- Email: [email protected]
Pedigree analysis of the 5′ flanking region of the insulin gene in familial diabetes mellitus
Pedigree
Analysis
of the 5’ Flanking Region of the Insulin Gene in Familial Diabetes Mellitus
Adrian S. Dobs, John A. Phillips III, Richard L. Mallonee, Non-insulin-dependent the 5’ flanking phenotype from
of NIDDM
7 pedigrees
mellitus (NIDDM)
has been reported
insulin gene. We
by studying multiple pedigrees.
(5 white,
insulin-dependent digestion
diabetes
region of the human
2 black) ranging
diabetes mellitus (IDDMI,
with Sst I, and electrophoresis.
Christopher
to be associated with an insertion
have attempted
We evaluated
D. Saudek, and Robert L. Ney
to examine
NIDDM
insertion
polymorphism.
(P = 0.98). When
(P = 0.58). In examining
and 80 were nondiabetics
Of these,
52 subjects
the DNA was blotted to nitrocellulose
sibships were analyzed the 7.6 kb allele occurred
in 16 of 36 ND, 1 of 5 IDDM,
filters and hybridized
insertion phenotype
polymorphism
flanking
the
insulin
previously
and 10 of 18
in 6 of 13 ND. 3 of 5 IDDM, and 8 of 12 NIDDM the 7.6 kb allele was documented
(P = 0.59). In these individuals and in the multiple families studied the
gene
showed
Mendelian
inheritance
and assorted
independently
of the
of diabetes mellitus.
Q 1986 by Grune
& Stratton,
inc.
R
ECENT characterization of the human insulin geneId has enabled detailed evaluation of how this gene may contribute to the etiology of diabetes mellitus (DM). Several laboratories have reported a polymorphic region of heterogeneous length occurring about 500 base pairs (bp) 5’ to the human insulin gene origin of mRNA transcription (Fig 1).7-9 The genomic DNA fragment corresponding to this region has been sequenced and found to primarily contain a 14-l 5 bp repetitive sequence rich in guanine and cytosine oligoare due to nucleotides.7.9 The length polymorphisms insertions of varying sizes. Bell et al” categorized these lengths into Class I alleles of 570 bp, Class II alleles of 1320 bp, and class III of 2470 bp in size. The allelic frequencies vary among ethnic groups with Blacks often having a genotype that includes the rare Class II alleles. Population studies of unrelated individuals have suggested that the frequency of these alleles may vary among phenotypes. Rotwein et al”,‘2 and Owerbach et alI3 have reported frequencies of the insertion Class III alleles significantly higher in non-insulin-dependent diabetes mellitus (NIDDM) than in insulin-dependent diabetes mellitus (IDDM) or nondiabetic (ND) people. Bell et al” observed a higher frequency of the short alleles (Class I) in Caucasian IDDM and N IDDM groups compared to ND controls. The level of significance for the NIDDM was, however, only marginal. Their analysis of composite data revealed no association between the presence or absence of Class I or CIass III inserts and NIDDM. Owerbach et al” have subsequently stated that the large insertions are primarily seen in individuals with atherosclerosis and may only secondarily be associated with diabetes. These studies employed unrelated individuals to derive allelic frequency data. Since families would Metabolism,
10 had
to a o?P-labeled
to the common
in 32 of 57 ND, 3 of 5 IDDM,
subjects. including 12 spouses from the pedigrees,
and 16 of 31 NIDDM
had NIDDM,
(ND). DNA was extracted from leukocytes and after
In these families, the 7.6 kb allele occurred
72 unrelated
in
to the
142 individuals (120 white, 22 black), 80 of whom were
in size from 4 to 37 members.
insulin gene probe. Two alleles of 6.0 and 7.6 kb in size were detected, the latter corresponding described
polymorphism
linkage of this polymorphism
Vol35, No
1
(January),
1986:
pp
13- 17
theoretically have identical mutations as the cause of their disease, pedigree analysis is a superior method to derive information of linkage phenomena. Owerbach et alI5 and Hitman et alI6 studied only one family each. The latter group revealed no linkage between the size of DNA inserts and diabetes. This paper reports experiments done not only on unrelated individuals but individuals belonging to multiple pedigrees. We are the first to report 7 pedigrees in order to examine for linkage between the presence of the polymorphic insert and the phenotype of DM. MATERIALS
AND
METHODS
Study Population We analyzed a total of 142 patients (120 white, 22 black) of whom 52 subjects had NIDDM, 10 IDDM, and 80 were ND (Table 1). Eighty of these patients were from 7 pedigrees (pedigree B and F black; A, C, D, E Caucasian) ranging in size from 4 to 37 members each with 2 to 7 diabetics per kindred. Of the ND, 25 had documented normal oral glucose tolerance tests and 37 had randomly determined plasma glucose values of less than 110 mg/dL. The remaining 18 were family members who had no history of DM. The diagnosis of IDDM was established by a clinical history of ketosis.” The diagnosis of NIDDM was made if there existed persistent hyperglycemia and the absence of ketosis.” All patients with diabetes had been diagnosed at least one year prior to study.
From the Departments of Medicine and Pediatrics, The Johns Hopkins Universiiy Hospital, Baltimore. Presented in part at the American Federation of Clinical Research, 1984. Washington, DC, and appeared in abstract form in Clinical Research 32:393A. Supported in part by project 52T32AM07109, National Institutes osHealth. Address reprint requests to Dr A.S. Dobs, Traylor Building 720, The Johns Hopkins Hospital, 720 Rutland Ave. Baltimore, MD, 21205. 0 1986 by Grune & Stratton, Inc. 0026&0495/86/3501-0003$03.00/0
13
DOBS ET AL
14
3.2 kb
6kb
-5
-4
-3
-2
-I
Statistical analysis The association of the insertional allele to the diagnosis of DM was tested using a contingency table in which the presence of one or more insertional alleles was compared to the diagnosis of DM. A chi-square test of independence was then performed.*’ In addition, the 95% confidence limit and odds ratio were calculated as described previously.22~23 Using the formula (a + S) (d + .5)/(b + .5) (c + .05), a is the number of diabetics with inserts, b the number of diabetics without the insert, c nondiabetics with the insert, and d nondiabetics without inserts. There is considered to be no relation between the variables if the 95% confidence limit includes one. A Fisher’s exact test was done for blacks because n c 40.”
m pgH IG900bp I
0
2
3
4
Fig 1. Endonuclease Restriction Map. Partial map of the human insulin gene modified from Bell et al’ and Rotwein et al” including the exons in the black areas and intervening sequences in the open areas. The variable region in the 5’ flanking area is designated by the dashed triangle. The dotted bar indicates the genomic insulin gene probe used. The open bar indicates the 6 kb allele observed when genomic DNA is digested with Sst I in the absence of an insertion.
DNA was isolated from the nuclei of leukocytes contained in 20 mL of peripheral blood collected with an EDTA anticoagulant by a procedure similar to Bell et al* The yield of DNA from fresh whole blood was 20-60 ugDNA/mL.
Analysis of the Insulin Gene Five pclgaliquots of DNA were digested with Ssr I according to the manufacturer’s specifications (Bethesda Research Laboratories, Rockville, Md). The DNA fragments were separated by electrophoresis at 15 mv constant current for 16 hours, in 0.9% (wt/vol) agarose gels prepared in 0.04M Tris acetate(pH 8.0)/0.025 moI/L sodium acetate/O.002 mol/L EDTA. The DNA fragments were then transferred from the agarose gels to nitrocellulose filters as described by Southern.” The filters were baked at 70 OC for 6 hours and prehybridized and hybridized as previously described.” The DNA fragment used,as the hybridization probe was a 900 bp Pst I fragment of the genomic human insulin gene (Fig 1) cloned into the plasmid pBR322 and grown in Escherichiu coli HBlOl cells.6 Following excision and isolation of the insert from the vector, the insert was labeled with cu’*P ATP and a’*P CTP by “nick translation” with a specific activity of 224 x IO* counts/min/ug.2”
Table 1. Characteristics
analysis of the insulin gene
Autoradiogram patterns of DNA from 142 individuals after restriction endonuclease digestion with Sst I and hybridization with the pgH1 probe revealed two size groups of alleles. One was a 6.0 kb (-) allele and the other was a 7.6 kb (+) allele (Fig 2). This size fragment would correspond to a class III allele.
ND
IDDM
NIDDM
Total
80
10
52
142
Sex: male
49
7
19
75
female
31
3
33
67
Race: white
79
9
32
120
1
Agex+SEMy
40 * 22
range y
l-74
Seven kindreds were studied to determine if the polymorphic flanking region and the phenotype of DM were genetically linked (Fig 3). The 7.6 kb allele occurred in 32 of 57 ND(56%), 3 of 5 IDDM (60%), and 10 of 18 NIDDM (55%), (P = 0.98). Comparing within sibships, the 7.6 kb allele occurred in 6 of 13 ND (46%), 3 of 5 IDDM (60%), and 8 of 12 NIDDM sibs (66%), (P = 0.58). In these families there were 4 adults (mean = 52, range 35-66 years) who were homozygous for the 7.6 kb fragment. Of these, 3 had oral glucose tolerance tests. One test was normal, while the other 2 were impaired having a normal fasting blood sugar and returning to 98 and 145mg% at 2 hoursI In addition there were 11 NIDDM (mean = 52, range 35-74 years) who were homozygous for the 6.0 kb allele.
1
2
3
4
5-6
of Population Studied
n
black
Restriction-endonuclease
Pedigree Analysis
DNA Isolation
Feature
RESULTS
Kb
1 24* 3-40
14
20
22
54 + 14
44
33-62
-
7.6 kbW 6.0 kbl) Fig 2. Typical autoradiogram. The numbers at the far left refer to the fragment sizes in kb. Lane 1,4,6-heterozygous for the 6.0 and 7.6 alleles: Lane 2,5-homozygous for 6.0; Lane Column 3-homozygous for 7.6.
15
PEDIGREE ANALYSIS OF THE INSULIN GENE
grees B through E with the + allele and B, C, and E with the - allele. In summary, for the majority of individual pedigrees and for the group of families analyzed as a whole, there was Mendelian inheritance of the insertional allele, but inconsistent segregation of the allele and the phenotype of DM. Length Polymorphism in Unrelated Individuals
Fig 3. Pedigrees. Seven pedigrees studied. Age is written below each figure. Black areas refer to subjects with NIDDM. Dotted areas with IDDM. and clear areas are unaffected. (- 1allele without an insertion: I + 1allele with an insertion.
Since DM is a heterogenous disorder we also analyzed our pedigrees individually. Each family shown in Fig 3 was examined for cosegregation of NIDDM or IDDM with the + or - insulin alleles. When we assumed an autosomal dominant mode of inheritance, recombination, ie, no linkage, of the alleles with the disease, was seen in all pedigrees except for A and F. In family F, since there is no + allele present, recombination or cosegregation cannot be proven. Similarly in families A and F, cosegregation may exist with the allele. When we assumed an autosomal recessive mode of inheritance, then recombination was seen in pedi-
We analyzed 144 alleles from 72 unrelated individuals, including 12 spouses from the pedigrees. The most common alleles were 6.0 and 7.6 kb. Two larger alleles of 9.0 kb correlating with 2 insertions, were also observed in 2 NIDDM. The presence of the 7.6 kb allele was observed in 16 of 36 ND (44%), 1 in 5 IDDM (20%), and in 16 of 31 NIDDM (51%), respectively (P > 0.59; Table 2). When the black individuals were analyzed separately the presence of 7.6 kb allele was observed in 1 of 2 ND, 0 of 1 IDDM, and 9 of 20 NIDDM (Fisher’s exact test P = 1.OO). The allelic frequencies of the 7.6 kb and 6.0 kb alleles were, respectively, 0.236 and 0.764 in ND, 0.125 and 0.875 in IDDM, and 0.312 and 0.688 in NIDDM (P = 0.201; Table 3). Thus in unrelated individuals there was no evidence of increased frequency of the 7.6 allele in ND v those with DM. DISCUSSION
There are conflicting data in the literature regarding the association of the 1.6 kb insert in the 5’ flanking region of the insulin gene and the phenotype of NIDDM.‘,” We are the first to have undertaken multiple family studies to determine if the insertion and the phenotype of NIDDM were genetically linked. Using a human insulin gene probe we detected a 1.6
Table 2. Presence or Absence of the Insertion in at Least One Allele and Its Association All Pedigrees ND
n
P
57
32
25
5
3
2
NIDDM
18
10
8
Sibships ND
13
6
7
5
3
2
12
8
4
36
16
20
5
1
4
31
16
15
2
1
1
1
0
1
9
11
IDDM
IDDM NIDDM Unrelated’ ND IDDM NIDDM Blacks + ND IDDM NIDDM
20
Abbreviations: (+ I. insertion allele present: (-),
1.10
1.79
.984
,583
.95 ? .532
2.74
+ ,846
1.242
r .480
.826
& 1.23
With Diabetes
-
I I I 1
CL
95%
.33 t 2.69
P
NS
NS
,207
- 3.18
NS
0.19
-
NS
10.17
absence of insertion allele: X2. Chi-square; OR, Odds ratio + standard error; 95% CL., 95%
confidence limits. l
.03
OR + SE
Including 12 spouses from pedigrees.
? When n < 40 Fisher’s Exact Test was performed.
DOBS ET AL
16
Table 3. Genotype and Allelic Frequencies for the Insertion et the 5’ Flanking Region of the Human Insulin Gene in Unrelated Individuals Alklic
Frequency
Frequency
n
ND
36
47
20
15
1
.54
.40
IDDM NIDDM
4 32
31 52
3 16
1 12
0 4
.75 .50
.25 .37
Abbreviations: t-j,
Qnqtype
Genotypic Age x
-/-
+/-
+/i
-/-
+1-
iI+
X2
-
+
,027
.764
.236
-
,125 .312
.278
,125
.a75 ,688
P
Sign
,203 -
NS
-
common 6.0 kb allele; (+), an allele with an insertion; X2 = chi-square contingency table 3 and 2.
kb insertion polymorphism corresponding to a class III allele, in the 5’ flanking region which is similar to that found by Bell et al,‘,* Rotwein et al,” and Owerbach et al.13 When allelic frequencies of the insertion polymorphism in unrelated individuals were compared there were no statistically significant differences in the frequencies of the insert observed in IDDM, NIDDM, or normals. When separated by race this lack of significance remained. In our 7 pedigrees, we found, as did Owerbach et alI5 in their one pedigree, that the insert is a stable genetic element, inherited in Mendelian fashion. In our pedigree analyses we found independent assortment of the insertion polymorphism and the phenotypes IDDM and NIDDM. In lieu of the heterogeneity of DM, there may exist a subset of diabetic families in which linkage can be documented. However, in our studies, evaluation for linkage of each separate pedigree suggested that in the majority of cases, the phenotypes of NIDDM and IDDM segregated independently of either the + or insulin allele when either an autosomal dominant or recessive mode of inheritance was hypothesized. This holds tiue for one of the black families B, as well as the white pedigrees. Pedigree F is uninformative. There are several limitations to our study. First, the insertion may have varying frequencies in different ethnic or racial groups. We have attempted to distinguish the effect of race on the prevalence of the polymorphism. Although our data suggested no relation, a larger sample size of several racial groups might be more informative. We did attempt controlling for race when we analyzed the data by sibships and found no association between the insert and DM. Most recently Knowler et al” published some elegant studies on Pima Indians and concluded that the insertional class III polymorphism had no importance in the genesis of NIDDM. A second problem is that the mean age of the
controls are younger than those with NIDDM. Perhaps with time those normal individuals with the insert will become diabetic. We realize that controlling for age can be a difficult problem when utilizing pedigree analysis as the study design. However, this disadvantage must be balanced against the advantages of studying families, ie, the improved ability to test for linkage of an allele and a phenotype. Finally, although the insulin alleles clearly fell into 2 principal groups, one referred to as the 6.0 kb fragment and the other the 7.6 kb fragment using the Sst I enzyme, there was a small variation in fragment size within each group. In that the relative sizes of the 2 fragments remained the same, we feel that the +500 bp was due to technical qualities of gel electrophoresis. Interestingly, the insertional polymorphism appears to have little effect on the expression of the insulin gene in in vitro systems,” which would fit with the results of our in vivo family studies. The region of significant gene regulation of the 5’ end of the human insulin gene appears to be up to -258 bp from the site of mRNA transcription, much closer than the location of the insertion.” In summary, in our population there exists a highly polymorphic area in the 5’ flanking region of the insulin gene. This 1.6 kb insertional polymorphism has a Mendelian inheritance pattern but from our multiple pedigree analysis it does not show statistical association with or genetic linkage to the phenotypic expression of DM and thus is not a reliable marker for the disease. ACKNOWLEDGMENT We are indebted to M. Pass and J. Thrower for manuscript preparation and M. Fuches for excellent technical assistance. ASD gratefully acknowledges Drs M.A. Levine, H.H. Kazazian, and R. Pyeritz for their guidance and suggestions.
REFERENCES 1. Lebo RV, Kan YW, Cheung MC, et al: Assigning the polymorphic human insulin gene to the short arm of chromosome 11 by chromosome sorting. Human Genetics 60:10-15, 1982 2. Owerbach D, Bell GI, Rutter WJ et al: The insulin gene is located on chromosome 11 in humans. Nature 286:82-84, 1980
3. Harper human insulin 11. Proc Nat1 4. Bell GI, gene. Nature
ME, Ullrich A, Saunders GF: Localization of the gene to the distal end of the short arm of chromosome Acad Sci USA, 78:44584460, 198 1 Pictet RL, Rutter WJ, et al: Sequence of the insulin 284:26-32, 1980
PEDIGREE ANALYSIS
OF THE INSULIN
17
GENE
5. Owerbach D, Bell GI, Rutter WJ, et al: The insulin gene is located on the short arm of chromosome 11 in humans. Diabetes 30:267-270, 198 1 6. Ullrich A, Dull TJ, Gray A, et al: Genetic human insulin gene. Science 209:612-615, 1980
variation
in the
7. Bell GI, Selby MJ, Rutter WJ: The highly polymorphic region near the human insulin gene is composed of simple randomly repeating sequences. Nature 295:31-35, 1982 8. Bell Gl, Karam JH, Rutter WJ: Polymorphic DNA region adjacent to the 5’ end of the human insulin gene. Proc Nat1 Acad Science USA 78:5759-5763, 198 1 9. Ullrich A. Dull TJ, Gray A, et al: Variation in the sequence and modification state of the human insulin gene flanking regions. Nucleic Acids Res 10:2225-2240, 1982 10. Bell GI, Horita S, Karam JH: A polymorphic locus near the human insulin gene is associated with insulin dependent diabetes mellitus. Diabetes 33:176-183, 1984 1 I Rotwein PS, Chirgwin J, Province M, et al: Polymorphism the 5’ flanking region of the human insulin gene: a genetic marker N IDDM. N Engl J Med 3086-7 I, 1983
in for
12. Rotwein P, Chyn J, Chirgwin J, et al: Polymorphism in the 5’ flanking region of the human insulin gene and its possible relation to Type 2 diabetes. Science 2 13: 1117-l 120, I98 I 13. Owerbach D, Nerup J: Restriction fragment length polymorphism of the insulin gene in diabetes mellitus. Diabetes 31:275-277, 1982 14. Mandrup-Paulsen T, Owerbach SA, Mortensen SA, et al: DNA-sequences flanking the insulin gene on chromosome 1 I confer risk of atherosclerosis. Lancet I (8371):25&252, 1984 15. Owerbach
D, Poulsen S, Billesbolle
P, et al: DNA insertion
sequences near the insulin gene affect glucose regulation. Lancet l(i):880-883, 1982 16. Hitman GA, Jowett NI, Williams LG, et al: Polymorphisms in the 5’ flanking region of the insulin gene-non-insulin diabetes. Clin Sci 66:383-388, 1984 17. National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose insulin tolerance. Diabetes 28:1039-1057, 1979 18. Southern EM: Detection of specific sequences among DNA fragments separated by gel electrophoresis. J Mol Biol 98:503-517, 1975 19. George DL, Phillips III JA, Francke U, et al: The genes for growth hormone and chorionic somatomammotropin are on the long arm of human chromosome 17 in region q2l - qter. Hum Genet 57:138-141, 1981 20. Rigby PW, Dieckmann M, Rhodes C, et al: Labeling deoxyribonucleic acid to high specific activity in vitro by Nick Translation with DNA Polymerase I. J Mel Bio 113:237-251, 1977 21. Snedecor GN, Cochran WG: Statistical Methods. Ames. Iowa, Iowa State University Press, 1980, p 73 22. Snedecor GN, Cochran WG: Statistical Methods. Ames, Iowa, Iowa State University Press, 1980, pp 50-53 23. Haldane JBS: The estimation and significance of the logarithm of a ratio of frequencies. Ann Hum Genet 20:309, 1955 24. Fisher RA: Statistical Methods for Research Workers. Edenburgh, Oliver and Boyd, 1941, p 21 25. Knowler WC, Pettitt DJ, Vasquez B, et al: Polymorphism in the 5’ flanking region of the human insulin gene. J Clin Invest 74:2129-2135, 1984 26. Walker MD, Edlund T, Boulet AM. et al: Cell specific expression controlled by the 5’-flanking region of insulin and chymotrypsin genes. Nature 306:557-561, 1983
Analysis
of the 5’ Flanking Region of the Insulin Gene in Familial Diabetes Mellitus
Adrian S. Dobs, John A. Phillips III, Richard L. Mallonee, Non-insulin-dependent the 5’ flanking phenotype from
of NIDDM
7 pedigrees
mellitus (NIDDM)
has been reported
insulin gene. We
by studying multiple pedigrees.
(5 white,
insulin-dependent digestion
diabetes
region of the human
2 black) ranging
diabetes mellitus (IDDMI,
with Sst I, and electrophoresis.
Christopher
to be associated with an insertion
have attempted
We evaluated
D. Saudek, and Robert L. Ney
to examine
NIDDM
insertion
polymorphism.
(P = 0.98). When
(P = 0.58). In examining
and 80 were nondiabetics
Of these,
52 subjects
the DNA was blotted to nitrocellulose
sibships were analyzed the 7.6 kb allele occurred
in 16 of 36 ND, 1 of 5 IDDM,
filters and hybridized
insertion phenotype
polymorphism
flanking
the
insulin
previously
and 10 of 18
in 6 of 13 ND. 3 of 5 IDDM, and 8 of 12 NIDDM the 7.6 kb allele was documented
(P = 0.59). In these individuals and in the multiple families studied the
gene
showed
Mendelian
inheritance
and assorted
independently
of the
of diabetes mellitus.
Q 1986 by Grune
& Stratton,
inc.
R
ECENT characterization of the human insulin geneId has enabled detailed evaluation of how this gene may contribute to the etiology of diabetes mellitus (DM). Several laboratories have reported a polymorphic region of heterogeneous length occurring about 500 base pairs (bp) 5’ to the human insulin gene origin of mRNA transcription (Fig 1).7-9 The genomic DNA fragment corresponding to this region has been sequenced and found to primarily contain a 14-l 5 bp repetitive sequence rich in guanine and cytosine oligoare due to nucleotides.7.9 The length polymorphisms insertions of varying sizes. Bell et al” categorized these lengths into Class I alleles of 570 bp, Class II alleles of 1320 bp, and class III of 2470 bp in size. The allelic frequencies vary among ethnic groups with Blacks often having a genotype that includes the rare Class II alleles. Population studies of unrelated individuals have suggested that the frequency of these alleles may vary among phenotypes. Rotwein et al”,‘2 and Owerbach et alI3 have reported frequencies of the insertion Class III alleles significantly higher in non-insulin-dependent diabetes mellitus (NIDDM) than in insulin-dependent diabetes mellitus (IDDM) or nondiabetic (ND) people. Bell et al” observed a higher frequency of the short alleles (Class I) in Caucasian IDDM and N IDDM groups compared to ND controls. The level of significance for the NIDDM was, however, only marginal. Their analysis of composite data revealed no association between the presence or absence of Class I or CIass III inserts and NIDDM. Owerbach et al” have subsequently stated that the large insertions are primarily seen in individuals with atherosclerosis and may only secondarily be associated with diabetes. These studies employed unrelated individuals to derive allelic frequency data. Since families would Metabolism,
10 had
to a o?P-labeled
to the common
in 32 of 57 ND, 3 of 5 IDDM,
subjects. including 12 spouses from the pedigrees,
and 16 of 31 NIDDM
had NIDDM,
(ND). DNA was extracted from leukocytes and after
In these families, the 7.6 kb allele occurred
72 unrelated
in
to the
142 individuals (120 white, 22 black), 80 of whom were
in size from 4 to 37 members.
insulin gene probe. Two alleles of 6.0 and 7.6 kb in size were detected, the latter corresponding described
polymorphism
linkage of this polymorphism
Vol35, No
1
(January),
1986:
pp
13- 17
theoretically have identical mutations as the cause of their disease, pedigree analysis is a superior method to derive information of linkage phenomena. Owerbach et alI5 and Hitman et alI6 studied only one family each. The latter group revealed no linkage between the size of DNA inserts and diabetes. This paper reports experiments done not only on unrelated individuals but individuals belonging to multiple pedigrees. We are the first to report 7 pedigrees in order to examine for linkage between the presence of the polymorphic insert and the phenotype of DM. MATERIALS
AND
METHODS
Study Population We analyzed a total of 142 patients (120 white, 22 black) of whom 52 subjects had NIDDM, 10 IDDM, and 80 were ND (Table 1). Eighty of these patients were from 7 pedigrees (pedigree B and F black; A, C, D, E Caucasian) ranging in size from 4 to 37 members each with 2 to 7 diabetics per kindred. Of the ND, 25 had documented normal oral glucose tolerance tests and 37 had randomly determined plasma glucose values of less than 110 mg/dL. The remaining 18 were family members who had no history of DM. The diagnosis of IDDM was established by a clinical history of ketosis.” The diagnosis of NIDDM was made if there existed persistent hyperglycemia and the absence of ketosis.” All patients with diabetes had been diagnosed at least one year prior to study.
From the Departments of Medicine and Pediatrics, The Johns Hopkins Universiiy Hospital, Baltimore. Presented in part at the American Federation of Clinical Research, 1984. Washington, DC, and appeared in abstract form in Clinical Research 32:393A. Supported in part by project 52T32AM07109, National Institutes osHealth. Address reprint requests to Dr A.S. Dobs, Traylor Building 720, The Johns Hopkins Hospital, 720 Rutland Ave. Baltimore, MD, 21205. 0 1986 by Grune & Stratton, Inc. 0026&0495/86/3501-0003$03.00/0
13
DOBS ET AL
14
3.2 kb
6kb
-5
-4
-3
-2
-I
Statistical analysis The association of the insertional allele to the diagnosis of DM was tested using a contingency table in which the presence of one or more insertional alleles was compared to the diagnosis of DM. A chi-square test of independence was then performed.*’ In addition, the 95% confidence limit and odds ratio were calculated as described previously.22~23 Using the formula (a + S) (d + .5)/(b + .5) (c + .05), a is the number of diabetics with inserts, b the number of diabetics without the insert, c nondiabetics with the insert, and d nondiabetics without inserts. There is considered to be no relation between the variables if the 95% confidence limit includes one. A Fisher’s exact test was done for blacks because n c 40.”
m pgH IG900bp I
0
2
3
4
Fig 1. Endonuclease Restriction Map. Partial map of the human insulin gene modified from Bell et al’ and Rotwein et al” including the exons in the black areas and intervening sequences in the open areas. The variable region in the 5’ flanking area is designated by the dashed triangle. The dotted bar indicates the genomic insulin gene probe used. The open bar indicates the 6 kb allele observed when genomic DNA is digested with Sst I in the absence of an insertion.
DNA was isolated from the nuclei of leukocytes contained in 20 mL of peripheral blood collected with an EDTA anticoagulant by a procedure similar to Bell et al* The yield of DNA from fresh whole blood was 20-60 ugDNA/mL.
Analysis of the Insulin Gene Five pclgaliquots of DNA were digested with Ssr I according to the manufacturer’s specifications (Bethesda Research Laboratories, Rockville, Md). The DNA fragments were separated by electrophoresis at 15 mv constant current for 16 hours, in 0.9% (wt/vol) agarose gels prepared in 0.04M Tris acetate(pH 8.0)/0.025 moI/L sodium acetate/O.002 mol/L EDTA. The DNA fragments were then transferred from the agarose gels to nitrocellulose filters as described by Southern.” The filters were baked at 70 OC for 6 hours and prehybridized and hybridized as previously described.” The DNA fragment used,as the hybridization probe was a 900 bp Pst I fragment of the genomic human insulin gene (Fig 1) cloned into the plasmid pBR322 and grown in Escherichiu coli HBlOl cells.6 Following excision and isolation of the insert from the vector, the insert was labeled with cu’*P ATP and a’*P CTP by “nick translation” with a specific activity of 224 x IO* counts/min/ug.2”
Table 1. Characteristics
analysis of the insulin gene
Autoradiogram patterns of DNA from 142 individuals after restriction endonuclease digestion with Sst I and hybridization with the pgH1 probe revealed two size groups of alleles. One was a 6.0 kb (-) allele and the other was a 7.6 kb (+) allele (Fig 2). This size fragment would correspond to a class III allele.
ND
IDDM
NIDDM
Total
80
10
52
142
Sex: male
49
7
19
75
female
31
3
33
67
Race: white
79
9
32
120
1
Agex+SEMy
40 * 22
range y
l-74
Seven kindreds were studied to determine if the polymorphic flanking region and the phenotype of DM were genetically linked (Fig 3). The 7.6 kb allele occurred in 32 of 57 ND(56%), 3 of 5 IDDM (60%), and 10 of 18 NIDDM (55%), (P = 0.98). Comparing within sibships, the 7.6 kb allele occurred in 6 of 13 ND (46%), 3 of 5 IDDM (60%), and 8 of 12 NIDDM sibs (66%), (P = 0.58). In these families there were 4 adults (mean = 52, range 35-66 years) who were homozygous for the 7.6 kb fragment. Of these, 3 had oral glucose tolerance tests. One test was normal, while the other 2 were impaired having a normal fasting blood sugar and returning to 98 and 145mg% at 2 hoursI In addition there were 11 NIDDM (mean = 52, range 35-74 years) who were homozygous for the 6.0 kb allele.
1
2
3
4
5-6
of Population Studied
n
black
Restriction-endonuclease
Pedigree Analysis
DNA Isolation
Feature
RESULTS
Kb
1 24* 3-40
14
20
22
54 + 14
44
33-62
-
7.6 kbW 6.0 kbl) Fig 2. Typical autoradiogram. The numbers at the far left refer to the fragment sizes in kb. Lane 1,4,6-heterozygous for the 6.0 and 7.6 alleles: Lane 2,5-homozygous for 6.0; Lane Column 3-homozygous for 7.6.
15
PEDIGREE ANALYSIS OF THE INSULIN GENE
grees B through E with the + allele and B, C, and E with the - allele. In summary, for the majority of individual pedigrees and for the group of families analyzed as a whole, there was Mendelian inheritance of the insertional allele, but inconsistent segregation of the allele and the phenotype of DM. Length Polymorphism in Unrelated Individuals
Fig 3. Pedigrees. Seven pedigrees studied. Age is written below each figure. Black areas refer to subjects with NIDDM. Dotted areas with IDDM. and clear areas are unaffected. (- 1allele without an insertion: I + 1allele with an insertion.
Since DM is a heterogenous disorder we also analyzed our pedigrees individually. Each family shown in Fig 3 was examined for cosegregation of NIDDM or IDDM with the + or - insulin alleles. When we assumed an autosomal dominant mode of inheritance, recombination, ie, no linkage, of the alleles with the disease, was seen in all pedigrees except for A and F. In family F, since there is no + allele present, recombination or cosegregation cannot be proven. Similarly in families A and F, cosegregation may exist with the allele. When we assumed an autosomal recessive mode of inheritance, then recombination was seen in pedi-
We analyzed 144 alleles from 72 unrelated individuals, including 12 spouses from the pedigrees. The most common alleles were 6.0 and 7.6 kb. Two larger alleles of 9.0 kb correlating with 2 insertions, were also observed in 2 NIDDM. The presence of the 7.6 kb allele was observed in 16 of 36 ND (44%), 1 in 5 IDDM (20%), and in 16 of 31 NIDDM (51%), respectively (P > 0.59; Table 2). When the black individuals were analyzed separately the presence of 7.6 kb allele was observed in 1 of 2 ND, 0 of 1 IDDM, and 9 of 20 NIDDM (Fisher’s exact test P = 1.OO). The allelic frequencies of the 7.6 kb and 6.0 kb alleles were, respectively, 0.236 and 0.764 in ND, 0.125 and 0.875 in IDDM, and 0.312 and 0.688 in NIDDM (P = 0.201; Table 3). Thus in unrelated individuals there was no evidence of increased frequency of the 7.6 allele in ND v those with DM. DISCUSSION
There are conflicting data in the literature regarding the association of the 1.6 kb insert in the 5’ flanking region of the insulin gene and the phenotype of NIDDM.‘,” We are the first to have undertaken multiple family studies to determine if the insertion and the phenotype of NIDDM were genetically linked. Using a human insulin gene probe we detected a 1.6
Table 2. Presence or Absence of the Insertion in at Least One Allele and Its Association All Pedigrees ND
n
P
57
32
25
5
3
2
NIDDM
18
10
8
Sibships ND
13
6
7
5
3
2
12
8
4
36
16
20
5
1
4
31
16
15
2
1
1
1
0
1
9
11
IDDM
IDDM NIDDM Unrelated’ ND IDDM NIDDM Blacks + ND IDDM NIDDM
20
Abbreviations: (+ I. insertion allele present: (-),
1.10
1.79
.984
,583
.95 ? .532
2.74
+ ,846
1.242
r .480
.826
& 1.23
With Diabetes
-
I I I 1
CL
95%
.33 t 2.69
P
NS
NS
,207
- 3.18
NS
0.19
-
NS
10.17
absence of insertion allele: X2. Chi-square; OR, Odds ratio + standard error; 95% CL., 95%
confidence limits. l
.03
OR + SE
Including 12 spouses from pedigrees.
? When n < 40 Fisher’s Exact Test was performed.
DOBS ET AL
16
Table 3. Genotype and Allelic Frequencies for the Insertion et the 5’ Flanking Region of the Human Insulin Gene in Unrelated Individuals Alklic
Frequency
Frequency
n
ND
36
47
20
15
1
.54
.40
IDDM NIDDM
4 32
31 52
3 16
1 12
0 4
.75 .50
.25 .37
Abbreviations: t-j,
Qnqtype
Genotypic Age x
-/-
+/-
+/i
-/-
+1-
iI+
X2
-
+
,027
.764
.236
-
,125 .312
.278
,125
.a75 ,688
P
Sign
,203 -
NS
-
common 6.0 kb allele; (+), an allele with an insertion; X2 = chi-square contingency table 3 and 2.
kb insertion polymorphism corresponding to a class III allele, in the 5’ flanking region which is similar to that found by Bell et al,‘,* Rotwein et al,” and Owerbach et al.13 When allelic frequencies of the insertion polymorphism in unrelated individuals were compared there were no statistically significant differences in the frequencies of the insert observed in IDDM, NIDDM, or normals. When separated by race this lack of significance remained. In our 7 pedigrees, we found, as did Owerbach et alI5 in their one pedigree, that the insert is a stable genetic element, inherited in Mendelian fashion. In our pedigree analyses we found independent assortment of the insertion polymorphism and the phenotypes IDDM and NIDDM. In lieu of the heterogeneity of DM, there may exist a subset of diabetic families in which linkage can be documented. However, in our studies, evaluation for linkage of each separate pedigree suggested that in the majority of cases, the phenotypes of NIDDM and IDDM segregated independently of either the + or insulin allele when either an autosomal dominant or recessive mode of inheritance was hypothesized. This holds tiue for one of the black families B, as well as the white pedigrees. Pedigree F is uninformative. There are several limitations to our study. First, the insertion may have varying frequencies in different ethnic or racial groups. We have attempted to distinguish the effect of race on the prevalence of the polymorphism. Although our data suggested no relation, a larger sample size of several racial groups might be more informative. We did attempt controlling for race when we analyzed the data by sibships and found no association between the insert and DM. Most recently Knowler et al” published some elegant studies on Pima Indians and concluded that the insertional class III polymorphism had no importance in the genesis of NIDDM. A second problem is that the mean age of the
controls are younger than those with NIDDM. Perhaps with time those normal individuals with the insert will become diabetic. We realize that controlling for age can be a difficult problem when utilizing pedigree analysis as the study design. However, this disadvantage must be balanced against the advantages of studying families, ie, the improved ability to test for linkage of an allele and a phenotype. Finally, although the insulin alleles clearly fell into 2 principal groups, one referred to as the 6.0 kb fragment and the other the 7.6 kb fragment using the Sst I enzyme, there was a small variation in fragment size within each group. In that the relative sizes of the 2 fragments remained the same, we feel that the +500 bp was due to technical qualities of gel electrophoresis. Interestingly, the insertional polymorphism appears to have little effect on the expression of the insulin gene in in vitro systems,” which would fit with the results of our in vivo family studies. The region of significant gene regulation of the 5’ end of the human insulin gene appears to be up to -258 bp from the site of mRNA transcription, much closer than the location of the insertion.” In summary, in our population there exists a highly polymorphic area in the 5’ flanking region of the insulin gene. This 1.6 kb insertional polymorphism has a Mendelian inheritance pattern but from our multiple pedigree analysis it does not show statistical association with or genetic linkage to the phenotypic expression of DM and thus is not a reliable marker for the disease. ACKNOWLEDGMENT We are indebted to M. Pass and J. Thrower for manuscript preparation and M. Fuches for excellent technical assistance. ASD gratefully acknowledges Drs M.A. Levine, H.H. Kazazian, and R. Pyeritz for their guidance and suggestions.
REFERENCES 1. Lebo RV, Kan YW, Cheung MC, et al: Assigning the polymorphic human insulin gene to the short arm of chromosome 11 by chromosome sorting. Human Genetics 60:10-15, 1982 2. Owerbach D, Bell GI, Rutter WJ et al: The insulin gene is located on chromosome 11 in humans. Nature 286:82-84, 1980
3. Harper human insulin 11. Proc Nat1 4. Bell GI, gene. Nature
ME, Ullrich A, Saunders GF: Localization of the gene to the distal end of the short arm of chromosome Acad Sci USA, 78:44584460, 198 1 Pictet RL, Rutter WJ, et al: Sequence of the insulin 284:26-32, 1980
PEDIGREE ANALYSIS
OF THE INSULIN
17
GENE
5. Owerbach D, Bell GI, Rutter WJ, et al: The insulin gene is located on the short arm of chromosome 11 in humans. Diabetes 30:267-270, 198 1 6. Ullrich A, Dull TJ, Gray A, et al: Genetic human insulin gene. Science 209:612-615, 1980
variation
in the
7. Bell GI, Selby MJ, Rutter WJ: The highly polymorphic region near the human insulin gene is composed of simple randomly repeating sequences. Nature 295:31-35, 1982 8. Bell Gl, Karam JH, Rutter WJ: Polymorphic DNA region adjacent to the 5’ end of the human insulin gene. Proc Nat1 Acad Science USA 78:5759-5763, 198 1 9. Ullrich A. Dull TJ, Gray A, et al: Variation in the sequence and modification state of the human insulin gene flanking regions. Nucleic Acids Res 10:2225-2240, 1982 10. Bell GI, Horita S, Karam JH: A polymorphic locus near the human insulin gene is associated with insulin dependent diabetes mellitus. Diabetes 33:176-183, 1984 1 I Rotwein PS, Chirgwin J, Province M, et al: Polymorphism the 5’ flanking region of the human insulin gene: a genetic marker N IDDM. N Engl J Med 3086-7 I, 1983
in for
12. Rotwein P, Chyn J, Chirgwin J, et al: Polymorphism in the 5’ flanking region of the human insulin gene and its possible relation to Type 2 diabetes. Science 2 13: 1117-l 120, I98 I 13. Owerbach D, Nerup J: Restriction fragment length polymorphism of the insulin gene in diabetes mellitus. Diabetes 31:275-277, 1982 14. Mandrup-Paulsen T, Owerbach SA, Mortensen SA, et al: DNA-sequences flanking the insulin gene on chromosome 1 I confer risk of atherosclerosis. Lancet I (8371):25&252, 1984 15. Owerbach
D, Poulsen S, Billesbolle
P, et al: DNA insertion
sequences near the insulin gene affect glucose regulation. Lancet l(i):880-883, 1982 16. Hitman GA, Jowett NI, Williams LG, et al: Polymorphisms in the 5’ flanking region of the insulin gene-non-insulin diabetes. Clin Sci 66:383-388, 1984 17. National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose insulin tolerance. Diabetes 28:1039-1057, 1979 18. Southern EM: Detection of specific sequences among DNA fragments separated by gel electrophoresis. J Mol Biol 98:503-517, 1975 19. George DL, Phillips III JA, Francke U, et al: The genes for growth hormone and chorionic somatomammotropin are on the long arm of human chromosome 17 in region q2l - qter. Hum Genet 57:138-141, 1981 20. Rigby PW, Dieckmann M, Rhodes C, et al: Labeling deoxyribonucleic acid to high specific activity in vitro by Nick Translation with DNA Polymerase I. J Mel Bio 113:237-251, 1977 21. Snedecor GN, Cochran WG: Statistical Methods. Ames. Iowa, Iowa State University Press, 1980, p 73 22. Snedecor GN, Cochran WG: Statistical Methods. Ames, Iowa, Iowa State University Press, 1980, pp 50-53 23. Haldane JBS: The estimation and significance of the logarithm of a ratio of frequencies. Ann Hum Genet 20:309, 1955 24. Fisher RA: Statistical Methods for Research Workers. Edenburgh, Oliver and Boyd, 1941, p 21 25. Knowler WC, Pettitt DJ, Vasquez B, et al: Polymorphism in the 5’ flanking region of the human insulin gene. J Clin Invest 74:2129-2135, 1984 26. Walker MD, Edlund T, Boulet AM. et al: Cell specific expression controlled by the 5’-flanking region of insulin and chymotrypsin genes. Nature 306:557-561, 1983