Characterization of two cold-sensitive mutants of the beta-galactosidase from Lactobacillus delbruckii subsp. bulgaricus.

Methoxylamine mutagenesis of the beta-galactosidase gene from Lactobacillus delbrückii subsp. bulgaricus was used to generate cold-sensitive variants. Two variants, P429S and L317F, were characterized kinetically in order to determine the enzymatic consequences of these mutations. The kinetic parameters Km and Vmax on the synthetic substrate o-nitrophenyl-beta-D-galactopyranoside have been determined over a temperature range of 11-45 degrees C. Only the Vmax of the two variants was significantly different than the wild-type enzyme over the temperature range studied. The Vmax of the L317F variant is reduced proportionately at all temperatures compared to the wild-type enzyme while the value of Vmax for the P429S mutant deviates from wild-type only at lower temperatures (in 2 mM Mg2+). This temperature-dependent effect on the Vmax of P429S can be suppressed by increasing the Mg2+ concentration. The results suggest that the binding of this essential metal ion is altered in the P429S variant such that its dissociation is increased by lowering the temperature.

8 To whom correspondence should be addressed.

MATERIALS AND METHODS
Plasmid Constructions-The wild-type P-galactosidase gene from L . delbriickii subsp. bulgaricus B131 (Centre International de Recherche Daniel Carasso, BSN group, Le Plessis-Robinson, France) was cloned previously in our laboratory (Schmidt et al., 1989). An XbaI restriction site was introduced at nucleotide 441 by site-directed mutagenesis (Zoller and Smith, 1982). Then, the 3.4-kilobase pairXba1-BamHI fragment carrying the promoterless P-galactosidase gene was ligated with a 418base pair fragment containing the alkaline phosphatase (phoA) promoter (Oka et al., 1985) into the vector pBR322 (Bolivar et al., 1977) that had been previously digested with EcoRI and BamHI. The resultant plasmid was used for mutagenesis of the gene and expression of the 6-galactosidase variants.
Chemical Mutagenesis and Variant Selection-Up to 1 pg of plasmid DNA was added to 150 pl of methoxylamine hydrochloride (Aldrich), prepared according to Kadonaga and Knowles (1985), and incubated in the dark at 50 "C. Aliquots were removed every 30 min for up to 5 h, 50 pl of ethylene glycol and 20 pl of 3 M sodium acetate, pH 5.5, were added, and the DNA precipitated by ethanol. The mutagenized plasmids were resuspended in 10 pl of 10 nm Tris-HC1, pH 8.0, with 1 nm EDTA and used to transform E. coli JM109 cells (lac-;Yanisch-Peron et al., 1985). Bacteria were spread on a nitrocellulose filter (Schleicher & Schuell) lying on top of a Luna-Bertani agar (Difco Laboratories) plate containing 50 pg of carbenicillin (Sigma) and grown overnight at 37 "C. Two replica filters were made and the duplicate colonies on these filters were grown as described above. One replica filter was then placed at 4 "C and after 24 h transferred to a cold agar plate containing 40 pg/ml5-bromo-4-chloro-3-indolyl-~-~-galactopyranoside (Sigma). A duplicate filter was placed on an 5-bromo-4-chloro-3-indolyl-~-~-galactopyranoside agar plate at 37 "C. Desired mutants had near wild-type activity at 37 "C and reduced activity at 4 "C relative to wild-type as detected visually by the rate of blue color formation.
Mutation Site Determination-The mutations of two selected coldsensitive variants, Cs2l and Cs4, were mapped to the 1.2-kilobase pair Ne01 fragment by swapping wild-type and mutant gene fragments. The nucleotide base substitutions present in these two variants was determined by DNA sequencing (Sanger et al., 1977) of the entire NcoI fragment. The appropriate mutations from cs2 and Cs4 were introduced in the wild-type gene by oligonucleotide site-directed mutagenesis (Zoller and Smith, 1982).
Enzyme Purification-Wild-type and variant P-galactosidases were expressed from the corresponding genes in E. coli JM109 cells and purified from cytoplasmic extracts as detailed earlier (Schmidt et al., 1989). The enzymes were judged to be homogeneous as only one band was visible after SDS-PAGE analysis. All of our studies were done with these purified enzymes.
meric subunits of the E. coli lacZ P-galactosidase, and the subunits of Molecular Mass Determinations-The molecular mass of the monothe L. delbriickii subsp. bulgaricus enzyme and its variants, were determined by SDS-PAGE (Power et al., 1986). The apparent molecular mass of the native states and the urea dissociated monomers of each of these enzymes was determined by gel filtration liquid chromatography on two Superose-12 columns (Pharmacia LKB Biotechnology Inc.) connected in series. Initial Rate Determinations-A 10-pl aliquot of an appropriate dilution of sample was added to a 1-cm path length disposable cuvette containing 980 pl of reaction buffer (25 rn Bis-Tris, 200 rn sodium acetate (NaOAc), pH 7.0, 2 rn MgSO,) and 10 pl of 0.5 M NPG in dimethyl sulfoxide in accordance with previously described procedures (Estell et al., 1985). The appearance of o-nitrophenol was monitored at 410 nm (e = 9609 ~-l ) for 15-60 s at 25 "C using a Hewlett-Packard 8451 diode array spectrophotometer. Substrate depletion was kept to less than 5% of the initial substrate concentration during the course of the assay. A unit of enzyme activity was defined as the amount of enzyme required to give 1 absorbance unit change at 410 n d m i n (M,l,Jmin).
The total protein was determined by the BCA protein determination method (Smith et al., 1985), using bovine serum albumin as the standard. Specific activity was determined from the initial rate at 25 "C in the presence of 50 m~ NPG substrate and saturating magnesium concentration.
Progress curve determinations were performed as described previously (Estell et al., 1985), using a thermostated HP spectrophotometer. Appearance of o-nitrophenol was followed at 400 nm for low substrate concentrations ( E = 9609 M -~) and 460 nm for high substrate concentrations (e = 4311 M -~) . NPG at 0.35-0.47 rn was used for the high substrate concentration, and 0.0660.073 rn for the low substrate concentration. Replicate assays were run over the temperature range of 11-45 "C.
Determination of Magnesium Binding Constants-Samples of purified wild-type and P429S mutant enzymes were dialyzed overnight at 4 "C uersus 150 volumes of 25 m~ Bis-Tris, 200 rn NaOAc, 10 m~ EDTA, pH 7.1, and then twice uersus 100 volumes of 25 rn Bis-Tris, 200 m M NaOAc, 1 rn EDTA, pH 7.1. Magnesium sulfate and EDTA were used to fix the magnesium concentrations in the individual reactions (Portzehl et al., 1964) between 0.03 p~ and 10 m~, pH 7.1. Two assumptions were made in the determination of free magnesium: 1) only the forms of the ligand with three and four negative charges will bind magnesium at pH 7.1 (Portzehl et al., 1964); and 2) the affinity of EDTA for magnesium does not change substantially over the temperature range used in these experiments. The assays were carried out in replicas of 4-5 in the presence of at least 5 different substrate concentrations below KD and 5 above. Assays were performed from 5 "C (necessitating the use of He gas to prevent condensation on the cuvette walls) to 45 "C. The enzyme was added to pre-equilibrated buffer, and allowed to further equilibrate at the appropriate temperature for 30 min. Substrate was added and the reaction quantitated as described above.
The initial rate determinations were averaged and fit to a two-state binding equation using the methods of Marquardt (1963). This equation assumes one rate (REM) in the presence of substrate (So) for the enzyme with metal bound, (EM): and one rate (RE) for the enzyme with no metal bound, ( E 1: with the measured rate (u) being derived as the sum of these two rates:

[MI + K#)I[~E&'(KEM + So)I}ET
where: kE = kc,, for the rate with no metal bound; KE = K, for the rate with no metal bound; kEM = kc,, for the rate with metal bound; KEM = K , for the rate with metal bound; M = magnesium concentration; KD = dissociation constant for magnesium binding; ET = total enzyme concentration.
Lactose Hydrolysis-The hydrolysis of lactose at 10 mg/ml in 10 m M sodium succinate, pH 6.7, in the presence and absence of 20 rn MgSO,, was measured using high performance liquid chromatography methods. Identical reactions were first preincubated at 9 and 33 "C; purified wild-type, Cs2, or Cs4 mutant enzyme was then added and the reaction allowed to proceed for 1-24 h. At appropriate time intervals, samples were taken and quenched with an equal volume of 20 rn H,SO,. The sugar profile of the quenched samples was stable over 4%96 h. The samples were chromatographed using an IBM LC/9533 liquid chromatograph equipped with a LC/9505 SE automated sampler and system 9000 computer. The samples were injected onto a 300 x 7.8-mm Bio-Rad Aminex HPX-87H column preceded by a Cation H' guard column. The column was run isocratically in 10 m~ H,SO, at 0.6 d m i n and 30 "C. Peaks were detected by monitoring refractive index. This system allowed the base-line resolution of allolactose, lactose, glucose, galactose, and succinate. Succinate serves as an internal standard in each experiment. Standard injections of lactose, glucose, galactose, and succinate were reproducible to within 5%.

RESULTS
Generation a n d Identification of Variants-Methoxylamine mutagenesis of the @-galactosidase gene from L. delbriickii subsp. bulgaricus resulted in the generation of a variety of low temperature-sensitive variants. The nucleotide sequence alterations discovered in two of these mutants, Cs2 and Cs4, were introduced in the wild-type gene by oligonucleotide site-directed mutagenesis to make the cold-sensitive variants L317F and P429S, respectively. The mutant Cs2 appeared to be more cold-sensitive on plate screens than the corresponding variant L317F due to a suppressor mutation that reduced the expression of the enzyme in E. coli (data not shown). All characterization of the variants reported here was performed on protein isolated from the site-directed mutants L317F and P429S.
Molecular Mass-The molecular mass of denatured L. delbriickii subsp. bulgaricus @-galactosidase subunits from both the wild-type a n d the mutants, L317F and P429S, appear on SDS-PAGE to be approximately 110,000 Da. This is in agreement with the estimated molecular mass inferred from the DNA sequence (Schmidt et al., 1989). However, their native molecular mass, as measured by gel filtration on Superose-12 columns, appear closer to 220,000 Da (Fig. L4). The elution profile does not change with the addition of 5 mM EDTA to the buffer. This molecular mass as well as the absence of heterogeneity in the N-terminal sequence (Schmidt et al., 1989) suggests that the enzyme exists as a dimer of two identical subunits. This contrasts with the tetrameric structure of the E. coli l a c 2 P-galactosidase. The possibility that the L. delbriickii subsp. bulgaricus enzyme is a monomer in its native form but chromatographs aberrantly during gel filtration was eliminated by further studies in the presence of 8 M urea. Work on the E. coli enzyme has shown that treatment with 8 M urea causes the tetramer to dissociate into four identical monomers. This has been shown with both native gels and sedimentation studies (Zipser, 1963). Gel filtration, in the presence of 8 M urea using the E. coli enzyme as an internal control, shows clearly that the urea-treated L. delbriickii subsp. bulgaricus enzyme runs identically to the E. coli monomer (Fig. lB), suggesting that the native form of the enzyme is indeed a dimer made up of identical subunits.
Specific Actiuity-The NPG hydrolysis specific activities of the wild-type, P429S, and L317F enzymes were determined to be 850, 760 (90% of wild-type), and 570 unitdmg (67% of wildtype) of protein, respectively. In comparison, the E. coli enzyme had a specific activity of 690 units/mg (81% of the L. delbriickii subsp. bulgaricus wild-type enzyme).
The specific activities of the enzymes did not change significantly during storage through the duration of this study. There was also essentially no differences in the specific activities measured from different preparations. Thus, it is unlikely that the lower specific activities of the mutants represents the presence of a constant fraction of inactive enzyme.
Determination  (K-I) can be generated from the data (Segal, 1975) for the wild-type and L317F enzymes (Fig. 2). As shown, the slope of this line (-E,/R) is essentially equivalent for both of these enzymes, however, VmaxlET for wild-type is -1.7 times greater than Vmax/E~ for the L317F enzyme at all temperatures investigated.

Fraction Number
In contrast to L317F, the values of VmaJE* and the slope of the Arrhenius plot for the P429S variant essentially matches that of the wild-type at temperatures greater than 20 "C, suggesting that the enzymes are similar at these elevated, permissive temperatures. However, the Arrhenius plot for P429S does not appear to be linear and curves downward a t lower temperatures (Fig. 3): Lactose Hydrolysis-Enzymatic hydrolysis of lactose was performed in order to correlate the synthetic substrate data (NPG) to the natural substrate lactose. The reaction profiles of the wild-type enzyme and the two mutants were compared at identical enzyme concentrations in the presence or absence of 20 mM MgS04 at 9,33, and 41 "C. The results were plotted as reaction time uersus the area of the eluted substrate, products, or succinate control peaks. A reaction profile of the hydrolysis of Characterization of ltoo Cold-sensitive P-Galactosidases  lactose by the wild-type enzyme is shown in Fig. 4 (42 "C, no Mg2+). The appearance of allolactose over time demonstrates that the L. delbriickii subsp. bulgaricus enzyme carries out transgalactosylation like the E. coli P-galactosidase (Fig. 4).
In the absence of magnesium, at 42 "C, the initial reaction rates are similar, although a slightly decreased rate is seen for the P429S variant (compare the initial slopes in Fig. 5). At 9 "C, the wild-type and L317F initial rates are essentially identical while very little lactose hydrolysis is seen for the mutant. This demonstrates that the mutant P429S is also cold-sensitive for the hydrolysis of its natural substrate, lactose, in the absence of magnesium. In the presence of 20 m~ M$+, the initial rates for all three enzymes are approximately the same and similar to the initial rates observed for wild-type and L317F in the absence of Mgz' (Fig. 5). For the P429S enzyme, Mg2' appears to suppress its low temperature lability suggesting that its cold sensitivity is a function of Mg2' binding to the enzyme.

Magnesium Binding Constants-
The discovery that M$+ could suppress the cold lability of P429S led to the determination of the magnesium ion binding constants for the wild-type and P429S enzymes as a function of temperature. The KO for wild-type P-galactosidase remains constant at 1.0 p~ from 11 to 37 "C, at which point it increases rapidly (above 55 "C the loss of activity becomes irreversible). For comparison, the KO for Mg2' of E. coli P-galactosidase has been reported to be 0.15 VM (Edwards et al., 1990b) and 2.0 p~ (Edwards et al., 1990a). Similar to the observed behavior of the E. coli enzyme, exhaustively chelated (dialysis uersus 10 mM EDTA) L. delbriickii subsp. bulgaricus /.?-galactosidase still retains a low but measurable 1 4 % activity (data not shown). This agrees with the k,,,IK,,, value at 25 "C for E. coli P-galactosidase in the absence of Mgz' which is 2% of its value in 1 mM Mgz' (Cupples et al., 1990).
In direct contrast to wild-type, the KO for Mg2+ of the P429S  Fig. 6. This clearly shows the contrast between the varying KD of the P429S and the constant KD of the wild-type enzyme. Both plots give good linear fits up until about 40 "C where there is an abrupt change so that KO now increases with temperature. This is most likely caused by the onset of the thermal denaturation of the protein. In addition, P429S appears more heat labile than wild-type, rapidly inactivating at temperatures above 48 "C at all Mg2+ concentrations tested (data not shown).

DISCUSSION
In this study, we have generated and characterized two apparent cold-sensitive variants of the P-galactosidase enzyme from L. delbriickii subsp. bulguricus. There are a variety of conditions which could result in cold sensitivity of an enzyme: 1) a cold-dependent change in the reaction mechanism; 2) low temperature denaturation of the tertiary or quaternary struc- ture (Brandts, 1964;Privalov et al., 1986); or 3) low temperature dissociation of a n essential cofactor. In addition, in our initial screens, variants with reduced protein expression could also appear as being cold-sensitive. From our results, one of the mutants studied, L317F, has a slightly reduced activity compared to wild-type on the synthetic substrate NPG at all temperatures tested. Thus, although L317F appears to be somewhat cold-sensitive on plate screens (data not shown), it must really be considered a n "activity" mutant and not truly "cold-sensitive" (cold-sensitive meaning having approximately wild-type activity at ambient temperature and reduced activity at lower temperatures). By these criteria, another mutant, P429S, is cold-sensitive on both the synthetic substrate NPG as well as lactose, its natural substrate. Our work demonstrates that this cold sensitivity correlates to a change in the affinity of the mutant enzyme for essential magnesium ions. In fact, the cold temperature effect can be totally suppressed by the addition of high concentrations of magnesium ions in vitro. The reversible nature of this change, and the absence of a temperature-sensitive K,,, for NPG, suggests that the P429S enzyme does not undergo a large structural alteration or denaturation at these temperatures. Assuming that the cold sensitivity of P429S is entirely due to the dissociation of M e , and that the activity of the enzyme without bound M e is negligible, then V, , as a function of temperature ( t ) can be derived from Equation 3 above: where KD(t) can be calculated from the linear fit of the van't Hoff plot shown in Fig. 6. The theoretical curve determined from Equation 4 fits the data generated at 2 mM Mg2+ quite well as shown in Fig. 3. Thus, the cold sensitivity of P429S can be explained by a reduction in the affinity of this variant for Mg2+ where the enzyme with no metal bound is essentially inactive. The wild-type enzyme, unlike the P429S variant, is not cold sensitive since its KO for M e has no strong temperature dependence in the range of interest (Fig. 6). It is interesting that the KD for M e of the wild-type E. coli P-galactosidase has been reported to decrease with increasing temperature similar to that of P429S (Edwards et al., 1990b). Thus, one might expect the wild-type E. coli enzyme to be cold sensitive as well. Of course, in this case, much lower Mg2+ concentrations would be required to observe the effect (the KO values for the E. coli and P429S P-galactosidases are 0.15 and 530 p~, respectively, at 25 "C). The zero slope of the van't Hoff plot for the wild-type P-galactosidase suggests that the binding of magnesium ions to the enzyme is driven by entropy. It seems likely, that upon the binding of Mg2+, a significant number of water molecules would be released from the solvated magnesium ions and from the Mg2+ binding pocket on the protein causing the entropy change necessary as a driving force. Binding of magnesium ions to the P429S variant requires added energy in addition to the entropy change as reflected in the positive slope of the van't Hoff plot. Thus, the P429S mutation might skew the configuration of the binding site reducing the affinity of this variant for Mg2+ (the KO for P429S is over 500 times larger than wild-type at 25 "C).
The P429S mutation is not in a highly conserved region but is in a short segment (<40 amino acids) that connects highly conserved Regions I11 and IV (Schmidt et al., 1989). It is possible that one or both of these regions is part of a Mg2+ binding site. In fact, Edwards et al. (1990a) have found that substitutions in Region IV a t one of the putative active site residues (E461) in the E. coli P-galactosidase can significantly decrease the affinity of the protein for Mg2+ (a 1500-fold increase in KD for the E461H variant a t 25 "C).
The mutagenesishelection procedure outlined here allows the identification of potentially different cold-sensitive phenotypes. Mutants P429S and L317F represent two of these cases. In both variants, dramatic effects result from what is apparently the small localized change introduced by a single amino acid substitution. Cold denaturation seems unlikely as neither aggregation nor precipitation of either variant at any temperature is observed, even a t very high ionic strengths. The decrease of activity at low temperatures is fully reversible and time independent. As with the E. coli enzyme, magnesium ions are not required for the quaternary structure of the protein at ambient temperatures as evidenced by the gel filtration studies run in 5 r n~ EDTA (data not shown). Also, importantly, the K, values for both the variants remain constant over the full temperature range examined. We conclude that L317F is crippled catalytically (specifically in its V, , ) while the cold sensitivity of P429S is due to an increase in the essentially inactive " M efree" enzyme concentration as the temperature is lowered.