Functional Synergism between the Most Common Polymorphism in Human Alanine:Glyoxylate Aminotransferase and Four of the Most Common Disease-causing Mutations*

The autosomal recessive disorder primary hyperoxaluria type 1 (PH1) is caused by a deficiency of the liver-specific pyridoxal-phosphate-dependent enzyme alanine:glyoxylate aminotransferase (AGT). Numerous mutations and polymorphisms in the gene encoding AGT have been identified, but in only a few cases has the causal relationship between genotype and phenotype actually been demonstrated. In this study, we have determined the effects of the most common naturally occurring amino acid substitutions (both normal polymorphisms and disease-causing mutations) on the properties, especially specific catalytic activity, of purified recombinant AGT. The results presented in this paper show the following: 1) normal human His-tagged AGT can be expressed at high levels in Escherichia coli and purified in a correctly folded, dimerized and catalytically active state; 2) presence of the common P11L polymorphism decreases the specific activity of purified recombinant AGT by a factor of three; 3) AGTs containing four of the most common PH1-specific mutations (G41R, F152I, G170R, and I244T) are all soluble and catalytically active in the absence of the P11L polymorphism, but in its presence all lead to protein destabilization and aggregation into inclusion bodies; 4) naturally occurring and artificial amino acid substitutions that lead to peroxisome-to-mitochondrion AGT mistargeting in mammalian cells also lead to destabilization and aggregation in E. coli; and 5) the PH1-specific G82E mutation abolishes AGT catalytic activity by interfering with cofactor binding, as does the artificial K209R mutation at the putative site of cofactor Shiff base formation. These results are discussed in the light of the high allelic frequency (∼20%) of the P11L polymorphism and its importance in determining the phenotypic manifestations of mutations in PH1.

Primary hyperoxaluria type 1 (PH1; MIM number 259900) 1 is a rare autosomal recessive disorder characterized by the excessive synthesis and excretion of oxalate and glycolate, and the progressive accumulation of insoluble calcium oxalate in the kidney and urinary tract (1)(2)(3). PH1 is caused by a functional deficiency of the liver-specific pyridoxal-phosphate-dependent enzyme alanine:glyoxylate aminotransferase (AGT; Enzyne Nomenclature 2.6.1.44) (4). Over 30 mutations have been identified in the AGT gene (AGXT), the most common being G170R and I244T replacements, which are found in approximately 30% and 9% of disease alleles, respectively (5)(6)(7). Numerous intragenic polymorphisms have also been identified, the most common being P11L and I340M replacements, which co-segregate with allelic frequencies of 20% in normal European and North American populations and 50% in PH1 patients (5,8). The combined presence of the P11L and I340M polymorphisms defines the minor AGXT allele, whereas their absence defines the major allele (9).
At the enzyme level, PH1 is extremely heterogeneous. Some patients have no AGT catalytic activity or immunoreactivity, and some have immunoreactive AGT but no AGT catalytic activity. However, about a third of patients have both catalytic activity and immunoreactivity (9). In these patients, who can have AGT levels as high as 30 -50% of the mean normal level, disease is due not to the absence of AGT but to it being targeted to the wrong intracellular compartment. In normal human hepatocytes, AGT is localized exclusively to the peroxisomes (10), but in these latter PH1 patients it is instead localized mainly to the mitochondria (11). Such organelle-to-organelle protein mistargeting is unparalleled in human genetic disease. AGT mistargeting is due to the combined effects of the P11L polymorphism and the PH1-specific G170R mutation (8). The P11L replacement generates a functionally weak N-terminal mitochondrial targeting sequence, the effective strength of which is impaired because it is unable to maintain newly synthesized AGT in an import compatible (i.e. monomeric) conformation (12,13). The additional presence of the G170R replacement increases the functional efficiency of this polymorphic mitochondrial targeting sequence by slowing down the rate with which AGT dimerizes, possibly by providing a molecular environment within which the polymorphic N terminus of AGT can adopt an ␣-helical secondary structure typical of mitochondrial targeting sequences (13,14).
Apart from the synergistic effects of the P11L polymorphism and the G170R mutation on AGT mistargeting, only one other (isolated) mutation has been studied in any detail (15,16). In no other cases have the effects of mutations on the properties of AGT been determined directly to explain the associated PH1 enzymic phenotypes. In this study, we have attempted to rectify this situation by expressing a variety of His-tagged normal, polymorphic, and mutant AGT constructs in Escherichia coli, purifying them on Ni 2ϩ resin columns and determining their various properties, including specific catalytic activity and, in some cases, also K m , pH optima, pyridoxal phosphate binding, stability, and dimerization. Our results demonstrate not only the remarkable synergistic effect that the common P11L polymorphism has on the phenotypic expression of four of the most common PH1-specific mutations, accounting for ϳ40% of disease alleles in European and North American patients but also that this polymorphism alone might not be without deleterious consequences for the general population.

MATERIALS AND METHODS
Expression Constructs-The expression constructs used in this study are described in Table I and the PCR primers used in their construction  are listed in Table II. The partial restriction map of AGT cDNA in relationship to the various polymorphisms and mutations is shown in Fig. 1. Preparation of the His-AGT and His-AGT[Leu 11 ,Arg 170 ,Met 340 ] expression plasmids has been described previously (14).
AGT-His was made as follows. The HindIII/BamHI fragment of H1-2 (17), which contains the complete open reading frame of human AGT cDNA, was cloned into pBluescriptKSϩ in which the multiple cloning ApaI site had been previously deleted (18) to give pAGT⌬ApaI. The 3Ј region of pAGT⌬ApaI was amplified by PCR using the internal primer P1 and the mutagenic primer P2, which adds SalI and BamHI sites to the 3Ј end of the coding region. The P1/P2 PCR product was digested with ApaI and BamHI and cloned back into ApaI/BamHI-digested pAGT⌬ApaI to give pAGTSalI. The NcoI/SalI fragment of pAGTSalI was then cloned into the NcoI/SalI sites of the E. coli expression vector pTrcHis2A (Invitrogen) to give pAGT-His, which consists of the complete open reading frame of normal human AGT together with a C-terminal AVDHHHHHH tag. All PCR products were sequenced to check for errors.
E. coli Expression and Purification-Constructs were transformed into E. coli JM109 and 40 -200-ml cultures grown at 37°C to an A 600 of 0.4 -0.6. Expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside, and the cells were grown at 37°C for a further 4 -6 h. In some experiments the expression phase was carried out at 30°C. Cells were harvested, and the His-tagged proteins were purified using the Xpress Protein Purification System (Invitrogen) according to the manufacturer's instructions. Soluble AGT was batch purified under native conditions, eluting from the Ni 2ϩ resin between 350 and 500 mM imidazole. Imidazole was removed and protein concentrated using Centriprep 10 Concentrators (Amicon). Purified proteins were stored in 20 mM sodium phosphate buffer, pH 6.0, 0.5 M NaCl, and 0.5 mM pyridoxal phosphate. Purity varied between 95 and 98%, with yields typically about 10 mg/liter culture.
Enzyme Assays-Enzyme assays were carried out as described previously (20), except that the AGT assay was modified by using the following concentrations: alanine (150 mM), glyoxylate (10 mM), pyridoxal phosphate (150 M) and increasing the pH to 8.0 (21).
Enzyme Kinetics-Enzyme assays for the kinetic analyses were carried out under V o conditions. The K m for glyoxylate was determined over a substrate range of 0 -2.5 mM with a fixed alanine concentration of 150 mM. The K m for alanine was determined over a substrate range of 0 -150 mM with a fixed glyoxalate concentration of 10 mM. The data were used to generate a Lineweaver-Burke plots, and the K m value was obtained from the linear equations.
Absorption Spectrometry-Free pyridoxal phosphate was removed from the purified protein preparations using the Centriprep 10 Concentrators by washing four times in 20 mM sodium phosphate buffer, pH 6.0, 0.5 M NaCl. Protein concentration was determined using the Coomassie Plus Protein Assay Reagent (Pierce). Samples were diluted to a protein concentration of 1 mg/ml buffer, and the absorption spectra were recorded between 250 and 500 nm against buffer blanks.
Chemical Cross-linking-Photo-induced cross-linking was carried out as described by Fancy and Kodadek (22) as follows. Purified recombinant His-tagged AGT (2.5 g) was cross-linked in 20 l of buffer (containing 15 mM sodium phosphate, pH 7.5, 150 mM NaCl, and 125 M TBPR). Immediately following the addition of 2.5 mM ammonium persulphate, the sample was irradiated at a distance of 10 cm by a 100 watt tungsten microscope light source for 20 s. TBPR-treated samples were mixed with equal volumes of 4% SDS and 10% 2-mercaptoethanol in 0.125 M Tris buffer, pH 6.8, heated to 95°C for 5 min, and electrophoresed on a 8% reducing polyacrylamide gel. Separated products were visualized by silver staining and immunoblotting using polyclonal rabbit anti-human AGT antibody. Miscellaneous Reagents-The thiol agents p-chloromercuribenzoate and iodoacetamide, the disulphide reducing agents dithiothreitol and 2-mercaptoethanol, and the specific AGT inhibitor amino-oxyacetic acid were obtained from Sigma. Table I) could be expressed to high levels in E. coli (a number of examples are shown in Fig.  2). With most of the "normal" constructs, the majority of the immunoreactive AGT protein could be recovered from the soluble fraction in a catalytically active form, confirming the absence of any requirement for eukaryotic-specific post-translational modification. However, for most constructs, detectable amounts could also been found in the insoluble pellet. Some constructs (see below) were almost undetectable in the soluble fraction (Ͻ0.2% of the total soluble cell protein) but instead were recovered almost entirely from the insoluble pellet. Quantitative densitometry of silver-stained gels (an example is shown in Fig. 3) indicated that for the normal constructs expression levels ranged between 12 and 24% of the total soluble cell protein. All constructs that expressed in the soluble fraction could be rapidly purified on Ni 2ϩ resin columns to give a purity of 95-98% (Fig. 3). Those constructs that expressed at a level of Ͻ0.2% of the total soluble cell protein could not be purified.

Properties of Normal and Polymorphic Purified Recombinant
His-tagged AGT-The specific activity of purified AGT-His (i.e. that encoded by the major AGXT allele) varied between 2778 and 3345 mol/h/mg protein (mean ϭ 3112) in three consecutive batches. The specific activity of His-AGT was generally about 70% of that of AGT-His. Surprisingly, the specific activity of AGT[Leu 11 ,Met 340 ]-His (i.e. that encoded by the minor AGXT allele) was only about 46 -50% of that of AGT-His (Fig.  4). This lower specific activity appeared to be due entirely to the presence of the P11L polymorphism rather than the I340M polymorphism, because the former on its own gave specific activities of 27-41% normal, whereas the latter gave values of 76 -117% of normal values (Fig. 4).
Because of the very different assay methods used, it is very difficult to compare directly the specific activity of recombinant human AGT purified in the present study with that purified by chromatographic methods from human liver. Nevertheless, our values are somewhat higher than some published values (e.g. ϳ650 mol/h/mg protein (23)) but lower than others (e.g. ϳ6000 mol/h/mg protein (24)). 5Ј-CTTCTCCTTCTCCCTGGACA-3Ј 5Ј-AGTGCTGCTGATCTTAACCCA-3Ј To check whether purified recombinant His-tagged AGT is functionally similar to AGT in human liver cells in situ or AGT purified from human liver, we determined the K m for alanine and glyoxylate, pyridoxal phosphate binding, pH optima, and the ability to dimerize. In general, the enzymological properties were as expected. K m for alanine was 9.1 mM for His-AGT and 9.4 mM for AGT-His, whereas for glyoxylate it was 0.23 mM for His-AGT and 0.39 mM for AGT-His. The K m values for alanine were similar to those found previously for AGT in liver samples (7.7-13.5 mM) (21,25,26) but rather larger that found previously with AGT purified from human liver (1 mM) (23). The K m values for glyoxylate measured in the present study, however, were significantly less than found previously in liver samples in some papers (1.3 and 2.5 mM) (21,25) but similar to those found in others (0.21 mM) (26). It has been reported previously (20,21,23,25) that AGT in liver extracts and AGT purified from human liver shows marked substrate inhibition at glyoxylate concentrations above 20 mM, although under some conditions the inhibition is not particularly marked (27). In the present study, purified His-AGT showed no significant inhibi-tion with glyoxylate concentrations up to 50 mM (data not shown). His-AGT, His-AGT[Leu 11 ,Met 340 ], AGT-His, and AGT[Leu 11 ,Met 340 ]-His had broad pH optima (7.5-8.5) that were similar to those reported previously for AGT in whole liver extracts or purified from liver (23,25).
Previous studies have shown that AGT also has serine:pyruvate aminotransferase activity, which although much lower than its alanine:glyoxylate aminotransferase activity might be important for its gluconeogenic role (28). In the present study, the serine:pyruvate aminotransferase activity of purified recombinant AGT-His, AGT[Leu 11 ]-His, AGT[Met 340 ]-His, and AGT[Leu 11 ,Met 340 ]-His varied between 3 and 5% of the AGT activity. Previous studies have shown that glutamate:glyoxylate aminotransferase has AGT activity, whereas AGT does not have glutamate:glyoxylate aminotransferase activity (23,29). In this study, purified recombinant AGT-His had negligible amounts (Ͻ0.2%) of glutamate:glyoxylate aminotransferase activity.
Human AGT has six cysteinyl moieties (residues 84, 173, 178, 253, 366, and 687), which with the exception of Cys 366 are conserved in all mammals so far studied. To test whether they contribute to the structure or catalytic activity of AGT, we determined the effects of various thiol and disulphide reagents on the specific activity of AGT-His. Neither dithiothreitol, 5-mercaptoethanol, p-chloromercuribenzoate, nor iodoacetate, all at a concentration of 1 mM, had any effect on the specific activity of purified AGT-His (data not shown), suggesting that either the conserved cysteinyl residues do not form crucial disulphide bonds and the free thiols are not crucial for stability or activity or alternatively that these residues are not accessible to the reagents. Previous studies had shown that p-chloromercuribenzoate also had only a minimal effect on the activity of AGT purified from human liver (23). Amino-oxyacetic acid has been shown to be a powerful inhibitor of AGT activity in human liver samples (30). In the present study, 1 mM aminooxyacetic acid inhibited the activity of purified AGT-His by 97.5%.
AGT is homodimeric in human liver in situ, when purified from human liver and when translated in vitro with a subunit molecular mass of 43 kDa (13,31). Purified recombinant AGT-His can be very efficiently photo-cross-linked using TBPR (22) to give a diffuse band of ϳ86 kDa, confirming its dimeric status (Fig.  5). The efficiency of AGT[Leu 11 ]-His and AGT[Leu 11 ,Met 340 ]-His cross-linking was similar to that of AGT-His when the purified protein was prepared from cells grown at 30°C. However, significantly less AGT[Leu 11 ]-His was cross-linked when prepared from cells grown at 37°C (Fig.  5, lane 5). This is compatible with our previous findings, using the much less efficient cross-linkers BS3 and sulfo-MBS, that in vitro translated AGT[Leu 11 ] cross-linked with lower efficiency than AGT (13,14). The temperature dependence of AGT[Leu 11 ]-His dimerization in the present study might reflect subtle differences in folding, a possibility that might also contribute to the lower specific activity of AGT[Leu 11 ]-His compared with AGT-His (see above).
Effects of Mutations on the Specific Activities of Purified Recombinant His-tagged AGT-Four of the mutations in this study (G41R, F152I, G170R, and I244T) segregate with the minor AGXT allele (see references quoted in Table I). The latter three have never been found on the major allele. On the other hand, G41R has been found on the major allele, but only in one patient (32). In PH1 patients, G41R, F152I, and I244T are associated with absence, or near absence, of immunoreactive AGT protein and catalytic activity, and G170R is associated with peroxisome-to-mitochondrion AGT mistargeting (see Table I).
When these mutations were expressed on the background of the minor AGXT allele, none of the protein could be recovered from the soluble fraction; instead almost all was aggregated into the insoluble pellet (Figs. 2 and 4) Interestingly, an artificial construct (AGT[Leu 10 ,Leu 11 ]-His), which is also targeted to the mitochondria in transfected COS cells (14), was also unstable and aggregated in the E. coli expression system (Figs. 2 and 4).
Surprisingly, all four of these mutations expressed on the background of the major allele were stable (i.e. expressed at normal or near normal levels in the E. coli soluble fraction),  (Table I). Normal residues (encoded by the major AGXT allele) are shown in white boxes, normal polymorphic residues (encoded by the minor AGXT allele) are in light gray boxes, and mutant residues are in dark gray boxes (naturally occurring PH1 mutations are in single boxes, and artificial mutations not found naturally are in double boxes). Also shown is whether or not the allele has been found naturally in normal or PH1 individuals, the approximate percentage allelic frequency (values not in parentheses indicate the normal population, and values in parentheses indicate PH1 patients), the presence or absence of catalytic activity in human liver (for the naturally occurring constructs only), the ability of the purified recombinant constructs to bind pyridoxal phosphate (either by direct measurement or implied because of the detection of catalytic activity), and the principle subcellular distribution of the constructs in human liver or transfected COS cells.  (Fig. 4). These latter values were similar to, or even greater than, the specific activity of normal AGT encoded by the minor AGXT allele.
Unlike the mutations above, the fifth mutation in this study (G82E) segregates with the major AGXT allele. Patients who are homozygous for G82E have normal amounts of correctly targeted immunoreactive AGT but almost zero catalytic activity (33). AGT containing this mutation (on the background of the major allele) was stably expressed at high levels in the soluble fraction in E. coli and could easily be purified. However, it had no catalytic activity (specific activity Ͻ2% of the control value) (Fig. 4), a result that is compatible with the PH1 enzymic phenotype. Unlike most of the other constructs, which when purified had a distinct yellow coloration, His-AGT[Glu 82 ] was colorless, suggesting that it had not bound its cofactor pyridoxal phosphate despite saturating amounts being present throughout the purification procedure. This was confirmed by demonstrating the loss of the characteristic absorption peak at 420 nm typical of pyridoxal phosphate in a Shiff base linkage (Fig. 6) (34). So far G82E is the only known example of a mutation in PH1 that abolishes catalytic activity by preventing cofactor binding. This is likely to be an irreversible step because our previous studies on the liver of a PH1 patients homozygous for G82E showed that increasing the pyridoxal phosphate concentration by 100 times made no difference to the level of enzyme activity (20).
The actual Shiff base binding site for pyridoxal phosphate in rat AGT has been suggested to be Lys 209 (34,35). In the present study, we have confirmed the importance of this residue in human AGT by showing not only that its substitution by arginine abolished catalytic activity (Fig. 4) but also that it prevents pyridoxal phosphate binding. Not only was purified AGT[Arg 209 ]-His colorless, but also the absorption spectrum showed a loss of the characteristic absorbance peak at 420 nm (Fig. 6). For reasons that are not immediately clear, the spectrum of His-AGT was different from that of AGT-His with a shift of the Shiff base peak from 420 to 410 nm, suggesting differences in cofactor binding.

Validity of the Expression and Purification
System-The normal C-and N-terminally His-tagged AGT constructs encoded by the major AGXT allele had properties such as solubility, catalytic activity, cofactor binding, molecular size, and dimerization as expected from similar studies on nontagged AGT using a variety of other systems. In addition, notwithstanding the predictable finding that N-terminal tags prevent mitochondrial targeting and C-terminal tags prevent peroxisomal targeting, our previous studies have shown that His tags do not affect the behavior of constructs in transfected tissue culture cells (14). This suggests not only that the protein folds similarly in prokaryotic cells to how it does in mammalian cells but also that the His-tag does not significantly interfere with the process. This was not a foregone conclusion because, for example, C-terminal His tags have been shown to interfere with the folding and processing of ␤-lactamase in Bacillus licheniformis (36). Thus, the observations on His-tagged constructs in the present study are probably relevant to the situation of AGT in  (34). The spectrum of AGT-His (B) is almost identical to that obtained from pyridoxal phosphate saturated rat AGT (34). The shift in the Shiff base peak (from 420 to 410 nm) and the presence of a peak at 330 nm with His-AGT (A) suggests that location of the His tag at the N terminus, as opposed to the C terminus, does alter to some extent the interaction with cofactor, a finding that might explain the 30% decrease in specific activity (see Fig.  4). The absence of a peak at 420 nm and the presence of a peak at 330 nm with AGT[R 209 ]-His (D) and His-AGT[Glu 82 ] (C), together with the loss of yellow color (see text), suggest a marked interference with cofactor binding. situ (i.e. in human hepatocytes).
Population Genetics of the P11L Polymorphism-One of the more surprising findings from the present study is that polymorphic AGT encoded by the minor AGXT allele has less than half of the specific activity of that of AGT encoded by the major allele. The presence of the P11L polymorphism alone reduces the specific activity of AGT by up to 75%. This might contribute to the wide range of AGT activities found in the normal population (20). Unfortunately, the liver-specific nature of AGT expression in humans precludes a comprehensive analysis of the relationship between the presence of the minor and major AGXT alleles and AGT activity in situ. Although the minor allele contains two amino acid substitutions (P11L and I340M), as well as a 74-base pair duplication in intron 1 (37) and a characteristic VNTR in intron 4 (38), it is only the P11L polymorphism that appears to have any affect on AGT function, including peroxisome-to-mitochondrion mistargeting, inhibition of dimerization, and decreased specific activity. In fact, as far as specific activity is concerned, it appears that the I340M polymorphism is able to, at least partially, normalize some of the consequences of the P11L polymorphism.
The evolutionary direction of the P11L replacement is clearly in the Pro 3 Leu direction because Pro 11 is conserved in all mammals studied so far, including the closely related great apes, such as chimpanzee and gorilla (39). However, the direction of the I340M replacement is less obvious, because it is poorly conserved. However, in the closest relative to humans in which the sequence in this region is known (i.e. the marmoset), isoleucine is present at this position as it is in most humans (40).
The explanation of the high frequency of the minor AGXT allele in European and North American populations (ϳ20%) is unclear. Interestingly, it appears to be present at a much lower frequency (ϳ2%) in Japanese populations (38). The question arises as to whether the presence of the P11L polymorphism is neutral or whether it provides the bearer with a selective advantage or disadvantage. It has been speculated previously that P11L might be evolutionarily advantageous. Whereas AGT encoded by the major allele is 100% peroxisomal, that encoded by the minor allele is 95% peroxisomal and 5% mitochondrial (in a homozygous individual). The latter distribution might be more compatible with our modern omnivorous (as opposed to our ancestral herbivorous) lifestyle (41). Conversely, P11L might be disadvantageous and only exist at high frequency because of, for example, an evolutionary bottleneck. Low AGT activity, as predicted for individuals homozygous for the minor allele, might be associated with a decrease in the capacity to transaminate (detoxify) glyoxylate to glycine (3). This might lead to mild hyperoxaluria, a frequent finding in idiopathic calcium oxalate stone disease (42,43). Whether the presence of the P11L polymorphism leads to an increased susceptibility to this very common, almost certainly multifactorial condition remains to be seen.
Any potential disadvantage that might result from the presence of P11L might be at least partially attenuated by the additional presence of I340M (the specific activity of AGT[Leu 11 ,Met 340 ] is higher than that for AGT[Leu 11 ]). This might explain the tight but not complete linkage between these two polymorphisms. A recent study of PH1 patients (7) showed that although P11L and I340M co-segregated in 44% of alleles, they were found separately in 2%. A similar frequency of linkage disruption is likely also in the normal population, so that 1% of normal alleles might possess P11L in the absence of I340M. The consequences for the individual of P11L unattenuated by I340M are currently unknown.
Functional Synergism between the P11L Polymorphism and Four of the Most Common PH1-specific Mutations-The functional synergism between the P11L polymorphism and the most common PH1-specific mutation (G170R) with respect to the inhibition of AGT dimerization and peroxisome-to-mitochondrion mistargeting is well recognized, as is its molecular basis (8,13). However, it was surprising to find that a functional synergism also existed between this polymorphism and three other mutations (G41R, F152I, and I244T). From this and our previous studies, it seems likely that F152I, G170R, and I244T would be relatively innocuous (i.e. without pathological phenotype) in the absence of P11L. G41R, on the other hand, did lead to a significant but not complete reduction in AGT specific activity (7% of normal) when present on the major allele. This finding is compatible with G41R having been found in one patient (homozygously) on the background of the major allele (32). This patient, however, was relatively mildly affected (as predicted from a residual 7% of activity), whereas when G41R was present on the minor allele (heterozygously with F152I or G170R) the patients were severely affected (19) (as predicted from the present studies).
There are few good examples in which the effects of diseasespecific mutations are modified by the presence of polymorphisms encoded by the same allele. Probably the best known is the different pathological consequences of the D178N mutation in the prion protein gene depending on the presence or absence of a M129V polymorphism (44). If residue 129 is methionine, then patients suffer from fatal familial insomnia, whereas if valine is present at this location then patients suffer from Creutzfeld-Jakob disease. There are no known consequences of the polymorphic variation in the absence of the D178N mutation. Unfortunately, the suggestion in the present study that there would be no clinical phenotype resulting from the presence of the F152I, G170R, and I244T mutations, except on the background of the minor AGXT allele, cannot be verified until appropriate individuals can be identified (presumably from the normal population).
Because most other mutations currently recognized at the AGXT locus are very rare, it is not known whether they segregate with the major or minor alleles. Therefore, it is not known whether any other mutation would require the presence of the P11L polymorphism to give a phenotype. The much higher allelic frequency of the P11L polymorphism in PH1 patients (50%) relative to the normal population (20%) can be easily explained by its segregation with the two most common mutations (G170R at 30% and I244T at 9%) together with a predicted frequency of 20% in the remaining mutant alleles. This would suggest that there is no other high frequency mutation in the AGXT gene that requires the presence of the P11L polymorphism.
Diagnostic Implications-The prediction that the most common mutations found in PH1 patients (F152I, G170R, and I244T) are only likely to lead to pathological phenotypes when associated with the P11L polymorphism means that the correct molecular diagnosis of PH1 cannot be guaranteed unless both mutation and polymorphism are identified on the same allele. Recombination between these mutations and P11L will inevitably have occurred in some individuals. If such alleles were to be identified in patients with idiopathic calcium oxalate stone disease, for example, then diagnostic confusion could ensue unless the presence or absence of P11L is also determined.
Instability and Aggregation of AGT Constructs Targeted to Mitochondria-The instability of the constructs containing amino acid replacements known to lead to AGT mistargeting to mitochondria (AGT[Leu 11 ,Arg 170 ,Met 340 ] and AGT[Leu 10 , Leu 11 ])and their aggregation in the insoluble pellet was a surprising finding. Both express reasonably well in transiently transfected COS cells in which they are efficiently targeted to the mitochondria (14,18). In addition, the naturally occurring PH1 mutant construct AGT[Leu 11 ,Arg 170 ,Met 340 ] can reach 30 -50% of the mean normal level in the livers of homozygous patients (45). Both the AGT[Leu 11 ,Arg 170 ,Met 340 ]-His and AGT[Leu 10 ,Leu 11 ]-His constructs are predicted to fold and dimerize more slowly than normal AGT, possibly because of the presence of an ␣-helical secondary structure at the N terminus (14). Therefore, it is possible that they would be more likely to aggregate in a prokaryotic environment. The relatively high levels of AGT[Leu 11 ,Arg 170 ,Met 340 ] in the livers of PH1 patients could conceivably result from some sort of compensatory mechanism at the level of transcription and/or translation, or protection from cytosolic aggregation/degradation by import into mitochondria.