Hypermethioninemias of Genetic and Non-Genetic Origin: A Review

This review covers briefly the major conditions, genetic and non-genetic, sometimes leading to abnormally elevated methionine, with emphasis on recent developments. A major aim is to assist in the differential diagnosis of hypermethioninemia. The genetic conditions are: (1) Homocystinuria due to cystathionine b-synthase (CBS) deficiency. At least 150 different mutations in the CBS gene have been identified since this deficiency was established in 1964. Hypermethioninemia is due chiefly to remethylation of the accumulated homocysteine. (2) Deficient activity of methionine adenosyltransferases I and III (MAT I/III), the isoenzymes the catalytic subunit of which are encoded by MAT1A. Methionine accumulates because its conversion to S-adenosylmethionine (AdoMet) is impaired. (3) Glycine N-methyltrasferase (GNMT) deficiency. Disruption of a quantitatively major pathway for AdoMet disposal leads to AdoMet accumulation with secondary down-regulation of methionine flux into AdoMet. (4) S-adenosylhomocysteine (AdoHcy) hydrolase (AHCY) deficiency. Not being catabolized normally, AdoHcy accumulates and inhibits many AdoMet-dependent methyltransferases, producing accumulation of AdoMet and, thereby, hypermethioninemia. (5) Citrin deficiency, found chiefly in Asian countries. Lack of this mitochondrial aspartate–glutamate transporter may produce (usually transient) hypermethioninemia, the immediate cause of which remains uncertain. (6) Fumarylacetoacetate hydrolase (FAH) deficiency (tyrosinemia type I) may lead to hypermethioninemia secondary either to liver damage and/or to accumulation of fumarylacetoacetate, an inhibitor of the high Km MAT. Additional possible genetic causes of hypermethioninemia accompanied by elevations of plasma AdoMet include mitochondrial disorders (the specificity and frequency of which remain to be elucidated). Non-genetic conditions include: (a) Liver disease, which may cause hypermethioninemia, mild, or severe. (b) Low-birth-weight and/or prematurity which may cause transient hypermethioninemia. (c) Ingestion of relatively large amounts of methionine which, even in full-term, normal-birth-weight babies may cause hypermethioninemia.

KEY WORDS: hypermethioninemia; methionine; S-adenosylmethionine; adenosyltransferase; homocysteine; homocystinuria; cystathionine beta-synthase; glycine; methyltransferase; S-adenosylhomocysteine; hydrolase; citrin; tyrosinemia type I; fumarylacetoacetate; deficiency; liver; disease; prematurity; low-birth-weight

INTRODUCTION

This article will discuss the known forms of human hypermethioninemia with emphasis on genetic abnormalities, their metabolic, genetic, diagnostic, and clinical aspects, but, for completeness, including some non-genetically deter- mined conditions. The chief aim is to provide a framework that may assist in the differential diagnosis of hyperme- thioninemia. At present, six genetic conditions leading to abnormal methio- nine elevations are known: methionine adenosyltransferase (MAT) I/III defi- ciency, homocystinuria due to cystathi- onine beta-synthase (CBS) deficiency, deficiencies of glycine N-methyltrans- ferase (GNMT), S-adenosylhomocys- teine hydrolase (AHCY), citrin, and fumarylacetoacetate hydrolase (FAH) (tyrosinemia type I). Several of these have been known for some time and extensively reviewed elsewhere. For them, the reader will be referred to those reviews and emphasis will be put on recent developments, and on the extent and pathophysiology of the methionine elevations. The author apologizes to the many workers whose contributions are covered only inadequately because of the need to keep this presentation to a manageable length. In order to put the conditions to be discussed in perspective, the normal metabolism of methionine is summarized in Figure 1 [Mudd et al., 1964; Baric et al., 2004]. Deficient activities of MAT I/III, GNMT, AHCY, and CBS are discussed in the sequence of the normal metabolic flow: methionine ! AdoMet ! AdoHcy ! homocysteine! cystathionine.

METHIONINE ADENOSYLTRANSFERASE I/III (MAT I/III) DEFICIENCY

Abnormal elevations of methionine due to defects in the conversion of methionine to AdoMet were first discovered in the early 1970s, soon after the introduc- tion of screening of newborn infants for abnormally high blood methionine. Such screening was carried out to identify babies with CBS deficiency, a genetic defect that leads not only to homocystinuria, but also to elevated plasma methionine (see Cystathionine Beta-Synthase (CBS) Deficiency Sec- tion). Infants with methionine eleva- tions, but without abnormalities of homocystine were discovered in France [Gout et al., 1977] and in the United States [Gaull and Tallan, 1974], and MAT activity was found to be low in extracts of biopsied liver [Gaull and Tallan, 1974; Finkelstein et al., 1975a]. During the next decade similar defi- ciencies were proven in liver extracts from six additional patients [Gaull et al., 1981; Hase et al., 1984; Gahl et al., 1987]. During that period it was also established that three major forms of chromatographically separable MAT are present in mammals: MAT I and MAT III, tetrameric and dimeric holoenzymes composed of the same subunit encoded by the gene MAT1A; and MAT II, containing two catalytic subunits encoded by a second gene, MAT2A. Of these isozymes, MAT III has the highest Km for methionine and is most strongly stimulated by dimethylsulfoxide (details reviewed in Kotb and Geller [1993]; consensus nomenclature sug- gested by Kotb et al. [1997]). In post- partum liver MAT I and MAT III are the predominant forms with smaller amounts of MAT II also present. For the eight patients in question with proven deficient hepatic MAT activities chromatographic studies to separate these forms were not carried out. However, the facts that the residual activities were relatively lower when assayed at higher methionine concen- trations and were not stimulated by dimethylsulfoxide [Finkelstein et al., 1975a; Gahl et al., 1987] suggested that the major losses of MAT activity were due to decreases in MAT III. In agree- ment, MAT activity was normal in a variety of non-hepatic tissues which express almost solely MAT II [Tallan and Cohen, 1976; Gaull et al., 1981; Gahl et al., 1988]. The amino acid sequence encoded by MAT1A was soon established [Horikawa et al., 1989; Alvarez et al., 1991; Sakata et al., 1993] and mutations in that gene were identi- fied in each of the six patients studied molecularly among the eight in whom hepatic MATactivity had been shown to be low [Ubagai et al., 1995; Chamberlin et al., 1996, 2000; Hazelwood et al., 1998] so that definitive diagnosis is now possible by molecular analyses without liver biopsy. Because the extent of the loss of MAT I activity relative to that of MAT III is not clearly defined by identification of the underlying muta- tion(s) in MAT1A, it is customary to characterize such patients as ‘‘MAT I/ III-deficient.’’

Metabolite Abnormalities: Occurrence and Cause

As shown in Table I, MAT I/III deficiency is characterized by hyper- methioninemia. The extent of the elevation depends upon the specific mutation(s) causing the block in con- version of methionine to AdoMet. For example, patients with homozygous truncating mutations and no residual MAT I/III activity have plasma methionines of 600 –2,541 mM [Gahl et al., 1987; Chamberlin et al., 1996; Hazel- wood et al., 1998], whereas for 28 heterozygotes with the dominant R264H mutation [who have perhaps 30% of normal activity, Chamberlin et al., 1997] the mean plasma methio- nine was 188 mM (range 45 –400 mM) [Mudd et al., 2001b]. A feature that distinguishes MAT I/III deficiency from GNMT, AHCY, and CBS defi- ciencies is that AdoMet is not elevated when methionine is abnormally high (Table I). This is diagnostically useful: all individuals in whom plasma AdoMet is not elevated in the presence of elevated methionine for whom molecular studies of their MAT1A genes have been carried out have had inactivating mutations in that gene. However, plasma AdoMet is often within, or only slightly below, the reference range. Individuals with trun- cating mutations that leave no residual MAT I/III activity must be forming AdoMet by use of MAT II. In one such individual, homozygous for MAT1A 185X [Hazelwood et al., 1998], the hepatic concentration of AdoMet was slightly low, 18 mmol/kg wet weight (reference range 35– 70 mmol/kg wet weight). Balance studies of that man showed that he was forming AdoMet at a rate of at least 18.4 mmol/day, a rate comparable to that of control young adult males [Mudd and Poole, 1975; Mudd et al., 1980; Gahl et al., 1988]. However, in spite of his excessively high bodily methionine content, he was recycling some 8.8 mmol/day of homo- cysteine back to methionine, an amount far in excess of that for a normal male ingesting excessive methionine. Ado- Met normally acts through two com- plementary regulatory effects to adjust the proportion of homocysteine being either recycled to methionine, or being irreversibly catabolized to cystathionine (see Fig. 1): (a) by inhibiting 5,10- methylenetetrahydrofolate reductase, thereby limiting the formation of methyl-THF, a substrate for homocys- teine remethylation [Kutzbach and Stokstad, 1967]; (2) by stimulating CBS activity, leading to increased conversion of homocysteine to cystathionine [Fin- kelstein et al., 1975b] (Fig. 1). Excessive recycling of homocysteine to methio- nine in the hypermethioninemic patient with the low hepatic AdoMet is thus understandable, and illustrates the major importance of AdoMet regulation in maintenance of bodily methionine bal- ance.

Some patients with hypermethio- ninemia due to proven MAT I/III deficiency, or with metabolic abnormal- ities typical of that deficiency, have objectionable breath odor due to the presence of dimethylsulfide [e.g., see Gahl et al., 1987]. This compound is formed by the methionine transamina- tion pathway in which methionine is converted successively to 4-methylthio- 2-oxobutyrate, 3-methylthiopropio- nate, methanethiol, and dimethylsulfide (Fig. 1). Assays of these metabolites in a series of 30 such patients showed that the pathway becomes prominent only dur- ing, or after, the first year of life, and when a threshold of 300– 350 mM plasma methionine is exceeded [Mudd et al., 1995b].

Patients with severe MAT I/III deficiency tend to have slight elevations of plasma tHcy, up to as high as 59 mM [Lagler et al., 2000; Stabler et al., 2002]. Although the mechanism(s) underlying this effect are not completely established, less than normal stimulation of CBS activity due to low AdoMet is thought to play a major role [Stabler et al., 2002]. This tendency is important because it may lead to erroneous diagnosis of combined methionine and tHcy eleva- tions as CBS deficiency (see Cystathio- nine Beta-Synthase (CBS) Deficiency Section) [Stabler et al., 2002]. Assays of plasma AdoMet and cystathionine may be the most useful means to distinguish between MAT I/III and CBS deficiency (see Table I).

Pathophysiology

MAT activity has been found in all organisms in which it has been sought with the exception of a few endosym- biotic or infectious forms that take up AdoMet from their endosymbionts or hosts [Merali et al., 2000]. Many organ- isms have two or more genes encoding MATs [for an extensive list of examples see Sa´nchez-Pe´rez et al., 2004]. For mammals the two genes, MAT1A and MAT2A, encode proteins with 85% amino acid identity, but the genes are on different chromosomes and are expressed with differing tissue patterns. At the time of writing, at least four humans have been identified as homo- zygotes for MAT1A insertions or dele- tions that lead to truncated subunits (c.539insTG, p.185X [Chamberlin et al., 1996; Hazelwood et al., 1998]; c.1943delTG, p.350X [Chamberlin et al., 1996]; c.827insG, p.351X [Cham- berlin et al., 1996]). These individuals presumably have total absence of MAT I/III activity. The first two are clinically normal, whereas the last two developed demyelination of the brain. Reports of seven further MAT I/III-deficient patients with CNS abnormalities have now been published [Chamberlin et al., 1996, 2000; Ito et al., 2003; Tada et al.,2004; Mahokol et al., 2009; Faghfoury et al., 2009]. As discussed below in the section on ‘‘Adverse effects of extreme hypermethioninemia,’’ very high concentrations of methionine may cause severe edema of the brain. Such edema occurred in a 5-year-old MAT I/ III-deficient boy being treated with betaine because of a mistaken diagnosis of CBS deficiency [Tada et al., 2004], but has not been reported in other MAT I/III-deficient patients not receiving betaine. Indeed, whether or not there is a cause- and effect relationship between MAT I/III deficiency and CNS abnormalities such as mental retardation and myelination abnormal- ities has not been established unequiv- ocally, and such a relationship would be puzzling because in the brain nor- mally MAT2A is expressed, so that presumably at least a normal activity of MAT II might be present in that organ of MAT I/III-deficient individuals. Nevertheless, in the one such indi- vidual in whom cerebrospinal fluid metabolites were assayed, AdoMet was abnormally low in spite of a CSF methionine concentration some 60-fold above the mean control level [Surtees et al., 1991]. Therefore, one possibility is that there is a quantitatively significant transport of AdoMet from the liver where it is normally synthesized by MAT I/III to the brain. Another possibility to connect MAT I/III defi- ciency to CNS damage would be that one or more compounds normally synthesized chiefly in AdoMet-depend- ent reactions in the liver are important to sustain CNS function. Candidates might include phospatidylcholine and its choline containing derivatives. Neither of these working hypotheses was strongly supported by assays of AdoMet and choline in a few MAT I/ III-deficient patients with and without CNS abnormalities [Mudd et al., 2000b]. Very recently Reytor et al. [2009] opened a new possibility when they reported that MAT I/III subunits are found in small amounts in many tissues in addition to liver, including brain. Mutant MAT I/III subunits thus might play a more direct role in CNS pathology.

Molecular Bases

Overall, MAT activity is the result of two successive reactions catalyzed by each single holoenzyme molecule: (1) transfer of the adenosyl moiety of ATP to methionine, forming AdoMet and enzyme-bound tripolyphosphate (PPPi); (2) cleavage of the PPPi to pyrophosphate (PPi) and inorganic phosphate (Pi) through a tripolyphos- phatase (PPPase) activity, followed by release of the PPi and Pi [Mudd, 1962, 1963]. Of the 37 MAT1A mutations reported to the present time in hyper- methioninemic humans, five truncating mutations and a splicing variant are presumed to have no residual activity [Chamberlin et al., 1996, 2000]. For 13 additional mutations, overall MAT activities were assayed after expression in COS-1 cells or E. coli [Ubagai et al., 1995; Chamberlin et al., 1996, 2000; Kim et al., 2002]. Eighteen further mutations were recently expressed in E. coli and assayed for both overall MAT and PPPase activities [Ferna´ndez-Iri- goyen et al., 2010]. Of the latter 18 mutations, 16 had severely reduced overall MAT activity; in 5, both MAT and PPPase activities were impaired; and in 7, MAT activity was low, but PPPase activity was close to normal. That loss of PPPase activity led con- sistently to loss of MAT activity is in accord with the evidence that the PPPi produced by the initial reaction remains bound to the active site of the enzyme, thereby preventing further AdoMet synthesis, until it is removed after hydrolysis or, far more slowly, by dissociation [Mudd, 1973; Markham et al., 2009]. Most MAT1A mutations behave as Mendelian recessives, but MAT1A R264H is exceptional in caus- ing relatively mild hypermethioninemia even in heterozygotes because the subunit containing the aberrant histi- dine associates with a WT subunit, inactivating its MAT activity [Cham- berlin et al., 1997; Pe´rez-Mato et al., 2001].

Newborn Screening

As already mentioned, infants with MAT I/III deficiency are being detected by the screening for elevated blood methionine currently carried out in several countries and states in the USA in an effort to identify infants with CBS deficiency. Experiences in programs for which the cut-off points for methionine elevation have recently been lowered have now been reported from Taiwan and Galicia, Spain: Among the 1,701,591 newborns screened in Taiwan by bacterial inhibition assay between January 1, 1991 and June 30, 2003 with a cut-off point of 134 mM, only a single case of CBS deficiency was found, whereas eight individuals with homo- zygous, compound heterozygous, or apparently heterozygous MAT1A muta- tions were identified. Of the latter, three were heterozygotes for R264H [Chien et al., 2005], the mild Mendelian dominant [Chamberlin et al., 1997; Pe´rez-Mato et al., 2001]. In 2003 MS/ MS replaced bacterial inhibition assay, and between January 1, 2004 and June 30, 2007 among 323,935 babies screened with cut-off points initially of 45 mM, then 54 mM, no further cases of CBS deficiency were found, but 11 more with MAT1A mutations were identified, of whom 5 were R264H heterozygotes [Chien et al., 2008]. Similarly, in Galicia, Spain, after an expanded screening program was intro- duced in June 2000 using MS/MS with cut-off values for methionine between 48 and 56 mM, among 140,818 new- borns one case of CBS deficiency was identified and five heterozygotes for MAT1A R264H were discovered [Couce et al., 2008]. If these experiences are representative of those that may occur in other regions, it appears that MAT I/III deficiency will predominate among the patients identified by screen- ing for methionine with lowered cut-off points, and that R264H heterozygosity will be relatively especially numerous. R264H heterozygosity is clinically benign, as shown by the facts that MAT activities in liver extracts of three R264H heterozygous siblings were nor- mal (although the Km value for methionine was low, indicating a lack of the normal high Km form, MAT III) [Nagao and Oyanagi, 1997]; and that R264H heterozygous parents and grandparents have had no health problems attributable to this genetic abnormality [Blom et al., 1992; Chamberlin et al., 1997; Nagao and Oyanagi, 1997; Couce et al., 2008]. Further recent findings indicate that heterozygosity for additional MAT1A mutations may also cause mild hyper- methioninemia [Ferna´ndez-Irigoyen et al., 2010; S.H. Mudd, F. Corrales, Y.-H. Chien, unpublished observa- tions]. Taken together these facts raise the question of whether or not detection of R264H heterozygotes and other heterozygosities with similar metabolic effects is a useful outcome of newborn screening.

Management

The optimal management of MAT I/III deficiency remains to be defined. Many of the known MAT I/III-deficient patients are clinically unaffected, even those with mutations other than R264H. However, having been detected chiefly by newborn screening, they are usually relatively young so their long-term prognoses are unknown. As already mentioned, at least nine have manifested CNS problems, usually those with the most severe MAT I/III defi- ciencies as judged by plasma methionine levels. However, expression and assay of the activities of numerous MAT1A mutations has not shown a pattern that permits prediction of which patents might develop CNS problems [Ferna´n- dez-Irigoyen et al., 2010]. Mice with MAT1A knocked out are predisposed to liver injury [Lu et al., 2001] and may develop hepatocellular carcinoma by age 18 months [Martinez-Chantar et al., 2002]; yet, to date, liver carcinoma has not been found in MAT I/III-deficient patients. Taken together these results provide no definitive criteria to deter- mine which, if any, of these patients should be treated. Dietary methionine restriction has lowered the plasma methionine in some cases [Nagao and Oyanagi, 1997; Stabler et al., 2002;Chien et al., 2005], but not markedly, or with a delay in others [Hase et al., 1984; Gahl et al., 1988]. However, the possibility that the adverse clinical effects of MAT I/III deficiency are due chiefly not to the elevated methionine, but rather to AdoMet depletion (see dis- cussion in Mudd et al. [1995b]) raises serious doubt if methionine restriction is advisable. If there is residual MAT I/III activity, lowering methionine may fur- ther decrease the synthesis of AdoMet. For patients with no MAT I/III activity who depend upon MAT II for AdoMet synthesis, lowering the methionine enough to jeopardize flux through MAT II (Km’s of various tissue forms of MAT II range from 2.2 to 23 mM [Kotb and Geller, 1993]) would certainly be dangerous. An alternative treatment might be supplementation with Ado- Met. During administration of AdoMet, myelination abnormalities seen on MRI of the brain mitigated in one case [Surtees et al., 1991], a girl homozygous for MAT1A 351X [Chamberlin et al., 1996]. Further experience with AdoMet administration is needed to determine when, and for whom, such treatment is indicated.

GLYCINE

N-METHYLTRANSFERASE (GNMT) DEFICIENCY

GNMT deficiency was discovered in humans only in 2001, and, at the time of writing, only three such patients have been identified [Mudd et al., 2001a; Augoustides-Savvopoulou et al., 2003]. The initial patient, an Italian girl, was investigated at age 2.3 years because of a history of elevated serum transaminases since age 1 year and moderate hepato- megaly. She was found to have abnor- mally high plasma methionine, with normal tHcy and tyrosine. However, on a normal diet plasma AdoMet was extremely elevated to 1,149 nM (refer- ence range 93 16 nM), whereas AdoHcy was, at most, very slightly elevated to 56 nM (reference range 15– 45 nM) (Table I). Thus, deficient activity of either MAT I/III or CBS was ruled out by, respectively, the elevated AdoMet (as discussed above), or the normal tHcy (see Cystathionine Beta- Synthase (CBS) Deficiency Section). Retrospectively, AHCY deficiency is also ruled out by the close to normal AdoHcy levels (although at the time AHCY deficiency was not known). Assays of methionine in several patients with comparable or greater elevations of serum transaminases showed that the mild hepatic abnormalities in the patient were unlikely to cause her hyperme- thioninemia, and further studies pro- duced evidence against a defect of methionine transamination or other hypothesized abnormalities that might affect methionine levels. Importantly, it was found that the plasma sarcosine (N- methylglycine) concentration was not elevated, as several lines of evidence showed would be expected when Ado- Met was very high and the activity of GNMT normal (Table I) [Mudd et al., 2001a]. Because GNMT activity could not be detected in readily available tissues from normal control subjects (peripheral white blood cells, stimulated lymphocytes, or cultured skin fibro- blasts), direct sequencing of the GNMT gene was carried out, and revealed that the patient was a compound heterozy- gote for two mutations in that gene not attributable to common polymor- phisms, Leu49Pro and His176ASn [Luka et al., 2002]. The older brother of the patient, clinically unremarkable at age 9 years except for moderate hep- atomegaly and slightly elevated serum transaminases was found also to have elevated plasma methionine and other metabolic findings similar to those in the initial patient and to be a compound heterozygote for the same two GNMT mutations [Mudd et al., 2001a]. The third known patient, a Greek boy, had a similar history of elevated serum trans- aminases, initially found at age 2 years during a routine biochemical and hem- atological workup. The transaminase abnormalities persisted and at age 4 years he was found to be hypermethionine- mic, to have a metabolic pattern very similar to that of the first two patients (Table I) and to be homozygous for a novel GNMT mutation, Asn140Ser [Augoustides-Savvopoulou et al., 2003].

Metabolite Abnormalities: Occurrence and Cause

As discussed above, GNMT deficiency is characterized metabolically by eleva- tions of methionine and AdoMet, with AdoHcy and tHcy usually normal, but with sarcosine (N-methylglycine) not being elevated, as it may be in other conditions with grossly elevated Ado- Met (Table I). The elevation of AdoMet and the failure of sarcosine to be elevated are readily explained because GNMT normally serves to dispose of excess AdoMet by transferring a methyl group to glycine to form the N-methyl deriv- ative. GNMT has several properties that, together, suit it well to serve this function, including: (a) The substrate, glycine, is readily available and non- essential, and the product, sarcosine is rapidly reconverted to glycine and a one- carbon folate-bound unit; (b) GNMT activity increases in livers of rats fed high methionine diets; (c) The catalytic rate is positively cooperative as a function of AdoMet concentration; (d) GNMT activity is quite resistant to the inhibition by AdoHcy that affects most AdoMet- dependent methytrasnferases; and (e) GNMT activity is inhibited by 5- methyltetrahydrofolate. Thus, when AdoMet is high and inhibits methyle- netetrahydrofolate reductase, the level of 5-methyltetrahydrolfolate will decrease, relieving inhibition of GNMT activity (for detailed citations on these points see Mudd et al. [2001a] and Luka et al. [2009]). The mechanism of abnormal elevation of methionine in GNMT deficiency is not completely clear. Inhibition of MAT I and MAT II by AdoMet [Kotb and Geller, 1993] almost certainly plays a role, but other feedback regulatory phenomena may also con- tribute [Mudd et al., 2001a].

GNMT knock-out mice have been generated by two groups by replacement of either the first exon with part of the promoter region [Luka et al., 2006] or exons 1– 4 and part of exon 5 [Liu et al., 2007]. With both models the metabolic changes mimic those seen in humans with GNMT deficiency: methionine increases several-fold in both liver [Luka et al., 2006] and serum [Liu et al., 2007;Martinez-Chantar et al., 2008]; AdoMet increases in liver from 37 to 48 nmol/g to 1,334–3,453 nmol/g [Luka et al., 2006; Liu et al., 2007] and in serum by about twofold [Martinez-Chantar et al., 2008], whereas AdoHcy in liver decreases slightly [Luka et al., 2006] or does not change [Liu et al., 2007]; and tHcy remains the same [Liu et al., 2007]. Together, this pattern of abnormalities in both humans and mice show that, when GNMTactivity is severely deficient, the many other mammalian methyltransfer- ases that use AdoMet (as many as 60 [Brosnan and Brosnan, 2006] or more [Katz et al., 2003]) do not, together, suffice to dispose of the AdoMet that forms on a normal diet.

Pathophysiology
At last report at the time of writing, the three known humans with GNMT deficiencies were clinically well at ages 141 and 191 years (R. Cerone, personal
communication) or 12 years (P. Augoustides-Savvopoulou, personal communication), although each has had persistent mild to moderate eleva- tions of serum transaminases [Mudd et al., 2001a; Augoustides-Savvopoulou et al., 2003]. The GNMT knock-out mice, however, have not only elevated serum transaminases [Liu et al., 2007; Martinez-Chantar et al., 2008; Varela- Rey et al., 2010], but also develop steatosis and liver fibrosis [Martinez- Chantar et al., 2008; Varela-Rey et al., 2010] or store excessive glyco- gen [Liu et al., 2007], and, most seriously, hepatocellular carcinomas [Martinez-Chantar et al., 2008; Liao et al., 2009].

Molecular Bases

The three mutations known to occur in GNMT-deficient humans have been expressed in E coli and characterized [Luka and Wagner, 2003]. Each leads to decreased activity: Asn140Ser to 0.5% that of wild-type; Leu49Pro to 10%; and His176Asn to 75%. Possible reasons for these changes in activity were discussed [Luka and Wagner, 2003].

Newborn Screening and Management

Given the relative infrequency of iden- tified cases of GNMT deficiency and the fact that, to date, they are not seriously adversely clinically affected, it appears there is no pressing need for newborn screening for this condition. It remains unknown whether such patients have methionine elevations within a few days of birth. If further experience shows they do not, assay of plasma AdoMet might identify such patients, but would probably be impractical because plasma would be needed and because of the relative instability of AdoMet.

Therapeutic interventions have not been tried, but a promising approach is suggested by the recent finding that development of fatty liver and hepatic fibrosis in GNMT knock-out mice is prevented by administration of nicoti- namide, a substrate for an alternative AdoMet-dependent methyltransferase. Not only did this intervention restore hepatic AdoMet to the control level, but also DNA hypermethylation was prevented, and expression of a variety of genes involved in fatty acid meta- bolism, oxidative stress, inflammation, cell proliferation, and apoptosis were normalized [Varela-Rey et al., 2010].

S-ADENOSYLHOMOCYSTE- INE HYDROLASE (AHCY) DEFICIENCY

AHCY deficiency is the most recently discovered genetic cause of human hypermethioninemia. The initial case, a Croatian boy, was investigated because of slow psychomotor development and severe muscular hypotonia. He was found to have elevated plasma trans- aminases and creatine kinase (chiefly the muscle isoenzyme), low plasma albu- min, and prolonged prothrombin time. At age 12.7 months MRI of the brain showed a pattern of myelination corre- sponding to that normally found at age 2– 3 months. Plasma methionine was elevated to 477– 784, but tHcy, cysta- thionine, and sarcosine were normal or only very slightly elevated (Table I). Tyrosine was normal. Assay of plasma AdoMet and AdoHcy revealed that AdoHcy was almost 150-fold elevated to 5,044 nm and that AdoMet was some 30-fold above normal at 2,971 nM, immediately suggesting a deficiency of AHCY activity, the enzyme that nor- mally catabolizes AdoHcy. Because AdoHcy is a strong inhibitor of most AdoMet-dependent methyltransferases, accumulation of AdoHcy would be expected to lead to accumulation of AdoMet. AHCYactivity was 5– 10% of control activities in extracts of red blood cells and cultured skin fibroblasts, and only about 3% in liver extracts, and the patient was found to be a compound heterozygote for two AHCY mutations, a nonsense codon that introduces a stop at amino acid 112, Trp112X, and a point mutation, Tyr143Cys [Baric et al., 2004]. Subsequently five further AHCY-deficient patients have been ascertained: (a) Two younger siblings of the initial patient have been reported to have the same pattern of metabolite abnormalities and to be compound heterozygotes for the same two AHCY mutations [Baric et al., 2005; Cuk et al., 2007]. (b) A man who had been found on newborn screening to have mildly elevated blood methionine (200 mM) that persisted at levels of 135– 900 mM, in whom MATactivity in a liver extract was normal, and who also had general- ized muscle weakness and elevated serum creatine kinase levels, was restudied at age 26 years after the reports of the initial AHCY-deficient patients were published. He was found to have markedly elevated plasma AdoHcy and AdoMet and to be a compound hetero- zygote for Tyr149Cys and a novel mutation, Ala89Val, in the AHCY gene [Buist et al., 2006]. (c) A baby born in Texas with fetal hydrops and severe hypotonia/myopathy and found to have elevated plasma methionine accompa- nied by elevations of plasma AdoHcy and AdoMet was revealed to be a compound heterozygote for two novel AHCY mutations, Arg49Cys and Asp86Gly. Retrospective examination of the newborn screening record and stored blood spots from her older sibling (who had also had fetal hydrops but died at age 25 days, before the reports of AHCY deficiency became available) showed that that girl, too, had had newborn hypermethioninemia and car- ried the same two AHCY mutations [Grubbs et al., 2010].

Metabolite Abnormalities: Occurrence and Cause

As stated above, the elevations of AdoHcy and AdoMet (Table I) are to be expected because AHCY is the only enzyme in mammals that catabolizes AdoHcy, and because of the generalized inhibition by AdoHcy of the enzymes utilizing AdoMet as a methyl donor. The accumulation of methionine is similar to that occurring in GNMT deficiency when AdoMet, but not AdoHcy, is elevated (see above section), and it is likely that in each case feedback regu- lation by AdoMet is responsible.

Pathophysiology

All known patients with AHCY-defi- ciency have had more-or-less severe myopathy and elevations of creatine kinase. Because the quantitatively pre- dominant uses of AdoMet are for the synthesis of phosphatidylcholine by phosphatidylethanolamine methyltrans- ferase and creatine by guanidinoacetate methyltransferase [Mudd et al., 2007], it is likely that inhibition of the production of these products contributes to the myopathy, or, at least, to hypotonia. Delayed mental development and low or marginally low levels of albumin have been frequent in untreated patients [Baric et al., 2004; Buist et al., 2006; Grubbs et al., 2010], as have been prolonged prothrombin time and retarded brain myelination [Baric et al., 2004; Grubbs et al., 2010]. The brains of the two young, most severely affected patients, the sibs born with hydrops fetalis, on MRI, and on post-mortem examination of the younger sister, showed mild dilatation of the lateral ventricles, markedly diminished myeli- nation, and hypoplasia of the ventral pons, corpus callosum, and mega cis- terna magna [Grubbs et al., 2010]. No model mouse with only AHCY knocked out has been reported, but embryonic death occurs in a mouse with a deletion including the AHCY gene, supporting the conclusion that severe AHCY defi- ciency has damaging effects in utero.

Molecular Bases

The four point mutations found to date in AHCY-deficient humans have each been expressed in E. coli and their kinetic and structural effects thoroughly inves- tigated and discussed by Vugrek and his colleagues. Tyr143Cys reduces enzy- matic activity by 65– 75%, makes the enzyme thermosensitive, and effects the oxidation state of NAD, the enzyme- bound cofactor [Beluzic et al., 2006]. Ala89Val leads to ≥70% loss of enzy- matic activity and changes in electro- phoretic mobility suggestive of changes in the overall charge of the mutant tetrameric complex [Beluzic et al., 2008]. Arg49Cys protein looses 93% of its activity, and forms intermolecular disulfide bonds leading to inactive aggregates. The loss of the negative charge in Asp86Gly leads to 84% loss of activity and to formation of enzymati- cally inactive macromolecular structures [Vugrek et al., 2009]. That the five identified AHCY mutations each behave as Mendelian recessives is shown by the fact that the six parents of the known patients, each a heterozygote for one of the mutations in question, have had normal plasma levels of AdoHcy, AdoMet, and methionine [Baric et al., 2004; Buist et al., 2006; Grubbs et al., 2010].

Newborn Screening

Screening of newborns for AHCY- deficiency has not been reported. If an attempt to do so is considered, assay of plasma AdoHcy might be useful. A simple and rapid immunoassay for that purpose is available [Capdevila et al., 2007]. An alternative screening ap- proach that might broaden the clinical spectrum of AHCY-deficiency might focus on infants with unexplained muscle weakness.

Management

For the initial patient and his two affected sibs, management has been aimed chiefly at reducing the inhibitory effects of AdoHcy by lowering the concentration of that compound through limitation of dietary methio- nine. To overcome any depletion of phosphatidylcholine and/or creatine that may be adversely affecting muscle function, supplements of these two compounds have been used. Although plasma AdoHcy levels have decreased, they have not normalized during methionine restriction severe enough to lower plasma methionine levels to 16 18 or 7 2 mM [Baric et al., 2005]. However, on the whole, the results have been somewhat encouraging: treatment of the initial patient, starting at age 13 months, was accompanied by gains in strength and striking improvement in brain myelination, but creatine kinase remained elevated, albumin mostly low normal, and prothrombin time short- ened but was still often above normal. The younger sibs of this boy were started on treatment soon after birth and have been clinically far less affected [Baric et al., 2005; Cuk et al., 2007]. On the other hand, similar management of the younger girl born with fetal hydrops did not lead to dramatic improvement: even the most stringent dietary methionine restriction attempted (15 mg/kg/day) did not lower her plasma AdoHcy to <26 times the mean reference value. She continued to deteriorate and died on day-of-life 122.

CYSTATHIONINE

BETA-SYNTHASE (CBS) DEFICIENCY

Homocystinuria due to CBS deficiency was the first genetic condition shown to cause hypermethioninemia. Because many aspects of this condition have been reviewed extensively [see, e.g., Kraus and Kozich, 2001; Mudd et al., 2001b; Andria et al., 2005; Jhee and Kruger, 2005], the present article will cover the major features relatively briefly with emphasis devoted to more recent devel- opments.

In 1962 excessive levels of urinary homocystine (i.e., homocystinuria) were found among mentally retarded children in Northern Ireland and Wis-consin [Field et al., 1962; Gerritsen et al., 1962] and in another with dislocated optic lenses seen at the Wills Eye Hospital in Philadelphia [Spaeth and Barber, 1965]. Elevations of plasma methionine were present in the same individuals [Gerritsen et al., 1962; Carson et al., 1963; Gerritsen and Waisman, 1964a; Spaeth and Barber, 1965] (Table I). Assays of MATand CBS activities in extracts of a liver biopsy soon showed that the underlying genetic abnormality was deficiency of CBS [Mudd et al., 1964]. Extensive experi- ence with CBS-deficient patients has shown that the most frequent clinical abnormalities are dislocation of the optic lenses, mental retardation, early throm- boembolic events, and skeletal abnor- malities including osteoporosis, genu valgum, and thinning and lengthening of the long bones [reviewed in Mudd et al., 2001b]. The metabolic abnormal- ities of some patients are B6-responsive (i.e., both their tHcy and methionine levels decrease markedly during admin- istration of pharmacological doses of vitamin B6, a precursor of pyridoxal 5'- phosphate, the cofactor for CBS), whereas others do not so respond (i.e., are B6-non-responsive) [Barber and Spaeth, 1969]. By 1985 it was possible to gather information on 629 patients (including 231 classified as B6-respon- sive and 231 as B6-non-responsive) [Mudd et al., 1985]. In untreated patients the clinical effects were usually more severe, or occurred at younger ages in B6-non-responders than in B6-res- ponders: Dislocation of optic lenses occurred in 50% of B6-non-responders by age 6 years; in 50% of B6-responders by age 10 years. The median IQ for B6- non-responders was 56 and only 4% of them had IQ’s of 90 or above; for B6- responders the median IQ was 78% and 22% had IQ’s of 90 or above. Throm- boembolic events usually happened after a lag of 8–12 years and, for the entire group, had occurred in about 25% by age 16 years, and 50% by age 29 years. There was a significant, but not marked, differ- ence between the time-to-event curves for B6-non-responders and B6-res- ponders. However, as discussed below, the data for the B6-responders may not have included patients who either have none of the adverse effects of CBS- deficiency or who suffer only throm- boembolic episodes in their 20s or later—such patients were unlikely to have been diagnosed at that time as CBS- deficient.

Metabolite Abnormalities: Occurrence and Cause

The hypermethioninemia of CBS defi- ciency is readily explained as a result of excessive remethylation of homocys- teine, the primary metabolite abnor- mally accumulated. Both methyl-THF- and betaine-dependent reactions may play a role in this remethylation. A few untreated patients were not hyperme- thioninemic (perhaps 1% of B6-non- responders and 10% of B6-responders) [Mudd et al., 1985].

In normal human plasma total homocysteine (tHcy) (reference range usually 5–14 mM) is found chiefly in four forms: 1– 2% as the thiol, homo- cysteine itself; 82–83% combined in disulfide linkage with cysteines of pro- teins [chiefly albumin Sengupta et al., 2001]; and the remaining 15–16% as the free disulfides, homocystine, and cys- teine-homocysteine disulfide [Mansoor et al., 1992; Mudd et al., 2000a]. In CBS-deficient individuals with elevated tHcy the percentage of the thiol, homocysteine, rises, reaching 10– 25% as tHcy reaches 150– 400 mM [Mansoor et al., 1993].

As shown in Table I, both AdoHcy and AdoMet may be extremely elevated in untreated CBS-deficient patients. The elevation of AdoHcy is due to the fact that AHCYactivity is reversible and the equilibrium favors AdoHcy forma- tion. Therefore, when tHcy is high, AdoHcy also tends to be elevated. Two factors contribute to the AdoMet ele- vation: inhibition by AdoHcy of meth- yltransferases using AdoMet; and the tendency of AdoMet to rise when methionine is high and MAT I/III activity is normal.

Additional metabolic abnormalities in CBS-deficiency include low levels of cystathionine in plasma (Table I) and in brain [Gerritsen and Waisman, 1964b].
The latter organ in humans normally contains a high amount of cystathionine [Tallan et al., 1958] due to the presence of ample CBS activity [Finkelstein, 1990], and very low [Finkelstein et al., 1980; Ishii et al., 2004], but not completely absent [Vitvitsky et al., 2006] activity of cystathionine gamma- lyase. Finally, total plasma cysteine also tends to be low in CBS deficiency [Wiley et al., 1989; Mansoor et al., 1993]. A potentially useful means of diagnosis of B6-non-responsive CBS- deficiency has recently been reported [Krijt et al., 2010]. These workers developed a method to assay CBS activity in 20 ml of human plasma using the stable isotope substrate, 2,3,3-2H serine. Median activity in control plasma samples was 404 nmol/h/L (range 66– 1,066, n ¼ 57); in B6-non-responsive CBS-deficient patients, 0 nmol/h/L (range 0– 9; n ¼ 26); in B6-responsive
patients, 16 nmol/h/L (range 0–358; n ¼ 28). The later range overlapped that of the control subjects.

Pathophysiology

Among the major clinical abnormalities found in CBS deficiency, the cause of the dislocated optic lenses may be the most clearly defined. This dislocation is a result of degenerative changes in the zonular fibers that hold the lenses suspended from the ciliary body [Streeten, 1994]. These fibers are com- posed chiefly of microfibrils of fibrillin, a protein containing 47 epidermal growth factor-like (EGF) domains, 43 of which are the calcium-binding (cbEGF) type, each of which has six cysteine residues. Fibrillin therefore is exceptionally rich in half-cystine residues with the total zonular protein containing from 38 to 83 such half-cystines per 1,000 amino acid residues [Mudd et al., 2001b]. Marfan syndrome, another genetic disorder in which dislocation of the optic lenses is prevalent, is due to mutations in the gene encoding fibrillin-1 which cause dis- ruption of the zonular fibers [Whiteman et al., 2006; Robinson et al., 2006], and Hubmacher et al. [2005, 2010] recently demonstrated that treatment of fibrillin- with homocysteine leads to the formation of disulfide bonds and makes the protein more susceptible to proteo- lytic degradation. The abnormalities of bone shape and size in CBS deficiency closely resemble those of Marfan syn- drome. Reduction of disulfide bonds within fibrillin-1 cbEGF domains by elevated homocysteine, resulting in the loss of native structure and protein misfolding, may play a role in such bony changes [Whiteman et al., 2006]. These effects of homocysteine are not unique to fibrillin-1, and it has been suggested that other cbEGF-containing proteins may be implicated in the pathogenic mechanisms of CBS deficiency [Hutch- inson et al., 2005].

Osteoporosis is common in CBS deficiency [Mudd et al., 2001b], but may or may not be sporadically present in Marfan’s Syndrome [Giampietro et al., 2007]. As originally proposed by Victor McKusick, this abnormality may be the result of interference by homocysteine with the synthesis of collagen cross-links [McKusick, 1966]. In agreement, Lubec et al. [1996] found decreased amounts of collagen type I cross-links in CBS- deficient patients.

Less well established is the patho- physiology of the tendency to undergo early vascular events, a tendency not seen in Marfan syndrome. Mechanisms involving platelets, effects on endothelial or non-endothelial vascular cells, coag- ulation disorders, or binding to lip- oproteins have been put forth and discussed [Mudd et al., 2001b; Lentz, 2005], as have been oxidative stress [Wilcken et al., 2000; Jacobsen, 2000; Lentz, 2005], impairment of endothe- lium-dependent arterial relaxation by elevated dimethylarginine inhibition of nitric oxide synthesis [Lentz, 2005], and possible homocysteine targeting of a variety of proteins by formation of disulfide bonds [Glushchenko and Jacobsen, 2007]. Murine models of hyperhomocysteinemia are now avail- able to provide further insights into vascular effects [Dayal and Lentz, 2008; Zhang et al., 2009].

The several postulated mechanisms of neurological abnormalities in CBS deficiency have been discussed [Mudd et al., 2001b]. More recently homocysteine activation of the N- methyl-D-aspartate (NMDA) receptor leading to death of neurons has received much experimental attention [reviewed in Poddar and Paul, 2009]. Hydrogen sulfide is coming to be recognized as a neuromodulator and neuroprotectant [Kimura, 2010], and it had been thought that, in brain, CBS activity is the major producer of hydrogen sulfide [Kamoun, 2004], seeming to raise the possibility that CBS deficiency could by that means lead to CNS abnormalities. However, it has recently been reported that brain contains 3-mercapto-sulfurtransferase (3MST), an alternative catalyst of hydro- gen sulfide formation. CBS is found chiefly in astrocytes, 3MST in neurons, and their relative contributions to hydrogen sulfide production remain to be established [Kimura, 2010], so the possibility of CNS problems secondary to lack of this gas seems still to be open. An additional mechanism for homocysteine toxicity has been eluci- dated by the extensive studies of Jaku- bowski and his colleagues who have shown that: (a) Homocysteine thiolac- tone, an intramolecular thioester of homocysteine,is synthesized by methionyl-tRNA synthetase. This error– editing step prevents incorpora- tion of homocysteine into proteins. The rate of formation of the thiolactone increases as the level of methionine rises. The thiolactone may N-homocysteiny- late protein lysine residues [Jakubowski, 2000]. (b) A homocysteine thiolactonase activity, paraoxonase I, hydrolizes the thiolactone, thereby detoxifying it [Jakubowski et al., 2001; Jakubowski, 2010]. (c) CBS mutations increase N- homocysteinylated protein levels in humans [Jakubowski et al., 2008]. (d) There are possible pathological conse- quences to N-homocysteinylation of proteins, including immune responses to the damaged proteins, cellular tox- icity, and an increase in the frequency of strokes [Jakubowski, 1999, 2006].

Despite these many possible patho- genetic effects, there are grounds for supposing that the pathophysiology of CBS deficiency may be even more complicated. Watanabe et al. produced mice with CBS knocked out (i.e.,genotype Cbs—/—). The mice usually die within 5 weeks of birth and have severe liver damage [Watanabe et al., 1995]. To test the effects of mutant forms of CBS on the knock-out phenotype, Wang and coworkers produced mice that expressed CBS solely as human I278T under the control of a zinc- regulated metallothionine-driven pro- moter (genotype Tg-278), then used these mice to engineer the production of mice with transgenic knock-out geno- type Tg-278 Cbs—/—. Surprisingly, even after weaning from zinc administration, the mutant I278T transgene was able to entirely prevent the neonatal mortaliy of Cbs—/— animals, although the mean serum tHcy of the Tg-278 Cbs—/— animals was 250 mM. They did develop facial alopecia, moderate liver steatosis, and were slightly smaller than hetero- zygous littermates [Wang et al., 2005], and had, in addition, osteoporosis, endoplasmic reticulum stress in liver and kidney, and a 20% reduction in survival time [Gupta et al., 2009]. When such mice were compared to transgenic mice that differed in carrying the normal human form of CBS (genotype Tg- hCBSCbs—/—) [Wang et al., 2004], it was found that even after discontinuation of zinc supplementation the Tg-hCBS Cbs—/— mice had none of the adverse effects found in the Tg-278 Cbs—/— mice and had a moderately lower mean tHcy level of 169 25 mM compared to the 296 28 mM of the Tg-278 Cbs—/— mice (due chiefly to a lowering of the free homocystine in the serum) [Gupta et al., 2009]. Taken together, these results suggest threshold effects for homocysteine toxicity [Gupta et al., 2009].
Maclean et al. [2010] have pointed out that results obtained with knock-out mice for CBS deficiency may not always reflect the condition in humans. As one example, human homozygotes for CBS I278T respond to B6 administration by lowering plasma tHcy, whereas mice expressing solely I278T as a source of CBS do not [Chen et al., 2006]. However with respect to human I278T homozygosity, an increasing number of CBS-deficient individuals have been ascertained chiefly because of thromboembolic events occurring no sooner than the third decade of life. When untreated, these persons have levels of tHcy from 105 to 446 mM, are not mentally retarded, usually do not have skeletal abnormalities, and often do not have dislocated optic lenses [Gaustadnes et al., 2000a,b; Maclean et al., 2002; Sokolova et al., 2001; Linnebank et al., 2001, 2003; Skovby et al., 2010; Magner et al., 2010]. Homozygosity for CBS I278T, a mutation that is B6-repsonsive in humans and is prevalent in northern Europe and the Czech Republic, has been the major CBS mutation among these patients. In Denmark, based on the frequency of heterozygosity for CBS I278T determined by molecular screen- ing of newborns, some 270 I278T homozygotes are predicted to be present, yet only 10 have been identified in the only two genetic centers that make such molecular diagnoses. It was sug- gested that the majority of such homo- zygotes may be clinically normal, and that a similar situation might exist for CBS R369C homozygosity [Skovby et al., 2010]. Considered together with the favorable outcomes of tHcy-low- ering regimens in CBS-deficient patients, even though tHcy levels are not reduced to normal (see Management Section under Cystathionine Beta-Syn- thase (CBS) Deficiency Section), these lines of evidence suggest the pathophysi- ology of CBS deficiency may be subject to threshold effects [Gupta et al., 2009] and may also be influenced by factors in addition to the tHcy level.

Molecular Bases

CBS in mammals is a pyridoxal 5'- phosphate-dependent tetramer com- posed of subunits, each 63 kDa in size [Skovby et al., 1984], 551 amino acids long, and organized in a three-domain structure [Taoka et al., 1999]. The middle domain contains the catalytic core that binds pyridoxal 5'-phosphate and the substrates. The N-terminal domain binds heme [Meier et al., 2001], a trait not found in non-verte- brate forms of CBS [Jhee and Kruger, 2005], nor necessary for enzyme func- tion [Taoka et al., 1998; Oliveriusova et al., 2002], but thought to play a critical role in proper CBS folding and assembly [Janosik et al., 2001; Majtan et al., 2009]. The C-terminal domain binds AdoMet which has long been known to stimulate CBS activity [Finkelstein et al., 1975b]. The enhanced rate is due to an increase in the turnover rate [Kery et al., 1998; Taoka et al., 1998]. This region includes two ‘‘CBS domains,’’ and its deletion produces an activated enzyme that is no longer stimulated by AdoMet [Kery et al., 1998; Oliveriusova et al., 2002; Zou and Banerjee, 2003]. These results, together, suggest that the C-terminal domain inhibits, and that AdoMet bind- ing reduces this inhibition [Janosik et al., 2001; Shan et al., 2001; Evande et al., 2002].

More than 150 human mutations in CBS are now known (http://cbs.lf1.cu- ni.cz/index.php) [CBS mutation data- base]. Many are private or have been found only rarely, but several are more common and/or very regional. Exam- ples of the latter include T191M in the Iberian Peninsula [Urreizti et al., 2003]; R369C in Norway and the Czech Republic, with predicted birth frequen- cies of homozygosity of 1:16,000 in Norway [Refsum et al., 2004] and 1:40,000 in the Czech Republic [Janosik et al., 2009]; G307S, the mutation that prevails among the Irish [Gallagher et al., 1995]: R336C, the mutation found in homozygous form in as many as 1:1,800 babies in the inbred population of Qatari descent [Gan-Schreier et al., 2010]; and, as mentioned above, I278T, especially common in northern Europe, but also the prevalent mutation in both the United Kingdom and the United States [Moat et al., 2004]. Analysis of the haplotypes bearing I278T revealed that this mutation has probably originated repeatedly and independently in the past, quite likely by recurrent gene conversion [Vyletal et al., 2007]. Anal- yses of the structural and other effects of a variety of CBS mutations have been made, based either on the crystal struc- ture of a truncated form of CBS [Meier et al., 2001; Taoka et al., 2002; Meier et al., 2003] or on structural models of the full-length enzyme [Yamanishi et al., 2007]. Very recently extensive studies of the properties of 27 mutants which together comprise 70% of the known alleles in CBS-deficient patients were reported. Mutants formed on average only 12% of tetramers, and their median activity was only 3%, of wild-type. About half were not stimulated by AdoMet. Buried mutations formed less tetramers than solvent-exposed mutants, and had far lower activity. Taken together these studies show that mis- folding plays an important role in the pathogenesis of CBS mutations [Kozich et al., 2010].

What determines the responsive- ness or non-responsiveness of a CBS- deficient individual to B6 administra- tion? It should be noted that this judg- ment is often made on the basis of whether or not B6 administration leads to substantial lowering of homocysteine and its derivatives without defining a level below which such metabolites must drop before a patient is classified as responsive. If such a level is defined, its value may determine whether certain patients are classified as responsive or noresponsive. For example, using a cut- off value of 20 mM for tHcy, Kluijtmans et al. classified 3 of 10 I278T homo- zygotes as non-responsive, although their tHcy values on treatment were only marginally higher than 20 mM (28– 32 mM), and completely normalized upon extended treatment. These work- ers concluded that ‘‘all homozygotes for I278T responded very successfully to homocysteine-lowering treatment’’ [Kluijtmans et al., 1999]. In spite of such minor uncertainties, at least two conclusions appear to be well justified: (a) Responsiveness requires the presence of at least a small amount of CBS activity. In early assays, using a sensitive assay of CBS activity in extracts of cultured skin fibroblasts from CBS-deficient patients, activities were detected in cells from 24 of 25 B6-responsive patients, but in cells of only 1 of 10 B6-non-responders. The activities in the responders, measured without added pyridoxal-phosphate, ranged from 0.1% to 10% of the mean control value. Similarly, Krijt et al. [2011] recently found that the median CBS activity in plasma of B6-responders was 16 nmol/h/L, whereas the median for non-responders was 0 nmol/h/L, with the control median 404 nmol/h/L. In agreement, in the current [Sep- tember 1, 2010] CBS mutation database (http://cbs.lf1.cuni.cz/index.php) each of the eight individuals listed who are homozygotes or compound heterozy- gotes for deletions and/or truncating CBS mutations are B6-non-responsive.(b) The particular CBS mutation(s) present in an individual determine his/ her responsiveness. A listing of the B6- responsiveness of a variety of CBS genotypes was presented in 1999 [Kraus et al., 1999]. To update and extend this listing the very extensive list of alleles found in the current CBS database (see above) was examined. The least equiv- ocal indications are presumably to be found among individuals who are both homozygous for their CBS mutations and have been classified with respect to their responses to B6. Table II lists the results for such individuals (left col- umns), stratified so as to show at the top mutations for which the evidence is at least suggestive of B6-responsiveness, followed by mutations that may be B6- non-responsive, then mutations for which the available evidence provides less certain indications. Clearly I278T is B6-responsive with 17 homozygotes responsive. The two classified as only partially responsive are patients reported by Kluijtmans et al. [1999] who, as noted above, concluded that these patients were, in reality, fully B6-responsive. The responsiveness of I278T is further supported by the data for I278T hetero- zygotes: in compound heterozygous form with G347S (a mutation that was non-responsive in a homozygote) the I278T allele seems to provide respon- siveness; and, when present in an wide variety of compound heterozygotes for which homozygotes for the other allele have not been classified for B6-respon- siveness, the I278T mutation is accom- panied by responsiveness. Additional mutations for which the indications of B6-responsiveness are relatively strong are R266K and perhaps A114V. G307S and T191M are unequivocally B6-non- responseive, with C162Y probably so (although less evidence is available). It is interesting that the only two compound same parents, and thus, that compound heterozygous CBS genotypes also deter- mine phenotypic B6 responsiveness.

That multiple mechanisms may be involved in responsiveness to B6 has been exemplified by Kruger and cow- orkers through comparisons of muta- tions R266K and I278T. Although homozygosity for either of these is clearly associated with B6-responsive- ness in humans, expression of human R266K in yeast with its endogenous CBS gene deleted promoted growth in cysteine-free medium, whereas expres- sion of human I278T did not [Kruger and Cox, 1995; Kim et al., 1997]. Purified R266K enzyme had reduced affinity for pyridoxal 5'-phosphate, whereas I278T did not. As detailed above, mice stably expressing human I278T as their sole CBS protein did not markedly lower their serum tHcy during administration of large doses of B6. Possible explanations for these differ- ences between R266K and I278T and between the behavior of I278T in humans and mice have been set forth and discussed [Chen et al., 2006]. Taken together, these lines of evidence suggest that responsiveness takes place via CBS itself, not by some alternative pathway, and depends upon the presence of at least minimal residual CBS activity; is determined chiefly by the particular mutations (or mutations) present; that there may be more than one mechanism of response; and that, at least for some mutations, response may require the intracellular environment usually pro- vided by human, but not mouse, tissue. However, despite the availability of the results of extensive and informative studies characterizing a wide variety of mutations, it appears that at this time no set of criteria have emerged upon which predictions of the B6 responsiveness of novel missense CBS mutations might be firmly based.

Newborn Screening

Screening of newborns in an effort to detect CBS-deficient patients has been carried out in many states of the USA and in other countries since the 1970s when evidence began to be available that early treatment prevented or mitigated the most severe clinical effects of CBS deficiency. Because of technical prob- lems in screening for elevation of homocysteine or its derivatives the screening has been based on detection of elevations of methionine. A compila- tion of the results for the first 20 or more years showed that the rates of detection ranged from 1:58,000 in Ire- land (where B6-non-responsive defi- ciency due to G307S is especially common), down to 1:1,000,000 in Scotland [Mudd et al., 1995a]. Among the major disadvantages of this approach is that it became clear that the majority of cases with B6-responsive deficiencies were being missed—for example, in an international survey, among the 55 patients discovered by newborn screen- ing, only 13% were B6-responsive, whereas in the total survey population 43.7% were B6-responsive [Mudd et al., 1985]. Had the large group of B6- responsive patients that, as recently realized and discussed above, are prob- ably not being currently ascertained, been added to the total of B6-responders in the total survey population, that discrepancy would probably have been far greater. Newborn screening pro- grams, for example, those in Taiwan and Galicia, Spain have recently lowered the cut-off values at which methionine elevations are flagged, thus possibly increasing the sensitivity for detection of CBS-deficient patients. It is still too early to judge whether the rate of detection of CBS deficiency will be increased, but it is already apparent that most mildly hypermethioninemic babies detected have deficiencies of clinically harmless MAT I/III activity rather than of CBS activity [Chien et al., 2008; Couce et al., 2008]. Screening for elevated tHcy would likely avoid these disadvantages and also serve to detect patients with homocysteine remethyla- tion disabilities that lead to elevated tHcy but low methionine. A reliable and rapid method to screen blood spots for tHcy was recently developed in Qatar, moti- vated by the uniquely high prevalence of B6-non-responsive CBS deficiency in natives of that country. During 3 years, 14 cases were detected among 46, 406 babies (1:3,289). Of these, 13 were homozygous for CBS R336C and therefore B6-non-responsive, and 7 had methionine values <65 mM (the 99.95% cut-off level). Thus, screening for methionine with a cut-off of 65 mM would have failed to detect 7 of 13 B6- non-responders. With respect to exten- sion of tHcy screening to areas in which CBS deficiency is less common, the
authors point out that a two-tier arrangement with initial screening for methionine with the cutoff lowered to 40 mM, followed by second tier testing for tHcy would have detected all the B6- non-responsive patients in question. Of all samples, only 0.5% would have passed to the second tier. Whether this approach would do equally well for B6-responders was left open [Gan- Schreier et al., 2010]. A two-tier approach for screening for homocys- teine remethylation disorders in which finding of low methionine in blood spots by tandem mass spectroscopy is followed by assay of tHcy is currently being evaluated in Minnesota [Tortorelli et al., 2010]. A similar approach using methionine elevation [as suggested by Gan-Schreier et al., 2010] seems now to be feasible, but only further experience will show if the overall results of such a program would be beneficial.

Management

Considerable experience with treatment of CBS deficiency is now available. As discussed in more detail elsewhere, B6 administration may be sufficient for responders. For B6-non-responders, dietary methionine restriction is the basic step taken to lower tHcy. Betaine administration is often used for non- responders [Mudd et al., 2001b], and can improve metabolic control when added to optimum dietary management [Singh et al., 2004]. Because betaine adminis- tration to CBS KO mice does not lower tHcy, or relieve their hepatopathy, caution should be used in extension of results obtained with such model animals to humans [Maclean et al., 2010].

By 1985, sufficient data were avail- able to indicate that methionine restric- tion initiated early in life prevented mental retardation in both B6-respon- sive and -non-responsive children; that either such treatment or administration of B6 to late-detected responders decreased the rate of occurrence of lens dislocation; and that B6 given to late- detected responders decreased the rate of initial thromboembolic events [Mudd et al., 1985]. Since then further experi- ence has extended the evidence by showing that: early-treated, B6-non- responsive, children who complied with methionine-restricted diets had full- scale IQ’s statistically significantly higher than their unaffected siblings [Yap et al., 2001c]; that early detected, metabol- ically well-controlled B6-non-respon- sive patients had no evidence of lens subluxation and all had bilateral 20/20 vision [Mulvihill et al., 2001]; and that early and good treatment significantly prevents osteoporosis and osteopenia [Yap et al., 1999]. A large multicenter study has reported results for 158 patients, with mean age 29.4 years (4.5–70) during 2,822 patient-years of treatment (average 17.9 years/patient). Based on comparison with the previ- ously published time-to-event curves for untreated patients [Mudd et al., 1985] during those years 112 vascular events were expected, but only 17 occurred, for a relative risk of 0.09 (95% CI 0.036–0.228; P < 0.0001) [Yap et al., 2001b].

These highly encouraging results were obtained in spite of the fact that the treatments used in the various partic- ipating centers lowered, but did not normalize the levels of patients’ homo- cysteine and its derivatives, the means of which were still several times higher than those in the respective normal popula- tions. As mentioned previously, recent evidence suggests that many homozy- gotes for the CBS I278T B6-responsive mutation may not be being ascertained, and that therefore the reference time-to- event curves, especially those for B6- responders [Mudd et al., 1985] are probably in need of revision [Skovby et al., 2010]. If this were done, the overall relative risk calculated in the multicenter study would probably be less favorable. However the fact that of the 2,822 patient-years covered in that report, 1,578 were for B6-non-responders [Yap et al., 2001b], suggests that the relative risks of vascular events for such patients are far less during tHcy-low- ering treatment, even if tHcy remains several times above the reference range. With regard to pregnancy in CBS- deficient women, evidence has been reported that with suitable management most such pregnancies, even those in B6-non-responders, may proceed nor- mally to term and usually result in normal babies [Yap et al., 2001a; Levy et al., 2002; Pierre et al., 2006].

A different and innovative possible mode of treatment for CBS deficiency, namely manipulation of the quality control system used by cells to deal with misfolded proteins, has been brought forward recently. The quality control systems in question consist of both molecular chaperones (heat-shock pro- teins) that prevent protein misfolding and aggregation by promoting refolding, and proteases that degrade misfolded protein molecules [Smith et al., 1998; Mayer and Bukau, 2005]. Initially it was found that osmolyte chemical chaper- ones, glycerol, trimethyamine-N-oxide, dimethylsulfoxide, proline, or sorbitol could restore the ability of yeast express- ing CBS I278T or three other CBS mutants to grow on cysteine-free media; and that I278T activity produced either in E. coli or in a coupled in vitro transcription/translation reaction was increased by addition of glycerol [Singh et al., 2007]. Members of the same group then found that, in yeast, increases in the activity or steady-state levels of I278T resulted from either (a) ethanol induc- tion of the heat-shock protein, Hsp70, that plays a positive role in proper I278T folding; or (b) from deletion of the gene encoding another heat-shock protein, Hsp26, that forms a complex with I278T that is degraded by a ubiquitin/ proteasome-dependent mechanism [Singh and Kruger, 2009]. These studies were then extended by showing that CBS activity could be restored in the CBS-deficient yeast carrying 17 of 18 disease-causing CBS mutations by expo- sure of the yeast to ethanol, proteasome inhibitors, or deletion of Hsp26. Pro- teasome inhibitors could also restore significant CBS activity to CBS alleles expressed in cultured fibroblasts derived from homocystinuric humans and in a mouse model expressing I278T [Singh et al., 2010]. Similarly, Kopecka´ et al. [2010] studied 27 mutations represent- ing 70% of known CBS alleles (only four of which overlapped with those studied by Singh et al. [2010]) and found that 14 responded by at least 30% increases in amounts of correctly assembled teramers and enzymatic activities upon expression in E. coli in the presence of at least one of the chaperones used (d-aminolevulinic acid, betaine, taurine, or glycerol). Eight of the mutant proteins studied by Kopecka´ et al. were purified and char- acterized by Majtan et al. after expres- sion in E. coli in the presence of chemical chaperones (ethanol, dimethylsulfoxide, or trimethylamine-N-oxide). The tetra- meric mutant enzymes fully saturated with heme had specific activities the same as, or higher than, that of the wild- type enzyme, although the responses to stimulation by heating or by AdoMet were variable [Majtan et al., 2010]. Taken together, these findings convinc- ingly show that misfolding is a prevalent cause of loss of activity in mutant CBS proteins and suggest that, if doses can be found of proteasome inhibitors, HSP70 inducing compounds, or chaperoning compounds that are both effective and can be used without adverse side-effects, the use of such compounds may provide useful and novel treatments for CBS deficiency [Kopecka´ et al., 2010; Singh et al., 2010].

TYROSINEMIA TYPE I (FUMARYLACETOACETATE HYDROLASE (FAH) DEFICIENCY)

FAH is the last enzyme in the pathway for tyrosine catabolism (Fig. 2). Defi- ciency of this activity leads to tyrosine- mia type I [Lindblad et al., 1977; Berger et al., 1981; Gray et al., 1981; Kvittingen et al., 1981], a severe disorder of child- hood that, if untreated, may lead to liver failure, neurological crises, rickets, and hepatocellular carcinoma. It has also been termed ‘‘hereditary tyrosinemia’’ or ‘‘congenital tyrosinosis.’’ For patients presenting before 6 months of age,forms of MAT separated chromato- graphically [van Faassen and Berger, 1990] that MAT III, the dimeric iso- zyme with the highest Km for methio- nine, is sensitive to inhibition by fumarylacetoacetae, whereas MAT I, the tetrameric isozyme is less so, and MAT II, the chief non-hepatic form, is insensitive [Berger, 1985]. In the pres- ence of such inhibition, it would be predicted that plasma AdoMet would not be elevated in the presence of elevated methionine as much as occurs in normals, but no report of such an assay has been published.

Pathophysiology

Although the pathophysiology of this deficiency is not completely under- stood, it seems likely that the metabolites accumulated upstream of the block at FAH may be important. These include maleylacetoacetate and fumarylacetate and their degradation products, succi- nylacetoacetic acid and succinylacetone [Mitchell et al., 2001]. As detailed above, fumarylacetate inhibition of MAT III may cause the hypermethioninemia. Succinylacetone may play a prominent role in episodes of porphyria-like peripheral neuropathy due to its pro- found inhibition of delta-aminolevu- linic acid dehydratase, thereby blocking heme biosynthesis [Sassa and Kappas, 1983]. The accumulation of tyrosine,itself, in FAH deficiency may be due to an unexplained secondary deficiency of 4-hydroxyphenylpyruvate dioyxgenase that leads to the accumulation of 4-hydroxyphenylpyruvate, a compound that exerts product inhibition of tyro- sine aminotransferase [Canellakis and Cohen, 1956].

Molecular Bases

The 26 mutations that had been identi- fied in FAH by 1997 were described in detail by St-Louis and Tanguay [1997]. Since then, at least 22 novel mutations have been reported [Bergman et al., 1998; Dreumont et al., 2001; Arranz et al., 2002; Heath et al., 2002; Cassiman et al., 2009; Park et al., 2009]. These mutations, spread over most of the 14 exons that encode the 419 amino acids that comprise the catalytic subunit, include 29 amino acid replacements, 5 nonsense codons, 4 deletions, and 10 putative splicing defects. Structural and functional analyses of several of the missense mutations have been reported [Bergeron et al., 2001].

Geographical/ethnic clustering of several of the FAH mutations has been found: Mutation, IVS12 + 5G>A, the prevalent one in the Saguenay-Lac-St- Jean region of northeast Quebec prov- ince in Canada with carrier rates ranging from 1/16 to 1/21 due to a founder effect [De Braekeleer and Larochelle,
1990], was found in many European patients and in one from Iran, indicating it is an ancient mutation. Other ethni- cally clustered mutations include W262X among ethnic Finns, Q64H among ethnic Pakistanis, and D233V among those of Turkish descent [St- Louis and Tanguay, 1997], IVS6nt-1(g- t) in patients of Mediterranean origin [St-Louis and Tanguay, 1997; Bergman et al., 1998; Arranz et al., 2002], and IVS12 + 5(g-a) in northwestern Europe [Bergman et al., 1998].

An unusual and interesting aspect of FAH deficiency was first suggested by Kvittingen and associates when they found that livers from FAH-deficient patients had mosaic patterns due to nodules staining immunohistochemi- cally for FAH. Liver samples from various areas contained fumarylacetoa- cetate activities ranging from 6 to 32 mmol/min in one patient and 0.4– 4 mmol/min in another (control range 33–66 mmol/min) [Kvittingen et al., 1993]. Subsequently, similar mosaic patterns were shown to be present in 15 of 18 livers from FAH-deficient patients of various ethnic origins, and for each of three differing mutations, in the FAH-positive regions a mutant AT nucleotide was shown to have reverted to a normal GC pair [Kvittingen et al., 1994]. In a cohort of 26 French- Canadian FAH-deficient patients, those classified as having more severe disease (based upon their having undergone liver transplantation before age 2 years) had a mean surface area of reverted tissue of 1.6%, whereas those classified as having less severe disease (having had liver transplantations at ages more than 6 years) had a mean surface of reverted tissue of 36%. Within reverted nodules hepatocytes had a normal appearance and showed no dysplasia. Hepatocellular carcinoma was present only in FAH- negative regions [Demers et al., 2003]. Thus, reversion of FAH mutations is a common event in FAH-deficient patients and the extent of the reverted tissue affects the severity of the clinical manifestations, offering an explana- tion of the facts that the condition is clinically very heterogeneous, even within affected sibships, and that no clear genotype– phenotype correlations have been observed. The high rate of reversion in FAH deficiency and its beneficial effects may reflect chance occurrences in a genetically unstable environment (due possibly to the metab- olites accumulating abnormally in FAH deficiency), followed by a net survival advantage of hepatocytes with at least a partial correction of their enzyme defi- ciency [Kvittingen et al., 1994; Russo et al., 2001].

Newborn Screening

Screening of newborns for FAH-defi- ciency has been in effect in Quebec since 1970. Initially screening was based on detection of elevated tyrosine and ele- vated a-fetoprotein, but this has evolved to assay of succinylacetone, used since 1997 as the first round screening step [Mitchell et al., 2001]. Recently MS/ MS screening for succinylacetone, acyl- carnitines, and amino acids has success- fully identified newborns [la Marca et al., 2009; Al-Dirbashi et al., 2010].

Management

Treatment of FAH deficiency has been reviewed recently [Ashorn et al., 2006; McKiernan, 2006; Santra and Baumann, 2008]. Treatment historically consisted of dietary restriction of phenylalanine and tyrosine and, when it became available in the 1980s, liver transplanta- tion. Since 1991 nitisinone (2-[2-nitro- 4-trifluoromethylbenzoyl]-1,3-cyclo- hexanedione) has been the treatment of choice. This compound inhibits 4- hydroxyphenylpyruvate dioxygenase, limiting the formation of homogentisate and the toxic metabolites formed from it, maleylacetoacetate, fumarylacetoace- tate, and succinylacetone, that accumu- late in the presence of FAH deficiency. It thus has beneficial effects on acute liver failure, chronic liver disease, develop- ment of hepatocellular carcinoma, renal tubular dysfunction, cardiomyopathy, and neurological disease. Dietary restric- tion of phenylalanine and tyrosine is also required. The major concern appears to be that this regimen may not completely prevent the development of hepatocellular carcinoma [Santra and Baumann, 2008].

POSSIBLE MALEYLACETOACETATE ISOMERASE (MAAI) DEFICIENCY

A case of possible maleylacetoacetate isomerase (otherwise known as gluta- thione transferase zeta, MAAI/GSTZ) [Ferna´ndez-Can˜o´n et al., 2002; Lim et al., 2004] deficiency was described in 1988 in a child with liver failure, renal tubular disease, psychomotor retarda- tion, and elevated levels of tyrosine (679 mM), methionine (999 mM), phe- nylalanine (350 mM), and cystathionine (38 mM) who died at age 1 year. FAH and 4-hydroxyphenylpyruvate dioxyge- nase activities were normal in a liver biopsy, but MAAI activity was not detected in a (presumably post-mortem) liver sample and was very low in cultured skin fibroblasts. Succinylacetone was not detected in his urine [Berger et al., 1988], but in the presence of normal FAH activity that compound might be removed by hydrolysis [Ferna´ndez- Can˜o´n et al., 1999]. However, several limitations cause lingering doubt as to whether or not this was indeed a case of MAAI deficiency [Ferna´ndez-Can˜o´n et al., 1999; Mitchell et al., 2001]: (a) molecular studies of the boy were not reported (the MAAI gene was cloned only 10 years later in 1998 [Ferna´ndez- Can˜o´n and Pen˜alva, 1998]); (b) no inactivating mutations of MAAI were found in four cases of unexplained severe infantile liver disease with similarities to FAH deficiency [Ferna´ndez-Can˜o´n et al., 1999]; and (c) mice with MAAI knocked out remained generally clin- ically well on normal diets [Ferna´ndez- Can˜o´n et al., 2002; Lim et al., 2004].

CITRIN DEFICIENCY

Citrin is a mitochondrial solute carrier protein active as an aspartate– glutamate exchanger. It was first identified by Kobayashi et al. [1999] during a genome-wide search using homozygos- ity mapping to determine the genetic cause of the form of citrullinemia termed adult-onset type II citrullinemia (CTLN2), a condition characterized clinically by onset at ages 11–79 years of sudden disturbances of consciousness, behavioral aberrations, disorientation, loss of memory, floppy tremor, drowsi- ness, and possibly coma [Saheki and Kobayashi, 2002; Saheki et al., 2002]. Mutations in a novel gene, SLC25A13, located at chromosome 7q21.3, were identified as the cause, and the gene was shown to encode a protein, citrin, the sequence of which indicated it to be a member of the mitochondrial solute- carrier family [Kobayashi et al., 1999]. Shortly thereafter the clinical spectrum recognized for citrin deficiency ex- panded when homozygosity or com- pound heterozygosity for mutations in the same gene, SLC25A13, were identified in three infants: a boy inves- tigated because of elevated methionine (134 mM) on neonatal mass screening; his younger brother with evidence of liver dysfunction on day-of-life 3; and a girl with hypergalactosemia detected during neonatal screening. At ages 25– 36 days, each were hypermethioninemic (229 –465 mM; reference range 13– 32 mM) with elevated plasma citrullines (286 –547 mM; reference range 5– 37 mM), and elevated bilirubin levels and other findings suggestive of intra- hepatic cholestasis [Ohura et al., 2001]. Furthermore, compound heterozygos- ity for SLC25A13 mutations was reported in three additional children in whom the only abnormality during newborn screen was elevated galactose in one, but each of whom had been noticed to be jaundiced at ages 39–147 days. Studies prompted by the jaundice showed elevated methionine, 426 mM, in only one of these children [Tazawa et al., 2001]. The form of citrin deficiency presenting clinically in neo- nates and due to SLC25A13 mutations then came to be known as neonatal intrahepatic cholestasis caused by citrin deficiency (NICCD) [Saheki et al., 2002]. The relationship between NICCD and CTLN2 became more clear when Tomomasa et al. studied a boy who, as an infant had elevated bilirubin, alkaline phosphatase, and g- glutamyl transpeptidase levels; who had citrulline levels of 48 mM (reference range 11– 37 mM) at 6 months, 208 mM at age 12 years, and 815 mM at age 16 years; and at age 16 years suddenly developed behavioral aberrations in- cluding shouting and violence and was found to be a homozygote for an SLC25A13 mutation [Kasahara et al., 2001; Tomomasa et al., 2001]. With further experience, it is now generally considered that citrin (SLC25A13) deficiency manifests itself in two clinical forms: the infantile form, NICCD, and the late-onset form, CTLN2. Usually NICCD is not severe and symptoms disappear within a year, although a few [perhaps 3% Yeh et al., 2006] have developed liver failure requiring liver transplantation at 10–12 months of age. However, some apparently recovered NICCD patients develop severe CTLN2 a decade or several decades later, although, which ones and what proportion of them will do so remains unclear [Saheki and Kobayashi, 2002; Kobayashi et al., 2003; Lu et al., 2005]. Of course, some citrin-deficient patients may remain undiagnosed, yet go on to adverse outcomes. Citrin deficiency is found most often in patients from Asia. Many aspects of the condition have been reviewed extensively in recent publica- tions [see, e.g., Tabata et al., 2008; Kobayashi and Saheki, 2008].

Metabolite Abnormalities: Occurrence and Cause

Hypermethioninemia is frequent in NICCD, but not in CTLN2. Elevations of methionine may be accompanied by elevations of galactose, citrulline, threo- nine, tyrosine, phenylalanaine lysine, and arginine. Table IV lists plasma methionine values reported for infants from several countries of Asia: Japan [Tamamori et al., 2004; Ohura et al., 2007], Korea [Ko et al., 2007a,b], China [Song et al., 2009a], Taiwan [Yeh et al., 2006], and the patients of various ethnicity (French-Canadian, South-East Asian, Han, Arabic, Pakistani, and Northern European) identified by Dimmock et al. [2009]. The elevations of methionine may be only moderate, or quite severe, but are not present in all cases, either during newborn screening or later. Methionine may be normal on newborn screen, but rise later to abnormal levels. As mentioned above, in most cases the methionine values, as well as those of the other elevated metabolites tend to normalize by about a year of age, although, as mentioned above, a portion of the patients do go on to develop clinically adverse outcomes.In spite of the extensive studies of the pathophysiology of citrin deficiency (see below), to the best of the knowledge of the present author no hypothesis as to the cause of the transient hypermethio- ninemia so often characteristic of NICCD has been published.

Pathophysiology

Citrin is a Ca2+-binding protein located on the inner membrane of mitochon- dria, facing the intermembrane space. Although present chiefly in post-natal liver, this protein is found also in many other tissues. Citrin catalyzes the exchange of mitochondrial aspartate for glutamate plus a proton from the cytosol [Palmieri et al., 2001]. The metabolic roles of this exchange in the urea cycle and in glycolysis and lactate metabolism have been discussed in detail [Satru´stegui et al., 2010] and elucidated to some extent by mouse models in which either citrin [Sinasac et al., 2004], or both citrin and glycerol-3-phosphate dehydrogenase [Saheki et al., 2007] have been knocked out. In brief, aspartate from the mitochondria is used by cytosolic argininosuccinate synthetase to form argininosuccinate by condensa- tion of citrulline with aspartate, an important step in the urea cycle. In citrin deficiency, the defective export of aspartate contributes to the elevations of citrulline and ammonia. Citrin is needed also for the malate–aspartate shuttle which serves to inport NADH into the mitochondria and consequently citrin deficiency interferes with gluconeogen- esis from reduced substrates. Liver damage in NICCD is indicated by raised serum total bile acids and conjugated bilirubin with mild elevations of serum transaminases, and decreases of total serum protein and vitamin K-dependent coagulation factors [Tamamori et al., 2004; Tazawa et al., 2004; Yeh et al., 2006; Ohura et al., 2007]. Histologically, the livers of patients with NICCD show a combination of mixed macrovesicular and microvesicular fatty hepatocytes together with fatty liver, necroinflam- matory reaction, and iron deposition. It was suggested that these histological changes are characteristic enough to be useful in establishing a diagnosis of NICCD in cases of infantile cholestasis [Kimura et al., 2010].

Molecular Bases

As of 2008, 52 mutations had been identified in SLC25A13 [Kobayashi et al., 2008]. These included not only individuals from Japan [Kobayashi et al., 1999; Yasuda et al., 2000; Yamaguchi et al., 2002; Takaya et al., 2005], but also from Taiwan [Lee et al., 2006; Tsai et al., 2006; Yeh et al., 2006; Sheng et al., 2010], Korea [Ko et al., 2007a,b], and China [Song et al., 2008, 2009a,b] so that citrin deficiency is particularly prevalent in East Asia [Kobayashi et al., 2003; Saheki et al., 2004; Lu et al., 2005; Tabata et al., 2008]. Regional specificity of the mutation types within East Asia has been noted [Kobayashi et al., 2008; Tabata et al., 2008]. In Japan, based on a carrier rate of 1:65, the frequency of homozygotes and com- pound heterozygotes has been estimated to be 1:17,000, almost the same as the incidence of NICCD (1:17,000– 34,000), and higher than the incidence of CTLN2 (1:100,000–230,000), indicating that most individuals with NICCD remain healthy without devel- oping CTLN2 [Tabata et al., 2008]. Citrin-deficiency has also been identi- fied in Israel [Ben-Shalom et al., 2002; Luder et al., 2006], the United Kingdom (including a child of Pakistani origin) [Hutchin et al., 2009], and in the United States [Dimmock et al., 2007], and the Czech Republic [Tabata et al., 2008].

Newborn Screening

In an attempt to ascertain the optimal means of screening newborns for citrin deficiency, Tamamori and coworkers assayed bile acids, galactose, and carried out amino acid chromatography on dried blood spots obtained during new- born screening from 20 children proven to have citrin deficiency. Citrulline values in 19 of 20 such children were more than 2 SD above the values in control children To offset the influence of a reduction in total blood amino acids, these authors recommended refinement of the method by using the ratio of citrulline to serine or leucine plus isoleucine. Use of that improvement would have permitted detection of all the cases of NICCD in question [Tamamori et al., 2004].

MITOCHINDRIAL ABNORMALITIES— POSSIBLE CAUSES OF HYPERMETHIONINEMIA

The author has participated in studies of two children with different and fatal mitochondrial abnormalities who had greatly elevated plasma methionines, in these individuals accompanied by severely elevated plasma AdoMet con- centrations. Further studies showed these findings were not attributable to any of the known causes of hyper- methioninemia discussed above (S.H. Mudd, C. Wagner, Z. Luka, S. Stabler, unpublished observations). A trans- porter has been described that imports AdoMet from the cytoplasm, where it is synthesized, into mitochondia [Horne et al., 1997], and there are a variety of mitochondrial AdoMet-dependent methyltransferases in humans (http:// www.broadinstitute.org/pubs/MitoCarta/ human.mitocarta.html). Taken together with the frequent early hypermethioni- nemia of citrin deficiency, a possible working hypothesis is that in some instances of mitochondrial abnormal- ities (the specificity and frequency of which remain to be characterized). AdoMet, therefore tends to accumulate in cytoplasm and plasma, and secondarily causes hypermethioninemia (as occurs with the elevated AdoMet of GNMT and AHCY deficiencies). AdoMet so derived from non-hepatic tissues might have limited access to the liver and so would not be disposed of in the usual manner by GNMT, located chiefly in that organ.

LIVER DISEASE

Methionine Concentrations in Plasma

Hypermethioninemia has long been known to occur in liver disease. Com- piled in Table V are data reported after quantitation of amino acids by column chromatography became available, listed in descending order according to the mean plasma methionine concentration of each patient group. Although these data clearly show a tendency for elevated levels of plasma methionine to occur in liver disease, they show also that in some instances the mean values are not significantly raised above the control means—even though some values in the patients are far above the upper limit of the ranges for control subjects. Thus the hypermethioninemia occurs only in a subset of patients. For example, even in the severely affected patients studied by Wu et al. [1955], two of seven in deep coma and two of four in hepatic precoma had normal values; and among the cirrhotic patients studied by Kinsell et al. [1948] [data not included in Table V because neither a mean nor range for liver patients is shown in the article] over 90% of the methionine values for liver patients fell within the designated control range (upper limit 99 mM). By contrast, and illustrating the variability of the methionine values, only 3 of the 14 alcoholic or post- hepatitic cirrhotic patients studied by Horowitz et al. [1981] had methionine concentrations in the normal range. Methionine is not the only amino acid often to have an altered plasma concen- tration in liver disease. As discussed by Morgan et al. [1982], and exemplified by the extensive data in their article, typical (but not invariant) changes are increased concentrations of tyrosine and/or phenylalanine together with increased methionine, and decreases in valine, isoleucine, and leucine. In more serious liver disease mean cystathionine is above normal in urine [Ma˚rtensson et al., 1992] and in serum [Look et al., 2000; Medici et al., 2010], and mean tHcy is slightly raised [Medici et al., 2010].

Pathophysiology

A plausible explanation for the tendency for plasma methionine to be elevated in liver disease was suggested by studies by Kinsell et al. of the kinetics of changes in plasma methionine after an intravenous load of methionine. In normals, max- imum plasma levels were attained within 15 min of the load, and between 30 and 180 min steady declines occurred in the plasma concentrations. In patients with liver disease, peaks were reached at similar times, but the rates of decrease between 30 and 180 min were slower. Because urinary excretion of methio- nine contributed only insignificantly to the declines, it was suggested that metabolic utilization accounted for the declines and that patients with liver damage were, on average, utilizing methionine at rates equal to only 58% of the rate in controls, a statistically highly significant decrease [Kinsell et al., 1948]. Subsequently, Horowitz et al. also found decreases in the rates of removal of methionine from plasma in cirrhotic patients following oral methionine loads. Expressed in terms of half-life (t1) for elimination of plasma 2 methionine, normals had times of 146 10 min, whereas cirrhotic patients had times of 458 81 min. Likewise, following loading by a two-step prime- continuous infusion of methionine, cirrhotic patients had t1’s prolonged to 282 90 min compared to 187 25 in
normals. The prolongations correlated (r ¼ 0.715) with the severities of the liver disease judged by the Child– Pugh scores [Marchesini et al., 1992]. In the above studies, doses of methionine large enough to markedly raise the baseline plasma methionine concentrations were administered. To estimate the kinetics with methionine concentrations main- tained at basal levels, Russmann et al. [2002] administered to seven control subjects and six patients with alcohol- induced cirrhosis an oral dose of L-[2H3-methyl-1-13C]methionine small enough that the plasma concentrations of the subjects remained unchanged. As shown in Table V, the patients had a mean plasma methionine level twice that of the controls. The rate of transsulfura- tion was then followed by the produc- tion of 13CO2 [produced from the [1-13C]a-ketobutyrate formed from [13C] cystathionine; see Fig. 1]; that of remethylation, by replacement of the labeled methyl group by an unlabelled one. The results were interpreted as showing that the amount of methionine passing through the complete methio- nine-AdoMet-AdoHcy-homocysteine- methionine cycle was about the same in the controls and patients, with the rate being driven in the patients by the higher concentration of methionine. On the other hand, the rate of transsulfuration was estimated to be markedly impaired in the cirrhotic patients. Taken together with the fact that the activity of MAT in liver of cirrhotics is decreased to 38– 56% of control activity [Martin Duce et al., 1988] due specifically to loss of the tetrameric form, MAT I, with the lower Km for methionine [Cabrero et al., 1988], these studies provide a basis for understanding the elevations (usually relatively small) of plasma methionine in the less serious forms of liver disease: the decrease in MAT I will tend to decrease the flux from methio- nine to AdoMet, but the flux may be restored by an increase in the concen- tration of methionine. That formulation would fit with the fact that the mean concentration of AdoMet in livers of the six cirrhotics studied by Cabrero et al. [1988], 17.8 3.1 mM, was the same as that in six controls,17.3 2.6 mM. Mild methionine elevations with the specified liver MAT activities of 38–56% of normal would fit also with the facts that of 13 hetero- zygotes for inactivating mutations in MAT1A all but two had plasma methio- nine values within the reference range (13 –43 mM) [Ferna´ndez-Irigoyen et al., 2010]. Further information as to the effect of partially deficient MAT I/III activity is provided by the fact that plasma methionines ranged from 45 to 400 mM (mean 188 mM) among 28 heterozygotes for the Mendelian dominant R264H MAT1A mutation [Mudd et al., 2001b]. These individuals have been estimated to have perhaps 30% of normal activity [Chamberlin et al., 1997].

Additional mechanisms may well be involved in the decreased rate of trans- sufuration documented in cirrhotics by Russmann et al. [2002], and in the marked/extreme elevations of plasma methionine present in at least some patients with the most severe forms of liver insufficiency (Table V). For exam- ple, Avila and coworkers reported that in 26 cirrhotic livers mRNA for MAT1A was very low or non-detectable in 31% and below normal in a further 58%, so that only 11% had normal levels. MAT2A was not induced in any of these livers. For GNMT, methionine synthase, BHMT, and CBS, mRNA concentra- tions were normal in only 4%, 0%, 11%, and 27%, respectively. Cystathionine gamma-lyase activity may also be defi- cient [Ma˚rtensson et al., 1992; Look et al., 2000; Medici et al., 2010]. Patients with severe loss of one type of mRNA tend to have severe losses of the other types [Avila et al., 2000], so that the abnormalities of sulfur amino acid metabolism in question may usually be explained by one-or- another combination of these enzyme activity deficiencies.

Probable Transient and/or Non-Genetic Cases

It has been known for more than 40 years that even normal-birth-weight, full- term infants whose dietary intakes of methionine are unusually high become hypermethioninemic. For example, Snyderman et al. [1968] reported that milk protein intakes of 9 g/kg/day led to methionine elevations of 2–35 times normal; and Levyet al. [1969] found that protein intakes of 7 g/kg/day led to plasma methionines of 315– 1,206 mM. Very low-birth-weight babies even on normal diets may have plasma methio- nines over 67 mM [Zytkovicz et al., 2001], and such babies [Valman et al., 1971; Gaull et al., 1977; Ten Hoedt et al., 2007], or some with presently undefined abnormalities [Mudd et al., 2003], when taking in higher amounts of protein or methionine are especially likely to become hypermethioninemic.

Mild hypermethioninemia that is transient and decreases to within the normal range within several months may occur in normal babies fed normal diets—for examples, see Tsuchiyama et al. [1982]. However, it is noted that most such cases have not been sequenced for MAT1A heterozygosities that may produce a rather similar sequence—for example, P357L [Ferna´ndez-Irigoyen et al., 2010].

Adverse Effects of Extreme Hypermethioninemia

Although elevations of methionine to the levels that usually occur in the situations covered in this review seem not to cause serious adverse clinical effects, very extreme elevations may indeed have such effects. For example,as detailed by Braverman et al. [2005] brain edema with abnormal brain MRI findings developed in three CBS-defi- cient children and an MAT I/III- deficient boy whose plasma methionines rose to at least 960 –3,000 mM when given betaine, and two of the infants whose methionines rose to 2,154 and 6,830 mM due to excessive dietary methionine intakes also had brain edema or MRI abnormalities typical of such edema. The CNS abnormalities in these cases disappeared when plasma methio- nine levels were lowered [Mudd et al., 2003]. Furthermore, a fatal episode involving extreme hypermethioninemia occurred in an adult woman participat- ing in an experimental study that involved administration of an acute oral dose of methionine of 100 mg/kg. Extensive experience has shown that that dose is usually well tolerated [Gra- ham et al., 1997; Boers, 1998; Krup- kova´-Meixnerova´ et al., 2002]. The available evidence suggests that this woman was erroneously given a dose perhaps 10 times that high. Within hours her plasma methionine rose to 5,760 mM, far higher than usually occurs in comparable subjects given 100 mg/kg (e.g., 1,107 53 mM) and her plasma AdoMet rose to 1,089 nM. She became demented, apneic, and pulseless, and died days later as a result of aspiration pneumonia [Cottington et al., 2002]. Taken together, these experiences warn that extraordinarily high plasma methio- nine levels should be avoided. Although the precise level is not clearly established, and it is possible there is a synergistic effect between betaine and methionine in this regard, caution suggests avoiding methionine concentrations of much more that 1,000 mM S-Adenosyl-L-homocysteine (certainly 2,000 mM) and being alert for neuro- logical abnormalities if such levels are attained.