Mol Gen Genet (1990) 220:353-360 © Springer-Verlag 1990
Interdependence and nodule specificity of c/s-acting regulatory elements in the soybean leghemoglobin lbc3 and N23 gene promoters Jens Stougaard, Jan-Elo Jorgensen, Tove Christensen, Astrid Kiihle, and Kjeld A. Marcker Department of Molecular Biology and Plant Physiology, University of Aarhus, C.F. MMlers All6 130, DK-8000 Aarhus C, Denmark Summary. The qualitative and quantitative contributions of four separate cis-acting D N A elements controlling the root nodule-specific soybean leghemoglobin lbc3 gene were analyzed in transgenic Lotus corniculatus plants. Expression from internal deletions in the 5' region between positions - 4 9 and - 1956 was monitored from a CAT reporter gene. The strong positive element (SPE; - 1090, - 947) responsible for high-level expression was demonstrated to be an organ-specific element by deleting proximal nodule-specific control elements. Deletion of the downstream qualitative organ-specific element (OSE; - 1 3 9 , - 1 0 2 ) containing the putative nodulin consensus sequences 5'AAAGAT and 5'CTCTT resulted in a low expression level. Efficient SPE enhancement is therefore dependent on the organ-specific element, which by itself does not enhance expression. This quantitative effect of the immediate upstream region carrying the consensus sequences was also found in hybrid promoter studies using the soybean nodulin N23 gene promoter, suggesting the involvement of these motifs in a regulatory mechanism for nodulin genes. Deletion of the lbca negative element (NE, - 1 0 2 , - 4 9 ) linking the SPE and OSE onto the TATA box did not lead to unregulated expression. These results indicate that interaction between positive, negative and neutral qualitative elements controls lbc3 expression. Binding of the nuclear protein NAT2 at the lbc~ weak positive element (WPE; - 2 3 0 , - 1 7 0 ) is probably not directly required for this mechanism. Key words: Nitrogen fixation - Plant gene regulation - Nodulin genes - 5' promoter analysis - D N A regulatory motifs
Introduction Several types of cis-acting elements are involved in the regulated expression of nuclear encoded eucaryotic genes. Gene analysis has identified strong positive elements, enhancers, organ- or cell-specific elements, and common promoter elements (Maniatis etal. 1987; Schell 1987). In addition, a silencer exerting position-independent negative effects on transcription has been identified in yeast (Brand et al. 1985). In plant genes, positive elements are generally located in the distal 5' upstream region. Short "constitutive" enhancer elements with little or no tissue specificity are located in the 5' distal regions of the 35S and octopine syn-
Offprint requests to: J. Stougaard
thase promoters (Ow et al. 1987; Ellis et al. 1987a). Strong positive elements linked to specific control elements are present in the upstream regions of developmentally or lightregulated genes and also in genes regulated by environmental stimuli (Timko et al. 1985; Simpson et al. 1986; Baumann etal. 1987; Chen etal. 1988; Poulsen and Chua 1988). Qualitative sequence elements directing selective gene expression in organs and tissues or responding to stimuli are generally located in the immediate upstream region of the respective promoters (Kiihlemeier et al. 1987; Walker et al. 1987; Ellis et al. 1987b; Lipphardt et al. 1988). The regular promoter elements like the CAT and TATA boxes are located close to the transcription initiation site, constituting the basic minimal promoter required for initiation of transcription (An et al. 1986; Fang et al. 1989). Transcription of mammalian genes is controlled by similar elements, although the number, location, and functions seem more diverse (Maniatis etal. 1987; Schirm etal. 1987; Wirth and Baltimore 1988; Str~ihle et al. 1988). We have analyzed the interaction between D N A sequence motifs involved in the root nodule-specific expression of the soybean leghemoglobin Ibc3 gene. By combination of a 5" and a 3" deletion series, we have systematically removed elements from a 2-kb 5' region which directs highlevel expression of the lbc3 gene in nodules (Stougaard et al. 1986). In comparative experiments, the differentially regulated nodulin N23 promoter (Jorgensen et al. 1988) was analyzed using hybrid-promoter constructs. Previously a 5' deletion analysis of the Ibca promoter delineated a strongpositive element (SPE) required for maximal expression between positions - 1 0 9 0 and - 9 4 7 , and a weaker positive element (WPE) sufficient for residual promoter activity between - 2 3 0 and - 1 7 0 (Stougaard et al. 1987; Fig. I). Two binding sites for a trans-acting nuclear protein factor (NAT2) from root nodules overlap the weak element of lbc~ (Jensen et al. 1988). The exact function of the protein factor in the ~molecular interactions directing the expression is, however, unknown. An organ-specific element (OSE; - 1 3 9 , --102) and a negative element (NE; - 1 0 2 , - 4 9 ) were further identified in the Ibe~ immediate upstream region (Stougaard et al. 1987). Presence of the 5'AAAGAT and 5'CTCTT motifs, conserved among nodulin genes (Sandal et al. 1987), in the OSE, and at similar positions in the N23 upstream region, suggested these sequences as core OSE motifs. Inability of the lbca and N23 minimal promoter regions to direct detectable expression from reporter genes further suggested a qualitative control
354 function for these motifs (Stougaard et al. 1987; Jorgensen et al. 1988). Here we report on experiments designed to elucidate the character and relative importance of the various elements and further discuss models for the regulatory mechanisms controlling the activation of nodulin genes. Materials and methods
Nucleic acid manipulation. Standard techniques described in Maniatis et al. (1982) were used for D N A manipulations. DNA-modifying enzymes were used according to the manufacturer's instructions. The 5' and 3' deletion series were generated by Bal31 treatment as described previously (Stougaard et al. 1987), except that XhoI linkers were ligated to the 3' deletion ends. Extraction and analysis of R N A was according to Marcker et al. (1984) and Stougaard et al. (1986).
Deletion constructs. The 3' deletion series was constructed in the YEpLbCAT101AXBA plasmid, derived from the YEpLbCAT transcriptional fusion (Jensen et al. 1986), by substituting a SaII linker for the XbaI fragment between position - 9 4 7 of 5' lbe3 and position 703 on the YEP24 vector plasmid. The EcoRI site seven nucleotides upstream of the A T G codon was replaced by a unique BglII linker site in YEpLbCAT101AXBA. The 3' deletions were obtained by Bal31 treatment of Bg/II-linearized YEpLbCATI01AXBA plasmid and recircularized by ligation of XhoI linkers. Internal deletions (AB,C,D,E,F,G,I,J,M) were subsequently constructed by subcloning selected 5' deletions (Stougaard et al. 1987) SalI/NcoI into the XhoI/NcoI sites of the 3' deletions, thereby reconstituting the CAT coding sequence and the lbc3 3' region. The internal deletions were inserted into the SalI site of pIV24 (J. Stougaard, unpublished work) linking the lbc3 (-1956, - 9 4 7 ) upstream region onto the internal deletions. Deletions AX,Z,P,Y were constructed by cloning the appropriate upstream regions in a polylinker upstream of the 5' deletions. T h e " minimal" 35S promoter construct ( - 9 0 ) 35S-CAT was derived from the 35S promoter (Jefferson et al. 1987) using the EcoRV site at --90. The 5' tbc3 restriction fragments IV, V, VIII, and X originate from MboII, AhaIII, and BanlI digests flushed with Klenow polymerase. Fragment XV originates from a SalI/XhoI digest of a 3' deletion. Constructions were checked by sequencing using the method of Hattori and Sakaki (1986). The D N A sequence of the complete 2-kb 5'Ibc~ region (Christensen etal. 1989) is filed with the EMBL genebank as accession no. X15061.
Transformation of plants. Gene constructions were transferred into Agrobaeterium rhizogenes as described by Van Haute et al. (1983). The AR14 strain carrying 35S-GUS in the TL segment (Hansen et al. 1989) was used as vector throughout this study. Composite plants with untransformed shoots on transformed "hairy roots" were produced according to Hansen et al. (1989). The transformation regeneration procedure (Petit et al. 1987) produced whole transformed plants. Biochemical assays. Activity from the chloramphenicol acetyl transferase (CAT) reporter was determined in five or more transgenic plants, as previously described (Stougaard et al. 1986, 1987). fl-glucuronidase (GUS) activity was measured by the fluorometric assay of Jefferson et al. (1987).
Results The soybean Ibc3 and N23 chimeric genes were transferred to Lotus corniculatus using Agrobacterium vector strains. Composite plants with A. rhizogenes-incited roots replacing the normal roots (Hansen et al. 1989) were generated for all lines. Whole transformed plants (Petit et al. 1987) were analyzed for lines where values from leaf and stem are given in Figs. I and 2; five or more independent transformants were analyzed for each line. The A. rhizogenes AR14 strain (Hansen et al. 1989) carrying the constitutively expressed 35S-GUS marker gene (Jefferson et al. 1987) was used as vector for all the chimeric CAT genes under study. Activity from the 35S-GUS gene therefore served as both a transformation marker and a control gene (see Figs. 1, 2, 4, and 6). The threefold variation in GUS activities reflects the copy number and insertion site effects generally observed in transgenic plants (Jones et al. 1985; Stougaard etal. 1987). Due to uncoordinate expression of marker and reporter genes, CAT activities could not be normalized. NAT2 BINDING SITES
-1090
-230-170 /-102-49 -139
-9/,7
~
g(D~
~ ~
B 1956H~-J_WPE -
-230
TATAA
~ ~
I
H OSE -170 -139 -102
-1~2 A C -1~2
NE
~
TATAA
IcA!
IY -49 I
AB
-7'7
A D -1~2 -l~Z /X N I
/~E
-9'7 I
-214
-139
GUS activity
CAT activity
nrnot/jag prot. hrs
cpm/pg prot.hrs
R
N
LS
N
0.20 ~ B 0 12200 0 0.56 ~C 0 5300 0 Z:XD 0 9300 0 0.36 0.53 Z~xN 0 /.3300 0 0.24 /X E 0 277900 lbc35'3'-CAT 0 222000 o 35S-CAT 8000 26700 12700 Fig. 1. A Organization of regulatory DNA elements in the 2-kb 5' region of the soybean leghemoglobin lbc3 gene. Binding sites for the NAT2 transacting factor are indicated by 1 and 2. Minimal promoter elements (CCAA, CACCC) and nodulin consensus sequences (AAAGAT, CTCTT) are indicated above the regulatory elements in B. See Christensen et al. (1989) for the complete nucleotide sequence and Stougaard et al. (1987) and Metz et al. (1988) for sequence comparisons of lb gene promoters. B Effect of internal deletions within the OSE and NE domains. Bars indicate the deleted sequences. Activity from the CAT reporter gene in roots (R) and nodules (N) are shown below, together with GUS activities from the 35S-GUS marker gene. Data from leaf and stem (LS) of whole transformed plants are included where available. CAT activities from the complete lbca5'3'-CAT and the constitutive 35SCAT genes are shown for comparison
355
-246
Binding site 1 -230 ¥
Binding site 2 -170 -223 -176 t
ATT? CTTAAATTATTTATTT, -161
t-< (D i
Ln O .£3
l -230
A E -2~ A G -214 i
&U
-170
-139
b
-
-102
AU
AD
AC
AN
A CAT probe
-1~9 -170
I
1
B
-159
-280
Ubiquitin
A I -2;o -2~o
AJ -3~s -3~2 A M -3;6
probe
-2~0 5
/%E AG hu AI
CATactivity
GUS activity
cpm/pg prot,hrs.
nrnol/IJg prot. hrs.
R 0 0 0 0
N LS 277900 121000 135700 0
/%J
255500 0 157500
/%M
0 117800 0
N 0.24 0.17 0.28 0.36
0.23 0.31
Fig. 2. Internal deletions in the weak positive element (WPE) removing binding sites for the NAT2 transacting factor. Binding site sequences are shown above the WPE element and deleted sequences below as bars. CAT activity from the reporter gene is given for roots (R) and nodules (N), together with GUS activity from the marker gene. Where available, data from leaf and stem (LS) of whole transgenic plants are included
The lbc3 organ-spec~'c and negative elements ( - 139,
-
-
49)
H y b r i d - p r o m o t e r studies using the 35S enhancer to reactivate inactive 5' deletions have previously located an OSE ( - 139, - 102) and an N E ( - 102, - 4 9 ) in the Ibc3 immediate upstream region (Stougard et al. 1987). The OSE containing the putative nodulin consensus motifs 5 ' A A A G A T and 5 ' C T C T T was required in these hybrid constructs for nodule-regulated expression, but was unable to express the lbe3 p r o m o t e r in the absence o f the heterologous enhancer. A qualitative function for the OSE was therefore suggested. The N E a p p e a r e d to suppress expression from reactivated 5' deletions where the OSE was not present (Stougaard et al. 1987). W e have now constructed internal deletions by removing one or b o t h o f these elements. C o m b i n a t i o n o f a 5' and 3' deletion series replaced sequences (shown with bars in Fig. 1) with a J(hoI/SalI hybrid site C C T C G A C C , except for AN, which was constructed by removing a BclI restriction fragment from the 5' region. Activity from C A T reporter gene fusions was measured in transgenic roots and nodules to determine the effect o f removing the specific control elements while leaving the upstream elements intact. Deletion o f the entire (--139, --49) region in AB or most o f the region in AC (--132, --77) results in a low level o f C A T activity in transgenic nodules (Fig. 1) and a reduced level o f steady state C A T m R N A (Fig. 3). C A T activity was not observed in roots
5
15
15
5
lag RNA Iooded
Fig. 3. A Northern analysis of RNA from root nodules of transgenic plants. Total RNA extracted from the AU, AD, AC, and AN lines was separated in agarose gels, transferred to nitrocellulose, and hybridized against CAT coding sequence. Five micrograms of RNA was loaded from the lbc35'3'-CAT control line and from the AU, AN lines. Fifteen micrograms RNA was loaded from the AD and AC lines. The lbc35'3'-CAT line represents the average expression level for the complete gene and was also used as control for CAT activities (Fig. ]). B Control hybridization showing the constitutive expression of the ubiquitin genes. The human ubiquitin cDNA (Wiborg et al. 1985) was used as probe; the hybridization signal corresponds to the trimer or tetramer ubiquitin polyprecursor. or leaves. The OSE and N E together are therefore not solely responsible for organ-specific control o f the lbca gene; other upstream elements also contribute specific control sequences. Deletion of the OSE in AD ( - - 139, - 102), leaving upstream elements and a minimal promoter, not only results in a more than tenfold reduction of C A T activity compared to A E ( - 2 1 4 , -- 139) or to the complete lbc35'3'-CAT gene, but also diminishes the steady state level o f C A T m R N A (Fig. 3). The level o f activity corresponds to the level originally observed after 5' deletion o f the SPE (Stougaard et al. 1987). This was unexpected, as the OSE itself was unable to activate the p r o m o t e r (Stougaard et al. 1987). The OSE is therefore required for efficient transcription initiation, but organ specificity is clearly also controlled by other elements. R e m o v a l o f the N E in A N (--97, --50) reduces the level of C A T activity and steady state m R N A in nodules. Constitutive expression in roots and leaves which could be expected after removal o f a putative repressor function was not detected. One possible explanation for the lowered expression level is the loss of basic p r o m o t e r elements in this deletion.
The lbc3 weak positive element (-230, -170) and the NA T2 binding sites The 5' deletion analysis o f the Ibc3 gene p r o m o t e r delimited a W P E between --230 and - 1 7 0 . Deletion o f this element reduced the expression o f the C A T reporter gene from 2% to below detection (Stougaard et al. 1987). Gel retardation with nuclear extracts from soybean nodules defined two binding sites (1 and 2) overlapping the borders o f the W P E (Jensen et al. 1988; Fig. 2). Both binding sites interact with
356 the same protein factor (NAT2) (Jacobsen et al. 1989), which in soybean is detected only in root nodules. The nodule-specific occurrence of this factor suggests a role for the weak element in mediating organ-regulated expression. Conservation of the sequence of binding site 2 in all soybean leghemoglobin genes and in the leghemoglobin genes from Sesbania rostrata (Metz et al. 1988) further indicates a possible role in general control mechanisms of leghemoglobin genes. To investigate these possibilities, we constructed deletions of the lbe3 binding sites. Complete or partial removal of binding site 2 (AE, AG) or binding site 1 (AI, AJ) did not affect CAT activity significantly (Fig. 2). Deletion of both binding sites in AU also had no dramatic effect on the activity of the receptor gene. When measured as CAT activity, no major effect on expression efficiency or organ specificity was detected deleting the WPE and the NAT2 binding sites. The AU plant lines were also analyzed at the R N A level (Fig. 3). R N A was extracted from nodules of regenerated transgenic plants, and steady state m R N A levels were determined by Northern analysis, using CAT coding sequence as probe. Total R N A from nodules of five different transgenic plants were mixed. The hybridization signal was compared with the control line carrying the complete lbea5'3'-CAT gene also used as control for CAT activities. No significant effect of deleting the NAT2 binding sites was detected at the R N A level. The weak positive element therefore seems to be a dispensable element, and the NAT2 factor is possibly a component of the unfolded chromatin structure near the accessible gene promoter.
(see Stougaard et al. 1987) and suggests that no indispensable regulatory elements are located between the OSE and SPE. Deletions AZ (--665, --102) and ziP ( - 6 6 5 , - 4 9 ) were constructed to investigate the importance of the promoter and immediate upstream elements. Results in Fig. 4 demonstrate that removal of the OSE from the minimal promoter in AZ reduced the CAT activity to a very low level. The OSE present in AX is therefore required for highlevel expression. Direct interaction between the SPE and the TATA box in AP results in a nodule-selective expression of approximately 5% of maximal level, demonstrating the effect of removing the negative element from AZ. Maximal promoter activity therefore appears to require interaction between the SPE and the OSE to overcome the negative regulation. Hybrid-promoter studies using a C a M V 35S ( - - 9 0 ) promoter Analysis of internal deletions and 5' deletions (Stougaard et al. 1987) indicates that no indispensable positive elements 150 bp !
I
o
o
-90, ~ ICAT ~-
Interdependence and nodule specificity of c/s-acting regulatory elements in the soybean leghemoglobin lbc3 and N23 gene promoters Jens Stougaard, Jan-Elo Jorgensen, Tove Christensen, Astrid Kiihle, and Kjeld A. Marcker Department of Molecular Biology and Plant Physiology, University of Aarhus, C.F. MMlers All6 130, DK-8000 Aarhus C, Denmark Summary. The qualitative and quantitative contributions of four separate cis-acting D N A elements controlling the root nodule-specific soybean leghemoglobin lbc3 gene were analyzed in transgenic Lotus corniculatus plants. Expression from internal deletions in the 5' region between positions - 4 9 and - 1956 was monitored from a CAT reporter gene. The strong positive element (SPE; - 1090, - 947) responsible for high-level expression was demonstrated to be an organ-specific element by deleting proximal nodule-specific control elements. Deletion of the downstream qualitative organ-specific element (OSE; - 1 3 9 , - 1 0 2 ) containing the putative nodulin consensus sequences 5'AAAGAT and 5'CTCTT resulted in a low expression level. Efficient SPE enhancement is therefore dependent on the organ-specific element, which by itself does not enhance expression. This quantitative effect of the immediate upstream region carrying the consensus sequences was also found in hybrid promoter studies using the soybean nodulin N23 gene promoter, suggesting the involvement of these motifs in a regulatory mechanism for nodulin genes. Deletion of the lbca negative element (NE, - 1 0 2 , - 4 9 ) linking the SPE and OSE onto the TATA box did not lead to unregulated expression. These results indicate that interaction between positive, negative and neutral qualitative elements controls lbc3 expression. Binding of the nuclear protein NAT2 at the lbc~ weak positive element (WPE; - 2 3 0 , - 1 7 0 ) is probably not directly required for this mechanism. Key words: Nitrogen fixation - Plant gene regulation - Nodulin genes - 5' promoter analysis - D N A regulatory motifs
Introduction Several types of cis-acting elements are involved in the regulated expression of nuclear encoded eucaryotic genes. Gene analysis has identified strong positive elements, enhancers, organ- or cell-specific elements, and common promoter elements (Maniatis etal. 1987; Schell 1987). In addition, a silencer exerting position-independent negative effects on transcription has been identified in yeast (Brand et al. 1985). In plant genes, positive elements are generally located in the distal 5' upstream region. Short "constitutive" enhancer elements with little or no tissue specificity are located in the 5' distal regions of the 35S and octopine syn-
Offprint requests to: J. Stougaard
thase promoters (Ow et al. 1987; Ellis et al. 1987a). Strong positive elements linked to specific control elements are present in the upstream regions of developmentally or lightregulated genes and also in genes regulated by environmental stimuli (Timko et al. 1985; Simpson et al. 1986; Baumann etal. 1987; Chen etal. 1988; Poulsen and Chua 1988). Qualitative sequence elements directing selective gene expression in organs and tissues or responding to stimuli are generally located in the immediate upstream region of the respective promoters (Kiihlemeier et al. 1987; Walker et al. 1987; Ellis et al. 1987b; Lipphardt et al. 1988). The regular promoter elements like the CAT and TATA boxes are located close to the transcription initiation site, constituting the basic minimal promoter required for initiation of transcription (An et al. 1986; Fang et al. 1989). Transcription of mammalian genes is controlled by similar elements, although the number, location, and functions seem more diverse (Maniatis etal. 1987; Schirm etal. 1987; Wirth and Baltimore 1988; Str~ihle et al. 1988). We have analyzed the interaction between D N A sequence motifs involved in the root nodule-specific expression of the soybean leghemoglobin Ibc3 gene. By combination of a 5" and a 3" deletion series, we have systematically removed elements from a 2-kb 5' region which directs highlevel expression of the lbc3 gene in nodules (Stougaard et al. 1986). In comparative experiments, the differentially regulated nodulin N23 promoter (Jorgensen et al. 1988) was analyzed using hybrid-promoter constructs. Previously a 5' deletion analysis of the Ibca promoter delineated a strongpositive element (SPE) required for maximal expression between positions - 1 0 9 0 and - 9 4 7 , and a weaker positive element (WPE) sufficient for residual promoter activity between - 2 3 0 and - 1 7 0 (Stougaard et al. 1987; Fig. I). Two binding sites for a trans-acting nuclear protein factor (NAT2) from root nodules overlap the weak element of lbc~ (Jensen et al. 1988). The exact function of the protein factor in the ~molecular interactions directing the expression is, however, unknown. An organ-specific element (OSE; - 1 3 9 , --102) and a negative element (NE; - 1 0 2 , - 4 9 ) were further identified in the Ibe~ immediate upstream region (Stougaard et al. 1987). Presence of the 5'AAAGAT and 5'CTCTT motifs, conserved among nodulin genes (Sandal et al. 1987), in the OSE, and at similar positions in the N23 upstream region, suggested these sequences as core OSE motifs. Inability of the lbca and N23 minimal promoter regions to direct detectable expression from reporter genes further suggested a qualitative control
354 function for these motifs (Stougaard et al. 1987; Jorgensen et al. 1988). Here we report on experiments designed to elucidate the character and relative importance of the various elements and further discuss models for the regulatory mechanisms controlling the activation of nodulin genes. Materials and methods
Nucleic acid manipulation. Standard techniques described in Maniatis et al. (1982) were used for D N A manipulations. DNA-modifying enzymes were used according to the manufacturer's instructions. The 5' and 3' deletion series were generated by Bal31 treatment as described previously (Stougaard et al. 1987), except that XhoI linkers were ligated to the 3' deletion ends. Extraction and analysis of R N A was according to Marcker et al. (1984) and Stougaard et al. (1986).
Deletion constructs. The 3' deletion series was constructed in the YEpLbCAT101AXBA plasmid, derived from the YEpLbCAT transcriptional fusion (Jensen et al. 1986), by substituting a SaII linker for the XbaI fragment between position - 9 4 7 of 5' lbe3 and position 703 on the YEP24 vector plasmid. The EcoRI site seven nucleotides upstream of the A T G codon was replaced by a unique BglII linker site in YEpLbCAT101AXBA. The 3' deletions were obtained by Bal31 treatment of Bg/II-linearized YEpLbCATI01AXBA plasmid and recircularized by ligation of XhoI linkers. Internal deletions (AB,C,D,E,F,G,I,J,M) were subsequently constructed by subcloning selected 5' deletions (Stougaard et al. 1987) SalI/NcoI into the XhoI/NcoI sites of the 3' deletions, thereby reconstituting the CAT coding sequence and the lbc3 3' region. The internal deletions were inserted into the SalI site of pIV24 (J. Stougaard, unpublished work) linking the lbc3 (-1956, - 9 4 7 ) upstream region onto the internal deletions. Deletions AX,Z,P,Y were constructed by cloning the appropriate upstream regions in a polylinker upstream of the 5' deletions. T h e " minimal" 35S promoter construct ( - 9 0 ) 35S-CAT was derived from the 35S promoter (Jefferson et al. 1987) using the EcoRV site at --90. The 5' tbc3 restriction fragments IV, V, VIII, and X originate from MboII, AhaIII, and BanlI digests flushed with Klenow polymerase. Fragment XV originates from a SalI/XhoI digest of a 3' deletion. Constructions were checked by sequencing using the method of Hattori and Sakaki (1986). The D N A sequence of the complete 2-kb 5'Ibc~ region (Christensen etal. 1989) is filed with the EMBL genebank as accession no. X15061.
Transformation of plants. Gene constructions were transferred into Agrobaeterium rhizogenes as described by Van Haute et al. (1983). The AR14 strain carrying 35S-GUS in the TL segment (Hansen et al. 1989) was used as vector throughout this study. Composite plants with untransformed shoots on transformed "hairy roots" were produced according to Hansen et al. (1989). The transformation regeneration procedure (Petit et al. 1987) produced whole transformed plants. Biochemical assays. Activity from the chloramphenicol acetyl transferase (CAT) reporter was determined in five or more transgenic plants, as previously described (Stougaard et al. 1986, 1987). fl-glucuronidase (GUS) activity was measured by the fluorometric assay of Jefferson et al. (1987).
Results The soybean Ibc3 and N23 chimeric genes were transferred to Lotus corniculatus using Agrobacterium vector strains. Composite plants with A. rhizogenes-incited roots replacing the normal roots (Hansen et al. 1989) were generated for all lines. Whole transformed plants (Petit et al. 1987) were analyzed for lines where values from leaf and stem are given in Figs. I and 2; five or more independent transformants were analyzed for each line. The A. rhizogenes AR14 strain (Hansen et al. 1989) carrying the constitutively expressed 35S-GUS marker gene (Jefferson et al. 1987) was used as vector for all the chimeric CAT genes under study. Activity from the 35S-GUS gene therefore served as both a transformation marker and a control gene (see Figs. 1, 2, 4, and 6). The threefold variation in GUS activities reflects the copy number and insertion site effects generally observed in transgenic plants (Jones et al. 1985; Stougaard etal. 1987). Due to uncoordinate expression of marker and reporter genes, CAT activities could not be normalized. NAT2 BINDING SITES
-1090
-230-170 /-102-49 -139
-9/,7
~
g(D~
~ ~
B 1956H~-J_WPE -
-230
TATAA
~ ~
I
H OSE -170 -139 -102
-1~2 A C -1~2
NE
~
TATAA
IcA!
IY -49 I
AB
-7'7
A D -1~2 -l~Z /X N I
/~E
-9'7 I
-214
-139
GUS activity
CAT activity
nrnot/jag prot. hrs
cpm/pg prot.hrs
R
N
LS
N
0.20 ~ B 0 12200 0 0.56 ~C 0 5300 0 Z:XD 0 9300 0 0.36 0.53 Z~xN 0 /.3300 0 0.24 /X E 0 277900 lbc35'3'-CAT 0 222000 o 35S-CAT 8000 26700 12700 Fig. 1. A Organization of regulatory DNA elements in the 2-kb 5' region of the soybean leghemoglobin lbc3 gene. Binding sites for the NAT2 transacting factor are indicated by 1 and 2. Minimal promoter elements (CCAA, CACCC) and nodulin consensus sequences (AAAGAT, CTCTT) are indicated above the regulatory elements in B. See Christensen et al. (1989) for the complete nucleotide sequence and Stougaard et al. (1987) and Metz et al. (1988) for sequence comparisons of lb gene promoters. B Effect of internal deletions within the OSE and NE domains. Bars indicate the deleted sequences. Activity from the CAT reporter gene in roots (R) and nodules (N) are shown below, together with GUS activities from the 35S-GUS marker gene. Data from leaf and stem (LS) of whole transformed plants are included where available. CAT activities from the complete lbca5'3'-CAT and the constitutive 35SCAT genes are shown for comparison
355
-246
Binding site 1 -230 ¥
Binding site 2 -170 -223 -176 t
ATT? CTTAAATTATTTATTT, -161
t-< (D i
Ln O .£3
l -230
A E -2~ A G -214 i
&U
-170
-139
b
-
-102
AU
AD
AC
AN
A CAT probe
-1~9 -170
I
1
B
-159
-280
Ubiquitin
A I -2;o -2~o
AJ -3~s -3~2 A M -3;6
probe
-2~0 5
/%E AG hu AI
CATactivity
GUS activity
cpm/pg prot,hrs.
nrnol/IJg prot. hrs.
R 0 0 0 0
N LS 277900 121000 135700 0
/%J
255500 0 157500
/%M
0 117800 0
N 0.24 0.17 0.28 0.36
0.23 0.31
Fig. 2. Internal deletions in the weak positive element (WPE) removing binding sites for the NAT2 transacting factor. Binding site sequences are shown above the WPE element and deleted sequences below as bars. CAT activity from the reporter gene is given for roots (R) and nodules (N), together with GUS activity from the marker gene. Where available, data from leaf and stem (LS) of whole transgenic plants are included
The lbc3 organ-spec~'c and negative elements ( - 139,
-
-
49)
H y b r i d - p r o m o t e r studies using the 35S enhancer to reactivate inactive 5' deletions have previously located an OSE ( - 139, - 102) and an N E ( - 102, - 4 9 ) in the Ibc3 immediate upstream region (Stougard et al. 1987). The OSE containing the putative nodulin consensus motifs 5 ' A A A G A T and 5 ' C T C T T was required in these hybrid constructs for nodule-regulated expression, but was unable to express the lbe3 p r o m o t e r in the absence o f the heterologous enhancer. A qualitative function for the OSE was therefore suggested. The N E a p p e a r e d to suppress expression from reactivated 5' deletions where the OSE was not present (Stougaard et al. 1987). W e have now constructed internal deletions by removing one or b o t h o f these elements. C o m b i n a t i o n o f a 5' and 3' deletion series replaced sequences (shown with bars in Fig. 1) with a J(hoI/SalI hybrid site C C T C G A C C , except for AN, which was constructed by removing a BclI restriction fragment from the 5' region. Activity from C A T reporter gene fusions was measured in transgenic roots and nodules to determine the effect o f removing the specific control elements while leaving the upstream elements intact. Deletion o f the entire (--139, --49) region in AB or most o f the region in AC (--132, --77) results in a low level o f C A T activity in transgenic nodules (Fig. 1) and a reduced level o f steady state C A T m R N A (Fig. 3). C A T activity was not observed in roots
5
15
15
5
lag RNA Iooded
Fig. 3. A Northern analysis of RNA from root nodules of transgenic plants. Total RNA extracted from the AU, AD, AC, and AN lines was separated in agarose gels, transferred to nitrocellulose, and hybridized against CAT coding sequence. Five micrograms of RNA was loaded from the lbc35'3'-CAT control line and from the AU, AN lines. Fifteen micrograms RNA was loaded from the AD and AC lines. The lbc35'3'-CAT line represents the average expression level for the complete gene and was also used as control for CAT activities (Fig. ]). B Control hybridization showing the constitutive expression of the ubiquitin genes. The human ubiquitin cDNA (Wiborg et al. 1985) was used as probe; the hybridization signal corresponds to the trimer or tetramer ubiquitin polyprecursor. or leaves. The OSE and N E together are therefore not solely responsible for organ-specific control o f the lbca gene; other upstream elements also contribute specific control sequences. Deletion of the OSE in AD ( - - 139, - 102), leaving upstream elements and a minimal promoter, not only results in a more than tenfold reduction of C A T activity compared to A E ( - 2 1 4 , -- 139) or to the complete lbc35'3'-CAT gene, but also diminishes the steady state level o f C A T m R N A (Fig. 3). The level o f activity corresponds to the level originally observed after 5' deletion o f the SPE (Stougaard et al. 1987). This was unexpected, as the OSE itself was unable to activate the p r o m o t e r (Stougaard et al. 1987). The OSE is therefore required for efficient transcription initiation, but organ specificity is clearly also controlled by other elements. R e m o v a l o f the N E in A N (--97, --50) reduces the level of C A T activity and steady state m R N A in nodules. Constitutive expression in roots and leaves which could be expected after removal o f a putative repressor function was not detected. One possible explanation for the lowered expression level is the loss of basic p r o m o t e r elements in this deletion.
The lbc3 weak positive element (-230, -170) and the NA T2 binding sites The 5' deletion analysis o f the Ibc3 gene p r o m o t e r delimited a W P E between --230 and - 1 7 0 . Deletion o f this element reduced the expression o f the C A T reporter gene from 2% to below detection (Stougaard et al. 1987). Gel retardation with nuclear extracts from soybean nodules defined two binding sites (1 and 2) overlapping the borders o f the W P E (Jensen et al. 1988; Fig. 2). Both binding sites interact with
356 the same protein factor (NAT2) (Jacobsen et al. 1989), which in soybean is detected only in root nodules. The nodule-specific occurrence of this factor suggests a role for the weak element in mediating organ-regulated expression. Conservation of the sequence of binding site 2 in all soybean leghemoglobin genes and in the leghemoglobin genes from Sesbania rostrata (Metz et al. 1988) further indicates a possible role in general control mechanisms of leghemoglobin genes. To investigate these possibilities, we constructed deletions of the lbe3 binding sites. Complete or partial removal of binding site 2 (AE, AG) or binding site 1 (AI, AJ) did not affect CAT activity significantly (Fig. 2). Deletion of both binding sites in AU also had no dramatic effect on the activity of the receptor gene. When measured as CAT activity, no major effect on expression efficiency or organ specificity was detected deleting the WPE and the NAT2 binding sites. The AU plant lines were also analyzed at the R N A level (Fig. 3). R N A was extracted from nodules of regenerated transgenic plants, and steady state m R N A levels were determined by Northern analysis, using CAT coding sequence as probe. Total R N A from nodules of five different transgenic plants were mixed. The hybridization signal was compared with the control line carrying the complete lbea5'3'-CAT gene also used as control for CAT activities. No significant effect of deleting the NAT2 binding sites was detected at the R N A level. The weak positive element therefore seems to be a dispensable element, and the NAT2 factor is possibly a component of the unfolded chromatin structure near the accessible gene promoter.
(see Stougaard et al. 1987) and suggests that no indispensable regulatory elements are located between the OSE and SPE. Deletions AZ (--665, --102) and ziP ( - 6 6 5 , - 4 9 ) were constructed to investigate the importance of the promoter and immediate upstream elements. Results in Fig. 4 demonstrate that removal of the OSE from the minimal promoter in AZ reduced the CAT activity to a very low level. The OSE present in AX is therefore required for highlevel expression. Direct interaction between the SPE and the TATA box in AP results in a nodule-selective expression of approximately 5% of maximal level, demonstrating the effect of removing the negative element from AZ. Maximal promoter activity therefore appears to require interaction between the SPE and the OSE to overcome the negative regulation. Hybrid-promoter studies using a C a M V 35S ( - - 9 0 ) promoter Analysis of internal deletions and 5' deletions (Stougaard et al. 1987) indicates that no indispensable positive elements 150 bp !
I
o
o
-90, ~ ICAT ~-
