The final step in methane formation. Investigations with highly purified methyl-CoM reductase (component C) from Methanobacterium thermoautotrophicum (strain Marburg)

Methyl-coenzyme M reductase (= component C) from Methanobacterium thermoautotrophicum (strain Marburg) was highly purified via anaerobic fast protein liquid chromatography on columns of Mono Q and Superose 6. The enzyme was found to catalyze the reduction of methylcoenzyme M (CH3-S-CoM) with N-7-mercaptoheptanoylthreonine phosphate (H-S-HTP = component B) to CH4. The mixed disulfide of H-S-CoM and H-S-HTP (CoM-S-S-HTP) was the other major product formed. The specific activity was up to 75 nmol min-1 mg protein-1. In the presence of dithiothreitol and of reduced corrinoids or titanium(III) citrate the specific rate of CH3-S-CoM reduction to CH4 with H-S-HTP increased to 0.5-2 mumol min-1 mg protein-1. Under these conditions the CoM-S-S-HTP formed from CH3-S-CoM and H-S-HTP was completely reduced to H-S-CoM and H-S-HTP. Methyl-CoM reductase was specific for H-S-HTP as electron donor. Neither N-6-mercaptohexanoylthreonine phosphate (H-S-HxoTP) nor N-8-mercaptooctanoylthreonine phosphate (H-S-OcoTP) nor any other thiol compound could substitute for H-S-HTP. On the contrary, H-S-HxoTP (apparent Ki = 0.1 microM) and H-S-OcoTP (apparent Ki = 15 microM) were found to be effective inhibitors of methyl-CoM reductase, inhibition being non-competitive with CH3-S-CoM and competitive with H-S-HTP.     

Electrophysiological evidence for the coexistence of alpha1 and alpha6 subunits in a single functional GABA(A) receptor

The subunit combinations alpha1beta2gamma2, alpha6beta2gamma2, and alpha1alpha6beta2gamma2 of the GABA(A) receptor were functionally expressed in Xenopus oocytes. The properties of the resulting ion currents were characterized by using electrophysiological techniques. The concentration-response curve of the channel agonist GABA for alpha1alpha6beta2gamma2 showed a single apparent component characterized by an EC(50) of 107 +/- 26 microM (n = 4). It was different from the one for alpha1beta2gamma2, which had an EC(50) of 41 +/- 9 microM (n = 4), that for alpha6beta2gamma2, with an EC(50) of 6.7 +/- 1.9 microM (n = 5), and those for alpha1beta2 and alpha1alpha6beta2. There was no appreciable functional expression of alpha6beta2. Allosteric responses of alpha1alpha6beta2gamma2 to diazepam were intermediate to those of alpha1beta2gamma2 and alpha6beta2gamma2, and allosteric responses to flumazenil were comparable to the ones for alpha1beta2gamma2. The inhibition by furosemide of the currents elicited by GABA in alpha1alpha6beta2gamma2 [IC(50) = 298 +/- 116 microM (n = 7), assuming only one component] was not identical with inhibition of alpha6beta2gamma2 (IC(50) = 38 +/- 2 microM, n = 4), alpha1beta2gamma2 (IC(50) = 5,610 +/- 910 microM, n = 5), or a mixture of these components (assuming two components). These findings indicate unambiguously the formation of functional GABA(A) receptors containing two different alpha subunits, alpha1 and alpha6, with properties different from those of alpha1beta2gamma2 and alpha6beta2gamma2. Furthermore, we provide evidence for the facts that in the Xenopus oocyte (a) the formation of the different receptor types depends on the relative abundance of cRNAs coding for the different receptor subunits and (b) that functional dual subunit combinations alphabeta do not form in the presence of cRNA coding for the gamma subunit.     

The respiratory nitrate reductase from Paracoccus denitrificans. Molecular characterisation and kinetic properties

The respiratory nitrate reductase from Paracoccus denitrificans has been purified in the non-ionic detergent Nonidet P-40. The enzyme comprises three polypeptides, alpha, beta and gamma with estimated relative molecular masses of 127 000, 61 000 and 21 000. Duroquinol or reduced-viologen compounds acted as the reducing substrates. The nitrate reductase contained a b-type cytochrome that was reduced by duroquinol and oxidised by nitrate. A preparation of the enzyme that lacked both detectable b-type cytochrome and the gamma subunit was obtained from a trailing peak of nitrate reductase activity collected from a gel filtration column. Absence of the gamma subunit correlated with failure to use duroquinol as reductant; activity with reduced viologens was retained. It is concluded that in the plasma membrane of P. denitrificans the gamma subunit catalyses electron transfer to the alpha and beta subunits of nitrate reductase from ubiquinol which acts as a branch point in the respiratory chain. A new assay was introduced for both nitrate and quinol-nitrate oxidoreductase activity. Diaphorase was used to couple the oxidation of NADH to the production of duroquinol which acted as electron donor to nitrate reductase. Under anaerobic conditions absorbance changes at 340 nm were sensitive to nitrate concentrations in the low micromolar range. This coupled assay was used to determine that the purified enzyme had Km(NO-3) of 13 microM and a Km of 470 microM for ClO-3, an alternative substrate. With viologen substrates Km(NO-3) of 283 microM and Km(ClO-3) of 470 microM were determined; the enzymes possessed a considerably higher Vmax with either NO-3 or ClO-3 than was found when duroquinol was substrate. Azide was a competitive inhibitor of nitrate reduction in either assay system (Ki = 0.55 microM) but 2-n-heptyl-4-hydroxyquinoline N-oxide was effective only with the complete three-subunit enzyme and duroquinol as substrate, consistent with a site of action for this inhibitor on the b-type cytochrome. The low Km for nitrate observed in the duriquinol assay is comparable with the apparent Km(NO-3) recently reported for intact cells of P. denitrificans [Parsonage, D., Greenfield, A. J. & Ferguson, S. J. (1985) Biochim. Biophys. Acta 807, 81-95]. This similarity is discussed in terms of a possible requirement for a nitrate transport system. The nitrate reductase system from P. denitrificans is compared with that from Escherichia coli.     

Kinetic and electrophoretic properties of native and recombined isoenzymes of human liver alcohol dehydrogenase

Ten, electrophoretically distinct, molecular forms of alcohol dehydrogenase have been isolated from a single human liver by affinity and ion-exchange chromatography. The starch gel electrophoresis patterns after the dissociation-recombination of the forms are consistent with the hypothesis that they arise from the random combination of alpha, beta 1, gamma 1, and gamma 2 subunits into six heterodimeric and four homodimeric isoenzymes. Large differences in kinetic properties are observed for the homodimeric isoenzymes, alpha alpha, beta 1 beta 1, gamma 1 gamma 1, and gamma 2 gamma 2. At pH 7.5, the Km value of beta 1 beta 1 for ethanol is 0.049 mM and that of alpha alpha is 4.2 mM. Forms gamma 1 gamma 1 and gamma 2 gamma 2 do not obey Michaelis-Menten kinetics at pH 7.5 but exhibit negative cooperativity with Hill coefficients of 0.54 and 0.55 and [S]0.5 values of 1.0 and 0.63 mM, respectively. However, all isoenzymes display Michaelis-Menten kinetics for ethanol oxidation at pH 10.0 with Km values ranging from 1.5 to 3.2 mM. The maximum specific activity of beta 1 beta 1 is considerably lower than that of the other three homodimers at both pH 7.5 and 10.0. The Km values of the four homodimers for NAD+ at pH 7.5 range from 7.4 to 13 microM and those for NADH, from 6.4 to 33 microM. Ki values for NADH range from 0.19 to 1.6 microM. At pH 7.5, the kinetic properties of alpha alpha and beta 1 beta 1, prepared in vitro from dissociated and recombined alpha beta 1, are similar to those of the native homodimers. The forms gamma 1 gamma 1 and gamma 2 gamma 2, prepared from dissociated and recombined alpha gamma 1 and beta 1 gamma 2, respectively, exhibit negative cooperativity with Hill coefficients that are similar to those seen with the respective native homodimers.     

Roles of the narJ and narI gene products in the expression of nitrate reductase in Escherichia coli

Nitrate reductase, released and purified from membrane fractions of Escherichia coli, is composed of three subunits. Formation of the enzyme depends on induction of the nar operon, narGHJI, which is composed of four open reading frames (ORF). Previous studies established that the first two genes in the operon narG and narH encode the alpha and beta subunits, respectively, while formation of the gamma subunit, cytochrome bNR, depends on expression of the promoter distal genes. The narJ and narI genes were subcloned separately into plasmids where each was under the control of the nar promoter. Expression of these plasmids in a mutant which forms only alpha and beta subunits revealed that expression of the narI gene is sufficient to restore normal levels of cytochrome bNR, but expression of both genes is required for assembly of fully active, membrane-bound nitrate reductase. The amino acid composition, the N-terminal sequence, and the sequence of cyanogen bromide fragments derived from the isolated gamma subunit corresponds to that expected for a protein produced by the narI ORF. A protein corresponding to the narJ ORF did not appear to be associated with the purified nitrate reductase complex or with the complex immunoprecipitated from Triton X-100-solubilized membrane preparations. We conclude that narI encodes the gamma subunit (cytochrome bNR) and that while the product of the narJ gene is required for assembly of fully active membrane-bound enzyme it is not tightly associated with the active enzyme.     

Effects of acyclovir and its metabolites on purine nucleoside phosphorylase

Acyclovir (9-(2-hydroxyethoxymethyl)guanine), the clinically useful antiherpetic agent, is an "acyclic" analogue of 2'-deoxyguanosine. Purine nucleoside phosphorylase partially purified from human erythrocytes did not catalyze detectable phosphorolysis of this drug or any of its metabolites (less than 0.07% of the rate with Guo). However, these compounds were competitive inhibitors of this enzyme with Ino as the variable substrate. Acyclovir per se was a relatively weak inhibitor. Its Ki value (91 microM) was much greater than that for its 8-hydroxy metabolite (Ki = 4.7 microM) but less than that for its carboxylic acid metabolite (9-carboxymethoxy-methylguanine) (K'i = 960 microM). The phosphorylated metabolites of acyclovir were more potent inhibitors than were their guanine nucleotide counterparts. At a phosphate concentration of 50 mM, the apparent Ki values for the mono- (120 microM), di- (0.51 microM), and tri (43 microM)-phosphate esters of acyclovir were 1/2, 1/1200, and 1/26 those for dGMP, dGDP, and dGTP, respectively. The concentration of phosphate did not markedly affect the Ki value of acyclovir but dramatically affected those of its phosphorylated metabolites and their nucleotide counterparts. Decreasing phosphate to a physiological concentration (1 mM) decreased the apparent Ki values for the mono-, di-, and triphosphate esters of acyclovir to 6.6, 0.0087, and 0.31 microM, respectively. Inhibition of the enzyme by acyclovir diphosphate was also influenced by pH. This metabolite of acyclovir is the most potent inhibitor of purine nucleoside phosphorylase reported to date. It has some features of a "multisubstrate" analogue inhibitor.     

Dissimilatory sulfite reductase (desulfoviridin) of the taurine-degrading, non-sulfate-reducing bacterium Bilophila wadsworthia RZATAU contains a fused DsrB-DsrD subunit

A dissimilatory sulfite reductase (DSR) was purified from the anaerobic, taurine-degrading bacterium Bilophila wadsworthia RZATAU to apparent homogeneity. The enzyme is involved in energy conservation by reducing sulfite, which is formed during the degradation of taurine as an electron acceptor, to sulfide. According to its UV-visible absorption spectrum with maxima at 392, 410, 583, and 630 nm, the enzyme belongs to the desulfoviridin type of DSRs. The sulfite reductase was isolated as an alpha2beta)gamma(n) (n > or = 2) multimer with a native size of 285 kDa as determined by gel filtration. We have sequenced the genes encoding the alpha and beta subunits (dsrA and dsrB, respectively), which probably constitute one operon. dsrA and dsrB encode polypeptides of 49 (alpha) and 54 kDa (beta) which show significant similarities to the homologous subunits of other DSRs. The dsrB gene product of B. wadsworthia is apparently a fusion protein of dsrB and dsrD. This indicates a possible functional role of DsrD in DSR function because of its presence as a fusion protein as an integral part of the DSR holoenzyme in B. wadsworthia. A phylogenetic analysis using the available Dsr sequences revealed that B. wadsworthia grouped with its closest 16S rDNA relative Desulfovibrio desulfuricans Essex 6.     

Consequence of the presence of two different beta subunit isoforms in a GABA(A) receptor

The major isoforms of GABA(A) receptors are thought to be composed of two alpha, two beta and one gamma subunit(s). GABA(A) receptors containing two beta1 subunits respond differently to the anticonvulsive compound loreclezole and the general anaesthetic etomidate than receptors containing two beta2 subunits. Receptors containing beta2 subunits show a much larger allosteric stimulation by these agents than those containing beta1 subunits. We were interested to know how receptors containing both beta1 and beta2 subunits, in different positions respond to loreclezole and etomidate. To answer this question, subunits were fused at the DNA level to form dimeric and trimeric subunits. Concatenated receptors (alpha1-beta1-alpha1/gamma2-beta1, alpha1-beta2-alpha1/gamma2-beta1, alpha1-beta1-alpha1/gamma2-beta2 and alpha1-beta2-alpha1/gamma2-beta2) were expressed in Xenopus ooctyes and functionally compared in their response to the agonist GABA and to the positive allosteric modulators, loreclezole and etomidate. We have shown that (I) in the presence of both beta1 and beta2 subunits in the same pentamer (mixed receptors) direct gating by etomidate is similar to exclusively beta1 containing receptors; (II) In mixed receptors, stimulation by etomidate assumed characteristics intermediate to exclusively beta1 or beta2 containing receptors, but the values for the concentrations < 10 microM were always much closer to those observed in alpha1-beta1-alpha1/gamma2-beta1 receptors; and (III) mixed receptors show no positional effects.     

Antisense suppression of GABA(A) receptor beta subunit levels in cultured cerebellar granule neurons demonstrates their importance in receptor expression

GABA(A) receptors in the CNS are pentameric molecules composed of alpha, beta, gamma, delta, epsilon and theta subunits. Studies on transfected cells have shown that GABA(A) receptor beta subunit isoforms can direct alpha1 subunit localization within the cell. To examine the role of selected subunits in governing GABA(A) receptor expression in neurons, cultures of rat cerebellar granule cells were grown with antisense or sense oligodeoxynucleotides (ODNs) specific for the alpha 1, beta 2 or gamma 2 subunits. These subunits are all expressed in granule neurons where they are thought to contribute to an abundant receptor type. Following ODN treatment, subunit expression and distribution were examined by western blotting, immunocytochemistry and RT-PCR. Treatment of the cultures with the antisense, but not the corresponding sense, ODNs reduced the levels of the targeted subunit polypeptides. In addition, the beta 2 antisense ODN reduced the level of the alpha1 subunit polypeptide without altering the level of its mRNA. In contrast, treatment with the beta 2 subunit antisense ODN did not alter gamma 2 subunit polypeptide expression, distribution or mRNA level. These findings suggest that the alpha1 subunit requires a beta subunit for assembly into GABA(A) receptors in cerebellar granule neurons.     

Rhodopsin and the retinal G-protein distinguish among G-protein beta gamma subunit forms

The beta gamma subunits of G-proteins are composed of closely related beta 35 and beta 36 subunits tightly associated with diverse 6-10 kDa gamma subunits. We have developed a reconstitution assay using rhodopsin-catalyzed guanosine 5'-3-O-(thio)triphosphate (GTP gamma S) binding to resolved alpha subunit of the retinal G-protein transducin (Gt alpha) to quantitate the activity of beta gamma proteins. Rhodopsin facilitates the exchange of GTP gamma S for GDP bound to Gt alpha beta gamma with a 60-fold higher apparent affinity than for Gt alpha alone. At limiting rhodopsin, G-protein-derived beta gamma subunits catalytically enhance the rate of GTP gamma S binding to resolved Gt alpha. The isolated beta gamma subunit of retinal G-protein (beta 1, gamma 1 genes) facilitates rhodopsin-catalyzed GTP gamma S exchange on Gt alpha in a concentration-dependent manner (K0.5 = 254 +/- 21 nM). Purified human placental beta 35 gamma, composed of beta 2 gene product and gamma-placenta protein (Evans, T., Fawzi, A., Fraser, E.D., Brown, L.M., and Northup, J.K. (1987) J. Biol. Chem. 262, 176-181), substitutes for Gt beta gamma reconstitution of rhodopsin with Gt alpha. However, human placental beta 35 gamma facilitates rhodopsin-catalyzed GTP gamma S exchange on Gt alpha with a higher apparent affinity than Gt beta gamma (K0.5 = 76 +/- 54 nM). As an alternative assay for these interactions, we have examined pertussis toxin-catalyzed ADP-ribosylation of the Gt alpha subunit which is markedly enhanced in rate by beta gamma subunits. Quantitative analyses of rates of pertussis modification reveal no differences in apparent affinity between Gt beta gamma and human placental beta 35 gamma (K0.5 values of 49 +/- 29 and 70 +/- 24 nM, respectively). Thus, the Gt alpha subunit alone does not distinguish among the beta gamma subunit forms. These results clearly show a high degree of functional homology among the beta 35 and beta 36 subunits of G-proteins for interaction with Gt alpha and rhodopsin, and establish a simple functional assay for the beta gamma subunits of G-proteins. Our data also suggest a specificity of recognition of beta gamma subunit forms which is dependent both on Gt alpha and rhodopsin. These results may indicate that the recently uncovered diversity in the expression of beta gamma subunit forms may complement the diversity of G alpha subunits in providing for specific receptor recognition of G-proteins.     

Two invariant tryptophans on the alpha1 subunit define domains necessary for GABA(A) receptor assembly.

Two invariant tryptophan residues on the N-terminal extracellular region of the rat alpha1 subunit, Trp-69 and Trp-94, are critical for the assembly of the GABA(A) (gamma-aminobutyric acid, type A) receptor into a pentamer. These tryptophans are common not only to all GABA(A) receptor subunits, but also to all ligand-gated ion channel subunits. Converting each Trp residue to Phe and Gly by site-directed mutagenesis allowed us to study the role of these invariant tryptophan residues. Mutant alpha1 subunits, coexpressed with beta2 subunits in baculovirus-infected Sf9 cells, displayed high affinity binding to [(3)H]muscimol, a GABA site ligand, but no binding to [(35)S]t-butyl bicyclophosphorothionate, a ligand for the receptor-associated ion channel. Neither [(3)H]muscimol binding to intact cells nor immunostaining of nonpermeabilized cells gave evidence of surface expression of the receptor. When expressed with beta2 and gamma2 polypeptides, the mutant alpha1 polypeptides did not form [(3)H]flunitrazepam binding sites though wild-type alpha1 polypeptides did. The distribution of the mutant receptors on sucrose gradients suggests that the effects on ligand binding result from the inability of the mutant alpha1 subunits to form pentamers. We conclude that Trp-69 and Trp-94 participate in the formation of the interface between alpha and beta subunits, but not of the GABA binding site.     

Reconstitution of the voltage-sensitive calcium channel purified from skeletal muscle transverse tubules

The purified calcium antagonist receptor of the voltage-sensitive calcium channel from skeletal muscle transverse tubule membrane consists of three subunits: alpha with Mr 135 000, beta with Mr 50 000, and gamma with Mr 33 000. Purified receptor preparations were incorporated into phosphatidylcholine (PC) vesicles by addition of PC in 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate and removal of detergent by molecular sieve chromatography. Forty-five percent of the alpha, beta, and gamma polypeptides and the [3H]dihydropyridine/receptor complex were recovered in association with PC vesicles. The rate of dissociation of the purified and reconstituted dihydropyridine/receptor complex was identical with that in T-tubule membranes, and allosteric modulation by verapamil and diltiazem was retained. The reconstituted calcium antagonist receptor, when occupied by the calcium channel activator BAY K 8644, mediated specific 45Ca2+ and 133Ba2+ transport into the reconstituted vesicles. 45Ca2+ influx was blocked by the organic calcium antagonists PN200-110 (K0.5 = 0.2 microM), D600 (K0.5 = 1.0 microM), and verapamil (K0.5 = 1.5 microM) and by inorganic calcium channel antagonists (La3+ greater than Cd2+ greater than Ni2+ greater than Mg2+) as in intact T-tubules. A close quantitative correlation was observed between the presence of the alpha, beta, and gamma subunits of the calcium antagonist receptor and the ability to mediate 45Ca2+ or 133Ba2+ flux into reconstituted vesicles. Comparison of the number of reconstituted calcium antagonist receptors and functional channels supports the conclusion that only a few percent of the purified calcium antagonist receptor polypeptides are capable of mediating calcium transport as previously demonstrated for calcium antagonist receptors in intact T-tubules.     

Potent inhibitory effects of the 5'-triphosphates of (E)-5-(2-bromovinyl)-2'-deoxyuridine and (E)-5-(2-bromovinyl)-1-beta-D-arabinofuranosyluracil on DNA polymerase gamma

(E)-5-(2-Bromovinyl)-2'-deoxyuridine 5'-triphosphate (BrVdUTP) and (E)-5-(2-bromovinyl)-1-beta-D-arabinofuranosyluracil 5'-triphosphate (BrVarafUTP), which are known as specific inhibitors of herpes simplex viral (type 1 and 2) DNA polymerase, were found to be strong inhibitors of DNA polymerase gamma from human KB and murine myeloma cells. In fact BrVdUTP and BrVarafUTP were found to be stronger inhibitors of DNA polymerase gamma than of other DNA polymerases having viral (herpes simplex virus or retrovirus) origin or cellular (eukaryotic alpha and beta, or prokaryotic) origin. The mode of inhibition of DNA polymerase gamma by BrVdUTP and BrVarafUTP was competitive with respect to dTTP, the normal substrate. Whereas BrVdUTP was an efficient substrate for DNA polymerase gamma and other DNA polymerases that were examined, BrVarafUTP failed to serve as a substrate for DNA synthesis. Ki values for BrVdUTP (40 nM) and BrVarafUTP (7 nM) with DNA polymerase gamma, as determined with (rA)n.(dT) as the template.primer, were much smaller than the Km values for dTTP (0.16 microM and 0.71 microM for murine and human DNA polymerase gamma, respectively). Thus, the affinity of BrVdUTP or BrVarafUTP for DNA polymerase gamma was much stronger than that of dTTP.     

Identification of a gamma subunit associated with the adenylyl cyclase regulatory proteins Ns and Ni

The subunit composition of the Ns and Ni, the human erythrocyte stimulatory and inhibitory regulatory proteins of adenylyl cyclase, respectively, were analyzed by a sodium dodecyl sulfate-containing discontinuous urea and polyacrylamide gradient gel electrophoresis system designed for the study of low molecular weight polypeptides. This system disclosed that these proteins, in addition to their known alpha and beta subunits, contain an additional small peptide of apparent molecular weight of 5,000 (5K). This "5K peptide" is also present in preparations of another protein which we termed "40K protein" on the basis of its hydrodynamic behavior and whose primary protein constituent is the Mr 35,000 beta subunit of the above regulatory proteins. Analyzing Ni, the 5K peptide was functionally related to the protein by showing that its apparent Stokes radius changes from 5.9 to 5.1 nm after treatment with guanyl-5'-yl imidodiphosphate and magnesium in parallel with the alpha and beta subunits. These data are interpreted as evidence for the existence of a third subunit associated with the regulatory proteins of adenylyl cyclase. We call this subunit gamma and propose a minimum subunit structure for these proteins of the alpha beta gamma type.     

Assembly of a functional F1 of the proton-translocating ATPase of Escherichia coli

Assembly of the F1 portion of the proton-translocating ATPase of Escherichia coli was examined in vivo. Analysis of strains lacking genes which specify the Fo polypeptides a, b, and c showed that the F1 subunits were able to assemble into a complex in the absence of the Fo subunits. In addition we have investigated the effects of mutations in the individual genes which specify the F1 polypeptides on the assembly process. Mutations of the uncA(alpha), uncG(gamma), or uncD(beta) genes result in a defective assembly of the F1 complex. In contrast, mutations in the uncH(delta) or uncC(epsilon) genes did not prevent assembly of the core alpha beta gamma complex. In these cases, however, the partial F1 complexes were incapable of restoring energy-linked functions to F1-depleted membranes.     

The sodium channel from rat brain. Separation and characterization of subunits

Procedures are described for separation of the alpha, beta 1, and beta 2 subunits of the voltage-sensitive sodium channel from rat brain by gel filtration in sodium dodecyl sulfate (SDS) before and after reduction of intersubunit disulfide bonds or by preparative SDS-gel electrophoresis. Partial proteolytic maps of the SDS-denatured subunits indicate that they are nonidentical polypeptides. They are all heavily glycosylated and contain complex carbohydrate chains that bind wheat germ agglutinin. The apparent molecular weights of the separated subunits were estimated by gradient SDS-gel electrophoresis, by Ferguson analysis of migration in SDS gels of fixed acrylamide concentration, or by gel filtration in SDS or guanidine hydrochloride. For the alpha subunit, SDS-gel electrophoresis under various conditions gives an average Mr of 260,000. Gel filtration methods give anomalously low values. Removal of carbohydrate by sequential treatment with neuraminidase and endoglycosidase F results in a sharp protein band with apparent Mr = 220,000, suggesting that 15% of the mass of the native alpha subunit is carbohydrate. Electrophoretic and gel filtration methods yield consistent molecular weight estimates for the beta subunits. The average values are: beta 1, Mr = 36,000, and beta 2, Mr = 33,000. Deglycosylation by treatment with endoglycosidase F, trifluoromethanesulfonic acid, or HF yields sharp protein bands with apparent Mr = 23,000 and 21,000 for the beta 1 and beta 2 subunits, respectively, suggesting that 36% of the mass of the native beta 1 and beta 2 subunits is carbohydrate.