A Krox binding site regulates protease nexin-1 promoter activity in embryonic heart, cartilage and parts of the nervous system

ELSEVIER

Mechanisms of Development O0 (1996) 139--150

A Krox binding site regulates protease nexin-1 promoter activity in embryonic heart, cartilage and parts of the nervous system H e n r i k E r n # , P a t r i c k KiJry 2, F l o r e n c e M. B o t t e r i , D e n i s M o n a r d * Friedrich Miescher-lnstitut, P.O. Box 2543, Basel, Switzerland Received 13 November 1995; revision received 25 July 1996; accepted 29 August 1996

Abstract

The rat protease nexin-1 (PN-1) promoter contains a G C G ~ G binding site for the transcription factors Krox-24, Krox-20 and NGFI-C. Mutations of this site abolished binding of Krox-24 in vitro. The wildtype protease nexin-1 promoter expressed flgalactosidase similarly to the expression of protease nexin-I mRNA. When the function of this Krox site was tested in vivo using transgenic F0 embryos, mutation had two opposite effects, fl-Galactosidase expression increased in cartilage and heart at both stages Ell.5 and E13.5, but was abolished in nerves of the central and peripheral nervous system at stage E13.5. These results suggest that Krox factors are among the important transcription factors regulating protease nexin-1 expression and thereby extracellular proteolytic activity in embryonic heart, cartilage and parts of the nervous system.

Keywords: Krox-24; PN-1 promoter; Transgenic mice; C6 rat glioma cells

1. Introduction The regulation of gene expression is a key issue in understanding the mechanisms controlling developmental processes. Information on the direct interaction of transcription factors with their target genes has for the most part been obtained from in vitro assays or studies on transcription in cultured cells. Recently, transgenic mouse technology was used to study the in vivo roles of transcription factors and the regulation of their target genes (Cheng et al., 1993; Sham et al., 1993; "fee and Rigby, 1993). Protease nexin-1 (PN-1) is a serine protease inhibitor with neurite promoting activity, widely expressed in the nervous system as well as in non neuronal tissues such as reproductive organs, lung, kidney, cartilage, heart and muscle (Mansuy et al., 1993; Reinhard et al., 1994). PN-1 is up-regulated after lesions in the central nervous system (CNS) (Hoffmann et al., 1992; Scotti et al., 1994) and in the peripheral nervous system (PNS) (Meier et al., 1989). * Corresponding author. 1 Present address: F.cole Normale Sup6deure, C.N.R.S. LIRA 1414, 46, Rue d'UIm 75005, Paris, France. 2 Present address: UMDS, Developmental Neurobiology, Guy's Hospital, London, SE1 9RT, UK.

The promoter of the rat PN- 1 gene contains, in addition to several Spl sites, eight potential binding sites for the Krox transcription factors (Ern¢ and Monard, 1993). So far, different factors belonging to this family have been characterized: Krox-24, of which two forms of different molecular weight are known, Krox-24 s2 and Krox-24 ss (Lemaire et al., 1990) (also known as egr-1; (Cao et al., 1990) and NGFI-A (Milbrandt, 1987)), Krox-20 (Chavrier et al., 1988), NGFI-C (Crosby et al., 1991; Crosby et al., 1992), egr-3 (Patwardhan et al., 1991) and Wilms' tumor factor (WT-1) (Rauscher et al., 1990). For reasons of simplicity we will refer to these factors and their consensus site as Krox. Krox-24 is expressed in muscle, cartilage and bone during development. (MacMahon et al., 1990; Watson and Milbrandt, 1990). At the same developmental stages, mouse PN-1 has been shown to be expressed in these structures (Mansuy et al., 1993). RT-PCR analysis demonstrated that Krox-24 is expressed in the sciatic nerve in early postnatal rats (Watson and Milbrandt, 1990). In the adult rat, Krox-24 is expressed in many different neurons (Mack et al., 1990; Worley et al., 1991) and in the olfactory bulb of the mouse (Brennan et al., 1992), a structure which contains high levels of PN-1 throughout life (Reinhard et al., 1988). PN-1 is expressed in different neurons of developing and adult CNS and PNS as well as

0925-4773/96/$15.00 © 1996 Elsevier Science Ireland Ltd. All fights reserved PII S 0 9 2 5 - 4 7 7 3 ( 9 6 ) 0 0 6 0 5 - 3

140

H. ErnO et al, / Mechanisms of Development 60 (1996) 139-150

in the mesonephric tubules, showing overlap with the expression patterns of Krox-20, N G F I - C and WT-1 (Reinhard et al., 1988; Armstrong et al., 1992; Crosby et al., 1992; Mansuy et al., 1993). Altogether, these findings suggest that the binding sites in the PN-1 promoter may interact with members of the Krox transcription factor family, and that these proteins may regulate PN-I expression in vivo. To test this hypothesis, binding of Krox factors to the PN-1 promoter was studied in vitro. A high affinity G C G G G G G C G site was identified and the consequences of mutating this site were analyzed in vivo.

2. Results 2.1. Characterization o f transcription factors binding to the rat PN-1 p r o m o t e r in vitro

Transient transfection experiments demonstrated that the PN-1 promoter fragment (-245 to +105) is transcriptionally active in C6 rat glioma cells (Ern¢ et al., 1996). Several factors in nuclear extracts from these cells or in extracts made from embryonic mouse brains, were found to bind to this GC-rich part of the PN-1 promoter (sequence shown in Fig. 1A) by DNase I footprinting and gel shift analysis. Computer analysis of the PN-1 promoter sequence showed that it contained several potential binding sites for S p l and eight sites homologous to the TG - C - G - T / g - G / A - G - G - C / a / t / - G - G / T sequence recently identified by Swirnoff and Milbrandt (1995) as the consensus site for the Krox transcription factors. DNase I footprint assays were performed with purified human S p l and extracts from E. coli expressing Krox2482. Two different promoter fragments were used: a wildtype fragment (-245 to +105) and the same fragment but with a mutation of the G C G G G G G C G site at -41 and - 4 2 to GCGGT'I'GCG. Such a mutation was shown to block binding of Krox-24 factor to this site (Nardelli et al., 1992a; Nardelli et al., 1992b). Using S p l on the wildtype PN-1 promoter two large footprints spanning from - 8 5 to - 5 4 and - 5 1 to - 3 0 were observed, both regions contain several S p l consensus binding sites. The

mutated promoter fragment displayed no S p l footprint in the region from -51 to - 3 0 , indicating abolished S p l binding upon mutation, whereas the more upstream S p l footprint (-85 to - 5 4 ) was untouched by the mutation (Fig. 1B, Spl). Note the different set of hypersensitivity bands between the wildtype and mutant promoter also in the absence of protein (Fig. 1B, S p l , lane 4 compared to lane 5). The PN-1 promoter was then analysed by footprinting with Krox-2482 expressed in E. coli. When the wildtype PN-1 promoter fragment was used, three Krox-24 footprints could be detected: from - 1 8 0 to -165, - 7 0 to - 5 3 and -51 to -30. The region from - 7 5 to - 5 3 contains two variants of the Krox consensus binding site: CCGG G G G C G at - 6 9 and G C G G G G G A G at - 6 3 , whereas the -51 to - 3 0 footprint contains a G C G G G G G C G site at 44. Mutating this site from G C G G G G G C G to G C G G T I ' G C G removed the strong K r o x - 2 4 footprint from -51 to -30 leaving the other two footprints (-180 to -165 a n d - 7 0 t o - 5 3 ) untouched (Fig. 2B, Krox-24). To test the effect of the mutation in non saturating conditions, a 206 bp fragment (-121 to + 85) encompassing the G C G G G G G C G site at - 4 4 with or without the mutation was used in gel shift assays with recombinant Krox-24 (Fig. 1C). This mutation prevented high affinity binding of Krox-2482 (Fig. 1C, lanes 8, 9 and 10) and Krox-20 (data not shown). Phosphorlmager analysis showed that the G C G G G G G C G to G C G G T T G C G mutation lowered the Krox-24 binding to the -121 to +85 fragment by 95%, despite the presence of the other binding sites on the fragment, indicating strong differences in the affinity of Krox-24 for different sites as previously observed by Swirnoff and Milbrandt (1995). When high amounts of protein (up to 5/~1) were used, additional complexes of high mobility were found (data not shown), confirming that additional Krox binding site(s) are present in the PN-1 promoter as identified by the footprint analysis in Fig. lB. However, the G C G G G G G C G consensus site at - 4 4 seemed to be necessary for high affinity binding of Krox-24. Since both Krox-20 and Krox-24 bind to the same sequence, it was of interest to see whether both factors would also bind to this

Fig. 1. (A) Sequence of the proximal PN-1 promoter (-245 to +105), numbering is done as in Em¢ and Monard (1993). The GCGGGGGCG site is bold and underlined. The transcriptional start site is shown by an arrow and the first exon is underlined. (B) Footprint analysis. The promoter fragments were endlabelled at the BamHl site at +105. The asterisk marks the position of the GCGGGGC~G to GCGGTTGCG mutation and the bars indicate the positions of the footprints. Spl: Lanes 1, 2, 3 and 4 show the wildtype promoter fragment and lanes 5,6 and 7 the mutant promoter fragment. One microliter of purified human Spl was used in lanes 2, 3, 6 and 7. Krox-24: Lanes 1,2 and 3 show the wildtype promoter fragment and lanes 4 and 5 the mutant promoter fragment. Lanes 2, 3 and 5 are footprint reactions performed in the presence of 5/~1 recombinant Krox-2482, lanes 1 and 4 are control reactions using 5 I~1E. coli extract. The marker (M) is the PN-1 promoter fragment sequenced using the Maxam-Gilbert GA reaction. (C) The GCGGGGGCG to GCGGTTGCG mutation prevents high affinity binding of Krox-24. The (-121 to + 85) fragments was used with approximately 0.5/~g probe per reaction. Lanes 1 to 5 are the wildtype promoter fragment and lanes 6 to 10 the mutated. Lanes 1 and 6, 0.5/~1 extract from control E. coli; lanes 2 and 7, no extract. Lanes 3-5 and 8-10, increasing amounts of E. coli extract containing Krox-2482 (lanes 3 and 8, 0.2/~1; lanes 4 and 9, 0.5/~1; lanes 5 and 10, 1/~1, respectively). (D) Binding of recombinant Krox-20 and Krox-24 to the GCGGGGGCG site in the PN-1 promoter. A 22 bp oligonucleotide (Krox-PN-1) corresponding to the sequence from the PN-I promoter -51 to -29 was used as a probe. Type and amount ~1) of extract added are indicated in the top rows, and the presence or absence of a 50-fold excess of unlabelled competitor is indicated. Krox-PN-1, same oligonucleotide as the probe; Krox, Krox specific; Spl, Spl specific; 0, no competitor oligonucleotide. The sequences of the oligonucleotides are shown in Section 4.

H. Ern# et al. / Mechanisms of Development 60 (1996) 139-150

141

A -245

C

5'ctcgagcccggccggtctggccgcgctttctgg

6

1 2 3 4 5

7

8

9 10

ggacccgacctccaccgccccagcgagaggg cagcatccggcgaccgcgggtcggcaggggg cgtcctaagtcccctgcggtggagagaccttgcg gccggctgccacacaaaggcggcggcgggaa ggcggggcggggcgggccgggggcggggga

-44

-36

ggcaggaaggaaCa~ClCOgcggcggt r-" gataaagcccccgcgctgcctggccggctaota caataattacaccaoaocacoaactacaaocat caccaccocct0tcccccaccoccococcoac ctccttcoccoctccooccaccacctooooatcc3'

.105

Spl

B

wt

Krox-24

mt

mt l

.

+

+

-

+

+

+ + - +

1 234567M -245

E.coli

-245

-145

E.col! rec. Krox-20

extract

E.coli rec. Krox-24

amount

2

1

2

1

1

1

1

2

1

1

competitor

0

0

0

~

~

~

0

0

~

"~

-165

-101

-101

-70

-70

-53 -37

D

12345

-53 "k

'k

-37

-25 -25

O aw

BB

,~

to~

1

142

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

G C G G G G G C G site in gel shift assays using a doublestranded oligonucleotide probe ( 5 ' - G G A A G G G G C G G G G G C G G C G G C G G - 3 ' , Krox-PN-1) corresponding to the (-51 to - 2 9 ) region. As expected Krox-20 and Krox-24 both bound specifically to this site since the binding could be competed away by adding unlabelled Krox oligonucleotides, but not by unlabelled S p l oligonucleotides (Fig. 1D). The effect of mutations in the high affinity Krox site at - 4 4 was further tested by transient transfection of C6 rat glioma cells which express PN-1 (Ern¢ and Monard, 1993), S p l and Krox-24 but not Krox-20, as determined by immunoblot assay and immunocytochemistry (data not shown). A 4.4 kb promoter fragment (-4.300 to +105) with or without the mutation was cloned upstream of the firefly luciferase gene. The GCGGq~FGCG mutation caused an approximately 2-fold increase in luciferase levels (Fig. 2A). To confirm that this effect was mediated by a factor belonging to the Krox family, a second mutation o f the G C G G G G G C G site at - 4 4 to G C T G G G G C G , also known to block Krox binding (Christy and Nathans, 1989), was tested. This mutation increased luciferase expression as much as the G C G G T F G C G mutation (Fig. 2A). The binding of nuclear proteins derived from C6 rat glioma cells to the wildtype and mutant PN-1 promoters was analyzed by DNaseI footprinting (Fig. 2B). This showed a large complex footprint on the wildtype PN-1 promoter fragment spanning from - 1 5 0 to +25. The G C G G T r G C G mutation introduced a DNase I hypersensitivity site at - 4 0 both with and without nuclear extract from C6 rat glioma cells. At the same time the band at - 4 9 was enhanced when the mutant promoter was incubated with nuclear extract from C6 rat glioma cells. The mutation did not change the footprint pattern outside the - 5 3 to - 3 0 area. Taken together the footprint analysis suggests that the mutation changed binding to the G C G G G G G C G site at - 4 4 but also indicates a change in protein/DNA interactions in the region immediately upstream of this site. The next step was to identify the binding factors present in nuclear extracts from C6 rat glioma cells. W e there-

fore performed gel shift assays using the Krox-PN-1 oligonucleotide as probe. The results showed that these cells contain at least three proteins or protein complexes, arbitrarily named a, b and d (Fig. 2C, lane 1) and that the mutation G C G G T F G C G blocked high affinity binding of nuclear proteins from C6 rat glioma cells to this site (Fig. 2C, lane 2), as shown above with the recombinant Krox proteins (Fig. 1B,C). The identity of the factors was investigated by the Shift-Western technique. The top band a was recognized by a specific antibody raised against Krox-24 s8 (Fig. 2D, lane 4) and band b was recognized by an antibody against Krox-2482 (Fig. 2D, lane 3). The identity of the lowest band d is not known, since it was not recognized by antibodies against Krox-20 (Fig. 2D, lane 2), Krox-24, egr-3, WT-1 or S p l (data not shown). In addition S p l can also be ruled out because of its large molecular weight. Taken together, these results showed that both forms of Krox-24 bind to the PN-1 promoter in vitro and that the G C G G G G G C G site at - 4 4 downregulates PN- 1 transcription in C6 rat glioma cells. The next step was to investigate the effect of an increase in the levels of Krox-24 on reporter gene expression controlled by the wildtype and mutant PN-1 promoter in C6 rat glioma cells. Co-transfection of the wildtype promoter plasmid with a Krox-24 expression vector increased reporter gene expression in a dose dependent manner. Co-transfection of Krox-24 did not increase transcription from the mutant promoter, consistent with the finding that the G C G G G G G C G site is necessary for high affinity binding (Table 1). The same results were obtained with co-transfection of plasmids expressing Krox-20 or N G F I - C (data not shown). In nuclear extract from embryonic mouse brain (E13.5) the presence of transcription factors binding to this site was analyzed by gel shift experiments using the Krox-PN-1 oligonucleotide as probe. Several factors (A to F) bound specifically to this probe (Fig. 3A, lane 2). To determine which factors were binding, competition experiments with unlabelled S p l or Krox consensus oligonucleotides were performed (for the sequences see Section 4). Addition of unlabelled Krox-PN-1 oligonucleotides to the binding reaction competed away all bound

Fig. 2. Effect of point mutations in the Krox site at --44 in the PN-i promoter. (A) C6 rat glioma cells were transiently transfected with the wildtype or mutated PN-I promoters (-4.300 to +105) controlling the luciferase expression; 'mock': transfection with a promoterless construct. Activities are given in light units per minute. Transfeetions were performed in duplicate. One representative experiment out of five is shown. (B) DNase I footprinting analysis of C6 rat glioma cell transcription factors binding to mutant and wiidtype PN-I promoter fragments, endlabelled at the BamHI site at +105. Lanes I and 2 show the wildtype promoter fragment and lanes 3 and 4 the mutant promoter fragment. Twenty micrograms of nuclear extract were used in lanes ! and 4. The same volume of Dignam nuclear extract buffer without protein was added to the control reactions (lanes 2 and 3). The marker (M) shows the promoter fragment sequenced using the Maxam-Gilbert GA reaction. The asterisk marks the position of the GCGGGGGCG to GCGGTTGCG mutation and the bars indicate the positions of the footprints. (C) Gel shift analysis of nuclear factors from C6 rat glioma cells binding to the Krox site at -44 in the PN-I promoter. One microgram of extract was used in each lane using a 4% polyacrylamide gel. Krox-PN-1 (i) and mutant Krox-PN-I (2) oligonueleotides were used as probes. (D) Shift-Western analysis performed as described in Section 4. Lane 1 is the autoradiogram of the DE81 chromatography paper binding the oligonucleotide probes. Under BA85 lanes 2, 3 and 4 are stripes of the nitrocellulose filter after immunoblotting and incubation with the antibodies. Lane 2 using an antibody against Krox-20, lane 3 using an antibody against Krox-2482 and lane 4 using an antibody against Krox-2488. Five micrograms of nuclear extract were used in each lane of a 5% polyacrylamide gel. The bands are arbitrarily named a, b and d.

H. Erno et al. / Mechanisms of Development 60 (1996) 139-150

A

B 20000 ¸

143

C6 rat glioma wt

15000

m u m

,~x~

-24s

1

mt l

4M

5000

0

mock

C_,C~G

C~_~C~

Gco~c~co

-165

Construct

C -101 a_

b_ d_

1

D

-75

2

-53

.

BA85

DF.81

ii,

.i

I

-44 a

~

bd_

1

2

3

4

-25

"k

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

144 Table 1

Co-transfection of Krox-24 and mutant and wildtype PN-I promoter plasmids

Wildtype promoter Mutant promoter

0.0/~g Krox-24 plasmid

0.3/~g Krox-24 plasmid

0.5/zg Krox-24 plasmid

100 ± 2 100 ± 18

344 _+21 79 ± 19

1031 ± 360 87 ± 17

1.5/,tg of the mutant or wildtype promoter plasmids were transfected into C6 rat glioma cells together with different amounts of Krox-24 expression vector (0, 0.3 or 0.5/ag). The total amount of DNA in the transfection assay was kept constant by addition of corresponding amounts of the expression plasmid without Krox-24 cDNA. The data are given as arbitrary units. The level obtained in the absence of the Krox-24 overexpression is defined as 100. This corresponds to 4899 light units for the wildtype promoter and to 30 895 light units for the mutant promoter. Transfection and luciferase assays were performed as described in Section 4. Transfections were performed in duplicate. proteins (Fig. 3 A , lanes 3 and 4). C o m p e t i t i o n with an increasing m o l a r excess o f S p l oligonucleotides s h o w e d that only the l o w e r two bands E and F but not the upper four bands A to D could be c o m p e t e d a w a y (Fig. 3A, lanes 5 and 6). All bands c o u l d be c o m p e t e d away by the presence o f K r o x c o n s e n s u s o l i g o n u c l e o t i d e competitors

(Fig. 3, lanes 7 and 8). This demonstrated that the proteins A to D were of the Krox type, w h i l e the E and F proteins were o f the S p l type. T h e A to D bands were strongly reduced w h e n mutant K r o x - P N - 1 oligonucleotides (5'-GG A A G G G G C G G ' I T G C G G C G G C G G - 3") were used as probe (data not shown). This s h o w e d that a functional Krox binding site is also necessary for high affinity binding of the factors present in the e m b r y o n i c m o u s e brain. In order to identify the proteins binding to the G C G G G G G C G site in e m b r y o n i c m o u s e brain, ShiftWestern assays were performed. H o w e v e r , no signal could be seen with any of the antibodies (results not shown). O n e reason for this might be that the antibodies detecting the rat Krox proteins do not r e c o g n i z e the m o u s e proteins. T o test this, a Shift-Western assay was p e r f o r m e d using K r o x - 2 4 s2 expressed in E. coli and nuclear extracts f r o m C6 rat g l i o m a cells, adult rat brain, adult m o u s e brain. The antibody against Krox-2482 recognized the band b in the C6 rat g l i o m a nuclear extract, the band B in rat brain nuclear extract, but not the B band in the m o u s e brain extract nor Krox-2482 expressed in E. coli (Fig. 3B). N o t e the identical migration o f the c o m plexes formed by nuclear proteins f r o m rat and mouse, suggesting that the c o m p l e x e s are f o r m e d by the h o m o l o gous proteins. The faster migration o f the E. coli Krox-

A 1

ii:i

2

34

5 •

6

78

....

ABCD-

B E

1

2

3

4

1

2

3

4

- -

F-

Krox.PN-1

$pl

Krox

DE81

BA85

Fig. 3. (A) Binding of nuclear factors from E13.5 mouse brain to the Krox binding site -44 of the rat PN-1 promoter sequence. Gel shift assays were carried out using the same pair of 22 bp oligonucleotides (Krox-PN-l) as in Fig. ID. Five micrograms of nuclear extract were used per binding reaction. Lane 1, no extract; lane 2, no competitor; lanes 3 and 4, cold probe (Krox-PN-1) as competitor; lanes 5 and 6, Spl specific competitor; lanes 7 and 8, Krox specific competitor (shown in Section 4). The triangles indicate increasing amounts of cold competitor oligonucleotides (50 and 100 fold excess). The bands are arbitrarily named A, B, C, D, E and F. (B) Shift-Western assay using the antibody against Krox-2482 with recombinant Krox-24 (lane 1), nuclear extracts from C6 rat glioma cells (lane 2), adult rat brain (lane 3) and adult mouse brain (lane 4). The gel shift and Shift-western assays were performed as in Fig. 2C,D. DE81 is the autoradiogram of the DE81 chromatogra~phy paper binding the oligonucleotide probes and BA85 is the nitrocellulose paper after immunoblotting. The arrow indicates the position of Krox-248~.

H. Era# et al. / Mechanisms of Development 60 (1996) 139-150

145

Table 2 Summaryof the effects of mutating the Kroxbinding site -44 in the 4.4 kb PN-I promoter Tissue

Embryos Stage El 1.5 wildtype (3)

Stage E11.5

GCGGTTC,-CG (2)

Stage E13.5 wildtype (8)

Stage E13.5 GCGGTTGCG (4)

Heart Cartilage

0 0

2 2

0 0

3 3

Olfactory projection Ocular motor nerve Ganglias Network Lumbarplexus

0 3 1 2 3

0 2 0 2 2

4 4 4 4 5

0 0 0 0 0

Otic vesicle

3

2*

0

0

The number of embryos showingfl-galactosideseexpression was scored. The total numberof embryosexpressingfl-galaetosidaseis shown in parentheses under each type of promoterconstruct. Only tissues showing an effect of the mutation are included. Transgenlcembryos which did not express fl-galactosidnseare not included in the table. *fl-Galactosidaseexpressiondriven by the mutant promoteris much strongerin and around this structure comparedto the wildtypepromoter. 24 az protein, compared to the mammalian Krox-2482 protein, is possibly due to lack of posttranslational modifications of the recombinant protein. 2.2. The in vivo role of the GCGGGGGCG site at -44 To test the in vivo function of this site we cloned the 4.4 kb wildtype PN-1 promoter fragment and a corresponding fragment mutated at the --44 Krox binding site (GCGGGGGCG to G C G G T r G C G ) in front of the lacZ gene. The fl-galactosidase expression driven by these promoters was analysed in 11.5 and 13.5 days old transgenie F 0 embryos. 2.2.1. In vivo expression of the lacZ gene controlled by the wildtype 4.4 kb PN-1 promoter The activity of the wildtype PN-1 promoter showed a common pattern of expression, displayed by the majority of the embryos (Table 2). In the CNS, the mammillary body of the diencephalon and the met-/mesencephalic boundary showed strong fl-galactosidase expression at both stages E l l . 5 (Fig. 4A) and E13.5 (Fig. 5A), as did projections or tracts emerging from these structures. Expression was seen at both developmental stages in several peripheral areas such as the dorsal posterior surface ectoderm of the paws, the most frontal part of the face, the ventral ectodermal ridge (VER) of the tail and the tip of the genital tubercle (Figs. 4A and 5A). However, the extent and strength offl-galactosidase expression in the paws was very sensitive to the integration site of the transgene. Out of eight wildtype embryos of stage E13.5, three displayed strong staining in the paws, three were weak and two had no staining. In the nervous system, the ocular motornerve was fl-

galactosidase positive at both stages E l l . 5 and E13.5 (Figs. 4A and 5A). The olfactory epithelium and projections from this structure towards the olfactory bulb, most likely being the axons of the olfactory nerves, were strongly stained at E13.5 (Fig. 5A). The periphery of the embryos contained two types of positive cells: (a) cells of the branchial and lumbar plexii with the fibers projecting from these structures distally into the limbs and (b) a network of elongated cells localized in the skin (Figs. 4A and 5A). At stage E13.5 expression was also found in cells of the trigeminal and sympathetic ganglia. Many cells in the dorsal root ganglia (DRG) also expressed flgalactosidase (Fig. 5A). Expression in these ganglionic structures was only seen when the fl-galactosidase expression was strong in other parts of the embryo, such as the brain. With exception of the paws the 4.4 kb PN-1 promoter fragment used drove fl-galactosidase expression in a reproducible pattern independent of integration site and this pattern is in agreement with the known PN-1 expression in mouse embryos at stages E11.5 and E13.5 (Mansuy et al., 1993). For example no expression at either stage was seen in liver and telencephalon where PN-1 is not expressed. However, in heart and cartilage no general flgalactosidase expression pattern could be seen using the wildtype promoter despite the fact that PN-1 mRNA is expressed in these tissues (Mansuy et al., 1993). 2.2.2. Mutation of the GCGGGGGCG site a t - 4 4 in the 4.4 kb PN-1 promoter abolishes /~galactosidase expression in nerves and increases expression in cartilage and heart Comparison with the common expression pattern seen in transgenic Fo embryos carrying the wildtype promoter

146

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

showed that the mutated Krox site at - 4 4 had two different and opposite effects on fl-galactosidase expression (Figs. 4A,B and 5A,B, Table 2). First, the GCGGTTGCG embryos showed strong fl-galactosidase expression at both developmental stages in heart and in facial, respectively skeletal cartilage (Figs. 4B and 5B). No expression was seen in the same embryonic structures with the wildtype promoter at either stage E11.5 or E13.5 (Figs. 4A and 5A, Table 2). The fl-galactosidase expression in the heart was confined to the cells in the wall of the left ventricle giving rise to cardio-myocytes (Fig. 4C). Expression in cartilage was observed in the limbs at E11.5 (Fig. 4D), in the tips of the ribs, in the nostrils at E13.5 (Fig. 5B) and in the head at both stages E11.5 and E13.5 (Figs. 4B and 5B). A comparison of the embryos at stage E11.5 showed a much stronger fl-galactosidase expression in and around the otic vesicle with the mutant promoter, than with the wildtype promoter. Second, the transgenic F 0 embryos with the G C G G T I ' G C G mutation showed no or reduced fl-galactosidase expression in olfactory projections and ocular nerves, or in the plexii at E 13.5 (Fig. 5B, Table 2), although this mutated promoter drove expression in some of these structures at E11.5 (Fig. 4B, Table 2). The mutant promoter was still driving flgalactosidase expression to the met-/mesencephalic boundary, although the projections from these structures were no longer stained (Fig. 5B).

3. Discussion 3.1. Interaction between Krox factors and the PN-1 promoter in vitro

The in vitro data presented here demonstrates that the G C G G G G G C G site at --44 in the PN-1 promoter is recognized by recombinant Krox-20 and Krox-24 factors (Fig. 1). In addition, C6 rat glioma cells were shown to express Krox-2482 and Krox-2488 proteins binding to the same site (Fig. 2C, lanes 3 and 4). NGFI-C expressed in COS-7 cells did also bind to the Krox site as detected by gel shift assays (data not shown). Whether WT-1 and egr3 would be able to bind to this site remains to be shown. Transfection of reporter constructs with wildtype and mutated Krox site at - 4 4 showed that it is involved in PN1 promoter activity in C6 rat glioma cells (Fig. 2A). On

the other hand, co-transfection of Krox-24 and PN-1 promoter/reporter constructs in C6 rat glioma cells showed that increased levels of Krox-24 will increase transcription from the PN-1 promoter and that this effect is dependent on a functional G C G G G G G C G site at -44 (Table 1). The following comments can be made when considering these results. Krox-24, like other members of the family, was originally described as a positive factor (Lemaire et al., 1990). However, it has recently been shown that Krox-24, in addition to its positive transactivating domains, contains a domain which acts as a negative regulator and that this latter domain interacts with a cellular co-factor (Gashler et al., 1993; Russo et al., 1993). This co-factor has been cloned and shown to be a nuclear protein called NAB1 which inhibits the ability of Krox-24 and Krox-20 to increase transcription (Russo et al., 1995). Thus, the function and consequently the effect of mutations in the Krox binding site depends on the type and on the amount of co-factor which interacts with the Krox protein in a given cell-type. This could explain why in the C6 rat glioma cells the reporter activity increases following the mutation. On the other hand, the increase in reporter gene activity observed after co-transfection of Krox-24 (Table 1) could mean that the increased level of Krox-24 would compete away the Krox-24/NAB1 complex and thus result in increased transcriptional activity from the PN-1 promoter. A further hypothesis is that mutations preventing binding of the Krox factors, could allow other factors, such as Spl, to bind at nearby sites and thereby increase transcription from the PN-1 promoter. This has been described for the mouse adenosine deaminase gene where mutation of a Krox site increased transcription due to an increased binding of Spl (Ackerman et ai., 1991). A similar mechanism might be proposed for the PN-1 promoter since C6 rat glioma cells express both the Spl and Krox-24 proteins (data not shown) and the PN-1 promoter binds purified Spl at several different sites including one overlapping with the GCGGGGGCG site at 4 4 (Fig. 1B, Spl). A change in binding was in fact observed in footprints with nuclear extracts from C6 rat glioma cells at a site immediately upstream of the mutated site (Figs. IB and 2B). This might indicate increased binding of Spl or of another nuclear factor interacting with these GC-rich sequences.

Fig. 4. PN- 1 4.4 kb promoter driven fl-galactosidase expression in transgenic E 11.5 embryos. (A) Wildtype 4.4 kb PN-1 promoter. (B) GCGGTTGCG mutated 4.4 kb PN-1 promoter. (C) Close-up of the heart of a GCGGTTGCGembryo. (D) Close-up of the limb of a GCGGTTGCGembryo. The promoter constructs are outlined below the corresponding embryo, the asterisks show the mutations, me, met-/mesencephalic boundary; m, mammillary body; o, ocular motornerve; otv, otic vesicle; pn, plexus and network; c, cartilage; h, heart. One representative embryo for each promoter construct is shown. Fig. 5. PN-I 4.4 kb promoter driven fl-galactosidase expression in transgenic E13.5 embryos. (A) Wildtype 4.4 kb PN-I promoter. (B) GCGGTTGCG mutated 4.4 kb PN-1 promoter. The promoter constructs are outlined below the corresponding embryo, the asterisks show the position of the mutations, me, met-/mesencephalic boundary; m, mammillary body; o, ocular motornerve; on, olfactory nerve; tg, trigeminal ganglia; sg, sympathetic ganglia; drg, dorsal root ganglia; pn, plexus and network; c, cartilage; nc, nostril cartilage; h, heart. One representative embryo for each promoter construct is shown.

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

147

!

4

C

A

B

D

~z

A

J

B I.AOZ

I.AOZ

148

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

The C6 rat glioma cells were used as a model in this detailed study because they express decent amounts of PN1, Spl and Krox-24. The fact that nuclear extracts from mouse and rat brain contain transcription factors interacting with this site suggests it to be relevant in normal neural tissue as well. In summary, the in vitro data showed that the Krox site is functional and that it binds Krox-24. Consequently, it was expected that Krox-24 and other Krox factors would also regulate PN-1 transcription in vivo. 3.2. The Krox site at - 4 4 regulates promoter activity in vivo

As transgenic embryos expressing reporter genes under the control of different promoter fragments are an interesting tool to approach the regulation of the gene of interest in vivo, we have chosen to analyse the reporter gene activity driven by the PN-1 promoter in embryos of stages E l l . 5 and 13.5, where sufficient data about the expression of the PN- 1 gene is already available (Mansuy et al., 1993). It is generally assumed that activity of a transgenic promoter on reporter gene expression in transgenic mice is sensitive to the integration site. We have observed such effects analyzing the PN-1 promoter activity and therefore we only discuss here the effect of mutating the Krox site at -44 in tissues where the promoter activity was consistently observed in the majority of transgenic embryos. The results summarized in Table 2 show that mutation of the Krox site has two opposite and tissue-specific effects; an up-regulation of reporter-gene expression in heart and cartilage at both E l l . 5 and E13.5, and a complete down-regulation in specific parts of the nervous system at the later stage. The reason for the different tissue-specific effects of the same mutation can, as discussed above, be explained by differential action of the Krox factors and/or by different combinations of factors interacting with Krox proteins or binding to the Krox site, respectively to other adjacent sites. The two opposite tissue specific effects of the GCGGT'I'GCG mutation on the reporter activity in vivo are consistent with our findings obtained with the C6 rat glioma cells and with the report that Krox-24 up-regulates expression from the PDGF-A promoter in human embryonic kidney 293 cells but down-regulates this same promoter in NIH3T3 fibroblasts (Wang et al., 1992). In cartilage, Krox-24 has been shown to be expressed at stage E14.5 and other stages of mouse development (MacMahon et al., 1990). One would therefore expect a change in fl-galactosidase expression in this tissue after mutation of the PN- 1 Krox site. Indeed, the mutation does cause an up-regulation in fl-galactosidase expression (Figs. 4 and 5). Interestingly, this is similar to the effect seen in C6 rat glioma cells which also express Krox-24. In the heart, no data are yet available on the expression of

Krox-24 in embryonic development, but we detected Krox-24 expression in mouse heart of stage E13.5 by RTPCR analysis (data not shown). High levels of Krox-24 mRNA are found in this organ at postnatal day 1 and adult stages (Watson and Milbrandt, 1990). Krox-24 has been shown to regulate the expression of rat a-cardiac myosin heavy chain gene in primary cultures of cardiac myoblasts (Gupta et al., 1991) and NAB1, the repressor of the Krox-24 transactivation, was found to be expressed in the heart (Russo et al., 1995). The increased reporter gene expression observed after the mutation, could therefore be due to decreased binding of the Krox-24/NAB 1 inhibitory complex, leading to an increase in transcriptional activity of the mutated PN- 1 promoter. On the other hand, embryonic mouse heart was shown to express Spl at both stages E8.5 and E12.5 (Saffer et al., 1991), consequently the increase in promoter activity could also be explained by an increased binding of Spl in the vicinity of the mutated site. It is interesting to note that neither embryonic heart nor cartilage express fl-galactosidase driven by the wildtype promoter, although these tissues were shown to contain PN-1 mRNA (Mansuy et al., 1993). This discrepancy could be explained by the absence of an enhancer element in the 4.4 kb promoter fragment regulating expression in these tissues. However, the GCGGTTGCG mutation seems to be, at least partially, sufficient to compensate for this. The disappearance of fl-galactosidase expression in structures of the CNS and PNS at E13.5 (Fig. 5 and Table 2) cannot be attributed with certainty to Krox-24 or any other member of this family, because sufficiently detailed data on Krox expression in these embryonic tissues are not available. Nevertheless, in the olfactory nerve, the sciatic nerve and ocular motornerves the GCGGTTGCG mutation clearly abolished fl-galactosidase expression at stage E13.5 (Fig. 5B). Later in development members of the Krox family are expressed at several of these sites. In adult mice and rats, Krox-24 is found in different types of brain neurons, among these, neurons in the olfactory epithelium and olfactory bulb (Brennan et al., 1992; Mack et al., 1990). The sciatic nerve was shown by RT-PCR to express Krox-24 at postnatal day 1, while DRG and superior cervical ganglia express little or no Krox-24 (Watson and Milbrandt, 1990). In addition to Krox-24, NGFI-C is yet another possible candidate for the effect in the peripheral nervous system, since it was shown by RTPCR to be expressed in the adult DRG, superior cervical ganglia and sciatic nerve (Crosby et ai., 1992), structures in which fl-galactosidase expression is turned off by the GCGGTTGCG mutation. No effect of the mutation was seen in structures where Krox-20 is expressed, as for example in the boundary cap cells of the dorsal root ganglia (Figs. 4 and 5; Sham et al., 1993). So far, our findings indicate that Krox-20 binds to the PN-1 promoter in vitro, but does not seem to regulate PN-I expression in vivo. It is surprising that the mutation of the Krox site at -44

H. Emlt et al. / Mechanisms of Development60 (1996) 139-150

149

4. Experimental procedures

plier (Biorad). The BA85 nitrocellulose membrane was cut in strips and analyzed with the following rabbit polyclonal antibodies: (588) directed against murine egr-I s2 (e.g. Krox-2482); (675) directed against murine egr-1 ss (e.g. Krox-24ss); (C-14) directed against murine egr-2 (e.g. Krox-20); (C-24) directed against egr-3; (C-19) directed against WT-1 and (PEP2) directed against Spl. The antibodies were used according to the manufacturer (Santa Cruz Biotechnology, Inc.). For detection of the proteins the ECL system from Boehringer-Mannheim was used. DNase I footprint assays were performed by incubating 200 ng of the promoter fragments (endlabelled at the +105 BamHI site) with the recombinant proteins or crude nuclear extracts in final volume of 50 #1 gel shift buffer supplemented with 4% polyvinyl-alcohol for 10 min on ice. Then an equal volume of 0.5#g/ml DNase I (Worthington) in 10 mM MgCI2, 4 mM CaCI2 was added. After 2 min incubation the reaction was stopped by the addition of phenol/chloroform. Then the DNA was ethanol precipitated and analyzed on a 6% sequencing gel. Purified human Spl was purchased from Promega, USA.

4.1. Gel shift and DNase I footprinting assays

4.2. Recombinant DNA and mutagenesis

Nuclear extracts from embryonic mouse brain, adult rat and mouse brain were prepared by the method of Schreiber et al. (1989), nuclear extracts from C6 rat glioma cells were according to Dignam et al. (1985). For the gel shift assays the following oligonucleotides were used (upper strand shown):

For site directed mutagenesis a fragment containing the PN-1 promoter from the SalI site (-383) to the BamHI site (+105) was cloned into the M13mpl8 vector. Site directed mutagenesis was performed according to Kunkel (1985) using the following mutagenic primers:

leads to a loss of reporter gene activity in specific structures of the nervous system at E13.5 but not at Ell.5 (Fig. 5 and Table 2). This could mean that different factors regulate PN-1 expression during different windows of development and that the role of the Krox site becomes predominant only at the later stage. In conclusion, we have shown that a functional GCGG~G site at -44 is important for tissue-specific activity of the PN-1 promoter. Mutation of this site leads to altered reporter gene expression in several embryonic structures which are known to express Krox-24 and/or members of the Krox transcription factor family. We therefore propose that PN-1 is a target gene for Krox-24, and possibly for other members of this family of transcription factors as well. Thus, Krox-24 may be one among other not yet identified transcription factors which regulate extracellular proteolysis in several different tissues during development. It remains to be shown whether other protease inhibitors and their target proteases are also regulated by members of the Krox family.

Krox-PN-1:5 ' - G G A A ~ ~ G C ~ G C ~ G G - 3 ' mutantKrox-PN-l:5 ' - G G A A ~ T T G C ~ ~ - 3 ' Kroxspecific:5'-GGATCCAGCGG~GACTTG-3' Spl specific:5 " - G A T C G A T C ~ ~ G A T C - 3 " The oligonucleotides were endlabelled by polynucleotide kinase and [7,-32p]ATP. Larger fragments were endlabelled with DNA polymerase I (Klenow enzyme) and [a-32p]dATP. Extracts from E. coli expressing Krox-20 or Krox-24 s2 and control E. coli extracts were kindly provided by J. Nardelli, Paris, France. The binding reaction was performed as described by Nardelli et al. (1992b) and the DNA/protein complexes were analyzed on a 4% native polyacrylamide gel. Gel shift assays with nuclear extracts were performed by incubating the labelled probes with 1 #g protein in presence of 50 mM NaCI, 1 mM DTI', 1 mM EDTA, 1 #M ZnClz, 50#g/ml poly-d[IC] and 5% glycerol (gel shift buffer). Shift-Westerns assays with nuclear extracts were performed using 5 pg protein per reaction performing a standard gel shift assay on a 5% polyacrylamide gel. The gel was then transferred to a BA85 nitrocellulose membrane placed on top of a DE81 chromatography paper by semi-dry electrotransfer as described by the sup-

5'-GCGGTTGCGGC~GC~GT-3' 5'-CAGGAA~CT~GC~GGCGGTGATAAAGC-3' After mutagenesis, the PN-1 promoter fragment was cloned into the luciferase vector (Erno and Monard, 1993) together with the upstream promoter sequences from the site HindlII (--4.300) to the SalI site (-383). The pSVflgal plasmid (Promega) was used for constructing the PN-1 promoter/lacZ reporter plasmids. The pSVflgal plasmid was digested with NcoI and HindlII, thereby removing the SV40 early promoter, and a linker with a 5' HindIII site and a 3' BgllI site was introduced instead. The PN-1 promoter fragments were then cloned into this plasmid using the HindlII site at (-4.300) and the BamHI site at (+105) of the promoter sequence. 4.3. Transfections and luciferase assays

C6 rat glioma cells were grown in Dulbecco's modified Eagle medium, supplemented with 10% fetal calf serum. The cells were transfected using the lipofection kit of Boehringer-Mannheim. Two days after transfection the cells were harvested and luciferase assays performed as described (ErnO and Monard, 1993). Transfections were performed in duplicate with two preparations of

150

H. Ern¢ et al. / Mechanisms of Development 60 (1996) 139-150

each plasmid. Luciferase activity was measured in triplicate.

4.4. Production and analysis of transgenic mice DNA (1-2 pl, conc. 2.5/,g/ml), a 8.5kb fragment containing the rat PN-1 promoter (from the HindlII site at (-4.300) to the BamHI site at (+105)) with or without point mutations, cloned upstream of the lacZ gene, was microinjected into the pronuclei of fertilized mouse oocytes. Those were obtained from C57BL/6 × BALB/C female mice mated to C57BL/6 males (Hogan et al., 1986). After 11.5 or 13.5 days of development, the F0 embryos were isolated, fixed and stained according to Zachgo and Gossler (1993). Placental DNAs were tested for transgene integration by Southern blot analysis. After fixation and staining, the embryos were dehydrated, cleared in benzylbenzoate/benzylalcohol (2:1) and photographed. Acknowledgements

We would like to thank A. Gossler for the protocols for whole mount staining of transgenic embryos, J. Nardelli and P. Charnay for recombinant Krox-20 and Krox24, H. van der Putten for assistance with transgenic mouse techniques, C. Lapize and S.A. Leuenberger for performing the Southern blot analysis and J.P. Jost, Y. Nagamine, M. Volovitch and F. Meins for critical reading of the manuscript. We thank P. Charnay for providing Krox-24 expression plasmid and thank the members of his laboratory for helpful discussions. H.E. was supported in part by a grant from the Danish Research Academy. References Ackerman, S.L., Minden, A.G., Williams, G.T., Bobinis, C. and Yeung, C.-Y. (1991) Proc. Natl. Acad. Sci. USA 88, 7523-7527. Armstrong, J.F., Pritchard-Jones, K., Bickmore, W.A., Hastie N.D. and Bard, J.B. (1992) Mech. Dev. 40, 85-97. Brennan, P.A., Hancock, D. and Keverne, E.B. (1992) Neuroscience 49, 277-284. Cao, X., Koski, R.A., Gashler, A., McKiernan, M., Morris, C.F., Gaffney, R., Hay, R.V. and Sukhatme, V.P. (1990) Mol. Cell. Biol. 10, 1931-1939. Chavrier, P., Zerial, M., Lemaire, P., Almendral, J., Bravo, R. and Chamay, P. (1988) EMBO J. 7, 29-35. Cheng, T.-C., Wallace, M.C., Merlie, J.P. and Olson, E.N. (1993) Science 261,215-217. Christy, B. and Nathans D. (1989) Proc. Natl. Acad. Sci. USA 86, 8737-8741. Crosby, S.D., Puetz, J.J., Simburger, K.S., Fahmer, T.J. and Milbrandt, ,j. (1991) Mol. Cell. Biol. 11, 3835-3841. Crosby, S.D., Veile, R.A., Donis-Keller, H., Baraban, ,J.M., Bhat, R.V., Simburger, K.S. and Milbrandt, J. (1992) Proc. Natl. Acad. Sci. USA 89, 4739-4743.

Dignam, J.D., Lebowitz R.M. and Roeder, R.G. (1985) Nucleic Acids Res. 11, 1475-1489. Ern0, H. and Monard, D. (1993) Gene Expression 3, 163-174. Era0, H., Kiiry, P., Nitsch, C., Jost., J.P. and Monard, D. (1996) Mol. Cell. Neurosci., in press. Gashler, A.L., Swaminathan, S. and Sukhatme, V.P. (1993) Mol. Cell. Biol. 13, 4556-4571. Gupta, M.P., Gupta, M., Zak, R. and Sukhatme, V.P. (1991) J. Biol. Chem. 266, 12813-12816. Hoffmann, M.C., Nitsch, C., Scotti, A.L., Reinhard, E. and Monard, D. (1992) Neuroscience 49, 397-408. Hogan, B., Costantini, F. and Lacy, E. (1986) Manipulating the mouse embryo. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Kunkel, T.A. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. Lemaire, P., Vesque, C., Schmitt, J., Stunnenberg, H., Randk, P. and Charnay, P. (1990) Mol. Cell. Biol. 10, 3456-3467. Mack, K., Day, M., Milbrandt, J. and Gottlieb, D.I. (1990) Mol. Brain Res. 8, 177-180. MacMahon, A.P., Champion, J.E., McMahon, .J.A. and Sukhatme, V.P. (1990) Development 108, 281-287. Mansuy, I.M., van der Putten, H., Schmid, P., Meins, M., Botteri, F.M. and Monard, D. (1993) Development 119, 1119-1134. Meier, R., Spreyer, P., Ortmann, R., Hard, A. and Monard, D. (1989) Nature 342, 548-550. Milbrandt, J. (1987) Science 238, 797-799. Nardelli, J., Gibson, T., and Charnay, P. (1992a) Nucleic Acids Res. 20, 4137-4144. Nardelli, J., Gibson, T., Vesque, C. and Charnay, P. (1992b) Nature 349, 175-178. Patwardhan, S., Gashler, A., Siegel, M.G., Chang, L.C., Joseph, L.J., Shows, T.B., LeBeau, M.M. and Sukhatme, V.P. (1991) Oncogene 6, 917-928. Rauscher, F.J., Morris, J.F., Tournay, O.E., Cook, D.M. and Curran, T. (1990) Science 250, 1259-1262. Reinhard, E., Meier, R., Halfter, W., Rovelli, G. and Monard D. (1988) Neuron l, 387-394. Reinhard, E., Suidan, H.S., Pavlik , A. and Monard, D. (1994) J. Neurosci. Res. 37, 256-270. Russo, M., Matheny, W.C. and Milbrandt, J. (1993) Mol. Cell. Biol. 13, 6858-6865. Russo, M., Sevetson, B.R. and Milbrandt, ,j. (1995) Proc. Natl. Acad. Sci. USA 92, 6873-6877. Saffer, J.D., .jackson, S.P. and Annarella, M.B. (1991) Mol. Cell. Biol. l l, 2189-2199. Schreiber, E., Matthias, P., MUller, M.M. and Schaffner, W. (1989) Nucleic Acids Res. 17, 6419. Scotti, A.L., Monard, D. and Nitsch, C. (1994) J. Neurosci. Res. 37, 155-168. Sham, M.H., Vesque, C., Nonchev, S., Marshall, H., Frain, M., Gupta, R.D., Whiting, J., Wilkinson, D., Charnay, P. and Krumlauf. R. (1993) Cell 72, 183-196. Swimoff, A.H. and Milbrandt, J. (1995) Mol. Cell. Biol. 15, 2275 2287. Wang, Z.Y., Madden, S.L., Deuel, T.F. and Rauscher, F.J. (1992) J. Biol. Chem. 267, 21999-22002. Watson, M.A. and Milbrandt, J. (1990) Development I l 0 173-183. Worley, P.F., Christy, B.A., Nakabeppy, Y., Bhat, R.V., Cole, A.J. and Baraban, ,J.M. (1991) Proc. Natl. Acad. Sci. USA 88, 5106-5110. Yee, S.-P. and Rigby, P.W.J. (1993) Genes Dev. 7, 1277-1289. Zachgo, J. and Gossler, A. (1993) In Joyner, A.L. (ed.), Gene Targeting: A Practical Approach. Oxford University Press. pp. 181-228.