmiddelen innamen, op de volgende tijden op een dag om: 0800, 1100, 1600 en 2200 uur. Oe bindings-kapaciteit van imipramine aan plaatjesmembranen werd gemeten met behulpvanO.8 - 8 nM 3H-imipramine en non-specifieke binding werd bepaald door 50 f.LM desipramine. Alle bloedprodukten van ieder proefpersoon werden tegelijkertijd bepaald. Oe intraassay variatie co-efficient was 6.3 per cent. Geen belangrijke dag-verschillen in bindings-kapaciteit of affiniteit (B max of~) werden gekonstateerd. Oe circadische variatie werd onderzocht met de toepassing van COSINOR analyse (De Met et al., 1989). Oe aanwezigheid van een circadisch ritme werd hiermee niet aangetoond, zelfs niet wanneer per individu alleen het variabele deel van de binding werd onderzocht. Oe intra-individuele bindingsvariatie was aanzienlijk met een gemiddelde variatie van 29 per cent voor Bmax en 38 per cent voor~. Oe reden van deze variatie in onduidelijk maar zou verklaard kunnen worden door de aanwezigheid van "occulte" binding plaatsen aan de membraan, of door het effect van een endogene substantie die de binding verandert. Oe onderlinge relatie tussen Bmax en ~ kan ook een faktor zijn. Geopperd werd dat de lage bindingsaffiniteit niet de reden kan zijn van de aanzienlijke variatie in Kd. Het nut van imipramine bindings-kapaciteit als biologische marker van depressie kan beperkt zijn bij een dergelijke variatie van bindingsparameters in een individu.
respectable levels of interassay reliability in a given center, there are marked variations in the binding levels reported for control subjects by different groups. For example, Schneider et al., (1985) found Bmax values of 2656 fmol/mg Pr. in their control subjects, while healthy controls examined by Rehavi et al., (1984) had Bmax values of 423 fmol/mg/Pr. Such differences have been ascribed to a wide variety of technical factors, but Received: 14. 8. 1990 Revised version: 27.12. 1990 Accepted: 14. 1. 1991
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
Summary
Circadian Variation in Platelet /mipramine Binding During the Day in Healthy Subjects
3H-imipramine binding was measured at 4°C using eight concentrations (0.8 - 8.0 nM) of the radioactive ligand (specific activity 20 Ci/mM, Amersham, UK). The incubation mixture contained a known amount of protein, elose to 200 !J.g, in a total volume of 500 !J.. Specific binding was defined as the difference between total binding and that which remained in the presence of 50!J.M desipramine. FoUowing a one-hour incubation, platelet membranes were washed rapidly with 16 ml cold assay buffer and filtered through Whatman GF/F g1ass fiber filters under vacuum. Filters were transferred to scintillation vials, air dried overnight, and extracted in a liquid scintillation cocktail before measurement for radioactivity. Estimates of maximum receptor binding (B max ) and the dissociation constant (.K.J) were obtained using an iterative, nonlinear curve fitting technique (Mclntosh and Mclntosh, 1980), which has been shown to be superior to Scatchard plots (Scatchard. 1949; Cooke and Mclntosh. 1986). The intra-assay coefficient of variation in binding was 6.3 percent, which was similar to that reported by others (Arora et a1., 1986; Paul et a1., 1980). Nonspecific binding represented 30 percent oftotal binding. Circadian variation in the .K.J and Bmax was evaluated by COSINOR analysis, as described by DeMet et a1. (1989), using multiple linear regression, with XI = cos(21t x hr/24) and X2 - sin(21t x hr/24) as the independent variables. The extent of intraindividual variation in binding was measured by calculation of the coefficient of variation in binding for each subject. Statistical analysis was performed on a Macintosh SE computer using the Statview SE+graphics program (Feldman et a1., 1988).
Results Method
Subjects and procedure Twelve healthy subjects free of current psychiatric or medical disorder and not taking any medication were studied during an eight-week period from late May 1988 (i. e. during the southern hemisphere winter). The subjects consisted of six males and six females, and had a mean age of 32.8 years (range 22.4 - 58.5 years, sd 10.7 years). All were day workers and were studied on a weekday, during their usual activities. All gave informed consent. Blood sampies were drawn into EDTA containing vacutainers at 8 a. m., II a. m., 4 p. m., and 10 p. m. on the same day from any one subject. All sampies from a given subject were assayed at the same time to eliminate the possible effect of interassay variation on the binding values for each subject.
Imipramine binding assay Platelet-rich plasma (PRP) was prepared by centrifugation at 100 g max for 10 mins at room temperature. The PRP was then stored at - 20°C for up to four weeks. Platelet membranes were prepared by an adaptation of the technique of Barber et a1. (1971), and the binding assay followed the method of Briley (1979). Whole platelets were retrieved from PRP by centrifugation at 16500 g max for 10 minutes at room temperature and washed in buffer I (5 mM Tris, 20 mM EDTA, 150 mM NaCI, pH 7.5). The platelets were then preloaded with g1ycerol by spinning the cells through an isotonic 0 - 40% g1ycerol gradient at 1465 g max for 30 minutes at room temperature and then pelleting them at 5860 g max for 10 minutes at room temperature. The pelleted cells were hypo-osmotically Iysed by vortexing in buffer 11 (5 mM Tris, 5 mM EDTA, pH 7.5). Platelet membranes were then centrifuged at 39000 g max for 10 minutes at 4°C and washed in buffer III (70 mM Tris, pH 7.5). The membranes were stored in assay buffer(50 mM Tris, 120 mM NaCI, 5 mM KCI, pH 7.4) at aconcentration ofO.5 - 1.0 mg/mI and were kept at - 20°C for up to one week. Protein was assayed according to the method of Lowry et a1. (1951), as modified by MarkweIl et a1. (1978), using bovine serum albumin as the standard protein.
77
The binding parameters at the different sampling times are detailed in Table 1. There was no c1ear difference in the number ofbinding sites (B max values) or the binding dissociation constant (~) between the different sampling times (single-factor ANOVA, repeated measures, p > 0.05). No significant circadian variation in binding consistent with the COSINOR model was found in either Kd ofB max . However, as the baseline binding values and amplitude ofvariation varied considerably between subjects, it appeared appropriate to explore the effects of correction for these factors. As a first step, to compensate for differences in baseline binding values, the mean binding over the day was calculated for each individual and the individual's binding values at the four sampling times during the day were expressed as apercentage of this mean value. Repeating the regression analysis at this stage still provided no evidence of significant circadian variation. Finally, in order to maximize the variable portion ofthe binding and compensate for differences in baselines, the individuals' mean binding values over the day (as described above) were subtracted from their binding values at each sampling time. Again, no significant evidence of circadian variation emerged from this process. The variation in Bmax in one ofthe subjects over a shorter time period on a different occasion is shown in Fig. 1, and was also not suggestive of a circadian rhythm. While Table 1 indicates that mean binding values are relatively stable, at least between 8 a. m. and 11 a. m. (the period during which most studies of imipramine binding in humans have been performed), it does not provide a satisfactory demonstration of intraindividual variation in binding. This is c1early of importance in assessing the significance of imipramine binding studies. To quantify this variabil-
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
even when an identical assay was used in different centres, as in the WHO study (Bech et al., 1988), there were significant differences in binding levels between different groups of control subjects. This suggests the importance of certain subject variables. A number of these have been examined previously, but there have been few studies examining circadian variation in binding. Wirz-Justice et al. (1983) found that Bmax was highest at the end of the dark period in a number of rat brain nuclei, but although similar observations have been described by another member of this group (Kraeuchi et al., 1986), these findings were not confirmed by Di Lauro et al. (1986). Galzin et al. (1987) found Bmax was higher in the dark phase in rabbit platelets studied at different times of day over one month. Baron et al. (1988), in a re-examination of previously reported data, found no consistent pattern of circadian variation in a substantial number of patients studied on a single occasion, and Gentsch et al. (1989) found no difference between moming and aftemoon sampIes taken from both depressed and healthy patients on the same day. In contrast to the animal studies, Nankai et al. (1986) found a significant circadian variation in binding in six subjects studied over one day, with higher binding levels in the light phase. Montero et al. (1989), however, while reporting a significant decrease in Bmax during the night, were not able to establish a c1ear common circadian rhythm for alI subjects. Such conflicting results require c1arification in normal subjects, with an assay of established reliability. The present study examined the variation in binding during one day in 12 healthy subjects.
Pharmacopsychiat. 24 (/99/)
P. M. EI/is, Ruth Beeston, R. R. Cooke, G. Mel/sop
Pharmacopsychiat. 24 (1991)
Bmax in the dark phase, included a sampie at 3 a. m., but found that the maximum decrease in Bmax occurred at 9 p. m., which is within the period investigated in the present study. The conditions under which their subjects were studied were not stated.
1300
tca
1200
I
1000
g
1100
t:.
J···················l
900
111
E
ID
'{ ..
800 700
600
500 +--~-r--~-"""'T"---r--""T"""--r----, 1700 1800 1900 2000 2100 2200 2300 2400
Time Fig. 1 B max vatues (tmol/mg Pr) in one subject during the evening. Error bars show standard deviation.
ity. the coefficient of variation of the binding values for each subject was calculated. The mean of these coefficients of variation for Bmax was 29 percent (sd 10 percent; range 16 - 52 percent) and for K
Circadian Variation in Platelet Imipramine Binding During the Day in Healthy Subjects P. M. Ellis 1, Ruth Beeston2, R. R. Cooke 2, G. Mellsopl Department of Psycho1ogical Medicine l • WeUington School ofMedicine. University ofOtago and Division ofLaboratory Services 2, Wellington Hospital, Wellington, New Zealand
Platelet tritiated imipramine binding values in healthy controls vary considerably from study to study. A possible contributor to such variation might be a circadian rhythm afTecting binding, although previous studies of this have been contradictory. Platelet imipramine binding was examined in 12 healthy, medication-free subjects studied at 8 a. m., 11 a. m., 4 p. m., and 10 p. m. during one day. lmipramine binding was determined on platelet membranes, using 0.8 - 8 nM 3H-imipramine, and nonspecific binding was defined by 50 f.LM desipramine. All sampIes from a given individual were assayed simultaneously. Tbe intra-assay coefficient of variation was 6.3 percent. Tbere was no evidence of significant differences in binding capacity or affinity (Bmax or Kd) at different times of day. Circadian variation was explored using COSINOR analysis (DeMet et al., 1989). Tbere was no evidence of circadian variation in binding using this model, even when only the variable portion of binding was considered for each individual. Intraindividual variation in binding was substantiaI, with a mean coefficient ofvariation of29 percent for Bmax and 38 percent for~. Tbe possible basis of this variation is unclear, but may retlect the presence of "occult" binding sites in the membrane, or the effect of endogenous modulators ofbinding. Tbe interrelationship of Bmax and Kd may also be a factor. It was considered that low-affinity binding did not account for a significant part of the variation in ~ in this assay. Tbe utility of imipramine binding as a biological marker of depression may be limited by such levels of intraindividual variation in binding parameters.
Introduction A decrease in the level of binding of tritiated imipramine to platelets has been repeatedly found in groups of depressed patients, and it has been suggested that this may prove to be a useful biological marker of depression (Langer and Schoemaker, 1988). However, despite the achievement of Pharmacopsychiat. 24 (1991) 76 - 80 C Georg Thieme Verlag Stuttgart· New York
Samenvatting Oe bindingswaarden van 3H-imipramine aan bloedplaatjes in gezonde kontrolle groepen varieren aanzienlijk in verschilIende onderzoeken. Een mogelijke bijdrage tot zo'n variatie kan een circadisch ritme zijn dat de binding beinvloedt. Vorige studies echter zijn hierover niet eensluidend. Oe bindings-kapaciteit van imipramine aan plaatjes werd onderzocht in 12 gezonde proefpersonen, die geen geneesmiddelen innamen, op de volgende tijden op een dag om: 0800, 1100, 1600 en 2200 uur. Oe bindings-kapaciteit van imipramine aan plaatjesmembranen werd gemeten met behulpvanO.8 - 8 nM 3H-imipramine en non-specifieke binding werd bepaald door 50 f.LM desipramine. Alle bloedprodukten van ieder proefpersoon werden tegelijkertijd bepaald. Oe intraassay variatie co-efficient was 6.3 per cent. Geen belangrijke dag-verschillen in bindings-kapaciteit of affiniteit (B max of~) werden gekonstateerd. Oe circadische variatie werd onderzocht met de toepassing van COSINOR analyse (De Met et al., 1989). Oe aanwezigheid van een circadisch ritme werd hiermee niet aangetoond, zelfs niet wanneer per individu alleen het variabele deel van de binding werd onderzocht. Oe intra-individuele bindingsvariatie was aanzienlijk met een gemiddelde variatie van 29 per cent voor Bmax en 38 per cent voor~. Oe reden van deze variatie in onduidelijk maar zou verklaard kunnen worden door de aanwezigheid van "occulte" binding plaatsen aan de membraan, of door het effect van een endogene substantie die de binding verandert. Oe onderlinge relatie tussen Bmax en ~ kan ook een faktor zijn. Geopperd werd dat de lage bindingsaffiniteit niet de reden kan zijn van de aanzienlijke variatie in Kd. Het nut van imipramine bindings-kapaciteit als biologische marker van depressie kan beperkt zijn bij een dergelijke variatie van bindingsparameters in een individu.
respectable levels of interassay reliability in a given center, there are marked variations in the binding levels reported for control subjects by different groups. For example, Schneider et al., (1985) found Bmax values of 2656 fmol/mg Pr. in their control subjects, while healthy controls examined by Rehavi et al., (1984) had Bmax values of 423 fmol/mg/Pr. Such differences have been ascribed to a wide variety of technical factors, but Received: 14. 8. 1990 Revised version: 27.12. 1990 Accepted: 14. 1. 1991
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
Summary
Circadian Variation in Platelet /mipramine Binding During the Day in Healthy Subjects
3H-imipramine binding was measured at 4°C using eight concentrations (0.8 - 8.0 nM) of the radioactive ligand (specific activity 20 Ci/mM, Amersham, UK). The incubation mixture contained a known amount of protein, elose to 200 !J.g, in a total volume of 500 !J.. Specific binding was defined as the difference between total binding and that which remained in the presence of 50!J.M desipramine. FoUowing a one-hour incubation, platelet membranes were washed rapidly with 16 ml cold assay buffer and filtered through Whatman GF/F g1ass fiber filters under vacuum. Filters were transferred to scintillation vials, air dried overnight, and extracted in a liquid scintillation cocktail before measurement for radioactivity. Estimates of maximum receptor binding (B max ) and the dissociation constant (.K.J) were obtained using an iterative, nonlinear curve fitting technique (Mclntosh and Mclntosh, 1980), which has been shown to be superior to Scatchard plots (Scatchard. 1949; Cooke and Mclntosh. 1986). The intra-assay coefficient of variation in binding was 6.3 percent, which was similar to that reported by others (Arora et a1., 1986; Paul et a1., 1980). Nonspecific binding represented 30 percent oftotal binding. Circadian variation in the .K.J and Bmax was evaluated by COSINOR analysis, as described by DeMet et a1. (1989), using multiple linear regression, with XI = cos(21t x hr/24) and X2 - sin(21t x hr/24) as the independent variables. The extent of intraindividual variation in binding was measured by calculation of the coefficient of variation in binding for each subject. Statistical analysis was performed on a Macintosh SE computer using the Statview SE+graphics program (Feldman et a1., 1988).
Results Method
Subjects and procedure Twelve healthy subjects free of current psychiatric or medical disorder and not taking any medication were studied during an eight-week period from late May 1988 (i. e. during the southern hemisphere winter). The subjects consisted of six males and six females, and had a mean age of 32.8 years (range 22.4 - 58.5 years, sd 10.7 years). All were day workers and were studied on a weekday, during their usual activities. All gave informed consent. Blood sampies were drawn into EDTA containing vacutainers at 8 a. m., II a. m., 4 p. m., and 10 p. m. on the same day from any one subject. All sampies from a given subject were assayed at the same time to eliminate the possible effect of interassay variation on the binding values for each subject.
Imipramine binding assay Platelet-rich plasma (PRP) was prepared by centrifugation at 100 g max for 10 mins at room temperature. The PRP was then stored at - 20°C for up to four weeks. Platelet membranes were prepared by an adaptation of the technique of Barber et a1. (1971), and the binding assay followed the method of Briley (1979). Whole platelets were retrieved from PRP by centrifugation at 16500 g max for 10 minutes at room temperature and washed in buffer I (5 mM Tris, 20 mM EDTA, 150 mM NaCI, pH 7.5). The platelets were then preloaded with g1ycerol by spinning the cells through an isotonic 0 - 40% g1ycerol gradient at 1465 g max for 30 minutes at room temperature and then pelleting them at 5860 g max for 10 minutes at room temperature. The pelleted cells were hypo-osmotically Iysed by vortexing in buffer 11 (5 mM Tris, 5 mM EDTA, pH 7.5). Platelet membranes were then centrifuged at 39000 g max for 10 minutes at 4°C and washed in buffer III (70 mM Tris, pH 7.5). The membranes were stored in assay buffer(50 mM Tris, 120 mM NaCI, 5 mM KCI, pH 7.4) at aconcentration ofO.5 - 1.0 mg/mI and were kept at - 20°C for up to one week. Protein was assayed according to the method of Lowry et a1. (1951), as modified by MarkweIl et a1. (1978), using bovine serum albumin as the standard protein.
77
The binding parameters at the different sampling times are detailed in Table 1. There was no c1ear difference in the number ofbinding sites (B max values) or the binding dissociation constant (~) between the different sampling times (single-factor ANOVA, repeated measures, p > 0.05). No significant circadian variation in binding consistent with the COSINOR model was found in either Kd ofB max . However, as the baseline binding values and amplitude ofvariation varied considerably between subjects, it appeared appropriate to explore the effects of correction for these factors. As a first step, to compensate for differences in baseline binding values, the mean binding over the day was calculated for each individual and the individual's binding values at the four sampling times during the day were expressed as apercentage of this mean value. Repeating the regression analysis at this stage still provided no evidence of significant circadian variation. Finally, in order to maximize the variable portion ofthe binding and compensate for differences in baselines, the individuals' mean binding values over the day (as described above) were subtracted from their binding values at each sampling time. Again, no significant evidence of circadian variation emerged from this process. The variation in Bmax in one ofthe subjects over a shorter time period on a different occasion is shown in Fig. 1, and was also not suggestive of a circadian rhythm. While Table 1 indicates that mean binding values are relatively stable, at least between 8 a. m. and 11 a. m. (the period during which most studies of imipramine binding in humans have been performed), it does not provide a satisfactory demonstration of intraindividual variation in binding. This is c1early of importance in assessing the significance of imipramine binding studies. To quantify this variabil-
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
even when an identical assay was used in different centres, as in the WHO study (Bech et al., 1988), there were significant differences in binding levels between different groups of control subjects. This suggests the importance of certain subject variables. A number of these have been examined previously, but there have been few studies examining circadian variation in binding. Wirz-Justice et al. (1983) found that Bmax was highest at the end of the dark period in a number of rat brain nuclei, but although similar observations have been described by another member of this group (Kraeuchi et al., 1986), these findings were not confirmed by Di Lauro et al. (1986). Galzin et al. (1987) found Bmax was higher in the dark phase in rabbit platelets studied at different times of day over one month. Baron et al. (1988), in a re-examination of previously reported data, found no consistent pattern of circadian variation in a substantial number of patients studied on a single occasion, and Gentsch et al. (1989) found no difference between moming and aftemoon sampIes taken from both depressed and healthy patients on the same day. In contrast to the animal studies, Nankai et al. (1986) found a significant circadian variation in binding in six subjects studied over one day, with higher binding levels in the light phase. Montero et al. (1989), however, while reporting a significant decrease in Bmax during the night, were not able to establish a c1ear common circadian rhythm for alI subjects. Such conflicting results require c1arification in normal subjects, with an assay of established reliability. The present study examined the variation in binding during one day in 12 healthy subjects.
Pharmacopsychiat. 24 (/99/)
P. M. EI/is, Ruth Beeston, R. R. Cooke, G. Mel/sop
Pharmacopsychiat. 24 (1991)
Bmax in the dark phase, included a sampie at 3 a. m., but found that the maximum decrease in Bmax occurred at 9 p. m., which is within the period investigated in the present study. The conditions under which their subjects were studied were not stated.
1300
tca
1200
I
1000
g
1100
t:.
J···················l
900
111
E
ID
'{ ..
800 700
600
500 +--~-r--~-"""'T"---r--""T"""--r----, 1700 1800 1900 2000 2100 2200 2300 2400
Time Fig. 1 B max vatues (tmol/mg Pr) in one subject during the evening. Error bars show standard deviation.
ity. the coefficient of variation of the binding values for each subject was calculated. The mean of these coefficients of variation for Bmax was 29 percent (sd 10 percent; range 16 - 52 percent) and for K
