Beta-lactoglobulin (BLG) is found in the milk of a wide variety of species including dogs and dolphins (1, 2) and is the major whey protein in the milk of ruminants. It is not, however, present in the milk of mice and humans and is a cause of allergy in human infants fed on cow's formula milk. The function of BLG is not clear although it is similar in structure to retinol-binding protein (3) and lipocalycins, suggesting that BLG may have a role in the transport of fatty acids and vitamin A. The ovine BLG gene promoter has been extensively characterised in both transgenic mice and mammary epithelial cells (MEC) in culture. This review will focus on the transcriptional regulation of BLG and the role of specific sequences and transcription factors in controlling BLG expression. The benefits and drawbacks of using the BLG promoter to direct expression of transgenes to the mammary gland, including ectopic and variegated expression, will also be discussed.
Hormonal requirements for induction of BLG expression
Mammary gland explants and primary cultures have been used to determine the minimal hormone requirements for ovine BLG expression. Quantification of BLG mRNA in explants from ewes in the first half of pregnancy showed that insulin, cortisol and prolactin were required to stimulate gene expression, that oestrogen and thyroid hormone had no additional effects, and that progesterone did not inhibit induction of the gene (4). In primary cultures of mammary acini from lactating sheep, BLG synthesis is maintained at high levels for 8 hours then declines by approximately 75% in the following 16 hours. However, this decline cannot be reversed by the addition of lactogenic hormones (5). In cultures of MEC, a cocktail of hormones including insulin, glucocorticoid and prolactin is routinely used to induce expression of BLG constructs.
BLG expression in transgenic mouse mammary glands
In sheep, BLG expression is tightly regulated and restricted to the mammary gland. Expression is induced mid-way through pregnancy, peaks in lactation and declines at the onset of involution (6). When BLG transgenes are introduced into the mouse, the temporal pattern of expression of BLG is more similar to that of the endogenous b-casein gene, being induced around day 10 of pregnancy, than the whey acidic protein (WAP) gene, which is induced later around gestation day 15. However, the regulation of expression is less tight, with low levels of RNA being detected in virgin animals (7).
The sheep BLG promoter has been extensively analysed, principally in Edinburgh. Originally, BLG transgenes were comprised of genomic constructs with 4.2kb of 5' promoter sequences and 1.7kb of 3' sequences flanking a 4.6kb gene containing 7 exons and 6 introns (8). The promoter of BLG has been extensively analysed by sequential resection of these 5' sequences and a surprisingly short region of the proximal promoter (410bp) is sufficient to drive high level expression of BLG in the mammary glands of transgenic mice (9). Truncation of 3' flanking sequences demonstrated that these are not essential. A similar resection analysis of BLG promoter constructs was performed in the mammary epithelial cell line HC11. These results essentially recapitulate the transgenic data (Figure 1). The basal level of BLG expression is substantially reduced (over 5-fold) by deletion of sequences distal to -800 whilst the hormonal responsiveness is retained. Deletion of sequences between -410 and -800 has a small effect on the induced level of expression whilst resection to -210 reduces expression approximately 3-fold. Expression is practically abolished by deletion of a further 100bp to -150. These results demonstrate that the essential lactogenic hormone response elements reside between -150 and -410 in the sheep BLG promoter.
Regulatory elements in the BLG promoter
This 'minimal' 410bp BLG promoter was further characterised by identifying transcription factor binding sites using the electrophoretic mobility shift assay (EMSA) and a series of overlapping oligonucleotides which spanned the region from -80 to -410. At least one protein-DNA complex was observed for each oligonucleotide, suggesting that a multiplicity of transcription factors (TFs) may bind to the BLG promoter (10). The identity of these transcription factors was determined by comparison of recognition sequences and supershift analysis with antibodies to a number of known TFs including nuclear factor 1 (NF1) and NFkB. Three sites were also found for a factor designated MPBF, now known to be STAT5. These STAT sites have different sequences with different affinities for STAT5 (Figure 2). A similar motif was found also in the b-casein promoter by Berndt Groner's laboratory and initially called MGF (11). In the mammary gland, STAT5 is activated by prolactin and the prolactin signal is transduced via the JAK/STAT pathway which is schematically illustrated in (Figure 3). Motifs for Ets factors and glucocorticoid response elements (GR) are also found within the minimal BLG promoter region.
Many of these TF binding sites are common to other milk protein promoters although it should be noted that the BLG promoter does not contain a sequence with good homology to the 'milk box', which is found in many milk protein gene promoters. This sequence spans the YY1 site, which represses expression of the rat b-casein gene (12), and may explain the high basal levels of BLG expression that we have observed (see below).
Role of specific transcription factors
In order to determine the role of individual TFs in regulating BLG expression, we abolished binding to selected sites and in various combinations using site-directed mutagenesis of STAT, NF1, NFkB and putative GR elements, which we had shown previously to bind these TFs.
Mutation of the 3 STAT sites (Figure 2) either individually or in all seven possible combinations was carried out and analysed in HC11 cells. Abolishing binding to a single site had no effect on the induction of BLG expression in response to the lactogenic hormones prolactin, dexamethasone and insulin (13). However, mutating 2 sites abrogated induction and mutation of all three elements completely abolished the hormone response. These results were essentially recapitulated in transgenic mice harbouring similar mutated BLG promoter constructs (13). This demonstrates that binding of a STAT factor, most likely STAT5, is not essential for expression of BLG in the mammary gland but mediates the hormonal response. In contrast, in Chinese hamster ovary cells which had been stably transfected with the long form of the prolactin receptor, a different hierarchy of sites was required for prolactin induction. In these non-mammary cells, the highest affinity site alone is essential for the response to prolactin (14). Remarkably, even in the context of the entire 4.2kb 5' region, this single site mediates the response to prolactin. It is clear from these results that additional factors are required for the mammary-specificity of expression.
A candidate factor is nuclear factor 1 (NF1). There are multiple binding sites for NF1 in the 410 bp BLG proximal promoter region (10). NF1 binds to the consensus palindromic motif TGGCA(N5)TGCCA although it can also recognise half sites. We have shown previously that variants of NF1 exist in the mammary gland and postulated that a mammary-specific form of NF1 may bind to sites in the BLG promoter. NF1 is encoded by multiple genes which are subject to differential splicing (15) thereby creating a multitude of isoforms which may have different functions. NF1 DNA-binding activity is present at significantly higher levels in primary mammary cells grown on extracellular matrix (ECM) compared to plastic or collagen but this activation is not prolactin-dependent (16). NF1 binding activity is therefore regulated differently from STAT5, whose activation is both prolactin and ECM-dependent. We have mutated putative NF1 binding sites in the BLG promoter. The effect of these mutations is to reduce the levels of BLG expression in HC11 cells and in transgenic mice. However, the reduction is minimal. This contrasts with the observations of Jeff Rosen's laboratory who mutated 2 NF1 sites in the proximal WAP promoter. Abolishing binding to the half site reduced WAP expression in transgenic mice by 90% and mutagenesis of the palindromic motif totally abrogated expression (17). The explanation for the apparent discrepancy in these results is most likely due to the presence of multiple NF1 sites in the BLG 410bp promoter region, some of which overlap. By analogy to STAT5, mutation of a single site is unlikely to have a remarkable effect.
Other TF binding sites have been mutated. A consensus NFkB site, which was shown in vitro to bind NFkB, does not have a role in regulating basal expression of BLG or in mediating the lactogenic hormone response in HC11 cells (18). Surprisingly, a mutation in a putative GRE, which completely abolished BLG basal expression in HC11 cells, had no effect in transgenic mice. The reason for this is not apparent. It may reflect differences in other transcrption factors or in the cellular environment. TF binding sites in the proximal BLG promoter are schematically mapped in (Figure 4).
It is almost certain that interplay between TFs is important for correct developmental regulation of BLG. We have shown that STAT5a and STAT5b form heterodimers which have different affinities for the 3 BLG motifs (19). Our mutational analysis has shown that at least 2 STAT sites are required for maximal BLG expression suggesting a synergy between motifs. Indeed, it has recently been shown that STAT dimers interact at adjacent sites to form tetramers and stabilise binding (20). It has been demonstrated that STAT5 and GR interact (B.Groner per.com) and it is very likely that NF1 will form complexes with other factors, including the basal transcription machinery.
The BLG promoters of other species have been characterised to some extent. A partial sequence of the bovine BLG promoter has been determined and shown to have a STAT binding site differing in only one position from the STM site in the sheep promoter. The possum promoter is also currently being characterised (J.Demmer pers.com).
Ectopic expression of BLG transgenes
One drawback of using the BLG promoter to drive expression in the mammary gland is that in some transgenic lines, low levels of BLG transgene expression can be detected in a number of tissues (21). This expression, unlike that in the mammary gland, is constitutive and is a feature of the transgene since ectopic expression is not detected in sheep. Transgenes comprising genomic BLG sequences are ectopically expressed in about 40% of lines (21). However, constructs comprising 4.2kb of BLG 5' sequences fused to a heterologous construct, SV40 T antigen, exhibited higher frequency of ectopic expression (K.Gordon and CJW) as evidenced by tumour development in a number of sites. This may reflect the nature of the transgene or the sensitivity of detection, since almost undetectable expression of the transgene in the mammary gland can result in tumour formation in other tissues. Generally, ectopic expression is not a major problem. However, in some situations a protein is produced which may have deleterious effects. One example is the ectopic expression of a viral oncoprotein (as we observed with our BLG/T antigen transgenic lines). Another is the expression of Cre recombinase in crosses with mice harbouring lox-flanked gene sequences. We have generated transgenic lines with BLG/Cre sequences and are currently investigating whether any of these lines exhibit ectopic expression of Cre, which could result in deletion of target sequences in tissues other than the mammary gland.
Position independant expression and transgene rescue
BLG genomic transgenes are expressed in a position-independent manner in the mammary gland (9). It is assumed that transgene arrays integrate into the host genome in a random manner, although it has been suggested that peri-centromeric regions may be favoured integration sites (22). The integration site often influences the level or pattern of expression of the transgene which may therefore be prone to position effects (23). However, expression of heterologous genes from the BLG promoter frequently results in either no expression (eg. BLG/CAT) or expression levels unrelated to copy number. In an attempt to overcome this problem, a transgene rescue strategy was devised (24) based on co-injection with genomic BLG constructs. This was successful in a number of cases but is not universally applicable suggesting that genomic BLG sequences insulate transgenes from position effects in some chromosomal locations but not in others. Similar experiments were carried out with WAP cDNA transgenes being 'rescued' by WAP genomic constructs and similar results were obtained (25).
Variegated, or mosaic, patterns of transgene expression is a phenomenon which is becoming more frequently observed. In the mammary gland, the levels of BLG in the milk of transgenic mice can vary by an order of magnitude between individual mice in the same transgenic line. The expression pattern, as revealed by in situ hybridisation, shows clusters of expressing cells surrounded by non-expressing cells which probably arise by clonal expansion (26). Similar observations have been made for WAP transgenes in the mammary gland and of five independent BLG/human serum albumin lines, four displayed variegated expression (27). It is interesting that lines of mice which exhibit variable expression in the mammary gland do not necessarily have variable expression in other tissues such as salivary gland and that the same transgene construct will not show variegated expression in other lines. What is the likely explanation of this phenomenon? In Drosophila, high transgene copy numbers and proximity to the centromere have been suggested to result in heterochromatisation of the transgene array leading to silencing of the locus (28). Variegated expression is not restricted to mammary gland and is an important consideration when interpreting the results of transgenic experiments, particularly when expression in all cells in required. For example, tissue- and temporal-specific deletion of a gene function using a system such as the lox/Cre recombinase system depends on Cre being expressed in all the target cells. We will analyse our lines of BLG/Cre transgenics by in situ hybridisation to check whether or not the transgene is exhibiting variegated expression.
The BLG promoter has been of value in identifying a number of TFs which may have a role in regulating milk protein gene expression and mammary gland development. However, the importance of some factors may not have been completely defined in large transgene arrays. In our STAT site mutagenesis analysis, one transgenic line harbouring only 1-2 copies of the trangene did not express BLG at detectable levels. In this context, loss of all 3 STAT sites completely abolished expression. A meaningful assessment of the role of TF binding sites may therefore require the development of techniques for the insertion of a single copy of a transgene.
Improvements in transgenic technology currently being developed include inducible expression systems. See this web site for more details!
I thank Tom Burdon and Jerome Demmer for being such excellent collaborators in the analysis of the transcriptional control elements in the BLG promoter. I also thank Maggie McClenaghan, Bruce Whitelaw and John Clark for valuable discussions of their work on variegated and ectopic expression and transgene rescue.
1. Jennes,R. (1982) In: Developments in Dairy Chemistry (ed. Fox,P.F.), Vol. 1, pp 87- 114. Applied Sciences Publication, New York.
2. Pervaiz,S. and Brew,K. (1985) Science 228 335-337. 3. Newcomer,M.E., Jones,T.A., Aquist,J., Sundelin,J., Eriksson,U., Rask,L. and Peterson,P.A. (1984) EMBO J. 3 1451-1454.
4. Osborne,R., Howell,M., Clark,A.J and Nicholas,K.R. (1995) J. Dairy Res. 62 321- 329.
5. Wheeler,T.T., Callaghan,M.R., Davis,S.R., Prosser,C.G. and Wilkins,R.J. (1995) Exp.Cell Res. 217 346-354.
6. Gaye,P., Hue-Delahaie,D., Mercier,J.C., Soulier,S., Vilotte,J.L. and Furet,J.P. (1986) Biochimie 68 1097-1107.
7. Harris,S., McClenaghan,M., Simons,J.P., Ali,S. and Clark,A.J. (1991) Dev.Genet. 12 299-307.
8. Ali,S. and Clark,A.J. (1988) J. Mol. Biol. 199 415-426. 9. Whitelaw,C.B.A., Harris,S., McClenaghan,M., Simons,J.P. and Clark,A.J. (1992) Biochem. J. 286 31-39.
10. Watson,C.J., Gordon,K.E., Robertson,M. and Clark,A.J. (1991) Nuc. Acids.Res. 19 6603-6610.
11. Schmitt-Ney,M., Doppler,W., Ball,R.K. and Groner,B. (1991) Mol.Cell. Biol. 11 3745-3755.
12. Meier,V.S., and Groner,B. (1994) Mol. Cell. Biol. 14 128-137. 13. Burdon,T.G., Maitland,K.A., Clark,A.J., Wallace,R., and Watson,C.J. (1994) Mol. Endo. 8 1528-1536.
14. Demmer,J., Burdon,T.G., Djiane,J., Watson,C.J., and Clark,A.J. (1995) Mol. and Cell. Endo. 107 113-121.
15. Santoro,C., Mermod,N., Andrews,P.C. and Tjian,R. (1988) Nature 334 218-224.
16. Streuli,C.H., Edwards,G., Whitelaw,C.B.A., Burdon,T.G., Delcommenne,M., Schindler,C., and Watson,C.J. (1995) J.Biol.Chem. 270 21639-21644.
17. Li,S. and Rosen,J,M. (1995) Mol. Cell. Biol. 15 2063-2070.
18. Burdon,T.G., Demmer,J., Clark,A.J., and Watson,C.J. (1994) FEBS Letters 350 177-182.
19. Philp,J.A.C., Burdon,T.G. and Watson,C.J. (1996) FEBS Letters in press.
20. Xu,X., Sun,Y-L. and Hoey,T. (1996) Science 273 794-797.
21. Farini,E. and Whitelaw,C.B.A. (1995) Mol. Gen.Genet. 246 734-738.
22. Dobie,K., Mehtali,M., McClenaghan,M. and Lathe,R (1996) submitted
23. Al-Shawi,R., Kinneard,J., Burke,J. and Bishop,J.O. (1990). Mol. Cell. Biol. 10 1192-1198.
24. Clark,A.J., Cowper,A., Wallace,R., Wright,G. and Simons,J.P. (1992) Bio/Technology 10 1450-1454.
25. McKnight,R.A., Wall,R.J. and Henninghausen,L. (1995) Transgenic Res. 4 39-
26. Dobie,K.W., Lee,M., Fantes,J.A., Graham,E., Clark,A.J., Springbett,A., ,R. and McClenaghan,M. (1996) Proc.Natl.Acad.Sci. USA 93 6659-6664.
27. Faerman,A., Barash,I., Puzis,R., Nathan,M., Hurwitz,D.R. and Shani,M. (1995) J. Histochem. and Cytochem. 43 461-470.
28. Dorer,D.R. and Henikoff,S. (1994) Cell 77 993-1002.
Beta-lactoglobulin, gene expression, Stat, mosaic gene expression
Submitted by: Christine Watson in October of 1996
University of Edinburgh
contributed: October, 1996