Comparative protein profiling of serum and plasma using an antibody suspension bead array approach

In the pursuit towards a systematic analysis of human diseases, array‐based approaches within antibody proteomics offer high‐throughput strategies to discover protein biomarkers in serum and plasma. To investigate the influence of sample preparation on such discovery attempts, we report on a systematic effort to compare serum and plasma protein profiles determined with an antibody suspension bead array. The intensity levels were used to define protein profiles and no significant differences between serum and plasma were observed for 79% of the 174 antibodies (targeting 156 proteins). By excluding 36 antibodies giving rise to differential intensity levels, cluster analysis revealed donor‐specific rather than preparation‐dependent grouping. With a cohort from a clinically relevant medical condition, the metabolic syndrome, the influence of the sample type on a multiplexed biomarker discovery approach was further investigated. Independent comparisons of protein profiles in serum and plasma revealed an antibody targeting ADAMTSL‐4, a protein that would qualify to be studied further in association with the condition. In general, the preparation type had an impact on the results of the applied antibody suspension bead array, and while the technical variability was equal, plasma offered a greater biological variability and allowed to give rise to more discoveries than serum.     

Serum Antibody Repertoire Profiling Using In Silico Antigen Screen

Serum antibodies are valuable source of information on the health state of an organism. The profiles of serum antibody reactivity can be generated by using a high throughput sequencing of peptide-coding DNA from combinatorial random peptide phage display libraries selected for binding to serum antibodies. Here we demonstrate that the targets of immune response, which are recognized by serum antibodies directed against sequential epitopes, can be identified using the serum antibody repertoire profiles generated by high throughput sequencing. We developed an algorithm to filter the results of the protein database BLAST search for selected peptides to distinguish real antigens recognized by serum antibodies from irrelevant proteins retrieved randomly. When we used this algorithm to analyze serum antibodies from mice immunized with human protein, we were able to identify the protein used for immunizations among the top candidate antigens. When we analyzed human serum sample from the metastatic melanoma patient, the recombinant protein, corresponding to the top candidate from the list generated using the algorithm, was recognized by antibodies from metastatic melanoma serum on the western blot, thus confirming that the method can identify autoantigens recognized by serum antibodies. We demonstrated also that our unbiased method of looking at the repertoire of serum antibodies reveals quantitative information on the epitope composition of the targets of immune response. A method for deciphering information contained in the serum antibody repertoire profiles may help to identify autoantibodies that can be used for diagnosing and monitoring autoimmune diseases or malignancies.     

The B10.A mouse B cell response to pigeon cytochrome c is directed against the same area of the protein that is recognized by B10.A T cells in association with the Ek beta:Ek alpha Ia molecule

An analysis of the fine specificities of the primary and hyperimmune antibody responses of B10.A mice to pigeon cytochrome c showed that both were qualitatively very similar. Small amounts of antibody appeared to be directed against the regions of serine 15 and/or glutamic acid 44. The remaining antibodies (greater than 70%) bound to the same complex topographic determinant (including residues 3, 103, and 104) on the back surface of pigeon cytochrome c which had been found to dominate the rabbit antibody response to this protein, and to be involved in Ia-restricted T cell stimulation. The mouse antibodies reacted very poorly with fragmented forms of the immunogen or with tobacco hornworm moth cytochrome c, even though both of these antigens had been shown previously to strongly stimulate pigeon cytochrome c-primed T cells. The specificities of the primary IgG responses of seven other mouse strains were found to be very similar, but not identical, to that of B10.A mice. The cytochrome c-specific antibodies in the hyperimmune serum were shown to bind to determinants involving residues that vary between pigeon and mouse cytochromes c. Comparison of the binding of the antibodies to the immunogen and to the corresponding host protein enabled the calculation of the proportion of the overall binding energy contributed by the variant residues. This was as low as 19 to 35% for the primary response, rose to 25 to 46% for the hyperimmune mouse antibodies, and reached 40 to 63% for hyperimmune rabbit antibodies. The remaining energy of interaction (37 to 81%) was necessarily contributed by the surface of the protein surrounding the variant residues, which is the same for the immunogen and the host protein. These results illustrate the relatively subtle differences in binding affinities which can distinguish self from non-self recognition by antibody molecules.     

Dehydroepiandrosterone (DHEA) measurements in cord blood

Antibodies to dehydroepiandrosterone (DHEA) can be generated by coupling DHEA to carrier proteins. This report describes the use of antibodies to three different DHEA-protein complexes for the measurement of DHEA in serum and cord plasma. Each of the three antibodies produces an assay with equivalent results in serum from adults and older children. However, with cord plasma samples, two of the antibodies detected much more cross-reacting material than was detected with the third antibody. In order to determine the basis for this discrepancy, a pool of cord plasma was extracted in a manner similar to that used in the assay procedure. The steroids in the extract were separated by chromatography on a Sephadex LH-20 column and individual fractions were assayed for DHEA with each of the antibodies. There were several peaks of cross-reacting material present in cord blood that were not present in comparable amounts in serum from adults. These results indicate that assays for DHEA need to be separately validated with samples from adults and from infants.     

Partial purification and characterization of heat stable protein phosphatase inhibitor-2 from rabbit reticulocytes

Previous studies indicated that the species of type 1 and type 2 protein phosphatases (PP-1, PP-2) in rabbit reticulocytes are similar to those of rabbit skeletal muscle and rabbit liver. Reticulocyte PP-1 was found to be selectively inhibited by the heat stable protein phosphatase inhibitor-2 (I-2) from rabbit skeletal muscle. Of interest was the observation that muscle I-2 appeared to regulate protein synthesis in reticulocyte lysates by inhibiting an eIF-2 alpha phosphatase with type 1 properties. In this study we have characterized reticulocyte inhibitor-2 (I-2) and find that its properties are similar to those of skeletal muscle I-2. (i) Both I-2 species are stable to boiling and to acid treatment, and have similar chromatographic profiles on DEAE-cellulose and on Blue Sepharose CL-6B. (ii) The two I-2 species migrate electrophoretically as 26-28,000 dalton polypeptides in SDS-acrylamide gels. (iii) Both skeletal muscle I-2 and reticulocyte I-2 selectively inhibit isolated reticulocyte PP-1 and endogenous PP-1 in the lysate. (iv) Reticulocyte I-2 co-chromatographs with PP-1 on DEAE-cellulose, and over 90% of lysate I-2 can be isolated from this partially purified PP-1. (v) Both inhibitor-2 species are active in the unphosphorylated state, but upon addition to lysates, both are phosphorylated by endogenous cAMP-independent protein kinase(s). In addition a preliminary analysis using a polyclonal antibody against muscle inhibitor-1 confirmed biochemical analyses which indicate that lysates are deficient in inhibitor-1.     

Inhibition of lymphocyte transformation responses to phytohemagglutinin by normal human plasma.

Marked variability in lymphocyte transformation responses to a suboptimal phytohemagglutinin concentration (0.1 μg/ml) was observed when peripheral blood mononuclear cells of normal subjects were cultured in media containing 15% autologous plasma. Low responses were related to the plasma and were caused by direct inhibition rather than an inability to support the response. This inhibitory activity varied greatly among different plasma specimens and was also found in human and animal serum. It appeared to be specific for suboptimal concentrations of phytohemagglutinin and could be overcome by increasing the mitogen concentration. The inhibitory activity was heat stable, was not dialyzable, and appeared to affect a very early stage of the response since a delay in addition of the inhibitory plasma reversed its effect. Our interpretation of these results is that human plasmas vary greatly with respect to their ability to bind phytohemagglutinin so that the addition of different plasmas to lymphocyte cultures stimulated by a limiting concentration of phytohemagglutinin results in marked variability of the results obtained.     

Heat Denaturation and Emulsifying Properties of Plasma Protein

The effects of pH and NaCl on the denaturation of plasma protein during heat treatment were investigated, as well as the relationship between protein structure and emulsifying properties. When the plasma protein solution (1% w/v) was heated at 80°C, precipitation was accelerated by the presence of NaCl. The measurement of SH groups, surface hydrophobicity and CD spectrum revealed that denaturation occurs easily by heat treatment in the neutral pH region and in the presence of NaCl. The emulsifying activity index (EAI) did not change much after heat treatment at pH 3 irrespective of the presence of NaCl, but it decreased about 60% after heat treatment at pH 7 in the absence of NaCl. Gel filtration patterns indicated that a high molecular weight peak arose upon heat treatment at neutral pH. We concluded that the decrease in EAI was owing to the polymerization of serum albumin and γ-globulin, which are the main components of plasma protein, and disulfide bonds participated in this process.     

Association of plasma antibodies against the inducible Hsp70 with hypertension and harsh working conditions

Autoantibodies against certain stress or heat shock proteins (Hsps) may play a role in the pathogenesis and/ or prognosis of some diseases. Using immunoblotting with human recombinant Hsps and univariate and multivariate logistic regression models, we have investigated the presence of antibodies against Hsp70, the inducible member of the 70-kDa family of heat shock proteins, and analyzed its possible association with hypertension and working conditions. Plasma and serum were collected from 764 steel mill workers from 6 work sites exposed to (1) severe noise; (2) severe noise and dust; (3) noise, dust, and heat; (4) noise and heat; (5) severe noise and heat; and (6) office conditions (control). Workers with prolonged exposure to stresses such as noise, dust, and high temperature and a combination of these in the workplace had a high incidence (26.6% to 40.2%) of antibodies to Hsp70 compared to the lowest incidence (18.6%) of antibodies to Hsp70 in the control group of office workers. Moreover, there was a statistical association of antibodies against Hsp70 with hypertension. The statistical correlation between the presence of antibodies to Hsp70 and hypertension is higher in the group of workers with blood pressure of 160/95 mmHg than in the 140/90-mmHg group after excluding possible effects of the workplace stresses. These results suggest that harsh workplace conditions can increase the production of antibodies against Hsp70 and that the presence of antibodies to this stress protein may be associated with hypertension. The precise mechanism for the elevation of antibodies against Hsps by environmental and workplace stresses and their relation to hypertension remains to be established.     

Differences in metabolite profile between blood plasma and serum

In metabolomic research, blood plasma and serum have been considered to possess similar compositions and properties. Their perceived equivalence has resulted in researchers choosing arbitrarily between serum and plasma for analysis. Here, routine serum and plasma were prepared and their low-molecular-weight compounds were determined using gas chromatography/time-of-flight mass spectrometry. Principal components analysis was applied to process the acquired data, and marked differences in metabolite profiles were observed between serum and plasma. Of the 72 identified compounds, 36 (50%) discriminate serum from plasma, with 29 and 7 metabolites showing a significantly higher abundance (t test, P < 0.05) in serum and plasma, respectively. Incubation of blood had distinct effects on the analyte peak areas, with the effects being more pronounced for plasma than for serum and more pronounced for a shorter incubation than for a longer incubation. These results highlight the importance in choosing serum or plasma as the analytical sample and in stipulating the incubation time. Because incubation affected the analyte peak areas less in serum than in plasma, we recommend serum as the sample of choice in metabolomic studies.     

Concentrations of total protein, albumin and immunoglobulins in undiluted uterine fluid of gynecologically healthy mares

Undiluted uterine fluid from 20 Warmblood/Standardbred mares (5 to 14 yr old) was recovered by absorption to an intrauterine tampon. The mares were considered gynecologically healthy based on a clinical examination including uterine swabs for cytology and bacteriology as well as endometrial biopsy examinations. The protein profiles (SDS-PAGE) and concentrations of total protein, albumin, and immunoglobulins (Ig) A and G in the uterine fluid were examined and compared with the same proteins in serum. Major peaks were identified on the obtained protein profiles, and there was a clear similarity between the serum profiles and uterine fluid profiles. Variability in protein concentrations among mares was considerably larger in uterine fluid than in serum. Concentrations of the various proteins in uterine fluid were 44 to 56% of those in serum, except for IgA, which had a similar concentration in both serum and uterine fluid. Concentration of the proteins corresponding to peak No. 3 (molecular weight 60 to 71 kDa) in uterine fluid was higher (P < 0.05) in younger mares than in older ones. Parity had no effect on the recorded protein concentrations. The present study of gynecologically healthy mares showed that there is a large individual variation in the protein composition of uterine fluid. The results suggest that age, but not parity, may affect this composition, and indicate further that there is considerable transudation to the uterine cavity.     

Selectivity analysis of single binder assays used in plasma protein profiling

The increasing availability of antibodies toward human proteins enables broad explorations of the proteomic landscape in cells, tissues, and body fluids. This includes assays with antibody suspension bead arrays that generate protein profiles of plasma samples by flow cytometer analysis. However, antibody selectivity is context dependent so it is necessary to corroborate on‐target detection over off‐target binding. To address this, we describe a concept to directly verify interactions from antibody‐coupled beads by analysis of their eluates by Western blots and MS. We demonstrate selective antibody binding in complex samples with antibodies toward a set of chosen proteins with different abundance in plasma and serum, and illustrate the need to adjust sample and bead concentrations accordingly. The presented approach will serve as an important tool for resolving differential protein profiles from antibody arrays within plasma biomarker discoveries.     

Comparison of biotin binding protein of pregnant rat serum with rat serum albumin

The purified biotin binding protein of pregnant rat serum was shown to be immunologically similar to rat serum albumin as assessed by a sensitive radioimmunoassay. In radioimmunoassay for rat biotin binding protein, the binding of [¹²⁵I] rat biotin binding protein to anti-chicken egg yolk biotin binding protein antibodies was displaced by both rat serum (10–100 nl) and purified rat serum albumin (0.1–10 ng). Similarly, in radioimmunoassay for rat serum albumin the binding of [¹²⁵I] rat serum albumin to either anti-rat serum albumin antibodies or anti-chicken egg yolk biotin binding protein antibodies was displaced by unlabelled rat biotin binding protein at comparable concentration range (0·5–10 ng). Significant fractions of radioiodinated rat biotin binding protein and rat serum albumin bound to antibodies to chicken egg yolk biotin binding protein. In immature rats, the circulating half-lives of rat biotin binding protein and rat serum albumin were determined to be 12 and 17 h respectively. The rat biotin binding protein and rat serum albumin were analysed by techniques that exploit their physicochemical properties. They displayed similar electrophoretic mobilities in alkaline as well as denaturing sodium dodecyl sulphate-polyacrylamide gels. However, in nonequilibrium pH gradient polyacrylamide gel electrophoresis, they resolved clearly. In two-dimensional tryptic peptide map analysis, the two proteins showed similarities as well as significant differences in the relative distribution patterns of their iodopeptides. These results showed that the primary structure of rat biotin binding protein and rat serum albumin were different in finer details despite the fact that they shared significant immunological cross-reactivity.     

Immunocytochemical localization of the gap junction 26 K protein in mouse liver plasma membranes

Specific binding sites for anti-26 K antibodies directed against the liver gap junction protein (26 K) were localized by immunoelectron microscopy in gap junction plaques purified from hepatic plasma membranes. Using immunofluorescence microscopy we found discrete fluorescent spots on plasma membranes in cross sections of liver tissues after incubation with anti-26 K antibodies. This is consistent with the notion of specific binding to gap junction plaques. Quantitative binding of anti-26 K antibodies was indirectly measured by the protein A-gold technique. We found that urea/detergent-treated, purified gap junction plaques bind 30-fold more anti-26 K antibodies than preimmune serum. Anti-26 K antibodies also bind specifically to native gap junction plaques within hepatic plasma membranes although only about one fifth as efficiently as to purified plaques. Possibly the anti-26 K antibodies raised after injection of SDS-denatured 26 K protein into rabbits recognize the cytoplasmic face of urea/detergent-treated plaques better than that of native plaques. Some, if not most, of the vesicular structures in preparations of purified plaques appear to be derived from split gap junction plaques and are probably sheets of gap junction hemichannels. In some vesicles the former cytoplasmic face of the hemichannels is turned outside, other vesicles have the former cell surface turned outside. The anti-26 K antibodies do not recognize any 26 K protein on the sheets of partially split gap junction plaques, on the heterogeneous vesicular structures, or on non-junctional areas of hepatic plasma membranes. These results suggest that the conformation of the 26 K protein in plaques must be different from that of the 26 K protein in earlier biosynthetic steps of plaque assembly.     

Isolation of albumin from whole human plasma and fractionation of albumin-depleted plasma.

The dye Cibacron Blue F-3-GA was conjugated to Sepharose to provide an affinity column for serum albumin. Passage of whole human plasma through a column of Cibacron Blue-Sepharose results in the removal of approx. 98% of the albumin. The latter can be quantitatively recovered by desorption with NaSCN. Albumin-depleted plasma can be readily resolved into discrete fractions by a combination of conventional biochemical techniques. In particular, the resolution of plasma proteins with properties similar to those of native human plasma albumin can readily be accomplished by ion-exchange chromatography of the Sepharose-dye-treated plasma on DEAE-cellulose.     

The 90-kDa heat shock protein (hsp-90) possesses an ATP binding site and autophosphorylating activity

The 90-kDa heat shock protein (hsp-90) is an abundant cytosolic protein believed to play a role in maintenance of protein trafficking and closely associated with several steroid hormone receptors. Incubation of highly purified hsp-90 with [gamma-32P]ATP results in its autophosphorylation on serine residues. There are several lines of evidence which suggest that this activity is due to a kinase intrinsic to hsp-90 rather than some closely associated protein kinases: 1) the phosphorylation persists after the removal of casein kinase II by heparin chromatography and after immunoprecipitation of hsp-90 with anti-hsp-90 antibodies. 2) The approximate kM for ATP of the reaction is 0.16 mM, higher than that of many other protein kinases. 3) Phosphorylation is not affected by a number of activators and inhibitors of other known kinases which might associate with hsp-90. 4) The phosphorylation displays a unique cation dependence being most active in the presence of Ca2+ and practically inactive with Mg2+, although the autophosphorylation in the presence of Mg2+ is activated by histones and polyamines. 5) The activity is remarkably heat-stable; incubation of hsp-90 for 20 min at 95 degrees C results in only a 60% decrease in autophosphorylation. 6) Finally, and most importantly, purified hsp-90 can be labeled with azido-ATP and it is able to bind to ATP-agarose. The binding shows similar cation dependence to the autophosphorylation. These data are in agreement with the presence of a consensus sequence for ATP binding sites in the primary structure of the protein similar to that observed in the 70-kDa heat-shock proteins. Our data suggest the 90-kDa heat shock protein possesses an enzymatic activity analogous in many respects to the similar activity of the 70-kDa heat shock proteins. This may represent an important, previously unrecognized function of hsp-90.     

Stress-induced aggregation profiles of GST-alpha-synuclein fusion proteins: role of the C-terminal acidic tail of alpha-synuclein in protein thermosolubility and stability

Alpha-synuclein is a well-known heat-resistant protein that does not aggregate upon heat treatment, whereas glutathione S-transferase (GST) is a heat-labile protein that easily precipitates as a result of thermal stress. This paper reports the role of the C-terminal acidic tail of alpha-synuclein in protein thermosolubility and stability. The region of alpha-synuclein that is responsible for the heat resistance was initially investigated using a series of deletion mutants, and the C-terminal acidic tail (residues 96-140) was found to be crucial for the thermosolubility of alpha-synuclein. The thermal behavior of the GST-alpha-synuclein fusion protein was next investigated, and the fusion protein was seen to be extremely heat-resistant. Using a series of GST-synuclein deletion mutants, the C-terminal acidic tail of alpha-synuclein was shown to play a critical role in conferring the heat resistance of the fusion proteins. Furthermore, the acidic tail appeared to protect the fusion protein from pH- and metal-induced protein aggregation, suggesting that the acidic tail can increase the virtual stability of the protein by protecting it from the aggregation induced by environmental stresses. Interestingly, the acidic tail also appeared to protect the GST enzyme from the thermal inactivation to a considerable extent. However, CD analysis of the heat-induced secondary structural changes of the GST-alpha-synuclein fusion protein revealed that the fusion protein is irreversibly denatured by heat treatment with a slightly lowered melting temperature (Tm). Thus, the results demonstrate that introducing an acidic tail into GST promotes the thermosolubility and virtual stability of the fusion protein, although it might be unfavorable for its intrinsic stability.     

Hepatocyte adhesion on plastic : Different mechanisms for serum- and fibronectin-mediated adhesion

Hepatocytes adhere well on plastic in the presence of serum or fibronectin and subsequent spreading is not prevented when protein synthesis was blocked by cycloheximide. Protein synthesis-independent spreading was also observed in cultures containing serum depleted of fibronectin by affinity chromatography. This indicates that serum-mediated adhesion is independent of fibronectin and suggests the existence of an adhesion factor other than fibronectin in serum. The involvement of different membrane components for fibronectin- and serum-mediated adhesion was demonstrated by experiments where the different adhesion-inhibiting activities of antisera raised against plasma membranes of rat liver and Morris hepatoma 7777 (Neumeier et al., FEBS lett 168 (1984) 241-244) were used. Whereas anti-liver antibodies inhibited both types of adhesion, anti-hepatoma antibodies were only able to prevent fibronectin-mediated adhesion. This indicates again that two different mechanisms are responsible for fibronectin- and serum-mediated adhesion. Fractionation of fetal calf serum (FCS) by size exclusion HPLC revealed that proteins of molecular weights of 60-80 kD promoted attachment and spreading of hepatocytes. Spreading was not perturbated by anti-hepatoma antibodies, indicating that an adhesion factor of 60-80 kD is responsible for serum-mediated adhesion. 'Serum-spreading factor', also called vitronectin, from human plasma has been described as having a similar molecular weight. The purified factor was found to mediate hepatocyte adhesion which was not inhibited by anti-hepatoma antibodies. This suggests that serum-mediated adhesion depends on an adhesion factor present in FCS, which is similar to or identical with vitronectin.     

Phosphorylation of cellular proteins in Rous sarcoma virus-infected cells: analysis by use of anti-phosphotyrosine antibodies.

The protein substrates for the tyrosine protein kinases in cells transformed by avian sarcoma viruses were analyzed by gel electrophoresis in combination with immunoblotting or immunoprecipitation by antibodies against phosphotyrosine. We found that greater than 90% of phosphotyrosine-containing cellular proteins can be immunoprecipitated by these antibodies. The level of phosphotyrosine-containing cellular proteins detectable by this method markedly increased upon transformation with Rous sarcoma virus, and more than 20 distinct bands of such proteins were found in lysates of Rous sarcoma virus-transformed cells. Most of these phosphotyrosine-containing proteins had not been identified by other methods, and their presence appeared to correlate with morphological transformation in cells infected with various Rous sarcoma virus mutants and Y73, PRCII, and Fujinami sarcoma viruses. However, considerably different patterns were obtained with cells infected with nontransforming Rous sarcoma virus mutants that encode nonmyristylated src kinases, indicating that most substrates that correlate with transformation can only be recognized by p60v-src associated with the plasma membrane.     

Studies of distribution,location and dynamic properties of EGFR on the cell surface measured by image correlation spectroscopy

In this work, we have studied the distribution and dynamic properties of Epidermal Growth Factor (EGF) receptors in the plasma membrane of fixed and live cells as well as the extent of co-localization of this transmembrane protein with proteins specific for three-membrane microdomains: membrane rafts, caveolae and clathrin-coated pits. This was achieved using a family of image-processing tools called image correlation spectroscopy (ICS), image cross-correlation spectroscopy (ICCS) and dynamic image correlation spectroscopy (DICS). Our results indicate that EGFR is diffusely distributed on the cell surface at 37°C and aggregates as the temperature is lowered to 4°C. This aggregation takes place within 15 min and is reversible. Changes in temperature also affect the diffusion of EGFR by two orders of magnitude. The dynamic properties of EGFR are similar to the dynamic properties of a GPI-anchored protein known to be present in membrane rafts, which motivated us to explore the extent of co-localization of EGFR with this membrane raft protein using ICCS. Our results indicate that more than half of the EGFR population is present in membrane rafts and smaller percentages are present in caveolae and clathrin-coated pits.     

Study of thermal properties and heat-induced denaturation and aggregation of soy proteins by modulated differential scanning calorimetry

The thermal properties and heat-induced denaturation and aggregation of soy protein isolates (SPI) were studied using modulated differential scanning calorimetry (MDSC). Reversible and non-reversible heat flow signals were separated from the total heat flow signals in the thermograms. In the non-reversible profiles, two major endothermic peaks (at around 100 and 220 degrees C, respectively) associated with the loss of residual water were identified. In the reversible profiles, an exothermic peak associated with thermal aggregation was observed. Soy proteins denatured to various extents by heat treatments showed different non-reversible and reversible heat flow patterns, especially the exothermic peak. The endothermic or exothermic transition characteristics in both non-reversible and reversible signals were affected by the thermal history of the samples. The enthalpy change of the exothermic (aggregation) peak increased almost linearly with increase in relative humidity (RH) in the range between 8 and 85%. In contrast, the onset temperature of the exotherm decreased progressively with increase in RH. These results suggest that the MDSC technique could be used to study thermal properties and heat-induced denaturation/aggregation of soy proteins at low moisture contents. Associated functional properties such as water holding and hydration property can also be evaluated.