INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Collagens are among the most abundant extracellular matrix proteins in vertebrate organisms. They maintain the structural integrity of tissues and mediate a wide variety of cell-matrix interactions. Type I collagen is a heterotrimer composed of two polypeptides encoded by the Col1A1 and Col1A2 genes. The transcriptional regulation of these two genes (reviewed in references 1, 10, 21, 28, 42, and 46) is of special interest because they are expressed at widely different levels which reflect the tissue-specific and developmental regulation of type I collagen synthesis. Furthermore, transcription of the Col1A1 and Col1A2 genes is responsive to cues generated by injury and repair and is modulated by a variety of cytokines, hormones, and pharmacological agents (1, 10, 28, 42, 46). Finally, the expression of type I collagen genes is disturbed in fibrotic disorders such as pulmonary fibrosis, cirrhosis, and scleroderma (13, 14, 18). Although both transcriptional and posttranscriptional mechanisms are involved in regulation, the concordance between mRNA levels and type I collagen synthesis suggests that the predominant mode of control is transcriptional (46).

Cell-specific expression of Col1A1 is conferred by a modular arrangement of promoter elements in the 5' flanking region of the gene (41, 45). The lethal phenotype of homozygous Mov 13 mice, which contain the Moloney murine leukemia retrovirus (MMLV) within the first intron of Col1A1, was the first indication that this intron might also be important in regulating expression of the gene (47). Subsequently, it was demonstrated that the retroviral insertion led to transcriptional inactivation of the gene in mouse embryo cell lines (24). It was also shown that the Col1A1 gene was transcribed in odontoblasts and osteoblasts and that in these cells, the first intron along with the integrated retrovirus was spliced out (30, 31, 48). These findings implied that incorrect splicing could not account for the lethality of the mutation in homozygous mice and that different cis-acting elements function in the regulation of Col1A1 in different cells. Since some of these elements could be placed in the first intron, numerous studies, using both transfection and transgenic approaches, have been designed to investigate transcriptional regulation by the first intron and to identify possible responsible regulatory elements, but these experiments have resulted in conflicting conclusions (reviewed in reference 8). While some studies have demonstrated a role for the intron in regulation and have identified cis-acting sequences that bind Sp1 and AP1 as important elements (6, 9, 29, 33-35, 44, 51, 52), other studies have indicated that the intron does not regulate expression of the gene (40, 53, 54). Thus, the role of the first intron in the regulation of the Col1A1 gene remains uncertain and controversial.

In recent years, the study of gene function has been assisted greatly by gene-targeting techniques (15, 38). Several methods for the introduction of mutations in the mouse genome have been described (3, 19, 23, 25, 55, 58, 59). Despite the potential advantages, relatively few studies have used these approaches to determine the role of cis-acting sequences (7, 20, 37, 50, 57). In this report, we describe the generation of a mouse model for the study of regulation mediated by the first intron of the Col1A1 gene. A double-replacement procedure (59), also termed tag and exchange (3), was used to create a mutated Col1A1 allele that contains a large deletion within the first intron and a silent mutation within exon 7; the latter generates a new XhoI restriction site. In heterozygous mice, reverse transcription (RT)-PCR followed by XhoI restriction analysis can be used to quantify the relative abundance of transcripts derived from the two alleles. In this study, we describe the approach and demonstrate that an intact first intron is required for normal transcription of Col1A1 mRNA in the lungs and muscle of adult mice.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Generation of mutant mice. Mouse 129 genomic clones, containing fragments of the Col1A1 gene, were kindly provided by H. Wu and were assembled to form a 12.8-kb EcoRI-SphI sequence. This sequence and two targeting constructs derived from it are shown in Fig. 1. In the Col-HPRT construct, the ~3-kb HPRT (hypoxanthine phosphoribosyltransferase) gene under the control of the PGK promoter (49) replaces an internal 4.1-kb SmaI fragment, creating a targeting construct with ~5.5 kb of 5' identity and ~3.2 kb of 3' identity with the endogenous allele. The Col-Int targeting construct contains a 1,283-bp deletion within the first intron and a single base alteration within exon 7 which generates an XhoI restriction endonuclease cleavage site. Although this change in exon 7 alters the genomic sequence and the sequence of the Col1A1 transcript, it preserves the identity of the amino acid residue at that position.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Map of the murine wild-type and mutant alleles and of the targeting constructs used in the double-replacement procedure. The relevant 14-kb EcoRI-EcoRI fragment of the Col1A1 allele, and its derivatives in the Col-HPRT and Col-Int alleles, are shown. The fragments from which probes 1 and 2 were derived and the sizes, in kilobases, of the restriction fragments (lines with double-headed arrows) which were used in the genotyping of ES cells and mice are also shown. Exons 1 to 7 are not drawn to scale, and exons 3' to exon 7 are omitted for the purpose of clarity. Locations of the restriction enzyme sites BamHI (B), EcoRI (E), SmaI (Sm), SphI (S), XhoI (X), and XbaI (Xb) are indicated.

View larger version (31K): [in this window] [in a new window]   FIG. 2.   Identification of targeted ES cell clones and procedure for determining the genotype of mice. (A) Representative Southern blot, hybridized with probe 1, showing EcoRI-restricted fragments diagnostic for wild-type (lane 1) and Col-HPRT-targeted (lane 2) ES cell clones. (B) Representative Southern blot hybridized with probes 1 and 2, showing EcoRI- and BamHI-restricted fragments diagnostic for wild-type (lane 1) and Col-Int-targeted (lane 2) ES cell clones. (C) A portion of the Col1A1 locus containing exons 6 to 9 (shown as rectangles) is illustrated. Also indicated are the locations of primers P1 and P1' (shown as arrowheads) and of the XhoI restriction site within the Col-Int allele. (D) Genotyping of wild-type (lane 1), heterozygous (lane 2), and homozygous (lane 3) mutant mice was accomplished by PCR analysis of DNA using primers P1 and P1'. After restriction with XhoI, the 750 bp of amplified genomic DNA gives rise to fragments of 520 and 230 bp, only if derived from the mutant allele.

COS cell culture, transfection, and PCR analysis. COS cells were cultured in high-glucose DMEM supplemented with 10% fetal bovine serum (GIBCO-BRL), 2 mM L-glutamine, penicillin G (100 U/ml), streptomycin, and nonessential amino acids (0.1 mM each; GIBCO-BRL); 1.5 × 10⁶ cells were transfected with 20 µg of DNA in a BRL Cell-Porator (800 µFa, 150 V). Stable transfectants were selected by means of cell culture in medium containing G418 (800 µg/ml), added 24 h after electroporation.

View larger version (54K): [in this window] [in a new window]   FIG. 3.   The mutation within the first intron does not hinder its capacity to be spliced correctly. (A) The relevant segment (exons 1 to 10 are shown as boxes) of the two alleles is shown, along with the positions of priming by P2 and P2', the deletion within intron 1, and the locations of the XhoI and EagI restriction enzyme sites. (B) RNA isolated from COS cells transfected with pcDNA3 (lane 5), wild-type Col1A1-pcDNA3 (lanes 1, 2), or Col-Int-pcDNA3 (lanes 3, 4) plasmid DNA was subjected to RT-PCR using primers P2 and P2'. Amplified DNA was electrophoresed on a 4% polyacrylamide gel and stained with ethidium bromide. Positions of the 200- to 600-bp DNAs in the size ladder (lane 6) are indicated.

Isolation of cells from lung and skin. Lung and back skin (taken from neonatal mice) were minced finely and were incubated for 30 min in a 2-mg/ml solution of collagenase (CLS1; Worthington Biochemical Corporation) at 37°C. Cells were cultured in high-glucose DMEM supplemented with 15% fetal bovine serum (GIBCO-BRL), 2 mM L-glutamine, penicillin G (100 U/ml), streptomycin (100 µg/ml), amphotericin B (Fungizone; 0.25 µg/ml), and nonessential amino acids (0.1 mM each; GIBCO-BRL).

Quantification of allele-specific expression. The StrataScript RT-PCR kit (Stratagene Cloning Systems) was used to perform RT-PCR on RNA extracted by the guanidinium thiocyanate method (16). For quantitation of allele-specific transcripts, 300 ng of primer P3' (5' CCGGGCTTGCCAGCTTCCCATCATC 3') was used for synthesis of first-strand cDNA from 5 µg of RNA; 5 µl of the reaction products containing cDNA (the volume used varied in the experiments reported in Fig. 6) was then subjected to 25 cycles of amplification at an annealing temperature of 65°C in the presence of 100 ng of primers P3 (5' CCACGCATGAGCCGAAGCTAACCCC 3') and P3' and 1 mM MgCl₂. For a determination of the relative abundance of the XhoI-resistant and XhoI-sensitive fragments (indicative of transcription from the wild-type and mutant alleles, respectively), [³²P]dCTP (1 µCi/sample) was added to the samples and a 26th amplification cycle was run. The addition of [³²P]dCTP prior to the final amplification cycle prevents erroneous estimates of transcript abundance attributable to heteroduplex formation between the two species of transcripts, since only DNAs synthesized during this cycle, which will not be heteroduplexed, will have incorporated the isotope and contribute to quantification. Labeled samples were restricted with XhoI, electrophoresed on a 4% acrylamide gel, dried, and subjected to phosphorimager analysis. Counts were determined for the 750- and 600-bp fragments only. To correct for the number of dCTP nucleotides in the smaller 150-bp XhoI fragment that was derived from the Col-Int allele, the counts obtained for the 600-bp DNA fragment were multiplied by 1.3. The values are expressed as a ratio of XhoI-sensitive fragment abundance relative to XhoI-resistant fragment abundance. Thus, a value of 1 is expected if the two alleles are transcribed equally well, and a value of 0.5 is expected if the mutant allele is transcribed half as well as the wild-type allele.

RNase protection. RNase protection assays were conducted by use of the Hyb-Speed RPA kit (Ambion). For these experiments, a 305-bp fragment of the murine Col1A1 cDNA containing exons 2, 3, and 4 and the first 42 bp of exon 5 was cloned into pBluescript KS (+); 400-bp [³²P]CTP labeled transcripts, containing the Col1A1 sequence, were generated from this clone by in vitro transcription with T3 RNA polymerase and were used in RNase protection experiments. Samples were simultaneously hybridized in the same tube with the Col1A1 probe and with a mouse -actin probe which was generated by using the clone provided in the kit. For each sample, the amount of Col1A1 transcript was normalized to the amount of protected -actin. Protected fragments were resolved on a 5% acrylamide gel and subjected to phosphorimager analysis for quantification.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Generation of Col-Int mice. The introduction of subtle mutations into a locus by double replacement, utilizing a mouse HPRT minigene/HPRT-deficient ES cell system, has been described previously (39). The replacement of one of the two Col1A1 alleles in E14TG2a ES cells with the Col-Int allele, which lacks 1,283 bp of the 1,462-bp first intron, was achieved in a similar manner (Fig. 1). Targeting of Col-HPRT to the Col1A1 locus was accomplished by selection for the presence of HPRT enzymatic activity, followed by the identification of homologous recombination events in ES clones by Southern analysis of DNA restricted with EcoRI (Fig. 2A). Whereas the Col1A1 allele yielded a hybridizing fragment of 14 kb (Fig. 2A, lane 1), the Col-HPRT allele gave rise to a 7.8-kb fragment (Fig. 2A, lane 2). Nine positive clones were identified among the 146 that were examined. A single target clone was expanded and utilized for the second step of the procedure, in which negative selection was used to accomplish the replacement of the Col-HPRT allele with the Col-Int allele (Fig. 1). For the identification of clones targeted by homologous recombination, ES cell clones were screened in two different ways. EcoRI- and XhoI-restricted DNA was hybridized with probe 1, which identifies a 13.2-kb hybridizing fragment from the Col1A1 allele and a 6.5-kb hybridizing fragment from the Col-Int allele (data not shown). In addition, DNA was restricted with EcoRI and BamHI and was hybridized with a mixture of probe 1 and probe 2. Under these conditions, wild-type ES cells yielded 7.8- and 5.6-kb fragments (Fig. 2B, lane 1), whereas targeted clones yielded hybridizing fragments of 7.8, 5.6, and 4.3 kb (Fig. 2B, lane 2). Genotyping of the mice was accomplished by PCR analysis of tail DNA with primers P1 and P1' (Fig. 2C). Whereas the wild-type allele gave rise to an XhoI-resistant 750-bp fragment (Fig. 2D, lanes 1 and 2), the corresponding fragment derived from the Col-Int allele was cleaved into 520- and 230-bp fragments (Fig. 2D, lanes 2 and 3).

The deleted intron is correctly spliced. We have generated mice bearing a 1,283-bp deletion within the first intron of the Col1A1 locus (Col-Int allele). Although splicing signals and sufficient intronic sequence were left intact, we wished to examine the capacity of the mutated first intron to be spliced correctly. For this purpose, we cloned a fragment of Col-Int, extending from the start of translation in exon 1 to an EagI site in exon 10, into the eukaryotic expression vector pcDNA3. A construct which contains the wild-type intronic sequence was generated for comparison. COS cells, which synthesize levels of Col1A1 transcript that are undetectable by Northern blot analysis (data not shown), were transfected with pcDNA3 and with two independently derived plasmid clones of each of the two constructs. As shown in Fig. 3B, RT-PCR using primers P2 and P2' amplified a 428-bp DNA fragment from both the wild-type (lanes 1 and 2)- and Col-Int (lanes 3 and 4)-transfected cells but not from cells transfected with pcDNA3 (lane 5). No additional fragment with a size of 606 bp, which would be expected if the deleted intron was not properly spliced, was detected in lanes 3 and 4. We therefore conclude that the deletion within the first intron does not affect its capacity to be spliced correctly.

Quantification of expression of Col1A1 and Col-Int in heterozygous mice. Col1A1 and Col-Int, the two alleles in a heterozygous mouse, are expected to produce transcripts that are identical in sequence except for the single base pair alteration which gives rise to the XhoI restriction enzyme cleavage site in the Col-Int allele. Furthermore, since the deletion in the first intron and the mutation are linked, all transcripts containing the XhoI site in exon 7 must be derived from the Col-Int allele. Thus, the relative abundance of XhoI-containing and XhoI-lacking transcripts should reflect the transcriptional activity of the two alleles. In this way, it should be possible to determine whether the first intron plays a regulatory role in the expression of the Col1A1 gene in different tissues.

View larger version (39K): [in this window] [in a new window]   FIG. 4.   Mixing experiments to demonstrate the accuracy of the RT-PCR/XhoI restriction digest assay for quantification of the ratio of abundance of Col-Int to Col1A1 transcripts. (A) The relevant segments (exons 1 to 10 are shown as boxes) of the two alleles are shown, along with the positions of priming by P3 and P3', the deletion within intron 1, and the location of XhoI within the Col-Int allele. (B) RT reaction was performed on RNA extracted from homozygous wild-type rib and mutant muscle. Assays were performed by mixing the RT reactions from the two genotypes in the denoted amounts (RT reactions were normalized to give equal activity per volume of the 750- and 600-bp fragments from wild-type and mutant reactions, respectively), amplifying by PCR, and digesting with XhoI. The observed ratios are compared to the ratios expected from the known compositions of the mixes.

Expression of the Col-Int allele in skin and muscle. The RT-PCR procedure described above was used to determine the ratio of allele-specific transcripts in skin and muscle of mice heterozygous for the Col-Int allele. The results indicate that the mutant allele transcript was as abundant as its wild-type counterpart in skin (Table 1). The age of the mouse did not seem to affect the level of expression of the mutant allele since the Col-Int/Col1A1 expression ratios were similar in mice ranging in age from 7 to 168 days (Table 1). Scattered determinations of this ratio in mice older than 6 months gave no indication of a reduction in this ratio (data not shown). The ratio of expression of the two alleles in cultured dermal cells derived from two different mice (1.03 ± 0.06 [n = 6] and 1.03 ± 0.07 [n = 3]) was also found to be in the range of that observed in skin. In contrast to skin, Col-Int/Col1A1 ratios in muscle showed a significant decline in older mice. Thus, the mutant and wild-type alleles were expressed equally well in young adults, but at some point between the ages of 2 and 6 months this ratio dropped to between 0.6 and 0.5 (Table 1).

View this table: [in this window] [in a new window]   TABLE 1.   Ratio of expression of Col-Int and Col1A1 alleles in skin and muscle

Expression of the Col-Int allele is reduced in the lung and is further reduced as a function of age. In contrast to skin and muscle, the level of expression of the mutant allele was considerably lower than that of the wild-type allele in lungs of young animals (Fig. 5 and 6). Furthermore, age was determined to influence the ratio of expression of the two alleles in this tissue (Fig. 5). Thus, the average of the Col-Int/Col1A1 ratios for nine premature mice aged 7 to 30 days was 0.74, that for two young adult mice aged 47 days was 0.58, and that for seven mature mice aged 82 to 254 days was 0.45. The correlation between age and ratio of expression in lung for the mice shown in Fig. 5, r = 0.80, was determined to be highly significant (P < 10⁴). Thus, it appears that in the lung, the influence of the intron on transcriptional regulation is evident at a young age and becomes more pronounced as the animals grow older. In three of the mice for which ratios of allelic expression in lung were determined to be 0.48, 0.52, and 0.62, the corresponding ratios in skin were 1.10, 1.08, and 1.03. Thus, it is clear that within the same animal, tissue-specific differences in the extent to which the deletion in the first intron influences allelic expression are observed. Interestingly, expression of the Col-Int allele was determined to be normal or slightly high (1.28 ± 0.06 [n = 3] and 1.12 ± 0.04 [n = 5]) in cultured cells derived from lungs of a 3- and a 4-month-old mouse. It is unclear whether the small increase in expression of the Col-Int allele is due to a lack of an inhibitory role for the first intron in cells subjected to culture conditions or whether it is due to a relative proliferation of some cells, among the many types present in lung (see Discussion), which are less dependent on the first intron for expression of the Col1A1 transcript. In any event, the data suggest that conclusions regarding the role of the first intron in regulation should consider the possibility that the modulatory effects of the first intron on Col1A1 gene expression may differ in vitro and in vivo.

View larger version (9K): [in this window] [in a new window]   FIG. 5.   Ratio of Col-Int to Col1A1 allelic expression in lung decreases with age. The relative abundance of the Col-Int transcript was plotted as a function of the age of the mouse. Average values and standard deviations of a minimum of three determinations are shown.

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Ratio of Col-Int to Col1A1 allelic expression in lung () and skin () as determined by RT-PCR. The relative abundance of the Col-Int transcript was determined as a function of increasing amounts of first-strand cDNA. Average values and standard deviations of determinations from three adult mice are shown.

Determination of Col1A1 transcript abundance by RNase protection. The results of the RT-PCR assays shown in Fig. 5 and 6 predict that lungs of adult wild-type mice should contain twice as much Col1A1 mRNA as lungs of homozygous Col-Int mice. To confirm the RT-PCR results, we measured the total amount of Col1A1 transcript in tissues of wild-type and homozygous Col-Int mice by RNase protection. The assays were conducted on RNA derived from sex-matched siblings of the same litter (5 to 6 months old) to reduce age- and sex-related variability. For each sibling set, the amount of Col1A1 transcript in a homozygous mutant mouse was compared to the amount detected in its wild-type or heterozygous littermate. The results of experiments with three sets of mice show that lungs of wild-type mice contained about 2.5 times the amount of Col1A1 transcript found in their homozygous mutant siblings (Fig. 7). Interestingly, lungs of heterozygous mutant mice were found to contain a similar excess of Col1A1 transcript (Fig. 7). Thus, the wild-type allele might be capable of compensating for reduced expression of the intron-deleted gene.

View larger version (13K): [in this window] [in a new window]   FIG. 7.   Determination of Col1A1 mRNA levels by RNase protection. Col1A1 mRNA levels in lungs of four heterozygous and three homozygous wild-type mice (5 to 6 months old) and their five homozygous mutant littermates were quantified by an RNase protection assay. The amount of Col1A1 mRNA in heterozygous and wild-type homozygous mice is expressed as the relative increase in abundance over that detected in a homozygous mutant littermate.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have used a gene-targeting approach to introduce, into the murine Col1A1 locus, a mutated Col1A1 allele (Col-Int) bearing a large deletion within the first intron and a single nucleotide substitution in exon 7. The nucleotide substitution generated a new XhoI restriction site. An RT-PCR assay, which takes advantage of the XhoI polymorphism and its linkage to the deletion in intron 1, was then developed for the quantification of allele-specific transcription. The RT-PCR/XhoI restriction digest assay was used to examine the relative abundance of allele-specific mRNA in different tissues of heterozygous Col-Int mice. Our data show that the shortened first intron, which is still spliced correctly, does not affect transcription of Col1A1 mRNA in skin but results in a substantial, age-dependent decrease in abundance of Col-Int mRNA in the lung and muscle. Thus, our results establish a developmental and tissue-specific role for the first intron in transcriptional regulation of the 1(I) collagen gene in mice. Additional effects of the intron in induced expression of Col1A1 are possible.

The role of the first intron in regulation of expression of the 1(I) collagen gene has been studied extensively both in transfected cells and in transgenic animals, but little agreement on the nature of the effect, or even on its existence, has been achieved (8, 11). A number of possible reasons for the conflicting results of previous experiments can be advanced. The majority of experiments were performed by transfection of promoter-reporter gene constructs in cultured cells. These experiments were restricted by the choice of the promoter, usually a 5' flanking sequence contiguous with the first exon and intron, or intronic sequences placed 5' or 3' to the promoter-reporter cassette, and by the type and culture conditions of the cells. The limitations of these experiments include the likely requirement for a precise geometry of the transcription initiation complex, which may not be accommodated by some constructs (9, 56), and distinct differences in the transcriptional activity of fibroblastic cells such as dermal fibroblasts and NIH 3T3 cells (9). The use of an 1(I) minigene in place of a promoter-reporter gene construct in transfection experiments (40) has averted some of these drawbacks, but even minigenes are limited in the extent to which 5' and 3' flanking sequences can be represented, and there are possible uncertainties in quantification that can result from differences in the levels of stability of the initial and mature transcripts in comparison with that of Col1A1 mRNA. The results of transgenic experiments have also been controversial (8, 35, 53, 54), in part because of the need to compare different constructs and the possible confounding effects of copy number and site of insertion on expression of the transgene.

The introduction of a deletion in the first intron of the Col1A1 gene by gene targeting, and the concomitant generation of a linked polymorphism, circumvent most of the shortcomings of transfection and transgenic approaches and, as was done in this study, enable the investigator to compare the expression of the mutant and wild-type collagen alleles accurately within the same tissue sample. Our observations provide partial explanations for some of the discrepancies that have been reported in the past. The large intronic deletion that was created resulted in, at best, a two- to threefold reduction in expression of the mutated allele, and this only in certain tissues of older mice. A difference of this magnitude would be difficult to detect in a comparison of two different tissues by a procedure such as RNase protection since, in this method, variations in measurements of replicate samples tend to be very high (35, 53). The recognition that the influence of the intron can be age dependent makes it imperative that the age of mice used in studies of this sort be carefully controlled and that expression be monitored as a function of age, a practice which has not always been followed. Finally, the finding that cultured cells derived from adult lung express the intron-deleted allele to an extent similar to the extent they express the wild-type allele, whereas this is not the case in the tissue, emphasizes the limitations of cell transfection and subsequent cell culture as a means of evaluating the regulation of transcription of the Col1A1 gene. However, since lung is a heterogeneous tissue in which many cells (smooth muscle, mesothelial, endothelial, and type II alveolar epithelial cells, as well as fibroblasts) are capable of type I collagen synthesis (5, 21), it is possible that the conditions used to derive cells from lung in our study favored the growth of cells which are less subject to regulation by the first intron. In addition, levels of expression of type I collagen genes are known to differ greatly between cells grown as a monolayer and cells grown within a collagen gel (17, 32). Growth of adult lung cells within a collagen gel might therefore restore the dependence of these cells on an intact intron for maximum expression of Col1A1.

In many ways the Col-Int allele is uniquely suited to evaluate the effects of the intron on expression of the Col1A1 gene since the intronic deletion occurs in the context of an otherwise wild-type locus. However, our studies would be unable to detect redundant elements that are present both within the intron and elsewhere in the gene. Such redundancy could account for the finding that a promoter-reporter transgene with a deletion in the first intron was poorly expressed in skin (35), whereas a requirement for the deleted sequence in skin was not found when a Col1A1 minigene was tested in transgenic mice (54) or in this study. Recently, indirect evidence was, in fact, obtained in stably transfected cells for an enhancer in the body of the Col1A1 gene (11). In addition, the deletion tested in our study was a large one which very likely includes many potential cis-acting elements. Some of these elements have been tested individually or as part of shorter sequences in cell transfection studies, and both positively and negatively acting elements have been identified (2, 6, 26, 29, 33, 34, 52; see reference 8 for a review). The phenotype identified thus far in homozygous Col-Int mice may therefore reflect the net result of deleting multiple elements with opposing effects on transcription. It is possible that more discrete deletions will reveal effects that differ in magnitude and in cell and tissue specificity.

Recently, in a study involving two populations of British women, reduced bone mineral density and a tendency for osteoporotic fractures were shown to be associated with a polymorphic (GT) Sp1-binding site in the first intron of the human Col1A1 gene (22). The polymorphism is predicted to reduce significantly the affinity of binding of Sp1 to the site. This finding, if confirmed, would provide additional evidence for a function of the first intron of Col1A1 in modulating tissue-specific expression of the gene. However, although this Sp1 site is located in a region of the intron that is well conserved between mice and humans (our observation), physiological differences between the two species, and the possibility that a point mutation and a large deletion may have different consequences, make it unlikely that homozygous Col-Int mice will develop osteoporosis.

Although different methods have been used to introduce mutations into the murine genome (3, 19, 23, 25, 55, 58, 59), relatively few studies have utilized gene-targeting techniques to examine regulatory sequences in the mouse (7, 20, 37, 50, 57). Reports by Fiering et al. (20) and McDevitt et al. (37) evaluated the importance of transcriptional elements for expression, whereas those by Takeda et al. (57), Serwe and Sablitzky (50), and Bories et al. (7) evaluated the role of transcriptional enhancer elements in controlling rearrangements in immunoglobulin genes. Interestingly, in most of these studies an inhibitory effect of the integrated Neo^r gene on transcription of the targeted gene was observed. Thus, the above-named studies illustrate the importance of using gene-targeting methods, such as double replacement, which do not otherwise perturb the endogenous gene. In this regard, it seems possible that the insertion of the 9-kbp MMLV 19 bp downstream from the 5' splice site in the first intron of the Col1A1 gene in homozygous Mov 13 mice (24, 31, 47) inhibits transcription of the gene as a consequence of the transcriptional activity of the MMLV genome. However, this possibility seems unlikely in view of evidence for increased methylation and altered chromatin conformation of Col1A1 sequences flanking the proviral insertion (12, 27, 43).

In conclusion, we have shown that the generation of a targeted, mutated allele containing both an altered, putative regulatory element, and a change in a restriction endonuclease sequence within an exon, can be used to determine the contribution of the regulatory sequence to the expression of the gene. Our findings confirm a role for the first intron in regulation of expression of the Col1A1 gene.