Polymorphisms in the growth hormone receptor: A case-control study in breast cancer
The human growth hormone receptor (GHR) mediates the effects of growth hormone (GH1), starting a signalling cascade that is involved in the regulation of proliferation, differentiation and apoptosis. Recently, an isoform of the GHR gene lacking exon 3 (GHRd3) was associated with accelerated responsiveness to growth hormone. In this study, we investigated the association of the GHRd3 polymorphism with breast cancer risk and performed a haplotype analysis with 3 additional single nucleotide polymorphisms (SNPs) (Gly186Gly, Cys440Phe and Ile544Leu) in the GHR coding region in a Polish cohort. We did not observe any effect of the 4 polymorphisms on breast cancer risk. Neither did the 3 most common haplotypes influence breast cancer risk. However, a rare haplotype (dGGC), containing the GHRd3 allele, was associated with a decreased breast cancer risk (OR 0.30, 95% CI 0.11–0.80). © 2005 Wiley-Liss, Inc.
Growth hormone (GH1) and its main effector, Insulin-like growth factor-1 (IGF-1), are mitogens in the breast, which act via their receptors and not only stimulate the normal mammary development but also play an important role in the development of breast cancer. Overexpression of these growth factors leads to an enhanced breast cancer risk.1, 2 Recent meta-analyses have confirmed that an increased IGF-1 level is a risk factor for breast cancer.3, 4 Besides its action as a pituitary hormone, GH1 is also locally produced in mammary epithelial cells where it can act in an autocrine/paracrine manner. Recent data show that autocrine GH1 production results in hyperproliferation of mammary carcinoma cells and enhanced transcriptional activation mediated by growth hormone receptor (GHR).5 In breast tissue, GHR is expressed in both epithelial and stromal cells, with a greater expression in the tumor than in the normal tissue.2, 6, 7
The human GHR consists of 9 coding exons out of which exons 3–7 encode the extracellular domain. A deletion polymorphism that results from homologous recombination skipping 22 residues encoded by exon 3 of the GHR gene has been reported (GHRd3).8, 9 The omission of the exon is due to a deletion between 2 retroviral sequences flanking exon 3 that mimic alternative splicing. The region is not highly conserved in different mammalian species and the GHRd3 allele is specific to humans.9 However, it has been hypothesized that a loss or retention of exon 3 could affect the receptor function. Children born small for gestational age and children with idiopathic short stature carrying at least 1 GHRd3 allele have shown a greater response to growth hormone administration and a higher growth acceleration.10 According to an in vitro analysis, GHRd3 containing cells induce a higher luciferase reporter activity when stimulated by GH1 than cells containing full length GHR. However, the binding affinity of GH1 to each receptor isoform was similar. A possible explanation could be that the removal of a part of the extracellular domain propagates subtle conformational changes that facilitate hormone-triggered activation of the receptor.10 Alternatively, the deletion may be in linkage disequilibrium (LD) with a functional polymorphism in the GHR gene or a near-by locus. Although the better GH1 response in this study was due to an exogenous GH1, a local autocrine production of GH1 in the mammary tissue might lead to a similar growth promoting effect through the GHRd3 isoform.
In this study, we investigated the GHRd3 polymorphism for an association with breast cancer risk. Additionally, we screened the coding region of the GHR gene for published polymorphisms and included 3 single nucleotide polymorphisms (SNPs) (Gly186Gly, Cys440Phe and Ile544Leu) for a haplotype analysis, to further characterize the importance of the GHR locus in breast cancer risk.
Material and methods
A case-control study was carried out using a Polish cohort enriched with familial and early age breast cancer cases. The cohort consisted of 350 breast cancer cases (mean age 45 years, range 29–67) and 530 regionally and ethnically matched healthy controls (mean age 41 years, range 18–79). The inclusion criteria for the cases were (i) at least 2 first-degree relatives with breast and/or ovarian cancer, regardless of age; (ii) breast cancer diagnosed below the age of 35 without family history; (iii) bilateral breast cancer, regardless of the family history; (iv) breast and ovarian cancer diagnosed in 1 patient, regardless of the family history and (v) breast cancer diagnosed below 50 years of age, regardless of family history.11, 12 The subjects corresponding to criteria (i)–(iv), 220 cases, were collected during the years 1997 to 2002 by the Chemotherapy Clinics and the Genetic Counselling Service (Gliwice, Poland) and the subjects corresponding to criterion (v), 130 cases, were collected between December 2002 and March 2004 by the Surgery Clinics (Gliwice, Poland). No information about the number of cases belonging to each category (i)–(iv) was available. All cases were unrelated. They were tested for 4 founder mutations in BRCA1 and 2 in BRCA2 and were found to be negative. Five of the mutations were frameshift mutations that were detected using Allele Specific Amplification PCR with 3 primers out of which the third one binds only the mutated allele. The sixth mutation was a base pair substitution and restriction fragment length polymorphism was used for detection. These mutations account for more than 90% of the BRCA1/2 mutations in the Polish population.13
About 90% of the patients and the controls approved the participation to the study and provided a blood sample. The study was approved by the ethical committee of the University of Heidelberg.
Analysis of the GHRd3 polymorphism
To detect the 2.7 kb deletion polymorphism including exon 3, we used a simple multiplex PCR assay using primers G1, G2 and G3 [GenBank: AF155912] as described earlier by Pantel et al.8 We used 5 ng of genomic DNA in a 10 μl PCR reaction using 1× PCR buffer (Invitrogen, Paisley, UK), 1.5 mM MgCl2 (Invitrogen), 0.11 μm dNTP mixture (Invitrogen), 0.15 μM of each primer and 0.3 U platinum Taq polymerase (Invitrogen). The PCR programme was as follows: 94°C for 5 min, 35 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 90 s and a final extension at 72°C for 7 min. PCR was performed in a GeneAmp PCR System 9700 thermocycler (Applied Biosystems, Foster City). We analyzed the amplification products by electrophoresis on a 1% agarose gel stained with ethidium bromide. We confirmed about 10% of the results by direct sequencing as described earlier.14
Genotyping using Taqman assays
The polymorphisms Gly186Gly, Cys440Phe and Ile544Leu were investigated using the allelic discrimination method. Taqman assays were ordered from Applied Biosystems as Assays-on-Demand (C_264405_10, C2389455_20 and C2841422_10, respectively). The reaction was performed in 5 μl using 5 ng of genomic DNA, 2.7 μl 2×Taq-Man Universal Master Mix (Applied Biosystems) and 0.03 μl 20× Assay-Mix per reaction. PCR was performed at 50°C for 2 min, 95°C for 10 min and 35–45 cycles at 92°C for 15 s and 60°C for 1 min. PCR was performed in a GeneAmp PCR System 9700 thermocycler and the number of cycles was dependent on the genotype clustering. The samples were read and analyzed in an ABI Prism 7900HT sequence detection system using SDS 1.2 software (Applied Biosystems). We confirmed ˜10% of the results by direct DNA sequencing as described earlier.14
The observed genotype frequencies in the breast cancer cases and controls were tested for Hardy–Weinberg equilibrium (HWE) and the difference between the observed and expected frequencies was tested for significance using the χ2 test. The distribution of the genotypes of the polymorphisms studied followed HWE. Statistical significance for the difference in the genotype and haplotype frequencies between the breast cancer cases and controls were determined by the χ2 test. Whenever the expected number of cases was smaller than 5, Fisher's exact test was used. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated for associations between genotypes as well as haplotypes and breast cancer. The haplotype analyses were carried out using the SNPHAP program created by David Clayton, which can be downloaded from the internet (www.gene.cimr.cam.ac.uk/clayton). If genotype data for one or more polymorphisms was not available, the sample was excluded from the haplotype analysis. LD between the SNPs was evaluated using the Haploview program (http://www. broad.mit.edu/mpg/haploview/documentation.php). Haplotype effects were estimated by logistic regression analysis using Statistical Analysis Software (Version 9.1.3, SAS Institute, Cary, NY).
No effect of the GHRd3 polymorphism on breast cancer risk
We investigated the GHRd3 polymorphism in a cohort of 350 Polish breast cancer cases and 530 matched controls. The number of cases and controls in Tables I and II may differ between different analyses because of unsuccessful amplification of a few samples. The allele and genotype frequencies were in concordance with other Caucasian populations.8, 10 We did not observe any appreciable difference in the genotype distribution between the breast cancer cases and controls (Table I). Adjustment for age did not change the results.
Table I. Genotype and Allele Distribution of the Polymorphisms in the GHR Gene in Polish Breast Cancer Patients and Controls
| wt/wt||189 (0.55)||272 (0.52)||1|| |
| wt/del||130 (0.38)||221 (0.42)||0.85 (0.64–1.13)||0.25|
| del/del||26 (0.08)||33 (0.06)||1.13 (0.66–1.96)||0.65|
| del%||26.4||27.3|| || |
| GG||186 (0.53)||269 (0.52)||1|| |
| GA||138 (0.39)||212 (0.41)||0.94 (0.70–1.26)||0.68|
| AA||26 (0.07)||39 (0.08)||0.96 (0.57–1.64)||0.89|
| A%||27.1||27.9|| || |
| GG||345 (0.99)||518 (0.98)||1|| |
| GT||4 (0.01)||11 (0.02)||0.55 (0.17–1.73)||0.30|
| TT||0 (0.00)||0 (0.00)||–|| |
| T%||0.57||1.04|| || |
| AA||98 (0.28)||152 (0.29)||1|| |
| AC||175 (0.50)||269 (0.51)||1.01 (0.74–1.39)||0.96|
| CC||76 (0.22)||106 (0.20)||1.11 (0.75–1.64)||0.59|
| C%||46.8||45.6|| || |
No effect of GHR coding SNPs on breast cancer risk
To search for SNPs with a potential influence on GHR properties and to perform a haplotype analysis, we screened the coding region of the GHR gene for the so far reported SNPs (according to the NCBI dbSNP database) in a small sample set of 23 breast cancer samples. We confirmed the existence of 5 SNPs and choose 3 of them (Gyl86Gly, Cys440Phe and Ile544Leu) for an investigation in a larger cohort. The Cys440Phe polymorphism was in 100% LD with the Pro579Thr SNP and therefore, we only investigated the Cys440Phe SNP. The Ser491Ser SNP was rare (allele frequency ˜2%) and we did not proceed with it. The allele frequencies were concordant with the ones published by NCBI except for the Cys440Phe polymorphism with a reported frequency of 6% in a multinational cohort (NCBI database). In our cohort, this SNP was very rare with a variant allele frequency of ˜1%. We did not detect any differences in the allele and genotype distributions of the studied SNPs (Table I). Adjustment for age did not change the results.
We performed a haplotype analysis with the 4 polymorphisms using logistic regression analysis (Table II). We found 3 common haplotypes, while the remaining 5 had a frequency of less than 4%. The overall P-value for a haplotype effect was 0.22, suggesting that the haplotypes do not affect breast cancer susceptibility. None of the common haplotypes (GGGC, GGGA and dAGA) showed any effect on breast cancer risk. The haplotype dGGC was associated with a significantly decreased breast cancer risk. However, only 5 cases and 23 control individuals carried this haplotype. As already observed in our small sequenced sample set, the SNPs were tightly linked. We observed a LD of 87% and 86% between the GHRd3 polymorphism and Gly186Gly and Ile544Leu, respectively. The LD between Gly186Gly and Ile544Leu was 98%.
Table II. Haplotype Distribution and Their Association with Breast Cancer Risk
|GGGC||313 (0.45)||431 (0.42)||1|
|GGGA||173 (0.25)||273 (0.27)||0.87 (0.69–1.11)|
|dAGA1||165 (0.24)||241 (0.24)||0.94 (0.74–1.21)|
|GAGA||20 (0.03)||32 (0.03)||0.86 (0.48–1.53)|
|dGGA1||8 (0.01)||6 (0.01)||1.84 (0.63–5.34)|
|dGGC1||5 (0.01)||23 (0.02)||0.30 (0.11–0.80)|
|dATA1||4 (0.01)||11 (0.01)||0.50 (0.16–1.59)|
|GAGC||0 (0.00)||3 (0.01)||–|
It has been shown earlier that exogenous GH1 administration results in a greater growth response in GHRd3 carriers, while endocrine GH1 does not have any significant effect on growth acceleration.10 We hypothesized that autocrine production of GH1 by local tissues may influence the transduction of growth hormone signalling in the GHRd3 carriers in a similar way like the exogenous administration. As has been shown, autocrine GH1 production results in a hyperproliferation of mammary carcinoma cells and enhanced transcriptional activation mediated by GHR.5
However, we did not observe any effect of the GHRd3 polymorphism on breast cancer risk in the present study. We had a power of 80% at a significance level of 0.05 to detect an OR of 1.5. We used a cohort enriched for familial cases and early age cases because it has been shown earlier that selection of cases based on the family history of the same disease increases the power to detect low-penetrance variants.15, 16 There can be several explanations for our negative finding. The deleted region is not located close to any binding motifs and the ligand binding capacity is not altered by the deletion.9 Nor seems the intramolecular disulfide bond pattern be affected by the loss of exon 3, suggesting no severe alteration in the protein folding.10 Nevertheless, the newly generated junction between exons 2 and 4 causes a loss of an asparagine-mediated N-glycosylation site that may affect glycosylation of the extracellular part of the receptor.17 GH1 has also an affinity to bind to the related prolactin receptor (PRLR) and in mice increased activation of PRLR, and not GHR, leads to tumour formation.18 The 2 receptors have a high homology except that in PRLR the exon 3 region is absent.19 In an early study, it was suggested that the deletion of exon 3 would increase the degree of similarity of GHR to PRLR and that this might have a functional effect on GH1 signalling.20
We also investigated polymorphisms Gly186Gly, Cys440Phe and Ile544Leu for their contribution to the risk of breast cancer. The Cys440Phe SNP was in 100% LD with the Pro579Thr SNP. Both of these amino acid substitutions were predicted to be possibly damaging according to the PolyPhen database (http://tux.embl-heidelberg.de/ramensky/polyphen.cgi). However, the carrier frequency observed in our study was very low. The Gly186Gly SNP is, according to the UniProtKB/Swiss-Prot database [Swiss-Prot:P10912], located at a turn position in the protein structure but the SNP does not change the amino acid. The Ile544Leu SNPs is conserved in different species but because of the minor amino acid exchange no effect on the protein function would be expected. We did not observe any effect on breast cancer risk by any of the studied SNPs. Logistic regression analysis revealed an overall P-value of 0.22 for a haplotype effect, suggesting that the haplotypes do not have a major effect on breast cancer. However, we found 1 rare haplotype, containing the GHRd3 allele, to be associated with a protective effect on breast cancer risk (dGGC: OR 0.30, 95% CI 0.11–0.80). Further studies are required to show whether the haplotype dGGC has any effect on GHR properties and further to the risk of breast cancer, or whether the haplotype identifies a founder effect on the gene or a near-by locus, which interferes with the function of GH1.
To our knowledge, this first breast cancer study on polymorphisms in the GHR gene detected no associations between the 4 polymorphisms (GHRd3, Gly186Gly, Cys440Phe and Ile544Leu) and breast cancer risk. However, in a haplotype analysis with these SNPs, the rare haplotype dGGC showed a slightly protective effect.
This study was supported by the grants from State Committee for Scientific Research (PBZ-KBN-040/P04/2001 and 3P05C 05825 to E.G) and a grant from EU (LSHC-CT-2004-503465 to E.G and K.H).