Neurogenetics

, Volume 6, Issue 1, pp 17–23

Investigation of hormone receptor genes in migraine

Authors

  • Natalie J. Colson
    • Genomics Research CentreSchool of Health Science, Griffith University
  • Rod A. Lea
    • Genomics Research CentreSchool of Health Science, Griffith University
    • Institute of Molecular SystematicsSchool of Biological Sciences, Victoria University of Wellington
  • Sharon Quinlan
    • Genomics Research CentreSchool of Health Science, Griffith University
  • John MacMillan
    • Queensland Clinical Genetics ServiceRoyal Children’s Hospital
    • Genomics Research CentreSchool of Health Science, Griffith University
Original Article

DOI: 10.1007/s10048-004-0205-0

Cite this article as:
Colson, N.J., Lea, R.A., Quinlan, S. et al. Neurogenetics (2005) 6: 17. doi:10.1007/s10048-004-0205-0

Abstract

Migraine is a common neurological condition with a complex mode of inheritance. Steroid hormones have long been implicated in migraine, although their role remains unclear. Our investigation considered that genes involved in hormonal pathways may play a role in migraine susceptibility. We therefore investigated the androgen receptor (AR) CAG repeat, and the progesterone receptor (PR) PROGINS insert by cross-sectional association analysis. The results showed no association with the AR CAG repeat in our study group of 275 migraineurs and 275 unrelated controls. Results of the PR PROGINS analysis showed a significant difference in the same cohort, and in an independent follow-up study population of 300 migraineurs and 300 unrelated controls. Analysis of the genotypic risk groups of both populations together indicated that individuals who carried the PROGINS insert were 1.8 times more likely to suffer migraine. Interaction analysis of the PROGINS variant with our previously reported associated ESR1 594A variant showed that individuals who possessed at least one copy of both risk alleles were 3.2 times more likely to suffer migraine. Hence, variants of these steroid hormone receptor genes appear to act synergistically to increase the risk of migraine by a factor of three.

Keywords

Androgen receptor geneAssociationMigraineProgesterone receptor gene

Introduction

Migraine is a common neurological condition that affects up to 18% of the population and imparts a significant burden on society [1]. Migraine is defined as a primary headache disorder characterised by recurrent attacks of disabling head pain, which may be accompanied by nausea and emesis, and neurological disturbances [2]. The pathophysiology of migraine is not fully understood, and in the absence of any biological marker, diagnosis must be based on subjective criteria alone [2]. In 1988 the International Headache Society published their classification and diagnosis criteria, which has since been revised [3], distinguishing different varieties of migraine, the two most frequent being migraine with aura, and migraine without aura [4]. Both subtypes of migraine are together referred to as typical migraine and recent research indicates that both have a similar etiology [5].

Migraine is known to have a significant genetic component, although the mode of transmission and the type and number of genes involved remains unclear [6]. It is understood that many genes may provide a significant, although small, contribution to each individual’s migraine susceptibility [7]. Therefore investigation of genes involved in known migraine pathways and triggering mechanisms may provide valuable insights into the genetic susceptibility to this disease.

There is considerable evidence to indicate a role for the female sex hormones in migraine susceptibility [8, 9, 10, 11]. There is no gender difference in migraine occurrence prior to puberty, however migraine develops in three times as many women than men during the adult years [11]. In many women, migraine worsens around the time of menstruation, and may cease altogether after menopause or during pregnancy [9]. The exact mechanism of the role of hormones in migraine has not yet been established. Estrogen withdrawal has been suggested, however treatment involving stabilisation of estrogen levels has not been effective in all cases [9]. Epstein et al. [12] have suggested a role for hormonal variation in all women with migraine, but also a role for factors additional to the hormonal environment. This study considers the hypothesis that genes involved in hormonal pathways may play a role in migraine susceptibility, and investigates this hypothesis by analysing hormone receptor variation using an independent cross-sectional association approach. We have previously reported a role for the estrogen receptor 1 (ESRα) gene exon 8 G594A polymorphism in migraine susceptibility in two independent case/control groups [13]. In this study we investigated the androgen receptor (AR) gene (GenBank accession M34233) exon 1 CAG repeat [14], and the progesterone receptor (PR) gene (GenBank accession AY525610) PROGINS insert in the same study groups.

The human AR gene is located on chromosome Xq11–12 and in humans is expressed in various organs including the brain in both males and females. A polyglutamine tract encoded by CAG repeats occurs in exon 1 of the androgen receptor gene [15]. Expansion of this repeat is considered to have an inhibitory effect on the transactivation function of the AR due to the interaction of this region with various co-activators. Short fragments are associated with enhanced receptor function [16]. Reduced activity has been demonstrated to decrease negative feedback to the hypothalamus, resulting in increased serum androgen levels [17]. Abnormal expansions of polyglutamine tracts in the central nervous system have been shown to cause neurodegenerative diseases such as Huntington’s disease and spinocerebellar ataxia type 1 [18]. It has been suggested that the effect of polyglutamine repeat length may be gene-specific. The transactivation activity of the AR may be unaffected on genes that determine sexual differentiation, but compromised on genes necessary for normal neuronal function [18]. Alleles of different sizes within the normal range of the AR CAG repeat have been associated with androgen-dependent prostate cancer [19], and arterial vasoreactivity in males [20]. In this study the AR CAG repeat polymorphism will be examined for a potential association with migraine.

The human PR gene is located on chromosome 11q22. Progesterone receptor expression is up-regulated by estrogen and down-regulated by progesterone in most target tissues [21]. The PR is found in various regions of the human brain including serotonin neurons [22]. Similar to the ESR, the PR can undergo ligand-independent activation and is involved in various intracellular signalling pathways [23]. The PROGINS polymorphic Alu insertion is a 306 bp insertion that occurs within intron 7 of the progesterone receptor gene in some individuals [24]. The G to T substitution in exon 4, which causes a valine to leucine change in the hinge region of the receptor, and a synonymous C to T substitution in exon 5 are linked to the Alu insertion. This complex of PR gene polymorphisms is designated the PROGINS [25, 26]. The PROGINS polymorphism is considered to have a deleterious effect on progesterone receptor expression, through recombination or mis-splicing [24, 27]. Due to the role of the PR in the central nervous system (CNS), this study examined the PROGINS variant for a possible association with migraine.

Methods

Subjects

This research was approved by the Griffith University Ethics Committee for experimentation on human subjects. All 1,150 participants in the study gave their informed consent prior to participation. All subjects were of Caucasian origin, and were recruited from the east coast of Australia through the Genomics Research Centre’s patient clinic. All were interviewed, and completed a detailed questionnaire on personal and family medical history, migraine symptoms, age of onset, frequency, severity, treatment and response, and migraine triggers as previously described [28, 29]. Migraine was diagnosed by a clinical neurologist as either migraine with aura (MA) or migraine without aura (MO), based strictly on the widely accepted criteria specified by the International Headache Society [4]. Due to a proposed similar genetic etiology [5], all participants who suffered from MA, MO, or both subtypes of migraine were classified as suffering from typical migraine, and were analysed as one group, as well as in the sub-categories. The first study population (which will hereafter be referred to as population 1) was comprised of 275 migraineurs and 275 unrelated control individuals. The controls were matched for sex, age (±5 years), and ethnicity to avoid the potential bias of population stratification, and were recruited in parallel at a similar time and from a geographical locations as the case group. The second independent study population (which will hereafter be referred to as population 2) consisted of 300 migraineurs similarly diagnosed and matched with 300 controls. All participants provided a blood sample from which DNA was extracted by a modification of the salting out method used by Miller et al. [30]. All experiments complied with the current laws in Australia.

Genotyping

All markers were amplified by using the polymerase chain reaction (PCR). For the AR gene CAG repeat, the primers used were those previously described by Sleddens et al. [15] and were ordered with the forward primer labelled with a fluorescent phosphoramidite TET. This enabled detection on the ABI 310 Genescan (Applied Biosystems) by capillary electrophoresis. The 20 μl PCR reaction mix contained 40 ng genomic DNA, 0.5 μm of each primer, 1× PCR buffer, 3.75 mm MgCl2, 0.2 mm dNTPs and the DNA polymerase. The thermocycler conditions were 94°C for 4 min, 35 cycles of 94°C for 1 min, 59°C for 1 min, and 72°C for 30 s, followed by a final step of 72°C for 2 min. After visualisation of the PCR products on 2% agarose gels, the samples were run on the ABI 310 Genetic Analyser. It was expected that the allele sizes would range from 171 to 234 bp in length, with the different allele sizes determined by the number of CAG repeats. As this is an X-linked marker, all male samples carry only one copy of this allele.

For the PR gene PROGINS marker, the primers used were those previously described by Lancaster et al. [31]. The 20 l PCR reaction mix contained 30 ng genomic DNA, 0.25  m of each primer, 1× PCR buffer, 1.5 mm MgCl2, 0.2 mm dNTPs and the DNA polymerase. The thermocycler conditions were 94°C for 4 min, 30 cycles of 94°C for 30 s, 51°C for 30 s and 72°C for 45 s, with a final step of 72°C for 2 min. The PCR products were loaded on a 2% agarose gel using a 100 bp DNA ladder for comparison. A fragment size of 173 bp indicated an allele that did not contain the PROGINS insertion. A fragment size of 479 bp indicated an allele that contained the PROGINS insertion. A heterozygote had both sized fragments. A gel electrophoretogram of six migraineur samples and a negative control appears in Fig. 1. Internal controls using random repeat samples and negative controls were used to confirm the genotypes and to reduce the possibility of genotyping errors resulting in a spurious positive association. Our genotyping error rate is estimated to be <5%.
Fig. 1

Analysis of the PR gene PROGINS marker. Agarose gel showing the PCR products amplified from genomic DNA isolated from six migraineurs and a negative control. Lane 1 DNA markers, lanes 2–7 migraneur samples, lane 8 negative control

Statistical analysis

Genotype data and allele frequencies were compared between the migraine and unaffected groups using standard chi-square analysis or CLUMP analysis using the Monte Carlo approach in the case of the AR multiallelic marker. Monte Carlo analysis may be used to analyse markers that result in sparse contingency tables [32]. As recommended by Sham and Curtis [32], we have presented the T1 statistic, which calculates a chi-squared statistic of the raw contingency table, and the T4 statistic, the maximised chi-squared statistic of all possible 2×2 tables. The CLUMP program was run over 5,000 simulations to estimate P values.

Results

To analyse whether variations in two migraine candidate gene loci were associated with typical migraine, we tested the CAG repeat in exon 1 of the AR gene, and the PROGINS insert in the PR gene by independent cross-sectional association analysis. For all genotype analyses internal controls were carried out using random repeat samples and negative controls.

AR CAG repeat

As the AR gene occurs on the X chromosome, only one copy exists in males. Therefore all analyses were performed on allele frequencies. The results of the analysis of the AR variant showed a borderline difference between the affected and control groups producing a T1 χ2 value of 26.46, and a P value of 0.048. The T4 χ2 value of 13.02 with a P value of 0.13, which was achieved by clumping together alleles 1, 2, 3, 5, 6, 9, 10, 11, 13, 15, 16, and 17, was not significant. Comparisons of the MA versus control groups (T1 χ2=18.74, P =0.28; T4 χ2=10.21, P =0.21), and male case versus control groups (T1 χ2=13.31, P =0.42; T4 χ2=8.16, P =0.52), and female case versus control groups (T1 χ2=19.48, P =0.25; T4 χ2=8.49, P =0.40) were not significant. A significant result was seen in the MO versus control group comparison (T1 χ2=33.26, P =0.01; T4 χ2=16.22, P =0.03). The frequency distribution appears in Table 1. The data were dichotomised based on the mode (17 CAG repeats) and a 2×2 contingency table was generated (Table 2).
Table 1

Distribution of AR CAG repeats. MA migraine with aura, MO migraine without aura

Allele counts

Total

Allele

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

No. of repeats

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

Cases

1

4

2

2

17

39

54

58

67

39

53

26

19

5

6

4

1

397

Male

0

1

0

0

5

6

8

9

9

11

5

1

5

1

2

1

1

65

MA

0

1

0

0

3

2

5

7

4

8

3

1

3

1

0

0

1

39

MO

0

0

0

0

2

4

3

2

5

3

2

0

2

0

2

1

0

26

Female

1

3

2

2

12

33

46

49

58

28

48

25

14

4

4

3

0

332

MA

1

2

0

2

7

19

30

35

39

20

40

15

10

4

2

2

0

228

MO

0

1

2

0

5

14

16

14

19

8

8

10

4

0

2

1

0

104

Controls

0

1

0

3

9

27

53

64

51

27

39

28

8

15

2

0

0

327

Male

0

0

0

0

1

4

5

10

5

4

5

5

2

3

1

0

0

45

Female

0

1

0

3

8

23

48

54

46

23

34

23

6

12

1

0

0

282

Table 2

Distribution of AR CAG alleles around the published mean of 17 repeats

Group

CAG n≤17

CAG n >17

Total

Cases

177

220

397

Male

29

36

65

MA

18

21

39

MO

11

15

26

Female

148

184

332

MA

96

132

228

MO

52

52

104

Controls

157

170

327

Male

20

25

45

Female

137

145

282

Total cases vs. controls

χ 2=0.85

P =0.36

The results of the chi-square analysis showed no significant difference in frequencies between the cases and the controls ( P =0.36), the MA group versus the control group ( P =0.83), the MO group versus the control group ( P =0.58), and both the male case group versus the controls ( P =0.35), and the female case group versus the control group ( P =0.22). Published allele frequencies vary somewhat between different ethnic groups, but the range in the normal population is expected to fall between 11 to 31 CAG repeats [18]. The mean number of CAG repeats detected in this study was 19, and the range was between 10 and 26.

PR PROGINS insert

The results of analysis of the PR PROGINS variant in the study population 1 consisting of 275 migraineurs and 275 unrelated control individuals showed that the PROGINS allele was over-represented in the migraine group compared to the healthy controls (genotype frequencies χ2=6.50, P =0.04; allele frequencies χ2=5.65, P =0.02). The results of the subgroup analysis showed a significant difference in the MO (genotype frequencies χ2=13.08, P =0.001; allele frequencies χ2=7.06, P =0.008) and the female subgroups (genotype frequencies χ2=10.64, P =0.005; allele frequencies χ2=8.1, P =0.004), but not the MA (genotype frequencies χ2=2.25, P =0.33; allele frequencies χ2=2.47, P =0.12) and male subgroups (genotype frequencies χ2=0.41, P =0.82; allele frequencies χ2=0.27, P =0.60). The frequency distribution appears in Table 3.
Table 3

Distribution of PR PROGINS polymorphism frequencies in migraineurs and controls of population 1. 11 No PROGINS insert, 12 heterozygote for the PROGINS insert, 22 homozygote for the PROGINS insert

Group

Genotypes

N (alleles)

Alleles

11

12

22

1

2

Migraine

173 (75%)

55 (23%)

4 (2%)

464

401 (86%)

63 (14%)

Male

43 (64%)

22 (33%)

2 (3%)

134

108 (81%)

26 (19%)

Female

130 (79%)

33 (20%)

2 (1%)

330

302 (92%)

38 (8%)

MA

113 (80%)

27 (19%)

4 (3%)

288

253 (88%)

35 (12%)

MO

60 (68%)

28 (32%)

0 (0%)

176

148 (84%)

28 (16%)

Control

182 (84%)

31 (15%)

3 (1%)

432

395 (91%)

37 (9%)

Male

44 (68%)

20 (31%)

1 (1%)

130

108 (83%)

22 (17%)

Female

138 (91%)

11 (7%)

2 (2%)

302

287 (95%)

15 (5%)

Total cases vs. controls

χ 2=6.50

P =0.039

χ 2=5.65

P =0.017

This marker was also investigated in the independent follow-up population (population 2) of 300 migraineurs and 300 controls. The results showed a significant difference in the genotype (χ 2=7.92, P =0.019) and allele frequencies (χ 2=8.78, P =0.003) in the total group analysis, and in the MA subgroup (genotype frequencies χ2=7.28, P =0.026; allele frequencies χ2=7.91, P =0.005). Similar results were seen in both the male (genotype frequencies χ2=5.27, P =0.07; allele frequencies χ2=5.87, P =0.02) and the female subgroups (genotype frequencies χ2=4.81, P =0.09; allele frequencies χ2=4.31, P =0.04), although they did not reach statistical significance. Analysis of the MO subgroup did not reach statistical significance (genotype frequencies χ2=3.53, P =0.17; allele frequencies χ2=3.11, P =0.08). The frequency distribution appears in Table 4. The allele frequencies in both study populations did not deviate from Hardy Weinberg equilibrium at P =0.22 and P =0.13 respectively. The published allele frequencies vary somewhat, but a recent analysis of this variant in 21 diverse human populations reported an average allele frequency of 11% and a heterozygosity of 0.188 [27].
Table 4

Distribution of PR PROGINS polymorphism frequencies in migraineurs and controls of population 2. 11 No PROGINS insert, 12 heterozygote for the PROGINS insert, 22 homozygote for the PROGINS insert

Group

Genotypes

N (alleles)

Alleles

11

12

22

1

2

Migraine

215 (78%)

54 (19%)

8 (3%)

554

484 (87%)

70 (13%)

Male

27 (69%)

8 (20%)

4 (11%)

78

62 (79%)

16 (21%)

Female

188 (79%)

46 (19%)

4 (2%)

476

422 (89%)

54 (11%)

MA

176 (77%)

45 (20%)

6 (3%)

454

397 (87%)

57 (13%)

MO

39 (78%)

9 (18%)

2 (4%)

100

87 (87%)

13 (13%)

Control

228 (87%)

32 (12%)

3 (1%)

526

488 (93%)

38 (7%)

Male

35 (85%)

6 (15%)

0

82

76 (93%)

6 (7%)

Female

193 (87%)

26 (12%)

3 (1%)

444

412 (93%)

32 (7%)

Total cases vs. controls

Χ 2=7.92

P =0.019

Χ 2=8.78

P =0.003

In order to analyse whether the PROGINS variant exerts a dominant or recessive effect on migraine susceptibility, we investigated the effect of the genotype risk groups (12/22; 22 only) and found that the 12/22 genotype was significantly over-represented in the total migraine subgroup of both populations (24%) compared to the total control subgroup (14%) ( χ 2=13.94, P =2×10−4). Odds ratios were calculated using the Mantel Haenszel method of combining the datasets [33], comparing those who carried the PROGINS allele and those who did not. The results indicated that those who carried the PROGINS allele were 1.8 times more likely to suffer from migraine than those who did not carry this allele (OR=1.77, 95% CI=1.23–2.55).

Interaction analysis

Estrogen, progesterone and their receptors play a complex, interdependent role in the CNS. Because we have found a positive association of the PR PROGINS insert in this study, as well as a previously reported association of the ESR1 G594A polymorphism with migraine in the same study population [13], we undertook an interaction analysis to determine if the possession of both risk genotypes confers an increased risk of migraine. The results of this analysis showed that 30% of all migraineurs carried at least one copy of the risk allele from both ESR and PR genes, compared to only 12% of controls. To determine the magnitude of the increased risk of migraine conferred specifically by the risk alleles from both the ESR and PR genes, odds ratios were calculated after dichotomising the genotype frequency data into risk (possessing at least 1 copy of the risk allele from each gene) and no risk (possessing 0 copies of the risk alleles) groups (see Table 5). Comparing the total migraine group against the controls (populations 1 and 2 together) under this grouping scheme produced an OR of 3.2 with a 95% CI of 1.9–5.3. Therefore, it appears that the PROGINS allele of PR acts synergistically with the 594A allele of ESR 1 to increase the risk of migraine. That is, these alleles act in combination to increase the risk of migraine by a factor of three, which is greater than the independent effects of these genetic variants on disease susceptibility.
Table 5

Distribution of individuals who carried none or both risk alleles

Migraine diagnosis

No risk alleles

At least one risk allele from each gene

Migraine

132 (70%)

57 (30%)

Control

191 (88%)

26 (12%)

χ 2=20.53

P =3×10−5

Discussion

The steroid hormones are key regulators of growth, differentiation and function, targeting a wide range of tissues including the male and female reproductive systems, the cardiovascular and skeletal systems, as well as the CNS. The predominant biological effects of the steroid hormones are mediated through hormone receptors, which are expressed in a wide range of target tissues [34]. Steroid receptors are classically known as ligand-activated transcription factors binding with high affinity to hormone responsive genes when activated by hormone [11]. As well as ligand-mediated activation, ligand-independent activation of the ESR, PR, and AR are well documented [23]. Estrogens, progestins, and androgens have direct effects on the CNS, presumably via receptor sites on neuronal membranes [35, 36]. They are known to interact with neurotransmitter receptors and play a key role in the regulation of ion channels [11]. Both the ESR1 and the PR have been detected in serotonin neurons of primates and it has been suggested that they may play an important role in serotonin receptor expression [22, 37].

The role of the steroid hormones in migraine susceptibility is well recognised, but poorly understood. Their complex role in the CNS, in particular their influence in key migraine pathways such as serotonin metabolism [11] and pain processing [38], suggest that genes involved in hormone pathways are promising targets for migraine candidate gene studies. In this study we investigated variants of the steroid hormone receptor genes for a role in migraine susceptibility using a population-based cross-sectional association analysis. We have previously reported a positive association with the ESR1 G594A exon 8 SNP in two independent case-control populations [13]. In this study we tested markers in two other genes involved in hormonal regulation; the AR CAG repeat and the PR PROGINS insert.

The results of the AR CAG repeat total group analysis showed an interesting borderline result, although this was primarily due to the MO subgroup. However, after applying the Bonferroni correction for multiple testing, which set the level of significance at 0.01 (i.e. 0.05/5), the overall results did not reach statistical significance. Furthermore, there was no significant difference in allele frequencies when the data were dichotomised and analysed using the chi-square statistic.

Of interest was the significant association of the PR PROGINS insert with migraine susceptibility. Statistical significance was reached in the first study population of 275 migraineurs compared to healthy controls, and also in the independent follow-up group of 300 migraineurs compared to controls. Furthermore, analysis of genotype risk groups showed that the PROGINS insert allele was significantly over-represented in migraineurs, and those who carried this allele were 1.8 times more likely to suffer migraine. We note that subgroup analysis showed dissimilar results in the MA/MO analyses in the two independent populations, however the small numbers in the MO subgroup would have reduced the statistical power, and may have contributed to this anomaly. Under the hypothesis of similar genetic etiology in the migraine subgroups, it would be expected that the PROGINS insert would confer a risk in both subgroups. Genotyping additional markers at this locus, particularly those that lie in a haplotype block with this locus, should provide further evidence for a role of this gene in migraine susceptibility. Furthermore, since it has been suggested that the PROGINS variant has an impact on gene expression, detailed functional analyses will be required to confirm that this variation plays a direct role in migraine susceptibility.

We have previously reported a significant association of the ESR 1 G594A SNP with migraine [13]. We now report a significant association of the PROGINS insert in the same population. As both genes play a complex, interrelated role in the CNS, and have been independently implicated in migraine susceptibility, we also analysed the impact on migraine risk of carrying susceptibility alleles for both genes. The results of our analysis showed that individuals who carried a copy of both PR and ESR 1 risk alleles were 3.2 times more likely to suffer from migraine, an effect that is greater than the independent effects of these genetic variants on disease susceptibility. These results present evidence for an interactive role of both variants in migraine susceptibility.

In conclusion, the results of our analyses provide evidence for a role of the PROGINS insert in the progesterone receptor gene in migraine susceptibility. This genetic variant has been suggested to result in anomalous transcription of the PR gene [39], therefore follow-up studies, including functional analyses, are clearly warranted to demonstrate causal variation at this locus. Furthermore, we have shown evidence that the PROGINS allele interacts synergistically with the ESR 1 594A allele to increase the risk of migraine by a factor of three.

Acknowledgements

This work was supported by funding from the National Health and Medical Research Council (NHMRC) of Australia and Griffith University. Dr Rod Lea is supported by an NHMRC C.J. Martin Fellowship. Experiments comply with the current laws in Australia.

Copyright information

© Springer-Verlag 2005