Influence of TFAP2B and KCTD15 genetic variability on personality dimensions in anorexia and bulimia nervosa

Abstract Introduction TFAP2B and KCTD15 are obesity‐related genes that interact to regulate feeding behavior. We hypothesize that variability in these loci, isolated or in combination, could also be related to the risk of eating disorders (ED) and/or associated psychological traits. Methods We screened 425 participants (169 ED patients, 75 obese subjects, and 181 controls) for 10 clinically relevant and tag single‐nucleotide polymorphisms (SNPs) in KCTD15 and TFAP2B by the Sequenom MassARRAY platform and direct sequencing. Psychometric evaluation was performed with EDI‐2 and SCL‐90R inventories. Results The KCTD15 rs287103 T variant allele was associated with increased risk of bulimia nervosa (BN) (OR = 4.34 [1.47–29.52]; p = .003) and with scores of psychopathological scales of these patients. Haplotype *6 in KCTD15 was more frequent in controls (OR = 0.40 [0.20–0.80], p = .009 for anorexia nervosa), while haplotype *4 in TFAP2B affected all three scales of the SCL‐90R inventory in BN patients (p ≤ .01). Epistasis analyses revealed relevant interactions with body mass index of BN patients (p < .001). Genetic profiles in obese patients did not significantly differ from those found in ED patients. Conclusions This is the first study that evaluates the combined role of TFAP2B and KCTD15 genes in ED. Our preliminary findings suggest that the interaction of genetic variability in these loci could influence the risk for ED and/or anthropometric and psychological parameters.

intake, such as those recently evidenced by genome-wide association studies (GWAS) and follow-up studies in obese individuals (Waalen, 2014), have not yet been explored in the ED setting. Two such genes are the Transcription Factor AP-2 Beta (TFAP2B) and the Potassium Channel Tetramerization Domain Containing 15 (KCTD15).
The transcription factor AP-2 family encodes critical regulatory factors for neural gene expression and development that also exert a tight regulation of monoaminergic genes (Damberg, 2005). There is a polymorphic region (rs370476693) in the second intron of TFAP2B gene consisting of a variable number of [CAAA] repeats of four or five times, with less number of repeats associated with increased gene expression (Ivorra et al., 2011). This polymorphism has been related to different personality dimensions in women and adolescents (Damberg et al., 2000(Damberg et al., , 2003Nilsson, Sjoberg, Leppert, Oreland, & Damberg, 2009;Prichard, Jorm, Mackinnon, & Easteal, 2007), including one study in patients with binge-eating disorder (BED) (Damberg, Garpenstrand, Hallman, & Oreland, 2001).
The KCTD15 gene is a member of the KCTD family with an antagonist role in neural crest formation (Dutta & Dawid, 2010;Zarelli & Dawid, 2013). A number of GWAS have identified SNPs in or near de KCTD15 gene as putative regulators of body mass index (BMI) (Mei et al., 2012;Paternoster et al., 2011;Thorleifsson et al., 2009;Willer et al., 2009). Interestingly, there is recent evidence in animal models that TFAP2B and KCTD15 homologues co-localize in areas of the brain where they interact to regulate feeding behavior and reward (Williams et al., 2014) (Figure 1).
Genetic alterations in the complex neural system aimed to keep energy balance by food intake are known to trigger and maintain obesity, but the genes involved in the neuronal control of weight regulation might also be relevant for ED (Day, Ternouth, & Collier, 2009;. Indeed, we and others have shown that variability in these genes is also associated with ED-related behaviors (Gamero-Villarroel et al., 2014Muller et al., 2012;Rybakowski et al., 2007). According to this background, we hypothesized in the present work that TFAP2B and KCTD15 variants, either isolated or in combination, may influence personality dimensions in anorexia nervosa (AN) or bulimia nervosa (BN) patients. We also investigated whether this effect was different between females with AN and those with BN, who may share more features with obese individuals. A secondary goal was to evaluate differences in genotype and haplotype frequencies between ED patients and both control and obese subjects.

| Participants
One-hundred and sixty-nine unrelated consecutive female patients with ED (AN = 106, BN = 63), participated in the study. The patients attended the collaborating Eating Disorders Unit at the Mental Health Outpatient Clinic in Badajoz (Spain), and were diagnosed by one psychiatrist and one psychologist using the ED section of the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (Association, 1994). Diagnosis was later reevaluated to comply with the new DSM-V guidelines. Anthropometric (weight, height, and BMI) and psychological parameters (see below) were collected. Diagnosis was blind to genotype. Exclusion criteria, determined upon screening, included dementia, mental retardation, schizophrenia, Turner's syndrome, other neurological disorders, and underlying endocrine pathologies. A total of 181 healthy women from the same geographical area as the patients (Health District of Badajoz, Spain) were also recruited. Interviews were carried out to guarantee that they had never been diagnosed as having any psychiatric disorder or received any psychiatric treatment. All these individuals showed normal BMI values and had no history of ED. Finally, a group of 75 female obese patients with no ED, recruited among patients of the Infanta Cristina University Hospital in Badajoz (Spain) was included for comparison purposes. These subjects were selected from families from the same region as the ED patients that were studied to detect cases of morbid obesity. Obese individuals included reached a weight F I G U R E 1 Hypothesis for the involvement of TFAPB and KCTD15 in the regulation of feeding behavior.
(1) TFAP2B and KTCD15 co-localize in Arcuate (ARC) and Ventromedial (VMN) nucleus of the hypothalamus, where KCTD15 would act like a scaffold for the sumoylation of TFAP2B.
(2) This post-translational modification (via E1-3 enzymes and ATP) would then change TFAP2B function and allow it to induce catecholamines signaling to trigger an orexigenic signal that would eventually lead to consummatory behavior >3 standard deviations above the mean before 14 years of age and referred at least two other morbid obesity cases among first or second degree relatives. Grade 3 overweight (morbid obesity) was considered to be BMI ≥ 40. Individuals with signs and/or symptoms of syndromic obesity were excluded from the study.
All the participants were white Spanish individuals who gave written informed consent. The study protocol was approved by the Ethics Committee of the University of Extremadura and was conducted in accordance with the Declaration of Helsinki and its subsequent revisions.

| SNPs selection
European population (CEU) single-nucleotide polymorphism (SNP) data were downloaded from the 1000 Genome website (https:// www.ncbi.nlm.nih.gov/variation/tools/1000genomes/). We analyzed the coding sequence and adjacent 3′-and 5′-untranslated regions of the TFAP2B and KCTD15 genes (ref. NC_000006.12, NC_000019.10) and tag SNPs were assigned using Haploview v4.2 software (Daly Lab at the Broad Institute, Cambridge, MA, USA; Table 1). The minor allele frequency considered was 10%, and pair-wise tagging with a maximum r 2 of .80 was applied to capture common variations. Thus, a set of four tag SNPs was selected for TFAP2B and five tag SNPs for

| Genotype analysis
Blood samples from all participants were collected and stored at −80°C until analysis. Genomic DNA was isolated from peripheral blood leukocytes in 2-ml aliquots of whole-blood samples with a

| Study of psychological traits
ED patients were evaluated on the first visit to the collaborating Eating Disorders Unit by experienced clinicians. The patients completed the Eating Disorders Inventory Test 2 (EDI-2) and the Symptom Checklist 90 Revised self-reported questionnaires. EDI-2 was designed to assess ED-related cognitive and behavioral characteristics by initially measuring eight main subscales (Drive for Thinness, Bulimia, Body Dissatisfaction, Ineffectiveness, Perfectionism, Interpersonal Distrust, Interoceptive Awareness, and Maturity Fears), to which other three (Asceticism, Impulse Regulation, and Social Insecurity) were added in a second version (Garner, 1991). The EDI-2 test has been validated in the Spanish population showing high internal consistency between the different subscales (Guimera & Torrubia, 1987). The Symptom Checklist 90 Revised inventory, which has also been validated in Spaniards (Derogaitis, 2002), assesses a broad range of psychopathological symptoms. Three general indexes (Global Severity Index, GSI; Positive Symptom Distress Index, PSDI; and Positive Symptom Total, PST) are obtained upon completion of the questionnaire (Derogaitis, 1977).

| Statistical analyses
Differences of quantitative variables between groups of study were assessed with the Student's t-test. Single-marker analyses were carried out using logistic regression models adjusted for age using the SNPassoc R package (Gonzalez et al., 2007). Statistical significance threshold after correcting for the 10 SNPs assayed was set at 0.005.
For the multiple-marker analysis, the p-value was adjusted by the

| RESULTS
Clinical and descriptive characteristics of healthy subjects and obese and ED patients are summarized in Table 2. Control subjects displayed significantly higher weight and BMI than AN patients (p < .05).
Likewise, obese individuals had higher weight and BMI than all the other groups, including BN patients (p < 10 −5 ). In addition, BN patients scored significantly higher than AN patients in the different scales measured in the psychometric evaluation (Table 2).

| Single-marker analyses
Minor allelic frequencies for the polymorphisms assayed are shown in

| Multiple-marker analyses
The frequencies of haplotypes identified in the different groups of study and linkage disequilibrium information are shown in Supplementary Table S1 and Supplementary Figure S1, respectively.
Not all the possible allele combinations were found in the popula- T A B L E 2 Descriptive and clinical variables of patients with anorexia nervosa (AN) or bulimia nervosa (BN), obesity, and healthy controls correlated with lower GSI, PST, and PSDI scores in these women.
After correcting for the number of haplotypes, p-values retained significance in the GSI and PSDI scales (

| Obese patients
Out of the eight SNPs assessed in the obesity and control groups, two were associated with obesity risk. was significantly less frequent in the obese group than in controls (Supplementary Table S1).

| DISCUSSION
There are many shared personality dimensions and risk factors between obesity and ED. These include body dissatisfaction, low selfesteem, anxiety, depression, substance abuse, dieting, binge-eating, and a history of sexual/physical abuse (Day et al., 2009). While some of these traits are the obvious result of the disease, it is likely that the direction of causality runs both ways. Indeed, psychopathological symptoms have been shown to alter eating behavior which in turn leads to obesity (Lee, 2007). Two genes recently associated with obesity have been TFAP2B and KCTD15, for which a combined function that affects eating behavior has been described (Williams et al., 2014;Zarelli & Dawid, 2013). Following the rationale that argues for a connection between ED and obesity, we could expect that genetic alterations in central pathways involved in the modulation of eating behavior and food intake that may lead to obesity, could also be linked to ED.
Our results showed that the rs287103 T variant allele in the KCTD15 gene was not only associated with increased risk of BN, but it was also found to modulate psychological features in these women.
Furthermore, haplotype *6, which was observed to decrease the risk for ED, contained the "protecting" rs287103 C major allele. In contrast, no associations were found for AN risk, which is consistent with the findings by Brandys et al. (2010), who failed to find a significant effect of a KCTD15 polymorphism previously associated with obesity in AN patients; as well as with several GWAS, which have not identified any KCTD15 SNP related to AN risk (Boraska et al., 2014;Hinney et al., 2017;Wang et al., 2011). In this regard, several loci in or near KCTD15 have repeatedly been associated with obesity in a number of GWAS and replication studies (Mei et al., 2012;Paternoster et al., 2011;Thorleifsson et al., 2009;Willer et al., 2009). The cerebral location of the gene indicates that its association with weight increase would likely be based on changes in feeding behavior. Indeed, a recent study has confirmed an association of KCTD15 SNPs and dietary behaviors in children (Lv et al., 2015) and another report has related variant transcripts of KCTD15 to fatness traits in animals . According to our findings, it seems plausible that SNPs in this gene could also help develop aberrant conduct patterns that favor the establishment of ED, should other susceptibility factors or personality dimensions were present. Of note, both SNPs and haplotypes in KCTD15 were associated in our sample with a number of traits that often appear coupled with ED.  behaviors displayed by patients with psychiatric disorders are mediated by the contribution of genes to psychological traits (Cloninger, 1999;Ebstein, Benjamin, & Belmaker, 2000). TFAP2B gene variants have previously been related to different personality dimensions in women, namely anxiety, psychasthenia, guilt, depression, impulsiveness, indirect aggression or addictive behavior (Damberg et al., 2000(Damberg et al., , 2003Hensch et al., 2008;Nilsson et al., 2009;Nordquist et al., 2009); features that are often present in women with ED. Furthermore, a recent study has also observed an association between TFAP2B SNPs and obesity-related traits (Albuquerque, Nobrega, Rodriguez-Lopez, & Manco, 2014). Therefore, a link between variability in this gene and ED is also plausible. To the best of our knowledge, there is only one report in the ED setting that addresses this question, although it only analyzed one polymorphism. The authors showed a higher frequency of the [CAAA] 5 long allele in women with binge-eating disorders than in controls (Damberg et al., 2001). We could not confirm an important role of the [CAAA] 4-5 SNP (rs370476693) in the modulation of psychopathological symptoms in ED patients; however, we did observe that rs552393576, an A-to-G transition located within the first repeat of this polymorphism, was indeed associated with certain traits.
Consequently, it is tempting to speculate that some of the associations reported for the [CAAA] 4-5 polymorphism (Damberg et al., 2000) could be due to rs552393576, which was not searched for in any of the aforementioned clinical studies. But, what would be the precise mechanism linking this gene to these behaviors and specifically to ED? These personality features are expressed by the involvement of a pattern of neuronal adaptations induced by central genes such as dopamine or serotonin, which has been consistently associated with ED-coupled personality dimensions . Dopaminergic and serotonergic genes in the midbrain have AP-2 binding sites; therefore it is likely that these transcription factors, aside from its established role during development, can also influence mood and personality in adults through this neurotransmitter system (Damberg, 2005;Damberg et al., 2001). In the same manner, variability in the KCTD15 locus could play its role by altering TFAP2B post-translational modifications leading to orexigenic signaling and consummatory behavior, as shown in Figure 1.
The analysis of obese patients in our sample could confirm that KCTD15 and TFAP2B variants may be related to obesity risk, a finding that has also been suggested by a number of GWAS and replication studies (Albuquerque et al., 2014;Bauer et al., 2009;Lv et al., 2015;Willer et al., 2009). Moreover, the genetic profile of AN, BN and obese patients did not differ significantly, only rs2817420 was differently distributed between the groups. A fact that stresses this similarity in TFAP2B and KCTD15 genetic variability is that the frequency of haplotype *6 in KCTD15 was significantly higher in controls but was similar between obese and ED patients.
An interesting finding of the present work was that the interaction between genetic variability in TFAP2B and KCTD15 was linked to BN risk as well as correlate with weight and BMI in these patients. These results support previous reports suggesting a combined function of this two genes (Zarelli & Dawid, 2013), and their alleged implication in the regulation of feeding behavior (Williams et al., 2014). It is of note that, overall, the impact of these obesity-related genes was more evident in BN patients than in the AN group, which makes sense if we bear in mind that alterations in the CNS resulting in behaviors leading to obesity (e.g., anxiety, binge-eating, etc.) could also lead to bulimia should additional personality traits were present. Their connection with AN would however be more intricate.
Several limitations have to be considered in this study. First, the relatively small size of the population studied might affect the generalizability of these results. However, the fact that the study was performed in the only primary care center specialized in ED in the area, allowed all the patients to be diagnosed and followed in the same facility by the same clinicians, which reduced the chance that the findings may be due to population structure. Second, we could not determine one of the tag TAFP2B SNPs and therefore the relevance of the region tagged by this variant could not be assessed. In addition, some of the reported associations for individual SNPs were based on the heterozygous genotype, which probably decrease their clinical significance.
Finally, it would have been informative to have EDI-2 and SCL-90R scores for the control population, as well as data on TFAP2B 5-SNP haplotypes, which were unfortunately unavailable.
This is the first study that evaluates the combined role of TFAP2B and KCTD15 genes in ED. Our findings, preliminary as they are, given the described limitations, suggest that the interaction of genetic variability in these loci could influence the risk for ED and/or be associated with anthropometric and psychological parameters in these patients.
Further studies with larger and homogeneous populations of patients are nevertheless warranted to confirm the reported associations.