Identification of Protein-Protein Contacts between α/β-Type Small, Acid-soluble Spore Proteins of Bacillus Species Bound to DNA*

Small, acid-soluble spore proteins (SASP) of the α/β-type from several Bacillus species were cross-linked into homodimers, heterodimers and homooligomers with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) in the presence of linear plasmid DNA. Significant protein cross-linking was not detected in the absence of DNA. In all four α/β-type SASP examined, the amino donor in the EDC induced amide cross-links was the α-amino group of the protein. However, the carboxylate containing amino acid residues involved in cross-linking varied. In SASP-A and SASP-C ofBacillus megaterium two conserved glutamate residues, which form part of the germination protease recognition sequence, were involved in cross-link formation. In SspC from Bacillus subtilis and Bce1 from Bacillus cereus the acidic residues involved in cross-link formation were not in the protease recognition sequence, but at a site closer to the N terminus of the proteins. These data indicate that, although there are likely to be subtle structural differences between different α/β-type SASP, the N-terminal regions of these proteins are involved in protein-protein interactions while in the DNA bound state.

Between 5 and 10% of the total protein in spores of the Bacillus and Clostridium species of bacteria is ␣/␤-type small, acid-soluble spore protein (␣/␤-type SASP) 1 (1,2). These proteins are encoded by four to seven monocistronic genes in each species, and their amino acid sequences are highly conserved both within and between Bacillus species (1,2). The ␣/␤-type SASP are nonspecific DNA-binding proteins which are synthesized only within the forespore compartment during sporulation (3,4). Typically, two major ␣/␤-type SASP accumulate to high levels within the spore, while the minor ␣/␤-type SASP are found at much lower levels. The level of total ␣/␤-type SASP in spores is sufficient to saturate the spore chromosome, and the binding of these proteins to spore DNA is the major determinant of spore resistance to UV radiation and a significant determinant of spore heat resistance (1,2). Bacillus subtilis spores which lack the two major ␣/␤-type SASP (␣ and ␤) are much more sensitive to UV radiation and heat than are wild type spores (5). During the first few minutes of spore germina-tion, ␣/␤-type SASP are quickly degraded by a sequence-specific protease termed germination protease (GPR) (1,2).
Structural studies of purified ␣/␤-type SASP and ␣/␤-type SASP⅐DNA complexes have shown that significant changes in these proteins' structure occur upon binding to DNA, as ␣/␤type SASP are predominantly unfolded in solution but acquire significant ␣-helical content upon binding to DNA 2 (6). The ␣/␤-type SASP cover 4 -6 base pairs of DNA, and binding of these proteins to DNA is highly cooperative, particularly to DNAs bound with low affinity. (7). Electron micrographs of ␣/␤-type SASP⅐DNA complexes indicate that the protein forms a helical coat along the DNA (8), suggesting that there are extensive interactions between ␣/␤-type SASP when bound to DNA, although these proteins are monomers in solution 3 (9). Consequently, it is possible that interactions between adjacent ␣/␤-type SASP along the DNA backbone may be important for the ␣/␤-type SASP/DNA binding interaction.
To determine which regions of the proteins are involved in interactions between ␣/␤-type SASP bound to DNA, we have performed protein cross-linking studies with the zero-length cross-linking reagent 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC). We have identified EDC-catalyzed protein cross-links in four different ␣/␤-type SASP from Bacillus species, and the identification of these cross-links has yielded new insights into the interaction of ␣/␤-type SASP on DNA.
E. coli strains were routinely grown in 2ϫ YT medium (16 g of tryptone, 10 g of yeast extract, 5 g of NaCl per liter) at 37°C with shaking. For the overexpression of cloned genes encoding ␣/␤-type SASP, the medium was supplemented with 100 g/ml ampicillin (JM107) or 200 g/ml ampicillin and 0.5% glucose (BL21). B. megaterium was sporulated at 30°C in supplemented nutrient broth, and spores were harvested and cleaned as described previously (12).
Polymerase Chain Reaction Amplification and Site-directed Mutagenesis-Oligonucleotides were designed to polymerase chain reaction amplify a 512-base pair fragment containing the gene encoding Bce1 (13) from B. cereus genomic DNA; the amplified fragment contained the gene's ribosome binding site and transcription terminator (13). The upstream primer, BCE1-1 (5Ј-AAAGGATCCTTATTATT-TCATAATTTGTAGC; complementary to nucleotides 119 -140) (13) and downstream primer, BCE1-2 (5Ј-AAAGGATCCTTTTAAGTATGCTTT-TTCCTGC; complementary to nucleotides 592-613) (13), each contained BamHI restriction sites and 5Ј-flanking sequences (underlined residues) for cloning purposes. The BamHI-digested polymerase chain * This work was supported by National Institutes of Health Grant GM19698. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The E10K mutant form of Bce1 was generated with the Transformer TM site-directed mutagenesis kit from CLONTECH according to manufacturer's instructions. Phosphorylated primers complementary, except for designed mismatches (underlined bases), to the unique AlwNI restriction site of pET3 (5Ј-CCTGTTACTAGTGGATGCTGC) and the Gly 6 -Gly 15 coding region of the gene encoding Bce1 (5Ј-GGAAGTCG-TAATAAAGTATTAGTTCGAGGC) (13) were used with plasmid pPS2532 as a template to synthesize a mutagenized plasmid lacking the AlwNI site, and with a lysine codon replacing the codon for glutamate 10 of bce1. The mutagenized plasmid was digested with AlwNI prior to transformation into the mismatch repair deficient E. coli strain (mutS::Tn10, Tet r ) supplied with the mutagenesis kit. Mutagenized plasmid was enriched by plasmid isolation, digestion with AlwNI and retransformation into E. coli strain JM83. One clone was isolated and the identity of the mutagenized plasmid, termed pPS2734, was confirmed by DNA sequencing.
Cross-linking of ␣/␤-Type SASP with EDC-␣/␤-Type SASP (0.5 mg/ml) with or without cesium chloride gradient-purified, EcoRI-linearized pUC19 plasmid DNA (100 g/ml) were incubated in 1 ml of 5 mM sodium phosphate (pH 7.5) at 24°C for 20 min prior to addition of EDC to 25 mM. The 5:1 (w/w) ratio of protein to DNA is sufficient to saturate the DNA with ␣/␤-type SASP, although SASP-A binds more weakly than do the other ␣/␤-type SASP tested (17). Under these conditions, approximately 50% of the ␣/␤-type SASP are bound to the DNA. The cross-linking reactions were incubated for 30 min at 24°C, followed by dialysis in Spectrapor 3 tubing against 1 liter of 10 mM sodium phosphate (pH 7.5) at 4°C for 18 h. Dialyzed cross-linking reactions were frozen, lyophilized, dissolved in sample buffer and run on Tris-Tricine SDS-PAGE (18). Gels were stained with Coomassie Blue, destained, and monomeric and cross-linked proteins were excised with a clean razor blade. Proteins were electroeluted from polyacrylamide gel slices into 50 mM NH 4 HCO 3 , 0.1% SDS using Elutrap TM separation chambers (Schleicher & Schuell). Gel purified proteins were frozen, lyophilized, dissolved in 100 l of MilliQ-H 2 O, and precipitated with 800 l of cold acetone. Precipitated proteins were washed with 500 l of cold acetone and dissolved in freshly prepared 8 M urea prior to trypsin digestion.
Cross-linked Peptide Purification and Analysis-EDC-treated proteins (ϳ20 -40 g) were digested with trypsin (Worthington, 5 g) in 100 l of 0.2 M NH 4 HCO 3 , 10 mM CaCl 2 , 1.2 M urea at 37°C for 15-18 h. Tryptic digests were run on reverse-phase high performance liquid chromatography (HPLC) using a Waters 680 gradient controller, two Waters 501 pumps, a Waters U6K injector, and a Vydac protein C 4 column (3.9 ϫ 150 mm). Tryptic digests were loaded onto the reverse phase column in 100% buffer A (0.06% trifluoroacetic acid) followed by 5 min of washing with 100% buffer A. Peptides were eluted at a flow rate of 1 ml/min with a discontinuous linear gradient as follows: 5-30 min, 0 -30% buffer B (0.052% trifluoroacetic acid in 80% acetonitrile); 30 -50 min, 30 -40% buffer B; 50 -70 min, 40 -100% buffer B. Peptides were detected by their UV absorption at 214 nm with a Waters 481 spectrophotometer, and fractions containing peptides were collected with an Isco 2150 peak separator and an Isco Foxy fraction collector.
Molecular masses of HPLC-purified peptides were determined by matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry using a Perseptive Biosystems Linear MALDI-TOF instrument. External calibration consisting of two standards (angiotensin, 1297.5 Da; and ACTH(18 -39), 2466.7 Da) was used for all determinations, resulting in Ϯ0.15% mass accuracy. Peptides from HPLC fractions (1 l, ϳ1-10 pmol/l) were mixed and dried on the instrument stage with an equal volume of ␣-cyano-4-hydroxycinnamic acid (10 mg/ml) in 50% acetonitrile. Amino acid analysis was conducted as described previously (9). Peptide sequences were determined with an ABI model 492A Procise automated protein sequencer.

RESULTS
EDC Cross-linking of ␣/␤-Type SASP Is DNA-dependent-In an effort to identify the interacting regions of ␣/␤-type SASP bound to DNA, we decided to use protein cross-linking to trap interacting amino acid residues for subsequent biochemical analysis. Most ␣/␤-type SASP do not contain cysteine residues (1) and therefore several protein cross-linking reagents which rely upon thiol chemistry could not be used for this study. The ␣/␤-type SASP are small proteins (6 -7.6 kDa) which are monomeric in solution 3 (9) and appear to interact with one another only when bound to DNA. We were interested in regions of close contact between ␣/␤-type SASP and therefore decided to use cross-linking reagents with short-or zero-length linker arms. EDC, a water-soluble carbodiimide, gave efficient crosslinking of ␣/␤-type SASP only in the presence of DNA (see below). Consequently, we chose this reagent for further work. Protein cross-linking with EDC usually involves the formation of an amide bond between either an N-terminal ␣-amino or lysine ⑀-amino group and the carboxyl side chain of aspartate/ glutamate residues. Therefore, in contrast to cross-linkers that contain flexible linker arms several angstroms in length, EDC induced cross-links should occur only between residues that are in very close proximity to one another.
The proteins chosen for this study were SASP-A and SASP-C from B. megaterium, SspC from B. subtilis, and Bce1 from B. cereus (Fig. 1). Most of the variability between these proteins occurs near the N termini which vary both in length and amino acid sequence ( Fig. 1) (1). All ␣/␤-type SASP lack the N-terminal methionine residue which is the only residue that is removed post-translationally ( Fig. 1) (1). SASP-A and SASP-C are major ␣/␤-type SASP in spores, whereas SspC and Bce1 are minor proteins. For all four proteins little or no protein-protein cross-linking was detected in the absence of added DNA, while significant protein-protein cross-linking was detected in reactions containing ␣/␤-type SASP and DNA (Fig. 2, and data not FIG. 1. Amino acid sequence alignment of some ␣/␤-type SASP. Amino acid sequences are from Setlow (1) and are given in the one-letter code with the amino-terminal residue called residue 1; note that the N-terminal methionine residues are removed post-translationally. The arrow indicates the peptide bond cleaved by the GPR. Trypsin cleavage occurs at peptide bonds following the underlined residues. Amino acid residues above asterisks (*) are completely conserved in all ␣/␤-type SASP identified thus far from Bacillus, Sporosarcina, and "Thermoactinomyces" species.
shown). The extent of protein-protein cross-link formation and the number of higher order oligomers formed varied for each protein tested (Fig. 2). By overloading polyacrylamide gels, decamers could be easily detected in cross-linking reactions with Bce1 and DNA, whereas only small amounts of trimer were detected in reactions with SASP-A and DNA ( Fig. 2 and data not shown). The efficiency of protein cross-linking corresponded roughly to the affinity of each protein for linear plasmid DNA (Bce1 Ͼ SspC Ϸ SASP-C Ͼ SASP-A) as determined by DNase protection assays (17) (data not shown), although the observed cross-linking efficiency of SASP-A was lower than expected.
Different ␣/␤-Type SASP Interact on DNA-The DNA dependence of cross-link formation between ␣/␤-type SASP suggested that the EDC-generated protein-protein cross-links are formed only between ␣/␤-type SASP that are adjacent to one another on the DNA backbone. These in vitro experiments used only a single ␣/␤-type SASP. However, there are multiple ␣/␤type SASP in spores, with two proteins present at high levels. Consequently, an obvious question is whether the different ␣/␤-type SASP interact when bound to DNA. To obtain data pertinent to this question we analyzed protein-protein crosslink formation in reactions with two different ␣/␤-type SASP bound to DNA. SASP-A and SASP-C from B. megaterium were chosen for the initial hetero-cross-linking experiments because they are the two major ␣/␤-type SASP found in spores of B. megaterium (9). These proteins also differ sufficiently in molecular mass (SASP-A ϭ 6,260.1 Da and SASP-C ϭ 7,423.3 Da) to allow resolution of the three possible dimeric forms by Tris-Tricine SDS-PAGE. Electrophoretic analysis of cross-linking reactions containing SASP-A, SASP-C, and DNA revealed the presence of a new predominant band that migrated at the position expected for a SASP-A/SASP-C heterodimer (Fig. 3, lane A ϩ C), and this band is indeed a SASP-A/SASP-C heterodimer (see below). Titration experiments demonstrated that the ratio of SASP-A to SASP-C that produces the most heterodimer is ϳ3:1 (w/w) (data not shown). This latter ratio approximates the relative levels of these two proteins in B. megaterium spores (9). Heterodimers were also formed between SASP-A and SspC from B. subtilis (data not shown). However, in contrast to the SASP-A/SASP-C cross-linking reaction in which the SASP-A/SASP-C heterodimer was the predominant cross-linked product (Fig. 3), SASP-A/SspC het-erodimer formation was much less efficient than formation of the SspC homodimer in these reactions (data not shown). Since SspC and SASP-C have similar affinities for linear plasmid DNA in solution, the difference in their formation of heterodimers with SASP-A is presumably due to differences in the amino acid sequences of SspC and SASP-C.
Identification of Cross-links between ␣/␤-Type SASP-There is presently very little detailed structural information available on ␣/␤-type SASP or the complex they form with DNA. Therefore, identification of the amino acid residues involved in EDCdependent cross-link formation was undertaken to determine which regions of ␣/␤-type SASP are involved in protein-protein interactions that occur in the DNA bound state. Purified monomeric and oligomeric ␣/␤-type SASP from EDC cross-linking reactions were digested with trypsin and the products resolved by reverse phase-HPLC. Two types of differences should be detected between the HPLC tryptic maps of dimeric (or oligomeric) and monomeric (but EDC treated) ␣/␤-type SASP. First, the digests of ␣/␤-type SASP dimers should show decreases (ϳ50%) in the relative yield of some peptide(s) as compared with the monomer, because amino acid residues within this peptide(s) will be in a cross-linked peptide in the dimer. Second, there should be a new peptide peak(s) in HPLC tryptic maps of ␣/␤-type SASP dimers, which should be the peptide containing the cross-link. Detailed analyses, including mass spectrometry, amino acid analysis and amino acid sequencing of the latter peptides should then allow both the unambiguous identification of the peptides in the cross-link, as well as the specific amino acid residues involved. Intramolecular crosslinks could also be formed by EDC, as ␣/␤-type SASP go from an unfolded to a more ordered structure on binding to DNA. Intramolecular cross-links could be found within both monomeric and oligomeric proteins, and this modification could be detected by comparing HPLC tryptic maps of EDC treated monomers and untreated protein. However, we never saw evidence for intramolecular cross-link formation in these analyses (data not shown).
HPLC analysis identified two unique, closely eluting peptides in the tryptic digest of dimeric SASP-A (Fig. 4B, peptides  labeled 1 and 2) which were not present in the digest of the SASP-A monomer (Fig. 4A). A substantial reduction in the amount of one peptide was also noted in the digest of dimeric SASP-A when compared with that of monomeric SASP-A (Fig.  4B, peptide labeled 3). No other significant differences were observed between digests of the monomeric and dimeric species. Because the relative amounts of all other peptides appeared to be approximately the same between digests of monomeric and dimeric SASP-A, these data suggested that the cross-link occurred between an amino acid residue in peptide 3 and a residue within a small peptide which has very little UV absorbance. The tryptic digests of the SASP-C monomer and dimer exhibited differences that were very similar to those seen with SASP-A (data not shown). HPLC analysis of the tryptic digest of the SASP-A/SASP-C heterodimer also identified two unique peptides which were not present in digests of SASP-A or SASP-C monomers (data not shown). Both of these unique peptides from the SASP-A/SASP-C heterodimer had HPLC retention times that differed from those of the putative crosslinked peptides identified from the SASP-A and SASP-C homodimers (data not shown).
Only two additional significant tryptic peptides were detected in the Bce1 dimer that were not present in the Bce1 monomer (Fig. 5, A and B). One of these peptides (Fig. 5B, peptide labeled with an asterisk) was an oxidized form of Bce1 tryptic peptide Lys 55 -Arg 66 which contained a methionine sulfoxide residue (data not shown). The other unique peptide, presumably the cross-linked peptide, eluted early in the HPLC gradient (Fig. 5B, peptide labeled 1). No obvious reduction in the level of any major peptide peak was observed when the HPLC profile of the tryptic digest of monomeric Bce1 was compared with that of dimeric Bce1, suggesting that the crosslink occurred between amino acid residues from two small tryptic peptides. In contrast to SASP-A, SASP-C, and Bce1, analysis of the tryptic digest of the SspC dimer identified only one unique peptide in comparison to the digest of the SspC monomer (data not shown). However, as was found with SASP-A and SASP-C, the amount of one major peptide was decreased significantly in the digest of the SspC dimer as compared with the digest of the monomer (data not shown). Presumably this large peptide is involved in cross-link formation with a rather small peptide.
The relatively high efficiency of SspC and Bce1 cross-linking ( Fig. 2) also allowed the purification and analysis of crosslinked trimeric and tetrameric species of these proteins. The HPLC profiles of the tryptic digests of the trimeric and tetrameric species of both SspC and Bce1 were essentially identical to the tryptic map of the dimeric forms, with the exception of greater reductions in the larger peptide partner in the crosslink in the higher oligomers of SspC (data not shown). These data suggest that identical EDC catalyzed cross-links occur between each protein in higher oligomers of cross-linked ␣/␤type SASP.
Elucidation of Amino Acid Residues Involved in Cross-link Formation-Various types of information were used to determine the amino acid residues involved in cross-link formation in the different ␣/␤-type SASP. For SASP-A, SASP-C, and SspC, mass spectrometry and amino acid analysis identified the large peptides whose level was decreased in tryptic digests of the dimeric species as Tyr 20 -Arg 37 , Phe 29 -Arg 46 , and Ser 8 -Lys 27 , respectively. This identified one probable partner in the major cross-link formed in these three proteins. Determination of the mass of each peptide tentatively identified as a crosslinked species from tryptic digests of both homo-and heterodimers (Table I), as well as amino acid analyses (data not shown) allowed assignment of the two tryptic peptides in the various cross-links. In all cases, the site of cross-linking was tentatively identified as between the ␣-amino group of the protein and an acidic group on a separate tryptic peptide. The peptides in the two new peaks from tryptic digests of crosslinked dimers of either SASP-A or SASP-C had virtually iden-  5. A and B, HPLC analysis of cross-linked Bce1. Polyacrylamide gel-purified monomeric and dimeric Bce1 were digested with trypsin and peptides resolved by reverse-phase HPLC as described under "Experimental Procedures." A, tryptic digest of EDC-treated monomeric Bce1. B, tryptic digest of EDC-treated dimeric Bce1. The cross-linked peptide is labeled 1. The peak labeled with an asterisk (*) is peptide L55-R66 containing an oxidized methionine. tical observed molecular masses (Table I), suggesting that each peak contained the same two tryptic peptides linked together, but cross-linked at a different site. Analyses of the two unique peptides isolated from the tryptic digest of the SASP-A/SASP-C heterodimer predicted that the cross-links were between the ␣-amino group of one protein and a tryptic peptide in the other protein (Table I).
Amino acid sequence analysis definitively identified the cross-linked peptides from the four ␣/␤-type SASP as well as the SASP-A/SASP-C heterodimer (Table II). Only one amino acid sequence was obtained from the cross-linked peptides, consistent with the involvement of the ␣-amino group of each protein in these cross-links. Sequencing of the two cross-linked peptides from SASP-A confirmed that one of the peptides involved in the cross-link was indeed Tyr 20 -Arg 37 . The differences in molecular mass and amino acid composition between the cross-linked peptides and Tyr 20 -Arg 37 were consistent with the cross-linking of Tyr 20 -Arg 37 to Ala 1 -Lys 5 of SASP-A (Table  I and data not shown). Peptide 1 from the SASP-A dimer had a reduced yield of glutamate in cycle 6, while peptide 2 had a dramatically reduced yield of glutamate in cycle 2 (Table II). Thus, both of these glutamate residues are involved in crosslinks to the amino terminus of the protein, and reduced yields of these glutamate residues were observed instead of blank cycles, because the two cross-linked peptides were not completely separated from one another by HPLC (Fig. 2B). A similar result was obtained in sequencing unique peptide 2 from the SASP-C dimer, as a reduced yield of glutamate was found in cycle 2 (Table II). We presume that unique peptide 1 from SASP-C dimer, which had the same mass and amino acid composition as peptide 2, contains a cross-link to the glutamate residue at position 6 in Phe 29 -Arg 46 . Sequencing of the two unique peptide peaks from the SASP-A/SASP-C heterodimer revealed cross-links at the same residues as in the SASP-A and SASP-C homodimers (Table II). However, the decreases in glutamate yields were not as pronounced as with the cross-linked peptides from the SASP-A and SASP-C homodimers, indicating that each of the unique HPLC peaks from the heterodimers were mixtures of two different cross-linked peptides (Table II).
As was found with the cross-linked peptide from SASP-A and SASP-C, only a single amino acid sequence was obtained with the cross-linked peptide from SspC and Bce1 (Table II). This is again consistent with the involvement of the ␣-amino group of the latter two proteins in cross-link formation. However, the sequencing results indicated that the glutamate residues involved in cross-link formation in SASP-A and SASP-C are not involved in formation of the major cross-link found in SspC and Bce1. Instead, blank sequencing cycles were obtained at positions corresponding to aspartate 13 of SspC and glutamate 10 of Bce1 (Table II and Fig. 1). These residues are at identical positions in ␣/␤-type SASP amino acid sequence alignments (Fig. 1). However, the residues at this position are not as highly conserved as are other residues in ␣/␤-type SASP (1).
Generation and Analysis of Bce1 E10K -The acidic residues identified as sites of cross-link formation in Bce1 and SspC are not highly conserved among ␣/␤-type SASP (1), and therefore their involvement in EDC cross-linking was somewhat unexpected. Consequently, we decided to investigate the role these specific acidic residues play in DNA binding and to determine whether in Bce1 the observed cross-link between the ␣-amino group and glutamate 10 was the only one generated by EDC. A site-directed mutant form of Bce1 was generated in which glutamate 10 was changed to a lysine residue. This change should not destroy Bce1/DNA binding, because most other ␣/␤type SASP contain either a lysine (e.g. SASP-A and SASP-C) or a glutamine residue at this position (1). This type of mutagenesis was not conducted with SASP-A or SASP-C, because previous work has shown that changes in the glutamate residues involved in cross-link formation in these proteins significantly diminish DNA binding (19).
Purified Bce1 E10K bound to DNA and gave approximately twice the protection against DNase I to linear plasmid as did equivalent amounts of Bce1 (data not shown). Bce1 E10K also displayed DNA-dependent EDC cross-linking, but with much lower efficiency (Ͻ20% based on Coomassie staining) than wild type Bce1 (Fig. 6). The large difference in cross-linking efficiency between Bce1 and Bce1 E10K indicates that glutamate 10 is the major site of EDC cross-link formation in Bce1, but that there is at least one additional cross-linking site. Therefore, cross-linked dimeric Bce1 E10K was purified and analyzed to identify the residues involved in cross-linking. HPLC analysis of a tryptic digest of the Bce1 E10K dimer revealed two unique peptides when compared with the digest of monomeric Bce1 E10K (data not shown). One of these peptides contained an EDC-induced modification but not a cross-link, while the other peptide was identified as a cross-linked peptide by mass spectrometry and amino acid analysis. The cross-link appeared to be between the ␣-amino group of the protein and one of the two conserved glutamate residues which participate in the crosslinks in SASP-A and SASP-C (data not shown).

DISCUSSION
Previous studies have demonstrated that ␣/␤-type SASP bind to DNA in a cooperative fashion and that protein-protein interactions probably occur during binding (7). The data reported in this communication confirm that direct protein-protein interactions occur between ␣/␤-type SASP bound to DNA. The protein-protein interaction detected by EDC cross-linking is not seen in the absence of DNA, which is consistent with previous data that suggest that ␣/␤-type SASP exist as monomers in solution 3 (9). It appears that ␣/␤-type SASP/DNA binding involves the polymerization of the monomeric proteins along the backbone of the DNA double helix. This is suggested both by direct visualization of ␣/␤-type SASP⅐DNA complexes  (1) and (2) according to the order of their elution during reverse-phase HPLC (Figs. 4B and 5B, and data not shown).
by electron microscopy (8) and by the ability of ␣/␤-type SASP to protect the DNA backbone from attack by hydroxyl radicals and orthophenanthroline-Cu 2ϩ (17). It further appears that ␣/␤-type SASP assemble in a polarized fashion along the DNA backbone, because EDC cross-linked trimers and tetramers of SspC and Bce1 appear to contain only one type of cross-link, which is consistent with a head-to-tail arrangement. However, it is not clear whether DNA polarity itself has any role in organizing ␣/␤-type SASP binding.
The DNA binding region of ␣/␤-type SASP has been postulated to be within the C-terminal half of these proteins. This region of ␣/␤-type SASP is more highly conserved than the N-terminal region, and a 29-amino acid residue synthetic peptide corresponding to residues Thr 43 -Phe 70 of SspC has been shown to bind to DNA and change the DNA's UV photochemistry (6). However, the binding of this synthetic peptide to DNA was much weaker than that of full-length SspC (6). The EDC cross-link sites found in ␣/␤-type SASP suggest that the Nterminal region of these proteins is involved in protein-protein interactions. These protein-protein interactions presumably increase the binding affinity of ␣/␤-type SASP for DNA, and therefore account, at least in part, for the difference in DNA binding affinities between the synthetic peptide and full-length SspC. The N-terminal region of ␣/␤-type SASP is the most variable region of these proteins, with variations in both sequence and length ( Fig. 1) (1). In fact, many ␣/␤-type SASP are virtually identical proteins with the exception of their N-terminal regions. This is best exemplified by SASP-A and SASP-C of B. megaterium, which are essentially identical with the exception of the longer N terminus of SASP-C (Fig. 1). However, despite their similarity in primary sequence, SASP-C has a higher affinity for DNA in solution (9). In light of the results reported here, it appears that the N-terminal regions of ␣/␤type SASP are involved in protein-protein interactions, and therefore could be a major factor in determining the binding affinity of these proteins for DNA in solution. It is important to note that other regions of ␣/␤-type SASP may also be involved in protein-protein interactions, which may not be detected because the reagent used in this study cross-links only amino and carboxyl groups.
The identification of cross-linked acidic residues within the GPR recognition sequences of SASP-A and SASP-C is particularly interesting because previous in vitro studies have demonstrated that ␣/␤-type SASP are resistant to GPR cleavage when bound to DNA (20). The data presented in this communication suggest that the N termini of SASP-A and SASP-C are close to the GPR cleavage site, while these proteins are in the DNA bound state. Thus, the GPR cleavage site may be inaccessible to the GPR protease due to steric interference by the N terminus of an adjacent protein. However, other structural changes are probably also important because purified cross-linked dimeric and trimeric SspC (which both contain unmodified GPR cleavage sequences) are partially resistant to GPR cleavage (data not shown). Thus, EDC cross-linking may stabilize a protein conformation of ␣/␤-type SASP in which GPR cleavage is inhibited.
The identification of two different sites of DNA dependent cross-link formation in the ␣/␤-type SASP examined was unexpected based upon the large degree of primary sequence conservation between members of this protein family. Although the position corresponding to Asp-13 and Glu-10 in SspC and Bce1, respectively, is far removed from the GPR recognition site in primary sequence, it is possible that these two regions are near one another in the three dimensional structure of ␣/␤-type SASP bound to DNA. Because SASP-A and SASP-C contain a lysine residue at the position corresponding to Asp 13 -Glu 10 , these proteins are unable to form cross-links with the ␣-amino group at this position and instead form cross-links with the glutamate residues of the GPR recognition sequence. The observed cross-links between the ␣-amino group and the two glutamate residues of the GPR recognition sequence in SASP-A and SASP-C, indicate that the N terminus of each protein is fairly mobile and may interact electrostatically with an acidic patch that is formed by the glutamate residues of the GPR recognition sequence. This acidic patch may also contain Asp 13 and Glu 10 in SspC and Bce1, respectively, and could explain the apparent shift in the cross-link formation site from Glu 10 in Bce1 to the GPR recognition site in Bce1 E10K . Although, we believe that we have identified the major sites of EDC cross-linking in each ␣/␤-type SASP studied, we cannot of course exclude the possibility that other minor cross-linking sites exist which were not detected by our HPLC analysis.
Another property of ␣/␤-type SASP that has been established in this study is that different ␣/␤-type SASP make functional protein-protein contacts with one another in the DNA bound state. Indeed, proteins from different species were found to interact as demonstrated by cross-linking of SspC from B. subtilis to SASP-A from B. megaterium. However, it appears that the interaction between SspC and SASP-A is not as favorable as the interaction between SASP-C and SASP-A. In fact, heterodimers of SASP-A and SASP-C were the predominant cross-linking products when the ratio of the two proteins approximated the in vivo ratio. Preferential heterodimer formation may be due to preferential association of the two proteins while bound to DNA, or merely to a greater probability of successful cross-linking occurring between these two particular proteins. Because SASP-A and SASP-C had identical EDC cross-linking sites, which were different from those of SspC and Bce1, it is reasonable to assume that the precise nature of the protein-protein interactions in these two groups of ␣/␤-type SASP is similar, yet distinct from one another. Thus, subtle variations in protein structure may allow some ␣/␤-type SASP to interact with one another more easily than others. These apparently minor differences in primary sequence between ␣/␤type SASP may be important for efficient binding to different regions of the spore chromosome, and account for the need to maintain multiple ␣/␤-type SASP in each species.
We are currently using another cross-linking strategy to identify amino acid residues in ␣/␤-type SASP which make close contacts with DNA. However, the structural information obtained during these types of studies is limited and efforts to obtain a high resolution structure of an ␣/␤-type SASP⅐DNA complex are ongoing. A high resolution structure of an ␣/␤-type SASP⅐DNA complex should confirm the results obtained in cross-linking experiments and also illustrate the nature of the change in DNA conformation which underlies the change in UV photochemistry of spore DNA and ultimately spore UV resistance.