hypnotic trance (i.e. the capability to accept suggestion in order to alter perception ... Sebastiani et al., 2005), also in the non hypnotic state and in the absence of ...
Hypnotizability and Polymorphisms of the COMT gene: an association study
INDEX INTRODUCTION HYPNOSIS: A PHYSIOLOGICAL AND NEUROPSYCHOLOGICAL PERSPECTIVE Hypnosis and hypnotizability Hypnosis: a brief introduction Hypnotizability as a cognitive trait The link between hypnotizability and attention Dopamine: a strong candidate gene to be the key-neurotransmitter Classification and functions Synthesis and processing HYPNOSIS: A GENETIC PERSPECTIVE Hypnotizability as a heritable trait The human COMT gene COMT SNPs: structural effects COMT SNPs: clinical associations with neuropsychiatric disorders COMT and schizophrenia COMT and eating disorders COMT and drugs addiction
1 1 1 1 3 4 9 10 12 15 15 17 19 24 25 26 27 i
COMT and bipolar disorders COMT and obsessive-compulsive disorder COMT and attention deficit hyperactivity disorder COMT and Parkison’ s disease COMT and chronic pain disorders Conclusions about COMT clinical associations Population and evolutionary genetics perspective of the COMT gene
27 28 28 29 29 30 32
AIM OF THE STUDY
35
SUBJECTS AND METHODS
36 36 37 37 37 38 43 43 43
Hypnotic susceptibility scale Population and genotyped samples Molecular techniques DNA extraction and amplification COMT sequencing Statistical data analysis Statistical tests Commingling analysis
RESULTS Distribution of hypnotizability in the population Genotype and allele frequency distributions of two SNPs of the COMT gene Haplotype analysis Haplotypic heterogeneity test
44 44 51 54 56 ii
DISCUSSION
57
CONCLUSION
61
AKNOWLEDGEMENTS
62
REFERENCES
62
Appendix A: Areas of interest in the
I
Nervous System Appendix B: Stanford Hypnotic Susceptibility Scale, Form A
VI
Appendix C: DNA extraction protocol
X
Appendix D: “Amplifying” PCR protocol and PCR products purification protocol
XI
Appendix E: Sequencing
XIV
Appendix F: Genotype and haplotype data
XVIII
analysis Appendix G: Analysis of the distributions of hypnotizability
XXVIII
Appendix H: Supplementary data tables
XXXVI
iii
INTRODUCTION HYPNOSIS: A PHYSIOLOGICAL AND NEUROPSYCHOLOGICAL PERSPECTIVE Hypnosis and hypnotizability Hypnosis: a brief introduction Despite a long history behind, hypnosis has always been an amusing matter of debate among scientists. Indeed, there still is an intense ongoing disputation about whether hypnosis should be considered an altered state of consciousness or an attentive receptive concentration in a wakeful state. Notwithstanding that, the hypnotic process can be described, from a behavioral viewpoint, as the behavior that we observe in a hypnotizable individual after a hypnotic induction procedure. This procedure involves, as specified by the American Society of Psychological Hypnosis: 1. an introduction to the operation, during which the subject is told that suggestions for imaginative experience will be presented; 2. a real induction, i.e. an extended initial suggestion for using one’s imagination, that can contain further elaborations of the introduction; 3. a phase in which the subject is encouraged by the hypnotist to respond to suggestions for changes in subjective experience (like feeling a non real sensory stimulus), alterations in perception (such as analgesia), sensation, emotion, thoughts or behaviour. This definition (Green & al., 2005) agrees with the description of hypnosis from a first person experiential perspective, that considers it as the set of changes in the subjective experience induced by suggestions and characterized by mental ease, absorption, reduced
1
self-orientation and automaticity (Rainville & al., 2003). In fact, in the “experiential model” we identify a beginning in a relaxed condition of mental (and often physical) ease, in combination with an absorbed and sustained focus on objects of attention, after being suggested so by the hypnotist. This relaxation/reduction of active attention supports an altered self-regulation, a lack of monitoring reality and perceived stimuli, which in turn determines the uncensored acceptance of what is being said by the hypnotist, that is no more verified by the own proper critical capability. In this way a sense of automaticity emerges, wherein thinking is no longer felt as preceding action but action is felt as preceding thought. By doing so, suggested changes in perception, imagination and behaviour are simply felt as automatically occurring. Of course, such a deep involvement of very different cerebral activities (from sensorial perception and attention to imagination, emotion and others) implies the commitment of widespread brain networks, composed by several neural structures and different kinds of neurotransmission, often engaged in more than one function. As a confirmation, from a neuropsychological point of view (thanks to the use of various brain imaging techniques), we can notice that the initial focusing of attention on an object engages the left-sided part of the frontal supervisory attentional systems (located in the frontal cortical lobe and in the limbic circuit, see Appendix A) and, after the eyes closure, the suggestion for imagery and emotion produces a shift of activation to posterior right-sided systems in the brain, with a concomitant inhibition of frontal functions. This is compatible with the suspension of both reality testing and critical analysis abilities (Gruzelier, 1996). The potential usefulness of hypnosis in the medical field is worth of a quotation: various studies demonstrated the important role of hypnosis in the control of inflammatory processes (such as skin hypersensitivity reactions or IgA production, Gruzelier, 2002) or the value of hypnosis
2
training for the suggestion of analgesia in the treatment of chronic pain or cancer patients (Jensen, 2009). Generally speaking, the efficacy of hypnosis in improving quality of life, alleviating pain and distress and counteracting the adverse effects of some therapies (such as chemotherapy in cancer patients) is beyond doubt (Gruzelier, 1996). Hypnotizability as a cognitive trait In the latest decades, a great emphasis was given to the hypnotic trait hypnotizability, that is, the cognitive ability enabling subjects to enter hypnotic trance (i.e. the capability to accept suggestion in order to alter perception and memory). A number of psychological instruments (scales) can measure it. The most used are the Stanford Scales (form A, B, or C), standardized 12-items measures of a subject’s response to suggestion following hypnotic induction. The suggestions include lowering of a hand while imaging holding a heavy weight, hypnotic dreaming, age regression, auditory hallucinations, negative visual hallucinations and other actions (further detailed in Appendix B). Each suggestion evaluated by the hypnotist as successfully completed (according to objective criteria) is counted as a point. At the end of the hypnotic suggestions, the final cumulative score (over a maximum overall score of 12) determines the classification of an individual as highly, intermediately or scarcely hypnotizable. Based on these scales, hypnotizability can be considered a complex phenotypic character, whose expression is by definition determined by multiple genetic as well as environmental factors and shows a characteristic distribution in the population. At variance with earlier reports showing a Gaussian distribution of the hypnotizability scores in the healthy population, other different distributions (depending on the scale used for the measurement) have been published (Balthazard & Woody, 1989).
3
Moreover, hypnotic susceptibility is associated with many behavioral/psychological differences, and subjects with high (Highs) and low (Lows) hypnotizability may be located at opposite extremes of a continuum of interindividual differences (Carli & al., 2008). Interestingly, the discrepancy between highly susceptible individuals (usually called “highs”) and barely susceptible ones (“lows”) involve several differences in cognitive capabilities, such as focused attention and absorption in mental images (Crawford, 1994, 1997; Tellegen & Atkinson, 1974) and proneness-to/efficacy-of imagery (albeit this relation is not yet recognized as univocal and it doesn’t seem to be linear, Glisky & al., 1995, Spiegel & al., 1998, Crawford, 1994, 1997). In addition, hypnotizability-related differences have been detected in sensori-motor integration (Santarcangelo & al., 2008a, b; Carli & al., 2008) and cardiovascular functions (Santarcangelo et al., 2008c; Sebastiani et al., 2005), also in the non hypnotic state and in the absence of specific suggestions. In particular, a more efficient endothelial function (i.e. a more efficient flow-mediated dilation of vessels, likely due to a larger NO bioavailability) has been observed in Highs during mental stress and nociceptive stimulation (Jambrik & al., 2004; 2005). The link between hypnotizability and attention Among the several traits associated with hypnotizability, one of the most strictly connected cognitive capacities is represented by focused attention. Attention is described as the process of selecting for active processing certain aspects of our physical environment (e.g. objects) or ideas stored in our memories (Raz, 2005). It permits voluntary control over thoughts, feelings and action as a mean of self-regulation. Furthermore, it constitutes a multidimensional phenomenon, divisible in four main dimensions (Crawford, 1994):
4
• focused and sustained attention (i.e. the ability to focus and sustain attention over time without distractions); • selective attention (i.e. the ability to select and discriminate between stimuli); • divided or dual attention (i.e. the ability to divide attention between two tasks); • ambient attention (i.e. the ability to respond to other stimuli while attending to a task). Interestingly, several reports in the past decades led to identification of a striking connection between focused/sustained attention and hypnotizability: particularly, the extremely focused component of attention (i.e. the ability to attend a task so fully that noise and irrelevant environmental stimuli are completely ignored) is closely correlated with hypnotic susceptibility (Crawford & al., 1991, 1993). More specifically, neuroimaging methods (such as Electroencephalography, Cerebral Blood Flow and others) have shown the role of the far fronto-limbic attentional system (see Fig. 1 for localization in the brain) in the hypnotic modulation of the functional relation between prefrontal cortex and the anterior cingulate areas, which is responsible for the active gating of incoming sensory stimuli (Crawford, 1991; Crawford & Brown, 1993).
Figure 1. The prefrontal cortex is the seat of the far fronto-limbic attentional system, which includes also nuclei belonging to the limbic system.
5
From a neuroanatomical perspective, attention can be considered as a system of brain networks, each one carrying out three distinct but interrelated functions (Raz, 2005, Raz & al., 2006): • alerting (that changes a individual’s internal state in preparation for perceiving a stimulus); • orienting (directing the individual’s point of reference to a sensory object); executive control (i.e. selective attention and conflict resolution, which implies choosing among conflicting actions). Each one of these networks has specific anatomical areas and neuromodulators of reference. While cholinergic systems in the basal forebrain (see Fig. 2) play a fundamental role in orienting, norepinephrine system arising in the locus coeruleus (LC, see Fig. 3) is involved in alerting, and anterior cingulated cortex (ACC) and dorsolateral prefrontal cortex (DLPFC, see Fig. 4) dopaminergic systems are engaged in the executive control. •
Figure 2. The basal forebrain, densely innervated with cholinergic synapses.
6
Figure 3. LC, an important aggregate of noradrenergic neurons located in the midbrain, involved in alerting.
Figure 4. (a) DLPFC and (b) ACC (highlighted in blue), whose dopaminergic systems are engaged in the executive control.
7
Being hypnosis often labeled as attentive receptive concentration (Frischholz & al., 1987), several reports have been conducted to evaluate the potential link between attention and hypnosis, by means of the Stroop task (Stroop, 1935). In this task experienced readers are asked to name the ink color of a displayed word and it is typically representative of individuals’ conflict resolution abilities, Indeed, subjects are slower and less accurate in responding to the ink color of an incompatible color word (e.g. the word “green” displayed in red ink, see Fig. 5) than in identifying the ink color of a control item (e.g. the word “home” written in blue ink), thus experiencing the classical Stroop Interference Effect (SIE). Amir Raz demonstrated that effective posthypnotic suggestion consistently cancelled the SIE in highs (Raz & al, 2002), suggesting that the amelioration of the conflict resolution capability must operate via a top-down cognitive mechanism, modifying the processing of input words. Hence, since the SIE classically activates the dorsal part of ACC, a natural connection between this structure, hypnotizability and the activity of monitoring conflict among potential responses has been strongly supported. Other studies utilizing Stroop-like tasks have also demonstrated the commitment of the lateral prefrontal cortex in the same function, furnishing further support to Crawford’s discoveries. More precisely, DLPFC areas are involved in the working memory (i.e. a component of the short-term memory, representing specific information over time), while ACC areas modulate the detection of conflicts.
Figure 5. Example of a Classical Stroop test image.
8
As a further confirmation of these findings, the anterior cingulus is deeply related to hypnotic analgesia and (along with the thalamus and other nuclei, see Appendix A) to the regulation of vigilance and state of consciousness during hypnosis (Rainville & al., 2003). Hypnotic susceptibility has been considered to arise from a more flexible executive attentional control by the supervisory frontal attentional system (Spiegel, 2003). More recent research, however, has put into question the higher ability of Highs to direct attention toward a chosen object and has suggested, on the contrary, that their apparent greater ability to focus attention might consist of a great disability to withdraw attention from its current focus (Egner & al, 2005; Jamieson & Sheehan 2004). This would indicate a weakening of the frontal executive function and the release of lower-level cognitive systems from the supervision of higher-level activities.
Dopamine: a strong candidate to be the key neurotransmitter The cited areas are densely innervated by dopaminergic neurons (particularly the DLPFC) or belonging to important dopaminergic circuits (like the ACC in the limbic system). In addition, dopaminergic synapses are widely distributed in the frontal cortex (See Fig. 6). Thus, since the beginning of research about the neurobiology of hypnotizability, dopamine has been the principal candidate neuromodulator to connect hypnosis with attention. Indeed, the hypothesis that hypnotizability could require cortical activation involving dopaminergic and frontal lobe activity was advanced by Spiegel since 1978 and confirmed by the finding of a strong statistical association between high concentrations of homovanilic acid (HVA, one of the principal dopamine metabolites in the nervous system) in the cerebrospinal fluid and the ability to experience hypnosis (Spiegel & al., 1992). The dopamine theory of hypnosis, however, is still under
9
debate, as a few neurophysiological studies confirm it (Lichtenberg & al., 2004; 2008b), while others do not (Lichtenberg & al., 2008a), in line with recent neuropsychological research (Egner et al., 2005; Jamieson and Sheehan, 2004).
Figure 6. Dopaminergic neurotransmission in the frontal cortex.
Classification and functions Dopamine is part of catecholamines, a class of neurotransmitters characterized by a catechol group (a benzene ring with two hydroxyl group, see Fig. 7a) and composed by epinephrine and norepinephrine as well, which are involved in numerous functions in both the CNS and the PNS. While norepinephrine and mostly epinephrine have an important function (as a neurotransmitter and as a hormone, respectively) in the activation of the Sympathetic system, in the CNS the arousing action of norepinephrine (principally in the LC) and dopamine (in the frontal cortex, hypothalamus and anterior cingulus, see Appendix A) prevails. More precisely, dopaminergic neurons constitute three principal brain systems (see Fig. 7b), each with a main functional specialization: • the nigrostriatal pathway (connecting the substantia nigra to basal ganglia), committed to the motor control;
10
• the mesocortical/mesolimbic system (connecting the ventral tegmental area to the frontal lobe and the limbic system, respectively), essential to the normal cognitive functions of the DLPFC and to the behavioral responses to stimuli activating feelings of reward; • the tuberoinfundibular pathway (made up of the hypothalamus and the pituitary gland), fundamental in the endocrine system regulation. As a confirmation of this roles’ subdivision, dysfunctions in these pathways are at the basis of specific clinical conditions, like: • the Parkinson’s disease, a neurodegenerative illness mainly implying the progressive loss of the motor skills, due to dopamine deficit in the nigrostriatal pathway; • the Attention deficit hyperactivity disorder, owing to the reduction of the dopamine levels in the mesocortical system; • drugs addiction, schizophrenia and other behavioral disorders, related to alterations of the mesolimbic pathway; • endocrine dysfunctions, due to defects in the tuberoinfundibular regulation. The mesolimbic/mesocortical pathway, involving most of the brain structures engaged by the hypnotic process, play an essential role in the hypnosis.
Figure 8.(b) Characteristical catechol group of the catecholamines.
11
Figure 7.(b) Dopaminergic systems in the brain.
Synthesis and processing The importance of dopamine derives also from being the precursor of the other catecholamines. Indeed, these amines have a common biosyntethic pathway (containing five enzymes, see Fig 8b) that, starting from tyrosine, first converts it to L-dihydroxyphenylalanine (i.e. L-DOPA, through tyrosine hydroxilase), then to dopamine and CO2 (through the DOPA decarboxilase). In the LC noradrenergic neurons, dopamine can be further converted to norepinephrine (by dopamine β-hydroxilase), which is in turn methylated to adrenaline in adrenal medulla cells (by phenylethanolamine-N-methyltransferase). The degradation of dopamine is mainly executed by two classes of enzymes:
monoamine
oxidases
(MAO)
and
Catecol
O-
methyltransferases (COMT). The formers are involved in the conversion of dopamine to the corresponding aldehyde, next oxidized to the respective carboxylic acid (i.e. dihidroxyphenyl-acetic acid, DOPAC) whereas the latters essentially catalyze the transfer of a methyl group (from S-adenosyl-methionine) to a hydroxyl group on
12
the catechol nucleus, so inactivating the methylated molecule (either DOPAC or dopamine, see Fig. 8a). Nonetheless, MAO seem to be mainly active in the norepinephrine catabolism in noradrenergic neurons, whilst COMT probably play a major role in the dopamine degradation in the CNS, assuming a grater importance in the frontal cortex, where it is responsible for more than 60% of dopamine degradation (Karoum et al. 1994).
Figure 8. (a) Main degradation pathways of the catecholamines.
13
Figure 8. (b) Biosynthetic pathway of the catecholamines.
14
HYPNOSIS: A GENETIC PERSPECTIVE Hypnotizability as a heritable trait In the last decade hypnotizability has been the object of an increasing interest from a genetic point of view. In fact, considering behaviour as a set of complex genetic characters, a new science was born, at the edge between psychology and genetics, i.e. behavioral genetics, with the purpose of discovering the genetic underpinnings of behavioral traits. Great support to this field was given by the Human Genome Project, which permitted the identification of more than 20,000 genes in the human genome, and by the HapMap project, that has allowed the mapping of 1,7 millions polymorphic sites: these findings illuminate how genes influence disease development, aid scientists looking for genes associated with particular illnesses, contribute to the discovery of new treatments both in medical and, more specifically, behavioral genetics. The research about the hereditary basis of hypnotic susceptibility started more than 30 years ago, with some twin studies. Twins are universally considered the “gold standard” in behavioral genetics, as they allow to clearly recognize the heritability of a genetic trait by assessing the phenotypic concordance in monozygotic (MZ) as well as dizygotic (DZ) twins (with the formers showing a higher concordance than the latters). These pioneering reports, mainly conducted by A.H. Morgan, showed a correlation of 0.63 for MZ and 0.08 for DZ twins (Morgan, 1973) or of 0.52 for MZ and 0.18 for DZ twins (Morgan & al., 1970), thus giving a first insight on a heritable component of hypnotizability. This finding was later reinforced by the discovering
15
of a concordance rate of 78,3% in 60 pair of Russian identical twins (Bauman & Bul’, 1981). More recently, it has been possible to apply a candidate gene approach, mainly as a consequence of the developments of the human genome project. Interest has been mainly focused on dopaminerelated genes that are involved in dopamine metabolism (owing to the fundamental role of dopamine in attentive-hypnotic brain structures, see previous paragraphs). As already seen, one such gene is COMT, given its predictive performance on prefrontal executive cognition and working memory tasks (Weinberger & al., 2001) and a demonstrated influence on frontal lobe functioning (Egan & al., 2001; Malhotra & al., 2002). A missense mutation (Val158Met) has been early detected, which is associated to differences in the COMT activity, thus representing an interesting and promising field of research in the genetics of neuropsychiatric traits (for a more detailed review of COMT clinical association see next paragraphs). This polymorphism has been specifically investigated in hypnotizability association studies. Lichtenberg & al., (2000, 2004) showed that individuals from the Ashkenazi Jew population homozygous for the high-activity allele (Val 158) were less hypnotizable than subjects with the other genotypes. However, the correlation was statistically significant only in females (i.e., not in males). In another study (Raz & al., 2006), Caucasian healthy volunteers were tested for known polymorphisms of several dopamine-related genes, including COMT, DRD3 and DRD4 (encoding MAOA (encoding
dopamine receptors D3 and D4, respectively), Monoamine oxidase A) and DAT (i.e. the
“dopamine transporter” gene). Among these loci, only the COMT
16
gene was found to be associated; heterozygous (Val/Met) individuals were more highly hypnotizable than both the homozygous Val/Val (in line with Lichtenberg’s findings) and the homozygous Met/Met subjects (in contrast with Lichtenberg’s findings). Thus, although a hypothesis of a possible role of the low activity enzyme form (Met allele) in raising dopamine levels in the DLPFC and being responsible for hypnotizability can be proposed, various questions about this phenomenon still remain to be answered (e.g., why Val/Met individuals seem to be more susceptible than Met/Met subjects? why some sex-related differences emerged in the genetic association?). In conclusion, the genetic underpinnings of hypnotic susceptibility remain largely unclear and ambiguous, firstly because hypnotizability (as all behavioral traits) is a complex character, whose environmental as well as genetic susceptibility factors have not yet been completely deciphered, and secondly, due to the small number of studies in this field of research.
The human COMT gene In humans, the COMT protein exists as two length variants, soluble (S-COMT) and membrane-bound (MB-COMT), which are encoded by a single six-exons gene localized to chromosome 22q11.1-q11.2 (Lundstrom & al., 1995; Tenhunen & al., 1994). The S-COMT has 221 amino-acid residues; its translation begins at exon 3 and is regulated by the P1 promoter (i.e. the promoter S in Fig. 10) located in intron 2 (Tenhunen & al., 1994), so producing a 1,3 Kb transcript. The MB-COMT has additional 50 amino-acid residues encoding a transmembrane domain at the N-terminal end of the protein and is regulated by the P2 promoter (i.e. the promoter MB in Fig. 9), located
17
upstream to exon 1, so giving a 1,5 Kb transcript. Except for the 50 amino-acids additional sequence, MB-COMT is otherwise identical to S-COMT. In most tissues, the two length variants of the COMT protein are expressed from both the RNA transcripts, albeit not equally. In fact, while the long mRNA is present in all tissues studied, and is capable of producing both S-COMT and MB-COMT from two different initiation sites (Lundstrom & al 1995; Tenhunen & al 1994, see Fig. 9), the short mRNA is present in greater amounts than the long mRNA in a variety of tissues, but is not able to produce the membrane-bound isoform. Thus, the S-COMT protein predominates in these tissues, such as liver, kidneys and blood (Lundstrom & al 1995). On the other hand, in the brain, the primary translated protein product is MB-COMT and the presence of S-COMT protein is mostly from translation from the secondary initiation site on the long mRNA (Lundstrom et al 1995; Tenhunen et al 1994). More specifically, it has been reported that S-COMT is seen predominantly in glial cells, whereas MB-COMT is seen in striatal neurons postsynaptic to dopaminergic neurons and seems to play a fundamental role in the prefrontal cortex, where the inactivation of dopamine seems to rely preferentially on this enzyme (Turnbridge & al., 2007).
Figure 9. The human COMT gene: structure and transcripts. ATG and TGA represent the start and stop codons, respectively. Filled and open boxes correspond to translated and untranslated regions.
18
COMT SNPs: structural effects Until now, several SNPs (Single Nucleotide Polymorphisms, i.e. allelic variants of a DNA sequence differing each other for a single base, with a frequency of at least 1% among the population) across the human COMT gene have been genotyped (see Table 1), with different allele frequencies among the human populations. Although the majority of these variations do not have any phenotypical effect, some have been reported to modify gene’s expression/protein’s structure and function, though further studies are needed to clarify their roles. For instance, the G allele of the rs2097603 SNP (better known as the “P2 promoter SNP”, due to its peculiar position) has been associated to a slightly reduced enzyme activity, probably owing to down-regulation in the gene transcription (Shifman & al., 2002). However, more are the SNPs residing in coding regions that can produce a change in the peptide sequence, sometimes altering the natural protein folding and functionality. Actually, the so called “non synonymous SNPs” (i.e. always resulting in a amino-acid substitution, usually because the base affected is the first or the second one in the codon of belonging, see Table 2), can alter the protein structure and perhaps the enzyme activity, depending on the position of the residue in the 3-D structure, on the chemical properties of the interchanging residues and on their importance in the protein activity.
19
SNP ID rs59938675 rs4485648 rs9606186 rs5646848 rs5748489 rs2097603 rs2020917 rs737865 rs933271 rs174675 rs5993882 rs5746849 rs740603 rs4646312 rs165722 rs6269 rs4633 rs6267 rs740602 rs4818 rs4680 rs769224 rs174699 rs362204 rs165599 rs165728 rs165815 rs887199 rs887204
Alleles G/A C/T G/C C/T C/A A/G C/T T/C C/T C/T T/G G/A G/A T/C C/T G/A C/T G/T G/A G/C G/A G/A T/C Del/G* G/A T/C C/T G/A G/A
location in the gene 5'-UTR 5'-UTR 5'-UTR 5'-UTR 5'-UTR 5'-UTR (P2 promoter) 5'-UTR intron 1 intron 1 intron 1 intron 1 intron 1 intron 1 intron 1 exon 2 intron 2 (P1 promoter) exon 3 exon 3 exon 3 exon 4 exon 4 exon 5 intron 5 intron 5 exon 6 exon 6 3'-UTR 3'-UTR 3'-UTR
Table 1. Main genotyped SNPs in the human COMT gene. (*Deletion SNP).
20
Residue
SNP ID
22
rs6267
42
rs13306281
50
rs61910731
52
rs5031015
96
rs8192488
108
rs4680
149
rs769224
Variation type Nonsynonymous Nonsynonymous Nonsynonymous Nonsynonymous Nonsynonymous Nonsynonymous Nonsynonymous
Alleles
Alternate residues
G/T
Ser/Ala
G/A
Met/Val
G/T
Leu/Val
G/A
Ala/Thr
C/T
Ala/Val
G/A
Val/Met
G/A
Pro/Leu
Table 2. Main non-synonymous SNPs of the human COMT gene.
One of the most functionally important coding SNPs is Ala22/72Ser (in S-/MB-COMT, respectively, and universally known as rs6267, located in the exon 3), whose effect is the decrease of the COMT activity by 30% (Lee & al., 2005). Indeed, residue 22 belongs to α2 helix and is positioned in a surface loop between α2 and α1 helixes in the enzyme structure (see Fig. 10a), 13 Å from the SAM-binding site and ~20 Å from the catechol-binding site (Rutherford & Daggett, 2009). Therefore, due to its peculiar position, the A22/72S variation leads to an expansion of the overall structure and to the opening of a large cleft in the protein (see Fig. 11a). This repacking around the polymorphic site is translated throughout the protein, significantly increasing the COMT affinity for SAM but also the solvent exposure
21
of the catechol-binding site (Rutherford & Daggett, 2009). Although these structural change are in agreement with the mutation’s detrimental effect on substrate binding, some reports suggest that the putative Ser allele effects are only due to being in high LD with other minor SNPs affecting COMT mRNA secondary structure and protein translation efficiency (Nackley & al., 2009). Therefore, more studies are needed to clarify the role of this mutation. In opposition to the still ambiguous role of A22/72S, another coding SNP, Val108/158Met (in S-/MB-COMT, respectively, and universally known as rs4680, residing in the exon 4), is generally considered more important from a structural and phenotypic point of view. This substitution causes a reductive effect on the enzyme activity more pronounced than that hypothesized for A22S. Indeed, the thermal stability of 158Met-COMT is significantly lower than the Val158COMT, resulting in a notable reduction of the protein levels, with a consequent 40% decrease of the in vivo enzyme activity (Chen & al., 2004). This effect is easily explicable in light of the position of the 108/158 residue in the protein structure (see Fig. 10 and 11b). In fact, it is located in the same area of the residue 22 (precisely in the surface loop between α5 helix and the contiguous β3 chain), constantly in touch with it, ~16-20 Å far from the active site (just like the residue 22). Moreover, it packs against a secondary structure element (i.e. β3) containing substrate-binding residues on its distal end (exactly like residue 22 does with α1). In such a “structural context”, a substitution of a small residue, Valine, with the larger Methionine, causes a distortion of interactions and packing within the polymorphic site and a subsequent translation of these alterations throughout the protein,
22
resulting in a big increment of the overall solvent-accessible surface area (see Fig. 11b). Despite this mechanism is the same generated by the Ala>Ser substitution at the residue 22, the Val>Met variation mainly distorts the SAM-binding site, so affecting the binding to the methyl donor. By contrast, the Met158 allele seems to be overexpressed when compared with Val158, both in brain and in lymphoblasts (Bray & al., 2003).
Figure 10. S-COMT structural model.
23
Figure 11. (a) Structural effects of Ala22/72Ser (rs6267) on S-COMT folding. The most important residues in this structural change are highlighted in fuchsia.
Figure 11. (b) Structural effects of Val108/158Met (rs4680) on S-COMT folding. Red and green residues form the catechol- and the SAM-binding sites, respectively; while the fuchsia residue represents Val158Met.
COMT SNPs: clinical associations with neuropsychiatric disorders Because Ala22/72Ser and Val108/158Met sites are the only known substitutions leading to protein alterations, other possible substantive changes have been largely ignored in the search for association of COMT with dopamine-related illnesses. In particular, among these diseases, the association of COMT with neuropsychiatric disorders constitutes
an
interesting
and
promising
field
of
research.
24
Additionally, in the last years a novel approach is being used, trying to focus the attention on haplotypes (i.e. combinations of alleles characterized by a high LD among them, due to their physical proximity on the chromosomes, which are then transmitted through generations as a “unique inheritance-pack”). Thus, by investigating about associations between COMT haplotypes and several diseases, more SNPs - both coding and non-coding SNPs, both synonymous and non-synonymous SNPs - can be examined at the same time, with a clearer and larger perspective. COMT and schizophrenia The COMT gene has long been considered a strong candidate for schizophrenia susceptibility, not only because of its role in dopamine clearing but also because it maps in the 22q11 region, whose deletion constitutes the molecular basis of the velocardiofacial syndrome (VCFS), a complex disorder characterized by elevated rates of psychosis and schizophrenic-like behaviours. Indeed, the findings suggest that, because of haploinsufficiency of the COMT gene, individuals with VCFS are exposed to a high level of prefrontal dopamine, and this may contribute to their high rate of psychosis and to other psychiatric disorders. Despite these converging lines of evidence, studies on the Val158Met polymorphism and its possible association with schizophrenia produced conflicting results (Hosák, 2007). On the one hand, a number of reports corroborate the thesis of a larger frequency of the Met allele or of the Val allele in the schizophrenic patients; however, several studies report no association between rs4680 and schizophrenic state (Hosák, 2007). By contrast, results of investigations looking for associations of the Val>Met SNPs
25
with aggressive and violent behaviour in schizophrenic patients are more consistent (Hosak, 2007). The research on the Ala22/72Ser SNP has given clearer outcomes, as a correlation between the Ser allele and risk for schizophrenia, and especially with aggressive behaviours (Lee & al., 2005) was evident, despite the small number of studies conducted on this SNP. Moreover, a haplotype-based approach resulted in the discovering of a highly significant association of schizophrenia with a particular haplotype (G_G_G, composed by SNPs rs737865 in intron 1; rs4680 and rs165599 in exon 6). Interestingly, through a separate analysis of the variations, the associations with schizophrenia shown by the noncoding SNPs were found more significant than the association shown by the exon 4 Val158Met site, so suggesting that other sites in the gene or nearby may contribute to the susceptibility. In particular, regulatory region variation may be involved (Shifman & al., 2002). Then, whilst evidence linking neural functioning and behavioral output has been promising, association and linkage studies have so far failed to provide conclusive evidence of a strong link between COMT genotypes and schizophrenia. COMT and eating disorders With concern to eating disorders (ED), modern brain imaging studies have shown that neurotransmitting regulation differs between case and control groups. As these disturbances seem to involve serotonin and dopamine systems, a hypothesis of possible genetic inheritance of ED can be proposed. Despite some works report that the COMT high activity Val allele (G) is more often transmitted from parents to ill daughter than the low activity Met allele (A), the largest European
26
investigation did not find any association of COMT-rs4680 alleles with anorexia nervosa (Hosak, 2007; Mikołajczyk, 2006). A combined genotype-based approach, analyzing both rs4633 (His102His) and rs4680 (Val158Met), showed more interesting results: the joined GGCT (rs4680-rs4633) genotype increased the risk of having ED over fivefold and over sevenfold the risk of having bulimia. Moreover, haplotype CT was found three times more often in ED women than in controls (Mikolajczyk, 2006). COMT and drugs addiction Addictive drugs increase the brain’s dopaminergic transmission, particularly affecting one of the principal dopaminergic systems in the brain, the mesolimbic pathway (see the Dopamine paragraph). In light of this, substance dependence is another promising target for research into the influence of
the Val158Met polymorphism. As a
confirmation, several relationships between rs4680 and different kind of substance dependences have been found. Despite this, these correlations were once again contrasting, as they go from the association between the Met allele and alcohol/cocaine abuse to the link between the Val allele and methamphetamine/heroin addiction (Hosák, 2007). COMT and bipolar disorder (BD) Case-control studies conducted on bipolar patients similarly showed a striking ambiguity in the results. In fact, while some report detected no allelic or genotypic association between bipolar disorder and the Val158Met SNP, others found a strong statistical association between the low activity allele and specifical clinical conditions characterizing the disease. Actually, ultra-ultra rapid cycling BD patients (i.e.
27
experiencing distinct shifts in mood within a 24/48-hour period) displayed a higher Met allele frequency if compared to rapid cycling subjects (i.e. having four or more major depressive, manic, hypomanic episodes per year) and non-rapid cycling subjects (Hosák, 2007). COMT and obsessive-compulsive disorder (OCD) As regards also obsessive-compulsive disorder, there is unclearness about the presence of an association with COMT, with some investigations reporting no statistical correlations and others reporting a mildly significant association between the Met allele and susceptibility to OCD (Hosák, 2007). COMT and Attention deficit hyperactivity disorder (ADHD) Similarly, concerning to ADHD, a number of contrasting issues was reported: a Val158 allele-based susceptibility hypothesis, with a particular influence on the impulsive-hyperactive type of ADHD is in contrapposition with the “no association” hypothesis (Hosák, 2007). Therefore, even in this case COMT SNPs seem to better explicate differences between distinct clinical conditions of the same disease, rather than differences in the susceptibility to the illness. This suggestion is also supported by haplotype-based investigations involving SNPs rs6269, rs4633, rs4818, and rs4680 and reporting a trend for association with the hyperactivity/impulsivity scale for all the SNPs, peaking at the marker rs6269. Particularly, the high COMTactivity
haplotype
was
associated
with
the
highest
hyperactivity/impulsivity score in the ADHD sample (Halleland & al., 2009).
28
COMT and Parkinson’s disease Recent
studies
established
that
genetic
factors
modify
the
susceptibility to sporadic Parkinson's disease (PD). It has been suggested that variations in the COMT gene might increase the risk of PD, but so far the results of this candidate-gene studies have been conflicting. On the one hand, some reports found no association between rs4680 and DP (Kalinderi & al., 2008); on the other hand, a recent study found an intense correlation of the high activity haplotype G_C_G_G (rs6269: A>G; rs4633: C>T; rs4818: C>G and rs4680: A>G) with the late onset PD, and an even stronger association between low (A_C_C_G) and medium (A_T_C_A) activity haplotypes with the control group. This suggests a possible functional association of the former haplotype with PD risk (Bialecka & al., 2008). COMT and chronic pain disorders By means of a haplotype-based approach, the COMT gene has been also investigated for a potential role in chronic pain disorders. Indeed, SNPs resulting in a reduced enzyme activity have been related to fibromyalgia (Vargas-Alarcon & al., 2007), temperomandibular disorder (Diatchenko & al., 2005) and other clinical pain-related conditions (Nackley & al, 2009). Specifically, the same haplotype investigated for associations with PD - rs6269:A>G; rs4633C>T; rs4818:C>G; and rs4680:A>G - was also investigated for possible associations with fibromyalgia, so allowing to find a strong correlation between the low activity haplotype (A_C_C_G, “high pain sensitivity haplotype”) and fibromyalgia risk and symptoms severity (Vargas-Alarcon & al., 2007). As a
29
confirmation, focusing on the three SNPs of the same haplotype residing in the coding region (rs4633, rs4818 and rs4680), individual carrying at least one C_C_G (“high pain sensitivity”, HPS) haplotype showed
a
higher
risk
for
developing
fibromyalgia
and
temperomandibular disorders than C_G_G (“low pain sensitivity” haplotype, LPS) carriers (Diatchenko & al., 2005) . This correlation was significant both for homozygous HPS subjects and for heterozygous HPS/APS (i.e. “average pain sensitivity” haplotype, T_C_A) subjects. These results are explicable in light of the inverse correlation between pain sensitivity and COMT activity: while the LPS haplotype (C_G_G, corresponding to the high activity haplotype G_C_G_G investigated in Vargas-Alarcon & al., 2007) showed the maximum enzyme functionality, the APS haplotype (T_C_A, corresponding to the medium activity haplotype A_T_C_A) and the HPS haplotype (C_C_G, corresponding to the low activity haplotype A_C_C_G) exhibited a 3-fold and a 20-fold decrease in the enzymatic activity, respectively (Nackley & al.; 2009). These reductions are probably due to the Val>Met substitution in the “APS-COMT” and to a loss of translational efficiency in the “HPS-COMT”, owing to the formation of a longer and more stable mRNA secondary structure (Nackley & al.; 2009). Conclusions about COMT clinical associations Therefore, these outcomes shed a new light on the research about the possible phenotypic effects of whole haplotypes on COMT expression, structure and function. This new approach also includes synonymous SNPs, like rs4633 and rs4818 (i.e. His102His and
30
Leu136Leu), and non-coding SNPs (e.g. rs6269), in addition to wellknown non-synonymous variations like rs4680. Additionally, evidence for the role of Val158Met in schizophrenia, substance
dependence,
bipolar
disorder,
obsessive-compulsive
disorder, eating disorders, attention deficit hyperactivity disorder and many other mental and nervous illnesses are notably inconsistent and contrasting, making it unlikely that the COMT gene plays an important role in these mental disorders, although a minor effect cannot be excluded (Hosák, 2007). For this reason, future studies on the COMT gene in mentally ill subjects should be stratified by clinical subtypes of the disorder. That is, studies of endophenotypes instead of the complex disorder seem to be a more promising research strategy. Endophenotypes (a psychiatric concept, often used in the analysis of complex mental disorders to identify phenotypic subsets of the global clinical condition, to be better correlated with genotypes) represent simpler clues to genetic underpinnings than the disease syndrome itself, thus resulting in more straightforward and successful genetic analysis. Therefore, the COMT gene is probably not ‘‘a gene for’’ any mental disorder, but a gene contributing (along with many other genes and environmental factors) to the disorder and influencing its clinical expression, and the Val158Met polymorphism appears to have pleiotropic effects on human behavior.
31
Population and evolutionary genetics perspective of the COMT gene From an evolutionary and population genetics point of view, the human COMT gene represents an interesting target of investigation. Indeed, in the last years several studies aimed to the comprehension of both structural and molecular aspects of COMT SNPs have proposed some insights into the evolutionary history of the gene, explaining its current variability among populations. Through the comparison of the human COMT protein with its hortologous products in other species - such as mouse, rabbit and pig it has been shown that these species exhibit a Leucine residue in the position 108/158 of the peptide sequence (Chen & al., 2004), instead of a Val residue in non-human primates (Palmatier & al., 2004) and of the Val/Met variation in Homo sapiens. Therefore, the Met158 allele is deemed a unique mutation in the human COMT evolutionary lineage, which has not been seen thus far in any other species (Chen & al., 2004). If we consider that, among the three possible alleles cited, the more active form is the Leu-COMT, followed by Val-COMT and Met-COMT, these observations strikingly suggest that there could have been an evolutionary pressure to lower the COMT activity, perhaps correlating with the emergence of higher cortical dopaminerelated functions. This is consistent with the property of the Met allele, which confers improved prefrontal functions, especially in working memory, but is inconsistent with the association of the same allele with an improved risk of cathecolestrogen-related cancers (suggested by an increasing number of studies in the last years) and of neuropsychiatric
disorders
(see
previous
paragraph).
These
considerations suggest that individuals heterozygous for the
32
Val158Met polymorphism may have a selective advantage, in that intermediate COMT activity minimizes the risk for deadly or severely disabling diseases, while slightly improving cortical processing (Rutherford & Daggett, 2009). Despite depending on several not yet confirmed conditions (like a certain link between rs4680 and neuropsychiatric disorders), this fascinating theory is supported by the high frequencies of Val158Met heterozygous individuals in Caucasian populations (heterozygosity near 50%, Palmatier & al., 1999). Furthermore, a notable variability of the COMT enzyme activity among ethnic groups is reported, as a result of a significant worldwide variance in the frequencies of both the 158-alleles (ranging from 0.01 to 0.62). These percentages are nearly equal in Europeans, with the Met allele a little less frequent, while the Val allele is much more common in populations from all other parts of the world (i.e. Africa, South America and Australia, Palmatier & al., 1999). The larger study of population genetics ever conducted at the worldwide level on COMT (with a sample of 2,385 subjects in 45 populations) made clearer the complex global pattern of genetic variation and LD of this gene, yielding the following evidences (Mukherjee & al., 2008): the intensity of LD fluctuates along the COMT gene and among the samples, most regions of the gene show a general tendency of decreasing heterozygosity with increasing distance from Africa;
33
while the upstream and downstream regions of the COMT locus exhibit high LD for all the populations, the coding region exhibits high values of LD for Europeans and Native Americans, but low or no LD for East Asians and Africans, respectively; specifically, European populations display high LD in the two most important haplotypes from a functional point of view, namely the “P2 promoter” (from rs9606186 through P2 promoter SNP) and the “Val158Met” haplotypes (i.e. rs4633_rs6267_rs4818_rs4680); particularly focusing on the Val158Met haploblock, the four SNPs (spanning only ~1 Kb) form five over sixteen possible haplotypes, but only three of these are common in populations of
non-African
ancestry
(C_G_G_G,
C_G_C_G
and
T_G_C_A); in the European populations, the ancestral haplotype C_G_G_G has a frequency of ~30%, while haplotypes C_G_C_G and T_G_C_A, show incidences of ~15% and ~48%, respectively. All of this knowledge about population and evolutionary genetics of the 22q11.1-q11.2 region could be very useful for future clinical and association studies investigating the COMT gene.
34
AIM OF THE STUDY The present work is based on the hypothesis that a genetic polymorphism of the COMT gene is associated to variation among individuals for susceptibility to hypnosis. This gene codes for a Catecol O-methyltransferase (MB-COMT) that participates in degrading dopamine in the CNS. A non-synonymous SNP (rs4680) in this locus has been extensively investigated in relation to several behavioral and cognitive traits (schizophrenia, drug addiction, bipolar disorder, obsessive-compulsive disorder, eating disorders, attention deficit hyperactivity disorder and other traits); it has also been reported to be associated to hypnotizability in two studies, though results were in part conflicting. We first examined the phenotypic distribution of the hypnotizability trait in the general population, in a sample of 1043 subjects, by commingling analysis, and confirmed the existence of a subgroup of individuals with a particularly high score (called “Highs”); then, we selected for genetic analysis 27 such highly hypnotizable subjects as well as 19 individuals with the opposite phenotype (“Lows”). 57 Newborns were selected as a sample representative of the general population. By direct sequencing of a portion of exon 4, we determined the COMT rs4680 genotype distribution, and the haplotype composition considering another close SNP (rs4818).
35
SUBJECTS AND METHODS Hypnotic susceptibility scale In this study, individuals’ hypnotic susceptibility was measured by means of the Standford Hypnotic Susceptibility Scale, Form A (SHSS:A, Weitzenhoffer & Hilgard, 1959). It consists of 12 items of progressive difficulty (from postural sway and eye closure to visual hallucinations and amnesia, see Appendix B for further details), administered during the hypnotic suggestion. For every task, a score of either “1” or “0” can be assigned, relying on being considered by the hypnotist as successfully accomplished or not (according to objective criteria). At the end of the hypnotic suggestions, each examined subject is assigned a final overall score (over a maximum score of 12, depending on the number of hypnotic suggestions considered successfully completed), so determining its classification as highly, intermediately or scarcely hypnotizable (called “High”, “Intermediate” and “Low”, respectively; see Table 3).
SHSS: A
Lows
Intermediates
Highs
≤3
> 3 and < 9
≥9
score Table 3. Individuals’ hypnotizability classification, based on the SHSS: A score.
Thus, the higher is the overall final score, the more responsive the subject is to hypnosis.
36
Population and genotyped samples Our historical series of subjects evaluated for hypnotizability included a total of 1043 subjects. Recruitment and evaluation were performed in two main locations, Pisa and Siena, respectively. Among individuals with extreme hypnotizability phenotypes (based on SHSS:A), 46 healthy Caucasian volunteers - all belonging to the “Pisa” sample - were selected for the study, which followed the rules of the Declaration of Helsinki. The groups were constituted by 27 Highs (score ≥ 9/12; age = 24 ± 3.2 years old; 18 females and 9 males) and 19 Lows (score ≤ 3/12; age = 25 ± 2.8 years old; 13 females and 6 males). These subjects had previously signed an informed consent approved by the local Ethics Committee. In addition, a group of 57 newborns (umbilical cords samples, provided by the Immunohemathology Unit Bank at the Azienda Ospedaliera-Universitaria Pisana) were genotyped (but obviously not tested for hypnotic susceptibility), as representative of the distribution of the COMT genotypes in the general population.
Molecular techniques The three groups of study were genotyped for SNPs in a portion of interest (210 bp in the exon 4 of the human COMT gene, see Fig. 12b). DNA extraction and amplification Genomic DNA was isolated by a commercial kit in according to manufacture’s instructions (QIAamp1 250 DNA Blood kit, QIAGEN GmbH, Hilden, Germany) from Highs’ and Lows’ peripheral blood leukocytes (see Appendix C for further details). The same was done with umbilical cords samples from Newborns. For privacy requirements, blood samples were coded anonymously. The DNA extracted from 500 µl of blood was diluted with 200 µl of H2O, quantified by UV measurement at OD 260 nm and stored at -20°C until further
37
processing. At the needed moment, the DNA sample was restored at a normal temperature, then underwent a Polymerase Chain Reaction (see Appendix D for laboratory protocols), aimed at amplifying the target region in the COMT gene, i.e. a portion of 210 bp in the exon 4, in which the SNP rs4680 resides. In the first amplification and sequencing we encountered some troubles. In fact, by firstly utilizing a couple o primers already used in Daniels & al., 1996 (see Table 3a), the sequencing generated genotype data that showed no adherence to the Hardy Weinberg Equilibrium. This was noticeable without the use of any statistical instrument, provided that heterozygous frequencies were ~85% for Newborns, 95100% for Lows and 44-48% for Highs (data not shown). A χ2 statistical test confirmed our insight. This could be due to a wrong annealing of the primers, resulting in the amplification of a region different from the COMT exon 4. Therefore, we repeated the PCR using two new primers (see Table 4) and amplifying a smaller region, yielding reliable genotype data (as we will see in the Results). Primer "old" primers "new" primers
Sequence
TGTAAAACGACGGCCAGT 21M13COMT (F) ACTGTGGCTACTCAGCTGTG CAGGAAACAGCTATGACC T4COMT (R) CCTTTTTCCAGGTCTGACAA COMT-F
CTCCAAGTTCCCCTCTCTCCACCTG
COMT-R
GTTGGGGCTCACCTCCAAGAGAAGC
Table 4. Primers used in the “amplifying PCR”. “Old” primers (published in Daniels & al., 1996) show tail-sequences (in red), that were later used as primers in the “sequencing PCR”. By contrast, in the sequencing PCR following the amplifying PCR with “new” primers, an apposite sequencing-primer (typically shorter than normal primers, See Table A caption in the Appendix E) was used.
COMT sequencing After DNA amplification and later purification, the target portion (see Fig. 12a and b) was sequenced (see Appendix E for further details). A
38
peculiar software stored sequence data of every sample in a directory and, by comparison with a COMT sequence of reference (see Fig. 2 in Appendix E), the program identified two SNPs in the region of interest (see Fig. 12a, b and 13a, b): a transversion CG at position 408 of the MB-COMT ORF, the 3rd base of the codon 86/136 (in S-/MB-COMT, respectively), i.e. the synonymous SNP Leu86/136Leu (variation ID: rs4818); a transition GA at position 472 of the MB-COMT ORF, the 1st base of the codon 108/158 (in S-/MB-COMT, respectively), i.e. the non-synonymous SNP Val108/158Met (variation ID: rs4680). The “rs” ID of the genotyped variations were obtained by crosschecking against the National Center of Biotechnology Information (NCBI) database.
39
domain” highlighted in green.
40
Figure 12. (a) SNP identification in the portion of interest: MB-COMT polypeptide sequence (in FASTA format) and typed SNPs, with the “exon 4
sequence is delimited by red inked letters.
end of the sequence). The codons affected by the typed SNPs are highlighted in green, while the exon 4
Figure 12. (b) SNP identification in the portion of interest (spanning from the dark pink G through the
41
Figure 13. Electropherograms of (a) rs4818 SNP and (b) rs4680, computed by the SeqScape® software. In each row (i.e. each individual) presenting the variation (C/G), we can see either two different intermediate peaks (meaning heterozygosis for the SNP) or a single major peak (meaning homozygosis for one of the alternative bases).
42
Statistical data analysis All data manipulation and analysis was carried out in Excel® (Microsoft Corporation®), using our own calculation procedures, or specific add-in routines. Statistical tests The G2-test was used to evaluate heterogeneity between phenotype and genotype distributions (see Appendix F ); adherence of the genotype data to Hardy Weinberg Equilibrium (HWE), for both the typed SNPs, was tested by χ2 analysis. The Fisher exact test was used to test heterogeneity of the allele frequencies between pairs of samples. Estimation of two-site haplotype frequencies was obtained by the Expectation-Maximization algorithm (see Appendix F). Linkage Disequilibrium (LD), was estimated through the D parameter and Lewtonin’s D’. Heterogeneity of the haplotype frequencies among groups was tested by assessing the overlap of their Confidence Intervals (CI, see Appendix F for further details). Commingling Analysis The commingling analysis is a statistical method for distinguishing an admixture of two or more component distributions within a general distribution. For this analysis, an Excel worksheet was set up, which allowed us to maximize the likelihood of the data, varying the parameters of the models. The Solver routine of Excel was used to maximize the likelihood. The Akaike Information Criterion was used to evaluate the appropriateness of different models (see Appendix G for further details).
43
RESULTS Distribution of Hypnotizability in the population A total of 1043 subjects were scored for hypnotizability. Figure 15a shows the distribution of the trait in the total sample subdivided by gender. Scores were heterogeneous (P = 0.004, G2-test) between males (n = 410) and females (n = 633). Females showed a slightly higher mean score (2.95 vs. 2.22), due to a lower frequency of scores “0” and “1” and higher frequency of high scores (“9 to 12”). We also tested the sample for possible age heterogeneity by subdividing it into two groups (fig. 15b): over 20 (age > 20) and under 20 (age ≤ 20). The G2-test was significant (P = 0.02), with the scores “0” and “1” being the most heterogeneous between the two groups (score “0”: under 20 = 58%, over 20 = 43%; score “1”: under 20 = 5%, over 20 = 12%); however, jointly considering the two scores reduced substantially the level of heterogeneity (P = 0.23). As a further check of data consistency, we tested the score distribution between the two subsamples of Pisa and Siena (N = 680 and 363, respectively): the G2test did not show significant heterogeneity (P = 0.07 see Fig 15c). Despite the significance of heterogeneity tests, the differences between subsamples were small, allowing us to consider for analysis the entire sample of 1043 subjects as a single group. The shape of the distribution shows the highest peak (the mode) on the “0” value (including 47.2% of the sample), and decreasing peaks for increasing scores, until the score “5” (2.3%). Above this score, the distribution is almost flat, each score including an average of about 3% of the sample. However, a second mode is present at score “10”, and a third mode at score “6”. Considering the thresholds that defines the “Lows” and the “Highs” (lower than 4, and higher than 8, respectively) we see
44
that 73.2% of the sample belongs to the first group and 13.2% belongs to the second group.
Sex-related hypnotizability distribution in the population of study 60
F
50
M
40 % 30
20 10 0 0
1
2
3
4
5
6
7
8
9
10
11
12
SHHS: A
Figure 15.(a) Sex-related distribution of hypnotizability in the total sample (N=1043).
Age-related hypnotizability distribution in the population of study
%
70 60 50 40
20
30 20 10 0 0
1
2
3
4
5
6 7 SHHS: A
8
9
10
11
12
Figure 15.(b) Age-related distributions of hypnotizability in the total sample (N=448).
45
Distribution of hypnotizability in the subsamples 60
Siena
Pisa
subjects (%)
50 40 30 20 10 0 0
1
2
3
4
5
6
7
8
9
10
11
12
SHSS: A
Figure 15.(c) Distribution of hypnotizability in “Pisa” (N = 680) and “Siena”(N = 363) subsamples. The curves are consistent with the general distribution.
To discriminate between several possibilities concerning the number of component distributions within the total sample, we applied a commingling analysis, a statistical method used for distinguishing an admixture of two or more component distributions within a general distribution. The analysis was based on the assumption that the component distributions were Poissonians; this is mainly suggested by the shape of the observed scores at the lowest values (see also the Discussion section for a more detailed justification). We started by hypothesizing an admixture of two component distributions, which implies estimating the values of three independent parameters by maximum likelihood, i.e. the mean values of the two Poissonians and their weights (Table 6a). As shown in table 6b and figure 16a, the fit of this model to the data was unacceptable, the G2 value being 182.5 (P