Inverted-U-shaped correlation between dopamine receptor availability in striatum and sensation seeking

Abstract

Sensation seeking is a core personality trait that declines with age in both men and women, as do also both density and availability of the dopamine D_2/3 receptors in striatum and cortical regions. In contrast, novelty seeking at a given age relates inversely to dopamine receptor availability. The simplest explanation of these findings is an inverted-U-shaped correlation between ratings of sensation seeking on the Zuckerman scale and dopamine D_2/3 receptor availability. To test the claim of an inverted-U-shaped relation between ratings of the sensation-seeking personality and measures of dopamine receptor availability, we used PET to record [¹¹C]raclopride binding in striatum of 18 healthy men. Here we report that an inverted-U shape significantly matched the receptor availability as a function of the Zuckerman score, with maximum binding potentials observed in the midrange of the scale. The inverted-U shape is consistent with a negative correlation between sensation seeking and the reactivity (“gain”) of dopaminergic neurotransmission to dopamine. The correlation reflects Zuckerman scores that are linearly linked to dopamine receptor densities in the striatum but nonlinearly linked to dopamine concentrations. Higher dopamine occupancy and dopamine concentrations explain the motivation that drives afflicted individuals to seek sensations, in agreement with reduced protection against addictive behavior that is characteristic of individuals with low binding potentials.

The Zuckerman scale rates the personality trait of sensation seeking. The neurobiological theory of personality holds that particular traits are determined by intrinsic properties of monoaminergic neurotransmitter systems in the brain (1). Thus, linkage studies reveal associations between sensation-seeking personality and polymorphisms of dopamine-related genes (2), and receptor images show an association between aspects of personality and the availability of binding sites for dopamine D_2/3 receptor ligands in human striatum (3, 4).

Studies of addictive behavior suggest that low availability of dopamine D_2/3 receptors in striatum disposes toward addiction, whereas high receptor availability protects against addiction (5). Other studies show that high-sensation or novelty seekers are at particular risk of addiction (6 –8). These observations suggest that low dopamine receptor availability is a biological marker, both of a specific personality and of an increased risk of addiction.

For individuals of similar age, Suhara et al. (9) reported a negative correlation of novelty seeking with dopamine receptor availability, but the study included no subjects at the lowest end of the novelty-seeking spectrum. In contrast, the closely related trait of sensation seeking appears to peak after puberty and then to decline as a function of age (10). Measures of both density and availability of the dopamine D_2/3 receptors decline with age above 18 (3, 11, 12). Jointly, these disparate observations make sense if dopamine receptor availability rises in concert with sensation seeking at lower Zuckerman scores but falls in opposition to sensation seeking at higher Zuckerman scores. The findings predict that highly sensation-seeking individuals have low receptor availability that would not protect against addiction. Together, the measures predict an inverted-U-shaped correlation between receptor availability and Zuckerman scores in which the midrange of the Zuckerman scale yields the highest receptor availability.

Here we test the prediction of an inverted-U-shaped correlation between dopamine receptor availability and ratings on the Zuckerman scale by determining the receptor availability in men with a range of sensation-seeking scores. We used positron emission tomography (PET) of [¹¹C]raclopride to determine the availability of dopamine D_2/3 receptors in brain of 18 healthy men who completed the Zuckerman sensation-seeking questionnaire.

Theory and Methods

The concept of gain (13 –15) is central to the understanding of receptor reactivity. As generally understood by Lodge and Grace (14) in terms of dopaminergic neurotransmission, gain is the increase in the number of neurons, and hence the number of active terminals that release dopamine in response to a stimulus. As a result, more terminals interact with more sites in response to a phasic event. The present treatment of neurotransmitter concentrations is consistent with the concept of gain as the relative increase in receptor occupancy given a constant initiating event. The keys to the quantitative analysis of gain are the ligand concentration and receptor density as related in the binding potential of the ligand. The binding potential of a radioligand is the ratio between bound and unbound radioligand molecules that reflects the number of receptor sites available for additional binding. A family of formulations equally applies to all ligands of the receptor (16), including endogenous competitors,

where by convention is the fractional binding potential of a ligand (17), relative to the number of unbound (nondisplaceable; ND) ligand molecules in the tissue. The ligand’s half-saturation concentration, , is measured relative to the ligand’s aqueous volume of distribution in a solvent, (water, plasma, or tissue), the reciprocal of which is the “free” fraction of the nondisplaceable ligand in the solvent, . is the maximum number of binding sites that can be occupied, B is the total number of binding sites actually occupied by ligands, is the number of unoccupied binding sites, BP ₀ is the theoretically highest achievable binding potential in the absence of occupying ligands, henceforth termed the binding capacity, and is the IC₅₀ of the ligand in the presence of competing ligands.

When the control site is a receptor, the differential , or “reactivity,” equals the slope of the Michaelis-Menten curve at the concentration C, where the receptor availability is B/C. The ratio of the two measures defines the logarithmic gain or control coefficient,

where γ is the gain or control coefficient, and σ is the occupancy of ligands. As a variable that ranges between 0 and 1, the logarithmic gain relates directly to the occupancy of the transmitter (σ) that increases when the gain declines. This result presents the binding potential as a product of a trait, that is, the theoretically maximum binding potential in the total absence of bound ligand , and a state, namely the logarithmic gain or control coefficient (γ), where the maximum binding potential reflects the total number of receptors, and the gain reflects the ligand concentration and hence occupancy at the receptors. The binding potential incorporates a unique value of the gain at a given binding capacity, , where Z is the state index. The equation predicts that binding potentials reflect two processes, one that sets a binding capacity and another that adjusts the gain. There the gain is the coefficient that translates neurotransmitter concentration to occupancy, , where χ is the concentration relative to the half-saturation constant ( or ).

We scored sensation-seeking men in the entire range of a previously studied normally distributed sample of 243 Danish individuals (mean 19.7, SD 5.2) on the 40-point Zuckerman sensation-seeking scale (18). This ranking is based on the response to questions pertaining to individual proclivity to engage in novel or risky activities. We recruited participants by local advertisement. At first contact, we accepted only right-handed candidates who took no medication for central nervous disorders and had no metallic foreign bodies. We screened the candidates for signs of psychological or psychiatric symptoms by means of a formal Structured Clinical Interview for the DSM-IV (SCID). The first 18 men who passed the first contact were aged 30.1 years (SD = 7.1, range 21–42) and none was excluded by subsequent screening. The men gave written informed consent to the study as approved by the official Central Danish Regional Science Ethics Committee. The men reclined on the bed of an ECAT HR47 tomograph, with the head comfortably immobilized in midfield of view with a custom-made head holder. A brief attenuation scan used radiation from a rotating rod source. We gave [¹¹C]raclopride i.v. as a single bolus at high specific activity (mean 87 GBq/μmol) at an average dose of 4.2 MBq (67 pmol) raclopride per kg body weight and then completed 60 min of dynamic emission recording. The dose of raclopride represented less than 1% of the Michaelis constant of raclopride at dopamine D_2/3 receptors in the brain and hence occupied fewer than 1% of the D_2/3 receptors.

The 22 frames of image records increased in duration from 30 s to 10 min. After attenuation and decay correction, sinograms were reconstructed by filtered backprojection with an 8-mm Hanning filter. The dynamic sequences were realigned and corrected framewise for head motion, and then summed and registered to the MNI stereotaxic brain (MNIB) using the AIR (19) and mutual information maximization (20) algorithms. Frame-averaged images were transformed to the MNIB, normalized to the average activity of each subject’s cerebellum volume of interest (VOI), and averaged across the 18 subjects, as presented as Fig. S1, to confirm adequate head motion correction. Time-radioactivity curves were then extracted by masking VOIs for left, right, and bilateral putamen (5.0, 4.7, and 9.7 cm³), left, right, and bilateral caudate nucleus (4.2, 4.3, and 8.5 cm³), bilateral ventral striatum (2.3 cm³), and cerebellum (48.3 cm³), based on the MNIB as above.

We determined the binding potentials of the radioligand in selected VOIs by means of linear graphical analysis (21) of the tracer accumulation in the interval 20–60 min after the i.v. injection, with the cerebellum as a nonbinding reference region.

Results

Binding potentials and Zuckerman scores are shown in Fig. 1. Linear regression revealed no correlation between the binding potential and the Zuckerman scores. Linear regression of binding potentials restricted to Zuckerman scores greater than 15 revealed significant negative correlations in ventral striatum and left putamen (slopes −0.026 and −0.027; P < 0.03 and P < 0.02). The quadratic expression of an inverted-U shape significantly fitted binding potentials in the entire range of sensation seeking. We tested the significance of the quadratic regression by means of the F test, assuming that observed values of binding potentials are independent and normally distributed with the same variance and mean . The F test tests the null hypothesis that . The quadratic regression was significant in ventral striatum and putamen of both hemispheres, separately as well as in combination, as listed in Table 1.

Fig. 1.

Average Zuckerman scores and binding potentials of the dopamine D_2/3 receptor ligand [¹¹C]raclopride (ordinate) in ventral striatum, total, right, and left caudate, and total, right, and left putamen of men with a range of Zuckerman scores.

Table 1.

Goodness of fit of quadratic regression to binding potentials versus Zuckerman score

Region	Residual SE on 15 degrees of freedom	Multiple R2	Adjusted R2	F statistic on 2 and 15 degrees of freedom	P value
Ventral striatum	0.2107	0.3468	0.2597	3.982	0.041
Caudate, total	0.2927	0.2578	0.1588	2.605	0.11
Caudate, left	0.2971	0.1413	0.02677	1.234	0.32
Caudate, right	0.3242	0.327	0.2372	3.644	0.051
Putamen, total	0.185	0.4204	0.3431	5.44	0.017
Putamen, left	0.1859	0.3833	0.3011	4.661	0.027
Putamen, right	0.2202	0.3859	0.3041	4.714	0.026

We tested regression of the quadratic equation toward the individual voxel values of the binding potential as shown in Fig. 2 for the coefficient . The significance of this regression is shown in Fig. S2.

Fig. 2.

Representative transaxial brain maps of magnitude of quadratic coefficient calculated voxelwise by least-squares optimization of the quadratic function , where Zuckerman’s sensation-seeking score is the explanatory variable and the binding potential of each voxel is the dependent variable. The color codes the negative coefficient of the inverted-U convexity. Eight pairs of sections with z axis coordinates are shown with and without superimposition on the standard MRI brain atlas of the MNI for precise anatomical identification. The color scale covers the range of negative coefficient values between 0.001 and 0.01.

We expressed the quadratic equation in terms of its roots to assess the separate linear components that define the quadratic form of the inverted-U shape,

where is the binding potential, a is a coefficient, and are the roots, and z is the Zuckerman score. This product is consistent with the factors of Eq. 2. Separately, however, the observation that the dopamine receptor binding capacity and Zuckerman scores both decline as functions of age is consistent with a simple linear relation between the binding capacity and the Zuckerman score. The form of this relationship for a relevant range of Zuckerman scores is

where is the ordinate intercept for , and is the rate of incline of the binding capacity with the Zuckerman score, equal to for the root . As a function of the Zuckerman score derived from Eqs. 3 and 4, the gain is

where is the intercept and is the rate of change of the gain with the Zuckerman score, equal to for the root . At a given arbitrary Zuckerman score value of Z, the gain is then and the binding capacity at this Zuckerman score is , where the binding potential is according to Eq. 2. By insertion of Eqs. 4 and 5, Eq. 3 yields the complete quadratic equation,

where the coefficient inside the curved brackets is the product in Eq. 3.

The regression of Eq. 6 to the binding potentials observed for the Zuckerman score of each subject is shown in Fig. 3. In the regression, the value of the Zuckerman score to which the average binding potential refers in Eq. 6 was chosen as the population average of 20. The results of the regression in Table 2 show a linear relation between the Zuckerman score and the binding capacity. For each unit increase of the Zuckerman score, the binding potential in the ventral striatum has risen about 20% of the baseline, and in putamen 11–14% of the baseline, with an average of 12% for total putamen. The gain separately has declined by 2% in ventral striatum and putamen for each unit increase of the Zuckerman score. As the binding capacity is known to be close to 30 pmol cm⁻³ at the normal average Zuckerman score of 20 (16), the capacity can be estimated to rise slightly less than 2 pmol cm⁻³ for each unit increase of the Zuckerman score. For the gain and related occupancy, the estimated decline of the gain and incline of the occupancy is 2–3 percentage points for each unit increase of the Zuckerman score, depending on the occupancy at a Zuckerman score of 20. The increase of dopamine concentration necessary to obtain this increase of occupancy with a linear increase of the binding capacity is a nonlinear increase of the concentration as a function of the Zuckerman score. Normalization of the binding potentials to the binding potential at the population average value of the Zuckerman score used for reference yielded a single relationship potentially valid for all three regions obeying the inverted-U shape (ventral striatum and left and right putamen),

where the ratio is the normalized binding potential. The regression of this equation to the combined normalized data, shown in Fig. 4, yielded the estimates listed in Table 3 of a single set of parameters.

Fig. 3.

Binding potentials in putamen and ventral striatum in relation to Zuckerman scores. Regression of Eq. 6 to measured binding potentials in three subdivisions of striatum, left and right putamen and ventral striatum. Abscissa: Zuckerman score. Ordinate: binding potentials. Regression of Eq. 6 yields estimates of binding potential at selected Zuckerman scores as well as relative decline or increase of gain and binding capacity as related to Zuckerman scores.

Table 2.

Results of regression of Eq. 6 to binding potentials (Fig. 2)

Regions	Ventral striatum	Total putamen	Left putamen	Right putamen
Degrees of freedom	15	15	15	15
BPND(Z) (Z = 20)	2.4 ± 0.070	3.1 ± 0.062	3.1 ± 0.062	3.1 ± 0.074
k1	−0.021 ± 0.0023	−0.019 ± 0.0020	−0.019 ± 0.0020	−0.019 ± 0.0024
k2	0.21 ± 0.20	0.12 ± 0.066	0.11 ± 0.058	0.14 ± 0.093
R2	0.35	0.42	0.38	0.39
Absolute sum of squares	0.67	0.51	0.52	0.73
Sy.x	0.21	0.18	0.19	0.22
Normality of residuals
D'Agostino and Pearson omnibus K2	1.0	3.5	0.42	1.5
P value	0.6017	0.1716	0.8115	0.4637
Runs test
Points above curve	9	9	10	10
Points below curve	9	9	8	8
Number of runs	8	10	8	11
P value (runs test)	0.2380	0.6008	0.2514	0.7822
Deviation from model	Not significant	Not significant	Not significant	Not significant
Covariance matrix
BPND(Z) and k1	−0.65	−0.65	−0.65	−0.66
BPND(Z) and k2	0.45	0.48	0.49	0.48
k1 and k2	−0.85	−0.89	−0.89	−0.88

Fig. 4.

Binding potentials normalized to binding potential at selected Zuckerman score in relation to Zuckerman scores. Regression of Eq. 7 to normalized binding potentials in three subdivisions of striatum where the inverted-U-shaped correlation is significant. Abscissa: Zuckerman scores. Ordinate: normalized binding potentials. Regression of Eq. 7 yields a single set of estimates of rates of decline and increase of gain and binding capacity, as related to the Zuckerman score.

Table 3.

Binding potentials normalized to Z = 20 (Fig. 3)

Estimate	Value
Parameters
Normalized BPND(Z) (Z = 20)	1.0
k1	−0.020
k2	0.15
SEs
Normalized BPND(Z) (Z = 20)	0.013
k1	0.0012
k2	0.056
95% confidence intervals
Normalized BPND(Z) (Z = 20)	0.97–1.0
k1	−0.022 to −0.017
k2	0.034–0.26
Goodness of fit
Degrees of freedom	51
R2	0.35
Absolute sum of squares	0.25
Sy.x	0.070

Discussion

Known as the Yerkes-Dodson Law (22), the inverted-U-shaped relationship between performance and arousal was first introduced by Wundt (23) and later elaborated by Eysenck (24). When arousal exceeds a certain threshold, the law holds that performance decreases (22). The highest performance is achieved in the midrange of arousal (midseason form). The inverted-U shape can be explained by the effects of two separate factors: the upslope indicating the effect of arousal, and the downslope indicating the specific effects of arousal on cognitive processing. Although the results of much research are consistent with this correlation (25), a single cause has not been established. Although performance and arousal are unlikely to represent single mechanisms, results of several studies specifically implicate relations between cognitive functioning, impulse traffic, and prefrontal activity on the one hand with dopamine signaling and dopamine concentrations on the other (25 –27).

The Zuckerman scale of sensation seeking has close ties to the personality scales of both Eysenck and Cloninger (28). The present results lend support to an inverted-U-shaped association between this sensation-seeking personality trait and dopamine receptor availability, the latter a potentially important factor in addiction. The associations between receptor binding and personality score were equally significant in putamen and ventral striatum. We detected no effects of age in this study, which is easily explained by the restricted age range of subjects (22–44 years).

The absence of a history of neuropsychiatric disease in the men introduces a potential bias, because of a suspected link between higher sensation seeking and neuropsychiatric pathology. Hence, the more highly sensation-seeking men in the present study may not be representative of unselected populations. The primary measure of binding potential has confounds, too. Dopamine receptor availability is a composite variable, proportional to the number of receptors and inversely proportional to the apparent affinity of [ ¹¹C]raclopride for the relevant receptors. The apparent affinity in vivo is itself a function of the intrinsic affinity of the ligand, reduced by competition from endogenous dopamine, which occupies 10–40% of striatal dopamine receptors at baseline (29, 30). Values of cannot be resolved into the separate components of and the product , in which incorporates information about the receptor affinity and the concentrations of competing ligands. Studies of varying pharmacological blockade or differing specific activities in humans indicate a of 30 pmol g^-1 and a product close to 10 pmol g^-1 for [¹¹C]raclopride binding in striatum (12, 30) (also see ref. 16).

When changes of dopamine are measured directly in animals, treatment with amphetamine raises both the dopamine concentration and the product, consistent with increased occupation of the dopamine receptors by endogenous dopamine (31). The role of dopamine in reward is evident in studies in which amphetamine self-administration is disrupted by dopamine-selective (but not noradrenaline-selective) receptor antagonists (32 –34). The effects disappear when dopamine antagonists are coadministered with amphetamine or cocaine (35, 36).

The present experimental design involved no perturbation of the brain or any direct measures of dopamine. Hence, the association of higher sensation seeking with lower dopamine receptor availability can reflect multiple factors, including lower receptor density, higher receptor affinity, higher volume of distribution, higher dopamine with increased receptor occupancy, or a combination of these factors. Yet Zuckerman (37) determined the relationship between sensation seeking and the activity of monoamine oxidase and concluded that sensation seeking is linked to elevated monoamine concentrations associated with low monoamine oxidase activity.

The concept of gain is closely related to the observations of responsivity of dopaminergic neurotransmission as influenced by the differential tonic and phasic release of dopamine from neurons identified in studies of firing patterns (38). On the assumption that the combination of additional dopamine receptors and additional dopamine overcomes the reduction of the gain that results from higher dopamine concentrations, and with the present results of an inverted-U-shaped relation between binding potential and Zuckerman scores, we used Eq. 2, 3, and 7 and the regression results listed for total putamen in Table 2 to reconstruct the hypothetical relations between the properties of the dopaminergic neurotransmission that directly or indirectly relate to the Zuckerman score. To determine these properties, we adopted average values for the density of the dopamine D_2/3 receptors of 30 pmol g^-1 in the striatum (16) and the occupancy of dopamine (σ) at the receptors of 40% (29, 30), and we assumed a constant inherent affinity of the radioligand that we derived from the average variables and the estimate of the binding potential (3.1; Table 2) at the Zuckerman score of 20 , although its value need not be known numerically to determine the general relationships described below.

We calculated the properties of dopaminergic neurotransmission for the entire range of Zuckerman scores from the average variables. The properties included the maximum binding capacity according to Eq. 4–6 , the gain according to Eqs. 2 and 3 , the occupancy according to Eq. 2 , the bound dopamine , the normalized dopamine concentration , and the predicted binding potential according to Eq. 6, as shown in Fig. 5. The relations depend entirely on the assumptions, including that of a constant affinity. It is possible to recreate the relations with varying affinity for dopamine rather than varying receptor density, or combinations thereof, but the parallel changes of receptor density and Zuckerman scores would not be upheld in such a case. Also, although the calculated relations imply a continuum, the values apply to separate individuals in whom the mechanisms that underlie the curves shown in Fig. 5 are speculative.

Fig. 5.

Relationship between derived variables of dopaminergic neurotransmission and the Zuckerman score. Abscissa: Zuckerman score. Ordinates: maximum binding capacity , calculated from Eq. 4, logarithmic gain (γ), calculated from Eq. 5, receptor occupancy (σ), calculated from Eq. 2, bound endogenous ligand , calculated from maximum binding capacity and occupancy, endogenous transmitter concentration relative to inhibitory constant , calculated from occupancy, and measured Zuckerman score (Z).

The present findings contribute to the evidence that links the personality trait of sensation seeking to dopaminergic neurotransmission. Sensation seeking may be associated with increased dopaminergic transmission (hyperdopaminergic trait), such that more highly sensation-seeking individuals use more dopamine in combination with more numerous receptors to achieve the same effect as normal-range sensation seekers. We argue that increased occupancy is revealed by the relatively low [¹¹C]raclopride binding potentials in striatum of more highly sensation-seeking individuals. The elevated occupancy may explain how high sensation seeking and low binding potentials are consistent with increased risk of addiction in some humans (6–8, 39).

In summary, we show that the availability of striatal dopamine receptors is lower in healthy male subjects that are either more or less highly sensation seeking than men with average sensation seeking, as measured by the Zuckerman inventory. We show that this inverted-U shape of the relationship can result from a specific combination of coupled changes of dopaminergic neurotransmission and dopamine receptor properties. The demonstration is consistent with the claim that elevated sensation seeking is associated with low receptor availability and high dopamine occupancy in the presence of high dopamine concentration and receptor density. Based on theoretical considerations of the binding potential, a gradual increase of the sensation-seeking score in proportion to dopamine occupancy is associated with combined increases of receptor density and bound dopamine in the men. The theoretical relations (Fig. 5) explain but do not prove the observations. We make the testable proposition that sensation seeking is a hyperdopaminergic reaction to dopamine receptor occupancy by dopamine, which is greatest in people who express the highest need to seek sensational activities (40).