![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 15, No. 12, pp. 1506-1517, June 15, 2001
Department of Neuroscience, University of Virginia, Charlottesville, Virginia 22908, USA
 |
ABSTRACT |
|---|
 Top
 Abstract
 Introduction
Results
Discussion
Materials and methods
References
|
|---|
We report the successful transfer of a fully functional lac operator-repressor gene regulatory system to the mouse. The key component is a lac repressor transgene that resembles a typical mammalian gene both in codon usage and structure and expresses functional levels of repressor protein in the animal. We used the repressor to regulate the expression of a mammalian reporter gene consisting of the tyrosinase promoter embedded with three short lac operator sequences and the tyrosinase coding sequence. Pigmentation of the mouse was controlled by the interaction of the lac repressor with the regulatable Tyrosinase transgene in a manner that was fully reversible by the lactose analog IPTG. Direct control of mammalian promoters by the lac repressor provides tight, reversible regulation, predictable levels of de-repressed expression, and the promise of reversible control of the endogenous genome.
[Key Words: lac repressor; gene regulation; transgenic mice; tyrosinase; pigmentation; albino]
| |
Introduction |
|---|
|
Top
Abstract
Introduction
Results
Discussion
Materials and methods
References
|
|---|
The mouse is the most widely used experimental animal to model mammalian development and disease. In recent years, technological advances in genetic manipulation of the mouse genome have made it even more valuable as a research tool, and a great deal of information has been learned from conventional knockout and transgenic experiments. Despite this, it can be difficult to draw definite conclusions from these studies. Embryonic lethality might preclude the possibility of analyzing adult phenotype, or activation of compensatory systems confuse the analysis. Tight, reversible control of gene expression would greatly broaden the possible experimental questions that can be addressed. To gain such control, the ideal system would enable a target gene to be switched on and off repeatedly, without affecting the expression of nontargeted genes. With this ideal in mind, we have developed the lac repressor regulatory system for use in the mouse.
The lac operon of Escherichia coli consists of a set
of genes coordinately regulated by lactose (Jacob and Monod 1961
). The regulatory components of the system are the lac repressor and its DNA-binding sequence, the lac operator. In the absence of lactose, lac repressor occupies the lac operators and
prevents transcription. Lactose causes a conformational change in the
repressor, and it vacates the operators, allowing RNA polymerase to
gain access to the promoter and initiate transcription.
Hu and Davidson (1987) were the first to use lac elements to
control reporter-gene expression reversibly in mammalian cells. Their
results were extended by Figge et al. (1988), who demonstrated that the
lac repressor was able to gain access to the mammalian chromosome to regulate a stably integrated reporter gene.
Our goal has been to adapt the lac regulatory system to
control gene expression in the mouse. In an earlier paper, we reported that transgenes containing the bacterial-coding sequence for the lac repressor downstream of the 
-actin promoter were
heavily methylated and only transcribed in the testis of transgenic
mice (Scrable and Stambrook 1997). Methylation and silencing in mice also was observed by Wyborski et al. when the bacterial lac
repressor sequence was downstream of the F9-1 polyoma promoter
(Wyborski et al. 1996). To create an active transgene, we changed the
primary DNA sequence of the bacterial lacI gene to resemble a
mammalian coding sequence more closely and still code for the same
amino-acid sequence. This synlacI gene was transcribed widely
(Scrable and Stambrook 1997). However, we could not detect the synlacI
protein product, either by Western blot or immunocytochemically, in the tissues of SynlacI transgenic mice or in cells transfected with the
synlacI transgene. We also could not detect any effect of the
repressor from synlacI on reporter-gene activity, either in embryonic mouse cells from transgenic animals, or in melanoma, neuronal, or kidney cell lines transfected with the synlacI
transgene (data not shown).
In this paper, we describe how we changed the DNA sequence and gene structure of lacI so that it expresses functional lac repressor protein ubiquitously in mice. We show that the lac repressor can repress the activity of a reporter gene, which subsequently can be derepressed by IPTG in the drinking water of an adult or the mother of an embryo or nursing pup. The lac operator-repressor system brings the temporal dimension of mammalian gene expression in the animal under experimental control that is both reliable and predictable, and should make it possible to introduce even lethal mutations into the mouse genome routinely and to study them at the organismal level.
| |
Results |
|---|
|
Top
Abstract
Introduction
Results
Discussion
Materials and methods
References
|
|---|
Reversion of four bases to the wtlacI sequence corrects a cryptic splice site in synlacI
To determine why no protein was observed from the synlacI
transgenes, we analyzed the RNA transcripts for evidence of processing errors. We consistently observed a single transcript on Northern blots
of RNA from animals transgenic for the bacterial, or wtlacI (W) construct, but a doublet in RNA from animals with the
re-encoded, or synlacI (S) transgene (Scrable and Stambrook
1997; Fig. 3, below). The sizes of the two transcripts with
synlacI suggested that the upper band represented an unspliced
transcript still containing an intron from the human -actin
promoter, and the lower band, the spliced transcript. On careful
examination of synlacI and wtlacI transcripts, we
found that the processed synlacI transcript was actually
slightly smaller than the wtlacI transcript. Incorrect
splicing at a cryptic splice acceptor site downstream of the start
codon could explain both the smaller size of the RNA and the absence of
protein product. We developed an RT-PCR assay, with primers that flank
the region encompassing the intron in the -actin promoter and the
translation start site in the lac repressor coding region
(Fig. 1A, arrows). The difference in size
between the correctly spliced W product and the correctly spliced S
product (513bp vs. 499 bp) is due to a difference in the size of the
region linking the promoter to the coding region (see Materials and
Methods). As can be seen in Figure 2A, the predominant product in RNA from the W animal is the correctly spliced
product, whereas the predominant product in the S animal is smaller.
This suggests that an aberrantly large region is spliced out of the
synlacI transcripts.
|
|
We prepared a series of chimeric repressors made by exchanging the 5' region of the wtlacI or synlacI constructs, as represented schematically in Figure 1A (5'C1-5'C4) and analyzed the splicing patterns of the constructs in transfected cells. Only one of these (5'C1) produced the correctly spliced product as the predominant PCR product (Fig. 2A). We correlated the splicing pattern with the presence of specific regions from W and S, and we found that the first 36 bp of the synlacI coding sequence was the only region consistently associated with incorrect splicing. Within these first 36 bp of coding sequence, the wtlacI and synlacI constructs differ by only four bases, as shown in Figure 2B.
Analysis of potential splice sites was performed using the Walkers
sequence analysis program (Rogan et al. 1998). The only potential donor
site in the transgenes is the splice donor site in the -actin
promoter. For the splice acceptor, in addition to the site in the
-actin promoter, there is a potential site just downstream of the
four bases that differ between W and S in the lacI coding
region (Fig. 2B). Based on the size of the smaller product identified
by our RT-PCR assay, this potential site is the acceptor site utilized
in synlacI. We cannot explain why this acceptor site is used
rather than the splice-acceptor site in -actin, which by current
models should be stronger.
Reversion of a 3' region of the coding sequence to wtlacI corrects a translational block in synlacI
To test whether the constructs coded for functional lac
repressor protein, we introduced each into Rat2 fibroblasts by
transient transfection along with the regulatable target gene,
SVOZ (Fig. 1B). SVOZ consists of a modified SV40
early promoter with a single lac operator between TATA and the
transcription start site (Brown et al. 1987), and the -galactosidase
(lacZ) reporter gene (which was shown to be regulatable with
the lac repressor and IPTG in Liu et al. (1989). Reporter-gene
activity was assessed under the following four conditions: (1)
SVOZ alone (the basal level of activity); (2) SVOZ + IPTG (the effect of IPTG on unrepressed activity); (3) SVOZ + lac repressor (repression); and (4) SVOZ + lac repressor + IPTG (derepression). In all experiments, there was no significant difference between conditions 1 and 2, indicating that IPTG alone had no effect on the unrepressed level of expression. The average of conditions 1 and 2 was taken as the baseline expression for each transfection. The numbers of blue cells under conditions 3 and
4 were expressed as percent of the unrepressed baseline.
The results for the constructs W, S, 5'C1, and 5'C2, are shown in Figure 2C. As expected from our previous studies, W encodes a functional repressor whereas S does not. However, 5'C1, which was found to splice correctly, does not encode a functional repressor. And surprisingly, 5'C2, which gives predominantly an incorrectly spliced product like S, does code for functional repressor, although the level of repression is not as tight as with W (31% vs. 7%). These results indicated that, in addition to splicing, there was a second problem with the synlacI coding region that affected the expression of functional repressor.
We made a second set of chimeras in which 3' regions of the W and S coding sequences were exchanged. These constructs are represented schematically in Figure 1A (3'C1-3'C4). As represented graphically in Figure 2D, both 3'C2 and 3'C4 had repressor activity, whereas 3'C1 and 3'C3 did not. These data indicate that the problematic region in synlacI is in the 150 bp between the 5' EcoRV site and the 3' PvuII site. Within this region, W and S differ by only 16 bases. We could find no coding error, cloning artifact, or other mutation in this region that could account for the lack of repressor function. We used site-directed mutagenesis to change all 16 bases in synlacI, in groups of two to five to the wtlacI sequence. None of these smaller changes could restore repressor function on their own (data not shown).
We hypothesized that loss of repressor function was the result of a
synlacI sequence element located between the EcoRV
and PvuII restriction sites that blocked RNA transcripts from
being translated. This hypothesis is supported by our finding that a synlacI construct that splices correctly (5'C1) is
expressed strongly at the RNA level, but is only barely detectable at
the protein level (data not shown). Two mechanisms that could account
for this translational block are: (1) nuclear retention of the RNA such
that it does not reach the cytoplasm to be transcribed; or (2) tight
secondary structure of the RNA such that the translation machinery
cannot access the primary sequence. The first hypothesis was ruled out
by finding that equal levels of lacI message were seen in
the total and cytoplasmic fractions of both wtlacI and synlacI RNA extracts on Northern blot (data not shown). The
second hypothesis was ruled out by analyzing the secondary structure predicted for each RNA sequence (Zuker et al. 1991) in which no significant differences were found (data not shown). Consequently, we
can conclude that the expression of functional lac repressor activity is blocked by the region of the synlacI sequence
between the EcoRV and PvuII sites, but the mechanism
responsible for this block is not known.
Restructuring optimizes transcription of CpG-rich lacI transgenes in the mouse
Having identified a modified synlacI coding sequence that was correctly spliced and translated in transfected cells (3'C4), we derived transgenic mice with this construct. We analyzed lacI expression by Northern blot, and the results are shown in Figure 3. We found that 3'C4 transgenes were expressed only in testis, resembling transgenes composed entirely of bacterial coding sequence. (See Materials and Methods for details on the mouse lines for each transgene).
|
We knew from our previous studies that the content of CpG dinucleotides
in the coding sequence was a major determinant of RNA expression in the
animal (Scrable and Stambrook 1997). The bacterially derived W
construct contains 97 CpG dinucleotides in the coding region and 5'
linker and is expressed only in the testis of transgenic mice. Removal
of all but five of these CpG dinucleotides resulted in the S construct,
which was expressed in all tissues (Scrable and Stambrook 1997). To
examine the arrangement of CpG dinucleotides in each transgene more
closely, we created CpG-density maps of each one and then aligned them
with the CpG-density map of the -actin locus (Fig. 3). In the
endogenous -actin locus there is a CpG island, which extends from
the proximal promoter into the -actin coding region. The high CpG
content of W downstream of the -actin promoter results in
recapitulation of the CpG island in the -actin locus, but both S and
3'C4 are relatively devoid of CpG in this same region. However, S is
expressed like -actin and 3'C4 is expressed like W.
Next, we made two constructs in which either W or 3'C4 coding regions
were flanked with noncoding regions derived from the rabbit -globin
gene (constructs M and R, respectively). This added CpG-poor sequences
to the transgenes, and moved the respective lacI coding
regions farther away from their promoters. As shown in Figure 3, the M
line does have a more widespread expression pattern than the W line,
although it is not ubiquitous. However, the combination of the
relatively CpG-poor coding region of 3'C4 and the flanking globin
sequences (R) results in global lacI transcription (Fig. 3).
One possible explanation for these different expression patterns is
that an open structure that allows for transcriptional activity relies
on a downstream element in the -actin coding region that is free of
CpG dinucleotides. There are several segments downstream of the
promoter that are devoid of CpG. Two of these -actin elements, and
their analogous regions in each of the lacI constructs, are
indicated by the dashed lines in Figure 3. In W, both of these elements
are CpG-rich and transgene expression is restricted to testis. In S,
the regions are CpG-poor, and expression is nearly ubiquitous. In 3'C4,
replacement of the sequence between the EcoRV and
PvuII sites in S with the corresponding sequence in W changed
the distal element from one devoid of CpG to one that is CpG-rich, and
the expression pattern from one that is ubiquitous to one restricted to
the testis. Flanking the coding region of 3'C4 with the globin
sequences shifted CpG-poor sequence into the two regions, and resulted
in nearly ubiquitous expression, similar to the expression pattern of
the endogenous -actin gene.
The lac repressor is widely expressed and functional in the LacIR transgenic mice
The key to using the lac repressor system in mice is expression of functional levels of repressor protein. We prepared tissue protein extracts and tissue sections from mice transgenic for the R transgene (lacIR) and analyzed the expression of the lac repressor by Western blot (Fig. 4A) and immunohistochemistry (Fig. 4B). There is no evidence of a translational block in any tissue tested (brain, thymus, heart, lung, liver, spleen, kidney, testis, muscle, and skin). The strong nuclear staining of the protein seen with immunohistochemistry is a result of a nuclear localization signal sequence appended to the carboxy terminal of the lac repressor (See Materials and Methods).
|
To determine if the lac repressor in this line is functional,
we prepared cultures of primary embryonic cells from R animals and
transfected them with the regulatable SVOZ construct. The embryo-derived cells exhibited strong lac repressor activity, as there was very little -galactosidase expression in the absence of
IPTG in the culture media (Fig. 5). A
titration curve of the amount of IPTG needed to derepress expression
shows that as little as 50 µM IPTG in the culture media allows for
full levels of expression (Fig. 5A), which correlates well with the
level needed to derepress transcription at the bacterial lac
operon (Cho et al. 1985). At 20 mM IPTG in the media, we observed some
toxicity, indicated by decreasing numbers of
-galactosidase-positive cells (Fig. 5B). Although the level at
which toxicity was observed agrees well with the findings of Figge et
al. (1988), we found that full derepression was achieved at a much
lower concentration of IPTG.
|
Target gene activity is controlled by LacIR and IPTG in the mouse
We tested the lac operator-repressor system in mice using a
regulatable version of a well-characterized visible marker gene, tyrosinase. Tyrosinase is the protein product of the albino
(c) locus (Kwon et al. 1987), and is the enzyme that catalyzes the first step in melanin biosynthesis. The target transgene consists of
the wildtype murine tyrosinase cDNA under the control of the murine
tyrosinase promoter modified to contain lac operator
sequences. The major transcription start site in the tyrosinase
promoter is 83 bp upstream of the start codon. To maintain the
endogenous spacing of promoter elements in the critical region between
the start of transcription and the start of translation, we used
PCR-based, site-directed mutagenesis to change 25 bp of the endogenous
sequence to create a primary lac operator centered at 59 bp
upstream of the start of translation. Additional operators were
inserted 176 bp and 526 bp upstream of the primary operator (Fig. 1B).
Mice containing this modified Tyrosinase transgene resemble
pigmented animals previously described (Yokoyama et al. 1990; Methot et
al. 1995) that had been microinjected with an unregulatable version of
the same transgene. We established two lines of pigmented Tyrosinase transgenic mice with the regulatable transgene. The TyrlacO-25 line displays a himalayan pigmentation pattern (as
shown in Figs. 6A and 7C), and the
TyrlacO-43 displays a light pigmentation pattern, similar to
those described in Methot et al. (1995).
|
|
We crossed mice transgenic for the Tyrosinase transgene to mice transgenic for LacIR. In double transgenics, the lac repressor should bind to the operator sequences located in the tyrosinase promoter, block transcription of tyrosinase, and revert pigmented animals to albino. An example of a TyrlacO-25, LacIR double-transgenic mouse is shown in Figure 6A next to a mouse transgenic for TyrlacO-25 alone. The coat of the double transgenic is unpigmented and indistinguishable from that of a nontransgenic albino. Treatment of a double transgenic animal with 10 mM IPTG in the drinking water derepressed tyrosinase expression, resulting in a phenotype indistinguishable from that of the mouse transgenic for TyrlacO-25 alone (Fig. 6A).
The stringency of repression and derepression is illustrated by the pigmentation of the eye. Figure 6B shows dissected eyes, and Figure 6C shows sections through the eyes of a nontransgenic albino, a Tyr transgenic, a Tyr, LacIR double transgenic, and a Tyr, LacIR double transgenic mouse treated with IPTG. Repression of target transgene expression is accompanied by an absence of melanin in the retinal pigment epithelium (RPE) of the double-transgenic animal (Fig. 6C). The entire RPE is devoid of melanin, as can be seen by the completely unpigmented appearance of the eye in whole mount (Fig. 6B). Derepression by IPTG is accompanied by a restoration of pigmentation in the RPE to levels indistinguishable from the nonrepressed state (Fig. 6B,C). We obtained similar results with both lines of regulatable Tyrosinase mice (data not shown for TyrlacO-43 adults, but see Fig. 7). This indicates that regulation is neither insertion site specific nor simply fortuitous, but controlled by the lac repressor acting specifically on a target gene with lac operator sequences in its promoter. The albino mutation is a single basepair change in the coding sequence of tyrosinase, which causes a single amino acid change in the protein. Because the mutant allele is both transcribed and translated, we have not been able to assay promoter activity quantitatively at the molecular level. Nevertheless, we can infer from its effect on pigmentation that the tyrosinase promoter is in fact regulated by the lac operator-repressor system tightly, in a biologically relevant manner.
These results also show that IPTG can be introduced into the drinking water and circulate in the mouse at a level sufficient to derepress target gene expression. This level appears to be completely nontoxic. At the time of this writing, we have had TyrlacO, LacIR double-transgenic mice on 10 mM IPTG in their drinking water for up to 8 months with no deleterious effects.
Regulation is functional during embryogenesis, and reversible
We next wished to determine if lac elements could regulate
pigmentation during embryogenesis and if IPTG could act
transplacentally. Figure 7A and B shows the embryonic and newborn
pigmentation of the TyrlacO-43 line. At E9, tyrosinase
activity in the embryonic eye begins to deposit pigment in the
developing retinal pigment epithelium. At E12.5, the mouse RPE clearly
is pigmented (Drager 1985). As shown in Figure 7A,
TyrlacO-43 transgenic mice recapitulate this developmental
event. A distinct band of pigmentation surrounding the central lens is
seen in the Tyrosinase transgenic embryo that is not seen in
the nontransgenic albino (Fig. 7A). The lac repressor blocked
pigmentation during embryogenesis in the double-transgenic embryo, but
not when the mother was treated with IPTG during pregnancy (Fig. 7A).
These results clearly demonstrate not only that lac regulatory
sequences function well during embryogenesis, but also that IPTG can
cross the placenta to alter the phenotype of developing animals.
Finally, we tested the reversibility of the system by switching the Tyrosinase transgene on after it had been off, or off after it had been on, in the same animal. In Figure 7B, the phenotypes of eyes of newborn mice are compared to the phenotypes of embryonic eyes of mice of the same genotype. The mother of the IPTG-treated TyrlacO, LacIR double transgenic shown in Figure 7B was not started on IPTG in her drinking water until E12.5 of the pregnancy. As is shown in Figure 7A, this double-transgenic pup would have shown the albino phenotype at E12.5. The fully pigmented eyes seen at birth in this animal demonstrate that even after a period of silencing, derepression by IPTG was able to switch tyrosinase expression on. Figure 7C illustrates that reversibility is also possible in the opposite direction. The TyrlacO-25, LacIR double-transgenic pup shown on the left at P8 was taken off IPTG at P9. Removal of IPTG caused reversion of the phenotype to albino, as shown in the picture of the same animal as an adult on the right. As expected, the eyes remain pigmented due to the low turnover of cells and melanosomes in the RPE.
| |
Discussion |
|---|
|
Top
Abstract
Introduction
Results
Discussion
Materials and methods
References
|
|---|
We have described modifications made to the lac operator-repressor system that have allowed it to be used to regulate gene expression in the mouse. We have expressed the lac repressor in mice at levels that provide tight and reversible experimental control of gene expression. We have inserted lac operators into the mammalian target promoter so that it is directly controlled by the lac repressor. These changes greatly improve the reliability and predictability of the lac system over other regulatory systems used in mice. This system is a technological advance that should increase significantly the kinds of experimental questions that can be addressed in a mammalian model system.
A re-encoded, restructured version of the bacterial lacI gene can transit embryogenesis without being silenced and can function like a normal mammalian gene
Experiments using the lac operator-repressor system in
cultured mammalian cells (Brown et al. 1987; Hu and Davidson 1987; Figge et al. 1988; Liu et al. 1989) suggested that bacterial sequences could be processed into proteins and function in the compartmentalized eukaryotic cell as effectively as they had in the simpler prokaryotic cell. In contrast to results in cultured cells, bacterial sequences in
animals were expressed sporadically (Scrable and Stambrook 1997) or
with the assistance of toxic chemicals such as the demethylating agent
5-azacytosine (Wyborski et al. 1996).
To use the lac repressor in animals, bacterial and viral
elements had to be changed into their mammalian counterparts, or eliminated entirely. The most time-consuming challenge was to adapt the
lacI coding sequence so that it would go through sexual reproduction and embryogenesis without losing the ability to be transcribed. Embryogenesis is the great dividing line between cultured
cells and animals. Genome-wide changes occur during gametogenesis, at
cleavage, and at implantation, when epigenetic modifications such as
methylation completely alter the chromatin environment of individual
genes. The discovery that transgenes are imprinted with
gamete-of-origin-specific methylation patterns similar to those laid
down on endogenous genes (Swain et al. 1987) highlights the fact that
artificially introduced sequences also must contain the requisite
signals that will allow them to reemerge from embryogenesis functionally intact.
We think that a large part of the failure to produce mice expressing functional lac repressor prior to now can be attributed to a still incomplete understanding of what all of those signals are. We had to determine empirically those sequences that changed the splicing pattern of the RNA in a manner not predicted by sequence analysis. We identified a region of the synlacI sequence that for unknown reasons causes a block between transcription and translation. We identified structural elements that are critical for reliable transgene expression in the mouse, but whose mechanisms remain largely unknown. Thus, although it is possible to remodel a bacterial gene so that it will be expressed reliably in the context of a mammalian genome, our work with lacI demonstrates the difficulty of the task given the current state of knowledge of the molecular underpinnings of mammalian gene structure.
Insertion of lac operators into a mammalian promoter make it directly regulatable by the lac repressor
The target for the lac repressor is its DNA binding
sequence, the lac operator. The lac operator is a
short (~25 bp) sequence that easily can be incorporated into an
existing promoter without compromising the efficacy of the promoter
itself. In principle, this makes it possible to confer regulatability
on virtually any mammalian gene simply by inserting operator sequences
at appropriate sites in its promoter region. Our target promoter was a
genomic fragment of DNA shown to be the promoter of the murine
Tyrosinase gene (Kwon et al. 1987). It was modified at three
sites, inserting a little more than 100 bp over 2.2 kb stretch of DNA,
with lac operators positioned in a way that simulated the
overall structure of the regulatable promoter in the lac
operon. The pigmentation patterns of our two founder lines both closely
resembled the patterns of pigmentation observed in animals that had
been microinjected with the unmodified Tyrosinase transgene
(Yokoyama et al. 1990; Methot et al. 1995). These results indicate that
the modifications we made to the promoter when we inserted lac
operators did not have a significant, nonspecific impact on promoter
activity. Tyrosinase activity originating at the transgene locus, as
determined physiologically by the deposition of melanin, followed a
normal pattern during development in the absence of repressor, further
supporting our contention that operator insertion only minimally
affected unrepressed promoter activity, if at all.
The lac operator-repressor system can reliably and predictably control the phenotype of the mouse
Reversible regulation of pigmentation in the transgenic mouse by elements from the lac operon of E. coli is the first successful demonstration that bacterial sequences can be used to create trans-operons in mice that function analogously to their bacterial counterparts. The normal or "ground" state of the regulated promoter is repressed, or inactive. In the case of our TyrlacO, LacIR double-transgenic mice, this results in a total lack of pigment and a phenotype indistinguishable from that of a nontransgenic albino. IPTG derepresses the regulated gene: tyrosinase is switched on and the phenotype of the derepressed transgenic animal becomes indistinguishable from that of the nonrepressed transgenic animal.
This all-or-none switch is achieved at a nontoxic dose of IPTG. We have
seen full derepression in cells from the LacIR transgenic
animals with as little as 50 µM IPTG, which agrees with the levels
needed in E. coli for derepression of the endogenous lac operon (Cho et al. 1985). This is in contrast to a
previous result suggesting that the levels of IPTG needed for full
expression with the lac system are toxic (Figge et al. 1988).
The lac operator-repressor system has advantages over the currently available gene regulatory systems for the mouse
To overcome the problems of embryonic lethality and developmental
compensation that hinder the utility of traditional knockout strategies, a system was developed that allows target genes to be
excised from the genome in a tissue-specific and timing-controlled manner. The Cre/loxP recombination system (Gu et al. 1993,
1994) is based on the cre recombinase of bacteriophage P1 that
catalyzes site-specific recombination between loxP sites. An
advantage of cre-mediated excision is that the experimental
manipulation can be performed on an endogenous locus. However, excision
is a one-time event, which cannot be reversed. The lac system
is fully reversible. In addition, by incorporating lac
operators into a mammalian promoter by homologous recombination, it
should be possible to regulate the expression of endogenous loci, as
has already been done at the CDC2 locus in cultured cells (Itzhaki et
al. 1997). Thus, with the lac system, it should be possible to
create reversible mouse models of disease and elucidate gene function
in their natural context.
Reversible systems that use mammalian regulatory ele- ments to control gene expression in cells and animals (such as heavy metal or hormone responsive systems) have been used for some time. They suffer from the fact that, by definition, the inducing agent also will have effects on endogenous genes. Regulatory systems based on elements from the bacterial tet and lac operons, on the other hand, are exquisitely specific for the gene of interest in the context of the mammalian genome.
To use the tet system of gene regulation in mammalian cells,
Gossen and Bujard converted the tet repressor into the
tet transactivator. This fusion of the tet repressor
and the activating domain of the herpes simplex virus VP16
transcriptional activator, made it necessary to couple the tet
operator to a viral promoter permanently (Gossen and Bujard 1992).
Binding of repressor to operator serves to align the VP16 fusion
partner with its specific binding site in the viral promoter, and it is
the binding of VP16 to the viral promoter that activates transcription.
The tet transactivator (tTA) or reverse transactivator (rtTA)
no longer functions as a repressor, and instead functions analogously
to the typical mammalian transcription factor. Tissue-specific gene
expression can be achieved by having a specific mammalian promoter
drive expression of the tet transactivator (or rtTA).
The ground state of the minimal viral promoter is one that is less
active than the induced state, and binding of the transcription factor
increases the activity of a promoter that is already in the active
configuration. This has made the tet system particularly suited for those applications in the mouse in which the phenotype to be
controlled depends on the relative levels of an active transgene. For
example, hyperplasia (Guo et al. 1999; Xie et al. 1999), prion disease
(Tremblay et al. 1998), cardiomyopathy (Redfern et al. 1999), and
Huntington's disease (Yamamoto et al. 2000) all have been successfully
modeled in mice using the tet-responsive minimal CMV promoter
and the tTA or rtTA. The tet system also has been used to
study the molecular mechanisms of memory formation (Mayford et al.
1996; Mansuy et al. 1998a,b) and the role of FGF7 in lung development
and injury repair in the adult mouse (Tichelaar et al. 2000).
The drawbacks of using an activator-responsive promoter, however,
become apparent in situations where it is desirable to switch the
promoter between an inactive state and an active state that results in
physiological levels of tissue-specific expression. The endogenous
endothelin receptor-B locus was replaced with tet-regulated elements to determine the role of the receptor during embryogenesis (Shin et al. 1999). In addition to the enormous effort such an undertaking must have required, the use of the tet system to
control gene expression in this way was not entirely predictable and
induction was incomplete.
Tet relies on viral promoter elements, and in the mouse,
nonmammalian promoters very frequently lead to erratic expression of
downstream coding sequences (Furth et al. 1991). Low-level leakiness
(Shockett et al. 1995) and heterogeneous expression (Redfern et al.
1999) have been problems with use of the minimal CMV promoter, and the
VP16 activating domain has been found to be toxic to cells (Baron et
al. 1997). Our approach of directly targeting the lac
repressor to operator sequences incorporated into mammalian promoters
completely eliminates the necessity of using viral promoters or viral
DNA-binding proteins in conjunction with a prokaryotic-based regulatory
system. A target gene made up entirely of mammalian elements is more
likely both to maintain functionality after transiting embryogenesis
and to be expressed in the appropriate tissue at the appropriate level.
This lends the system a particularly strong element of predictability
that other prokaryotic-based systems cannot match.
In summary, by modifying both the target promoter and the gene encoding the lac repressor, we have been able to adapt a regulatory system used in bacteria to control the transcription of genes so that it can function analogously in the complex environment of the mouse. The LacIR mouse described in this report expresses the lac repressor ubiquitously, so it can be used in the future to regulate other promoters with the same degree of control we have demonstrated for our regulatable Tyrosinase transgene. Targeting endogenous promoters should move the system to the next level, where endogenous loci can be switched on and off repeatedly to create reversible models of human disease and normal development in the mouse.
| |
Materials and methods |
|---|
|
Top
Abstract
Introduction
Results
Discussion
Materials and methods
References
|
|---|
Construction of lac repressor genes
The lac repressor constructs W, S, 5'C1, 5'C2, 5'C4, 3'C1,
3'C2, 3'C3, and 3'C4 are driven by a 4.3-kb promoter region from the human -actin gene up to the AluI site at 
7 (Leavitt et al. 1984).
Followed by a short linker of either 44 bp (gatcagtcga cctgcagccc
aagcttgata tcgaattcgg atct) in W, 5'C1, 5'C4, 3'C1, 3'C2, 3'C3, and
3'C4, or 30 bp (gatcagtcga cctgcagccc aagcttcacc) in S and 5'C2. 5'C3
contains the 4.3-kb promoter up to the start of translation with no
polylinker. All of the above constructs contain the polyadenylation
signal sequence from the bovine growth hormone gene (Woychik et al.
1982) connected to the 3' end of the construct by a BamHI and
EcoRI linker region (taggatcccc gggctgcagg aattc).
Coding regions for the original wtlacI (W) and
synlacI (S) constructs are as previously described (Scrable
and Stambrook 1997). 5'C1 and 5'C2 were made by switching the linker
region and the first 36 bp of the coding region between wtlacI
and synlacI using the BsrFI site shared by both
constructs. 5'C1 contains the wtlacI linker and first 36 bp of
the coding region, and then the synlacI coding region. The
nuclear localization signal sequence (NLS) that had been attached to
the synlacI coding sequence was removed by PCR mutagenesis, so
that 5'C1 codes for a protein identical in amino-acid sequence to the
endogenous lac repressor. 5'C2 contains the synlacI
linker and first 36 bp of the coding region, and then the 3'
wtlacI coding region through the stop site. 5'C3 is identical to the endogenous -actin promoter up to the ATG start site, then contains the original synlacI coding region. This was created by PCR mutagenesis to remove the linker region present in S and replace
the 6 bp missing between the AluI site and +1. 5'C4 contains the linker region from W and the entire synlacI coding region, with no NLS. This was created by PCR mutagenesis of 5'C1 to return the
four bases in the beginning of the coding region that differ between W
and S back to the synlacI sequence (see Fig. 2B). 3'C1 contains the wtlacI sequence from the start of translation up to the EcoRV site at +800 (which W and S have in common) and
the synlacI sequence after the EcoRV site. The SV40
NLS is attached to the 3' end of the 3'C1 coding region with the linker region (agcagcctgaggcct), as described (Fieck et al. 1992), and was
created by PCR mutagenesis of the existing NLS linker region described in Scrable and Stambrook (1997). 3'C2 is identical to 5'C1 up to the
EcoRV site, then identical to W downstream. 3'C3 is identical to 5'C1 up to the PvuII site at +950 from the start of
translation, then identical to W downstream. 3'C4 is identical to W
upstream of the PvuII site, then identical to 3'C1 downstream.
Constructs M and R contain the human -actin promoter blunted at the
AscI site at 70 followed by the rabbit -globin intron 2 from the blunted NcoI site through the EcoRI site in
exon 3. The lacI coding region is inserted at the
EcoRI site. The M coding region is identical to W, and the R
coding region is identical to 3'C4. The rabbit -globin fragment
continues downstream of the lacI coding region to include the
rest of exon 3 and intron 3 with a polyadenylation signal sequence. The
polyA signal sequence from SV40 also is present at the 3' end. All of
the -globin sequences and the SV40 polyA signal sequence are as
described (Katsuki et al. 1988).
RT-PCR assay for splice site use in lac repressor transcripts
Total RNA was extracted from testis of W and S transgenic animals,
or Rat 2 fibroblasts transfected by calcium phosphate with the
indicated lacI construct using TRI Reagent (Molecular Research Products, Inc.). RNA was DNase treated with RQ1 DNase (Promega) and 1 µg reverse transcribed with AMV-RT. cDNA was subjected to 30 rounds
of amplification (95°C for 30 sec, 60°C for 30 sec, 72°C for 1 min.) using a primer in the -actin promoter (5'acagagcctcgcctttg3') and a primer in the lacI coding sequence
(5'tgcaggcagcttccaca3'). PCR products were run on a 4% polyacrylamide
gel in 1X TBE and transferred to Hybond -N+ membrane (Amersham) by
semi-dry electrophoresis in NAQ transfer solution (0.08 M Tris-HCl,
0.118 M Borate, 2.4 mM EDTA, pH8.3) at 220 mA for 1 h. The resultant
Southern blot was UV crosslinked and then prehybridized and hybridized
according to the methods described in Scrable and Stambrook (1997).
Rat2 transfection assay for lac repressor function
Rat 2 fibroblasts were transfected with 2.5 µg pSVOZ DNA and 2.5 µg of the indicated lac repressor construct DNA (or pBSSK carrier DNA) per 3 × 105 cells by standard calcium phosphate-mediated transfection. Growth media was DMEM, 0.1 units/mL penicillin/ 0.1 µg/mL streptomycin (Life Technologies), 5% FCS (Hyclone); (with 20 mM IPTG, if indicated). Two days after transfection, cells were fixed in 0.5% gluteraldehyde, incubated with X-gal containing solution (0.5mg X-gal in dimethylformamide, 44 mM HEPES, 3 mM potassium ferrocyanide, 3 mM potassium ferricyanide, 1.5 mM NaCl, 0.13 mM MgCl2 at pH > 7) at 37°C overnight, and the number of blue cells in each well recorded. Data shown in Figure 2C and D represent the average of three or four transfections for each construct (except S, which was only tested once in this Rat 2 assay, although similar results were obtained with over 10 transfections into other cell types).
Nucleic acid extraction and blotting
Northern blots were prepared as described in Scrable and Stambrook
(1997), except that RNA was transferred to Biodyne A nylon membrane
(Pall). DNA was extracted from tail biopsies using the simplified
method as described in Laird et al. 1991. Southern blots were prepared
and analyzed according to the methods given in Scrable and Stambrook (1997).
Detection of lac repressor protein by Western blot and immunohistochemistry
A panel of monoclonal antibodies to the lac repressor was created by injecting a LacI-TrpE fusion protein into mice. For Western blots, total protein was extracted into lysis solution (50 mM Tris at pH 7.5, 0.15 M NaCl, 1% Nonidet P40), containing protease inhibitors (0.25% sodium deoxycholate, 1 mM PMSF, 2 mM EGTA, 1 µM leupeptin, 0.2 µM aprotinin, 0.8 mM N-ethylmalemide, 2 µM pepstatin A). Protein concentration was determined by Lowry's assay, and 30 µg run on a 12% SDS-PAGE gel. The proteins were transferred to nitrocellulose membrane with semi-dry electrophoresis, and blocked in 5% dried milk in PBS overnight. The blot was incubated with biotinylated anti-lacI antibody 5F8 (25 µg/mL in 1% BSA/TBST) for 1 h at 37°C, labeled with peroxidase (ABC reagent, Vector) and visualized with chemiluminscence (SuperSignal, Pierce) on a ChemiImager (Alpha Innotech Corp.).
For immunohistochemistry, mice were given a lethal dose of Nembutol sodium, and perfused with 4% paraformaldehyde for 30 min. Tissues were placed in 20% sucrose overnight at 4°C, frozen, sectioned at 30 µm, and thaw-mounted onto Superfrost Plus (Fisher) slides. Sections were incubated with biotinylaed anti-lacI antibody 9A5 (3 µg/mL in 1% BSA/0.3% Triton-X100 in PBS) overnight at 4°C, labeled with peroxidase (ABC reagent, Vector), and visualized with DAB.
Preparation of primary mouse embryo cells
Pregnant females were euthanized on day E13.5 (where E0.5 was the day a vaginal plug was observed). The embryos were dissected out and a small section frozen for genotyping. Embryonic tissue was minced and placed in 2-mL dissociation solution [2 mg/mL Collagenase B, 2 U/mL RQ1 DNase in RPMI 1640 media (GIBCO)] at 37°C for 2 h, triturating the solution after 1 h. Cells were spun at 175g, washed one time with Hank's BSS, plated with growth media [DMEM, 0.1 units/mL penicillin/0.1 µg/mL streptomycin (GIBCO), 10% FCS (Hyclone)], and transfected by calcium phosphate.
Construction of the regulatable Tyrosinase transgene (TyrlacO)
The regulatable TyrlacO transgene is based on the
construct TYBS described in Yokoyama et al. (1990). The first
lac operator was created by site-directed mutagenesis (ExSite,
Stratagene). 25 bp of the endogenous promoter sequence (from 72 to
48) was changed to make a 29 bp operator centered at 59, identical
in sequence to the primary operator of the lac operon
(gtggaattgt gagcggataa caatttcac) (Lewis et al. 1996). Two additional
operators with the same sequence were inserted as part of a 47 bp
fragment (agatct gtggaattgtgagcggataacaatttcac ggatccagatct) into the
BsrGI site at 203 and the EcoRV site at 548 of
the promoter.
Production of transgenic mice
Production of the W and S lines is described in Scrable and
Stambrook (1997). The rest of the transgenic lines described were produced by microinjection into the outbred ICR line (Harlan) using
standard procedures. We made two transgenic founders for the 3'C4
transgene; both showed the testis-only expression pattern shown in
Figure 3. One founder line was established for the M construct. Three
founders were transgenic for R; two (lines 1 and 3) exhibited the
ubiquitous expression shown in Figure 3, one (line 13) had more limited
expression that ranged from low to moderate in various tissues. Eight
founders were transgenic for TyrlacO; an F1 generation was
produced from all eight, and two of those established pigmented
transgenic lines (lines 25 and 43). Of the animals indicated as
Tyrosinase transgenic, those in Figure 6A were homozygous for
TyrlacO, and all others were hemizygous for
TyrlacO. All lacI transgenic mice described were
hemizygous for lacI.
Analysis of eye pigmentation and IPTG treatment of mice
For adult eyes, mice were given a lethal dose of Nembutol sodium, perfused transcardially (1.25% paraformaldehyde, 1.5% gluteraldehyde, in 0.1 M phosphate at pH 7.4); eyes were dissected out and photographed. They then were embedded in parafin, sectioned at 10 µm, dewaxed in Xylene, hydrated in decreasing concentrations of ethanol, and reacted in cresyl violet (0.5% in 20% ethanol, pH to 2.5 with glacial-acetic acid) for 8 min, dehydrated, cleared, and mounted in DPX. For embryonic eyes, pregnant females were euthanized at E12.5, and the embryos removed. The head of each embryo was fixed in 2% paraformaldehyde in 0.1 M phosphate (pH 7.4), and the lower half was taken for genotyping.
Tyrosinase activity was derepressed by replacing the drinking water with a 10 mM solution of IPTG (changed every four days).
| |
Acknowledgments |
|---|
We are grateful to P. Stambrook for support during the early stages of the project; to M. Dalton, L. Abramova, and B. Bernier for assistance with mouse genotyping and excellent technical support; to M. Kimura and P. Overbeek for the plasmids pBstN and pTYBS, respectively; to J. Ellis for assistance with splice site analysis; to E. Talley, G. Owens, I. McCaffrey, B. Maier, S. Medrano, and M. Burton for helpful comments. This work was supported by a grant from the National Center for Research Resources, NIH grant RR11102. C.A.C. was a recipient of an NRSA predoctoral fellowship, NIH grant MH12406.
The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.
| |
Footnotes |
|---|
Received February 27, 2001; revised version accepted April 25, 2001.
1 Corresponding author.
E-MAIL hs2n{at}virginia.edu; FAX (804) 982-4380.
Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/gad.892001.
| |
References |
|---|
|
Top
Abstract
Introduction
Results
Discussion
Materials and methods
References
|
|---|
