Histochemical localization of IGF-I and IGF-II mRNA in the rat between birth and adulthood

Insulin-like growth factors (somatomedins) are mitogenic peptides of 7·5×10³M_r which have been unequivocally demonstrated in a variety of embryonic, fetal, postnatal and adult mammalian tissues. Their primary structure in the human has been known for 10 years (Rinderknecht & Humbel, 1978a,b). In general terms, insulin-like growth factor II (IGF-II) is preponderant during early development whilst insulin-like growth factor I (IGF-I) is characteristic of postweaning childhood and of adult tissues (Sara & Hall, 1984; Graham et al. 1986; Brown et al. 1986; Beck et al. 1987; Murphy et al. 1987). In early studies, IGF-1 and -II were demonstrated by radioimmuno and radioreceptor assay. Some discrepancies in published work have clearly arisen from the relative sensitivities of the various assay procedures used, the ability of IGFs to cross-react with each other’s receptors, their relationship to insulin, their species specificity and the fact that many observations were based on cultured cells derived from normal and abnormal lines.

More recently, following the isolation of cDNA clones, the human and rat IGF-II genes have been defined (Dull et al. 1984; Irminger et al. 1987; Frunzio et al. 1986; Soares et al. 1986) and the human and rat IGF-I genes have also been characterized (Ullrich et al. 1984; Bell et al. 1985; Rotwein et al. 1986; Shimatsu & Rotwein, 1987). cDNA and oligonucleotide probes have been constructed which have made it possible to demonstrate somatomedin mRNAs by Northern blotting and, at a cellular level, by in situ hybridization (Han et al. 1987; Beck et al. 1987). As a result, multiple molecular weight species of pre-pro-IGF-I and -II mRNA have been defined in both human and rat tissues, though there is some disagreement in the literature as to the exact sizes of the various transcripts (Lund et al. 1986; Beck et al. 1987; Graham et al. 1986; Brown et al. 1986; Soares et al. 1985; Frunzio et al. 1986; Soares et al. 1986). For IGF-II, some of this heterogeneity has been shown to be because the gene is transcribed from two different promoters, generating two classes of molecules with different 5 ′ (untranslated) exons (Irminger et al. 1987). Preliminary results in the rat also suggest that different 3 ′ extensions arise through the use of alternative 3 ′ cleavage/polyadenylation sites (Frunzio et al. 1986; Stylianopoulou et al. 1988). Multiple mechanisms, including alternative RNA splicing and alternative polyadenylation, are involved in the generation of IGF-I mRNAs (Bell et al. 1986; Rotwein et al. 1986).

The regulation of hepatic IGF-I secretion is well understood. The growth factor, as well as its larger carrier protein (M_r 150×10³), is secreted in response to pituitary growth hormone, and serum levels in man rise greatly when somatic growth becomes growth hormone regulated (Sara & Hall, 1984). By contrast, the regulation of IGF-II production and the smaller (M_r 40 –60×10³) serum carrier protein has not been established. It may be that its biosynthesis is controlled largely by substrate availability, although it has been shown that placental lactogen stimulates IGF-II synthesis in fetal fibroblasts (Adams et al. 1983) .

When insulin-like growth factors were described, some confusion existed concerning nomenclature but this has been rationalized by Doughady et al. (1987). It is now clear that human and rat IGF-II are highly homologous and a substance termed multiplication stimulation activity (MSA) secreted by the BRL-3A buffalo rat liver cell line is generally accepted to be rat IGF-II (Marquardt et al. 1981). These polypeptides consist of 67 amino acid residues (Rinderknecht & Humbel, 1978b) and their sequences share amino acids in 45 positions with IGF-I, which consists of 70 amino acid residues (Rinderknecht & Humbel, 1978a). Human and mouse IGF-I are also highly homologous with each other and with a more recently identified rat IGF-I (Rubin et al. 1982). Somatomedin C is a basic peptide derived from human serum which probably contains an extension at the C terminus and consequently has a molecular size slightly larger than that of IGF-I (Hall & Sara, 1983) but is otherwise identical with it. Somatomedin A is also an IGF-I-related peptide (Zumstein et al. 1983; Engberg et al. 1984) . The B and A domains of the IGFs are 40% identical to the B and A chains of human insulin (Hall & Sara, 1983). Comprehensive reviews of the physiological actions of insulin-like growth factors during development have been published by Underwood & d’Ercole (1984) and by Gluckman (1986).

Using oligonucleotide probes, we have recently described the localization of IGF-I and -II mRNAs in the developing rat embryo and fetus by Northern blotting and in situ hybridization (Beck et al. 1987). We now describe the postnatal distribution of these mRNAs in various rat tissues and note, in particular, the temporal differences in the control of IGF-II mRNA transcription between liver, muscle and brain (choroid plexus).

Rats were killed by cervical dislocation on the day of birth and at 4, 10, 14, 17, 18, 20, 23 and 34 days postnatally. Tissues from adult animals were also studied. Segments of the abdomen and brain (or pieces of liver, kidney, pancreas and brain in the case of adult animals) were rapidly frozen in OCT compound at −70 °C and 5 μm cryostat sections were fixed, prehybridized and stored in ethanol vapour at 4 °C using the methods described by Beck et al. (1987).

The IGF-II message was located by a 30 mer oligodeoxyribonucleotide probe complementary to the C-terminal region of the peptide of pre-pro-IGF-II mRNA. The sequence (5 ′-CTG ATG GTT G CT GGA CAT CTC CGA AGA GGC-3 ′) is complementary to the mRNA for the amino acid sequence 147 to 156 of the rat precursor protein (Whitfield et al. 1984). For IGF-I mRNA a 30 mer probe (5 ′-CTC AGC CCC GCA AAG GGT CTC TGG TCC AGC-3 ′) complementary to the mRNA sequence encoding amino acid −1 to +9 of mouse pre-pro-IGF-I (prepeptide and B domain) was used (Bell et al. 1986). The sequences were selected to eliminate crossreactivity; they were synthesized by the solid-phase phosphoramidite procedure (Caruthers et al. 1982) and, for each, a ‘sense’ probe was also made to serve as a control in the in situ hybridization experiments: 100 ng samples of each probe were 5 ′ end labelled with γ³²P-ATP as previously described (Beck et al. 1987).

Sections were hybridized with the appropriate probes for 24 h following a previously published method (Beck et al. 1987). At each stage of development, the corresponding ‘sense’ probe and an independently tested 30 mer rat insulin probe were used as controls.

Slides were exposed to Kodak CX-Omat AR film for 18 h or to Cronex high resolution (Dupont) X-ray film for 72 h in an X-ray cassette without an intensification screen. Subsequently, autoradiography was carried out using Ilford K5 emulsion (Beck et al. 1987) and dark-field photography with matching bright-field exposures were obtained using a Leitz diaplan photomicroscope.

Tissues were removed from animals immediately after sacrifice, frozen in liquid nitrogen and stored at −70 °C. Total tissue RNA was prepared using a modification of the method of Auffray & Rougeon (1980) as previously described (Beck et al. 1987). RNA concentrations were determined from absorbance measurements at 260 nm and checked by comparing ethidium bromide fluorescence of samples run on agarose gels.

RNA samples (100 μg each) were electrophoresed on a 1·2 % agarose gel containing formaldehyde (Maniatis et al. 1982) and transferred to Hybond N (Amersham, UK). Hybridization to 5 ′-end-labelled IGF-I and IGF-I1 probes and autoradiography were carried out as previously described (Beck et al. 1987). Molecular weights of bands were estimated from a concurrently run RNA molecular weight ladder (BRL, UK).

RNA samples from the following tissues and postnatal ages were examined: liver (days 4, 10, 17, 20, 23, adult); brain and muscle (days 10, 17, 20, 23, adult) and kidney (day 10).

Fig. 1 shows the results obtained with the IGF-I probe. Four distinct mRNA species (7·5, 4·7, 2·0 and 1·2 kb) were detected in the liver with a progressive increase in the relative amounts of all species from the 4-day-old neonate to the adult rat. Comparatively much smaller quantities of all four species were also seen in the muscle (but without any progressive change with age) and in the kidney of the 10-day-old rat. Distinct IGF-I transcripts were not seen in the brain at any age in the time allowed for exposure, though there was a suggestion of activity at the early stages examined, particularly on days 4 and 10.

Transcripts of sizes 4·0, 2·4, 1·75 and 1·25 kb were detected in the liver and muscle by the IGF-II probe (Fig. 2). In both the liver and muscle, the highest levels were observed in the youngest rats examined (4-day-old rat for the liver and 10-day-old rat for muscle). Furthermore, in both tissues there was a coordinate decline in levels of all species with age. However, an important difference between the two was observed: extinction of the transcripts was seen to occur in the muscle later than their disappearance from the liver (Fig. 2). In the brain only the 4·0 kb transcript was clearly seen in the time allowed for exposure and there was no age-related change in expression. No transcripts were seen in the kidney of the 10-day-old rat.

In the neonate tissue, expression of IGF-I and -11 mRNAs is similar to that previously described for the fetal rat (Beck et al. 1987). Liver, skeletal muscle, perichondrium, periosteum (Figs 3, 4) and the sclera of the eye hybridize strongly for IGF-II mRNA. The pia mater and choroid plexuses are also strongly positive (Fig. 5), but all other ectodermally derived structures (CNS, PNS, skin, epithelium etc.) are negative. A very low level of hybridization is demonstrable in some loose connective tissues (e.g. the cores of the intestinal villi) but the bronchial and gut epithelia, which showed activity in the fetal period, are now negative. In the oesophagus, it is interesting to note that the striated component of the muscularis externa is positive whilst the smooth muscle that constitutes the same layer in the forestomach is negative (Fig. 6), as is all smooth muscle elsewhere. With the exception of the liver, the viscera after completion of histogenesis are negative. IgF-I mRNA is very faintly detectable on autoradiographs in the neonatal liver and just possibly in skeletal muscle and perichondrium.

The distribution of mRNAs for IGF-I and -II in selected tissues is shown in Table 1, at various ages postnatally and illustrated in Figs 7 and 8. In situ evidence for the existence of IGF-II mRNA in the liver disappears between 18 and 20 days postnatally but skeletal muscle in the sections continues to hybridize faintly with the probe for IGF-II mRNA until after 34 days (Fig. 9A). Here our results are at variance with Brown et al. (1986), who reported IGF-II mRNA to be undetectable in muscle by day 11 postnatally. Perichondria! cells and chondrocytes remain positive during growth but, when mitosis ceases, both become negative. During endochondrial ossification the zone of cartilaginous proliferation is positive but the zones of maturation and calcification are negative (Figs 10, 11). Osteocytes are negative and so are the osteoblasts of the inner layer of the periosteum. The spindle-shaped cells in the outer layer of the periosteum remained positive (though decreasingly so) at all the stages examined (Fig. 12) except in the adult.

Yet another pattern of expression is shown by pia mater and choroid plexus. Here hybridization occurred at a constant level at all ages investigated, including the adult (Fig. 13). Sections through neural tissue failed to hybridize with the probe for IGF-II mRNA throughout, as did all the major viscera as indicated by the results obtained from kidney and pancreas (Fig. 3).

In contrast to IGF-II mRNA, the probe for IGF-I mRNA produced unequivocal in situ hybridization in one organ only, namely the liver (Fig. 14). Here a low level of activity could be demonstrated from the neonatal period onwards; this became stronger at about 10 days and was stronger still in the adult. These results closely reflect our observations using Northern blotting (Fig. 12).

Control sections failed to hybridize with IGF-I and -Il ‘sense’ probes and gave a positive reaction over pancreatic islets only with the insulin probe.

The oligonucleotide probe for IGF-I mRNA is complementary to a conserved segment of mouse IGF-I mRNA and has been used for in situ hybridization in a previous study (Beck et al. 1987). It detected all the transcripts that were detected by Lund et al. (1986) in the rat using a full-length mouse IGF-I cDNA probe.

The sizes of the four IGF-II transcripts detected by our IGF-II oligonucleotide probe agree well with the major transcripts detected in the rat by Brown et al. (1986) using a rat IGF-11 cDN A probe. We could not clearly define the ∼6kb transcript seen by them, which they have subsequently measured more accurately to be about 5 kb (Frunzio et al. 1986), though our results suggest the presence of an additional transcript larger than 4 kb. We were also unable to separate a 3·8 kb transcript from the 4·0 kb transcript. We have achieved better fractionation than previously in the gels obtained in this study and have, therefore, altered our earlier estimate of the sizes of transcript (Beck et al. 1987).

For in situ hybridization, adequate tissue controls at each stage were provided by structures that did not hybridize with the probe and both positive and negative probe controls, made using an insulin probe and the ‘sense’ sequence of the IGF-I and -II probes, respectively, produced the expected results.

Graham et al. (1986) found strong signals for IGF-II in the liver of rats between 2 and 15 days old but none from 25- to 65-day-old animals. Our results show that IGF-II mRNA in the liver rather suddenly falls to below demonstrable levels between 18 and 20 days after birth. This timing corresponds precisely with a morphological and functional change in the epithelium of the small intestine which, over this period of time, loses its capacity for large-scale endocytosis of macromolecules and for transepithelial transport of intact IgG from ingested milk to the neonatal circulation (reviewed by Patt, 1977). This well-known phenomenon of gut ‘closure’ is coincident with a surge in plasma corticosterone levels (Walker et al. 1986) and can be precipitated at earlier stages of postnatal life by injection of cortisone acetate (Halliday, 1959; Clark, 1971; Carlile & Beck, 1983). This raises the possibility that glucocorticoids might also control the expression of IGF-II by the rat liver. We have observed (manuscript submitted) rapid and complete extinction of liver IGF-II mRNA following injection of cortisone acetate into 9-day-old rat pups. The region of the rat IGF-II gene between the two promoters EI(1) and EI(2) (Soares et al. 1986) contains two sequences with strong homology to the consensus glucocorticoid-responsive element (Jantzen et al. 1987). Although, in most cases, glucocorticoids act to induce gene expression, there are several well-established precedents for the repression of gene expression by glucocorticoids (Eberwine & Roberts, 1984; Turcotte et al. 1985; Camper et al. 1985; Siminoski et al. 1987). This repression can also be tissue specific: the rat gene for phosphoenolpyruvate carboxykinase is repressed by glucocorticoids in adipose tissue but induced in the liver and kidney (Nechushtan et al. 1987). It seems possible, therefore, that IGF-II mRNA expression in the liver comes to an end under natural conditions as the result of glucocorticoid action, whereas no such control is exerted in striated muscle or in the pia/ choroid plexus. In these latter tissues, IGF-II gene expression continues after 20 days and, in the case of the pia mater and choroid plexus, remains demonstrable in the adult.

The presence of IGF-II mRNA in cells of the skeletal system appears to be related to their developmental stage. It has been shown to be present in precartilaginous mesenchyme and in the perichondrium of growing cartilage (Beck et al. 1987). Our present observations indicate that it is present in chondrocytes which are actively proliferating but ceases to be demonstrable when cartilage has reached its adult morphology and is growing slowly or not at all. IGF-II is not demonstrable in osteoblasts or osteocytes but remains present in the outer periosteal layer of growing bone. It appears, therefore, that the role of IGF-II in skeletal maturation is closely related to cell proliferation. By contrast, the persistence of IGF-II in skeletal muscle until 34 days after birth suggests a relationship to maturation unconnected with mitosis. The differing correlation between IGF-II mRNA synthesis and cell division or differentiation in the various tissues studied indicate that it is probably an oversimplification to consider the localized production of the hormone in these terms.

IGF-II has been reported in cerebrospinal fluid (Sara et al. 1982; Hasselbacher & Humbel, 1982). Our previous findings (Beck et al. 1987) suggested that it is secreted by the choroid plexus and is thus able to cross the blood/brain barrier. Similar observations have recently been reported by Stylianopoulou et al. (1988) as well as Hynes et al. (1988). Its presence in the CSF, particularly in the adult, supports the suggestion that its action is not entirely confined to stimulating cell division, even though it undoubtedly promotes cell replication in a variety of cultured cells (Zapf et al. 1978). IGF-II may, therefore, have a role in the maintenance of nondividing cells in the central nervous system, even though it is not actively secreted by nervous tissue. The matter is further complicated by reports (Sara & Hall, 1984) of a distinct fetal IGF derived from cultures of fetal brain cells and only weakly cross-reacting with antisera against IGF-I and -II.

Our observations indicate a distribution of IGF-II mRNA in the neonate which closely corresponds to that previously described in the late fetal stages (Beck et al. 1987). We did not demonstrate in situ hybridization in a number of tissues for which other workers have obtained positive results, albeit at low levels, by Northern blotting. Thus Lund et al. (1986) found IGF-II mRNA in rat brain and Brown et al. (1985) described activity in the brain stem, cerebral cortex and hypothalamus, whilst Frunzio et al. (1986) quoted low amounts of IGF-II mRNA in adult rat thymus, heart, kidney, hypothalamus, brain stem and cerebral cortex. Murphy et al. (1987) showed that IGF-II mRNA expression in the adult rat was most abundant in the brain; expression in the heart was about 30% of this level whilst all the other tissues examined showed levels less than 10 % of those in the brain. In man Irminger et al. (1987) have shown the presence of IGF-II mRNA in the adult hypothalamus, adrenal gland and kidney. In the reports describing IGF-II mRNA activity in the brain, it is possible that pia mater and/or choroid plexus were present in the tissue examined since details of the isolation procedures were not given. In the other tissues, the low levels of activity determined by Northern blotting appear to be beyond the level of sensitivity of the in situ hybridization technique. Results obtained by others from human tissues, which are at variance with those reported here, may be due to species difference.

IGF-I mRNA activity is demonstrable in the liver from the neonatal period onwards by both Northern blotting and in situ hybridization. No evidence of in situ hybridization was obtained in any other tissue but Northern blotting indicated the presence of small amounts of message in muscle which is constant throughout the neonatal period and also in the kidney at 10 days (the only stage sampled). Although there are references to autocrine or paracrine secretion of IGF-I (Murphy et al. 1987; Han et al. 1987; Lund et al. 1986; Gluckman, 1986; d’Ercole et al. 1984; Rothstein, 1982), it is clear that this must occur at a low level. Our results agree with those of Murphy et al. (1987) who examined a variety of organs and showed by Northern blotting that IGF-I mRNA in peripheral tissues was less than 2 % of the level seen in the liver. Lund et al. (1986) did not detect IGF-I expression in the adult brain but found some in fetal rat brains between 14 and 17 days of gestation. Our observations on the early neonatal period suggests that IGF-I expression in the brain is largely switched off at, or soon after, birth.

Our results draw attention to the fact that IGF-I and -II mRNAs are independently regulated and that the increase of one is not temporally related to the disappearance of the other. Furthermore, although multiple (identical) IGF-II mRNAs are widespread in a variety of neonatal organs, the developmental control of IGF-II gene expression varies from tissue to tissue.

We are grateful to Dr G. W. Tregear, Professor J. P. Coghlan and the Howard Florey Institute, Melbourne, for the design and synthesis of the oligonucleotide probes, and to MRC and the Wellcome Trust for grants in aid of research. N.J.S. is supported by an MRC Clinical Training Fellowship.