Selective Expression and Developmental Regulation of the Ancestral Rat Insulin II Gene in Fetal Liver
Stephen J. Giddings and Lynn R. Carnaghi Washington University Medical Service (111JC) St. Louis Department of Veterans' Affairs Medical Center John Cochran Division St. Louis, Missouri 63106
Previous studies have indicated that high levels of insulin synthesis occur in the yolk sac of fetal rats. Because the yolk sac is an early site for synthesis of several tissue-specific proteins synthesized by liver later in development, these studies were performed to determine whether insulin gene expression also occurs in fetal liver. To this purpose, liver RNA obtained on consecutive days of rat fetal development from embryo day (E) 13 to E21 was evaluated for the presence of insulin or insulin-like mRNA species using Northern hybridization with a uniformly labelled rat insulin II genomic antisense RNA probe. Two species were detected. The larger was approximately 2.4 kilobases in length, was very low in abundance, and was present only during the earliest days studied (E13-15). The second species was approximately 720 bases in length, increased in abundance between days E13-16, and decreased between days E16-21. Maximum abundance of this mRNA was 0.3 pg/Mg total liver RNA, or 1/10th to 1/ 20th the abundance of total insulin mRNA in adult rat pancreas. Sequencing of multiple cloned products of E15 rat liver cDNA amplified by polymerase chain reaction using insulin I or II gene-specific primers indicated that the bands detected on Northern hybridization were (ancestral) rat insulin II gene transcripts. Analysis of products of polymerase chain reactions also indicated that the duplicated rat insulin I gene was not expressed in fetal liver. The content of insulin mRNA in fetal liver is sufficient to suggest that the liver may be a significant source for insulin at specific times during fetal development. (Molecular Endocrinology 4: 1363-1369,1990)
lated to show that insulin production is not limited to the pancreas during fetal development. Muglia and Locker (2) first suggested the presence of a mRNA encoding insulin in rat yolk sac and confirmed that observation in a subsequent report (3). De Pablo et al. (4, 5) presented evidence for both insulin synthesis in nonpancreatic tissues and insulin actions in early chick embryos. Others have made similar observations in rodents (6). Recent studies have shown that fetal and neonatal insulin receptors have greater intrinsic tyrosine kinase activity than those in the adult (7, 8), that insulin can mediate the phosphorylation of insulin-like growth factor-l (IGF-I) receptor in the fetus via interaction with insulin receptors (8), and that insulin-IGF-l receptor complexes may exist in vivo as heterodimers (9). All of these findings point to the need for a reassessment of the potential role(s) of insulin during embryogenesis and fetal development. In previous studies we confirmed the presence of insulin synthesis in rat yolk sac (10). In that report we demonstrated that only the ancestral rat insulin II gene, but not the duplicated rat insulin I gene, was expressed, that its expression preceded in time and exceeded in quantity that in fetal pancreas. Because yolk sac expresses several genes also expressed by fetal liver and other endodermally derived tissues [IGF-II (11), transthyretin (12), a-fetoprotein (13), retinol-binding protein (14), and apoproteins-A-l and -A-IV (15), among others] , fetal rat liver was assessed for the presence of insulin mRNA. Results indicate that fetal liver expresses the ancestral insulin II gene, but not the duplicated insulin I gene, similar to yolk sac and different from fetal pancreas (10, 16). Hepatic insulin II gene expression is regulated differently during development than that in yolk sac or pancreas; hepatic insulin II mRNA reaches maximum levels between embryo days (E) 15-17, declines during late fetal development (days E18-21) and is undetectable in the adult. The abundance of this mRNA suggests that between days E13 and E16, liver produces a significant portion of total fetal insulin.
INTRODUCTION While insulin production by nonpancreatic tissues in the adult remains controversial (1), evidence has accumu0888-8809/90/1363-1369S02.00/0 Molecular Endocrinology Copyright © 1990 by The Endocrine Society
1363
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 1364
Vol 4 No. 9
RESULTS Analysis of Fetal Hepatic Insulin or Insulin-Like mRNAs by Northern Hybridization While screening several tissues for the presence of insulin mRNA using both Northern hybridization and ribonuclease protection assays (10), a signal for insulin was detected in day E18 rat liver RNA (data not shown), similar to that reported by Muglia and Locker (2). To further document and characterize the presence of insulin or insulin-like mRNAs in liver during late embryonic (days E13-15) and fetal (days E16-E21) periods in the rat, liver samples were collected on subsequent days of development between days E13-21, using extreme care to avoid contamination by either membranes or pancreas, as described in Materials and Methods. Aliquots of total fetal and adult hepatic RNA, day E20 pancreatic RNA and adult parotid gland RNA were electrophoresed, transferred to membranes, and hybridized with a rat insulin II cRNA probe, as described in Materials and Methods. Results are shown in Fig. 1A. In contrast to yolk sac and fetal pancreas, the abundance of a 720-base long hybridizing moiety was greatest between days E15 and E17. The autoradiogram was subjected to scanning densitometry, and the relative intensities of signals from each sample were plotted in Fig. 1B. The abundances of total insulin mRNA in yolk sac and pancreas, determined previously
(10), have been included here for comparison. The signal from 20 /xg day E16 hepatic RNA was equivalent to that from 200 ng day E20 pancreatic RNA. The absolute abundance of insulin mRNA in this day E20 pancreas sample, determined previously (10), was 30 pg/nQ. Thus, fetal liver at most contains approximately 0.3 pg insulin or insulin-like mRNA/jug total RNA. Also detected in this experiment was a 2.4-kilobase band, similar in size to that reported by Muglia and Locker (2, 3). The developmental pattern of expression for this species is different from that of the 720-base moiety. It was most abundant on day E13 and diminished thereafter, becoming undetectable by day E16. Sequence Determination of Hepatic Insulin Gene Transcripts in Fetal Hepatic cONAs Amplified by Polymerase Chain Reaction To establish conclusively the identity of the RNA species detected by Northern hybridization, single stranded cDNA was created from day E15 hepatic RNA using oligo-dT as a primer (see Materials and Methods). These cDNAs were then amplified using rat insulin Iand ll-specific primers (see Fig. 2). Because the primer sequences flank introns in genomic DNA, amplification products of genomic DNA can be distinguished by size from products of cDNA amplification. Amplification products of rat insulin I and II cDNA should be 285 bases long, those from rat insulin I genomic DNA should
B Pancreas IE20I
Liver 13
14
15
ib
17
IB
19 20 21 Ad
M Q02 0.2 2.0
10.000
1.000 -2400
10
11
12
13
14
15
16
17
18
19
20
21
22
DAY OF GESTATION
Fig. 1. Detection of Hepatic Insulin mRNA by Northern Hybridization Aliquots (20 ng) of days E13-21 and adult rat liver total RNA and day E20 rat pancreatic RNA (0.02, 0.2, and 2 Mg) were electrophoresed on 0.6 M formaldehyde-2% agarose horizontal slab gels, transferred to nylon membranes, hybridized with rat insulin II cRNA probe, and washed, and autoradiography was performed as described in Materials and Methods. The autoradiogram is shown in A. The intense smear in the marker lane (M) was caused by cross-hybridization of plasmid-derived sequences in the cRNA probe to the RNA markers. The approximate sizes of species detected are indicated on the right side of the autoradiogram. Exposure time was 10 days at - 7 0 C with an intensifying screen. The graph in B compares the abundance of insulin mRNA in fetal liver (triangles) to its abundance in extraplacental membranes (circles) and fetal pancreas (squares), determined previously (10). The abundance of hepatic insulin mRNA was determined by densitometry, as described in Materials and Methods.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
Insulin Gene Expression in Fetal Liver
1) 1st strand cONA synthesis
2) Amplification Round 1 Rl I 5' primer Rl II 5 primer
1365
3-
_
5
-AAAAAAAA. 3'
' 3 GCTACAATCATAGACATC -ifGCTACAGTCGGAACCATC TT-
Subsequent Rounds
GCTACAATCATAGACATC # 5' 3 primer:
3r -AAAAAAAA. . CCTAGAAGTCTGGAACCGT 5'
GCTACAGTCGGAACCATC• 5' 5' 3 primer: • 285b
ACTGGAAGTCTGGAACCGT 5' 1
3) Gel Purify
4) Phosphorylate and Blunt
nserted sequence
5) Clone
6) Sequence
Fig. 2. Schematic Diagram of Cloning and Sequencing Strategy of Rat Insulin II Gene Transcripts from Fetal Liver cDNA First (-) strand cDNA synthesis was performed using oligodeoxythymidylate as primer. Reaction products were added to polymerase chain reactions containing insulin I- or insulin ll-specific primers. Amplification products were isolated by gel purification, ends were made blunt, and fragments were cloned and sequenced as indicated in the diagram and in Materials and Methods.
be 404 bases long, and those from rat insulin II genomic DNA should be 903 bases long. Results are shown in Fig. 3. Controls included cDNA from adult rat liver (containing no insulin mRNA), cDNA created from cell line 1046-38 RNA (containing only rat insulin I mRNA) (16a) and cDNA from extraplacental membranes (containing only rat insulin II mRNA, 10). Products of sham cDNA syntheses, including all components of the cDNA synthetic reactions except reverse transcriptase, were amplified to control for the potential presence of products arising from contamination of RNA samples by small amounts of genomic or other DNA. Products of reactions with rat insulin I primers are shown in the upper lanes of the gel shown in Fig. 3. Products of reactions with rat insulin II primers are shown in the lower lanes. Lanes containing amplification products of sham cDNA syntheses (those lacking reverse transcriptase) are marked - ; lanes containing amplification products of complete cDNA synthetic reactions (those to
which reverse transcriptase was added) are marked +. The three right-most sample lanes in both the upper and lower portions of the gel contain amplification products of plasmids encoding rat insulin I (I), rat insulin II (II), and no added template (N). Rat insulin I primers amplified rat insulin I, but not rat insulin II genomic DNA; conversely, rat insulin II primers amplified rat insulin II, but not rat insulin I genomic DNA, indicating the specificity of the primers. There were no amplification products detected in the sham cDNA synthetic lanes (those marked - ) . In the cDNA amplifications with rat insulin I primers, an appropriately sized product was detected in cell line cDNA, which contains rat insulin I, but not rat insulin II mRNA; no product was detected in rat yolk sac cDNA, which contains only rat insulin II mRNA (10), or in cDNA from adult liver. Fetal liver cDNA amplification produced a single band, which was not the correct length for insulin cDNA or genomic DNA. Its identity is discussed below.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
Vol 4 No. 9
MOL ENDO-1990 1366
AL
CL
YS
N H S
-404 -285
Fig. 3. Agarose gel analysis of rat insulin I- and ll-specific oligonucleotide-primed amplification products. Pairs of 19base oligonucleotides encoding analogous regions of the rat insulin I and rat insulin II genes (shown in Fig. 2) were used to prime amplification by polymerase chain reaction of cDNAs from fetal day 15 rat liver, adult rat liver, rat insulinoma-derived cell line 1046-38, and rat yolk sac RNA. Amplifications and cDNA syntheses are described in Materials and Methods. Primer sequence flanked intron insertion regions, so that amplification products of genomic DNA would be larger than those from cDNA (404 vs. 285 bases for rat insulin I and 903 vs. 285 bases for rat insulin II). Lanes containing products of sham reverse transcriptase reactions, which contained all reaction components except reverse transcriptase, are marked (-) as are lanes containing products from complete reverse transcription reactions (+). Products of rat insulin l-primed amplifications were electrophoresed in the upper lanes of the gel, and products of rat insulin ll-primed reactions in the lower lanes. The sources of cDNAs used for each amplification are indicated above the sample lanes. AL, Adult rat liver; FL, day E15 fetal liver; CL, rat insulinoma-derived cell line 1046-38; YS, day E20 rat yolk sac. The right lanes (upper and lower) contain products of amplifications that used plasmids encoding rat insulin I genomic DNA (in the lanes marked I), rat insulin II genomic DNA (lanes marked II) as templates, or no added template (N). The markers used were H/nfl digestion products of pSP 64 (in lanes marked H) and Sspl digestion products of bacteriophage phi X 174 (lanes marked S). Products were electrophoresed in a 2% agarose-0.04 M Tris-acetate-0.002 M EDTA gel for 8 h at 2 V/cm, stained with ethidium bromide, destained, and photographed.
The results with rat insulin II primers are shown in the lower lanes in Fig. 3. Control reactions using sham cDNA synthetic products gave no signal. The only bands detected of correct size in the cDNA amplifications were obtained using fetal liver and rat yolk sac cDNA as templates. These results are consistent with the interpretation that fetal liver contains mRNA for rat insulin II. The presence of rat insulin II mRNA and the absence of rat insulin I mRNA in fetal liver RNA were
confirmed in ribonuclease protection experiments using assay conditions that discriminate between rat insulin I and rat insulin II (10) (data not shown). Because the results with fetal liver cDNA amplified with rat insulin I primers could not be explained easily and to prove that the rat insulin ll-primed fetal liver amplification product did, in fact, encode appropriate insulin II cDNA sequence, both insulin I- and insulin llprimed amplification products from fetal liver cDNA were cloned, and isolates from multiple colonies sequenced. Seven clones from rat insulin l-primed fetal liver cDNA amplification products were isolated. Six gave identical restriction digests using three separate enzymes. Two of the six clones with the same restriction pattern and the single isolate with a different restriction pattern were sequenced in both orientations. The clones with identical restriction digests contained the same sequence. None of the three clones encoded insulin; all were delimited on both 5' and 3' ends by rat insulin 13' primer sequence and most likely represented a mRNA not significantly related to insulin, present in fetal but not adult liver. Nine clones from the rat insulin ll-primed products were isolated. Digests with San 11 of eight clones gave patterns consistent with that for rat insulin II cDNA cloned in both orientations into the Sma\ site of pGEM 1. The restriction pattern of the ninth clone could not be explained. Seven of these clones were sequenced. Six of the clones sequenced (those with easily explained restriction patterns) contained sequence for insulin II inserted in 5'-3' or 3'-5' orientation. The clone giving the uninterpretable BanW restriction pattern contained two tandemly linked sequences for insulin II. Each insert was delimited at the 5' and 3' ends by rat insulin II 5' and 3' primers and encoded an appropriate portion of rat insulin II cDNA. The sequence is indicated in Fig. 4. Bases not homologous to rat insulin I are indicated by asterisks. The sequence obtained from each clone was identical to that for rat insulin II reported previously (17) with one exception, a silent A for C substitution in the codon for alanine at amino acid 18 of the C peptide. This substitution was also found in rat insulin II mRNA obtained from yolk sac (10). These results indicate that fetal liver contains mRNA for rat insulin II, but not for rat insulin I.
DISCUSSION
These experiments indicate that the ancestral rat insulin II, but not the duplicated rat insulin I gene, is expressed in fetal liver and that its expression is regulated developmentally. For several reasons, it is unlikely that these findings are the result of either pancreatic or extraplacental membrane contamination. First, the size of hepatic insulin mRNA is different from that in pancreas. Second, fetal hepatic RNA contains only rat insulin II mRNA. These findings eliminate pancreatic contamination as an explanation for the presence of insulin
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
1367
Insulin Gene Expression in Fetal Liver
16 !IVS 1 65 I15 GCTACAGTCG GAAACCATCA GCAAGCAGGT CATTGTTCCA ACATGGCCCT GTGGATCCGC TTCCTGCCCC TGCTGGCCCT GCTCATCCTC TGGGAGCCCC * * * * * * * 165 215 GCCCTGCCCA GGCTTTTGTC AAACAGCACC TTTGTGGTTC TCACTTGGTG GAAGCTCTCT ACCTGGTGTG TGGGGAGCGT GGATTCTTCT ACACACCCAT **
•
*
•
*
*
*
*
!IVS 2 265 300 GTCCCGCCGC GAAGTGGAGG ACCCACAAGT GGCACAACTG GAGCTGGGTG GAGGCCCGGG GGCAGGTGAC CTTCAGACCT TGGCA
*
*
*
*
*
c
* *
Fig. 4. Sequence of rat insulin II cDNA obtained from day E15 fetal liver. Single stranded cDNA was synthesized using oligo(dT) as a primer from day E15 liver total RNA. Complementary DNA was subjected to 35 cycles of amplification using rat insulin II genespecific primers, shown in Fig. 2. Products identical to those shown in the lower FL+ lane of Fig. 3 were isolated from agarose gels by electroelution and inserted by blunt end ligation into pGEM 1. Sequence was obtained from seven separate clones and verified by sequencing in both orientations. The sequence is numbered with reference to the pancreatic transcription start site (1). Points of insertion for introns are indicated above the sequence. Sequences corresponding to oligonucleotide primers are underlined. Bases not homologous between rat insulin I and rat insulin II are indicated by asterisks, and the base different from the previously determined rat insulin II sequence (17) is indicated below the experimentally determined sequence.
mRNA in liver. Third, the relative abundance varies in a pattern that is inversely proportional to the abundance in yolk sac, making contamination by that organ less likely. Fourth, while the low abundance in liver relative to that in yolk sac at times late in gestation [15 pg//xg in day E20 extraplacental membranes (10) vs.
Stephen J. Giddings and Lynn R. Carnaghi Washington University Medical Service (111JC) St. Louis Department of Veterans' Affairs Medical Center John Cochran Division St. Louis, Missouri 63106
Previous studies have indicated that high levels of insulin synthesis occur in the yolk sac of fetal rats. Because the yolk sac is an early site for synthesis of several tissue-specific proteins synthesized by liver later in development, these studies were performed to determine whether insulin gene expression also occurs in fetal liver. To this purpose, liver RNA obtained on consecutive days of rat fetal development from embryo day (E) 13 to E21 was evaluated for the presence of insulin or insulin-like mRNA species using Northern hybridization with a uniformly labelled rat insulin II genomic antisense RNA probe. Two species were detected. The larger was approximately 2.4 kilobases in length, was very low in abundance, and was present only during the earliest days studied (E13-15). The second species was approximately 720 bases in length, increased in abundance between days E13-16, and decreased between days E16-21. Maximum abundance of this mRNA was 0.3 pg/Mg total liver RNA, or 1/10th to 1/ 20th the abundance of total insulin mRNA in adult rat pancreas. Sequencing of multiple cloned products of E15 rat liver cDNA amplified by polymerase chain reaction using insulin I or II gene-specific primers indicated that the bands detected on Northern hybridization were (ancestral) rat insulin II gene transcripts. Analysis of products of polymerase chain reactions also indicated that the duplicated rat insulin I gene was not expressed in fetal liver. The content of insulin mRNA in fetal liver is sufficient to suggest that the liver may be a significant source for insulin at specific times during fetal development. (Molecular Endocrinology 4: 1363-1369,1990)
lated to show that insulin production is not limited to the pancreas during fetal development. Muglia and Locker (2) first suggested the presence of a mRNA encoding insulin in rat yolk sac and confirmed that observation in a subsequent report (3). De Pablo et al. (4, 5) presented evidence for both insulin synthesis in nonpancreatic tissues and insulin actions in early chick embryos. Others have made similar observations in rodents (6). Recent studies have shown that fetal and neonatal insulin receptors have greater intrinsic tyrosine kinase activity than those in the adult (7, 8), that insulin can mediate the phosphorylation of insulin-like growth factor-l (IGF-I) receptor in the fetus via interaction with insulin receptors (8), and that insulin-IGF-l receptor complexes may exist in vivo as heterodimers (9). All of these findings point to the need for a reassessment of the potential role(s) of insulin during embryogenesis and fetal development. In previous studies we confirmed the presence of insulin synthesis in rat yolk sac (10). In that report we demonstrated that only the ancestral rat insulin II gene, but not the duplicated rat insulin I gene, was expressed, that its expression preceded in time and exceeded in quantity that in fetal pancreas. Because yolk sac expresses several genes also expressed by fetal liver and other endodermally derived tissues [IGF-II (11), transthyretin (12), a-fetoprotein (13), retinol-binding protein (14), and apoproteins-A-l and -A-IV (15), among others] , fetal rat liver was assessed for the presence of insulin mRNA. Results indicate that fetal liver expresses the ancestral insulin II gene, but not the duplicated insulin I gene, similar to yolk sac and different from fetal pancreas (10, 16). Hepatic insulin II gene expression is regulated differently during development than that in yolk sac or pancreas; hepatic insulin II mRNA reaches maximum levels between embryo days (E) 15-17, declines during late fetal development (days E18-21) and is undetectable in the adult. The abundance of this mRNA suggests that between days E13 and E16, liver produces a significant portion of total fetal insulin.
INTRODUCTION While insulin production by nonpancreatic tissues in the adult remains controversial (1), evidence has accumu0888-8809/90/1363-1369S02.00/0 Molecular Endocrinology Copyright © 1990 by The Endocrine Society
1363
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 1364
Vol 4 No. 9
RESULTS Analysis of Fetal Hepatic Insulin or Insulin-Like mRNAs by Northern Hybridization While screening several tissues for the presence of insulin mRNA using both Northern hybridization and ribonuclease protection assays (10), a signal for insulin was detected in day E18 rat liver RNA (data not shown), similar to that reported by Muglia and Locker (2). To further document and characterize the presence of insulin or insulin-like mRNAs in liver during late embryonic (days E13-15) and fetal (days E16-E21) periods in the rat, liver samples were collected on subsequent days of development between days E13-21, using extreme care to avoid contamination by either membranes or pancreas, as described in Materials and Methods. Aliquots of total fetal and adult hepatic RNA, day E20 pancreatic RNA and adult parotid gland RNA were electrophoresed, transferred to membranes, and hybridized with a rat insulin II cRNA probe, as described in Materials and Methods. Results are shown in Fig. 1A. In contrast to yolk sac and fetal pancreas, the abundance of a 720-base long hybridizing moiety was greatest between days E15 and E17. The autoradiogram was subjected to scanning densitometry, and the relative intensities of signals from each sample were plotted in Fig. 1B. The abundances of total insulin mRNA in yolk sac and pancreas, determined previously
(10), have been included here for comparison. The signal from 20 /xg day E16 hepatic RNA was equivalent to that from 200 ng day E20 pancreatic RNA. The absolute abundance of insulin mRNA in this day E20 pancreas sample, determined previously (10), was 30 pg/nQ. Thus, fetal liver at most contains approximately 0.3 pg insulin or insulin-like mRNA/jug total RNA. Also detected in this experiment was a 2.4-kilobase band, similar in size to that reported by Muglia and Locker (2, 3). The developmental pattern of expression for this species is different from that of the 720-base moiety. It was most abundant on day E13 and diminished thereafter, becoming undetectable by day E16. Sequence Determination of Hepatic Insulin Gene Transcripts in Fetal Hepatic cONAs Amplified by Polymerase Chain Reaction To establish conclusively the identity of the RNA species detected by Northern hybridization, single stranded cDNA was created from day E15 hepatic RNA using oligo-dT as a primer (see Materials and Methods). These cDNAs were then amplified using rat insulin Iand ll-specific primers (see Fig. 2). Because the primer sequences flank introns in genomic DNA, amplification products of genomic DNA can be distinguished by size from products of cDNA amplification. Amplification products of rat insulin I and II cDNA should be 285 bases long, those from rat insulin I genomic DNA should
B Pancreas IE20I
Liver 13
14
15
ib
17
IB
19 20 21 Ad
M Q02 0.2 2.0
10.000
1.000 -2400
10
11
12
13
14
15
16
17
18
19
20
21
22
DAY OF GESTATION
Fig. 1. Detection of Hepatic Insulin mRNA by Northern Hybridization Aliquots (20 ng) of days E13-21 and adult rat liver total RNA and day E20 rat pancreatic RNA (0.02, 0.2, and 2 Mg) were electrophoresed on 0.6 M formaldehyde-2% agarose horizontal slab gels, transferred to nylon membranes, hybridized with rat insulin II cRNA probe, and washed, and autoradiography was performed as described in Materials and Methods. The autoradiogram is shown in A. The intense smear in the marker lane (M) was caused by cross-hybridization of plasmid-derived sequences in the cRNA probe to the RNA markers. The approximate sizes of species detected are indicated on the right side of the autoradiogram. Exposure time was 10 days at - 7 0 C with an intensifying screen. The graph in B compares the abundance of insulin mRNA in fetal liver (triangles) to its abundance in extraplacental membranes (circles) and fetal pancreas (squares), determined previously (10). The abundance of hepatic insulin mRNA was determined by densitometry, as described in Materials and Methods.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
Insulin Gene Expression in Fetal Liver
1) 1st strand cONA synthesis
2) Amplification Round 1 Rl I 5' primer Rl II 5 primer
1365
3-
_
5
-AAAAAAAA. 3'
' 3 GCTACAATCATAGACATC -ifGCTACAGTCGGAACCATC TT-
Subsequent Rounds
GCTACAATCATAGACATC # 5' 3 primer:
3r -AAAAAAAA. . CCTAGAAGTCTGGAACCGT 5'
GCTACAGTCGGAACCATC• 5' 5' 3 primer: • 285b
ACTGGAAGTCTGGAACCGT 5' 1
3) Gel Purify
4) Phosphorylate and Blunt
nserted sequence
5) Clone
6) Sequence
Fig. 2. Schematic Diagram of Cloning and Sequencing Strategy of Rat Insulin II Gene Transcripts from Fetal Liver cDNA First (-) strand cDNA synthesis was performed using oligodeoxythymidylate as primer. Reaction products were added to polymerase chain reactions containing insulin I- or insulin ll-specific primers. Amplification products were isolated by gel purification, ends were made blunt, and fragments were cloned and sequenced as indicated in the diagram and in Materials and Methods.
be 404 bases long, and those from rat insulin II genomic DNA should be 903 bases long. Results are shown in Fig. 3. Controls included cDNA from adult rat liver (containing no insulin mRNA), cDNA created from cell line 1046-38 RNA (containing only rat insulin I mRNA) (16a) and cDNA from extraplacental membranes (containing only rat insulin II mRNA, 10). Products of sham cDNA syntheses, including all components of the cDNA synthetic reactions except reverse transcriptase, were amplified to control for the potential presence of products arising from contamination of RNA samples by small amounts of genomic or other DNA. Products of reactions with rat insulin I primers are shown in the upper lanes of the gel shown in Fig. 3. Products of reactions with rat insulin II primers are shown in the lower lanes. Lanes containing amplification products of sham cDNA syntheses (those lacking reverse transcriptase) are marked - ; lanes containing amplification products of complete cDNA synthetic reactions (those to
which reverse transcriptase was added) are marked +. The three right-most sample lanes in both the upper and lower portions of the gel contain amplification products of plasmids encoding rat insulin I (I), rat insulin II (II), and no added template (N). Rat insulin I primers amplified rat insulin I, but not rat insulin II genomic DNA; conversely, rat insulin II primers amplified rat insulin II, but not rat insulin I genomic DNA, indicating the specificity of the primers. There were no amplification products detected in the sham cDNA synthetic lanes (those marked - ) . In the cDNA amplifications with rat insulin I primers, an appropriately sized product was detected in cell line cDNA, which contains rat insulin I, but not rat insulin II mRNA; no product was detected in rat yolk sac cDNA, which contains only rat insulin II mRNA (10), or in cDNA from adult liver. Fetal liver cDNA amplification produced a single band, which was not the correct length for insulin cDNA or genomic DNA. Its identity is discussed below.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
Vol 4 No. 9
MOL ENDO-1990 1366
AL
CL
YS
N H S
-404 -285
Fig. 3. Agarose gel analysis of rat insulin I- and ll-specific oligonucleotide-primed amplification products. Pairs of 19base oligonucleotides encoding analogous regions of the rat insulin I and rat insulin II genes (shown in Fig. 2) were used to prime amplification by polymerase chain reaction of cDNAs from fetal day 15 rat liver, adult rat liver, rat insulinoma-derived cell line 1046-38, and rat yolk sac RNA. Amplifications and cDNA syntheses are described in Materials and Methods. Primer sequence flanked intron insertion regions, so that amplification products of genomic DNA would be larger than those from cDNA (404 vs. 285 bases for rat insulin I and 903 vs. 285 bases for rat insulin II). Lanes containing products of sham reverse transcriptase reactions, which contained all reaction components except reverse transcriptase, are marked (-) as are lanes containing products from complete reverse transcription reactions (+). Products of rat insulin l-primed amplifications were electrophoresed in the upper lanes of the gel, and products of rat insulin ll-primed reactions in the lower lanes. The sources of cDNAs used for each amplification are indicated above the sample lanes. AL, Adult rat liver; FL, day E15 fetal liver; CL, rat insulinoma-derived cell line 1046-38; YS, day E20 rat yolk sac. The right lanes (upper and lower) contain products of amplifications that used plasmids encoding rat insulin I genomic DNA (in the lanes marked I), rat insulin II genomic DNA (lanes marked II) as templates, or no added template (N). The markers used were H/nfl digestion products of pSP 64 (in lanes marked H) and Sspl digestion products of bacteriophage phi X 174 (lanes marked S). Products were electrophoresed in a 2% agarose-0.04 M Tris-acetate-0.002 M EDTA gel for 8 h at 2 V/cm, stained with ethidium bromide, destained, and photographed.
The results with rat insulin II primers are shown in the lower lanes in Fig. 3. Control reactions using sham cDNA synthetic products gave no signal. The only bands detected of correct size in the cDNA amplifications were obtained using fetal liver and rat yolk sac cDNA as templates. These results are consistent with the interpretation that fetal liver contains mRNA for rat insulin II. The presence of rat insulin II mRNA and the absence of rat insulin I mRNA in fetal liver RNA were
confirmed in ribonuclease protection experiments using assay conditions that discriminate between rat insulin I and rat insulin II (10) (data not shown). Because the results with fetal liver cDNA amplified with rat insulin I primers could not be explained easily and to prove that the rat insulin ll-primed fetal liver amplification product did, in fact, encode appropriate insulin II cDNA sequence, both insulin I- and insulin llprimed amplification products from fetal liver cDNA were cloned, and isolates from multiple colonies sequenced. Seven clones from rat insulin l-primed fetal liver cDNA amplification products were isolated. Six gave identical restriction digests using three separate enzymes. Two of the six clones with the same restriction pattern and the single isolate with a different restriction pattern were sequenced in both orientations. The clones with identical restriction digests contained the same sequence. None of the three clones encoded insulin; all were delimited on both 5' and 3' ends by rat insulin 13' primer sequence and most likely represented a mRNA not significantly related to insulin, present in fetal but not adult liver. Nine clones from the rat insulin ll-primed products were isolated. Digests with San 11 of eight clones gave patterns consistent with that for rat insulin II cDNA cloned in both orientations into the Sma\ site of pGEM 1. The restriction pattern of the ninth clone could not be explained. Seven of these clones were sequenced. Six of the clones sequenced (those with easily explained restriction patterns) contained sequence for insulin II inserted in 5'-3' or 3'-5' orientation. The clone giving the uninterpretable BanW restriction pattern contained two tandemly linked sequences for insulin II. Each insert was delimited at the 5' and 3' ends by rat insulin II 5' and 3' primers and encoded an appropriate portion of rat insulin II cDNA. The sequence is indicated in Fig. 4. Bases not homologous to rat insulin I are indicated by asterisks. The sequence obtained from each clone was identical to that for rat insulin II reported previously (17) with one exception, a silent A for C substitution in the codon for alanine at amino acid 18 of the C peptide. This substitution was also found in rat insulin II mRNA obtained from yolk sac (10). These results indicate that fetal liver contains mRNA for rat insulin II, but not for rat insulin I.
DISCUSSION
These experiments indicate that the ancestral rat insulin II, but not the duplicated rat insulin I gene, is expressed in fetal liver and that its expression is regulated developmentally. For several reasons, it is unlikely that these findings are the result of either pancreatic or extraplacental membrane contamination. First, the size of hepatic insulin mRNA is different from that in pancreas. Second, fetal hepatic RNA contains only rat insulin II mRNA. These findings eliminate pancreatic contamination as an explanation for the presence of insulin
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 04:03 For personal use only. No other uses without permission. . All rights reserved.
1367
Insulin Gene Expression in Fetal Liver
16 !IVS 1 65 I15 GCTACAGTCG GAAACCATCA GCAAGCAGGT CATTGTTCCA ACATGGCCCT GTGGATCCGC TTCCTGCCCC TGCTGGCCCT GCTCATCCTC TGGGAGCCCC * * * * * * * 165 215 GCCCTGCCCA GGCTTTTGTC AAACAGCACC TTTGTGGTTC TCACTTGGTG GAAGCTCTCT ACCTGGTGTG TGGGGAGCGT GGATTCTTCT ACACACCCAT **
•
*
•
*
*
*
*
!IVS 2 265 300 GTCCCGCCGC GAAGTGGAGG ACCCACAAGT GGCACAACTG GAGCTGGGTG GAGGCCCGGG GGCAGGTGAC CTTCAGACCT TGGCA
*
*
*
*
*
c
* *
Fig. 4. Sequence of rat insulin II cDNA obtained from day E15 fetal liver. Single stranded cDNA was synthesized using oligo(dT) as a primer from day E15 liver total RNA. Complementary DNA was subjected to 35 cycles of amplification using rat insulin II genespecific primers, shown in Fig. 2. Products identical to those shown in the lower FL+ lane of Fig. 3 were isolated from agarose gels by electroelution and inserted by blunt end ligation into pGEM 1. Sequence was obtained from seven separate clones and verified by sequencing in both orientations. The sequence is numbered with reference to the pancreatic transcription start site (1). Points of insertion for introns are indicated above the sequence. Sequences corresponding to oligonucleotide primers are underlined. Bases not homologous between rat insulin I and rat insulin II are indicated by asterisks, and the base different from the previously determined rat insulin II sequence (17) is indicated below the experimentally determined sequence.
mRNA in liver. Third, the relative abundance varies in a pattern that is inversely proportional to the abundance in yolk sac, making contamination by that organ less likely. Fourth, while the low abundance in liver relative to that in yolk sac at times late in gestation [15 pg//xg in day E20 extraplacental membranes (10) vs.
