Abstract: Congo red (CR) binding, monitored by characteristic yellow-green birefringence under crossed polarization has been used as a diagnostic test for the presence of amyloid in tissue sections for several decades. This assay is also widely used for the characterization of in vitro amyloid fibrils. In order to probe the structural specificity of Congo red binding to amyloid fibrils we have used an induced circular dichroism (CD) assay. Amyloid fibrils from insulin and the variable domain of Ig light chain demonstrate induced CD spectra upon binding to Congo red. Surprisingly, the native conformations of insulin and Ig light chain also induced Congo red circular dichroism, but with different spectral shapes than those from fibrils. In fact, a wide variety of native proteins exhibited induced CR circular dichroism indicating that CR bound to representative proteins from different classes of secondary structure such as α (citrate synthase), α + β (lysozyme), β (concavalin A), and parallel β-helical proteins (pectate lyase). Partially folded intermediates of apomyoglobin induced different Congo red CD bands than the corresponding native conformation, however, no induced CD bands were observed with unfolded protein. Congo red was also found to induce oligomerization of native proteins, as demonstrated by covalent cross-linking and small angle x-ray scattering. Our data suggest that Congo red is sandwiched between two protein molecules causing protein oligomerization. The fact that Congo red binds to native, partially folded conformations and amyloid fibrils of several proteins shows that it must be used with caution as a diagnostic test for the presence of amyloid fibrils in vitro. Congo red (CR) binding, monitored by characteristic yellow-green birefringence under crossed polarization has been used as a diagnostic test for the presence of amyloid in tissue sections for several decades. This assay is also widely used for the characterization of in vitro amyloid fibrils. In order to probe the structural specificity of Congo red binding to amyloid fibrils we have used an induced circular dichroism (CD) assay. Amyloid fibrils from insulin and the variable domain of Ig light chain demonstrate induced CD spectra upon binding to Congo red. Surprisingly, the native conformations of insulin and Ig light chain also induced Congo red circular dichroism, but with different spectral shapes than those from fibrils. In fact, a wide variety of native proteins exhibited induced CR circular dichroism indicating that CR bound to representative proteins from different classes of secondary structure such as α (citrate synthase), α + β (lysozyme), β (concavalin A), and parallel β-helical proteins (pectate lyase). Partially folded intermediates of apomyoglobin induced different Congo red CD bands than the corresponding native conformation, however, no induced CD bands were observed with unfolded protein. Congo red was also found to induce oligomerization of native proteins, as demonstrated by covalent cross-linking and small angle x-ray scattering. Our data suggest that Congo red is sandwiched between two protein molecules causing protein oligomerization. The fact that Congo red binds to native, partially folded conformations and amyloid fibrils of several proteins shows that it must be used with caution as a diagnostic test for the presence of amyloid fibrils in vitro. Congo red circular dichroism 3,3′-dithiobis(sulfosuccinimidylpropionate) dithiobis(succinimidylpropionate) initials of the patient with light chain amyloidosis whose sequence information was used to generate synthetic recombinant proteins attenuated total reflectance Fourier transform infrared spectroscopy polyacrylamide gel electrophoresis small angle x-ray scattering interleukin-2 Stokes radius In the 1920s Benhold (1Benhold H. Muenchen. Med. Wochenschr. 1922; 69: 1537Google Scholar) and Divry (2Divry P. J. Neurol. Psychiatr. 1927; 27: 643Google Scholar) established that Congo red bound to amyloid in tissue sections and demonstrated its characteristic yellow-green birefringence under crossed polarizers. Since then this birefringence has been used as a diagnostic for amyloid fibrils. The birefringence assay is not a simple one, for example, the tissue sections need to be of a required thickness to show birefringence, reviewed by Elghetany and Saleem (3Elghetany M.T. Saleem A. Stain Technol. 1988; 63: 201-212Crossref PubMed Scopus (137) Google Scholar) and Westermark et al. (4Westermark G.T. Johnson K.H. Westermark P. Methods Enzymol. 1999; 309: 3-25Crossref PubMed Scopus (237) Google Scholar). Congo red binding is not specific for amyloid in the tissue sections, but the assays are performed under extreme conditions with 50–80% ethanol, high salt and alkaline pH conditions to yield binding to amyloid (5Puchtler H. Sweat F. Levine M. J. Histochem. Cytochem. 1962; 10: 355Crossref Google Scholar). Despite these extreme conditions, binding to collagen fibers and cytoskeletal proteins results in false-positive results (4Westermark G.T. Johnson K.H. Westermark P. Methods Enzymol. 1999; 309: 3-25Crossref PubMed Scopus (237) Google Scholar). Due to the difficulties with the birefringence assay for in vitro amyloid fibrils, Klunk and co-workers (6Klunk W.E. Pettegrew J.W. Abraham D.J. J. Histochem. Cytochem. 1989; 37: 1293-1297Crossref PubMed Scopus (113) Google Scholar, 7Klunk W.E. Pettegrew J.W. Abraham D.J. J. Histochem. Cytochem. 1989; 37: 1273-1281Crossref PubMed Scopus (566) Google Scholar) have developed simpler filtration based assays followed by measuring the concentration of free Congo Red to quantify dye binding. The filtration assays would not detect CR1 bound to soluble monomers or oligomers as they are not large enough to be trapped by 0.2-μm filters. Large particles, such as amyloid fibrils, are retained on the filters accounting for the loss of free dye molecules, whereas any native protein molecules bound to CR would pass through the filter pores. Thus the filtration assay is not affected by possible binding of CR to native soluble conformations of protein. Benditt and co-workers (8Benditt E.P. Eriksen N. Berglund C. Proc. Natl. Acad. Sci. U. S. A. 1970; 66: 1044-1051Crossref PubMed Scopus (66) Google Scholar) analyzed spectral probes such as absorbance red shift, optical rotary dispersion, and circular dichroism (CD) to describe the Cotton effect (9Cotton A. Ann. Chim. Phys. 1896; 8: 347Google Scholar) responsible for birefringence. They used human albumin, poly-l-lysine with different conformations, and amyloid fibrils as substrates for binding of Congo red. The random coil conformation of poly-l-lysine did not show spectral changes, but both helical and β conformations induced Congo red CD bands as well as optical rotary dispersion spectra similar to those observed with the amyloid fibril samples. Edwards and Woody (10Edwards R.A. Woody R.W. Biochem. Biophys. Res. Commun. 1977; 79: 470-476Crossref PubMed Scopus (43) Google Scholar, 11Edwards R.A. Woody R.W. Biochemistry. 1979; 18: 5197-5204Crossref PubMed Scopus (85) Google Scholar) demonstrated that induced circular dichroism can be used as a probe for Cibacron blue and Congo red bound to dehydrogenases such as liver alcohol dehydrogenase, yeast alcohol dehydrogenase, lactic dehydrogenase, and kinases including phosphoglycerate kinase and porcine adenylate kinase. Edwards and Woody (11Edwards R.A. Woody R.W. Biochemistry. 1979; 18: 5197-5204Crossref PubMed Scopus (85) Google Scholar) believed that Congo red bound to the coenzyme-binding sites of the enzymes based on the similarities between dye and coenzyme structures. Congo red has also been shown to bind other native proteins including cellular prion protein (12Caughey B. Brown K. Raymond G.J. Katzenstein G.E. Thresher W. J. Virol. 1994; 68: 2135-2141Crossref PubMed Google Scholar), elastin (13Kagan H.M. Hewitt N.A. Franzblau C. Biochim. Biophys. Acta. 1973; 322: 258-268Crossref PubMed Scopus (13) Google Scholar), RNA polymerase (14Woody A.Y. Reisbig R.R. Woody R.W. Biochim. Biophys. Acta. 1981; 655: 82-88Crossref PubMed Scopus (16) Google Scholar), and human prostatic phosphatase (15Kuciel R. Mazurkiewicz A. Acta Biochim. Pol. 1997; 44: 645-657Crossref PubMed Scopus (6) Google Scholar). Structure FT1 of Congo red suggests that binding to protein could occur through a combination of both hydrophobic and electrostatic interactions. An additional complication is that CR has been reported to form linear ribbon-like micelles (16Attwood T.K. Lydon J.E. Hall C. Tiddy G.J.T. Liquid Crystals. 1990; 7: 657-668Crossref Scopus (73) Google Scholar,17Skowronek M. Stopa B. Koniczny L. Rybarska J. Piekarska B. Szneler E. Bakaalarski G. Roterman I. Biopolymers. 1998; 46: 267-281Crossref Google Scholar). In an attempt to understand the binding specificity of Congo red we have used induced circular dichroism as an assay for binding of CR to native proteins, partially folded protein conformations, and amyloid fibrils, using native proteins from a variety of different secondary structure classes. The results suggest a mechanism of binding of Congo red to native proteins involving the intercalation of Congo red between protein molecules leading to oligomerization of the protein. Fibril were grown in vitro with a 0.5 mg/ml SMA (a recombinant amyloidogenic variable domain of Ig light chain made by Stevens and co-workers (18Stevens P.W. Raffen R. Hanson D.K. Deng Y.L. Berrios-Hammond M. Westholm F.A. Murphy C. Eulitz M. Wetzel R. Solomon A. Protein Sci. 1995; 4: 421-432Crossref PubMed Scopus (107) Google Scholar)) and 1 mg/ml bovine insulin solution at pH 2 in 20 to 50 mm HCl and 100 mm NaCl, that were agitated using a magnetic stirrer in a 37 °C incubator for a day. CD spectra were collected between 650 and 300 nm, with 1-nm step size, 10 s averaging time, in an Aviv 62 DS spectropolarimeter. Induced CD spectra were obtained using a split quartz cell with each compartment having a path length of 5 mm and total path length of 10 mm. The protein and Congo red solutions of twice the final concentration were added in each compartment to obtain a control spectrum. Induced circular dichroism was only observed upon mixing the Congo red and the protein solutions before collecting the spectrum again. The induced CD assay was performed at pH 7.5 using a 1-cm path length rectangular cell with final Congo red concentration of 30 to 40 μm with protein concentration of 0.1 to 0.2 mg/ml. At pH 2.0 the induced CD assay was performed using a 10-cm path length circular CD cell with final Congo red concentrations of 3–4 μm as the solubility of Congo red was much lower at pH 2. Induced CD measurements for fibril samples were obtained with a 10-cm path length circular cell using very dilute samples (80 to 400 nm), as at higher concentrations the fibril samples precipitated as red particles. Hydrated (H2O) thin film spectra were collected using a Nicolet 800 FTIR spectrometer equipped with a liquid nitrogen-cooled MCT detector and purged with dry air. All samples were scanned in an out-of-compartment horizontal ATR accessory (SPECAC) with a high throughput 73 × 10 × 6 mm, 45° trapezoidal germanium crystal (the internal reflectance element). To collect spectra protein samples with and without Congo red were applied (40 μl of 1 mg/ml protein solution and 500 μm Congo red) and dried while being spread constantly with a spatula. Data processing was done with GRAMS32 (Galactic Industries) and SAFAIR software as previously described (19Khurana R. Fink A.L. Biophys. J. 2000; 78: 994-1000Abstract Full Text Full Text PDF PubMed Scopus (107) Google Scholar). Water vapor components were subtracted until the region between 1800 and 1700 wavenumbers was featureless. Equation 1 was used to fit the data obtained at 40 μm Congo red and varying the protein concentration from 0 to 20 μm and measuring the CD spectra of protein bound dye molecules. C=nCm[L][L]+KdEquation 1 Where C is the change in circular dichroism signal at a particular wavelength measured upon addition of protein ([L]) in varying concentration, Cm is the maximal CD change obtained by varying the protein concentration at a constant Congo red concentration, n is the number of binding sites for Congo red on the protein, and Kdis the dissociation constant for binding of Congo red to protein. 10 μmDTSSP (3,3′-dithiobis(sulfosuccinimidylpropionate)), a bifunctional, cleavable cross-linking reagent, was added to the protein solution in the absence and presence of Congo red and incubated for 1 h. DTSSP (Pierce) is a sulfonated derivative of DSP (dithiobis(succinimidylpropionate)) that is water soluble, unlike DSP, and, like DSP, can be cleaved under reducing conditions (the presence of dithiothreitol or β-mercaptoethanol). The cross-linked protein was then precipitated using 10% acetic acid, and spun in a centrifuge. The pellet was resuspended in SDS loading buffer and separated on either a nonreducing or a reducing 8–25% SDS-PAGE gel. Small angle x-ray scattering (SAXS) measurements were done using Beam Line 4-2 at the Stanford Synchrotron Radiation Laboratory. X-ray energy was selected at 8980 eV (Cu edge) by a pair of Mo/B4C multilayer monochromator crystals. Scattering patterns were recorded by a linear position-sensitive proportional counter, which was filled with an 80% Xe, 20% CO2 gas mixture. Scattering patterns were normalized by incident x-ray fluctuations, which were measured with a short length ion chamber before the sample. The sample-to-detector distance was calibrated to be 230 cm, using a cholesterol myristate sample. The measurements were performed in a 1.3-mm path length observation static cell with 25-μm mica windows. To avoid radiation damage of the protein samples during the SAXS measurements, the scavengerN-tert-butyl-α-(4-pyridyl)-nitrone-N′-oxide was added to a final concentration of 10 mm. Background measurements were performed before and after each protein measurement and then averaged for background subtraction. All SAXS measurements were performed at 23 ± 1 °C. The radius of gyration (Rg ) was calculated according to the Guinier approximation (41Glatter O. Kratky O. Small Angle X-ray Scattering. Academic Press, New York1982Google Scholar),lnI(Q)=ln I(0)−Rg2Q2/3Equation 2 where Q is the scattering vector given byQ = (4π sinθ)/λ (2θ is the scattering angle, and λ is the wavelength of x-ray).I(0), the forward scattering amplitude, is proportional ton·ρc 2·V 2, where n is the number of scatterers (protein molecules) in solution; ρc is the electron density difference between the scatter and the solvent; andV is the volume of the scatter. This means that the value of forward-scattered intensity, I(0), is proportional to the square of the molecular weight of the molecule (41Glatter O. Kratky O. Small Angle X-ray Scattering. Academic Press, New York1982Google Scholar). Thus,I(0) for a pure N-mer sample will therefore be N-fold that for a sample with the same number of monomers since each N-mer will scatterN 2 times as strongly as monomer, but in this case the number of scattering particles (N-mers) will be N times less than that in the pure monomer sample. The relationship between RS (Stokes radius) andRg (radius of gyration) is quite sensitive to the shape and compactness of protein, asRS /Rg = P 1/3[5/(P 2 + 2)]1/2 , where P = a/b, anda and b are the semiaxis of the revolution of the ellipsoid and the equatorial radius of the ellipsoid, respectively (20Luzatti V. Witz J. Nikolaieff A. J. Mol. Biol. 1961; 3: 367-372Crossref Scopus (106) Google Scholar, 21Kumosinski T.F. Pessen H. Methods Enzymol. 1985; 117: 154-182Crossref PubMed Scopus (39) Google Scholar, 22Shi L. Kataoka M. Fink A.L. Biochemistry. 1996; 35: 3297-3308Crossref PubMed Scopus (72) Google Scholar). In order to estimate the RS value for the oligomeric form of β-lactoglobulin the following approach was used. SAXS data show that in the presence of 1 mm Congo red β-lactoglobulin forms 28-mers, which corresponds to particles with the molecular mass of 515,200 Da (= 28 × 18, 400 Da). The value of RS of a globular protein with the molecular mass of 515,200 Da is 71.4 Å. This was calculated from an empirical equation log = 0.369·log(M) − 0.254 (23Uversky V.N. Biochemistry. 1993; 32: 13288-13298Crossref PubMed Scopus (447) Google Scholar) based on the intrinsic viscosity data (24Tanford C. Adv. Protein Chem. 1968; 23: 121-282Crossref PubMed Scopus (2426) Google Scholar). Here RSN is the Stokes radius of native (N) protein. It is known that for an ideal spherical particlea = b, P = a/b = 1.00, andRS /Rg = 1.29, whereas for globular proteins the average value of this ratio is about 1.25 (25Gast K. Damaschun H. Misselwitz R. Muller-Frohne M. Zirwer D. Damaschun G. Eur. Biophys. J. 1994; 23: 297-305Crossref PubMed Scopus (98) Google Scholar). Amyloid fibrils made in vitro from purified proteins such as bovine insulin and Ig light chain (SMA), were tested for binding to Congo red using the apple-green birefringence under crossed polarization (data not shown) and induced CD signal (Fig.1 A). Low concentrations of fibril solutions were used for induced CD measurements, as higher concentrations yielded red precipitates. Both fibril samples showed major induced CD bands with positive maxima in the vicinity of 570 nm and negative maxima in the vicinity of 500–525 nm. The spectrum for the induced CD of insulin fibrils was red-shifted relative to that for the light chain fibrils. The shape of the induced CD bands for Congo red bound to fibrils is different from that described by Bendittet al. (8Benditt E.P. Eriksen N. Berglund C. Proc. Natl. Acad. Sci. U. S. A. 1970; 66: 1044-1051Crossref PubMed Scopus (66) Google Scholar) for tissue-extracted amyloid. The differences are most likely due to "contaminants" such as glycosaminoglycans, serum amyloid protein etc. present in the ex vivo extracts. Native insulin and native SMA show induced Congo red CD bands that differ in shape and intensities from the CD bands induced upon binding to the fibrils (Fig. 1 B). Instead of a positive and a negative peak observed for the corresponding amyloid fibrils the native proteins show a broad positive band between 500 and 600 nm, that possibly has several components. The peculiar CD band shape with a maxima and a minima is related to the special nature of birefringence (26Taylor D.L. Allen R.D. Benditt E.P. J. Histochem. Cytochem. 1974; 22: 1105-1112Crossref PubMed Scopus (43) Google Scholar) observed for amyloid fibrils upon binding to Congo red. The positive peaks induced by the native proteins were 10-fold smaller in intensity compared with their corresponding amyloid fibrils. A probable explanation is that amyloid fibrils have more binding sites for Congo red than the native protein. The intensity of the induced Congo red CD bands upon binding to native SMA was much smaller than that for native insulin. It is possible that the hexameric insulin (under native conditions) has more Congo red binding sites, leading to a larger induced CD signal, compared with native SMA. Native proteins such as lysozyme (α + β), concavalin A (β), and citrate synthase (all α) (Fig.2) showed induced Congo red CD signals, thus revealing binding of Congo red. A number of other native proteins from different structural classes, interleukin-2 (all α), malate dehydrogenase (α/β), β-lactoglobulin, and apomyoglobin (α) also showed induced Congo red CD signals (data not shown). Other than the observation that the major induced CD bands were in the vicinity of 525 nm, and quite broad, no consistent patterns were apparent. Circular dichroism bands in the visible region were not observed for protein alone, or Congo red alone. This was demonstrated by using split cells where one side of the cell contains the protein solution and the other side of the cell contains Congo red solution, no CD bands were observed until the protein and Congo red solutions were mixed together. Interestingly, no Congo red CD bands were induced in the presence of unfolded protein (e.g. acid unfolded apomyoglobin; Fig. 5), amorphous aggregates of P22 tail spike protein, or Ig light chain variable domain, or inclusion bodies. These results are supported by a report that there are no Congo red CD or optical rotary dispersion signals with the random coil conformation of poly-l-lysine (8Benditt E.P. Eriksen N. Berglund C. Proc. Natl. Acad. Sci. U. S. A. 1970; 66: 1044-1051Crossref PubMed Scopus (66) Google Scholar). Whereas the induced CD bands indicate a specific orientation of Congo red, which is assumed to be due to binding of the dye to the protein, the absence of an induced CD band does not necessarily mean the absence of dye binding, but rather the lack of specific orientation responsible for the induced CD bands.Figure 5Native apomyoglobin shows an induced CD band upon binding to Congo red (—). Acidic unfolded apomyoglobin at pH 2 with no salt (···) shows no CD bands. Acidic partially folded intermediates stabilized at pH 2 in the presence of salt (—··—··) and at pH 4 (- - - -) show induced CD bands that are different from the ones obtained from native apomyoglobin.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Given that unfolded protein did not induce Congo red CD bands, we conclude that significant secondary structure, and probably a collapsed conformation is required to induce a specific orientation of Congo red molecules responsible for CD bands. Indeed Turnell and Finch (27Turnell W.G. Finch J.T. J. Mol. Biol. 1992; 227: 1205-1223Crossref PubMed Scopus (154) Google Scholar) observed a Congo red molecule intercalated between the antiparallel β-strands of two insulin molecules, using x-ray diffraction of a crystalline complex. However, our results suggest that Congo red binding is not limited to β-sheet proteins; rather, all classes of proteins, α-helical, β, and α/β, also bind Congo red resulting in the observed induced CD bands. Congo red binding to dehydrogenases of the α/β class has been reported previously, and is thought to reflect specific interactions with coenzyme-binding sites, due to structural similarities between CR and the coenzyme (10Edwards R.A. Woody R.W. Biochem. Biophys. Res. Commun. 1977; 79: 470-476Crossref PubMed Scopus (43) Google Scholar). β-Helical proteins also induced Congo red CD bands (Fig.3). Interestingly, the right-handed β-helical proteins including pectate lyase (28Yoder M.D. Keen N.T. Jurnak F. Science. 1993; 260: 1503-1507Crossref PubMed Scopus (398) Google Scholar), and p22 tailspike protein (29Steinbacher S. Seckler R. Miller S. Steipe B. Huber R. Reinemer P. Science. 1994; 265: 383-386Crossref PubMed Scopus (277) Google Scholar) induced different Congo red CD bands, with positive ellipticity, compared with the left-handed β-helical protein LpxA (30Raetz C.R. Roderick S.L. Science. 1995; 270: 997-1000Crossref PubMed Scopus (294) Google Scholar), which induced two negative Congo red CD bands. This suggests that the positive or negative CD bands may reflect the underlying chirality of the CR-binding site. In view of the fact that the induced CD spectra demonstrate that proteins from all classes of secondary structures bind Congo red, Congo red binding is clearly not restricted to the crossed-β structures present in amyloid fibrils. Benditt and co-workers (8Benditt E.P. Eriksen N. Berglund C. Proc. Natl. Acad. Sci. U. S. A. 1970; 66: 1044-1051Crossref PubMed Scopus (66) Google Scholar) have also shown binding of Congo red to both α and β conformations of poly-l-lysine. It is possible that the shape of the induced Congo red CD bands has specific clues as to which secondary structures in the proteins Congo red dye is bound, but more work is needed to understand these distinctions. The lack of correlation between the shape of the induced CD band and the protein secondary structure suggests that binding sites for CR in individual proteins are more related to their specific environment rather than to a particular type of secondary structure. Since it has been suggested that Congo red bound specifically to crossed β structures present in amyloid fibrils (31Glenner G.G. Eanes E.D. Page D.L. J. Histochem. Cytochem. 1972; 20: 821-826Crossref PubMed Scopus (156) Google Scholar), or to β-sheets in native proteins (32Roterman I. Rybarska J. Koniczny L. Skowronek M. Stopa B. Piekarska B. Belarski G. Comput. Chem. 1998; 22: 61-70Crossref Scopus (39) Google Scholar), and our results indicated that all α proteins also bind Congo red, we sought to confirm that Congo red binding does not induce changes in secondary structure,e.g. from α to β. To test this we collected infrared spectra of interleukin-2 (IL-2) (a four-helix bundle protein) in the absence and presence of Congo red (Fig.4 A). The spectrum of Congo red alone was featureless in the amide I and II regions where proteins show specific conformational-sensitive bands, and no secondary structure changes were observed in the IL-2 FTIR spectrum upon binding to Congo red (Fig. 4 A). Lysozyme, an α/β protein, forms a red precipitate upon binding to Congo red. To test if this precipitate involves formation of new β-structure with low wavenumber amide I peaks, as observed for many protein aggregates, 2R. Khurana, K. A. Oberg, S. Sheshadri, L. Shi, and A. L. Fink, submitted for publication. we examined it with ATR-FTIR. The FTIR spectra of soluble free lysozyme and precipitated Congo red-bound lysozyme are compared in Fig.4 B. No significant increase in β-structure was observed in the precipitated Congo red-bound lysozyme compared with native lysozyme. The minor differences observed between 1610 and 1580 cm−1 are probably due to interaction of Congo red with specific side chains, as these bands have significant contributions from side chains and are not indicative of protein secondary structure changes. Thus it is clear that binding of CR does not result in induction of β-structure. 1,8-Anilinonaphthalene sulfonate (33Stryer L. J. Mol. Biol. 1965; 13: 482-495Crossref PubMed Scopus (1329) Google Scholar) and its dimer bis-1,8-anilinonaphthalene sulfonate (34Takashi R. Tonomura Y. Morales M.F. Proc. Natl. Acad. Sci. U. S. A. 1977; 74: 2334-2338Crossref PubMed Scopus (50) Google Scholar) are commonly used as probes of hydrophobic regions in native and partially folded proteins. Due to the similarity of the structures of Congo red and bis-1,8-anilinonaphthalene sulfonate we decided to also test differential binding of Congo red to native and partially folded conformations. Apomyoglobin exists in its native conformation at pH 7, the acid unfolded state at pH 2 in the absence of salts and as a partially folded intermediate at pH 4 (35Barrick D. Baldwin R.L. Biochemistry. 1993; 32: 3766-3790Crossref Scopus (232) Google Scholar, 36Griko Y.V. Privalov P.L. Venyaminov S.Y. Kutyshenko V.P. J. Mol. Biol. 1988; 202: 127-138Crossref PubMed Scopus (309) Google Scholar), and at pH 2 in the presence of salt (37Goto Y. Fink A.L. J. Mol. Biol. 1990; 214: 803-805Crossref PubMed Scopus (167) Google Scholar). No induced Congo red CD bands were observed for the acid unfolded form, but the native (pH 7) and partially folded conformations at pH 4 and 2 with 500 mm KCl showed different induced spectra (Fig. 5). Binding of the dye to the native conformation of apomyoglobin is not surprising, since the protein is known to bind a variety of hydrophobic molecules in the vacant heme-binding site. Increased binding of Congo red has been observed for the molten globule intermediate compared with the native conformation for human prostatic phosphatase (7 to 8 molecules of Congo red bind to the intermediate conformation as opposed to 1.6 dye molecules to the native protein) (15Kuciel R. Mazurkiewicz A. Acta Biochim. Pol. 1997; 44: 645-657Crossref PubMed Scopus (6) Google Scholar). Consequently it appears that CR-binding sites are present in partially folded intermediates. This is not surprising since such intermediates are known to have exposed hydrophobic patches and bind hydrophobic molecules. Thus it is likely that different binding sites for CR may exist in native and partially folded intermediate states. The number of molecules of Congo red bound per molecule of native β-lactoglobulin was estimated from an analysis of the data obtained by varying the concentration of protein from 0 to 20 μm while keeping the concentration of Congo red constant at 40 μm (Fig.6 A). The ellipticity at 530 and 450 nm were plotted against β-lactoglobulin concentration (Fig.6 B) and the data were fitted to Equation 1. The analysis showed that 1.52 ± 0.05 molecules of Congo red bound per molecule of β-lactoglobulin, similar to the value obtained by Kuciel and Mazurkiewicz (15Kuciel R. Mazurkiewicz A. Acta Biochim. Pol. 1997; 44: 645-657Crossref PubMed Scopus (6) Google Scholar) for human prostatic phosphatase. Since we could not measure the concentration of unbound protein for technical reasons, a more accurate analysis involving Scatchard plots was not possible. A likely mechanism would involve three protein molecules with two Congo red molecules intercalated between them. The interaction of Congo red with protein molecules may involve a complex of multiple protein molecules with intercalated Congo red molecules. Intercalation as a mechanism of binding of Congo red molecules between peptide chains has also been suggested by Stopa and co-workers (38Stopa B. Gorny M. Konieczny L. Piekarska B. Rybarska J. Skowronek M. Roterman I. Biochimie ( Paris ). 1998; 80: 963-968Crossref PubMed Scopus (54) Google Scholar). To test if Congo red binding involved oligomerization of protein molecules we added DTSSP, a cleavable cross-linker, to the protein solution in the absenc