‘Repair’ of the chorionic somatomammotropin-A ‘enhancer’ region reveals a novel functional element in the chorionic somatomammotropin-B enhancer

(\J(olecular and Csllular Endocrinology Molecular and Cellular Endocrinology

119 (1996) 1 - 10

‘Repair’ of the chorionic somatomammotropin-A ‘enhancer’ region reveals a novel functional element in the chorionic somatomammotropin-B enhancer Aristides Lytras, Rama Mohan Surabhi, J. Feng Zhang, Yan Jin, Peter A. Cattini* Department of Ph_vsiology, Universily of Manitoba. 730 William Ar?enue, Winnipeg, Manitoba, Canada R3E 3J7

Received 17 January 1996; revised 2 February 1996; accepted 2 February 1996

Abstract Human chorionic somatomammotropin (CS) synthesis results from the independent expression of two homologous genes, CS-A and CS-B. A transcription enhancer factor-l (TEF-1) element and an upstream 81 bp modulatory domain, containing repressor (RF-l) and derepressor (DF-I) activities, are important for efficient CS-B enhancer function in transfected placental JEG-3 cells.

The equivalent region of the CS-A gene is not active. Although the TEF-1 element is conserved between the CS-A and CS-B genes, a single base substitution is present in the DF-1 element and two more are located between the RF-l and DF-1 sites in a region we term AF-1. Repair of the DF-1 site increased CS-A enhancer function -70-fold, but repair of previously uncharacterized AF-1 sequences was also required for full (CS-B like) enhancer activity. A 5 bp disruption of AF-1 sequences in the CS-B enhancer region, resulted in a 97% loss of stimulatory activity. The AF-1 sequences showed no intrinsic enhancer activity, however, they were able to significantly repress heterologous promoter activity stimulated by a TEF-I enhancer element. A high affinity/specificity interaction between JEG-3 nuclear protein and AF-1 sequences was confirmed by gel mobility shift assay. By comparison to ‘wild type’ AF-1 sequences, this interaction was competed to a lesser extent by both RF-1 and DF-1 elements, but not by mutated AF-1 sequences. The major protein binding to AF-1 sequences was estimated to be 23 kDa by UV crosslinking. These data indicate that enhancer activity can be generated by modulating binding events proximal to the TEF-1 element in the CS-A ‘enhancer’ region and that coordinated binding of AF-1 and DF-1 are required for efficient (CS-B) enhancer activity. kvwords:

Chorionic somatomammotropin:

Gene regulation; Enhancer; Placenta; Growth hormone family

1. Introduction Human chorionic somatomammotropin (CS), also known as placental lactogen, is a member of the physiologically and clinically important growth hormone (GH) gene family that includes both pituitary GH and prolactin [l]. CS is expressed at high levels in the syncytiotrophoblast of the placenta during pregnancy. The detection of specific high affinity receptors suggests an important role for CS in the regulation of maternal and fetal processes during pregnancy [2-61, including its effect controlling glucose availability to the fetus

* Corresponding 774 9517. 0303-7207/96/$15,00 PII

author. Tel.: + 1 (204) 789 3735 Fax: + 1 (204)

[7,8]. Production of CS results from the independent expression of two genes, CS-A and CS-B, which are contained at a single locus on chromosome 17 [1,9]. A 22 base pair (bp) transcription enhancer factor-l (TEF1) element [lo, 1 l] in a 241 bp fragment of the CS-B 3’-flanking DNA was shown to be important for efficient promoter activity in placental cells after gene transfer [9]. This element was shown to bind a 30 kDa protein, termed CSEF-1, as well as 55 kDa TEF-1 from placental BeWo cell line nuclear extracts [12]. Although the TEF-1 site used alone possesses enhancer activity, it does not contain all the information required for the complete enhancer activity seen with the 241 bp fragment [13]. We showed that additional repressor (RF-l) and derepressor (DF-1) sequences were contained in a

8 1996 Elsevier Science Ireland Ltd. All rights reserved

9n~n7-72n7(9h~n1777-u

modulatory domain of about 81 bp located upstream of the TEF-1 site [13]. The importance of these sequences. particularly those associated with the DF-l element. were shown recently in three independent studies to be required for efficient CS-B enhancer activity [I3 151. From the sequences examined, only mutations of the TEF-I and DF-l elements of CS-B essentially eliminated enhancer function [13,15]. resulting in at least a 40-fold lower activity than the full length enhancer and at least 5-fold lower than the activity of the TEF-I element when used alone [li]. The importance of the DF-1 site was also suggested by the high degree of similarity, in terms of structure and activity, between some sequences in the DF-I element (5’-TAGGATGT3’) and those that constitute a binding site for the rfs-I protooncogene product (5’-gAGGATGT-3’) in the 3’ enhancer of the T-cell receptor gene [ 13,161. The 22 bp TEF-I element is conserved between the CS-A and CS-B genes, yet the CS-A ‘enhancer’ region was inactive in human placental choriocarcinoma JEG3 cells after gene transfer [17]. A comparison of the 241 bp CS-B enhancer region with the equivalent region 01 the CS-A sequences revealed six nucleotide differences [13,14]. None of these differences falls within the 22 bp TEF-1 enhancer element, but does include three nucleotide differences in the 81 bp modulatory domain. One base change is present in the (JI.Pregion of the DF-I (DF- 1jets) element (A to G; 5’-TAGGgTGT-3’) and the other two occur between the RF-l and DF-I sites in a region of the 81 bp modulatory domain we refer to as AF-I. Here, we demonstrate that ‘repair’ of the DF-1; rts site in the CS-A ‘enhancer’ region results in activation of the CS-A enhancer, but that the level of stimulation observed ( c 70-fold) is significantly less than the _ 300-fold effect seen with the equivalent CS-B enhancer region. However, comparable levels of CS-A and CS-B enhancer activity were observed when the AF-1 region was also ‘repaired’. Since the AF-1 region was essentially uncharacterized and represented a novel regulatory sequence, we tested it for activity in the context of the CS-B enhancer. Disruption of the AF-1 region resulted in a 97”/;1 loss of CS-B enhancer activity. An interaction between AF- 1 and TEF-1 was also suggested by the ability of an isolated AF-1 region to repress heterologous promoter activity stimulated by a TEF-I enhancer element. The presence of a high affinity/specificity interaction between JEG-3 nuclear protein and AF-1 DNA was confirmed by gel mobility shift assay and implicated the involvement of a 23 kDa protein. Competition studies suggested that AF-I might interact with additional components (RF-l or DF-1) in the modulatory domain. These results indicate that an interaction between factors associated with AF-I and DF-1 sites in the 81 bp modulatory domain as well as the downstream TEF-1 site. is required for efficient CS enhancer activity.

2. Materials

and methods

Oligonucleotide primers were synthesised by the University of Calgary, DNA Synthesis Laboratory. Alberta. Canada or through the Gene Technology Group. DNA Synthesis Laboratory at the University of Manitoba, Manitoba, Canada. .?.2. Cloning

qf’ CS-A and CS-B enhancer .wqurtw~s hi*

PCR Upstream and downstream primers, 5’-CACTGTTAGTCTACATTTCAG-3’ and 5’-ACCAGTACAACAGCTGTGAAC-3’. respectively, were synthesized and used to amplify 265 base pair (bp) regions overlapping and flanking the 241 bp CS-A or CS-B enhancers from I /lg of PstI digested human genomic DNA. After an initial step at 94°C for 5 min, the amplification (in IO mM Tris-HCl, pH 8.3, 1.5-2.5 mM MgCl?, 50 mM KCl. 200 LLg/rnl gelatin. 0.2 mM of each dNTPs, 1 /IM of each primer and 2 U Tcq polymerase in a final volume of 50 ,LII) was obtained with 35 cycles of denaturation at 94°C for 1 min, annealing at 42°C for 40 s and extension at 72°C for 1 min. In the final cycle the extension time was increased to 7 min. The products were resolved by 4% agarose gel electrophoresis and visualised by ethidium bromide staining. The PCR generated CS enhancer fragments were subcloned in the 5’nzllI of pUCll9 [ 181 and sequenced as previously described [13]. Excision of these enhancer fragments, from the pUCll9 vectors. by AccI and Poll11 digestions and blunting of the protruding ends was followed by subcloning of the generated 241 bp to the Sal1 site downstream of the chloramphenicol acetyl transferase (CAT) gene directed by the SV40 promoter which we term SVp (pCAT-Basic Vector; Promega, Wisconsin, USA).

Site-directed mutagenesis [19,30] was done to ‘repair’ or disrupt CS enhancer sequences from 180 ng of plasmid template. After an initial step at 95°C for 5 min. amplification was done as described above with minor modifications (370 nM for each primer in a total volume of 0.1 ml; 27-33 cycles; and annealing at 36-46°C for 40 s). The products were resolved by I --4% agarose gel electrophoresis and, when necessary, were isolated by electroelution or excised from a gel slice (Qiaex gel extraction kit, Qiagen, Germany). Mixing, heat denaturation. vortexing and annealing of the subfragments followed by amplification with the two external sequencing primers resulted in full length mutated products.

A. Lytras et al. / Molecular and Cellular Endocrinology 119 (1996) l-10

The pUCl19 vector containing the 265 bp ‘enhancer’ fragment was used as template for (‘repair’) mutagenesis of the CS-A ‘enhancer’ fragment. Oligonucleotide primers corresponding to sequences within the 265 bp fragment but bearing 1 or 2 bp mismatches and identical to the CS-B sequences, were used for amplification of ‘repaired’ subfragments in separate reactions. These primers (AF- 1 primer 1: 5’-CG/“GCAATTTCT/‘GCTGCAAATTTGAG-3’; AF- 1 primer 2: 5’-CTCAAATTTGCAGCAGAAATTGCCG-3’; DF- 1 primer 1: 5’-GAGATGCCTAGGATGTTTTCT-3’; DF- 1 primer 2: 5’-AGAAAACATCCTAGGCA TCTC-3’; mismatches are underlined) were used in combination with pUCl19 sequencing primers corresponding to vector sequences flanking the polylinker. The PCR products were digested with Ace1 and PvuII and subcloned directly into a pUCl19 vector for sequencing and confirmation of the mutation. The 241 bp AccI/PvuII fragment (isolated from a 1022 bp fragment of the CS-B gene provided by Dr. G.F. Saunders, M.D. Anderson Cancer Center, Houston, TX) subcloned in pUC19 was used as the template for (disruption) mutagenesis of the CS-B enhancer region. Oligonucleotide primers corresponding to sequences within the 241 bp fragment but bearing 5 bp mismatches were used for amplification of mutated (m) subfragments in separate reactions. These primers (AF1m primer 1: 5’-CGGCAATTGACTATGCAAATTTGAG-3’; AF-lm primer 2: 5’-CTCAAATTTGCATAGTCAATTGCCG-3’; mismatches are underlined) were used in combination with pUC19 sequencing primers corresponding to vector sequences flanking the polylinker. The PCR products were digested with EcoRI and Hind111 and subcloned into a pUCl19 vector for sequencing and confirmation of the mutation. Both pUC19 or pUCl19 containing the wild type or mutated 241 bp CS enhancer fragments were digested with either AccI/PuuII or HindIII, blunting with Klenow and EcoRI to release the CS-A and CS-B fragments, respectively. Similarly, SVp was digested and modified at sequences downstream of the CAT gene to provide linearized vector with one BarnHI (blunted) and one Sal1 end. Ligation of the blunt ends of the vector and enhancer fragments was followed by repair/blunting of the Sal1 and EcoRI ends and subsequent ligation/circularization to produce the insertion of the 241 bp fragments 3’ of the CAT gene in the correct orientation. A hybrid CAT gene (SVp.TEF) directed by the SV40 promoter and containing the TEF-1 sequences (common to CS-A and B) was described previously (called SV4Op35) [13]. The 25 bp AF-1 upstream and downstream primers were annealed and inserted into SVp.TEF in the forward orientation at the BamHI site (after blunting) upstream of the TEF-1 element, main-

3

taining the 5’ to 3’ positioning of the two elements in the 241 bp enhancer fragment to generate SVp.AF + TEF. In addition, the AF-1 sequences ‘were inserted downstream of the CAT gene in SVp at the X&I site (after blunting) to generate SVp.AF. All constructs were confirmed by sequencing. The firefly luciferase gene directed by the cytomegalovirus promoter (CMVp.luc) was described elsewhere [ 131. 2.4. Cell culture, gene transfer and reporter gene analyses Human choriocarcinoma JEG-3 cells were obtained from the American Type Culture Collection. Cells were grown in RPM1 1640 supplemented with 8% fetal bovine serum (FBS)’ in monolayer culture in a humidified atmosphere of 95% air and 5% carbon dioxide. Cells were plated at l-2 x 10” per 100 mm dish and transfected by the calcium phosphate/DNA precipitation method essentially as described previously [21]. Briefly, cells were transfected with 15 @g of hybrid CAT plasmid DNA and 1.0 pug of CMVp.luc in 10% FBSDMEM. After 24 h. cells were refed with growth medium and maintained for up to 48 h before processing. Cells were lysed and protein concentration was determined as previously described [22]. CAT activity was measured using a modification of the two-phase fluor diffusion assay [22]. Values for CAT activity were determined by regression analysis to give cpm/min per mg of cell lysate protein. Luciferase activity per mg cell lysate protein was determined using a ‘Luciferase Assay System’ (Promega Corp., Wisconsin. USA) according to manufacturers instructions and a photon counting luminometer (ILA911 Luminometer, Tropix Inc., Bedford, MA, USA). For quantitation, CAT values were divided by luciferase activity and CAT activity was expressed as the mean fold increase over the basal SV40 promoter activity plus or minus standard error of the mean from at least nine determinations (n) from three separate experiments. The range for basal SV40 promoter activity (SVp) determined from transfected JEG3 cells used in this study was 328.9 _+ 75.14 cpm/min per mg protein (n = 10). 2.5. Statistical analysis Statistical analysis of the data was done using the unpaired t-test as well as by a one-way analysis of variance (ANOVA) and the Bonferroni multiple comparisons post-hoc test. The results were accepted if Bartlett’s test for homogeneity of variances indicated that the difference between standard deviations from each test group was not significant. When this difference was shown to be significant, analysis was done using the Mann-Whitney test (non-parametric). In all cases, a value was considered statistically significant if P was determined to be < 0.05.

Nuclear extracts were prepared from JEG-3 cells according to published protocols [23] and dialysed against 20 mM Hepes (pH 7.9), 20% v/v glycerol, 0.1 M KCl, 0.2 mM ethylenediaminetetraacetic acid, 0.5 mM dithiothreitol and 1 mM PMSF). Following determination of protein concentration [24], nuclear extracts were aliquoted and maintained at - 70°C. The gel mobility shift assay was performed essentially as described by Baldwin [25]. Nuclear extract (20 j/g) was incubated with “P-end-labelled AF-1 double stranded oligonucleotide fragment (annealed AF-1 primers 1 and 2; 0.5~~. 1.O ng; l-2 x lo4 cpm) in the presence of 2 jig of poly dl-dC. Incubation of the reaction (in 10 mM Hepes+ NaOH. pH 7.9, 50 mM KCl. 0.5 mM ethylenedi10% aminetetraacetic acid. glycerol, 1 mM dithiothreitol) on ice for 15 min and at room temperature for 20 min was followed by electrophoresis in non-denaturing 4% polyacrylamide gels with 1:60 bis to acrylamide crosslinking ratio. For competition. synthetic 25 bp wild type AF-1 (annealed AF-1 primers 1 and 2) or mutated AF-I (annealed AF-1 m primers I and 2) fragments as well as previously described [I 31 TEF-1, RF-l and DF- I double stranded oligonucleotides (lo-500-fold molar excess over radiolabelled fragment) were added to nuclear extracts on ice prior to the addition of the radiolabelled DNA. -7.7. U V Crosshking Nuclear extracts (10 /Lg) from JEG-3 cells were incubated with radiolabelled double stranded AF-1 oligonucleotide (5 ng, 1.4 x 10’ cpm/ng) in the presence of 4 Llg poly(dI-dC) with or without competitor oligonucleotides (50- or 250-fold molar excess). The binding reaction (in 10 mM Hepes-NaOH. pH 7.9. 50 mM KCl, 0.5 mM ethylenediaminetetraacetic acid, IO’%, glycerol and 1 mM dithiothreitol) was carried out in a 96-well microtiter plate on ice for 20 min and then at room temperature for 20 min before being placed back on ice. The reaction mixture was irradiated for 50 min in UV cross-linker (Stratagene), keeping a distance of 50 mm between the plate and the UV lamp. Sodium dodecyl sulphate (SDS) sample buffer was added to the binding reaction and boiled for 5 min before electrophoresis in a 10% discontinuous SDS-polyacrylamide gel. The gel was dried for subsequent autoradiography in the presence of an intensifying screen. The molecular weight of crosslinked complexes was determined through the use of prestained molecular weight standards (phosphorylase B, 112.0; bovine serum albumin, 84.0; ovalbumin. 53.2; carbonic anhydrase, 34.9; soybean trypsin inhibitor, 28.7; and lysozyme, 20.5 kDa; Bio-Rad Laboratories, California. USA). The molecular weight of DNA binding protein

was estimated following molecular mass of the crosslinking.

subtraction of 15.4 kDa, DNA element employed

the for

3. Results

Genomic sequences corresponding to the 3’ flanking 241 bp CS-B enhancer region [9] as well as the equivalent region of the CS-A gene were inserted downstream of a hybrid CAT gene directed by the SV40 promoter. In addition, mutations of the CS-A ‘enhancer’ fragment were made to ‘repair’ (make like CS-B) the DF-1 ;‘rts site and/or the remaining two substitutions located between the RF-l and DF-1 sites in the 81 bp modulatory domain (Fig. 1A) described previously [13]. The effect of these CS ‘enhancer’ elements on SV40 promoter activity was tested in placental JEG-3 cells after gene transfer. The results are expressed as the fold stimulation relative to an ‘enhancerless’ promoter (SVp). for which activity was arbitrarily set to 1.O (Fig. IB). The presence of the CS-A ‘enhancer’ fragment resulted in only a threefold increase in promoter activity and, thus, possessed less than 2% of the stimulatory activity observed with the equivalent CS-B enhancer region. ‘Repair’ of the DF-liets site in the CS-A fragment resulted in a 73-fold enhancement (P < 0.0001, II = 12). however this represented only 27% of the activity seen with equivalent CS-B sequences. These data were not significantly different from those obtained following ‘repair’ of the AF-1 region (between RF- 1 and DF-1) in the CS-A fragment. which resulted in a 39-fold enhancement corresponding to 15% of 24 1 bp CS-B enhancer activity. In contrast, repair of both DF-1 ,c’ts and AF-1 sequences resulted in enhancer activity (255.5 f 36.8 fold, II = 25) that was not significantly different from that seen with equivalent CS-B enhancer sequences (268.1 + 41.9 fold, n = 11). -1.2. Mod$cution of’ nuckotides 4644 (AF- 1 site) irl tlw CS-B enhancer rlimiwtes stimulutog~ jimction To assess whether AF-1 sequences were of functional significance in the context of the CS-B enhancer, a 5 bp substitution was made in the 241 bp CS-B enhancer fragment in the centre of the AF-1 region, encompassing one of the nucleotides that differs in the equivalent region of the CS-A gene (Fig. 2A). Similar 5 bp mutations of the DF-1 and RF-l sites, done previously [13], were also included in the study for comparison (Fig. ?A). The effect of wild type and modified CS-B enhancers on SV40 promoter activity was tested in pla-

A. Lvtras et al.

: Molecular and Cellular Etldocrinology 119 (1996) 1~ 10

cental JEG-3 cells after gene transfer. The results are expressed as the fold stimulation relative to an ‘enhancerless’ promoter (SVp), for which activity was arbi-

A OLIGO.

SX!QWNCE/MUTATION 14

RF-1

A RF-1 sits 18 -~TCT*CA*CAGCTCATCARCTT9gtgtGG wt QFr -:::::::::::::::::::::::::::::::::A::::::: AFr -::::.z::::::::::::::::::::::::::::G::::::: AFr/DFr-::::::::::::::::::::::::::!::::::G:::::::

5

33

CS-A

AF-1

acgtc

gacta

57

CG~CTGCTFcAAATTTGAG t tacg GAGATGCCTAGGATGTTTTCT

55 DF-1

36

CTCATCAACTTGGTGTGGACGGC

75

CS-A

wt -:C::::::::::::!::::::::::A:::::::::::::: DFr AFr -:T:::::::::::::::::::::::G:::::::::::::: AFr/DFr-:T:::::::::::::::::::::::A::::::::::::::

B

B

iWJpflrc*T

__AF-1

11X-l&EFl~

@

Wp.CS-A (AFr/DFr)

Hybrid Genes

Hybrid Genes Fig. 1. Effect of site-directed mutagenesis ‘repair’ of sequences in the 81 bp modulatory domain of the CS-A ‘enhancer’ region. (A) The sequence of the 8 I bp modulatory domain of the wild type (wt) CS-A gene is shown. Single base pair differences (corresponding to CS-B sequence) are shown and represent the point mutations made. Colons indicate conserved sequences in the CS-A and CS-B genes. Note, the sequence for the 81 bp modulatory domain of CS-B (wt) will be identical to CS-A (AFr/DFr). The location of the RF-I and DF-I sites, as identified by mutagenesis previously, are shown in lower case letters. The underlined sequences correspond to the AF-1 region identified in this study. (B) Wild type and mutated 241 bp CS enhancer fragments were tested downstream of a hybrid CAT gene directed by the SV40 promoter in JEG-3 cells. CAT activity was corrected for DNA uptake and expressed as the mean fold stimulation over basal SV40 promoter (SVp) activity plus or minus standard error of the mean from at least nine determinations.

Fig. 2. Elfect of site-directed mutagenesis of sequences in the 81 bp modulatory domain of the CS-B enhancer region. (A) The sequences corresponding to the RF-I, AF-I and DF-I regions generated as ohgonucleotides. in the 81 bp modulatory domain of the 241 bp CS-B enhancer are shown. The mutations made to each of these regions is shown in lower case above the bases to be mutated. Note the imperfect repeat, separated by 10 bp, composed of 7 bp and 8 bp (underlined). (B) Wild type and mutated 241 bp CS-B enhancer fragments were tested downstream of a hybrid CAT gene directed by the SV40 promoter in JEG-3 cells. CAT activity was corrected for DNA uptake and expressed as the mean fold stimulation over basal SV40 promoter (SVp) activity plus or minus standard error of the mean from at least 9 determinations.

trarily set to 1.0 (Fig. 2B). Disruption of the AF-1 region essentially eliminated ( > 97%) CS-B enhancer activity and was not significantly different from the loss in activity seen following modification of the DF-I site. The - 2-fold increase in activity observed with a disrupted RF-l site was statistically significant (P < 0.0001, n = 18).

3

AF-1 30x

Fig. 3. Characterization incu[bated without

without

AF- I oligonucleotide.

of AF-I

binding

activities

10x

in placcntnl

in the praence

JL(;-.J

50x

250x

100x

500x

100x

500x

z” 3

nuclear- cxtraca.

Radiolabelled AF-I oligonucleotide (Fig. ?A ) uas tb) or with a IO- (c). 30. (d) and IOO- (e) fold excess of ‘cold’ oligonuclcotidc. In a separate assay. radiolabelled AF-I oligonucll eotide

(AF-lm)

(r) or with nuclear extract in the presence of poly dIdC’ without or 15. (h). 50. (I). 250- tm) fold molar excess of TEF-I

or IOO- (p), and 500. (q) fold molar excess of DLI

of J EG-3

15x

of poly did<’ uithout

or a IOO- tfl fold excess of mutated AF-I

mot jility band and a high afinity/specilicity preparations

-

75~

$ u ZF

hijklmnopqr

(a) or with nuclear extracl

AF- I oligonucleotide,

oliglonucleotide.

loox

bcdefg

a

was incubated

100x

25~

DF-1

RF-1

TEF-1

AF-1

AF-1 m

interaction

with the AF-I

tg) or with a IO- (h). 35. (i) 75. (j) fold excess of ‘cold‘ or IOO- tn). 500- to) fold molar excess of RF-I

oligonucleotide.

oligonucleotide. quenccs.

The open arrow

rcspectivcly.

and closed arrow

head indicate

which were observed consistently

a lower

with dif Teerent

nuclear extract.

Since previous studies had demonstrated only partial or no nuclease protection of the AF-I site, we used gel mobility shift assays with radiolabelled AF-I oligonucleotides and nuclear extracts from human placental JEG-3 cells in the presence of excess poly dldC. to assess possible DNA-protein interactions. Shifted bands. representing multiple complexes, were observed (Fig. 3). One band (closed arrow head) was competed effectively and reproducibly by low levels of ‘cold’ AF- 1 oligonucleotide suggesting a high affinity/specificity interaction (Fig. 3. lanes a-e). This result was confirmed using mutated AF-1 fragment (corresponds to the 5 bp modification in the AF-lm primer sequences) as a competitor. The high affinity/specificity band was not competed with a lOO-fold molar excess of mutated AF-1 fragment (Fig. 3. lane f). Since AF-I sequences modify enhancing activity (Fig. 2B), we examined whether TEF-1, RF-I or DF-I oligonucleotides could compete for AF- 1 complexes with JEG-3 nuclear proteins. to test whether AF-I

complexes include TEF-I. RF-l and DF-I protein(s) or cofactor(s). Interestingly. although the complex with the highest affinity/specificity was unaffected (closed arrow head), TEF- 1 oligonucleotide could compete with high affinity for a lower mobility AF-I complex (open arrow: Fig. 3, lanes k-m). In contrast, both DF-I and RF-l oligonucleotides were able to compete for the high affinity/specificity complex, although to a lesser extent than observed with wild type AF-I sequences (closed arrow head; Fig. 3, lanes n-q). Also, DF-I and RF-I oligonucleotides. like AF-1, did not compete efficiently for the lower mobility complex that was competed by the TEF-1 (open arrow; Fig. 3, lanes n 9).

3.3. U V crosslinking

of’ A F- I DNA /protein

cow@srs

UV crosslinking was done with JEG-3 nuclear protein and the AF-I element in the presence or absence of ‘cold’ AF-I oligonucleotide competitor. Following SDS-polyacrylamide gel electrophoresis of the products and autoradiography, the size of the (specifici competed) major protein/AF-I DNA complex was esti-

A. Lytras et al. 1 Molecular und Cehlar

tii

T! t

a

1

AF-1 250x

50x

b

c

Endocrinology 119 (1996) I-10

-

d

2 8 Pa8

8

TEF-1 250x

50x

f

g

-

h

Fig. 4. SDS-polyacrylamide gel electrophoresis/autoradiography of UV crosslinked JEG-3 nuclear protein and the AF-I element. Radiolabelled AF-I oligonucleotide was incubated without (a,e) or with nuclear extract in the presence of poly dIdC without (d,h) or with a 250- (b) and 50(c) fold excess of ‘cold’ AF-I oligonucleotide. or with a 250- (f) and 50- (k) fold excess of ‘cold’ TEF-I oligonucleotide. Complexes of 38 and 27.5 kDa are indicated.

mated to be 38 kDa (Fig. 4). A minor JEG-3 protein/ AF-1 DNA complex with a molecular mass of 27.5 kDa was also observed (Fig. 4). Neither of these complexes were competed as efficiently with TEF-1 oligonucleotide (Fig. 4). Subtraction of the molecular mass associated with the AF-1 element (15.4 kDa) from the values obtained for the major and minor complex suggests that proteins of about 23 kDa and 12 kDa can associate with the AF-1 site. 3.5. AF-I

sequences have u direct negntive TEF- 1 enhancer activity

after gene transfer and the results are expressed as the fold stimulation relative to an ‘enhancerless’ promoter (SVp), for which activity was arbitrarily set to 1.0 (Fig. 5). There was no significant effect of AF-1 sequences on heterologous promoter activity in the absence of the TEF-1 element. However, the 43-fold increase in promoter activity observed with the TEF-1 enhancer sequence (P < 0.0001, y1 = 21), was reduced significantly by 65% in the presence of the AF-1 sequences (P < 0.0001, n = 24).

effect on 4. Discussion

We examined whether the AF-1 element might possess intrinsic enhancer activity or act directly on the TEF-1 element to confer efficient enhancer activity. A double stranded AF-1 oligonucleotide, corresponding to the region located between the RF-l and DF-1 sites in the 81 bp modulatory domain (Fig. 2A), was generated by anealing AF-1 primers 1 and 2 (see Section 2). The AF- 1 oligonucleotide was introduced downstream of a CAT gene directed by the SV40 promoter to generate SVp.AF, or in combination with the synthetic 22 bp TEF-1 enhancer element to generate SVp.AF + TEF (Fig. 2A). Activity was assessed in JEG-3 cells

The results of this study confirm that the 241 bp CS-A ‘enhancer’ region is not functional in transfected JEG-3 cells in spite of the presence of an intact TEF-1 site. Further, the stimulatory function of the CS-A ‘enhancer’ region is less than that observed with the CS-A/CS-B TEF-1 element used alone, indicating the presence of a repressor activity in the 241 bp fragment. A comparison of the 241 bp CS-A and CS-B enhancer sequences (nucleotide positions l-241) revealed six differences (nucleotides 34, 43, 67, 87, 151 and 203). None of these differences occur in the 22 bp TEF-1

enhancer element, however. with the exception of nucleotides 34 and 43. they all fall within the limits of nuclease protected regions [ 14.151. Mutational analysis of the nuclease protected regions (nucleotides 8OG85. 153.-158 and 195-205) downstream of DF-1 and distinct from TEF-I, suggested they may be involved in enhancer activity as a 1.4 -2.9-fold decrease in stimulation was observed [15]. However, full CS-B enhancer activity was obtained with a fragment that excluded sequences downstream of nucleotide 148 indicating that the region downstream of the TEF-1 element may not be important for placenta-specific enhancer activity [14]. Mutation of the DF-I region at nucleotides 64-68 and 70--~75. resulted in a 40-fold reduction of enhancer activity representing a loss of 97.5% and 87-100% of CS-B enhancer activity, respectively (Fig. 2B) [13,15]. This low level of activity is comparable to that seen when the 241 bp CS-A ‘enhancer’ region was used (Fig. 1B). However, since the ets region of the DF-1 element

SV40 promoter ::KcAT/A SV40 promoter

l q q

TEF-1

q

::

SVP SVp.AF SVp.TEF SVp.AF+TEF

1

50 1

T

_I

01

Hybrid Genes Fig. 5. Effect of AF-I sequences on heterologous promoter activity in the absence or presence of a TEF-I enhancer. Oligonucleotides containing AF-I sequences (Fig. ?A) were tested downstream of a hybrid CAT gene directed by the SV40 promoter (SVp) in JEG-3 cells in the absence (SVp.AF) or presence (SVp.AF + TEF) of a 22 bp synthetic TEF-1 element. The hybrid gene SVp.TEF. containing the TEF-I element downstream of the CAT gene and directed by the SV40 promoter. was also included for comparison. CAT activity was corrected for DNA uptake and expressed as the mean fold stimulation over basal SV40 promoter (SVp) activity plus or minus standard error of the mean from at least 6 determinations.

in the CS-A sequences differs by a single nucleotide (Fig. IA). we questioned whether ‘repair’ of the DF-I)// cts site in the context of the CS-A ‘enhancer’ would be suficient to restore enhancer activity to levels seen with equivalent CS-B sequences. ‘Repair’ of the DF-I region within the CS-A ‘enhancer’ did result in the generation of a potent enhancer (Fig. 1B). comparable to the TEF-1 element used alone (Fig. 5). This indicates that repair of the DF-1 element in the CS-A ‘enhancer’ was sufficient to overcome repression by RF-l. ‘Repair’ of the CS-A ‘enhancer’, however, did not result in full activity when compared to the CS-B enhancer (Fig. 1B) suggesting that remaining nucleotide differences between CS-A and CS-B account for the reduced enhancer activity. Nucleotides 34 and 43 are located between the RF-I and DF-1 sites in a region of the 81 bp modulatory domain we term AF-1 (Fig. IA). Although nucleotide 34 was included in our RF-l oligonucleotide fragment described previously [13], nuclease protection studies suggest the 3’ boundary of the RF-l element is at nucleotide 27 [15]. No mutagenesis or deletion data were available for the AF-1 region. Thus, we chose to investigate whether ‘repair’ of the essentially uncharacterized AF-1 region in the CS-A gene would influence enhancer activity. ‘Repair’ of the AF-1 site within the CS-A ‘enhancer’ resulted, as in the case of the DF-I ‘repair’. in stimulatory activity comparable with the TEF-1 element when used alone (Fig. 5). This was in spite of the presence of the modified, by comparison to CS-B sequences, DF-1 site (Fig. IB). This suggested that by repairing either AF- 1 or DF- 1 in the context of the CS-A ‘enhancer’, a relief of a repressing effect on the function of TEF-1 was observed. This partially contradicts the result observed in the context of the CS-B enhancer, where in the presence of an intact AF-1 region, a 5 bp substitution of DF-1 sequences results in elimination of enhancer activity (Fig. 2B). In that case, however, the modification of DF-1 was far more extensive than the one nucleotide difference between the CS-A and CS-B enhancer regions. Thus, in the context of a CS-A ‘enhancer’ it may be that a ‘repaired’ AF-1 region can compensate for the loss of function of the DF-I region due to the one nucleotide substitution. In contrast, this is not the case for the CS-B enhancer where the mutation introduced into the DF-1 site was far more extensive. Regardless, the restoration of full (CS-B like) enhancer activity following repair of both the AF-1 and DF-1 sites in the CS-A ‘enhancer’ (Fig. lB), suggests that the AF-1 and DF-1 regions cooperate to confer derepressing activity. If this were the case, mutation of the AF-I region within the context of the CS-B enhancer should compromise derepression even in the presence of an intact DF-1 site. Within the context of the CS-A ‘enhancer’. where only two nucleotide differ-

A. Lytras et al. / Molecular and Cellular Endocrinology 119 (1996) I-10

ences at a distance of eight nucleotides are observed within the AF-1 region, repair of DF-1 results in enhancer activity. This suggests that a ‘repaired’ DF-1 element can compensate for this ‘minimal’ AF-1 modification. Thus, to secure complete disruption of the AF-1 site we introduced a substitution of 5 bp within the AF-1 region of the CS-B enhancer. As predicted, the AF-1 mutation alone essentially eliminated CS-B enhancer activity in placental cells in a similar manner to that observed with a disrupted DF-1 region (Fig. 2B) [13,15]. We showed previously that DF-1 and RF-l participate in the formation of a common complex or compete for common protein factors [13]. Although TEF-1 oligonucleotides could compete for the major DNAprotein complex formed on the 81. bp modulatory domain, that includes RF-l, AF-1 and DF-1 (Fig. lA), direct competition of complexes formed on either RF-l or DF-1 elements alone by TEF-1 oligonucleotides was not observed [13]. This suggested that there is either no direct interaction between the DF-1 (or RF-l) and the TEF-1 region or that the oligonucleotide competitors might be incorporated in the complex without causing the removal of proteins participating in this complex [13]. If the former were the case, then other sequences within the 81 bp modulatory domain might participate in complexes containing placental JEG-3 nuclear factors that can be directly associated with the TEF-1 element. Although the AF-1 complex with the highest affinity/specificity was unaffected, TEF- 1 oligonucleotides could compete with high affinity for a lower mobility complex that was formed on the AF-1 sequences (Fig. 3, lanes k-m). A possible explanation is that a TEF-1 like protein interacts with the AF-1 oligonucleotide. A comparison of core sequences in the TEF-1 site (5’-CTGGAATGTGG-3’) with part of the AF- 1 element (S’CTGcAAatTtG-3’) reveals some similarity. In contrast, both DF-1 and RF-l oligonucleotides could compete for the high affinity/specificity complex but not for the lower mobility complex that was competed by TEF-1 (Fig. 3, lanes n-q). This is consistent with the lack of direct interaction of DF-1 (or RF-l) and TEF-1. Fragments containing AF-1 and DF-1 elements (nucleotides 23-103) or a TEF-1 element (nucleotides 98148) were shown to possess moderate enhancing activity [14]. When used together, however, these two fragments resulted in a synergistic increase in enhancing activity [14]. We examined whether the AF-1 element of CS-B might act synergistically with the TEF-1 element to confer efficient enhancer activity. Surprisingly, introduction of AF-1 upstream of the TEF-1 element at the 3’ end of the CAT gene repressed TEF-1 enhancer activity 2.9 fold (or 65%; Fig. 5). In contrast, both the substitution present in the AF-1 region of the CS-A gene as well as the mutation we introduced in the CS-B

9

enhancer fragment resulted in loss of derepression (Fig. 1B and Fig. 2B). Although this result appears contradictory, the ‘repair’ of the CS-A ‘enhancer’ and the mutational analysis of the CS-B enhancer suggest that derepression requires both the AF-1 and DF-1 regions. Thus, if repression as well as derepression activities are present within the AF-1 region, derepression activity is not apparent when DF-1 is absent. Of course, additional sequences might also be necessary. The results from this study, specifically the ‘repair’ of the CS-A ‘enhancer’ region, demonstrate the importance of an interaction between elements/factors in the 81 bp modulatory domain and the downstream TEF-I site to obtain efficient enhancer activity. They also reveal the involvement of previously uncharacterized AF-1 sequences which were able to directly influence TEF-1 enhancer function on a heterologous promotor (Figs. 1 and 5). Our data indicate that a high affinity protein/DNA interaction occurs at the AF-1 site (Fig. 3) and likely involves a 23 kDa protein (Fig. 4). Although our data do not exclude the possibility that AF-1 and TEF-1 interact directly, we were unable to demonstrate that TEF-1 participates in the high affinity/specificity AF-1 complex (Fig. 3). The data suggest that CS-B but not CS-A possesses a functional enhancer. Paradoxically, a reverse transcriptase-PCR analysis of CS-A and CS-B RNA levels during pregnancy suggests that levels are equivalent at 8 weeks of pregnancy and by term CS-A levels are 5-fold more abundant than CS-B [26]. This increase in CS-A RNA expression is also reflected by the number of CS-A versus CS-B clones detected in a term placenta cDNA library [l]. Possible explanations include a dramatic difference in CS-A versus CS-B RNA stability or reflect the use of choriocarcinoma JEG-3 cells as a placental model system. Certainly, the level of CS synthesis and the pattern of placental hormone gene expression is not the same as in the syncytiotrophoblast of a term placenta [27]. However, an assessment of the chromatin containing the GH/CS locus in placental cells using accessibility to deoxyribonuclease I digestion, indicates that, like the CS-B sequences, hypersensitive sites are located downstream of the CS-A gene [28]. The results of our study indicate that enhancer activity can be generated by modulating binding events proximal to the TEF-1 element in the CS-A ‘enhancer’ region, in this case through modification of DF-1 and AF-1 sequences. At first glance, the rescue of enhancer activity could be dismissed as simply allowing factors to bind at these sites. However, since nuclease protection of, at least, DF-1 and TEF-1 were identified on the CS-A ‘enhancer’ region, it is more likely that our modifications to AF-1 and DF-1 sites allow a ‘functional’ interaction to occur. Thus, it is also possible that stimulation (or repression) might be induced by generation of a ‘functional’ interaction, not by mutation of

DNA as in this study, but through modification of proteins or novel protein-protein interactions that are absent in the placental culture system.

Acknowledgements This work was supported by a Medical Research Council of Canada grant (MT-10853). A. Lytras was the recipient of a Manitoba Health Research Council Studentship and P.A. Cattini is the recipient of a Medical Research Council of Canada Scientist Award. We would like to thank M.E. Bock and V.S.K. Chen for excellent technical assistance.

References [I] Chen EY, Liao Y-C. Smith DH. Barrera-Saldana HA, Gelinas RE. Seeburg PH. The growth hormone locus: nucleotide sequence, biology and evolution. Genomics 1989: 4: 479- 497. [2] Hill DJ. Freemark M. Strain AF, Handwerger S, Milner RDG. Placental lactogen and growth hormone receptors in human fetal tissues: relationship to fetal plasma. human placental lactogen concentrations and fetal growth. J Clin Endocrinol Metab 1988: 66: 128331290. M, Comer M. Purification of a distinct placental PI Freemark lactogen receptor, a new member of the growth hormone:prolactin receptor family. J Clin Invest 1989; 83: 883889. I41 Urbdnek M. MacLeod JN, Cooke NE. Liebhaber SA. Expression of a human growth hormone (hGH) receptor isoform is predicted by tissue-specific alternative splicing of exon 3 of the hGH receptor gene transcript. Mol Endocrinol 1992; 6: 2799287. M. Kirk K. Pihoker C. Robertson MC. Shiu RPC. [51 Freemark Driscoll P. Pregnancy lactogens in the rat conceptus and fetus: circulating levels distribution of binding. and expression of receptor messenger ribonucleic acid. Endocrinology 1993: 133: 1830&1842. [61 Urbanek M. Russell JE. Cooke NE. Liebhaber SA. Functional characterization of the alternatively spliced. placental human growth hormone receptor. J Biol Chem 1993: 268: 19025 19032. S. Clinical counterpoint: the physiology of placental [71 Handwerger lactogen in human pregnancy. Endocr Rev 1991; 12: 3299336. F. PI Brelje TC, Scharp DW. Lacy PE. Ogren L, Talamantes Robertson M. Friesen HG, Sorenson RL. Effect of homologous placental lactogens, prolactins, and growth hormones on islet B-cell division and insulin secretion in rat, mouse. and human islets: implication for placental lactogen regutation of islet function during pregnancy. Endocrinology 1993; 132: 8799887. SL. Saunders GF. Human placental [91 Walker WH, Fitzpatrick lactogen transcription enhancer. J Biol Chem 1990; 265: l294012948. I. Xiao JH. Rosales R, Staub A, Chambon P. The [lOI Davidson HeLa cell protein TEF-I binds specifically and cooperatively to two SV40 motifs of unrelated sequence. Cell 1988; 54: 93l~-943. C. Kanno M. Nomiyama H, Chambon P. CooperaIllI Fromental tivity and hierarchical levels of functional organization of the SV40 enhancer. Cell 1988; 54: 943 -953.

NL. Involvement of a protein distinct [I21 Jiang S-W. Eberhardt from transcription enhancer factor-1 (TEF-1) in mediating human chorionic somatomammotropin gene enhancer function through the GT-IIC enhanson in choriocarcinoma and COS cells. J Biol Chem 1995: 270: 13906- 13915. somatomammotropin 1131 Lytras A. Cattini PA. Human chorionic gene enhancer activity is dependent on the blockade of a repressor mechanism. Mol Endocrinol 1994; 8: 478. 489. P. Oury C, Peers B, Morin A, Belayew A. Martial [I41 Jacquemin JA. Characterization of a strong tissue-specific enhancer downstream from the three human genes encoding placental lactogen. Mol Cell Biol 1994: 14: 933103. NL. The human chorionic somatomam[I51 Jiang S-W. Eberhardt motropin gene enhancer is composed of multiple DNA elements that are homologous to several SV40 enhansons. J Biol Chem 1994; 269: 10384410392. LR, Lindsten T, Thompson CB, [IhI Ho IC, Bhat NK, Gottschalk Papas TS. Leiden JM. The Tr2 nuclear protein binding site from the human T cell receptor a enhancer functions as both a T cell-specific transcriptional activator and repressor. Science 1990; 250: 814-818. MG, Saunders GF. Transcriptional en[I71 Rogers BL. Sobnosky hancer within the human placental lactogen and growth hormone multigene cluster. Nucleic Acids Res 1986; 14: 764777659. of single-stranded plasmid DNA. [I81 Vieira J. Messing J. Production Methods Enzymol 1987; 153: 3 I I. [I91 Higuchi R, Krummel B, Saiki RK. A general method of in Vitro preparation and specific mutagenesis of DNA fragments: study of protein and DNA interactions. Nucleic Acids Res 1988; 16: 7351-7367. M. Construction of PO1 Vallette F, Mege E. Reiss A. Adesnik mutant and chimeric genes using the polymerase chain reaction. Nucleic Acids Res 1989: 17: 723733. NL. Regulated expression of chimaeric PII Cattini PA, Eberhardt genes containing the 5’-flanking regions of human growth hormone-related genes in transiently transfected rat anterior pituitary tumor cells. Nucleic Acids Res 1987; 15: 1297- 1309. [22] Nickel BE, Kardami E, Cattini PA. Differential expression of human placental growth hormone variant and chorionic somatomammotropin in culture. Biochem J 1990: 267: 653658. [3] Dignam JD. Lebovitz RM, Roeder RG. Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res 1983: I 1: 1475.m 1489. [24j Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976; 72: 2488254. AS. Analysis of sequence-specific DNA-binding P51 Baldwin proteins by the gel mobility shift assay. DNA Protein Eng Tech 1990; 2: 7376. WI MacLeod JN, Lee AK. Liebhaber SA, Cooke. NE. Developmental regulation and alternative splicing of the placentally expressed transcripts from the human growth hormone gene cluster. J Biol Chem 1992; 267: 14219- 14226. P71 Lytras A, Bock ME. Dodd JG, Cattini PA. Detection of placental growth hormone variant and chorionic somatomammotropin RNA expression in human trophoblastic neoplasms by reverse transcriptase-polymerase chain reaction. Endocrinology 1994: 134: 2461-2461. WI Jimtnez G. Ford AM, Enver T. Boronat A. Multiple changes in chromatin structure precede the transcriptional activation of the human growth hormone locus in placenta1 cells. Mol Cell Endocrinol 1993; 96: 53 -60.