N-terminal Amino Acid Residues Mediate Protein-Protein Interactions between DNA-bound α/β-Type Small, Acid-soluble Spore Proteins fromBacillusSpecies

Abstract The binding of α/β-type small, acid-soluble spore proteins (SASP) to DNA of spores of Bacillus species is the major determinant of DNA resistance to a variety of damaging treatments. The primary sequence of α/β-type SASP is highly conserved; however, the N-terminal third of these proteins is less well conserved than the C-terminal two-thirds. To determine the functional importance of residues in the N-terminal region of α/β-type SASP, variants of SspC (a minor α/β-type SASP from Bacillus subtilis) with modified N termini were generated and their structural and DNA binding properties studied in vitro andin vivo. SspC variants with deletions of up to 14 residues (∼20% of SspC residues) were able to bind DNA in vitroand adopted similar conformations when bound to DNA, as determined by circular dichroism spectroscopy and protein-protein cross-linking. Progressive deletion of up to 11 N-terminal residues resulted in proteins with progressively lower DNA binding affinity. However, SspCΔ 14 (in which 14 N-terminal residues have been deleted) showed significantly higher affinity for DNA than the larger proteins, SspCΔ 10 and SspCΔ 11. The affinity of these proteins for DNA was shown to be largely dependent upon the charge of the first few N-terminal residues. These results are interpreted in the context of a model for DNA-dependent α/β-type SASP protein-protein interaction involving the N-terminal regions of these proteins.

The binding of ␣/␤-type small, acid-soluble spore proteins (SASP) 1 to DNA of spores of Bacillus species is the primary mechanism by which spore DNA is protected from the damaging effects of heat, peroxides, and UV radiation (1,2). The ␣/␤-type SASP are nonspecific DNA-binding proteins, whose synthesis is developmentally regulated such that these proteins are only synthesized within the developing forespore during sporulation (3,4). The ␣/␤-type SASP accumulate to very high levels, which are sufficient to saturate the spore chromosome, and DNA within this nucleoprotein complex is protected from a variety of environmental insults (1,2,4). Accordingly, the ␣/␤-type SASP are important determinants of long term spore survival, and spore resistance to heat, peroxides, and UV radiation. Spores of Bacillus subtilis that lack the majority of their ␣/␤-type SASP, termed ␣ Ϫ ␤ Ϫ spores, are much more sensitive to these treatments than are wild-type spores (1,2). The ␣/␤-type SASP are encoded by a number (4 -7) of monocistronic genes in Bacillus species. Generally, in each species there are two ␣/␤-type SASP that are expressed at very high levels (the major ␣/␤-type SASP), and a variable number of other ␣/␤-type SASP expressed at much lower levels (the minor ␣/␤-type SASP) (4). The amino acid sequences of these proteins are highly conserved, both within and between species; however, significant differences in DNA binding affinity between ␣/␤-type SASP have been demonstrated (5)(6)(7). During spore germination, the ␣/␤-type SASP are rapidly degraded by a sequence-specific endoproteinase, termed the germination protease (GPR), which recognizes and cleaves within a pentapeptide sequence found within all ␣/␤-type SASP (4).
The interaction between ␣/␤-type SASP and DNA has been studied in detail, and several features of ␣/␤-type SASP-DNA binding have been characterized (5, 6, 8 -10). DNA structure changes from B-DNA to an A-like conformation upon binding to ␣/␤-type SASP (7,8), whereas ␣/␤-type SASP undergo a transition from random coil to a largely ␣-helical conformation upon binding to DNA (6). The binding interaction is significantly cooperative, with ␣/␤-type SASP having a 50-to ϳ600-fold greater affinity for contiguous DNA binding sites than for noncontiguous sites depending upon the bound polynucleotide and the salt concentration (5,6). The binding cooperativity is thought to be due at least in part to protein-protein interactions between adjacent, DNA-bound ␣/␤-type SASP (8,11). Contacts between the ␣-amino group and the carboxylate side chains of three acidic residues found within the N-terminal 40 -50% of a variety of ␣/␤-type SASP have been identified previously using a zero-length cross-linking reagent (11). The N-terminal third of ␣/␤-type SASP varies both in length and amino acid sequence, and its sequence is generally less well conserved than the C-terminal two-thirds of these proteins ( Fig. 1) (4). Because individual ␣/␤-type SASP have significantly different affinities for DNA, it seems reasonable to speculate that these differences in binding affinity could be due to differences in the length and sequence of the N-terminal third of these proteins. Consequently, we have investigated the effects of modifications of the N-terminal sequence upon the DNA binding properties of ␣/␤-type SASP. This analysis has further led to a model for ␣/␤-type SASP protein-protein interaction, which involves an electrostatic interaction between the positively charged N-terminal region of one protein and an acidic region on an adjacent, DNA-bound protein.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-The Escherichia coli strains used were JM83 (ara ⌬(lac-proAB) rpsL 80 lacZ⌬M15) (12) and BL21(DE3) (T7 RNA polymerase under control of the lac promoter) (13). The B. subtilis strains used were all derivatives of strain 168. E. coli strains were routinely grown in Terrific broth (24 g of yeast extract, 12 g of tryptone, and 4 ml of glycerol per 900 ml, plus 100 ml of 170 mM KH 2 PO 4 , 720 mM K 2 HPO 4 ) at 37°C with shaking. For the overexpression of cloned genes encoding ␣/␤-type SASP from pET11d vectors (13), the medium was supplemented with 200 g/ml ampicillin and 0.5% glucose. B. subtilis was transformed to kanamycin resistance with pUB110 derivatives as described previously (14).
Construction of sspB Promoter-sspC N-terminal Deletion Variant Gene Fusions-To achieve high level expression of SspC variants in spores of B. subtilis, the genes encoding the variants were fused to the strong, sporulation-specific promoter of the sspB gene, which is expressed at the same time in sporulation as sspC (16). An NcoI restriction site was first introduced at the initiating methionine codon of the sspB gene by the PCR megaprimer method (17). The first round of PCR used the upstream primer SSPB1 (5Ј-ACGGCTAAGCTTTTTTTATT-TCTC), complementary to nt 173-196 of the sspB sequence (15), and the downstream primer SSPB-NCOI (5Ј-GAGTTTTGGTTAGCCATGGG-TAAAATCTCC), complementary to nt 357-386 of the sspB sequence (15) to generate a DNA fragment containing the sspB promoter with a site-directed mutation (underlined residue) creating the NcoI site. This PCR product was agarose gel-purified and used as the upstream primer in a second round of PCR with the downstream primer SSPB2 (5Ј-GGATCCCTTTTTTTCTAGGATATGTGGAGCAGG), complementary to nt 639 -665 of the sspB sequence (15) to generate a complete sspB gene with an added 3Ј BamHI site (underlined residues). This fragment was ligated into SmaI-digested pUC19 to generate plasmid pPS2798. The identity of the insert in plasmid pPS2798 was confirmed by DNA sequencing. NcoI-BamHI fragments of all sspC variants were ligated to NcoI-BamHI-digested plasmid pPS2798 generating the sspB promoter-sspC gene fusions.
To introduce the sspB promoter-sspC gene fusions into B. subtilis, these fragments were first subcloned into a pBluescript-pUB110 fusion plasmid (pPS2952). Plasmid pPS2952 was constructed by ligating a pBluescript (Stratagene, La Jolla, CA) derivative (a SmaI-HincII fragment was deleted) to a pUB110 derivative (plasmid pUB-B, containing the sspB gene at the unique EcoRI site (10)) at their BamHI sites. The sspB promoter-sspC-containing pPS2798 plasmids were first digested with BamHI, the ends filled with the large fragment of E. coli DNA polymerase I, and then digested with HindIII to remove HindIII-endfilled BamHI fragments containing the sspB promoter-sspC gene fusions. These fragments were ligated to plasmid pPS2952 that had been digested with HindIII and HpaI. Finally, the plasmid pPS2952 clones containing the sspB promoter-sspC gene fusions were digested with BamHI to remove pBluescript DNA, the pUB110 derivative fragments agarose gel-purified, religated, and used to transform B. subtilis strain PS356 (␣ Ϫ ␤ Ϫ ) (18) to kanamycin resistance as described (14).
Purification of ␣/␤-type SASP and Polynucleotides-All SspC vari-ants were overexpressed in E. coli strain BL21(DE3) from pET11dderived plasmids as described previously (13). SspC and SspC ⌬11-D13K were extracted from dry ruptured E. coli cells with 3% acetic acid, 30 mM HCl and purified as described previously (19). SspC ⌬10 and SspC ⌬11 were acid-extracted as described above, but were purified by ion exchange chromatography on a QAE-cellulose column equilibrated in 10 mM Tris-HCl (pH 8.0) at 4°C. Proteins were eluted by a linear salt gradient from 0 to 200 mM NaCl in 10 mM Tris-HCl (pH 8.0) at 4°C. SspC ⌬5 , SspC ⌬11-D13N , and SspC ⌬14 were acid-extracted as outlined above, but were purified by ion exchange chromatography on a CMcellulose column equilibrated in 10 mM NaOAc (pH 5.5) at 4°C. Proteins were eluted by a linear salt gradient from 0 to 200 mM NaCl in 10 mM NaOAc (pH 5.5) at 4°C. Column fractions containing purified ␣/␤-type SASP were pooled, concentrated by lyophilization, and dialyzed exhaustively against 10 mM sodium phosphate (pH 7.5). All proteins were Ͼ95% pure, as judged by SDS-PAGE and staining with Coomassie Blue. The concentrations of protein stock solutions were determined by quantitative amino acid analysis. Plasmid DNA (pUC19) was purified by two rounds of CsCl equilibrium density centrifugation, and linearized by digestion with EcoRI. Poly(dG)⅐poly(dC) and poly(dA-dT)⅐poly(dA-dT) were obtained commercially (Sigma). All polynucleotides were dialyzed exhaustively in Spectra/Por 3 tubing (molecular mass cut-off 3500 Da) against 10 mM sodium phosphate (pH 7.5).
Analysis of SspC Variants in Vitro and in Vivo-DNA binding was assessed in vitro by DNase I protection assays (5) and by circular dichroism (CD) spectroscopy (6). DNA binding of purified ␣/␤-type SASP was assessed by the ability of protein to protect EcoRI-linearized pUC19 plasmid from DNase I digestion as described previously (5,19). All CD measurements and spectra were obtained on a Jasco 715 spectropolarimeter with a Jasco PS-150-5 power supply, and all recorded spectra were the average of three scans (6). Far UV (protein conformation) spectra were obtained from solutions containing 25 M ␣/␤-type SASP and 115 M (in bp) poly(dG)⅐poly(dC) in 10 mM sodium phosphate (pH 7.5) at 21°C. Difference spectra corresponding to the ␣/␤-type SASP component of the complex were obtained by subtracting the spectrum of free DNA from the spectra of ␣/␤-type SASPpoly(dG)⅐poly(dC) complexes as described (6,23,24). Secondary structure deconvolution was carried out on data from difference spectra using a webbased neural network algorithm, K2D (25).
Thermal denaturation of SspC variant-DNA complexes was performed on pre-equilibrated, stirring solutions containing 5 M SspC variant complexed with 23 M (in bp) poly(dG)⅐poly(dC) in 10 mM sodium phosphate (pH 7.5) (6). This amount of DNA is sufficient to bind all wild-type SspC (6). Denaturation was monitored by measuring ellipticity at 222 nm at 0.5°C intervals as a function of temperature from 20°C to 90°C. The midpoint of each transition (defined as T M ) was determined by taking the first derivative of ellipticity at 222 nm with respect to the inverse of the absolute temperature as described (26). Experimental error was estimated at Ϯ1°C, based on duplicate measurements of poly(dG)⅐poly(dC) complexes with SspC, SspC ⌬5 , and SspC ⌬11 .
Equilibrium binding titrations by CD spectroscopy were performed as described using EcoRI-linearized pUC19 plasmid, poly(dG)⅐poly(dC), and poly(dA-dT)⅐poly(dA-dT) (6). A DNA binding site size of 4 bp was directly determined for SspC, SspC ⌬5 , SspC ⌬11-D13N , and SspC ⌬11-D13K from stoichiometric forward titrations of poly(dG)⅐poly(dC) in 10 mM sodium phosphate (pH 7.5) at 21°C, as was found previously for a number of ␣/␤-type SASP with several DNAs (5, 6). Stoichiometric DNA binding conditions could not be obtained for SspC ⌬10 , SspC ⌬11 , Asp-N fragment, and SspC ⌬14 , although data from forward titrations of poly(dG)⅐poly(dC) with these proteins under no salt conditions were entirely consistent with the same 4-bp site size. Mean residue ellipticity, [] 222 , values for DNA-bound protein were determined from complexes containing 25 M protein and 115 M (in bp) poly(dG)⅐poly(dC) under no salt conditions after correction for DNA contributions to ellipticity at 222 nm. Further addition of poly(dG)⅐poly(dC) above 115 M resulted in no additional change in ␣/␤-type SASP ellipticity at 222 nm. The binding site size (n) and the mean residue ellipticities at 222 nm for free ([ u ] 222 ), and DNA-bound ([ b ] 222 ), ␣/␤-type SASP were used to calculate the percentage of bound and free protein at each point in forward titrations according to a two-state model as described (6). The McGhee-von Hippel model was fit to these data by an iterative least squares method (using KaleidaGraph 3.0.2) to obtain apparent binding constants (K) for the interaction with pUC19, and intrinsic binding constants (K) and cooperativity factors () for the interaction with poly(dA-dT)⅐poly(dA-dT) (6). Errors in the fits of K and were determined by the model fitting program. The experimental error in K determinations was estimated to be Ϯ15% as determined by duplicate titrations of pUC19 DNA with SspC and poly(dA-dT)⅐poly(dA-dT) with SspC ⌬11-D13K .
Spores of B. subtilis that overexpress variants of SspC were prepared by nutrient exhaustion in 2ϫ SG medium (27). Spores were purified by sonication and distilled water washing at 4°C (28), and all spores used were Ͼ98% pure. Purified spores were analyzed for resistance to both wet heat and UV radiation as described (19,29).

SspC N-terminal Variants Bind to DNA in Vitro-Amino
acid sequence alignment of all identified ␣/␤-type SASP from Bacillus, Sporosarcina, and "Thermoactinomyces" species show that these proteins are most highly conserved in the C-terminal two-thirds of the protein ( Fig. 1) (4), and part of this region is thought to be involved in directly contacting the DNA (30). In contrast, the sequence of the N-terminal one third of ␣/␤-type SASP is less well conserved with variations in both length and amino acid sequence (4). The identification of protein-protein contacts between amino acid residues found within the Nterminal 40 -50% of ␣/␤-type SASP in the DNA-bound state (11) suggests that the N termini of ␣/␤-type SASP mediate protein-protein interactions which may be important for binding cooperativity, and therefore the apparent binding affinity of the DNA binding interaction. To explore this question in more detail, a number of N-terminal variants of SspC, a minor ␣/␤type SASP from B. subtilis, were produced and their DNA binding properties analyzed. SspC was chosen for this study because it is easily overexpressed and purified from both B. subtilis spores and E. coli, and its interaction with and effects upon DNA have been studied extensively both in vitro and in vivo (5,9,10). Furthermore, SspC binds to DNA with a higher affinity than many other ␣/␤-type SASP, and also has one of the longest N termini of all identified ␣/␤-type SASP (4,15).
Variants were designed in which blocks of approximately five amino acid residues were progressively removed from the N terminus of SspC (Fig. 1). In addition, SspC ⌬11 was constructed to assess the effect of an N-terminal deletion that removes an asparagine residue conserved in all ␣/␤-type SASP from Bacillus species and their close relatives (Fig. 1). Each deletion variant was designed such that the protein would contain an N-terminal alanine residue after post-translational removal of the initiating methionine residue (Fig. 1), since alanine is the N-terminal residue of most ␣/␤-type SASP (4,31). It was possible to overexpress in and purify SspC ⌬5 , SspC ⌬10 , SspC ⌬11 , and SspC ⌬14 from either E. coli, or from spores of B. subtilis. However, smaller SspC variants (SspC ⌬21 and SspC ⌬25 ) could not be overexpressed in E. coli nor in spores of B. subtilis (data not shown); the reason for this is not clear, but these proteins may not bind to DNA well and therefore they may be very unstable in vivo, since DNA binding greatly stabilizes ␣/␤-type SASP against protease digestion (32). In addition to these genetically engineered N-terminal SspC variants, the C-terminal fragments of SspC from endoproteinase Asp-N and GPR digests were also purified and analyzed (Fig. 1). The ability of SspC N-terminal variants and C-terminal proteolytic fragments to bind to DNA was first assessed by DNase I protection assays. All SspC N-terminal variants as well as the C-terminal Asp-N fragment of SspC gave approximately the same degree and pattern of DNase protection to linear pUC19 plasmid DNA as did wild-type SspC, whereas the C-terminal GPR fragment of SspC gave no DNase protection to pUC19 (data not shown). These data indicate that up to 14 residues of SspC (ϳ20% of , and GPR digest C-terminal peptide (GPR C-frag). Amino acid residues marked by an asterisk (*) are completely conserved in all ␣/␤-type SASP identified from Bacillus, Sporosarcina, and "Thermoactinomyces" species (4, 31); underlined acidic residues have been identified as sites of intermolecular EDC protein cross-linking (11). The downward-pointing arrow indicates the peptide bond cleaved by GPR.
total residues) may be removed without abolishing DNA binding.
All ␣/␤-type SASP are essentially random coils in the absence of double stranded-DNA, but become ␣-helical upon binding to DNA (6). These changes in protein conformation are conveniently monitored by CD spectroscopy, which was used to confirm the DNA binding properties of the SspC N-terminal variants (Fig. 2). The far UV CD spectra of all N-terminal SspC variants were indicative of unstructured polypeptides ( Fig. 2 and data not shown), but upon addition of poly(dG)⅐poly(dC) each variant except the GPR C-terminal fragment of SspC became significantly ␣-helical ( Fig. 2 and data not shown). The ␣/␤-type SASP bind to poly(dG)⅐poly(dC) very tightly (5,6), and the vast majority of the ␣/␤-type SASP (ϳ98%) is bound to DNA under these conditions (25 M ␣/␤-type SASP and 115 M (in bp) poly(dG)⅐poly(dC)). Therefore, the secondary structure content of the DNA-bound protein could be estimated using [] values from difference spectra in which the spectrum of free poly(dG)⅐poly(dC) had been subtracted from the spectrum of the complex (6,25). The estimates of secondary structure content suggest that the great majority of the removed residues are not part of regular secondary structure in the full-length protein as the calculated number of ␣-helical residues is similar (44 -50 residues) in each deletion variant (data not shown).
Thermal Stability and Equilibrium Binding Affinity of SspC Variant-DNA Complexes-The DNase I protection assays mentioned above indicate that all SspC N-terminal variants (with the exception of the C-terminal GPR fragment) are able to bind to plasmid DNA. However, these experiments are performed at protein concentrations in excess of the dissociation constants for the complexes, and thus it is difficult to detect differences in ␣/␤-type SASP-DNA binding affinities using this assay. Therefore, we conducted more sensitive and quantitative studies of SspC N-terminal variant-DNA binding, which were monitored by CD spectroscopy. It has been demonstrated previously that corrected [] 222 values from ␣/␤-type SASP-DNA mixtures can be used to determine the relative concentrations of DNA-bound and unbound ␣/␤-type SASP in solution (6). Thus, [] 222 values are an index of ␣/␤-type SASP-DNA binding, which can been used to determine the thermal stabilities (T M ) and equilibrium binding constants (K) for ␣/␤-type SASP-DNA interactions (6).
The thermal stability of SspC N-terminal variant complexes with poly(dG)⅐poly(dC) was determined by monitoring the CD at 222 nm for each protein-DNA complex as a function of temperature. Each ␣/␤-type SASP-poly(dG)⅐poly(dC) complex underwent a sharp thermal dissociation transition over a characteristic temperature range, and the midpoint of each transition (defined as T M ) was determined (Table I) (6). The thermal stability hierarchy was as follows: (Table I). This result was unexpected in that the deletion of additional residues from SspC ⌬11 to SspC ⌬14 actually increased the stability of the ␣/␤type SASP-poly(dG)⅐poly(dC) complex.
We next wanted to determine the apparent equilibrium binding constants (K) for the interaction between SspC N-terminal deletion variants and DNA to confirm the relative binding affinities suggested by the thermal stability studies. Intrinsic binding constants (K) and cooperativity factors () may be extracted from equilibrium binding titrations of homogeneous synthetic polynucleotides by fitting the McGhee-von Hippel model of cooperative, nonspecific protein-nucleic acid binding to the experimental data (6,33). The cooperativity factor () is a dimensionless thermodynamic parameter, which describes the relative affinity of a protein ligand for a ligand-contiguous binding site versus an isolated binding site (33). Therefore, differences between SspC variants in both their intrinsic binding constants and binding cooperativity may be detected and quantitated; this is potentially important as the N termini of ␣/␤-type SASP are thought to be involved in DNA-dependent protein-protein interactions (11). Initial equilibrium binding studies were conducted with the N-terminal variants and poly(dG)⅐poly(dC); however, because the binding cooperativity is low with this polynucleotide and the interactions are very tight (6), it was impossible to accurately determine K and values. The interaction between ␣/␤-type SASP and poly(dA-dT)⅐poly(dA-dT) is much more cooperative than with poly(dG)⅐poly(dC) (6). Unfortunately, most of the SspC N-terminal variants did not bind to poly(dA-dT)⅐poly(dA-dT) at concentrations that could be accurately measured by CD spectroscopy. However, we were able to conduct titrations of poly(dA-dT)⅐poly(dA-dT) with SspC and SspC ⌬5 , and determined a significant decrease in both K and for the interaction with SspC ⌬5 compared with that with SspC (Table II). To obtain binding constants for all the N-terminal variants, additional equilibrium titrations were conducted using linear pUC19 plasmid DNA, which tends to bind to ␣/␤-type SASP more tightly  than poly(dA-dT)⅐poly(dA-dT). Apparent equilibrium binding constants (K) from these titrations confirmed the hierarchy of binding affinities as determined by thermal stability assays, although individual fits of K and were not possible because the McGhee-von Hippel model was not designed to describe binding to polynucleotides of heterogeneous sequence (Table  II).
Protein-Protein Contacts between DNA-bound SspC N-terminal Variants-Previously, we used chemical cross-linking to identify amino acid residues that form close contacts between adjacent DNA-bound ␣/␤-type SASP (11). These residues are located within the N-terminal 40 -50% of the proteins, and all identified cross-links occurred between the ␣-amino group of the protein and either of the two glutamate residues of the GPR recognition sequence, or a relatively nonconserved acidic residue (aspartate 13 in SspC) found closer to the N terminus of ␣/␤-type SASP (11) (Fig. 1). Therefore, the N terminus of a DNA-bound protein appears to be interacting with negatively charged residues on an adjacent DNA-bound protein. This proposed protein-protein interaction could partially explain the unexpected increase in binding affinity seen in SspC N-terminal variants as additional amino acid residues are removed from SspC ⌬11 to SspC ⌬14 (Table II). SspC ⌬10 , SspC ⌬11 , and the Asp-N C-fragment all contain a negatively charged aspartate residue near the N terminus (Fig. 1), which could interfere with the postulated DNA-dependent protein-protein interactions through electrostatic repulsion. This N-terminal aspartate residue is absent in SspC ⌬14 and may therefore account for the increase in DNA binding affinity over SspC ⌬10 , SspC ⌬11 , and the Asp-N C-fragment. To determine if the SspC N-terminal variants made the same type of protein-protein contacts when bound to DNA, these proteins were cross-linked in the presence of DNA and the cross-linking sites identified.
All N-terminal variants except the GPR C-fragment were cross-linked by EDC in a DNA-dependent fashion, although a very small amount of cross-linking was also seen in the absence of DNA (Fig. 3). DNA-independent cross-linking is probably nonspecific and occasionally occurs during lyophilization of the samples if EDC is not completely removed by prior dialysis. The smaller SspC N-terminal variants tended to cross-link with somewhat lower efficiency than wild-type SspC, whereas the cross-linking of SspC ⌬14 was significantly reduced compared with the larger variants (Fig. 3). To determine whether cross-linking was occurring between the same amino acid residues in the SspC N-terminal variants as in wild-type SspC, EDC cross-linked and monomeric proteins were purified by SDS-PAGE, digested with trypsin, and the total digests analyzed by MALDI-TOF mass spectrometry. MALDI maps of tryptic digests of wild-type SspC EDC dimer showed three additional peptides not seen in tryptic digest mass maps of EDC-treated SspC monomer (Fig. 4, A and B). The mass of one of these peptides (Fig. 4B, labeled 2745.12 Da) corresponded to that of a previously characterized EDC cross-linked peptide in SspC (Ala 1 -Arg 5 ϫ Ser 8 -Lys 27 ), which contains an isopeptide cross-link between the ␣-amino group of alanine 1 and the ␤-carboxyl of aspartate 13 (Table III) (11). The mass of a second unique peptide (Fig. 4, labeled 2490.75 Da) corresponded to that predicted for a peptide containing a cross-link between the ␣-amino group of alanine 1 and the ␥-carboxyl of either glutamate 29 or 33 (Ala 1 -Arg 5 ϫ Leu 28 -Arg 45 ) (Table III); these glutamate residues are located within the GPR recognition and cleavage site (Fig. 1). This type of cross-link has previously been identified in other ␣/␤-type SASP (SASP-A and SASP-C from B. megaterium, and Bce1 E10K from Bacillus cereus) treated with EDC (11), and therefore it was presumed that this peptide has the same structure. The mass of the third unique peptide was consistent with a peptide containing a cross-link between the ␣-amino group of alanine 1 and the C-terminal carboxyl group (Fig. 4, labeled 1802.55 Da). This third crosslinked species is probably a minor cross-linking product, as it has never been detected in reverse phase-HPLC tryptic maps of cross-linked SspC (11); therefore, we have not been able to purify this peptide nor confirm its structure. MALDI tryptic digest maps of EDC-cross-linked SspC ⌬5 and SspC ⌬10 indicated that the same amino acid residues were cross-linked as in wild-type SspC (Table III and data not shown). The MALDI tryptic map of SspC ⌬11 dimer only showed two cross-linked peptides corresponding to cross-links between the ␣-amino group and aspartate 13 and the C terminus (Table III). It is possible that the other expected cross-linked peptide from SspC ⌬11 failed to ionize efficiently and therefore was not detected by mass spectrometry. Only two cross-linked peptide species were expected in SspC ⌬14 dimer trypsin digests because the aspartate residue involved in cross-linking in the other proteins (aspartate 13 in wild-type SspC) has been deleted in this variant (Fig. 1). We were only able to detect one crosslinked peptide, corresponding to a cross-link between the ␣-amino group and the C terminus, in the trypsin digest of SspC ⌬14 dimer. However, we hypothesize that additional crosslinking occurs between the ␣-amino group and the glutamates Protein-Protein Interactions in ␣/␤-Type SASP of the GPR sequence, and that this cross-linked peptide probably does not ionize efficiently during mass spectrometry. These data, in conjunction with the estimates of DNA-bound N-terminal variant secondary structure content, suggest that the nature of the protein-protein interactions formed by the N-terminal variants while bound to DNA is very similar to those formed by wild-type SspC.
The ␣-Amino Group and N-terminal Charge ␣/␤-type SASP Are Important for DNA Binding-Amino acid residues within the N-terminal half of ␣/␤-type SASP have been shown to be in close contact with one another while in the DNA-bound state by cross-linking and all identified cross-links involved the ␣-amino group of the N terminus (11). The data presented thus far in this report also suggest that residues in the N-terminal region of ␣/␤-type SASP are important in determining the strength of protein-protein interactions and therefore DNA binding affinity. Therefore, we sought to determine whether the ␣-amino group of ␣/␤-type SASP is important for binding of these proteins to DNA. The N-terminal ␣-amino group of proteins can be specifically converted to an ␣-keto group by a nonenzymatic transamination reaction with glyoxylate and copper(II) ions, resulting in an oxidative deamination of the ␣-amino group (20,21). Several ␣/␤-type SASP were deaminated quantitatively as determined by polyacrylamide gel electrophoresis at acid pH (34), which demonstrated a net loss of positive charge (data not shown). The transamination reaction was specific as no other modifications were detected by reverse phase-HPLC peptide mapping of trypsin and endoproteinase Glu-C digests of unmodified and deaminated ␣/␤-type SASP (data not shown). DNase I protection assays showed that deaminated SspC conferred somewhat less protection to plasmid DNA than did untreated SspC (Fig. 5, lanes 3 and 6). CD-based equilibrium binding studies also demonstrated that N-terminal deamination significantly reduced the affinity of SspC for pUC19 with a Ͼ6-fold decrease in the apparent binding constant (Table III). The effect of N-terminal deamination was even more dramatic for another ␣/␤-type SASP, SASP-A, a major ␣/␤-type SASP from B. megaterium, which binds less tightly to DNA than does SspC; deaminated SASP-A provided much less DNase I protection to plasmid DNA than untreated SASP-A (Fig. 5). N-terminal deamination of SspC ⌬10 and SspC ⌬11 had the same effect upon the ability to provide DNase I protection to plasmid DNA as was seen with SASP-A ( Fig. 5 and data not shown), whereas the effect was less dramatic comparing SspC ⌬5 and deaminated SspC ⌬5 (Fig. 5).
To further confirm that charged residues at the N terminus of ␣/␤-type SASP are important for high affinity DNA binding, the aspartate 13 residue of SspC ⌬11 was changed to either an asparagine or a lysine residue. These changes were chosen because asparagine (or glutamine) and lysine residues are found at this position in other ␣/␤-type SASP (4, 31). Far UV CD spectroscopy of the resulting proteins, SspC ⌬11-D13N and SspC ⌬11-D13K , bound to an excess of poly(dG)⅐poly(dC) indicated that these proteins contain essentially the same amount of secondary structure as SspC ⌬11 in the DNA-bound state (data not shown). As predicted, SspC ⌬11-D13N -and SspC ⌬11-D13Kpoly(dG)⅐poly(dC) complexes were significantly more thermostable than the SspC ⌬11 -poly(dG)⅐poly(dC) complex (Table I). Accordingly, SspC ⌬11-D13N and SspC ⌬11-D13K also bound to DNA with significantly higher affinity than SspC ⌬11 (Table II). SspC ⌬11-D13N had ϳ10-fold greater affinity for pUC19 plasmid DNA than did SspC ⌬11 , whereas SspC ⌬11-D13K bound to pUC19 and poly(dA-dT)⅐poly(dA-dT) more tightly than even wild-type SspC (Table III). EDC cross-linking of SspC ⌬11-D13K and subsequent MALDI mass spectrometry analysis confirmed that this protein (and presumably also SspC ⌬11-D13N ) makes the same DNA dependent protein-protein contacts as the other SspC N-terminal variants (Table III).
Analysis of the Effects of SspC N-terminal Variants on Spore Properties-The in vitro analysis of SspC N-terminal variants described above demonstrates that up to 14 amino acid resi-dues may be deleted from the N terminus of SspC while still maintaining DNA binding ability, albeit at lower affinities than full-length protein. Therefore, we wanted to determine whether SspC N-terminal variants are able to confer heat and UV resistance on spores in vivo, as does the full-length protein.
SspC N-terminal variant proteins were overexpressed in ␣ Ϫ ␤ Ϫ spores of B. subtilis (which lack 75-80% of their ␣/␤-type SASP and are therefore very sensitive to heat and UV radiation) (18), and the resistance properties of these spores studied (Fig. 6, A  and B). SspC, SspC ⌬5 , SspC ⌬10 , and SspC ⌬11 were expressed at similar high levels in these spores. Spores expressing SspC, SspC ⌬5 , and SspC ⌬10 were essentially equally resistant to UV radiation, whereas spores expressing SspC ⌬11 were slightly, yet significantly, more sensitive to UV radiation (Fig. 6A). Similar results were obtained when heat resistance at 85°C was determined, except that spores expressing SspC ⌬5 and SspC ⌬10 appeared to be less resistant to heat killing than spores expressing wild-type SspC (Fig. 6B). The resistance properties of spores expressing SspC ⌬14 could not be directly compared with the other spore strains because SspC ⌬14 was only expressed at ϳ50% the levels of wild-type SspC (data not shown). These data indicate that, although SspC ⌬5 and SspC ⌬10 have a lower affinity for DNA in vitro, they still function nearly as well as wild-type SspC in vivo. a Amino acid residues are numbered according to their position in wild-type SspC. These peptides were detected only within EDC-treated cross-linked proteins and were absent in tryptic digests of EDC-treated monomeric proteins.
b These masses were determined using the negative ion mode, which resulted in underestimation of all peptide masses.

DISCUSSION
The results reported here indicate that a substantial portion of the N terminus of SspC (ϳ20% of total amino acid residues) may be deleted without destroying DNA binding function. These residues are therefore probably not directly involved in contacting the DNA, consistent with the low level of sequence conservation in this region of ␣/␤-type SASP. However, SspC N-terminal deletion variants do show reduced binding affinity for DNA, indicating that the deleted residues contribute significantly to the SspC-DNA binding interaction. Our inability to overexpress shorter SspC N-terminal deletion variants, and the failure of the GPR C-terminal fragment of SspC to bind DNA, suggest that proteins smaller than SspC ⌬14 do not interact with DNA very strongly, and are presumably rapidly degraded in vivo. Additionally, it appears that the N terminus of wild-type SspC is considerably flexible because protein crosslinking demonstrated that the N termini of the SspC N-terminal variants all make similar close contacts with acidic residues on adjacent DNA-bound proteins. Indeed, heterotypic cross-links have also been identified between SASP-A and SASP-C from B. megaterium, whose N termini vary greatly in length (4,11). These conclusions are consistent with the variability in N-terminal length and sequence seen in the amino acid alignment of all identified ␣/␤-type SASP; in fact, the first significantly conserved amino acid residue in wild-type SspC is asparagine 12 (Fig. 1). However, although not conserved or required for DNA binding, residues glutamine 2 through asparagine 11 in wild-type SspC are important in that they significantly increase the affinity of this ␣/␤-type SASP for DNA.
Thermal stability and equilibrium binding analysis of the SspC N-terminal variant-DNA complexes indicate that SspC ⌬14 binds to DNA with higher affinity than SspC ⌬10 , SspC ⌬11 , and the large Asp-N fragment of SspC. The DNA binding affinity of these proteins seems to be related to the net charge in the N-terminal regions. These findings, in conjunction with the EDC cross-linking data showing that the N terminus of each protein is in close contact with negatively charged residues on adjacent DNA-bound protein (11), may be explained by a model of DNA-dependent protein-protein interaction which involves a significant electrostatic interaction. According to this model, a flexible and positively charged N terminus from each DNA-bound protein interacts with a proposed acidic patch (formed by aspartate 13 and glutamates 29 and 33) found on a neighboring DNA-bound protein (Fig. 7). If the N terminus is arbitrarily defined as the first five amino acid residues, the net charge of the N terminus of wild-type SspC at pH 7.0 is ϩ2, and is conferred by arginine 5 and the ␣-amino group (Fig. 1). By the same criteria, most of the N-terminal deletion variants of SspC have a less positively charged N terminus: SspC ⌬5 is ϩ2, SspC ⌬10 is 0, SspC ⌬11 is 0, Asp-N C-fragment is 0, and SspC ⌬14 is ϩ1 (Fig. 1). These changes in N-terminal charge are due to the removal of arginines 5 and 7, and the increasing proximity of aspartate 13 to the N terminus as amino acid residues are deleted from wild-type SspC (Fig. 1). In particular, the presence of the negatively charged aspartate 13 residue at the N terminus (in SspC ⌬10 , SspC ⌬11 and the Asp-N fragment of SspC) appears to interfere with proteinprotein interaction and therefore with DNA binding affinity, because when this residue is removed (along with leucines 14 and 15) in SspC ⌬14 , the DNA binding affinity is significantly increased. Therefore, we hypothesize that electrostatic repulsion between the now N-terminal aspartate 13 and the acidic FIG. 6. Heat and UV radiation resistance of spores expressing SspC N-terminal deletion variants. Spores of various strains were purified, and their resistances to UV radiation at 254 nm (panel A) and heat at 85°C (panel B) resistance were determined as described under "Experimental Procedures." The following spores were tested: patch on an adjacent DNA-bound SspC ⌬11 (or SspC ⌬10 ) contributes to the lower DNA binding affinity. Consistent with this model, derivatives of SspC ⌬11 that contain either an asparagine or a lysine residue in place of aspartate 13 have significantly higher affinity for DNA than SspC ⌬11 . The model presented above (Fig. 7) is probably applicable to all ␣/␤-type SASP; the evidence for this is as follows. First, DNA-dependent EDC cross-links identical to those found in SspC have been identified in three other ␣/␤-type SASP from different species (11). Second, higher ␣/␤-type SASP-DNA binding affinity roughly corresponds to a net positively charged N terminus (5,6,11). Third, deamination of SASP-A and SspC and its variants reduces the N-terminal net positive charge and results in significantly lower DNA binding affinity. Of course, we cannot exclude the possibility that the N-terminal ␣-keto group in deaminated ␣/␤-type SASP actually disrupts binding interactions. However, the electrostatic component of ␣/␤-type SASP protein-protein interactions has been shown to be thermodynamically significant because the cooperativity parameter () of the ␣/␤-type SASP-DNA interaction decreases with increasing salt concentration (6), suggesting that DNA dependent protein-protein interactions are destabilized by salt.
Although the deletion of 10 amino acid residues from the N terminus of SspC resulted in a protein (SspC ⌬10 ) that showed significantly lower affinity for DNA in vitro, it was still able to confer almost full resistance to UV radiation and significant protection from heat to ␣ Ϫ ␤ Ϫ spores. This is not particularly surprising because SASP-␣, the ␣/␤-type SASP in B. subtilis most responsible for spore UV resistance, has a much lower affinity for DNA than does wild-type SspC (5,7). In fact, it is thought that high affinity for DNA is not necessary for ␣/␤-type SASP function because the major ␣/␤-type SASP are present at very high concentrations (ϳ1-2 mM) within the spore core (4). At these concentrations, ϳ80 -90% of ␣/␤-type SASP would be DNA-bound at binding constants as low as 1.0 -2.0 ϫ 10 4 M 1 . In addition, the spore core is very dehydrated compared with the corresponding cell cytoplasm, and most of the divalent cations are probably chelated by the enormous level of dipicolinate in the spore core (1); both of these spore core environmental factors should tend to favor ␣/␤-type SASP-DNA interaction. However, although SspC ⌬11 has a similar or even higher affinity for DNA in vitro compared with SspC ⌬10 , it was less able to confer UV and heat resistance to ␣ Ϫ ␤ Ϫ spores. The reason for this is not clear, but may be due to the removal of the very highly conserved asparagine 12 residue, which is found in all ␣/␤-type SASP identified from Bacillus species (4,31). Perhaps this conserved asparagine residue plays an significant structural role in vivo.
It appears that the function of amino acid residues glutamine 2 through asparagine 11 in wild-type SspC is to increase the DNA binding affinity of the protein. Although this increased affinity is not necessary for ␣/␤-type SASP function in vivo, it may reflect the actual role SspC plays in wild-type (␣ ϩ ␤ ϩ ) spores. SspC is a minor ␣/␤-type SASP and is only present as ϳ10% of total ␣/␤-type SASP within wild-type B. subtilis spores (4), with SASP-␣ and SASP-␤ comprising most (ϳ80%) of the remaining ␣/␤-type SASP (18). SspC has a higher affinity for DNA than both SASP-␣ and SASP-␤, and therefore SspC may have been selected to bind regions of the spore chromosome that are not bound efficiently by SASP-␣ or SASP-␤. Although there is no significant decrease in spore heat or UV radiation resistance in the laboratory when the sspC gene is inactivated, the role for SspC proposed above could confer a selective advantage to spores. This proposed function for SspC may be generalizable because all Bacillus examined to date contain a number of minor ␣/␤-type SASP (4,31).
The results in this report also have practical value in aiding in the designing of a minimal high affinity ␣/␤-type SASP for biophysical studies. Because all ␣/␤-type SASP are largely unstructured in the absence of double-stranded DNA (6), a well defined ␣/␤-type SASP-oligonucleotide complex will be required to obtain a high resolution structure. This complex should be small to avoid potential problems with degeneracy arising from the nonspecific nature of ␣/␤-type SASP-DNA binding. In addition, only small complexes (Ͻ25 kDa) are routinely tractable by multidimensional NMR methods. For these reasons, the identification of minimal ␣/␤-type SASP that maintain high affinity for DNA is desirable. This study indicates that ϳ1.2 kDa of SspC may be removed while retaining DNA binding, although there is an attendant loss of binding affinity. However, we have been able to produce a truncated ␣/␤-type SASP, SspC ⌬11-D13K , which binds to DNA with higher affinity than even wild-type SspC. Small, double-stranded oligonucleotides that bind to ␣/␤-type SASP have previously been identified by CD spectroscopic studies (6), and these oligonucleotides are currently being tested with SspC ⌬11-D13K in an attempt to determine the solution structure of an ␣/␤-type SASP-DNA complex by multidimensional NMR studies.