Human alpha1-acid glycoprotein (AGP) or orosomucoid (ORM) is a major acute phase protein that is thought to play a crucial role in maintaining homeostasis. Human AGP is the product of a cluster of at least two adjacent genes located on HSA chromosome 9. Using a range of restriction endonucleases we have investigated DNA variation at the locus encoding the AGP genes in a panel of healthy Caucasians. Polymorphisms were identified using BamHI, EcoRI, BglII, PvuII, HindIII, TaqI and MspI. Non-random associations were found between the BamHI, EcoRI, BglII RFLPs. The RFLPs detected with PvuII, TaqI and MspI were all located in exon 6 of both AGP genes. The duplication of an AGP gene was observed in 11% of the indiviuals studied and was in linkage disequilibrium with the TaqI RFLP. The identification and characterization of these polymorphisms will prove useful for other population and forensic studies.
Human alpha1-acid glycoprotein; RFLP; linkage disequilibrium
Identification and characterization of polymorphisms at the HSA a 1-acid glycoprotein (ORM*) gene locus in Caucasians
Catherine M. Owczarek1, Aleksander L. Owczarek2 and Philip G. Board 3
1Centre for Functional Genomics and Human Disease, Monash Institute of Reproduction and Development, Monash University, Melbourne, Australia.
2Department of Mathematics and Statistics, University of Melbourne, Parkville, Australia.
3Molecular Genetics Group, John Curtin School of Medical Research, Australia.
Send correspondence to Catherine M. Owczarek. Monash Institute of Reproduction and Development, Monash University, Melbourne, Victoria 3168, Australia. E-mail: email@example.com.
Human a1-acid glycoprotein (AGP) or orosomucoid (ORM) is a major acute phase protein that is thought to play a crucial role in maintaining homeostasis. Human AGP is the product of a cluster of at least two adjacent genes located on HSA chromosome 9. Using a range of restriction endonucleases we have investigated DNA variation at the locus encoding the AGP genes in a panel of healthy Caucasians. Polymorphisms were identified using BamHI, EcoRI, BglII, PvuII, HindIII, TaqI and MspI. Non-random associations were found between the BamHI, EcoRI, BglII RFLPs. The RFLPs detected with PvuII, TaqI and MspI were all located in exon 6 of both AGP genes. The duplication of an AGP gene was observed in 11% of the indiviuals studied and was in linkage disequilibrium with the TaqI RFLP. The identification and characterization of these polymorphisms will prove useful for other population and forensic studies.
Key words: Human a1-acid glycoprotein, RFLP, linkage disequilibrium.
Received: March 5, 2002; accepted: March 25, 2002.
HSA a1-acid glycoprotein (AGP, orosomucoid, ORM) is an abundantly expressed plasma protein whose levels rise dramatically during the acute phase response. A member of the lipocalin protein family, it is thought to function mainly as a transport protein for basic drugs although several other functions have been ascribed (Flower 1996). The expression of the ORM protein product in most individuals is controlled by two genes, AGP1 and AGP2 (Dente et al. 1987; Merritt and Board 1988), that are closely linked on HSA chromosome 9q31-q34.1 (Webb et al. 1987). A third gene, structurally identical to AGP2 has been reported to exist in some individuals (Dente et al. 1987), and duplication of the AGP1 gene has been demonstrated to occur in the Japanese population at an appreciable frequency (Nakamura et al. 2000). Considerable variation in the ORM polypeptide chain has been described. In addition to the two common alleles ORM1*F and ORM1*S (Johnson et al. 1969) a large number of variants have been identified in different populations (Yuasa et al. 1993).
In this study we have used RFLP analysis to investigate DNA variation at the AGP gene locus. We have demonstrated the existence of RFLPs in the region upstream of the AGP gene locus as well as polymorphisms within the AGP gene cluster and have examined linkage disequilibrium between these sites. A duplication of one of the AGP genes was observed in the population studied and was strongly linked with the presence of a TaqI polymorphism.
MATERIALS AND METHODS
Genomic DNA samples
The samples of DNA used for screening for RFLPs were from a set of 97 unrelated Caucasian blood donors recruited at the Canberra Red Cross Blood Transfusion Centre. An additional 20 random controls were obtained from healthy staff members at the John Curtin School of Medical Research. Family material was obtained from healthy Caucasian volunteers.
Genomic RFLP analysis
High molecular weight genomic DNA was extracted from the buffy coat from 10 mL peripheral blood (Grunebaum et al. 1984). Approximately 10 mg of genomic DNA was digested with the following enzymes according to the manufacturer's specifications: BamHI, EcoRI, BglII, PvuII, HindIII, MspI, and TaqI and electrophoresed through a 0.8% agarose gel. After Southern blotting of the DNA (Reed and Mann 1985) on to Gene-Screen Plus (Dupont) nylon membranes, the filters were hybridized overnight at 65 °C with an a 32P-dCTP labeled a1-AGP cDNA probe (Board et al. 1986).
HARDY-WEINBERG AND LINKAGE DISEQUILIBRIUM ANALYSIS
Standard c2 tests were used to compare observed genotype frequencies with those expected under the Hardy-Weinberg equilibrium (Weir 1996). In order to test for linkage disequilibrium between the alleles of the different polymorphisms, contingency tables were used, with standard c2 tests and Fisher's exact tests, producing identical results. When using Fisher's exact test we used a two-sided p-value that was the minimum of 1 and twice the one sided p-value. Because the data cannot distinguish the two possible double heterozygotes gametic frequencies could not be inferred (Weir 1996), and hence the standard test for linkage equilibrium using hapolotype frequencies was not possible. In such cases various alternative approaches are possible. However, because in all cases when the data was analyzed using 2x2 contingency tables between the two less common alleles, the results were extreme in one direction or the other the inferences of linkage disequilibrium/equilibrium were unequivocal.
Polymorphisms detected by BamHI, EcoRI and BglII
When human genomic DNA digested with BamHI was hybridized to the a1-AGP cDNA probe, two constant bands: 4.5 kb and 2.5 kb and two variable bands: 14.8 kb (B1) and 13.6 kb (B2) were observed (Figure 1a). Co-dominant segregation was observed in Families D and M for the B1 and B2 alleles of the BamHI RFLP (Figure 3). Mapping data (Merritt and Board 1988) indicated that these hybridizing bands corresponded to exons 1-5 of AGP2 and exon 6 of both AGP genes respectively placing the polymorphic BamHI site approximately 11 kb upstream of the AGP1 gene.
Digestion of human DNA with EcoRI detected a two-allele polymorphism with bands at 12.6 kb (E1) and 11.6 kb (E2) and invariant bands at 16.8 kb and 6.9 kb (Figure 1b). Co-dominant segregation of the EcoRI RFLP was demonstrated in Family D (Figure 4). The 6.9 kb invariant band contains exon 1 of AGP1 and exons 1-5 of AGP2 (Merritt and Board 1988). Hybridization with a probe specific to exon 1 of AGP1/2 (data not shown) indicated that the polymorphic EcoRI site was located upstream of AGP1.
BglII digestion also detected a two-allele polymorphism. Fragment lengths of 9.8 kb (Bg1) and 8.5 kb (Bg2) with invariant bands at 12.0 kb, 5.2 kb 0.8 kb and 0.7 kb were observed. Genetic transmission of the BglII RFLP was observed in Family D (Figure 3). The position of the polymorphic BglII site was determined from the nucleotide sequence (Merritt and Board 1988) and additional mapping experiments (data not shown). The 0.8 kb and 0.7 kb fragments corresponded to exons 2-3, and exons 4-5 respectively of both AGP genes. The 5.2 kb fragment contained exon 6 of AGP1 and exon 1 of AGP2 plus intergenic sequence. The polymorphic BglII band was detected with an exon 1-specific probe indicating that it was located upstream of the AGP1 gene.
Polymorphisms detected by TaqI, HindIII, PvuII and MspI
TaqI digestion generated a two allele polymorphism consisting of either a 3.02 kb band (T2) or a 2.88 kb band (T1) with invariant bands at 4.5 kb, 1.4 kb 1.2 kb 0.84 kb and 0.285 kb (Figure 2a). Co-dominant segregation for the TaqI RFLP was observed in two informative families (Figure 3). Hybridization experiments and analysis of sequence data (Merritt and Board 1988) indicated that the TaqI polymorphic site located on an exon-6 containing fragment.
When human genomic DNA was digested with HindIII and hybridized to the a1-AGP cDNA probe two bands at 4.6 kb (AGP1) and 6.9 kb (AGP2) were detected (Figure 2b) but the 6.9 kb band relative to the 4.6 kb band was more intense in 7 out of 65 individuals examined (11%) (Table I). The greater intensity of the 6.9 kb HindIII band relative to the 4.6 kb band in some individuals has been previously noted (Dente et al. 1985; Merritt and Board 1988) and correlates with an extra AGP gene (Dente et al. 1987; Nakamura et al. 2000). Individuals were scored either as 1-2 (AGP1-AGP2 on each chromosome) or 1-2-2' indicating the presence of an extra AGP gene on one or both chromosomes (since homozygotes and heterozygotes would be indistinguishable under the conditions used in this study). Co-dominant segregation was observed in Family M (Figure 3) where the father had the intense 6.9 kb HindIII band and the mother had 6.9 kb and 4.5 kb HindIII bands of equal intensity. Two siblings have inherited the intense 6.9 kb HindIII band and the other two have HindIII bands of equal intensity indicating that the father was heterozygous for the presence of a third AGP gene and the mother was homozygous for the more common two AGP gene arrangement. Family M had the same pattern of inheritance for the TaqI RFLP.
A complex polymorphism was detected in human genomic DNA digested with PvuII (Figure 2c). A total of four bands of different intensities were detected. Alleles P1 and P2 were defined by 1.8 kb and 1.6 kb bands respectively and invariant bands were observed at 1.46 kb, 1.38 kb and 0.69 kb. Hybridization experiments (data not shown) indicated that the polymorphic PvuII site was present in fragments containing exon 6 of either AGP gene. Because of the duplicated and sometimes triplicated genes in the AGP gene locus it was not possible to determine exactly which AGP gene contained the polymorphic allele/s or what the allelic distribution was in a given individual. In the case of equal intensities of bands P1 and P2 could be the consequence of several possible arrangements of the P1 and P2 alleles e.g. P1 could arise from AGP1 and P2 from AGP2 on one chromosome (or vice versa). The same number of P1 and P2 alleles will be present on the other chromosome but the arrangement could either be reversed or the same. Five phenotypic classes were therefore assigned based on the intensity of each allele (P1, P2) relative to each other on an autoradiogram (Figure 2c).
Co-dominant segregation was observed in Families R and M (Figure 3). In Family R, the father was 2P1 2P2 and the mother P1 3P2. One child was 2P1 2P2 indicating that a P1 and a P2 allele are co-segregating in the father and two P2 alleles are co-segregating in the mother. The two other children in this family were 2P1 2P2. Family M demonstrated Mendelian inheritance of a different phenotypic class where the father was 3P1 P2 and the mother was 2P1 2P2. The phenotypic class of the offspring indicated that a P1 and a P2 allele must be co-segregating in the mother whilst the father is P1/P1 on one chromosome and P1/P2 on the other.
Digestion of human genomic DNA with MspI and hybridization to the a1-AGP cDNA probe also resulted in a complex band pattern. The variant fragments were designated M1 to M5 in order of decreasing size: 4.4 kb (M1), 4.3 kb (M2), 3.2 kb (M3), 2.9 kb (M4), and 2.8 kb (M5). The polymorphic fragments all hybridized to an exon-6 specific probe. A total of 47 random individuals were screened and several different arrangements of the variant MspI fragments were observed. Some representative combinations are presented in Figure 2d. The various alleles that these fragments represent were not examined for deviation from Hardy-Weinberg equilibrium since extensive family studies would be necessary in order to determine the correct number of alleles in a particular individual. However, a segregation pattern consistent with Mendelian inheritance was observed in Families R and M (Figure 3). In Family R each allele were interpreted as being alleles at a separate locus. The father had bands M1, M3 and the mother had bands M1 and M2. These segregated independently in each parent to give siblings with allele distributions of M1 M3, M1 M1, and M3 M2. The pattern of inheritance in Family M was more complex. Fragment M1 in the father segregated independently from bands M2, M3, M4, and M5. The mother was homozygous for M1. Two of the siblings have inherited an M1 band from either parent. The remaining two siblings derived an M1 band from their mother and the M2, M3, M4 and M5 bands from their father. The bands appeared to be transmitted as a single allele although they most likely represented multiple closely linked sites on a single chromosome. Interestingly, the M2, M3, M4, M5 band arrangement had the same pattern of inheritance as the HindIII and TaqI polymorphisms in this family.
The allele frequencies for the BamHI, EcoRI, BglII, TaqI, PvuII RFLPs and occurrence of an extra AGP gene in a sample of Caucasians are given in Table I. The observed frequency distribution of genotypes for the BamHI, EcoRI, BglII and TaqI RFLPs did not differ significantly from those expected on the basis of a Hardy-Weinberg equilibrium (Table I). As mentioned above, tests were carried out using the standard c2 statistic. As each RFLP was concordant with Hardy-Weinberg equilibrium using that statistic the more conservative Fisher's exact test was unnecessary (despite some small expected counts).
The distribution of genotypes for the PvuII RFLP was examined for deviation from "Hardy-Weinberg equilibrium" under the assumption that the probability of the occurrence of similar alleles at each of the polymorphic sites is identical. In this case the null hypothesis of Hardy-Weinberg equilibrium (Table I) amongst the alleles at the two loci is rejected on the basis of a c2 statistic using a standard significance level of 0.05. It must be remembered that this disequilibrium could be due to one or more causes. Firstly, one or more of the loci separately may be out of Hardy-Weinberg equilibrium, or secondly, that the probability of the occurrence of similar alleles at each of the polymorphic sites is not identical, or thirdly, that the two loci are in linkage disequilibrium. The experimental data does not allow reasonable discrimination between these alternatives since the two loci cannot be distinguished.
ANALYSIS OF LINKAGE DISEQUILIBRIUM BETWEEN POLYMORPHIC SITES
The distributions of the alleles of the polymorphic loci were analyzed for linkage disequilibrium. The null hypothesis that each of the polymorphic sites were in linkage equilibrium was tested by both the c2 statistic and using Fisher's exact test on 2x2 contingency tables. The 2x2 tables were constructed from the 3x3 tables given in Table II by simply summing the second and third columns and rows respectively. For example, the BamHI and BglII 2x2 contingency table had entries 40 and 1 (being the sum of 1 and 0) in the first row, and 1 (being the sum of 1 and 0) and 10 (being the sum of 9, 0, 1 and 0) in its second row. This 2x2 table then gives the counts of the combinations of the absence and presence of the less-common allele of the two RFLPs in its entries. This approach was necessary because of the very low expected counts in the third columns and rows, due to the small frequencies of the homozygotes of the less common allele. P-values from the c2 statistic with 1 degree of freedom and Fisher's exact test are both given. In all cases the two tests lead to the same conclusions. This is because when linkage equilibrium is observed using the c2 statistic Fisher's exact test, being a more conservative test, naturally leads to the same conclusion, and where linkage disequilibrium is inferred the results are so extreme, with p-values of less than 10-6 that no ambiguity occurs. Linkage disequilibrium was observed between the following RFLPs: BamHI and EcoRI; BamHI and BglII; BglI and EcoRI (Table II).
In order to test for linkage disequilibrium between the RFLP genotypes and the presence of multiple AGP genes (as detected by a 6.9 kb HindIII band that was more intense relative to the 4.5 kb HindIII band) 2x2 contingency tables were constructed from the 3x2 tables given in Table III in an analogous fashion to that described above. These 2x2 tables were then tested for association using, as above, the c2 statistic and Fisher's exact test. Linkage disequilibrium was observed between the presence of the relatively intense 6.9 kb HindIII band (1-2-2') and hence extra AGP gene(s), and the TaqI RFLP (Table III).
Since the initial studies (of Johnson et al. 1969) there have been many reports of genetic variation at the human a1-acid glycoprotein or ORM locus. In this study we have investigated DNA variation in and around the human AGP genes in a Caucasian population using RFLP analysis. RFLPs were detected with the use of restriction enzymes BamHI, EcoRI, BglII, PvuII, HindIII, MspI, and TaqI.
RFLPs detected by enzymes BamHI, EcoRI, BglII, were located at least 11 kb upstream of the AGP gene cluster and were in linkage disequilibrium with each other. This group of polymorphic loci did not deviate from a random association with the TaqI RFLP that was located within the AGP gene cluster. Two complex polymorphisms were detected within non-coding regions of the AGP gene cluster using PvuII and MspI. Interestingly, the PvuII, MspI, and TaqI polymorphisms could be detected with an exon 6-specific probe indicating a higher degree of recombination in this region of the AGP genes.
Previous studies have suggested that the two AGP gene array seen in the majority of individuals arose as a consequence of gene duplication subsequent to the divergence of humans from rodents (Merritt et al. 1990). Individuals containing three AGP genes have been reported (Dente et al. 1987; Merritt and Board 1988; Nakamura et al. 2000). These three gene arrays (AGP1-AGP2-AGP2 or AGP1-AGP1-AGP2) represent polymorphisms in the populations studied and are the result of further crossover events that must have occurred relatively recently since there were no changes in the duplicated genes studied. In the Caucasian population studied here linkage disequilibrium was observed between the presence of an intense 6.9 kb HindIII fragment (1-2-2' and hence multiple AGP1 or AGP2 genes) and the TaqI RFLP. The simplest explanation for the origin of the TaqI polymorphism would be a point mutation that caused the loss of the TaqI site in the region between the AGP1 and AGP2 genes. However, if one considers that unequal crossing over events generated three member AGP gene arrays (Dente et al. 1987; Merritt and Board 1988; Merritt et al. 1990; Nakamura et al. 2000) it is possible that a crossover event could cause the loss of a TaqI site. In this study individuals who may be homozygous for the presence of a third AGP gene would be indistinguishable from heterozygotes since increased intensity of the 6.9 kb band relative to the 4.5 kb was used as the basis for scoring. Interestingly, however, Family M, (Figure 3.) was informative for both the TaqI polymorphism and the presence of an extra AGP gene and those individuals that were heterozygous for the TaqI polymorphism were also heterozygous for an extra AGP gene. Furthermore, in the population studied there were no T2T2 individuals suggesting that there were no individuals homozygous for an extra AGP gene. Further sequence analysis of the AGP locus from the individuals studied would be required to confirm the genetic basis for observed linkage between the TaqI polymorphism and the presence of multiple AGP genes and to determine if the particular duplicated gene was AGP1 or AGP2.
The HSA orosomucoid polymorphisms (Yuasa et al. 1993; Yuasa et al. 1997) have been widely studied in a range of populations. The results presented in this survey provide evidence for further variation at the AGP gene locus and the polymorphisms described may be potentially useful as genetic markers in a variety of forensic, linkage and population studies.
- Board PG, Jones IM and Bentley AK (1986) Molecular cloning and nucleotide sequence of human a1-acid glycoprotein cDNA. Gene 44:127-131.
- Dente L, Ciliberto G and Cortese R (1985) Structure of the human a1-acid glycoprotein gene: Sequence homology with other acute phase proteins. Nucl. Acid Res. 13:3941-3952.
- Dente L, Pizza MG, Metspalu A and Cortese R (1987) Structure and expression of the genes coding for human a1-acid glycoprotein. EMBO J 6:2289-2296.
- Flower DR (1996) The lipocalin protein family: structure and function. Biochem J 318:1-14.
- Grunebaum L, Casenave J-P, Camerino G, Kloepfer C, Mandel J-L, Tolstoshev P, Jaye M, De la Salle H and Lecocq J-P (1984) Carrier detection of hemophilia B by using a restriction site polymorphism associated with the coagulation factor IX gene. J Clin Invest 73:1491-1495.
- Johnson AM, Schmid K and Alper CA (1969) Inheritance of human a1-acid glycoprotein (orosomucoid) variants. J Clin Invest 48:2293-2299.
- Merritt CM and Board PG (1988) Structure and characterization of a duplicated human a1-acid glycoprotein gene. Gene 66:97-106.
- Merritt CM, Easteal S and Board PG (1990) Evolution of human a1-acid glycoprotein genes and surrounding Alu repeats. Genomics 6:659-665.
- Nakamura H, Yuasa I, Umetsu K, Nakagawa M, Nanba E and Kimura K (2000) The rearrangement of the human a1-acid glycoprotein/Orosomucoid gene: Evidence for tandemly triplicated genes consisting of two AGP1 and one AGP2. Biochem Biophys.Res Commun 276:779-784.
- Reed KC and Mann DA (1985) Rapid transfer of DNA from agarose gels to nylon membranes. Nucl Acids Res 13:7202-7221.
- Webb GC, Earle EA, Merritt CM and Board PG (1987) Localization of human a1-acid glycoprotein genes to 9q31-34.1. Cytogenet Cell Genet 47:18-21.
- Weir BS (1996) Genetic Data Analysis II, 2nd ed. Sinauer Associates, Sunderland, MA.
- Yuasa I, Weidinger S, Umetsu K, Suenaga K, Ishimoto G, Eap BC, Duche J-C and Baumann P (1993) Orosomucoid System: 17 additional orosomucoid variants and proposal for a new nomenclature. Vox Sang 64:47-55.
- Yuasa I, Umetsu K, Vogt U, Nakamura H, Nanba E, Tamaki N and Irizawa Y (1997). Human orosomucoid polymophis: molecular basis of the three common ORM1 alleles ORM1*F1, ORM1*F2, and ORM1*S. Hum Genet 99:393-398.
Publication in this collection
02 Aug 2002
Date of issue
25 Mar 2002
05 Mar 2002