Somatomedin-C/Insulin-like Growth Factor4 and Insulin-like Growth Factor-I1 mRNAs in Rat Fetal and Adult Tissues*

Somatomedin-C or insulin-like growth factor I (Sm-C/IGF-I) and insulin-like growth factor I1 (IGF-11) have been implicated in the regulation of fetal growth and development. In the present study 3ZP-labeled com- plementary DNA probes encoding human and mouse Sm-C/IGF-I and human IGF-I1 were used in Northern blot hybridizations to analyse rat Sm-C/IGF-I and IGF-I1 mRNAs in poly(A+) RNAs from intestine, liver, lung, and brain of adult rats and fetal rats between day 14 and 17 of gestation. In fetal rats, all four tissues con- tained a major mRNA of 1.7 kilobases (kb) that hybridized with the human Sm-CflGF-I cDNA and mRNAs of 7.5, 4.7, 1.7, and 1.2 kb that hybridized with the mouse Sm-C/IGF-I cDNA. Adult rat intestine, liver, and lung also contained these mRNAs but Sm-C/ IGF-I mRNAs were not detected in adult rat brain. These findings provide direct support for prior

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Recipient of a Career Development Award from the Juvenile Diabetes Foundation during the course of these studies. To whom reprint requests should be addressed. suggest that a role for IGF-I1 in the adult rat, particularly in the central nervous system, cannot be excluded.
T h e s o m a t o m~n s or insulin-like growth factors (Sm/ IGFs') are peptides that are structurally homologous with proinsulin and that stimulate cellular proliferation and the production of differentiated cell products in a wide range of cell types (I). Two somatomedins, somatomedin-C or insulinlike growth factor I (Sm-C/IGF-I) and insulin-like growth factor I1 (IGF-IJ), have been purified from human plasma (2)(3)(4). Counterparts of these somatomedins have been identified and characterized in the rat (5)(6)(7). Evidence that somatomedins may play a role in fetal development (reviewed in 8 and 9) is based on their ability to stimulate mitosis of fetal cells, their binding to specific receptors in fetal tissues, and their secretion by cultured fetal cells and explants. In the rat, IGF-I1 (also termed multiplication-stimulating activity) concentrations in fetal serum are 20-100-fold higher than in maternal serum and decline within days after birth (10). In contrast, serum concentrations of Sm-C/IGF-I are low in the rat fetus and rise in the immediate postnatal period (11). Serum concentrations, however, may not reflect the production of these peptides in specific tissues. It also may not be appropriate to extrapolate data obtained under culture conditions to the in vivo situation. Although immunoreactive Sm-CIIGF-I is secreted in vitro by cultured cells and organ explants from a variety of fetal mouse tissues (E?), the amount secreted as well as responsivity of fetal cells to exogenous somatomedin varies with culture conditions (13). Thus, the role of the different somatomedins in fetal development is not clear and information about their precise sites of synthesis in the fetus in vivo is lacking.
In the present study we have investigated the mRNAs encoding Sm-C/IGF-I and IGF-I1 in several fetal and adult rat tissues to determine directly whether there is synthesis of somatomedins during development. cDNAs encoding human Sm-C/IGF-I (14), mouse Sm-C/IGF-I (provided by Dr. G. Bell, Chiron Corp., Emeryville, CA) and human IGF-I1 (15) were used as probes to analyze corresponding rat mRNAs. The structure of the rat Sm-C/IGF-I mRNAs is currently not known, and the studies reported here provide information about the size and abundance of Sm-C/IGF-I mRNAs in fetal and adult rat tissues. The structures of three rat IGF-I1 cDNAs, derived from polyadenylated (poly(A+)) RNA from The abbreviations used are: Sm/IGFs, somatomedin insulin-like growth factors; kb, kilobases; SDS, sodium dodecyl sulfate; UT, untranslated region.

14539
the Buffalo rat liver cell line BRL-3A, were reported recently (16)(17)(18). While each of these cDNA sequences are very similar in the regions encoding the IGF-I1 protein precursor, differences in the reported 5"untranslated regions (17-18) indicate the existence of multiple rat IGF-I1 mRNAs. Bento-Soares et al. (18) observed multiple mRNAs in neonatal rat tissues that hybridized to a rat IGF-I1 cDNA probe and provided evidence that their characterized cDNA corresponds to a 3.4-kb IGF-I1 mRNA. In their study the 3.4-kb mRNA was detected in neonatal but not adult tissues. The present study provides further information about the expression of IGF-I1 mRNAs in rat fetal and adult tissues.

EXPERIMENTAL PROCEDURES
Animals and Tissues for RNA Extraction-Sprague-Dawley pregnant female rats and mature 200-250 g male rats were purchased from Charles River Breeding Laboratories. Pregnant female rats were killed by decapitation, fetuses were removed, killed by decapitation, and fetal tissues dissected for RNA extraction. Gestational age of fetal rats was timed as days postcoitum for the pregnant females. Male rats also were killed by decapitation and used as a source of adult rat tissues for RNA extraction. Immediately after collection, tissues were frozen in liquid nitrogen and then stored at -100 "C until use.
cDNA Probes-cDNAs encoding human Sm-C/IGF-I (14). mouse Sm-C/IGF-I, and a human IGF-I1 variant (15) were used as probes to analyze rat somatomedin genomic DNA fragments and mRNAs. The human and mouse Sm-C/IGF-I cDNA sequences are 94% homologous in coding regions with very limited homology in the 5'untranslated (UT) regions. A short 3' UT region of the mouse cDNA shows 78% homology with a corresponding region of the longer 3' UT of the human probe? The coding sequences of the human IGF-I1 cDNA (15) and published rat IGF-I1 cDNA sequences are 87% homologous but share very limited homology in UT regions (16)(17)(18). cDNA inserts were labeled with 3zP by nick translation (19) to specific activities of 1o7-1OS cpmlpg.
Poly ( Total RNA was pelleted by centrifugation of the homogenate over a cushion of 5.7 M CsCl as described by Ullrich et al. (23). The RNA pellet was dissolved by heating to 65 "C in high salt buffer (0.5 M NaCI, 10 mM Tris-HC1, pH 7.0, 0.25% SDS). Poly(A+) RNA was purified from total RNA by a modification of the oligo(dT)-cellulose affinity chromatography procedure of Aviv and Leder (24). Briefly, total RNA in high salt buffer was incubated batch fashion with oligo(dT)-cellulose (Type 3, Collaborative Research; 0.5 g of cellulose added to total RNA from 1 g of tissue) for 1 h at room temperature. The cellulose suspension was then poured into a sterile 2-ml polypropylene dispo-column (Bio-Rad). The high salt buffer flow-through was collected. The cellulose was washed in 50 ml of high salt buffer at room temperature or until absorbance at 260 nm of 1 ml of collected high salt eluate was returned to base line. Poly(A+) RNA was eluted from the oligo(dT)-cellulose column by washing with low salt buffer (10 mM Tris-HC1, pH 7.0, 0.125% SDS) heated to 68 "C. Six low-salt fractions of 1.5 ml were collected, and absorbance at 260 nm was measured to monitor elution of poly(A+) RNA. Routinely, poly(A+) * G. Bell, personal communication.
RNA eluted in the first three fractions. Low salt fractions containing poly(A+) RNA were pooled, and a one-twentieth volume of 4 M potassium acetate, pH 5.5, and three volumes of 95% ethanol were added. Poly(A+) RNA was dissolved in sterile water to give a final concentration of 0.5 pglpl based on absorbance at 260 nm (absorbance of 1 = 40 pg/ml RNA (25)). Recovery of RNA was assessed by comparison of absorbance of 260 nm of total RNA pelleted over CsCl with the sum of absorbances at 260 nm of the flow-through from the oligo(dT)-cellulose column and the eluted poly(A+) RNA. Recoveries varied between 60 and 105%. Analyses of Poly(A+) RNAs-Poly(A+) RNAs and unlabeled Sm-C/ IGF-I and IGF-I1 cDNA inserts (standards) were denatured in glyoxal and dimethyl sulfoxide and size-fractionated on 1% agarose gels as described by Thomas (26). After electrophoresis, samples were transferred to Genescreen (New England Nuclear) by the blotting method of Southern (21). Hybridization of Northern blots with 32P-labeled Sm/IGF cDNAs was as described above for genomic blots. Washing and autoradiography was also as described for genomic blots except that the final 0.1X SSC wash was at 60 'C. Relative abundance of Sm/IGF mRNAs was estimated by densitometric scanning of autoradiograms (DARWIN densitometry system, Darwin Instruments, Inc., Winston-Salem, NC). Purified Sm/IGF cDNA inserts of known concentration, electrophoresed on the same gels as poly(A+) RNA samples, provided internal standards to allow comparison of the relative amounts of IGF mRNAs on different blots. Analyses were performed on poly(A+) RNAs because Sm/IGF mRNAs are too low in abundance to be quantified in total RNA. One cycle of oligo(dT)cellulose chromatography enriches for poly(A+) RNA but does not remove all ribosomal RNA. Poly(A+) RNA samples used for quantitative Northern blot analyses were therefore analyzed further by hybridization with a 32P-labeled cDNA probe encoding human ubiquitin (27) to establish that, for all poly(A+) RNA preparations compared, the amounts of RNA loaded onto blots and the degree of contamination with ribosomal RNA were comparable (Fig. 2). Abundance of Sm/IGF mRNAs were assessed as picogram per microgram of poly(A+) RNA rather than as abundance per gram of starting tissue because of variability in RNA recovery among preparations.

Rat Sm-CIIGF-I and IGF-11 Genomic DNA Fragments-
Southern blots of rat genomic DNA digested with a number of restriction endonucleases revealed that the sizes of the rat genomic DNA fragments that hybridized with the human and mouse SmTC/IGF-I cDNAs were the same and differed from those that hybridized with the human IGF-I1 cDNA (Fig. 1).

Sm-CIIGF-I mRNAs in Rat Fetal and Adult Tissues-
Poly(A+) RNAs from adult rat intestine, liver, lung, and brain, and from fetal tissues between day 14 and day 17 of gestation, were analyzed by Northern blot hybridization with 32P-labeled fragments of X DNA that were used as molecular weight markers. Lanes 2, 5, and 8 show rat genomic fragments that hybridized with the mouse Sm-C/IGF-I cDNA; Lanes 3, 6, 9, 11, and 13 show fragments that hybridized with the human Sm-C/IGF-I cDNA; Lanes 4, 7, 10, 12, and 14 show fragments that hybridized with the human IGF-I1 cDNA. Exposure time for the autoradiogram was 72 h at -100 "C with intensifying screens.
human and mouse Sm-C/IGF-I cDNAs. Hybridization with the human Sm-C/IGF-I cDNA revealed a major mRNA of estimated size 1.7 f 0.05 kb (mean and standard deviation of estimated size from six different blots) in poly(A+) RNAs from all rat fetal tissues (Fig. 2, toppanels). Using the human Sm-C/IGF-I cDNA, the 1.7-kb mRNA was found also to be the major hybridizing mRNA in adult intestine, liver, and lung. For adult brain, it is not clear whether the mRNA is absent or below the detection limit of the hybridization procedures.
The abundance of the 1.7-kb mRNA was assessed by densitometric scanning of the hybridization signals and comparison with the signal intensities obtained for different amounts of unlabeled human Sm-C/IGF-I cDNA insert on the same blots. Abundance of the 1.7-kb mRNA showed no dramatic differences in the different fetal and adult tissues (Fig. 3, top  panel). Levels were, however, consistently higher in poly(A+) RNAs from fetal than from adult tissues, especially in the brain where the 1.7-kb mRNA was not detectable. Poly(A+) RNAs from intestine and liver of both fetal and adult rats showed consistently higher levels of the 1.7-kb mRNA than those from lung and brain.
Hybridization of poly(A+) RNAs from fetal and adult rat tissues with the mouse Sm-C/IGF-I cDNA probe revealed a different pattern of hybridizing mRNAs than obtained with the human probe (Fig. 2, middle panels). Several  The relative abundance of the mRNAs that hybridized to the mouse Sm-C/IGF-I cDNA was estimated by densitometric scanning of the 7.5-kb mRNA (Fig. 3, bottom panel). No major differences in abundance of the 7.5-kb mRNA were observed in poly(A+) RNAs from fetal liver, lung, intestine, or brain although levels were consistently higher in intestine than in other tissues. Poly(A+) RNAs from adult intestine, brain, and lung showed no major differences in abundance of the 7.5-kb mRNA compared with fetal tissues. Adult liver poly(A+) RNAs, however, contained 10-25-fold higher levels of the 7.5kb mRNA than those from fetal liver and from other adult tissues. The abundance of Sm-C/IGF-I mRNAs was compared only in poly(A+) RNAs that showed similar intensities of hybridizing ubiquitin mRNAs as illustrated in the bottom panels of Fig. 3. This ubiquitin control served to validate that similar amounts of poly(A+) RNAs were applied to blots.

IGF-11 mRNAs in Rat Fetal and Adult
Tissues-Multiple mRNAs that hybridized to the human IGF-I1 cDNA probe were found in poly(A+) RNAs from fetal intestine, liver, lung, and brain (Fig. 4A). The estimated sizes of the five distinct IGF-I1 mRNAs detected in fetal tissues were 4.7 & 0.6.3.9 & 0.3, 2.2 f 0.2, 1.75 f 0.06, and 1.2 f 0.06 kb (mean + S.D. of estimates from six blots). The 4.7-and 3.9-kb IGF-I1 were detected in poly(A+) RNAs from all fetal tissues. The signal intensities of the other IGF-I1 mRNAs were more variable in poly(A+) RNAs from the same fetal tissues at different times of gestation (Fig. 4A) and even at the same time (data not shown). Because of this variability, no attempt was made to quantify the relative abundance of the different IGF-I1 mRNAs. Some general trends did emerge, however, from densitometric scanning of the IGF-I1 mRNAs detected in six liver, four lung, nine brain, and three intestine poly(A+) RNA preparations, each prepared from different samples of pooled fetal tissues a t day 14-18 of gestation. The abundance of IGF-I1 mRNAs was consistently higher in poly(A+) RNAs from fetal intestine and liver than from lung and brain. The 4.7and 3.9-kb mRNAs were the major hybridizing mRNAs and the 2.2-kb mRNA was the minor one in all fetal tissues at all times. Fetal intestine poly(A+) RNAs contained the highest proportion of the 1.75-kb mRNA, and fetal lung contained the highest proportion of the 1.2-kb mRNA.
IGF-I1 mRNAs were detected in some but not all of the poly(A+) RNA preparations from adult rat tissues. The brain was the only adult tissue in which IGF-I1 mRNAs were consistently detected (eight different brain poly(A+) RNA preparations were analyzed, and IGF-I1 mRNAs were present in each). The 3.9-kb mRNA was the major hybridizing mRNA and the 4.7-kb mRNA was present (Fig. 4B). The 4.7-and 3.9-kb IGF-I1 mRNAs were detected in five of six poly(A+) RNA preparations from adult intestine but with signal intensities lower than for brain (Fig. 4B shows an example). Some poly(A+) RNAs from adult liver and lung (four of eight liver preparations and two of eight lung preparations) contained a barely discernible hybridizing band at 4.7 kb (Fig. 4B). As long exposure times were necessary to see these hybridization signals, the possibility cannot be ruled out that these signals above the poly(A+) RNA lanes correspond to the time in fetal development between day 14 and 17 for the RNA samples shown. Arrows at the right point to five distinct hybridizing IGF-I1 mRNAs which were consistently observed in fetal poly(A+) RNAs. Numbers at the right are size estimates (kb) for the rat IGF-I1 mRNAs. There is an additional hybridizing mRNA of estimated size 3.1 kb in the fetal liver mRNA preparations shown, and these were the only poly(A+) RNA preparations that contained this mRNA. Exposure of the autoradiogram was for 16 h at -100 "C with intensifying screens. B, shows hybridization of the human IGF-I1 cDNA probe with IGF-I1 mRNAs in adult rat tissues. M = labeled molecular weight markers (sizes in kb at lejt); 0.05,0.01,0.005 represent ng of unlabeled human IGF-I1 cDNA insert; the other four lanes correspond to 40 pg of poly(A+) RNA from adult rat tissues: I = intestine, Li = liver, L = lung, and B = brain. Numbers at left correspond to the sizes of the hybridizing IGF-I1 mRNAs (kb). Exposure of autoradiography was for 48 h at -100 "C with intensifying screens.
were attributable to weak cross-hybridization of probe with 28 S ribosomal RNA which migrates closely to the 4.7-kb IGF-I1 mRNA.

DISCUSSION
Human and mouse cDNAs-encoding Sm-C/IGF-I and a human cDNA-encoding IGF-I1 were used to analyze mRNAs encoding these growth factors in poly(A+) RNAs from rat tissues. Rat Sm-C/IGF-I mRNAs have not been characterized but there is marked homology between Sm-C/IGF-I and IGF-I1 at the protein level (1). To assess whether there was crosshybridization of Sm-C/IGF-I cDNAs with IGF-11-coding se-quences, the human and mouse Sm-C/IGF-I probes were tested initially in hybridizations with rat genomic DNA. Sm-C/IGF-I cDNAs hybridized with different genomic fragments than did the human IGF-I1 probe, and thus provided evidence that there was not cross-hybridization of Sm-C/IGF-I cDNAs with rat IGF-11-coding sequences. The hybridization of both human and mouse cDNAs with the same rat genomic DNA fragments provide evidence for specific hybridization of these probes with rat Sm-C/IGF-I-coding sequences.
Mouse and human Sm-C/IGF-I probes both hybridized with mRNAs from fetal intestine, liver, lung, and brain and from adult intestine, liver, and lung but not brain. The mouse probe hybridized with several mRNAs of estimated sizes 7.5, 4.7, 1.7, and 1.2 kb. The human probe hybridized with a major 1.7-kb mRNA, and with long exposures the 7.5-and 4.7-kb mRNAs were also detected. Because our data on hybridization with rat genomic digests do not establish the number of rat Sm-C/IGF-I genes, the multiple rat Sm-C/IGF-I mRNAs detected with the mouse and human Sm-C/IGF-I cDNAs could be either variants in splicing of a single Sm-C/IGF-I gene transcript or products of different genes.
Our finding of different patterns of hybridizing rat Sm-C/ IGF-I mRNAs with the mouse cDNA compared with the human cDNA was surprising as these probes are 94% homologous within the coding sequences. The 1.7-kb mRNAs detected with the human and mouse Sm-C/fGF-I cDNAs are probably not the same mRNAs as they differ in their relative hybridization signal intensities in the same poly(A+) RNAs from fetal and adult rat tissues (Fig. 3). We have preliminary evidence that the major 1.7-kb mRNA in fetal and adult rat tissues is recognized by the 3' UT region of the human Sm-C/IGF-I cDNA based on hybridization with a 237-base pairs BarnHI/PstI fragment corresponding entirely to 3' UT (data not shown). The 3' UT of the human Sm-C/IGF-I cDNA is longer than that of the mouse cDNA and there is only 78% homology between the corresponding regions? The possibility cannot be excluded, therefore, that the major 1.7-kb mRNA detected in fetal and adult rat tissues with the human Sm-C/ IGF-I probe is a gene transcript unrelated to Sm-C/IGF-I that shows fortuitous cross-hybridization with the 3' UT of the human probe. Evidence against this possibility, however, is provided by the hybridization of human and mouse Sm-C/ IGF-I cDNAs with the same rat genomic fragments. Different and/or additional genomic fragments should be detected with the human Sm-C/IGF-I cDNA if the 1.7-kb mRNA detected with this probe corresponds to a transcript of a gene other than rat Sm-C/IGF-I. As the structural relationships among the multiple mRNAs detected with the mouse and human Sm-C/IGF-I cDNAs will not be established until information is available about the sequences of cloned rat Sm-C/IGF-I cDNAs and genomic fragments, the term Sm-C/IGF-I-related mRNAs has been used for further discussion of the multiple mRNAs detected here in poly(A+) RNAs from rat tissues using human or mouse Sm-C/IGF-I cDNA probes.
D'Ercole et al. (12) reported that explant cultures of fetal mouse intestine, liver, lung, brain, and other tissues secreted significant amounts of immunoreactive Sm-CIIGF-I into media. The observations here of Sm-C/IGF-I-related mRNAs in fetal rat intestine, liver, lung, and brain provide direct support for the conclusion that multiple fetal tissues synthesize Sm-C/IGF-I. Our findings that adult rat liver, lung, and intestine synthesize Sm-C/IGF-I-related mRNAs are also in agreement with another report (28) that immunoreactive Sm-C/IGF-I is produced in multiple tissues in the adult rat. The abundance of the 7.5-kb Sm-C/IGF-I-related mRNA detected with the mouse cDNA probe was 10-50-fold higher in poly(A+) RNAs from adult rat liver than other adult tissues. These data provide further evidence for the contention that the liver is a major site of synthesis of Sm-C/IGF-I and of circulating Sm-C/IGF-I in the adult rat (28,29). To our knowledge, immunoreactive Sm-C/IGF-I concentrations in rat intestine have not been reported previously. Our findings of Sm-C/IGF-Irelated mRNAs in adult as well as fetal rat intestine suggest that rat intestine is a site of Sm-C/IGF-I synthesis. AS Sm-C/IGF-I is a mitogen (1) and intestinal mucosa is characterized by high rates of cell division (30), further investigations of the biological role of Sm-C/IGF-I in intestine may be of significance. Sm-C/IGF-I mRNAs were not consistently detected in adult rat brain in the present study, whereas low levels of immunoreactive Sm-C/IGF-I were found previously in rat brain (28). This suggests that the abundance of Sm-C/ IGF-I mRNAs in rat brain are below the detection limit of the blot hybridization procedures employed.
Five distinct IGF-I1 mRNAs of estimated sizes 4.7, 3.9, 2.2, 1.75, and 1.2 kb were found in poly(A+) RNAs from rat fetal intestine, liver, lung, and brain. These observations suggest that synthesis of IGF-I1 by a number of tissues in the fetal rat contributes to the high levels of IGF-I1 found in fetal rat serum (10,11). The IGF-I1 mRNAs found in fetal rat tissues are similar in size to those found by  in neonatal rat tissues using a rat IGF-I1 cDNA probe. The multiple rat IGF-I1 mRNAs may be either splicing variants of a single rat IGF-I1 gene transcript or derived from different IGF-I1 genes. An understanding of the role of IGF-I1 in rat fetal development will require information about the structures of the different IGF-11 mRNAs and the factors that regulate their production.
Poly(A+) RNAs from adult rat intestine, liver, and lung contained lower levels of IGF-I1 mRNAs than poly(A+) RNAs from the same fetal tissues. Some adult poly(A+) RNA preparations contained no detectable IGF-TI mRNA. These findings are consistent with previous observations that serum concentrations of IGF-I1 are very low in adult rats (10,11). When IGF-11 mRNAs were detected in adult tissues, only the 4.7-kb or 4.7-and 3.9-kb mRNAs were detected. This is in contrast to the findings of  that adult rat liver contained a 1.75/1.6-kb mRNA doublet that hybridized with the rat IGF-I1 cDNA. As both rat and human IGF-I1 cDNAs show hybridization with IGF-I1 mRNAs of similar size (181, it seems unlikely that probe differences account for this discrepancy. One possibility is that the rat IGF-I1 cDNA used by Bento-Soares et al. cross-hybridizes with 1.7-kb Sm-C/IGF-I-related mRNAs that are present in high abundance in adult rat liver. The 4.7-and 3.9-kb IGF-I1 mRNAs were consistently detected in poly(A+) RNAs from adult rat brain, whereas Sm-C/IGF-I mRNAs were not consistently detectable. These findings in adult rat brain are in agreement with previous reports of higher levels of IGF-I1 than Sm-C/IGF-I in adult human brain (31). Receptors for IGFs have been reported in human brain (32,33) and IGF-I1 has been shown to stimulate neurite o u t~o~h in human neurob~astoma cells (34). Our findings, that Sm-C/IGF-I and IGF-I1 mRNAs are produced in fetal rat brain and that IGF-I1 mRNA is produced in adult rat brain, suggest that investigations of the cellular sites of synthesis of brain Sm/IGFs and of the factors that regulate synthesis at different developmental stages, may provide clues as to the role of the Sm/IGFs in the central nervous system.