Protein Science (1996), 5:517-523. Cambridge University Press. Printed
`Copyright 0 1996 The Protein Society
`
`in the USA.
`
`Control of aggregation in protein refolding:
`The temperature-leap tactic
`
`YANSHENG XIE AND DONALD B. WETLAUFER
`Department of Chemistry and Biochemistry, University of Delaware, Newark, Delaware 19716
`(RECEIVED October 31, 1995; ACCEPTED December 13, 1995)
`
`Abstract
`The kinetics of renaturation of bovine carbonic anhydrase I1 (CAII) were studied from 4" to 36", at the relatively
`high [CAII] of 4 mg/mL. Following dilution to 1 M guanidinium chloride, aggregate formation is very rapid and
`reduces the formation of active enzyme. The CAII activity yield at 150 min, 20" (-6OVo), is greater than that at
`either 4" or 36". However, if refolding is conducted at 4", aggregation is reduced dramatically and 37% yield is
`obtained at 120 min. If the solution is then rapidly warmed to 36", the yield rises rapidly to 95% at 150 min. This
`is an example of the "temperature leap" tactic. These results can be understood on the basis of two slow-folding
`intermediates whose kinetics have been studied. Only the first of these forms aggregates. Kinetic simulations show
`that, at 4", the first intermediate is depleted after 120 min, and the second intermediate rapidly isomerizes to ac-
`tive enzyme on warming. A series of experiments was conducted where the initial (120 min) folding temperature
`was systematically varied, followed by a "leap" to 36" for 30 additional minutes. With initial incubations from
`4" to 12", the final yield is >90%, drops rapidly from 12" to 20°, and decreases more gradually to -45% at 36".
`The overall results qualitatively fit the simple idea of ordinary temperature-accelerated reactions in competition
`with hydrophobic aggregation, which is strongly suppressed in the cold. Qualifications are
`discussed for the
`temperature-leap approach to find application in refolding other proteins.
`Keywords: cold suppression of aggregation; protein folding; temperature leap
`
`Protein refolding, as in the renaturation of inclusion bodies, is
`
`a matter of both intrinsic and practical interest. It is commonly
`found that formation of inactive aggregates accompanies refold-
`ing, except at very low protein concentrations. Recovery of a
`protein from large volumes of very dilute solutions adds diffi-
`
`culty and expense to its production. General references may be
`found in recent publications (Fischer et al., 1993; Pain, 1994;
`Wetlaufer & Xie, 1995).
`We recently published studies of the much-studied carbonic
`anhydrase I1 (CAII) refolding system (Wetlaufer & Xie, 1999,
`in which we showed that high yields of native proteins are ob-
`tained at high [CAII], in spite of substantial aggregate forma-
`tion. We also showed that yield may be further increased by a
`variety of surfactant promoters. All these studies were carried
`out at 20".
`We here explore the temperature-dependence of refolding and
`aggregation during refolding at high [CAII]. The refolding of
`CAII at high dilution has been studied over a wide temperature
`range (Semisotnov et al., 1990). Those studies provided kinetic
`data and confirmed a reaction scheme that has been useful in
`
`Reprint requests to: D.B. Wetlaufer, Department of Chemistry and
`Biochemistry, University of Delaware, Newark, Delaware 19716; e-mail:
`31806@brahms.udel.edu.
`
`interpreting our results. An elaboration of this reaction scheme
`to include concomitant formation of aggregates is due to Cleland
`et al. (1992), who showed that aggregation is due only to the first
`of two folding intermediates.
`Our results show the suppression of aggregation as well as the
`rate of refolding at low temperatures. They also show that, af-
`ter a time interval sufficient to deplete the concentration of the
`first folding intermediate, a rise in temperature results in rapid
`conversion of the second intermediate to native protein. We call
`this the "temperature leap" tactic, which is not to be confused
`with "temperature jump."
`Suppression of aggregation during refolding at low temper-
`atures has been noted occasionally (Mitraki et al. 1987; Brems,
`1989; van der Vies et al., 1992). However, it does not appear to
`have been systematically studied and exploited as in the present
`work.
`
`Results
`As a preliminary survey of the effect of temperature on CAII
`refolding, the experiments of Figure 1 were carried out. Using
`the relatively low concentration of 0.50 mg CAII/mL, the time
`course of refolding was determined at six temperatures ranging
`from 4" to 44". The initial rates of refolding are seen to increase
`
`517
`
`APOTEX EX1029
`
`Page 1
`
`

`
`518
`
`Y. Xie and D.B. Wetlaufer
`
`monotonically from the lowest temperature to 28", remain the
`same at 36", and decrease slightly at 44". The yields at 30 min
`follow a somewhat similar pattern, increasing to a maximum
`at 28", decreasing slightly at 36", and decreasing substantially
`at 44". Indeed, the 44" regeneration has nearly reached its max-
`a plateau at
`imum at 5 min, with the later points describing
`-55% activity.
`The results of Figure 1 show that for refolding at [CAII] =
`0.50 mg/mL, the optimum temperature is -30". However, one
`of our main objectives is to uncover conditions that favor a high
`percent renaturation at high protein concentrations. Our earlier
`work has shown that refolding results at low [CAII] are poor
`predictors of folding at high [CAII]. For both these reasons, it
`seemed advisable to carry out a temperature survey at a much
`in Figure 2 ,
`higher [CAII], 4.0 mg/mL. The results, shown
`present a somewhat different pattern from those of Figure 1. At
`these higher concentrations of CAII, the yield curve increases
`from 4" to 20", but decreases from 20" to 36". Comparing the
`recoveries in this figure with those at the same temperatures in
`Figure 1, we see that those of Figure 2 are all depressed, and that
`the 36" recovery is most strongly depressed. In the 4" refolding,
`there continues to be an appreciable rate of renaturation at
`150 min. For this reason, we continued one experiment overnight
`and found >90% activity at 20 h. We have not explored recov-
`eries for time intervals > 150 min at other temperatures.
`In addition to measuring the time course of activity at high
`[CAII], we determined the progress curves of the turbidity of so-
`
`lutions refolding under identical conditions. The results (Fig. 3)
`show a maximum in turbidity forming within the dead time from
`manual mixing to measurement (a few seconds). The initial tur-
`bidity value increases more than twofold from 4" to 20", but is
`virtually unchanged from 20" to 36". Also noteworthy is the
`change in trend as the temperature is increased from 20" to 36".
`At the two lower temperatures, turbidity decreases after the ini-
`tial dilution and mixing, whereas at 36", the turbidity increases
`with time. This difference suggests a significant change in the
`system between 20" and 36". A strong turbidity (by visual ob-
`servation) persists for >20 h in samples incubated at 20", as in
`
`.,
`
`TIME, MIN.
`
`4"; 0, 12"; 0,
`Fig. 1. Refolding of CAI1 at various temperatures:
`20"; 0 , 28"; A, 36"; V, 44". All renaturations were carried out at
`0.50 mg C A I I h L , 1.00 M GdmCl.
`
`7 5 1
`
`> 00
`
`0
`
`0 20 "C
`
`..____.. ..A 36 'c
`
`4 OC
`
`0
`
`50
`
`l o 0
`TIME, MIN.
`
`I50
`
`2
`
`.,
`
`Fig. 2. Refolding at high CAII concentrations: 4.0
`1 .OO M GdmCI, at the indicated temperatures:
`
`mg C A I I h L ,
`4"; 0, 20"; A, 36".
`
`Figure 2. It therefore appears that aggregate dissociation at 20"
`is composed of two processes, one with t l I z -40 min, and the
`other essentially irreversible on our time scale.
`Another way to explore the nature of the 36" refolding is by
`examining the effects of surfactant additives that enhance re-
`folding at 20" (Wetlaufer & Xie, 1995). Figure 4 shows the ac-
`tivity recovery at 36", high [CAII], with and without n-hexanol
`or CHAPS. The concentrations of these surfactants employed
`enhance the 150-min yield of active CAII by -25%
`in regener-
`ations that were identical except that temperature was 20". It is
`immediately evident in Figure 4 that these two surfactants d o
`not influence the time course of renaturation at 36".
`Reflection on the data of Figures 2 and 3 led to the follow-
`ing experiment. Refolding was initially carried out at 4", where
`the turbidity (aggregate) formation is much lower than at room
`
`0 l
`0
`
`50
`
`100
`TIME, MIN.
`
`I
`150
`
`Fig. 3. Turbidity formation during CAI1 refolding at different temper-
`atures. All experiments were conducted at 4.0 mg C A I I h L , 1.00 M
`GdmCl. W , 4"; 0, 20"; A, 36".
`
`Page 2
`
`

`
`Control of aggregation in protein folding
`
`I
`
`* 75
`
`-M-
`control
`0 CHAPS
`-0"- hexanol
`
`.. ...
`
`0
`
`50
`
`100
`TIME, MIN.
`
`I50
`
`200
`
`(4.0 mg/mL) at 36 "C in 1.00 M
`Fig. 4. Refolding kinetics of CAI1
`GdmCI. 0, control; with surfactant additives: 0, CHAPS (6 X lO-'M);
`M).
`and 0, hexanol
`
`temperature. After a substantial time interval, the refolding so-
`lution was rapidly heated to 36". The time courses of both the
`turbidity and activity were determined (Fig. 5 ) up to 120 min.
`The time course of activity is essentially the same as shown in
`Figure 2. In the time interval 120-150 min the activity increases
`rapidly to approach 100% recovery. We call this a temperature
`leap refolding, to reflect both the rapid temperature change and
`the dramatic increase in yield that follows. The time course in
`turbidity for the first 120 min is very similar to that shown in
`Figure 3; a slight increase follows the temperature rise.
`To further explore the temperature leap approach, we carried
`out a series of such experiments. The 120-min initial folding was
`
`519
`
`carried out at temperatures ranging from 4" to 36", and was in
`every case followed by a 30-min incubation at 36". Limitation
`of time and materials led us to assay activity only at 150 min (ac-
`tually duplicate assays at 150 and 153 min). The results, shown
`in Figure 6, display a sigmoidal dependence of 150 min activity
`on temperature. Yields >90% were obtained from 4" to 12",
`then rapidly decrease to 60% at 20", and decrease more slowly
`from 20" to 36".
`The experiment of Figure 7 was conducted to test whether pro-
`tein aggregates formed at room temperature redissolve at low
`temperature, and can be re-activated by the temperature-leap
`tactic. CAI1 refolding was initiated by dilution at 20°, allowed
`to proceed for 50 min, and then chilled to 4" for an additional
`70 min. At 120 min after folding initiation, the temperature was
`raised to 36". Frequent measurements of both activity and tur-
`bidity were made over the time course of the experiment. Fig-
`ure 7 shows a spontaneous turbidity decrease during the first
`50 min, with no significant change in response to the tempera-
`ture drop, and only a marginal increase following the tempera-
`ture rise at 120 min. No appreciable change in activity is seen
`associated with either the temperature drop or the temperature
`rise.
`
`Discussion
`Ideally, before beginning temperature-dependent kinetic stud-
`ies, we should have in hand the family of equilibrium curves
`showing the extent to which denaturation depends on [guani-
`4-44"
`dinium chloride] [GdmCI] at temperatures across the
`temperature range we have here explored. We have indeed con-
`ducted equilibrium denaturation studies (via AAbsorbance at
`291 nm) at 20", 28", and 36" (Xie, 1996), and find our 28" re-
`sults to be in close agreement with those (at 25") of Wong and
`Tanford (1973). The curves move progressively to lower [GdmCl]
`as the temperature is increased, but the extent of denaturation
`is always <5% at 1.00 M GdmC1. The rate of equilibration is
`so slow at 20" that we did not attempt to determine denaturation
`
` a
`
`3
`
`150
`
`
`
`0.00
`
`...... 0 ......
`
`Activity
`
`urbidity
`
`15
`
`0
`0
`
`I
`30
`
`4 oc
`
`60
`
`9
`
`0
`
`1
`
`TIME, MIN.
`
`Fig. 5. Temperature-leap refolding of CAI1 at high protein concentra-
`tion (4.0 mg CAWmL), 1.00 M GdmCI. The first 120 min of the refold-
`ing was carried out at 4"; the sample was then transferred to a
`36"
`thermostat. Activity assays and turbidity measurements were made on
`separate portions of the same sample.
`
`" I 0
`
`30
`20
`10
`TEMPERATURE, 'C
`
`Fig. 6 . Activity recovery at 150 min in a series of temperature-leap ex-
`periments (as in Fig. 5). where the first 120-min refolding is conducted
`at the temperature plotted; the sample is then transferred to a 36" ther-
`mostat and the activity assayed at 150 min.
`
`Page 3
`
`

`
`7''zo
`
`a-#4 1.10
`
`0
`0
`
`30
`
`60
`
`I
`90
`
`TIME. MIN.
`
`* * * r
`I so
`
`0
`0.70
`
`
`
`Fig. 7. Effect of temperature drop followed by temperature leap on
`CAII refolding. Temperatures are indicated
`in each panel. [CAII] =
`3.2 mg/mL, [GdmCI] = 1.00 M. This experiment was carried out at
`3.2 mg/mL CAII rather than 4.0 mg/mL (as in Figures 2-6 inclusive).
`
`curves at lower temperatures. We believe it a reasonable assump-
`of denaturation at 1.00 M
`tion that the equilibrium extent
`GdmCl will be less than 5% from 20" down to 4".
`Our results can usefully be discussed in terms of the reaction
`scheme shown in Figure 8, based on the work of Cleland et al.
`(1992) and several other laboratories (Wong & Tanford, 1973;
`Ikai et al., 1978; Stein & Henkens, 1978; Dolgikh et al., 1984;
`Semisotnov et al., 1990; Wetlaufer & Xie, 1995.) Figure I shows
`that the rate of active enzyme formation increases with temper-
`ature up to -30". At 36" and 44", the existence of a competing
`
`Aggregate
`
`e t c g
`
`Fig. 8. Scheme for refolding of CAII in GdmCI, modified from Cleland
`et al. (1992). based extensively on earlier investigations. When the un-
`folded protein (U) in 5 M GdmCl is rapidly diluted to GdmCl I 1 M,
`the first intermediate, If, is formed rapidly (tl/2 = 0.03 s). At high di-
`lution of CAII, I' isomerizes to I" ( f I l 2 = 120 s), which subsequently
`rearranges ( t l I 2 = 550 s) to form N , the native protein. At higher pro-
`tein concentrations, self-association of I' yields dimer I; and trimer I ; ,
`both of which may continue to associate to form micron-size aggregates.
`Partial reversibility of the aggregate-forming reactions (Wetlaufer & Xie,
`1995; also evident in the present work)
`is indicated by an en-
`semble of kinetic coefficients. Once some aggregate is formed, equili-
`I' and aggregate. The
`bration may also take place directly between
`kinetic parameters cited were obtained for refolding in 0.60 M GdmCl
`at 23" (Semisotnov et al., 1990).
`
`Y. Xie and D.B. Wetlaufer
`
`nonproductive process is evident. We have earlier shown (Wet-
`laufer & Xie, 1995) that the rate of CAII reactivation at 20" de-
`creases significantly as [CAII] is increased above 0.10 mg/mL.
`It therefore seems likely that association of folding intermedi-
`ate I' (forming I;, I ; , etc.) takes place in most or all of the ex-
`results in
`periments of Figure 1, and that such association
`decreasing the yields at higher temperatures. For a reaction
`
`scheme of the complexity of Figure 8, it is evident that the data
`of Figure 1 alone cannot furnish meaningful rate parameters.
`On the basis of the above summary, it might seem counter-
`intuitive to carry out another temperature survey at substantially
`higher protein concentration. However, other extrapolations in
`a refolding system at least as complex as Figure 8 had proved
`unreliable (Wetlaufer & Xie, 1995). With the motivation of de-
`veloping approaches that permit efficient refolding at high pro-
`tein concentration, a second survey of refolding temperatures
`was conducted at an eightfold higher protein concentration
`(Fig. 2). A 150-min time interval was chosen both to be practical
`both for large-scale preparative operations and for laboratory
`investigation. The results at 20" are in good agreement with what
`we have earlier reported, with -60% reactivation at 150 min,
`much higher than one would expect from Cleland and Wang's
`(1990) diagram of refolding regimes. It is not surprising that the
`4" yield curve in Figure 2 is substantially lower than at 20".
`The two curves make a qualitatively similar pattern to that of
`the same temperatures
`in Figure 1. However, the depression
`of the 36" curve below the 20" curve was somewhat unexpected,
`although it may be argued that this is qualitatively what one
`would expect from a hydrophobic mechanism dominating the
`oligomer and aggregate-forming reactions (Kauzmann,
`1959;
`Smith & Lauffer, 1967; Tanford, 1980).
`It should be noted that the turbidity measured in Figures 3
`and 5 is most probably due to micron-size aggregate particles,
`I ; , I ; etc. (Cleland &
`with very minor contributions from
`Wang, 1990). Although turbidity as we measured does not di-
`rectly measure aggregate mass, it is reasonable to suggest that
`turbidity decrease does reflect dissolution of some of the aggre-
`gates. Viewed in this perspective, the kinetics of turbidity change
`at 20" and at 4" suggest more than one rate process of aggre-
`gate dissolution. We recognize this in Figure 8, where we iden-
`tify the kJ,,+,,, as a collection of aggregate dissolution rate
`constants. The remainder of what we know about aggregate dis-
`solution is: (1) the process is spontaneous at temperatures 520";
`(2) at some temperature between 20" and 36", aggregate growth
`dominates dissolution; and (3) dissolution of aggregate is un-
`affected either by surfactant promoters (Wetlaufer & Xie, 1999,
`or by lowering the temperature.
`If a hydrophobic mechanism is primarily responsible for a
`lower 150-min yield at 36", it might be possible to interfere with
`that mechanism by employing surfactants. We have earlier
`found (Wetlaufer & Xie, 1995) that 8 of 18 surfactants provided
`some increase in 150-min yield at 20", under the same refold-
`ing conditions as here used for Figures 2 and 3. The experiments
`of Figure 4 were designed to test the effect of surfactants on
`folding at 36" at high [CAII]. Concentrations of the two sur-
`factants were chosen that gave maximal increase in yield at 20".
`It is clear that neither hexanol nor CHAPS alters the refolding
`progress curve at 36". Although these results are not inconsis-
`tent with a hydrophobic basis for aggregate formation, they do
`indicate a significant difference between the reaction paths at
`20" and 36".
`
`Page 4
`
`

`
`Control of aggregation in protein foiding
`
`521
`
`Note that the amount of unrecovered activity at 20" is -40%
`In reflecting on the data of Figures 2 and 3, we see that re-
`(Fig. 2), whereas at 4", it is -5% (from Fig. 5, using the 150-min
`folding at 4", although slower in producing native enzyme, has
`yield). If we let these values represent the amount of aggregate
`a greatly reduced amount of aggregate as measured by turbid-
`at the two temperatures, the 4" simulation might be expected
`ity. We infer that the equilibria forming oligomers and aggre-
`results of
`to be a better approximation to the experimental
`gate are substantially shifted toward the monomer, 1', at low
`Figure 2 than the 20" simulation. Using activation energies of
`temperatures. The work of Semisotnov et al. (1987, 1990) has
`18 kcal/mol for both kl,,Iv and klm,N, we find that these rate
`provided values for ku,l,, kl,,I-, and kl"+, as well as deter-
`coefficients increase 6x from 4" to 20", and further increase
`mined the activation energies for the latter two coefficients.
`5x for the temperature rise from 20" to 36". Inspection of Fig-
`From this information, we generated the kinetics simulations
`ure 9B shows that, after 120 min at 4", most of the protein will
`shown in Figure 9 (Xie, 1996). Because the U + I' reaction is
`be divided between I" and N , with I' depleted. With aggregates
`very fast on the timescale of Figure 9, intermediate I' initially
`forming only from I' and its oligomers (Cleland et al, 1992; Wet-
`represents all the material. Although the values of the kinetic
`laufer & Xie, 1995), a rise in temperature to 36" would be ex-
`coefficients were obtained at 0.60 M GdmC1 instead of the
`pected to effect a rapid conversion of I" to N , with little change
`1 .OO M GdmCl used in our experiments, these coefficients ap-
`in the amount of aggregate. This actually happens, as seen in
`pear in the Semisotnov studies to be independent of denaturant
`Figure 5 . Thus, the premises of the simulation appear to be es-
`k,,,, has not been
`concentration. The activation energy for
`sentially correct. However, we note a qualitative difference be-
`determined, but any reasonable
`value would not change the
`tween the time course of activity at 4", which rises steeply to 15%
`appearance of Figure 9B, because kU,/. is 2-3 orders of mag-
`in the early minutes (Fig. 2), in contrast with the simulation
`nitude greater than kl.,I,, and kl.,,.
`(Fig. 9B), which shows N rising only gradually after an initial
`We therefore consider it valid to use the Semisotnov et al.
`lag period. We suspect that some of the rapid rise seen at early
`parameters to approximate the kinetics of the several species at
`times at 4" is due to refolding during the sampling and assay time
`various temperatures in our refolding system. The most impor-
`interval. This overestimates the true activity of a rapidly regen-
`tant exception is that the Semisotnov et al. data do not account
`erating solution (see Methods).
`for oligomerization and aggregate formation, additional pro-
`We have thus far ignored the kinetics and equilibria describ-
`cesses that are clearly evident at our high protein concentrations.
`ing formation and dissociation of oligomers and aggregates. It
`is worth noting that, under our conditions, these reactions (ex-
`cept k,,,-aRR) are substantially faster than I"-+N. It also appears
`that the aggregate formation is somewhat more temperature-
`dependent than I" -+ N . For the present treatment, it appears
`acceptable not to attempt to incorporate detailed estimates of
`the nonproductive component of Figure 8.
`Because Figure 6 represents the results of a competition be-
`tween productive reactions (+ Native) and aggregation, and
`because aggregation is [CAI[]-dependent, the details of the ex-
`perimental curve should depend on [CAII]. Specifically, at higher
`[CAII], we would expect the steep (12-20") segment of the curve
`to move to lower temperatures, and the shallow (20-36") seg-
`ment to reflect lower activity recovery.
`Figure 6 shows "final" activity recovered after a temperature
`leap from a range of initial 120-min incubations. This pattern
`is strongly reminiscent of that shown by the reversible temper-
`ature dependence of TMV protein aggregation to form rodlike
`polymers (Smith & Lauffer, 1967). In this well-studied system,
`a strong hydrophobic component of the association energetics
`has been inferred (Lauffer, 1975). It is well-known that hydro-
`phobic interactions are weakened by lowering the temperature
`below ambient (Kauzmann, 1959; Tanford, 1980). It is less well-
`known that GdmCl is less effective in weakening hydrophobic
`interactions in the cold than at room temperature (Wetlaufer
`
`et al., 1964). In the case of the experiments summarized in Fig-
`ure 6, it appears that the former of these two effects dominates.
`
`There is an interesting similarity between the early reactions
`of Figure 8 and a recent view of the folding kinetics of small pro-
`teins (Sosnick et al., 1994). These authors summarize evidence
`for the view that the rate-limiting step in cytochrome c renatur-
`ation is the correction of a very early misfolding, involving the
`interaction of non-native His residues with heme. This presents
`a formal kinetic parallel to Figure 8, where substantial reduc-
`tions of rate and yield arise from oligomerization and aggrega-
`tion of 1'. The work of Semisotnov et al. (1987, 1990) strongly
`
`I'
`
`I "
`N
`
`I so
`
`0
`
`30
`
`60
`
`00
`
`I20
`
`TIME, MIN
`
`Fig. 9. Modeling the kinetics of CAlI refolding at high dilution as three
`consecutive first-order reactions. A: At 23", computed from the rate
`coefficients ku,,,, kl.,l-, and klVdN obtained by Semisotnov et al.
`(1990). B: At 4", computed via the Arrhenius equation using activation
`energies of 18 kcal/mol (Semisotnov et al., 1990) for both k,,-/" and
`kl--.N. The rate coefficient kU,/. was considered to be temperature-
`independent in this simulation, because kU+/, > kl,,/..
`
`.-." """.
`
`60
`90
`TIME, MIN.
`
`I20
`
`I SO
`
`B, 4 "C
`
`0.00 '
`
`0
`
`30
`
`I M I
`
`0.75
`
`Page 5
`
`

`
`5 22
`
`Y. Xie and D. B. Wetlau fer
`
`implicates prolyl peptide isomerization as rate-limiting for both
`the I'+ I" and I" + N reactions. The presence of certain non-
`native prolyl isomers is therefore a key co-determinant of the
`aggregation-prone structure of 1'. It appears that non-native pro-
`lyl residues may commonly be associated with aggregation-prone
`folding intermediates.
`How applicable will the temperature leap tactic be to refold-
`ing other proteins? If the refolding of a protein involves an early
`intermediate prone to aggregation, and at least one later non-
`
`aggregating intermediate, the temperature leap tactic is likely to
`be useful. This requires that k,.,I- > k,,,4N, permitting accu-
`mulation of nonaggregating I" at low temperature. It appears
`that the aggregation-prone intermediate is generally identifiable
`with the so-called molten globule state (Ptitsyn & Uversky,
`1994), or a still-earlier intermediate (Uversky & Ptitsyn, 1994).
`The molten globule state is characterized as a somewhat larger-
`than-native particle with greater-than-native solvent exposure
`of nonpolar side chains. Association of such species by hydro-
`phobic interactions would not be surprising. A molten globule
`intermediate has been shown for several proteins (Fink et al.,
`1991; Ptitsyn & Uversky, 1994); it appears to be quite wide-
`spread. It seems likely that some of these proteins will have a
`later intermediate as described above, and that the temperature
`leap tactic will find additional favorable cases for application.
`It should be clear, however, that detailed knowledge of the
`refolding kinetics is not a necessary prerequisite to explore the
`usefulness of this tactic. The oxidative refolding
`of reduced
`disulfide proteins might possibly benefit from the temperature-
`leap approach. However, additional issues, such as the substan-
`tial temperature-dependence of thiol pK, values, should be
`considered.
`Neither this work nor the previous studies with surfactant pro-
`moters (Wetlaufer & Xie, 1995) were intended to be optimiza-
`tions. If one were to carry out an optimization, a number of
`adjustable parameters could be explored systematically. These
`would include pH, protein concentration, denaturant concen-
`tration, temperature and time of the cold incubation, tempera-
`ture and time of the post-leap incubation, and the
`possible
`addition of surfactant promoters.
`
`Materials and methods
`
`Materials
`Bovine carbonic anhydrase I1 (MW = 29,000, PI = 5.9), puri-
`fied by electrophoresis, was purchased from Sigma Chemical
`Co. (St. Louis, Missouri). The isozyme distribution was checked
`by IEF in our laboratory. CHAPS, p-nitrophenyl acetate, and
`Tris (hydroxymethyl) amino methane (Trizma base, molecular
`biology grade) were also obtained from Sigma Chemical Co.
`Ultrapure guanidinium chloride was the product of Schwartzl
`Mann Biotech (Cleveland, Ohio). N-hexanol was obtained from
`Aldrich Chemical Co. (Milwaukee, Wisconsin). Acetonitrile
`(HPLC grade) was a product of Fisher Scientific (Pittsburgh,
`Pennsylvania). All solutions were prepared with MilliQ water,
`and buffers were filtered through a 0.45-p membrane before use.
`
`Methods
`Protein concentrations were measured by 280 nm absorbance
`as detailed earlier (Wetlaufer & Xie, 1995). Guanidinium chlo-
`
`ride concentrations were determined refractometrically (Nozaki,
`1972).
`Refolding was carried out by rapid manual dilution of dena-
`tured CAII in 5.00 M GdmCl to 1.00 M GdmCl and the de-
`sired protein concentration, with a dilution buffer composed of
`50 mM tris sulfate, pH 7.5. Buffer pH measurements were made
`at *2", and not corrected for change in pH with temperature.
`All activity assays were conducted at 20.0 -t 0.3". The activity
`assay involved catalysis of p-nitrophenyl acetate hydrolysis at
`20", with corrections as detailed by Wetlaufer and Xie (1995).
`The assays in this paper were conducted without EDTA as a
`quench to suppress refolding during the assay. For rapidly re-
`generating solutions, and perhaps for the 4" regenerations, this
`leads to somewhat higher estimates of
`activity than what
`is
`found with EDTA in the assay solutions. However, when activity
`is changing very slowly with time, the same level of CAII activ-
`ity is found with and without EDTA (Wetlaufer & Xie, 1995). For
`the refolding experiments, all containers were pre-equilibrated
`at the desired temperature before mixing the diluting buffer with
`CAII in 5 M GdmC1.
`re-
`The turbidity of refolding solutions was estimated and
`ported as apparent absorbance at 330 nm in cells of 10.0-mm
`pathlength.
`
`Acknowledgments
`
`We thank Prof. H.F. Gilbert for critical discussions. This work was sup-
`ported by a grant-in-aid from Genentech, Inc., and by the College of
`Arts and Science of the University of Delaware.
`
`References
`
`Brems DN. 1988. Solubility of different folding conformers of bovine growth
`hormone. Biochemistry 27:4541-4546.
`Cleland JL, Hedgepeth C, Wang DIC. 1992. Polyethylene glycol enhances
`refolding of bovine carbonic anhydrase B. Reaction stoichiometry and
`refolding model. J Biol Chem 267:13327-13334.
`Cleland JL, Wang DIC. 1990. Refolding and aggregation of bovine carbonic
`anhydrase B: Quasi-elastic light scattering analysis. Biochemistry 29:
`11072-1 1078.
`Fink AL, Calciano LJ, Goto Y, Palleros D. 1991, Conformation states in
`acid-denatured proteins. In: Nall BT, Dill KA, eds. Conformations and
`forces in protein folding. Washington, DC: AAAS.
`Fischer B, Sumner I, Goodenough P. 1993. Isolation, renaturation, and for-
`mation of disulfide bonds of eukaryotic proteins expressed in Escherichia
`coli as inclusion bodies. Biotechnol Bioeng 41:3-13.
`Kauzmann W. 1959. Some factors in the interpretation of protein denatur-
`ation. Adv Prot Chem 14:l-57.
`Lauffer MA. 1975. Entropy-driven processes in biology: Polymerization of
`fobacco mosaic virus protein and similar reactions. New York: Springer-
`Verlag.
`Mitraki A, Betton JM, Desmadril M, Yon JM. 1987. Quasi-irreversibility
`in the unfolding-refolding transition of phosphoglycerate kinase induced
`by guanidine hydrochloride. Eur J Biochem 163:29-34.
`Nozaki Y. 1972. The preparation of guanidine hydrochloride. Methods En-
`zymol26:43-50.
`Pain RH. 1994. Mechanismsof protein folding. Oxford: IRL Press. Oxford
`University Press.
`Ptitsyn OB. 1995. How the molten globule