JOURNAL OF BACTERIOLOGY, Dec. 1985, p. 1332-1336 Vol. 164, No. 3 0021-9193/85/121332-05$02.00/0 Copyright ©D 1985, American Society for Microbiology Porphyrin Binding by the Surface Array Virulence Protein of Aeromonas salmonicida W. W. KAY,* B. M. PHIPPS, E. E. ISHIGURO, AND T. J. TRUST Department of Biochemistry and Microbiology, University of Victoria, Victoria, British Columbia V8W 2 Y2, Canada Received 5 June 1985/Accepted 26 September 1985

Congo red binding by virulent A-layer-containing (A') and avirulent A-layer-deficient (A-) strains of Aeromonas salmonicida was examined. Congo red binding to A' cells was enhanced by salt and thus hydrophobically driven, but at low Congo red concentrations binding was salt independent. Congo red was bound by A' cells by a kinetically distinct mechanism (Kd, 0.25 ,uM) which was absent in A- isogenic strains. Purified A-layer protein ("A protein") protein A also bound Congo red with similar affinity (Kd, 0.40 ,IM). Congo red binding was structurally specific; it was not influenced by a wide variety of compounds including amino acids and nucleotides and only weakly inhibited by structurally similar dyes. However, protoporphyrin IX and hemin were strong competitive inhibitors of Congo red binding. Protoporphyrin and hemin were bound only by A' strains (Kds of 0.41 and 0.63 ,uM, respectively). Furthermore, binding of these porphyrins was strongly inhibited by Congo red but weakly inhibited by hematoporphyrin. Purified A protein also bound protoporphyrin IX and hemin with affinities similar to those of A' cells (Kds of 0.94 and 0.41 ,uM, respectively).

Aeromonas salmonicida is a pathogen of a wide variety of used as complete media. Cells were grown at 20°C to late log fish and is the causative agent of the systemic disease phase on a reciprocating shaker. Cultures were routinely furunculosis, a commercially important disease in cultured harvested by centrifugation at 10,000 x g and washed once salmonids (6). Virulent strains possess a cell surface protein with 100 mM Tris hydrochloride buffer, pH 7.0 (TB). array designated as the A-layer (23), the principal compo- CR and porphyrin binding to whole cells. Cells were nent of which is a 49,000Mr protein arranged tetragonally resuspended to an optical density at 650 nm of 1 ml-' (6 x and contiguously over the cell surface (4, 11). Isogenic 108 cells ml-') in TB in 1.5-ml Eppendorf tubes. Various strains devoid of A-layer protein ("A protein") (A-) are concentrations of CR were added and allowed to bind for 10 avirulent (8, 11), and the role of the A-layer as a virulence min at room temperature. Cells were pelleted on an Ep- factor stems at least partly from the protection if affords pendorf centrifuge by centrifugation for 1 min. CR binding to against the bactericidal activity of serum (14). However, the cells was determined subtractively by assaying the residual A-layer also endows the cell with an enhanced ability to dye in the supernatant spectrophotometrically at 480 nm. In associate with macrophages (22) and to resist the effects of a competition experiments with interfering chromogenic com- variety of proteases (unpublished data), thus potentially pounds, the chromogen was first allowed to bind to the cells facilitating the spread of the infection. for 10 min; the cells were then centrifuged for 45 s, washed Virulent A-layer-containing strains of A. salmonicida can once with TB, pelleted once more, resuspended in TB, and be detected by resistance to bacteriophage whose receptors assayed for residual chromogen contamination of the super- reside beneath the A-layer (8), or, more conveniently, by the natant. In separate tubes the cells were resuspended in TB ability of A' cells to hydrophobically bind the aromatic containing the various amounts of CR, incubated for 10 min, sulfonated diazo dye, Congo red (CR), during growth on and pelleted for 1 min as before. solid media (7). CR media have been used for some time to CR and porphyrin binding to A protein. A Protein was permit differentiation of virulent and avirulent strains of a purified from outer membrane preparations as previously variety of gram-negative bacteria (13, 15, 20), but the mo- described (16) except that it was solubilized using 2 M lecular basis of CR binding is unknown. guanidine-hydrochloride in the presence of 0.1 mM The ability of Yersinia species and Shigella flexneri to phenylmethylsulfonyl fluoride. Various concentrations of absorb CR as well as hemin has been correlated with CR or porphyrins were mixed with 70 ,ug (1.4 nmol) of pure virulence (9, 15, 17), and it has been suggested that the A protein in 0.1 M TB (pH 7.0) at a total volume of 1 ml. ability to bind CR is related to the ability to sequester iron After 5 min, duplicate 0.5-ml samples were applied to (17). In this study we expand on these observations by Sephadex G-25 columns of 0.5-ml bed volume made from a demonstrating that CR and porphyrins are both bound by the 1-ml Eppendorf pipettor tip. The pipettor tip was inserted same site on the A protein of A. salmonicida. into the lid of a precut 1.5-ml Eppendorf tube into which a separate small hole had been drilled to prevent an air-tight MATERIALS AND METHODS seal. This column plus tube was then centrifuged for 10 s at Bacterial strains. A. salmonicida A451 is an A' virulent maximum speed on an IEC clinical bench top centrifuge strain. Isogenic A- derivatives A451-3 and A451-25 are stock equipped with a swing-out head. The contents of the tube laboratory strains. A451-3 is an A- strain containing normal containing A protein plus bound ligand (700 RI) were then lipopolysaccharide, and A451-25 is a strain deficient in both assayed for ligand concentration spectrophotometrically at A protein and the lipopolysaccharide o-oligosaccharide. 480 nm for CR, 460 nm for protoporphyrin IX, and 400 nm Media and growth conditions. L broth and L agar were for hemin. All samples were run in duplicate with duplicate protein-free controls at each ligand concentration tested. * Corresponding author. Chemicals. CR dye was obtained from MCB Manufactur- 1332 VOL. 164, 1985 PORPHYRIN BINDING BY PROTEIN OF A. SALMONICIDA 1333

I_ 1.0 E 4 4' Z75 o0 0 '0 0 5 0 b\\0 0.5 {) 025 , . In 1 0 . 0 2 Z co gx _ 0~~~- 0fD LF

C'g25 10 20 ~~~~~0 Ia- 1- 2 N , 0 1 2 3 4 0 5 10 15 20 -2 -1 CR (ug ml ) CR FREE (mM) FIG. 1. Effect of (NH4)2SO4 on the concentration-dependent FIG. 3. Kinetics of CR binding by purified A protein from A. binding of CR by A451 cells. Washed cells of A451 in TB were salmonicida. A protein was purified by a modification of the method incubated with CR in the presence (0) or absence (0) of 2 M of Phipps et al. (16). A protein-bound CR was separated from free (NH4)2SO4 for 10 min. Cells were collected by centrifugation, and CR by rapid centrifugation through Sephadex G-25. Inset: a the residual CR left in the supernatant was measured. Controls were Scatchard plot of the binding data where N is the stoichiometric run without cells at each CR concentration. Inset: a ieplot of the ratio of bound ligand to total protein and LF is the free-ligand same data but the total CR bound was plotted as a function of CR concentration. concentration. we examined further the binding of CR at low CR levels in ing Chemists, Inc., Cincinnati, Ohio. Protoporphyrin IX, the absence of salt. hemin, and hematoporphyrin were from Sigma Chemical Kinetics of CR binding to A' cells. A. salmonicida A451 Co., St. Louis, Mo. All other reagents were commercially (A') cells possess a kinetically distinct high-affinity binding available. of CR with an apparent Kd of 0.25 ,uM (Fig. 2). At higher CR concentrations no saturation was observed (Fig. 2, inset), RESULTS suggesting some nonspecific binding of CR to the cells at these ligand concentrations. Furthermore, A- cells of Effect of (NH4)2SO4 on CR binding to A' cells. At higher A451-3 had no high-affinity CR binding at low dye concen- CR levels, the binding of CR to A' cells was strongly trations, suggesting that A protein is responsible for binding. enhanced by the presence of 2 M (NH4)2SO4, whereas at 2 ,ug ml-' CR the added salt had little effect. At low dye Kinetics of CR binding to purified A protein. We developed a rapid dye-binding assay for CR to A protein since technical concentrations there was an apparent salt-independent bind- ing of CR which also suggested saturation kinetics (Fig. 1, problems such as dye precipitation hindered accurate bind- inset). Since salt enhances nonspecific hydrophobic binding, ing kinetics by more traditional binding assay methods such

TABLE 1. Effect of various hydrophobic compounds and dyes on /0.4 0 CR binding to A. salmonicida A451 2.0 CR CR w- Compounda binding Compound" binding (%)b (%) E VI /0.2 J Phenylalanine 83.3 Bromcresol green 108.8 0 1.5 Tryptophan 97.1 Bromcresol purple 79.4 0 0 Tyrosine 97.1 Bromthymol blue 87.3 E Methionine 92.7 Nitroblue tetrazolium 117.3 c Leucine 91.3 Methylthiazole 131.8 1.0 00 "-i / °~~~0.2 0.4 0.6 tetrazolium 9 Isoleucine 79.7 Fast violet 108.0 z 00 Valine 75.4 Fast garnet 100.0 00 ATP 102.6 Fast blue 108.0 0 0.5 oo NAD 102.6 Ruthenium red 120.0 Imidazole 83.3 Evans blue 79.5 , Methylxanthine 77.3 Trypan blue 69.2 I 110.6 Procion red MXSB 63.1 - Benzidine n Ia,-2 3 Aminoacridine 131.8 Procion red HE38 69.0 v Jr__ FREE (NM) Aminonaphthylsulfonate 72.7 Coomassie brilliant blue 55.7 FIG. 2. Kinetics of CR binding by A451 (A+) cells and A451-3 a All of the compounds listed in the left column had no interfering cells. Washed were absorption at 480 nm and thus were added directly to the cells at 200 ,ug ml-' (A-) cells incubated with various concentrations with 2 ,ug of CR ml-'. Compounds listed in the right column had absorption of CR and unbound CR was measured in the supernatants after interferences and were incubated at 200 ,ug ml-' with the cells, which were centrifugation. Results are plotted as double reciprocals of bound then harvested, washed, and assayed for binding of 2 pg of CR ml-'. versus free CR for A' cells (0) and isogenic A- cells (0). Inset b All assays were in duplicate. CR binding at 2 pg ml-' (2.9 ,uM) in the represents binding at higher CR concentrations for A+ cells. absence of added compounds was taken as 100%o. 1334 KAY ET AL. J. BACTERIOL. as equilibrium or flow dialysis. Rapid separation of the protein-dye complex by brief centrifugation through small 1.5 Sephadex G-25 columns proved to be an accurate and rapid assay method. Figure 3 depicts the CR binding kinetics A obtained by this method. Purified A protein also bound CR avidly with affinity comparable to that of whole cells (Kd, 0.4 1.0 ,uM). A Scatchard plot of the data (Fig. 3, inset) indicated that there were approximately two molecules of CR bound per monomer of A protein. Specificity of CR binding to A' cells. We reasoned that in all likelihood CR was acting as a structural analog for an . 0.5 Y unknown molecule with a unique binding site on the A O 0 protein of A' cells. From a survey of 28 potential candi- oE ~ dates, including amino acids, nucleotides, various hydropho- bic molecules, and several dyes (Table 1), it was apparent v0 that the majority were ineffective competitive inhibitors of 1.5 CR binding. Only the last five structurally similar dyes were somewhat inhibitory, Coomassie brilliant blue being the strongest. These data strongly suggested that CR was highly specific in its binding to A protein on A' cells and did not 1.0 °/ / complement a mono- or dinucleotide fold (5). 0~~~~~~~~~~~ Inhibition of CR binding by porphyrins. Since a correlation between CR absorption and hemin uptake during growth has 00,0~~~~~ been observed with virulent cells of other pathogens (13, 15, 0.5 W <....-v ~ 1 20), we decided to measure the effect of various porphyrins on CR binding to A' cells. Under the conditions presented in Table 1, that is, at a 100-fold excess of competitor to CR, no significant CR binding was observed in the presence of 10 15 protoporphyrin IX or hemin. Hemin is protoporphyrin IX 0 4 with a ferric ion ligand. Figure 4 shows the concentration- FREE (uM) dependent inhibition of CR binding by protoporphyrin IX, FIG. 5. Kinetics of hemin and protoporphyrin IX binding to hemin, and hematoporphyrin. Protoporphyrin IX was A451 (A') cells (0) and the isogenic A- mutants A451-3 (0) and strongly inhibitory, hemin was moderately inhibitory, and A451-25 (A). The experiment was as described in the legend to Fig. hematoporphyrin had little effect. Conversely, CR was 2 except that protoporphyrin (A) and hemin (B) were the ligands. shown to be a potent inhibitor of protoporphyrin and hemin binding (data not shown). Protoporphyrin IX, hemin, and hematoporphyrin binding strains A451-3 and A451-25. The binding of protoporphyrin to A' cells. Both protoporphyrin IX and hemin were avidly was of higher affinity (Kd, 0.41 ,uM) than that of hemin (Kd, bound by A' cells of A. salmonicida A451. Figure 5 depicts 0.63 ,M) but no kinetic component of hematoporphyrin the kinetics of binding for both of these compounds to A451 binding could be observed over a wide concentration range, and the complete absence of specific binding by the A- suggesting a very low binding affinity relative to the former two porphyrins. Protoporphyrin IX and hemin binding to A protein. Using the centrifugation binding assay developed for CR, we also observed high-affinity binding of both protoporphyrin IX 100 N (Kd, 0.94 ,uM) and hemin (Kd, 0.41 p.M) to A protein (Fig. 6). *0 As with whole cells, virtually no binding of hematoporphyrin K0 could be observed. 875 - DISCUSSION The binding of CR to A. salmonicida cells apparently zZ50 'l- proceeds by at least two mechanisms. In differential media D I\ (i.e., CR agar) at high concentrations of 30 to 50 ,g ml-' (7, 0 17) in the presence of some salt, much of the binding is cc undoubtedly hydrophobic and nonspecific and results in a 25 . large number of dye molecules bound per cell. At high CR concentrations, for instance at greater than 5 ,uM, dye binding is amplified due to the formation of stacked aggre- FI25020 40 gates (3). CR binding at high dye concentrations (Fig. 2, 500 inset) likely reflects this. The binding of CR to A' A. INHIBITOR (uig ml 1) salmonicida cells during growth on CR media likely pro- FIG. 4. Inhibition of CR binding to A451 (A') cells by ceeds by this mechanism. However, at low CR levels porphyrins. Washed cells of A451 were incubated with various binding appears to proceed by a salt-independent mechanism concentrations of either hematoporphyrin (0), hemin (A), or of surprisingly high affinity. It is possible that CR bound to protoporphyrin IX (0). After centrifugation and one wash with TB, the high-affinity site acts as an nucleation site for further the binding of CR (1 ,ug ml-') was measured. binding at higher CR concentrations. VOL. 164, 1985 PORPHYRIN BINDING BY PROTEIN OF A. SALMONICIDA 1335

1.0 molecules of protoporphyrin occupy a single site. One could view this site as a hydrophobic cleft into which four El protoporphyrin molecules stack. Hemin containing a bulky 0 iron atom could foreseeably prevent this. The chemical nature of this site is under further investigation. What then is the physiological role of porphyrin binding to A protein? It is tempting to speculate that such binding represents an initial stage in iron transport. Avirulent Shi- 7.// gella, Vibrio cholerae, Escherichia coli, and Neisseria men- ingitidis mutants which have lost the ability to bind CR have an enhanced virulence upon supplementation with iron (15). Perhaps the CR binding factor of these strains also binds hemin. It is also possible that the hemin binding site is a siderophore receptor site, since A' strains ofA. salmonicida have been shown to synthesize a unique siderophore (2). -2 -1 1 2 However, an uncharacterized siderophore from the related FRE E(uM) bacterium Aeromonas hydrophila (R. Byers, personal com- FIG. 6. Kinetics of hemin and protoporphyrin binding to purified munication) did not compete for porphyrin binding to A' A protein ofA. salmonicida. The expenment was as described in the cells (unpublished data). Another possibility is that the legend to Fig. 3 except that hemin (0) and protoporphyrin (0) were A-layer strips iron from lactoferrin or transferrin using the ligands. protoporphyrin IX bound to A protein to sequester ferric ion. Strains of N. meningitidis (18, 19) and A. salmonicida (11) have been shown to require cell contact with an iron chelator such as transferrin. However, unlike most iron- The A protein component of the A-layer appears to be the transport systems in the system described here A protein is CR-binding component since A- strains are unable to bind not derepressed by iron limitation. It is equally likely that CR at low concentrations and purified A protein binds CR virulent bacterial strains such as A. salmonicida either have with high affinity. In outer membrane protein gel profiles of a unique heme requirement or scavenge heme from their A- strains, only the A protein is missing (11). It is well hosts during invasion. Whatever the role of porphyrin bind- known that aromatic dye molecules tend to bind to the ing, it is worthwhile to note that there exists a high degree of active-site regions of some globular proteins (5). This may be conservation in the primary structure of A proteins from a the result of nonspecific interactions, such as general hydro- variety of A. salmonicida strains of diverse properties, phobicity, or by some specific arrangement of charge origins, and pathogenesis (12), underlining the importance of groups, hydrophobic regions, and hydrogen bond donors this function. and acceptors. Due to their similarity to the biological heterocyclic bases, dyes have been used as analogs to nucleotides, dinucleotides, acetyl coenzyme A, and even folic acid (1, 10, 21). Indeed, CR has been shown to interact CONGO RED with the nucleotide binding site of several dehydrogenases (3). The A. salmonicida A protein appears to possess a unique binding site for the porphyrins, protoporphyrin IX and hemin. This site is effectively mimicked by CR and is the first instance we know of in which a commercial dye has been shown to mimic a porphyrin binding site. It is surprising, at first, that CR is so effective and specific. However, when molecular models are constructed there emerges a striking similarity in structure between CR and protoporphyrin (Fig. 7). The important features are a hydrophobic edge, and two appropriately spaced single-charge residues per molecule. Also, the overall dimensions of the two molecules are approximately the same. There is likely a hydrophobic binding domain on A protein. Thus hematoporphyrin, in which the vinyl groups of protoporphyrin IX are substituted with hydroxyl groups, does not bind to A' cells or to protein A and does not compete with CR or porphyrin binding. Also, the number and probably the positioning of the sulfonic acid residues is important since Trypan blue with four sulfonic acid residues and Evans blue with three were only weak competitive inhibitors of CR binding. Estimates of the num- ber of ligand binding sites by Scatchard analysis of the PROTOPORPHYRIN IX binding data for soluble protein A is confusing. While hemin FIG. 7. Structures of CR and protoporphyrin IX, arranged to binds at one molecular site per monomer, CR binds at two maximize similar conformations. All double bonds are not included and protoporphyrin IX at approximately four. It is possible for sake of simplicity. In the foreground is the common hydrophobic that two molecules of CR fit the binding site, perhaps as a edge and in the background are the two negative-charge residues stacked dimer, but it is difficult to rationalize how four common to both molecules. 1336 KAY ET AL. J. BACTERIOL.

ACKNOWLEDGMENTS 11. Kay, W. W., J. T. Buckley, E. E. Ishiguro, B. M. Phipps, J. P. L. Monette, and T. J. Trust. 1981. Purification and This work was supported by a Strategic Grant from the Natural disposition of a surface protein associated with virulence of Sciences and Engineering Research Council of Canada. Aeromonas salmonicida. J. Bacteriol. 147:1077-1084. We are grateful to G. R. MacCharles for excellent technical 12. Kay, W. W., B. M. Phipps, E. E. Ishiguro, R. W. Olafson, and assistance and G. A. Poulton, R. J. Belland, and H. M. Atkinson for T. J. Trust. 1984. Surface layer virulence A-proteins from helpful suggestions and discussions. Aeromonas salmonicida strains. Can. J. Biochem. Cell Biol. 62:1064-1071. LITERATURE CITED 13. Maurelli, A. T., B. Blackmon, and R. Curtiss III. 1984. Loss of 1. Baird, J. K., R. F. Sherwood, R. J. G. Carr, and A. Atkinson. pigmentation in Shigella flexneri 2a is correlated with loss of 1976. Enzyme purification by substrate elution chromatography virulence and virulence-associated plasmid. Infect. Immun. from procion dye-polysaccharide complexes. FEBS Lett. 43:397-401. 70:61-66. 14. Munn, C. B., E. E. Ishiguro, W. W. Kay, and T. J. Trust. 1982. 2. Chart, H., and T. J. Trust. 1983. Acquisition of iron by Role of surface components in serum resistance of virulent Aeromonas salmonicida. J. Bacteriol. 156:758-764. Aeromonas salmonicida. Infect. Immun. 36:1069-1075. 3. Edwards, R. A., and R. W. Woody. 1979. Spectroscopic studies 15. Payne, S. M., and R. A. Finkelstein. 1977. Detection and of Cibacron blue and Congo red bound to dehydrogenases and differentiation of iron-responsive avirulent mutants on Congo kinases. Evaluation of dyes as probes of the dinucleotide fold. red agar. Infect. Immun. 18:94-98. Biochemistry 18:5197-5204. 16. Phipps, B. M., T. J. Trust, E. E. Isbiguro, and W. W. Kay. 1983. 4. Evenberg, D., and B. Lugtenberg. 1982. Cell surface of the fish Purification and characterization of the cell surface virulent A pathogenic bacterium Aeromonas salmonicida. II. Purification protein from Aeromonas salmonicida. Biochemistry 22:2934- and characterization of a major cell envelope protein related to 2939. auto-agglutination, adhesion and virulence. Biochim. Biophys. 17. Prpic, J. K., R. M. Robins-Browne, and R. B. Davey. 1983. Acta 684:249-254. Differentiation between virulent and avirulent Yersinia entero- 5. Glazer, A. N. 1970. On the prevalence of "nonspecific" binding colitica isolates by using Congo red agar. J. Clin. Microbiol. at the specific binding sites of globular proteins. Proc. Natl. 18:486-490. Acad. Sci. USA 65:1057-1063. 18. Simonson, C., D. Breener, and I. W. De Voe. 1982. Expression 6. Herman, R. L. 1968. Fish furunculosis 1952-1966. Trans. Am. of a high-affinity mechanism for acquisition of transferrin iron Fish. Soc. 97:221-230. by Neisseria meningitidis. Infect. Immun. 36:107-113. 7. Ishiguro, E. E., T. Ainsworth, T. J. Trust, and W. W. Kay. 1985. 19. Simonson, C., T. Trivett, and I. W. De Voe. 1981. Energy- Congo red agar, a differential medium for Aeromonas salmoni- independent uptake of iron from citrate by isolated outer cida strains, detects the presence of the cell surface array membranes of Neisseria meningitidis. Infect. Immun. involved in virulence. J. Bacteriol. 164:1233-1237. 31:547-553. 8. Ishiguro, E. E., W. W. Kay, T. Ainsworth, J. B. Chamberlain, 20. SurgalHa, M. J., and E. D. Beesley. 1969. Congo red agar plating R. A. Austen, J. T. Buckley, and T. J. Trust. 1981. Loss of medium for detecting pigmentation of Pasteurella pestis. Appl. virulence during culture of Aeromonas salmonicida at high Microbiol. 18:834-837. temperature. J. Bacteriol. 148:333-340. 21. Thompson, S. T., and E. Stellwagen. 1976. Binding of Cibacron 9. Jackson, S., and T. W. Burrows. 1956. The pigmentation of Blue F36A to proteins containing the dinucleotide fold. Proc. Pasteurella pestus on a defined medium containing haemin. Br. Natl. Acad. Sci. USA 73:361-365. J. Exp. Pathol. 37:570-576. 22. Trust, T. J., W. W. Kay, and E. E. Ishiguro. 1983. Cell surface 10. Jacobsberg, L. B., E. R. Kantrowitz, and W. N. Lipscomb. 1975. hydrophobicity and macrophage association of Aeromonas Interaction oftetraiodofluorescein with aspartate transcarbamyl- salmonicida. Curr. Microbiol. 9:315-318. ase and its isolated catalytic and regulatory subunits. J. Biol. 23. Udey, L. R., and J. L. Fryer. 1978. Immunization of fish with Chem. 250:9238-9249. bacterins ofAeromonas salmonicida. Mar. Fish. Rev. 40:12-17.