Histone H1 Proteins Act As Receptors for the 987P Fimbriae of Enterotoxigenic Escherichia coli*

The tip adhesin FasG of the 987P fimbriae of enterotoxigenic Escherichia coli mediates two distinct adhesive interactions with brush border molecules of the intestinal epithelial cells of neonatal piglets. First, FasG attaches strongly to sulfatide with hydroxylated fatty acyl chains. This interaction involves lysine 117 and other lysine residues of FasG. Second, FasG recognizes specific intestinal brush border proteins that migrate on a sodium-dodecyl sulfate-polyacrylamide gel like a distinct set of 32–35-kDa proteins, as shown by ligand blotting assays. The protein sequence of high performance liquid chromatography-purified tryptic fragments of the major protein band matched sequences of human and murine histone H1 proteins. Porcine histone H1 proteins isolated from piglet intestinal epithelial cells demonstrated the same SDS-PAGE migration pattern and 987P binding properties as the 987P-specific protein receptors from porcine intestinal brush borders. Binding was dose-dependent and shown to be specific in adhesion inhibition and gel migration shift assays. Moreover, mapping of the histone H1 binding domain suggested that it is located in their lysine-rich C-terminal domains. Histone H1 molecules were visualized on the microvilli of intestinal epithelial cells by immunohistochemistry and electron microscopy. Taken together these results indicated that the intestinal protein receptors for 987P are histone H1 proteins. It is suggested that histones are released into the intestinal lumen by the high turnover of the intestinal epithelium. Their strong cationic properties can explain their association with the negatively charged brush border surfaces. There, the histone H1 molecules stabilize the sulfatide-fimbriae interaction by simultaneously binding to the membrane and to 987P.

The tip adhesin FasG of the 987P fimbriae of enterotoxigenic Escherichia coli mediates two distinct adhesive interactions with brush border molecules of the intestinal epithelial cells of neonatal piglets. First, FasG attaches strongly to sulfatide with hydroxylated fatty acyl chains. This interaction involves lysine 117 and other lysine residues of FasG. Second, FasG recognizes specific intestinal brush border proteins that migrate on a sodium-dodecyl sulfate-polyacrylamide gel like a distinct set of 32-35-kDa proteins, as shown by ligand blotting assays. The protein sequence of high performance liquid chromatography-purified tryptic fragments of the major protein band matched sequences of human and murine histone H1 proteins. Porcine histone H1 proteins isolated from piglet intestinal epithelial cells demonstrated the same SDS-PAGE migration pattern and 987P binding properties as the 987P-specific protein receptors from porcine intestinal brush borders. Binding was dose-dependent and shown to be specific in adhesion inhibition and gel migration shift assays. Moreover, mapping of the histone H1 binding domain suggested that it is located in their lysine-rich C-terminal domains. Histone H1 molecules were visualized on the microvilli of intestinal epithelial cells by immunohistochemistry and electron microscopy. Taken together these results indicated that the intestinal protein receptors for 987P are histone H1 proteins. It is suggested that histones are released into the intestinal lumen by the high turnover of the intestinal epithelium. Their strong cationic properties can explain their association with the negatively charged brush border surfaces. There, the histone H1 molecules stabilize the sulfatide-fimbriae interaction by simultaneously binding to the membrane and to 987P.
Enterotoxigenic Escherichia coli (ETEC) 1 cause diarrhea in mammals by expressing at least one type of enteroadhesive fimbria and one type of enterotoxin. By adhering to intestinal epithelial cells, localized multiplication of an ETEC strain can progress to mucosal surface colonization and concomitant effective enterotoxin delivery. This was illustrated with hostspecific ETEC strains in both animals and human volunteers (1)(2)(3)(4)(5). ETEC are the most important etiologic agents of both neonatal and postweaning diarrhea in pigs (6,7). 987P-fimbriated ETEC cause diarrhea in neonatal piglets in the United States (8), Europe (9,10), Asia (11)(12)(13)(14)(15), and Central and South America (16 -18). In addition to this widespread distribution, the recent identification of human ETEC strains with 987Plike fimbriae (19,20) on various continents highlights the evolutionary adaptability of these fimbriae (21).
The 987P fimbria consists of the helical arrangement of protein subunits along a filamentous axis (22,23). It is a heteropolymeric structure that is made up of one major subunit, FasA, and two minor subunits, FasF and FasG (24). Fimbriae are not produced in the absence of any of these subunits (25). Electron microscopy and export studies indicated that FasG is the first exported subunit followed by FasF and FasA (26). Because fimbriae grow from the base, FasG was proposed to be the tip subunit and FasF, a linker molecule. The 987P system has one outer membrane protein (FasD) proposed to fold in a ␤-barrel structure and act as the 987P usher molecule (27). 987P was the first fimbrial system found to utilize two different subunit-specific periplasmic chaperones. The FasB chaperone interacts with the major fimbrial subunit, FasA (28), but not with FasG, the adhesin subunit. Conversely, FasC interacts only with FasG, but not FasA. A third chaperone-like protein, FasE, was also located in the periplasm and was shown to be required for optimal export of FasG. In contrast to most fimbriae, 987P do not agglutinate mammalian red blood cells but bind to receptors only found on the relevant piglet intestinal cells (29). Age-related resistance to 987P-fimbriated ETEC was related to the appearance of new receptors in the mucus of post-neonatal pigs (29,30). These receptors were proposed to inhibit 987P-mediated bacterial colonization by competing with membrane-anchored receptors. Glycolipid receptors were identified and further characterized as sulfatide and hydroxylated ceramide monohexoside (31,32). FasG was shown to be responsible for 987P fimbrial binding to the former molecule, whereas FasA mediated 987P binding to the latter receptor (31). Alanine-scanning mutagenesis of fasG identified several positive-charged residues as involved in sulfatide recognition (33). One fasG mutant (K117A) did not bind to sulfatide, suggesting that the substituted lysine residue and possibly the other charged residues communicate with the sulfate group of sulfatide by hydrogen bonding and/or salt bridges.
In addition to sulfatide, a group of piglet brush border proteins of ϳ32-35 kDa were described to act as 987P protein receptors in ligand blotting assays (29). This was confirmed by showing that FasG was the 987P adhesin responsible for this interaction (24). All the allelic FasG adhesins with reduced binding to sulfatide still interacted like wild-type FasG with the protein receptors (33). Studies with truncated FasG molecules indicated that at least two FasG segments (Glu-211-Ser-220 and Asp-20 -Ser-41) were involved in protein receptor recognition (34). Different FasG residues participated in sulfatide and protein receptor binding. Thus, the FasG adhesin was proposed to harbor two different binding domains for its two types of receptors. In this study the porcine intestinal protein receptor for the 987P fimbriae was identified, and the domain most important for the interaction with 987P was characterized.

Bacterial Strains and Plasmid
Constructs-E. coli strains and plasmids used in this study are listed in Table I. Standard procedures were used to prepare new constructs.
Media and Reagents-Wild-type E. coli strain 987 was grown in minimal medium E supplemented with pantothenic acid and glycerol (35). Strain JM109 was used for recombinant DNA work. Strain B834(DE3) (Novagen, Madison, WI) was used for 35 S-labeling of the 987P fimbriae. Strain BL21(DE3) was used for the expression of a porcine histone H1 protein and partial fragments of it. Cultures for colony isolation or plasmid purification were grown in LB medium. When appropriate, media were supplemented with the following antibiotics: ampicillin (200 g/ml), chloramphenicol (30 g/ml), tetracycline (10 g/ml). Culture media were purchased from Difco. Restriction and modification enzymes were from New England Biolabs, Inc. (Beverly, MA). Unless specified, reagents were purchased from Sigma.
Preparation of 987P Fimbriae and Brush Borders of Intestinal Epithelial Cells-Fimbriae expressed on the bacterial surface were prepared by heat extraction, and brush border vesicles were prepared from piglet enterocytes as described earlier (24).
Polyacrylamide Gel Electrophoresis, Western and Ligand Blotting-Bacterial pellets, isolated fimbriae, purified brush border vesicles, or histone H1 proteins were separated by SDS-PAGE in the absence of reducing agents. Gel migration shift assays were undertaken with 987P-histone H1 or 987P-poly-L-lysine mixtures (4.5:1 concentration ratios) in native 5% polyacrylamide gels (0.75 mm thick, 8 cm long) by using a phosphate buffer (0.1 M, pH 7) and reversing the electrodes at the power supply so that the positively charged histone H1 proteins migrated to the negative cathode (30 mA constant current, 4 -8 h) (36). For Western blotting, the histone H1 proteins were probed with a sheep anti-calf histone H1 polyclonal antibody (Fizgerald, Concord, MA) or with a rabbit anti-porcine histone polyclonal antibody (prepared by Cocalico Biologicals Inc., Reamstown, PA, using purified pig histone H1 proteins as antigen) using horseradish peroxidase (HRP)-conjugated secondary antibodies and enhanced chemiluminescence (ECL) for detection (27). For the ligand blotting assays, the blots were probed with isolated 987P fimbriae (10 g in PBS, 0.5% BSA) or with HRP-conjugated concanavalin A. Bound fimbriae were detected with 987P quater-nary structure-specific monoclonal antibody E11 (23), HRP-conjugated secondary antibodies, and ECL. For some experiments, isolated 35 Slabeled 987P fimbriae were prepared from E. coli B834(DE3)/pBKC1, pBKC2, and used for fluorography, as described previously (34). For the ligand blotting inhibition assay, nitrocellulose strips with SDS-PAGE separated histone H1 (5 g) were incubated with different concentrations of inhibitors for 1 h at room temperature. After several washing steps, the strips were incubated with 35 S-labeled fimbriae.
Amino Acid Sequencing-Brush border vesicle proteins were separated by SDS-PAGE. Protein bands recognized by the 987P fimbriae were mapped by a ligand blotting assay of one of two gels run in parallel. The two major protein bands labeled by ligand blotting were cut out of the gel and separately subjected to in-gel proteolysis with trypsin. A blank control band from a parallel gel run without brush border proteins was subjected to the same treatment. Tryptic fragments of the two proteins and the mock control were separated by microbore reversed phase HPLC. Of the two analyzed protein bands giving the same HPLC profile, only samples from HPLC peaks of the top major band were analyzed by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) (Proteonomics Facility, Dr. D. W. Speicher, The Wistar Institute, University of Pennsylvania). Several samples specific for the brush borders (i.e. not present in the HPLC profile of the mock control band) and showing only one distinct peak by MALDI-MS were submitted to sequencing by Edman degradation (Applied Biosystems, Procise Protein Sequencing system, Model 494).
Isolation of Porcine Histone H1 Proteins-Histone H1 proteins were isolated from intestinal epithelial cells of neonatal piglets by using a standard protocol (37,38). Briefly, epithelial cells were obtained by gently scraping the mucosa of washed segments of the small intestines of 1-3-days-old piglets and placing the material in iced-cold PBS, pH 7.2, containing a mixture of protease inhibitors (Roche Applied Science). The mucosal cells were washed 3 times, and histone H1 proteins were extracted from the mucosa with 5% HClO 4 at 4°C for 30 min. The treated cells were pelleted by centrifugation (14,000 ϫ g, 10 min) at 4°C. The histone H1 proteins were obtained by precipitation in trichloroacetic acid (20%, final concentration) at 4°C overnight. The precipitated histone H1 proteins were concentrated by centrifugation (14,000 ϫ g, 20 min, 4°C), washed with cold acetone three times, dried in a vacuum desiccator and dissolved in deonized H 2 O to determine the protein concentration (39). Rat histones were from a rat nuclear extract. The quality of each batch was tested by SDS-PAGE to confirm at least 90% purity. Protein aliquots were stored at Ϫ70°C for later use.
Binding Assay in Microtiter Plates-Fimbriae (1 mg in 1 ml of PBS) were biotinylated with 2 l of 10 mg/ml NHS-LC-LC-Biotin (Pierce) in deionized H 2 O for 2 h at room temperature. The fimbriae were dialyzed in PBS to remove unreacted biotin. Ninety-six-well enzyme-linked immunosorbent assay plates (polyvinyl chloride; Falcon, BD Biosciences) were coated with porcine histone H1 proteins or BSA as a control (5 g/ml 0.1 M PBS, pH 7.2, 100 l/well) and incubated at 37°C for 3 h. The plates were blocked with 5% BSA in PBS at 37°C for 1.5 h, washed 4 times with PBS, and incubated with serial dilutions of biotinylated 987P fimbriae in PBS, 0.5% BSA, 0.01% Tween 20 for 2 h at room temperature. After four washes with PBS, 0.01% Tween 20, 100 l of streptavidin-HRP conjugate diluted 1/2000 in PBS, 0.5% BSA was added to each well and incubated for 30 min at room temperature. After the last washing step, bound fimbriae were detected by using o-phenylenediamine as the chromogenic substrate analogue and reading the absorbance at 450 nm (37). Protease Treatments of Brush Border Vesicles or Histone H1-Isolated brush borders (10 mg protein/ml, 50 l) were digested for 1 h at 56°C with proteinase K (Roche Applied Science, 10 g) in 100 l of digestion buffer P (10 mM Tris, pH 7.8, 5 mM EDTA, 0.5% SDS) or for 2 h at 37°C with TPCK-treated trypsin from bovine pancreas (25 g) in 100 l of digestion buffer TC (100 mM NH 4 HCO 3 , pH 8.0, 0.1% SDS). Histone H1 (36 g) was digested for 4 h at 37°C with TLCKtreated chymotrypsin from bovine pancreas (0.5 g) in 40 l of digestion buffer TC or for 24 h at 37°C with Staphylococcus aureus V8 protease (0.6 unit) in 40 l of digestion buffer V (50 mM NH 4 HCO 3 , pH 8.0, 1.25 mM CaCl 2 , 1.275 M urea). The digested brush border or histone H1 proteins were retaken in 2 ϫ SDS sample buffer for SDS-PAGE, as described above.
Cloning and Expression of Porcine Histone H1 in E. coli-The following steps were taken to clone a full-length porcine histone H1 cDNA. Two expressed sequence tags (ESTs) in the pig data base (AW415122 and B1185028) contained DNA that was similar to known human and murine histone H1 sequences. These EST sequences were used to design two similar pairs of oligonucleotide primers for PCR (UH1-2, 5Ј-CCCCCAGTGTCCGAGCTCATCAC-3Ј, and LH1-2, 5Ј-GGTCTG-TACCAAGGTGCCCTTGCTCAC-3Ј; UH1-3, 5ЈCCTCCGGTGTCCGA-GCTCATCAC-3Ј, and LH1-3, 5Ј-AGTCTGCACCAGGGTGCCCTTGCT-CAC-3Ј. Using a ZAPII library of pig jejunal cDNA (40), two PCR fragments (both 177 bp) were obtained and cloned into plasmid pGEM-T (Promega, Madison, WI). The cloned DNA fragments were sequenced and shown to be 91.6 -100% similar to the corresponding region of the EST sequences. In a second step, the two 177-bp PCR fragments were purified and labeled in one reaction mix (Random Primer fluorescein labeling kit; PerkinElmer Life Sciences) to probe the ZAPII library by plaque hybridization as recommended (Stratagene, La Jolla, CA). From about 10 5 plaques, three positive plaques could be purified successfully in the following screenings. The inserts of the corresponding in vivo excised phagemids were sequenced, and one insert was shown to harbor the 5Ј end of a histone H1 ORF. This 384-bp DNA fragment was amplified with a stop codon at its 3Ј end using PCR and cloned into plasmid pET22b (Novagen) to get pZS384 (from amino acid residue Met-1 to Pro-128) (Table I). Third, a full-length porcine histone H1 cDNA was cloned by reverse transcription-PCR (Marathon cDNA amplification kit; Clontech, Palo Alto, CA) using primers designed to include the start and stop codons of porcine histone H1, as suggested by the EST sequences (UH1, 5Ј-ATGTCGGAGACCGCTC-CAGTG-3Ј, and LH1, 5Ј-CTATTACTTCTTCTTGGAGACGGCTTTCT-3Ј), and isolated mRNA from piglet enterocytes as template (Micro-Trak 2.0 mRNA isolation system; Invitrogen). The amplified cDNA was cloned into pGEM-T (pZSH1) and shown to represent a full-length 669-bp ORF by DNA sequencing. To express histone H1 or various continuous domains of it in E. coli, four PCR products incorporating histone H1 DNA were amplified from pZSH1 as template. For this three pairs of primers including NdeI at their 5Ј end and BamHI at their 3Ј end were used to amplify and clone corresponding fragments into the expression plasmid pET22b. The three resulting constructs were pZS669 (full-length histone H1), pZS444 (from residues Met-1 to Phe-148), and pZS212 (from residues Ala-36 to Leu-108). A fourth construct, pZS468, was prepared by inverse PCR (41) using pZS669 as template and two primers designed to delete residues Lys-30 to Asn-109 from the full-length histone H1 (Table I). The histone H1 protein and protein fragments were expressed in E. coli BL21(DE3) and analyzed by SDS-PAGE as described previously (26,34).
Immunohistochemistry and Electron Microscopy-Several 1-cm-long intestinal segments were collected from neonatal piglets and fixed immediately after euthanasia following IACUC guidelines. For immunohistochemistry, tissues were fixed in 10% buffered formalin, dehydrated, and paraffin-embedded. Glass slide-mounted serial sections were prepared for indirect immunohistochemistry using mouse antihuman histone H1 monoclonal IgG 2a (clone AE-4; Upstate, Charlottesville, VA). All reactions were performed at room temperature. Tissue sections were treated with H 2 O 2 for 10 min to inactivate endogenous peroxidases and incubated with the primary antibody for 30 min followed by biotinylated anti-mouse antibody for 15 min at room temperature, HRP-conjugated streptavidin for 15 min, and 3,3Ј-diaminobenzidine as chromogen for 3 min. The counterstaining was obtained using Mayer's hematoxylin; the slides were then dehydrated, and coverslips were applied. The negative controls were obtained using N-Universal Negative Control (cocktail of mixture of nonspecific mouse IgG, IgM, IgG 2a , IgG 2b , and IgG 3 , DakoCytomation, Carpinteria, CA) instead of the primary antibodies. Transmission electron microscopy was undertaken at the Biomedical Imaging Core, University of Pennsylvania.
Intestinal tissue was fixed in 4% paraformaldehyde, embedded in LR-White (Electron Microscopy Sciences, Hatfield, PA), and UV-cured at Ϫ20°C. 90-nm-thick sections were picked up on nickel grids for postembedding immunostain with the mouse anti-human histone H1 monoclonal IgG 2a (clone AE-4 described above) followed by an anti-mouse antibody gold particle (10-or 15-nm diameters) conjugate. A mouse monoclonal IgG 2a was used as the isotype (negative) control (Caltag Laboratories, Burlingame, CA). Images were captured in Jeol JEM 1010 by an AMT 12-HR-aided Hamamatsu CCD camera (42). All the supplies for electron microscopy were from Electron Microscopy Sciences (Fort Washington, PA).

Binding of 987P Fimbriae to Protein Receptors of Brush
Border Vesicles-As shown previously, the 987P fimbriae, used as ligand, bind to a group of intestinal brush molecules migrating like globular proteins of 32-35 kDa (Fig. 1A, lanes 1 and 4). The proteinaceous nature of these receptors was confirmed by their susceptibility to the proteolytic activities of trypsin or proteinase K (Fig. 1A, lanes 2, 3, 5, and 6). N-Glycosidase F treatment of the isolated brush border vesicles did not affect the 987P binding activity and the migration properties of the receptor bands on SDS-polyacrylamide gels (Fig. 1B, lanes 5  and 6). To control for the activity of the enzyme, a known N-glycosylated protein (RNase B) was treated with the same N-glycosidase and probed with concanavalin A, a lectin known to recognize the N-glycosylated moiety of RNase B. In contrast to the untreated RNase B, N-glycosidase F-treated RNase B appeared smaller by SDS-PAGE analysis and was not recognized by concanavalin A on a ligand blot (Fig. 1B, lanes 1-4), indicating that the enzyme was active. Moreover, dithiothreitol treatment of the isolated brush border vesicles did also not affect the 987P binding activity and the migration properties of the receptor bands (Fig. 1C, lanes 3 and 4). The activity of dithiothreitol was controlled with the major 987P fimbrial subunit which contains one disulfide bridge and migrates as an apparently larger protein when reduced (Fig. 1C, lanes 1 and  2). These results indicated that the studied intestinal 987P protein receptors do not involve N-linked carbohydrates or  1-3) and ligand blot assay (lanes 4 -6) of protease-treated piglet BBV proteins using 987P fimbriae as the ligand followed by quaternary structure-specific anti-987P monoclonal antibody E11 and a horse-radish peroxidaseconjugated secondary antibody to label the bound ligand. Lanes 1-4, untreated; lanes 2 and 5, trypsin-digested; lanes 3 and 6, proteinase K-digested BBV proteins. B, Coomassie Blue-stained gel of RNase B (control, lanes 1 and 2) and ligand blots of RNase B (lanes 3 and 4) and BBV proteins (lanes 5 and 6). The proteins were untreated (lanes 1, 3, and 5) or digested with peptide N-glycosidase F (lanes 2, 4, and 6) before running the gel and probed with the HRP-conjugated lectin concanavalin A (lanes 3 and 4) or with 35 S-labeled 987P fimbriae (lanes 5 and 6). C, Coomassie Blue-stained gel of 987P fimbriae (control, lanes 1 and 2) and ligand blot of BBV proteins using 35 S-labeled 987P fimbriae as the ligand (lanes 3 and 4). The proteins were boiled for 5 min in the absence (lanes 1 and 3) or presence (lanes 2 and 4)  disulfide bridges that are important for their 987P fimbriaespecific receptor activity.
Identification of the Protein Receptors for 987P-The identity of the 987P protein receptors was determined by protein sequencing. For this, brush border proteins from piglet intestinal epithelial cells were separated by SDS-PAGE, and the two major protein bands recognized by the 987P fimbriae were individually digested with trypsin. Tryptic fragments of the two protein bands separated by microbore reversed phase HPLC showed a similar peak profile. This indicated that the two protein bands represented partially degraded or posttranslationally modified variants of the same protein. Thus, only the upper protein band was studied further. Samples of separate and distinctly unique peaks were analyzed by MALDI-MS. The samples containing a single peak by MALDI-MS were submitted to protein sequencing. Two samples contained one peptide each that could be sequenced successfully. The peptide sequences (ALAAAGYDVEK and LGLK) were shown by BLAST to match the sequences of several human and murine histones H1 and of the only sequenced porcine histone H1 (the testis-specific H1 variant) as well as two histone segments found in the pig expressed sequence tags (ESTs) data base. An additional sequenced segment could be assembled to match some histone H1 sequences (SETAP), although this information was from one of the samples containing 2-5 peptides that could not be sorted into distinct sequences because of similarity of signals. Based on their molecular masses, histone H1 molecules would be expected to migrate as 22-24-kDa proteins on SDS-polyacrylamide gels. However, consistent with their high pI, their migration pattern doesn't relate to their effective molecular masses as shown for example with murine histone H1, which migrate as proteins of 32-35 kDa (38). To confirm the identity of the porcine histone H1 proteins in the intestinal brush borders and in the purified protein fraction, the migration properties of these proteins were compared on SDS-polyacrylamide gels. For this, porcine histone H1 proteins were isolated from piglet intestinal epithelial cells by using a standard protocol (38). The purity of the isolated proteins was evaluated by SDS-PAGE (Fig. 2 lanes 2-4). Purified porcine histone H1 proteins (Fig. 2, lane 2) co-migrated with the rat histone H1 (Fig. 2, lane 1) as 32-35 kDa proteins, matching the previously described migration properties of the 987P protein receptors. The identity of the isolated porcine histone H1 proteins was confirmed by Western blot analysis using anti-calf histone H1 polyclonal antibodies (Fig. 2, lanes  5-7). The antibody reacted only with purified porcine histone H1 proteins (Fig. 2, lane 6) and the corresponding bands in the total porcine brush border vesicle (BBV) extract (Fig. 2, lane 7) but not with the porcine histone H1-depleted BBV extract (Fig.  2, lane 5), confirming the identity of the purified proteins and the specificity of the antibodies. Similar results were obtained with polyclonal rabbit antibodies prepared against purified porcine histone H1 proteins (data not shown).
Binding of 987P to Histone H1 Proteins Purified from Porcine Intestinal Epithelial Cells-To verify that the porcine histone H1 proteins behave like the 987P protein receptors, the migration properties of these proteins were compared on SDS-polyacrylamide gels. 987P fimbriae bound to the isolated pure porcine histone H1 proteins (Fig. 3A), showing the same pattern of at least three bands of 32-35-kDa proteins on both the SDS-PAGE (lane 1) and the ligand blot probed with 987P 5-7) of total BBV proteins, purified histone H1, or histone H1-depleted BBV protein fractions are shown. Lane 1, purified rat histone H1 proteins; lanes 2 and 5, porcine histone H1-depleted BBV extract; lanes 3 and 6, porcine histone H1 proteins purified from BBV; lanes 4 and 7, total porcine BBV proteins. The Western blot was probed with sheep anti-calf-histone H1 antibody, and the ligand blot was probed with 987P fimbriae and anti-987P antibodies. Molecular mass markers are indicated on the left .   FIG. 3. 987P adhesion assays. A, Coomassie Blue-stained SDS-PAGE gel of porcine histone H1 proteins purified from BBV (lane 1) and ligand blot of the same material probed with 987P fimbriae and 987P quaternary structure-specific monoclonal antibody E11. Molecular mass markers are indicated on the left. B, binding of biotinylated 987P fimbriae (•), 987P-like CS18 fimbriae (f), or BSA (ࡗ) to microtiter wells coated with purified piglet histone H1. Binding of the 987P fimbriae was also tested with coated poly-L-lysine (E). C, gel migration shift assay. Histone H1 (lanes 1-6) or poly-L-lysine (lanes 7-9) polypeptides were preincubated for 1 h with 987P fimbriae (lanes 1, 2, and 7), BSA (lanes 3, 4, and 8), or PBS (lanes 5, 6, and 9). Untreated (lanes 1, 3, 5, 7, 8, and 9) or 5 min-boiled samples (lanes 2, 4, and 6) were separated on 5% nondenaturing polyacrylamide gels using reversed electrodes. The gels were stained with Coomassie Blue. fimbriae (lane 2). This result confirmed that the previously described 987P protein receptors are histone H1 proteins. The binding of 987P fimbriae to porcine histone H1 was ligand concentration-dependent, as shown in solid-phase binding (Fig.  3B, filled circles) and binding inhibition assays (Fig. 3E). Moreover, unlike the 987P fimbriae, the 987P-like CS18 fimbriae, which mediated the binding of human ETEC to enterocytes (21,43), did not bind significantly to histone H1 proteins (Fig. 3B,  filled squares). Because histone H1 molecules have a high content of lysine residues, it is possible that the 987P fimbriae interact non-specifically with these residues through salt bridges. To evaluate the affinity of 987P for arrays of positively charged residues, parallel binding assays were undertaken with histone H1 and poly-L-lysine molecules of 15-30 kDa (Fig.  3B, open circles). Although 987P fimbriae bound to the latter molecules, binding was significantly better to histone H1 when both lysine-rich molecules were fixed to a solid phase. In addition, only the histone H1 proteins and not the poly-L-lysine molecules were inhibited by the 987P fimbriae to migrate into a native polyacrylamide gel (Fig. 3C, lanes 1 and 7). The BSA or PBS controls had no significant effects on the migration properties of both lysine-rich molecules (Fig. 3C, lanes 3 and 8 or 5  and 9, respectively). Thus, the presence of lysine arrays is not sufficient to explain the 987P-histone H1 interaction. Interestingly, treating the histone-fimbriae mix at 100°C released only little H1 histone proteins (Fig. 3C, lane 2), suggesting that the association with fimbrial proteins was not affected by the treatment or reformed quickly after it. As a control, showing that protein degradation due to potential protease contaminants was not responsible for this result, no adverse effect on the amounts of 987P subunits and histone H1 proteins were detectable at the end of the incubation (Fig. 3D, lanes 1, 2, and 3). Similarly, the amount of BSA (control) was not reduced by the presence of the histone H1 proteins (Fig. 3D, lanes 4 and 5). Taken together, the data indicated that the 987P fimbriae adhere to porcine histone H1 and that this interaction is specific.

FIG. 2. Histone H1 proteins and the BBV receptor proteins for 987P. Coomassie Blue-stained gels (lanes 1-4) and Western blot (lanes
Localization of Porcine Histone H1 on Intestinal Brush Borders-To determine whether histone H1 molecules, known to be nuclear proteins, might effectively act as receptors for bacterial fimbriae, their cellular localization was evaluated microscopically. Immunohistochemical labeling with a monoclonal IgG 2a antibody specific for human histone H1 showed strong labeling of most nuclei (Fig. 4A), in comparison to control labeling with a mixture of nonspecific murine antibodies that included a nonspecific isotype antibody (Fig. 4, B and E). Although this expected result confirmed the specificity of the antibodies, light cytoplasmic labeling for the histone H1 proteins was a surprising finding. Patches of histone H1 accumulations in the cytoplasm were clearly detectable at higher magnification (Fig. 4, C and D). The compartmentalization of this staining was further visualized by the absence of specific labeling in the exocytic granules of goblet cells. Most importantly, a thin layer of histone H1 antigen appeared to cover the brush borders, some areas showing better covering (Fig. 4C). Electron microscopy with the same anti-histone H1 monoclonal antibody and gold particle-conjugated secondary antibodies was undertaken to better evaluate the suggested localization of histone H1 on the intestinal surface. In addition to the expected accumulation of gold particles in the nuclei, both the microvilli and the cytoplasm were consistently labeled with gold particles (Fig. 5, A-C). Only rare gold particles were detectable in controls labeled with the nonspecific isotype monoclonal antibody (Fig. 5D), supporting the specificity of the observed histone H1 labeling pattern. Comparable results were obtained with the polyclonal sheep anti-calf and rabbit anti-porcine histone H1 polyclonal antibodies (data not shown). Similar data were also obtained from different small intestinal segments. Taken together, these results indicated that histone H1 is present on the intestinal surface of neonatal piglets.
Cloning and Sequencing Porcine Histone H1-To better characterize porcine histone H1 and prepare tools aimed at studying the nature of the 987P-histone H1 interaction, the specific cDNA was cloned and sequenced as described under "Experimental Procedures." In contrast to the mRNAs of the replication-dependent histones, the replacement histone mRNAs are polyadenylated (44). Thus, some porcine histone H1 cDNA was first cloned from a cDNA phage library by using partial sequence information from the porcine EST data base. Two PCR-amplified 177-bp DNA fragments were used to probe the library. One phage was found to carry histone H1 DNA after sequence analysis. In addition to extraneous DNA, the derived phagemid contained the 5Ј end of a histone H1 ORF. This partial ORF was cloned by PCR with a stop codon at its 3Ј end (pZS384) (GenBank TM accession number AY489287). Based on this sequence and the two available EST sequences, the cDNA sequence for a full-length ORF of a porcine histone H1 was determined (GenBank TM accession number AY489288) using freshly prepared RNA from piglet intestinal epithelial cells as template for reverse transcription-PCR. A similar sequence was obtained from piglet liver and kidney mRNA used in parallel for comparative purposes (GenBank TM accession number AY489289). The predicted protein sequence of piglet enterocyte histone was most similar to murine H1.3 (93% residue identities) followed by human H1.4, human H1.3, and murine H1.4 (data not shown).
987P Binding to Porcine Histone H1 Fragments-Histone H1 molecules can be divided into three domains, a central globular domain flanked by mostly unstructured tails (45). Current models suggest that the former domain binds to nucleosome, whereas the basic shorter N-and longer C-terminal ends bind to linker DNA. To determine whether one or more of these domains interact directly with the 987P fimbriae, protein fragments encompassing various portions of porcine histone H1 were prepared by protease treatments or DNA cloning and expression techniques. Large C-terminal fragments of histone H1 were prepared by endoproteolysis with chymotrypsin or the S. aureus V8 protease. Both proteases target amino acid residues found exclusively in the N-terminal end and globular domain of histone H1. As shown by SDS-PAGE, both proteases left intact histone H1 fragments of 16.5-17 kDa (Fig. 6, lanes 2  and 3). Based on the known sequences of mammalian histone H1 proteins, these fragments represent the histone H1 lysinerich C termini. Interestingly, 987P fimbriae bound efficiently to both fragments, as shown in ligand blots (Fig. 6, lane 5 and 6), suggesting that the 987P binding interaction of the histone H1 proteins involves their lysine-rich C termini. To study the role of the N terminus of histone H1 proteins for the interaction with the 987P fimbriae, two gene fragments that include the DNA for the N terminus of a porcine histone H1 and the complete gene were cloned and expressed in E. coli, as shown by SDS-PAGE (Fig. 7, lanes 1-3). Although the 987P fimbriae bound to the full-length recombinant histone H1 protein in a ligand blot assay (Fig. 7, lane 8), they did not interact significantly with recombinant histone H1 trimmed of their C termini (Fig. 7, lanes 6 and 7), indicating that the N terminus of histone H1 is not needed for 987P recognition. In contrast, the fimbriae adhered efficiently to the recombinant histone H1 protein that had only its globular domain deleted, keeping its C terminus intact (Fig. 7, lanes 5 and 10). Finally, the binding of 987P fimbriae to the full-length recombinant histone H1 protein indicated that posttranslational modifications of the histones occur- ring only in the eukaryotic host were not necessary for binding. Taken together, the results indicated that the 987P fimbriae recognize and adhere to the lysine-rich tail of porcine histone H1.

DISCUSSION
Protein receptors of ϳ32-35 kDa for the 987P fimbriae of ETEC were originally detected on intestinal brush borders and mucus of pigs (29). Later, the 987P minor subunit FasG was identified as the specific ligand molecule for the porcine protein receptors (24). The observed interaction involved at least two continuous stretches of the FasG sequence (34). Here, studies were aimed at characterizing the protein receptors. Protein sequencing suggested that these receptors were histone H1 proteins. In addition, to show that the 987P-histone H1 interaction was specific, the presence of histone H1 on the intestinal mucosa of neonatal piglets was confirmed by biochemical and microscopic means. Thus, unlike a previously reported intestinal rabbit glycoprotein whose carbohydrate portion was shown to be important for the 987P-receptor interaction (46), the porcine receptors were non-glycosylated proteins that, in addition, were highly basic and lack cysteines. This result was consistent with our findings that dithiothreitol or N-glycosidase F did not affect the structure and the binding properties of the 987P protein receptors, as evaluated by SDS-PAGE and ligand blotting.
Finding histone H1 on the cell surface was surprising but not without precedent. Nuclear proteins were located on the surface of various cells, including monocytes and lymphocytes (47)(48)(49)(50). It was suggested that the extracellular histones released from apoptotic cells bind to the negatively charged surfaces of neighboring viable cells (51). Intestinal epithelial cells undergo a rapid turnover, replicating in the crypts and migrating to the tip of the villi, where the bulk of cell loss occurs mainly by apoptosis (52). Apoptotic cultured epithelial cells release histones, including histone H1, from nucleosomes (53). Moreover, epithelial cells from intestinal villi that detach from the basement membrane undergo apoptosis and release histone H1 proteins (54). The surface of brush border membranes of porcine intestinal cells is expected to contain a significant amount of negatively charged molecules, including sulfatide (31,32). It is suggested that the histone H1 proteins mediate the binding to brush borders by ionic interactions, as previously modeled with small basic peptides and membranes containing acidic lipids (55). Because of the low levels of trypsin (56) and the presence of trypsin inhibitors in the intestines of neonatal piglets (57) and maternal colostrum (58), surfacebound histone H1 is not expected to be degraded by trypsin. Additional mechanisms appear to protect the histone H1 from intestinal proteases, since the 987P protein receptor was also observed on the brush borders of older pigs (29). For example, the presence of sulfatide in the brush border membrane of piglets should have an inhibitory activity on trypsin, as shown with various acidic lipids and sulfoglycolipids (59).
In this study 987P was confirmed to bind specifically to purified porcine histone H1 in a dose-dependent manner. Using labeled ligand, the interaction could be inhibited in a dose-dependent manner with unlabeled 987P. Moreover, the ligandreceptor interaction occurred both with solid-phase or soluble histone H1. A porcine histone H1 gene was cloned in an expression vector to help map its binding domain. The deduced protein sequence of the open reading frame of the gene was most similar to the human and murine histone H1.3 and H1.4 proteins. The 987P fimbriae bound to the recombinant protein, indicating that posttranslational modifications of histone H1 proteins were not needed for their receptor activities. All the   1-5) and a ligand blot (lanes 6 -10) of histone H1 polypeptide Met-1- Pro-128 (pZS384, lanes 1 and 6), histone H1 polypeptide Mer-1- Pro-148 (pZS444, lanes 2 and 7), full-length histone H1 (pZS669, lane 3 and 8), negative control (pET22b, lanes 4 and 9), and histone H1 lacking residues Lys-30 to Asn-109 (pZS468, lanes 5 and 10) are shown. The binding of 987P fimbriae was identified with monoclonal antibodies E11 using ECL. Molecular mass markers are indicated on the left. histone H1 isoforms, also called linker histones, have a central globular domain flanked by a shorter N-and a longer C-terminal domain that contain predominantly lysine residues (45,60). Using several protease and recombinant fragments of histone H1, the lysine-rich C-terminal tail was found to be largely responsible for 987P binding.
In addition to histone H1 acting as a receptor for 987Pfimbriated ETEC, it is interesting to note that another nuclear protein, namely nucleolin, was recently identified as a cellular receptor for intimin, the major adhesin of enterohemorrhagic E. coli O157:H7 (61). Moreover, thyroglobulin was found to bind to extracellular histone H1 on the surface of macrophages, resulting in the cellular uptake of thyroglobulin (38). Histone molecules were shown to translocate across cellular membrane and penetrate epithelial cells in an ATP-independent manner (62). Whether brush border-bound histone H1 gets internalized by the porcine intestinal epithelial cells remains to be investigated. However, it is noteworthy that in this study, as described previously (54), significant amounts of histone H1 proteins were observed in the cytoplasm of piglet enterocytes. Whether the cytoplasmic histone resulted from nuclear release, intracellular uptake, or both remains to be determined. As recently revealed, after DNA damage in cells, nuclear histone H1.2 is released into the cytoplasm, where it induces apoptosis by releasing cytochrome c in a Bak-dependent pathway (63). It is tempting to speculate that the uptake of extracellular histone H1 proteins, which might include histone H1.2, could play an amplifying or synchronizing role for the apoptosis of neighboring enterocytes along the intestinal villi.
Various additional functions have been attributed to extranuclear histone H1, including the stimulation of myoblast proliferation (64,65). Extracellular histones and some of their proteolytic fragments released from epithelia were shown to have strong antimicrobial activities (54, 66 -69), not unlike other cationic peptides such as defensins or polymyxin. Whether the binding of 987P fimbriae to histones covering intestinal brush borders also serves to protect the bacteria by interdicting the access of the histones to their target of antimicrobial action, namely the bacterial outer membrane, is an intriguing possibility that is being investigated.
In summary, this study identified a third 987P-specific receptor on the biologically relevant intestinal cell surface (24,29,31). These interactions are mediated by the 987P adhesin FasG, which recognizes hydroxylated sulfatide and histone H1, and the major subunit FasA, which binds to hydroxylated ceramide monohexoside, most likely, galactosyl cerebroside. Bacterial expression of multiple ligands on the same fimbrial structure acting synergistically is consistent with a model involving a succession of primary and secondary binding sequences of events for successful bacterial colonization (70). It is hoped that further studies on the 987P binding domains and their relative importance for fimbrial adhesion will help to design new therapeutic and prophylactic agents against colonizing bacterial pathogens.