Selective amplification of polymorphic dihydrofolate reductase gene loci in Chinese hamster lung cells

Severe folate restriction results in depletion of and alteration in the composition of the intracellular folate pool, moderate sensitization to methotrexate and trimetrexate, upregulation of endogenous DHFR activity, and overexpression of metallothionein II and folate receptor alpha that, upon folate repletion, confer drug resistance to CHL cells

DC-3F/FA3 cells (FA3) were derived from antifolate-sensitive CHL cells by selection for growth in folate-free media containing 15 pM [6S]-5CHOFH4. These cells undergo a 30-fold decrease in intracellular folates, overexpress folate receptor alpha (FR alpha) and metallothionein II, and display increased sensitivity to the dihydrofolate reductase (DHFR) targeted anti-folates methotrexate (MTX) and trimetrexate (TMTX), which can be attributed primarily to the folate pool status. Upon folate repletion by growth in 15 nM [6S]-5CHOFH4, they display a 5- and 10-fold increase in resistance to both drugs, respectively, even though folate pools are restored by only 43%. Enforced overexpression of FR alpha in transfectants cultured in nanomolar folate did not confer resistance to MTX but did support a modest 2-fold increase in resistance to TMTX. Enforced overexpression of MTII had a similar effect, but when both were overexpressed together no increase in resistance beyond that conferred by each one separately was noted, suggesting that both confer resistance to TMTX through a common downstream mechanism. Analysis of three independent low folate selected clones, FA3, FA7, and FA14, showed that each had a 5- to 6-fold increase in DHFR activity accompanied by a similar increase in DHFR protein level. However, no differences were detected in the DHFR gene copy number or in the steady-state amount of DHFR mRNA, suggesting that a posttranscriptional mechanism was responsible for the increase in DHFR expression.

A genetic polymorphism within the third poly(A) signal of the DHFR gene alters the polyadenylation pattern of DHFR transcripts in CHL cells

Two polymorphic dihydrofolate reductase (DHFR) alleles, termed 20 K and 21 K, exist in Chinese hamster lung cells. Three major transcripts of different lengths are transcribed from each allele, and the expression of these transcripts differs dramatically between the alleles as a result of differential utilization of three poly(A) sites. Transcripts from the 20 K allele are preferentially polyadenylated at the first poly(A) site, while those from the 21 K allele are preferentially polyadenylated at the third site. In this study, transient expression experiments were used to demonstrate that a 2.1 kb genomic fragment containing the three DHFR poly(A) sites is sufficient to reproduce the allele-specific polyadenylation pattern on transiently expressed CAT-DHFR transcripts in COS cells. Site-directed mutagenesis allowed identification of the sequence elements which are responsible for this allele-specific polyadenylation. These studies indicate that a single-base change in the third poly(A) signal sequence, which alters the consensus AAUAAA signal in the 21 K allele to a weak AAUAAU signal in the 20 K allele, is primarily responsible for the dramatic difference in polyadenylation between the two alleles. Thus, as a result of this single-base change in the third poly(A) signal sequence, utilization of the first poly(A) site, located 1.2 kb upstream, changes dramatically.

Differential utilization of poly (A) signals between DHFR alleles in CHL cells

The Chinese hamster cell line, DC-3F, is heterozygous at the DHFR locus, and each allele can be distinguished on the basis of a unique DNA restriction pattern, protein isoelectric profile and in the abundancy of the DHFR mRNAs it expresses. Although each allele produces four transcripts, 1000, 1650 and 2150 nucleotides [corrected] in length, the relative distribution of these RNAs differs for each; the 2150 nt mRNA represents the major (60%) species generated from one allele, while the 1000 nt mRNA is the major species generated from the other. The allele that predominantly expresses the 2150 nt transcript is preferentially overexpressed when DC-3F cells are subjected to selection in methotrexate. We have analyzed the 3' ends of both DHFR alleles and have found that the three major mRNAs arise by readthrough of multiple polyadenylation signals. A four base deletion in one allele changes the consensus polyadenylation signal AAUAAA to AAUAAU, resulting in the utilization of a cryptic polyadenylation signal lying 21 bp upstream. Surprisingly, this mutation in the third polyadenylation signal appears to affect not only the utilization of this signal, but also the efficiency with which the first signal, located 1171 bp upstream from the third site, is utilized.

Amplified allele of the human N-myc oncogene in neuroblastomas

The N-myc amplification unit was investigated using two restriction fragment length polymorphisms for SphI and PvuII which we previously found to occur in the 2nd intron and in the 3' region of the human N-myc oncogene, respectively. All the three haplotypes which are dominant in the Japanese population were observed in the amplified DNA after analyses of a total of 22 DNA samples of fresh neuroblastomas with N-myc amplification. We conclude that only one of the alleles was amplified and that either allele could be amplified with respect to both the SphI and PvuII polymorphisms.

The role of thymidylate synthase in the response to fluoropyrimidine-folinic acid combinations

A panel of human colorectal tumor cell lines has been examined to determine the role of TS in the response to fluoropyrimidine antimetabolites. Among these cell lines, the response to FdUrd does not correlate with the levels of TS. In cell lines HCT 116 and RCA, which are poorly responsive to FdUrd, structural alterations in TS have been identified. In HCT 116, two TS polypeptides are present: a common form, occurring in all the cell lines and a variant form. The variant TS polypeptide has a reduced affinity for the TS ligands, FdUMP and CH2H4PteGlu, relative to the common TS polypeptide. Clonal populations of HCT 116 that overproduce each form have been isolated. Clones that overproduce the variant polypeptide are 4-fold less responsive to TS-directed cytotoxic agents than those that overproduce the common; thus, the presence of the variant TS is associated with a reduced response to TS-directed cytotoxic agents. The response of cell line RCA to FdUrd is dependent upon the extracellular CF concentration: response increases as CF is increased. RCA contains a TS enzyme with reduced affinity for CH2H4PteGlu, relative to cell line C, which is sensitive to FdUrd at all CF concentrations. Both cells form high chain-length polyglutamates of CH2H4PteGlu at CF concentrations in which the response to FdUrd differs by 4-fold. In RCA, the TS structural gene is variant, relative to the other cell lines. This variation may underlie the altered enzyme affinity for CH2H4PteGlu and the sensitivity to modulation of FdUrd response by CF.

Discrete human dihydrofolate reductase gene transcripts present in polysomal RNA map with their 5' ends several hundred nucleotides upstream of the main mRNA start site

The 5' ends of dihydrofolate reductase (DHFR)-specific transcripts have been mapped in the 5'-flanking region of the amplified DHFR gene of the human methotrexate-resistant cell line 6A3 by primer extension and S1 protection experiments. The main 5' end, at position -71 relative to the first nucleotide of the DHFR reading frame, corresponds to the recently identified main transcription initiation site for the DHFR gene and pertains to transcripts representing approximately 99% of the DHFR-specific polysomal polyadenylic acid-containing RNA, and including the previously described DHFR mRNAs with sizes of 3.8, 1.0, and 0.8 kilobases. At least six other minor 5' ends have been mapped to nucleotide positions -449 to -480 upstream of the DHFR gene and pertain to approximately 1% of the DHFR-specific polysomal polyadenylic acid-containing RNA. These upstream initiating transcripts appear to include five major discrete species with sizes of 4.3, 3.8, 3.1, 2.1, and 1.0 kilobases and four minor ones with sizes of 7.3, 5.0, 1.4, and 0.8 kilobases. These species, with the exception of those of 3.1- and 2.1-kilobase sizes, also have been found in VA2-B cells, the parental line of 6A3, and in HeLa cells. The upstream initiating transcripts present in all three cell lines are increased in amount in 6A3 cells as compared with the other cell lines, in about the same proportion as the three identified DHFR mRNAs.

Regular pattern of karyotypic alterations accompanying gene amplification in Djungarian hamster cells: study of colchicine, adriablastin, and methotrexate resistance

Chromosomal analysis of 26 Djungarian hamster cell lines obtained from 11 independent clones and possessing different levels of resistance to colchicine or adriablastin as a consequence of gene amplification revealed regular patterns in the karyotypic changes that accompanied the development of drug resistance. Usually the sequence of karyotypic changes was as follows: first an additional chromosome 4 appeared: then single unpaired small chromatin bodies (SCBs) arose; later in the middle part of the long arm of one of three chromosomes 4 long homogeneously staining regions (HSRs) and double minute chromosomes (DMs) were formed; and finally in the most resistant variants large clusters of SCBs appeared. The emergence of the clusters of the SCBs correlated well with the occurrence of autonomously replicating, amplified DNA sequences. In contrast to DNA of the HSRs the DNA of the SCBs could replicate outside the S-phase of the cell cycle. When kept in a non-selective medium, the cells gradually lost their resistance to colchicine: 1%-4% of the cells lost the capacity to form colonies in the selective medium independently of the pattern of location in them of amplified genes (in chromosomal HSRs. SCBs, or DMs). Loss of drug resistance was accompanied by disappearance of the chromosomal HSRs, SCBs, and DMs. Chromosomal analysis of the set of methotrexate-resistant Djungarian hamster cell lines indicated the following karyotypic evolution: first the additional material on the distal part of one of two chromosomes 3 appeared; then the light HSRs were formed on the distal part of one of two chromosomes 4; later clusters of SCBs and HSRs arose on the distal part of the short arm of chromosome 3. Probably the amplification of different genes is characterized by specific patterns of karyotypic alterations.

Nucleotide sequence of the 3' exon of the human N-myc gene

We have analyzed a 3.8 kb Eco RI fragment of genomic DNA obtained from the amplified N-myc gene of human neuroblastoma cell line BE(2)-C. This fragment contains an exon with an open reading frame encoding approximately 170 amino acids of the carboxy-terminal end of the putative N-myc protein. Comparison of the inferred amino acid sequence of this peptide with that of the 3' domain of the human c-myc protein shows that locally conserved but dispersed regions of homology exist throughout the lengths of these peptides, while hydropathy plots indicate that the physical properties implied by their primary sequences are strikingly similar. Based upon these and other considerations, it is suggested that the 3' domains of c-myc and N-myc may potentially share related functions.

Discrete human dihydrofolate reductase gene transcripts present in polysomal RNA map with their 5' ends several hundred nucleotides upstream of the main mRNA start site

The 5' ends of dihydrofolate reductase (DHFR)-specific transcripts have been mapped in the 5'-flanking region of the amplified DHFR gene of the human methotrexate-resistant cell line 6A3 by primer extension and S1 protection experiments. The main 5' end, at position -71 relative to the first nucleotide of the DHFR reading frame, corresponds to the recently identified main transcription initiation site for the DHFR gene and pertains to transcripts representing approximately 99% of the DHFR-specific polysomal polyadenylic acid-containing RNA, and including the previously described DHFR mRNAs with sizes of 3.8, 1.0, and 0.8 kilobases. At least six other minor 5' ends have been mapped to nucleotide positions -449 to -480 upstream of the DHFR gene and pertain to approximately 1% of the DHFR-specific polysomal polyadenylic acid-containing RNA. These upstream initiating transcripts appear to include five major discrete species with sizes of 4.3, 3.8, 3.1, 2.1, and 1.0 kilobases and four minor ones with sizes of 7.3, 5.0, 1.4, and 0.8 kilobases. These species, with the exception of those of 3.1- and 2.1-kilobase sizes, also have been found in VA2-B cells, the parental line of 6A3, and in HeLa cells. The upstream initiating transcripts present in all three cell lines are increased in amount in 6A3 cells as compared with the other cell lines, in about the same proportion as the three identified DHFR mRNAs.

Expression of the amplified domain in human neuroblastoma cells

Screening of a partial cDNA library prepared from the human neuroblastoma cell line BE(2)-C with genomic DNA probes containing sequences representative of the amplified domain of that cell line allowed us to identify cloned transcripts from an active gene within the domain. The gene BE(2)-C-59 is amplified ca. 150-fold and encodes a 3.0- and a 1.5-kilobase RNA transcript, both of which are overproduced in BE(2)-C cells. A survey of a large variety of human tumor cell types indicated that this gene is amplified to varying degrees in all neuroblastoma cell lines and a retinoblastoma cell line that exhibit obvious cytological manifestations of DNA sequence amplification, i.e., homogeneously staining regions and double-minute chromosomes. The BE(2)-C-59 gene is not amplified, however, in other nonrelated tumor types, even those containing amplified DNA. Although the functional significance of this specific gene amplification in neuroblastoma cells remains unknown, an indication that it may relate to the malignant phenotype of these cells follows from the remainder of our data which show that the amplified BE(2)-C-59 gene shares partial homology with both the second and third exons, but not the first exon, of the human c-myc oncogene.

Expression of the amplified domain in human neuroblastoma cells

Screening of a partial cDNA library prepared from the human neuroblastoma cell line BE(2)-C with genomic DNA probes containing sequences representative of the amplified domain of that cell line allowed us to identify cloned transcripts from an active gene within the domain. The gene BE(2)-C-59 is amplified ca. 150-fold and encodes a 3.0- and a 1.5-kilobase RNA transcript, both of which are overproduced in BE(2)-C cells. A survey of a large variety of human tumor cell types indicated that this gene is amplified to varying degrees in all neuroblastoma cell lines and a retinoblastoma cell line that exhibit obvious cytological manifestations of DNA sequence amplification, i.e., homogeneously staining regions and double-minute chromosomes. The BE(2)-C-59 gene is not amplified, however, in other nonrelated tumor types, even those containing amplified DNA. Although the functional significance of this specific gene amplification in neuroblastoma cells remains unknown, an indication that it may relate to the malignant phenotype of these cells follows from the remainder of our data which show that the amplified BE(2)-C-59 gene shares partial homology with both the second and third exons, but not the first exon, of the human c-myc oncogene.

Phenotypic expression in Escherichia coli and nucleotide sequence of two Chinese hamster lung cell cDNAs encoding different dihydrofolate reductases

Nucleotide sequence analysis of two cDNA clones, one shown to direct the synthesis in Escherichia coli of the pI 6.7 form of the 20,000-molecular-weight class of Chinese hamster lung cell dihydrofolate reductase, and the other shown to direct the synthesis of the pI 6.5 form of the 21,000-molecular-weight class of the enzyme, has revealed the following: (i) the differences in physical and enzymatic properties displayed by these two proteins are due to two variations in their respective amino acid sequences with the conversion of Leu to Phe at position 22 probably responsible for the differential sensitivity of these two enzymes to methotrexate and methasquin; (ii) the multiple mRNAs responsible for the synthesis of each of these proteins differ in size due, at least in part, to a length heterogeneity at their 3' ends; (iii) these two proteins are encoded by different genes; and (iv) the sequence AAATATA appears to be a major polyadenylation signal in one Chinese hamster lung cell dihydrofolate reductase gene and a minor signal in another.

Phenotypic expression in Escherichia coli and nucleotide sequence of two Chinese hamster lung cell cDNAs encoding different dihydrofolate reductases

Nucleotide sequence analysis of two cDNA clones, one shown to direct the synthesis in Escherichia coli of the pI 6.7 form of the 20,000-molecular-weight class of Chinese hamster lung cell dihydrofolate reductase, and the other shown to direct the synthesis of the pI 6.5 form of the 21,000-molecular-weight class of the enzyme, has revealed the following: (i) the differences in physical and enzymatic properties displayed by these two proteins are due to two variations in their respective amino acid sequences with the conversion of Leu to Phe at position 22 probably responsible for the differential sensitivity of these two enzymes to methotrexate and methasquin; (ii) the multiple mRNAs responsible for the synthesis of each of these proteins differ in size due, at least in part, to a length heterogeneity at their 3' ends; (iii) these two proteins are encoded by different genes; and (iv) the sequence AAATATA appears to be a major polyadenylation signal in one Chinese hamster lung cell dihydrofolate reductase gene and a minor signal in another.

Organization of a Chinese hamster ovary dihydrofolate reductase gene identified by phenotypic rescue

We have constructed a genomic DNA library from a methotrexate-resistant Chinese hamster ovary cell line (CHOC 400) in the cosmid vector pHC79. By utilizing a murine dihydrofolate reductase (DHFR) cDNA clone, we have identified 66 DHFR+ clones among the 11,000 colonies screened by colony hybridization. To isolate a recombinant cosmid containing the entire DHFR gene, we have tested these colonies for their ability to rescue a DHFR- Chinese hamster ovary cell line, using the spheroplast fusion method of gene transfer developed by W. Schaffner (Proc. Natl. Acad. Sci. U.S.A. 77:2163-2167, 1980). One clone (cH1) was able to transform DHFR- cells to the DHFR+ phenotype and was shown in hybridization studies to contain all of the gene except a small portion of the 3' untranslated region. We have mapped cosmid cH1 and several overlapping cosmids with a variety of restriction enzymes and have determined the approximate positions of the five (and possibly six) exons within the DHFR gene. Differences between the sizes of homologous genes in hamster cells (24.5 kilobases [kb]) and in mouse cells (31.5 kb) are shown to reside primarily in the length of the 3' intron, which is 8 kb in the hamster gene and 16 kb in length in the mouse gene. Our studies confirm the utility of cosmid libraries for the isolation of large genes, as previously shown by R. de Saint Vincent et al. (Cell 27:267-277, 1981). In addition, a cosmid that contains a functional DHFR gene will be a useful vector for the co-amplification and subsequent overexpression of other cloned genes.

Amplification of a cloned Chinese hamster dihydrofolate reductase gene after transfer into a dihydrofolate reductase-deficient cell line

We have transformed a dihydrofolate reductase (DHFR)-deficient Chinese hamster ovary cell line to the DHFR+ phenotype with a recombinant cosmid (cH1) containing a functional Chinese hamster DHFR gene (J.D. Milbrandt et al., Mol. Cell. Biol. 3:1266-1273, 1983). After exposure of cells to successive increases in methotrexate, we have isolated a resistant cell line (JSH-1) that grows in 1 microM methotrexate. We show here that JSH-1 contains 300 to 500 copies of the integrated cosmid and that these copies are located predominantly at one position on a chromosome identified as Z5a. Hybridization analysis of restriction digests of genomic DNA indicates that the cosmid has been integrated intact into the genome and that upon amplification, the original cosmid/genomic junction fragments are also amplified in JSH-1. Furthermore, the pattern of amplified bands observed in ethidium bromide-stained gels indicates that the unit amplified sequence (amplicon) may be as large as 120 to 135 kilobases and therefore includes considerable amounts of flanking DNA in addition to the 45 kilobases of integrated cosmid. We also show that the protein overproduced by the amplified cosmid in JSH-1 comigrates with the 21,000-dalton polypeptide characteristic of the methotrexate-resistant cell line (CHOC 400) from which cH1 was cloned. However, the DHFR mRNA species overproduced in JSH-1 appear to be larger than those detected in CHOC 400, indicating that not all of the normal transcription and processing signals are preserved in the integrated recombinant cosmid.

Specific DNA sequence amplification in human neuroblastoma cells

Southern blot analysis of a number of EcoRI-digested human neuroblastoma DNAs has revealed the presence of a family of discrete restriction fragments, the majority of which are amplified in most, but not all, of the neuroblastoma cell lines tested. None of these sequences is abundantly present in DNA from other human tumors of different tissue origins, including several either known or presumed to contain amplified DNA. Hence, these sequences appear to be specifically amplified by neuroblastoma cells. Hybridization with metaphase chromosomes in situ has localized these sequences to either the homogeneously staining regions or double-minute chromosomes of different neuroblastoma cell lines, indicating that these chromosomal structures, although present in cell lines established from different patients, share many sequences and may have a common, but as yet unknown, function.

Organization of a Chinese hamster ovary dihydrofolate reductase gene identified by phenotypic rescue

We have constructed a genomic DNA library from a methotrexate-resistant Chinese hamster ovary cell line (CHOC 400) in the cosmid vector pHC79. By utilizing a murine dihydrofolate reductase (DHFR) cDNA clone, we have identified 66 DHFR+ clones among the 11,000 colonies screened by colony hybridization. To isolate a recombinant cosmid containing the entire DHFR gene, we have tested these colonies for their ability to rescue a DHFR- Chinese hamster ovary cell line, using the spheroplast fusion method of gene transfer developed by W. Schaffner (Proc. Natl. Acad. Sci. U.S.A. 77:2163-2167, 1980). One clone (cH1) was able to transform DHFR- cells to the DHFR+ phenotype and was shown in hybridization studies to contain all of the gene except a small portion of the 3' untranslated region. We have mapped cosmid cH1 and several overlapping cosmids with a variety of restriction enzymes and have determined the approximate positions of the five (and possibly six) exons within the DHFR gene. Differences between the sizes of homologous genes in hamster cells (24.5 kilobases [kb]) and in mouse cells (31.5 kb) are shown to reside primarily in the length of the 3' intron, which is 8 kb in the hamster gene and 16 kb in length in the mouse gene. Our studies confirm the utility of cosmid libraries for the isolation of large genes, as previously shown by R. de Saint Vincent et al. (Cell 27:267-277, 1981). In addition, a cosmid that contains a functional DHFR gene will be a useful vector for the co-amplification and subsequent overexpression of other cloned genes.

Amplification of a cloned Chinese hamster dihydrofolate reductase gene after transfer into a dihydrofolate reductase-deficient cell line

We have transformed a dihydrofolate reductase (DHFR)-deficient Chinese hamster ovary cell line to the DHFR+ phenotype with a recombinant cosmid (cH1) containing a functional Chinese hamster DHFR gene (J.D. Milbrandt et al., Mol. Cell. Biol. 3:1266-1273, 1983). After exposure of cells to successive increases in methotrexate, we have isolated a resistant cell line (JSH-1) that grows in 1 microM methotrexate. We show here that JSH-1 contains 300 to 500 copies of the integrated cosmid and that these copies are located predominantly at one position on a chromosome identified as Z5a. Hybridization analysis of restriction digests of genomic DNA indicates that the cosmid has been integrated intact into the genome and that upon amplification, the original cosmid/genomic junction fragments are also amplified in JSH-1. Furthermore, the pattern of amplified bands observed in ethidium bromide-stained gels indicates that the unit amplified sequence (amplicon) may be as large as 120 to 135 kilobases and therefore includes considerable amounts of flanking DNA in addition to the 45 kilobases of integrated cosmid. We also show that the protein overproduced by the amplified cosmid in JSH-1 comigrates with the 21,000-dalton polypeptide characteristic of the methotrexate-resistant cell line (CHOC 400) from which cH1 was cloned. However, the DHFR mRNA species overproduced in JSH-1 appear to be larger than those detected in CHOC 400, indicating that not all of the normal transcription and processing signals are preserved in the integrated recombinant cosmid.

A human dihydrofolate reductase pseudogene and its relationship to the multiple forms of specific messenger RNA

The presence of dihydrofolate reductase (DHFRase)-specific sequences that, in contrast to the normal DHFRase gene, are not amplified in a methotrexate-resistant cell line, has been detected in the DNA from human sperm and from several human cell lines. DNA fragments containing some of these sequences have been isolated from a cosmid library of human sperm DNA. One of these fragments contains a DHFRase pseudogene (psi HD1) that completely lacks introns, has 92% sequence homology to the corresponding region of normal DHFRase complementary DNA, but exhibits several alterations that make it nonfunctional. The sequence analysis of the inserts of four different plasmids containing the reading frame and varying lengths of the 3' non-coding regions of human DHFRase-specific cDNAs has revealed that the 3' non-coding segments all are colinear in their corresponding portions. Furthermore, the data indicate that the cDNA of one of the plasmids is probably derived from the smallest of the three main human DHFRase messenger RNAs, the 0.8 X 10(3) base (0.8 kb) mRNA, the cDNA of two others, from the 1.0 kb mRNA, and the cDNA of the fourth, from a longer mRNA. These results are consistent with the idea that the multiple forms of DHFRase mRNA in human cells derive from the same gene by different transcription or RNA-processing events. Moreover, the sequence comparison between the psi HD1 and the different DHFRase cDNAs clearly indicates that, if an mRNA intermediate has participated in the formation of this pseudogene, a form of mRNA larger than the 1.0 kb mRNA, probably the 3.8 kb mRNA, must have been involved.

Structure of the dihydrofolate reductase gene in Chinese hamster ovary cells

Overlapping recombinant lambda 1059 phages carrying regions of the dhfr locus from the amplified Chinese hamster ovary (CHO) cell clone MK42 have been isolated. In addition, dhfr cDNAs from this cell line have been cloned into plasmid pBR322. Restriction analysis of these recombinant molecules has led to a map of the Chinese hamster dhfr gene. This gene has a minimum size of 26 kb and contains six exons as defined by hybridization to a combination of mouse and CHO cDNA probes. The latter probes reveal 3' exonic sequences that are not present in mouse cDNA. The CHO dhfr gene thus extends about 700 bp further 3' than in the mouse, consistent with the larger size of the hamster mRNA. At least five intervening sequences are present, of approximate sizes: 0.3, 2.5, 8.6, 2.6 and 9.4 kb. Four sequences from highly repeated families are situated in introns within the dhfr gene. The overall structure of this gene is strikingly similar to that of the mouse. Evolutionary conservation of interrupted gene structure among mammals thus extends to genes that code for household enzymes as well as specialized or structural proteins.