Reproductive neuroendocrine defects programmed by prenatal testosterone treatment between gestational days 60-90 are amplified by postnatal obesity in sheep.

Polycystic ovary syndrome (PCOS) is the leading cause of anovulatory infertility in women of reproductive age, and obesity can increase the severity and development of the PCOS phenotype. Prenatal testosterone (T) treatment between gestational days 30-90 advanced puberty and disrupted the reproductive and metabolic phenotype in female sheep, recapitulating attributes of women with PCOS, with postnatal obesity amplifying its severity. On the other hand, prenatal T treatment from gestational days 60-90 led to a much milder phenotype. We hypothesized that reproductive neuroendocrine defects programmed by prenatal T treatment between gestational days 60-90 are amplified by postnatal obesity in sheep. Suffolk ewes received T propionate (T; 100 mg) or corn oil (C; vehicle) twice weekly from gestational days 60-90. At 5 months of age, T lambs were assigned to either a maintenance (100% of NRC requirements) or overfed (130% NRC) diet and C lambs were fed the maintenance diet. We compared the timing of puberty (n = 15/group) determined by twice weekly measurement of progesterone concentrations, estradiol positive feedback responsiveness (n = 8/group) determined by assessing LH secretion in response to exogenous estradiol, periovulatory LH dynamics during the second breeding season (n = 8/group) following synchronization with two injections of PGF2α, and progesterone negative feedback (n = 8/group) determined by characterizing LH pulses during the mid-luteal phase between C, T-maintenance and T-overfed groups. Our findings indicate that postnatal obesity: 1) exacerbated reproductive defects and further deteriorated reproductive cyclicity during the second breeding season (adulthood); 2) did not amplify the impairment in estradiol positive feedback in delaying the timing and amplitude of the LH surge, although it reduced the total amount of LH secreted during the preovulatory LH surge; 3) amplified the reduced responsiveness to progesterone negative feedback manifested as an increase in LH pulse amplitude and peak. These observations, in addition to supporting our previous findings that prenatal T treatment results in reproductive neuroendocrine dysfunction and periovulatory disruptions, provide evidence that these neuroendocrine defects programmed between gestational days 60-90 are amplified by postnatal obesity in female sheep.

Induced prostaglandin release alters steroid concentrations but not pregnancy survival in cows.

Embryonic mortality (EM) is a major factor limiting reproductive efficiency in cattle, and despite negative connotations related to reproductive performance, prostaglandin F2α (PGF2α) is capable of being released by the uterus by Day 30 of gestation. Therefore, the objective was to evaluate differences in PGF2α release after an oxytocin challenge between cows with high circulating concentrations of pregnancy-associated glycoproteins (PAGs) vs low PAG because of the association of increased PAG concentrations with pregnancy success. At Day 30 of gestation, pregnant cows were divided into oxytocin treatment (OT; n = 13) and control (CON; n = 12) groups. Treatment cows were further subdivided by circulating PAG concentration (high PAG, n = 7; and low PAG, n = 6). Blood samples were collected every 30 min beginning 1 h before oxytocin administration and continuing for 4 h. Prostaglandin F2α metabolite (PGFM), progesterone, estradiol-17ß (E2), and PAG concentrations were quantified. The peak concentration of PGFM occurred 2 h after oxytocin injection in treatment animals and returned to baseline levels by 4 h. No correlations were observed between PAG and PGFM, progesterone, or E2 concentrations (P > 0.05). There was no difference in initial or final PGFM concentrations between groups (P > 0.05). Progesterone and E2 concentrations decreased in cows after treatment of oxytocin (P < 0.05); however, only progesterone returned to basal concentrations by the end of the sampling period. In summary, cows with high vs low PAG concentrations at Day 30 of gestation have a similar PGFM response to oxytocin challenge.

Nutritional control of puberty in the bovine female: prenatal and early postnatal regulation of the neuroendocrine system.

Puberty is a complex biological event that requires maturation of the reproductive neuroendocrine axis and subsequent initiation of high-frequency, episodic release of GnRH and LH. Nutrition is a critical factor affecting the neuroendocrine control of puberty. Although nutrient restriction during juvenile development delays puberty, elevated rates of body weight gain during this period facilitate pubertal maturation by programming hypothalamic centers that underlie the pubertal process. Recent findings suggest that maternal nutrition during gestation can also modulate the development of the fetal neuroendocrine axis, thus influencing puberty and subsequent reproductive function. Among the several metabolic signals, leptin plays a critical role in conveying metabolic information to the brain and, consequently, controlling puberty. The effects of leptin on GnRH secretion are mediated via an upstream neuronal network because GnRH neurons do not express the leptin receptor. Two neuronal populations located in the arcuate nucleus that express the orexigenic peptide neuropeptide Y (NPY), and the anorexigenic peptide alpha melanocyte-stimulating hormone (αMSH), are key components of the neurocircuitry that conveys inhibitory (NPY) and excitatory (αMSH) inputs to GnRH neurons. In addition, neurons in the arcuate nucleus that coexpress kisspeptin, neurokinin B, and dynorphin (termed KNDy neurons) are also involved in the metabolic control of puberty. Our studies in the bovine female demonstrate that increased planes of nutrition during juvenile development lead to organizational and functional changes in hypothalamic pathways comprising NPY, proopiomelanocortin (POMC, the precursor of αMSH), and kisspeptin neurons. Changes include alterations in the abundance of NPY, POMC, and Kiss1 mRNA and in plasticity of the neuronal projections to GnRH neurons. Our studies also indicate that epigenetic mechanisms, such as modifications in the DNA methylation pattern, are involved in this process. Finally, our most recent data demonstrate that maternal nutrition during gestation can also induce morphological and functional changes in the hypothalamic NPY system in the heifer offspring that are likely to persist long after birth. These organizational changes occurring during fetal development have the potential to not only impact puberty but also influence reproductive performance throughout adulthood in the bovine female.

Prolonged in vivo residence times of antibody fragments associated with albumin.

Antibody fragments can be expressed at a high level in microbial systems, but they may have limited therapeutic value because they are rapidly eliminated from the body. We demonstrate here that site-specific conjugation or binding of bacterially derived Fab' to the long-lived protein serum albumin allows full retention of the antibody's binding characteristics while imparting the albumin's longevity in vivo. In rats the area under the curve for Fab' conjugated to rat serum albumin was 17-fold greater than for the control of Fab' conjugated to cysteine. Again, a bispecific F(ab')(2) with specificity for rat serum albumin showed an area under the curve about 8-fold greater than did a F(ab')(2) without specificity to albumin. Genetic fusions of scFv to albumin were similarly long-lived and could be expressed in yeast to provide the basis of a cost-effective production system.

Improved efficiency of site-specific copper(II) ion-catalysed protein cleavage effected by mutagenesis of cleavage site.

The peptide sequence (N)DKTH(C) was previously investigated as a site for efficient, specific cleavage of a fusion protein by cupric ions using a humanized gamma1 Fab' as a model protein. Here we show that conservative mutations to three of the residues in the introduced cleavage site resulted in cleavage sites that were significantly improved. They were cleaved more efficiently by Cu(2+), such that cleavage reactions could be shorter, of lower pH or at a lower temperature. Some were even found to be measurably cleaved by Ni(2+). Use of these new cleavage sequences along with cupric ions may provide a more rapid and less harsh method for cost-effective, large-scale proteolytic cleavage of fusion proteins and peptides.

Efficient site specific removal of a C-terminal FLAG fusion from a Fab' using copper(II) ion catalysed protein cleavage.

The peptide sequence (N)DKTH(C) was investigated as a site for efficient, specific cleavage of a fusion protein by cupric ions using a humanised gamma1 Fab' as a model protein. The native upper hinge (N)DKTH(C) sequence was mutated to create a site resistant to cleavage by cupric ions and a (N)DKTH(C) sequence introduced between the hinge and a C-terminal FLAG peptide. Incubation of Fab' with Cu2+ at 62 degrees C at alkaline pHs resulted in removal of the FLAG peptide with efficiencies of up to 86%. Cleavage conditions did not detrimentally affect the Fab' protein. Use of the (N)DKTH(C) sequence along with cupric ions may provide a cost-effective method for large scale proteolytic cleavage of fusion proteins.

Cloning and overexpression in Escherichia coli of the gene encoding citrate synthase from the hyperthermophilic Archaeon Sulfolobus solfataricus.

The citrate synthase (CS) gene from the hyperthermophilic Archaeon Sulfolobus solfataricus has been cloned and sequenced. The gene encodes a polypeptide of 378 amino acids with a calculated polypeptide molecular mass of 42679. High-level expression was achieved in Escherichia coli and the recombinant citrate synthase was purified to homogeneity using a heat step and dye-ligand affinity chromatography. This procedure yielded approximately 26 mg of pure CS per liter of culture, with a specific activity of 41 U/mg. The enzyme exhibited a half-life of 8 min at 95 degrees C. A homology-modelled structure of the S. solfataricus CS has been generated using the crystal structure of the enzyme form the thermoacidophilic Archaeon Thermoplasma acidophilum with which it displays 58% sequence identity. The modelled structure is discussed with respect to the thermostability properties of the enzyme.

Improved protein refolding using hollow-fibre membrane dialysis.

We have used a cellulose acetate, hollow-fibre (HF) ultrafiltration membrane to refold bovine carbonic anhydrase, loaded into the lumen space, by removing the denaturant through controlled dialysis via the shell side space. When challenged with GdnHCl-denatured carbonic anhydrase, 70% of the loaded protein reptated through the membrane into the circulating dialysis buffer. Reptation occurred because the protein, in its fully unfolded configuration, was able to pass through the pores. The loss of carbonic anhydrase through the membrane was controlled by the dialysis conditions. Dialysis against 0.05 M Tris-HCl for 30 min reduced the denaturant around the protein to a concentration that allowed the return of secondary structure, increasing the hydrodynamic radius, thus preventing protein transmission. Under these conditions a maximum of 42% of carbonic anhydrase was recovered (from a starting concentration of 5 mg/mL) with 94% activity. This is an improvement over refolding carbonic anhydrase by simple batch dilution, which gave a maximum reactivation of 85% with 35% soluble protein yield. The batch refolding of carbonic anhydrase is very sensitive to temperature; however, during HF refolding between 0 and 25 degrees C the temperature sensitivity was considerably reduced. In order to reduce the convection forces that give rise to aggregation and promote refolding the dialyzate was slowly heated from 4 to 25 degrees C. This slow, temperature-controlled refolding gave an improved soluble protein recovery of 55% with a reactivation yield of 90%. The effect of a number of additives on the refolding system performance were tested: the presence of PEG improved both the protein recovery and the recovered activity from the membrane, while the detergents Tween 20 and IGEPAL CA-630 increased only the refolding yield.

Protein folding in vivo and renaturation of recombinant proteins from inclusion bodies.

Eukaryotic proteins expressed in Escherichia coli often accumulate within the cell as insoluble protein aggregates or inclusion bodies. The recovery of structure and activity from inclusion bodies is a complex process, there are no general rules for efficient renaturation. Research into understanding how proteins fold in vivo is giving rise to potentially new refolding methods, for example, using molecular chaperones. In this article we review what is understood about the main three classes of chaperone: the Stress 60, Stress 70, and Stress 90 proteins. We also give an overview of current process strategies for renaturing inclusion bodies, and report the use of novel developments that have enhanced refolding yields.

The pyruvate dehydrogenase complex of Saccharomyces cerevisiae is regulated by phosphorylation.

Mitochondria were isolated from Saccharomyces cerevisiae grown on different carbon sources prior to incubation with [gamma-32P]ATP. A major 46,000-M(r) phosphoprotein, corresponding in M(r) value to the E1 alpha subunit of the yeast pyruvate dehydrogenase complex (PDC), was detected only in mitochondria isolated from cells grown on a fermentable carbon source such as galactose. Immunoprecipitation with subunit-specific antiserum to the E1 component of mammalian or yeast PDC confirmed the identity of this polypeptide. PDC activity in isolated yeast mitochondria could be inactivated in an ATP-dependent fashion and reactivated in the presence of Ca2+ ions.

Dissociation and unfolding of the pyruvate dehydrogenase complex by guanidinium chloride.

The effect of guanidinium chloride (GdnHCl) on the pyruvate dehydrogenase complex (PDC) from bovine heart and its constituent enzymes has been studied. The overall activity of the complex is lost reversibly at low levels of GdnHCl (0.2 M) which cause 40-50% inactivation but no loss of overall secondary or tertiary structures of the individual enzymes; the inactivation of the complex is shown to be caused by dissociation of the E1 and E3 components from the E2/X core assembly. This provides an improved procedure for controlled dissociation of the complex and efficient recovery of its component enzymes in their native states. Higher concentrations of GdnHCl (up to 4 M) lead to the unfolding and irreversible inactivation of the separate enzymes of the complex with the E2/X core proving the most resistant to GdnHCl-induced unfolding. Neither the 60-meric E2/X core assembly nor the dimeric E3 component are dissociated into monomers in the presence of 6 M GdnHCl; the latter enzyme forms higher-M(r) aggregates under these conditions.

Comparison of three porcine restenosis models: the relative importance of hypercholesterolemia, endothelial abrasion, and stenting.

BACKGROUND: Porcine models of post-angioplasty restenosis commonly rely on hypercholesterolemia, endothelial abrasion, and intracoronary stenting to induce neointimal thickening. Although stenting clearly induces marked thickening, the influence of pre-stenting endothelial abrasion, and pre- and post-stenting hypercholesterolemia, on the degree and nature of post-stenting neointimal thickening is not clear. In order to assess this influence, we compared the quantity and quality of neointimal thickening in three stented swine restenosis models. METHODS: Twenty-three Hanford miniature swine completed one of three protocols. Model A animals (n = 9) were fed a cholesterol-raising diet, underwent endothelial abrasion of the left anterior descending (LAD) and circumflex (CFX) coronary arteries after 2 weeks on this diet, had balloon-expandable tantalum coil stents placed in the right coronary artery (RCA), LAD, and CFX after 9 weeks on the diet, and were killed 4 weeks later (total of 13 weeks on diet). Model B animals (n = 7) were also fed the cholesterol-raising diet, underwent stenting after 5 weeks on the diet, and were killed 4 weeks later (total of 9 weeks on diet). Model C animals (n = 7) were fed normal swine food, underwent stenting, and were killed 4 weeks later. Endothelial abrasion was not performed in models B and C. RESULTS: Quantitative angiography revealed no significant differences between models in the change of minimal lumen diameter (mm +/- SD) of stented vessels from post-stenting to pre-sacrifice (LAD: 1.05 +/- 0.74, 0.75 +/- 0.62 and 1.05 +/- 0.34; CFX: 1.00 +/- 0.65, 0.83 +/- 0.51 and 1.17 +/- 0.38; RCA: 0.99 +/- 0.35, 0.20 +/- 0.34, and 0.94 +/- 0.80 for models A, B, and C, respectively; all P = NS). Likewise, morphometric analysis showed no differences in percentage area stenosis (% +/- SD) over the same time (LAD: 55 +/- 15, 44 +/- 24, and 42 +/- 16; CFX: 54 +/- 12, 55 +/- 17, and 40 +/- 15; RCA: 39 +/- 20, 34 +/- 11, and 26 +/- 13 for models A, B, and C, respectively; P = NS). The neointima in each model predominantly consisted of smooth muscle cells and collagen matrix. CONCLUSIONS: The degree and nature of coronary artery neointimal thickening 4 weeks after stenting in normolipemic swine are similar to those in stented swine after 9 weeks on a high-cholesterol diet or 13 weeks on a high-cholesterol diet and early endothelial abrasion. The insertion of an intracoronary stent appears to be the major stimulus to neointimal thickening in these swine models of post-angioplasty restenosis.

Studies of activated GPIIb/IIIa receptors on the luminal surface of adherent platelets. Paradoxical loss of luminal receptors when platelets adhere to high density fibrinogen.

The accessibility of activated GPIIb/IIIa receptors on the luminal surface of platelets adherent to damaged blood vessels or atherosclerotic plaques is likely to play a crucial role in subsequent platelet recruitment. To define better the factors involved in this process, we developed a functional assay to assess the presence of activated, luminal GPIIb/IIIa receptors, based on their ability to bind erythrocytes containing a high density of covalently coupled RGD-containing peptides (thromboerythrocytes). Platelets readily adhered to wells coated with purified type I rat skin collagen and the adherent platelets bound a dense lawn of thromboerythrocytes. With fibrinogen-coated wells, platelet adhesion increased as the fibrinogen-coating concentration increased, reaching a plateau at about 11 micrograms/ml. Thromboerythrocyte binding to the platelets adherent to fibrinogen showed a paradoxical response, increasing at fibrinogen coating concentrations up to approximately 4-6 micrograms/ml and then dramatically decreasing at higher fibrinogen-coating concentrations. Scanning electron microscopy demonstrated that the morphology of platelets adherent to collagen was similar to that of platelets adherent to low density fibrinogen, with extensive filopodia formation and ruffling. In contrast, platelets adherent to high density fibrinogen showed a bland, flattened appearance. Immunogold staining of GPIIb/IIIa receptors demonstrated concentration of the receptors on the filopodia, and depletion of receptors on the flattened portion of the platelets. Thus, there is a paradoxical loss of accessible, activated GPIIb/IIIa receptors on the luminal surface of platelets adherent to high density fibrinogen. Two factors may contribute to this result: engagement of GPIIb/IIIa receptors with fibrinogen on the abluminal surface leading to the loss of luminal receptors, and loss of luminal filopodia that interact with thromboerythrocytes. These data provide insight into the differences in thrombogenicity between surfaces, and may provide a mechanism for purposefully passivating platelet-reactive artificial surfaces.

Thromboerythrocytes. In vitro studies of a potential autologous, semi-artificial alternative to platelet transfusions.

In an attempt to overcome the limitations and drawbacks of using fresh platelets for transfusion therapy of thrombocytopenic patients, we have performed in vitro experiments on an autologous, semi-artificial alternative to platelet transfusions. Based on our previous studies of the interactions of unactivated and activated platelets with beads coated with peptides of various lengths, all of which contained the arginine-glycine-aspartic acid (RGD) cell recognition sequence, the peptide Ac-CGGRGDF-NH2 was chosen for covalent coupling to erythrocytes. A heterobifunctional crosslinking reagent (N-maleimido-6-aminocaproyl ester of 1-hydroxy-2-nitrobenzene-4-sulfonic acid) was used to crosslink via the peptide's free sulfhydryl group and the erythrocyte's surface amino groups. Approximately 0.5-1.5 x 10(6) peptide molecules bound per erythrocyte after 2 h of incubation, and most of the peptides appeared to crosslink to glycophorin A. The resulting cells, termed thromboerythrocytes, interacted selectively with activated platelets to form mixed aggregates. Studies with fluid phase RGD peptides and monoclonal antibodies indicated that the RGD peptides on the thromboerythrocytes interacted with the GPIIb/IIIa receptors on activated platelets. Thromboerythrocytes could also bind to platelets adherent to collagen. There was minimal erythrocyte hemolysis during the formation of thromboerythrocytes and studies of thromboerythrocyte osmotic fragility and cellular deformability showed no significant changes from control erythrocytes. Whereas there is a 20:1 ratio of erythrocytes to platelets in the circulation of normal individuals, the erythrocytes from as little as 50 ml of blood could be transformed into the equivalent of 2 U of platelets by numbers (equivalent to 18 U of platelets by mass), and reinfused into the same individual within several hours. These data encourage us to proceed to in vivo studies to assess the hemostatic efficacy of thromboerythrocytes in thrombocytopenic animals.

Platelet fibrinogen and vitronectin in Glanzmann thrombasthenia: evidence consistent with specific roles for glycoprotein IIb/IIIA and alpha v beta 3 integrins in platelet protein trafficking.

To assess the individual contributions of the platelet glycoprotein (GP) IIb/IIIa receptor and the alpha v beta 3 vitronectin receptor to platelet levels of fibrinogen and vitronectin, we analyzed the platelets from two groups of Glanzmann thrombasthenic patients: Iraqi-Jews, whose platelets lack both receptors, and Arab patients in Israel, whose platelets lack GPIIb/IIIa, but have normal or increased numbers of alpha v beta 3 vitronectin receptors. The platelets from both thrombasthenic groups had profound deficiencies of fibrinogen, but the defect in the Iraqi-Jewish patients' platelets appeared to be slightly more severe. This finding indicates that GPIIb/IIIa is the major determinant of platelet fibrinogen, presumably acting by receptor-mediated uptake, and that the alpha v beta 3 vitronectin receptor plays little or no role. Arab patients' platelets have normal amounts of platelet vitronectin, whereas Iraqi-Jewish patients' platelets have nearly five times as much vitronectin as control or Arab patients' platelets. To account for these data, we propose a working hypothesis in which vitronectin is synthesized in megakaryocytes and the alpha v beta 3 vitronectin receptor is involved in transport of the protein out of megakaryocytes and/or platelets. Collectively, these observations suggest that in addition to their recognized roles in cell adhesion and in the interaction of cells with extracellular proteins, integrin receptors may be important in protein trafficking into, and perhaps out of, platelets.

Clarification of the identity of the major M2 autoantigen in primary biliary cirrhosis.

1. In primary biliary cirrhosis, the major M2 autoantigen, reacting with antimitochondrial antibodies in sera from greater than 90% of patients, has been identified as the E2 component of the pyruvate dehydrogenase complex. However, two recent reports suggest that alternative polypeptides may be major autoantigens. 2. The evidence that a 75 kDa subunit of complex I of the respiratory chain is a major autoantigen (Frostell, Mendel-Hartvig, Nelson, Totterman, Bjorkland & Ragan, Scand. J. Immunol. 1988; 28, 157-65) is refuted. The findings of Frostell et al. can be explained by contamination of complex I with the pyruvate dehydrogenase complex, evidence for which is presented here. 3. Inspection of the partial amino acid sequence of an unidentified mitochondrial autoantigen (Muno, Kominami, Ishii, Usui, Saituku, Sakakibara & Namihisa, Hepatology 1990; 11, 16-23) shows that it is the E1 beta-subunit of the pyruvate dehydrogenase complex, previously identified as a major autoantigen, and not a 'new' alternative major autoantigen. 4. These findings substantiate previous work showing that the mitochondrial M2 autoantigens identified so far in primary biliary cirrhosis are all polypeptide components of the pyruvate dehydrogenase complex or the other related 2-oxo acid dehydrogenase complexes.

The unfolding and attempted refolding of citrate synthase from pig heart.

The unfolding of the dimeric enzyme citrate synthase from pig heart in solutions of guanidinium chloride (GdnHCl) was studied. Data from fluorescence, circular dichroism (CD) and thiol group reactivity studies indicated that the enzyme was almost completely unfolded at GdnHCl concentrations greater than or equal to 4 M. On dilution of GdnHCl, essentially no reactivation of the enzyme occurred. The implications of this finding for the process of folding and assembly in vivo of this and other mitochondrial enzymes are discussed. Exposure of the enzyme to high pH (9-10) led to only a small loss of secondary structure and partial reactivation could be observed on readjustment of the pH to 8.0.

The unfolding and attempted refolding of mitochondrial aspartate aminotransferase from pig heart.

The unfolding of the mitochondrial isoenzyme of aspartate aminotransferase from pig heart in solutions of guanidinium chloride (GdnHCl) has been studied. By a number of criteria (enzyme activity, protein fluorescence, c.d., thiol-group reactivity), the enzyme was judged to be almost completely unfolded in 2 M-GdnHCl. On dilution of the GdnHCl, no re-activation of the enzyme could be observed, whether or not pyridoxal 5'-phosphate and dithiothreitol were present. The behaviour of the mitochondrial isoenzyme is in marked contrast with that of the cytoplasmic isoenzyme [West & Price (1989) Biochem. J. 261, 189-196], despite the similarities in the amino acid sequences and tertiary structures of the two isoenzymes. The implications of these findings for the process of folding and assembly of the mitochondrial isoenzyme in vivo are discussed.

The unfolding and refolding of cytoplasmic aspartate aminotransferase from pig heart.

The unfolding of cytoplasmic aspartate aminotransferase from pig heart in solutions of guanidinium chloride (GdnHCl) was studied. Data from protein fluorescence, c.d. and thiol-group reactivity indicated that the enzyme was unfolded in 6 M-GdnHCl. Spectroscopic studies showed that this unfolding was accompanied by dissociation of the pyridoxal 5'-phosphate cofactor. On dilution of the GdnHCl, re-activation of the enzyme occurred in reasonable yield, provided that dithiothreitol and pyridoxal 5'-phosphate were present. The regain of activity obeyed second-order kinetics. In the absence of added dithiothreitol and pyridoxal 5'-phosphate, substantial formation of high-Mr aggregates occurred.

The unfolding and refolding of glutamate dehydrogenases from bovine liver, baker's yeast and Clostridium symbosium.

The unfolding behaviour of the hexameric glutamate dehydrogenases from bovine liver, Clostridium symbosium and baker's yeast in solutions of guanidinium chloride (GdnHCl) was studied. Changes in Mr studied by light-scattering indicate that, in each case, the hexamer dissociates to form trimers, which then dissociate to monomers at higher concentrations of GdnHCl. Dissociation to trimers is accompanied by a reversible loss of enzyme activity, but no gross structural changes can be detected by fluorescence or c.d. Dissociation to monomers is accompanied by large structural changes, and the loss of activity cannot be reversed by dilution. The parallel behaviour of all three enzymes shows that the previously noted inability of the isolated subunits of the bovine liver enzyme to refold [Bell & Bell (1984) Biochem. J. 217, 327-330] is not a result of any modification of the enzyme as a result of import into mitochondria, since the C. symbosium and baker's-yeast enzymes do not undergo any such post-translational translocation.