Each adult transgenic male expressed the thymidine kinase gene in testes and brain and ... the site of integration in transgenic mice (1, 5, 6, 12, 17, 18,.
Vol. 13, No. 4
MOLECULAR AND CELLULAR BIOLOGY, Apr. 1993, p. 2214-2223
0270-7306/93/042214-10$02.00/0
Copyright © 1993, American Society for Microbiology
Transcriptional Insulation of the Human Keratin 18 Gene in Transgenic Mice NICKOLAY NEZNANOV, IRMGARD S. THOREY, GRACE CECENA, AND ROBERT G. OSHIMA* Cancer Research Center, La Jolla Cancer Research Foundation, La Jolla, California 92037 Received 2 October 1992/Returned for modification 12 December 1992/Accepted 13 January 1993
Expression of the 10-kb human keratin 18 (K18) gene in transgenic mice results in efficient and appropriate tissue-specific expression in a variety of internal epithelial organs, including liver, lung, intestine, kidney, and the ependymal epithelium of brain, but not in spleen, heart, or skeletal muscle. Expression at the RNA level is directly proportional to the number of integrated K18 transgenes. These results indicate that the K18 gene is able to insulate itself both from the commonly observed cis-acting effects of the sites of integration and from the potential complications of duplicated copies of the gene arranged in head-to-tail fashion. To begin to identify the K18 gene sequences responsible for this property of transcriptional insulation, additional transgenic mouse lines containing deletions of either the 5' or 3' distal end of the K18 gene have been characterized. Deletion of 1.5 kb of the distal 5' flanking sequence has no effect upon either the tissue specificity or the copy number-dependent behavior of the transgene. In contrast, deletion of the 3.5-kb 3' flanking sequence of the gene results in the loss of the copy number-dependent behavior of the gene in liver and intestine. However, expression in kidney, lung, and brain remains efficient and copy number dependent in these transgenic mice. Furthermore, herpes simplex virus thymidine kinase gene expression is copy number dependent in transgenic mice when the gene is located between the distal 5'- and 3'-flanking sequences of the K18 gene. Each adult transgenic male expressed the thymidine kinase gene in testes and brain and proportionally to the number of integrated transgenes. We conclude that the characteristic of copy number-dependent expression of the K18 gene is tissue specific because the sequence requirements for transcriptional insulation in adult liver and intestine are different from those for lung and kidney. In addition, the behavior of the transgenic thymidine kinase gene in testes and brain suggests that the property of transcriptional insulation of the K18 gene may be conferred by the distal flanking sequences of the K18 gene and, additionally, may function for other genes. 20). The position-independent, efficient copy number-dependent expression of a gene in transgenic mice was first achieved by the inclusion of a large regulatory region, called the locus control region (LCR), located about 20 kb upstream of the first gene of the human beta-globin locus (20). Subsequent studies have shown that a powerful, tissuespecific enhancer of less than 2 kb within the LCR ensures expression levels comparable to those of endogenous genes, although strict copy number dependence is not necessarily retained (16, 37). Similarly, a regulatory region of the lysozyme gene acts as an LCR. In this case, the LCR was dissected into one fragment that contained the transcriptional enhancer activity and another retaining the characteristic of position-independent and copy number-dependent expression. The latter fragment included nuclear matrix attachment region activity (38). DNase-hypersensitive sites and tissue-specific transcriptional enhancer activity are also associated with a 3'-flanking region of the CD2 gene, which confers copy number-dependent expression in transgenic mice (17). One of the smallest genomic fragments so far associated with LCR activity is the 585-bp fragment of the 5' end of a human class I histocompatibility antigen (12). In contrast to the known transcriptional regulatory activity of these examples, a small (630-bp) segment of the first intron of the adenosine deaminase (ADA) gene has been implicated recently in the position-independent and gene copy-proportional expression of ADA-chloramphenicol acetyltransferase (CAT) reporter constructions in transgenic mice (5). This fragment is predominantly Alu-type repetitive DNA, is not DNase hypersensitive, and lacks regulatory activity detectable by standard transient expression methods.
Eukaryotic genes are regulated by the interaction of transcription factors with their cognate DNA binding sites and with each other. However, the ability of many transcriptional regulatory elements to function relatively independently of their orientation and position raises the question of how neighboring transcription units may be insulated from each other. In addition, the discovery of regulatory elements located a great distance from the proximal promoter of some genes and the existence of elements that participate in the regulation of more than one gene (10, 20) stimulate consideration of how regulatory element activity is restricted. Distinctive chromosomal domains may be defined by special chromatin structures as have been discovered in Drosophila melanogaster (24), by nuclear matrix attachments regions that may physically define chromosomal loops (14, 19), or by additional mechanisms that designate regions or single genes for repressive chromatin condensation. However, even within a single chromosomal loop that may span 100 kb or more, multiple genes presumably must be regulated precisely. The potential complication of cis-acting regulatory elements is illustrated in transgenic mice. The integration of foreign genes commonly leads to very dramatic differences in both the level of expression and sometimes the tissue specificity of expression (35). These differences are attributed to cis-acting regulatory elements flanking the sites of integration (4). A few genes are expressed independently of the site of integration in transgenic mice (1, 5, 6, 12, 17, 18, *
Corresponding author. 2214
VOL. 13, 1993
TRANSCRIPTIONAL INSULATION OF THE HUMAN K18 GENE IN MICE
The keratin 18 (K18) gene codes for a type I keratin intermediate filament protein that is expressed first just prior to the blastocyst stage and later in a variety of embryonic and adult simple (single-layered) epithelia, including intestine, lung, liver, kidney, and the ependymal cell layer of the brain (1). Analysis of three transgenic mouse lines carrying the human K18 gene revealed appropriate, tissue-specific, and copy number-dependent expression even though no matrix attachment regions were detected within the gene (1). In this study, we have confirmed and extended those initial results by characterizing the expression of additional transgenic mice carrying either the full 10-kb K18 gene or one of two fragments of the gene. We show the 3'-flanking 3.5 kb of the K18 gene is essential for position-independent, copy number-dependent expression in liver and intestine but not in lung, kidney, or brain. Furthermore, the herpes simplex virus (HSV) thymidine kinase (TK) gene flanked by the distal 5' and 3' regions of the K18 gene is expressed in a position-independent, copy number-dependent manner in transgenic mice without the tissue specificity of the K18 gene. The flanking sequences of the K18 gene may act as local transcriptional insulators that may be important for preventing inappropriate interaction of neighboring transcription units. MATERIALS AND METHODS Transgenic mice. Transgenic mice were prepared by standard procedures as previously described (1) by the Transgenic Mouse Facility at the La Jolla Cancer Research Foundation. Strain FVB/N mouse eggs were injected and transferred to CD-1 foster mothers. Founder animals identified by dot blot hybridization of tail DNAs were sacrificed without further breeding (K18-Xmn mice) or were mated once and positive second-generation animals were sacrificed for RNA and DNA analysis (K18, K18-Nsi, and NNTK mice). When founder animals were analyzed, mosaic individuals, identified by immunofluorescent staining of K18 in gut and liver, were excluded from further analysis. Nucleic acid analysis. Transgene copy number was determined by dot blot hybridization with the K18 cDNA (33) or TK coding sequences followed by quantitation in an Ambis radioactivity image analyzer or by liquid scintillation counting using appropriate plasmid DNAs diluted in nontransgenic DNA and previously determined transgenic mouse DNAs as standards. DNA concentrations were determine fluorimetrically (28). To minimize variation due to DNA loading or retention by the filter, each DNA spot was normalized by rehybridizing the stripped filter with a random-primed total mouse DNA probe. The average signal of all dots was considered 2 ,ug of DNA. The copy number of each line, as indicated by comparison of the radioactive signals of at least duplicate sets of dots for each animal with those of standards on the same filter, were adjusted for loading as indicated by the second round of hybridization. Southern blots were performed on DNA samples of each mouse to determine the integrated arrangement of the transgenes and to confirm the dot blot quantitation. In all cases reported, the predominant integrated organization was a head-to-tail tandem array of duplicated unit-length gene fragments. Because of the difficulty of determining the exact arrangement of each transgenic integration site, the dot blot estimation of copy number was used as the quantitative estimate, with no adjustment for flanking partial or rearranged genes. This may result in overestimation of the competent copy number of integration events which contain
2215
small numbers of transgenes, but it eliminates the need for subjective judgment of whether any particular hybridizing fragment represents an active or inactive gene. RNA was purified from dissected mouse tissues with the use of guanidine isothiocyanate and CsCl2 ultracentrifugation (13). K18 RNA was quantitated by Northern (RNA) blot analysis and an RNase protection assay which used a T7 RNA polymerase transcript of a 431-bp fragment of the K18 gene (XhoI at nucleotide [nt] 2284 to XhoII at nt 2715) (27) cloned into the BSKSM13+ vector. The protected K18 signal was normalized to the signal obtained for the endogenous mouse keratin 8 (mK8 or EndoA) RNA, which was measured simultaneously in the same RNA sample. The mK8 probe was derived from the SP6 RNA polymerase transcription of a 295-bp SmaI fragment of the EndoA alpha-1 gene (39, 40) that spans the major transcriptional start site of the gene, resulting in a probe of approximately 295 nt, of which 135 nt is protected by mK8 RNA. The mean mK8 value of all samples of the same organ was considered 10 ,ug of RNA. The K18 values were compared with standard synthetic K18 RNA standards and then adjusted to reflect the signal from 10 ,ug of total RNA. Northern blot filters were analyzed by hybridization with a random-primed K18 cDNA probe under conditions sufficiently stringent to exclude cross-hybridization with mK18, the mouse homolog (final washes in 0.1 x SSPE-0.1% sodium dodecyl sulfate at 65°C). After appropriate exposures, the filters were stripped of probe by heating to about 98°C and rehybridized with a probe for the ribosomal protein L32 RNA. Densitometer tracings of the K18 signals were compared with synthetic standard K18 RNAs run on the same filters. The estimated abundance of K18 RNA was then adjusted for the actual load of RNA as indicated by the L32 hybridization signal. To measure the TK RNA by RNase protection, a 417-bp fragment of the HSV TK gene defined by EcoRI and EcoRV sites was subcloned into pGEM1. An antisense RNA probe was made by transcription of the EcoRI-digested plasmid with SP6 RNA polymerase. An additional probe was made from SP6 transcription of a larger fragment of the TK gene that extended to the BamHI site at the 5' end of the TK gene fragment (Fig. 1). Standard synthetic TK RNA was made from a third TK fragment of 532 bp defined by PstI (nt 429) and SacI (nt 961) cloned into pGeml and transcribed by SP6 polymerase. TK RNA was standardized to the ribosomal protein L32 RNA (15) as described by others (11). TK enzyme activity was measured by a standard assay (23) but with 0.4 mM TTP included to inhibit endogenous TK (2). Tissue extracts were prepared from frozen tissues by homogenization in 500 ,ul of 50 mM Tris-HCl (pH 7.5)-5 mM mercaptoethanol-5 ,uM thymidine. Insoluble material was removed by centrifugation at 10,000 x g for 45 min at 4°C. Protein concentration was determined by the method of Bradford (7). RESULTS The position-independent, copy number-dependent expression of K18. Previous studies indicated that each of three lines of K18 transgenic mice expressed the K18 transgene in appropriate adult tissues, at levels directly proportional to the copy number and with the same efficiency as in the endogenous mouse homolog (1). To confirm and extend these studies, additional transgenic mice were prepared from the full 10-kb K18 gene fragment (Fig. 1) and two smaller fragments that represent simple deletions of the 5' and 3' flanking portions of the K18 gene. The K18-Nsi mice re-
2216
NEZNANOV ET AL.
MOL. CELL. BIOL.
10kb Nsi
H
K18
r=
1II
Xm B
X
1
2
U 4 5 6
3
H
MY
*E
rmA
7
8.5kb
Kl 8-Nsi
Xm
6.5kb
Kl 8-Xmn B
B
(X) RI B RV
NNTK ATG
Not
I AATAAA 9.3kb
FIG. 1. Schematic representation of the K18 gene and the fragments used to generate transgenic mice and the NNTK transgenic construction. The K18 gene is shown by the thick dark line, with exons represented by filled boxes. Locations of two Alu sequence elements are shown by the open boxes, with the direction of putative transcription indicated within the boxes. DNase-hypersensitive sites are indicated by the vertical arrowheads. K18-Nsi transgenic mice were generated by injection of the 8.5-kb fragment defined by digestion with NsiI (Nsi) and HindIII (H). K18-Xmn mice received the 6.5-kb fragment generated by digestion with HindIII and XmnI (Xm). The NNTK vector represents the HSV TK gene (TK) flanked by the 5'- and 3'-flanking regions of the K18 gene. The TK gene is represented by the open box. Positions of the ATG translation initiation codon and the AATAAA polyadenylation signal are indicate below the map. Additional restriction enzyme sites are indicated for EcoRI (RI), BamHI (B), BglII (Bg), EcoRV (RV), Notl (Not), and XhoI (X). The XhoI site in parentheses was destroyed during construction.
ceived an 8.5-kb fragment missing the first 1.5 kb of the 5'
flanking sequences. This deletion removes one of two Alu repetitive elements and two apparently conserved distal sequence elements but retains the DNase-hypersensitive sites in the 5' end (31). K18-Xmn mice received a 6.5-kb fragment which lacked the 3' 3.5 kb of the K18 isolate. This deletion leaves 155 nt of DNA downstream of the last exon of K18. Previous studies by transient expression of transfected constructions had not identified any essential regulatory elements within these distal flanking regions of the K18 gene (32). The arrangements of the transgenes were evaluated by Southern blot analysis. Figure 2A shows typical data for four of the K18 and two of the K18-Nsi (5'-deleted) transgenic mouse lines. The K18 fragments generate the expected 10-kb fragment, while the K18-Nsi lines generate the expected 8.5-kb fragment and additional fragments of approximately 4.7 and 12.5 kb consistent with additional head-to-head and tail-to-tail arrangements of some of the transgenes. Fragments other than those expected for both the K18 and K18-Nsi mice likely represent single-copy elements of the genes positioned at the distal ends of the integrated arrays or a minority of copies which have suffered deletions or other rearrangements. However, in all cases except K18 Nsi-1, the major hybridizing fragment was consistent with a tandem head-to-tail array (data not shown). Because of the ambiguity in counting distal fragments and potential head-to-head and tail-to-tail arrays as active genes, copy number was estimated by quantitative dot blot analysis (see Materials and Methods). The abundance of K18 RNA was estimated by both Northern blot analysis and RNase protection analysis, using simultaneous analysis of either endogenous mK8 RNA or the RNA for ribosomal protein L32 for normalization. Previous results have confirmed that overexpression of K18 does not affect mK8 RNA expression (1, 25, 26). Figure 2B shows the result of a typical RNase protection analysis of
liver and intestine RNAs from representative K18 and K18Nsi lines. The levels of K18 RNA from different individual animals from the same strain (Fig. 2B, lanes 1 and 2 or lanes 8 and 9) were very similar. The complementary endogenous mK8 RNA that was used as an internal standard was very highly expressed in intestine. However, the absolute levels of K18 RNA were similar in liver and intestine in any one line. Northern blot analysis of RNAs from different tissues of both K18 and K18-Nsi mice confirmed that intestine, liver, kidney, and lung were the tissues with highest levels of K18 RNA in both types of mice (data not shown). Overall, no qualitative or quantitative difference in the level or tissue specificity of the K18-Nsi mice was detected. Both the K18 and K18-Nsi fragments are expressed proportionally higher with increasing numbers of transgenes in liver (Fig. 3A), intestine (Fig. 3B), kidney, and lung (1) (additional data not shown). These results confirmed that K18 gene expression is copy number dependent and independent of the site of integration. In addition, the distal 1.5 kb of the 5'-flanking sequence of the K18 genomic fragment, deleted in the K18-Nsi mice, is dispensable for the correct adult regulation of the gene. Copy number-dependent expression in liver and intestine requires sequences flanking the 3' end of the K18 gene. In contrast to the results of K18-Nsi transgenic mice, deletion of the 3'-flanking sequences of the K18 gene in K18-Xmn mice resulted in a dramatic loss of expression in adult liver. Figure 4 shows the results of a Northern blot analysis of K18 RNA from different K18-Xmn lines of mice. K18 RNA was very low in the livers of all lines of the K18-Xmn mice (Fig. 4A, lanes 2, 7, 12, 17, 22, and 27). The results of estimation of liver RNA levels of the K18-Xmn mice are compared with those of K18 and K18-Nsi mice in Fig. 3A. In addition, the levels of intestinal K18 RNA were found to deviate from the linear relationship of RNA level and copy number found for K18 and K18-Nsi mice (Fig. 3B). However, expression in
TRANSCRIPTIONAL INSULATION OF THE HUMAN K18 GENE IN MICE
VOL. 13, 1993
A.
B. Intestine
Liver
-
11 Nsi line: Gene copy/cell Std 1 2 3 4 6 1 2 0 1 4 6 10 20
K18TG line:
I
2217
1
Kl8 line:
__
lI
4 4
1 2
3 6 4
..
4 1
Liver Intestine m I m Nsi line: I II
--
2 3 6
1
2 1
2
-23
-9 6 -
- 6.6
hKl 8
-4 4
- 2.4 -
*
2.0
-*
-4- mK8
-0.56 1
2 3 4 5 6 7 8 90
11 12 13
14
DNA
1
2 3 4
5
6 7 8 9
10
11 12 13 14 15 16
RNA
FIG. 2. Southern blot analysis of K18 and K18-Nsi transgenic mouse DNAs. Five micrograms of DNA was digested with BglII, separated by agarose gel electrophoresis, transferred to a charged nylon filter, and hybridized with the K18 cDNA probe. DNAs from four K18 transgenic mouse lines (K18TG lines 1, 2, 3, 4, and 6) and two K18-Nsi lines (Nsi lines 1 and 2) are indicated at the top. Gene copy standards are represented by the cloned K18 gene diluted in normal mouse DNA; 10.7 pg of plasmid DNA represents the one-gene-copy standard. Std, radioactive HindIII-digested lambda DNA fragments (indicated at the right in kilobases). (B) RNase protection analysis of 10 pg of total RNA from the livers or intestines of the indicated transgenic mice. RNA was hybridized to 2 x l05 cpm of both the K18 and mK8 probes, digested with RNase A and T1, and analyzed by acrylamide gel electrophoresis and autoradiography. Positions of the protected fragments of human K18 (hKl8) and the endogenous mK8 RNAs are indicated at the right.
lung, kidney, and brain was strong and increased with increasing copy number (Fig. 3C). Furthermore, expression in spleen remained undetectable (Fig. 4A lanes 5, 15, 20, and 30; RNAs in lanes 10 and 25 were degraded). The range of gene copies within the mice from which intact spleen RNA was isolated was sufficient to allow us to conclude that K18 was not expressed in spleen tissue. Similarly, expression in heart was extremely low (data not shown). Thus, the tissue specificity of expression in K18-Xmn mice was the same as for K18 and K18-Nsi mice and for the endogenous mK18 gene (1). Table 1 compares the efficiency of expression for each of the three types of transgenic mice in liver, intestine, and kidney. Estimates of the abundance of K18 RNA were divided by the gene copy number. The values for each line of one construction were then averaged to generate a mean and a standard deviation reflecting the variation of the different individual lines of mice. Both the mean and variation of the values for liver, intestine, and kidney of K18-Nsi mice were close to those of K18 transgenic animals. The values for liver RNA of K18 mice represent approximately 150 K18 RNA molecules per gene per cell. However, the efficiency of expression of K18 RNA in the livers of K18-Xmn mice was only 8% of those of the other two constructions. In addition, the variation among K18-Xmn was four to six times greater. Not only is the level of K18 RNA low in the livers of K18-Xmn mice, but the residual levels of expression are not proportional to copy number. The K18 gene was expressed
only slightly less efficiently in intestines of K18-Xmn mice. However, the variation between individual lines was much greater than for K18 or K18-Nsi mice. This finding suggests that intestinal K18 expression is much more sensitive to its site of integration or its neighboring transcription units in these mice than K18 or K18-Nsi mice. In contrast, both the average expression per gene and the variation between lines were similar in kidney for all three types of transgenic mice. The 3' end of the K18 genomic fragment is essential for copy number-dependent expression in liver but not in kidney, lung, or brain. The 3'-flanking sequences of the K18 gene also appear necessary for the insulation of the K18 gene in intestine. Expression of the HSV TK gene is copy number dependent and integration site independent when flanked by K18 distal 5' and 3' sequences. To test whether the characteristic of position-independent, copy number-dependent expression was independent of the regulation of the K18 gene, the HSV TK gene, including both promoter and coding elements, was inserted between the distal flanking sequences of the K18 gene (Fig. 1, vector NNTK). This vector lacks the proximal promoter and all internal regulatory elements (27, 32) of the K18 gene. Two nonmosaic male and two nonmosaic transgenic female mouse founders were identified. Both males were sterile and were sacrificed for analysis. Male progeny of two female founders were analyzed. One of these strains, NNTK-26, integrated transgenes into two different places.
2218
NEZNANOV ET AL.
MOL. CELL. BIOL.
Liver
A
z * K18 *
Kl 8-Nsi
O Kl 8Xmn
0
10
20
30
40
50
60
Genes/Cell
B
Intestine 200-
150
z 100'
0o v-
50 -
10
0
20
30
40
50
60
Genes/Cell
Kl 8-Xmn
c 350
0
300-
H
250-
z
200-
c0
150-
v-b ye
a
a
100'-
100 A
Kidney
a
x
A Brain
50-
O Lu ng
50 0
10
20
30
40
50
60
Genes/Cell FIG. 3. Copy number dependence of K18, K18-Nsi, and K18Xmn transgenic mouse lines. K18 RNA (picograms per 10 pg of total RNA) is plotted as a function of the number of K18 genes estimated for each line. K18 RNAs were determined by both RNase protection and Northern blot experiments. Data shown for K18 and K18-Nsi mice are derived in part from the experiments shown in Fig. 2B. Gene copy number was determined by quantitative dot blot analysis. (A) Liver RNA. Note the low and gene copy-independent level of K18 RNA from the livers of K18-Xmn mice. (B) Intestine RNA. Note the deviation of RNAs from K18-Xmn mice from linearity and from the values of the K18 and K18-Nsi mice. (C) Other organs for K18-Xmn mice. Note the relatively linear relationship of K18 RNA and copy number in each of the organs.
Subsequent breeding of the founder resulted in the segregation of two different integration sites, thus generating an additional different line. Southern blot analysis of the five NNTK lines is shown in Fig. 5. Two different animals of lines NNTK-26A and NNTK-26B were included (Fig. 5, lanes 4 and 5 and lanes 6 and 7). The common 9.3-kb fragment detected in each line indicates that the predominant form of integration was the expected head-to-tail array. Both the intensities of the 9.3-kb fragments and the presence of smaller fragments found in the NNTK-26B mice confirm the different integration sites responsible for the segregation of the NNTK-26A and NNTK-26B lines of mice. In a survey of different tissues, TK RNA was detected only in testes and, at a much lower level, in brain. However, like K18 and K18-Nsi transgenic mice, every NNTK transgenic mouse expressed detectable levels of transgenic RNA. Representative results of an RNase protection assay are shown in Fig. 6. Two major sets of protected fragments were detected in testes. The larger protected fragment of 420 nt was specific for the TK RNA, as shown by the tRNA control (lane 7) and additional assays of various positive and negative tissues (data not shown). The smaller fragment of about 150 nt corresponds to the internal, cryptic promoter previously reported for other transgenic mice containing constructs that used the coding portion of the TK gene as a reporter gene (3). The levels of both RNAs increased linearly with increased copy number, as shown in Fig. 7A. This linear relationship was also true for the lower amounts of the longer TK RNA found in the brains of transgenic animals (Fig. 7C). No transcripts from the cryptic promoter were detected in brain (data not shown). Expression of TK in the testes of transgenic animals leads to sterility due to defective sperm (3, 8). None of the NNTK transgenic males was able to generate progeny, suggesting that each expressed TK enzyme in the testes. In confirmation of this suggestion and the RNA analysis, testes TK enzyme activity increases linearly with copy number (Fig. 6B).
DISCUSSION K18 expression in adults is restricted primarily to singlelayered epithelial tissues of diverse functions and embryonic origins. The promiscuous expression of the cloned K18 gene in transfected cultured cells that do not express endogenous mK18 (26) has led to the hypothesis that the tissue-specific expression of the gene may involve cis-acting mechanisms acquired during normal development which are bypassed by direct transfection (26, 34). The demonstration that transgenic K18 is expressed in the correct cells of complex organs such as kidney as well as in liver, lung, and intestine but not in spleen or muscle (1) confirms that the cloned gene contains sufficient genetic information for tissue-specific expression if provided with a normal developmental environment. Results for the additional K18 transgenic mice described here confirm this conclusion and explore the unexpected finding that K18 expression is unusually efficient and insensitive to the sites of integration of the transgenes. Mice from every transgenic line containing either the full 10-kb K18 fragment or the 5'-deleted, 8.5-kb K18-NsiI fragment expressed K18 RNA at levels proportional to the number of transgenes. In addition, the efficiency of expression from each copy of K18 is very similar and comparable to that of the endogenous mK18 beta-1 coding gene (1). As the sites of integration for each line are different, we conclude that K18 expression is integration site independent and copy number dependent. The efficient expression of K18 in
TRANSCRIPTIONAL INSULATION OF THE HUMAN K18 GENE IN MICE
VOL. 13, 1993
2219
K18-Xmn line
A 4
12
6
10
li
li lu k s b
i lu k s b
li lu k s b
li lu k s b
li lu k s b
U
I.
12 13 14 15 16 1718 19 20 21
22 232425 26
K18-*-
0
1
B
15
13
C
2 3 4 5 6
4
C
li
L32--
li lu k
s
b
7 8 9 10
12
| r
li
11
lu
k
6 s
b
li
lu k
10 s
b
li
2 3 4 5 6
7 8 9 1011
12 13 14 15 16
I. 27 28293031
15
13 s
b
i
lu k
s
b
li
lu k
s
b
l
_*
_
,*0 1
lu k
li lu k s b
1718 19 20 21 22 23 24 25 26
27 28 2930 31
FIG. 4. Northern blot analysis of the RNAs from various organs of K18-Xmn transgenic mice. Panels A and B represent the fluorographically enhanced, autoradiographic X-ray film signal obtained after the sequential hybridization of a single filter holding the RNAs from six different K18-Xmn transgenic lines with the K18 and ribosomal protein L32 gene probes. (A) K18-Xmn transgenic lines are indicated by the numbers above each set of lanes and the control (C) K18TG1, heterozygous transgenic liver RNA. Organs are indicated above each lane and abbreviated as follows: li, liver; lu, lung; k, kidney; s, spleen; b, brain. (B) L32 signal.
transgenic mice suggests that nearly all full copies of the gene are equally active. Some K18 copies may be partially deleted or rearranged, as suggested by extra bands in the
Southern blot analysis. However, the majority of copies of the K18 transgenes in each mouse line appear full length. While it is possible that some copies of the gene are particularly active while others are silent, the analysis of K18 DNase-hypersensitive sites in transgenic livers supports the suggestion that most, if not all, of the copies of K18 are active (31). This view implies that each K18 copy is transcriptionally insulated from its site of integration and from each other.
The K18 properties of position-independent, copy number-dependent expression require different sequences in different tissues. In liver and intestine, transcriptional insulation requires the 3.4 kb of DNA flanking the immediate 3' end of K18. However these sequences are not necessary for efficient, insulated expression of K18 in kidney, lung, and brain. The concept of insulating a modestly active gene from neighboring regulatory elements might require a buffer or boundary sequence on both sides of a transcription unit, as has been described for the specialized chromatin structure sequences of D. melanogaster (24) and the A elements of the chicken lysozyme gene (6, 38). In transgenic mice, the
TABLE 1. Efficiency of K18 expression in transgenic mice pg of K18 RNA/10 jig of RNA/gene)a (% of K18 in
Line
K18 K18-Nsi K18-Xmn
Liver
liver)b Kidney
Intestine
Mean + SD
n
Mean ± SD
n
Mean ± SD
n
12.4 4.0 (100 32) 13.2 + 2.6 (106 + 20) 1.0 ± 1.2 (8 + 120)
5 6 7
8.2 2.9 (66 35) 5.8 ± 0.8 (47 ± 14) 4.5 ± 5.7 (36 ± 127)
5 6 5
6.5 ±2.3 (52 35) 9.0 ± 4.2 (72 ± 47) 7.7 + 4.2 (62 ± 54)
4 6 6
a K18 RNA was estimated by either RNase protection or Northern blot anhlysis. The average signal for the ribosomal protein L32 RNA was used as the standard of 10 ,ug of total RNA. K18 estimates were adjusted to 10 ,ug of total RNA and divided by the copy number. The number of transgenic lines contributing to the mean and standard deviation is indicated by n. The values for K18 transgenic lines in liver represent approximately 150 K18 RNA molecules per cell per gene, based on the estimate of approximately 33 pg of total nucleic acid per cell (30) and 6 pg of DNA per cell. b In liver, this value is approximately equal to the endogenous mK18 values (1). The standard deviations are expressed as a percentage of the absolute values.
2220
NEZNANOV ET AL.
MOL. CELL. BIOL.
NNTK liri
JO 264;A ,26426g 26B6 12 30. 2
3
41
5 (;
A 83
Probe 1 72 n0
3
325 rit Std
Protected:
420 't
50 rt
HSV TK
9.3Kb -*
Ir atIi B
to
3 14-31Kb STD
P atg R~~~~~
atg RV
41 12 30 A B
probe
234561 78
_,
-1 596 820
-652
.w 420.
- 544
2
42
454
357f
-3
*
.
common head-to-tail tandem duplication of transgenes could provide such elements on both sides of most genes of the cluster. If the flanking insulating element functions independently of orientation, the duplicated 5'-flanking region of the K18 gene could serve this 3'-flanking function in K18-Xmn mice in kidney, lung, and brain. Unfortunately, the absence of single-copy K18 transgenic animals prevents a conclusion concerning the minimal requirements for copy number dependence. However, further analysis of the sequence requirements of the 5'-flanking element in the absence of the 3'-flanking elements will be possible, at least for kidney and lung expression. The K18 3.4-kb 3'-flanking sequence may contain a liverspecific regulatory element not needed for expression in other tissues. Such an element might be similar to the powerful, tissue-specific transcriptional enhancer activity of the globin LCR, which is a necessary but not necessarily sufficient requirement for copy number-dependent expression of the locus in transgenic mice (16, 37). In the case of K18, expression in other tissues may have different regulatory element requirements, just as mechanisms of expression of other genes use different subsets of regulatory elements in different tissues (29, 42). While transient expression analysis of the K18 gene failed to detect activator sequences in the 3'-flanking DNA, the analyses were not performed with primary liver cells (32). Nevertheless, even if a liver-specific regulatory element exists in the 3'-flanking region, K18 RNA was detectable, albeit at much lower levels, in the livers of each of the K18-Xmn mice. However, the residual level of liver K18 RNA did not increase with increasing copy number. Whether loss of copy number dependence and loss of efficient liver expression are due to separable genetic elements remains to be determined. However, in intestine this may be the case because the average level of expression per gene for K18-Xmn lines is compara-
-
NNTK Lirie
B
I
FIG. 5. Southern blot analysis of the DNAs of NNTK transgenic mice. In lanes 1 to 7, 10 p.g of DNA from the indicated lines of mice was digested with BglII, separated by agarose gel electrophoresis, blotted to a charged nylon membrane, and hybridized with a random-primed fragment of the 5' end of the HSV TK gene (BamHI to BglII; Fig. 1). Lanes 8 and 9 contained the equivalent of six and two copies per cell of the NNTK fragment used for the generation of the transgenic mice mixed with 5 ,ug of normal mouse DNA and digested with BglII. The 9.3-kb size of the major hybridizing band is the expected size for head-to-tail duplications of the injected fragment. The probe hybridizes to the 3.1-kb 5' portion of the nonduplicated NNTK fragment. STD, standard.
i
325 Std
193
176
. 156 150 Cryptic
129
FIG. 6. TK RNA analysis. (A) A partial map of the 5' end of the HSV TK gene is shown by the heavy line. Putative transcriptional initiation sites and direction are indicated by the bent arrows. The portions corresponding to the probe or the nonradioactive standard RNA and observed protected fragments are shown by the thin, horizontal arrowed lines. B, BamHI; R, EcoRI; RV, EcoRV; atg, methionine codons within the reading frame of TK. (B) Total RNA from testes of the five indicated lines of transgenic mice was hybridized to the TK probe, digested with RNases A and T1, and analyzed by acrylamide gel electrophoresis in 8 M urea and autoradiography; 10 ,ug of RNAs for each line except 26B (lane 6, 6 ,ug) was used. RNA amounts were standardized independently by hybridization with ribosomal protein L32 (data not shown). Size markers are shown at the left in nucleotides. Lane 7 represents the tRNA control; lane 8 represents 4 pg of the synthetic standard TK RNA (Std). The major protected fragment representing the previously reported testes-specific, internal promoter is indicated at the right (Cryptic).
ble to that of K18 mice, but the great variation from line to line suggests that intestinal expression (but not kidney, lung, or brain expression) is greatly influenced by the particular site of integration. When combined with K18 distal flanking sequences, HSV TK gene expression is copy number dependent and position independent though strictly restricted in tissue specificity.
TRANSCRIPTIONAL INSULATION OF THE HUMAN K18 GENE IN MICE
VOL. 13, 1993
Testes TK RNA
A < 0.6
.
z
0.4 .'Z
75 0.2
A Ah. 0
0.0 0
15 10 5 TK Genes/Cell
Testes TK Enzyme
B
A
A A A
5
0
10 15 20 TK Genes/Cell
Brain TK RNA
C 8
0
