Biochem. J. (1970) 120, 255-261Printed in Great Britain

255

Molecular Weight of Escherichia coli p-Galactosidase in ConcentratedSolutions of Guanidine Hydrochloride

ROBERT P. ERICKSON*Laboratory of Chemical Biology, National Institute of Arthritis and Metabolic Diseases, National Institutes

of Health, Bethesda, Md. 20014, U.S.A.

(Received 5 June 1970)

The molecular weight of Escherichia coli f-galactosidase was determined in 6M-and 8 M-guanidine hydrochloride by meniscus-depletion sedimentation equilibrium,sedimentation velocity and viscosity. Sedimentation equilibrium revealed hetero-geneity with the smallest component having a molecular weight of about 50 000. Atlower speeds, the apparent weight-average molecular weight is about 80000. Byuse of a calculation based on an empirical correlation for proteins that are randomcoils in 6M-guanidine hydrochloride, sedimentation velocity gave a molecularweight of 91 000, and the intrinsic viscosity indicated a viscosity-average molecularweight of 84000. Heating in 6M-guanidine hydrochloride lowered the viscosity of/3-galactosidase in a variable manner.

The molecular weight of f-galactosidase (f-D-galactoside galactohydrolase, EC 3.2.1.23) purifiedfIrom Escherichia coli has previously been deter-mined in 6M-guanidine hydrochloride and foundto be in general accord with the 'inonomer' mole-cular weight of 135000 (Craven, 1967; Ullmann,Goldberg, Perrin & Monod, 1968a). The latterstudy utilized low-speed sedimentation equilibria,which would poorly distinguish multiple com-ponents, and an enmpirical correlation of apparentmolecular weights in guanidine hydrochlorideLlimc-oMapp.(1l-p)] to actual nmolecular weightsdetermined for several well-studied proteins, thusneglecting variable buoyancies. This result hasbeen used to argue that the 'monomer' is onepolypeptide chain (Ullmann, Jacob &Monod, 1968b)despite the fact that it is easily dissociable intosmaller subunits by detergent, dilute acid or dilutebase (Wallenfels, Sund & Weber, 1963; Steers,Craven & Anfinsen, 1965a). We find the molecularweiglhts of f-galactosidase determined in con-centrated solutions of guanidine hydrochloride tobe significantly lower than is compatible with asingle chain of molecular weight 135 000. Apreliminary report of these results has been publish-ed (Erickson & Steers, 1969a). Evidence for blockedN-terminal amino acids in /3-galactosidase aspyrrolidonecarboxylic acid has been reportedelsewhere (Erickson & Steers, 1969b).

* Present address: Department of Pediatrics, Univer-sity of California Medical Center, San Francisco, Calif.94122, U.S.A.

MATERIALS AND METHODS

Purification of f-galactosidase. The ,-galactosidaseused in this study was prepared from the regulator.constitutive strain of E. coli K,2 3300 by the procedure ofMarchesi, Steers & Shifrin (1969). Briefly, the bacteriallysate was fractionated with (NH4)2SO4 and the 20-40%-saturation fraction was subjected to gel filtration onSepharose 4-B (Pharmacia Fine Chemicals, Uppsala,Sweden) in buffer B (0.1M-NaCl-0.0lM-MgCl2-0.05M-2-mercaptoethanol-0.05M-tris-HCl buffer, pH 7.4). Themajor peak of activity was further purified by ion-exchange chromatography on DEAE-cellulose (WhatmanDE23; Reeve Angel, Clifton, N.J., U.S.A.) with a com-bination pH-NaCl gradient. Sometimes a SephadexG-200 (Pharmacia) purification step was interposedbetween the first gel-filtration and the ion-exchangesteps. Purity was checked by polyacrylamide disc gelelectrophoresis. Amidoschwarz (Allied Chemical Corp.,New York, N.Y., U.S.A.) staining revealed no conitaminat-ing bands when 100-200k,g quantities of the purifiedenzyme were subjected to electrophoresis on 5% poly-acrylamide gels at 3 mA/tube for 3 h. The purified proteinwas stored under 50% (w/v) (NH4)2804 at 4°C.

Preparation of solutions of ,B-galactosidase in concentratedaqueous guanidine hydrochloride. Spectral-quality guani-dine hydrochloride was obtained from Aldrich ChemicalCo., Milwaukee, Wis., U.S.A., or from Mann ResearchLaboratories, New York, N.Y., U.S.A. A fl-galactosidasesolution in buffer B was diluted to the desired concentra-tion with (or directly dialysed against) 6M- or 8M-guanidinehydrochloride-0.1 M- 2 - mercaptoethanol- 0.1 M - tris - HCIbuffer, pH 7.5, and dialysed against three lOOml portionsofthe buffer for 3 days. The resultant solution (and controlsamples of dialysate) were filtered with pressure through0.7-1 ,um millipore filters (Millipore Filter Corp., Bedford,

R. P. ERICKSON

Mass., U.S.A.) Concentrations in the usual buffers werecalculated by using an extinction coefficient at 280nmof 2.09mg/ml for a 1cm light-path (Craven, Steers &Anfinsen, 1965). By difference spectroscopy in theCary model 15 recording spectrometer, an extinctioncoefficient at 280nm of 2.207mg/ml (1cm light-path)was found for fi-galactosidase in 6M- or 8m-guanidineand this value was used to calculate the proteinconcentrations of guanidine solutions.

Sedimentation-equilibrium measurement8. High-speedsedimentation-equilibrium experiments were performedas described -by Yphantis (1964) at 20°C, double-sectorduraluminum-filled epoxy or six-sector carbon-filledepoxy centrepieces and sapphire windows being used.[Base fluid FC43 (Beckman, Palo Alto, Calif., U.S.A.)was used in the two- but not the six-sector experiments.]Optical blanks were obtained by a preliminary run withwater in the cells. One window assembly was then removed(without dismantling the windows or cell centrepiece),water was carefully removed and the cell was dried overP205. The cell was then ready for the experimental run.The concentrations across the cell were recorded with theinterference optical system after 24h at each increasingspeed. Equilibrium had been achieved, as exposures at18h were occasionally taken and were not measurablydifferent from those at 24h. The Spinco model E instru-ment equipped with electronic speed control was used andphotographic plates were measured with a Nikon 6Ccomparator. Densities of concentrated aqueous solutionsof guanidine hydrochloride were taken from Kawahara &Tanford (1966). The effective specific volume, b, was takenas 0.75ml/g, 0.01 ml/g less than the previously determinedpartial specific volume of 0.76 (Wallenfels, Sund & Weber,1963) following Tanford (Tanford, Kawahara & Lapanje,1967). A computer program in Basic fitted linear least-squares slopes for plots of Inc versus r2, determined trueslopes for the second component by subtracting theextrapolated first slope from the observed second slopewhen polydispersity was indicated by breaks in the Incversus r2 plots, calculated molecular weights, and deter-mined amounts of each component by integrating areasunder the curve of c versus r2.

Sedimentation-velocity determinations. The duralumi-num-filled epoxy, capillary artificial-boundary, double-sector cell was used with sapphire windows. The deter-minations were carried out at 56000rev./min and 20°C byusing schlieren optics in the analytical ultra-centrifugeequipped with electronic speed control. Measurements ofboundary movement were made with the Nikon 6Ccomparator.

Viscosity measurements. Viscosities were determined at20.0±0.010C in an Ostwald viscometer of the dilutiontype. Protein concentrations were checked on the finaldilution of the series. Some determinations were madewith a four-chamber Ubbelohde viscometer that allowedvariation in shear forces.

RESULTS

Sedimentation-equilibrium experiments by themeniscus-depletion method gave weight-averagevalues for the molecular weight of fl-galactosidaseof 70000-80000 in concentrated guanidine hydro-

Table 1. Sedimentation-equilibrium-determinedmolecular weights of [3-galactosidase in 8M-guanidinehydrochloride

Results are given ±5.E.M. for the plot of Inc versus r2; ifonly one value is listed the plot is unbroken; if two valuesare listed the second molecular weight is that of thesecond component as determined by subtraction of theextrapolated first slope from the observed second slope.

Molecular weight

Speed Unheated (approx.(rev./min) 0.3mg/ml)36000 -32000 53000±1060

212000± 850028000 70000±70024000 84000± 840

-3.4-3.6

-4.0-4.2

-4.6-4.8

c -5.2-4

Heated at 100°C for20min (0.3mg/ml)

53000± 106049000± 2450156000 ± 9350

/-5.81-

-6.4

0 ,

42.5 43.0r2 (cm2)

44.0

Fig. 1. Sedimentation equilibrium of ,-galactosidase(0.25mg/ml) in 6M-guanidine hydrochloride after 24h at36000rev./min at 20°C.

chloride at moderate speeds (Table 1). At higherspeeds heterogeneity became apparent (Fig. 1).Here three segments of the curve are distinguish-able; the three curves are distinguishable onmultiple experiments. The first component endsat one fringe, so that the value calculated is ofquestionable accuracy. However, it is given inTable 2 and used in Fig. 3. If the first slope is con-sidered to be erroneous and the second slope is takento be the initial slope, a molecular weight of 70 000 iscalculated for this slope. That the 70000 valuerepresents the weight average of several com-ponents and the first slope truly represents theinitial component becomes clearer when a solutionof 1 mg/ml is run under the same conditions(Fig. 2). The initial slope is now much clearer andextends for two fringes although we have obscured

1970256

Vol. 120 MOLECULAR WEIGHT OF ,B-GALACTOSIDASE IN GUANIDINE

Table 2. Sedimentation-equilibrium-determined molecular weights of fl-galactosidase in 6M-guanidinehydrochloride

Results are given ±S.E.M. for the plot of In c versus r2; ifonly one value is listed the plot is linear; if two valuesare listed, the second molecular weight is that of the second component as determined by subtraction of theextrapolated first slope from the observed second slope.

Molecular weight

Speed (rev./min)4000036000

Heated at 100°C for 20min(0.25mg/ml)52000 49000

0.25mg/ml

50000± 2000167000± 27900

-3.4F-4.0p-

-4.6F

-5.2

-5.8

-6.4

-6.849.

I S e

.3 50.0 51.0r2 (cm2)

Fig. 2. Sedimentation equilibrium of fl-galactosidase(1mg/ml) in 6M-guanidine hydrochloride after 24h at36000rev./min at 200C.

the third slope. A molecular weight of 35000 isindicated by the first slope, and subtraction of theextrapolated first slope from the second observedslope shows that the component contributing tothe second slope has a mnolecular weight of about117000 (Table 2; non-heated, lmg/ml). Some ofthe values were determined after heating thesolution of /-galactosidase in concentrated guani-dine hydrochloride and an identically treatedportion of the final dialysate (for the blank) at 1000Cfor 20min. This was because ofa change in viscositynoted after such treatment (see below). No signi-ficant changes in sedimentation-equilibrium valueswere noted and the values are included with theother determinations.

Although only two protein concentrations were

used in these experiments, variable amounts ofprotein were sedimented to the bottom of theultracentrifuge cell as a function ofthe speed so thata range of concentrations was achieved in thesolutions. Since the volume of solvent remained thesame, the concentration of the component of lowestmolecular weight is measured by the determination

9

of the quantity in solution from integrating the area

under the curve of c versus r2 with boundaries ofthemeniscus and base. The plot ofthese concentrationsagainst the molecular weight shows little depend-ence of molecular weight on concentration (Fig. 3),although there is a fair amount of scatter. (Ingeneral, sedimentation-equilibrium determinationsin concentrated solutions of guanidine hydrochlo-ride are less precise than those in usual solventsbecause of variations in density with exposure ofsolvent to the atmosphere, leakage due to 'creep' ofthe solutions between windows and centrepieces,etc.) However, the concentration determined bythis integration is for the whole cell rather than theconcentration at the point where the values generat-ing the molecular-weight value are obtained. Assuch, the plot is preferable to one of initial proteinconcentrations for estimating concentration effectsbut not as good as point weight-average molecularweights would be. The latter, however, cannot beobtained over any significant concentration range

for the multicomponent curves herein obtained.The results shown in Tables 1 and 2 and Fig. 3

leave little doubt that the smallest component ofP-galactosidase in concentrated solutions of guani-dine hydrochloride has a inolecular weight of about50000. The molecular weight of the second com-

ponent is much more variable. Of the four valuesin Tables 1 and 2, the range is from 116000 to212 000 with a mean of about 160 000. This is mostcompatible with an undissociated-monomer weightof 135000. Since a third component is observedbeyond this in some experiments (Fig. 1), and it ishighly unlikely that there is any component largerthan undissociated monomer in these solutions, it ispossible that the second component represents a

fragment of P-galactosidase remaining when a

50000 dalton piece is dissociated from the 135000dalton monomer. On this interpretation, the secondcomponent should have a molecular weight ofabout 90000 and the third component would beundissociated monomer ofmolecular weight 135 000.The large discrepancy between a theoretical

Bioch. 1970, 120

1mg/ml42000± 295035000± 1400117000±5850

257

R. P. ERICKSON

c; 60

o 50, 40

e w 30

; 20x 10

- 0

FH-.--F I-F

70

600 so

F

A A A

_

30 F

1 2 3 4 5 6 7 8Quantity of component

20 _

10

Fig,. 3. Concentration dependence of the molecular weightof the component of f-galactosidase of lowest molecularweight in concentrated guanidine hydrochloride solutionis.The abscissa represents the absolute quantity of thecomponent in the solution phase, Q = frbase. c( r)dr2.Results are from heated or non-heated enzyme, (AtI- or8M-guanidine hydrochloride.

0 0.2 0.4 0.6 0.8Conen. of ,B-galactosidase (g/100nml)

.0

Fig. 5. Viscosity measurements of f-galactosidase in6 m-guanidine hydrochloride. 0, Uniheated; 0, heated for20min at 100°C; A, heated for 20min at 100IC 10 daysbefore viscosity determination.

0._'.

0- - 0.8

(: 0.6

_0$- C. 0.4

no

I nl_Fr .

40

1 ~~ ~ ~

0 0.2 0.4 0.6 0.8 1.0Conen. of f-galactosidase (g/100 ml)

Fig. 4. Sedimentation velocity of fl-alactosidase in6M-guanidine hydrochloride at 20°C.

second-component mnolecular weight of 90000 andthe observed value is only partly explainable bythe error amplification that occurs during thesubtraction of the extrapolated first slope from theobserved second slope.Only a single peak was visible when f-galacto-

sidase was run in sedimentation-velocity experi-ments in 6 M-guanidine hydrochloride. For thesevery low sedimentation values, we would not expectthe several components to be resolved as individualpeaks. The values of 80, the apparent sedimentationcoefficient, obtained at 20°C are plotted againstconcentration in Fig. 4. An 8243 guanidine hydrochloride

of 0.95 results when the plot is linearly extrapolatedto zero protein concentration. It is possible thatwe are ignoring an upward deflexion at low con-

centrations of fl-galactosidase but nonlinear plotswere not noted by Tanford et al. (1967). (They doemphasize the poor precision of the results for s

valtues in concentrated solutions of guanidinehydrochloride.) The number, n, of amino acid

residues per chain in a randomn-coil peptide inguanidine hydrochloride may be calculated fromTanford's empirical equation (Tanford et al. 1967):

80/(1 - 'p) = 0.286n° 473

where p is the density of the solvent and O' theeffective specific volume. From this formula, thenumber of residues calculated is 790. Taking aresidue-average molecular weight of 114.9 (Craven,Steers & Anfinsen, 1965), this corresponds to amolecular wveight of 91000. The deficienciesinherent in this approach include the decreasedaccuracy for the apparent sedimentation coefficientwhen several components are being 'averaged' inthe peak, amplification of any error in partialspecific volume (q) by the high solvent-densityterm (p), and the amplification of error in the experi-mental value by the exponential relationship of thenumber of residues, n, to it. The latter factor alsoeffects the relationship of molecular weight toviscosity in concentrated guanidine hydrochloridesolutions.The intrinsic viscosity measured in 6M-guanidine

hydrochloride was high (Fig. 5). On multipledeterminations, the intrinsic viscosity of fl-galacto-sidase in 6M-guanidine hydrochloride was60 + 3 ml/g. The inclusion of 10mM-EDTA or10mM-magnesium chloride in the solvent or re-duction and carboxymethylation (Steers, Craven &Anfinsen, 1965b) of the protein were without effecton the viscosity. Again, the number, n, of aminoacid residues per chain in a random-coil peptide inguanidine hydrochloride may be calculated from asecond empirical formula (Tanford et al. 1967):

[ii] = 0.761 N0.66

258 1970

1.I

Vol. 120 MOLECULAR WEIGHT OF 3-GALACTOSIDASE IN GUANIDINEwhere [vi] is the intrinsic viscosity. Fronm thisformula, the number of residues calculated is 740corresponding to a molecular weight of 84000. Thisis a viscosity-average molecular weight.

WN'hen ,B-galactosidase is heated for 20min at100°C in 6m-guanidine hydrochloride-0.1 M-2-mercaptoethanol-0.IM-tris-HCl buffer, pH7.5,there is a marked change in the intrinsic viscosity(Fig. 5). A similar volume of solvent is heated in thesame way for the blank. This result is reproduciblebut the degree of decreased viscosity is variable.The fall in viscosity is not found ifthe concentrationof 3-galactosidase is greater than 8-lOmg/ml whenit is heated. The change in viscosity is not due toloss of the volatile 2-mercaptoethanol. Theaddition of more mercaptoethanol to the proteinsolution and the blank after heating did not changethe results and reduced carboxymethylated /3-galactosidase also showed a similar fall in intrinsicviscosity after being heated in 6m'-guanidinehydrochloride. Such a change in viscosity withheating in 6M-guanidine hydrochloride solutionsis not found with bovine serum albumin. Deter-minations in a four-chamber Ubbelohde viscometerat different shear forces showed a trend for the /3-galactosidase heated in 6m-guanidine hydrochlorideto be less viscous at lower shear forces, but this wasnot observed with the non-heated solutions. Thisrepresented a total variation of 5% and is un-explained, if it is indeed significant. The change inviscosity with heatinginthepresence ofconcentratedguanidine hydrochloride seems reversible, as theinitial intrinsic viscosity gradually returns whilethe ,B-galactosidase remains in the guanidinehydrochloride solution (Fig. 5). In 6M-guanidinehydrochloride the recovery takes about 10 days.

DISCUSSION

The measurements of the molecular weight of/3-galactosidase in solution in weak acid, weak baseand detergent might be artifically low because ofcharge effects; careful measurements at low con-centrations are needed. This source of error seemsunlikely to apply to the sedimentation-equilibriumresults in concentrated solutions of guanidinehydrochloride and the concentration plot of Fig. 3argues against such an effect. The sedimentation-velocity method determines the molecular weightof a component if the molecule follows random-coilbehaviour in concentrated guanidine hydrochloride.The value of 91000 determined might represent theaverage for several components. Thus the values of80 000-90 000 daltons found as averages bymeniscus-depletion sedimentation-equilibrium atlower speeds, by sedimentation velocity and byviscosity determinations might represent a mixtureof about 35% 50000 dalton material, about 35%

90000 dalton mnaterial and about 30% 135 000dalton material.The integrated areas under the c versus r2 curves

always show that the order of quantities of com-ponents is first> second> third. This is not un-expected, as each heavier component in turn wouldlose a greater proportion by sedimentation to thebottom of the cell. The presence of undissociatedmonoiner is suggested by the change in viscosity onheating in concentrated guanidine hydrochloride.The viscosity determination should result in aviscosity-average molecular weight if there areseveral components in the solution of 6M-guanidinehydrochloride. The method could give too low aresult if the protein(s) is incompletely uncoiled.However, the decrease in viscosity on heating inguanidine hydrochloride suggests that dissociationis incomplete. This average molecular weight of84000 is close to that determined at lower speedsduring the sedimentation-equilibrium studies.

These results are not in agreement with those ofUllmann et al. ( 1968a) or the single determination byGoldberg & Edelstein (1969). However, Goldberg &Edelstein (1969) detected approximately equiinolarconcentrations of 43000 dalton and 109000 daltoncomponents in complemented ,B-galactosidase.Their result for complemented /3-galactosidase isvery similar to these results with native /3-galacto-sidase. The discrepancy is unlikely to be due to adifference in strains of E. coli (3300 is a PasteurInstitute strain) but there could be strain differencesin quantity of interpeptide-chain bonds. It isunlikely that our preparations of /3-galactosidasehave been modified by proteolytic enzymes. Thenolecular weight of these preparations remains135000 in 8m-urea. This should be compared withtrypsin-treated /-galactosidase which retains fullactivity and has a sedimentation coefficient of 16Sfrom velocity measurements, but is dissociated intofour small components in 8M-urea (Givol, Craven,Steers & Anfinsen, 1966). The point has previouslybeen made that ,B-galactosidase is dissociable intocomponents smaller than the monomer by treat-ments known to break ester bonds (Craven, 1967).Such results suggest that the /3-galactosidase mono-mer is actually a protomer of several polypeptidechains held together by covalent bonds other thanthe peptide ones. If so, these 'ester' bonds must beat least partially dissociable by guanidine hydro-chloride or be present in variable amounts, e.g. aproportion of /-galactosidase protomer would havefewer interpeptide-chain bonds.

Difference spectra and fluorescence-emissionspectra show that the optically active amino acids,tyrosine and tryptophan, are fully exposed to thesolvent in 4M-guanidine hydrochloride after 3h(R. P. Erickson, unpublished work). In 6- and8m-guanidine hydrochloride the spectra are stable

259

260 R. P. ERICKSON 1970almost immediately; there is a slight increase inquenching of the absorption spectra as the con-centration of guanidine increases from 4 to 8M.These results suggest that only a small part of thefl-galactosidase protomer, if any, is resistant todenaturationbyconcentrated solutions ofguanidine.Although all of the proteins studied for random-

coil behaviour in 6M-guanidine hydrochloride byTanford's group were denatured at 250C, denatura-tion was not complete until 12h for bovine serumalbumin (Tanford, Kawahara & Lapanje, 1967).Harrington's group found that myosin had to beheated at 550C for unspecified times to be com-pletely dissociated in 6M-guanidine hydrochloride(Woods, Himmerfarb & Harrington, 1963). Norm-ally iodinated thyroglobulin is not completelydevoid of ordered structure by the criteria ofviscosity, optical rotatory dispersion and fluoresc-ence polarization when reduced with 15mm-2-mercaptoethanol in 5.8M-guanidine hydrochlorideat neutral pH (Edelhoch & Steiner, 1966). Further,synthetic poly-L-leucine shows helical stability in7.2M-guanidine hydrochloride at temperatures upto 95°C (Auer & Doty, 1966). Avidin retains anormal spectrum in 6.4M-guanidine hydrochlorideif complexed with biotin; biotin will reverse thedenaturation spectra only at 3M-guanidine hydro-chloride (Green, 1963). Thus, ,B-galactosidase wouldnot be unique if it partially resists denaturation inconcentrated guanidine hydrochloride.

Presumptive genetic evidence for multipleproducts of the z gene, the structural gene for ,B-galactosidase, was provided byUllmann et al. (1965)and Ullmann, Jacob & Monod (1967) with studiesof complementation between different z-genedeletions, a complementation mediated byseparablepolypeptides. Recently this group studied /3-galactosidase purified from strains bearing operator-distal point mutations, s908, u366, sexducted withF-lac bearing an operator-proximal deletion, B9(Ullmann et al., 1968b). They found that the /l-galactosidase purified from these complementarydiploids differed from the wild-type enzyme in heatstability and for the quantity of complementingpeptides extractable. A proportion of the 'com-plemented' enzyme was always similar to thewild-type. The authors concluded that the com-plemented ,B-galactosidase was produced by associa-tion between peptides corresponding to differentfragments of the wild-type chain with redundancyof some amino acids, i.e. an unused 'tail' mustproject from the folded monomer (Goldberg, 1969).The wild-type monomer was considered to be asingle polypeptide chain. However, the presenceof a heterogeneous population of complemented,B-galactosidase molecules, some wild-type and somealtered, argues for an alternative explanation.Association of non-mutant polypeptide chains

would result in wild-type enzyme when four suchprotomers formed the tetramer, and a tetramercontaining even one protomer created by associa-tion of a point-mutant chain and wild-type chains,complemented to native configuration by the otherprotomers, would show altered characteristics.A genetic argument against a polycistronic z

gene, i.e. in favour of a polypeptide of molecularweight 135 000, was the polarity gradient extendingacross the z gene for transacetylase production(Newton, 1967). However, Martin (1966) found apolarity gradient across the C gene of the histidineoperon in Salmonella, a gene known to determinetwo polypeptide chains, and finer analysis revealsseveral peaks in the z gene polarity gradient (Zipser,Zabell, Rothman, Grodzicker & Noritski, 1970).Also, recent evidence for reinitiation sites in the zgene (Grodzicker & Zipser, 1968; Newton, 1969)argues that it is polycistronic. These sites maycorrespond to initiation sites of chains with theblocked N-terminal pyrrolidonecarboxylic acid(Erickson & Steers, 1969b).

I thank Dr D. Robard for preparing the computerprogram, Dr Harvey Pollard for help with the Yphantiscells and Dr P. Charlwood for helpful comments on themanuscript. I am also grateful for the encouragementand support of Dr E. Steers, jun., and Dr C. B. Anfinsen.

REFERENCES

Auer, H. & Doty, P. (1966). Biochemistry, Easton, 5,1716.Craven G. (1967). Cold Spring Harb. Symp. quant. Biol.

31, 186.Craven, G., Steers E., jun. & Anfinsen, C. B. (1965). J.

biol. Chem. 240, 2468.Edelhoch, H. & Steiner, R. F. (1966). Biopolymer8, 4,

999.Erickson, R. P. & Steers, E., jun. (1969a). Genetics,

Au8tin, 61, s16.Erickson, R. P. & Steers, E., jun. (1969b). Biochem.

biophy8. Res. Comm. 37, 736.Givol, D., Craven, G. R., Steers, E., jun. & Anfinsen, C. B.

(1966). Biochim. biophye. Acta, 113,120.Goldberg, M. E. (1969). J. molec. Biol. 46, 441.Goldberg, M. E. & Edelstein, S. J. (1969). J. molec. Biol.

46, 431.Green, N. M. (1963). Biochem. J. 89, 609.Grodzicker, T. & Zipser, D. (1968). J. molec. Biol. 38,

305.Kawahara, K. & Tanford, C. (1966). J. biol. Chem. 241,

3228.Marchesi, S., Steers, E., jun. & Shifrin, S. (1969). Biochim.

biophy8. Acta, 181, 20.Martin, R. (1966). J. molec. Biol. 21, 357.Newton, A. (1967). Cold Spring Harb. Symp. quant. Biol.

31, 181.Newton, A. (1969). J. motec. Biol. 41, 329.Steers, E., jun., Craven, G. & Anfinsen, C. (1965a). J. biol.

Chem. 240, 2478.

Vol. 120 MOLECULAR WEIGHT OF /3-GALACTOSIDASE IN GUANIDINE 261Steers, E., jun., Craven, G. & Anfinsen, C. (1965b). Proc.

natn. Acad. Sci. U.S.A. 54, 1174.Tanford, C., Kawahara, K. & Lapainje, S. (1967). J. Am.

chem. Soc. 89, 729.Ullmann, A., Goldberg, M. E., Perrin, D. & Monod, J.

(1968a). Biochemistry, Easton, 7, 261.Ullmann, A., Jacob, F. & Monod, J. (1967). J. molec. Biol.

24, 339.Ullmann, A., Jacob, F. & Monod, J. (1968b). J. molec.

Biol. 32, 1.

Ullmann, A., Perrin, D., Jacob, F. & Monod, J. (1965).J. molec. Biol. 12, 918.

Wallenfels, K., Sund, H. & Weber, K. (1963). Biochem. Z.338, 714.

Woods, E. F., Himmerfarb, S. & Harrington, W. J. (1963).J. biol. Chem. 238, 2374.

Yphantis, D. A. (1964). Biochemistry, Easton 3,297.

Zipser, D., Zabell, S., Rothman, J., Grodzicker, T. &Noritski, M. (1970). J. molec. Biol. 49, 251.