ABSTRACT
Two new
[beta]
-lactoglobulin (BLG)/human serum albumin (HSA) hybrid gene vectors were
constructed and tested for expression in COS-7 cells and in transgenic mice. The HSA sequences were inserted between
the second and sixth BLG exons. Transient transfection experiments with these
vectors as well as a series of additional vectors with either the BLG 5
'
- or 3
'
- intragenic sequences revealed that sequences within BLG exon 1/intron
1/exon 2 abrogated BLG-directed HSA expression
in vitro
, regardless of the presence of HSA introns or the origin of the 3
'
polyadenylation signal. In contrast, the same BLG expression cassette enabled
the efficient expression of HSA cDNA or minigene in the mammary gland of
transgenic mice with subsequent secretion of the corresponding protein into the
milk of 56 and 82%, respectively of the mouse strains at levels up to 0.3
mg/ml. Previous attempts to express HSA cDNA inserted into exon 1 of the BLG gene had failed [Shani,M., Barash,I., Nathan,M., Ricca,G., Searfoss,G.H., Dekel,I., Faerman,A., Givol,D. and Hurwitz,D.R. (1992)
Transgenic Res
. 1, 195-208]. The new BLG expression cassette conferred more stringent tissue
specific expression than previously described BLG/HSA constructs [Barash,I., Faerman,A., Ratovitsky,T., Puzis,R., Nathan,M. Hurwitz,D.R. and Shani,M. (1994)
Transgenic Res
. 3, 141-151]. However, it was not able to insulate the transgenes from the
surrounding host DNA sequences and did not result in copy number dependent
expression in transgenics. Together, the
in vitro
and
in vivo
results suggest both positive and negative regulatory elements within the BLG
intragenic sequences evaluated. The new BLG construct represents an extremely
valuable vector for the efficient expression of cDNAs in the mammary gland of
transgenic animals.
The promoters of the major milk protein genes can be used to target the
expression of any gene to the mammary gland of transgenic animals resulting in
the production of desired proteins in the milk. In ruminants, the major whey
protein is [beta]-lactoglobulin (BLG). The ovine BLG gene has been cloned, sequenced (
1
,
2
) and expressed at high levels in the mammary gland of transgenic mice (
3
-
5
). Although mice and sheep carrying minigenes composed of genomic sequences of
human [alpha]1-antitrypsin (
6
,
7
) or human serum albumin (HSA) (
5
,
8
) behind the BLG promoter secreted high levels of the corresponding proteins,
almost no expression was detected in the mammary gland of transgenic mice
carrying the cDNA of these genes (
5
,
9
).
In the generation of transgenics, the use of a vector comprised of a cDNA sequence encoding the desired protein has certain advantages over the
use of a vector comprised of genomic sequences. The cDNAs are usually more
available than the genomic sequences and their significant shorter length makes
vector construction easier. However, their poor performance in directing
protein expression in transgenic mice, attributed to increased sensitivity to
chromosomal position effect (
10
) and difficulties in maintaining active chromatin configuration (
11
), has been problematic. The present study was undertaken to design a BLG vector
that enables efficient expression of the HSA cDNA in the mammary gland.
Detailed analysis of the native BLG gene in sheep and the BLG transgene in
transgenic mice identified three strong DNase I hypersensitive sites in the
proximal promoter region, between -1800 and +100 bp from the transcription initiation site (
12
). No hypersensitive sites were detected within the BLG structural or 3' end region of the gene (
12
). DNase I hypersensitive structures have been strongly correlated with the
presence of binding sites of regulatory elements such as silencers and
enhancers (
13
), and efforts to identify mammary specific transcriptional factors that interact with these sequences is underway. These results, however, do not
exclude the possible existence of autonomous regulatory elements within the BLG
structural gene or elements capable of activating gene expression by interacting with sequences at the proximal promoter or at the 3' flanking region. This idea is based on the fact that the BLG promoter
alone is unable to drive the expression of heterologous genes in a position-independent and copy number dependent manner unlike the expression of the
entire gene (
6
,
12
).
We have previously demonstrated a correlation between HSA expression in COS-7 cells transiently transfected with a series of BLG/HSA constructs
comprised of different complements of the native HSA introns and the efficiency
of HSA secretion into the milk of transgenic mice (
8
). In the absence of a reliable cell line capable of mimicking the entire
repertoire of gene regulation mechanisms in the mammary gland, this
in vitro
tissue culture system provided a useful tool for predicting the potential of
individual BLG/HSA constructs for expression
in vivo
.
Using transiently transfected COS-7 cells and transgenic mice we analyzed the expression of an HSA cDNA
introduced into a new BLG expression cassette. This cassette consisted of the
BLG promoter as well as BLG intragenic sequences from exon 1 through mid exon 6
through the end of the gene (exon 7) and 3'-flanking sequences. Thus potential regulatory elements were
preserved not only in the promoter region but also within the BLG exon 1/intron
1/exon 2 and exon 6/intron 6/exon 7 and the 3' end of the gene. We found that the HSA cDNA was expressed in a high
percentage (56%) of transgenic strains at levels similar to that obtained in
transgenics generated from a vector with an HSA minigene in the identical BLG
expression cassette. However, it appears that the presence of HSA introns did
improve upon the frequency of expression with 82% of the transgenic strains
producing HSA in their milk. In addition, the presence of BLG intragenic
sequences resulted in a much greater tissue specificity of expression than
previous BLG/HSA vectors.
In vitro
expression in COS-7 cells from homologous vectors (including an SV40 enhancer) were
completely abrogated. A series of additional
in vitro
vectors demonstrated that sequences within BLG exon 1/intron 1/exon 2 exerted
the dominant inhibitory activity.
The stimulation of expression
in vivo
, the greater tissue specificity of expression
in vivo
and the inhibition of expression
in vitro
, together suggest the presence of both positive and negative regulatory
elements within the BLG intragenic sequences evaluated.
Vector p835 is a modified p585 BLG vector (
5
) in which the BLG sequences between the first part of BLG exon 2 and the middle
of BLG exon 6 (including TAG termination codon) were deleted and replaced with
a short fragment of HSA cDNA sequences starting from 10 bp upstream of the HSA
ATG translational initiation codon through the native
Bst
EII site 12 bp downstream of the ATG. The natural BLG ATG translational
initiation codon, as well as a second potential initiation codon in BLG exon 1,
were converted into non-initiating ATT and ATC sequences, respectively, during this construction
process.
Detailed description of the construction of vectors p658enh, p659enh, p691enh,
p660enh, p652enh, p812enh and p698enh (illustrated in Fig.
2
A) were presented in our previous reports (
5
,
8
). Briefly, they all share 3 kb of the BLG 5'-flanking regulatory region in which the SV40 enhancer sequences
were cloned 900 bp upstream of the transcription initiation site. The HSA
minigenes or cDNA are inserted into the first untranslated exon of the BLG gene
and all the constructs were terminated with the SV40 polyadenylation sequences.
BLG/HSA vectors displayed in Figure
2
B were generated from constructs p838 and p839 by fragment switching with a
previously described vector [designated p652(1-6)*enh, 8], resulting in the introduction of the SV40 enhancer sequences ~900 bp upstream of the BLG transcription initiation site. These
vectors were designated p838enh and p839enh, respectively. Fragment switching
between vectors p838enh, p839enh and vectors previously described (
5
,
8
,
15
) generated the remaining vectors. Vectors p845enh and p846enh carry the BLG 5' sequences of p838enh and p839enh (including BLG exon 1/intron 1/exon 2),
HSA minigene including introns 1,2+12-14 or 1,2+7-14, respectively, and an SV40 poly(A) site which replaced the BLG
3'-sequences downstream of the
Nco
I site [BLG exon 6/intron 6/exon 7, (
8
)]. A third set of vectors was constructed that maintained the BLG 3' sequences (exon 6/ intron 6/ exon 7) but replaced the BLG 5' sequences (exon 1/intron 1/exon 2) with the BLG promoter/exon 1
sequences used previously in all earlier vectors (
5
,
15
) shown in panel A. Vector p847enh carries an HSA minigene with introns 1,2+12-14 and vector p848enh carries an HSA minigene with introns 1-6+12-14. Vector p849enh, carries the Adenovirus Major Late
Promoter and SV40 enhancer (
15
). The promoter itself is 402 bp and is followed by the 173 bp of the Adenovirus
Major Late leader sequences (L1, L2 and most of L3). The SV40 enhancer (179 bp)
was inserted at the very 5'-end of the major late promoter at an
Eco
RV site. Vector p849enh maintains BLG exon 1/intron 1/exon 2, the HSA minigene
with introns 1,2+12-14 and BLG exon 6/intron 6/exon 7. Vector p601enh contains the SV40
enhancer in front of the Adenovirus Major Late Promoter and HSA minigene
containing the first HSA intron in its native position. This vector is
terminated by the SV40 poly(A) site.
Mice used in this study were of the FVB/N strain. Plasmids containing fusion
genes comprised of the sheep BLG promoter and intragenic sequences and the HSA
cDNA or minigene were restricted with
Sal
I, and the appropriate fragments were prepared and microinjected as described (
5
). Transgenic animals were identified by Southern blot analysis of genomic DNA
prepared from tail biopsies using the respective fragments as probes (
5
).
Total RNA from tissues of lactating (10 days) transgenic mice was isolated by
the LiCl/urea procedure (
14
) essentially as described earlier (
5
). RNA (10 [mu]g) was fractionated on MOPS/formaldehyde agarose gels and blotted onto
Nytran membranes (DuPont, Boston, MA). Hybridization was performed at 42oC according to the manufacturer's protocol with an HSA cDNA probe,
radiolabeled with [
32
P]dATP by random priming (Boehringer Mannheim, Indianapolis, IN). The presence
of equal amounts of RNA from the various tissues was confirmed by the intensity
of the ribosomal RNA bands.
Milk was collected from nursing transgenic females 10-12 days after parturition. HSA was analyzed by Western analysis of 1:5 diluted
milk samples using anti-HSA monoclonal antibody (Ceder Lane Laboratories, Hornby, Ontario, Canada)
iodinated to a specific activity of 8.0 [mu]Ci/[mu]g (
5
). The levels of HSA secreted into the milk were determined by dot-blot assay compared to HSA standards (
5
).
Genomic DNA (10 [mu]g) extracted from tail biopsies of transgenic mice was digested with
Bam
HI, subjected to electrophoresis on 0.8% agarose gels and hybridized first with
a 1350 bp DNA probe derived from the 5' end of the BLG gene and then re-probed with a [beta]-actin cDNA. The relative densitometric values of the
signals were determined.
Mammalian COS-7 cells were transiently transfected with the various BLG/HSA constructs
using the calcium phosphate technique (5 prime-3 prime Inc., Boulder, CO) as described earlier (
5
). Following metabolic labeling, HSA secreted to the medium was detected by
immunoprecipitation using rabbit anti-HSA antibodies (DAKO immunoglobulins, Denmark) as described (
5
,
8
).
A generic BLG vector (p835) possessing BLG intragenic sequences was constructed
and its ability to direct high level expression of cDNA products to the mammary
gland was determined. The vector composed of 3 kb of the 5'-BLG promoter sequences in conjunction with a 5'-portion (the entire BLG exon 1 and intron 1 and part
of BLG exon 2) and a 3'-portion (part of BLG exon 6 downstream of the TAG termination
codon, all of BLG intron 6 and exon 7) of the BLG intragenic sequences, as well
as BLG 3'-flanking sequences is shown in Figure
1
A and was constructed as detailed in Materials and Methods. The natural BLG ATG translation initiation
site as well as a second potential ATG initiation site within exon 1 was
converted to ATT and ATC non-initiating sequences, respectively. No other potential initiating codons
are present in the 5'-portion of the BLG sequences utilized.
We had previously demonstrated that the level of
in vitro
expression of HSA in COS tissue culture cells supported by earlier BLG/HSA
vectors, lacking the BLG intragenic sequences of the new vectors, is modulated
by the specific complement of HSA introns included (
8
). In the present study, these earlier
in vitro
vectors (Fig.
2
A) served as references for comparison with the newly constructed BLG/HSA
vectors shown in Figure
2
B. The high level of expression of HSA supported by vector p652enh, with HSA
introns 1-6, was set at 100% to which all other relative expression levels were
compared. Very low levels (1%) of HSA were secreted from cells transfected with
vector p658enh, carrying the HSA cDNA inserted into the untranslated BLG exon
1. Significantly higher levels of HSA were secreted from cells transfected with
all of the earlier BLG/HSA vectors, comprised of the various BLG/HSA minigenes,
through a wide range of levels depending upon the specific combination of HSA
introns included (Fig.
2
A). Thus addition of the first intron of the HSA gene (p659enh) resulted in low
levels of expression (6%) while the inclusion of HSA minigenes with introns 2
(p691enh) and intron 1 and 2 (p660enh) resulted in moderately high (75 and 50%) levels of expression relative to the p652enh standard, respectively. Vector
p812enh, containing HSA introns 1+2+12-14, and vector p698enh, containing HSA introns 1+2+7-14 in their native positions, were expressed at the very high
levels of 110 and 156% of the control p652enh vector, respectively. These
results indicate that intronic sequences from both the 5' and the 3' ends of the HSA gene contribute to high level of expression.
When COS cells were transfected with our new BLG/HSA vector p838enh, comprised
of the HSA cDNA positioned between the BLG 5'- and 3' intragenic sequences, no expression of HSA could be
detected (Fig.
2
). This lack of detectable expression should be compared to the low level (1%)
expression with our earlier HSA cDNA vector p658enh. Thus, the BLG intragenic
sequences, including BLG introns 1 and 6 could not improve upon the expression
from a vector with the HSA cDNA. Surprisingly, no HSA could be detected in the
medium of cultures transfected with vector p839enh, containing HSA introns
1+2+12-14 (Fig.
2
). This was unexpected in view of the high level of expression with analogous
vector p812enh, containing the same HSA gene sequences which was attributed to
the presence of HSA intronic sequences (
8
). The present results demonstrate that sequences within the BLG 5'-intragenic region (exon 1/intron 1/exon 2) and/or the BLG 3' intragenic regions (exon 6/intron 6/exon 7) and/or the
native sequences 3' of the BLG gene abrogated HSA expression from identical HSA gene
sequences. It did not appear likely that the BLG 3' intragenic or 3' flanking sequences in themselves had this abrogating effect since
earlier vectors possessing these sequences supported similar levels of HSA
expression as did vectors comprised of the SV40 poly(A) signal at their 3' end (
5
).
In an attempt to identify the element or combination of BLG gene elements
responsible for the suppression of HSA expression from p838enh and p839enh,
additional new transient BLG/HSA vectors were constructed and tested by
transfection in COS cells. In two new vectors (p845enh and p846enh), the BLG 5' sequences (promoter/exon 1/intron 1/exon 2) were maintained and the BLG
3' sequences including exon 6/intron 6/exon 7 were replaced with SV40
poly(A) signal sequences. In both new vectors, the sequences downstream of the
Nco
I site in HSA exon 7 in vector p839enh were replaced with HSA sequences which
included either HSA introns 12-14 and an SV40 poly(A) site (p845enh) or HSA introns 7-14 and an SV40 poly(A) site (p846). Both of these vectors
supported only marginal levels (2% of p652enh) of HSA expression (Fig.
2
B), even though vectors p812enh and p698enh, which are identical to p845enh and
p846enh except for the absence of the BLG 5'-intragenic sequences, supported very high levels (110 and 156%,
respectively, of p652enh level) of HSA.
A second set of vectors, p847enh and p848enh, was constructed that maintained
the BLG 3' sequences including exon 6/intron 6/exon 7 but replaced the BLG 5'-intragenic sequences (exon 1/intron 1/exon 2) with the BLG
promoter/exon 1 sequences used in all earlier vectors (e.g. Fig.
2
A). Thus vector p847enh carried an HSA minigene with introns 1+2+12-14 and vector p848enh carried an HSA minigene with introns 1-6+12-14. Both of these vectors were able to support moderately
high (32 and 59%, respectively, of p652enh level) expression of HSA in
transient assays, though at lower levels than the equivalent vectors carrying
an SV40 poly(A). The results with these four new vectors indicate that although
the BLG 3'-intragenic sequences may play a minor role in reducing expression,
BLG 5'-intragenic sequences within BLG exon 1/intron 1/exon 2 exert a
major inhibitory effect on HSA expression in the context of BLG/HSA vectors.
To determine if the BLG intragenic sequences were also inhibitory to
in vitro
expression of HSA minigenes utilizing a promoter other than the BLG
promoter/SV40 enhancer combination, we constructed vectors p601enh and p849enh
(Fig.
2
B). Vector p601enh is comprised of the SV40 enhancer in conjunction with the
Adenovirus Major Late promoter with its leader sequences (Ad MLP) (
15
), the HSA minigene with intron 1, the SV40 poly(A) signal and no BLG sequences.
Vector p601enh supported much higher levels of HSA expression than did
analogous BLG/HSA vector p659enh (68% compared with 6% of p652 level)
demonstrating how very effectively the Ad MLP is able to drive
in vitro
expression of HSA. Vector p849enh is also comprised of an HSA minigene (introns
1+2+12-14) driven by the Ad MLP and lacks the BLG promoter, but it maintains the
BLG 5'- and 3'-intragenic sequences. The level of HSA expression from
p849enh was very low and only 3% of p652enh level.
We conclude from these experiments that sequences within BLG exon 1/intron
1/exon 2 was the cause of the suppression of HSA expression.
To test the role of DNA sequences in the BLG 5'- and 3'-intragenic regions on HSA expression
in vivo
, vectors p838, carrying HSA cDNA and vector p839, carrying HSA introns 1+2+12-14, were constructed from the new generic vector p835 (Fig.
1
). These former two vectors are identical to the
in vitro
constructs p839enh and p839enh, respectively, except for the absence of the
SV40 enhancer sequences. The levels of HSA secreted into the milk and the
expression of HSA RNA in different tissues of a number of transgenic mouse
strains carrying these constructs are shown in Table
1
.
Five of the nine transgenic strains carrying the p838 BLG/HSA, lacking HSA introns, secreted detectable levels of HSA in the milk. In six of
these strains HSA mRNA was detected in the mammary gland. The four transgenic
strains which expressed the highest levels of HSA in their milk also presented
the highest levels of HSA mRNA in their mammary glands. Strain #145 which expressed only an
extremely low level of HSA in the milk did not present detectable HSA mRNA in
the mammary gland while a couple of strains (#151, #154) with no detectable HSA
in the milk did show low levels of HSA mRNA in their mammary gland. This
apparent discrepancy may be due the different sensitivities of the assays. The
levels of HSA detected in the milk of transgenics were moderately high but did
not exceed 0.3 mg/ml. This level was detected in the milk of three independent
transgenic strains (#130, #147 and #155). Significantly, lower levels of HSA
were detected in the other two strains.
The milk of nine of the eleven strains, carrying construct p839, contained detectable levels of HSA. Surprisingly, despite the presence of HSA
intron sequences, the level of HSA did not exceed that seen in mice carrying the cDNA based vector p838 and also reached a
maximum of 0.3 mg/ml. HSA mRNA was detected in the mammary gland of six of
eleven transgenic strains carrying p839. Comparable variations in HSA secreted
into milk and in HSA mRNA detected in the mammary gland was observed among transgenics carrying the intronless construct p838 as well as the
HSA minigene p839, with the five strains expressing the highest levels of HSA
in milk showing the highest levels of HSA mRNA in the mammary gland. As seen
with p838, at the lower levels of expression the levels seen in milk did not
correlate with the levels of mRNA seen in mammary tissue.
Table 1
The HSA cDNA probe utilized in this study hybridizes to both the endogenous
mouse and exogenous human albumin mRNAs. However, HSA mRNA could be easily
distinguished from the endogenous mouse albumin mRNA due to its slower
migration on 0.8% agarose gels (
5
,
16
). To test the stringency of tissue specificity of expression of the two new
constructs we analyzed HSA mRNA in various tissues of transgenic mice. Skeletal
muscle, kidney, salivary gland and brain were chosen, since they appear to be
favorable tissues for ectopic expression of the previously reported BLG/HSA
vectors in transgenic mice (
5
,
16
). As shown in Table
1
and Figure
3
, the only tissue analyzed expressing HSA RNA other than the mammary gland was
skeletal muscle. While five of the six strains presenting HSA mRNA in skeletal
muscle also presented RNA in mammary tissue, the sixth strain did not express
HSA in its milk. No correlation could be drawn between the level of expression
in the muscle and in the mammary gland. However, although in three strains
(#145, #137 and #149) the level of HSA RNA in the muscle was higher than in the
mammary gland, in the majority of the strains expression was higher in the
mammary gland. Interestingly, this ectopic expression in skeletal muscle was
common to mice carrying either the cDNA or the HSA minigene based constructs,
suggesting that the more stringent tissue specific expression relative to
earlier gene constructs tested (
5
,
16
) was due to the presence of the additional BLG sequences in the new vectors. As
expected, endogenous mouse albumin gene transcripts were detected in the kidney
of ~17% of the tested mice (data not shown,
5
).
Many transgenes are strongly influenced by the site of integration in the host
genome. Expression can also be influenced by the number and arrangement of the
integrated copies of the transgene. To determine whether the additional BLG
sequences employed in this study conferred position independent expression of the BLG/HSA constructs,
we performed a statistical correlation analysis between the level of HSA
secreted into milk and the relative copy number of the BLG/HSA transgene in independent strains carrying
constructs p838 and p839. The copy number was calculated by densitometry of the
BLG/HSA Southern signals relative to the single copy [beta]-actin internal control gene signal in each sample.
r
values, calculated for the correlation between HSA secreted into the milk and
the relative copy number of the integrated BLG/HSA sequences, were insignificant [-0.4 for p838 (
n
= 8) and -0.33 for p839 (
n
= 8)]. A similar insignificant
r
value (
r
= -0.3) was also calculated for the correlation between HSA secretion and
copy number for all transgenic strains tested (
n
= 16).
Most identified regulatory elements are located in genomic sequences 5' of the transcription start site (e.g.
17
-
19
). For example, the positive MGF (
18
) and negative YY1 (
19
) transcriptional factors that bind to the promoter of the [beta]-casein gene and the milk protein binding factor (MPBF) that binds
the proximal promoter sequences of the sheep [beta]-lactoglobulin gene (
20
).
There are, however, increasing number of reports demonstrating the presence of
intragenic regulatory regions: the glucocorticoid responsive elements within
the first intron of the human growth hormone gene (
21
); the thyroid responsive element in the third intron of the rat growth hormone
gene (
22
); intragenic regulatory elements within genes coding for the growth factors IGF-I (
23
), PDGF-B (
24
), the prosthetic binding protein (
25
) and calbinding (
26
); within the oncogenes c-myc (
27
), c-myb (
28
) and c-fos (
29
); the elastin (
30
); the collagen family (
31
-
35
) and many other genes. Most of these regulatory elements were identified either
directly, by their affect on gene expression, or indirectly, by localizing
DNase I hypersensitive sites (reviewed by
13
). Perhaps due to its close proximity to the promoter region, the first intron
of several genes contains the majority of putative and confirmed intragenic regulatory sequences. Another region of interest,
especially with respect to the regulation of milk proteins genes, is the 3'-UTR. It was demonstrated that sequences within the 3'-UTR of the Whey acidic protein gene contributed to its
high level, copy number dependent expression in transgenic mice (
36
).
Previously we had shown that HSA gene intron sequences play a key role in regulating HSA expression driven by the BLG promoter both
in vitro
and
in vivo
(
5
,
8
). Accordingly, an HSA cDNA based construct was poorly expressed
in vitro
and did not secrete HSA into the milk of transgenics (
5
,
8
), whereas specific combinations of HSA introns made major positive
contributions to the levels of HSA expression both
in vitro
and
in vivo
.
In an attempt to design a mammary specific vector capable of driving high level
and position independent cDNA expression in transgenic mice, we evaluated the
effect of sequences downstream of the BLG promoter (within exon 1/intron 1/exon
2 and exon 6/intron 6/exon 7 and BLG 3'-flanking sequences) on HSA expression. An HSA minigene was also
inserted into the same vector to determine the contribution of HSA intron
sequences on the expression of the human protein.
No HSA expression could be detected in COS-7 cells transiently transfected with the new p838enh BLG/HSA vector
comprised of the HSA cDNA. This was not overly surprising since our previous
BLG/HSA cDNA vector, p658enh, supported only marginal expression. However, it
does support our conclusion (
5
,
8
) that specific elements within HSA introns have a positive effect on
in vivo
expression since the BLG introns, contained with the BLG intragenic sequences, were not able to stimulate
expression from vector p838enh. What was completely unexpected was that even
the presence of HSA introns 1+2+12-14 with the new p839enh vector, was not sufficient to stimulate
expression, or even to support detectable expression, from this vector. This
suggested that elements within BLG 5'- and/or 3'-intragenic sequences were inhibiting
in vitro
expression. Only marginal
in vitro
expression was obtained with vectors p845enh and p846enh while moderately high
levels of expression were obtained with vectors p847enh and p848enh. This
indicates that the BLG 5'-intragenic sequences predominantly exert the inhibitory effect and
that the abrogation of HSA expression was independent of the presence of HSA
intron sequences or the origin of the poly(A) site. The fact that the BLG
intragenic sequences also suppressed HSA expression from HSA minigene driven by
the Adenovirus Major Late promoter (vector p849) demonstrates that the
inhibitory effect exerted by the BLG 5'-intragenic sequences within exon 1/intron 1/exon 2 was not just
specific for the SV40 enhancer/BLG promoter but was inhibitory to other
promoters.
In striking contrast to the repression of HSA expression
in vitro
, the new constructs, p838 and p839, promoted HSA expression in the mammary
gland of transgenic mice whether the vector was comprised of HSA cDNA or HSA
minigene. Moreover, the new BLG vector provided a permissive environment for
the expression and secretion of moderately high levels of HSA into the milk of
up to 0.3 mg/ml even with an HSA cDNA. Previous studies demonstrated that no
expression could be detected when the same HSA cDNA was driven by the BLG
promoter alone (
5
,
8
). In addition, HSA was expressed in the milk of a high percentage (56%; 5 of 9)
of transgenic strains generated from the new HSA cDNA vector, p838. Introns are
known to increase levels of gene expression in transgenic mice in general (
11
) and in the milk of transgenics specifically (
5
-
7
). While the new p838 vector is comprised of the HSA cDNA and lacks any HSA
introns, it does none the less possess BLG introns 1 and 6. It therefore
appears that the high percentage of transgenic strains generated from p838
which express could be the result of the positive regulation effect of BLG
intragenic sequences (including introns) on expression
in vivo
.
Expression levels in p839 transgenic milk did not exceed those found in p838 transgenic milk (0.3 mg/ml). This suggests that the major
stimulatory effect is due to the presence of the BLG intragenic sequences. The
in vitro
and
in vivo
results taken together suggest that sequences within exon 1/intron 1/exon 2 of
the BLG gene exhibit both positive and negative cell-specific regulation of the transgene. A similar phenomenon of cell-specific positive and negative regulatory elements was also
demonstrated for the PDGF-B gene (
24
) and the [alpha](1) collagen gene (
33
). The moderately high level of HSA expression may be the result of the
in vivo
interactions between the putative positive and negative regulatory elements
contained with the BLG intragenic sequences incorporated into these vectors.
Our results demonstrating expression from an HSA cDNA driven by vector p838
differ from those reported for the AAT-C vector comprised of the first BLG intron and the [alpha]-anti trypsin cDNA (
37
). In that study, only one of seven transgenic strains expressed any level of [alpha]-anti trypsin mRNA. In addition, that mouse strain did not secrete
the protein into its milk. However, while both our p838 and their AAT-C vectors were comprised of the BLG promoter, BLG exon 1/intron 1/exon 2,
followed by their respective cDNA, the two vectors differed considerably in
other regards. While in vector p838, translation initiation codons were altered, in vector AAT-C, the BLG ATG translational initiation site was unaltered and an in-frame TGA termination codon was introduced into BLG exon 2.
Therefore, any resultant mRNA was bicistronic and a very inefficient
translational re-initiation would have been required in order to translate the anti-trypsin open reading frame. In addition, in the AAT-C vector, the anti-trypsin cDNA was followed by the entire BLG cDNA
downstream from within BLG exon 5. In vector p838, the HSA cDNA was followed by
BLG genomic sequences downstream from within BLG exon 6 and included BLG intron
6. These vector differences may well have contributed to
in vivo
expression differences.
The complete BLG gene is expressed exclusively in the mammary gland of
transgenic mice. However, the BLG 5' sequences by themselves do not possess sufficient elements for this
strict tissue-specificity. Thus, transgenic strains carrying earlier BLG/HSA constructs
lacking the BLG intragenic sequences of the new vectors, express HSA RNA in a
variety of tissues, irrespective of the level of HSA expression in the mammary
gland (
16
). It has been reported that conserved sequences within the first and last introns of the bovine [alpha]s1 casein gene are needed for the tissue- and stage-specific expression of a [alpha]s1-CAT construct in the mammary gland of transgenic
mice (
38
). The results with the new BLG vectors clearly demonstrate that sequences
within the BLG intragenic 5' and/or 3' sequences confer more stringent tissue specific expression than
the BLG promoter alone. Ectopic expression was found only in skeletal muscle of mice carrying the HSA cDNA
vector or the HSA minigene vector. Taken together, our results demonstrate that
in addition to enabling HSA cDNA expression, BLG intragenic sequences
contribute to a better tissue specificity. Both of these improvements are
especially important in the application of transgenesis to the production of
valuable pharmaceutical proteins in the milk of transgenic farm animals.
Constructs based on cDNAs are easier to prepare than genomic constructs and
eliminating ectopic expression could prevent unexpected effects of biological
active gene products. The new BLG vector, however, did not confer position
independent and copy number dependent expression of the HSA transgenes. This is
consistent with the reports on the majority of regulatory elements of milk
protein genes (
6
,
39
,
40
), with the exception of rat Whey acidic protein gene (
36
,
41
). The fact that the native BLG gene is expressed in all our transgenic strains at levels >1.0 mg/ml (
5
) indicates that BLG sequences other than those used previously and in this study are able to insulate hybrid transgenes from host genome integration site effects. The improved BLG expression cassette provides the basis for identifying those regulatory sequences. In addition, the generic BLG expression vector similar to vector p835 may allow high level expression of cDNAs other
than HSA in the milk of transgenics in a highly tissue specific manner.
Present addresses:
+
Life Technologies, PO Box 6009, Gaithesburg, MD 20877, USA and
[sect]
ALG Company, 734 Forest Street, Marlboro, MA 01752, USA
Construct
Strain
HSA in milk
HSA RNA accumulation
(mg/ml)
M.G.
Sk. Musc.
Kidney
Salivary G.
Brain
838
130
0.3
++++
+++
0
0
0
838
155
0.3
+++++
+
0
0
0
838
147
0.3
++++
0
0
0
0
838
146
0.005
++
0
0
0
0
838
145
0.0005
0
++
0
0
0
838
151
U.D.
+
+
0
0
0
838
154
U.D.
+
0
0
0
0
838
127
U.D
0
0
0
0
0
838
150
U.D.
0
0
0
0
0
839
141
0.3
++++
0
0
0
0
839
133
0.3
+++
0
0
0
0
839
140
0.15
++
0
0
0
0
839
134
0.15
++
0
0
0
0
839
149
0.15
++
+++
0
0
0
839
138
0.01
0
0
0
0
0
839
144
0.015
0
0
0
0
0
839
139
0.01
0
0
0
0
0
839
137
0.005
+
+++
0
0
0
839
132
U.D.
0
0
0
0
0
839
135
U.D.
0
0
0
0
0
REFERENCES
Return