Selection of Mutations and In Silico Analysis
To gain clinically as well as mechanistically relevant insight into the role of misfolding in CBS deficiency we first selected a series of naturally occurring CBS mutations for further laboratory studies. The set of nine most prevalent mutations having a frequency of at least 10 alleles in the CBS Mutation Database was selected; these included the p.A114V, p.R125Q, p.E144K, p.T191M, p.R266K, p.I278T, p.G307S, p.W409_G453del, and p.D444N. This set was expanded by additional 18 less frequent mutations known to be localized in different domains of the CBS protein [Meier et al., 2001
]. For the list of mutations see Table 1; location of the mutations in the CBS active core and in the modeled carboxyterminal domain is shown in . The final series reflects the molecular pathology of CBS deficiency as these 27 mutations are present on about 70% of all pathogenic CBS alleles known in humans to date.
Fig. 1 Location of studied mutations in three-dimensional CBS structure. A: Mutations located at the surface of the active core are shown in magenta, pyridoxal 5′-phosphate (PLP) is shown in yellow. B: Mutations buried within the active core are indicated (more ...)
Based on our previous work [Singh et al., 2007
], we hypothesized that CBS mutations of residues buried in protein globule may have more severe effect on folding than those exposed to the solvent. After selecting the mutants we performed an in silico analysis with calculation of the solvent-accessible surface area. Based on previously published criteria we considered mutant residues with relative and absolute solvent-accessible surface area higher than 9% and/or 40 A2
, respectively, as being solvent exposed while the rest of the mutations were considered being buried in the globule [Mirkovic et al., 2004
]. shows the predicted solvent accessibility and demonstrates that mutations belonging to these two classes are represented almost equally in the series of 27 variant proteins.
Total Amount of CBS Mutant Proteins Expressed in E. coli and Their Solubility
To test this assumption we first optimized a method for extracting water-soluble as well as aggregated and water-insoluble CBS from bacteria expressing wild -ype and four mutant enzymes (for details, see the Methods section). Boiling in 3% SDS was subsequently used for the entire series of 27 mutant proteins to determine the amount of CBS antigen in both the particulate (i.e., pellet after centrifugation) and nonparticulate (i.e., supernatant) fractions of sonicated bacterial extracts. The sum of CBS signal in both fractions is referred to as the total SDS-soluble CBS antigen.
Using the SDS extraction we observed that all 27 mutant proteins were present in detectable amounts not only in the supernatants but more importantly also in the particulate fractions of the bacterial extracts. The presence of CBS antigen in both fractions is shown in ; however, this publication gel does not allow inferring either on relative proportions of the mutant proteins in the supernatant and particulate fractions or on their relation to the wild-type enzyme (for details, see the Methods section and legend to ). The signal of total SDS-soluble CBS antigen of the mutant proteins was generally somehow decreased with a median of 66% of the wild-type CBS (see Supp. Table S2
; gels used for quantification are not shown), although some variant proteins were present in increased quantities. These data suggest that compared to the wild-type enzyme the majority of mutant CBS proteins are less stable in E. coli
; however, we cannot exclude that some of them are also less soluble in 3% SDS.
Figure 2 Amounts of mutant CBS antigen in fractions of bacterial extracts. A: Mutant proteins expressed at 37°C; B: Results of expression of mutant proteins at 18°C. Top part of panels. The amount of SDS-soluble CBS antigen in particulate (Pel, (more ...)
Proportion of the SDS-soluble CBS antigen, which is present in the particulate fraction, indicates the propensity of the mutant protein to form high molecular weight aggregates and inclusion bodies. The particulate fraction of the wild-type CBS lysates contained only 12% of antigen of the total SDS-soluble signal, while more of the signal in the pellet was observed for the mutant proteins (median value was 18% of the total SDS-soluble antigen, with a range of 5 to 50%; see Supp. Table S2
). These data lend support to our hypothesis that the mutant proteins in this study were in general more prone to formation of large molecular weight aggregates and/or inclusion bodies. This part of our study further indicates that the previously reported analyses of mutants CBS in the supernatants of centrifuged bacterial extracts [Kozich et al., 1993
] may have underestimated the steady-state amounts of mutant proteins present in bacteria, and that extraction by 3% SDS is needed to recover the insoluble aggregated molecular species.
Impaired Quaternary Structure of Mutant Proteins as a Measure of Their Misfolding
The published data strongly support the hypothesis that CBS deficiency may be a conformational disease caused by misfolding of mutant enzymes. Detailed studies of folding, however, require protein purification, which would be especially difficult to accomplish for the most severely affected CBS enzymes. Because misfolded polypeptide chains cannot be correctly assembled and tend to form aggregates, we examined the quaternary structure of the mutant proteins in the nonparticulate fraction of crude bacterial extracts as a surrogate marker of their folding status. To assess the oligomeric structure of the mutant CBS enzymes we used electrophoresis in gradient polyacrylamide gels under nondenaturing conditions followed by Western blotting.
Visual inspection of blots showed variable amounts of fuzzy and putatively misassembled CBS, and largely differing amounts of clearly demarcated CBS tetramers or higher order oligomers as shown in . Subsequent quantitative analysis of the blots revealed that 88% of the total water-soluble antigen of the wild-type CBS is present as tetramers/oligomers. In contrast, these correctly assembled fractions were substantially decreased in our set of mutant proteins (median tetrameric/oligomeric signal was only 31% of the total water-soluble CBS antigen, range 1–92%; data are given in Supp. Table S2
). Because the CBS antigen in the nonparticulate fraction of mutant proteins was also decreased to a median 60% compared to the wild-type, the resulting net yield of tetramers/oligomers of the mutant proteins was severely reduced (median 12% of tetramers compared to the wild-type CBS). There was no clear correlation between the location of mutant proteins in different domains and their misfolding with the exception of mutant proteins in the carboxyterminal domain that were mostly assembled correctly. In summary, the above analyses demonstrated that a substantial proportion of CBS mutant proteins formed severely reduced amounts of correctly assembled tetramers; these findings add further support to the notion that misfolding is an important pathogenic mechanism in CBS deficiency.
Different Mobility of CBS Mutant Proteins in Native Westerns
The above-mentioned Western blot analysis revealed slight but consistent differences in migration distance of the putative tetramers/oligomers of many mutant proteins in comparison to the wild-type CBS tetramer (see ). These mobility alterations may have resulted from charge differences, conformational changes of tetramers, or even different number of subunits in the differently migrating fractions. To explore the latter hypothesis we used the Ferguson plot to determine the molecular weight of four fractions seen for wild-type enzyme, and the molecular weight of fast migrating (K102N, S466L), slow migrating (E302K, G305R, and G307S) and fast/slow migrating (R125Q) mutant proteins. For results see Supp. Figure S1
The apparent molecular weights of sharply demarcated and thus presumably correctly assembled oligomer fractions seen for the wild-type enzyme were 177, 398, 512, and 724kDa. Considering the limitations of the technique employing crude extracts with subsequent Western blotting and keeping in mind that CBS is known to form tetramers, the results suggest that in our electrophoretic system the wild-type CBS is present predominantly in the form of tetramers, and in small amounts also as octamers, dodecamers, and hexadecamers. Two fractions of the R125Q mutant protein were predicted to be composed of tetramers and octamers, respectively. There was only a small difference in slopes, and thus in estimated molecular weights of the remaining fast (162 and 166 kDa) and slow (199–213 kDa) migrating mutant proteins as demonstrated in Supp. Figure S1
. Consequently, the estimated number of CBS subunits ranged between 2.7 and 3.5 per fraction, which is congruent with tetrameric composition of these fractions. In summary, the results of Ferguson plot do not support a hypothesis that different mobility of mutant proteins compared to wild-type CBS is caused by changes in the number of enzyme subunits. Thus, the observed varying mobility of CBS mutant proteins results most likely from subtle changes in the shape and/or charge of the assembled tetramers.
Catalytic Activity of Mutant Proteins
To enable measurements of low residual activities of mutant proteins we first developed a sensitive method for the analysis of the CBS reaction product—cystathionine. The radioactive substrate used in previous studies, 14C-labeled serine, was replaced by 2,3,3- 2H-labeled serine, and the amount of 3,3- 2H-labeled cystathionine produced was determined using LC-MS/MS (Krijt et al., unpublished). Using this technique, we were able to measure reproducibly activities as low as 0.3 nmol cystathionine/hr/mg of protein, which represents about 0.2% of the average wild-type CBS activity in E. coli extracts.
Using the sensitive new assay we observed an extreme variation of activities ranging from 0 to 245% of the wild-type enzyme with a median activity of 3% of the wild-type CBS (for details, see ). Only three mutant proteins—p.G305R, p.T262R, and p.N228K—did not exhibit any measurable residual activity above 0.3 nmol/mg/hr, whereas the remaining 24 mutant proteins had above-threshold activity in at least two of the three independent expression experiments. The activities determined in the present study are for the majority of mutant proteins congruent with the previously published values listed in . In contrast to effects on folding location of the mutant residues appears to play a more important role in affecting their catalytic activity of variant proteins. There was a tendency for mutations located in the active site, heme binding pocket, and elsewhere in the active core to exhibit rather low activities while mutations inside or on the surface of the first CBS domain and at the dimer–dimer interface were in many cases as active as the wild-type enzyme (for details, see ). These data suggest that mutations in different domains of the enzyme may have substantially variable impact on its catalytic activity, and that they may not affect folding and activity simultaneously.
Activities of Mutant Enzymes in Nonparticulate Water-Soluble Fractions of Bacterial Extracts
Effect of AdoMet and AdoHcy on the Catalytic Activity of Mutant Proteins
To get an insight into the possible pathophysiological and therapeutic implications of AdoHcy and AdoMet in CBS deficiency we studied the effect of these two compounds on the activity of mutant proteins in nonparticulate fractions of bacterial extracts. Blood concentration of AdoHcy is increased in patients with CBS deficiency [Orendac et al., 2004
] due to its reverse synthesis from homocysteine and adenosine with the help of AdoHcy hydrolase, an enzyme that can be in vitro blocked with different inhibitors. AdoMet is an allosteric activator of CBS that can possibly stimulate the residual activity of mutant proteins and that can be administered as a drug in the form of tosylate.
Activity of the wild-type CBS enzyme increased 3.9 × and 1.3 × in the presence of 0.5 mM AdoMet and 0.5 mM AdoHcy, respectively (see ). Because activation of CBS by AdoHcy has not yet been reported, we explored whether it could not have been caused by contamination of this compound by AdoMet. However, LC-MS/MS analysis showed that the AdoHcy standard contained less than 1% of AdoMet, indicating that the 1.3 × increase in wild-type CBS activity in crude bacterial extracts can be attributed to AdoHcy itself.
We classified without formal statistical testing the response of mutant proteins to AdoMet and AdoHcy in this simple screening system as follows: clear activation similar to the wild-type enzyme, clear inhibition, or absence of expected activation. About one-half of the mutant enzymes in the panel exhibited an unusual response to either one of the adenosyl compounds: the p.E176K was inhibited by AdoMet, the p.E302K was inhibited by AdoHcy, and the p.H65R, p.T191M and p.C165Y appeared to be inhibited by both adenosyl compounds. In addition, mutations p.E302K, p.W409_G453del, p.I435T, p.L539S, p.E144K, p.G148R, p.G307S, p.I278T, and p.S466L were not activated by AdoMet to an extent that was usually observed for the wild type CBS (see Table 2). These data show that responses of about half of mutant proteins to AdoMet and AdoHcy may differ substantially from those of the wild-type CBS. The relevance of these findings for regulation of the activity of mutant proteins and possibly for treatment of CBS deficiency is further explored in the Discussion section.
Effect of Low Temperature on Properties of Mutant Proteins
Low temperature during expression has been shown to enhance folding of various mutant proteins that are prone to misfolding. To examine whether any of the CBS mutations under study may be rescued by more folding-permissive conditions we expressed all 27 mutant proteins at 18°C. Compared to expression at 37°C the lower temperature has not dramatically changed the total amount of SDS-soluble mutant CBS enzymes and only slightly decreased the proportion of signal in the particulate fraction (median of 18% at 37°C vs. 15% of total signal at 18°C; see Supp. Tables S2
). However, the median amount of correctly folded oligomers in the water-soluble nonparticulate fraction detected on native-PAGE increased about two times from 31% of total signal at 37°C to 68% of total signal at 18°C (Supp. Table S2
). Several mutant proteins responded to low expression temperature with a dramatic (i.e., more than approximately threefold) increase of the proportion of tetramers compared to the proportion of tetramers/oligomers of the wild-type CBS, namely, the p.R266K, p.R369C, p.R125Q, p.P78R, p.V180A, and p.I435T. However, this folding promoting effect of low temperature was not universal and some mutant proteins produced conversely less tetramer. All these observations indicate that for some mutant proteins more folding-permissive conditions facilitate the attainment of correct tertiary structure, which subsequently increases the assembly of tetramers.
In addition, the low expression temperature increased the median catalytic activity of mutant proteins from 3% of wild type at 37°C to a median of 13% of wild type at 18°C (see ). The mutant proteins p.R266K, p.R125Q, p.R369C, p.V180A, and p.P78R exhibited more than approximately twofold increase in activity relative to the wild type, which correlated well with the increased amount of tetramer. These five mutant proteins may be considered true folding mutants as their residual activity directly correlates with the amount of tetramer. In summary, the lower expression temperature substantially increased activity of five mutant proteins, further supporting the hypothesis that deficient CBS activity may be in some instances caused by misfolding that may be alleviated by folding permissive conditions.
Solvent Exposure of Mutant Residues as a Determinant of Folding and Activity
We hypothesized that topology of mutations in respect to solvent exposure/buriedness may have an impact on folding and assembly of mutant CBS molecules. Therefore, we compared in the next step the typical properties of CBS enzymes carrying buried versus solvent-accessible mutations by comparing median values in Table 2 and in Supp. Table S2
. We have observed similar stability of these two classes of mutant proteins in E. coli
as evidenced by identical amounts of total SDS-soluble antigen. However, the propensity of buried mutations to correctly fold and assemble is impaired, as clearly demonstrated by the lower proportion of tetrameric/oligomeric fractions in the total CBS antigen in native gels with a median of only 14% of total antigen for buried mutations versus 36% for the solvent accessible ones. More importantly, the buried mutations are about 23 times less active with median activity of only 0.4% of the wild-type CBS versus 9.4% of the wild-type enzyme for solvent-exposed mutations. In addition, the buried mutations are not activated by AdoMet (the median activation of buried mutations of 1.0 × contrasts with the 2.5 × activation of solvent exposed mutation) and seem to be inhibited by AdoHcy (median activation of buried mutations is 0.8 ×, while it is 1.3 × for solvent-exposed mutations). These differences between buried and solvent-exposed mutations appear to be further augmented by the expression at 18°C. In summary, these comparisons strongly suggest that buried mutations may lead to much severe impairment of CBS structure and function than the solvent-exposed ones.
Expression of several mutant proteins at 18°C resulted in an increase in their catalytic activity as well as in the amount of tetramers/oligomers; we therefore proposed that these variant proteins may be considered true folding mutations. To evaluate this phenomenon in a more systematic way we have plotted the normalized activity of mutant proteins against the normalized amount of tetramers/oligomers using the combined data from five independent expression experiments. Supp. Figure S2
demonstrates that among the solvent-exposed mutations there are indeed examples of such increase of activities mirroring the increased amounts of tetramers (e.g., p.A114V, p.H65R, p.K102N, p.P78R, p.R125Q, and p.V180A). However, changes in amounts of tetramers of buried mutations were not usually accompanied by corresponding differences in activities. That led us to speculate that the solvent exposure of mutations may be indeed an important determinant of their propensity to misfold and to lose the catalytic activity. We have next performed a regression analysis in which we combined data for all buried and all solvent-exposed mutant proteins, respectively. This analysis has demonstrated that the activity of solvent-exposed mutations correlates quite strongly with the amount of oligomers (r2
= 0.53; see ), and suggests that for this type of mutations enhanced folding achieved by, for example, low temperature or chaperones may result in increased residual activity. In contrast, the mutations buried in the enzyme globule did not exhibit such properties, as there was only a weak correlation between the activity and the amount of oligomers (r2
= .11; see ). The above-described regression analysis suggests that solvent-exposed CBS mutations may be more likely to respond to interventions aimed at correcting their misfolding.
Figure 3 Solvent exposure and misassembly of mutants. (A) buried mutations; (B) solvent exposed mutations. Data on normalized activities and tetramer amounts shown in Supp. Figure S2 are here combined separately for all solvent exposed and all buried mutations; (more ...)