The Plant Cell, Vol. 2, 85-94, January 1990 O 1990 American Society of Plant Physiologists

HMG I-Like Proteins from Leaf and Nodule Nuclei lnteract with Different AT Motifs in Soybean Nodulin Promoters

Karin Jacobsen, Niels Bech Laursen, Erik 0stergaard Jensen,' Anne Marcker, Carsten Poulsen, and Kjeld A. Marcker

Department of Molecular Biology and Plant Physiology, University of Aarhus, C.F. Mdlers AIIe 130, DK-8000 Aarhus C, Denmark

Three different nuclear factors recognizing short AT-rich DNA sequences were identified in different organs of soybean. One factor (NAT2) was found to be present in mature nodules, another factor (NAT1) was detected in roots and nodules, and a third one (LAT1) was only observed in leaves. All three factors recognized several DNA sequences in the promoter region of the soybean nodulin N23 gene. Footprinting, deletion, and point mutation analyses revealed different binding properties for all three factors and further showed that even single base pair substitutions had a dramatic effect on binding affinity. The LATl and NAT1 factors were released from chromatin by extraction with a low-salt buffer and were soluble in 2% trichloroacetic acid, implying a relationship to high- mobility group (HMG) proteins. DNA binding studies futther indicated a functional relationship of these factors to the human HMG I protein. Purification of the LATl factor from leaf nuclei revealed the presence of two polypeptides with molecular masses of 21 kilodaltons and 23 kilodaltons, respectively, binding the same DNA sequence with equal affinity.

INTRODUCTION

The high-mobility group (HMG) proteins are a heteroge- neous group of eukaryotic chromosomal proteins defined on an operational basis. They are proteins extracted from chromatin by 0.35 M NaCl and soluble in 2% trichloroacetic acid (Goodwin, Sanders, and Johns, 1973). Characteriza- tion of the HMG proteins has generally revealed a high content of basic and acidic amino acid residues (approxi- mately 25% of each sidegroup) and a relatively large number of prolines (approximately 7%) (Johns, 1982).

The HMG proteins in mammals can be divided into two subgroups: the high-molecular-mass HMG proteins (HMG 1 and 2) and the low-molecular-mass proteins (HMG 14, 17, Y, and I) (Johns, 1982; Lund et al., 1983). HMG I is analogous to the a-protein isolated from African green monkey (Strauss and Varshavsky, 1984). This protein is present in appreciable concentrations only in nuclei iso- lated from rapidly proliferating cells.

HMG proteins have been isolated from representatives of all eukaryotic kingdoms (for a review see Mayes, 1982). In plants, HMG proteins have been partially characterized in wheat, barley, maize, and Arabidopsis thaliana (Spiker, 1984; Vincentz and Gigot, 1985; Moehs, McElwain, and Spiker, 1988). Amino acid composition, peptide maps, and immunological data suggest a substantial difference be-

' To whom correspondence should be addressed.

tween plant and animal HMG proteins (Spiker and Everett, 1987).

Unlike the other major HMG proteins, which primarily bind single-stranded DNA, HMG I preferentially binds dou- ble-stranded DNA in vitro. Solomon, Strauss, and Var- shavsky (1 986) demonstrated by DNase I footprinting that HMG I binds with an approximately equal affinity to any run of six or more A-T base pairs and to many, if not all, runs of five A-T base pairs. Therefore, it was proposed that HMG I recognizes an aspect of B-DNA conformation, most likely a specific configuration of the minor groove rather than recognizing a specific nucleotide sequence. However, binding of HMG I to bovine interleukin 2 cDNA and to several genes from mouse indicated a more specific recognition sequence for this DNA-binding protein (Reeves et al., 1987; Russnak, Candido, and Astell, 1988).

The function of HMG I has not been elucidated. How- ever, the preferred nucleosome phasing frame detected in a-satellite DNA from African green monkey is precisely bordered by a-protein binding sites, suggesting that the a-protein might function as a nucleosome positioning or phasing protein (Strauss and Varshavsky, 1984). It has also been suggested that HMG I is involved in nuclear scaffold-DNA interactions in vivo (Solomon et al., 1986).

We have recently detected a nuclear factor isolated from soybean nodules that recognizes two specific AT-rich se-

86 The Plant Cell

quences in the promoter region of the leghemoglobin c3(Ibc3) gene (Jensen et al., 1988). In this report we demon-strate that this factor, which now is denoted NAT2, alsorecognizes DMA sequences in the promoter region ofanother nodulin gene, N23. In addition, this region is rec-ognized by two other factors that also bind to AT-rich DNAsequences. One factor is present in roots and nodules(NAT1), and the other is detected in young leaves (LAT1).Point mutation analysis and deletion studies show a sub-stantial difference between the recognition sequences ofthe NAT1 and NAT2 factors, whereas the sequence re-quirements for LAT1 and NAT2 binding are almost identi-cal. Purification and identification of the proteins in theLAT1 complex revealed two homologous polypeptides of21 kD and 23 kD, respectively, that bind the same nucleo-tide sequence with equal affinity.

On the basis of operational criteria and DNA bindingproperties, we propose that NAT1 and LAT1 are HMG-likeproteins with properties analogous to HMG I from mam-malian cells.

RESULTS

An N23 Promoter Fragment Forms Two Complexes withNodule Nuclear Extract (NAT1 and NAT2) and One(LAT1) with Leaf Nuclear Extract

The promoter region of the soybean nodulin N23 gene(Wong and Verma, 1985; Sandal, Bojsen, and Marcker,

1987) was analyzed for binding to leaf and nodule nuclearproteins. Five restriction fragments spanning approxi-mately 1000 bp of the proximal promoter region wereincubated with nodule and leaf nuclear extracts. All frag-ments except one bound proteins from both extracts andexhibited similar complex patterns in the gel retardationassay (data not shown). However, one fragment (fragmentN23-4), spanning the region from —407 to —295, exhibiteda stronger complex formation with leaf and nodule nuclearextracts than the other fragments. Thus, fragment N23-4was selected for further studies. Nuclear proteins preparedfrom leaf, uninfected roots, and nodules at different stagesof development were analyzed for binding to fragmentN23-4. Figure 1 shows that this fragment forms two dif-ferent complexes (NAT1 and NAT2) with nuclear proteinsfrom mature nodules. The complexes migrating slightlyslower than NAT1 and NAT2 might be multiple interactionsof the factors with their respective target DNA sequences(Figure 1). The DNA binding activity responsible for theformation of the upper complex (NAT2) is not present inleaves and only barely detectable in young nodules androots. Sequence requirements and the mobility of thecomplex in a retardation gel resemble the properties of thepreviously described nodule-specific factor (Jensen et al.,1988). Thus, it appears that the NAT2 factor is identical tothis factor. The faster migrating complex (NAT1) is presentin uninfected roots and is only slightly enhanced duringnodule development. Only the LAT1 complex having amobility corresponding to NAT1 could be formed whenusing nuclear extract from leaves. However, when adding

BNODULE LEAF

11 13 15 17 19 31

NAT1-

Daysafterinoculation

-NAT 2

-NAT1 -LAT1

Figure 1. Binding of Soybean Leaf, Root, and Nodule Nuclear Extracts to an N23 Promoter Fragment.

(A) Retardation gel of fragment N23-4 after incubation in the presence of 2 /jg of root nuclear extract.(B) Binding of the same fragment to 2 /jg of nodule nuclear extracts prepared from nodules of various developmental stages.(C) Gel retardation assay of 2 jug (a) and 4 /ig (b) of leaf nuclear extract using fragment N23-4 as a probe. The DNA-protein complexesare denoted NAT1, NAT2, and LAT1.

HMG l-Like Proteins from Soybean 87

increasing amounts of leaf nuclear extract, several com-plexes are formed, probably because of multiple interac-tions of LAT1 with fragment N23-4 (Figure 1, right-mostlane).

Delimitation of Binding Sites on Fragment N23-4

The binding sites on fragment N23-4 were further delimitedby DNase I footprinting as shown in Figure 2. In thepresence of leaf and nodule nuclear extract, two specificsites on the strand corresponding to the upper strand inFigure 3A were protected from DNase I activity (Figure2A). The positions and extent of the protected sequenceswere deduced by comparison to the accompanyingMaxam-Gilbert A-G reaction ladders. Interestingly, thecomplementary strand was only protected from DNase Idigestion by leaf extract in the regions corresponding tothe binding sites on the opposite strand (Figure 2B),whereas nodule extract did not protect any sequences onthat strand (data not shown). The results of the DNase Ifootprinting experiments and the sequences of the pro-tected regions are shown in Figure 3A. The extent of thetwo binding sites was difficult to assess because of theAT-rich nature of the flanking sequences, which are poorlycleaved by DNase I.

To delimit further the sequence sufficient for binding, a5' deletion series of fragment N23-4 was constructed(Figure 3B). Each deletion was tested by gel retardationassay for the formation of NAT1, NAT2, and LAT1 com-plexes, and the result is presented in Figure 3C. The longdeletion (A329) abolished the formation of all three com-plexes, whereas deletion A344 retained the ability to bindLAT1 and NAT2. In contrast, additional sequences wererequired for NAT1 complex formation (A371). However,the binding of NAT1 to A371 is not as strong as the bindingto the full-length fragment N23-4, suggesting that thesequence upstream to —371 affects binding affinity. Thedeletion analysis suggests that LAT1 and NAT1 are differ-ent factors, despite the identical mobilities of the respectivecomplexes in a gel retardation assay. Furthermore, the 5'sequence requirements for LAT1 and NAT2 binding sitesboth appear to be delimited by position -344. Accordingly,a double-stranded oligonucleotide (O344/329) containing thesequence between -344 and -329 on fragment N23-4was synthesized and cloned into the BamHI site of pUC19.The oligonucleotide was excised by EcoRI and Hindllldigestion and analyzed for protein binding. Figure 3Cshows that one nodule complex equivalent to NAT2 andone leaf complex (LAT1) was obtained. In agreement withthe deletion analysis, we did not observe any NAT1 com-plex. Similar results were also observed with anothersynthetic 34-bp oligonucleotide. This oligonucleotide cor-responds to the upstream binding site on fragment N23-4(-407 to -374, data not shown).

B.

i*I

to « _ -

Figure 2. DNase I Protection Assay.

(A) End-labeled upper strand of fragment N23-4 (see Figure 3A)was incubated in the presence of 5 M9 to 10 ng of leaf or nodulenuclear extracts prior to DNase I treatment.(B) Similar footprinting analysis of the complementary strand.A + G and C + T refer to Maxam-Gilbert reactions. Regionsprotected by leaf nuclear extract are marked by open bars andregions protected by nodule nuclear extract are shown by closedbars.

88 The Plant Cell

A.-407

I-371 -3U

I-329

I-295

AAAATTGATATTTTATATAATATATTAAGTCTCTTTAAAATTCTTGTAAAAAAAGACATTTTTAAATAATAAAATAAAGCAACTCTTAATTTTAATGAAACATCCCTTTGTTTTTTAACTATAAAATATATTATATAATTCAGAGAAATTTTAAGAACATTTTTTTCTGTAAAAATT TAT TAT TT TAT T TCGTTGAGAATTAAAAT TACT T TGTAGGGAAACAA

B.A329

A3U

A371

03U/329

c.A329L N

A3UL N

A371L N

03U/329L N

LAT1-* MFJ LAT1*

Figure 3. Delimitation of Binding Sites on Fragment N23-4.

(A) DMA sequence of fragment N23-4. Numbers indicate the distance from the transcription start point. Bars represent the protectedregions defined by DNase I footprinting analysis using leaf extract (open) or nodule extract (closed).(B) Bars represent the DMA sequences of N23-4 present in the respective deletion constructs and the oligonucleotide 0344/329.(C) Gel retardation analysis of the deletion constructs incubated in the presence of leaf (L) or nodule (N) nuclear extracts. The NAT1,NAT2, and LAT1 complexes are indicated.

Single Base Pair Transversion Affects Binding ofLAT1 and NAT2

The sequences of three NAT2 binding sites have beencharacterized previously in the promoter regions of thegenes encoding soybean Ibc3 (Jensen et al., 1988) andSesbania rostrata Srglb2 and Srglb3 (Metz et al., 1988). Acomparison of these sequences implied a core consensussequence of 14 bp -T(T/A)AAT(T/A)(T/A)TTTATTT. Whenthe two NAT2 binding sequences described in this studywere included in the comparison, no common sequencemotif was apparent, apart from the requirement for A-T

base pairs. This suggested a rather degenerated sequencespecificity for protein binding. To study the constraint ofthe recognition sequences for NAT2 and LAT1, a numberof oligonucleotides were synthesized that differed in singlepositions compared with the previously described soybeanIbc3 binding site (binding site 1, Jensen et al., 1988). Eachof the mutant oligonucleotides was cloned into the BamHIsite of pUC19, excised, and end-labeled. The ability ofeach fragment to form complexes was analyzed by gelretardation assay, and the binding affinity was determinedas the ratio between the amount of probe present in thecomplexes and the total amount of probe (Table 1). Some

HMG I-Like Proteins from Soybean 89

Table 1. Binding Affinity of LATl and NAT2 for Mutated Wild- Type Binding Sites

Binding affinity for

Oligonucleotide NAT2 LATl

Wild-type ATATTAATATTTTATTT 100% 100% A4 . . . A . . . . . . . . . . . . . 94% 128% A1 2 . . . . . . . . . . . A . . . . . 86% 100% A1 6 A - 56% 113% T6 .....I........... 35% 73% 0344,329 TCT . . . T . T . A . . . . . . 31% 119% 3‘tr . . . . . . . A . . . . . . 5% 139%

...............

Nodule and leaf nuclear extracts were incubated with labeled DNA fragments containing the sequences listed above. Only nucleo- tides differing from the Ibc3 wild-type sequence are shown. The amount of probe present in the LATl and NAT2 complexes and the unbound DNA fragments was measured by scintillation count- ing. The binding affinity was determined as the ratio between the amount of probe in the complexes and total amount of probe. The binding affinity of the wild-type fragment was set to 100%.

of the single base pair transversions (A4, A12) did not affect the affinity for binding of either LATl or NAT2 to the mutant oligonucleotide. However, one particular A to T transversion (T6) resulted in a significant decrease in bind- ing affinity for both LATl and NAT2. The oligonucleotide derived from the downstream binding site on fragment N23-4 (0344,329) bound much stronger to LATl than to NAT2. This tendency was even more pronounced when using an oligonucleotide truncated at the 3’ end (3’tr). Neither of the factors was capable of binding to single- stranded AT-rich DNA (not shown). Consequently, LATl and NAT2 share a common binding site but they have different sequence requirements for high-affinity binding.

Partia1 Purification and Identification of the Peptides Forming the LATl Complex

Because no activity could be detected in old leaves, crude nuclear extract was prepared from apical expanding young leaves as described in Methods. Heat treatment of the crude nuclear extract at 90°C for 10 min did not abolish the binding activity of LAT1, and this was therefore used as an initial purification step (Figure 4B, lane 1). The heat- treated extract was applied to a column of DEAE-cellulose (DE52), and the flow-through was collected. No detectable activity was bound to the column (compare Figure 48, lanes 1 and 2). Finally, the DE52 flow-through was loaded onto a heparin-Sepharose column. All activity was bound to the heparin-Sepharose and could be eluted by a contin- uous 150 to 1 O00 mM NaCl gradient. Two polypeptides of 21 kD (LATlb) and 23 kD (LATla) co-eluted with the

binding activity (Figure 4A). To demonstrate that these two peptides possess LATl binding activity, DNA hybridization to protein gel blots was performed. The peptides in crude nuclear extract and heparin-Sepharose-purified extract were separated by SDS-PAGE, renatured, and transferred to a nitrocellulose membrane by diffusion. The membrane was incubated in the presence of a 3’P-labeled DNA frag- ment containing the LATl binding site (N23-4). After wash- ing and autoradiography, two distinct bands corresponding to the 21 -kD and 23-kD polypeptides (LATl a and LATl b) were detected in both extracts, as shown in Figure 5.

A 68.0 + 13.0 -m

25.7 +

18.1 -m

1L.3 -b

B. l i I \ \

Figure 4. Purification of LATl .

(A) Silver-stained SDS polyacrylamide gel showing leaf nuclear proteins present in the various stages of purification. LATla and LATl b indicate the two peptides described in the text. Numbers refer to the molecular mass (in kilodaltons) of the marker proteins. (B) Gel retardation assay of the protein fractions shown in (A) using N23-4 as a probe.

HMG l-Like Proteins from Soybean 89

Table 1. Binding Affinity of LAT1 and NAT2 for Mutated Wild-Type Binding Sites

Bindingaffinity for

Oligonucleotide NAT2 LAT1Wild-type A T A T T A A T A T T T T A T T TM A

A19 ... A.

A16 ...............s.re .....T.... .......O344/329 T C T - - - T - T - A - - - - - -

3'tr - - - - - . - A . . - . - . A A G

100%

cco/

OCOA

31%COA

100%•1 OQO/

100%1 1 *^°/

119%1 ̂ Q%,

Nodule and leaf nuclear extracts were incubated with labeled DMAfragments containing the sequences listed above. Only nucleo-tides differing from the Ibc3 wild-type sequence are shown. Theamount of probe present in the LAT1 and NAT2 complexes andthe unbound DNA fragments was measured by scintillation count-ing. The binding affinity was determined as the ratio between theamount of probe in the complexes and total amount of probe.The binding affinity of the wild-type fragment was set to 100%.

of the single base pair transversions (A4, A12) did notaffect the affinity for binding of either LAT1 or NAT2 to themutant oligonucleotide. However, one particular A to Ttransversion (T6) resulted in a significant decrease in bind-ing affinity for both LAT1 and NAT2. The oligonucleotidederived from the downstream binding site on fragmentN23-4 (6344/329) bound much stronger to LAT1 than toNAT2. This tendency was even more pronounced whenusing an oligonucleotide truncated at the 3' end (3'tr).Neither of the factors was capable of binding to single-stranded AT-rich DNA (not shown). Consequently, LAT1and NAT2 share a common binding site but they havedifferent sequence requirements for high-affinity binding.

Partial Purification and Identification of the PeptidesForming the LAT1 Complex

Because no activity could be detected in old leaves, crudenuclear extract was prepared from apical expanding youngleaves as described in Methods. Heat treatment of thecrude nuclear extract at 90°C for 10 min did not abolishthe binding activity of LAT1, and this was therefore usedas an initial purification step (Figure 4B, lane 1). The heat-treated extract was applied to a column of DEAE-cellulose(DE52), and the flow-through was collected. No detectableactivity was bound to the column (compare Figure 4B,lanes 1 and 2). Finally, the DE52 flow-through was loadedonto a heparin-Sepharose column. All activity was boundto the heparin-Sepharose and could be eluted by a contin-uous 150 to 1000 mM NaCI gradient. Two polypeptides of21 kD (LAT1b) and 23 kD (LAT1a) co-eluted with the

binding activity (Figure 4A). To demonstrate that these twopeptides possess LAT1 binding activity, DNA hybridizationto protein gel blots was performed. The peptides in crudenuclear extract and heparin-Sepharose-purified extractwere separated by SDS-PAGE, renatured, and transferredto a nitrocellulose membrane by diffusion. The membranewas incubated in the presence of a 32P-labeled DNA frag-ment containing the LAT1 binding site (N23-4). After wash-ing and autoradiography, two distinct bands correspondingto the 21-kD and 23-kD polypeptides (LAT1a and LAT1b)were detected in both extracts, as shown in Figure 5.

A.68.0-63.0-

25.7-

18.4-

14.3-

B.

*M|

FreeDNA

Figure 4. Purification of LAT1.

(A) Silver-stained SDS polyacrylamide gel showing leaf nuclearproteins present in the various stages of purification. LAT1a andLAT1b indicate the two peptides described in the text. Numbersrefer to the molecular mass (in kilodaltons) of the marker proteins.(B) Gel retardation assay of the protein fractions shown in (A)using N23-4 as a probe.

90 The Plant Cell

B.South-Western

-LAT1a-LAT1b

Figure 5. Identification of the LAT1 Peptides by DNA Hybridiza-tion to Protein Gel Blots.

(A) Silver-stained SDS polyacrylamide gel containing crude nu-clear leaf extract and heparin-Sepharose-purified extract.(B) Autoradiography of the proteins shown in (A) blotted onto anitrocellulose membrane and probed with fragment N23-4. Num-bers refer to the molecular mass (in kilodaltons) of the markerproteins.

extraction and purification procedure was applied to leafand nodule nuclei. Gel retardation assays demonstratedthat two of the binding activities (LAT1, NAT1) were re-leased by 350 mM NaCI and were soluble in 2% trichlo-roacetic acid (TCA) (Figure 6). The LAT1 activity precipi-tated by 25% TCA migrates slower than normally in theretardation gel because of a concentration of the activity.A dilution of this fraction resulted in a normal migrationpattern. These proteins, therefore, meet the operationalcriteria used for defining HMG proteins. The resemblanceto human HMG I was further studied by binding of purifiedHMG I to some of the mutant oligonucleotides describedabove. The Ibc3 wild-type oligonucleotide and the oligo-nucleotide 0344/329 containing the downstream binding siteon fragment N23-4 did bind HMG I (Figure 7). The singlebase pair transversion oligonucleotide mutant (T6), bindingweakly to LAT1 and NAT2, also showed strong binding toHMG I, whereas the point mutation in A16 resulted in low-affinity binding of HMG I. Thus, HMG I recognizes the wild-type LAT1 /NAT2 binding sites, but the optimal binding sitesequence is not completely identical to the recognitionsequences of LAT1 and NAT2.

Binding of HMG I to AT-rich DNA sequences involves aminor groove contact (Solomon et al., 1986). To analyzewhether LAT1, NAT1, and NAT2 interact with the DNA in

LEAF EXTRACT NODULE EXTRACT

^ J\ov-

0\0

Some minor DNA binding peptides were also detected.However, it is likely that these peptides represent proteo-lytic breakdown products of the two major ones. No bind-ing could be detected when using the fragment A329,deleted for the binding site (data not shown). Amino-terminal amino acid sequence determination of the twomajor peptides resulted in two identical 30-residue se-quences. This indicates that LAT1a and LAT1b are veryhomologous or that one is a modified version of the other.

tjjj •*- NAT2H

-LAT1 NAT1-

_ FreeDNA

LAT1 and NAT1 Are HMG l-Like Proteins

The AT-rich binding sequences recognized by LAT1,NAT1, and NAT2 resemble the binding sites for the mam-malian HMG I protein. Furthermore, the ability of the soy-bean proteins to form multiple complexes with DNA inretardation gels was also observed for HMG I-DNA com-plexes (Elton, Nissen, and Reeves, 1987). To investigatea possible relationship to HMG I, the established HMG

Figure 6. Binding Analysis of Proteins from Leaf and NoduleNuclear Extract Precipitated by TCA.

Fragment N23-4 was incubated with crude nuclear extracts andproteins precipitated by 2% TCA and 25% TCA, respectively, andanalyzed by gel retardation assay. The NAT1, NAT2, and LAT1complexes are indicated. nl_AT1 corresponds to multiple interac-tions of LAT1 with the probe.

HMG l-Like Proteins from Soybean 91

roO

}HMGI-DNACOMPLEXES

Figure 7. Binding of Human HMG I to Mutated LAT1 and NAT2Binding Sites.

Pure HMG I protein (100 ng) was incubated with the mutantoligonucleotides described in Table 1 and analyzed by gel retar-dation assay.

the same way, we compared the relative competitionabilities of double-stranded poly(dA-dT), poly(dl-dC), andpoly(dG-dC) in a gel retardation assay. Double-strandedpoly(dG-dC) did not compete for formation of any of thecomplexes, even when present in a 1000-fold excesscompared with the probe. However, a 100-fold excess ofeither double-stranded poly(dA-dT) or double-strandedpoly(dl-dC) was sufficient to out-compete NAT2 binding.Binding of LAT1 and NAT1 was abolished by a 1000-foldexcess of double-stranded poly(dl-dC) and a 100-fold ex-cess of double-stranded poly(dA-dT), respectively (datanot shown). This suggests that the 2-amino group ofguanine in the minor groove, which is absent in inosine,inhibits the binding of the soybean proteins to the DMA.Thus, it appears that minor groove contacts may be in-volved in the binding of all three factors to their respectivesoybean DMA sequences.

DISCUSSION

Organ-Specific DMA-Binding Proteins RecognizeDifferent AT Motifs

Gel retardation assays revealed the presence of at leastthree different nuclear DNA-binding factors in soybean thatrecognize AT-rich DNA sequences. One factor (NAT2) isonly present in appreciable amounts in mature nodules.The binding activity of another factor (NAT1) is slightlyenhanced during nodule development, but is also present

in uninfected roots. A third factor (LAT1) can be detectedin young leaves only. We previously failed to detect thissoybean factor because of the use of older leaves and adifferent nuclear protein preparation method (Jensen et al.,1988). The mobilities of LAT1 and NAT1 complexes areidentical in retardation gels, whereas the electrophoreticmobility of the NAT2 complex is much lower. The NAT2factor is identical to the previously identified nodule-spe-cific nuclear protein interacting with two AT-rich sequencesin the soybean Ibc3 promoter region. Recently, an NAT2binding site was also delimited in the promoter region ofthe S. rostrata Srglb2 and SrglbS leghemoglobin genes(Metzetal., 1988).

All three factors recognize several sequences in thepromoter region of N23, two of which are described in thisreport. Several binding sites were also detected in theupstream promoter region of the Ibc3 gene (data notshown). Binding of leaf and nodule nuclear extract to adeletion series of the promoter fragment N23-4, to thecorresponding oligonucleotide O344/327, and to mutatedoligonucleotides derived from an Ibc3 wild-type binding sitereveals significant differences in sequence requirementsfor the binding of NAT1, NAT2, and LAT1 factors. It wasalso shown that particular base pair substitutions have adramatic effect on binding affinity, whereas other muta-tions did not affect binding. Thus, the three different nu-clear proteins recognize very similar, but distinct, (AT)sequences, and LAT1 and NAT2 seem to be organ-spe-cific, whereas NAT1 is found in both nodules and roots.Recently, different organ-specific proteins were isolatedfrom S. rostrata leaf, root, and nodule nuclei, which alsointeract with an AT oligonucleotide sequence (Metz et al.,1988). An AT-rich DNA sequence from a soybean lectingene has been demonstrated to bind an embryo-specificnuclear protein (Jofuku, Okamuro, and Goldberg, 1987).This protein may be related to the proteins described here,and, therefore, we suggest that organs in legumes andpossibly in other plants as well contain specific AT-bindingnuclear proteins.

Promoter analyses in transgenic Lotus corniculatus haveshown that deletion of the two previously defined bindingsites for NAT2 located in the Ibc3 promoter abolished theresidual expression of the gene (Stougaard et al., 1987).Nodule-specific expression could be restored when fusing35S enhancer sequences to these truncated promoters.Internal deletions, removing both binding sites, did notsignificantly affect expression of the gene (J. Stougaard,unpublished data). However, we have recently detectedadditional NAT2 binding sites further upstream in the Ibc3promoter. These sites might be functional, and, therefore,it is still unclear whether interaction between NAT2 andthe Ibc3 promoter is essential for expression.

The sequences of the two previously determined NAT2sites on the Ibc3 promoter are very homologous to theSAR sequences that are possibly involved in binding DNAto the nuclear scaffold (Gasser and Laemmli, 1986). Prelim-

92 The Plant Cell

inary experiments have demonstrated that the Ibc, pro- moter region binds to the Drosophila nuclear scaffold (J. Falquet, E.@. Jensen, and U.K. Laemmli, unpublished re- sult). Therefore, it is possible that the NAT2 protein is involved in binding DNA to the soybean nuclear scaffold, but this remains to be investigated.

The purification of the LATl factor revealed two peptides of 21 kD and 23 kD binding to the LATl binding site on fragment N23-4, whereas a fragment with the binding site deleted failed to recognize either of the two peptides. The affinity of the two peptides for fragment N23-4 appeared to be almost identical. Therefore, we suggest that the LATl complex observed in retardation gels results from interaction of this sequence with either of the two peptides or from binding to a heterodimer. Amino-terminal sequence determination revealed that the sequence of the first 30 amino acid residues is identical in both peptides, indicating a high degree of homology or partia1 identity of the two proteins. However, it is unclear whether the two proteins are separate gene products or whether they represent different modifications of the same protein.

on this observation and their DNA binding properties, we suggest that LATl and NATl are HMG I-like proteins, although no amino acid sequence similarity between the N-terminal residues of HMG I and LATl could be detected (Lund et al., 1987).

I I t is possible that LATl and NATl are differently modified versions of the same protein. Cell-specific modifications of mammalian DNA-binding proteins have been noted in some cases (Sen and Baltimore, 1986). Although NAT2 hasmany properties in common with NATl and LAT1, NAT2 was never recovered in an active form after treatment with 2% TCA. Consequently, we cannot classify NAT2 as an HMG protein by that criterion. Because of the homology between SAR sequences and NAT2 binding sequences, it is conceivable that NAT2 is involved in the formation of the nuclear scaffold structure.

Therefore, we propose that NATl , NAT2, and LATl are involved in regulation of organ-specific expression most likely by modulating the conformation of chromatin. In this model, the change in chromatin structure involving these proteins is a prerequisite for the organ-specific interaction of transcription factors with the corresponding promoters.

Relationship to the HMG I Protein METHODS

The operational properties of the LATl and NATl factors are reminiscent of the properties of mammalian HMG proteins. Extraction of nuclei with 350 mM NaCl released significant amounts of binding activities, although the sub- sequent 500 mM NaCl extraction released a higher pro- portion of activity. However, 450 mM NaCl extraction was also needed to release HMG proteins from wheat nuclei (Spiker, 1984), suggesting that plant HMG proteins are more tightly bound to chromatin than is the case for HMG proteins in mammals. 60th LATl and NATl are soluble and active in 2% TCA, whereas the NAT2 activity is completely lost upon this treatment.

HMG I also binds to the NAT2 and LATl binding sites present on fragment N23-4 as well as to binding site 1 in the lbc, promoter. Our experiments using mutant oligo- nucleotides suggest that the sequence requirements for binding of HMG I to DNA are even more specific than previously indicated. These sequence requirements con- form more closely with those of LATl compared with the optimal binding sequence of NAT2. Minor groove contact is important for HMG I binding to DNA. Competition ex- periments using double-stranded poly(dA-dT), poly(d1-dC), and poly(dG-dC) revealed that such contact is also impor- tant for binding of NAT2 and to some extent also for binding of NATl and LAT1.

The LATl factor is present in much higher concentration in shoots and young leaves than in old leaves. NATl and NAT2 are detected in the proliferating nodule. Thus, all factors are isolated from organs containing a large propor- tion of proliferating cells. A similar expression pattern has been observed for HMG I (Goodwin et al., 1985). Based

Preparation of Nuclear Extracts

Nodule and root nuclear extracts were prepared as described previously (Jensen et al., 1988). Leaf nuclear extract was prepared from leaves of 3-week-old soybeans. Young leaves (1 O0 g) were harvested and chilled on ice. Tissue was disrupted in 1 L of NIB buffer [10 mM MOPS-NaOH, pH 6.1, 0.25 M sucrose, 10 mM MgCI,, 0.5 mM EDTA, 0.5% Triton X-1 00, 5 mM P-mercaptoeth- anol, and 0.8 mM phenylmethylsulfonyl fluoride (PMSF)] using a Virtus blender mounted with four razor blades. The homogenates were filtered through a layer of Miracloth, and nuclei were pelleted at 20009 for 10 min. The nuclei were washed in 100 mL of NIB and pelleted at 5009 for 1 O min. Finally, the nuclei were washed in NIB buffer without Triton X-100 and resuspended in 5 mL of extraction buffer NXB-200 (200 mM NaCI, 25 mM MOPS-NaOH, pH 6.1, 10 mM MgCI,, 10% glycerol, 0.8 mM PMSF, and 5 mM 0-mercaptoethanol). Following sonication, the nuclear debris and chromatin were sedimented at 70009 for 15 min and resuspended in 5 mL of NXB-800 buffer (like NXB-200, but containing 800 mM NaCI). The mixture was incubated 1 hr to 2 hr at 4OC, and nuclear debris was removed by a 20-min centrifugation at 10,OOOg. Ex- tracts containing nuclear proteins were aliquoted, frozen in liquid nitrogen, and stored at -70°C.

To investigate the relationship to HMG proteins, nuclei were prepared as described above. After sonication, the nuclear pro- teins were extracted in NXB-350 at 4OC for l hr. After centrifu- gation, the supernatants were stored on ice, the nuclear debris and chromatin were resuspended in NXB-500, and the remaining chromatin-attached nuclear proteins were extracted for 1 hr at 4°C. TCA was added to the nuclear extracts to a final concentra- tion of 2%, and, after 1 hr incubation (OOC), the insoluble proteins were pelleted at 10,OOOg for 20 min. The supernatants were

HMG I-Like Proteins from Soybean 93

adjusted to a final 25% TCA and the precipitation procedure was repeated. All TCA pellets were washed with cold acetone, lyoph- ilized, resuspended in NXB-100 buffer, and stored at -70°C.

Partia1 Purification of LATl

The NXB-800 leaf nuclear extract was heated to 90°C for 1 O min and centrifuged for 10 min at 10,OOOg. Four volumes of NXB-O were added to the supernatant and then applied to a DEAE- cellulose (DE52) column equilibrated in NXB-160. The flow- through fraction was collected and loaded onto a heparin-Sepha- rose column (1 mL). After washing the column with 10 mL of NXB-160, proteins were eluted with NXB buffer containing in- creasing NaCl concentrations from 160 mM NaCl to 1000 mM NaCI. Aliquots of the heparin-Sepharose fractions were analyzed on 20% denaturing SDS polyacrylamide gels and assayed for DNA binding activity by gel retardation assay.

Gel Electrophoresis of Proteins

SDS-PAGE was run according to Laemmli (1970) with a resolving gel of 20%. The proteins were stained with silver as described by Morrissey (1 981).

DNA Hybridization to Protein Gel Blots

DNA hybridization to protein gel blots was performed essentially as described by Lelong et al. (1989). The filter was incubated in the presence of 2 x 106 cpm fragment N23-4 in 5 mL of buffer. Poly(d1-dC) was replaced by salmon sperm DNA as nonspecific competitor.

Gel Retardation Assay

Gel retardation assays were carried out as described by Jensen et al. (1988). Sonicated salmon sperm DNA (0.4 pglassay) was included as nonspecific competitor.

DNA Probes for Gel Retardation Studies

The promoter region (-91 5 to +66) of the N23 gene was obtained from plasmid N23-CAT-3’lbc3 (Jargensen et al., 1988). Fragment N23-4 was excised from this plasmid by an Mboll digestion. After repair of the ends of the restriction fragment with T4 DNA polym- erase, the blunt-ended fragment was subcloned into a filled-in BamHl site of pUCl9. To create various deletions of fragment N23-4, the plasmid was cleaved at the unique Sal1 site and treated with Ba131. The deleted ends of the DNA fragments were repaired by using T4 DNA polymerase, and Sal1 linkers were attached with T4 DNA ligase. Plasmids were transformed into fscherichia coli DH1. Blunt-end synthetic oligonucleotides were also cloned into the filled-in BamHl site of pUC19. Fragments to be used for gel retardation assays were isolated from the appropriate plasmids by EcoRI/Sall or EcoRI/Hindlll double digestions. Fragments were end-labeled with 32P-dATP using Klenow polymerase, as de- scribed by Maniatis, Fritsch, and Sambrook (1982).

DNase I Footprinting Experiments

Fragment N23-4 was labeled with 32P using either polynucleotide kinase or Klenow DNA polymerase (Maniatis et al., 1982). The end-labeled fragments were separated from the vector by PAGE and recovered by isotachophoresis (Ofverstedt et al., 1984). Nuclear extracts were incubated with the labeled DNA fragments in the presence of salmon sperm competitor DNA as described for retardation assays. DNase I footprinting was carried out as described by Strauss and Varshavsky (1 984).

ACKNOWLEDGMENTS

We would like to thank Professor Saren G. Laland for a gift of pure human HMG I. This research was supported by the EEC contract BAP-O173 DK, the Danish FTU Programme, the Danish Biotechnology Programme, and the Danish Biomolecular Engi- neering Programme.

Received September 14, 1989; revised October 30, 1989.

REFERENCES

Elton, S.T., Nissen, M.S., and Reeves, R. (1987). Specific A-T DNA sequence binding of RP-HPLC purified HMG-I. Biochem. Biophys. Res. Commun. 143,260-265.

Gasser, S.M., and Laemmli, U.K. (1986). Cohabitation of scaffold binding regions with upstream/enhancer elements of three de- velopmentally regulated genes of D. melanogaster. Cell 46,

Goodwin, G.H., Sanders, C., and Johns, E.W. (1973). A new group of chromatin associated proteins with a high content of acidic and basic amino acids. Eur. J. Biochem. 38, 14-19.

Goodwin, G.H., Cockerill, P.N., Kellam, S., and Wright, C.A. (1 985). Fractionation by high-performance liquid chromatogra- phy of the low-molecular-mass high-mobility-group (HMG) chro- mosomal proteins present in proliferating rat cells and an inves- tigation of the HMG proteins present in virus transformed cells. Eur. J. Biochem. 149,47-51.

Jensen, E.O., Marcker, K.A., Schell, J., and de Bruijn, F.J. (1 988). lnteraction of a nodule specific, trans-acting factor with distinct DNA elements in the soybean leghaemoglobin lbc, 5’ upstream region. EMBO J. 7, 1265-1271.

Jofuku, K.D., Okamuro, J.K., and Goldberg, R.B. (1987). Inter- action of an embryo DNA binding protein with a soybean lectin gene upstream region. Nature 328, 734-737.

Johns, E.W. (1982). History, definitions, and problems. In The HMG Chromosomal Proteins, E.W. Johns, ed (London: Aca- demic Press), pp. 1-7.

Jorgensen, J.-E., Stougaard, J., Marcker, A., and Marcker, K.A. (1988). Root nodule specific gene regulation: Analysis of the soybean nodulin N23 gene promoter in heterologous symbiotic systems. Nucl. Acids Res. 16, 39-50.

521-530.

94 The Plant Cell

Laemmli, U.K. (1 970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,

Lelong, J.-C., Prevost, G., Lee, K., and Crepin, M. (1989). South Western blot mapping: A procedure for simultaneous charac- terization of DNA binding proteins and their specific genomic DNA target sites. Anal. Biochem. 179, 299-303.

Lund, T., Holtlund, J., Frederiksen, M., and Laland, S.G. (1983). On the presence of two new high mobility group-like proteins in HeLa S3 cells. FEBS Lett. 152, 163-167.

Lund, T., Dahl, K.H., M ~ r k , E., Holtlund, J., and Laland, S.G. (1987). The human chromosomal protein HMG I contains two identical palindrome amino acid sequences. Biochem. Biophys. Res. Commun. 146,725-730.

Maniatis, T., Fritsch, E.F., and Sambrook, J. (1 982). Molecular Cloning: A Laboratory Manual. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press).

Mayes, E.L.V. (1 982). Species and tissues specificity. In The HMG Chromosomal Proteins, E.W. Johns, ed (London: Aca- demic Press), pp. 9-40.

Metr, B.A., Welters, P., Hoffmann, H.J., Jensen, E.O., Schell, J., and de Bruijn, F.J. (1988). Primary structure and promoter analysis of leghemoglobin genes of the stem-nodulated tropical legume Sesbania rostrata: Conserved coding sequences, cis- elements and trans-acting factors. MOI. Gen. Genet. 214,

Moehs, C.P., McElwain, E.F., and Spiker, S. (1988). Chromo- soma1 proteins of Arabidopsis thaliana. Plant MOI. Biol. 11,

Morrissey, J.H. (1 981). Silver stain for proteins in polyacrylamide gels: A modified procedure with enhanced uniform sensitivity. Anal. Biochem. 117,307-310.

Ofverstedt, L.G., Hammerstram, K., Balgobin, N., Hjerten, S., Pettersson, U., and Chattopadhyaya, J. (1 984). Rapid and quantitative recovery of DNA fragments from gels by displace- ment electrophoresis (isotachophoresis). Biochim. Biophys. Acta 782, 120-1 26.

680-685.

181-191.

507-51 5.

Reeves, R., Elton, T.S., Nissen, M.S., Lehn, D., and Johnson, K.R. (1 987). Posttranscriptional gene regulation and specific binding of the nonhistone protein HMG-I by the 3’ untranslated region of bovine interleukin 2 cDNA. Proc. Natl. Acad. Sci. USA

Russnak, R.H., Candido, P.M., and Astell, C.R. (1988). Interac- tion of the mouse chromosomal protein HMG-I with the 3’ ends of genes in vitro. J. Biol. Chem. 263, 6392-6399.

Sandal, N.N., Bojsen, K., and Marcker, K.A. (1987). A family of nodule specific genes from soybean. Nucl. Acids Res. 15,

Sen, R., and Baltimore, D. (1986). lnducibility of K immunoglobulin enhancer-binding protein NF-KB by a posttranslational mecha- nism. Cell 47, 921-928.

Solomon, M.J., Strauss, F., and Varshavsky, A. (1986). A mam- malian high mobility group protein recognizes any stretch of six A-T base pairs in duplex DNA. Proc. Natl. Acad. Sci. USA 83,

Spiker, S. (1 984). High-mobility group chromosomal proteins of wheat. J. Biol. Chem. 259, 12007-12013.

Spiker, S., and Everett, K.M. (1987). Blotting index of dissimilar- ity: Use to study immunological relatedness of plant and animal high mobility group (HMG) chromosomal proteins. Plant MOI. Biol. 9, 431-442.

Stougaard, J., Sandal, N.N., Gron, A., Kühle, A., and Marcker, K.A. (1987). 5’ analysis of the soybean leghaemoglobin lbc3 gene: Regulatory elements required for promoter activity and organ specificity. EM60 J. 6, 3565-3569.

Strauss, F., and Varshavsky, A. (1984). A protein binds to a satellite DNA repeat at three specific sites that would be brought into mutual proximity by DNA folding in the nucleosome. Cell,

Vincentz, M., and Gigot, C. (1985). HMG-like protein in barley and corn nuclei. Plant MOI. Biol. 4, 161 -1 68.

Wong, S.L., and Verma, D.P.S. (1985). Promoter analysis of a soybean nuclear gene coding for nodulin-23, a nodule-specific polypeptide involved in symbiosis with Rhizoblum. EM60 J. 4,

84,6531-6535.

1507-1 51 9.

1276-1 280.

37,889-901.

2431 -2438.